Equity and Development - World Bank€¦ · Equity and Development ©2005 The International Bank for Reconstruction and Development / The World Bank ... reverse because of growth

2006world development report

Equity and Development

32204P

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

ed



A copublication of The World Bank

and Oxford University Press



©2005 The International Bank for Reconstruction and Development / The World Bank1818 H Street NWWashington DC 20433Telephone: 202-473-1000Internet: www.worldbank.orgE-mail: [email protected]

All rights reserved

1 2 3 4 08 07 06 05

A copublication of The World Bank and Oxford University Press.

Oxford University Press198 Madison AvenueNew York NY 10016

This volume is a product of the staff of the International Bank for Reconstruction and Devel-opment / The World Bank. The findings, interpretations, and conclusions expressed in thispaper do not necessarily reflect the views of the Executive Directors of The World Bank or thegovernments they represent.

The World Bank does not guarantee the accuracy of the data included in this work. Theboundaries, colors, denominations, and other information shown on any map in this work donot imply any judgement on the part of The World Bank concerning the legal status of any ter-ritory or the endorsement or acceptance of such boundaries.

Rights and PermissionsThe material in this publication is copyrighted. Copying and/or transmitting portions or all ofthis work without permission may be a violation of applicable law. The International Bank forReconstruction and Development / The World Bank encourages dissemination of its work andwill normally grant permission to reproduce portions of the work promptly.

For permission to photocopy or reprint any part of this work, please send a request withcomplete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers,MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com.

All other queries on rights and licenses, including subsidiary rights, should be addressed tothe Office of the Publisher, The World Bank, 1818 H Street, NW, Washington, DC 20433, USA;fax: 202-522-2422; e-mail: [email protected].

ISBN-10: 0-8213-6249-6ISBN-13: 978-0-8213-6249-5ISSN: 0163-5085eISBN: 0-8213-6250-XDOI: 10.1596/978-0-8213-6249-5

Cover image: Dream of a Sunday Afternoon in Alameda Park 1947–48 (fresco) by Diego Rivera.The mural is located in Museo Mural Diego Rivera, Mexico City. Reproduction authorized by the Instituto Nacional de Bellas Artes y Literatura–Mexico; Copyright © Photograph byFrancisco Kochen.

Library of Congress Cataloging-in-Publication Data has been applied for.

Contents

v

Foreword xiAcknowledgments xAbbreviations and Data Notes x vi

Overview 1Inequity within and across nations 4

Why does equity matter for development? 7

Leveling the economic and political playing fields 9

1 Introduction 18Equity and inequality of opportunity: the basic concepts 18

Inequality traps 20

A brief preview of the Report 23

Inequity within and across countries 25

focus 1 on Palanpur 26

2 Inequity within countries: individuals and groups 28Inequalities in health 29

Inequalities in education 34

Economic inequalities 36

The relationship between group differences and inequality 43

Agency and equity: inequalities of power 48

The inequality trap for women 51

3 Equity from a global perspective 55Examples and concepts 55

Global inequalities in health 56

Global inequalities in education 60

Global inequalities in income and expenditure 62

Part I

iii

vi CONTENTS

Global inequalities in power 66

A glimpse of the future 68

focus 2 on empowerment 70

Why does equity matter? 73

4 Equity and well-being 76Ethical and philosophical approaches to equity 76

Equity and legal institutions 78

People prefer fairness 80

Income inequality and poverty reduction 84

5 Inequality and investment 89Markets, wealth, status, and investment behavior 89

The evidence on underinvestment 96

Inequalities and investment 101

focus 3 on Spain 106

6 Equity, institutions, and the development process 107The distribution of power and institutional quality:circles vicious and virtuous 107

Institutions and political inequality matter for development:historical evidence 109

Institutions and political inequality matter for development:contemporary evidence 113

Transitions to more equitable institutions 118

Conclusion 124

focus 4 on Indonesia 126


7 Human capacities 132Early childhood development: a better start in life 132

Basic education: expanding opportunities to learn 135

Toward better health for all 141

Social protection: managing risk and providing social assistance 148

Summary 155

Part III

Part II

Contents vii

8 Justice, land, and infrastructure 156Building equitable justice systems 156

Toward greater equity in access to land 162

Providing infrastructure equitably 168

Summary 175

focus 5 on taxation 176

9 Markets and the macroeconomy 178How markets relate to equity 178

Achieving equity and efficiency in financial markets 179

Achieving equity and efficiency in labor markets 185

Product markets and trade reform 193

Macroeconomic management and equity 198

focus 6 on regional inequality 204

10 Achieving greater global equity 206Making global markets work more equitably 207

Providing development assistance to help build endowments 218

Transitions to greater equity 221

Summary 223

focus 7 on drug access 224

Epilogue 226

Bibliographic note 231

Endnotes 233

References 247

Selected Indicators 275

Measuring Equity 277

Selected world development indicators 289

Index 309

viii CONTENTS

Figures1 Wealth matters for the immunization of children 5

2 Opportunities are determined early 6

3 Life expectancy improved and became more equal—untilthe onset of the AIDS crisis 6

4 A long-run diverging trend in income inequality begins toreverse because of growth in China and India 7

5 Children’s performance differs when their caste is madesalient 8

2.1 Unequal opportunities persist across generations in Brazil 29

2.2 Unequal assets, unequal opportunities: AIDS orphans inSouthern Africa 33

2.3 Health improvements and greater health equity in Peru 34

2.4 Child test scores in Ecuador: the role of wealth, parentaleducation, and place of residence 35

2.5 Beware of intercountry comparisons of inequality! 38

2.6 Revisiting the Kuznets hypothesis for economic growth andinequality 44

2.7 Inequitable agencies and institutions in Pakistan 48

2.8 Legacies of discrimination and the reproduction of inequalities and poverty among the Batwa in Uganda 49

2.9 Sex ratios and “missing women” 51

3.1 Three competing concepts of inequality: global,international, and intercountry 57

4.1 A simple representation of different concepts of equity 78

4.2 Capuchin monkeys don’t like inequity either . . . 82

4.3 Worker perceptions of unfairness, product quality, andconsumer safety 83

6.1 Banking in the nineteenth century, Mexico and the UnitedStates 109

6.2 Growth with poor institutions does not last 113

6.3 Polarization, conflict, and growth 118

6.4 Aiding equitable growth in early modern Britain:the role of the Poor Laws 120

7.1 ECD programs are an essential ingredient for theattainment of education for all 134

7.2 School fees—an instrument of exclusion or accountability? 137

7.3 Desegregating Roma schools in Bulgaria:the Vidin model 138

7.4 Remedying education: the Balsakhi program in India 140

7.5 School vouchers: efficient and equitable? 141

7.6 Working with mothers to treat malaria 142

7.7 Poor people and ethnic minorities receive lower-quality care 143

7.8 Better maternal health in Malaysia and Sri Lanka 144

7.9 Mobilizing support for universal coverage in Thailand 146

7.10 Public works programs: key issues 152

7.11 Africa’s orphans and public action 155

8.1 Increasing legal literacy and public awareness:“My Rights” on Armenian public television 157

8.2 Affirmative action in India and the United States 158

8.3 State frameworks and customary institutions in South Africa 160

8.4 The impact of legal aid in Ecuador 160

8.5 Bogota, Colombia: civic culture program 161

8.6 Land reform in South Africa: picking up steam 164

8.7 Clarifying how customary rights fit with formal systems 166

8.8 Land and output tax combinations 168

8.9 Lagging infrastructure in Africa 170

8.10 The distributional impact of infrastructure privatization inLatin America: a mixed bag 171

8.11 The pro-poor agenda for urban water in Senegal 173

8.12 Addressing accountability and transparency intelecommunications in Brazil and Peru 174

9.1 Markets and development: policy, equity, and social welfarein China 180

9.2 Too much and too little regulation: Russia before and afterthe transition 182

9.3 Organizing in the informal economy 190

9.4 Employment protection legislation 191

9.5 Two cases of labor market reform: One comprehensive,one partial 192

9.6 Did the Russian 1998 crisis have equitable consequences? 201

10.1 International law, globalization, and equity 207

10.2 Making migrant worker schemes more developmentfriendly 210

10.3 Cotton subsidies are huge—and tenacious 212

10.4 Will improved working conditions in Cambodia’s textileindustry survive the end of the quota system? 214

10.5 Expanding access to antiretroviral drugs in South Africa 215

Boxes

6 Catching up through early interventions 11

7 Better to be close to economic opportunities 15

1.1 The interaction of political, economic, and socioculturalinequalities 20

2.1 Infant mortality varies across countries but also by mother’seducation within countries 30

2.2 Stunting levels of children born in rural versus urban areasare far from the same 30

2.3 Access to childhood immunization services depends onparents’ economic status 31

2.4 Stunting and underweight in Cambodia 33

2.5 Education levels vary across countries, but they also dependon gender of household head 36

2.6 Education levels vary by country and between rural andurban sectors 37

2.7 The share of inequality in years of schooling attributable to differences between males and females has been declining 37

2.8 Market capitalization controlled by the top 10 families inselected countries, 1996 38

2.9 Africa and Latin America have the world’s highest levels ofinequality 39

2.10 Between-group inequality decompositions: social group ofthe household head 40

2.11 Between-group inequality decompositions: education of thehousehold head 41

2.12 Location, education, and social groups can make adifference: regressions of total inequality on shares ofbetween-group inequality of different householdcharacteristics 43

2.13 Women work longer hours than do men 53

3.1 Vanishing twin peaks in life expectancy at birth 58

3.2 Life expectancy is highly correlated with income,particularly in poor countries 59

3.3 The distribution of years of schooling improved greatly inthe second half of the twentieth century 60

3.4 Mean years of schooling increased while inequality declinedacross birth cohorts 60

3.5 Gender disparities in years of schooling declined butremained significant in some regions 61

3.6 Incomes range broadly across countries and individuals 62

3.7 Since 1950, intercountry inequality increased whileinternational inequality declined 63

3.8 Unlike relative inequality, absolute inequality has beensteadily increasing 63

3.9 The inequality decline between countries was neutralized byincreases within countries 64

3.10 Inequality between countries became much moreimportant over the long run 65

3.11 Absolute poverty declined globally, but not in every region 66

3.12 There is no one-to-one relationship between voice andincome 67

4.1 The distribution of observed offers in ultimatum games 80

4.2 Views on inequality from the World Values Survey 84

4.3 Growth is the key to poverty reduction . . . 85

4.4 . . . and, on average, growth is distribution-neutral 85

4.5 The national growth incidence curves for Tunisia1980–1995 and Senegal 1994–2001 86

4.6 Greater inequality reduces the power of growth to reducepoverty 87

5.1 In rural Kerala and Tamil Nadu, the rich access most of thecredit and pay relatively low rates 90

5.2 Children’s performance differs when their caste is madepublic 96

5.3 Returns to capital vary with firm size: evidence from smallMexican firms 97

5.4 Inefficient allocation of resources; the example of theGounders vs. the outsiders 98

5.5 Average returns for switching to pineapples as an intercropcan exceed 1,200 percent 98

5.6 Profit-wealth ratios are highest for the smallest farms 99

6.1 Countries with more secure property rights have higheraverage incomes 108

6.2 Low population density in 1500 is associated with a lowerrisk of expropriation today 110

6.3 Worse environments for European settlers are associatedwith worse institutions today 110

6.4 A worse environment for settlers is associated with fewerconstraints on the executive at independence 111

6.5 Constraints on the executive are greater in Mauritius thanin Guyana 116

6.6 GDP per capita is rising in Mauritius, not in Guyana 116

6.7 Inequality in Britain began to fall around 1870 121

7.1 Children from better-off households have a big edge incognitive abilities by age three 133

7.2 Early childhood interventions are good investments 133

7.3 Catching up through early intervention 134

7.4 Boosting enrollments is not enough to overcome thelearning gap 139

7.5 Almost all countries spend more on social insurance thanon social assistance (percent of GDP) 149

8.1 Unequal initial land distributions go together with slowereconomic growth 163

8.2 Title to land increases investment and access to credit 165

Contents ix

x CONTENTS

8.3 Poor families did not benefit from an expansion of access inAfrica 172

8.4 Poorer households have lower-quality water and pay morein Niger 173

9.1 Poland’s stock market started slowly but then surpassed theCzech Republic’s 185

9.2 Patterns of employment and unemployment vary widelyacross African countries 187

9.3 Different labor market institutional setups can yield equallygood productivity growth paths: Scandinavia versus theUnited States 188

9.4 It’s better for household welfare to be close to economicopportunities 195

9.5 Weaker institutions are associated with macroeconomicvolatility and crises 199

9.6 Labor shares fall during crises and don’t fully recoverafterward 200

9.7 In Argentina, the wealthy had a way out during the crisis 201

10.1 Wage differentials are substantially larger today than at theend of the nineteenth century 208

10.2 More subsidies than aid 220

Tables2.1 Decomposition of inequality between and within

communities 42

2.2 Percentage of women who have ever experienced physical orsexual violence by an intimate partner 54

3.1 Increases in life expectancy at birth slowed downdramatically in the 1990s 58

3.2 Mean years of schooling increased continuously whileinequality declined 61

3.3 Mobility matrix in absolute country per capita incomes,1980 to 2002 66

5.1 The effect of income shocks on consumption,Côte d’Ivoire 92

5.2 Farm size productivity differences, selected countries 99

7.1 Examples of social protection programs 150

7.2 Targeting performance of conditional transfer schemes 153

9.1 Two pathologies in the interaction between equity andgrowth 179

9.2 Financial policy and institutions are often captured by thefew: case study evidence 181

9.3 Fiscal costs of selected banking crises 200

10.1 ODA as a share of GNI, 2002, 2003, and simulation for 2006 220

A1 Poverty 278

A2 Income/consumption inequality measures 280

A3 Health 282

A4 Education 284

Classification of economies by region and income 291

1 Key indicators of development 292

2 Poverty and income distribution 294

3 Economic activity 296

4 Trade, aid, and finance 298

5 Key indicators for other economies 300

Poverty reduction comes about through individuals, families and communities taking advan-tage of the opportunities available to them by working, investing and innovating to better theirlives. But we live in a world of extraordinary inequalities in opportunity, both within and acrosscountries. Even the basic opportunity for life itself is disparately distributed: whereas less thanhalf of one percent of children born in Sweden die before their first birthday, this is the case forclose to 15% of all children born in Mozambique. Within El Salvador, the infant mortality rateis 2% for children of educated mothers, but 10% for those whose mothers have no schooling.In Eritrea, immunization coverage is close to 100% for children in the richest fifth of the popu-lation, but only 50% for the bottom fifth.

These children can not be blamed for the circumstances into which they are born, yet theirlives—and their ability to contribute to the development of their nations—are powerfullyshaped by them. That is why the World Development Report 2006, the twenty-eighth in thisannual series, looks at the role of equity in the process of development. Equity is defined interms of two basic principles. The first is equal opportunities: that a person’s life achievementsshould be determined primarily by his or her talents and efforts, rather than by pre-determinedcircumstances such as race, gender, social or family background. The second principle is theavoidance of deprivation in outcomes, particularly in health, education and consumption levels.

For many if not most people, equity is of intrinsic importance as a development goal in itsown right. But this report goes further, by presenting persuasive evidence that a broad sharingof economic and political opportunities is also instrumental for economic growth and devel-opment. This is for economic reasons, because greater equity can lead to a fuller and more effi-cient use of a nation’s resources. It is also for political and institutional reasons: excessiveinequalities in power and influence can lead to political, social and economic institutions thatare less conducive to long-term growth. Few today’s prosperous societies, if any, developed byexcluding the majority of their people from economic and political opportunities.

The implication of this message for the work of the World Bank and others in the develop-ment community is that a focus on equity should be a central concern in the design and imple-mentation of policy for development and growth. This insight needs to be integrated into bothanalytical and operational work on core areas of development design, including the role andfunctioning of markets. Public action should seek to expand the opportunity sets of those who,in the absence of policy interventions, have the least resources, voice and capabilities. It shoulddo so in a manner that respects and enhances individual freedoms, as well as the role of marketsin allocating resources.

Equity in the international arena is also a central concern, and can play a powerful comple-mentary role to domestic action. In a globally interconnected world, leveling the internationalplaying fields, both economically and politically, will help domestic efforts to combine equitywith efficiency and growth.

In my view, the evidence that equity and economic efficiency as well as growth are comple-mentary in the long run helps to integrate the main two components of the World Bank’spoverty reduction strategy. The focus on broadening opportunities strongly supports the first

Foreword

xi

xii WORLD DEVELOPMENT REPORT 2005

pillar of the Bank’s development strategy, namely enhancing the investment climate for every-one. Together with the interdependence between the economic and political dimensions ofdevelopment it also reinforces the importance of empowerment. This report shows that thetwo pillars are not independent from each other in supporting development, but instead areintricately linked with one another. It is my hope that this report will have a real influence in theway that we and our development partners understand, design and implement developmentpolicies.

Paul D. WolfowitzPresidentThe World Bank

xiii

This Report has been prepared by a core team led by Francisco H.G. Ferreira and Michael Wal-ton, and comprising Tamar Manuelyan Atinc, Abhijit Banerjee, Peter Lanjouw, Marta Menén-dez, Berk Özler, Giovanna Prennushi,Vijayendra Rao, James Robinson, and Michael Woolcock.Important additional contributions were made by Anthony Bebbington, Stijn Claessens, Mar-garet Ellen Grosh, Karla Hoff, Jean O. Lanjouw, Xubei Lou, Ana Revenga, Caroline Sage, MarkSundberg, and Peter Timmer. The team was assisted by Maria Caridad Araujo, Andrew Beath,Ximena del Carpio, Celine Ferre, Thomas Haven, Claudio E. Montenegro, and Jeffery C. Tan-ner. The work was conducted under the general guidance of François Bourguignon.

Extensive and excellent advice was received from Anthony B. Atkinson, Angus Deaton, NailaKabeer, Martin Ravallion, and Amartya Sen, to whom the team is grateful without implication.Many others inside and outside the World Bank also provided helpful comments; their namesare listed in the Bibliographical Note. The Development Data Group contributed to the dataappendix and was responsible for the Selected World Development Indicators. Much of thebackground research was supported by a multidonor programmatic trust fund, the Knowledgefor Change Program, funded by Canada, the European Community, Finland, Norway, Sweden,Switzerland, and the United Kingdom.

The team undertook a wide range of consultations for this Report, which included work-shops in Amsterdam, Beirut, Berlin, Cairo, Dakar, Geneva, Helsinki, Hyderabad, London,Milan, Nairobi, New Delhi, Oslo, Ottawa, Paris, Rio de Janeiro, Stockholm, Tokyo, Venice, andWashington, D.C.; videoconferences with sites in Bogota, Buenos Aires, Mexico City, andTokyo; and an on-line discussion of the draft Report. The team wishes to thank participants inthese workshops, videoconferences, and discussions, which included researchers, governmentofficials, and staff of nongovernmental and private-sector organizations.

Rebecca Sugui served as executive assistant to the team, Ofelia Valladolid as program assis-tant, Madhur Arora and Jason Victor as team assistants. Evangeline Santo Domingo served asresource management assistant.

Bruce Ross-Larson was the principal editor. Book design, editing, and production werecoordinated by the World Bank’s Office of the Publisher under the supervision of SusanGraham and Monika Lynde.

Acknowledgments

NGO Nongovernmental organization ODA Official development assistanceOECD Organisation for Economic Co-operation and

Development PPA Participatory Poverty AssessmentPPP Purchasing-power parityPROMESA Promoción y Mejoramiento de la SaludSMEs Small and medium enterprisesTAC Treatment Action CampaignTIMSS Third International Mathematics and Science

StudyTRIPs Trade-related aspects of intellectual property

rightsU.N. United NationsUNCTAD United Nations Conference on Trade and

DevelopmentUNDP United Nations Development ProgrammeUNAIDS Joint United Nations Programme on HIV/AIDS UNICEF United Nations International Children’s

Emergency FundVAT Value added tax WHO World Health OrganizationWTO World Trade OrganizationWWII World War II

AbbreviationsThe following abbreviations are used in this Report:

AA Affirmative actionAIDS Acquired immune deficiency syndrome CCP Chinese Communist PartyDAC Development Assistance CommitteeDHS Demographic and Health SurveyECD Early child developmentEPL Employment protection legislationFDI Foreign direct investmentGDP Gross domestic product GHG Greenhouse gas GNI Gross national incomeHIPC Heavily Indebted Poor CountriesHIV Human immunodeficiency virusICOR Incremental Capital-Output RatioICRISAT International Crop Research Institute

in the Semi-Arid TropicsIDA International Development AssociationILO International Labour OrganizationIMF International Monetary FundIMS Intercontinental Marketing ServicesKDP Kecamatan Development ProjectMDG Millennium Development GoalsMMM Movement Militant MauricienMSF Médecins Sans FrontièresNAFTA North American Free Trade Agreement

Abbreviations and Data Notes

xiv

Data notesThe countries included in regional and income groupings inthis Report are listed in the Classification of Economies tableat the beginning of the Selected World Development Indica-tors. Income classifications are based on GNP per capita;thresholds for income classifications in this edition may befound in the Introduction to Selected World DevelopmentIndicators. Group averages reported in the figures and tablesare unweighted averages of the countries in the group, unlessnoted to the contrary.

The use of the word countries to refer to economies impliesno judgment by the World Bank about the legal or other statusof a territory. The term developing countries includes low- andmiddle-income economies and thus may include economies intransition from central planning, as a matter of convenience.The term advanced countries may be used as a matter of conve-nience to denote high-income economies.

Dollar figures are current U.S. dollars, unless otherwisespecified. Billion means 1,000 million; trillion means 1,000 bil-lion.

Consider two South African children born onthe same day in 2000. Nthabiseng is black,born to a poor family in a rural area in theEastern Cape province, about 700 kilometersfrom Cape Town. Her mother had no formalschooling. Pieter is white, born to a wealthyfamily in Cape Town. His mother completeda college education at the nearby prestigiousStellenbosch University.

On the day of their birth, Nthabiseng andPieter could hardly be held responsible fortheir family circumstances: their race, theirparents’ income and education, their urbanor rural location, or indeed their sex. Yetstatistics suggest that those predeterminedbackground variables will make a major dif-ference for the lives they lead. Nthabisenghas a 7.2 percent chance of dying in the firstyear of her life, more than twice Pieter’s 3percent. Pieter can look forward to 68 yearsof life, Nthabiseng to 50. Pieter can expect tocomplete 12 years of formal schooling,Nthabiseng less than 1 year.1 Nthabiseng islikely to be considerably poorer than Pieterthroughout her life.2 Growing up, she is lesslikely to have access to clean water and sani-tation, or to good schools. So the opportuni-ties these two children face to reach their fullhuman potential are vastly different fromthe outset, through no fault of their own.

Such disparities in opportunity translateinto different abilities to contribute toSouth Africa’s development. Nthabiseng’shealth at birth may have been poorer, owingto the poorer nutrition of her mother dur-ing her pregnancy. By virtue of their gendersocialization, their geographic location, andtheir access to schools, Pieter is much morelikely to acquire an education that willenable him to put his innate talents to fulluse. Even if at age 25, and despite the odds,Nthabiseng manages to come up with a

great business idea (such as an innovationto increase agricultural production), shewould find it much harder to persuade abank to lend her money at a reasonableinterest rate. Pieter, having a similarlybright idea (say, on how to design animproved version of promising software),would likely find it easier to obtain credit,with both a college diploma and quite pos-sibly some collateral. With the transition todemocracy in South Africa, Nthabiseng isable to vote and thus indirectly shape thepolicy of her government, somethingdenied to blacks under apartheid. But thelegacy of apartheid’s unequal opportunitiesand political power will remain for sometime to come. It is a long road from such a(fundamental) political change to changesin economic and social conditions.

As striking as the differences in lifechances are between Pieter and Nthabisengin South Africa, they are dwarfed by thedisparities between average South Africansand citizens of more developed countries.Consider the cards dealt to Sven—born onthat same day to an average Swedishhousehold. His chances of dying in thefirst year of life are very small (0.3 percent)and he can expect to live to the age of 80, 12years longer than Pieter, and 30 years morethan Nthabiseng. He is likely to complete11.4 years of schooling—5 years more thanthe average South African. These differencesin the quantity of schooling are com-pounded by differences in quality: in theeighth grade, Sven can expect to obtain ascore of 500 on an internationally compara-ble math test, while the average SouthAfrican student will get a score of only264—more than two standard deviationsbelow the Organisation for Economic Co-operation and Development (OECD) median.

1

Overview

Nthabiseng most likely will never reach thatgrade and so will not take the test.3

These differences in life chances acrossnationality, race, gender, and social groupswill strike many readers as fundamentallyunfair. They are also likely to lead to wastedhuman potential and thus to missed devel-opment opportunities. That is why WorldDevelopment Report 2006 analyzes the rela-tionship between equity and development.

By equity we mean that individualsshould have equal opportunities to pursue alife of their choosing and be spared fromextreme deprivation in outcomes. The mainmessage is that equity is complementary, insome fundamental respects, to the pursuitof long-term prosperity. Institutions andpolicies that promote a level playing field—where all members of society have similarchances to become socially active, politicallyinfluential, and economically productive—contribute to sustainable growth and devel-opment. Greater equity is thus doubly goodfor poverty reduction: through potentialbeneficial effects on aggregate long-rundevelopment and through greater opportu-nities for poorer groups within any society.

The complementarities between equityand prosperity arise for two broad sets ofreasons. First, there are many market fail-ures in developing countries, notably in themarkets for credit, insurance, land, andhuman capital. As a result, resources maynot flow where returns are highest. Forexample, some highly capable children, likeNthabiseng, may fail to complete primaryschooling, while others, who are less able,may finish university. Farmers may workharder on plots they own than on thosethey sharecrop. Some efficient developing-country producers of agricultural com-modities and textiles are shut out of someOECD markets, and poor unskilled workershave highly restricted opportunities tomigrate to work in richer countries.

When markets are missing or imperfect,the distributions of wealth and power affectthe allocation of investment opportunities.Correcting the market failures is the idealresponse; where this is not feasible, or fartoo costly, some forms of redistribution—of access to services, assets, or politicalinfluence—can increase economic efficiency.

The second set of reasons why equity andlong-term prosperity can be complementaryarises from the fact that high levels ofeconomic and political inequality tend tolead to economic institutions and socialarrangements that systematically favor theinterests of those with more influence. Suchinequitable institutions can generate eco-nomic costs. When personal and propertyrights are enforced only selectively, whenbudgetary allocations benefit mainly thepolitically influential, and when the distri-bution of public services favors the wealthy,both middle and poorer groups end up withunexploited talent. Society, as a whole, isthen likely to be more inefficient and to missout on opportunities for innovation andinvestment. At the global level, when devel-oping countries have little or no voice inglobal governance, the rules can be inappro-priate and costly for poorer countries.

These adverse effects of unequal opportu-nities and political power on development areall the more damaging because economic,political, and social inequalities tend to repro-duce themselves over time and across genera-tions. We call such phenomena “inequalitytraps.” Disadvantaged children from familiesat the bottom of the wealth distribution donot have the same opportunities as childrenfrom wealthier families to receive qualityeducation. So these disadvantaged childrencan expect to earn less as adults. Because thepoor have less voice in the political process,they—like their parents—will be less able toinfluence spending decisions to improvepublic schools for their children. And thecycle of underachievement continues.

The distribution of wealth is closely corre-lated with social distinctions that stratify peo-ple, communities, and nations into groupsthat dominate and those that are dominated.These patterns of domination persist becauseeconomic and social differences are rein-forced by the overt and covert use of power.Elites protect their interests in subtle ways, byexclusionary practices in marriage and kin-ship systems, for instance, and in ways thatare less subtle, such as aggressive politicalmanipulation or the explicit use of violence.

Such overlapping political, social, cultural,and economic inequalities stifle mobility.They are hard to break because they are so

2 WORLD DEVELOPMENT REPORT 2006

closely tied to the ordinary business of life.They are perpetuated by the elite, and ofteninternalized by the marginalized or oppressedgroups, making it difficult for the poor tofind their way out of poverty. Inequality trapscan thus be rather stable, tending to persistover generations.

The report documents the persistence ofthese inequality traps by highlighting theinteraction between different forms of in-equality. It presents evidence that the inequal-ity of opportunity that arises is wasteful andinimical to sustainable development andpoverty reduction. It also derives policyimplications that center on the broad conceptof leveling the playing field—both politicallyand economically and in the domestic andthe global arenas. If the opportunities facedby children like Nthabiseng are so muchmore limited than those faced by children likePieter or Sven, and if this hurts developmentprogress in the aggregate, then public actionhas a legitimate role in seeking to broaden theopportunities of those who face the mostlimited choices.

Three considerations are important atthe outset. First, while more even playingfields are likely to lead to lower observedinequalities in educational attainment,health status, and incomes, the policy aim isnot equality in outcomes. Indeed, even withgenuine equality of opportunities, one wouldalways expect to observe some differences inoutcomes owing to differences in preferences,talents, effort, and luck.4 This is consistentwith the important role of income differencesin providing incentives to invest in educationand physical capital, to work, and to takerisks. Of course outcomes matter, but we areconcerned with them mainly for their influ-ence on absolute deprivation and their role inshaping opportunities.

Second, a concern with equality of oppor-tunity implies that public action should focuson the distributions of assets, economicopportunities, and political voice, rather thandirectly on inequality in incomes. Policies cancontribute to the move from an “inequalitytrap” to a virtuous circle of equity and growthby leveling the playing field—through greaterinvestment in the human resources of thepoorest; greater and more equal access topublic services and information; guarantees

on property rights for all; and greater fair-ness in markets. But policies to level the eco-nomic playing field face big challenges.There is unequal capacity to influence thepolicy agenda: the interests of the disenfran-chised may never be voiced or represented.And when policies challenge privileges,powerful groups may seek to block reforms.Thus, equitable policies are more likely to besuccessful when leveling the economic play-ing field is accompanied by similar efforts tolevel the domestic political playing field andintroduce greater fairness in global gover-nance.

Third, there may be various short-run,policy-level tradeoffs between equity and effi-ciency. These are well recognized and exten-sively documented. The point is that the(often implicit) cost-benefit calculus thatpolicymakers use to assess the merits of vari-ous policies too often ignores the long-term,hard-to-measure but real benefits of greaterequity. Greater equity implies more efficienteconomic functioning, reduced conflict,greater trust, and better institutions, withdynamic benefits for investment and growth.To the extent that such benefits are ignored,policymakers may end up choosing too littleequity.

By the same token, however, those inter-ested in greater equity must not ignore theshort-term tradeoffs. If individual incen-tives are blunted by income redistributionschemes that tax investment and produc-tion too steeply, the result will be less inno-vation, less investment, and less growth.The history of the twentieth century is lit-tered with examples of ill-designed policiespursued in the name of equity that seriouslyharmed—rather than spurred—growthprocesses by ignoring individual incentives.A balance must be sought, taking intoaccount both the immediate costs to indi-vidual incentives and the long-term benefitsof cohesive societies, with inclusive institu-tions and broad opportunities.

While careful assessment of policy designin local contexts is always important, equityconsiderations need to be brought squarelyinto the center of both diagnosis and policy.This is not intended as a new framework. Itmeans integrating and extending existingframeworks: equity is central both to the

Overview 3

investment environment and to the agendaof empowerment, working through theimpact on institutions and specific policydesigns. Some may value equity for its ownsake, others primarily for its instrumentalrole in reducing absolute poverty, the WorldBank’s mission.

This report recognizes the intrinsic valueof equity but aims primarily to documenthow a focus on equity matters for long-rundevelopment. It has three parts.

• Part I considers the evidence on inequal-ity of opportunity, within and acrosscountries. Some attempts to quantifyinequality of opportunity are reviewedbut, more generally, we rely on evidenceof highly unequal outcomes acrossgroups defined by predetermined cir-cumstances—such as gender, race, fam-ily background, or country of birth—asmarkers for unequal opportunities.

• Part II asks why equity matters. It dis-cusses the two channels of impact (theeffects of unequal opportunities whenmarkets are imperfect, and the conse-quences of inequity for the quality ofinstitutions a society develops) as well asintrinsic motives.

• Part III asks how public action can levelthe political and economic playing fields.In the domestic arena, it makes the casefor investing in people, expanding accessto justice, land, and infrastructure, andpromoting fairness in markets. In theinternational arena, it considers levelingthe playing field in the functioning ofglobal markets and the rules that governthem—and the complementary provi-sion of aid to help poor countries andpoor people build greater endowments.

The remainder of this overview provides asummary of the principal findings.

Inequity within and across nationsFrom an equity perspective, the distribu-tion of opportunities matters more thanthe distribution of outcomes. But opportu-nities, which are potentials rather thanactuals, are harder to observe and measurethan outcomes.

Within-country inequities have many dimensionsDirect quantification of inequality ofopportunity is difficult, but one analysis ofBrazil provides an illustration (chapter 2).Earnings inequality in 1996 was dividedinto one share attributable to four predeter-mined circumstances that lie beyond thecontrol of individuals—race, region ofbirth, parental education, and paternaloccupation at birth—and a residual share.These four circumstances account foraround one-quarter of overall differences inearnings between workers. Arguably, otherdeterminants of opportunity are equallypredetermined at birth but not included inthis set—for example, gender, familywealth, or the quality of primary schools.Because such variables are not included inthe inequality “decomposition,” the resultshere can be seen as lower-bound estimatesof inequality of opportunity in Brazil.

Unfortunately, predetermined (and thusmorally irrelevant) circumstances deter-mine much more than just future earnings.Education and health are of intrinsic valueand affect the capacity of individuals toengage in economic, social, and politicallife. Yet children face substantially differentopportunities to learn and to lead healthylives in almost all populations, dependingon asset ownership, geographic location, orparental education, among others. Considerhow access to a basic package of immuniza-tion services differs for the rich and thepoor across countries (figure 1).

There is substantial inequality in accessbetween, for example, Egypt, where almosteveryone is covered (on the left), and Chad,where more than 40 percent of children areexcluded (on the right). Yet the disparitiescan be as large within some countries as theyare across all nations in the sample. InEritrea, for instance, the richest fifth enjoysalmost complete coverage, but almost half ofall children in the poorest fifth are excluded.

Significant gender differences also per-sist in many parts of the world. In parts ofEast and South Asia, notably in certain areasin rural China and northwest India, theopportunity to life itself can depend on onesingle predetermined characteristic: sex.These regions have significantly more boy


infants than girls, in part because of sex-selective abortion and differential care afterbirth. And in many (though not all) parts ofthe world, more boys than girls attendschool. The hundreds of millions of dis-abled children across the developing worldalso face very different opportunities thantheir able-bodied peers.

These inequities are usually associatedwith differences in an individual’s “agency”—the socioeconomically, culturally, and polit-ically determined ability to shape the worldaround oneself. Such differences createbiases in the institutions and rules in favorof more powerful and privileged groups.This is seen in realities as diverse as the lowchances for mobility of scheduled castes ina village in rural India and the frequentepisodes of discrimination against theQuichua people in Ecuador. Persistent dif-ferences in power and status between

groups can become internalized into behav-iors, aspirations, and preferences that alsoperpetuate inequalities.

Inequalities of opportunity are alsotransmitted across generations. The chil-dren of poorer and lower-status parentsface inferior chances in education, health,incomes, and status. This starts early. InEcuador, three-year-old children from allsocioeconomic groups have similar testscores for vocabulary recognition and areclose to a standard international referencepopulation. But by the time they are five, allhave faltered relative to the internationalreference group, except for those in therichest groups and with the highest levels ofparental education (figure 2). Such pro-nounced differences in vocabulary recogni-tion between children whose parents had 0to 5 years of schooling and those whoseparents had 12 or more years are likely to

Overview 5

Percentage not covered

70

60

50

40

30

20

10

0

Poorest

Wealthiest

Egyp

tJo

rdan

(*)

Colo

mbi

aRw

anda

Peru

Sout

h Af

rica

Ken

yaM

alaw

iBr

azil

Zam

bia

(*)

Viet

nam

Turk

eyGu

atem

ala

Tanz

ania

Indo

nesia

Turk

men

istan

(*)

Mor

occo

Ghan

aBe

nin

Philip

pine

sBa

ngla

desh

Com

oros

Boliv

iaPa

ragu

ayKa

zakh

stan

(*)

Yem

enBu

rkin

a Fa

soCa

mer

oon

Ugan

daIn

dia

Mau

ritan

iaHa

itiTo

goEt

hiop

ia

Cent

ral A

frica

n Re

publ

icM

adag

asca

rM

ozam

biqu

eGu

inea

Mal

iCa

mbo

dia

Pakis

tan

Eritr

eaNi

ger

Chad

Figure 1 Wealth matters for the immunization of children

Source: Authors’ own calculations from Demographic Health Survey (DHS) dataNote: * indicates that the poorest quintile have higher access to childhood immunization services than the wealthiest quintile.The continuous orange line represents the overall percentage of children without access to a basic immunization package in each country,while the endpoints indicate the percentages for the top and the bottom quintile of the asset ownership distribution.

carry over to their performance once theyenter primary school, and will likely persistthereafter. Intergenerational immobility isalso observed in rich countries: new evi-dence from the United States (where themyth of equal opportunity is strong) findshigh levels of persistence of socioeconomicstatus across generations: recent estimatessuggest that it would take five generationsfor a family that earned half the nationalaverage income to reach the average.5

Immobility is particularly pronounced forlow-income African Americans.

Global inequities are massiveIf unequal opportunities are large withinmany countries, they are truly staggering ona global scale. Chapter 3 shows that cross-country differences begin with the opportu-nity for life itself: while 7 of every 1,000American babies die in the first year of theirlives, 126 of every 1,000 Malian babies do.Babies who survive, not only in Mali but inmuch of Africa and in the poorer countriesof Asia and Latin America, are at muchgreater nutritional risk than their counter-parts in rich countries. And if they go toschool—more than 400 million adults indeveloping countries never did—theirschools are substantially worse than thoseattended by children in Europe, Japan, orthe United States. Given lower school qual-

ity, undernutrition, and the earnings a childcan generate by working instead of study-ing, many children leave school early. Theaverage person born between 1975 and1979 in Sub-Saharan Africa has only 5.4years of schooling. In South Asia, the figurerises to 6.3 years; in OECD countries, it is13.4 years.

With such differences in education andhealth, compounded by large disparities inaccess to infrastructure and other publicservices, it is not surprising that opportuni-ties for the consumption of private goodsdiffer vastly between rich and poor coun-tries. Mean annual consumption expendi-tures range from Purchasing Power Parity(PPP) $279 in Nigeria to PPP $17,232 inLuxembourg. This means that the averagecitizen in Luxembourg enjoys monetaryresources 62 times higher than the averageNigerian. While the average Nigerian mayfind it difficult to afford adequately nutri-tious meals every day, the average citizen ofLuxembourg need not worry too muchabout buying the latest generation cellphone on the market. Because of the muchgreater restrictions on the movement ofpeople between countries than withincountries, these inequalities in outcomesamong countries are likely to be much moreclosely associated with inequalities inopportunities than within countries.

Global inequality trends have varied.Between 1960 and 1980 there was a pro-nounced decline in the inequality in lifeexpectancy across countries, driven bymajor increases in the poorest countries inthe world (figure 3). This welcome develop-ment was due to the global spread of healthtechnology and to major public healthefforts in some of the world’s highest mor-tality areas. Since 1990, however, HIV/AIDS(predominantly in many African countries)and a rise in mortality rates in transitioneconomies (largely in Eastern Europe andCentral Asia) have set back some of the ear-lier gains. Because of the AIDS crisis, lifeexpectancy at birth has fallen dramaticallyin some of the world’s poorest countries,sharply increasing the differences betweenthem and richer societies.

Inequality in access to schooling has alsobeen falling around the world, within most


40 60 7050

100

90

80

70

60

110

Wealthiest 25%

Median score Median score

Poorest 25th%

40 60 7050Age in months Age in months

Wealthiest and poorest quartiles Maternal education

100

90

80

70

60

110

12 or more years

0–5 years

Figure 2 Opportunities are determined early Cognitive development for children ages three to five in Ecuador differs markedly across familybackgrounds

Source: Paxson and Schady (2005).Note: Median values of the test of vocabulary recognition (TVIP) score (a measure of vocabulary recognition in Spanish,standardized against an international norm) are plotted against the child’s age in months. The medians by exact monthof age were smoothed by estimating fan regressions of the median score on age (in months), using a bandwidth of 3.

0

0.05

2000

1990

1960

Population-weighted internationaldistribution of life expectancy, 1960–2000

27 39 51 63Years

75 87

Figure 3 Life expectancy improved andbecame more equal—until the onsetof the AIDS crisis

Source: Schady (2005).

countries as well as across them, as averageschooling levels rise in the vast majority ofcountries. This too is a welcome develop-ment, although concerns over the quality ofschooling provide reasons for guardingagainst complacency.

While our primary concern is withinequality of opportunities, the large dif-ferences in income or consumption acrosscountries surely affect the life chances facedby children born today in those differentnations. Trends in life expectancy at birthand years of schooling were converging, atleast until 1990, but a different pictureemerges for income and consumption.While the recent trends depend greatly onthe specific concept chosen (discussed ingreat detail in chapter 3), global incomeinequality has steadily increased over thelong run until the onset of rapid economicgrowth in China and India in the 1980s(figure 4).

It is possible to decompose totalinequality across individuals in the worldinto differences among countries and dif-ferences within countries. Between-coun-try differences were relatively small early inthe nineteenth century, but they came toaccount for a larger part of total inequalitytoward the end of the twentieth century. IfChina and India are excluded, globalinequalities have continued to rise, owingto the continuing divergence betweenmost other low-income countries and richcountries.

Why does equity matter for development?Why do these persistent inequalities—both within and across countries—matter?The first reason is that the interconnec-tions and resilience of these inequalitiesimply that some groups have consistentlyinferior opportunities—economic, social,and political—than their fellow citizens.Most people feel that such egregious dis-parities violate a sense of fairness, particu-larly when the individuals affected can dolittle about them (chapter 4). This is con-sistent with the teachings of much politi-cal philosophy and with the internationalsystem of human rights. The core moraland ethical teachings of the world’s lead-ing religions include a concern for equity,although many have also been sources ofinequities and historically have been linkedto unequal power structures. There is alsoexperimental evidence suggesting thatmany—but not all—people behave inways consistent with a concern for fair-ness, in addition to caring about how theyfare individually.

Important as these intrinsic reasons arefor caring about inequality of opportunitiesand unfair processes, the primary focus ofthis report is on the instrumental relation-ship between equity and development, withparticular emphasis on two channels: theeffects of unequal opportunities when mar-kets are imperfect, and the consequences ofinequity for the quality of institutions asociety develops.6

With imperfect markets, inequalities inpower and wealth translate into unequalopportunities, leading to wasted productivepotential and to an inefficient allocation ofresources. Markets often work imperfectlyin many countries, whether because ofintrinsic failures—such as those associatedwith asymmetric information—or becauseof policy-imposed distortions. Microeco-nomic case studies suggest that an ineffi-cient allocation of resources across produc-tive alternatives is often associated withdifferences in wealth or status (chapter 5).

If capital markets worked perfectly, therewould be no relation between investment

Overview 7

Mean log deviation1

0.8

0.6

0.4

0.2

019921980197019601950192919101890187018501820

0.33

0.50

0.83

0.33

0.36

0.69

0.05

0.37

0.42

Global inequality

Within-country inequality

Between-country inequality

Figure 4 A long-run diverging trend in incomeinequality begins to reverse because of growth in China and India

Source: Authors’ manipulation of data from Bourguignon and Mor-risson (2002).

and the distribution of wealth: anyone witha profitable investment opportunity wouldbe able to either borrow money to financeit, or to sell equity in a firm set up to under-take it. But capital markets in just aboutevery country (developed and developing)are very far from perfect: credit is rationedacross prospective clients, and interest ratesdiffer considerably across borrowers, andbetween lenders and borrowers, in waysthat cannot be linked to default risk orother economic factors affecting expectedreturns to lenders. For example, interestrates decline with loan size in Kerala andTamil Nadu in India, and across tradinggroups in Kenya and Zimbabwe, in ways notexplained by risk differences.7 In Mexico,returns to capital are much higher for thesmallest informal sector firms than forlarger ones.

Land markets also have imperfectionsassociated with a lack of clear titling, histo-ries of concentrated land ownership, andimperfect rental markets. In Ghana, lowersecurity of tenure among women leads toan inefficiently low frequency of land fal-lowing and, hence, to progressive declinesin land productivity.

The market for human capital is alsoimperfect, because parents make decisions on

behalf of their children and because theexpected returns to investment are influencedby location, contacts, and discrimination—on grounds of gender, caste, religion, orrace. Discrimination and stereotyping—mechanisms for the reproduction of inequal-ity between groups—have been found tolower the self-esteem, effort, and perform-ance of individuals in the groups discrimi-nated against. This reduces their potential forindividual growth and their ability to con-tribute to the economy.

Striking evidence of the impact ofstereotyping on performance comes from arecent experiment in India. Children fromdifferent castes were asked to complete sim-ple exercises, such as solving a maze, withreal monetary incentives contingent on per-formance. The key result of the experimentis that low-caste children perform on parwith high-caste children when their caste isnot publicly announced by the experi-menter but significantly worse when it ismade public (figure 5). If a similar inhibi-tion of talent occurs in the real world, thisimplies a loss of potential output owing tosocial stereotyping.

Economic and political inequalities areassociated with impaired institutionaldevelopment. The second channel throughwhich inequity affects long-run processesof development is the shaping of economicand political institutions (chapter 6). Insti-tutions determine the incentives and con-straints people face and provide the contextin which markets function. Different sets ofinstitutions are the outcome of complexhistorical processes that reflect the interestsand structure of political influence ofdifferent individuals and groups in a soci-ety. From this perspective, market im-perfections may arise not by accident butbecause they distribute income or power inparticular ways. In this view, there will besocial conflict over the institutions ofsociety and incentives for people who con-trol power to shape institutions in waysthat benefit them.

The central argument here is thatunequal power leads to the formation ofinstitutions that perpetuate inequalities inpower, status, and wealth—and that typi-


8

6

4

2

0Caste

notannounced

Castenot

announced

Casteannounced

Casteannounced

Casteannounced

andsegregated

High casteLow caste

Piece rate Tournament

Average number of mazes solved, by caste,in five experimental treatments

Figure 5 Children’s performance differs when theircaste is made salient

Source: Hoff and Pandey (2004).Note: The figure depicts the number of mazes correctly completedby low-caste and high-caste children from a set of Indian villagesin a number of different experiments. The difference between thefirst two and the last three columns refers to payouts: whetherchildren are paid per correct maze completed (piece rate) or onlyif they complete the most mazes (tournament).

cally are also bad for the investment, inno-vation, and risk-taking that underpinlong-term growth. Good economic insti-tutions are equitable in a fundamentalway: to prosper, a society must createincentives for the vast majority of thepopulation to invest and innovate. Butsuch an equitable set of economic institu-tions can emerge only when the distribu-tion of power is not highly unequal and insituations in which there are constraintson the exercise of power by officeholders.Basic patterns in cross-country data andhistorical narratives support the view thatcountries moving onto institutional pathsthat promoted sustained prosperity did sobecause the balance of political influenceand power became more equitable.

One example comes from a comparisonof the early institutions and of the long-term development paths of Europeancolonies in North and South America. Theabundance of unskilled labor prevalent inthe South American colonies—where eithernative Americans or imported Africanslaves were available in large numbers—combined with the technology of miningand large plantation agriculture to providethe economic base for hierarchical andextractive societies, in which land owner-ship and political power were highly con-centrated. In North America, by contrast,similar attempts to introduce hierarchicalstructures were foiled by the scarcity oflabor—except where agro-climatic condi-tions made slavery economically feasible,such as in the southern region of the UnitedStates. Competition for free labor in thenorthern areas of North America led to thedevelopment of less unequal land owner-ship patterns, a faster expansion of the fran-chise, and rapid increases in literacy andbasic education. The resulting economicand political institutions persisted overtime, with positive consequences for long-run economic development.

Leveling the economic and political playing fieldsSo a portion of the economic and politicalinequalities we observe around the world isattributable to unequal opportunities. This

inequality is objectionable on both intrinsicand instrumental grounds. It contributes toeconomic inefficiency, political conflict, andinstitutional frailty. What are the implica-tions for policy, and do they give rise to anagenda that is different from the povertyreduction agenda already embraced by theWorld Bank, other multilateral institutions,and many governments?

We argue that an equity lens enhancesthe poverty reduction agenda. The poorgenerally have less voice, less income, andless access to services than most other peo-ple. When societies become more equitablein ways that lead to greater opportunities forall, the poor stand to benefit from a “doubledividend.” First, expanded opportunitiesbenefit the poor directly, through greaterparticipation in the development process.Second, the development process itself maybecome more successful and resilient asgreater equity leads to better institutions,more effective conflict management, and abetter use of all potential resources in soci-ety, including those of the poor. Resultingincreases in economic growth rates in poorcountries will, in turn, contribute to areduction in global inequities.

One manifestation of the greater partic-ipation of the poor in economic growth isthe fact that the growth elasticity ofpoverty reduction falls with greater incomeinequality. In other words, the impact of(the same amount of) growth on povertyreduction is significantly greater when ini-tial income inequality is lower. On average,for countries with low levels of incomeinequality, a 1 percentage point growth inmean incomes leads to about a 4 percent-age-point reduction in the incidence of $1per day poverty. That power falls to close tozero in countries with high incomeinequality.8 Policies that lead to greaterequity thus lead to lower poverty—directlythrough expanding the opportunities ofthe poor and indirectly through higher lev-els of sustained development.

An equity lens adds three new—or atleast often neglected—perspectives to devel-opment policymaking:

• First, the best policies for povertyreduction could involve redistributions

Overview 9

of influence, advantage, or subsidiesaway from dominant groups. Highlyunequally distributed wealth associatedwith unduly concentrated politicalpower can prevent institutions fromenforcing broad-based personal andproperty rights, and lead to skewed pro-visioning of services and functioning ofmarkets. This is unlikely to changeunless voice and influence, and publicresources, shift away from the domi-nant group toward those with feweropportunities.9

• Second, while such equity-enhancingredistributions (of power, or access togovernment spending and markets) canoften be efficiency-increasing, possibletradeoffs need to be assessed in thedesign of policy. At some point, highertax rates to finance spending on moreschools for the poorest will create somuch disincentive to effort or invest-ment (depending on how the taxes areraised), that one should stop raisingthem. When making a policy choicealong such tradeoffs, the full value of thebenefits from equity enhancementshould be considered. If greater spend-ing on schools for lower-caste childrenmeans that, over the long term, stereo-typing will decline in society, with atten-dant increases in performance that areadditional to the specific gains fromgreater schooling today, these gainsshould not be ignored.

• Third, the dichotomy between policiesfor growth and policies specificallyaimed at equity is false. The distributionof opportunities and the growth processare jointly determined. Policies that affectone will affect the other. This does notmean that each policy needs to takeequity into account individually: forexample, the best way to deal withinequitable effects of a particular tradereform is not always through fine-tuningtrade policy itself (which might make itmore susceptible to capture) but throughcomplementary policies for safety nets,labor mobility, and education. The over-all package and the fairness of the under-lying process are what matter.

The analysis of development experienceclearly shows the centrality of overall politi-cal conditions—supporting the emphasison governance and empowerment in recentyears. However, it is neither the mandatenor the comparative advantage of theWorld Bank to engage in advice on issues ofpolitical design. In turning to policy impli-cations, we focus instead on the core areasof development policy, while recognizingthat policy design needs to take account ofthe broader social and political context, andthat accountability mechanisms influencedevelopment effectiveness.

Because economic policies are deter-mined within a sociopolitical reality, howpolicies are designed, introduced, orreformed matters as much as which specificpolicies are proposed. Policy reforms thatresult in losses for a particular group will beresisted by that group. If the group is pow-erful, it will usually subvert the reform. Thesustainability of reforms, therefore, maydepend on making information about itsdistributional consequences publicly avail-able and, perhaps, forming coalitions ofmiddle and poorer groups that stand togain from them to “empower,” directly orindirectly, relatively disadvantaged mem-bers of society.

How policies are implemented has atechnical aspect as well. Just as we empha-size that the full long-term benefits of redis-tributions need to be taken into accountwhen making policy choices, so must alltheir costs. A focus on equity does notchange the facts that asset expropriations—even in instances of historical grievances—may have adverse consequences for subse-quent investment, that high marginal taxrates create disincentives to work, or thatinflationary financing of budget deficitstends to lead to regressive implicit taxation,economic disorganization, and reducedinvestment and growth. In short, a focus onequity must not be an excuse for poor eco-nomic policy.

The report discusses the role of publicaction in leveling the economic and politi-cal playing field under four main headings.Three of the headings concern domesticpolicies: investing in human capacities;expanding access to justice, land, and infra-


structure; and promoting fairness in mar-kets. The fourth turns to policies for greaterglobal equity, in terms of access to markets,resource flows, and governance.

Throughout the discussion, the reportweighs a desire to be specific and practicalagainst the fact that the best specific policymix is a function of country context. Theeducational challenges facing Sudan aredifferent from those facing Egypt. Theoptimal sequencing of reform in the publicsectors of Latvia and Bolivia are unlikely tobe the same. The capacity for implement-ing health finance reform in China andLesotho are also different. So the detailed,specific policy advice always needs to bedeveloped at the country—or even subna-tional—level. Everything that is said belowtherefore retains some level of generalityand should be interpreted accordingly, andcautiously.

Human capacitiesEarly childhood development. In manydeveloping countries, the actions of thestate in providing services magnify—ratherthan attenuate—inequalities at birth. Aguiding principle is to shape public actionso that the acquisition of human capacitiesis not driven by circumstances of theirbirth, although it can reflect people’s prefer-ences, tastes, and talents.

Because differences in cognitive devel-opment start to widen from a very earlyage (see figure 2), early childhood develop-ment initiatives can be central to moreequal opportunities. Evidence supports theview that investing in early childhood haslarge impacts on children’s health andreadiness to learn and can bring importanteconomic returns later in life—oftengreater than investments in formal educa-tion and training.

An experiment in Jamaica focused onundersized children (ages 9 to 24 months)and found that they suffered from lowerlevels of cognitive development than thoseof normal height. Nutritional supplementsand a program of regular exposure to men-tal stimulation, helped offset this disadvan-tage. After 24 months, kids who receivedboth better nutrition and more stimulationhad virtually caught up developmentally

with children who started life at a normalheight (figure 6). This illustrates how deci-sive and well-designed public action cansubstantially reduce the opportunity gapsbetween those least privileged and the soci-etal norm. Investing in the neediest peopleearly in their childhoods can help level theplaying field.

Schooling. The process continues through-out the school system. Actions to equalizeopportunities in formal education need toensure that all children acquire at least abasic level of skills necessary to participatein society and in today’s global economy.Even in such middle-income countries asColombia, Morocco, and the Philippines,most children completing basic educationlack an adequate level of achievement, asmeasured by internationally comparabletest scores (chapters 2 and 7).

Access to schooling matters—especiallyin very poor countries—but, in many coun-tries, it is only a small part of the problem.Greater access needs to be complemented bysupply-side policies (to raise quality) anddemand-side policies (to correct for the pos-sibility that parents may underinvest in theeducation of their children for various rea-sons). There are no magic bullets for this, butincreasing teachers’ incentives, enhancingthe basic quality of the school’s physicalinfrastructure, and researching and imple-menting teaching methods to increase thelearning performance of students who do

Overview 11

85

110Development quotient

Baseline 6 12Months

18 24

105

100

95

90Control group

Nutritional supplement

Stimulation

Stimulation and nutritional supplementChildren of normalheight

Figure 6 Catching up through early interventions

Source: Grantham-McGregor and others (1991).Note: The development quotient is an index of progress on four behavioral and cognitive indicators of childhooddevelopment. Number of months refers to the time after entry into the program—generally at an age of nine months.

not do well when left to their own devices aresome of the suggestions on the supply side.

On the demand side, there is now a con-siderable body of evidence showing thatscholarships conditional on attendance havesignificant impacts. Such transfers work incountries from Bangladesh to Brazil, with theimpacts often greater for girls. There are alsopromising approaches to bring in excludedgroups—as in the Vidin model of reachingRoma in Bulgaria—and to bring up those leftbehind through remedial education—as inthe Balsakhi program using young women aspara-teachers in 20 cities in India. As arguedin World Development Report 2004, devel-oping the accountability of schools andteachers to students, parents, and the broadercommunity can help ensure effective serviceprovider behavior.

Health. Two areas stand out in reducinginequity and tackling economic distortionsin the provision of health services. First,there are many cases when the benefits spillover beyond the direct beneficiary in arange of areas of service provision: forimmunization, for water and sanitation,and for information on hygiene and childcare. Public assurance of provisioningmakes sense in these areas. Demand-sidesubsidies to provide incentives for maternaland child health increase use, offsettingpossible information problems as in Mex-ico’s Oportunidades program.

Second, insurance markets for cata-strophic health problems are beset with fail-ures. (Here “catastrophic” is in relation to thecapacity of the household to deal with thedirect costs and the loss in earnings.) The tra-ditional supply-side model of relying on pub-lic hospitals works badly, especially for poorand excluded groups. What can work better ispublic provisioning or regulation that pro-vides some insurance for all. Examplesinclude risk pooling in Colombia, healthcards in Indonesia and Vietnam, and Thai-land’s “30-baht” universal coverage scheme.As with education, these interventions needto be combined with incentives for providersto be responsive to all groups.

Risk management. Social protection sys-tems shape opportunities by providing peo-

ple with a safety net. In addition to ill health,macroeconomic crises, industrial restructur-ing, weather, and natural disasters can con-strain investment and innovation. The poor,with the lowest capacity to manage shocks,generally are the least well covered by risk-management structures, although in mostcountries many among the non-poor riskfalling into poverty. Broader social protec-tion systems can help prevent today’sinequalities—sometimes generated by badluck—from becoming entrenched and lead-ing to tomorrow’s inequities. Just as safetynets can spur households to engage in riskieractivities that can yield higher returns, theycan also help complement reforms that pro-duce losers.

Safety nets typically target three groups:the working poor, people viewed as unable towork or for whom work is undesirable, andspecial vulnerable groups. If safety nets aredesigned in a manner appropriate to the localrealities on the ground in each country, indi-vidual targeted interventions in these threecategories can be combined to provide aneffectively universal public insurance system.In such a system, each household that suffersa negative shock, and falls below some prede-termined threshold of living standards,would qualify for some form of state support.

Taxes for equity. Successful interventions tolevel the playing field require adequateresources. The main aim of good tax policy isto mobilize sufficient funding, while distort-ing incentives and compromising growth aslittle as possible. Because taxes impose effi-ciency costs by altering individual choicesbetween labor and leisure and consumptionand savings, most developing countries arelikely to be best served by avoiding high mar-ginal taxes on income and relying on a broadbase, especially for taxes on consumption.Public spending should play the primary rolein actively furthering equity. Nevertheless,there is some scope for making the overall taxsystem moderately progressive without largeefficiency costs. Societies that desire such anoutcome can consider simple exemptions forbasic foodstuffs, and an expanded role forproperty taxation, for example.

While the capacity of the tax administra-tion and the structure of the economy influ-


ence the ability to raise revenues, the qualityof institutions and the nature of the socialcompact are also critical. When citizens canrely on services actually being provided, theylikely are more willing to be taxed. Con-versely, a corrupt or kleptocratic state engen-ders little citizen trust in authority and littleincentive to cooperate. As a general rule, amore legitimate and representative state maybe a prerequisite to an adequate tax system,even as the notion of adequacy varies fromcountry to country.

Justice, land, and infrastructureThe development of human capacities willnot broaden opportunities if some peopleface unfair returns on those capacities andunequal protection of their rights, and haveunequal access to complementary factors ofproduction.

Building equitable justice systems. Justicesystems can do much to level the playingfield in the political, economic, and socio-cultural domains, but they can also rein-force existing inequalities. The report paysattention to both codified law and the waysin which the law is applied and enforced inpractice. Legal institutions can uphold thepolitical rights of citizens and curb the cap-ture of the state by the elite. They can equal-ize economic opportunities by protectingproperty rights for all and ensuring nondis-crimination in the market. They underpinand reflect the rules of the game in societyand thus are central to fair process—and tothe broad-based property rights and un-biased dispute resolution mechanisms soimportant for investment.

The law can also accelerate shifts innorms, and justice systems can serve as aprogressive force for change in the socialdomain by challenging inequitable prac-tices. For example, the U.S. Civil Rights Actof 1964 and Medicare in 1965 enforced thedesegregation of hospitals and led to largereductions in infant mortality for AfricanAmericans. Affirmative action programshave also been shown to reduce group-based differences in earnings and educa-tion. But they can become politicallyentrenched and limited to helping the bet-ter-off among disadvantaged groups.

Equity in laws and fairness in their imple-mentation involve striking a balance betweenstrengthening the independence of justicesystems and increasing accountability—especially to counter the risk that the pow-erful and wealthy might corrupt, influence,or ignore the law. Measures to make thelegal system more accessible—mobilecourts, legal aid, and working with custom-ary institutions—all help reduce the barri-ers that excluded groups face. Customaryinstitutions raise complex issues and mayincorporate inequities (for example, withrespect to gender), but they are too impor-tant to be ignored. South Africa is an exam-ple of a country that is pursuing a policythat balances recognition of customarypractices with the rights and responsibilitiesin state law.

Toward greater equity in access to land.Broader access to land does not necessarilyhave to come through ownership (chapter 8).Instead, improving the functioning of landmarkets and providing greater security oftenure for poorer groups may be a morefruitful area for policy—as in rural Thailandand in urban Peru. Redistributive landreform can make sense in some circum-stances in which land inequalities are extremeand the institutional context allows fordesigns that effectively redistribute land tosmaller farms and support this with comple-mentary services, without large transitionalcosts. But this can be difficult, and tradeoffsmay be large when property rights have ahigh degree of legitimacy.

Expropriating land (with compensation)is probably the most disruptive redistributioninstrument. Divesting state lands and recu-perating illegal settlements, possibly inexchange for titling a portion of the settle-ment, may be two cost-effective alternatives.Market or community-based approaches thatallow community members to obtain subsi-dized credit for land rentals or purchasesaccording to the willing-buyer-willing-sellerprinciple, as in Brazil and South Africa,appear promising. A land tax can be a usefulcomplement, generating revenues to pur-chase land to redistribute or encouragingredistribution by disproportionately taxinglarge or underused plots.

Overview 13

Providing infrastructure equitably. Access toinfrastructure—roads, electricity, water,sanitation, telecoms—is typically highly un-equal across groups. For many people indeveloping countries, lack of access toaffordable infrastructure services means liv-ing in isolation from markets and servicesand having intermittent or no supply ofpower or water for productive activities anddaily existence. This often results in a signifi-cant curtailment of economic opportunities.

While the public sector will in manycases remain the main source of funds forinfrastructure investments aimed at broad-ening opportunities for those who have thefewest, the efficiency of the private sectorcan also be harnessed. Although utility pri-vatizations have often been attacked forhaving unequal effects, the evidence indi-cates a more complex reality. Privatizationsin Latin America typically led to expansionsin access to services, particularly in electric-ity and telecommunications. In some cases,however, postprivatization increases inprices more than outweighed gains fromquality and coverage, leading to widespreadpopular discontent.

Privatizations are therefore a classic caseof a policy that may or may not make sense,depending on the local context. If the publicsystem is highly corrupt or inefficient, andone expects postprivatization regulatorycapacity to be adequate, it can be a usefultool. In other cases, poorly designed privati-zations may be captured, transferring pub-lic assets, at excessively low prices, into pri-vate hands.

Experience suggests that whether infra-structure services are provided by privateoperators or public utilities seems lessimportant for equity than the structure ofincentives facing providers and howaccountable these providers are to the gen-eral public. We argue that policymakers canimprove the equitable provision of infra-structure services by focusing on expand-ing affordable access for poor people andpoor areas—which often means workingwith informal providers and targetingsubsidies—and strengthening the gover-nance of the sector through the greateraccountability of providers and thestronger voice of beneficiaries.

Markets and the macroeconomyMarkets are central to shaping the potentialfor people to convert their assets into out-comes. When market transactions are influ-enced by the wealth or status of participants,they are both inequitable and inefficient—and can also influence the incentives fordifferent groups to expand their assets(chapter 9).

Financial markets. Captured banking sys-tems exchange favors: market power is pro-tected for a few large banks, which then lendfavorably to a few selected enterprises, whichmay not be those with the highest expectedrisk-adjusted returns. This may lie behind across-country association between greaterfinancial depth and lower income inequal-ity. Achieving more equal access to financeby broadening financial systems thus canhelp productive firms that were previouslybeyond the reach of formal finance.

These relations are only suggestive, how-ever, so the report draws on case studiesfrom middle-income economies, such asthe Republic of Korea, Malaysia, Mexico,and the Russian Federation, and poorereconomies, including Indonesia and Pak-istan, to provide more concrete evidence.These studies suggest an apparent paradox.Societies with extensive inequalities inpower and wealth, weak institutions, andcontrolled financial systems typically sufferfrom narrow financial sectors that are ori-ented to the influential and hide weak assetquality. Opening the financial system wouldseem to be an obvious solution. Liberaliza-tion, however, has also often been capturedby the influential or wealthy, in countriesranging from Mexico (in the early 1990s) totransition economies such as the CzechRepublic and Russia.

Gradual deepening and broadening thusneeds to be combined with stronger horizon-tal accountability (in regulatory structures),greater openness to societal accountability,and, where feasible, external commitmentdevices (such as the entry of Central Euro-pean and Baltic states to the EuropeanUnion). Programs targeted to the poor—such as microcredit schemes—can help butare no substitute for the overall broadening ofaccess.


Labor markets. Leveling the playing field inlabor markets consists of seeking the right(country-specific) balance between flexibilityand protection to provide more equal accessto equal employment conditions to as manyworkers as possible. Many countries havefairly extensive regulations and provisions forformal sector workers, and far fewer for “out-siders” in the unregulated (and often less safe)informal sector. There is usually a degree ofvoluntary movement between the sectors,and great diversity within the informal sectoritself, ranging from microentrepreneurs andsome of the self-employed with incomesabove formal sector workers to many withmuch worse employment conditions. Thismix leads to inadequate protection for poorerworkers, while regulations for formal workerscan reduce the flexibility of employment andoften are a poor deal for the workers them-selves, such as when job-related social secu-rity systems are inefficient.

Two broad labor market approaches arerelevant for equity. First, interventions inthe labor market should ensure effectiveapplication of the core labor standardsacross the whole market, implying no slaveor indentured work, no dangerous forms ofchild labor, and no discrimination. Workersshould be free to assemble and form associ-ations, and their unions should be free tohave an active role in bargaining. Second, inall areas the policy mix needs to be assessedin ways that balance protection (for allworkers) with allowances for the restructur-ing so central to dynamic growth andemployment creation.

Worker security is often provided by vari-ous excessively stringent forms of employ-ment protection legislation, which, in generalmake it costly to hire and, in some cases,make it even costlier to hire unskilled, young,and female workers—exactly those the lawsseek to protect. For many countries, less dis-tortionary and more inclusive policy alterna-tives are available, which may make the play-ing field more even in labor markets. Thesealternatives include unemployment insur-ance schemes (more likely in middle-incomecountries) and low-wage employmentschemes (ideally with an employment guar-antee), which can be applied successfully evenin poor countries or states.

Product markets. There is substantial het-erogeneity in the effects of opening a coun-try’s product markets to trade, at least in theshort to medium term. This can be due togeographic location, as illustrated by thevarying impact of trade liberalization inMexico (figure 7). This illustrates theimportance of interactions between domes-tic product markets and patterns of infra-structure provision. There are also oftenstrong interactions with skills in the labormarket. In many countries, opening totrade (often coinciding with opening to for-eign direct investment) has been associatedwith rising inequality in earnings in the pasttwo decades. This is especially so for mid-dle-income countries, notably in LatinAmerica. Opening to trade often boosts thepremium on skills as firms modernize theirproduction processes (skill-biased technicalchange, in the jargon of economists). This isbad for equity if the institutional contextrestricts the capacity of workers to shift intonew work—or limits future cohorts’ accessto education.

Macroeconomic stability. This report arguesthat there are two-way relationships betweeninequitable institutions and macroeconomiccrises, with mostly bad effects for equity and

Overview 15

Changes inhousehold welfare

> 5 percent4–5 percent2–4 percent0–2 percent

Figure 7 Better to be close to economic opportunitiesChanges in household welfare in Mexico, following trade liberalization in the 1990s

Source: Nicita (2004).

long-run growth. Weak and captured institu-tions are associated with a greater propensityfor countries to experience macroeconomiccrises. When crises occur, they can be costlyfor the poor, who have weaker instrumentsto manage shocks. In addition, crisis resolu-tion is often regressive, through a variety ofmechanisms (most of them not captured intraditional household survey instruments):declines in the labor share, at least for formalworkers; capital gains for those who get theirmoney out; and fiscal workouts that bail outthe influential at substantial cost. Suchbailouts must be paid for through somecombination of higher taxes and lowerspending. Because taxes are typically propor-tional and spending is often progressive atthe margin (notably in Latin America), thecost of bailouts is borne disproportionatelyby poorer groups. High inflation has alsobeen found to be both bad for growth andregressive in its impact.

A concern for equity would lead, in gen-eral, to a highly prudent stance on macroeco-nomic management and financial regulation.Populist macroeconomic policy, sooner orlater, is bad for equity and bad for growth.Policy design can increase equity through thepursuit of countercyclical fiscal policy, build-ing safety nets before a crisis, reducing riskylending, and supporting only smaller deposi-tors in bailouts. But, as in other policy areas,these responses need to be underpinned byinstitutional designs that combine greaterinstitutional freedom from political influence(such as independent central banks and auto-nomous financial regulatory agencies) withgreater information and debate in society.

The global arenaOne predetermined circumstance that mostpowerfully determines a person’s opportuni-ties for leading a healthy and productive life ishis or her country of birth. Global inequitiesare massive. Reducing them will depend pri-marily on domestic policies in poor countriesthrough their impact on growth and develop-ment. But global action can change externalconditions and affect the impact of domesticpolicies. In this sense, global and domesticactions are complementary.

We live in an integrated world in whichpeople, goods, ideas, and capital flow across

countries. Indeed, most policy advice given topoor countries over the last several decades—including that by the World Bank—hasemphasized the advantages of participatingin the global economy. But global markets arefar from equitable, and the rules governingtheir functioning have a disproportionatelynegative effect on developing countries(chapter 10). These rules are the outcome ofcomplex negotiating processes in whichdeveloping countries have less voice. More-over, even if markets worked equitably,unequal endowments would limit the abilityof poor countries to benefit from globalopportunities. Leveling the global economicand political playing fields thus requires moreequitable rules for the functioning of globalmarkets, more effective participation of poorcountries in global rule-setting processes, andmore actions to help build and maintain theendowments of poor countries and poorpeople.

The report documents some of the manyinequities in the functioning of global mar-kets for labor, goods, ideas, and capital.Unskilled workers from poor countries, whocould earn higher returns in rich countries,face great hurdles in migrating. Developing-country producers face obstacles in sellingagricultural products, manufactured goods,and services in developed countries. Patentprotection restricts access to innovations(particularly drugs) for poor countries, whilenew research is strongly oriented to thediseases of richer societies. Rich-countryinvestors often get better deals in debt crises.In most cases, more equitable rules wouldbring benefits to developed- and developing-country citizens. Benefits vary across marketsand countries, with those from greater legalmigration likely to be greatest (and to accruedirectly to migrants) and those from tradelikely to accrue mostly to middle-incomerather than the least developed countries.

The report discusses options to reduceinequities in the functioning of global mar-kets, including the following: allowing greatertemporary migration into OECD countries,achieving ambitious trade liberalizationunder the Doha Round, allowing poor coun-tries to use generic drugs, and developingfinancial standards more appropriate todeveloping countries.


The international laws that govern globalmarkets are the product of complex negotia-tions. In some cases, as for human rightscovenants, the processes generating the lawsare perceived to be fair. In other cases,processes and outcomes are perceived asunfair, even though the formal regulationsare equitable. Within the World Trade Orga-nization (WTO), for example, each countryhas a vote and each can block proceedings.Even so, WTO processes are at times per-ceived as unfair because of the underlyingpower imbalance between strong commercialinterests and the public interest, in bothdeveloped and developing countries. Theseimbalances manifest themselves, for instance,in the number of staff employed in Genevaby different WTO members. More effectiverepresentation of poor countries in globalinstitutions would help improve processesand may lead to more equitable rules.

The impact of reducing imperfections inglobal markets varies by country. The largerand fast-growing developing countries standto benefit significantly from freer globaltrade, migration, and capital flows, helpingthem sustain fast growth (while equitabledomestic policies both help underpin long-run growth and the broad internal sharing ofthis growth). Countries left behind in theglobal economy stand to benefit much lessfrom global markets in the short run and willcontinue to rely on aid. For them, globalaction that helps compensate for unequalendowments is truly essential. Action to buildendowments is primarily domestic, throughpublic investments in human development,infrastructure and governance structures. Butglobal action can support domestic policiesthrough resource transfers in the form of aid,which is not offset by debt repayments, andinvestments in global public goods, particu-larly global commons.

Aid levels need to be bolstered in line withthe commitments rich countries made at the2002 Monterrey Conference and concreteplans should be made to reach the target of

devoting 0.7 percent of gross national incometo aid. Larger volumes of aid will only help,however, if aid is effective in alleviating con-straints and spurring development in therecipient countries. Greater effectiveness canbe achieved by emphasizing results, movingaway from ex ante conditionality, and pro-gressively shifting design and managementfrom donors to recipients. Aid should not beundermined by debt, for debt reduction thatis not financed by additional resources canactually undercut effective aid programs.Innovative mechanisms to expand develop-ment assistance should be explored, includ-ing global taxes and private contributions.

Equity and developmentBringing equity to the center of developmentbuilds on and integrates the major emphasesin development thinking of the past 10 to 20years—on markets, on human development,on governance, and on empowerment. It isnoteworthy that this year equity is the focus ofboth this World Development Report and theHuman Development Report of the UnitedNations Development Programme. The pleafor a more level playing field in both the pol-itics and the economies of developing coun-tries serves to integrate the World Bank’stwin pillars of building an institutional cli-mate conducive to investment and empow-ering the poor. By ensuring that institutionsenforce personal, political, and propertyrights for all, including those currentlyexcluded, countries will be able to draw onmuch larger pools of investors and innova-tors, and be much more effective in provid-ing services to all their citizens. Greaterequity can, over the long term, underpinfaster growth. This can be helped by greaterfairness in the global arena, not least throughthe international community’s meeting itscommitments made at Monterrey. Fastergrowth and human development in poorercountries are essential to reducing globalinequity and to reaching the MillenniumDevelopment Goals.

Overview 17

Nthabiseng and Pieter—the hypotheticalSouth African children who opened thereport’s overview—are not unusual exam-ples of people who face highly disparateinitial opportunities. A girl born to a lower-caste family of nine in the slums of Dhakahas vastly different opportunities from aboy born to well-educated and affluentparents in the well-heeled neighborhoods.An AIDS orphan in rural Zimbabwe isalmost certain to have fewer chances andchoices in life than a compatriot born tohealthy and well-educated parents inHarare. Those differences are even greateracross borders: an average Swiss, American,or Japanese child born at the same instantas Nthabiseng will have incomparablysuperior life chances.

Such staggering inequalities in opportu-nity are intrinsically objectionable, andalmost every culture, religion, and philo-sophical tradition has developed argu-ments and beliefs that place great value onequity for its own sake. In addition, Part IIof this report will argue that we now haveconsiderable evidence that equity is alsoinstrumental to the pursuit of long-termprosperity in aggregate terms for society asa whole. But before one can describeinequity, or assess its impact on growth anddevelopment, a clear definition of the termis needed.

This introductory chapter presents ourworking definition of equity and briefly dis-cusses its main component—equality ofopportunity. It then turns from our centralnormative concepts to one of the report’skey positive concepts: inequality traps. Aninequality trap encapsulates the mutuallyreinforcing nature of various inequalities,which leads to their persistence and to aninferior development trajectory.

Equity and inequality ofopportunity: the basic conceptsWhat is equity? As with any normative con-cept, the word “equity”means different thingsto different people. It is a difficult concept,with a history of different interpretations,varying by country and academic discipline.Economists link equity to questions of distri-bution. Lawyers tend to think of principlesmeant to correct the strict application of thelaw, which may lead to an outcome judged tobe unfair in specific circumstances. Philoso-phers have produced the most headway in thethinking about equity. Indeed, the attributesthat would characterize a just and fair societylie at the foundation of Western political phi-losophy, from Plato’s Republic and Aristotle’sPolitics onward. Equity is also central to mostof the world’s great religions, including Bud-dhism, Christianity, Hinduism, Islam, andJudaism, as well as to most other faith tradi-tions. More recently, social choice theory, andthe closely related domain of welfare eco-nomics, have been concerned with the aggre-gation of preferences into some form of“social optimum.”

Summarizing such long-standing andnuanced characterizations is perilous, butthe common denominator of these manydifferent views is that equity relates to fair-ness, whether locally in families and com-munities, or globally across nations. We donot dwell on the different approaches toequity here, but we do elaborate on them inchapter 4, which reviews various categoriesof evidence in support of the intrinsicimportance of equity. For this report, wethink of equity as being defined in terms oftwo basic principles:

• Equal opportunity. The outcome of aperson’s life, in its many dimensions,

18

Introduction

1c h a p t e r

should reflect mostly his or her effortsand talents, not his or her background.Predetermined circumstances—gender,race, place of birth, family origins—andthe social groups a person is born intoshould not help determine whether peo-ple succeed economically, socially, andpolitically.1

• Avoidance of absolute deprivation. Anaversion to extreme poverty, or indeed aRawlsian2 form of inequality aversion inthe space of outcomes, suggests thatsocieties may decide to intervene to pro-tect the livelihoods of its neediest mem-bers (below some absolute threshold ofneed) even if the equal opportunityprinciple has been upheld. The roadfrom opportunities to outcomes can betortuous. Outcomes may be low becauseof bad luck, or even because of a person’sown failings. Societies may decide, forinsurance or for compassion, that itsmembers will not be allowed to starve,even if they enjoyed their fair share ofthe opportunity pie, but things some-how turned out badly for them.

The equal opportunity principle is con-ceptually simple: circumstances at birthshould not matter for a person’s chances inlife. But to measure inequality of opportu-nities is much harder. Chapter 2 briefly dis-cusses one approach, which decomposesobserved income inequality into one partthat can be attributed, in a statistical sense,to predetermined circumstances—such asrace, place of birth, and parental back-ground—and one part that cannot. Thefirst component captures a lower boundvalue for the opportunity share of incomeor earnings inequality. But it is generallyvery difficult to measure things like familybackground precisely: years of schoolingand broad occupational categories areimperfect proxies for a family’s endow-ments of human, physical, and social capi-tal.

A superior approach would be to capturethe inherently multidimensional and group-based nature of inequality of opportunity.How do the factors that determine a per-son’s chances in life—the access to healthand educational opportunities, the ability to

connect to the rest of the world, the qualityof the services available, the way institutionstreat them—relate to one another? And howdo these factors vary across groups? Such anapproach would require a focus not only onthe dispersion of univariate distributions(such as income inequality or life expectancy)but also on the correlations among them(how do health outcomes vary across socio-economic groups?). This is the approachtaken in most of chapter 2, which summa-rizes information on inequalities (with em-phasis on the plural) in the various buildingblocks of opportunity and on their interre-lationships.

In taking this route, the report recognizesthat predetermined circumstances, or mem-bership in prespecified groups, affect oppor-tunities in two ways:

• The circumstances of one’s birth affect theendowments one starts with, including allkinds of private assets, such as physicalwealth (including land and financialassets), family background (the human,social, and cultural capital of one’s par-ents), and access to public services andinfrastructure (sometimes referred to asgeographic capital).

• Group membership and initial circum-stances also affect how one is treated bythe institutions with which one mustinteract. Two individuals may both live inareas where formal labor markets exist,where courts are agile, and where a policeforce is present. But if these two (other-wise identical) people, because of theirgender, race, religion, sexual orientation,political beliefs, residential address, orany other morally irrelevant reason, aredifferently rewarded for the same work inthe labor market, are discriminatedagainst by the court of law, or are treatedwith bias by the police force, then therules are not being applied fairly. There-fore, these two people do not have thesame opportunity sets. Equity alsorequires fairness in processes.

Endowments that are less unequal,processes that are fair, and protection fromdeprivation are not always mutually con-sistent. At the policy level, there may be

Introduction 19

tradeoffs among them. Indeed, some policiesor institutions developed to further one ofthe principles may compromise the other.

For example, a policy of affirmativeaction that seeks to correct past inequitiesin the access to educational opportunitiesfor one group—to equalize endowments—may imply that individuals of greater merit(but from another group) are excluded, cre-ating unfair processes. For another exam-ple, the taxes needed to raise governmentrevenues to make transfers to poor individ-uals (desirable to avoid deprivation) expro-priate some fruits of the efforts of hard-working men and women. This might beseen as violating property rights or therights to appropriate the fruits of one’s ownlabor, again creating unfair processes.

Whenever such tradeoffs exist—which ismost of the time—no textbook policy pre-scription can be provided. Each society mustdecide the relative weights it ascribes to eachof the principles of equity and to the effi-cient expansion of total production (orother aggregate). This report will not pre-scribe what is equitable for any society. Thatis a prerogative of its members to be under-taken through decision-making processesthey regard as fair.

Inequality trapsIf people care about equity, and if politicalsystems aggregate people’s views into socialpreferences, why don’t the distributions weobserve represent optimal choices? Why doinequalities of opportunity persist, if theyare both unfair and inimical to long-termprosperity? And how do these inequalitiesreproduce themselves? The short answer isthat political systems do not always assignequal weights to everyone’s preferences.Policies and institutions do not arise from abenign social planner who aims to maxi-mize the present value of social welfare.They are the outcomes of political economyprocesses in which different groups seek toprotect their own interests. Some groupshave more power than others, and theirviews prevail. When the interests of domi-nant groups are aligned with broader col-lective goals, these decisions are for thecommon good. When they are not, the out-comes need be neither fair nor efficient.

The interaction of political, economic,and sociocultural inequalities shapes theinstitutions and rules in all societies. Theway these institutions function affects peo-ple’s opportunities and their ability to investand prosper. Unequal economic opportuni-ties lead to unequal outcomes and reinforceunequal political power. Unequal powershapes institutions and policies that tend tofoster the persistence of the initial condi-tions (figure 1.1).

Consider the status of women in patriar-chal societies. Women are often denied prop-erty and inheritance rights. They also havetheir freedom of movement restricted bysocial norms that create separate “inside” and“outside” spheres of activity for women andmen. These social inequalities have economicconsequences: girls are less likely to be sent toschool; women are less likely to work outsidethe home; women generally earn less thanmen. This reduces the options for womenoutside marriage and increases their eco-nomic dependence on men. The inequalitiesalso have political consequences: women areless likely to participate in important deci-sions within and outside the home.

These unequal social and economicstructures tend to be readily reproduced. Ifa woman has not been educated and hasgrown up to believe that “good, decent”women abide by existing social norms, sheis likely to transmit this belief to her daugh-ters and to enforce such behavior amongher daughters-in-law. An inequality trapmay thus prevent generations of womenfrom getting educated, restrict their partici-pation in the labor market, and reduce theirability to make free, informed choices andto realize their potential as individuals. This


Politicalinequalities

InstitutionsSocioculturalinequalities

Economicinequalities

Figure 1.1 The interaction of political, economic, andsociocultural inequalities

reinforces gender differences in power thattend to persist over time.

Similarly, the unequal distribution ofpower between the rich and the poor—between dominant and subordinate groups—helps the rich maintain control over re-sources. Consider an agricultural laborerworking for a powerful landlord. Illiteracyand malnourishment may prevent him frombreaking out of the cycle of poverty. But he isalso likely to be heavily indebted to hisemployer, which puts him under the land-lord’s control. Even if laws were in place thatwould allow him to challenge his landlord’sdictates, being illiterate, he would find it diffi-cult to navigate the political and judicial insti-tutions that might help him assert his rights.In many parts of the world, this distancebetween landlords and laborers is com-pounded by entrenched social structures:landlords typically belong to a dominantgroup defined by race or caste, tenants andlaborers to a subordinate group. Becausemembers of these groups often face severeconstraints from social norms against inter-marrying, group-based inequalities are per-petuated across generations.

Poor individuals in geographically isolatedregions and racial and ethnic minorities alsohave less political power and less voice inmany countries. This affects their ability topropose and implement policies that wouldreduce their disadvantage, even if such poli-cies might be growth-enhancing for the coun-try.3 The correlations between the unequaldistribution of assets, opportunities, andpolitical power give rise to a circular flow ofmutually reinforcing patterns of inequality.Such a flow, and its associated feedback loops,help inequalities persist over long periods—even if they are inefficient and deemed unfairby a majority of the population.4

Economic and political inequalities arethemselves embedded in unequal social andcultural institutions.5 The social networksthat the poor have access to are substantiallydifferent from those the rich can tap into. Forinstance, a poor person’s social network maybe geared primarily toward survival, withlimited access to networks that would linkhim or her to better jobs and opportunities.The rich, by contrast, are bequeathed withmuch more economically productive social

networks that maintain economic rank. Richparents can use their social connections toensure that their child gets into a goodschool, or they can call a few good friends tomake sure that their son gets a good job. Con-versely, poor parents are more subject tochance. Connections open doors and reduceconstraints.

Social networks are closely allied with cul-ture. (By “culture” we mean aspects of lifethat deal with relationships among individu-als within groups, among groups, andbetween ideas and perspectives). Subordinategroups may face adverse “terms of recogni-tion,” the framework within which they nego-tiate their interactions with other socialgroups.6 One obvious expression is explicitdiscrimination that can lead to an explicitdenial of opportunities and to a rationalchoice to invest less at the margin.

But the process may also be less overt. Aperson born into a low social class or asocially excluded group may adopt the domi-nant group’s value system.7 Religious beliefsmay propel this: women may take on gen-dered beliefs about their economic and socialrole, and low castes may absorb the uppercastes’ view of their “inferior” status. Inschools, a stigmatized group may face a“stereotype threat,” adopting the dominantgroup’s view of their ability to perform incognitive tests or in occupations historicallycontrolled by dominant groups.8 This canaffect a discriminated group’s “capacity toaspire.”9 It also implies that “voice,” the capac-ity of individuals to influence the decisionsthat shape their lives, is also unequally dis-tributed and that “effort” and “ability” are notnecessarily exogenous (predetermined).10

The existence of these inequality traps—with mutually reinforcing inequalities in theeconomic, political, social, and culturaldomains—has two main implications for thisanalysis. The first implication is that, becauseof market failures and of the ways in whichinstitutions evolve, inequality traps can affectnot only the distribution but also the aggre-gate dynamics of growth and development.This in turn means that, in the long run,equity and efficiency may be complements,not substitutes.11

Capital, land, and labor markets in devel-oping countries are imperfect. Informational

Introduction 21

asymmetries and contract enforcement prob-lems imply that some people with goodproject ideas (and thus a potentially highmarginal product of capital) end up con-strained in their access to capital. This, evenas other people earn a lower return on their(more abundant) capital. In agriculture,land market failures mean that some farm-ers exert too little effort on some plots(where they are sharecropping), and toomuch effort on other plots (which theyown).12 Investment in human capital canalso be allocated inefficiently, because ofintrahousehold disputes, because credit-constrained households lack the resourcesto keep their children healthy and in school,or because discrimination in the labor mar-ket reduces the expected returns to school-ing for some groups. What do such diversemarket failures have in common? Theycause differences in initial endowments—such as family wealth, race, or gender—tomake investment less efficient.

There also are political and institutionalreasons why equity and efficiency are long-term complements. Markets are not theonly institutions in society. The function-ing of states, legal systems, and regulatoryagencies—indeed, of all the institutionsthat assign and enforce property rights andmediate conflicts among citizens—is influ-enced by the distribution of political power(or influence, or voice) in society. Unequaldistributions of control over resources andof political influence perpetuate institu-tions that protect the interests of the mostpowerful, sometimes to the detriment ofthe personal and property rights of others.13

Those whose rights are not protectedhave little incentive to invest, perpetuatingpoverty and reproducing inequality. Con-versely, good institutions that protect andenforce personal and property rights for allcitizens have led to higher sustained eco-nomic growth and long-term prosperity.Equity can, once again, help societies growand develop.

This does not mean, of course, thatefficiency-equity tradeoffs have somehowbeen abolished. In some cases, equityenhancements bring immediate—as well aslong-run—benefits for efficiency. If wereduce discrimination against women in

one segment of the labor market, such asmanagement, and if this brings a new poolof talent into that segment, efficiency islikely to increase even in the short run.14 Inother cases, however, expanding the oppor-tunity sets for the disadvantaged mayrequire more costly redistribution. Tofinance better-quality schooling for thosewho have the least educated parents, andwho attend the worst schools, it may benecessary to raise taxes on other people.The basic economic insight that such taxa-tion distorts incentives remains valid. Suchpolicies should be implemented only to theextent that the (present) value of the long-run benefits of greater equity exceed theefficiency costs of funding them.15

The point is that some of these long-term benefits of pursuing greater equity areignored in the conceptual calculus of policydesign. The fact that better-schooled chil-dren who are poor and from a racial minor-ity will be more productive is usually takeninto account. But the fact that they mayacquire greater political voice and helpmake social institutions more inclusive—which, in turn, may increase the stake ofthat group in society, potentially leading to greater trust, less conflict, and moreinvestment—may not be. To the extent thatsuch indirect (but important) benefits ofequity-enhancing policies are ignored, toofew of them are pursued—even assuming apurely benevolent government.

By placing equity and fairness as centralelements of an efficient development strat-egy, developing countries will be better ableto reach sustainable growth and develop-ment trajectories. Such equitable growthpaths are likely to lead to faster reductionsin the many dimensions of poverty, the cen-tral objective of development everywhere.16

The second implication of the existenceof inequality traps is that no real-life policyor institution is entirely exogenous: noexisting organization or application of apolicy idea has been implemented on apurely technocratic basis. All policies andinstitutions exist because the political sys-tem has brought them into being or allowedthem to survive. The political systemreflects the distribution of power and voiceattained at a particular time and place. This


distribution is, in turn, influenced by thedistribution of wealth, income, and otherassets and outcomes in that society. Such“circular causality” for wealth, income,social and cultural capital, and power,mediated through institutions, evolvesthroughout time and history.

Acknowledging history and social andpolitical institutions is crucial to avoid pol-icy mistakes. But a fatalistic view of theworld is not only wrong, but also counter-productive. To propose policies withoutunderstanding history, or the specific con-text for developing these policies, oftenleads to failure. But this acknowledgment isnot equivalent to the view that no policiesshould be suggested at all. Such a view failsto recognize how purposeful social andpolitical action can achieve significant pol-icy and institutional changes—and wouldresult in fatalistic inaction.

History is not endlessly repetitive and, asthis report documents, many countries havetaken on the challenge of breaking inequalitytraps with some success. Groups have alsochanged their circumstances or changedsocial and political institutions. Considerthe civil rights movement in the UnitedStates, the democratic overthrow ofapartheid in South Africa, the more partici-patory budgeting practices in some Brazil-ian cities, and the reforms in access to land,education, and local government in theIndian state of Kerala. The challenge forpolicy is to ask when and how such changescan be supported.

A brief preview of the ReportPart I summarizes evidence on inequitywithin and across countries. Part II asks whyequity matters for development, both intrin-sically and instrumentally. Part III turns tothe policy implications. If unequal opportu-nities and absolute deprivation are inimicalto long-term prosperity—as well as intrinsi-cally objectionable—there is scope for policyand institutional reform aimed at leveling theeconomic and political playing fields.

An equity lens and the focus on levelingthe playing field add three basic points. First,redistributions from richer and more power-ful groups to poorer groups that face morelimited opportunities are sometimes neces-sary and should be pursued. Second, whenconsidering policy tradeoffs between equityand efficiency, the full long-term benefits ofequity—including on the development ofbetter and more inclusive institutions—needto be taken into account. Third, all categoriesof economic policy—macro and micro—have effects on both efficiency (and growth)and equity (and distribution). Because ourultimate goal is the reduction of povertythrough the equitable pursuit of prosperity,the policy suggestions in these chapters areconsistent with good poverty-reduction poli-cies, which the World Bank has been advocat-ing since at least the publication of the WorldDevelopment Report 1990.17 These sugges-tions are also in line with the 2000 WorldDevelopment Report’s three pillars of oppor-tunity, empowerment, and security.18

Introduction 23

Inequity within and across countries

IP A R T

CasteCaste in Palanpur defines opportunitiesand determines the activities villagers pur-sue, even independent of occupation, edu-cation, and other standard household char-acteristics. The three largest castes inPalanpur are Thakurs, Muraos, and Jatabs.

At the top is a martial caste known asthe Thakurs, which accounted for about aquarter of the population in 1993. Thakursare disproportionately represented in jobssuch as the army and police that accord wellwith their martial past. They are typicallyaverse to wage employment in the village,because this would place them in a subordi-nate position. Alert to nonfarm employ-ment opportunities outside the village, theyare well placed to take advantage of them,thanks to stronger information and socialnetworks.

Just below the Thakurs is a cultivatingcaste, the Muraos, also accounting for aquarter of the population. Muraos are tra-ditional cultivators who have continued tospecialize in agriculture. Very hardworking,they have seen a rapid rise in wealth andeconomic status in the village. While theymay still not enjoy the same social status asThakurs, they have become more prosper-ous and now challenge the previouslyunquestioned political and economic dom-inance of the Thakurs.

At the bottom are the scheduled castesknown as Jatabs, accounting for 12 per-cent of the population. Traditionally“untouchable” leather workers who nowengage primarily in agricultural wagelabor, Jatabs have not seen any of thesocial mobility of the Muraos. Theyremain a caste apart, with little or no land,poor education, and little access to non-farm employment outside the village.Despite some slight improvement over theyears, Jatabs continue to endure manyforms of discrimination, including thatfrom government officials.

The village of Palanpur, in the northIndian state of Uttar Pradesh, hasbeen the subject of intensive study

by a group of development economistsbetween the late 1950s and early 1990s.1

Researchers visited the village repeatedlyand collected detailed quantitative andqualitative information. While a single vil-lage study covering a specific period of timecannot be used to draw inferences aboutdevelopment in rural India as a whole, itdoes provide a distinct window into thekind of processes that can shape growthand equity over time.

The study documents modest economicprogress over time with slow growth in percapita incomes and some declines in incomepoverty. But alongside this sluggish growth isevidence of stagnation and even deteriora-tion along other dimensions of well-being.

Different groups of villagers, defined bysuch predetermined characteristics as casteor gender, face radically different opportuni-ties for economic and social mobility. Theireconomic endowments differ markedly, asdo their education, health, occupationalmobility, and capacity to influence andshape social and political institutions in thevillage. Disadvantage in one dimension ofopportunity is generally reinforced by disad-vantage in others, combined in a way thatperpetuate the stark inequities over genera-tions.

These deep-seated inequalities ofopportunity shape, and are shaped by, mar-ket imperfections in the village, resulting insuboptimal investments and impedinggrowth. Inequalities are also mirrored invillage institutions. State and central gov-ernment policies that were introduced inthe village were inevitably filtered through ahighly unequal distribution of power andinfluence. Rather than stimulating broadeconomic and social progress, public policyhas simply reproduced the prevailing pat-terns of inequality.

GenderGender inequalities in Palanpur are pro-nounced. In 1993 there were 84 females forevery 100 males, strikingly lower than inmost parts of the world (where the ratio isusually greater than one). Child mortalityrates are much higher among girls thanamong boys. As the researchers reported,“We witnessed several cases of infant girlswho were allowed to wither away and die incircumstances that would undoubtedlyhave prompted more energetic action in thecase of a male child.”2

Young girls leave their village to jointheir husband’s family. Marriage is “the giftof a daughter.” In the new household, thegirl is acutely vulnerable with no income-earning opportunities, no property, no pos-sibility of returning home permanently.Giving birth to a child improves her sta-tus—particularly if it’s a boy. But familyplanning practices are limited, leading tohigh fertility rates and short birth-spacing.Repeated pregnancies take an enormoustoll on women’s general health and puttheir lives at risk at the time of delivery. Oldage is strongly associated with widowhood,in part because of the typically large age dif-ference between husbands and wives. Tosurvive, widows depend overwhelminglyon adult sons.

The participation of women in the laborforce in Palanpur is extremely low. Of 313women age 15 or older in 1993, only 14 hadanything other than domestic work as theirprimary or secondary occupation. This lowfemale participation in the labor force andsociety, more generally, has extensive conse-quences. For example, the survival disad-vantage of girls compared with boys tendsto narrow only when adult women havewider opportunities for gainful employ-ment. Similarly, the virtual exclusion ofwomen from most representative institu-tions in Palanpur has limited the focus andquality of local politics and public action.

Inequality traps stifle economic development in a north Indian villageVillagers differ markedly from one another in the opportunities they have to improve their welfare and in their abilities touse the assets and endowments available to them. Mirrored in village economic and social institutions—and in the politi-cal processes for seeking change—these deep-seated inequalities have prevented the village from improving human devel-opment and accelerating economic growth.

Palanpur f o c u s 1 o n

often less strenuous and demanding than inagriculture. Access to nonfarm jobs is farfrom equal, however. Workers who wish toobtain a regular job generally have to paybribes and, more important, get a recom-mendation or introduction from a friend orrelative. Such rationing by personal con-tacts and influence implies that people withlow social status tend to be at a disadvan-tage in the competition for nonfarm jobs,even for given education levels, skills, andendowments.

The least advantaged segments of thelabor force in Palanpur are highly repre-sented in agricultural wage labor. Casualwage labor in agriculture can be describedas a “last-resort” occupation, one taken upby those who have no significant alterna-tive. Agricultural wage rates have risen overtime, but slowly, and there are prolongedperiods of seasonal unemployment.

Econometric analysis indicates that—controlling for a large number of house-hold characteristics (caste, demographiccharacteristics, education, land, and soon)—the probability of engagement inagricultural labor is 50 to 60 percent higherfor households that had engaged in thisoccupation a decade earlier. Occupationalinequalities thus result in income inequal-ity, and they persist over long periods.

Incomes, assets, and liabilitiesPer capita incomes in Palanpur have grownat around 2 percent a year between 1957–8and 1983–4 and income poverty fell fromaround 47 to 34 percent during this period.Incomes in the village are distributed aboutas unequally as they are in India as a whole,and income inequality has remained rela-tively stable over time.

An assessment of economic inequalitiesbased on wealth provides a different pic-ture. Ownership of durables has expanded,and the value of land and other productiveassets has grown, implying a significant risein gross wealth. But there has also been adramatic and uneven expansion of liabili-ties. Inequality in the distribution of netwealth has widened in Palanpur from aGini of around 0.46 in 1962–3 to a conserv-atively estimated 0.55 in 1990.

Many of the liabilities come from pub-licly provided and subsidized credit sourcesthat have expanded sharply over time, butthat have been associated with pervasivecorruption. Disadvantaged groups, such as

SchoolingInequalities in education are wide, decliningonly slowly. In the late 1950s, just under 20percent of males age seven or older, and only1 percent of females, were literate. By 1993,male literacy had risen to 37 percent andfemale literacy to just below 10 percent. Yeteducation is clearly of great value in Palan-pur. Years of schooling strongly increase thelikelihood that an individual will findemployment in a regular job outside the vil-lage. Among farmers, too, direct observationstrongly suggests that better-educated farm-ers in Palanpur have been crucial in techno-logical innovation and diffusion.

The perceived value of female educationis quite different from that for boys, becausegirls are expected to spend most of theiradult life in domestic work. Although thereis good evidence of the benefits of educa-tion in domestic activities, it is not clearthat the effects of maternal literacy on childhealth, for example, are recognized. Even ifbenefits are correctly perceived, they mightnot be of direct interest to the parents,because daughters are “transferred” fromthe village when they marry. Those whobear the costs of female education thusshare little in the benefits.

The upper-caste Thakurs have a view(adopted by many others) that education isnot important or even suitable for the lowercastes. Blatant forms of discriminationagainst children from disadvantaged casteshave disappeared from the schooling sys-tem, but subtler forms of discriminationhave remained—for example, the high-caste teacher considered any form of con-tact with Jatab children as “repulsive,”which likely affected his or her rapport withthem and probably discouraged their atten-dance.

WorkOccupational divisions in Palanpur havewidened as the village has shifted from anoverwhelmingly agricultural economy toone in which nonagricultural activities havecome to account for 30 to 40 percent of vil-lage income. In 1957–58 some 13 villagers(of 528) were employed in regular or semi-regular nonfarm jobs. By 1993, this numberhad increased more than four times to 57jobs (the total population had only dou-bled).

Outside jobs are associated with higherand more stable incomes, and the work is

the Jatabs, are the principal targets of fraud-ulent accounting practices that haveresulted in a dizzying accumulation ofdebts and dramatically raised the cost ofborrowing for such households. Thosewithout access to cheap formal credit haveto fall back on private moneylenders, athigh interest rates.

Collective inactionThe different bases of social division inPalanpur have led to multiple solidaritiesand oppositions. The village society ishighly fragmented, with few solid rallyingpoints for collective action, whether coop-erative or adversarial. The limited reach ofcollective action, in turn, is responsible forsome of the most serious failures of itsdevelopment. For example, the villageassembly (panchayat) is constituted everyfew years, but it rarely meets. In 1984 it wasmade obligatory that at least one womanparticipant be selected, but in Palanpur sheis never consulted and has never attendedany panchayat meetings. All decisions andresponsibilities are effectively taken by a vil-lage headman, who has always come fromone of the privileged groups. There also isample scope for self-serving patronage andfraud. Modern arrangements (elections,reserved seats for low castes, and women onthe panchayat) have not profoundly alteredthe elitist and nonparticipatory character oflocal politics in the village.

The dominance of privileged groupsover collective institutions has had far-reaching consequences. Between the late1950s and early 1990s, no fewer than 18types of government-provided programswere introduced to the village: a publicworks road-building program, free school-ing, free basic health care, old-age pensions,a fair-price shop, a farmer’s cooperative,and so on. Most of them remained non-functional, particularly when there was aredistributive component. Only programsthat enjoyed strong backing from the polit-ically advantaged in the village wereallowed to succeed. The authors of thestudy conclude, “There is little prospect ofmajor improvement in the orientation andachievements of government interventionwithout a significant change in the balanceof political power, both at the state and atthe local level.”3

Source: Drèze, Lanjouw, and Sharma (1998).

Focus on Palanpur 27

Across the world, individuals and groupsface highly unequal opportunities to betterthemselves economically and socially.Inequalities, as such, might not be of particu-lar concern if outcomes varied for reasonsthat had to do mainly with individual efforts.But, taking our cue from the first chapter, weare concerned here with systematic differ-ences in opportunities for individuals andgroups who differ only in skin color, caste,gender, or place of residence, predeterminedcharacteristics that can be argued to be“morally irrelevant.” As illustrated in focus 1,on the Indian village of Palanpur, when suchinequalities of opportunity are pronounced,they are often reproduced over time and notonly affect welfare directly but also act to sti-fle human development and economicgrowth.

On the basis of what predeterminedcharacteristics should groups be definedsuch that we would not want to see sys-tematic differences in their opportunities?Clearly there is no single answer. Roemer(1998) argues that society has to make thischoice through some kind of ethical andpolitical process. The circumstances couldinclude social origin variables outside anindividual’s control, such as sex, race, eth-nicity, caste, parental education and occu-pation, wealth, or place of birth. Cogneau(2005) notes that a society’s choice ofcircumstances establishes a direct linkbetween equality of opportunities and theintergenerational transmission of out-comes. In this chapter, we are largely com-pelled to let data availability dictate thegroup definitions we consider. We canthus present only a partial, and often rudi-mentary, picture of the full range ofinequity that might exist in a country.Because we wish not only to look within a

country but also to compare across coun-tries, we use group definitions of broadrelevance.

Although economic inequalities areclearly part of the story, this chapter goesbeyond incomes to emphasize inequalitiesin key dimensions of opportunity, such ashealth, education, and the freedom andcapacity of people to participate in andshape society. There is a special concernwith inequalities that tend to perpetuatedifferences across individuals and groupsover time, within and across generations.These result in “inequality traps” that arepervasive in many countries. Such inequal-ity traps reinforce our concern with equityon intrinsic grounds, but they can also beparticularly detrimental to the developmentprocess, because they act to curtail eco-nomic dynamism.

A key objective here is to show howinequalities combine, interact, and arereproduced through interlinked economic,political, and sociocultural processes. Indi-viduals and groups differ markedly in theirpower to influence these processes; indeed,they differ even in their capacity to aspireto such influence. The report emphasizesthat such “agency” is a dimension ofopportunity, alongside education, health,and wealth. And inequalities of agency arecentral in explaining how inequalities ofopportunity are transmitted over time(box 2.1).

This chapter presents evidence of ahigh degree of inequality of opportunityin many developing countries—inequali-ties manifest in a variety of dimensions,such as health, education, and income. Itthen focuses on the specific dimension ofinequality of power, or agency. Through-out the chapter, we emphasize that

28

Inequity within countries:individuals and groups

2c h a p t e r

inequalities in different dimensions caninteract with, and reinforce, one anotherover time. To highlight these connections,we end by focusing on the specific case ofgender inequity.

Inequalities in healthAlongside the intrinsic importance ofhealth as a dimension of welfare, poorhealth can directly influence an individ-ual’s opportunities—his or her earningscapacity, performance at school, ability tocare for children, participation in commu-nity activities, and so on. This importantinstrumental function of health impliesthat inequalities in health often translateinto inequalities in other dimensions ofwelfare. And these inequalities are repro-duced over time. We focus here on chil-dren, while recognizing that differences insocial status, wealth, and health also mat-ter for adults.

Demographic and Health Survey (DHS)data indicate that health status variessharply across population groups. To whatextent does it vary across population groupsdefined by characteristics that are predeter-mined and arguably have no moral rele-vance? We draw on DHS data from 60countries to examine how the health ofchildren varies across population groupsdefined by mother’s education, rural orurban residence, and parent’s economic sta-tus, proxied by an index of household own-ership of consumer durables. (We look fur-ther at cross-country differences in healthin chapter 3.)

Infant mortality. For these countries,infant mortality rates vary markedly—from a low of around 25 per 1,000 livebirths in Colombia and Jordan, to morethan 125 in Mali, Niger, and Mozambique(figure 2.1). But even where overall infantmortality rates are high, the figures forchildren whose mothers have a secondaryeducation or higher are dramaticallylower. The risk of death among childrenwith well-educated mothers in Mali, forexample, is about the same as that for theaverage child in Indonesia. And while theoverall infant mortality rate in Brazil lies

below 50 (estimates from 1996), the ratefor children whose mothers have not beeneducated is roughly twice as high. Furtheranalysis, not reported here, indicates thatinfant mortality rates are also sharply differ-entiated across population groups definedby rural-urban residence and economic sta-tus, proxied by asset ownership.

Stunting. Another dimension of health,extreme stunting (with height-for-agebelow three standard deviations from thereference population), also varies markedlyacross countries. Overall rates are as high as 30 percent in Pakistan and the Rep-ublic of Yemen, but negligible in Trinidadand Tobago and very low in Jordan, Arme-nia, Brazil, and Kazakhstan (figure 2.2).

Inequity within countries: individuals and groups 29

As a prelude to the themes in this chapter,we describe one attempt to quantify thelevel and persistence of inequalities ofopportunity in Brazil, based on nationallyrepresentative household survey data. Brazilwas selected for a reason.With a Gini coeffi-cient of per capita incomes just below 0.6and persistent over time, it is generally per-ceived to be one of the world’s mostunequal countries.*

Brazil’s main household survey, thePesquisa Nacional por Amostra de Domicilios(PNAD), included in 1996 a set ofsupplemental questions on the parents ofrespondents.This permitted an analysis ofthe intergenerational persistence ininequalities. Using four circumstance vari-ables (parental schooling, father’s occupa-tion, race, and region of birth), Bourguignon,Ferreira, and Menendez (2005) investigatedhow inequalities of opportunity generateinequality in current earnings across differ-ent cohorts of adult individuals. Applying aconceptual framework closely related tothat in chapter 1, they decomposed earn-ings inequality into a lower bound compo-nent attributable to the inequality of oppor-tunity—to the effect of the four observedcircumstance variables—and a residualcomponent, which would account for per-sonal effort, luck, measurement error, transi-tory income, and other unobservable char-acteristics.They found that the fourvariables accounted for more than a fifth ofthe total earnings inequality within gender

cohorts. Of the four, family background wasmost important.

This distribution of certain opportuni-ties and outcomes has persisted across gen-erations.When the authors estimatedeconometrically the relationship betweenschooling and race, region of origin,parental education, and father’s occupation,only the coefficient on parental educationseems to have fallen across cohorts. In otherwords, race, region of origin, and father’soccupation continue to predict an individ-ual’s education level. And even for educa-tion, mechanisms are at work to reproduceschooling levels across generations, espe-cially at the lower end of the distribution.

Brazil underscores the need to look at arange of outcomes (of which incomes areonly one, with education, health and servicesalso of great concern). It also underscores theneed to look at a range of processes—ofwhich income and economic wealth-basedmechanisms form only part, and for whichgroup-based interactions are as central ashousehold and individual conditions, behav-iors, and characteristics.

Source: Bourguignon, Ferreira, and Menendez(2005).* The perception of particularly high inequalityin Brazil may to some extent be a result of theway income is measured there. Alternativeapproaches to measuring inequality, based onother welfare indicators, indicate that Brazilmay be less of an outlier in Latin America thanpreviously believed. See box 2.5 and also DeFerranti and others (2004).

B O X 2 . 1 Unequal opportunities persist across generationsin Brazil

30

Infant mortality rate per 1,000 live births160

140

120

100

80

60

40

20

0

No education

Secondary or higher

Colo

mbi

aJo

rdan

Sri L

anka

Viet

nam

Para

guay

Philip

pine

sTh

aila

ndBo

tsw

ana

Sou

th A

frica

Peru

Nica

ragu

aBr

azil

Turk

eyGu

atem

ala

Indo

nesia

Egyp

tTu

nisia

Zim

babw

eGh

ana

Mor

occo

Sene

gal

Keny

aNi

geria

El S

alva

dor

Turk

men

istan

Indi

aBo

livia

Eritr

eaSu

dan

Nepa

lBa

ngla

desh

Cam

eroo

nTo

goCo

mor

osHa

itiUg

anda

Yem

enCa

mbo

dia

Zam

bia

Pakis

tan

Beni

nM

adag

asca

r

Cent

ral A

frica

n Re

publ

icGu

inea

Burk

ina

Faso

Chad

Côte

d’Iv

oire

Mal

awi

Ethi

opia

Rwan

daM

ali

Nige

rM

ozam

biqu

e

Figure 2.1 Infant mortality varies across countries but also by mother’s education within countries

Source: Authors’ calculations from Demographic Health Survey (DHS) data.Note: The continuous dark line represents the mean infant mortality rate in each country, while the endpoints of the whiskers indicate the infant mortality rates by different levels of mother’s education.

Rural

Urban

35

30

25

20

15

10

5

0

Percentage of children severely stunted (z-score < 3)

Trin

idad

and

Tob

ago

(*)

Jord

anAr

men

iaBr

azil

Kaza

khst

an (*

)Co

lom

bia

Dom

inic

an R

epub

licPa

ragu

ayTu

nisia

Turk

eyEg

ypt

Gabo

nTu

rkm

enist

anHa

itiSr

i Lan

kaNa

mib

iaM

oroc

coNi

cara

gua

Ghan

aBo

livia

Zim

babw

eSe

nega

lGu

inea

Beni

nUg

anda

Côte

d’Iv

oire

Mau

ritan

iaTa

nzan

iaBu

rkin

a Fa

soBa

ngla

desh

Rwan

daM

ali

Cam

bodi

aCh

adGu

atem

ala

Nepa

lNi

geria

Zam

bia

Mal

awi

Ethi

opia

Yem

enPa

kista

n

Figure 2.2 Stunting levels of children born in rural versus urban areas are far from the same

Source: Authors’ calculations from Demographic Health Survey (DHS) data.Note: The continuous dark line represents the percentage of severely stunted children in each country, while the endpoints of the whiskers indicate the percentages for urban and rural areas.* Indicates stunting level in urban areas are higher than in rural areas.

The difference between children born inrural and urban areas can be dramatic,particularly at higher overall stunting lev-els. In Guatemala, stunting rates for chil-dren in urban areas are around 10 percent,but in rural areas they are as much as threetimes higher. Children in Guatemala clearlyhave no choice in deciding whether theyare born in the countryside or the city, buttheir opportunities to achieve good healthare clearly much less assured in rural thanin urban areas. As for infant mortalityrates, stunting among children is alsosharply differentiated by mother’s educa-tion and household economic status.

Access to immunization. Children born infamilies whose asset ownership places themin the top quintile of the distribution ofeconomic status have a high probability of

access to health services, proxied here ashaving received at least one of three keychildhood vaccinations—bacille Calmette-Guérin; diptheria, pertussis, and tetanus; ormeasles (figure 2.3). This is so even incountries where the overall percentage ofchildren without any coverage is as high as40 percent. Conversely, children whose par-ents are in the bottom quintile are muchmore likely to lack access to such basichealth care. In Morocco, where roughly 5percent of children have not received evenone of these three vaccinations, the propor-tion for children in the poorest quintile iswell above 15 percent.

High-impact health services. The WorldBank (2003j), drawing on DHS data from30 low- and middle-income countries, findsthat the poor are considerably less likely


Percentage not covered

70

60

50

40

30

20

10

0

Poorest

Wealthiest

Egyp

tJo

rdan

(*)

Colo

mbi

aRw

anda

Peru

Sout

h Af

rica

Ken

yaM

alaw

iBr

azil

Zam

bia

(*)

Viet

nam

Turk

eyGu

atem

ala

Tanz

ania

Indo

nesia

Turk

men

istan

(*)

Mor

occo

Ghan

aBe

nin

Philip

pine

sBa

ngla

desh

Com

oros

Boliv

iaPa

ragu

ayKa

zakh

stan

(*)

Yem

enBu

rkin

a Fa

soCa

mer

oon

Ugan

daIn

dia

Mau

ritan

iaHa

itiTo

goEt

hiop

ia

Cent

ral A

frica

n Re

publ

icM

adag

asca

rM

ozam

biqu

eGu

inea

Mal

iCa

mbo

dia

Pakis

tan

Eritr

eaNi

ger

Chad

Figure 2.3 Access to childhood immunization services depends on parents’ economic status

Source: Authors’ calculations from Demographic Health Survey (DHS) data.Note: The continuous dark line represents the percentage of children without access to a basic immunization package in each country, while the endpoints of the whiskers indicate thepercentages for the top and the bottom quintile of the asset ownership distribution.* Indicates that the poorest quintile have higher access to childhood immunization services than the wealthiest quintile.

than the non-poor to have access to high-impact health services, such as skilled deliv-ery care, antenatal care, and complemen-tary feeding. Similarly, Wodon (2005) drawson household survey data from 15 Africancountries to indicate that, while virtually allurban households are within one hour’stravel time to a health center, the propor-tion in rural areas is generally only aroundhalf, and as low as 35–38 percent in Nigerand Ethiopia.

Disability. Data from a number of coun-tries suggest that disabled people aremuch more likely to be poor. Hoogeveen(2003) reports that in Uganda the proba-bility of poverty for urban dwellers livingin a household with a disabled head is 38percent higher than for those who live in ahousehold with an able-bodied head. TheSerbian Poverty Reduction Strategyreports that 70 percent of disabled peopleare unemployed. In a study drawing on 10 household surveys in eight countries,self-reported disability was found to bemore correlated with nonattendance atschool than other characteristics, includ-ing gender or rural residence.1 Sen (2004)emphasizes that the disabled face not onlyan “earnings handicap,” associated with alower probability of employment andlower compensation for their work, butalso a “conversion handicap.” By this hemeans that a physically disabled personrequires more income than an able-bodied person to achieve the same livingstandard.

Social inequalities damaging health. Notonly are health outcomes correlated withinequalities in other dimensions, but suchsocial inequalities can be argued to bedetrimental to individual health out-comes.2 In his comprehensive review ofthe literature, Deaton (2003) argues that,while it is certainly plausible that variousinequalities (such as those in power) causebad health, it is not clear that inequality ofincome is the main culprit. He providesevidence suggesting that, after controllingfor an individual’s income, income in-equality at the group level does not matterindependently for individual health. Thus,

the main inequalities that affect healthmay not be in the income space. He citesexamples of other key dimensions ofinequality: land ownership, women’sagency (health and fertility in India), anddemocratic rights (in England in the1870s and in the U.S. South in the 1960s).In general, an individual’s rank in the rele-vant hierarchy has been found to be im-portant to health in animal and humanexperiments. Repeated stress associatedwith insults and the lack of control thatcomes from low rank has a well-developedbiochemical basis.3

The consequences of poor health arereflected in education achievements, eco-nomic prosperity, and future generations.Consider the plight of AIDS orphans insouthern Africa, the stark inequalities ofopportunity they face, and the possible rolefor public action (box 2.2).

DHS data (figures 2.1–2.3) providedetailed insights into the relationshipbetween inequalities in health and somekey circumstance variables. But they arenot particularly well suited to capturingthe contribution of detailed spatial factors,such as place of birth, in overall inequality,because of the limited sample size. In oneattempt to get around this problem, childheight in Cambodia was estimated at thecommune level based on a statistical pro-cedure to combine DHS data with popula-tion census data.4 The study documentsconsiderable heterogeneity across Cambo-dia’s more than 1,600 communes in theprevalence of stunting and being under-weight among children under the age offive (figure 2.4). The analysis providesclear evidence that in Cambodia a child’sopportunities for good health have astrong spatial dimension to them. Yetclearly, no child is able to determine inwhich locality he or she is born.

TrendsAverage health in most countries improvedin the twentieth century (chapter 3).Deaton (2004) documents that improve-ments in health are likely to have accom-panied economic growth, but he alsoemphasizes the globalization of knowl-edge, facilitated by local political, eco-


nomic, and educational conditions. In the1980s and 1990s, however, progressslowed—a result of the worldwide HIV/AIDS epidemic and rises in cardiovascularmortality in Eastern Europe and formerSoviet Union countries.

How have inequalities in health evolvedwithin countries? Data from DHS providesome clues. For a subset of countries, mul-tiple rounds of DHS data are available todocument changes in infant mortality overtime. Of some 36 “spells” of health changethat could be identified, roughly 25 corre-sponded to improved health outcomes inthe form of lower infant mortality rates.Although overall health improved in these25 cases, the gaps between urban and ruralareas, between groups defined by mother’seducation, and between groups defined bydurable asset ownership did not univer-sally decline alongside the overall declines


It is hard to imagine people with fewer assets,through absolutely no fault of their own,thanAIDS orphans.Left to fend for themselves on thedeath of one or both parents from a progressivelydebilitating,heavily stigmatized,and costly-to-treat disease, their plight would be of concerneven if they numbered but a few. In southernAfrica,however, the United Nations Children’s Fund(UNICEF) estimates that there were 12.3 millionAIDS orphans in 2003,a veritable demographicgroup in their own right.By 2010,UNICEF projectsthat there will be 1.5 million AIDS orphans inSouth Africa; by 2014,1 million in Zambia.

An entire generation of Africans is emergingwho will have been raised, if they are lucky, bygrandparents or extended family members(themselves likely to be impoverished,overwhelmed, and suffering from the disease).At worst, they will grow up in child-headedhouseholds or in situations in which their basicrights to food, clothing, shelter, and adequatecare are routinely denied.

Wills and schoolingBeginning to overcome the huge disadvantagesthat AIDS orphans start life with requires specialattention on numerous fronts (box 7.11 consid-ers a variety of policy options). From a legalstandpoint, parents who know their death isimminent and who have young children need tobe encouraged (even if they are illiterate) to pre-pare enforceable wills that will protect the inher-itance rights of their children to ensure that sur-viving adults do not just forcibly take their land,

savings, or other valuables. From an educationstandpoint, it is vital to keep children in school,where the acquisition of even basic skills cangive them some viable prospect of being able tomove out of poverty.Where a child is the head ofa household and perhaps its sole income earner,however, the pressures to drop out of school areenormous. Numerous studies document signifi-cantly higher dropout rates of AIDS orphans. InKenya, one extreme example found that “52 per-cent of the children orphaned by AIDS were notin school, compared to 2 percent of non-orphans”(UNAIDS 2002, 135).

Attending school is also important from acivic perspective: it socializes children into thenorms and mores of society, and gives them theconfidence and capacity to participate morefully in it.Without such socialization, vulnerableyoung children are easy targets for those offer-ing them security and status through member-ship in a street gang, criminal network, or militiamovement. If AIDS orphans continue to stayfrom school at their current rate, comments onesenior U.N. official,“you will have a society wherekids haven’t been to school and therefore can’tfulfill even basic jobs...a society where a largeproportion can have antisocial instincts becausetheir lives have been so hard.You [will] have ageneration of children who will be more vulnera-ble to exploitation and to disease because theywon’t have the same sense of self-worth”(citedin Fleshman 2001, 1). Such children face the dis-mal prospect of failing to accumulate assetsbecause of the extreme burdens thrust on them

in their most formative years and the paucity ofopportunities available to them thereafter.

Avoiding infectionThe most immediate priority,however, is ensuringthat AIDS orphans do not themselves becomeinfected with the disease,thereby increasing thelikelihood that they will perpetuate the cycle.AIDSorphans face precisely such a risk,however,because the stigma of HIV/AIDS means that peopleoften assume that the children of parents who diedfrom AIDS must be infected,shunning,shaming,orexploiting them accordingly.Some AIDS orphanshave even been denied access to schools andhealth clinics because of the fear their verypresence generates.Children grieving the loss of aparent are also vulnerable to the sexual predationsof those putatively claiming to offer them comfort.Indeed,the desperation and apparent hopeless-ness of their circumstances—all the more so if itcoincides with a natural disaster such as drought—can drive AIDS orphans into prostitution.

The plight of AIDS orphans provides agraphic illustration of how cycles of disadvantagecan perpetuate themselves, and how social isola-tion and exclusion (especially at a young age) canpreclude the acquisition of assets and underminethe capacity to sustain participation in the insti-tutions that provide the best path out of poverty.

Sources: Avert.org (2004) http://www.avert.org/aidsorphans.htm. Accessed December 14, 2004.Fleshman (2001). Hargreaves and Glynn (2002),Lewis (2003), UNAIDS (2002), UNICEF (2003), USAID,UNAIDS, and UNICEF (2004).

B O X 2 . 2 Unequal assets, unequal opportunities: AIDS orphans in Southern Africa

Stunting and underweight(number of communes)

Stunting low/underweight low (260)Stunting low/underweight high (189)Stunting high/underweight low (365)Stunting high/underweight high (780)Incomplete data

2000 100 200

kilometers

Figure 2.4 Stunting and underweight in Cambodia

Source: Fujii (2005).

in infant mortality.5 The improvements inhealth were not necessarily shared across allgroups in the population.

As Cornia and Menchini (2005) note,mortality differentials across groups tendto narrow with an improvement of theaverage only if policies focus explicitly onequity. Without such a focus, improve-ments in the average may not translate todeclining group differences. For example,in the United States between the 1950s and

1990s, the overall decline in the infantmortality rate to 7.9 in 1994 was accompa-nied by an increase in the ratio of black towhite infant mortality rates from 1.6 in1950 to 2.2 in 1991. Inequality in healthdoes not inevitably fall as overall healthimproves, but such a virtuous process ispossible (box 2.3).

Inequalities in educationEducation is of great intrinsic importancewhen assessing inequalities of opportunity.It is also an important determinant of indi-viduals’ income, health (and that of theirchildren), and capacity to interact and com-municate with others. Inequalities in educa-tion thus contribute to inequalities in otherimportant dimensions of well-being.

Measuring inequality in education isnot easy. Census and survey data in mostcountries can generally yield statistics on,for example, years of schooling. But suchinformation does not capture well thequality of education and how that mightvary across individuals. Nor is it easy tocompare years of schooling across coun-tries, because those years might meansomething quite different from country tocountry.

Test results. Despite the measurement diffi-culties, there is considerable evidence ofinequalities of opportunity in education inthe developing world. Consider the differ-ences in test performance among Ecuado-rian children ages three to six years acrosspopulation groups defined by parentaleducation, region of residence, and wealth(box 2.4).

Test results among very young childrencapture well the inequality in opportunityin education, but such data are not readilyavailable for large numbers of developingcountries. So we look instead at the percent-age of household heads with no educationby gender and by urban-rural residence.

Male and female household heads. Theoverall percentage of household headswithout any education varies dramaticallyacross our sample of 60-odd countries(figure 2.5). In the high-income countries,the percentage rates are negligible. But at


Paxson and Schady (2004), drawing on mul-tiple rounds of DHS data, document thedeclining infant mortality rate in Perubetween the late 1970s and late 1990s. Ageneral downward trend exhibited a sharpsetback during the major economic crisisbetween 1988 and 1992, but resumed afterthe crisis.The downward trend remainedevident even after adjusting for age ofmother, recall period, education, and urbanstatus—indicating that the overall trenddecline in infant mortality was not attributa-ble only to general improvements in educa-tion, an aging population, or urbanization. Inaddition, the fact that infant mortality rosesharply around 1990, even after theseadjustments, supports the notion that thedecrease in household income and the col-lapse of public expenditures on health as aresult of the crisis were important.

Infant mortality rates in Peru variedmarkedly with the education level of themother in late 1970s and the 1980s (see fig-ure below). During the economic crisis,increases in mortality were largest amonginfants born to women with less education.After the crisis, the gap between infant mor-tality rates associated with different mater-nal education levels declined steadily, sug-gesting an overall decline in inequality inmortality alongside the decline in overallmortality rates.

There is some support for the view thatchanges in the amount and composition ofpublic expenditures on social programsdrove these improvements. Real totalexpenditures increased two and a half timesbetween 1991 and 2000, and such publicspending did not bypass the poor.

B O X 2 . 3 Health improvements and greater health equityin Peru

0.15

0.12

0.09

0.06

0.03

0.00

90th percentile or 11 years

75th percentile or 10 years50th percentile or 5 years25th percentile or 2 years10th percentile or 0 yearYears of maternal education

199819941990198619821978

Probability of dying in the first year

Adjusted infant mortality rates by maternal education

Source: Paxson and Schady (2004).

the other extreme, in Burkina Faso andMali, for example, the overall percentage ismore than 80 percent. What is similarlystriking is that, in most countries, the likeli-hood that the household head is uneducatedis dramatically higher than average whenshe is a woman. In the Laos People’s Demo-cratic Republic, for example, although theoverall percentage of household heads withno education is about 20 percent, the rate iscloser to 70 percent for female householdheads.

Rural and urban household heads. Simi-lar patterns can be observed for rural andurban areas (figure 2.6). In general, house-hold heads are far more likely to have noeducation when they are based in ruralareas than in urban areas. Even in coun-tries where the overall percentage withouteducation is very high, the rate in urbanareas can be dramatically lower. For exam-ple, in Burundi, the percentage of house-hold heads with no education in urbanareas compares with the national averagein Mexico, the Dominican Republic, andBrazil.

Access to teachers. A recent study of primaryschools and health clinics in Bangladesh,Ecuador, India, Indonesia, Peru, and Ugandahas identified teacher absenteeism as animportant, common, problem. The studyfound that higher income areas generallyhave lower teacher absentee rates thanpoorer areas.6 It also found that higher paidteachers, generally more educated and expe-rienced, appear equally or more likely to beabsent than contract or less remuneratedinstructors, perhaps because these instruc-tors sense a lower risk of being fired for theirabsence. And although salaries in rural areaswere often higher than in urban areas,teacher attendance in these areas was typi-cally lower than in urban areas. In most sur-veyed countries, the quality of infrastructureand the frequency of monitoring appeared tocontribute to lower absenteeism.

TrendsAnother way to assess inequalities ofopportunity in education is to calculate anoverall index of inequality for years of edu-

cation and to assess how much overallinequality of education can be attributed tomean differences between “morally irrele-vant” groups. Araujo, Ferreira, and Schady(2004) find that the inequality of adulteducation, measured by years of schoolingfor 124 countries, can be pronounced. Theyalso find that it is strongly (and inversely)correlated with mean years of schoolingacross countries.7

The data assembled by these authors alsoindicate that the inequality of education forspecific subgroups of the population canchange. While female schooling achieve-ments relative to male achievements weredramatically lower among the oldest cohorts,particularly in Sub-Saharan Africa, South


That education achievements varymarkedly by population groups—and thatthis can have profound implications—isbrought out forcefully in a recent study byPaxson and Schady (2005).They show thatcognitive development of Ecuadorian chil-dren ages three to six years, as measured bya test of vocabulary recognition (TVIP),varies significantly depending on thewealth of their household, their place of res-idence, the education of their mother, andthat of their father.The extent to whichthese circumstance variables are associatedwith performance on cognitive tests is typi-cally more pronounced for the olderchildren in their sample.

These socioeconomic characteristicsare significantly associated with cognitivedevelopment even after controlling forchild health and home environment. Theresearchers point to the striking evidencethat, in Ecuador, the youngest children,irrespective of wealth quintile or educa-tion of their parents, perform broadly aswell as their comparators. But as childrenin Ecuador get older, their cognitive devel-opment, relative to this benchmark, falterssignificantly. Only children in the top halfof the wealth distribution and with highlyeducated parents maintain their perform-ance relative to their comparators. By thetime they are six years old, most childrenin the sample are so far behind in theircognitive development that it is uncertainwhether and how they could ever catchup.

B O X 2 . 4 Child test scores in Ecuador: the role of wealth,parental education, and place of residence

40 60 7050

100

90

80

70

60

110

Wealthiest 25%

Median score

Median score

Poorest 25th%

40 60 7050

Age in months

Age in months

Wealthiest and poorest quartiles

Maternal education

100

90

80

70

60

110

12 or more years

0–5 years

Source: Paxson and Schady (2005).

Asia, and to a lesser extent the Middle Eastand North Africa, these disparities arenoticeably lower for the younger cohorts,particularly in Sub-Saharan Africa (figure2.7). Additionally, disparities in years ofschooling between urban and rural areashave been falling in some regions, moststrikingly in the Middle East and NorthAfrica and in Eastern Europe and CentralAsia. But in Sub-Saharan Africa there hasbeen little, if any, change. The (urban-rural)between-group contribution to inequalityin this region has hovered at around 30 per-cent across all the cohorts examined.

Economic inequalitiesAn individual’s consumption, his or herincome, or his or her wealth have all beenused as indicators of the command of an

individual over goods and services that canbe purchased in the market and that con-tribute directly to well-being. It is clear too,that individuals’ economic status can deter-mine and shape in many ways the opportu-nities they face to improve their situations.Economic well-being can also contribute toimproved education outcomes and betterhealth care. In turn, good health and goodeducation are typically important determi-nants of economic status.

An ideal measure of economic well-being for assessing inequality will capturean individual’s long-term economic status.But it is difficult to produce such a compre-hensive indicator accurately. In practice, it iscommon to work with measures of currentincome or consumption compiled fromhousehold survey data. While consumption


Percentage of household heads with no education

100

80

60

40

20

0

Unite

d Ki

ngdo

mCa

nada

Norw

ayUn

ited

Stat

es

Finla

ndGe

rman

ySw

eden

Kyrg

yzst

anKa

zakh

stan

Jam

aica

(*)

Tajik

istan

Rom

ania

Isra

elCh

ileM

oldo

vaPe

ruPa

ragu

aySr

i Lan

kaIta

lyTh

aila

ndGe

orgi

a (*

)Vi

etna

mAl

bani

a

Bosn

ia a

nd H

erze

govin

aCo

sta

Rica

Turk

eyIn

done

siaEc

uado

rPa

nam

aBo

livia

Colo

mbi

a

Repú

blic

a Bo

livar

iana

de

Vene

zuel

aEl

Sal

vado

r

Mex

ico

Dom

inic

an R

epub

licBr

azil

Lao

PDR

Pola

ndM

adag

asca

rTa

nzan

iaJo

rdan

Keny

aHo

ndur

asCa

mbo

dia

Nica

ragu

aCa

mer

oon

(*)

Guat

emal

aPa

kista

nYe

men

Côte

d’Iv

oire

Ghan

aHa

itiBa

ngla

desh

Tim

orBe

nin

Buru

ndi

Mor

occo

Guin

eaEt

hiop

iaM

aurit

ania

Mal

i

Burk

ina

Faso

(*)

Female

Male

Figure 2.5 Education levels vary across countries, but they also depend on gender of household head

Source: Authors’ calculations from household survey data.Note: The continuous dark line represents the percentage of household heads with no education in each country, while the endpoints of the whiskers indicate the percentages for male andfemale-headed households. * Indicates that female-headed households have higher average levels of education than male-headed households.


Percentage of household heads with no education

100

80

60

40

20

0

Cana

daUn

ited

Stat

es(*

)

Finla

ndSw

eden

Kaza

khst

anJa

mai

caTa

jikist

anRo

man

iaIs

rael

(*)

Chile

Mol

dova

Peru

Para

guay

Sri L

anka

(*)

Italy

Thai

land

Geor

gia

Viet

nam

Alba

nia

Bosn

ia a

nd H

erze

govin

aCo

sta

Rica

Turk

ey(*

)In

done

siaEc

uado

rPa

nam

aBo

livia

Colo

mbi

a

Repú

blic

a Bo

livar

iana

de

Vene

zuel

aEl

Sal

vado

r

Mex

ico

Dom

inic

an R

epub

licBr

azil

Lao

PDR

Pola

ndM

adag

asca

rTa

nzan

iaJo

rdan

Keny

aHo

ndur

asCa

mbo

dia

Nica

ragu

aCa

mer

oon

Guat

emal

aPa

kista

nYe

men

Côte

d’Iv

oire

Ghan

aHa

itiBa

ngla

desh

Tim

orBe

nin

Buru

ndi

Mor

occo

Guin

eaEt

hiop

iaM

aurit

ania

Mal

iBu

rkin

a Fa

so

Rural

Urban

Germ

any

Unite

d Ki

ngdo

m(*

)No

rway

(*)

Kyrg

yzst

an

Figure 2.6 Education levels vary by country and between rural and urban sectors

Source: Authors’ calculations from household survey data.Note: The continuous dark line represents the percentage of household heads with no education in each country, while the endpoints of the whiskers indicate the percentages for urban andrural households. * Indicates that rural households have higher average levels of education than urban households.

0.4

0.3

0.2

0.1

0.0

Between-group contribution to total inequality(proportion)

Other

East Asia

Europe andCentral Asia

South Asia

Latin America and Caribbean

Middle East andNorth Africa

Sub-Saharan Africa

1975–91970–41965–91960–41955–9Year of birth

1950–41945–91940–41935–9

Figure 2.7 The share of inequality in years of schooling attributable to differences betweenmales and females has been declining

Source: Araujo, Ferreira, and Schady (2004).

and income inequality are expected to cor-relate reasonably well with long-term well-being, it is unclear exactly how well theyactually do. And different measures of eco-nomic welfare—based on income, con-sumption, or wealth—can yield quite dif-ferent assessments of inequality (see alsobox 2.5).

For example, Sudjana and Mishra(2004), drawing on evidence produced byClaessens, Djankov, and Lang (2000), arguethat wealth inequality in Indonesia is farmore concentrated than suggested by com-parable figures based on consumption (fig-ure 2.8). In 1996 more than 57 percent ofthe stock market capitalization in Indonesiawas controlled by 10 families. This is instark contrast to neighboring countries,such as Singapore and Malaysia, but it is

only marginally higher than the figure forthe Philippines. More generally, Davies andShorrocks (2005) report estimates pub-lished by Merrill Lynch and Forbes thatsome 20 percent of the world’s millionairescome from the developing world. Similarly,Morck, Stangeland, and Yeung (2000) finda higher ratio of billionaire wealth to grossdomestic product (GDP) in Latin Americaand the Caribbean, and East Asia, but notIndia and South Africa (see chapters 6 and9 for further discussion). These figuresimply that the distribution of wealth may,on average, be more concentrated in devel-oping countries than in the developed.When wealth is associated with politicalinfluence, such inequalities also translateinto political capture and can provide awindow on this added dimension ofopportunity.

Bearing in mind the warnings offered inbox 2.5, figure 2.9 provides an approximate

picture of how economic inequality is dis-tributed across countries. The highest levelsof recorded inequality occur in Africa, thesecond highest in Latin America. But in-equality measures for Latin America comelargely from income data, while those inother regions, such as South Asia, comemainly from consumption data. As box 2.5illustrated, income data tend to producehigher measured inequality. Within regions,the data suggest that inequality can varymarkedly between countries: consumptioninequality in South Africa is extremely high,while in Mauritius it is lower even than inOECD countries.

How much overall economic inequalitywithin countries is attributable to differ-ences across population groups? Unlikehealth and education inequalities, the sys-tematic decomposition of income inequal-ity by population groups has long been sub-ject to analysis in the economics literature.


Because countries differ in their data collectionsystems, cross-country data on economicinequality are generally based on a variety ofindicators that are treated interchangeably.Thelack of a uniform basis for measuring economicinequality in different countries has seriousimplications for comparability.

One of the main sources of noncomparabil-ity of inequality is that some countries usehousehold income as indicator of well-beingwhile others use consumption expenditures(Atkinson and Brandolini 2001).These two indi-cators capture different aspects of economicwelfare, with the former perhaps seen better asa measure of welfare opportunity and the latteras a measure of welfare achievement. In mostcountries, measured inequality based onincome is higher than if it is based on consump-tion. But this is not inevitable, and the degree towhich the two indicators disagree varies fromcountry to country (see table to the right).

The problem of comparability is notconfined to the choice of welfare indicator. Animportant but underappreciated additionalissue is that, even for a given indicator, its defini-tion varies considerably across countries andeven within countries over time. Consumptioninequality based on different definitions of con-sumption can vary markedly, and will dependon a variety of factors, including the following:

• The length of the recall period over whichconsumption is recorded.

• The degree of disaggregation of consump-tion items.

• The methods for imputation of housing anddurables consumption.

Similarly, income inequality can vary dependingon whether income—

• Is intended to capture pre- or post-tax income,

• Includes actual and implicit transfers, and

• Refers to full income or earnings only.

Additional factors confounding comparabil-ity include differences in survey nonresponserates across countries (which are likely to affectmeasured inequality—see Korinek, Mistiaen,and Ravallion forthcoming). Differences acrosscountries in the availability of spatial priceindexes can also affect conclusions.Thomas(1987) demonstrates that adjusting for spatialprice variation can affect conclusions about thedegree of income or consumption inequality.Across countries there tends to be little unifor-mity in whether, and how, spatial price variationis accommodated.

Cross-country datasets on economicinequality generally incorporate some attemptsto improve comparability, but they typically fallfar short of achieving strict comparability.With-out a concerted effort to harmonize data collec-tion across countries, it is unlikely that suchglobal databases can be relied on to providemore than a tentative picture of differences ininequality across countries.

Inequality: summary measures in a selectionof countries: consumption versus income

Gini coefficient

Year Consumption Income

Panama 1997 0.468 0.621

Brazil 1996 0.497 0.596

Thailand 2000 0.428 0.523

Nicaragua 1998 0.417 0.534

Peru 1994 0.446 0.523

Morocco 1998 0.390 0.586

Vietnam 1998 0.362 0.489

Nepal 1996 0.366 0.513

Albania 1996 0.252 0.392

Bulgaria 1995 0.274 0.392

Russian Federation 1997 0.474 0.478

Bangladesh 2000 0.334 0.392

Source: Authors’ creation.

B O X 2 . 5 Beware of intercountry comparisons of inequality!

IndonesiaPhilippines

ThailandHong Kong, ChinaRepublic of Korea

SingaporeMalaysia

Taiwan, ChinaJapan

57.7%52.5%

46.2%32.1%

26.8%26.6%

24.8%18.4%

2.4%

Figure 2.8 Market capitalizationcontrolled by the top 10 families inselected countries, 1996

Source: Claessens, Djankov, and Lang (2000).

39

0.0 0.2 0.4 0.6 0.8

TurkmenistanGeorgiaEstoniaTurkey

FYR MacedoniaLatvia

MoldovaLithuaniaTajikistan

RussiaCroatiaAlbaniaPoland

KazakhstanKyrgyzstan

SloveniaBelarus

BulgariaRomania

SerbiaUzbekistan

ArmeniaUkraine

SlovakiaCzech Republic

Bosnia and HerzegovinaHungary

Azerbaijan

MalaysiaPapua New Guinea

SingaporePhilippinesCambodia

Thailand

China

Lao PDRMongoliaIndonesia

Rep. of KoreaVietnam

Japan

NamibiaBotswana

Central African RepublicSwaziland

LesothoSouth Africa

ZambiaMalawi

Gambia, TheZimbabwe

MadagascarCôte d’Ivoire

KenyaUganda

CameroonBurundiNigeria

Burkina FasoAngola

SenegalMozambique

MaliGhana

GuineaMauritania

BeninTanzania

NigerEthiopia

Mauritius Based on consumption

Sub-Saharan Africa

High-Income Economies

Latin America and the Caribbean

Middle East and North Africa

South Asia

East Asia and Pacific

Europe and Central Asia

0.0 0.2 0.4 0.6 0.8

Sri LankaNepalIndia

BangladeshPakistan

IranTunisia

MoroccoJordanAlgeria

IsraelEgypt

Yemen

BrazilBolivia

ParaguayColombiaEcuador

ChileArgentinaHonduras

El SalvadorMexico

Dominican RepublicPanama

GuatemalaCosta Rica

GuyanaSt. LuciaUruguay

R.B. de VenezuelaJamaica

NicaraguaTrinidad and Tobago

Peru

Hong Kong, ChinaPortugal

United StatesNew Zealand

GreeceUnited Kingdom

SpainCanada

AustraliaIreland

ItalySwitzerland

FranceLuxembourgNetherlands

AustriaGermanyNorway

DenmarkBelgiumSwedenFinland

Taiwan, China

Based on incomeBased on consumptionBased on income

Figure 2.9 Africa and Latin America have the world’s highest levels of inequalityIncome and expenditure Gini coefficients

Source: Authors’ calculations from household survey data.

These decomposition exercises seek tounderstand what share of inequality can beattributed to differences between groupsand what to inequality within groups. Thereare several attractions to studying certainpopulation groups in this way and to com-paring findings across countries.

Our interest here is to define groups bycircumstances we might consider “morallyirrelevant,” thereby gaining a window onthe importance of inequality of opportu-nity in the economic sphere. Additionally,decomposition results generally are far lesssensitive to differences in definitions ofunderlying welfare indicators than are

measured levels of inequality. In that sense,some of the difficulties with cross-countrycomparisons described in box 2.5 are atten-uated by subgroup decompositions.

Between-group shares of total inequalityWhile the “between-group” share of overallinequality is an appealing indicator of thesalience of differences across groups in theoverall assessment of inequality, there areconcerns about its interpretation.8 In par-ticular, empirical measures of between-group shares are generally found to bequite low (see figures 2.10 and 2.11).9 Theconventional presentation of between-group inequality is relative to total inequal-ity. Elbers and others (2005), however, notethat total inequality can be viewed as thebetween-group inequality that would beobserved if every household in the popula-tion constituted a separate group. Clearly,against such a benchmark, one wouldrarely observe a high share of between-group inequality.

Elbers and his colleagues propose analternative, comparing the actual between-group inequality with the maximum possi-ble inequality that would be obtained bykeeping the number of groups and their sizesat actual levels. For example, an assessmentof the contribution of gender differences toinequality compares actual between-genderinequality with the hypothetical between-gender inequality that would be obtained bysorting the income distribution so that allmales appeared at one end of the distribu-tion and all females at the other. This ratioprovides a measure of how far actualbetween-group inequality lies below themaximum between-group inequality that isfeasible given the existing configuration ofgroups.

Economic inequality can be decom-posed in a large sample of countries basedon several population breakdowns, two ofwhich are presented in figures 2.10 and2.11: social group and education of house-hold head. Such decompositions can followthe conventional decomposition method-ology, complemented by the Elbers andothers (2005) measure of feasible groupdecomposition.


0.10 0.2 0.3Proportion

0.4 0.5 0.6

ConventionalFeasible

NepalSri Lanka

Bangladesh

IsraelJordan

ParaguayGuatemala

BoliviaPanama

PeruBrazil

GuyanaNicaragua

St. Lucia

United StatesGermany

FranceLuxembourg

United KingdomCanadaBelgium

SwitzerlandAustralia

IrelandNorwaySwedenAustria

Finland

KyrgyzstanRomania

Vietnam

South Africa Madagascar

BeninCôte d’Ivoire

NigerGuinea

Sub-Saharan Africa

East Asia and Pacific


High-Income Economies



South Asia

Figure 2.10 Between-group inequality decompositions: social group of the household head


Different population breakdowns con-tribute to differing extents to overallinequality. In general, the conventional cal-culation of the between-group contribu-tion points to a fairly low share attributableto between-group differences. But in somecountries even the conventional share is

high. For example, in Paraguay, wheninequality is decomposed between groupsby language spoken at home, the conven-tional between-group share is approxi-mately 30 percent (figure 2.10). Andwhen inequality is decomposed for fivebroad education groups in Guatemala,


Proportion

Proportion

0.00 0.15 0.30 0.45

AzerbaijanTajikistan

ArmeniaUkraine

KazakhstanRussia

Bosnia & HerzegovinaMoldovaGeorgia

KyrgyzstanEstonia

MacedoniaAlbania

HungaryTurkeyPolandSerbia

RomaniaLithuania

VietnamLao PDR

PhilippinesEast Timor

Papua New GuineaIndonesia

Thailand

NigerEthiopia

MauritaniaNigeria

Burkina FasoTanzania

KenyaGuinea

MaliBenin

Côte d’IvoireUgandaBurundiSenegal

MadagascarCameroon

0.00 0.15 0.30 0.45

NepalPakistan

Sri LankaBangladesh

YemenIsrael

JordanMorocco

SurinameGuyana

St. LuciaTrinidad Tobago

R.B. de VenezuelaEl Salvador

Dominican RepublicUruguay

Costa RicaHonduras

BoliviaHaiti

JamaicaEcuador

ColombiaMexico

ParaguayChile

ArgentinaNicaragua

PeruPanama

BrazilGuatemala

BelgiumAustralia

AustriaCanadaNorwaySweden

SwitzerlandNetherlands

FinlandGermany

ItalyUnited Kingdom

IrelandUnited States

Taiwan, ChinaFrance

Luxembourg

Sub-Saharan Africa High-Income Economies



South Asia

East Asia and the Pacific




Figure 2.11 Between-group inequality decompositions: education of the household head


the between-group contribution is above40 percent (figure 2.11).

In most countries, the between-groupshare is noticeably higher for decomposi-tions based on the alternative, “feasible” cal-culation. Based on this approach, observedbetween-group differences are indeed sub-stantial in many countries—for the groupdefinitions here. To the extent that these cir-cumstances are judged “morally irrelevant,”the findings suggest that in economic life,just as in health and education, a substantialportion of observed inequality in manydeveloping countries can be linked toinequalities of opportunity.

Spatial differencesAs with inequalities in health, conventionalsurvey data cannot say much about the con-tribution of finely detailed spatial hetero-geneity to overall inequality—because of thelimited sample size. In an exercise analogousto that for health in Cambodia (figure 2.4), avariety of studies have applied statisticaltechniques to combine survey data withpopulation census data to produce tentativeestimates of inequality at the communityand district levels. Elbers and others (2004)document the contribution to overall esti-mated inequality of differences in meanconsumption for subdistricts in Ecuador,Madagascar, and Mozambique. Theydemonstrate that the between-subdistrictcontribution to total estimated inequalityranges from a low of 22 percent in Mozam-bique to more than 40 percent in Ecuador(table 2.1). Based on a similar approach,World Bank (2004e) reports between-com-mune differences in Morocco, accountingfor 40 percent of overall estimated con-sumption inequality. The general impres-sion is that spatial differences across locali-

ties account for a larger share of totalinequality as the number of localitiesincreases. The analysis confirms that forsome countries the spatial dimension ofinequality is of considerable importance.This conclusion carries over even morepowerfully at the global level, where thebetween-country contribution to globalinequality is dramatic (chapter 3).

Other studies and methodologies cor-roborate the finding that spatial differenceswithin countries are important. Using farm-household data for rural China, Jalan andRavallion (1997) identify “spatial povertytraps,” where poorer areas have lower provi-sions of essential public goods (such asroads) and, as a result, households in thearea experience lower productivity on theirinvestments. Various studies find spatialeffects on living standards, even after con-trolling for nongeographic household char-acteristics. Ravallion and Wodon (1999)demonstrate that place of residence is animportant determinant of poverty inBangladesh. They also note that importantspatial differences can be discerned evenwithin urban areas—households in the dis-trict of Dhaka are markedly better off thantheir counterparts in other urban districts.

Many studies suggest that spatial differ-ences in incomes are driven by policy. InChina, Kanbur and Zhang (2001) find ameasurable polarization between inland andcoastal regions where factors unrelated tophysical geography—development of heavyindustry in certain provinces, trade open-ness, and government investment in coastalregions—are associated with wideninginterregional inequality. Escobal and Torero(2003) compare coastal Peru with the high-lands and find that average per capita expen-ditures vary markedly and that this varianceis associated with fewer and weaker infra-structure services in the highlands.

The role of infrastructure is thus central.Although it is not disputed that physical geog-raphy can also influence poverty directly, theassociation between geographic variation inpoverty and geographic variation in infra-structure access is typically strong. Accord-ingly, it is argued that the influence ofregional geographic location on inequalitywill diminish as access to transport and


Table 2.1 Decomposition of inequality between and within communities

Level of Number of Within-group inequality Between-group inequality decomposition communities (percent) (percent)

Ecuador 1,579 58.8 41.2

Madagascar 1,248 74.6 25.4

Mozambique 424 78.0 22.0

Source: Elbers and others (2004).Note: Our communities in Ecuador are zonas in urban areas and parroquias in rural areas. Communities in Madagas-car are firiasana (communes) and in Mozambique they are administrative posts. The decompositions are performedusing the conventional methodology.

communications services improve; beinggeographically isolated will matter lessbecause infrastructure improvements willhelp compensate for distance.10

The relationship between groupdifferences and inequalityAs is clear from the discussion here, ourinterest in the contribution of group differ-ences to total inequality extends beyondnormative considerations of fairness andjustice. Differences between groups are alsothought to explain overall inequality out-comes, particularly the reproduction ofinequalities over time. The basic idea is thatbetween-group differences in incomeinequality, for example, will tend also to bemirrored in between-group differences inhealth and education inequalities—and inthe agency of groups in influencing theircircumstances (see below). These group dif-ferences will then reinforce one another.Group differences in education, for exam-ple, will translate into differences inincomes and in political voice and partici-pation. These inequalities will, in turn,affect health inequalities between groups,which are passed on to education inequali-ties and so on. “Inequality traps” are theresult. A corollary of this idea is that effortsto moderate overall inequality levels mightrequire a focus on reducing between-groupdifferences.

It is difficult to systematically documentthis instrumental role of group differences.Figure 2.12 illustrates one attempt. Overallinequality is correlated with the between-group share for the sample of countries infigures 2.10 and 2.11, controlling for regionand whether the underlying welfare indica-tor is income or consumption. Nothing inthe mechanics of the calculation forcesoverall inequality to be correlated with theshare attributable to between-group differ-ences. Yet, for this sample of countries,higher overall inequality is associated with alarger between-group share of overallinequality, which is attributable to therural-urban breakdown, to differencesacross social groups, to differences in edu-cation, and (weakly) to differences in broadoccupation class of the household head.11

One interpretation of these findings isthat between-group differences account for,and possibly explain, a non-negligible por-tion of overall inequality. This is consistentwith the broader theme of this report: thatgroup differences reinforce one anotherand in this way contribute to the replica-tion of inequality over time. But these sim-ple correlations, while suggestive, couldalso be pointing to other processes and ontheir own cannot exclude other competingexplanations.

Inequality and growth, economicstructure, and tradeSystematic exploration of the impact ofbetween-group shares on overall inequalityhas not, to date, been a major topic ofempirical investigation. A longer-standingquestion in economics has been howinequality evolves with economic growthmore generally. Pioneering work by Kuznetsin the 1950s launched an enormousamount of empirical work on this question,stimulating much debate. There is still no


Urban-rural

Overall inequality

Occupation of the household head

Between-group share

Overall inequality

Social group of the household head

Overall inequality

Education of the household head

Between-group share

Between-group share Between-group share

Overall inequality

Figure 2.12 Location, education, and social groups can make a difference: regressions of totalinequality on shares of between-group inequality of different household characteristics

Source: Authors’ calculations from household survey data.Note: Regressions include as controls (X) regional area dummies and a welfare measure (Y/C) dummy. The shares ofthe between component of inequality across gender and age of the household head, and regions within the countrywere not significant.

consensus on a systematic relationshipbetween the long-term growth processes ofindustrialization and urbanization—andoverall inequality (box 2.6).

Cross-country studies have also analyzedthe relationship between inequality and eco-nomic structure. Bourguignon and Morri-son (1990), for example, argue that “develop-ing countries which are comparativelyendowed with mineral resources and land(climate) tend to be less egalitarian than oth-ers, although the effect of the agriculturalcomparative advantage may be offset by thedistribution of land.” They also find that thelabor productivity difference between agri-culture and the rest of the economy is a pow-erful explanatory factor for differences inincome inequality in a number of developingcountries in the 1970s and 1980s.12

A large body of literature has alsoexplored the relationship between tradeopenness and inequality but has not reacheda consensus. For example, Dollar and Kraay(2002) and Dollar and Kraay (2004) find noeffect of trade openness on inequality, butLundberg and Squire (2003) do find suchan effect. Ravallion (2001) and Milanovic(2002) report that at low incomes opennessmay be inequality-increasing, but that thiseffect reverses at higher incomes.

TrendsThe discussion above highlights the manymechanisms for hypothesizing how aggre-gate economic growth, and the evolution ofdifferent sectors of the economy, can influ-ence economic inequality. Popular lines ofargument have emphasized Lewis-Kuznets


The starting point of the literature linking eco-nomic development and income inequalitydates to the well-known works of two NobelPrize winners, W. Arthur Lewis (1954) and SimonKuznets (1955). Lewis, in his classic 1954 article“Economic Development with Unlimited Sup-plies of Labor,” developed a theoretical model inwhich growth and accumulation in a dual econ-omy would start in the modern industrial sector,where capitalists would hire at a given wageand reinvest a share of their profits.The numberof traditional agricultural laborers willing tomove to this high-productivity, high-wage sec-tor was assumed to be unlimited. In this processof development, and as long as these assump-tions would prevail, inequality in thedistribution of income would increase as aver-age incomes rose.There would be a turningpoint after which inequality would fall again asthe surplus labor phase ends and the dualisticeconomy becomes a single-sector, fully industri-alized economy.

Although Kuznets did not explicitly modelthe intersectoral shifts of population as part ofthe development process, he did build on themto articulate his basic idea of an inverted-U rela-tionship between economic growth and incomeinequality (the “Kuznets curve”). In his presiden-tial address at the Annual Meeting of the Ameri-can Economic Association in 1954, he hypothe-sized that in the process of growth andindustrialization, inequality would first increase,because of the shift from agriculture and thecountryside to industry and the city, and thendecrease as returns across sectors equalized.Thedata Kuznets used to make this statement came

from a long-run series of inequality indicatorsfor England, Germany, and the United States,and from a single observation in time for threedeveloping countries—India, Ceylon (Sri Lankatoday), and Puerto Rico.These were the dataavailable at that time, and Kuznets was wellaware of the limitations of the empirical backingof his argument, in his own words, on “5 percentof empirical information and 95 percent specu-lation, some of it possibly tainted by wishfulthinking.”

Kuznets based his speculation primarily onlongitudinal data and called for in-depth casestudies of the economic growth of nations. Butmany subsequent studies simply usedaggregate cross-country data (often of not par-ticularly high quality) and reduced-form modelsto explore and support the hypothesis of aninevitable tradeoff between development andequality.The Kuznets curve became one of themost quoted stylized facts of the study ofincome distribution for nearly four decades.

Cross-country data can be misleading for dynamic processesWith the development of much larger data sets,such as the Deininger and Squire (1996) interna-tional inequality database (following on fromFields 1989), empirical “tests” of the Kuznetscurve were widely conducted. But it has becomeunderstood that the use of cross-country datato analyze what are essentially dynamicprocesses can be strongly misleading. Moreover,numerous studies have shown that theevidence in favor of the Kuznets curve is not atall robust to econometric specifications, sample

composition, and period of observation. See,among others, Bourguignon and Morrisson(1989), Fields and Jakubson (1994), Deiningerand Squire (1998), and Bruno, Ravallion, andSquire (1998). Bruno, Ravallion, and Squire(1998), while drawing in part on cross-countrydata, also analyzed one country—India—forwhich relatively long time-series data hadbecome available, and again found no sign thatgrowth increased inequality.

Why the Kuznets curve does not hold in prac-tice probably has to do with the fact that devel-oping countries do not generally satisfy theassumptions on migration processes and sectoraldevelopment underlying the Kuznets hypothesis.To explain international differences in inequalityof incomes, it is important that the link betweeneconomic inequalities and other factors, such aseconomic dualism, land, education, and regionaldifferences, be more carefully analyzed.

No straightforward relationship betweenincome and inequalityTo conclude, there is today something of a con-sensus that no straightforward relation betweenincome and inequality can be established. Asargued by Kanbur (2000) in his exhaustivereview of the Kuznets curve literature in theHandbook of Income Distribution: “it seems to usfar better to focus directly on policies, or combi-nation of policies, which will generate growthwithout adverse distributional effects, ratherthan rely on the existence or nonexistence of anaggregative, reduced form relationship betweenper capita income and inequality.”

Source: Authors’ creation.

B O X 2 . 6 Revisiting the Kuznets hypothesis for economic growth and inequality

type processes, the race between relative sup-ply and demand for skills along with house-hold adjustments to participation, educa-tion, and fertility; the transitions fromcontrolled to market-oriented economicsystems; and various forms of power andbargaining-related views of the world. Inthe end, and perhaps not surprisingly, it isdifficult to identify a single overarchingexplanation. Until recently, this did notseem to matter much because there was ageneral perception that inequality does notvary markedly over short periods.13 In ear-lier studies, few countries having data oninequality over multiple time periods indi-cated sharp changes.

For countries and regions. Empirical investi-gation of how inequality evolves in a countryis subject to concerns similar to those forcomparisons of levels (see box 2.5). But thereis a growing sense that the impression of sta-ble, unchanging income inequality may wellbe misleading. A few recent examples ofchanging inequality bear mentioning. First,careful work by Atkinson (2003) has docu-mented the evolution of inequality in OECDcountries during the second half the twenti-eth century. He finds that inequality in theUnited States has been rising steadily sincethe early 1970s (after seeing little change, andpossibly some decline, in the precedingdecades) and has risen dramatically in theUnited Kingdom since 1980. Between 1984and 1990, the Gini coefficient in the UnitedKingdom rose by 10 percentage points (butthen did not increase further)—an unprece-dented increase over such a short time. Else-where in the OECD, inequality changes havebeen less marked. But to the extent that theearly and middle decades of the twentiethcentury were associated with declininginequality in these countries, this trendseems to have halted by the century’s laterdecades.

Second, inequality in China wasmarkedly higher at the end of the 1990sthan it had been in the early part of the1980s. In general, the recent evidence inEast Asia suggests that inequality has risenfaster in the second round of high growthAsian economies—such as China and Viet-nam—than had been observed in the first

round—Hong Kong (China), Republic ofKorea, Malaysia, Singapore, and Taiwan(China). A complete picture of the factorsbehind this process is as yet unclear.Although it is likely that at least part of thestory is linked to intersectoral transfers, asemphasized by Lewis (box 2.6), Ravallionand Chen (2004) indicate that inequality inChina grew fastest during periods wheneconomic growth and poverty reductionwere slow. They argue that China provideslittle support for the view that risinginequality is inevitable with rapid economicgrowth and poverty reduction.

Third, South Asia has generally beenperceived as a region with relatively lowinequality. This probably is due, in part, toinequality being measured by consump-tion. In this region, too, the prevailing viewhas been that inequality changes little overtime. But the stylized fact of low and stableinequality in South Asia has also been chal-lenged. In India, the largest country in theregion, some uncertainty remains over howinequality has evolved, because of well-publicized issues concerning data compa-rability over time.14 The best available esti-mates suggest that inequality in India hasbeen rising, but with no solid assessment ofby how much.15

In Bangladesh, Nepal, and Sri Lanka,however, recent and reliable data show verylarge increases of inequality in the late1980s and 1990s. In Bangladesh, incomeinequality (as opposed to consumptioninequality) has been documented to haverisen from a Gini of 0.30 to 0.41 between1991 and 2000.16 In Sri Lanka, the increasein consumption inequality has been verysimilar, from 0.32 to 0.40 between 1990 and2002.17 And, in Nepal, the Planning Com-mission has produced estimates suggestingthat consumption inequality rose from 0.34to 0.39 between 1995–6 and 2003–4.18 Onlyin Pakistan is the evolution of inequalitynot clear, because of difficulties with datacomparability.

In other regions of the world, the recentpicture on inequality trends is more diffi-cult to summarize. For Latin America, DeFerranti and others (2004) indicate thatinequality increased in most countries, by asizable margin, during the “lost decade” of


the 1980s. But during the 1990s, inequalitycontinued to rise in only about half of thecountries in the region, and less rapidly. Theauthors note that, in Argentina, inequalityhas risen sharply in the growth period andduring the crisis years. In Brazil and Mex-ico, the 1990s witnessed some smalldeclines. In Eastern Europe and CentralAsia, changes in inequality during the early1990s, associated with the transition to themarket economy, have been difficult to doc-ument systematically because of data prob-lems, according to World Bank (2000c).Between 1998 and 2003, consumptioninequality declined in the former SovietUnion countries (with the exception ofGeorgia and Tajikistan), while there was noclear trend in eastern and southern Euro-pean countries (World Bank, 2005a). InAfrica and the Middle East, it is difficult topoint to broad trends, largely because ofconcerns with data comparability overtime.

To what extent does our examination oflevels and trends in income inequality bearon the themes of this report? This report ismost concerned about changes in inequali-ties in incomes, and other specific dimen-sions, if these dimensions are associatedwith changes in underlying inequalities ofopportunities. Rising income inequality inRussia during the 1990s, for example, is ofconcern precisely because of its strong asso-ciation with rising political influence andstate capture.

But this is not inevitably the case. Arecent study of income distribution dynam-ics in six East Asian and Latin Americancountries by Bourguignon, Ferreira, andLustig (2005) decomposes income distri-bution dynamics into the underlying driv-ing forces. They show that complex andcountry-specific interactions betweenpowerful underlying social and economicphenomena imply that distributional expe-riences must be assessed country by country.For example, improvements in education(equalizing opportunities) may be associ-ated in one case with falling income inequal-ity—Brazil or Taiwan, China—but inanother country with rising inequality—Indonesia or Mexico. Our assessment ofthe equity implications of changes in

income inequality will thus differ acrosscountries.

Across generations. Our assessment will alsodepend on the degree to which inequalitiesare transmitted across generations. Thestudy of intergenerational transmission ofwelfare is not straightforward, because ofthe scarcity of datasets containing informa-tion on various generations of adults in thesame family. Data from long panels are rare,and questions about family background ofindividuals are not always asked in surveys(the Brazil data described in box 2.1 are arare exception). Information about educa-tion or occupation for various generationscan be captured relatively easily in recallquestionnaires. But information aboutother dimensions, such as the incomes,earnings, or even health status of earliergenerations, is not easily remembered byindividuals (not least because they oftenchange during a lifetime). The scarcity ofintergenerational data is particularly strik-ing in developing countries. Even thoughthe persistence of inequalities across gener-ations is often thought to be much moreacute in developing countries, studies onintergenerational mobility in the develop-ing world remain few and far between.

Even when the data exist, differences inmethodologies and data often limit thescope for comparisons across countries.The most widespread measure of intergen-erational mobility in the economics litera-ture is the intergenerational earnings elas-ticity, or the elasticity of sons’ earningswith the earnings of their parents. Thismeasure generally comes from a log-linearregression of sons’ earnings (although itcould also be income or years of schooling)on fathers’ observed earnings (or its pre-dicted value using such other informationas education or occupation). The closer theelasticity is to zero, the more mobile thesociety is supposed to be. This elasticity hasbeen widely used in the U.S. literature,where longitudinal data are relativelyabundant. And for comparability, it hasalso been calculated in most other coun-tries’ recent studies.19

Until recently, estimates of the intergener-ational elasticity of earnings were thought to


be around 0.4 in the United States, suggest-ing a reasonably mobile society in incomes.20

More recently, however, Mazumder (2005)uses new data and recent econometric tech-niques to correct for transitory fluctuationsin earnings—he shows that the previous esti-mates of intergenerational elasticity werebiased downward by about 30 percent. Heargues that the true estimate is somewherearound 0.6 for the United States.

An intergenerational elasticity of 0.6 com-pared to 0.4 paints a dramatically differentpicture of mobility in American society. Forexample, it implies that a family whose earn-ings are half the national average wouldrequire five generations instead of threebefore it substantially closed the gap. Obvi-ously a difference of two generations, orabout fifty years, is quite substantial and sug-gests the need to examine policies that fostergreater mobility.21

In parallel analyses, estimates of inter-generational mobility in Canada, Finland,or Sweden, among others, have tended toreport elasticities closer to 0.2 or lower, sug-gesting that these societies are considerablymore mobile than the United States. A rela-tively early study of mobility in the UnitedKingdom (Atkinson, Maynard, and Trinder1983) reports an elasticity of 0.43, while amore recent study by Dearden, Machin, andReed (1997) estimates an elasticity of 0.57.These studies indicate that people in theUnited Kingdom are about as mobile asthose in the United States. Because of thedata limitations, only a few exceptionalstudies on intergenerational earnings elas-ticities for less-developed countries havebeen carried out. These provide evidence ofrelatively low mobility.22

In another literature review of cross-country differences in intergenerationalearnings mobility, Solon (2002) askswhether there is any link between cross-sectional inequality within a generationand the intergenerational transmission ofinequality. Although there is greater cross-sectional inequality in the United Statesand the United Kingdom than in Swedenor Finland, Canada also has relatively highinequality. The evidence needed to pro-vide a clear answer to this question istherefore still fragmentary, and only “con-

tinuing research (on international evi-dence of intergenerational mobility) willimprove our understanding of why theintergenerational transmission of eco-nomic status is strong in some countriesand weak in others.”23

The intergenerational transmission mech-anisms of inequalities will differ across coun-tries and within countries across differentpopulation groups. As described above,Mazumder (2005) points to rather low lev-els of intergenerational mobility in theUnited States. He also highlights an impor-tant racial dimension to this limited mobil-ity and finds evidence of substantial immo-bility at the ends of the distribution. Heshows that of the individuals whose fatherswere in the bottom decile of the earningsdistribution, 50 percent will be below thethirtieth percentile and 80 percent belowthe sixtieth percentile. He finds the evidenceto be consistent with the hypothesis thatsuch immobility “might be due to theinability of families to invest in their chil-dren’s human capital due to the lack ofresources.” By contrast, more than 50 per-cent of the individuals whose parents werein the top decile will remain above theeightieth percentile and two-thirds will beabove the median.

In another U.S. study, Hertz (2005) con-firms the findings of Mazumder (and oth-ers) on the size of the intergenerationalelasticity. He then shows evidence that it islargely driven by the especially low rate ofmobility of black families from the bottomof the income distribution. While only 17percent of whites born to the bottom decileof family income remain there as adults, thecorresponding figure is 42 percent forblacks. He also finds that “rags-to-riches”transitions from the bottom quartile to thetop were less than half as likely for black asfor white families. He further provides evi-dence that the black-white mobility gap isnot “appreciably altered by controlling forparents’ years of schooling.” Last, he pro-vides evidence that the incomes of blackchildren are unresponsive to small changesin parents’ incomes at the bottom of thedistribution.

To recap, summary measures suggest thateven in such developed countries as the


United States and United Kingdom there israther limited intergenerational mobilityacross generations. Research in these coun-tries has highlighted important hetero-geneities in the patterns of reproduction ofdifferent inequalities across populationsgroups. For most developing countries, rela-tively little is known about intergenerationalincome mobility. But given the acute group-based inequalities in many developing coun-tries, there appears to be little basis forexpecting much intergenerational mobility.

Agency and equity:inequalities of powerThe foregoing discussion has raised explic-itly the question of how inequalities aredetermined and reproduced. It has pointedto the potentially important role of groupdifferences in this process. This focus onprocess and the factors that account for thepersistence of inequality over time puts thespotlight on how much inequality is rootedin deeper institutions in society—institu-tions of governance, access to land, controlof labor, market regulation. Chapter 6 dealswith the emergence and effects of suchinstitutions in more detail. Here we turn todifferent kinds of evidence—and tradi-tions of analysis—to discuss the unequalcapacity of people to influence the formtaken by these institutions and the conse-quences of unequal institutions for contin-uing inequality in such capacities. For

poverty the inequalities in capacity to forgethe institution or society can be as impor-tant as inequalities in health, income, andeducation.24

A recent study of inequalities in gover-nance in four slums of Delhi found thataccess to formal government by slumdwellers is more available to the better offand to those who have good contact net-works.25 Community leaders in these slumsfacilitate access primarily to their castemembers, and slum dwellers are more likelyto delegate custodianship of their intereststo better-educated community leaders. Thestudy concludes that because access tobureaucracy and political representation forslum dwellers in Delhi is largely the pre-serve of the better off and better connected,decisions of formal policymakers do notseek to represent slum interests as a whole,producing interventions that do not targetthose in most need. The lack of broadly dis-tributed “voice” thus results in patterns ofresource allocation, and income generation,that are far from egalitarian.

The nature of this unequal capacity canbe captured through the sociological con-cept of agency. Agency refers to people’scapacity to transform or reproduce suchsocietal institutions. Some of this capacity isconscious—for example, when interestgroups lobby for a change in land tenurelegislation, or when women refuse to acceptlaws around marriage that systematicallydisadvantage them. Some of it is uncon-


A recently completed Human DevelopmentReport for Pakistan provides rich documenta-tion of the skewed distributional impact of cor-ruption (United Nations DevelopmentProgramme 2003).The report notes that corrup-tion raises the costs of getting things done—forsetting up a new business, for crossing borders,for obtaining a driver’s license. In Pakistan thesecosts fall most heavily on those least able toafford them: the poor. According to the PakistanHuman Development Report, 16.7 percent ofthe extremely poor reported paying a bribe torun their business enterprise, handing over anaverage of 6,800 rupees. Only 6.7 percent of thenon-poor paid a bribe, of 9,300 rupees. In ruralareas, the contrast is even starker: 20 percent of

the extremely poor paid a bribe, while only 4.3percent of the non-poor had to do so. In urbanareas, the extremely poor paid on average 8,700rupees in bribes, while the non-poor paid only1,200 rupees.

Similar patterns emerge for mediating dis-putes.The extremely poor not only pay a higherprice to seek a resolution than the non-poor, butalso they are less likely to receive a satisfactoryoutcome (38.5 percent versus 80.8 percent).Indeed, the fee the extremely poor must pay isoften higher than their annual householdincome, leaving many to choose to suffer theconsequences of a dispute even when they areclearly in the right. In addition, the extremelypoor receive less assistance from the police (the

most immediate representatives of the formaljustice system), who are involved in only 1 per-cent of their disputes but nearly 5 percent of thedisputes of the non-poor.The poor perceive thatthe police will be slow and inefficient inhandling their cases, and they frequently experi-ence outright harassment and intimidation.Even to register a case of kidnapping with thepolice requires paying a bribe. In thesesituations, it is hardly surprising that the poorfind it more expedient to take the law into theirown hands, creating in many urban areas a hostof new problems related to gang violence andvigilantism.

Source: United Nations Development Programme(2003).

B O X 2 . 7 Inequitable agencies and institutions in Pakistan

scious—for example, when people engagein land transactions without questioningthem, they reproduce the institutions ofland tenure and the markets in land. Whena disadvantaged group accepts its disadvan-tage as “taken for granted,” the effect is toallow the continuing existence of the rela-tionships that create such disadvantage.

The internalization of disadvantageleads to pernicious forms of agency thatperpetuate inequalities. From inequalitiesin agency come inequalities in power, voice,and self-confidence—a major part of ourstory (box 2.7). Inequalities of agency are asmuch products of dominant institutions assources of those institutional arrangements.Maintaining these arrangements bothreflects and produces the distribution ofpower among people. As for health, educa-tion, and income, though, this distributioncan change—and it has. Indeed, it has oftenchanged in relation to changes in theseother distributions.

Internalization of disadvantage and inequalities of agencyRecent work on urban slum dwellers inIndia26 (and elsewhere)27 suggests that a keyform of powerlessness for the poor involvesliving with “negative terms of recognition.”This concept highlights the conditions andconstraints under which the poor negotiate

the very norms that frame their social lives.Being so routinely treated with contempt bygovernment officials, employers, and fellowcitizens—and encountering such enormousobstacles to advancement—means thatexcluded groups can, over time, come tosubscribe to norms about themselves andtheir situation “whose social effect is to fur-ther diminish their dignity, exacerbate theirinequality, and deepen their lack of accessto material goods and services.”28

In these circumstances, the poor are notonly persistently and overtly discriminatedagainst. Their problems are further com-pounded and consolidated by their apparentcomplicity in it, their revealed “adaptivepreference”29 for menial occupations andascription to norms and subservient behav-iors that only legitimize and perpetuate theirpowerlessness. Dire material circumstances,rational expectations about their limitedprospects for upward mobility, and strongbeliefs about the legitimacy and immutabil-ity of their situation conspire to create avicious circle from which it may be very dif-ficult for the poor to escape (see box 2.8).30

Inequality traps may cause crime andviolence. First, people who perceive theirpoverty as permanent may be driven byhostile impulses rather than rational pur-suit of their interests. Second, sensitivityto inequality, especially by those feeling


The Batwa, who are described in many parts aspygmy peoples, live in Eastern Uganda, easternDemocratic Republic of Congo, and Rwanda.Batwa have been subject to negative stereotypessince at least 1751, when Edward Tysonconcluded that pygmies were not human butrather apes or monkeys.They suffer multipleasset depletion and wide ranging forms of dis-crimination, a situation that public actions haveat times made worse.Though longstanding forestdwellers, the British sought to expel them to cre-ate forest reserves in the 1930s. In 1991, theUganda National Park authorities increasedefforts to enforce this exclusion from forest areas.Although the World Bank—which was fundingsome of the park authorities’work—required thatthe government assess the impact on indigenouscommunities and follow defined compensationprocedures, these did not take sufficient account

of power differences among Batwa and otheraffected groups, nor consider Batwa preferences.All communities were viewed as uniform, a prac-tice that the authorities later recognized “did nottake into account Batwa realities and left themwith nothing”(Zaninka (2003), 170).

Non-Batwa locals have resisted efforts to pro-vide more appropriate compensation to theBatwa. A Participatory Poverty Assessment (PPA)highlighted persistent discrimination, describingthe Batwa as a “group of people who aredespised”and who “have no means of productionsuch as land, credit and training.They areregarded by other ethnic groups in Kisoro as apeople with no rights.”This leads to everyday andinstitutionalized forms of exclusion, with theBatwa suffering discrimination in access to bothpublic spaces and services.While some Batwarespond to this by organizing themselves, others

respond in ways that—however rational and self-protecting—often reproduce the extent to whichthey are excluded.The same PPA reports someBatwa children saying that they did not attendschool because it was so unfriendly to them.When asked what they wanted to do upon com-pleting school, one child replied that she wishedto be “a cleaner.”Discrimination and prejudicediminish the capacity to aspire to and imagine adifferent future.

Repudiation and discrimination can alsolead the Batwa to self-exclude from the publicsphere.The PPA notes that no Batwa attendedPPA exercises. Non-Batwa locals explained:“Batwa would never come to such meetings, sothere is no point in mobilizing them.”

Source: Moncrieffe (2005), citing ParticipatoryPoverty Assessment reports.

B O X 2 . 8 Legacies of discrimination and the reproduction of inequalities and poverty among theBatwa in Uganda

trapped at the bottom, may lead to higher-risk tactics like crime, when the expectedpayoffs from socially legitimate activitiesare poor. Third, people may be particularlysensitive to group-based inequalities. If, forexample, racial heterogeneity and incomeinequality are correlated and consolidatestatus distinctions in a society, this couldspell potential for violence. Finally, as Mer-ton (1938) elegantly states,

. . . when a system of cultural values empha-sizes, virtually above all else, certain commonsymbols of success for the population at largewhile its social structure rigorously restrictsor completely eliminates access to approvedmodes of acquiring these symbols for a con-siderable part of the same population, . . . anti-social behavior ensues on a considerablescale. (italics reflect original emphasis)

A lack of upward mobility in a society, com-bined with a high premium on economicaffluence, results in anomie—a breakdown ofstandards and values.31

Changing between-group inequalitiesof agency and institutional powerInequality of agency often leads to institu-tions that reproduce such inequality. Butthese relationships are not immutable.There are ample cases in which interven-tions—by civil society, reformist publicofficials, external actors, religious institu-tions, and others—have given more self-confidence and assertiveness to disadvantagedgroups, worked against the internalization ofdisadvantage, and created new channels forexcluded groups to exercise voice with greatereffect. These changes improve the terms ofrecognition for the powerless: they becomerecognized by more powerful groups whootherwise would not acknowledge them at all,leading to empowerment of disadvantagedgroups in economic, social, and politicalrealms.

Empowerment can occur in many ways.32

Change typically occurs through the interac-tion between the opportunities for actioncreated by dominant political structures andthe capacity of poorer or middle groups toengage. The “political opportunity struc-ture”—that shapes the possibilities foraction—is itself a function of the openness of political institution, the coherence and

positions of elites, and the effectiveness ofgovernments to implement approved coursesof action. The capacity of subordinate groupsis influenced by their “economic” capital—their education and economic resources—their “capacity to aspire,” and the closely asso-ciated capacity to organize.33

In Indonesia, the Kecamatan Develop-ment Project (KDP) illustrates changeoccurring through action from above andbelow: it aims to improve the terms of recog-nition and the political agency of marginalgroups, and to create new institutions forgreater agency to lead to material changes inpatterns of public investment. Consistentwith the ongoing process of democratizationin Indonesia, the source of change comesfrom public policy rather than nongovern-mental action, allowing the project to oper-ate on a large scale (see also focus 4 for exam-ples of change occurring at the local level).

A recent study34 of the efficacy of theKDP on challenging and changing theterms of recognition of participants sug-gests that it does provide villagers with a setof deliberative routines for more equitablymanaging the conflicts it inevitably trig-gers.35 These routines introduce marginalactors to more equitable spaces of engage-ment with more organized and influentialactors. But building this conflict manage-ment capacity among marginal groupsdepends on more than just forging collabo-rative routines. It also requires a set ofrules—defined by the KDP—that limit theunfair exercise of power by dominantgroups. With the KDP cultivating collabo-ration and tangible points of politicalpower for marginalized groups, the resultsinclude a well-functioning school or med-ical clinic but equally important a style ofgroup (re)definition and defense.

Changes in the agency of indigenouspeoples in Ecuador since the 1960s provideanother example in which mobilizationfrom below came to change national andlocal structures. These changes are clear atboth local and national levels. In the 1960sin the Andean province of Chimborazo,the indigenous Quichua people sufferedmultiple deprivations. They were subjectto everyday forms of violence and to dom-ination and racism in their interactions


with other ethnic groups and with author-ities.36 Power was concentrated in the tri-umvirate of landowner, priest, and localgovernment authority. Much indigenouslabor was tied to large rural estates onwhich labor relations were sometimes vio-lent and returns to labor manifestly unfair.Life expectancy was short, alcoholismsevere, and children’s access to educationand health acutely constrained.

At the start of this twenty-first century,indigenous people now occupy severalcounty mayorships and have a majority ofcouncilors in several counties. The provin-cial prefect is also Quichua. Similarly at anational level, former leaders of nationalindigenous people’s organizations are nowministers. And the national Confederationof Indigenous Nationalities of Ecuador hascontrol of the directorate of bilingual edu-cation, the indigenous development coun-cil, and the office of indigenous health. Italso played a big part in negotiating andadministering a World Bank and Interna-tional Fund for Agricultural Develop-ment–supported national Program for theDevelopment of Indigenous and Afro-Ecuadorian Peoples. By any calculation,power relationships have changed inEcuador, becoming more equitable, withindigenous people participating morecompletely (and more equitably) in theirsociety.

The inequality trap for womenUnequal opportunities in health, education,economic welfare, and political agency canbe readily observed in most developingcountries. The preceding sections haveemphasized that these different manifesta-tions of inequity are not generally inde-pendent from one another and that thisinterdependence can replicate inequalitiesover time. This interrelationship can bevividly illustrated by examining the natureand implications of the inequality that trapsmany women in developing countries.

Men and women around the world havestarkly different access to assets and opportu-nities, reinforced by unequal norms andsocial structures, perpetuating gender differ-ences over centuries. Gender inequity directlyaffects the well-being of women and deci-

sions in the home, affecting investments inchildren and household welfare (box 2.9).

Gender inequity is the archetypical“inequality trap.” Most societies have normsthat preserve the prevalent social order, delin-eating different roles and spheres of influencefor men and women. The male sphere is typi-cally outside the home in market work andsocial interactions that enhance the family’s


Gender inequity causes many societies todisplay some preference for male children.But the “son preference” is strong enoughto result in substantial excess female childmortality in parts of East and South Asia—leading to the phenomenon of whatAmartya Sen calls “missing women.” (Sen,1990). In China and India the practice offemale infanticide was noted at least acentury ago, and in the Republic of Koreaand India high juvenile sex ratios (the pro-portion of male to female children belowthe age of 4) have been documented sincethe first modern censuses were taken. Bycontrast there seems to be little son pref-erence in Southeast Asia or in most otherparts of the developing world.

The reasons for this seem to stem fromrigid patrilineal inheritance systems.Whilemost societies deny women inheritancerights, in other parts of the world there issome flexibility in these rules. In peasantEurope and Japan, for instance, womencould inherit land if their parents had nosons. Despite egalitarian laws, customarypractices in China, the Republic of Korea,and northwest India permit a man, if hedoes not have sons, to adopt one fromother male kin. In the past, it would alsohave been possible to take another wife.The driving motivation is to use whatevermeans possible to continue the male fam-ily line.Thus, girl children are undervalued.

During pregnancy, sex-selection maylead to aborting female fetuses, reflected insex ratios at birth that are more masculinethan the biological rate of 105 boys forevery 100 girls. Sex-selection can also hap-pen through infanticide, although the datamake it difficult to distinguish betweenselective abortion and infanticide.The third,and most common, mechanism is the neg-lect and other practices that result in highermortality rates for girls than boys duringearly childhood.37

In China, intense efforts by the govern-ment resulted in a brief improvement in thesex ratio during 1953–64 (see figure to theright). But since the 1980s it has steadily

risen. In the Republic of Korea stark declineshave become only apparent in the lastdecade—perhaps because ofimprovements in labor market opportuni-ties for women. India, as a whole, does nothave juvenile sex ratios that are far differentfrom many other parts of the world. Butnorthwest India has seen some particularlyworrying trends, with sex ratios sharply ris-ing between 1981 and 2001, much attribut-able to the higher incidence of sex-selection in abortion. Other parts of India,especially the south, have more equitablelabor markets and fewer restrictions onwomen’s mobility and inheritance.

Source: Das Gupta and others (2003).

B O X 2 . 9 Sex ratios and “missing women”

1.00 1.05 1.10 1.15 1.20 1.25

1953 1964 1982 1990 1995 2000

1949 1960 1970 1980 1990 1995 2000

1951 1961 1971 1981 1991 2001

1951 1961 1971 1981 1991 2001

China

Republic of Korea

Punjab and Haryana (India)

India

Juvenile (newborn to four years old) sex ratios inChina, the Republic of Korea, India, and Punjaband Haryana, 1950–2000

status and power. The female sphere is usuallyinside the home—looking after householdwork, rearing children, and contributing tothe stability of the household. So, women’sactivities serve primarily as inputs into thehousehold’s collective well-being, while menare ostensibly at its center—its breadwinnersand its link to the larger world where eco-nomic and social status are determined.

Marriage and kinship systems preservethese structures of patriarchy. Most societiesare “patrilocal,” with women moving fromtheir parents to their husband’s home aftermarriage. Marriage can therefore be thoughtof as a framework that serves to exchangewomen between households, and marriagedecisions are made with a view toward ensur-ing that this exchange of women promises themaximum gain to both households. Theman’s household is the point of reference—while the woman is simply an input into theprocesses for households controlled by mento generate economic and social returns.38

Inheritance tends to be consistent withthis pattern. Most societies are not justpatrilocal—they are also patrilineal, withinheritance and property rights primarilypassed on to men. The majority of countries,outside of Europe and Central Asia and LatinAmerica and Caribbean, restrict inheritancerights to women.39 Some countries have leg-islation that guarantees equality in inheri-tance laws. But these laws often are notenforced, and real authority over decisionson inheritance rests in the hands of villageelders and chiefs, who follow customarypractices that discriminate against women.

Most countries that have unequal inheri-tance laws also have unequal property rightsregimes.40 Indeed, the vast majority of landowners are men.41 Many societies compoundthis by denying women the right to divorce.This inequality in property rights regimespersists even in countries where agriculturalproduction depends heavily on women’slabor, such as many in Sub-Saharan Africa.In Cameroon, women make up more than51 percent of the population and do morethan 75 percent of the agricultural work, butthey are estimated to hold fewer than 10 per-cent of all land certificates.42 So, if womenwork on farms, they are usually working onfarms owned by men.

In addition to being denied inheritanceand property rights, women in many soci-eties face restrictions on their mobility. Forexample, in the state of Uttar Pradesh innorthern India close to 80 percent ofwomen require their husband’s permissionto visit a health center, and 60 percent haveto seek permission before stepping outsidetheir house.43 These mobility restrictionsmay be socially imposed, as with gunghatamong Hindus—or have religious sanc-tions, as with purdah among Muslims. Suchpractices are not just socially enforced, theycan be internalized by women who treatthem as marks of honorable behavior. Thesenorms are transmitted by parents to theirchildren, ensuring their continuity over gen-erations; in many societies, they are enforcedby older women in the community.44

Restrictions on mobility and rules of kin-ship and inheritance help shape social per-ceptions about women’s roles. If women aresocially and economically directed to focustheir attention and energy on activities in thehome, this is not just what men expect ofthem—it is also what other women expect ofthem. In much of the developing world,women’s participation in the labor market ismore a function of adversity than activechoice—because husbands cannot earn anadequate income or because of an unantici-pated shock, such as a child’s illness.Bangladeshi women described it this way,“Men work to support their families, womenwork because of need.”45 Women around theworld participate in a fair amount of market-based activity for a wage, but they have to con-tinue to perform most household chores (fig-ure 2.13). They thus face a time squeeze,spending more time at work, both in and outof the home, than men do.

Because social and economic factorsdetermine women’s life chances more in mar-riage than in labor markets, parents invest lessin their human capital. Throughout thedeveloping world, women are much lesslikely to be enrolled in secondary school oruniversity than men.46 So, they typically workin less lucrative occupations. Moreover, labormarkets may themselves be discriminatory,paying women less than men for the samework. For these reasons, even when womenparticipate in the labor market, they earn less


than men. Low earnings are a further disin-centive for women to enter the labor market,perpetuating traditional social roles.

Inequality in the homeFor a long time, economists did not ade-quately recognize that gender inequity has animpact in the home, and models of thehousehold assumed that decisions were takenby one person—with no room for differentchoices across spouses. The consequence ofthis world view is not just academic. It sug-gests, for instance, that policy interventionsthat attempt to alleviate poverty should notbother with targeting by gender—or suggeststhat taxes on a household will not affect theallocation of resources within it.

Economists now question this view,developing models of household decisionmaking that allow for inequality betweenspouses. The new models start with theassumption that households are efficient, inthe sense that they make decisions that max-imize the use of the household’s resources.With this assumption, the models show thata spouse’s share in household resources isdetermined by two factors. The first is thefallback option for the spouse in the event ofdivorce—laws of inheritance, property, anddivorce would matter here. Second is the rel-ative size of the spouse’s contribution to thehousehold’s income, which is determined bytheir opportunities in the labor market.47 Ifhusbands and wives have different prefer-ences, an increase in a woman’s outsideoptions or in her labor market opportuni-ties should reflect consumption choicesmore in line with her preferences.

Econometric work confirms that an in-crease in a woman’s relative worth and animprovement in her fallback options haveeffects on consumption patterns.48 Thehealth of Brazilian children improves whenadditional nonlabor income is in the hands ofwomen.49 In the United Kingdom, when leg-islation ensured that child support paymentswere made directly to mothers, expenditureson children’s clothing tended to rise.50 InBangladesh and South Africa, women bring-ing more assets into the marriage increasehousehold expenditures on children’s educa-tion.51 The patterns seem to indicate that,when women are better off, children seem to

benefit more than when men are better off.The most obvious way to explain bargainingand sharing is to assume that women intrin-sically care more about children than men do,but this risks being tautological.

Perhaps the explanation can benefit fromunderstanding that social and economic dif-ferences outside the household can matternot only for determining bargaining powerbut also for socially determined perceptionsof what men and women consider impor-tant. If men and women occupy different“outside” and “inside” spheres of influence, itseems to make sense that improvements inwomen’s incomes would have a greaterimpact on investments in the household.Improvements in the income of men, bycontrast, are more likely to result in socializ-ing activities outside the home and in pur-chases that reflect social status.

Another consequence of this separationbetween inside and outside roles is that


0 400400Minutes a day

Australia

Marketactivity

Marketactivity

Women Men

Nonmarketactivity

Nonmarketactivity

R.B. de VenezuelaUnited States

United KingdomPhilippines

NorwayNetherlandsNepal, urban

Nepal, ruralKenya, urban

Kenya, ruralItaly

IsraelIndonesia

GuatemalaGermany

FranceFinland

DenmarkColombia

CanadaBangladesh

Austria

Figure 2.13 Women work longer hours than do men

Source: United Nations Development Programme (1995).Note: Data refers to rural Bangladesh in 1990, urban Colombia 1983, rural Guatemala 1977, urban Indonesia 1992, ruralKenya 1988, urban Kenya 1986, rural Nepal 1978, urban Nepal 1978, rural Philippines 1975–77, urban Venezuela 1983,Australia 1992, Austria 1992, Canada 1992, Denmark 1987, Finland 1987-88, France 1985–86, Germany 1991–92, Israel1991–92, Italy 1988–89, the Netherlands 1987, Norway 1990–91, the United Kingdom 1985, and the United States 1985.

inequalities in the home are also manifestedin differences in access to information,which can be used to manipulate intra-household bargaining. In an ethnographicstudy of Bangladeshi garment workers,Kabeer (1997) found that men and womentried to control information about theirincomes from their spouses so that theycould make purchases without consultingthem. Women may also hesitate to shareinformation with their husbands, or to col-laborate efficiently in farming their plots ofland, to retain control over their property. Instudying the farms owned by men and thoseowned by women in Ghana, Udry (1996)found, keeping other things constant, thatwomen-owned farms were less productivethan those owned by men. When wives andhusbands are not sharing information, ormanipulating the flow of information, theyclearly are not using their resources opti-mally. In other words, intrahouseholdbehavior is not efficient—contradicting animportant assumption in economic models.

The widespread domestic violence in thefamily is another type of inefficiency. RecentWorld Health Organization (WHO) datashow that both physical and sexual violenceare widespread in diverse parts of the world(table 2.2). An important reason for domes-

tic violence is that it allows husbands toinstitute a regime of terror to control theirwives’ behavior. In India, Bloch and Rao(2002) find that husbands systematically useviolence as a means of extracting a largerdowry from their wives. This “instrumental”use of violence has widespread acceptanceamong both men and women. Surveys havefound that large percentages of respondentsin developing countries report that menhave the right to beat their wives when theyanswer back or disobey them.52

Gender inequity is thus the result of anoverlapping set of economic, social, cultural,and political inequalities that reinforce eachother. They cause women to have less access toproperty rights, wealth, and education—andlimit their access to labor markets and tospheres of activity outside the home. This, inturn, constrains their ability to influencehousehold decisions. Also limiting this influ-ence are asymmetries of information in thehousehold and the use of violence to controlwomen’s behavior. All of this maintains a cleardemarcation between the roles of women andmen, readily reproduced across generations.

There are some signs that changes inlabor markets and interventions by the statecan break this inequality trap. The develop-ment of the garment industry in Bangladeshhas resulted in a sharp and visible increase inwomen’s access to a lucrative labor market,expanding their ability to influence house-hold choices.53 Higher wages for womenseem to compensate for restrictive practices,such as purdah, by reducing limits onwomen’s physical mobility, and increasingtheir say in household decision making.54

Globalization has expanded opportunitiesfor women in Mumbai and increased theiraccess to schooling.55 A comparative study ofthe Philippines, Sumatra, and Ghana foundthat patterns of land inheritance and invest-ments in schooling have became more egali-tarian because of changes in labor marketopportunities for women.56 And althoughChina, Republic of Korea, and India startedout with similar discriminatory social struc-tures, intervention by the state has improvedgender equity much more in China than inRepublic of Korea or India.57


Table 2.2 Percentage of women who have everexperienced physical or sexual violence by anintimate partner

Physical violence Sexual violence

Bangladesh, rural 42

Brazil, urban 27 10

Ethiopia, rural 49 59

Namibia, urban 31 17

Peru, rural 62 47

Samoa 41 20

Serbia and Montenegro 23 6

Tanzania, urban 33 23

Thailand, rural 34 29

Source: Unpublished data from the WHO Multi-Country Study onWomen’s Health and Domestic Violence Against Women obtainedfrom a presentation by Claudia Garcia-Moreno at the World Bank’sConference on Gender-Based Violence. The final published com-parative report is forthcoming.Note: Data refer to different time periods. Brazil, Peru, and Thailandrefer to 2000. Reference period for Bangladesh, Ethiopia, Namibia,Samoa, Serbia and Montenegro, and Tanzania are unknown.

55

In examining the inequality of opportuni-ties within countries, the previous chapteremphasized people’s “predetermined cir-cumstances,” or life chances beyond theircontrol, as distinct from their “efforts” and“talents” as individuals. One of these cir-cumstances is a person’s place of birth. Inmany countries, access to basic publichealth services, for example, is significantlylower in rural areas than in urban areas.That can mean much for surviving the firstyear of life—the infant mortality rate in Riode Janeiro was 3.3 percent in 1996, less thanhalf the 7.4 percent in northeast Brazil.

But, just as being born in a village or acity is one circumstance that should beirrelevant to a person’s chances in life, beingborn in a specific country is another. Why isit objectionable for, say, Turkish women tohave inferior opportunities and outcomescompared with Turkish men, but not soobjectionable if the comparison is betweenTurkish men and English women? After all,in many dimensions of well-being, majordifferences in opportunities and outcomesexist between citizens of different countries,in some cases differences larger than thosebetween various groups within countries.

This chapter tries to answer two questions.First, how much does one’s country of birthdetermine one’s opportunities in life? Second,does one’s country of birth mean less for lifechances today than in the near or distantpast? To answer these questions, we discussinequalities in health, education, income, andpower in the global arena. We show that theinequalities between countries are staggeringdespite some improvements over time.

Examples and conceptsThere is no doubt that we live in a worldwith massive inequalities in the opportuni-

ties to live a free, healthy, and fulfilled life.As Angus Deaton writes,

We are living with appalling inequalities, inwhich the poor of the world die of AIDS, and,more broadly, where poor people around theworld die of diseases that are readily prevent-able elsewhere, including in the first-worldhospitals and clinics that serve the rich inpoor countries.1

In 2000 the life expectancy of a child bornin Sierra Leone (37 years) or Botswana (39years) was less than half that for a childborn in the United States (77 years).2 Theaverage educational attainment (uncondi-tional on quality of schooling) of an indi-vidual born in a Sub-Saharan countrybetween 1975 and 1979 is less than 6 years,but more than 12 years in OECD coun-tries. Inequalities in income are also highamong individuals in different parts of theworld.3

How do we view large average improve-ments in the world, set against this pictureof unacceptable inequalities between coun-tries? Sen (2001) describes the current stateof the world while making the case for afairer distribution of the fruits of globaliza-tion: “Even though the world is incompara-bly richer than ever before, ours is also aworld of extraordinary deprivation andstaggering inequality.” He argues thatwhether there have been some gains for allis not as important as whether the distribu-tion of gains has been fair. Inequalities inaffluence—and in political, social, and eco-nomic power among countries—are cen-tral to the debate on globalization. As longas the sharing of potential gains from glob-alization is viewed as unfair by many, theinequalities described in this chapter willbe deemed unacceptable. This, despite the

Equity from a globalperspective

3c h a p t e r

fact that absolute poverty has declined inthe last two decades—though by no meansuniformly.

To put global inequalities in well-being inperspective, it helps to examine two coun-tries at opposite ends of the spectrum—Mali, one of the world’s poorest countries,and the United States, one of the richest. Ababy born in Mali in 2001 had an approxi-mately 13 percent chance of dying beforereaching age one, with this chance decliningonly slightly (to 9 percent) even if the babywere born to a family in the top quintile ofthe asset distribution. By contrast, a babyborn in the United States the same year hada less than 1 percent chance of dying in itsfirst year. The picture for under-five mortal-ity is even more egregious: 24 percent ofchildren will not reach age five in Mali,compared with less than 1 percent of Amer-ican children. Even a child born into therichest quintile in Mali is more than 16times likely to die before age five than anaverage American child.

The picture does not improve for educa-tion. The average American born between1975 and 1979 has completed more than 14years of schooling (roughly the same formen and women, and in urban and ruralareas), while the average school attainmentfor the same cohort in Mali is less than twoyears, with women’s attainment less thanhalf that for men, and virtually zero in ruralareas. If one considers the quality of theeducation received, the inequalities in learn-ing achievement are possibly much larger.

It is not surprising, then, that many citi-zens of Mali, having survived immensehardships as children and without mucheducation, can barely eke out a living asadults, on average living on less than $2 aday ($54 a month) in 1994. By comparison,the average American earned $1,185 amonth, more than 20 times that for theaverage Malian.

While there is probably some consensusthat inequalities in health, education,income, and voice are large globally, there ismuch less agreement on whether thingshave been getting better or worse. Is one’scountry of origin more or less pertinenttoday to the life chances that she faces atbirth than it was 20, 50, or 200 years ago?

The debate on inequalities in variousdimensions of well-being and their relationto globalization rages on as you read thisreport.4 It is indeed harder to assert whetherinequalities increased or decreased overtime. Various questions have to be answeredfirst: inequality of what, over which timeperiod, using which concept of inequality?While there is some evidence of convergencein opportunities in health and educationand some divergence in incomes (or at leastlack of convergence), these results cannot bestated without many qualifications andcaveats. Box 3.1 introduces some underlyingconcepts that need to be clarified.

Global inequalities in healthThe unweighted and weighted internationaldistributions of life expectancy at birth(ignoring the distribution of life expectancyat birth within countries) both show a clear“twin-peakedness” in 1960.7 Data show that50 countries had life expectancies between35 and 45 years, 41 countries had lifeexpectancies between 65 and 75 years, andthere was relatively little mass in the middleof the distribution.

By 1980 the left-hand mode of the distri-bution had decreased considerably in size.The distributions began to look moreright-skewed, unimodal, especially in theweighted international distribution: 73countries had a life expectancy between 65and 75, compared with 31 countries between55 and 65, and 35 countries between 45 and55. But by 2000 the two modes become evi-dent once again, especially in the unweighteddistribution, although there is more mass inthe right mode of the distribution.

In 1980, the average life expectancy infour regions—Middle East and North Africa,East Asia (excluding China and Japan),South Asia, and Sub-Saharan Africa—wasbelow the world average.8 Between 1980 and2000, rapid increases in life expectancy in thefirst three of these regions were globallyinequality-reducing, while the decline of lifeexpectancy in Sub-Saharan Africa in the1990s boosted inequality by stretching thebottom tail of the distribution. By 2000,only South Asia and Sub-Saharan Africawere below the world average, with the dif-ference in life expectancy at birth between


Equity from a global perspective 57

On the welfare gains from globalization, thetwo sides of the debate often makestatements that are diametrically opposed, allthe while examining the same data. Whilethere are some differences in and problemswith data, the wide discrepancy in views onthe topic seems to stem from the fact that thetwo sides do not share the same values aboutwhat constitutes a just distribution of thegains from globalization.

Considered here are three differentconcepts of inequality, drawing from Milanovic(2005) and Ravallion (2004a). Both authors, andthe globalization debate in general, discussthese “competing concepts” in the domain ofincomes. But these concepts can be extended toother dimensions, such as health and education(especially for inequality between countries).The conclusions one would draw in each ofthese dimensions of well-being then depend onthe concept of inequality adopted. It is impossi-ble for the two sides to communicate withoutfirst making these concepts clear.

Has global income inequality increased ornot? Before we can answer this, we have todefine what we mean by global inequality andhow that differs from what we will call interna-tional and intercountry inequality.

Global inequality: forget countryboundaries, each person has his or her real incomeGlobal inequality is easy to define: simply forgetcountries, line up all citizens of the world, andcalculate the inequality in the distribution oftheir real incomes, adjusted for purchasingpower parity.5 The global inequality measuresthat belong to the general entropy class, such asa mean log deviation or Theil’s index, can beneatly decomposed into inequality attributableto inequalities between persons within eachcountry and the mean differences of incomebetween countries (Shorrocks 1980).

Within-country inequality is what the overallinequality in the world would be if there wereno differences in mean consumption acrosscountries but each country had its actualinequality level. Between-country inequality canbe interpreted as measuring what the level ofinequality in the world would be if everyonewithin each country had the same (the country-average) consumption level.Total inequality inthe world is the sum of these two parts, and theratios of the respective parts to total inequalityprovide a measure of the percentage contribu-tion of between-country and within-countryinequality to total inequality.

International inequality: each person hashis or her country’s mean incomeThroughout the rest of the report, we will referto this between-country inequality as interna-

tional inequality, the inequality in the distribu-tion of all of the world’s citizens, but with eachperson assigned the mean income of his/hercountry instead of his/her own income. Globalinequality is calculated by simply adding inter-national inequality to within-country inequality.

Intercountry inequality: each country hasone representative at its mean incomeThese two concepts, however, are not enough tosettle the debate.Think of the following state-ment in support of the argument that inequalityin the world has been increasing:“The GDP percapita of the richest country in the world wasabout 9 times that of the poorest around 1870compared with 45 times by 1990.”6 Notice thatwhile this statement seems to be referring tosomething akin to international inequality, thereis a subtle but very important difference: thesize of the richest or the poorest country playsno role in this statement.The statement remainsthe same whether the richest country is Palauand the poorest country is Jamaica, or whetherthey are China and India.

This is why a third concept is needed. In thisconcept, all countries of the world (instead of allcitizens) line up together, and each of them isassigned her mean income.We will call theinequality in this distribution (of roughly 200 orso countries of the world) intercountry inequal-ity. Milanovic (2005) refers to our intercountry,international, and global inequality concepts asConcept 1, Concept 2, and Concept 3 inequality,respectively (see figure below).

Why use intercountry inequalityThe implicit value judgment in using inter-country inequality instead of internationalinequality is that countries, not people, shouldget equal weight in assessing the fairness of

the division of the gains from globalization.The measures most widely quoted by the crit-ics of globalization treat each country as oneobservation, while decompositions of worldinequality into between-country and within-country components described above givepeople equal weight, whether they live inChina or Chad.

Note that in the globalization debate, thechoice of the measure of inequality can alsodepend on the question one is trying toanswer. If one is interested in the impact ofsome “globalizing” policies on growth or distrib-utional outcomes at the country level, it mightbe preferable to use a measure of intercountryinequality.

Why use international inequality—as we do in this report?Alternatively, if we are trying to determinewhether world poverty or inequality decreasedas a result of “globalizing” policies, then wemight be more inclined to examine measures ofinternational inequality.

No right or wrong choiceArguments can be made in favor of each ofthese two concepts when assessing trends ininequality between countries. This choice is nota matter of what is right or wrong. When itcomes to judging inequality, intelligent peoplecan disagree about whether countries or peo-ple should be weighted equally—somethingthat Ravallion (2004a) argues in detail. Thepoint: the judgments (or the questions of inter-est) that affect the choice of the inequality con-cept employed in empirical work matter greatlyto the assessment one can make about the dis-tributive justice of current globalizationprocesses.

B O X 3 . 1 Three competing concepts of inequality: global, international, and intercountry

Intercountry inequality:Three countries and three representatives

with mean incomes (height)

International inequality:Entire population included,

but with mean incomes

Global inequality:All individuals with their actual income

Three concepts of inequality illustrated

Sources: Milanovic (2005) and Ravallion (2004a).

these two regions having increased from 5.8years to 15.6. Between-country inequalitydeclined until the early 1990s and thenincreased back to its 1980 level by 2000. Thelarge decline in life expectancy at birth inSub-Saharan Africa more than offset theinequality-reducing effect of growth inSouth Asia in the 1990s.

Over a longer period (1820–1992) Bour-guignon, Levin, and Rosenblatt (2004a)show tremendous gains in life expectancy atbirth (rising from approximately 27 years to61 years), unequally distributed at first,then equalizing in three waves between latenineteenth century and 1990. Decades of

consistent improvements in life expectancyat birth came to a screeching halt in the1990s (table 3.1). Between-country inequal-ity among developing countries is as high asit has ever been since 1960.

So, there is some convergence in lifeexpectancy at birth over a long period,although there are significant losses in the1990s in Sub-Saharan Africa, mainly causedby AIDS, and in some European and CentralAsian countries.9 With the developed coun-tries reaching a biological limit at the top ofthe distribution and many regions catchingup to them, the inequality of life expectancyin the world will become more a function ofchanges in health and population growth inSub-Saharan Africa—barring a major healthcatastrophe elsewhere in the world. (Werevisit this issue at the end of the chapter.)But for now there remain two worlds withsignificantly different life expectancies: thegap in life expectancy between Sub-SaharanAfrica and Europe and North America in2000 is higher than it was in 1950.10

Health outcomes of even the rich citi-zens in poor countries remain well below


29 35 41 47 53 59 65 71 77 83 89

0.05

0.04

0.03

0.02

0.01

0.00

Estimated density, weighted

0.05

0.04

0.03

0.02

0.01

0.00

Estimated density, unweighted

27 33 39 45 51 57 63 69 75 81 87

29 35 41 47 53 59 65 71 77 83 89

0.05

0.04

0.03

0.02

0.01

0.00

0.05

0.04

0.03

0.02

0.01

0.00

27 33 39 45 51 57 63 69 75 81 87

29 35 41 47 53 59 65 71 77 83 89

0.05

0.04

0.03

0.02

0.01

0.00

0.05

0.04

0.03

0.02

0.01

0.00

27 33 39 45 51 57

µ = 50.2

µ = 53.4

µ = 62.3

µ = 61.0

µ = 66.4

µ = 64.8

1960

1960

1980

1980

2000

2000

63 69 75 81 87

Figure 3.1 Vanishing twin peaks in life expectancy at birth


Table 3.1 Increases in life expectancy at birth slowed down dramatically in the 1990s

1960 1970 1980 1990 2000

Mean 53.4 57.4 61.0 64.0 64.8

Coefficient of variation 0.233 0.203 0.183 0.173 0.194

Theil-T 0.027 0.021 0.017 0.016 0.020

Theil-L 0.028 0.022 0.018 0.017 0.021

Source: Schady (2005).Note: Theil-L and Theil-T are two inequality measures that belong to the general entropy class, with parameters 0 and1, respectively (unweighted).

the average in OECD countries. For exam-ple, for all countries with average per capitaGDP below the $2 a day threshold, the childmortality rate of the richest 20 percent ofthe population is more than 10 times theOECD average of six years.11 While this isalso certainly the case with many otherindicators, it is difficult to make statementsabout how large differences between coun-tries are in comparison with inequalitieswithin countries. Unlike the incomeinequality literature, there are no acceptedpractices for decomposing inequalities inhealth into within-country and between-country components.12

Consider, however, a simple thoughtexperiment. In the 45 developing countriesfor which a Demographic Health Surveywas available in 2000, 4.9 million infantdeaths could be prevented by bringing theirinfant mortality levels to the OECD aver-age. But if one eliminated the infant mortal-ity gap between the rich and the poorwithin each of the same countries by lower-ing the infant mortality rate for everyone tothe level of the top decile, 3.1 million infantdeaths could be prevented.13 While theaverage infant mortality rate for the rich inthese poor countries is almost five timeslarger than the OECD average, it seemsthat eliminating within-country differencesbetween the rich and the poor (by improv-ing the health of the poor), at least in thisparticular case, would get us about two-thirds of the way to the number of total pre-ventable deaths (by moving everyone to theOECD average).

So, while large differences in health out-comes remain between countries and withinthem, it is not possible to make definitivestatements about the relative weight ofthese components in global health inequali-ties. One can say, however, that there is noclear presumption that inequalities betweencountries dwarf those within them. Thisfinding, as we will see later in this chapter,stands in sharp contrast to that in incomesbut is congruent with that in education.

While technical change in private andpublic health knowledge may be moreimportant to account for the overall betterhealth,14 income may be important in thepoorest countries, through its impact on

the adoption of even inexpensive tech-niques, adequate nutrition, and water andsanitation infrastructure. Life expectancyincreases steeply with income among thepoorest countries (figure 3.2).15

But differences in income growth ex-plain less than a sixth of the variation inimprovements in life expectancy at birth.More important determinants are cleanwater, health systems, demand for ade-quately operated and equipped health sys-tems, and basic sanitary knowledge, the lat-ter two having much to do with education,particularly women’s education.16

While life expectancy at birth continuedto increase, and the infant and child mortal-ity rates declined, the last decade of thetwentieth century has seen a divergencebetween rich and poor countries.17 The dif-ficulties faced by Europe and Central Asiacountries during transition, and the spreadof HIV/AIDS and civil conflicts, were majorfactors in this, but they are not solelyresponsible.18 Cornia and Menchini (2005)cite changes in health spending, publichealth programs, and the structure and sta-bility of households as possible reasons forthe slowdown in health progress in devel-oping countries.


United States

JapanSpain

China

RussiaBrazil

Botswana

Rep. of KoreaGermany

Argentina

Gabon

Mexico

South AfricaNamibia

IndiaPakistan

0 30,00020,00010,000

GDP per capita, 2000, current PPP $

45

35

75

65

55

85

Life expectancy, 2000

Figure 3.2 Life expectancy is highly correlated with income, particularly in poor countries

Source: Deaton (2004).Note: The curve is nonparametrically fitted, weighted by population. The figure plots country life expectancy (usingcircles whose size is proportional to population) against GDP per capita in purchasing power parity (PPP) dollars at theturn of the twenty-first century.

Global inequalities in educationThe unweighted distribution of adultschool attainment was clearly left-skewedin 1960 (figure 3.3). This reflects the factthat many countries, particularly in Africaand Asia, had mean years of schooling closeto zero. The weighted distribution, like thatfor life expectancy at birth, was bimodal,with one peak around two years of meanschooling and a smaller peak around eightyears.

Many of these countries saw large in-creases in enrollments in the 1960s, 1970s,and 1980s. As these younger cohorts aged,the mean years of schooling in their coun-tries increased, and the skewness in theunweighted international distribution ofschooling disappeared. Note that the bi-modal distribution of schooling across per-sons (weighted by population) persisteduntil the 1990s and then gave way to a uni-modal distribution only by 2000.

By any measure the international distri-bution of years of schooling has undergonedramatic changes between 1960 and 2000.As mean levels have risen, inequality hasfallen, decade after decade (figure 3.4). Themean years of educational attainment forthe world almost doubled from 3.4 to 6.3(table 3.2). Sub-Saharan Africa, the MiddleEast and North Africa, and South Asiastarted with high inequalities (not shownhere) and reduced them over time—theMiddle East and North Africa region wasparticularly successful. Latin America andthe Caribbean and East Asia also had someinequalities, which they essentially elimi-nated for their youngest cohorts. Despite theprogress, mean levels of educational attain-


0 1 3 5 7 9 11 13

0.20

0.15

0.10

0.05

0.00

Estimated density, unweighted

0 1 3 5 7 9 11 13

0.20

0.15

0.10

0.05

0.00

Estimated density, weighted

0 1 3 5 7 9 11 13

0.20

0.15

0.10

0.05

0.00

0 1 3 5 7 9 11 13

0.20

0.15

0.10

0.05

0.00

0 1 3 5 7 9 11 13

0.20

0.15

0.10

0.05

0.00

0 1 3 5

µ = 3.3

µ = 4.4

µ = 4.6

µ = 5.6

µ = 6.3

µ = 6.6

1960

1960

1980

1980

2000

2000

7 9 11 13

0.20

0.15

0.10

0.05

0.00

Figure 3.3 The distribution of years of schooling improved greatly in the second half of the twentieth century


1.0

0.8

0.6

0.4

0.2

0.0

9

8

7

6

5

4

Inequality in years of schooling Mean years of schooling

Inequality in years of schooling

Mean years of schooling

1975–91970–41965–91960–41955–9Year of birth

1950–41945–91940–41935–9

Figure 3.4 Mean years of schooling increased while inequality declined across birth cohorts

Source: Araujo, Ferreira, and Schady (2004).Note: Inequality in years of schooling is measured using GE (0.5), that is, the general entropy class inequality measurewith an inequality aversion parameter of 0.5.

ment in Sub-Saharan Africa and South Asiaremain low even for the youngest cohorts.

While significant disparities remain ineducational attainment across countriesdespite evidence of significant catch-up bypoorer countries in the past half century,there is also large variation within countries(chapter 2). In fact, less than 20 percent ofthe inequality in educational attainmentbetween adults born between 1935 and1979 is attributable to that between coun-tries, a share that has been steadily decliningover time. While both inequality within andbetween countries is declining, the rate ofconvergence in country means has beenfaster.

The story remains the same when decom-posing inequality in educational attainmentinto inequalities between men and women.Roughly a quarter of global inequality in edu-cational attainment is attributable to differ-ences between men and women, but this gapis again declining over time, from 31 percentin the oldest cohort in our sample, to 16 per-cent in the youngest. But there are large dif-ferences in this convergence by region (figure3.5). While Latin America and the Caribbean,East Asia, and Europe and Central Asia seemto have reached gender parity in education,along with other developed countries, theprogress in South Asia, Sub-Saharan Africa,and the Middle East and North Africa hasbeen slower. Women still lag far behind menin educational attainment.

It should not be assumed that highattainment necessarily implies high achieve-ment, and vice versa. An analysis of the rela-tionship between attainment (measured bythe percentage of 25 to 34 year olds withupper-secondary education) and achieve-ment (measured by reading proficiency of15 year olds) in 27 OECD countries (plusBrazil) shows a rank correlation coefficientbetween these two variables of 0.57. It isclear that the rankings of countries accord-ing to these two indicators are not the same.The Republic of Korea and Japan (at the topof the OECD distribution) and Mexico, Por-tugal, and Turkey (at the bottom) have simi-lar ranks for both attainment and achieve-ment. But the Czech Republic, Norway, andthe United States do worse in achievementthan attainment. And achievement rankings

of Australia, Finland, and Ireland are muchhigher than their rankings in attainment.

Achievement differences between devel-oping countries and OECD countries re-main strikingly large. Using internationallycomparable assessments of reading, mathe-matics, and science, Pritchett (2004b)shows that developing countries do not justconstitute the lower tail of the learning dis-tribution, but that most actually do far worsethan the poorest performing OECD coun-tries. For example, children in Argentina,Mexico, and Chile perform about two(OECD) standard deviations below chil-dren in Greece—one of the poorest per-forming countries in the OECD. In readingcompetence (based on PISA 2001), the


Table 3.2 Mean years of schooling increased continuously while inequality declined

1960 1970 1980 1990 2000

Mean 3.38 3.82 4.67 5.55 6.30

Coefficient of variation 0.739 0.705 0.612 0.518 0.461

Theil-T 0.281 0.259 0.195 0.143 0.115

Theil-L 0.392 0.365 0.250 0.179 0.144

Source: Schady (2005).Note: Theil-L and Theil-T are two inequality measures that belong to the general entropy class, with parameters 0 and1, respectively (unweighted).

1975–91935–9 1945–9 1955–9 1965–91960–41940–4 1970–41950–4Year of birth

4

3

2

1

Male to female schooling ratio

OECDEast AsiaEurope and Central AsiaSouth AsiaLatin America and the CaribbeanMiddle East and North AfricaSub-Saharan Africa

Figure 3.5 Gender disparities in years of schooling declinedbut remained significant in some regions

Source: Araujo, Ferreira, and Schady (2004).

average Indonesian student performed atthe level of a French student at the seventhpercentile. Considering children who havenever attended school, those who enrolledbut dropped out, and those who completedgrade nine but whose test scores remainmore than one standard deviation belowthe OECD mean in mathematics, Pritchettfinds that 96 percent of 15 to 19 year olds inMorocco lack achievement in “adequatelearning.”19

Global inequalities in income and expenditureThe answers to basic questions—such aswhether income inequality has been increas-ing or decreasing—depends, among otherthings, on which concept of inequality isunder the microscope: intercountry inequal-ity (in the distribution of unweighted coun-try means), international inequality (in thedistribution of country means weighted bytheir population size), or global inequality(in the distribution of individual incomes).

We start the discussion by presenting themedian and mean incomes of selectedcountries by region for a range of years

between 1997 and 2002, as well as the dis-persion of those incomes within each coun-try (figure 3.6). Large differences acrosscountries and across people within coun-tries are striking. For example, an indi-vidual in the tenth percentile in the U.S.distribution enjoys a level of income higherthan an individual earning the meanincome in Brazil or Argentina.20 While aChinese individual living in a rural area hasa mean income similar to an average Cam-bodian, an urban Chinese enjoys a similarincome to an average Brazilian.21 A SouthAfrican at the bottom of the income distri-bution in her country earns as much as theaverage individual in Mali while a SouthAfrican at the ninetieth percentile of thatincome distribution enjoys a standard ofliving (in income) comparable to that of amedian Irish individual.

The difference in the evolution of inter-country (unweighted) and international(weighted) inequality between 1950 and2000—borrowing from Milanovic (2005),who calls this the “mother of all inequalitydisputes”—could hardly be more dramatic(figure 3.7). When countries are the unit of


Sub-Saharan Africa

East Asia

Latin America andthe Caribbean

Middle East andNorth Africa

Europe andCentral Asia

OECD countries

South Asia

MaliEthiopia

South AfricaCambodia

India, ruralPakistan

India, urbanChina, rural

IndonesiaChina, urban

NicaraguaArgentina

BrazilYemen

MoroccoIsrael

RussiaAlbaniaPolandIreland

DenmarkUnited States

25002000150010005000 225017501250750250

Income per capita (1993 PPP $)

Figure 3.6 Incomes range broadly across countries and individuals

Source: Authors’ calculations.Note: Years range from 1997 to 2002 as measured by adjusted (1993 PPP $) monthly per capita income (blue box) or consumption (orangebox). The lowest point of each line represents the income level at the tenth percentile, followed by that at the median, the mean (the twoedges of each box), and the ninetieth percentile (top of each line).

observation, (intercountry) inequality hasbeen undeniably increasing, especially sincethe 1980s. But international inequality hasbeen steadily declining, thanks mostly tothe income growth in some populous coun-tries, mainly China and India. Note thatintercountry inequality and internationalinequality without China and India trackeach other quite closely from 1980 onward,coinciding with the period of rapid growthin these two countries, the slower averagegrowth in other developing countries, andthe declines in measured output in EasternEurope and former Soviet Union countries.

If Luxembourg and Nicaragua, at oppo-site ends of the world income distribution,grew at the same annual rate of 2 percentper capita a year for the next 25 years, theper capita yearly incomes in Luxembourgwould increase from $17,228 (PPP-adjusted)to $28,264, an increase of more than$10,000 dollars. That of Nicaragua, by con-trast, would increase by a mere $375, from$573 to $940, during the same period.Atkinson and Brandolini (2004) note that“with annual per capita growth rates of 5percent in China and 2 percent in theUnited States, the absolute income gapbetween the two countries would widen fora further 41 years before starting to narrow,to finally disappear in 72 years.”

The evaluative judgments drawn aboutthe distributional changes associated withglobalization may depend crucially onwhether one thinks about inequality in

absolute terms or relative terms. There is noeconomic theory that tells us that inequalityis relative, not absolute. Again, as with inter-country and international inequality, it isnot that one concept is right and the otherone wrong. Nor are they two ways of meas-uring the same thing. Instead, they are twodifferent concepts. The revealed preferencesfor one concept over another reflect implicitvalue judgments about what constitutes afair division of the gains from growth.Those judgments need to be brought intothe open and critically scrutinized beforeone can take a well-considered position inthis debate.

An examination of international inequal-ity using absolute rather than relative meas-ures of inequality reveals a steady increaseover the long run, as well as in recentdecades—this latter finding contrasts withrelative international inequality trends. Atkin-son and Brandolini (2004) find that absoluteindexes of inequality, such as the AbsoluteGini and the Kolm Index22 (with variousparameters of inequality aversion), have beenincreasing steadily since 1970 (figure 3.8).23


1950 1970 19901960 19801955 1975 19951965 1985 2000

0.4

0.5

0.6Gini index

International inequality(weighted)

Intercountry inequality(unweighted)

International inequalitywithout China and India

Figure 3.7 Since 1950, intercountry inequalityincreased, while international inequality declined

Source: Milanovic (2005).

2000199519901985198019751970

250

Indexes, 1970 = 100

225

200

175

150

125

100

75

Kolm index (0.3)Kolm index (3.0)Absolute Gini indexGini indexTheil indexMean logarithmicdeviation

Absolutemeasures

Relativemeasures

Figure 3.8 Unlike relative inequality, absoluteinequality has been steadily increasing

Source: Atkinson and Brandolini (2004).

What happened to global inequality inthe past 20 years or so has been the subjectof fierce debate in the context of globaliza-tion and is perhaps the hardest question toanswer. Some authors24 claim that globalinequality increased slightly, while others25

argue that they have declined.Examining global inequality requires

knowledge of the distribution of inequal-ity within each country. Household sur-veys that collect such data are a relativelynew phenomenon, having become morecommon since the 1980s even in develop-ing countries. So, if we want to knowabout the distribution of income foreveryone in the world, we are confined toa much shorter time period. We haveselected three waves, similar to those usedby Milanovic (2005): 1986–1990, 1991–96,and 1997–current.

Global inequality (measured by themean log deviation) did not change signif-icantly over this period, although there is aslight decrease between 1993 and 2000(figure 3.9). The mean log deviation forthe world would have increased withoutChina and India, consistent with the con-sensus in the literature that internationalinequality declined in this period thankslargely to these two countries. But if globalinequality stayed roughly the same whileinternational inequality declined, inequal-ity within countries must have increasedby approximately the same amount—asubject that we discuss below.

Most of the world’s income inequalitycan be explained by the differences in coun-try means—that is, by international (orbetween-country) inequality. Our estimatesshow that the share of global inequality,which can be attributed to inequalitybetween countries, declined steadily from78 percent around 1988 to 74 percentaround 1993 and to 67 percent by around2000. With global inequality staying roughlythe same during this period, within-groupinequality increased at a somewhat steadypace (figure 3.9). These results are consis-tent with the evidence (in chapter 2) ofincreasing inequality within countries inmany parts of the world, includingBangladesh, China, the United Kingdom,and the United States.

The between-country share of globalinequality is also consistent with Milanovic(2005), who puts this figure at about 71 per-cent in 1998. It is possible that theMilanovic figures overestimate between-country inequality because he assigns allhouseholds in a decile the same incomeinstead of estimating a Lorenz curve (forpercentiles). Our results use slightlyimproved data from Milanovic in threeaspects. First, for many countries, we calcu-late our welfare measures using raw data atthe household level, while Milanovic (andmany others) use grouped data. Second, weincorporate more recent data for the cur-rent period, possibly providing an improve-ment in data quality, especially for EasternEuropean countries. Third, for the coun-tries with grouped data, we estimate Lorenzcurves instead of assigning everyone in thegroup with the same income.26 That mostof the global inequality in incomes isexplained by between-country inequalityseems to be a robust finding in the litera-ture, in stark contrast with the picture inhealth and education.

Over a much longer period (1820–1992)Bourguignon and Morrisson (2002) esti-mate that global inequality has beensteadily increasing, because of a rapidincrease in international inequality untilWorld War II, and then to smaller increasesin both within-country and internationalinequality between 1970 and 1992 (figure3.10).27 They also argue that international


Mean log deviation Percent

0.15

0.40

0.65

0.90

1.15

0

25

50

75

100

Between-country share (right axis)

Between-countryinequality (left axis) Within-country

inequality (left axis)

Global inequality(left axis)

2000199819961994199219901988

0.27

0.55

67

0.82

0.22

0.64

74

0.87

0.19

0.65

78

0.84

Figure 3.9 The inequality decline between countrieswas neutralized by increases within countries

Source: Authors’ calculations.

inequality was essentially negligible at theturn of the nineteenth century (accountingfor roughly 12 percent of global inequality),but that it increased very rapidly untilWorld War II, and then continued toincrease, but at a much slower pace. Withincountry inequality, however, reached itspeak around 1910 and declined dramati-cally between the two world wars (mainlybecause of equalizing forces in the now-developed countries), and started creepingback up only since the 1970s. The combinedeffect of these changes is an increase in theshare of international inequality fromroughly 10 percent in 1820 to more than 60percent by 1992.

In summary, while the world got richer,income inequality—relative and absolute,international and global—increased tremen-dously over a long period of time(1820–1992). But the story is less clear-cutfor a more recent time frame. In thepost–World War II era, intercountry inequal-ity (unweighted) has continued to increasewhile international inequality (weighted forpopulation) declined. International in-equality declined in the final decades of thetwentieth century, because the inequality-reducing effects of income growth inChina and South Asia more than offset theinequality-boosting effects of continuedsteady income growth in the now-developedcountries and the declining incomes in Sub-Saharan Africa.

Pritchett (1997), examining the periodbetween 1870 and 1990, argues that whilethere was convergence of incomes fortoday’s developed countries (what Maddi-son 1995 calls the “advanced capitalist”countries), the growth rates between devel-oped and developing economies show con-siderable divergence. He provides evidencethat “the growth rates of developed coun-tries are bunched in a narrow group, whilethose of less developed countries are allover with some in explosive growth andothers in implosive decline.”28

Further evidence of convergence amongrich countries and divergence between richand poor countries comes from Schultz(1998), who estimates that internationalinequality accounted for about two-thirdsof total inequality (measured by log vari-

ance) between 1960 and 1990; however,there also were large differences by region.Inequality between the countries of OECD(and the rest of Europe including Turkey)decreased by 50 percent during thisperiod, at the end accounting for only one-third of total inequality. During the sameperiod, international inequality in Sub-Saharan Africa nearly doubled, causing itsshare in total inequality to increase from20 percent to 36 percent. In both Sub-Saharan Africa and Latin America and theCaribbean, overall inequality levels remainhigh, while high-income countries showsigns of convergence.

One can also examine inequality trendsby focusing on the mobility of countriesrather than by taking an anonymousapproach to inequality comparisons. Poorcountries’ mobility from the bottom hasbeen limited in the past 25 years. With theexception of China, the six countries thatoccupied the bottom decile (population-weighted) in 1980—all in Sub-SaharanAfrica—had no growth worth noting.29

While there is significant upward mobil-ity between 1980 and 2002—the 97.08 per-cent entry in the first row of table 3.3 isChina—there is also troubling stagnationand downward mobility. Note that approxi-mately 8 percent of each of the second andthird income ranges fell into the bottomrange over these two decades. “It is clearthat no Pareto improvement has taken placein the world between 1980 and 2002, whichleaves room for different value judgments


Mean log deviation1

0.8

0.6

0.4

0.2

019921980197019601950192919101890187018501820

0.33

0.50

0.83

0.33

0.36

0.69

0.05

0.37

0.42

Global inequality

Within-country inequality

Between-country inequality

Figure 3.10 Inequality between countries becamemuch more important over the long run

Source: Authors’ manipulation of data from Bourguignon and Mor-risson (2002).

about the evolution of world welfare,inequality, and relative poverty.”30 Milanovic(2005) also point out the “downward mobil-ity” of many countries in the past 40 years orso. Those who do not share the view thatinequality between countries fell in the past20 to 25 years—that is, those who take the“unweighted” view of the world—may havesuch mobility concerns in mind.

Absolute poverty rates have declined inthe past 20 years or so, and a variety ofstudies have confirmed this trend (figure3.11).31 Overall, while there are roughly400 million fewer people who live on lessthan a $1 a day in 2001 than there were in1981, the number of poor people in Sub-Saharan Africa almost doubled, fromapproximately 160 million to 313 million.

While some populous countries, almostexclusively in Asia, such as Bangladesh,China, India, and Pakistan, made signifi-cant headway against extreme poverty,almost all increases in extreme poverty—especially in countries with high initialheadcount rates—took place in Sub-Saha-ran Africa.32 Among the larger countrieswith rising headcount rates are Nigeria,South Africa, and Tanzania.

If the poverty trends discussed here con-tinue, the Millennium Development Goal ofhalving the proportion of people living onless than $1 a day will be met. But only Eastand South Asia will reach this goal. We can-not be satisfied if this were to happen. Otherthings equal, we would prefer to see thepoverty rate falling at the same pace in allcountries. Currently, hundreds of millionsof people in numerous developing countrieslack the opportunity to avoid hunger, poorhealth, and low access to vital services, suchas education and clean water.33

Global inequalities in powerOne of the main arguments in the conclud-ing chapter of this report is that the rulesand processes in global markets can beunfair to developing countries. A country’spower in decision making in multilateralbanks is usually correlated with its eco-nomic strength. Even when each countryhas equal representation in an internationalbody, such as the United Nations system orthe World Trade Organization (WTO), pow-erful forces can chisel away at developing-country interests (through separate bilat-eral agreements, for example). And thecapacity of developing countries to makeinformed decisions can be limited.

Poor countries lack the financial and humancapital resources that would allow them to beequal participants in the international bodiesin which decisions are taken that affect themand, beyond that, in setting the rules underwhich the international system operates.34

In the International Bank for Recon-struction and Development (IBRD)—themarket-lending arm of the World Bank—acountry’s voting power depends on the per-centage of IBRD shares it holds. The largestshareholders are the United States with 16.4


Table 3.3 Mobility matrix in absolute country per capita incomes, 1980 to 2002

Income in 2002

Income in 1980 <710 711–1,100 1,101–2,890 2,891–10,000 10,001>

<710 1.28% 1.64% 0.00% 97.08% 0.00%

711–1,100 8.23% 3.89% 87.88% 0.00% 0.00%

1,101–2,890 8.09% 0.56% 59.08% 32.28% 0.00%

2,891–10,000 0.00% 0.00% 0.98% 90.84% 8.17%

10,001> 0.00% 0.00% 0.00% 3.99% 96.01%

Source: Bourguignon, Levin, and Rosenblatt (2004a).Note: Incomes are per capita (constant PPP dollars).

Headcount ($1 per day) in 2001

ZWE

ZAM

TZA

ZAF

SLE

SEN

R W A

NGA

NER

NAM

MOZ

MAU

MLI

MWI

MDG

LSO

KEN

GHA

GAM ETHCIV

CAF

CMR

BDI

BF A

BWA

VNMTHA

PHL

MON

MYSLAO

CHN

KHM

UZB

UKR TURRUS

VEN

NIC

MEX

GTM

COL BRAARG

YEM

MARJOR

EGY LKA

P AK

IND

BGD

AFREAPECALACMENASAR

80

60

40

20

0

40200 60 80

Headcount ($1 per day) in 1981

Figure 3.11 Absolute poverty declined globally, but not in every region

Source: PovcalNet (http://iresearch.worldbank.org/PovcalNet/jsp/index.jsp).

percent of the vote, Japan with 7.9 percent,Germany with 4.5 percent, France and theUnited Kingdom with 4.3 percent. Each has arepresentative on the Board of Directors. Bycontrast, all Sub-Saharan countries togetherhave two representatives and 5.2 percent ofthe vote. China and India both have 2.8percent of the vote.35 Country influence insetting the agenda for the institutions is notlimited to board membership. A 1998 study byFilmer and others (1998) shows that roughlytwo-thirds of the senior management–levelpositions at the World Bank are occupied bycitizens of Part I (mainly OECD) countries,although these countries account for lessthan one-fifth of the global population anda smaller share of the number of membercountries.

At the WTO, each member country hasone vote. Moreover, because decisions areby consensus, each country effectively hasveto power. So the WTO is, at least onpaper, perhaps the most democratic ofinternational organizations. In practice, theability of countries to influence the agendaand decisions depends crucially on theircapacity to be present, to follow negotia-tions, to be informed, and to understandfully the impact of the complex issues athand. A rough indicator of a country’scapacity is the size of its representation inGeneva. A study by Blackhurst, Lyakurwa,and Oyejide (2000) found that only 8 of the38 Sub-Saharan countries had close to five(the WTO average) resident delegates listedin the WTO directory. Worse, 19 of the 38countries—half of the Sub-Saharan WTOmembership—had no delegate resident inGeneva. Only Nigeria had a delegation thatdeals solely with the WTO.36

Even when country representation in theinternational arena is considered adequate,it is debatable whether the representativesof some countries are fully accountable totheir citizens. There are considerable differ-ences among countries in the extent towhich their political and legal institutionsprovide citizens with fair, transparent, andinclusive environments to enhance andleverage their assets. While there are numer-ous problems with trying to measure suchthings, Kaufmann, Kraay, and Mastruzzi(2004), in the most comprehensive compar-

ative assessment to date, integrate data col-lected by 25 separate sources constructed by18 (commercial and advocacy) organiza-tions. The authors used the data to providea common empirical basis to assess the rela-tive differences among countries of thequality of their “governance.”

Figure 3.12 summarizes information on“voice and accountability,” which refersbroadly to the extent to which citizens havefreedom of expression, a free press, andopen elections, using standardized meas-ures for selected countries (the same ones asin figure 3.6). The upper bar for each coun-try represents the country’s percentile rankin the “voice and accountability” distribu-tion with the intersecting black rule linerepresenting the confidence interval. Thelower bar is the average percentile score forthe income category to which the countrybelongs.37 The top of the “voice” rankings is


Denmark

Ireland

United States

Poland

South Africa

Israel

India

Brazil

Mail

Argentina

Nicaragua

Albania

Morocco

Indonesia

Russia

Cambodia

Yemen

Pakistan

Ethiopia

Voice andaccountabilityIncome category

10060200 40 80

Figure 3.12 There is no one-to-one relationshipbetween voice and income

Source: Kaufmann, Kraay, and Mastruzzi (2004).Note: “Voice and accountability” refers broadly to the extent towhich citizens have freedom of expression, a free press, and openelections based on a statistical compilation of responses on thequality of governance given by a large number of enterprise, citi-zen, and expert survey respondents in industrial and developingcountries, as reported by a number of survey institutes, think tanks,nongovernmental organizations, and international organizations.Countries’ relative positions on these indicators are subject tomargins of error that are clearly indicated. Consequently, precisecountry rankings should not be inferred from these data.

filled with wealthy countries, such as Den-mark, the United States, Ireland, and Israel.The voice ranking of such countries asSouth Africa, Poland, and especially Maliand India exceed their ranks in incomes.The opposite is true for China, Ethiopia,Pakistan, and the Russian Federation. Cam-bodia and the Republic of Yemen are bothquite poor and rank low in freedom ofexpression. It is clear that there is no one-to-one relationship between citizens’ voiceand average income at the country level.

A glimpse of the futureDespite improvements over time, inequali-ties among countries in various dimensionsremain unacceptably high. Each year 10.5million child deaths are preventable in thesense that these children would not havedied if they had been born in rich coun-tries.38 The mean educational attainmentlevel for adults born in 1975–79 in Sub-Saharan Africa remains at 5.4 years, com-pared with 10.1 years in Latin America andCaribbean and 13.4 years in developedcountries. Developing countries also facemassive challenges in influencing the globalrules and processes that determine out-comes, which matter greatly to the well-being of their citizens.

International inequalities in educationalattainment have been steadily declining.This is also true in health—one’s countryof birth 50 years ago was much more perti-nent to survival than it is today. In thissense, opportunities across countries areequalizing. But improvements in lifeexpectancy at birth have reversed since theearly 1990s, because of the devastatingeffects of HIV/AIDS and the difficult cir-cumstances facing citizens of some transi-tion economies. The world distribution ofincomes, by contrast, was becoming secu-larly more unequal from the early nine-teenth century until about the end of WorldWar II. Since the war, international in-equality between countries has decreasedimmensely, because of the fast growth inChina and India in more recent times, andglobal inequality has leveled off. BecauseChina and India are only two countries,intercountry inequality in incomes has con-tinued to increase.

What explains the convergence in healthand education and the lack of it in incomes?Deaton (2004) points out that, while gainsin income were undoubtedly important forimproving nutrition and funding betterwater and sanitation schemes, some coun-tries made progress in reducing child mor-tality even in the absence of economicgrowth. These improvements came fromthe globalization of knowledge, facilitatedby local political, economic, and educationconditions. A possible explanation for thedisconnect between the convergence ineducation and the divergence in incomes isthat education is not translating intohuman capital and that the rise in perworker schooling explains only a small partof the growth in output per worker.39

We have seen that the story of incomeinequality in the world has been a story offalling international inequalities and ris-ing within-country inequality. For global in-equality, these two effects are offsetting, andthe conclusion drawn depends on knowingwhich effect dominated. The decline in inter-national inequality is largely due to fastincome growth in China and South Asia.40

But as China and South Asia catch up to theworld average, their equalizing effect willdiminish. And if they continue to develop atsimilar rates to that in the past two decades,the effect of their growth will increase inter-national inequality.41 Without the offsettingeffect of declining international inequality,global inequality would also be on the riseagain unless inequality within countriesstarts to decline and Sub-Saharan Africaneconomies begin to experience healthygrowth. This suggests that the future ofworld income inequality will increasinglybe a function of economic growth in Africa(and some other low-income countriesunder stress), especially if the populationgrowth rates in Africa remain above theworld average. That both populationgrowth and economic growth in Africa havebeen stunted by the AIDS tragedy is doublydisturbing.

On whether today’s poor countries withstagnant economies will take off, someresearchers are optimistic. Lucas (2003)suggests that the countries that have not yetjoined the industrial revolution (which he


attributes to socialist planning, lawlessness,and corruption) will become the miracleeconomies of the future. He reckons thatthe growth rates in these catch-up countriesmay be quite high and that they will also gothrough a similar demographic transitionexperienced by today’s developed countries.The world population will stop rising andworld production growth will stabilize untilall countries, economically, start resemblingcountries like the United States, thanks tofree trade and the diffusion of technology.

Pritchett—who calls this idea “advantageto backwardness”42—remains more cau-tious. Conceding that such rapid gains inproductivity are a possibility, he argues that“the cases in which backward countries, andespecially the most backward of countries,actually gain significantly on the leader arehistorically rare.”43 He observes that thereare also forces for “implosive” declines inthese countries, suggesting that backward-ness may also carry “severe disadvantages.”

On health in Sub-Saharan Africa, theUN Population Division projects that lifeexpectancy at birth in Africa will declineover the next 5 to 10 years and then startclimbing again, reaching 65 years by about2050.44 These projections assume thatHIV/AIDS prevalence rates in Africa willpeak sometime before 2010 and thendecline over the next decades. But the JointUnited Nations Programme on HIV/AIDSestimated that 43 percent of pregnantfemales in 2000 were HIV positive inBotswana and 19 percent in South Africa.45

Thus, millions of babies are being infectedat birth, which is mostly preventable withproper interventions. Life expectancy inAfrica would not improve much, and cer-tainly not soon, if these assumed improve-ments in HIV/AIDS prevalence rates do notmaterialize.

Because South Asia has almost caught upto the world average in life expectancy, Sub-Saharan Africa will be the only region signif-icantly affecting health inequalities betweencountries, barring a major catastrophe else-where.46 So, improvements in life expectancyin Sub-Saharan Africa are the key to futuredeclines in international health inequalities.Chapter 2 documented within-countryinequalities in health opportunities for chil-dren born to poor or rich parents, educatedor uneducated mothers, in rural or urbanareas, and so on. Steep gradients in healthopportunities and outcomes exist alongthese dimensions in many countries. A confi-dent assessment of past and future trends inhealth inequality awaits future research.

If the trends that brought about thecatching up of many poor countries outsideof Africa continue in health, education, andincomes, the biggest challenges will remainin Africa and some poor countries in otherregions. Growth with equity needs to berevived in stagnating economies around theworld, and the AIDS tragedy (along withthe folly of preventable diseases) needs tobe addressed urgently, especially in Sub-Saharan Africa. These remain the biggestglobal challenges in development today.


These two initiatives have much in com-mon. They were both conceived as directand conscious efforts to break with theelite-dominated and clientelistic politics oflocal government by promoting redistribu-tive policies through broad popular partici-pation. Thus, they both shifted the politicalopportunity structure and involved actionto strengthen the agency of subordinategroups.2 Both have, in effect, comple-mented representative forms of democracywith participatory forms of democracy byopening institutions to the direct engage-ment of civil society. And both havestrengthened public authority and publicaction by increasing both the depth andscope of democratic decision making.

Deepening democracyThe evidence shows that these initiativeshave deepened democracy, expanding therange of social actors participating in thepolitical arena. In Porto Alegre, an esti-mated 100,000 adults have participated atsome point in the budgeting assemblies.Other cities that have adopted some formof the process have also experienced activeparticipation, including municipalitieswithout established civil societies. InKerala, nearly one in four householdsattended village assemblies in the first twoyears of the campaign, and despite rou-tinization of the process in subsequentyears, these assemblies continue to drawlarge numbers. Hundreds of thousands ofcitizens have undergone training in plan-ning and budgeting, and the committeesthat actually design and budget specificprojects have been composed primarily ofcivil society actors.

A redesign of institutional incentives andnew mobilizational efforts saw womenaccount for 40 percent of the participants invillage assemblies (a level otherwise unheardof in India) and the participation rate of dal-its (scheduled castes) has exceeded their

Local government is a critical domainfor the exercise of democratic rightsand for making effective public

choices. But several factors have conspiredagainst good governance, democracy, andequity at the local level in much of thedeveloping world. The social and economicpower of local elites has often translatedinto disproportionate influence over thepolitical process, and top-down, insulated,and nontransparent decision-makingstructures have made it difficult for ordi-nary citizens to have voice.1 Democraticdeepening in the developing world oftenbegins with the democratization of localgovernment, and that is precisely what twoparticipatory governance initiatives—in theIndian state of Kerala and in a variety ofmunicipalities in Brazil—have tried to do.

In 1996 the state government of Keralalaunched what is widely viewed to be themost ambitious initiative for democraticdecentralization in India: the People’s Cam-paign of Decentralized Planning. The gov-ernment not only devolved significantresources and authority to Kerala’s 1,214panchayats (village councils) and munici-palities, but it also promoted direct citizenparticipation by mandating village assem-blies and citizen committees to plan andbudget local development expenditures.

In Brazil, the city of Porto Alegrelaunched a participatory budgeting initia-tive in 1990 that has since been copied in atleast 400 municipalities throughout thecountry. The process begins with neighbor-hood assemblies in which citizens deliber-ate and set budgeting priorities, and endswith a citywide budget formulated by dele-gates directly elected by neighborhoodassemblies. The success in Porto Alegre hasseen its steady diffusion, with at least 100municipalities, including São Paulo, imple-menting variations of participatory budget-ing in 1996–2000, and some estimated 250municipalities in 2000–04.

representation in the population.3 Moreover,both these initiatives have created a newcadre of grassroots politicians who either didnot exist before (delegates in Brazil) or whopreviously had no powers (the 14,000elected panchayat councilors in Kerala). Thelocal public sphere—the sine qua non of anyvibrant democracy—has become moreextensive, more inclusive, and more mean-ingful.

Extending democracyThese initiatives have been marked by theextension of democracy, specifically publicdecision making in arenas of authority pre-viously dominated by private and stateelites. Municipal budgets in Brazil have longbeen the preserve of oligarchic parties andnarrow sectoral interests. Panchayats inKerala have long been little more than pas-sive recipients of top-down projectsdesigned and delivered by state bureaucra-cies. In both cases, citizens now have a voicein determining how public resources areallocated. In the most successful participa-tory budgeting cases, the entire budget ofthe municipality is discussed and approvedby delegates acting on priorities establishedby neighborhood assemblies, with citizensdeliberating on capital and operational seg-ments of the budget.

In some municipalities, direct partici-pation has been extended to thematicareas, such as economic development,public transportation, education, socialservices, and urban planning. In Kerala,panchayats have been given authority forup to 35 percent of the developmentbudget, a fivefold increase in theirresources base. Panchayats have ranked,designed, and implemented hundreds ofprojects a year across all development sec-tors. These have included housing for thepoor, small-scale irrigation, local roadsand infrastructure, agricultural projects,support services in health and education,

Popular participation and equitable transitions at the local level Promoting equity through public action requires changes in the existing configurations of power and influence. Becauseestablished institutions privilege certain interests and marginalize others, making governance institutions moredemocratic and more equity-enhancing calls for reforms that increase the possibilities for effective participation by tradi-tionally marginalized groups.

empowermentf o c u s 2 o n

the economy is otherwise growing, themost marginalized social groups—widows,slum dwellers, sex workers, the very poor—may continue to be excluded. What can bedone? The challenge is greater when theextent of commitment from above andmobilization from below is less than inthese two cases.

As discussed in chapter 2, the most mar-ginal groups are often stuck in more severeforms of an “inequality trap”—a situationcharacterized by dire material circum-stances, rational expectations about limitedmobility opportunities, and internalizedbeliefs regarding the legitimacy andimmutability of their circumstances. Break-ing out of such inequality traps andimproving the terms by which the poor are“recognized” by others starts with buildingboth a “capacity to aspire” and, equallyimportant, a “capacity to engage.”6 Thisincludes being able to envision and enactalternative futures, believing that it is desir-able and possible to move out of poverty,and being able to more meaningfully par-ticipate in forums where decisions affectingtheir welfare are made.

Acquiring a “capacity to aspire” is largelya product of developing more broadlyaccessible and equitable mechanisms forinteraction between the poor and elites,mechanisms that are reciprocally linked toattaining greater voice in associationalinteractions. It thrives in and throughgroup organizing and public dialogue, andthe opportunities these afford for practice,repetition, exploration, conjecture, andrefutation.

An association of sex workers in a Cal-cutta slum, for example, gave its individualmembers a voice, a public presence, and a

and a range of projects specifically tar-geted at women and dalits.

Enhancing equityThese initiatives have generally had equity-enhancing effects. In Porto Alegre, the bestknown and most documented case, there isclear evidence that expenditures on poorerareas of the city increased steadily with theintroduction of participatory budgeting. Inother large cities with participatory budget-ing, such as Belém and Belo Horizonte,expenditures have also targeted the poor. Astatistical analysis of all Brazilian munici-palities in 1997–2000 revealed that partici-patory budgeting cities had significantlyhigher expenditures on sectors that affectthe poor most directly.4

In Kerala, a large survey of key respon-dents found that “disadvantaged” groupswere the prime beneficiaries of targetedschemes. Case studies show that panchayatshave emphasized the need to bring allhouseholds up to a certain basic level ofwell-being, with a heavy emphasis on pro-viding sanitation facilities, decent housing,and safe water to needy families.

In both cases, there is also strong evidencethat the incidence of rent-seeking has fallensharply.5 The greater transparency of thebudgeting process alone has raised the trans-action costs of predation and patronage.

Empowering the most marginal groupsThe Kerala and Porto Alegre cases illustratethe value of improving the accessibility,transparency, and accountability of localgovernment. However, even when such ini-tiatives have been undertaken, and where

capacity to realize their interests and aspira-tions that would have been denied themotherwise.7 Young, female, often illegalimmigrants, contractually bound to worklong hours for ruthless bosses, and facingsure rejection by their families if they man-aged to escape and return home, the sexworkers had virtually no capacity to exer-cise their voices and realize their interests.Persistent efforts to organize the womeninto a union, however, eventually gave themthe confidence and competence to bringabout a change in condom use by clients.

In Indonesia, the Kecamatan Develop-ment Project (KDP), which operates in28,000 villages across the country, seeks toimproving the “terms of recognition” andpolitical agency of marginalized groups. 8

The project allocates grant money at thesubdistrict level, for which several groups ofpoor villagers (two of whom must bewomen) are invited to compete for fundson the basis of the presentation of a formalsubproject proposal. KDP’s procedures,institutions, and norms are largely decen-tralized, they focus on joint public problemsolving, they invite broad public participa-tion and scrutiny, and they occur in a moreor less continuous and institutionalizedway.9

Recent work assessing the impact ofKDP on local decision making finds thatKDP helps marginalized groups cultivateaccess to more constructive spaces and pro-cedures for addressing project and non-project conflicts.10 The beginning stage ofsuch a transformation—in which unequalgroups build the capacity to peacefullyengage one another—is a humble but non-trivial outcome for a development project.

Focus on empowerment 71

IN THE FIRST PART OF THIS REPORT, WE SUMMARIZED some of the evi-dence on inequalities in several dimensions. In addition to affectingwell-being directly, such dimensions as health, education, income,voice, and access to services shape the opportunities people face forfuture progress and achievement. We emphasized the interconnectionsbetween these various dimensions. Not only is there inequality in thedistributions of income, health status, and educational attainment,but—even more important—these indicators tend to be correlated.The rich tend to be both healthier and better educated than others.The poorest of the poor tend to have the lowest attainment in years ofschooling and some of the worst health indicators. These correlationsgenerally also extend to public services, with the poor gaining access toinfrastructure, electricity, water, sanitation, and garbage disposal muchlater than others, if at all.

Because education and wealth help a person gain influence in soci-ety, voice and political power are also generally thought to be correlatedwith economic well-being. The interaction between these mutuallyreinforcing economic, social, and political inequalities perpetuatesthem across generations. Chapter 2 discussed evidence indicating that a10 percent difference in economic status between two families in onegeneration tends to imply, on average, a 4 percent to 7 percent differ-ence in the next generation, depending on the country and measure-ment details. Opportunities clearly are not independent from socialand family background, or from group identity.

Do such disparities matter? Are people concerned with the largeobserved differences in access to education and health, and in eco-nomic opportunities, or merely with the fact that some people havelow absolute levels of income, years of schooling, and access to serv-ices? Should policymakers worry about the unequal opportunities

Why does equity matter?

IIP A R T

that arise from discrimination, unequalaccess to justice or other unfair processes?Should an institution like the World Bank,whose primary objective is to assist its clientcountries in eradicating extreme poverty,care about inequalities—in opportunities,outcomes, and processes—at all?

Opinions on these questions are wide-ranging. Support for equal opportunitieshas long been a theme in domestic policyin the United States, for instance. FranklinD. Roosevelt once said that “We know thatequality of individual ability has neverexisted and never will, but we do insist thatequality of opportunity still must besought.” 1 Some participants in the consul-tations for this report were even offendedthat the question “Does inequality mat-ter?” was asked at all, because they consid-ered its answer to be “Obviously, yes.” Oneparticipant felt that the very question indi-cated that “we are suffering [from] a terri-ble tolerance to horror.”2

The next three chapters in this reportaddress the following question: shouldgood development policy be concernedwith equity? Equity, as discussed in chapter1, is understood here as the pursuit of equalopportunities and the avoidance of severedeprivation. Equity is not the same as equal-ity in incomes, or in health status, or in anyother specific outcome. It is the quest for asituation in which personal effort, prefer-ences, and initiative—rather than familybackground, caste, race, or gender—account for the differences among people’seconomic achievements. A situation inwhich all institutions are color-blind andnonmarket institutions are equally respon-sive to the rich and the poor. In which per-sonal and property rights are enforcedequally for all. And in which all have accessto the public services and the infrastructureto leverage their productivity and theirchances of success in the markets.

The evidence we review here has beenassembled in disciplines ranging from eco-nomics and history to sociology andanthropology. On balance, this evidencesuggests that the pursuit of sustainable,long-term prosperity is inseparable from abroadening of economic opportunities and

political voice to most or all of society. Oneset of reasons for this arises from failures incapital, land, and labor markets. Those fail-ures imply that productive opportunitiesare not necessarily seized by those with thehighest potential returns on their talents orideas, but instead by those with greaterwealth, better connections, or larger landparcels. This would not happen if marketsworked perfectly, as resources would flow tothose with the most productive investmentprojects. But given that markets are not per-fect, scope arises for efficient redistributionschemes.

Chapter 5 documents cases in whichaggregate efficiency could be improved byredistributing wealth or power towardpoorer or marginal groups. Sometimes, theevidence of inefficiency is seen in differ-ences in marginal products of capital acrossfirms. We know that smaller entrepreneurspay interest rates much higher than themarginal product of capital accruing toother firms. We know that some farmersallocate effort between plots in a way that isnot socially efficient, because they own oneplot and sharecrop in another. We haveexperimental evidence suggesting thatgroups discriminated against performbelow their own capacity, either becausethey internalize the stereotype or becausethey expect to be treated unfairly. Each ofthese pieces of carefully researched empiri-cal evidence, and others discussed in chap-ter 5, provide reasons why more equitableeconomies would, in most cases, also bemore efficient.3

Chapter 6 complements this picture bylooking at historical evidence, suggestingthat large inequalities in political rights andpower give rise to exclusionary institutionsthat generally impair development processes.Greater political equality, by contrast, estab-lishes limits on predation by the most pow-erful in each society. This tends to lead toinstitutions that level the playing field andprovide opportunities for advancement andmobility to those from underprivilegedbackgrounds.

Such institutions seem to be associatedwith more sustained growth. One examplecomes from contrasting the exploitative


labor practices of the Spanish conquista-dores in the mining centers of their Ameri-can colonies from the sixteenth to the eigh-teenth century, with the greater freedomand opportunity afforded to early settlers inNorth America. Another example ofinequitable treatment of citizens by thestate, which was also enormously costly forefficiency, was the very high taxation ofpoor African farmers by state-owned orparastatal agricultural marketing boards inGhana, Nigeria, and Zambia, which pre-vailed a few decades ago.

Equity and fairness matter not onlybecause they are complementary to long-term prosperity. It is evident that manypeople—if not most—care about equityfor its own sake. Some see equal opportu-nities and fair processes as matters ofsocial justice and thus as an intrinsic partof the objective of development. In chapter4, we briefly review arguments and evi-dence suggesting that most societiesexhibit a pervasive and long-standing con-cern for equity.

Why does equity matter? 75

People from many cultures seem to share aconcern for equity that is reflected in reli-gious and philosophical traditions, as wellas in legal institutions, both national andinternational. Religions from Islam to Bud-dhism and secular philosophical traditionsfrom Plato to Sen have shown both a con-cern for equity and an aversion to absolutedeprivation. In modern legal institutions,equity remains a fundamental tenet of the-ory and practice.

That a concern with equity is so pervasiveacross cultures, religions, and philosophicaltraditions suggests that a fundamental pref-erence for fairness is deeply rooted in humanbeings. We review experimental evidenceshowing that many people place a monetaryvalue on “fairness” and are prepared to giveup real money if they feel that a process theyare involved in is unfair. Complementing thisevidence are data from opinion surveys, andsurveys on subjective well-being, suggestingthat higher inequality in incomes is, on aver-age, associated with lower aggregate levels ofsubjective well-being.

An empirical link between incomeinequality and poverty reduction reinforcesthe conceptual link between the aversion toinequality and the quest to avoid absolutedeprivation. We highlight the obvious factthat, if inequality falls during a growthspell, poverty generally falls by more than ifinequality had not changed. We also docu-ment the less obvious fact that higher-income inequality reduces the effectivenessof future economic growth in reducingabsolute income poverty.

Ethical and philosophicalapproaches to equityPerhaps the oldest manifestations of concernwith equity and the avoidance of deprivation

come from religion. Several major world reli-gions endorse the notions of social justiceand a duty toward the poor. Buddhists see aduty to care for the poor. Christians are to“love their neighbor as themselves.” TheHebrew word for “charity” is the same as theone for “justice.” One of Islam’s five pillars offaith is zakat, providing for the poor andneedy. The World Faiths Development Dia-logue (1999) states that “all religions wouldsee the extreme material poverty in the worldtoday as a moral indictment to contemporaryhumanity and a breach of trust within thehuman family.”And religious views on equityare not restricted to poverty. Despite varyinginterpretations, and a wealth of differences inperspective, a belief in the fundamental dig-nity of human beings is a theological tenet inmost major religions. While there are impor-tant differences in how this belief manifestsitself across faiths, and even among differentgroups within the main religions, some ana-lysts see a growing emphasis on this principleof equality within various faiths.1

Equity is also a key theme in secularphilosophical traditions. Western politicaland ethical philosophy, for instance, haslong been concerned with distribution. Inancient Greece, Plato argued that “if a stateis to avoid . . . civil disintegration . . . extremepoverty and wealth must not be allowed torise in any section of the citizen-body,because both lead to disasters.”2 Roman law,while discriminating against slaves, as in allancient empires, also laid the foundationsfor some of the principles of equality thatunderlie modern legal principles in manycountries. Those principles applied only toRoman citizens who were free, but in mod-ern nations they have become all-inclusive.

In the modern era, Western thinkingabout social justice was greatly influencedby utilitarianism—the idea, originally from

Equity and well-being

4c h a p t e r

76

Equity and well-being 77

Bentham (1789), that the social goal shouldbe to achieve “the greatest happiness for thegreatest number.” Although utilitarianswere essentially unconcerned with the dis-tribution of happiness, enjoining societiessimply to maximize the sum of utilitiesacross all individuals, the approach hasearned the somewhat misbegotten reputa-tion (at least among economists) of havingegalitarian implications.3

Modern theories of distributive justicehave largely moved beyond utilitarianism, inpart because of its fundamental lack of con-cern with the distribution of welfare. Since theearly 1970s, a number of influential thinkers,including John Rawls, Amartya Sen, RonaldDworkin, and John Roemer, have made sepa-rate and important contributions to the waywe think about equity. Although the theoriesof justice and social choice proposed by eachof them are different in important respects,they share much in common.

All four reject final welfare (or utility) asthe appropriate space in which to judge thefairness of a given allocation or system. Allacknowledge the importance of individualresponsibility in moving from resources tofinal outcomes, including welfare. All preferto see some combination of the set of liber-ties and resources available to individuals asthe right space to form a social judgment.All seem to appeal, at some stage, to the “veilof ignorance” argument, from Harsanyi(1955), that a fair allocation of resourcesshould be one that all “prospective membersof society” would agree on, before they knewwhich position they would occupy. Theyused this thought experiment to argue thatjustice implies equality in the allocation toall people of some fundamental concept,such as primary goods.

What Rawls, Sen, Dworkin, and Roemerdisagree on is what exactly this concept shouldbe. Rawls (1971) argued that social justicerequired that two basic principles shouldhold. The first “demands the most extensiveliberty for each, consistent with similar libertyfor others.”4 The second requires that oppor-tunities—which he related to the concept of“primary goods”—should be open to allmembers of society. Under the DifferencePrinciple, he proposes that the chosen alloca-tion should be one that maximizes the oppor-

tunities of the least privileged group. (TheDifference Principle is also known as Rawls’s“maximin” principle).

Sen (1985) thought that different peo-ple might have different “conversion fac-tors” from resources to actions and wel-fare. He argued that all goods, includingRawls’ “primary goods,” are inputs to aperson’s functionings—the set of actions aperson performs and of states the personvalues or enjoys. For Sen, the concept to beequalized across people is the set of possi-ble functionings from which a personmight be able to choose (which he called a“capability set”).

Dworkin (1981b) and Dworkin (1981a)argued that justice required that individualsshould be compensated for aspects of theircircumstances over which they had no con-trol, or for which they could not be heldresponsible. He argued for a distribution ofresources that compensated people forinnate differences that they could not havehelped, including differences in talent.

Roemer (1998) argued that equitydemanded an “equal opportunity policy.” Heacknowledged that individuals bear someresponsibility for their own welfare, but alsothat circumstances over which they have nocontrol affect both how much effort theyinvest and the level of welfare they eventu-ally attain. He argued that public actionshould therefore aim to equalize “advan-tages” among people from groups with dif-ferent circumstances at every point alongthe distribution of efforts within the group.

Despite important (but subtle) differ-ences, all four thinkers have contributed toshifting the focus of social justice from out-comes to opportunities. We also take a leaffrom Nozick (1974), who is usually regardedas an anti-egalitarian. He argued that theo-ries of justice generally placed an excessiveemphasis on outcomes, such as welfare,utility, or even capabilities. Nozick remindedus of the (obvious) fact that outcomes arethe result of processes and argued that thecorrect focus for a theory of justice shouldbe on the fairness of processes. If a particu-lar allocation departs from a fair initialstate, and is arrived at through a fairprocess, it should be judged to be fair—even if it is unequal.

The concept of equity we adopt in thisreport draws on the contribution of thesefour thinkers by focusing on opportunities,rather than on welfare, utilities, or someother corresponding individual outcome.We do not dwell on the fine distinctionsbetween Sen’s capabilities and Roemer’sopportunities. As in both frameworks, weacknowledge the central role of individualresponsibility and effort in determining out-comes. We focus on eliminating disadvan-tage from circumstances that lie largelybeyond the control of the individual but thatpowerfully shape both the outcomes and theactions in pursuit of those outcomes.

These different perspectives on what asocial optimum should be can shed light onan important point for this report—onepreviously mentioned in chapter 1. It is notfor us to advise countries on what exactlyconstitutes an equitable distribution intheir societies. Instead, our role is to pointout the inequities we can observe and tonote that reducing them may be perfectlyconsistent with—perhaps even necessaryfor—greater efficiency and prosperity in thelong run. Box 4.1 draws on a classic publiceconomics discussion of these philosophi-cal perspectives to illustrate this point.

Equity and legal institutionsThe concerns with equity that feature inmoral, religious, and ethical debates aroundthe world are reflected in real-world institu-tions, through which people have sought topromote justice throughout history. Chiefamong them are legal institutions, where“equity” has a distinct—and specific—inter-pretation as a set of principles intended toguide and correct the application of the law.According to Kritzer (2002), how these prin-ciples merge with the written, codified lawvaries across legal traditions, but the over-arching concept of “fairness” is a cross-cultural reality. And in practice, definitionsoften refer directly to shared values within aparticular community5 as well as to the beliefthat people should not suffer before the lawas a result of having unequal bargainingpower.6

In Western philosophy, Aristotle isregarded as the first author to distinguishbetween justice and equity.7 He found that


The figure below (adapted from figure 11-1 inAtkinson and Stiglitz 1980),* can help sum-marize what this report hopes to achieve—and what it does not. Suppose that a societyconsists of only two groups of people (1 and2) and let the axes of the diagram depict theopportunity levels of each group. Opportu-nity sets are obviously multidimensional, butimagine for simplicity that the various dimen-sions can be meaningfully conflated onto an“opportunity index,”O1 for type 1s, and O2 fortype 2s. Now let the curved frontier AC repre-sent the “opportunity possibility frontier”forthis society.** It reflects the maximum oppor-tunity indexes that types 1 and 2 can obtain,given the available resources and technology.The fact that it does not monotonicallydecline from A to C incorporates the fact thatwhen type 1 individuals have very limitedopportunities, type 2 people can also benefitfrom an expansion in type 1 opportunity sets,and vice versa. Over some ranges, improve-ments in the opportunity sets of the “poor-est”types can be Pareto-improving—that is,can benefit everyone. Put another way, thereis scope for efficient, growth-promoting redis-tribution.

Eventually, however, tradeoffs set in.Between points P and R, if society is on theopportunity possibility frontier, anyimprovement in type 1 sets must imply areduction in type 2 sets, and vice-versa.Points B, R, and E are translations to this“opportunity space” of the welfare conceptsassociated with Bentham, Rawls, and fullegalitarianism, respectively.

• If this society wished to maximize thesum of total opportunity indexes, itshould aim for point B.

• If, instead, it wanted to maximize theopportunity set of the “poorest” group, itshould aim for point R.

• If it insisted on absolute equality ofopportunities, it must lie along the 45-degree ray through the origin, and wouldaim for point E.

What this report will not attempt to dois to advise countries on which of these cri-teria of social justice, or indeed any other, anindividual society should aim for. Each ofthese three points can be defended by logi-cal arguments, under different degrees ofaversion to inequalities in opportunity.

What the report will try to do is as follows:

• Describe the inequalities in opportunityactually observed in society (at a pointsuch as X).

• Investigate whether some of those dis-parities (which in this diagram greatlyfavor type 2) might actually be prevent-ing the society from enjoying higheraggregate opportunities (and welfare, onanother space).

• Suggest possible policy and institutionalapproaches that might help move frompoints such as X to whichever point soci-ety considers equitable, on the opportu-nity possibility frontier.

*Atkinson and Stiglitz (1980) refer, in turn, to afigure 1 in Buchanan (1976).**In Atkinson and Stiglitz (1980), utilities areused instead of opportunities.While that dis-tinction is fundamental in almost everyrespect, it is not for the point being made here,namely that different social justice criteriaimply different optimal allocations.

B O X 4 . 1 A simple representation of different concepts of equity

C

E

Opportunity possibility frontier

B: Maximize sum of total opportunity (O1 + O2)

R: Maximize opportunities for group 1 (O1)

E: Absolute equality of opportunitiesbetween group 1 (O1) and group 2 (O1)

R

BP

X

O1

O2

A

An illustration of choices between the opportunities of two types of people

courts enact justice according to law—that is,by applying general rules that give an equi-table solution in the majority of cases. In somecases, however, the results are inequitable.Equity then rectifies law in so far as the law isdefective on account of its generality.8 TheRomans operationalized this concept ofequity by distinguishing between ius strictum(strict law) and ius aequum (equity), with thelatter used to interpret the law and to comple-ment it. Equity prevailed in instances of con-flict between the two.

In modern legal traditions, equity re-mains a fundamental tenet of legal theoryand practice. In common law systems, equitywas historically a separate branch of lawadministered by Chancery Courts.9 TheJudicature Act of 1873 in the United King-dom “fused” the courts of law and equity,doing away with a bifurcated system ofcourts, while establishing the supremacy ofequity in cases of conflict between equity andcommon law. Equitable principles, based onconscience and fairness, have continued todevelop and be applied in common law juris-dictions around the world to mitigate harshand unfair results produced by the applica-tion of formal legal rules in specific cases.10

In general, the use of equity as a sourceof law in the civil law traditions of the Euro-pean continent is more limited than in thecommon law tradition. Civil legal codes,which have their origins in the Enlighten-ment era, aim at integrating equity intoformal law—that is, by designing lawsaimed at producing equitable results.Equity is seen as part of law and, therefore,should be achieved by applying the formalrules. Provisions in the codes that referexplicitly to equity, however, are used tocorrect inequitable results of the applica-tion of other formal provisions, in a waywhich is similar to common law systems.

Both the common law and the codifiedsystems from the continental European tra-dition have spread to countries around theglobe, and equity is now a global legal con-cept. The legal systems of Latin Americancountries, such as Argentina, Brazil, andMexico, have approaches to equity similarto those in continental Europe, whileBangladesh, India, and Nigeria follow thecommon law tradition. Importantly, equity is

not a purely Western concept—it can befound in legal systems around the world,including those that do not share Europeanorigins.11 For example, the distinctionbetween justice and equity is also found inIslamic law, in which the former is referred toas adala, the latter as insaf, and in Jewish law,with the distinction between din and tsedek.12

In today’s more integrated world, legalunderstanding of equity has also influencedinternational law—serving as the basis forindividualized justice, creating specificprinciples of fairness and reasonableness, orbeing identified with international equi-table standards for sharing resources andredistributing wealth. Perhaps the foremostexample of the development of interna-tional principles of equity is the interna-tional human rights regime. Internationalhuman rights law is rooted in a commit-ment to protect the “equal and inalienablerights of all members of the human family,”which itself is considered to be the “founda-tion of freedom, justice and peace in theworld.”13

The U.N. Charter laid the foundation forcontemporary international human rightslaw. The preamble to the Charter states thatthe U.N. community “reaffirms faith in fun-damental human rights, in the dignity andworth of the human person, in the equalrights of men and women and of nationslarge and small.”14 The Universal Declara-tion of Human Rights, adopted by the Gen-eral Assembly of the United Nations onDecember 10, 1948, is viewed as the “sourceof inspiration and . . . the basis for the U.N.in making advances in standard setting ascontained in the existing human rightsinstruments.”15 It has become a highly visi-ble and widely recognized statement ofmoral, ethical, and political standards at theinternational level.16

The contemporary international humanrights regime comprises a broad array oflegal instruments,17 many operating underthe aegis of the United Nations. There alsoexist regional human rights regimes inEurope (European Convention on HumanRights and Fundamental Freedoms), theAmericas (Inter-American Convention onHuman Rights), and Africa (African Char-ter on Human and Peoples’ Rights). In


addition, the laws of some other interna-tional entities, such as the European Union,incorporate human rights norms (Treaty ofNice, Charter of Fundamental Rights of theEuropean Union). Together, these differentlegal instruments are aimed at protectingpeople against a variety of harms, includingpotential harm by their governments, andat committing to the fundamental princi-ples of equality and nondiscrimination.

People prefer fairnessDifferent cultures and religions around theworld may differ in important respects, butthey all share a concern with equity and fair-ness. This suggests something quite funda-mental about the value human beings placeon them. A fairly recent body of literature ineconomics sheds some light on these sharedhuman preferences. It has amassed convinc-ing evidence on individuals’ preferences forfairness, based on controlled laboratoryexperiments. In these experiments, individ-uals interact through behavioral games andplay with real money under tightly con-trolled conditions. Results from such experi-ments over the last 10 or so years reject thehypothesis in standard economic modelsthat all individuals are exclusively concernedwith their material self-interest. This newbody of literature is large and rich, but itsmain findings can be summarized underthree main points.

First, some people behave in ways clearlyinconsistent with the rational self-interesthypothesis. According to Fehr and Fisch-bacher (2003), such people regularly displaya willingness to engage in two specific formsof behavior: “altruistic rewarding, a propen-sity to reward others for cooperative, norm-abiding behavior, and altruistic punishment,a propensity to impose sanctions on othersfor norm violations” (785). These behaviorsare observed in contexts in which it is possi-ble to rule out individual motivation by adesire for reciprocity or a concern with rep-utation. While reciprocity and reputationare important additional determinants ofcooperation in many settings, the experi-mental evidence suggests that they are notthe only factors that influence cooperativebehavior.

A classic example is the UltimatumGame, in which a player (the Proposer) isasked to suggest a one-time division of acertain sum of money (say, $100) betweenhimself or herself and another player. Thesecond player (the Responder) has thepower to simply accept or reject the offer.Acceptance leads to the implementation ofthe offer, whereas rejection leads to a zeropayoff for both players. Monetary stakes arefor keeps, and neither player knows the realidentity of the other player. Both players aretold that they will never play with eachother again.

In such circumstances, standard gametheory predicts a unique equilibrium: theProposer should offer the smallest possibleamount, and the Responder should accept(since a penny is higher than zero). But timeand again, across hundreds of experimentsin highly heterogeneous cultural circum-stances and with amounts ranging from onehour’s to one week’s local wages, observedoffers are substantially higher and, even so,rejections are often observed.18 In manyexperiments, the modal (most frequent)offer is actually at 50 percent. Figure 4.1depicts the actual distribution of observedoffers in two sets of Ultimatum Games, onewith lower monetary stakes (bar on the left)and one with higher (bar on the right).

Second, people are heterogeneous. A sizablefraction of people in most experiments (20


0 1–10 11–20 21–30 31–40

Proportion of pie offered (%)

41–60 51–60 61–700.0

0.1

0.2

0.3

0.4Frequency

Accept

Reject Accept

Reject

Low monetary stake ($10 pie)High monetary stake ($100 pie)

Figure 4.1 The distribution of observed offers inultimatum gamesOffers and rejections in high- and low-stakesultimatum games

Source: Based on data from Hoffman, McCabe, and Smith (1996).

to 50 percent) engage in altruistic giving oraltruistic punishment—expending realresources in a way that is unambiguouslycostly to them, without any hope of elicitingpersonal gain if the other person is ration-ally self-interested. But the behavior of oth-ers (a majority) is consistent with rationalself-interest. This is brought out quitestarkly in the Dictator Game, a variant ofthe Ultimatum Game in which the respon-der is purely passive. The second player issimply a Receiver, with no right to reject theoffer. Positive offers in the Dictator Gameare observed, but they are both rarer andsmaller on average than under the Ultima-tum Game, in which the Responder may—always at a cost—punish the Proposer.These results point to the importance ofinvestigating more precisely the conditionsunder which people exhibit self-interestedand other-regarding behavior.

Third, fair-minded people can behave self-ishly, and self-interested people can behavefairly. Behaviors depend on the rules of thegame. In games where competitive pressuresare introduced, mimicking a competitivemarket, players tend to quickly convergetoward actions consistent with self-inter-ested behavior. An example is an UltimatumGame with Multiple Proposers. If theResponder can choose among various offersfrom different Proposers, with all noncho-sen Proposers receiving zero, observedbehavior quickly tends to the Nash equilib-rium. In the Nash equilibrium, all Proposersoffer the full amount—or very close to it—despite this giving rise to a very unequal dis-tribution, in which the Responder capturesthe entire surplus, and all Proposers get zero.In other settings, however—such as theRepeated Public Good Game with Punish-ment in Fehr and Gachter (2000)—even asmall number of altruistic players can sus-tain a cooperative equilibrium.

These findings have been interpreted tosuggest that a sizable fraction of humanbeings in most societies care not only abouttheir own individual opportunities and out-comes but also about “fairness.” There is alsobroad agreement that fairness consists of aconcern for others, although some authorssuggest that it is other people’s intentions

toward us that attract reward or punish-ment, while others think it is their outcomesor opportunities.19 These studies do notusually distinguish explicitly between out-comes or opportunities. But it is possible tospeculate that the aversion to very unequalpayoff distributions in the Ultimatum Gamearises from the arbitrary and unequal natureof the endowments (or power) implicit inthe initial allocation of the roles of Proposerand Responder.

The experiments also show that people’sviews of fairness are complex and do notdepend entirely on outcomes. Some playersare prepared to punish noncooperators untilthey receive less than other people, becauseof what they perceive to be the unfairness oftheir actions in the process of the game. Thisis consistent with our emphasis that observeddistributions of certain outcomes—such asincomes—are the product of complex pro-cesses, and that the primary interest forthose concerned with equity is not the out-come, but the fairness of the processes theyparticipate in over their lifetimes. Anincome distribution in which some peopleare much richer than others because, givensimilar chances, they have worked muchharder, may be regarded as fair. But the sameincome distribution may be regarded asunfair if it was generated by the richer grouphaving access to much better schools or jobs,solely because of the wealth or connectionsof their parents.

A separate but related point is made bythe social identity literature in social psy-chology (see Haslam 2001) and epidemiol-ogy (see Marmot 2004), which suggests thatindividual behavior and performance areheavily conditioned by group identity (forexample, caste, gender, occupation); bywhether those groups are seen as subordi-nate to others (for example, doctors andpatients, the status accorded minority eth-nic communities); and by whether theboundaries among groups are regarded aspermeable (for example, the rules shapingwhether and how employees get promo-tions, immigrants become citizens, and soon). Civil servants with low status and fewupward mobility prospects suffer fromhigher mortality.20 Employees of low-statusfirms undergoing a merger more readily


laboratory, it is plausible that large inequal-ities in real life also reduce their well-being(particularly if the inequalities are not seento reflect only differences in effort ormerit). This provides support for the statis-tical association that the subjective well-being literature finds between incomeinequality and self-reported happiness—asubject we now turn to.

Income inequality and subjective well-beingTo what extent do the concerns with equityshown in tightly controlled lab experimentsalso manifest themselves in the attitudes,feelings, and opinions of “regular people?” Arecent study of labor strife in the UnitedStates suggests that worker perceptions ofwhether they have been treated fairly orunfairly can affect their efforts, and thusproduct quality, in important ways (box 4.3).

Other studies investigate the associationsbetween the narrower concept of incomeinequality and measures of subjective well-being. One recent study of Europeannations and the United States relies on indi-vidual answers to the following question:“Taken all together, how would you saythings are these days—would you say thatyou are very happy, pretty happy, or not toohappy?”22 Based on the variation in answersto this question, across European countriesand U.S. states, and on objective income-inequality measures, Alesina, Di Tella, andMacCulloch (2004) find that “individualshave a lower tendency to report themselveshappy when inequality is high, even aftercontrolling for individual income, a largeset of personal characteristics, and year andcountry [. . .] dummies” (2009).23

One reason that inequality might makepeople less happy, even when controllingfor absolute income levels, is that it violatestheir sense of fairness. It may be that (atleast some) people feel that a very unequalincome distribution reflects unfair processesand unequal distributions of opportunity. A2001 study of Latin American countries byLatinobarometro, a reputable Chile-basedopinion survey company, asked respondentsthe following question: “Do you think thatthe income distribution in your country isvery fair, fair, unfair, or very unfair?” On


Research is under way on the roots ofhuman altruism and the aversion toinequity, whether cultural or genetic. Butthere is some evidence that aversion tounfairness is not just human. In a recent arti-cle in Nature, “Monkeys Reject Unequal Pay,”Brosnan and De Waal (2003) report theresults of exchange experiments withbrown capuchin monkeys (Cebus apella).The animals were given a token that theycould immediately redeem for food byreturning it to the experimenter.They wereplaced in adjoining compartments withvisual and vocal contact.

In the baseline treatment (the equalitytest), both specimens received a quarter ofa cucumber slice for each token exchanged.

In one treatment of interest (the“inequality test”), the first monkey receiveda grape, while the second was given theusual slice of cucumber.That capuchinmonkeys prefer grapes to cucumber sliceshad apparently been amply established byprevious research.The results were striking.Monkeys failed to exchange their tokens forfood around 5 percent of the time under

the equality test, but this rate rose to morethan 50 percent under the inequality test.

The refusal rate rose even further (tomore than 80 percent) under an alternativetreatment, known as “effort control.” In thistreatment, the first monkey received agrape with no effort—with no need to pickup a token and exchange it for food.Although few monkeys were used, thesedifferences were all statistically significant.

In both treatments, refusal ratesincreased over time, as the experiment wasrepeated (never more than once a day) manytimes. Interestingly, only female monkeyscompleted these tests, as earlier experimentssuggested that male capuchins are muchless sensitive to the distribution of rewards.

The authors concluded that “tolerantspecies with well-developed food sharingand cooperation, such as capuchins [. . .]may hold emotionally charged expectationsabout reward distribution and socialexchange that lead them to dislikeinequity” (Brosnan and De Waal 2003, 299).

Source: Brosnan and De Waal (2003).

B O X 4 . 2 Capuchin monkeys don’t like inequity either . . .

accept the new organizational structurebecause it benefits them individually, whilemembers of the higher-status firm are morelikely to resist change and act collectively interms of their premerger identity.21

The experimental and subjective well-being literature in economics and socialpsychology remind us that there is some-thing deep and fundamental about our tastefor fairness and equity. Such “human altru-ism,” argue Fehr and Fischbacher (2003) inNature, may be what accounts for the muchgreater complexity of cooperative patternsin human societies compared with those ofother animals (box 4.2). Equity, it seems,matters intrinsically and fundamentally forhuman beings.

Whatever the exact form of the truemotives of individuals, the main implica-tion for this report from this large body ofexperimental evidence is that a good pro-portion of people in most societies appearto dislike unfair outcomes and behaviors, somuch so that they are prepared to pay topunish those responsible for them. If peopleare prepared to pay real money to reduceinequalities that appear unfair to them in a

average, 89 percent of respondents re-garded the distribution in their countriesto be either unfair or very unfair. In 17 ofthe 18 countries surveyed, fewer than 20percent of respondents answered fair orvery fair.24

Such results may be particularly strong inLatin America, which is one of mostunequal regions in the world, but they arenot exclusive to that region. A recent analysisof several OECD countries (which are lessunequal than most developing countries)was based on data from the InternationalSocial Survey Program. To construct a proxymeasure of cross-national attitudes towardincome inequality, Osberg and Smeeding(2004) ask what a number of different pro-fessions25 “should earn” and what they “doearn.” They find that citizens of most high-income countries26 appear on average tohave similar attitudes toward inequality,generally thinking that less well-paid profes-sions should be paid more and that better-paid professions should be paid less.

Osberg and Smeeding (2004) findingsreinforce the view that the normative prefer-ences people have over distributions are

based not exclusively on actual incomes, butalso on processes, and that some differencesin outcomes may be considered fair (forexample, because of differences in effort),while others are not (for example, because ofdifferences in opportunities). People areclearly aware that income differentials canprovide incentives for work and investment,including in education, if they are coupledwith opportunities for reward to thoseactions. This comes across very clearly fromthe answers to one question in the latest waveof the World Value Survey, which splitrespondents around the world more or lessevenly into those who felt that incomeinequality was too high and those who felt itwas too low.

Income inequality and incentives:What do people say?The World Value Survey is a multicountrysurvey of individuals designed and spon-sored by the Inter-university Consortiumfor Political and Social Research, based atthe University of Michigan. The surveyaims to “enable a cross-national compari-son of values and norms on a wide variety


Do people change their behavior because theyfeel they are being treated unfairly, when theyare outside laboratories? And if so, is this likelyto have any serious consequence? A study ofindustrial relations in Illinois, United States, sug-gests that the answer to both questions is yes.

Since the 1940s, Firestone tire plants hadadhered to the industrywide bargain with laborunions. But when negotiations for a new contractbegan in 1994, Bridgestone/Firestone proposeddeviating from the industrywide bargain in a waythat worsened the terms for labor at a time whencompany profits were increasing.The companyproposed moving from an 8- to a 12-hour shiftthat would rotate between days and nights. Italso proposed cutting pay for new hires by 30percent.The union at the plant in Decatur, Illinois,called a strike and, shortly after the strike, thecompany hired replacement workers.

Labor strife at the Decatur plant closelycoincided with lower product quality and defec-tive tires. In August 2000, Firestone announcedthe recall of 14 million ATX and AT tires, most ofthem on the Ford Explorer.The U.S. governmentreported that Firestone tires under investigation

were related to 271 deaths and more than 800injuries.The most common source of failure ofthe recalled tires was tread separation, a suddendetachment of the rubber tread from the steelbelts that caused the tire to blow out.

Krueger and Mas (2004) compare the num-ber of claims of defective tires in the Decaturplant with those from the two other NorthAmerican plants where Firestone ATX tires wereproduced: Joliette, Quebec, and Wilson, NorthCarolina. Neither of these two plantsexperienced labor strife in the relevant period.Tires produced during the labor dispute(1994–96) at Decatur had a much higher failurerate than those produced at Joliette or Wilson,although before and after the dispute periodthe rate of claims was similar for tires manufac-tured in Decatur and in the other plants.Thepattern suggests that changes in technologycannot explain the rise in complaints againstDecatur tires, because no such rise occurred inJoliette or Wilson.

Nor does it appear that the lack of experienceof replacement workers is to blame.There was aspike in claims in the first half of 1994, around the

time concessions were demanded and the oldcontract expired, which occurred before replace-ment workers were hired.Through early 1995,when many replacement workers were makingtires, there were no excess claims in the Decaturplant. It was not until the end of 1995, when manyreturning strikers were working side by side withreplacement workers, that the excess number ofclaims became high. On the basis of this and amuch broader analysis, it appears likely that thechemistry between the recalled strikers and thereplacement workers, or the cumulative impact oflabor strife in general, created the conditions thatled to the production of many defective tires.

The authors of the study “recommend thatthe reader exercise caution in extending ourresults to other settings; our paper provides adetailed case study of only one firm in oneunique period of its history” (Krueger and Mas2004, 257). At least in that instance, it appearsthat perceptions of unfair treatment influencedworker attitudes—and product quality and thesafety of consumers.

Source: Krueger and Mas (2004).

B O X 4 . 3 Worker perceptions of unfairness, product quality, and consumer safety

of topics.” Four main waves have beenfielded since the early 1980s. In the latestwave, Inglehart and others 2004) asked rep-resentative samples of people in 69 coun-tries to place their views on a scale from 1 to10, where 1 implied agreement with thestatement that “Incomes should be mademore equal,” and 10 implied agreementwith the statement that “We need largerincome differences as incentives for indi-vidual effort.”

Figure 4.2a suggests considerable polar-ization on views about inequality. Themedian answer is 6, suggesting no strongagreement with the two polar statements.Yet almost 20 percent of all respondentswere in strong agreement with each of thetwo extreme views, represented by scores of1 and 10. Figure 4.2b shows a positive corre-lation between the score (which is nega-tively correlated with inequality aversion)and a respondent’s own income. This isconsistent with the evidence on the impor-tance of relative incomes for welfare: if youare richer, you are less inclined to favor areduction in income inequalities than if youare poorer.

The World Values Survey results cautionagainst any preconceived notion thatincome inequality is viewed everywhere asinherently undesirable. When asked aboutincome differences explicitly “as incentivesfor individual effort,” (many) people seemquite happy to have them and, indeed, to

want more of them (although this ten-dency was less pronounced in countrieswith either very low or very high levels ofinequality).

The balance of the survey evidence sug-gests that, although inequality in incomesseems to be associated with lower aggregatelevels of subjective well-being, there is con-siderable heterogeneity in opinions aboutwhether it should be reduced or not.Poorer people, and people in countries atvery high or very low inequality levels,seem likelier to favor a reduction in in-equality. People recognize that someinequality is important to generate incen-tives for investment and effort; however,when asked about relative pay scales acrossprofessions, they would on average prefersmaller differentials. While in Latin Amer-ica, for instance, a majority judges theincome distribution to be unfair, there is noworldwide agreement that income dispari-ties should be reduced everywhere. This isgenerally consistent with a view that whatmatters for ethical judgment is not income,but fair processes and opportunities.

Income inequality and povertyreductionTo the philosophical and legal argumentsfor equity, and to the survey-based andexperimental evidence that fairness mattersintrinsically to people, we add a final argu-ment: high levels of inequality make it moredifficult to reduce poverty. First, we high-light the fact that if inequality falls during agrowth spell, poverty generally falls by morethan it would have if growth had been dis-tribution-neutral. Second, we documentthe finding that the effectiveness of futureeconomic growth in reducing absoluteincome poverty declines with initial incomeinequality.

If inequality falls, poverty falls moreduring spells of growthBy raising the incomes and consumption ofpeople across the distribution of income,economic growth is the main driver ofpoverty reduction in the developing world.The negative association between the aver-age annual rate of change in poverty and


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

20

15

10

7.0

6.5

6.0

5.5

5

0

Income decilesPreference for inequality

Frequency (percent)

Income: more equal or more unequal? Views on inequality vary by incomeMean score (lower values indicatepreference for more equality)

Figure 4.2 Views on inequality from the World Values Survey

Source: Inglehart and others (2004).Note: Author’s calculations are based on data for the years 1999–2000. Preference for inequality ranges from agree-ment with 1, “Income should be more equal,” to agreement with 10, “We need larger income differences as incentivesfor individual effort.”

the average annual rate of growth in meanincomes is immediately clear from figure4.3, suggesting that countries experiencinghigher rates of economic growth can beexpected to reduce poverty much fasterthan those that grow more slowly.27 Theslope of the simple regression line, –2.4, isthe average total elasticity of poverty withrespect to economic growth. It implies that,without controlling for any characteristicsof the country, 1 percentage point growth ina country’s mean income can be expected toreduce the incidence of poverty in thatcountry by about 2.4 percentage points.

This powerful association between eco-nomic growth and poverty reduction isone of the central stylized facts of develop-ment economics. Its qualitative nature haslong been understood, and it has recentlybeen quantified by Ravallion and Chen(1997), Dollar and Kraay (2002), and others.Indeed, the growth-poverty relationship isprobably more powerful than surprising: itmerely reflects the fact that, on average,the growth in the incomes of the poor issimilar to the growth of mean incomes(figure 4.4). Put differently: aggregate eco-nomic growth is, on average, distribution-neutral.28

There is, however, considerable variationaround those averages. About half the totalvariation in poverty reduction is accountedfor by economic growth (see the explana-tory power of the underlying regression forfigure 4.3).29 The other half must reflectchanges in the underlying distribution ofrelative incomes. This happens because theincidence of economic growth (its distribu-tional pattern) can vary dramaticallyacross countries. Two countries with simi-lar rates of growth in mean incomes canhave very different growth profiles acrossthe population. As one would expect,reductions in inequality at a given growthrate add a “redistribution component” tothe “growth component,” leading to fasteroverall poverty reduction.

The contribution of inequality reduc-tion alongside growth is illustrated by acomparison of the growth incidence curves(GIC) for Tunisia (1980–2000) and Senegal(1994–2001) (figure 4.5). In both coun-tries, the average annual growth rate in the

mean incomes from the household surveywas close to 2.5 percent. In Tunisia, wherethe distribution of this growth was rela-tively more beneficial to the poor, the head-count index of poverty fell by 67 percent(from 30 percent to 10 percent). This cor-responds to an annual rate of decline inpoverty of 5.4 percent. In Senegal, wheregrowth was less pronounced for the bottomhalf than for the upper half of the distribu-


150100500Change in log mean consumption or income between surveys (x100)

–50–100–150

500Change in log poverty headcount index (x100)

400

300

200

100

0

–100

–200

–300

–400

y = –2.3841x + 2.3517t-stat = 9.30R2 = 0.5225N = 73

BUL

LVA

PER MDAARG

HUNMLI POLLAOCIV PAR

YEM VEN SVNMNG

COLTTOZWE BGD

GEOROM SLVNER

MDGVNMBDI

LSOHRV ZAF NGANICRUS ZMB UGAGHA

BWAUZB EGYETH INDBOL BFALKA BRA PHLKEN

DZAHONPANTUR MEX

MAULTU CMR ECU GMBSENCHNALB GUYURY GTM

AZE CRIMAR PAKIDNEST

TUNIRNTHA

LYSJAM

KAZ

0

Figure 4.3 Growth is the key to poverty reduction . . .


150100500Change in log mean consumption or income between surveys (x100)

–50–100–150

50Change in log Gini index (x100)

40

30

20

10

0

–10

–20

–30

–40

–50

–60

y = –0.0288x + 3.3717t-stat = 0.515R2 = 0.004N = 73

KAZJAM

LYSTHAIRN

TUN

EST

IDN

PAKMAR

CRI

AZE GTMURY

GUY

ALB

CHN

SEN

GMB

ECU

CMRLTU

MAU

MEX

TUR

PANHON

DZA

KEN

PHL

BRA

LKA

BFA

BOL

IND

ETH

EGY

UZB

BWA

GHA

UGA

ZMB

RUS

NIC

NGA

ZAF

HRVLSO

BDI

VNMMDG

NER

SLV

ROMGEO

BGD

ZWE

TTO

COL

MNG

SVN

VEN

YEM

PAR

CIV

LAO

POL

MLI

HUNARGMDA

PER

LVA

BUL

Figure 4.4 . . . and, on average, growth is distribution-neutral


tion, poverty fell by only 15 percent (from68 percent to 57 percent), corresponding toan annual rate of poverty reduction of 2.3percent. Although some of this difference isdue to the fact that the actual growth ratewas marginally higher in Tunisia (2.7 per-cent versus 2.3 percent in Senegal), much ofit is clearly due to the different patterns inthe incidence of growth, which is evident infigure 4.5.

This contribution of declines ininequality to poverty reduction holds moregenerally. According to Datt and Ravallion(1992), a decomposition of changes inpoverty into growth and inequality compo-nents has been widely applied. Redistribu-tion components are usually smaller thangrowth components and, because inequal-ity often rises, they often have the “wrong”sign. But when inequality falls, this helpsreduce poverty.

A second and separate point is that thepower of growth to reduce poverty declineswith higher initial income inequality. Areduction in inequality today thereforealso tends to have a future impact on theeffectiveness of (even distribution-neutral)growth in reducing poverty. This occursbecause the shape of most income distri-butions means that the growth elasticityof poverty reduction tends to be smallerin more unequal countries. Put anotherway, because the initial distributions ofincome are different, the rate of poverty

reduction in two countries with the samedistribution-neutral growth rate may wellbe different.

Perhaps the most flexible way to capturethe variation in growth elasticity withinequality across the sample of countriesavailable for these exercises is simply tocompute the total and the partial growthelasticity of poverty reduction for each sin-gle country (in a single spell per country)and to plot it against the initial Gini coeffi-cient (figure 4.6).30 A positive relationship isapparent for both partial and total elasticityconcepts, for all four poverty line/povertymeasure combinations.31 The absolute valueof the growth elasticity of poverty reductionfalls as countries become more unequal,both for the total and for the partial con-cepts. The slope of the line fitted throughpanel (a) suggests that a 10 percentagepoint increase in the Gini coefficient is, onaverage, associated with a decline of 1.4 inthe (absolute value) of the elasticity. Giventhat the average elasticity is 2.53, this is nota small effect.

The fact that very unequal countries(with a Gini coefficient near 0.6) have atotal elasticity near zero in this sampleshould not be overemphasized. It is caused,in part, by increases in inequality in someof these countries during the recordedgrowth spells. This is evident from the factthat the partial elasticity (which controlsfor changes in distribution) does not reachzero for the same sample. Growth still con-tributes to poverty reduction, even in high-inequality countries. The robust findingrelates to the sign of the slope of the line,not its exact intercepts: higher initialinequality means that growth reducespoverty by a lesser amount.

It has been argued that this is a mechan-ical result in that, given a fixed functionalform for the income distribution, greaterinequality results in slower poverty reduc-tion even if each individual’s income growsat the same rate. Indeed, as indicated here,distributional change is on average uncor-related with mean growth rates so that, onaverage, the poor see their incomes grow atthe same rate as other people’s. That doesnot, however, follow from any law of nature.


0 20 503010 70 9040 60 80 100 100

4

Annual growth rate, % Annual growth rate, %

3

2

1

0

0 20 50Population percentile Population percentile

3010 70 9040 60 80

3

4

2

1

0

–1–1

Growth incidence curve

Tunisia Senegal

Growth incidence curve

Growth rate in meanGrowth rate in mean

Figure 4.5 The national growth incidence curves for Tunisia 1980–1995 and Senegal 1994–2001

Sources: Ayadi and others (2004) for Tunisia and Azam and others (2005) for Senegal. These are two of 14 country casestudies from the World Bank’s “Operationalizing Pro-Poor Growth” Study.

Income distributions in individual coun-tries can and do change during spells ofgrowth (see figure 4.5).33 There is nomechanical rule that states that the incomesof the poor must grow at the same rate asthe rest of the population.34 If on averagethey do and if, given the shape of the empir-ical income distributions, the poverty elas-ticities are lower in countries with higherinitial inequality, this is an empirical fact.

The balance of the evidence does not,therefore, allow much room for doubt thatgrowth elasticities of poverty reduction arestronger in more equal societies. Inequality

reduces the effectiveness of economicgrowth in reducing poverty. This meansthat, if all else remains the same, a reductionin income inequality today has a doubledividend: it is likely to contribute to a con-temporaneous reduction in poverty, and itis likely to make future growth reducepoverty faster.

Evidently, the caveat “if all else remainsthe same” is of crucial importance. The dis-tribution of incomes is a reflection of thegeneral equilibrium of an economy, basedon the social, political, and institutionalstructures that condition its behavior.


0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65

0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65

Gini Index, initial period Gini Index, initial period

Gini Index, initial period Gini Index, initial period

Total elasticity Partial elasticity

Total elasticity

–16

Partial elasticity11

8

5

2

–1

–4

–7

–10

–13

–16

11

8

5

2

–1

–4

–7

–10

–13

a. Total elasticity vs. Gini (headcount ratio, $1) b. Partial elasticity vs. Gini (headcount ratio, $1)

c. Total elasticity vs. Gini (squared poverty gap, $2) d. Partial elasticity vs. Gini (squared poverty gap, $2)

y = 15.039x – 8.7066t-stat = 3.24R2 = 0.1468N = 63

y = 6.5129x – 4.8068t-stat = 1.97R2 = 0.0585N = 65

LAO URY

BGDKAZ IRN

EGY CIVMLI JAMINDR

MARCOL

YEM VEN

ZWE SENALB

VNMBDI

AZE BOLMNG

LYSTUNEST TURGEONER THAROM IDN

INDU

MDG

PAN

CMRBRA

ZAF

PAK GUYMAU BFAPHL

CHNUHONCRI ECU

MEX GMB

GTMBWAUGA RUSUZBNGA LSONIC KGZ

UKR ETHKENPAR

DZA

–20

–15

–10

–5

0

5

10

–20

–15

–10

–5

0

5

10

0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65

0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65

RUSUGA GTMGMBCRI BFA HON

MAUNGA BWA PANECUZWE

SENCHNU

LSOPAK

KENNICMEXINDU MDG GUY

CMRBDI BRABOLPHL

IDNINDR THANERPAR ETHEST

AZEMYSUZB

JAMTUNGEOROM COLDZA TUR ZAF

VNMBGDIRN KGZURYMNG KAZALB VENARG

MARMLI

UKRBUL CIVEGYLAO

YEM PER

y = 13.758x – 8.2332t-stat = 2.45R2 = 0.0911N = 62

y = 8.2813x – 6.0475t-stat = 2.97R2 = 0.1231N = 65

KENPAR

DZAETH

ECU LSONICNGA

UKRKGZ

UZB BWARUS GTMEGY UGAGMB

MEXCRICHNU

HONZAF

PHLPAK

GUYMAUBFA BRAIDN CMRTHA

MDG PANMYSEST

INDUAZE

TURBOLTUN

VNMMNGMAR

ALB YEMNER

SENCOLZWE JAM VENBDI

INDRBGD CIV IRN

MLIKAZLAO

GEOROM

MDA

URY

RUS

ZAFDZAVNMBDIBGD ECU KEN

ETHGTMGMB

CRI UGAMDGMAU HONCHNU BWAPAK BFAMEX PANNER COLGUY

LAO ZWESEN LSOCMRIDN

THAPHLINDR

PARNGA

UZB BRAINDU

MYSAZETUN

ESTBUL JAM

MNG ARGVEN

BOLNIC KGZTURUKRALBCIV

IRN

MLI

KAZ YEMMARGEO PERROM

URYEGY

Figure 4.6 Greater inequality reduces the power of growth to reduce poverty

Source: Authors’ calculations.Note: The figure shows the scatter plots of country-level elasticities against initial-year Gini coefficients. Panel (a) shows the total elasticity for the headcount measure of poverty incidence,with a $1 per day poverty line. Panel (b) shows the partial elasticity for the same measure and the same line. Panels (c) and (d) also show the total and partial elasticities respectively, but nowfor the squared poverty gap index FGT(2) and with respect to a $2 per day line.32

Simple-minded attempts to change the wayincomes are distributed, without takinginto account the effects of policies on theincentives of all agents in the economy, arebound to fail. We return to the issue ofappropriate policy design in part III of thisreport. All that can be said about the resultshere is that, if policies exist that can lead to aless unequal distribution of resources with-out major costs to the (static and dynamic)efficiency of resource allocation, such poli-

cies are likely to lead to faster povertyreduction in the future, for any amount ofgrowth that the economy generates.

It turns out, however, that some inequali-ties—not necessarily those of incomes—arealso detrimental to economic growth itself.Such inequalities in power, assets, and accessto markets and services are most likely to bethe ones on which policy can productivelyfocus. The next two chapters turn to a dis-cussion of these “inefficient inequalities.”


89

In a world in which markets worked per-fectly, investment decisions would have lit-tle to do with the income, wealth, or socialstatus of the decision maker. They would bedetermined by the returns an investmentpromises and by the market price of capital,adjusted for the extra risk it entails. If peo-ple had good investment opportunities, itreally would not matter whether they hadthe money—they could always borrowwhat they needed, and if the risk botheredthem, they could always sell shares in theirbusiness and buy safer assets with themoney from the sale.

However, for various reasons—mainlyeconomic but also political—markets arenot perfect. If borrowers can willfully defaulton their loans, lenders prefer to make loansto borrowers who can provide collateralassets. Private returns for politically con-nected firms can be higher than for thosewithout such connections, and so these firmsmay attract more capital, even though socialreturns may not be any greater.1 Members ofgroups subject to discrimination may ration-ally invest less in their human capital thanthey would in the absence of such explicit orsubtle stereotypes.

After we give up the idea that marketswork anywhere close to perfectly, the scopefor a direct link between investment andthe distribution of wealth or power widenssubstantially, in many instances leading tounderinvestment by those who have goodgrowth opportunities.2 Correcting themarket failures directly is often not feasi-ble, and in these cases certain redistribu-tions of wealth, power, and resources canserve as second-best alternatives.3 In otherwords, interventions to enhance equity canimprove efficiency.

One of the great advances in developmenteconomics in the past 15 years is the accre-tion of a substantial body of evidence ondocumenting how well (or badly) asset andfinancial markets work in developing coun-tries. The fact that these markets rarely meas-ure up to their ideal creates the possibilitythat wealth and social status, defined as one’sposition in society both in ascriptive identityand in connections, will have an importantinfluence on investment decisions. It seemsnatural to start with this evidence.

Markets, wealth, status,and investment behaviorThe market for creditIn a perfect credit market, there is a singleinterest rate and everyone can borrow orlend as much as they want at that rate. Thatindividuals can borrow as much as they wantat the current rate explains the presumptionof a separation between the wealth or statusof the investors and the amount they invest.Whether they are rich or poor, well-con-nected or just off the streets, an extra dollarof investment will be profitable for themonly if the return they get from it is morethan the interest rate. If the interest rate ishigher, they would be better off lending thatmoney if it was their own, or borrowing lessif it were someone else’s. So, two people withthe same return on investment would end upinvesting the same amount.4

How close are real markets to this idealmarket? Chambhar is a market town inSindh (Pakistan), on the east bank of theIndus. In 1980–81 farmers from the areaaround Chambhar got most of their creditfrom about 60 professional moneylenders.Based on detailed data from 14 of these

Inequality and investment

5c h a p t e r

lenders and 60 of their clients, Aleem(1990) calculated the average borrowinginterest rate charged as 78.5 percent. But ifthese farmers wanted to lend their money,the banking system would pay them onlyabout 10 percent. It is possible, however,that they may not have been depositing inthe banks. An alternative measure of thedeposit rate that is relevant for these farm-ers is the opportunity cost of capital to themoneylenders, 32.5 percent. In either case,it suggests a gap of at least 45 percentagepoints between the borrowing and lendingrates.

The borrowing rate also varied enor-mously across borrowers. The standard devi-ation of the interest rate was 38.1 percent,compared with an average lending rate of78.5 percent. In other words, an interest rateof 2 percent and an interest rate of 150 per-cent are both within two standard deviationsof the mean. One possibility is that these dif-ferences in interest rates reflect differences inthe default rate: perhaps the expected repay-ment was the same for everybody, becausethose who paid higher rates were more likelyto default. Also the expected repaymentcould have been equal to the actual interestrate paid to the depositors, if the default ratewas high enough. But default was rare: forindividual lenders, the median default ratewas between 1.5 percent and 2 percent, witha maximum of 10 percent.

The same pattern—high and variable bor-rowing rates, much lower deposit rates, andlow default rates—shows up in the “Sum-mary Report on Informal Credit Markets inIndia.”5 This report summarizes results fromcase studies commissioned by the AsianDevelopment Bank and carried out underthe National Institute of Public Finance andPolicy.

For the urban sector, the data are basedon various case surveys of specific classes ofinformal lenders. For the broad class ofnonbank financial intermediaries calledfinance corporations, the maximum depositrate for loans of less than one year is 12 per-cent. These corporations offer advances forone year or less at rates that vary from 48percent per year to the utterly astronomical5 percent per day. The rates on loans ofmore than one year varied between 24 per-cent and 48 percent. Default, once again, isonly a small part of the story: default costsexplain only 4 percent of total interest costs.For hire-purchase companies in Delhi, thedeposit rate was 14 percent and the lendingrate was at least 28 percent and could be ashigh as 41 percent. Default costs were 3 per-cent of total interest costs.

For the rural sector, interest rates arehigh, but they are also variable (figure5.1). This finding is based on surveys ofsix villages in Kerala and Tamil Nadu, car-ried out by the Centre for DevelopmentStudies, Trivandrum. The rich (with Rs100,000 or more in assets) get most of thecredit (nearly 60 percent) and pay a rela-tively low rate (33 percent), while thosewith assets between Rs 20,000 and Rs30,000 pay rates of 104 percent and getonly 8 percent of the credit. The averageinterest rate charged by professional mon-eylenders (who provide 45.6 percent ofthe credit) is about 52 percent.

While the average deposit rate is notreported, the maximum from all the casestudies is 24 percent, and in four of them it isno more than 14 percent. In the category ofprofessional moneylenders, about half theloans were at 60 percent or more, butanother 40 percent or so had rates below 36percent. Default rates were higher than in theurban sector, but they still cannot explainmore than 23 percent of the interest costs.


125 5000

100 4000

75 3000

55 2000

25 1000

0 00 –

5,0005,000 –10,000

10,000 –15,000

15,000 –20,000

20,000 –30,000

Asset group (Rs)

30,000 –50,000

50,000 –100,000

100,000and above

Cumulativeproportionof credit[left scale]

Average interest rate (%) [left scale]Average loan size (Rs) [right scale]

Figure 5.1 In rural Kerala and Tamil Nadu, the rich access most of the credit and pay relatively lowrates

Source: Dasgupta, Nayar, and Associates (1989)

The fact that credit access depends onsocial status is also shown by Fafchamps’(2000) study of informal trade credit inKenya and Zimbabwe. It reports an averagemonthly interest rate of just over 2.5 per-cent (corresponding to an annualized rateof 34 percent), but it also notes that the ratefor the dominant trading group (Indians inKenya, whites in Zimbabwe) is 2.5 percent amonth, while the blacks pay 5 percent amonth in both countries.6 Chapter 9 alsoprovides evidence that in many countries“insiders” effectively lobby to limit access tofinancial institutions and that lending isskewed toward the rich, consistent with theevidence in figure 5.1.

None of these facts is surprising. Contractenforcement in developing countries is oftendifficult, and it is not easy to get courts topunish recalcitrant borrowers.7 As a result,lenders often spend at lot to make sure thattheir loans get repaid: it is plausible thatthese are the resources that drive a wedgebetween the borrowing rate and the lendingrate. Indeed, Aleem (1990) shows that theresources spent by lenders to monitor bor-rowers explain the nearly 50 percentagepoint gap between the lending and borrow-ing rates in his data. It is easy to imagine thatborrowers who are easier to monitor willenjoy better rates, which would explain whylending rates vary so much.

These imperfections in credit marketshave immediate implications for the rela-tionship between wealth and investment.First, with the rate of interest on depositsmuch lower than that on loans, the oppor-tunity cost of capital for those who justwant to invest their own money is muchlower than the opportunity cost for thosewho have to borrow. This means that thewealthy will end up investing much morethan the indigent, even if they face exactlythe same returns on their investment. Sec-ond, the lower interest rates charged to richpeople reinforce this conclusion, becausethe rich then face a lower opportunity costwhen they too are borrowing. Third, insome cases, those who are unable to providecollateral will have no access to credit at anyinterest rate.

We would thus expect the poor to under-invest, certainly relative to the rich, but also

relative to what would happen if marketsfunctioned properly. The capital releasedbecause they underinvest is absorbed by thenon-poor, who may actually end up overin-vesting relative to how they would invest inperfect markets. This is the reason: becausethe poor cannot borrow, the non-poor can-not lend as much as they would like to (thisis why deposit rates in developing countriesare often very low). And because the non-poor cannot lend, it makes sense for them tokeep investing in their own firms, even whenthe returns are low.

Because the poor underinvest, and be-cause the opportunity cost of capital to thenon-poor is thus lower than it would other-wise be, the composition of the investorsalso changes. In particular, firms that wouldnot be viable if markets functioned per-fectly (for example, because the interest ratewould be too high) can survive and evenexpand because markets are the way theyare. In other words, the “wrong” firms endup investing.

The market for insuranceThe ideal insurance market is one in whichpeople bear no avoidable risks. In a setting inwhich a single village constitutes a separateinsurance market closed to the rest of theworld (so that only people in the village caninsure other people in the village, in somekind of mutual insurance arrangement),individual consumption should respondonly to aggregate (village-level) income fluc-tuations and not to fluctuations in theincome of specific individuals. Put in blunterterms, as long as aggregate consumption isunchanged, individual income fluctuationsshould not translate into fluctuations inindividual consumption. When insurancemarkets work well, risk considerationsshould not have a significant impact on thechoices people make, irrespective of theirwealth, given that what an individual doeshas little impact on aggregate uncertainty.

While a perfect insurance market is morecomplex than a perfect credit market, andthus harder to detect, there have beenattempts to test the prediction about theirrelevance of fluctuations in one’s ownincome. The Côte d’Ivoire Living StandardsMeasurement Surveys from 1985 to 1987

Inequality and investment 91

provide panel data on the income and con-sumption of nearly 800 households, witheach household tracked for two consecutiveyears (1985 and 1986 or 1986 and 1987). Intable 5.1, the relationship between changesin consumption and changes in incomes isreported separately for the three mainregions and separately for 1985–86 and1986–87. The first row of the first block foreach year reports the basic correlationbetween income and consumption: a fall inincome always hurts consumption, althoughthe coefficient varies between a low of 0.15(a $1 reduction in income means that con-sumption goes down by $0.15) to a high of0.46. The next row does the same thing, butnow there is a village dummy intended topick up any village-level changes in con-sumption. Remarkably, the coefficients onown income, which under perfect insuranceshould have fallen to zero after controllingfor village-level changes, barely budge.8

Not all the evidence is quite so pes-simistic. Townsend (1994) used detailedhousehold-level data from four villages,which were intensively studied by the Inter-national Crop Research Institute in theSemi-Arid Tropics (ICRISAT) in India, tosee whether the full insurance hypothesis isconsistent with the data. He found thatwhile the data did reject the exact predic-tion, it did not miss by very much. In otherwords, his evidence suggested that villagersdo insure each other to a considerableextent: movements in individual consump-tion in his data seem largely uncorrelatedwith movements in income.

Later work by Townsend, based on datahe collected in Thailand, turned out to beless encouraging.9 Some villages seemed to

be much more effective than others in pro-viding insurance to their residents.Townsend describes in detail how insurancearrangements differ across villages. While inone village there is a web of well-function-ing, risk-sharing institutions, the situationsin other villages are different. In one village,the institutions exist but are dysfunctional;in another, they are nonexistent; in a third,close to the roads, there seems to be no risk-sharing whatsoever, even within families.10

As for credit, the failure of insurancecould have something to do with informa-tional asymmetries. It is not easy to insuresomeone against a shock that he aloneobserves, because he has every incentive toalways claim that things had gone badly. Butas Duflo and Udry (2004) demonstrate,spouses in Côte d’Ivoire do not seem to bewilling to insure each other fully againstrainfall shocks that affect them differen-tially. Because rainfall obviously is observ-able, at least part of the problem has to beelsewhere. One possibility is limited com-mitment. People may be happy to claimwhat was promised to them when it is theirturn to be paid, and then default when thetime comes for them to pay. This may beparticularly easy in a setting in which thesocial relations between the sets of peoplewho are insuring each other are not partic-ularly close, perhaps explaining whyTownsend found no insurance in the villageclosest to the road.

Lack of insurance should have an effecton the pattern of investment. That manyinsurable risks are uninsured means thatone cannot invest without personally bear-ing a significant part of the concomitantrisk. Indeed, big corporations able to sell


Table 5.1 The effect of income shocks on consumption, Côte d’Ivoire

West Forest East Forest Savannah All Rural

OLS 1985–6

No dummies 0.290 (6.2) 0.153 (3.2) 0.368 (5.8) 0.259 (8.8)Village dummies 0.265 (5.7) 0.155 (3.5) 0.373 (5.7) 0.223 (7.7)

OLS 1986–7

No dummies 0.458 (8.8) 0.162 (5.3) 0.168 (4.0) 0.239 (10.4)Village dummies 0.424 (8.1) 0.173 (5.6) 0.164 (3.8) 0.235 (10.1)

Source: Adapted from Deaton (1997), table 6.5, 381.Note: Absolute value of t-statistics are shown in brackets. The first row of each panel shows the coefficient on income change of a regres-sion of consumption changes on income changes. The second row reports the same result when village dummies are included in theregression. OLS = Ordinary Least Squares.

their equity in organized equity marketsmay be the only players who can really hopeto diversify away a large part of the risk of aparticular project. Given this fact and thereasonable assumption that the poor aremore risk-averse than the rich, we are likelyto be in a perverse situation in which thepoor may also find it hardest to reduce theirexposure to risk. Thus, they are likely to shyaway from riskier and higher-return invest-ments, reinforcing the prediction that thepoor invest too little.

The market for landIn a perfect land market, individuals canbuy or lease as much land as they want foras long as they want at a price that dependsonly on the quality of the land (and thelength of the lease). The lease should be at afixed rent, so that the lessor is the residualclaimant on the produce of the land. Thatland can be freely bought and sold ensuresthat there is no particular advantage or dis-advantage to owning land compared withany other asset of similar value. That thelessor is a residual claimant means that theland is put to optimal use. Not so, however,in practice.

Many developing (and some developed)countries have regulations about who canbuy land and how much or how little theycan buy. Binswanger, Deininger, and Feder(1995) argue that almost every developingcountry today has gone through a phasewhen it had regulations intended to con-centrate landownership. By contrast, Besleyand Burgess (2000) provide a list of regula-tions from different states in India, each ofwhich is an attempt to limit the concentra-tion of ownership in land.

Governments also directly limit trans-actions in land, with the ostensible aim ofpreventing the accumulation of land in thehands of a few people. In Ethiopia in thelate 1990s, Deininger and others (2003)note that selling and mortgaging land wereagainst the law. While rentals were offi-cially allowed (after being disallowed fortwo decades), local leaders and govern-ments were free to restrict even theserental transactions in land. For example,the Oromia region allowed farmers to rentonly 50 percent of their holding and stipu-

lates maximum contract terms of 3 yearsfor traditional technologies and 15 years formodern technologies.

It is often unclear who has the right tosell a particular plot of land, when no singleperson or family has a clear, undisputed,legal title to the land. This ambiguity reflectsencroachments and land grabs in the evolu-tion of land rights, as well as the importanceof custom in governing land relations, espe-cially in Africa. The recent popularity ofland titling as a social intervention is adirect consequence.

Where lease contracts exist, they are notalways of the fixed-rent type, at least whenthe land is used for cultivation. Many coun-tries, including the United States, have along tradition of an alternative contractualform: sharecropping. Under sharecropping,the farmer gets only a fraction of the pro-duce, but he does not need to pay a fixedrent. As Alfred Marshall pointed out morethan one hundred years ago, this weakensincentives and reduces the productivity ofthe land, but the near universality of share-cropping suggests that it is a response to areal need. There is some disagreementamong economists about the exact natureof that need.11 It is plausible, however, thatthe need is related to the fact that farmersare often poor, and making them pay thefull rent when their crop does poorly is dif-ficult and probably not desirable.

Leaseholds in developing countries tendto be short-lived. The norm is either a yearor a season. Longer leases are not unknown,but they are rare. This might reflect the factthat custom, rather than law, secures most ofthese leases: perhaps it is too much to rely oncustom to enforce leases of arbitrary length.

The imperfect salability of land can, ofcourse, hurt anyone who owns it. But therural poor probably have more of theirwealth in land than most people, so makingland nonsalable might be particularly harshon them.

What tends to discourage investment inthe land is the lack of an explicit title, or theinsecurity of tenure more generally (caused,for example, by the short duration of leasesand the possibility that the landlord mightthreaten to take the land away at the end ofthe lease). It clearly helps if land is owned


by the person contemplating the invest-ment. That most who work in agriculturetend to be too poor to buy out the land theyare cultivating is thus a potential source ofunderinvestment.

The market for human capitalOne thing makes the market for humancapital different from all the other assetmarkets: many decisions about investing inhuman capital are made by parents (orother family members) for their children. Inother words, those making the decisions aredifferent from those who receive the humancapital. It is not hard to imagine why thisseparation might introduce important dis-tortions to the functioning of this market.Gary Becker’s classic formulation avoidsthis issue by assuming that the family canborrow against the child’s future income,turning the problem into a conventionalinvestment decision. Under that assump-tion, the amount invested will not dependon the family’s means.

In practice, however, although humancapital is an asset, it cannot be legallypledged or mortgaged, for the simple rea-son that pledging your human capitalwould be tantamount to selling yourselfinto slavery.12 This obviously constrainspeople’s ability to borrow money to financeinvestments in their education.

When parents cannot borrow againsttheir children’s future income—true most ofthe time in most developing countries—they may still hope that those children willtake care of them in their old age. The hopemight be that the children do grow up toreap the benefits of their parent’s investmentand that they will pay their parents back. Butchildren know that they have no legal obli-gation to do so. If they do repay their par-ents, it is because they love their parents orbecause society expects them to do so.

Investments in human capital may thusbe driven as much by parents’ sense of whatis the right thing to do, as by any calculationof costs and benefits. Once we accept this, itbecomes clear that children’s human capitalmay not be very different from any otherconsumption good—so richer families willtend to invest more in their children’s healthand education. And human capital decisions

may be more a product of culture and tradi-tion than of the cold calculation of benefits.Benefits are relevant, but the responsivenessto them may not be as large as one mighthave expected.

In the market for human capital, thereward should be based entirely on thehuman capital supplied, not on other attrib-utes of the person supplying the skills. Dis-crimination based on gender, caste, religion,or race obviously violates this, but so does asystem of job allocation based on contacts.Until very recently, job discrimination basedon gender was the norm all over the world,and the number of countries where suchdiscrimination is still either legally orsocially accepted is dwindling but signifi-cant. Even where such discrimination isexplicitly frowned on, there is some evi-dence of continuing discrimination. Thesame is true of race, caste, and religion. Mostdiscrimination—unless legally mandatedthrough affirmative action in favor of a his-torically disadvantaged group, such as lowcastes in India and African Americans in theUnited States—flies in the face of explicitlaws against it.

One reason discrimination is so hard toeliminate comes from its sheer insidious-ness. Beliefs about differences are embeddedin everyday attitudes and practices in a waythat neither the discriminator nor the dis-criminated against may be conscious of,even though these beliefs transform howthey both behave. This is what underlies thepower of the stereotype. In a telling exam-ple, Stone, Perry, and Darley (1997) asked allparticipants in a recent experiment (Ameri-can Caucasians, hereafter referred to aswhites) to listen to the same runningaccount of an athlete’s basketball perform-ance on the radio. Half the participants wereled to believe that the target player waswhite, half that he was African American.The results indicated that information wasless likely to be absorbed if it was discordantwith the prevailing U.S. stereotypes thatwhites are more academically talented thanAfrican Americans, and that African Ameri-cans are more athletically gifted. The whitetarget player was perceived as exhibiting lessnatural athletic ability but more “courtsmarts.” The African-American target player


was perceived as exhibiting fewer courtsmarts but more natural athletic ability.

Such biases have also been documentedin real-world settings. A recent study of theeffect of stereotyping on judgment findsthat prison inmates with more Afrocentricfeatures receive harsher sentences thanthose with less Afrocentric features, con-trolling for race and criminal history.13

Bertrand and Mullanaithan (2003) showevidence from a field experiment provingbeyond reasonable doubt that there is ahigh degree of African-American discrimi-nation in the United States. They sent thesame resumes to a large number of compa-nies under either a stereotypically whitename or a stereotypical African-Americanname, and found a 50 percent higher call-back rate when the name was white. Thedata say that having a white name is worthas much as eight additional years of jobexperience. Moreover, the discriminationtended to be greater when the resume cor-responded to someone who was better edu-cated, suggesting that investment in humancapital among African Americans probablyis significantly underrewarded.

A very different form of discriminationcomes from the allocation of jobs based oncontacts. Munshi (2003) presents persua-sive evidence that contacts are very impor-tant in the allocation of jobs for migrantlabor in the United States. The employmentprospects for Mexican migrants there, itturns out, are much better when they arefrom areas where there was an earlier out-flow of migrants. Quite remarkably, it helpsif migrants are from an area where therewas a drought several years ago, whichpushed out a cohort of migrants to theUnited States. These migrants then help thelater generations of migrants from that areato find jobs. This is the clincher: it does nothelp to be from an area where there was arecent drought.

The perception of discrimination, con-scious or not, can affect investments inhuman capital. Those who expect to be dis-criminated against in a particular labormarket—rightly or wrongly, consciously orotherwise—will tend to invest less inacquiring the type of human capital that themarket rewards. This could, perversely, gen-

erate self-reinforcing behavior. If membersof the discriminated group invest less intheir own education, or in searching foremployment, others might use this under-investment to confirm their prejudiceagainst that group.

Stereotypes can be self-fulfilling not onlybecause they influence perceptions of thetarget of the stereotype, but also becausethey influence the behavior of the individu-als who are stereotyped. Stone and others(1999) asked college undergraduate volun-teers to play a miniature golf course. Perfor-mance was measured by how many strokeswere needed to put the ball in the hole:fewer strokes meant better performance.The variable that the experimenters manip-ulated was the description of the task. Inone treatment, the task was described as a“standardized test of natural athletic abil-ity,” in the other as a “standardized test ofsports intelligence.” When the task wasdescribed as a test of natural athletic ability,the African-American participants per-formed better than the whites: they aver-aged 23.1 strokes to complete the 10-holegolf course, compared with 27.8 for thewhites. But when the task was described as atest of sports intelligence, the race gap wasreversed: African Americans averaged 27.2strokes, whites 23.3.

One way to interpret this behavior is thatsocial ideas—stereotypes about the talentsof different social groups—impose boundsfrom within. Under the rational, self-inter-est hypothesis, individuals change theirbehavior only when their preferences orexternal constraints change. But the behav-ior of real individuals depends as well onbelief systems that society impresses onthem. Negative stereotypes create anxietythat may interfere with performance: that iswhy the psychologist Claude Steele termedthis kind of behavior “stereotype threat.”14

The beliefs underlying the stereotypes, ifdeeply internalized, can affect early deci-sions about prospective careers, and atti-tudes toward society, by changing whatAppadurai (2004) calls a person’s “capacityto aspire.” The reader may recall the exam-ple (from chapter 2) of the Batwa girl whowanted to be a cleaner upon completingschool. Positive stereotypes, by contrast, can


boost self-confidence and lead individualsto expend greater effort.

Stereotypes influence behavior twice—through their impact on individuals’ self-confidence, and through their impact on theway individuals expect to be treated. Toexamine the effect of stereotypes on the abil-ity of individuals to respond to economicincentives, Hoff and Pandey (2004) under-took experiments with low- and high-castechildren in rural north India. The caste sys-tem in India can be described as a highlystratified social hierarchy in which groups ofindividuals are invested with different socialstatus and social meaning.

In the first experiment, groups composedof three low-caste (“untouchable”) andthree high-caste junior high school studentswere asked to solve mazes and were paidbased on the number of mazes they solved.In one condition, no personal informationabout the participants was announced. In asecond condition, caste was announced witheach participant’s name and village. In athird condition, participants were segre-gated by caste and then each participant’sname, village, and caste were announced inthe six-person group.

When caste was not announced, therewas no caste gap in performance (figure5.2). But increasing the salience of caste ledto a significant decline in the average per-

formance of the low caste, regardless ofwhether the payment scheme was piece rate(that is, participants were paid 1 rupee permaze solved) or tournament (that is, theparticipant who solved the most mazes waspaid 6 rupees per maze solved, while theother participants received nothing). Whencaste was announced, the low-caste childrensolved 25 percent fewer mazes on average inthe piece-rate treatments, compared withthe performance of subjects when caste wasnot announced. When caste was announcedand groups were composed of six childrendrawn from only the low caste (a pattern ofsegregation that for the low caste implicitlyevokes their traditional outcast status), thedecline in low-caste performance was evengreater. While we cannot be sure from thesedata what the children were thinking, somecombination of loss of self-confidence andexpectation of prejudicial treatment likelyexplains the result.

The expectation by the low-caste sub-jects of prejudicial treatment may berational given the discrimination in theirvillages. But the discrimination itself maynot be fully rational. Cognitive limitationsmay prevent others from judging stigma-tized individuals fairly. That people arebounded in their ability to process informa-tion creates broad scope for belief sys-tems—in which some social groups areviewed as innately inferior to others—toinfluence economic behavior. If such beliefspersist, it will generally be rational for thosediscriminated against to underinvest (withrespect to others) in the accumulation ofskills for which the return is likely to belower for them. This rational calculation isadditional to any reduction in their “capac-ity to aspire,” arising from the internaliza-tion of those beliefs.

The evidence onunderinvestmentHighly imperfect markets suggest consider-able scope for underinvestment.

Industry and tradeDirect estimates of marginal products showthat there are many unexploited investmentopportunities. For small Mexican firms


8

6

4

2

0Caste

notannounced

Castenot

announced

Casteannounced

Casteannounced

Casteannounced

andsegregated

High casteLow caste

Piece rate Tournament

Average number of mazes solved, by caste,in five experimental treatments

Figure 5.2 Children’s performance differs when theircaste is made public

Source: Hoff and Pandey (2004).Note: A vertical line in the figure indicates that the caste gaps arestatistically significant.

with less than $200 invested, the rate ofreturn reaches 15 percent per month, wellabove the informal interest rates available inpawn shops or through microcredit pro-grams (on the order of 3 percent a month)(figure 5.3).15 Estimated rates of returndecline with investment, but the ratesremain high—7 percent to 10 percent amonth for firms with investments between$200 and $500, and 5 percent for firms withinvestments between $500 and $1,000. Allthese firms are thus too small and couldreap large gains from increased investment.

Trade credit is an important form ofcredit everywhere, perhaps especially wherethe formal institutions of the credit marketare underdeveloped. Fisman (2001a) lookedat the relation between access to trade creditand capacity utilization for 545 firms inCôte d’Ivoire, Kenya, Tanzania, Zambia, andZimbabwe. He finds that firms that receivetrade credit from three main suppliers (onaverage, about one of the three suppliersprovides trade credit) have 10 percent bettercapacity utilization than firms that receiveno trade credit. The relation is muchstronger in industries in which it is impor-tant to carry large inventories.

Such studies present serious method-ological issues, however. The basic problemcomes from the fact that investment levelsare likely to be correlated with omitted vari-ables. For example, in a world without creditconstraints, investment will be positivelycorrelated with the expected returns toinvestment, generating a positive “abilitybias.”16 McKenzie and Woodruff (2003)attempt to control for managerial ability byincluding the firm owner’s wage in previousemployment. This goes only part of the way,however, if individuals choose to enter self-employment precisely because their expectedproductivity in self-employment is muchhigher than their productivity in anemployed job. Conversely, if capital is allo-cated to firms to avoid their failure, therecould be a negative ability bias.

Banerjee and Duflo (2004a) take advan-tage of a change in the definition of the“priority sector” in India to circumventthese difficulties. All banks in India arerequired to lend at least 40 percent of theirnet credit to the “priority sector,” which

includes small industry. In January 1998,the limit on total investment in plants andmachinery for a firm to be eligible for inclu-sion in the small industry category wasraised from Rs 6.5 million to Rs 30 million.The researchers first show that, after thereforms, newly eligible firms (those withinvestment between Rs 6.5 million and Rs30 million) received, on average, largerincrements in their working capital limitthan smaller firms. They then show that thesales and profits increased faster for thesefirms during the same period. Putting thesetwo facts together, researchers can estimatethe impact of the increased access to work-ing capital on the growth in profits. Allow-ing for the possibility that the firms in thepriority sector were paying less than thetrue cost of capital for the extra moneyfrom the bank, they estimate that thereturns to capital in these firms must be atleast 94 percent.

A different kind of evidence for underin-vestment comes from the fact that manypeople pay the high interest rates reportedearlier. Given that this money typically goesinto financing trade and industry, the pre-sumption is that the people borrowing atthese rates of often 50 percent or more musthave a marginal product of capital that iseven higher. But the average marginal prod-uct in developing countries seems to benowhere near 50 percent. One way to get atthe average of the marginal products is tolook at the incremental capital-output ratio(ICOR) for the country as a whole.17 For


0 100 200 300 400 500Nonrented capital (US$)

600 700 800 900 1000

30Monthly returns (%)

20

10

0

Figure 5.3 Returns to capital vary with firm size:evidence from small Mexican firms

Source: McKenzie and Woodruff (2003).

the late 1990s the International MonetaryFund (IMF) estimates the ICOR to be morethan 4.5 for India and 3.7 for Uganda. Theimplied upper bound on the average mar-ginal product is 22 percent for India and 27percent in Uganda.

That many firms in India have a mar-ginal product of 50 percent or more, whilethe average marginal product is only 22 per-cent or so, is strong prima facie evidence forthe misallocation of capital. The firms withthe marginal product of 50 percent andmore are clearly too small, while other firms(the ones who bring the average down to 22percent) must, in some sense, be too large.

A specific example of this kind of misal-location of capital comes from a study ofthe knitted garment industry in the south-ern Indian town of Tirupur.18 Two groupsof people operate in Tirupur: Goundersand outsiders. The Gounders, who issuefrom a small, wealthy, agricultural commu-nity from the area around Tirupur, movedinto the readymade garment industrybecause there were not many investmentopportunities in agriculture. Outsidersfrom various regions and communitiesstarted joining the city in the 1990s.

The Gounders, unsurprisingly, havemuch stronger ties in the local community,and thus better access to local finance. Butthey may be expected to have less naturalability for garment manufacturing than theoutsiders, who came to Tirupur preciselybecause of its reputation as a center for gar-ment export. The Gounders own about

twice as much capital as the outsiders onaverage. Gounder firms of all ages ownmore capital, although there is a strong ten-dency toward convergence as the firms age(figure 5.4a). The Gounders, despite own-ing more capital, lose their early lead insales by about the fifth year, and end up sell-ing less (figure 5.4b). In other words, out-siders invest less and produce more. Theyare clearly more able than the Gounders,19

but because they are less cash-rich and donot have the right connections, they end upworking with less capital.

AgricultureThere is also direct evidence of high rates ofreturns on productive investment in agricul-ture. In the forest-savannah in SouthernGhana, cocoa cultivation, receding for manyyears because of the swollen shoot disease,has been replaced by a cassava-maize inter-crop. Recently, pineapple cultivation forexport to Europe offered a new opportunityfor farmers in this area. In 1997 and 1998more than 200 households cultivating 1,070plots in four clusters in this area were sur-veyed every six weeks for about two years.Pineapple production dominates the tradi-tional intercrop (figure 5.5),20 and the aver-age returns associated with switching fromthe traditional maize and cassava intercropsto pineapple is estimated to be in excess of1,200 percent! Yet only 190 out of 1,070plots were used for pineapple. When theauthors asked farmers why they were notfarming pineapple, the virtually unanimousresponse was, “I don’t have the money,”21

although some heterogeneity in abilitybetween those who have switched to pine-apple and those who have not, cannot beentirely ruled out.

Evidence from experimental farms sug-gests that, in Africa, the rate of returns tousing chemical fertilizer (for maize) wouldalso be high. But the evidence may not berealistic if the ideal conditions of an experi-mental farm cannot be reproduced onactual farms. Foster and Rosenzweig (1995)show, for example, that the returns toswitching to high-yielding varieties wereactually low in the early years of the greenrevolution in India, and the returns wereeven negative for farmers without an educa-


1098765432101098765Years of experienceYears of experience

43210

3.5log capital stock (100,000 Rs)

Gounder firms of all ages own more capital

log sales value (100,000 Rs)

Gounders lose early lead in sales around fifth year

3.0

2.5

2.0

7.0

6.5

6.0

5.5

5.0

4.5

4.0

Gounders

OutsidersOutsiders

Gounders

Figure 5.4 Inefficient allocation of resources; the example of the Gounders vs. the outsiders

Source: Banerjee and Munshi (2004).

1008060Cumulative percent of plots

40200

20,000Per hectare profits (1,000 Cedis)

Pineapple profits

Non-pineappleprofits

15,000

10,000

5,000

0

–5,000

Figure 5.5 Average returns forswitching to pineapples as anintercrop can exceed 1,200 percent

Source: Goldstein and Udry (1999).

tion. This, despite the fact that these vari-eties had been selected precisely for havinghigh yields, in proper conditions. But theyrequired complementary inputs in the cor-rect quantities and timing. If farmers werenot able or did not know how to supplythem, the rates of returns were actually low.

Chemical fertilizer, however, is not a newtechnology, and the proper way to use it iswell understood. To estimate the rates ofreturns to using fertilizer on farms inKenya, Duflo, Kremer, and Robinson(2004), in collaboration with a small non-governmental organization (NGO), set upsmall randomized trials on people’s farms.Each farmer in the trial delimited two smallplots. On one randomly selected plot, a fieldofficer from the NGO helped the farmerapply fertilizer. Other than that, the farmerscontinued to farm as usual. The rates ofreturn from using a small amount of fertil-izer varied from 169 percent to 500 percent,depending on the year, although marginalreturns declined quickly with the quantityof fertilizer used on a plot of a given size.

Evidence for a different type of underin-vestment in agriculture is the negative size-productivity relationship, the idea that thesmallest farms tend to be the most produc-tive (table 5.2). The gap in the productivityof small and large farms within a countrycan be enormous: a factor of 5.6 in Braziland a factor of 2.75 in Pakistan.22 It issmaller in Malaysia (1.5), but a large farm inMalaysia is not very large. This is strongprima facie evidence that markets are some-how not allocating the right amount of landto those who currently farm the smallerplots.

The problem with this kind of evidence isthat it ignores the many reasons why the big-ger farm may be inherently less productive,for example, lower soil quality. Even so, sim-ilar (but somewhat less dramatic) resultsshow up even after controlling for differ-ences in land quality. The profit-wealth ratioin Indian ICRISAT villages is the highest forthe smallest farms, and when risk is compar-atively low, the gap is more than 3:1 (figure5.6). Because wealth includes the value ofthe land, the measure implicitly takes intoaccount differences in the quality of theland. It does so as long as land prices are a

reasonable measure of land quality, which,however, is not entirely clear. There are alsoresidual doubts about whether the returnsare well measured—it is possible that theland of the smaller farms is degrading faster,but the degradation is not being countedwhile calculating the returns.

For these same firms, when risk goes up,the average return goes down. In part thismay be inevitable, but it may also reflect thefact that the lack of insurance encouragespeople to avoid risky (but remunerative)choices.23 This is consistent with the fact thatprofitability falls faster for the poorer farm-ers (less able to self-insure) as the risk goesup. Specifically, a one-standard-deviation in-crease in the coefficient of variation of rain-fall leads to a 35 percent reduction in theprofit of poor farmers, a 15 percent reduc-tion in the profit of median farmers, and noreduction in the profit of rich farmers. Thestudy also finds that input choices areaffected by variability in rainfall, and in par-


Table 5.2 Farm size productivity differences, selected countries

Farm size Northeast Brazil Punjab, Pakistan Muda, Malaysia

Small farm 563 274 148(hectares) (10.0–49.9) (5.1–10.1) (0.7–1.0)

Largest farm 100 100 100(hectares) (500+) (20+) (5.7–11.3)

Source: Berry and Cline (1979).Note: 100 = land productivity in the largest farm size.

24

0.4

0.3

0.2

0.1

022201816141210

20th percentile

40th percentile

60th percentile80th percentile

Profit/wealth ratio

Monsoon onset, standard deviation (weeks)

Figure 5.6 Profit-wealth ratios are highest for thesmallest farms

Source: Rosenzweig and Binswansger (1993).Note: The standard deviation of the date of monsoon onset is ameasure of underlying risk. The onset date of the monsoon was thesingle most powerful of eight rainfall characteristics to explaingross farm output. The data come from the Indian ICRISAT villages.

ticular, poor farmers make less efficient inputchoices in a risky environment.

In related work, Morduch (1993) specifi-cally investigated how the anticipation of acredit constraint affects the decision toinvest in high-yielding variety seeds. Usinga methodology inspired by Zeldes (1989),he splits the sample into two groups—onegroup of landholders expected to have theability to smooth their consumption, andone group that owns little land, expected tobe constrained. He finds that the more con-strained group devotes a considerablysmaller fraction of land to high-yieldingvariety seeds for rice and castor.

Another consequence of the lack ofinsurance is that it may lead households touse productive assets as buffer stocks andconsumption smoothing devices, whichwould be a cause for inefficient investment.Rosenzweig and Wolpin (1993) argue thatbullocks (an essential productive asset inagriculture) serve this purpose in ruralIndia. They show, using ICRISAT data cov-ering three villages in semiarid areas inIndia, that bullocks, which constitute a largepart of households’ liquid wealth (50 per-cent for the poorest farmers), are boughtand sold quite frequently (86 percent ofhouseholds had either bought or sold a bul-lock in the previous year). Moreover, theybuy when they are flush with money andsell when they are broke.

Since people are not simultaneously sell-ing and buying land, they are not sellingthese animals because they no longer needthem for production. Indeed, from the viewpoint of production, most of these farmersshould own two bullocks and never sellthem. If they are selling, the reason is thatthey need the money for consumption. Thedata suggest that, for poor or midsize farm-ers, there is considerable underinvestmentin bullocks, presumably because of the bor-rowing constraints and the inability to bor-row and accumulate financial assets tosmooth consumption: almost half thehouseholds in any given year hold no bul-locks (most of the others own exactlytwo).24

There is also compelling evidence thatsharecroppers lack incentives. Binswanger

and Rosenzweig (1986) and Shaban (1987)both show that productivity is 30 percentlower in sharecropped plots, controlling forfarmers’ fixed effects (that is, comparing theproductivity of owner-cultivated and farmedland for farmers who cultivate both theirown land and that of others) and for landcharacteristics. Shaban (1987) shows thatall the inputs are lower on sharecroppedland, including short-term investments(fertilizer and seeds). He also finds system-atic differences in land quality (owner-cultivated land has a higher price perhectare), which could in part reflect long-term investment.

On the impact of security of property,Do and Iyer (2003) find that a land reformthat gave farmers the right to sell, transfer, orinherit their land-use rights also increasedagricultural investment, particularly theplanting of multiyear crops (such as coffee).Laffont and Matoussi (1995) use data fromTunisia to show that a shift from sharecrop-ping to owner cultivation raised output by33 percent, and moving from a short-termtenancy contract to a longer-term contractincreased output by 27.5 percent.25

Security of property rights is oftenlinked to the local power structure. Theconnection between inequalities in powerand underinvestment is nicely exemplifiedby the Goldstein and Udry (2002) study ofinvestment in land in a setting where land isallocated by custom (rural Ghana). Theyshow that individuals are less likely to leavetheir land fallow (an investment in long-run productivity of the land) if they do nothold a position of power within either thehierarchy of the village or the hierarchy ofthe lineage. The problem is that the landgets taken away from them when it is lyingfallow. Because women rarely hold thesepositions, women’s land is not left fallowenough and is much less productive thanmen’s.

Human capitalAccording to the report of the Commissionon Macroeconomics and Health (2001),returns to investing in health are on theorder of 500 percent. But these numbers,arrived at through cross-country growth


regressions, are not as easy to interpret aswhat would actually happen if someonewere to invest an extra dollar in health. Thatsaid, there clearly are examples of specifichealth interventions that have enormousprivate and social returns. There is substan-tial experimental evidence that supplemen-tation in iron and vitamin A increases pro-ductivity at relatively low cost.

• Basta, Soekirman, and Scrimshaw (1979)study iron supplementation among rub-ber tree tappers in Indonesia. Baselinehealth measures indicated that 45 per-cent of the study population was anemic.The intervention combined an iron sup-plement and an incentive (given to bothtreatment and control groups) to takethe pill on time. Work productivityamong those who got the treatmentincreased by 20 percent (or $132 a year),at a cost per worker-year of $0.50. Eventaking into account the cost of the incen-tive ($11 a year), the intervention sug-gests extremely high rates of returns.

• Thomas and others (2005) obtain lowerbut still high estimates in a larger experi-ment, also in Indonesia. They found thatiron supplementation experiments inIndonesia reduced anemia, increased theprobability of participating in the labormarket, and increased earnings of self-employed workers. They estimate that,for self-employed males, the benefits ofiron supplementation amount to $40 peryear, at a cost of $6 per year.26

• The cost-benefit analysis of a dewormingprogram27 in Kenya reports estimates ofa similar order of magnitude. Takinginto account externalities (because of thecontagious nature of worms), the pro-gram led to an average increase inschooling of 0.14 years. Using a reason-able figure for the returns to a year ofeducation, this additional schooling willlead to a benefit of $30 over the life of thechild, at a cost of $0.49 per child per year.Not all interventions have the same ratesof return, however. A study of Chinesecotton mill workers28 led to a significantincrease in fitness, but no correspondingincrease in productivity.

Measured returns to private investment ineducation tend not to be quite so high.Banerjee and Duflo (2004b) survey cross-country evidence, and conclude that—

Using the preferred data, the Mincerian ratesof returns seem to vary little across countries:the mean rate of returns is 8.96, with a stan-dard deviation of 2.2. The maximum rate ofreturns to education (Pakistan) is 15.4 per-cent, and the minimum is 2.7 percent (Italy).29

But most of the educational benefits ofdeworming mentioned above would becaptured by a child whose parents arewilling to spend $0.50 on the dewormingmedicine. This clearly offers a returnmuch higher than the measured Mincer-ian returns at affordable absolute cost,although they are not strictly compara-ble. Deworming does not require thechild to spend more years in school, butit does help the child get more out of theyears he or she is already spending inschool. However, when the dewormingmedicine was offered free to the chil-dren, the take-up was only 57 percent. Inthis sense, it is clear that at least somecauses of underinvestment have to befound in the way the family makes deci-sions, rather than in the lack ofresources.

The fact that a lack of connections altersthe nature of human capital investment isnicely demonstrated in a recent paper byMunshi and Rosenzweig (forthcoming).They show that, in India, trade liberaliza-tion increased returns to knowing the Eng-lish language in families with connectionsin the blue-collar sector compared withfamilies with no connections. However,there is a much bigger gap between girls andboys in the increase in enrollment in Eng-lish-medium schools. This is attributed tothe fact that girls never really expected toget these blue-collar jobs, while for theirbrothers, it depended on whether they hadthe right contacts.

Inequalities and investmentFour important points follow from thisbody of evidence: first, markets in develop-ing countries are highly imperfect, and


those who do not have enough wealth orsocial status tend to underinvest. Theresources underused because of this under-investment end up being used for some lessproductive purpose, reducing overall pro-ductivity. In the example from the knittedgarment industry in Tirupur, the Gounderswere overinvesting in their own relativelyunproductive firms, while the much moreproductive firms of the outsiders werestarved of capital. The land owned byGhanaian women was getting degraded,because they did not have the social statusneeded to hold on to the land during thefallowing period. This, once again, is a pureloss for society. The fact that other peoplewho do have status and can fallow theirland as needed is not, in any way, compen-sating for the loss of productivity on thelands of the powerless. This creates a strongpresumption that certain specific types ofredistribution, by empowering certain peo-ple or increasing their access to resourcesor contacts, can promote efficiency andequity.

Second, this hypothesis would imply a biasin favor of those kinds of redistribution thattarget the specific lack of access to resources orinfluence causing the inefficiency. In some sit-uations this will mean redistributing assets,but it also might mean redistributing access tocapital, perhaps by promoting microcredit,strengthening women’s land rights or accessto jobs and welfare programs, designingaffirmative action programs to break downstereotyping, and improving access to justicesystems.

Third, because investments build wealthand wealth makes it easier to invest in aworld where markets do not function verywell, a little help can go a long way. Startingthe right business might be the biggest chal-lenge: once started, the business might pro-pel itself forward without any further help.

Fourth, it is not clear that the beneficiar-ies from this kind of efficiency-promotingredistribution have to be the poorest of thepoor. Because the ideal is to promote pro-ductive investments, the target should bethose most likely to make these invest-ments. Whether the poorest are the rightpeople from this point of view is an empiri-cal question, and one for which the answer

might depend on the set of economicopportunities available.

The microcredit community, in particu-lar, has long debated this last issue in tryingto decide whether microcredit is best instru-ment for helping the poorest of the poor.This clearly turns partly on whether thepoorest are the ones who have the projectswith the highest returns, which could be thecase if the poor and the less poor have thesame kinds of production functions, and ifthere are diminishing returns to scale. If,instead, the most productive technology inthis area had a fixed cost of production but(say) diminishing returns otherwise, givingthe poorest access to more capital may notbe very productive: even with all the capitalthey can get, they may not be able to coverthe fixed cost. It may be more effective tohelp people who are slightly richer, becausewith some help they may actually be able tostart a business.

How good or bad is the assumption ofdecreasing returns in the production func-tion of an individual firm? As mentionedabove, McKenzie and Woodruff (2003) esti-mate a production function for small Mexi-can firms, suggesting strong diminishingreturns. Mesnard and Ravallion (2004) findweak diminishing returns using Tunisiandata. But estimating a production functionthat exhibits local increasing returns isinherently difficult. A firm is likely to grow(or shrink) quickly when it is in the regionof increasing returns. So we will observe fewfirms in this region, and be likely to rejecttoo often the assumption of local increasingreturns. Certainly the natural interpretationof the results in Banerjee and Duflo (2004a),showing close to 100 percent returns inmedium firms in India, is that there areincreasing returns over some range.

A corollary of this discussion is that theredistribution that maximizes productivitygrowth is not necessarily the one that hasthe strongest immediate effect on poverty.Nor is it the one that does most to reduceinequality. Indeed, except under very spe-cial circumstances, this discussion tells usnothing about the relation between someglobal measure of inequality and the effi-ciency of resource use or investment. Con-sider the case, discussed above, in which the


production function has a fixed cost butalso diminishing returns. If all firms areequal and the maximum they can eachinvest is less than the fixed cost, no one willbe able to start a firm. Increasing inequalitywill raise the productivity of capital bymaking it possible for some firms to pay thefixed cost. Because there are also diminish-ing returns, however, there will be a point atwhich any further increase in overallinequality would be counterproductive.

More generally, the effect of inequalitywill depend on the shape of the productionfunction, and the size of the investmentpotential of the average person relative tothe fixed cost. Obviously, the issue gets evenmore complicated if different firms have dif-ferent production functions and if produc-tivity is correlated with the owner’s wealth(as it might be if the owner’s education is animportant input into production and richerpeople tend to be more educated).

Several authors have tried to look for asystematic relation in cross-country databetween inequality and growth (presumablywhat investment is meant to achieve). Alengthy body of literature30 estimated a long-run equation, with growth between 1990 and1960, for example, regressed on income in1960, a set of control variables, and inequal-ity in 1960. Estimating these equationstended to generate negative coefficients forinequality. But there are obvious concernsabout whether such a relation could bedriven entirely by omitted variables. Toaddress this problem, Li and Zou (1998),Forbes (2000), and others used the timeseries dimension of the Deininger and Squiredata set to look (effectively) at the effect ofshort-run changes in inequality on changesin growth.31 The results change rather dra-matically: the coefficient of inequality in thisspecification is positive and significant.

A recent review paper by Voitchovsky(2004) concludes that both these effects arequite robust. Most studies that look at thecross-sectional relationship between inequal-ity and subsequent growth over a relativelylong period in cross-country data, and espe-cially those that use measures of assetinequality, find a negative relationship, oftensignificant.32 By contrast, most studies thatlook at the relationship between changes in

inequality and changes in growth, includingseveral studies that do the analysis at the sub-national level within the same country, find apositive effect.

Both Banerjee and Duflo (2003) andVoitchovsky (2004) conclude that there is noreason to give one of these sets of results pri-ority over the other. Indeed, both could beright. For example, in the short run, policiesthat allow large cuts in real wages mightencourage investment, but in the long run,the consequent increase in poverty mightmake it harder for the population to maintainits human capital. Or both could be wrong.Most important among the many reasons forboth the cross-sectional and the time seriesevidence to be misleading are the following:the possibility of a nonlinear relationshipbetween inequality and growth, problemswith comparability of cross-country data,and the difficult question of identifying thedirection of causality when both variables arelikely to influence one another.

This lack of clear-cut results is perhaps dis-appointing, but it is worth emphasizing thatour focus here has been on redressing spe-cific inequalities in productive opportunitiesrather than some overall measure of inequal-ity. Despite the great attention devoted to thequestion of a systematic relationship betweenoverall inequality and growth at the countrylevel, the body of evidence remains uncon-vincing. But there clearly are situations inwhich there is a strong presumption thatreducing a specific inequality would promotebetter investment.

One such example comes from Opera-tion Barga, a tenancy reform in the Indianstate of West Bengal in the late 1970s and1980s. It has been known, at least since thework of the great Victorian economistAlfred Marshall, that sharecropping pro-vides poor incentives and discourageseffort. In such an environment, a govern-ment intervention that forces the landlordsto give their sharecroppers a higher share ofthe output than the market would givethem should increase effort and productiv-ity. This is exactly what happened in WestBengal, India, when a Left Front govern-ment came to power in 1977. The tenant’sshare of output was set at a minimum of 75percent as long as the tenant provided all


inputs. In addition, the tenant was guaran-teed a large measure of security of tenure,which may have encouraged him or her toundertake more long-term investments onthe land. Survey evidence shows a substan-tial increase in both the security of tenureand the share of output going to the share-cropper. The fact that the implementationof this reform was bureaucratically driven,and proceeded at different speeds in differ-ent areas, suggests the possibility of usingvariation in the implementation of thereform to evaluate its impact. The evidencesuggests that there was a 62 percent increasein the productivity of the land.33

A different program, also promotingequity and efficiency, had to do with redress-ing the effects of intrafamily inequality. Along line of research claims that income andexpenditures are often controlled by themale members of the family and that thisleads to underinvestment, especially in thehealth and education of girls. One fallout ofdismantling the apartheid regime in SouthAfrica was the expansion of the SouthAfrican social pension program to the blackpopulation. Pension entitlements wouldaccrue to elderly males and females, andmany older women living alone were entitledto receive the benefit. In many cases, childrenof very poor parents were sent to live withgrandparents who began to receive thesepensions. Duflo (2003) compared the impactof these new transfers on the nutrition ofchildren living with their grandparents, sepa-rately for households in which the pensionwas given to the grandmother and those inwhich it was assigned to a grandfather.

For children born before the expansion,in 1990 and 1991, height-for-age was slightlylower in families in which the grandmotherwould eventually get the pension. For chil-dren born after the expansion, in 1992 and1993, the children are significantly taller(except for the newborns) in those families.There is no difference between noneligiblefamilies and families in which pension

money goes to the grandfather. (Boys areessentially unaffected.) The estimates sug-gest that receipt of the pension (which wasabout twice the per capita income amongblacks) was enough to help girls bridge halfthe gap in height-for-age between SouthAfrican and American children.

These examples show that it is possible toenhance both equity and efficiency simulta-neously. Judicious redistribution—of incometo grandmothers, of power to poor womenfarmers, of credit to entrepreneurs in smallfirms—can increase the productivity ofresources, such as land, human capital, andphysical capital. If markets fail, resources donot always flow to where their return is great-est, particularly if that happens to be in proj-ects run by people with limited wealth orinfluence. Careful microeconomic case studyevidence, some of which was summarized inthis chapter, suggests that certain forms ofredistribution can reduce waste and con-tribute to a better use of resources, while alsoreducing inequality of opportunity. In fact, itenhances efficiency precisely because itreduces inequality of opportunity.

This is not to say that one cannot easilyimagine certain types of redistribution thathurt efficiency. But given the near universal-ity of market failures and underinvestmentin poor countries, it should be possible, witha combination of good research and carefulthinking, to identify opportunities for redi-recting resources to poorer people who arein a position to make good use of them.

In making the case for improvementsin equity that are also efficiency-enhanc-ing, this chapter used mainly micro-economic evidence on markets, wealth,and agency of individuals. The next chap-ter uses a different set of historical, macro-economic, and institutional evidence toargue that complex historical processes,combined with inequalities in influence andpower, may lead to bad political and eco-nomic institutions, which severely impairthe development of poor countries.


new elections were held, with a victory forthe left. The threat of more radical policiesprompted a military uprising, supported bythe landed classes, much of the bourgeoisie,and the church. Spain became engulfed in athree-year civil war. The social polarizationlocked Spain into a zero-sum gain over thedistribution of wealth. There was littlepolitical space for compromise or reformistsolutions. The “haves” opposed all attemptsat even minimalist reform. The “have-nots”wanted radical change, not gradual reform.

The Franco regime:from autarky to growthWith the defeat of the Republican govern-ment by the Nationalist army in 1939, Gen-eralissimo Francisco Franco established anauthoritarian regime that lasted until hisdeath in 1975. The destruction caused by thecivil war depressed the Spanish economy. Percapita income fell to its 1900 level and didnot reach its 1918 level until 1950. The pro-portion of the active population in industrydeclined to 22 percent in 1940 (the level in1920) and the share of employed in agricul-ture rose above 50 percent. Growth averagedonly 1.2 percent a year in the 1940s.

Spain’s economic recovery was ham-pered, above all, by the autarkic and statistpolicies of the Franco regime. Inspired bythe corporatist ideologies of Italian Fascismand German Nazism, Franco’s regime gen-eralized a system of price controls andrationing and regulated foreign tradethrough quantitative controls. This inter-ventionist strategy extended to the laborand housing market. To quell one of themain forces that opposed the military insur-rection, Franco outlawed any independentlabor unions. Instead, workers and employ-ers had to affiliate in a national trade unionorganization. This repressive stance was“compensated” by strict labor legislationthat made it hard for employers to dismissworkers or to hire them through temporary

Before the civil war:social and economic polarizationUntil the second half of the twentieth cen-tury, Spanish contemporary history was atale of political and economic failure. Aftera period of territorial expansion and Euro-pean hegemony in the early modern ages,Spain lapsed into economic decline andcultural stagnation in the following cen-turies. During most of the nineteenth cen-tury, its industrial takeoff was blocked bypolitical instability, inefficient legal institu-tions, substantial inequalities, and a poorlyeducated population. In 1929, per capitaincome was $3,000 (in 1990 dollars)—two-fifths that of Britain and less than two-thirds that of France.

Spain was polarized by entrenched socialand economic inequalities. In a country stilleminently agrarian, the distribution of landwas very unequal. About 1 percent of theholdings occupied 50 percent of the land.Educational attainment remained low,strictly linked to circumstances of birth.Social mobility was almost nonexistent.Except for Catalonia and the Basque coun-try, which industrialized in the nineteenthcentury, Spain lacked a large middle class.

Against this backdrop of relative stagna-tion and high inequality, democratic insti-tutions were introduced in 1931—only thesecond time in Spanish history. They didnot last long. The brief democratic period(1931–36) was characterized by huge politi-cal instability and social agitation. The firstRepublican government pushed ahead witha strong reformist program: separation ofchurch and state; a single system of stateschools and a goal to universalize educa-tion; a process of land reform; a law todecentralize political power to Catalonia;and stepped up efforts to reform the army.

These reforms elicited a strong reactionfrom the right, which came to power in1933 and quickly moved to halt them. Twoand a half years later, in the spring of 1936,

contracts. Emphasis on permanent jobs andcheap housing was seen as a substitute forthe lack of direct social policies, an attemptby the regime to win legitimacy.

In the late 1950s, Spain eventuallymoved to break with this interventionistsystem. An acute political crisis—associatedwith a wave of strikes, an economic re-cession, and severe balance-of-paymentcrisis—led the government to adopt a stabi-lization plan in March 1959. In addition tofiscal and monetary restraint, the planincluded wide-ranging measures to liberal-ize the economy. It was an outright success.From 1960 to the outbreak of the first oilcrisis, output expanded at an average annualrate of more than 7 percent with very littleinteryear volatility. Per capita incomealmost tripled from about $3,000 (in 1990dollars) to $8,500 in 15 years. Productivitygrowth averaged 6 percent.

The transformation of the Spanish econ-omy led to significant structural changes inSpanish society. The combination of eco-nomic growth, industrial expansion andinternal migration produced a substantialdecline in the levels of interregionalinequality (from a standard deviation in percapita income of 0.37 in 1955 to 0.27 in1973). Interhousehold inequality alsodeclined considerably: the Gini coefficientfor wages and salaries of employees (agrar-ian and industrial) declined from 0.29 in1964 to 0.23 in 1973; the Gini coefficient forhousehold income fell from 0.39 in 1964 to0.36 in 1974. The income share of the threecentral deciles went up from about 51 per-cent to 59 percent in that decade.

Still, significant social and economicinequities remained. Although the illiteracyrate had fallen to 10 percent by 1970, only 6percent of the population had completedsecondary studies. Wages remained damp-ened by repressive labor institutions. Taxa-tion and public spending were low, andredistributive social programs nonexistent.

Equity and development in the Spanish transition to democracyIn the last half century, Spain has gone from authoritarianism and underdevelopment to democracy and wealth. Spain’shistory illustrates how the distribution of political agency and economic assets greatly influences the policy choices avail-able to a society. The fundamentals of economic and political structure influence and constrain the choices. But theprocess is not deterministic: political agency and policies can shift the underlying fundamentals (as happened in Spain inthe 1960s and 1970s) and open the space for new choices.

Spainf o c u s 3 o n

the 1960s and 1970s. Rapid industrializa-tion and urbanization deflated past con-flicts around the distribution of land. Theexpansion of literacy and the increase inproductivity and incomes generated a largemiddle class. Sustained growth defusedsocial conflict with the credible promise ofhigher incomes and more social mobility.In short, Spain had overcome the zero-sumgame it had been locked in for the past cen-tury and a half.

Economic growth resulted in a differenteconomic structure and better distribu-tional outcomes, supporting a swift andsuccessful transition to democracy. In turn,the transition to democracy changed therole and size of the public sector.

Democratization reinforced social de-mands for progressive and redistributivepolicies—especially for public infrastruc-ture, and education, health, and socialprograms. In 1979 more than 70 percentof Spaniards agreed with the statementthat “the distribution of wealth in thiscountry is totally unjust.” In 10 years,social expenditure almost doubled toreach 80 percent of the European average.Public expenditure in education steadilyincreased from 2 percent of GDP in 1975to 4.5 percent in 1995. By 2001 almost 50percent of the population had completed

Transitioning to democracy and building the welfare stateFollowing Franco’s death in 1975, KingJuan Carlos became the Spanish head ofstate. He immediately launched a process ofpolitical change. Employing the legal mech-anisms put in place by the very technocraticgeneration that had reformed the economyin the early 1960s, as well as pointing towide popular support for democracy, hesecured the consent of the old FrancoistCortes to establish a truly democratic par-liament elected through direct, competitiveelections.

The political reform was ratified withoverwhelming popular support in a ref-erendum in December 1976. Althoughconducted in a climate of uncertainty, par-ticularly over the reaction of the army andthe extent to which terrorist violence orlabor mobilization could disrupt the negoti-ations, democratic elections were held inJune 1977. After protracted negotiations, anew constitution was approved in 1978 withthe support of all parliamentary groups. Toreinforce the political pact in parliament,the government also struck a wide eco-nomic and social deal with employers andtrade unions that same year.

Spain’s democratization was rooted inthe new economic and social conditions of

secondary education—10 times morethan in the mid-1970s. An ambitious pub-lic investment program tripled the publichighway network, revamped and expandedmetropolitan transportation, and modern-ized the railroad system.

Spain’s transition to democracy and theresulting expansion of its welfare stateshows how a mutually reinforcing packageof policy and institutional choices leadingto greater equity helped underpin thedevelopment and modernization of theSpanish economy and its integration intothe European Union. It illustrates howpolitical and economic structures shape thepossibilities for policy choice, a theme ofchapter 6. But it also illustrates that specificpolicy choices matter—across social sec-tors, infrastructure, the workings of mar-kets, and international integration—andthat there can be important complementar-ities for both equity and dynamic growth,notably between greater social provisioningand greater reliance on markets. This takesus to the issue of practical policy design, thecentral theme of part III of this report.

Sources: Synthesized from Boix (2005), with references toGunther, Montero, and Botella (2004); North and Thomas(1973); and Revenga (1991).


107

Product, land, labor, and capital markets arecrucial for the allocation of resources anddevelopment. Market institutions, however,exist and function in the context of a wholeset of nonmarket and political institutions.The nature of these other institutions—andthe way they function—are influenced byinequalities in the political and social realm.

The most obvious of these other institu-tions are those that define and enforce prop-erty rights and contracts. People will notinvest if property rights are not well definedand enforced, or if they believe that the con-tracts they write will not be honored or thatcourts of law will not be fair. The state mustalso provide a whole set of other inputs apartfrom social order and fair contract enforce-ment. These include various types of publicservices and regulations. Lying behind well-functioning markets are legal systems, judges,policemen, and, ultimately, social groups andpoliticians.

This chapter considers the circumstancesand processes for creating institutions thatpromote prosperity. These circumstances areclosely related to the concerns of this report.In essence, societies that create institutions togenerate sustained prosperity are equitable inimportant ways. Because talent and ideas arewidely distributed in the population, it is cru-cial that the property of all people is secureand that there is equality before the law forall, not just for some. Predetermined circum-stances should not constrain anyone’s inno-vation or investment opportunities. Thisimplies that a good institutional environmentwill not block entry into new lines of businessand that the political system will provideaccess to services and public goods for all.Institutions must be equitable.

To take an extreme example, institutionswere severely inequitable in slave societies,

such as Haiti or Barbados in the eighteenthcentury. Even though property rights in landand people were well defined and even wellenforced (although subject to potential slaverebellions), most people had no propertyrights and were thus subject to expropriationby others, particularly their masters. For 95percent of society, there were no incentives toengage in socially desirable activities. A simi-lar, although somewhat less extreme, exampleof inequitable institutions is South Africaunder apartheid. Institutions there were goodfor the whites but left 80 percent of the popu-lation without incentives or opportunities toengage in economically productive activities.

The distribution of power and institutional quality:circles vicious and virtuous How do societies develop equitable non-market institutions? First, there must besufficient political equality—equality inaccess to the political system and in the dis-tribution of political power, political rights,and influence.

Poor institutions will emerge and persistin societies when power is concentrated inthe hands of a narrow group or an elite. Suchan elite may grant property rights to itself, butthe property rights of most citizens will beunstable. There may be equality before thelaw for a particular elite group, but not for themajority of people. Government policies mayfavor such an elite, granting them rents andmonopolies, but most people will beexcluded from entering profitable lines ofbusiness. The education system may investheavily in the children of such elites, but mostwill be excluded.

Many things determine the distribution ofpolitical power in society—the constitution,

Equity, institutions,and the development process

6c h a p t e r

the nature of checks and balances, and theability of different groups to solve collectiveaction problems. But economic inequalityoften underpins political inequality. In a soci-ety with large inequalities of assets andincomes, the rich will tend to have moreinfluence and an advantage in adapting anddistorting institutions to their benefit.

Because the distribution of power,through its impact on institutions, helps todetermine the distribution of income, thepossibility of vicious and virtuous circles isclear. A society with greater equality of con-trol over assets and incomes will tend tohave a more equal distribution of politicalpower. It will therefore tend to have institu-tions that generate equality of opportunityfor the broad mass of citizens. This will tendto spread rewards and incomes widely,thereby reinforcing the initial distribution ofincomes. In contrast, a society with greaterinequality of assets and incomes will tend tohave a less egalitarian distribution of powerand worse institutions, which tend to repro-duce the initial conditions.

The evidence in this chapter suggests thatthe first type of society will tend to be moreprosperous. We argue that societies prosper-

ous today are so because they have developedmore egalitarian distributions of politicalpower, while poor societies often suffer fromunbalanced distributions. We also considerhow some societies made the transition fromone equilibrium to the other.

Because institutions have distributionaleffects, conflict arises naturally. One set ofinstitutions will benefit some people, whileanother will benefit different people. Thus,there will be incentives for people to controlpower to create or keep the institutions thatbenefit them and to avoid or weaken theinstitutions that disadvantage them. If thegroups in conflict are defined along ascriptivelines, such as ethnicity, then this may induce amore severe form of conflict than whengroups are defined along other lines, or whenthere are cross-cutting cleavages. More polar-ized conflict seems to be an independentforce leading to bad institutions that can helpto explain the relatively weak performance insome societies (discussed below in a compar-ison between Guyana and Mauritius).

Political equality also matters for the qual-ity of public policy. The basic role of the stateis to provide public services. But politicianshave the correct incentives to provide publicservices only when they have to appeal to thebroad mass of citizens to attain power. Ifthey can win power with a small number ofkey supporters, or with few votes, they willtend to be clientelistic and more inclined tobuy votes or make individual exchanges ofpatronage for support without providing thegoods and services critical to raising the massof people out of poverty.

Some simple patterns in the cross-countrydata show that more egalitarian distributionsof political power and income are associatedwith sustained and enduring prosperity. Fig-ure 6.1 indicated that more secure propertyrights are associated with higher incomes.Crucially, however, better institutions andsecure property rights are associated withgreater political equality.

Although there is no perfect way ofmeasuring political equality, protectionagainst expropriation risk is highly corre-lated with measures of democracy andmeasures of “constraints on the executive”from the Polity IV database. This secondvariable is designed to capture the extent to


5

6

7

8

9

10

GDP per capita, PPP in 1995 (log)11

Average protection against risk ofexpropriation, 1985–95

3 4 5 6 7 8 9 10

AGO

ARG

AUS

BFA BGD

BHS

BOL

BRABWA

CAN

CHL

CIVCMRCOG

COLCRIDOM

DZAECU EGY

GAB

GHA

GIN

GMB

GNB

GTMGUY

HND

HTI

IDN

IND

JAM

KEN

LKA MAR

MDG

MEX

MLI

MLT

MOZ MWI

MYS

NER NGA

NIC

NZL

PAK

PANPER

PHLPNG

PRY

SDNSEN

SGP

SLE

SLV

STPTGO

TTO

TUN

TZA

UGA

URY

USA

VEN

VNM

ZAF

ZAR

ZMB

ZWE

AUTBEL

BGR

BHR

CHE

CHN

CZE

DNK

ESP

ETHETH

FINFRAGBR

GRC

HUN

IRL

IRN

ISLISR

ITA

JOR

JPN

KOR

KWT

LUX

MNG

NLDNOROMN

POL

PR T

QAT

ROM

RUS

SAU

SWE

SYR

THATUR

YEM

Figure 6.1 Countries with more secure property rights have higher average incomes

Sources: Political Risk Services, International Country Risk Guide (ICRG) and World Bank database.Note: The figure shows the relationship between GDP per capita in 1995 and a measure of the security of propertyrights, “protection against expropriation risk,” averaged over the period 1985 to 1995. The data on institutions comefrom Political Risk Services, a private company that assesses the risk that investments will be expropriated in differentcountries. These data, first used by Knack and Keefer (1995) and subsequently by Hall and Jones (1999) and Acemoglu,Johnson, and Robinson (2001, 2002a, 2004), are imperfect as a measure of the relevant institutions because they pertainto investments by foreigners only. Even so, they seem in practice to capture how stable property rights are in general.The findings are robust to using other available measures of related institutions.

which those who control political power areconstrained or checked by others. The typesof checks and balances and separation ofpowers written into the U.S. Constitutionare classic examples of such constraints.There is a negative correlation between con-straints on the executive and the Gini coeffi-cient of income distribution.

The simple correlations suggest comple-mentarities between a relatively egalitariandistribution of political power, good institu-tions, and prosperity, and a relatively egali-tarian distribution of economic resources.The correlations are consistent with manydifferent causal stories, but recent researchsuggests that one can tell a causal storyabout this data along exactly the lines we aresuggesting, which the rest of this chapterdiscusses. The different evolutions of bank-ing systems in Mexico and the United Statesin the nineteenth century provide a goodexample of the sort of historic argument werely on (box 6.1).

Institutions and politicalinequality matter fordevelopment: historical evidenceFigure 6.1 showed the relationship betweensecurity of property and prosperity for thewhole world, but to interpret this causally weneed to find a source of variation in institu-tions. Doing this is not easy, but Acemoglu,Johnson, and Robinson (2001) provide apartial answer. They show that the samebasic pattern holds for a smaller sample ofcountries—those colonized by Europeansafter 1492. Indeed, colonization of much ofthe world by Europeans provides somethingof a large natural experiment.

Beginning in the early fifteenth centuryand massively intensifying after 1492, Euro-peans conquered many other nations. Colo-nization transformed the institutions inmany diverse lands conquered or controlledby Europeans. Most important, Europeanscreated very different sets of institutions indifferent parts of their global empire, asexemplified most sharply by the contrastbetween the institutions in the northeast ofAmerica and those in the plantation societiesof the Caribbean. This experience persua-sively establishes the central role of institu-

tions in development. It also provides fairlyclear-cut evidence to support our conjecturesabout the joint evolution of prosperity andpolitical and economic equality.

Colonial origins of contemporaryinstitutionsAcemoglu, Johnson, and Robinson, buildingon the research of Engerman and Sokoloff(1997), explain that Europeans created goodinstitutions in some colonies, particularly theUnited States, Canada, and Australasia (what

Equity, institutions, and the development process 109

Much recent work on growth and develop-ment has focused on financial and capitalmarkets. A central issue is to understandwhy financial systems differ. For example,studies of the development of banking inthe United States in the nineteenth centurydemonstrate a rapid expansion of financialintermediation, which most scholars see asa crucial facilitator of the economy’s rapidgrowth and industrialization. Haber (2001)investigated the development of banks inthe nineteenth century in Mexico and theUnited States. He shows that “Mexico had aseries of segmented monopolies that wereawarded to a group of insiders” (24). In 1910“the United States had roughly 25,000banks and a highly competitive marketstructure; Mexico had 42 banks, two ofwhich controlled 60 percent of total bank-ing assets, and virtually none of which actu-ally competed with another bank.”

Why this huge difference? The relevanttechnology was certainly widely available,and it is difficult to see why the varioustypes of moral hazard or adverse selectionconnected with financial intermediationshould have limited the expansion of banksin Mexico but not the United States. Indeed,Haber shows when the U.S. Constitutionwas put into effect in 1789, the structure ofU.S. banking looked remarkably like thatarising later in Mexico. State governments,stripped of revenues by the Constitution,started banks as a way to generate tax rev-enues and restricted entry to generaterents.Yet this system did not last becausestates began competing among themselvesfor investment and migrants. As Haber(2001) puts it,

The pressure to hold population and business inthe state was reinforced by a second, related, fac-tor: the broadening of the suffrage. By the 1840s,most states had dropped all property and literacyrequirements, and by 1850 virtually all states . . .

had done so.The broadening of the suffrage, how-ever, served to undermine the political coalitionsthat supported restrictions on the number of bankcharters.That is, it created a second source of polit-ical competition—competition within states overwho would hold office and the policies they wouldenact (10).

The situation was very different in Mex-ico. After 50 years of endemic political insta-bility, the country became unified under thehighly centralized 40-year dictatorship ofPorfirio Díaz until the revolution in 1910.

In Haber’s argument, politicalinstitutions in the United States allocatedpolitical power to people who wantedaccess to credit and loans. As a result, theyforced state governments to allow freecompetitive entry into banking. In Mexico,political institutions were very different.There were no competing federal states,and suffrage was highly restrictive. As aresult, the central government grantedmonopoly rights to banks, which restrictedcredit to maximize profits.The granting ofmonopolies turned out to be a rational wayfor the government to raise revenue andredistribute rents to political supporters(North 1981).

Haber (2001) documents that marketregulation was not aimed at solving marketfailures, and it is precisely during this periodthat the huge economic gap between theUnited States and Mexico opened (onwhich see Coatsworth 1993, Engerman andSokoloff 1997). Haber and Maurer 2004examined in detail how the structure ofbanking influenced the Mexican textileindustry between 1880 and 1913.Theyshow that only firms with personal contactswith banks were able to get loans and thatsuch firms were less efficient. Even thougheconomic efficiency was hurt byregulations, those with political power wereable to sustain them.

B O X 6 . 1 Banking in the nineteenth century,Mexico and the United States

Crosby (1986) calls the neo-Europes), and badones in others (particularly in Latin Americaand Sub-Saharan Africa). These institutionshad a strong tendency to persist and thus,today, generate the results seen in figure 6.1.

Why did different institutions develop indifferent European colonies? The simplestanswer is that Europeans shaped the institu-tions in various colonies to benefit themselves.And because conditions and endowments dif-fered among colonies, Europeans con-sciously created different institutions. Thereare several important empirical regularitiesconnecting initial conditions to current out-comes. Of particular importance are initialpopulation density, the disease environ-

ment, and the factor endowments that influ-enced economic organization.1 There is astrong inverse relationship between popula-tion density in 1500 and current protectionagainst expropriation risk for former Euro-pean colonies (figure 6.2). And colonies withdisease environments that were worse forEuropean settlers also have worse institu-tions today (figure 6.3).

Other aspects of factor endowments aremore difficult to measure directly, but Enger-man and Sokoloff (1997) point out thatwhere the climate and soils were suitable forcrops such as sugarcane—which could begrown on large plantations with slave labor,such as northeastern Brazil—much worseinstitutions and more skewed distributions ofpolitical power evolved than in climateswhere wheat or other nonplantation cropscould be grown.

Why did Europeans introduce betterinstitutions in previously relatively unset-tled and healthy areas than in previouslydensely settled and unhealthy areas? Howdid factor endowments influence institu-tions? Europeans were more likely to intro-duce or maintain bad institutions wherethere were a lot of resources and rents toextract—gold, silver, and, most important,people to provide the labor. In places with alarge indigenous population, Europeanscould exploit the population through taxes,tributes, or employment as forced labor inmines or plantations. And where plantationcrops could be profitably grown, slave-based societies emerged. These types ofcolonization were incompatible with insti-tutions providing economic or civil rights orequality of opportunity to the majority ofthe population. So, a more developed civi-lization with a denser population structure,and particular climatic and agricultural con-ditions, made it more profitable for theEuropeans to introduce bad institutions.

In contrast, in places with little to extract,where plantation agriculture was not prof-itable, and in sparsely settled places wherethe Europeans became the majority of thepopulation, it was in their interests to intro-duce much better institutions. In addition,the disease environments differed markedlyamong the colonies, with obvious conse-quences for the attractiveness of European


Average protection against risk of expropriation, 1985–95

Population density in 1500 (log)–4 –3 –2 –1 0 1 2 3

3

4

5

6

7

8

9

10

4 5

AGO

ARG

AUS

BFA

BGD

BHS

BOL

BRABWA

CAN

CHL

CIV

CMR

COG

COLCRI

DOMDZAECU

EGY

GAB

GHAGIN

GMB

GNB

GTM

GUY

HND

HTI

IDN

IND

JAM

KEN

LBY

LKA

MAR

MDG

MEX

MLI

MLT

MMR

MOZMWI

MYS

NER

NGANIC

NZL

PPAKPANPERPHL

PNGPR Y

SDN

SEN

SGP

SLE

SLV

SOM

TGO

TTO

TUNTZA

UGA

URA

USA

VEN

VNM

ZAF

ZAR

ZMB

ZWE

Figure 6.2 Low population density in 1500 is associated with a lower risk of expropriation today

Source: Political Risk Services, International Country Risk Guide (ICRG) and Acemoglu, Johnson, and Robinson (2002).

3

4

5

6

7

8

9

Average protection against risk of expropriation, 1985–95

10

Settler mortality (log)2 3 4 5 6 7 8

AGO

ARG

AUS

BFA

BGD

BHS

BOL

BRA

CAN

CHL

CIV

CMR

COG

COLCRI

DOMDZAECUEGY

GAB

GHAGIN

GMB

GNB

GTM

GUY

HND

HTI

IDN

IND

JAM

KENLKA

MAR

MDG

MEX

MLI

MLT

MMR

MYS

NER

NGANIC

NZL

PAKPANPER

PNG

PRY

SDN

SEN

SGP

SLE

SLV

TGO

TTO

TUNTZA

UGA

URY

USA

VEN

VNMZAF

ZAR

Figure 6.3 Worse environments for European settlers are associated with worse institutions today

Source: Political Risk Services, International Country Risk Guide (ICRG) and Acemoglu, Johnson, and Robinson (2002).

settlement. When Europeans settled, theyestablished institutions under which theythemselves had to live.

This research suggests that most of thegap in per capita income between rich andpoor countries today is due to differences ininstitutions. More precisely, if one takes twotypical countries—in the sense that theyboth lie on the regression line—with highand low expropriation risk, such as Nigeriaand Chile, almost the entire difference inincome per capita between them can beexplained by the differences in the histori-cally shaped measure of the security ofproperty rights.2 The research also pre-sented regression evidence showing thatonce the effect of institutions on GDP percapita is properly controlled for, geographicvariables—such as latitude, whether or nota country is landlocked, the current diseaseenvironment—have no explanatory powerfor current prosperity.

Different types of societies thus devel-oped in different colonies with radicallydifferent implications for subsequent devel-opment. Crucially, the societies that emergedin the neo-Europes had distributions of eco-nomic resources and political power thatwere much broader. And they placed con-straints on the exercise of political powerand the ability of elites to adopt policiesfavorable to themselves but deleterious forsociety (figure 6.4).

Development and inequality in the Americas: A case study in colonial originsThe colonization of Latin America beganwith the discovery of the “Indies” by Colum-bus in 1492, the assault on Mexico by Cortésafter 1519, and the conquest of Peru by Piz-zaro after 1532. From the beginning, theSpanish were interested in the extraction ofgold and silver, and later in taking tributeand raising taxes. The colonial societies thatemerged were authoritarian, based on thepolitical power of a small Spanish elite whocreated a set of institutions to extract wealthfrom the indigenous population.

After Pizzaro conquered Peru, he imposedinstitutions to extract rents from the newlyconquered Indians. The main such institu-tions were the encomienda (which gave Span-

ish conquistadors the right to Amerindianlabor),3 the mita (a system of forced laborused in the mines), and the repartimiento (theforced sale of goods to Indians, typically athighly inflated prices). Pizzaro created 480encomenderos, under whose care the entireIndian population was placed. In othercolonies the situation was similar. Forinstance, in the territory of modern Colom-bia, there were about 900 encomenderos.4

The encomienda did not last for long inall parts of the empire because the SpanishCrown attempted to curtail it by the end ofthe sixteenth century. But the mita (fromthe Quechua word mit’a, meaning “turn”)became a central institution until inde-pendence, and forced labor lasted farbeyond this in most of Latin America (until1945 in Guatemala). The effects of theencomienda also persisted because the con-centration of political power that it wasassociated with led to the emergence oflarge landed estates.5 The feasibility andattraction of this type of economic systemwas determined by the higher populationdensities of indigenous people in manyparts of the Spanish empire and the extentto which such societies had already devel-oped into “complex societies.”6

Other institutions were designed to rein-force this system. For instance, indigenous


0.0

Constraint on the executive at independence1.0

0.8

Settler mortality (log)2 3 4 5 6 7 8

0.6

0.4

0.2

AFG

AGO

ARG

AUS

BDI

BEN

BFA

BGD

BOL

BRA

CAF

CANCHL

CIV

CMR

COG

COL

CRI

DJI

DOM

DZA

ECU

EGY

FJI

GHA

GIN

GMB

GNB

GTMGUY

HND

HTI

IDN

INDJAM

KENLAO

LKA

MAR

MDG

MEX

MLI

MMRMRT

MUS

MYSNER

NGA

NIC

NZL

PAK

PAN

PER

PNG

PRY

RWA

SDN

SENSGP

SLE

SLV

TCDTGO

TTO

TUN

TZA

UGA

URYUSA

VEN

VNM

ZAF

ZAR

Figure 6.4 A worse environment for settlers is associated with fewer constraints on theexecutive at independence

Source: Acemoglu, Johnson, and Robinson (2002a). The analysis indicates that that the same factors that gave rise togood institutions gave rise to a more egalitarian distribution of power. Without some measure of voice, it is impossiblefor a person’s property rights to be guaranteed or for them to have real access to the legal system to make sure thatcontracts are honored. A more egalitarian distribution of political power is also associated with a more egalitarian dis-tribution of economic resources. To get a better understanding of the mechanisms, we need to look further into histori-cal analysis.

people were not allowed to give testimonyin some cases, and in others the testimonyof 10 indigenous people was equal to that of1 Spaniard.7 Although indigenous peopledid use the legal system to challenge aspectsof colonial rule, they could not alter themain parameters of the system. In addition,the Spanish Crown created a complex webof mercantilistic policies and monopoliesfrom salt to gunpowder, from tobacco toalcohol and playing cards, to raise revenuesfor the state.

Spanish colonies that had small popula-tions of Amerindians, such as Costa Rica,Argentina, or Uruguay, seem to have fol-lowed different paths of institutional devel-opment. The sharp contrasts along manyinstitutional dimensions between CostaRica and Guatemala (where populationdensity was greater) have been much stud-ied. Although the formal political institu-tions of the Spanish empire were the sameeverywhere, the way they functioneddepended on the local conditions.8

The institutions that emerged in themain Spanish colonies greatly benefited theSpanish crown and the Spanish settler elite,but they did not promote prosperity inLatin America. Most of the population hadno property rights, nor incentives to entersocially desirable occupations or to invest.Europeans developed coercive regimesmonopolizing military and political powerand respecting few constraints on theirpower (unless imposed by the mothercountry in Europe).9

In North America, the initial attempts atcolonization were also based on economicmotives. British colonies were founded bysuch entities as the Virginia Company andthe Providence Island Company with the aimof profits. The model was not so differentfrom that of the Spanish or Portuguese (a sys-tem that other British colonizing entities,such as the East India Company, used to greateffect). Yet these companies made no money.Indeed, both the Virginia Company and theProvidence Island Company went bankrupt.Because of the absence of a large indigenouspopulation and complex societies, a colonialmodel involving the exploitation of indige-nous labor and tribute systems was simplynot feasible in these places.

Historical accounts show that initial con-ditions had a large impact on the institutionsthat the settlers built. Because there was lowpopulation density and no way to extractresources from indigenous peoples, earlycommercial developments had to importBritish labor. And, relative to much of thecolonial world, the disease environment wasbenign, stimulating settlement. Indeed, thePilgrim fathers decided to migrate to theUnited States rather than Guyana because ofthe high mortality rates in Guyana.10 Butthese same conditions made it impossible toprofitably exploit labor, whose bargainingpower forced elites to extend political rightsand create equal access to land and the law.These forces were reinforced by the fact thatplantation agriculture and slavery were notprofitable, at least in the northern UnitedStates and Canada.

These colonies ultimately provided ac-cess to land to a broad cross-section of soci-ety and the legal system became fairlyimpartial, ensuring secure property rightsfor smallholders and potential investors.The new institutions made investment pos-sible through financial development andsecure contracting and business relation-ships. Underpinning these institutions werefairly representative political institutionsand a fairly egalitarian distribution ofresources. As in Latin America, there was asynergy between economic and politicalinstitutions, but this time it was virtuous,not vicious. Institutions giving and protect-ing property rights for the mass of peopleand institutions of democratic politicscomplemented each other, ensuring anenvironment conducive to investment andeconomic progress.

Representative political institutions inVirginia were a direct result of the authori-ties realizing that, because of the differentconditions, the colonization strategy thatworked in Peru would not work in theUnited States. Virginia had many competingand fragmented tribes, not a large centraltribal empire. It had no gold or silver, andthe Indians, not used to paying tribute orengaging in forced labor, would not work.So, the settlers of Jamestown starved.11 Inresponse to these early failures, the VirginiaCompany tried various incentive schemes,


including a highly punitive, almost penal,effort to make money. Such efforts quicklycollapsed, however, and by 1619 the Com-pany had created an unusually representa-tive set of institutions for that era: a generalassembly with adult male suffrage.

The early history of the United Statesshows a possible path to good institutions.Early attempts to create an oligarchic soci-ety with close control of labor quicklycollapsed. What emerged instead was arelatively egalitarian society, with represen-tative institutions giving even the poorestcolonists access to the law and some politi-cal representation. This laid the basis foreconomic and social institutions thatunderpinned the takeoff of the UnitedStates in the nineteenth century and itsdivergence from the fortunes of much ofLatin America. Some countries with weakand unequal institutions have experiencedperiods of rapid growth, but these haveproved to be unsustainable over the longterm (box 6.2).

Institutions and politicalinequality matter for development:contemporary evidenceOur review of comparative history supportstwo conclusions. First, institutions, espe-cially those that underpin property rightsfor all and broad-based investment, have acausative influence on long-run develop-ment processes. And second, greater polit-ical equality can lay the basis for bettereconomic institutions. By greater politicalequality, we mean, in particular, checks onthe predatory behavior of political and eco-nomic elites, and the political need for thestate to be responsive to middle and poorerpopulation groups. The basis for greaterpolitical equalities is often associated withunderlying economic structures, althoughcausation can run both ways.

How does this perspective relate to thevariety of contemporary development expe-riences? It is consistent with the perspective


The elite had good investment opportunities inArgentina in the golden age from the 1870s tothe 1920s, in Czarist Russia in the decades lead-ing up to World War I, in Colombia in the halfcentury after 1900, and in the Côte d’Ivoire forthe first two decades after independence (Wid-ner 1993). Such situations are rarely sustainable,for three reasons. First, the possibilities for sus-tained growth are, by definition, limited becauseinstitutions exclude the majority of the popula-tion from effectively investing. Second, in therare situations in which elites manage to createarrangements so that they can benefit directlyfrom growth without the need to create goodinstitutions more generally, such arrangementstend to be fragile, vulnerable to shocks or crises.Third, bad institutions create power strugglesthat undermine growth, because they generatelarge rents for those who control power.

Consider the growth of Argentina in the halfcentury before 1930. After its independencefrom Spain in 1816, Argentina plunged into 50years of civil wars and conflicts over control ofthe country, mainly clashes between those incontrol of Buenos Aires and the littoral andthose in the interior.These conflicts abated afterthe 1853 constitution and the presidency of Bar-tolomé Mitre with a compromise between thePampas and the interior. Pampean mercantileand agrarian interests would be allowed to cre-

ate institutions to take advantage of the hugeeconomic opportunities emerging on worldmarkets, but the structure of the political rules,such as their overrepresentation in nationalpolitical institutions, guaranteed the interiorprovinces a large slice of the benefits (Samuelsand Snyder 2001).

Although the majority was excluded fromthe political system, the economy boomed withthe property rights of the Pampean elite guar-anteed. But the huge rents created by this sys-tem began to cause conflict. In the 1890s, theRadical Party emerged under Hipólito Yrigoyen,and after a series of revolts it was incorporatedinto the political system by the democratizingimpact of the Sáenz Peña Law in 1912.

Although Yrigoyen was elected president in1916, the traditional interests were confidentthat they could keep control of the polity andthe economy.They were mistaken. Significantchanges in the social structure had occurred,with rapid immigration from Europe, induced byeconomic success, and the associated urbaniza-tion.The vote share of the Conservativesdeclined rapidly and the prospect of a RadicalParty majority was a key factor behind the coupof 1930. Smith (1978) notes “this situation con-trasts sharply with that in Sweden and GreatBritain . . . where traditional elites continued todominate systems after the extension of

suffrage” (21). From this point onward politicalconflicts intensified, with a stream of coups andredemocratizations that lasted until 1983.Though among the richest countries in theworld in the 1920s, Argentina gradually slid backto being a developing country.

Argentina shows that, even with poor insti-tutions for political inclusion and conflict man-agement, growth is possible if elites have goodinvestment opportunities and can manage toforge compromises. But the booms eventuallyunravel. Even when elites, such as the agricul-turalists of the Argentine Pampas, face verygood investment opportunities, growth cannotbe sustained forever by agricultural exportbooms. Moreover, the rents created by badinstitutions create conflict without fundamen-tal balances of power in society. This meant thatdemocracy in Argentina after 1912 was unsta-ble. The unchecked power of PresidentYrigoyen in the 1920s induced a coup in 1930,as did that of Perón in the 1940s and in 1955and again in 1976 after his return from exile.Although temporary political solutions cansometimes ease conflict for a while, as they didin Argentina after 1853, in the absence ofbroader institutional inclusion, conflictultimately reemerges, undermining the incen-tives to investment.

B O X 6 . 2 Growth with poor institutions does not last

that institutions and governance are centralto a wide variety of development perform-ance, from growth to service delivery.12

While debate continues, an important thrustof this research has been to support the viewthat causation runs, at least in part, from bet-ter institutions to higher incomes, ratherthan the other way.13 What is additional tothis (ongoing) debate is the second part ofthe argument—that the nature and manage-ment of inequalities in power shapes the for-mation of institutions. Some cross-countryanalysis is suggestive: Rodrik (1999a) arguesthat the capacity of societies to manageadverse shocks—itself a crucial determinantof growth—depends on the depth of latentsocial conflict and the strength of conflictmanagement mechanisms.

To illustrate the argument, we continueto draw on comparative development expe-riences. We first look at East Asia, and thenlook at agricultural pricing polices inAfrica. We then examine in greater depththe comparative experience of Mauritiusand Guyana, countries that started withsimilar initial conditions, but then followedradically different development paths. Thisis also related to different experiences inmanaging polarization, which can be con-tributory factors for violent social conflict.

Shared growth in East Asia: the Republic of Korea, Taiwan(China), and IndonesiaElites may be forced by threats of social disor-der to promote the prosperity of most citi-zens. Indeed, societies that have a politicalnecessity to appeal to or appease middle andlower groups (initially the peasantry) cangrow substantially in the short run. Long-run prosperity, however, requires institution-alized, rather than contingent checks andbalances on elite power and capacities toadjust to changing circumstances. Theresponse of elites to social disturbancessometimes leads to solutions that perma-nently change the political equilibrium in abeneficial way, as may have happened withthe agrarian reforms in the Republic ofKorea and Taiwan, China, in the late 1940sand early 1950s. More often, however, thetransitory ability of citizens to act collectivelydissipates without elites having to propose

anything more than a transitory solution, asmay have been the case in Indonesia underthe New Order.

The rapid economic development of theRepublic of Korea after the mid-1960s wasnot due to a set of institutions put in placethrough a domestic balance of politicalpower. Instead, as in Indonesia under theNew Order regime, a precarious geopoliticalsituation, particularly after the rundown ofU.S. aid in the early 1960s, induced the Parkregime to create a pro-growth environ-ment.14 This at least led to a contingent com-mitment to good institutions, as it did underan authoritarian regime in Taiwan, China,where a fairly egalitarian distribution ofassets and incomes, perhaps eased the transi-tion in the 1990s toward democracy, agreater equality of political influence, andgood institutions. As in much of East Asia,there was a political necessity to deliverincome growth and services to the peasantry.

In Indonesia, Suharto’s New Order gov-ernment also recognized that economicgrowth was necessary to keep the regime inpower and that, to achieve this, good eco-nomic policies had to be in place. Thisinduced Suharto to delegate macroeconomicpolicy to technocrats and to respond to theoil booms wisely. It also led him to interveneto attempt to control corruption and excessesthat would put in jeopardy the underpin-nings of the regime.15

Yet this constraint, real though it was, atleast in the 1960s and 1970s, is only part ofthe story about Indonesian growth. Suhartomanaged to create a system that, while notintroducing good institutions, inducedinvestments and growth from which theregime could benefit. One of the secretsbehind this appears to have been the role ofSino-Indonesian businessmen, the cukongentrepreneurs. Many firms and businesseswere controlled by Indonesians of Chineseorigin who were very marginal politically.Suharto granted such businessmen mono-poly rights and placed members of the mil-itary and his supporters on their boards ofdirectors.16 Rock (2003) argues, “There islittle doubt that the . . . distortions in NewOrder microeconomic policies thwartedcompetition, rewarded cronies, and en-couraged substantial investment in uneco-


nomic projects” (10). Yet they also gener-ated wealth, economic growth, and rentsfor the regime. It was precisely the politicalmarginality of the cukong entrepreneursthat made them an attractive business part-ner for the regime.

The economic success of Indonesia after1966 elevated it into the class of an Asian“miracle economy.”17 The East Asia financialcrisis in 1997, however, exposed and exacer-bated Indonesia’s institutional weaknesses,plummeting the country into an economicand political crisis from which it is only nowbeginning to recover, doing so on the basis ofa new foundation of decentralization anddemocracy, which have progressively institu-tionalized greater relationships of accounta-bility between citizens and state. (See focus 4on Indonesia for a further discussion of therelationship between social and political con-text and policy choices.)

Agricultural pricing policies in AfricaAnother important example illustrating theconnections between institutions, the dis-tribution of political power and growthcomes from the seminal studies of priceregulation prices in agricultural markets inAfrica by Robert Bates.18 Bates (1981)demonstrated that poor agricultural per-formance in Ghana, Nigeria, and Zambiawas due to government-controlled market-ing boards systematically paying farmersprices much below world levels. The mar-keting board surpluses were given to thegovernment as a form of taxation. As aresult of this pernicious taxation, reachingup to 70 percent of the value of the crop inGhana in the 1970s, investment in agricul-ture collapsed, as did the output of cocoaand other crops. In poor countries with acomparative advantage in agriculture, thismeant negative rates of economic growth.

Why were resources extracted in this way?Although part of the motivation was to pro-mote industrialization, the main one was togenerate resources that could be either expro-priated or redistributed to maintain power.As Bates (1981) put it,

governments face a dilemma: urban unrest,which they cannot successfully eradicate

through co-optation or repression, poses aserious challenge to their interests . . . Theirresponse has been to try to appease urbaninterests not by offering higher money wagesbut by advocating policies aimed at reducingthe cost of living, and in particular the cost offood. Agricultural policy thus becomes a by-product of political relations between gov-ernments and urban constituents (33).

In contrast to the situation in Ghana,Nigeria, and Zambia, Bates (1981), Bates(1989) showed that agricultural policy inKenya over this period was much more pro-farmer. The difference was due to who con-trolled the marketing board. In Kenya, farm-ers were not smallholders, as they were inGhana, Nigeria, and Zambia, and concen-trated landownership made it much easierto act collectively. Moreover, farming wasimportant in the Kikuyu areas, an ethnicgroup closely related to the ruling politicalparty, the Kenya African National Union(KANU), under Jomo Kenyatta.19 Farmersin Kenya therefore formed a powerful lobbyand were able to guarantee themselves highprices. Even though the government ofKenya engaged in land reform after inde-pendence, Bates (1981) argued that—

80 percent of the former white highlands wereleft intact and . . . the government took elabo-rate measures to preserve the integrity of thelarge-scale farms . . . [which] readily combinein defense of their interests. One of the mostimportant collective efforts is the KenyaNational Farmer’s Union (KNFU) . . . Theorganization . . . is dominated by the large-scale farmers . . . [but] it can be argued thatthe KNFU helps to create a framework ofpublic policies that provides an economicenvironment favorable to all farmers (93–4).

Bates concluded that in Kenya “large farm-ers . . . have secured public policies that arehighly favorable by comparison to those inother nations” (95).

Bates demonstrated why economic poli-cies were better in Kenya than Ghana in the1960s and 1970s, but this advantage did notsurvive the coming to power of Daniel arapMoi in Kenya.20 The change in the ethnicbasis of the regime, from Kikuyu to Kalenjin,undermined the coalition that had sup-ported good agricultural policies, becausethe export farmers were not only large, but


also predominantly Kikuyu. As a result, eco-nomic performance declined precipitouslyin the 1980s and 1990s. The balance ofpower that sustained good policies in the1970s did not endure.

The contrasting experience of Mauritius and GuyanaMauritius and Guyana, in the 1960s, wereboth poor societies dominated by the pro-duction and export of sugarcane. They hadsimilar histories, factor endowments, socialand political cleavages, and institutions. Ifanything, Guyana, although slightly poorer,had better prospects, because of its proxim-ity to the large U.S. market. Yet Mauritiushas become one of the most dynamic andsuccessful (and equal) developing countries,industrializing and maintaining competitivedemocratic politics. Guyana slumped intodictatorship and poverty.

The divergence between Mauritius andGuyana since independence is a fascinatingexample of economic and political diver-gence in apparently similar societies (fig-ures 6.5 and 6.6).

What can explain this? Both countrieshave similar histories. Mauritius was takenfrom the French and Guyana from the Dutchduring the Napoleonic wars.21 In the nine-teenth century both developed sugarcaneeconomies and, after the abolition of slaveryin the British Empire in 1834, imported largenumbers of indentured laborers from India.Both have a similar population structure,with Indo-Guyanese and Indo-Mauritiansforming the majority of the population withsignificant minorities of people of African,European, and Chinese descent.

After World War II, both colonies weremoved by the British toward independencewith early elections for democratic legislativeassemblies dominated by pro-independencepolitical parties led by Seewoosagur Ram-goolam in Mauritius and Cheddi Jagan inGuyana. Both groups used extensive socialistrhetoric and proposed land reforms andfairly radical policies. Many of the politicalstruggles with British administrators overpostindependence institutions, such as theform of the electoral system, were fought oversimilar issues. As independence arrived how-ever, political forces re-formed into a situa-

tion in which parties led by Indo-Mauritiansand Indo-Guyanese faced a coalition of par-ties supported by the non-Indian population,led by Gaetan Duval in Mauritius and ForbesBurnham in Guyana. Yet, at independence,politics and economics diverged.

The Mauritian Labour Party won powerinitially and quickly abandoned its radicalpolicies—by the early 1970s, investment inthe export processing zone had begun. Thepolitical hegemony of the Labour Party wasquickly contested by a strong socialist party,the MMM (Mouvement Militant Mauricien)led by Paul Berenger and Dev Virahsawmy. Inresponse, the Labour Party entered a coali-tion with Duval and his PMSD (PartiMauricien Social Democrate) and the previ-ous opposition groups. The Labour Partydrew back from repressing the new politicalforces, allowed the MMM to contest the 1976election, and instead adopted social policies,such as the provision of universal secondaryeducation, to improve its popularity. It alsoquickly dropped populist macroeconomicpolicies and, in the late 1970s, implemented aserious stabilization program under the IMF.The final test of Mauritian institutions wasthe election of an MMM government for thefirst time in 1982. Once in power, the MMMabandoned its more radical policies, andwhen the broad political consensus for goodinstitutions became clear, the export process-ing zone boomed.

The contrast with Guyana is stark. Thefirst election on the eve of independence waswon by Burnham and his People’s NationalParty in a coalition against Jagan’s People’sProgressive Party. Burnham maintainedpower by increasingly fraudulent means,finally changing the constitution in 1980 tomake himself executive president. He assassi-nated opponents, most famously the radicaleconomist and political activist Walter Rod-ney in 1980. The economic policies of Burn-ham’s regime were a disaster. He expropriatedthe sugar plantations, creating highly ineffi-cient state industries, and he aggressivelypromoted his party members throughpatronage, particularly in the civil service. Theimplied or actual threat to property and per-son led to a huge diaspora of Indo-Guyanesefrom the country, including most of the pro-fessional and middle-class people. Only in the


0.0

Constraint on the executive

Mauritius

Guyana

1.0

0.8

1970 1980 1990 2000

0.6

0.4

0.2

Figure 6.5 Constraints on the executiveare greater in Mauritius than in Guyana

Source: Polity IV data set, downloaded from Inter-University Consortium for Political and SocialResearch. Variable described in Gurr (1997).

7.5

GDP per capita (log)

Mauritius

Guyana

9.5

9.0

8.5

8.0

1970 1980 1990 2000

Figure 6.6 GDP per capita is rising inMauritius, not in Guyana

Source: World Bank (2005g).

1990s did a democratized Guyana begin toslowly recover from this legacy. But the ethnicdivide endures, and the country continues tosuffer from weak governance, a lack of politi-cal transparency, and ethnic tensions thathamper economic and social development.

What can explain such divergent outcomesin such apparently similar circumstances? InGuyana, there were fewer constraints on theuse of power, and political conflict wasmore polarized, defined solely along ethniclines. And although both countries startedindependence as democracies, what themajority could do (or wanted to do) to theminority was limited in Mauritius, but notin Guyana.

In Mauritius, the British colonial statefaced a powerful and homogeneous Frenchplanter class that did not leave the islandafter Mauritius was annexed to Britain in1812. In the 1870s, when Britain wasreducing the autonomy of colonial admin-istrations, it was forced to create a legisla-tive assembly. Although this was initiallydominated by the planters, by the turn ofthe twentieth century the first Indo-Mau-ritians were elected. This was a clear signthat the greater political autonomy of theisland was allowing for a more open soci-ety with greater upward mobility of for-mer indentured laborers. The power of thecolonial state was checked, evident in thefact that Mauritian independence leaderswere able in the 1960s to negotiate postin-dependence institutions closer to the onesthey wanted.

This juxtaposition of different localinterests and the weakening of the legacy ofthe colonial state gave rise to a more bal-anced distribution of political power inMauritius. And from this situation morefluid interests emerged. Though ethnicidentities were certainly important in poli-tics, so were different cleavages, as is clearfrom the development of the MMM into apowerful political force and the coalition ofRamgoolam and Duval in the 1970s. Poli-tics became much less polarized than theymight have been.

In Guyana, there was no indigenousplanter class to check the power of the colo-nial state. After the departure of the Dutch,the plantations came to be owned by absen-

tee British companies. The authoritariantendencies of the colonial state were rein-forced by British military intervention, pro-moted in 1953 by the United States, toremove Jagan from power because of hissocialist tendencies. Guyanese politicians,unlike those in Mauritius, had far less abil-ity to get what they wanted from the colo-nial state. This meant that there were fewerindigenous checks on the exercise of power,and unfettered use of political power wasthe norm. The best example here is the elec-toral system. Britain imposed a propor-tional representation system on Guyanabecause it was afraid that the overrepresen-tation of large parties inherent in majoritar-ian systems would allow Jagan to win anabsolute majority in the 1964 election (thePeople’s Progressive Party won 42.6 percentof the vote in the 1961 election). This sys-tem facilitated Burnham’s rise to power.

Although the British tried to do the samething in Mauritius, political elites there heldout and forced a compromise: a system withrelatively large electoral districts with thethree politicians who got the most votesbeing elected and with the eight best “losers”from the entire country being elected to par-liament. This system maintained elementsof the majoritarian institutions that Maurit-ian leaders believed were essential to main-taining the country’s governability. Politicsin Guyana became completely defined alongethnic lines. This occurred because the pre-vious evolution of the economy, and thedominant power of colonial interests, leftlittle room for the varied interests thatemerged in Mauritius. While Guyana hasnot suffered outright social conflict, highlevels of polarization and weak conflictmanagement institutions can be contribu-tory factors to civil wars (box 6.3).

ImplicationsIn Mauritius, property rights are secure andthe country has experienced open demo-cratic politics. There has been intensiveinvestment in education and free access intoprofitable investment opportunities, illus-trated most clearly by the export processingzone. In Guyana, the opposite was true in the1970s and 1980s. The puzzle is why institu-tions have been so good in one case and so


bad in the other, given such apparently sim-ilar histories and circumstances.

But the two cases make sense in moredetail. The colonial history of Mauritiusdiverged from Guyana’s in significant waysthat allowed the development of a strongerdomestic political society. Mauritius resis-ted the colonial state more effectively and,ultimately, generated a more egalitariandistribution of political power and a lesspolarized structure of political conflict. InGuyana, however, there was no powerfuldomestic interest group that had a vestedinterest in opposing the colonial state or thatwas able to block the state from expropriat-ing land and other assets after independ-ence. The use of power was unconstrained,and politics were highly polarized alongethnic lines.

Indonesia shows that growth is possibleeven with underlying bad institutions whenelites can credibly make a contingent com-mitment to improve institutions and whenthey manage to forge mechanisms that indi-rectly benefit from encouraging the invest-ment opportunities of others. The accelera-tion of growth after 1966, and particularlythe pro-poor aspect of growth, was clearlydriven by the threat of communism and

rural social disorder. The spillover from theconflicts of 1965 and 1966 was a redistribu-tion of power toward the rural sector, withsustained, inclusive growth necessary forthe political survival of the regime.

Yet the redistribution of power inIndonesia was not institutionalized, unlikewhat occurred in the Republic of Korea, forexample. Moreover, it did not force the NewOrder Regime to improve institutions out-side the rural and education sectors,although the connection between promot-ing economic development and social ordermay well have helped the government tosustain its relationship with the cukongentrepreneurs. As the constraints on eco-nomic policies of the New Order Regimerelaxed in the 1990s, it appears to have beenmore difficult to avoid a massive and debili-tating upsurge in corruption and rent-seek-ing. Moreover, the collusive agreement thatthe state forged with the Sino-Indonesianentrepreneurs appears to have been veryfragile. It rested on shared expectationsabout the longevity of the relationship,expectations that clearly deteriorated withSuharto’s failing health and could not sur-vive the financial crisis in 1997.22

Transitions to more equitableinstitutionsSo far we have examined cases illustratingthe mechanisms that create good institu-tions and sustain prosperity. They involveinstitutions that allow for greater equality ofopportunity, and behind such a set of insti-tutions lies a relative balance of economicresources and political power. Such institu-tions have emerged in some societies but notothers. Although systems of institutionsoften tend to reinforce one another and per-sist for long periods, they also change.Countries with unequal distributions ofresources and political power become moreegalitarian and democratic, and previouslypowerless people gain power and influence.Although institutions are sometimes createdby colonialism or military conquest, theycan often evolve through good decisions,virtuous paths, and the intrinsic dynamicsof the development process, as in Mauritius.It is also possible that even transitory condi-


Researchers have long recognized thatdeep social divisions make it harder toimplement policies that benefit all. Gettinga more precise measure of the nature andextent of such divisions, however, hasproved problematic. For much of the1990s, scholars used a measure known as“ethno-linguistic fractionalization”—firstcompiled by Russian social scientists in the1960s—to show that economic growthwas slower, controlling for other factors, insocieties where there was a low probabilitythat two citizens drawn randomly from apopulation group were of the same ethnicgroup. Africa’s “growth tragedy” was, inpart, blamed on its high level of “fractional-ization” (Easterly and Levine 1997).

More recent work has sought to refinemeasures of social diversity by focusinginstead on polarization, or the extent towhich a small number of influential groupsdominate a society, thereby providing amore theoretically informed basis forexplaining the relationship between diver-

sity and conflict, and through this channel,economic growth (Esteban and Ray 1994).By this measure, a country with threegroups that comprise, respectively, 49 per-cent, 49 percent and 2 percent of the popu-lation will be more polarized than a coun-terpart country where those same groupscomprise 33 percent, 33 percent, and 34percent of the population.The polarizationmeasure is a far more robust predictor ofcivil conflict than either measures of theinequality of individual incomes orfragmentation.This statistical association isillustrated by the fact that, by this measure,9 of the 10 most polarized societies in theworld have experienced major civil conflictin the past few decades, including Eritrea,Guatemala, Nigeria, Sierra Leone, and Bosniaand Herzegovina (García-Montalvo and Reynal-Querol forthcoming).This is only oneinfluence on conflict, of course, and otherwork has emphasized the role of resourcedependence and state capacities (see WorldBank 2003h).

B O X 6 . 3 Polarization, conflict, and growth

tional solutions lead to permanent change,because growth unleashes transformationsthat induce beneficial changes in institu-tions. This message from modernizationtheory23 is precisely what may have hap-pened in the Republic of Korea.

The biggest challenge is to understandprocesses of change and to distill from themlessons about how poorer societies canundergo beneficial institutional transitions.This does not appear to have happened inArgentina (box 6.2) or Guyana, but it didhappen in Britain in the seventeenth, eigh-teenth, and nineteenth centuries and in Fin-land, Sweden, Spain, and the Republic ofKorea in the twentieth century. It also hap-pened in Mauritius. Here we briefly reviewthree such transitions: early modern Britain,Finland and Sweden in the early twentiethcentury, and China in the last 20 years. Thetransitions and policy choices in Spain arediscussed in focus 3 on Spain.

Early modern Britain Around 1500 most European countrieswere highly hierarchical feudal societiesruled by absolute monarchs whose powerswere endowed by God. The most prosper-ous places, such as the Italian city states ofVenice, Genoa and Florence, had escapedfeudalism and were ruled by republicangovernments strongly representing mercan-tile interests. The Netherlands also escapedintense feudalism and was relatively pros-perous, but it was part of the autocraticHabsburg Empire. Nevertheless, the differ-ences in income between the most and theleast prosperous places were relativelysmall. After 1500, this picture began tochange rapidly. First the Netherlands andthen Britain became much more prosper-ous than the rest of Europe, and theMediterranean world went into decline.

As North and Thomas (1973) argued, themost plausible explanation for these changesis the emergence of constitutional govern-ment in the Netherlands and Britain: diverg-ing prosperity within the early modern periodwas tied to the evolution of political institu-tions.24 Institutions improved because of achange in the distribution of resources andpolitical power. Indeed, there was a virtuouscircle of changes in institutions, the broader

distribution of resources and power, andsubsequent changes in institutions. Thesechanges included the collapse of feudalismand serfdom and the move to a free labormarket, the changes in land distribution, thecommercialization of agriculture and thedevelopment of interoceanic commerce.25

Yet, even after 1688, the political systemwas at root oligarchic. Further changeswere needed in the distribution of powertoward greater political equality to sustainBritain’s development path and eventuallydeliver a more egalitarian society. Eventhough Britain was a constitutional regime,it was a very limited democracy in 1800.Before the first reform act of 1832 set inmotion political liberalizations that culmi-nated in full democracy in 1918, fewer than10 percent of adult males could vote. Thereason for these changes seems to have beenthe effect of early industrialization andurbanization on the ability of the disenfran-chised to contest the power of politicalelites.26 British democratization in the nine-teenth century was the outcome of a seriesof strategic concessions by political elites toavoid social disorder.27

While the political system of the eigh-teenth century was consistent with individ-ual initiative, invention, and the start of theindustrial revolution in Britain, sustainedlong-run growth called for broad invest-ment, particularly in human capital. Suchinstitutions had to wait for mass democ-racy to begin to arrive after 1867.28 How-ever, the longer history of the Poor Lawsprovide an example of how provisioningfor adverse risks was also supportive ofgreater dynamism (box 6.4)—a theme wereturn to in chapter 7.

The types of political reforms in nineteenth-century Britain led to economic institutionsthat clearly influenced the distribution ofincome, most obviously the promotion ofeducation after 1867. But the same periodalso saw extensive labor market reforms thatstrengthened the bargaining power of laborand led to the rise of the Labor Party. After1906, the Liberal government of HerbertAsquith also began to introduce the basics ofa welfare state, further extended by the Laborgovernment after 1945. As Britain began toadopt institutions that promoted prosperity,


it was still a highly unequal society, andinequality almost certainly increased until theearly or mid-nineteenth century (figure 6.7).Although precise measures of inequality dif-fer depending on the sources, inequalityappears to have risen until the early and per-haps mid-nineteenth century.29 After about1870, there is wide consensus that inequalityfell substantially for the next century.

The fall in inequality after 1870 is closelycorrelated with the Second Reform Act of

1867, which was the first reform that reallyexpanded voting rights to working people.When democracy enfranchises the relativelypoor, they usually can use democracy to tilteconomic institutions and the distribution ofincome in society in their favor.30

Twentieth-century Finland and Sweden31

Finland and Sweden are popularly identi-fied as prosperous countries with generouswelfare states that, in some measure, areproducts of a small and ethnically homoge-nous population. But, a closer reading oftheir economic histories shows that theircontemporary “virtuous circles”—withgrowth and equity mutually reinforcing—are the outcome of a long and difficultpolitical struggle to establish institutionsand enact policies that provide broad eco-nomic opportunities and respond to theinherently wrenching social transitions ofpositive (economic growth, structuralchange) and negative shocks (macroeco-nomic crises, civil war).

Finland was part of Sweden in the MiddleAges, but following a war between Russia andSweden in 1808–09, it became part of theRussian empire. It experienced one of the lastEuropean famines in 1867–68, an event thatushered in major demographic and eco-nomic changes as entire regions were devas-tated. The Russian revolution of 1917 led to acollapse of imperial authority in Finland, andthe country soon declared its independence.But this immediately gave birth to a bloodycivil war between “white guards” (bourgeoisnationalists) and “red guards” (socialists loyalto Russia). More than 30,000 troops alonelost their lives.

In the aftermath, however, many progres-sive reforms laid the foundation for the mod-ern Finnish economy and society. Landreform—a major cause of the civil war—wasenacted almost immediately. A law passed in1918 allowed sharecroppers to buy their land,and amendments in 1922 facilitated the sub-sidized expansion of small farms. Progressiveincome and wealth taxation were in place by1920, soon followed by expansions ofwomen’s rights (although universal suffragein parliamentary elections had been in placesince 1906) and commitments by the central


Far from being a consequence of successfuleconomic growth, recent historical researchon seventeenth- and eighteenth-centuryBritain has found that widespread butunique institutions of social security were inexistence for several centuries before theindustrial revolution. Indeed, scholarsincreasingly argue that a previously under-estimated influence on Britain’s industrialrevolution, in fact, lies in its prior agriculturalrevolution.The principal comparator here iswith the immensely advanced Dutch ruraland trading economy of the sixteenth andseventeenth centuries. Many of the mostimportant technical innovations in Britishagriculture during this period, such as landdrainage engineering, new crop types, androtations, were directly borrowed from theDutch.Yet it was the British agricultural andservice economy that was increasingly out-pacing the Dutch as the seventeenth andeighteenth centuries progressed.Why?

Attention has recently been given toone major institutional difference betweenthe two countries—the nationwide systemof social security created in England by thePoor Laws, which gradually evolved duringthe course of the sixteenth century, culmi-nating in the famous Elizabethan statutes of1598 and 1601.This was a Christian human-ist response, imbued with a new optimismabout what government could and shouldbe able to achieve in the face of perceptionsof increased poverty amid plenty in a timeof population growth.The Poor Law wasmandated by the central state but—mostimportant for its practical effectiveness—itsimplementation was entirely locallydevolved: it was funded by a local tax onproperty in every parish, administered bylocal officials but also rigorously enforced bylocal magistrates. It went side by side with arelatively efficient nationwide populationregistration system, the Church of England’s

parish registers, which was instituted in1538.This placed the English population onan entirely different basis, in terms of socialsecurity, from that of the rest of Europe.

The comprehensive social security systemprovided by the Poor Laws had a number ofhighly significant economic consequences. Incombination with laws (dating from the thir-teenth century) granting complete alienabilityof land, it encouraged labor mobility andreduced the attachment to land holding asthe only form of security for peasants. Individ-uals had a relative certainty of being providedfor,wherever they moved to work in the econ-omy,no matter what their property-owner-ship status.Landlords and farmers could reapthe economic gains to be had from increasedfarm sizes, from enclosure,and from laying offworkers or changing their labor contracts tomore efficient weekly or day labor,withoutprovoking the same degree of peasantprotest as occurred on the continent.Butequally,employers in England had a strongincentive only to do this if it made economicsense because, through the Poor Law,theywould also have to reckon with their liabilityto pay for the families of the laid-off workers.

What the Poor Law created in Englandwas a public system of acknowledgment ofcollective responsibility for the basic subsis-tence of all, including for a strikingly non-moralistic approach to the support of singlemothers and their illegitimate children.Thecomparative evidence suggests a relativelack of correspondence in England—alonein all of Europe—between fluctuations inthe price of food and the death rate, andEngland—but not Ireland—was the firstnation in the world to cease to experiencefamine-related mortality.

Sources: Szreter (2005) drawing on Slack (1990),Wrigley (1998), Solar (1997), Solar (1995), King(1997), King (2000), Lees (1998).

B O X 6 . 4 Aiding equitable growth in early modernBritain: the role of the Poor Laws

government (not just local municipalities) toprimary education.

From the late 1940s until the early 1990sthe economy expanded steadily, with percapita incomes catching up with GreatBritain in the 1980s and Sweden in the1990s (from roughly half a century earlier).This success was a product of Asian-style“governed markets”: collaboration betweenthe state and private sector was harnessed torapidly industrialize an economy that, aslate as the 1950s, generated 40 percent of itsoutput from agriculture.32 A crucial coun-terpart to Finland’s activist industrial policy(based on high rates of capital accumula-tion and public saving, low interest rates oncredit, and major investments in manufac-turing infrastructure), however, was theconstruction of a welfare state to cushioncitizens of all ages against the unsettlingsocial changes wrought by such a rapid eco-nomic transformation.

Strong and credible political leadershipwas central to making this possible. In theaftermath of World War II, President UrhoKekkoken famously asked his nation, “Do wehave the patience to prosper?” Thereafter, heset about negotiating the arrangements(“social corporatism”) among industrialists,trade unions, and citizen groups that wouldenable all to act as complements. The Finnishmodel has its problems (high unemploy-ment), but it shows how state, market, andsociety can jointly generate the institutions,policies, and spaces needed to generate equi-table development outcomes.

Sweden is perhaps most closely associ-ated with the welfare state today. Less wellknown is the timing and sequencing ofevents putting it in place. Importantly, theSwedish welfare state was the product of,not a precursor to, the country’s transitionto modern economic growth. Indeed, it wasdesigned in response to the very problems(old-age security, unemployment) gener-ated by such growth. But to make suchgrowth possible, and to have in placesociopolitical conditions that would enablethe articulation of and sustained supportfor something like the welfare state (whensuch a system existed only in rather embry-onic forms elsewhere in the capitalistworld), it was vital that a prior set of equi-

table institutional arrangements be inplace.

In Sweden, these prior arrangementswere unusually favorable to upward mobil-ity by subordinate groups: a long history ofpeasant autonomy, a correspondingly weakaristocracy, and an emerging nation-stateable to secure support from farmers whilealso repudiating aristocratic claims on itspowers. Sweden was also the first country tohave a central bank (in 1668) and among thefirst to grant basic property rights. As such,“inclusion of the peasantry in the trans-formation of the agrarian economy andinstitutional arrangements that sustainedegalitarianism were to become fundamentalelements in the rise of the Swedish industrialmarket economy.”33 This was an economyincreasingly grounded in broad politicalrights and social opportunities.

But history is not destiny. Equitabledevelopment is as much a function of keychoices and decisions at pivotal historicjunctures. The Middle Ages, the industrialrevolution, and the tumultuous twentiethcentury unleashed sweeping forces onSwedish society. Some were leveling (risingagricultural productivity), others wrench-ing (mass unemployment). Each attempt torespond to these forces established thepolitical contours for subsequent attempts.Drawing on and extending the equitableinstitutional foundations during these piv-otal historic junctures have been the unify-ing elements of Sweden’s developmentstrategy. Its achievements to date have beenremarkable, even as twenty-first-centuryrealities present distinctive challenges to itswelfare state.


0.0

Gini coefficient

Lindert and Williamson (1982, 1983)

Bourguignon andMorrison (2002)

Williamson (1985)

0.8

0.6

1759 1801 1823 1867 1871 1881 1890 1901 1910 1929 1950 1960 1970 1980 1992

0.4

0.2

Figure 6.7 Inequality in Britain began to fall around 1870

Sources: Lindert and Williamson (1982), Lindert and Williamson (1983), Williamson (1985),and Bourguignon and Morrisson (2002).

The main implications of the Finnish andSwedish cases for today’s developing coun-tries are that economic growth and sociopo-litical equity can be powerfully reinforcing,and can be underpinned by institutionaltransitions. These cases should not be seen asblueprints for others to follow. Instead, theyshould be read as examples of how commit-ments to equity in a given context help lay thefoundations for short- and long-term pros-perity by consolidating virtuous circles link-ing institutions and incentives.

China in the late twentieth centuryEconomic development in China since 1978has been nothing short of spectacular. Withthe quadrupling of GDP per capita over thelast 25 years, China has transformed itselffrom a poor centrally planned economy to a lower-middle-income emerging marketeconomy. As a result, the number of peopleliving in poverty (under $1 per day) fell from634 million in 1981 to 212 million in 2001.34

From the perspective of this chapter, whatis interesting is that the world’s largest coun-try has undergone profound economictransformation without substantially chang-ing the political institutional structure, thatremains dominated by the Chinese Commu-nist Party. Yet institutional improvement didtake place in China along with economicreform. And the large increase in nonstateinvestment and free entry into profitableeconomic opportunities suggest that prop-erty rights are secure, despite the absence of aWestern-style judicial system.

While the particular institutional form isdifferent from other cases reviewed here, theexperience in China is broadly consistentwith the thesis of this chapter. The earlierdiscussion of equitable transitions in Britainand Scandinavian countries illustrated theargument that a successful economic systemdepends on the political system to assignand enforce property rights and contracts,and to protect the market from politicalencroachment. China’s recent history sug-gests that the starting point for reformsdoes not necessarily have to be in politicalinstitutions. Changes in economic institu-tions and in economic relations among lev-els of government can also establish crediblecommitment to a reform path and act as a

check on the discretionary use of power bythe central government. China’s experiencealso demonstrates that what is important forequitable development are credible checkson the arbitrary use of power, assurance ofproperty rights and fair treatment for abroad segment of society. The particularform that institutions take to deliver thesefunctions can vary, especially during periodsof transition.

The key to China’s equitable developmentwas the combination of initial conditionsand the economic reforms launched in 1978that unleashed entrepreneurial initiative andlegitimized the profit motive. China’s eco-nomic policies following the 1949 revolutionproved seriously flawed: they stifled incen-tives for investment and innovation. But thesocial policies of the Mao Tse-Tung periodleveled the distribution of assets in importantand durable ways. As a result, both land andhuman capital were equitably distributed onthe eve of reforms. With the adoption of therural household responsibility system, peas-ants became the immediate beneficiaries ofreform. This helped to reinforce equity, whileunleashing entrepreneurial initiative andboosting productivity.

The economic reforms launched in1978 aimed at decentralizing economicdecisions—to individual farm households,enterprise managers, local governments—so as to generate incentives for investmentand innovation. Importantly, the formthese policies took and the transitionalinstitutions that were created weredesigned to preserve the political supportfor reforms, by compensating potentiallosers.

The aftermath of the cultural revolution,and the recognition that China’s economyhad fallen behind—not least in relation to theEast Asian Tigers—led to a growing consen-sus on the need for and urgency of change,and paved the way for the economic reformsinitiated under Deng Xiaoping’s leadership.These reforms were inspired by the wide-spread recognition of the failure of centralplanning as an instrument for economicorganization, and reflected the need to deliveron economic growth for the legitimacy of thenew leadership. The political need for growthimplied a new focus on liberating markets


and incentives. The sequencing of reformsand the transitional institutional arrange-ments that accompanied the economicdecentralization, on the other hand, reflectedthe premium the leadership placed on socialand political stability.

The impetus for economic decentraliza-tion on the one hand, and the need for anintegrated national market on the other,helped to shape a dynamic relationshipbetween the central government and localgovernments that held them mutuallyaccountable and limited discretion on bothsides. Over time, the result of these policieswas to create a stake in new economic institu-tions for all the main actors, including thelocal governments which served as a crediblecheck on the powers of the central govern-ment in the economic domain. The reformsalso fueled the emergence of strong economiccenters, such as Guangdong province and theShanghai municipality. These centers nowwield considerable influence and bargainingpower relative to the central government andcan serve as important countervailing forces.

How did economic decentralization rein-force private incentives? According to Walderand Oi (1999), “For almost 20 years, reformin China has proceeded through the gradualreassignment of specific property rights fromhigher government agencies to lower govern-ment agencies, or from government agenciesto enterprises, managers, families, or individ-uals” (7). All of these reforms enhanced thepower of economic agents to make decisionsover economic activities in their respectivedomains, and boosted productivity throughbetter incentives. Farmers retained their earn-ings and therefore worked harder andinvested more. Township and village govern-ments had rights to the profits made bytownship and village enterprises (TVEs) andtherefore adopted policies that promotedbusiness. But because they had no revenueauthority, they did not have the ability to bailout poorly performing TVEs, which made forhard budget constraints and higher efficiency.

Higher levels of local governments (coun-try and province) acquired control over localenterprises and therefore also had a stake intheir performance. They were allowed toretain more local revenues through fiscal con-tracting and to have extrabudgetary funds,

which generated incentives that focused oncollection to provide local public goods thatattracted local investments. These changesprovided for significant autonomy from thecentral government and considerable inde-pendent authority over their economies.

The modern Chinese system includes a divi-sion of authority between the central and localgovernments. The latter have primary controlover economic matters within their jurisdic-tions. Critically, there is an important degreeof political durability built into the system.35

China’s reforms are also replete with inno-vative mechanisms for protecting potentiallosers during transitional periods. This ofteninvolved designing reforms that sustainedsources of income for incumbents, bykeeping important elements of pre-existingpricing and payment mechanisms, whileproviding incentives at the margin. “Thetransitional institutions [were] not createdsolely for increasing the size of [the] pie, [butalso] to reflect the distributional concerns ofhow the enlarged pie is divided and the polit-ical concerns of how the interests of those inpower are served.”36

Dual pricing at the start of reforms is aprime example. The system obliged farmersand enterprises to sell specified quantities tothe state at “plan” prices, while allowing themto obtain market prices for any above-quotaproduction. This maintained the planningsystem for those who benefited from it, whilecreating incentives for efficient production.Equally important, it allowed time for marketinstitutions to emerge, avoiding the institu-tional vacuum that plagued many transitioneconomies when state institutions were dis-mantled. Fiscal contracting guaranteed thecentral government a certain level of rev-enues,37 but it generated incentives for localgovernments to collect more because themarginal retention rate was much higher.Similarly, labor contracting allowed stateworkers to retain the guarantee of lifetimeemployment while introducing greater flexi-bility in labor policies for new contractualworkers. These arrangements made reforms awin-win game, ensuring social stability andthe support of those in power.

But there is a danger in such a strategy ofincrementalism: getting stuck in incomplete


reforms if local governments and incum-bents acquire too much power and are ableto block further progress. The prevalence ofinterprovincial barriers to trade in the1990s, with each province vying to boostprofits for the enterprises it owned, is anexample. But there are some checks and bal-ances in the system that help maintain thedirection and momentum of reforms. Theseinclude competition among local govern-ments, hard budget constraints for localgovernments, the central government’sinsistence on enforcing a unitary market,and a growing economy that reduces theeconomic influence of incumbents.

The struggle for the right balancebetween economic centralization and decen-tralization is constantly evident in many ofChina’s domains of intergovernmental rela-tions. The 1994 tax reforms recentralized fis-cal revenues, in part to ensure greaterregional equity in spending, and the centralgovernment continues to apply strict con-trols on deficit financing by local govern-ments. Constraints on labor mobility haveeased considerably over time, helping to cre-ate a more unified labor market, despite con-cerns by some provincial governments thatthis might aggravate problems of unemploy-ment for established urban residents.

There are also some more recent andmore permanent institutional changes thatreconfirm the government’s commitment tomarket-oriented reform. These includemechanisms that strengthen accountabilityat the local level and empower local popula-tions. Local elections are the most importantof these mechanisms but others include, forexample, recent regulations to eliminate nui-sance taxes on the rural population. Chinahas also successfully used the external com-mitment device of WTO accession to signalits resolve to move ahead with marketreforms and impose discipline on incum-bents. For example, it is no longer possiblefor every province to have its own inefficientautomobile factory erected behind trade bar-riers designed to provide local employmentand local taxes. More broadly, China’s desireto carve for itself an important place in theglobal order and to be recognized as aresponsible global power places constraintson the shape of its future policies.

Qian (2003) notes the following:

There is apparently a larger room than wethought for institutional innovation tosimultaneously address both the economicand political concerns, that is, to make areform efficiency improving and interestcompatible for those in power (305).

But there are many challenges ahead, someof which will not be amenable to win-winsolutions and therefore are likely to be polit-ically and socially more costly. Continuedreforms in the state enterprise and financialsectors, managing rural-urban migration,and addressing increasing regional disparities(see focus 6 on regional inequality) are someof these challenges. Macroeconomic policyand structural reforms will need to be under-pinned by further institutional improvementto ensure broader participation and account-ability so that the interests and desire of thepeople are better reflected in decision mak-ing, and to further strengthen the govern-ment’s capacity to lead market-orientedreform while maintaining economic andsocial equity.

ConclusionA few simple principles go a long waytoward unifying different developmentexperiences in the historic and the contem-porary worlds. There is little disagreementamong scholars that basic institutions, suchas security of property rights and equalitybefore the law, are keys to prosperity. Theseinstitutions lie behind the capital, financial,land, and labor markets that we saw inaction in chapter 5. Because talent and ideasare widely distributed in the population, aprosperous modern society requires themass of people to have incentives—and astate that can and will provide key com-plementary inputs and public goods. Ittherefore requires an underlying set ofinstitutions that generate the equality ofopportunity for individuals and assure theaccountability of politicians to all.

Why do some societies have such insti-tutions and not others? A relatively egali-tarian distribution of political powerunderpins the institutions that promoteprosperity. Institutions clearly have distrib-utional effects, and bad institutions often


arise because they benefit some group orelite. Good institutions arise when checksare placed on the power of elites and whenthe balance of political power becomes moreequal in society. Often, equality of politicalpower is supported by economic equality,and this connection gives rise to the possi-bility of both virtuous and vicious circles.

Growth certainly can occur in societiesin which these conditions do not apply. Butthe preponderance of evidence suggests thatsuch growth is unsustainable. This perspec-tive is consistent with historical narratives,basic patterns in cross-country data, andmore careful causal empirical work on thesources of prosperity.

The crucial question for the promotionof development is this: how can poor soci-eties improve their institutions and move

onto a dynamic path toward a virtuous cir-cle of equity and prosperity? The organiza-tion of society is highly persistent, but wehave seen many cases of transitions to bet-ter institutions. Sometimes, as in earlymodern Britain, economic changes lead tochanges in the distribution of power, whichpromotes a more equitable society and bet-ter institutions. Contemporary China fol-lows a similar pattern albeit with a differentconfiguration of institutions. In othertimes, as in the Republic of Korea andIndonesia, regimes are forced, by externalor internal threats, to change the trajectoryof their society in ways that become institu-tionalized. In still other times, such as Mau-ritius and Botswana, leaders make gooddecisions that lead to reinforcing paths ofbetter institutions and development.


“politics in command” in 1959 and pro-duced a ruinous inflation that broughtmuch of the population to near starvation inthe mid-1960s. It was with just cause thatGunnar Myrdal pronounced in AsianDrama, 1967, that “no economist holds outany hope for Indonesia.”

Indonesia’s rapid, pro-poor growth forthe 30 years after the fall of Sukarno aston-ished the development profession and,along with other countries in East andSoutheast Asia, Indonesia became the objectof intense analysis.1 In Indonesia, the weakstarting conditions significantly influencedhow the economic planners approached thetask of linking growth to the poor. Theydesigned a three-tiered strategy for pro-poor growth, which connected soundmacroeconomic policy to market activitiesthat were facilitated by progressively lowertransaction costs. Those policies were linkedto household decisions about labor supply,agricultural production, and investment inthe nontradable economy.

The extent to which the poor benefitedfrom growth depended on the array ofassets they controlled: their labor, humancapital, social capital, and other forms ofcapital, including access to credit.2 Appro-priate government policies also influencethose dimensions, especially in health andeducation. The “road to pro-poor growth”started from desperately poor economicconditions, weak institutions, and a decadeof political instability. It seemed that every-thing needed to be done at once. The keywas to focus on restarting and then sustain-ing rapid economic growth, empoweringpoor households to enter the market econ-omy, and reducing the costs and risks ofdoing so by investments to lower transac-tion costs.

The strategy worked for three decades:between 1967 and 1996, income per capitaincreased by 5 percent a year. The incomesof the bottom quintile of the income distri-bution, all individuals below the national

Indonesia presents an illuminatingexample of the long-term interactionsof the three basic themes of this report

on equity and development:

• The importance of market-drivenprocesses in determining the distribu-tion of opportunities and incomes.

• The role of political processes, and theengagement of the poor in theseprocesses, in determining the policyframework for market and asset accu-mulation.

• The overriding dominance of institu-tions in determining the long-run con-ditions of governance for markets andpolitics to operate.

These complex interactions require longperiods of developmental evolution toobserve and identify.

Indonesia has substantial varianceacross all three of these themes. There isenough independence in the variance foreach factor to sort out, if only roughly, whatis driving what. In chapter 6, the politicaldimension of the economic performance ofthe Suharto regime was discussed. Here, wediscuss the connections with policychoices.

Because Indonesia has been so impor-tant to the development profession, it hasbeen studied for a long time. The Dutchexploited the Netherlands East Indies fromthe seventeenth century to early in thetwentieth century. Then, under politicalpressure at home, the Dutch experimentedwith an “Ethical Policy” for the colony, andthe poor benefited significantly. During theGreat Depression, World War II, and thefight for Independence, the Indonesianeconomy deteriorated rapidly, and the poorsuffered disproportionately. Java was theoriginal home of the “dual economy” ana-lyzed by Boeke (1946) and formalized byLewis (1954). After declaring independencein 1945, President Sukarno eventually put

poverty line until the 1990s and all still sub-sisting on less than $2 a day, grew at thesame rate (or possibly slightly faster). Thedistribution of household expenditures hadbeen remarkably stable, with the overallGini coefficient staying within a narrowrange between 0.31 and 0.36.3 Ruralinequality had actually declined signifi-cantly since the 1970s, when access to landallowed substantial benefits to be reapedfrom the green revolution. By the mid-1980s, the labor market had become theprimary determinant of income in ruralareas.

But when the Asian financial crisis hit in1997 and President Suharto was forced toresign in the face of widespread rioting in1998, the country was entirely unprepared inpolitical or institutional terms to cope withthe rapid changes needed in corporate andpublic governance. The crisis sharply low-ered inequality, as urban real estate andfinancial markets collapsed. But the dra-matic reduction in GDP—over 13 percent in1998 alone—caused poverty rates to triple.Only after 2002 did poverty rates return tothe previous lows observed in 1996. By 2004they still had not returned to the trend rateof decline disrupted in 1998.

Explaining these trends in per capitaincomes and their distribution requires anunderstanding of how markets, politics,and institutions jointly shaped the rapid,pro-poor growth strategy, its subsequentcollapse, and current efforts to revive it.Any such explanation is bound to be con-troversial, and there is no formal modelbehind the story about to be told.4 But thestory is plausible and anchored in the his-torical record.

The story begins with two concerns ofthe emerging Suharto government in thelate 1960s. The first was the misery and dis-content of the rural masses, who had sup-ported Sukarno’s communist leanings andpopulist rhetoric. After a decade of activediscrimination against their livelihoods,

Growth, equity, and poverty reduction in an East Asian giant

Indonesiaf o c u s 4 o n

by putting down the riots and imprisoningthe student leaders. Then it mounted a seri-ous effort to make the economy more equi-table. The result, also stimulated by theworld food crisis in 1973–4, was a major shiftin priorities toward rural development and aspecific push toward increasing domesticrice production. Behind this push were theobjectives of stabilization and equity. Tolose control of the rice economy was to losecontrol of what mattered to Indonesiansociety.

The restructuring of Indonesia’s develop-ment approach after 1974, especially the pre-emptive devaluation of the rupiah in 1978,signaled the government’s determination toinclude the poor in the development process.The stability of the Gini coefficient seen fromthe late 1960s to 2004 should not be taken asthe result of market-driven forces in the faceof given technology, but as a conscious gov-ernment effort, led from the macroeconomicarena by the technocrats, to stimulate pro-poor growth.6 This effort succeeded in spec-tacular fashion until the mid-1990s, whencronyism and the growing influence ofSuharto’s children on economic decisionmaking caused the approach to unravel.

Part of the problem of post-Suhartogovernments has been their need to dis-tance themselves from this record of repres-sion and cronyism, despite three decades ofpro-poor growth. This tension brought thefailure of political and institutional devel-opment during the Suharto era to the fore.Questions about causality remain, particu-larly whether rapid, pro-poor growth canbe implemented by authoritarian regimes.Indonesia’s record, along with that of mostof East and Southeast Asia, indicates thatthey can. But is such growth sustainable?And which is more important for managinglong-run, pro-poor growth: good econom-ics or good institutions?

In Indonesia, there was no “chicken oregg” problem. Something had to be done atonce in view of widespread destitution andpolitical chaos, and the sequencing wasclear. Rapid, pro-poor economic growthwas imposed by an authoritarian regimeconcerned about its survival. But this sameregime also imposed on itself commitmentmechanisms to make the growth processmarket friendly to rural households and toChinese capitalists—that is, both ends ofthe economic system. Inflation was broughtunder control by a law requiring the

rural households were near starvation andthus an obvious source of opposition unlessthe new government could incorporatethem in its development plans. Second, thehyperinflation of the mid-1960s, the totaldisintegration of the market economy, andthe political chaos meant the entire popu-lation was ready for a more stable life. Astrategy that promised stability and ruralrecovery would win wide support (as itwould throughout densely settled East andSoutheast Asia).

This is the message that Suharto deliv-ered to his technocrats. This economicteam had engaged Suharto and other seniormilitary officials in economic training exer-cises at the Military College. The tech-nocrats were handed the macroeconomicportfolio and told to deliver on whatbecame known in Indonesia as the develop-ment trilogy—growth, equity, and stability.To many in the political and military arena,stability meant repressive measures to stifledissent, but to the technocrats it meantrestraining inflation (which they did inspectacular fashion in just three years) andstabilizing the rice economy, which was stilla quarter of GDP and providing half theaverage Indonesian’s daily calories. Theinstitutions built to provide this stability, inboth macro terms and in the food econ-omy, became essential to the Suhartoregime’s success.5

Thirty years of rapid economic growth,with equally rapid rates of poverty reduc-tion, was politically popular (the elasticityof reduction of the headcount povertyindex with respect to growth in per capitaincomes was about 1.3 during the Suhartoera). Every five years, the polling results forparliament were gleaned for signs of disap-pointment with the development program.Despite the heavy hand of Golkar, the pres-ident’s party, real information was flowingfrom villages up to the center through theseelections.

Almost despite the intentions of theSuharto regime, political institutions weretaking root (people expected to vote) andthese institutions provided feedback to thepolicy approach of the government. Therewere other feedback mechanisms as well,and the ones that threatened stability weretaken very seriously. After the 1974 riots inJakarta in reaction to the visibly wideningincome distribution, especially in urbanareas, the government responded brutally

national budget be balanced quarter byquarter—a law Suharto basically imposedon himself, but then touted to all con-stituents as a rule the government had tolive under. To build confidence among theChinese business community, the govern-ment opened the capital account in 1970when it unified the exchange rate. The flowof foreign exchange to and from Singaporeand Hong Kong was a sensitive barometerof the investment climate.

Thus the two constraints on the presi-dency, which Suharto felt personally andused as motivation for his bureaucracy andgovernment (not the same thing in Indone-sia), were the need for rural areas to partici-pate in growth, and the need to keep theinvestment climate highly favorable forSuharto’s business partners. The response toboth constraints was an economic package—low inflation, food price stability, an openeconomy, and massive investments in ruralinfrastructure—that generated rapid pro-poor growth. But another part of theinvestment climate, a part only for thosefavored business partners, involved speciallicenses, trade protection, and lucrativeaccess to domestic markets. This partunraveled the “open economy” part of thegrowth package.

The Suharto legacy, despite the deepcommitment to pro-poor growth, did notbuild the groundwork for a political andinstitutional framework that would ulti-mately support it. A deep tension developedbetween the institutional framework tokeep the open economy functioning effi-ciently and the political controls to keep thecronies’ businesses profitable. Without polit-ical feedback about these very same politicalcontrols, the regime was blindsided by theferocity of the opposition to its manage-ment of the Asian financial crisis. Thedepth of the crisis, both economic andpolitical, reflected the vacuum of institu-tions in place to cope with an alternativepolitical system.

The climb out of the chaos of 1998 mir-rors that from the 1965 era, but this timewithout order imposed from above. Theeagerness and skill with which the Indone-sian population has participated in thedemocratic process suggests that social andpolitical order will now be far more sustain-able. The challenge now is to translate thesame democratic process into rapid andsustainable pro-poor economic growth.

Focus on Indonesia 127

WHAT CAN BE DONE TO INCREASE EQUITY IN THE WORLD? Can this bedone in ways that also spur long-term prosperity? We read in part I thatthere are large inequalities of opportunity between people within coun-tries and—even more—between people in different countries. Theseinequalities are perpetuated through interlocking economic, political,and sociocultural mechanisms, creating inequality traps. Individualsfrom different groups and countries face a highly uneven playing field,both in their capacities to acquire endowments and aspire to a better life,and in their opportunities to reap returns from those endowmentsthrough market and nonmarket processes. Because differences betweencountries often exceed within-country differences, it is of particularimportance that national policies support, or are at least consistent with,the narrowing of international differences, notably through the growthprocess.

We argued in part II that many inequalities not only violate peo-ple’s concern for fairness, but actually have costs for the developmentprocess. The effects on development depend on specific forms ofinequality and their interactions with market imperfections and insti-tutions. Unequal opportunities are associated with inefficiencies andwasted economic potential. Pronounced inequalities in the distribu-tion of power are often associated with weak economic institutions,undermining the investment and innovation that is central to long-run growth. Greater equity is thus not only intrinsically desirable butalso is complementary to long-run growth and prosperity. For poorerand excluded groups, a focus on equity can bring a double benefit—abigger pie and a greater share.

But the scope for such a complementary relationship betweenequity and aggregate development is often not exploited. When exam-ining this, we suggest there are two kinds of pathology in policy

Leveling the economic and political playing fields

IIIP A R T

design. First, there is the pathology associ-ated with oligarchic dominance—institu-tions and policies that further the interest ofelites but not those of the whole society.This may take the form of extreme preda-tion and high-level corruption, as inMobutu’s Zaire or Haiti under the Duva-liers. Or it may take the form of enmeshedalliances between economic and politicalelites that favor rent-seeking, as in thePhilippines under Marcos, in much of LatinAmerica in past decades, and in more subtleforms in many countries of the world.

Second, there is a more complex pathol-ogy of policies pursued with the intent, or inthe name, of equity that have high efficiencycosts or perverse effects. Communist eco-nomic policy was disastrous for efficiency,even while many communist societies didmuch in social provisioning. Directedcredit—in India, for example—was intendedfor the poor (and reached some of the poor),but proved a high-cost strategy. Populistmacropolicy is always bad for growth, andalmost always bad for equity sooner orlater—witness Argentina during much of thesecond half of the twentieth century. Per-verse or growth-sapping effects of policiesunder this pathology can be caused byadverse consequences for incentives, unaf-fordable fiscal burdens, or the capture of thebenefits, often by middle groups, which“hoard opportunities” at a cost for othergroups and the overall growth process.

What can be done? At a fundamentallevel, the analysis underscores the centralityof shifting to a state that is more account-able, has checks on predatory behavior ofpolitical and economic elites, is responsiveto all citizens—especially from middle andpoorer groups—and has effective conflictmanagement mechanisms. In part II, wesketched cases of transitions in this direc-tion from history and contemporary expe-riences, and at the local level. The emphasisin the development community on issues ofgovernance and empowerment is entirelyconsistent with this perspective.

While such overall shifts are central todevelopment, the World Bank has neitherthe mandate nor the comparative advantageto discuss specifics of the design of political

institutions (even though action to supportempowerment of the poor is now empha-sized in the design of specific policies—seeNarayan 2002). In part III, we focus on a setof areas that do lie squarely in the arena ofdevelopment analysis and practice—in poli-cies affecting the sectors, markets, and in theglobal arena. This recognizes the influenceof the political and sociocultural context,but focuses rather on what an equity prism,based on the analysis of parts I and II, has tosay about the policy design to break inequal-ity traps and support aggregate growth Thelesson from part II is that this implies payingattention to specific inequalities and theirinteractions with markets, social structure,and power. This involves both issues of tech-nical design and mechanisms that providethe political underpinnings for change,notably through broader accountability,coalitions for change, or compensation oflosers. And while an overarching message isof the potential complementarity betweengreater equity and long-run prosperity,there will often be tradeoffs in specific areasand context. One cross-cutting area con-cerns the need to raise taxes to finance desir-able public spending. The design of taxinstruments is of great importance to mini-mize adverse efficiency effects, while alsopromoting equity where feasible.

We organize the discussion of domesticaction into three areas. First is building andprotecting people’s human capacities—from the start of people’s lives and throughadulthood and old age. Here we focus onequalizing from the bottom up—equalizingup the opportunities of the least advantagedin terms of skills, health, and risk manage-ment. There are certainly issues of equityamong more advantaged groups, but wegive priority to the disadvantaged (in partfor reasons of space). As seen in part II,there are major market imperfections inhuman capital formation and insurancethat affect poor or lower-status groupsmost, yet political action has also often beenbiased against these groups.

Second is ensuring equitable access to jus-tice and complementary assets. A fair andaccessible justice system is crucial for con-straining the power of the political and eco-


nomic elite, avoiding discrimination, andprotecting property rights and personal safetyfor all—with important implications for thewillingness to invest and innovate. Inequitableaccess to land and infrastructure—by wealth,location, or social group—is typical of devel-oping societies and often enmeshed withpolitical structures. Policy design can helpshifts to more equitable and often more effi-cient patterns (chapter 8).

Third is the domain of markets—finan-cial, labor, and product—that have a power-ful influence on the returns to people’sendowments. As chapters 5 and 6 argued,markets are typically far from ideal, work-ing in noncompetitive and discriminatoryways, whether because of intrinsic marketimperfections, or because power structureshave shaped them to serve the purposes ofthose in power. In these areas, and notablyin the case of finance, a primary concern isequalizing down, by reducing protectingprivileges of incumbents. Closely related isthe conduct of macroeconomic policy(chapter 9).

In the global arena, concern remains withindividuals—and the enormous, unjustifieddifferences in opportunity that people facethrough the morally irrelevant fact of coun-try of birth. The global playing field between

nation-states is uneven—and has uneveneffects on different groups within countries.There is substantial scope for making theplaying field more even. But as in the domes-tic arena, policy design involves both techni-cal questions (such as the details of migra-tion arrangements and the application anddesign of patent legislation) and the politicalunderpinnings of rules and institutions forglobal governance. We examine the potentialfor change both in the key global markets—for labor, products, ideas, and capital—andin the potential scope for designing aid inways that support (rather than undercut)domestic development, and through moreeffective and equitable management of theglobal commons (chapter 10).

The epilogue links the report’s perspectiveon equity to the thinking and agreements thathave evolved in the development communityin the past decade—captured, for example, inthe Millennium Declaration (2000) and theMonterrey Consensus (2002)—as well as theWorld Bank’s own strategic pillars of anenabling investment climate and promotingempowerment. We argue that an approach todevelopment that is deeply informed byequity is fundamental to the full integration ofthese frameworks into an effective develop-ment strategy.


Expanding people’s capabilities to leadfuller lives, the aim of all development,means investing in their education andhealth and in their ability to manage risks.But as chapters 5 and 6 discussed, failures inmarkets and governments conspire to gen-erate large inequalities in people’s opportu-nities to build their capabilities. Childrenfrom poorer families start out life withgreater disadvantages than their wealthierpeers, attend lower-quality schools, haveless access to health services, and are not asprotected from economic downturns andfamily crises. By the time they are adults,they are far less equipped to be productivemembers of society. Economic, political,and sociocultural inequalities fuel such dif-ferences in life chances, perpetuating themacross generations.

Public action can level the playing fieldand broaden opportunities by addressinginequalities in access to quality education,health care, and risk management. Well-designed policies will result in more equi-tably distributed opportunities to acquireendowments and boost overall productiv-ity. As potentially talented and productiveindividuals gain access to the services fromwhich they may have been excluded for rea-sons that have nothing to do with theirpotential, societies make gains throughgreater efficiency and greater social cohe-sion in the long run.

Still, there are challenges. Programs requireresources, administrative capacity, and politi-cal support. This means paying attention tothe design of tax systems, tailoring programintervention to context, and above all buildingconstituencies for change. We focus on level-ing the playing field mainly through aug-menting the capacities of those with thefewest opportunities, but we recognize that itmay be necessary to attack the undue influ-

ence of the powerful and the wealthy to beable to shape public policies to benefit the restof society. As we have seen, successful transi-tions are far more likely where the power ofthe excluded to influence public action hasbeen enhanced.

There are strong complementarities amongthe different investments in people. Betternourished children have higher cognitiveabilities. Well-educated parents, especiallymothers, invest more in their children’seducation and health. More educated indi-viduals are likely to be more resilient toshocks. Instruments to smooth consump-tion will spur people to take on not onlyhigher risk but also potentially higherreturn activities and prevent them from dis-investing in themselves (lowering foodintake, forgoing treatment) or in their chil-dren (pulling them out of school) in timesof shocks. And people with more humancapital and better risk management capabil-ities can reduce the variability and increasethe level of their incomes.

The policies we consider in this chapterare particularly important in arresting theintergenerational transmission of inequali-ties. We begin with a review of the rationaleand potential for early childhood develop-ment programs. We next consider broadereducation and health policies for expandingaccess to quality education and care, andfinally discuss transfer policies that helpmanage risks and provide for efficient andequity-enhancing redistribution.

Early childhood development:a better start in lifeBy the time poorer children in many countriesreach school age, they are at a significant dis-advantage in cognitive and social ability. TheEcuadorian study cited in chapter 2 docu-

Human capacities

7c h a p t e r

132

Human capacities 133

mented substantial differences at six years ofage, related to socioeconomic status andparental education. Differences in childhoodcognitive abilities are indeed apparent as earlyas 22 months of age. One study in the UnitedStates shows that by age three the gaps inlearning, measured by vocabulary, are alreadylarge among children from different socialgroups (figure 7.1).1 Cognitive learning isaffected by a child’s socioeconomic statusthrough health (malnutrition, iron andmicronutrient deficiency, parasite infections)and the quality of the home environment,including care-giving and cognitive stimula-tion.2 Scientific evidence on brain develop-ment supports this. Recent research findingsrevamp earlier thinking that assumed that thestructure of the brain was genetically deter-mined at birth and point to the determininginfluence of early experiences—from concep-tion to age six, and particularly the first threeyears—on the architecture of the brain andcapacities in adulthood.3

As a child ages, environmental effectsappear to accumulate. Poor cognitive andsocial abilities are associated with weakerfuture academic performance and loweradult economic and social outcomes,including poor health, antisocial behavior,and violence.4 These underachieving adultsinfluence the cognitive abilities of the nextgeneration of children, creating an inter-generational cycle of poverty and unequalopportunities.5 Studies using internation-ally comparable student achievement testsconfirm that socioeconomic background isthe overwhelming determinant of learningoutcomes, with schools accounting for nomore than 20 percent of the variation in testperformance.6

Benefits of early interventionsEarly interventions can substantially en-hance a child’s life chances and loosen theintergenerational grip of poverty and in-equality. In recent years, interest has ex-panded in early childhood development(ECD) in low- and middle-income devel-oping countries, paralleling greater atten-tion in developed countries.

Early childhood development programscomprise a range of interventions thatinclude providing nutritional supplements to

children, regularly monitoring their growth,stimulating the development of their cogni-tive and social skills through more frequentand structured interactions with a caringadult, and improving the parenting skills ofcaretakers. The evidence suggests that theseprograms can be highly effective in address-ing problems experienced later in schoolingand adulthood.

A recent study in the United States showsthat investments in the early years of life,before children reach the formal school sys-tem, give greater returns than later invest-ments (figure 7.2).7 Well-designed longi-tudinal studies—mainly from developedcountries—indicate that programs typicallyregister improvements for children inhealth, cognitive ability, academic perform-ance, and tenure within the school systemand, later in life, higher incomes, higherincidence of home ownership, lower propen-sity to be on welfare, and lower rates ofincarceration and arrest.8 This suggests astrong productivity case for investing inearly childhood development; the argu-ments for public subsidies to disadvantagedfamilies are compelling on both productiv-ity and equity grounds. As Heckman argues,

early interventions in children from disadvan-taged environments raise no efficiency-equitytrade-offs; they raise the productivity of indi-viduals, the workforce and society at large, andreduce lifetime inequality by helping to elimi-nate the factor of accident of birth.9

Studies of ECD programs in developingcountries also document strong benefits

0

1200

800

Cumulative vocabulary

24201612 28 32 36

Early vocabulary growth

Age, months

HighSES

MiddleSES

LowSES

Figure 7.1 Children from better-offhouseholds have a big edge incognitive abilities by age three

Source: Hart and Risley (1995).Note: SES refers to socioeconomic status.

Return per $ invested

Preschool programs

Schooling

Job training

Opportunitycost of funds

Post-schoolSchoolPreschool

0

2

0

4

6

8

6 18Age

r

Figure 7.2 Early childhood interventions are goodinvestments

Source: Carneiro and Heckman (2003).

for all children, with cost-benefit analysesshowing returns of $2–5 for every $1 in-vested. For example, preliminary resultsfor experimental nutritional interventionsat 6 to 24 months of age in rural Guatemalashow that consuming a nutritional supple-ment increased the probability of attend-

ing school by 5.6 percentage points and ledto higher completed schooling attainmentand higher adult cognitive achievement testscores.

Evidence is also mounting that interven-tions in early childhood particularly benefitpoor and disadvantaged children and fami-lies.10 In Jamaica, nutritional supplementa-tion and stimulation administered to under-nourished children between the ages of 9 to24 months—who are most likely to comefrom disadvantaged families—improvedtheir mental development.11 Malnourishedchildren who received a milk supplementdeveloped more than those who did not (fig-ure 7.3). Children who benefited from stim-ulation did even better, while the benefits ofsupplements and stimulation were additiveand allowed disadvantaged children exposedto them to almost catch up with the develop-ment trajectory of “normal” children over aperiod of 18 months. The results suggest thatECD programs may be one of the most cost-effective avenues for reaching the Millen-nium Development Goals for universal edu-cation and an important contributor to theattainment of gender parity in primary com-pletion (box 7.1). They also help mothersparticipate in the labor force—poor womenwho had access to free child care in the fave-las of Rio de Janeiro increased their incomeby as much as 20 percent—and improve theacademic performance of older children, asdocumented for a community nursery pro-gram in rural Colombia.12

Designing ECD programsInterventions to improve young children’scapacity to develop and learn can focus onimproving parents’ teaching and child careskills, delivering services directly to chil-dren, or improving child care services in acommunity. Programs may be establishedin homes, day-care centers, or communi-ties. The evidence suggests that three designfeatures are important for the full realiza-tion of benefits from ECD programs: start-ing early, having strong parental involve-ment, and focusing on child health(especially nutrition) and cognitive andsocial stimulation. The focus on healthleads to a virtuous cycle, because improvedhealth also helps increase cognitive and


There is sufficient evidence from studiesthroughout the world to make a case forplacing early childhood education amongthe key interventions to achieve the educa-tion Millennium Development Goals(MDGs).

Higher school enrollment. Colombia’sECD program PROMESA reportssignificantly higher enrollment rates in pri-mary school for children participating in theprogram, compared with children not par-ticipating. ECD programs in India (Haryana)and Guatemala resulted in a significantdecline in enrollment age for girls.

Less grade repetition. In Colombia’sPROMESA program, the Alagoas and Fort-aleza PROAPE study of Northeast Brazil andthe Argentina ECD study, children who par-ticipated in the programs repeated fewergrades and progressed better throughschool than did nonparticipants in similarcircumstances.

Fewer dropouts. In India’s IntegratedChild Development Services program inDalmau, attendance of children ages six toeight in primary school increased by 16 per-cent for children who had participated inthe program; dropout rates did not change

significantly for children from the highercaste, but fell a dramatic 46 percent for thelower caste and an astonishing 80 percentfor the middle-caste students. In Colombiathe third-grade enrollment rates forchildren who participated in the PROMESAprogram increased by 100 percent, reflect-ing their lower dropout and repetition rates.In addition, 60 percent of the children whoparticipated in the ECD program attainedthe fourth grade, compared with only 30percent of the comparison group.

Higher intelligence. ECD programsencourage young children to explore andfacilitate the social interaction thatpromotes cognitive development. Childrenwho participated in Jamaica’s First Home-Visiting Program, Colombia’s Cali Project,Peru’s Programa No Formal de EducaciónInicial (PRONOEI),and the Turkish EarlyEnrichment Project in low-income areas ofIstanbul averaged higher scores on intellec-tual aptitude tests than did nonparticipants.Evidence from other studies, however, sug-gests that these effects dissipate over time.

Sources: Chaturvedi and others (1987), Myers(1995),Young (2002).

B O X 7 . 1 ECD programs are an essential ingredient for theattainment of education for all

85

110Development quotient

Baseline 6 12Months

18 24

105

100

95

90Control group

Nutritional supplement

Stimulation

Stimulation and nutritional supplementChildren of normalheight

Figure 7.3 Catching up through early intervention

Source: Grantham-McGregor and others (1991).Note: The Development Quotient is an index composed of ratings in four behavioral and cognitive development indica-tors: locomotor (large-muscle activities, running, and jumping), hand-eye coordination, hearing and speech, and per-formance (shape recognition, block construction, and block patterns). “Months” refers to time after entry into the pro-gram, generally at around 9 months old.

social abilities.13 Overly formal programscan be too expensive for poor families, cul-turally irrelevant, and insensitive to fami-lies’ needs.14 They thus run the risk of beingabandoned even when they demonstratehigh returns.

What then are the impediments to thewidespread implementation of ECD pro-grams given that they are such good invest-ments? Political economy constraints arisefrom the difficulty of making a case forspending resources on a program with thepromise of (uncertain) benefits to come onlyyears in the future. Such a case is often madeby the immediate beneficiaries (parents ofschool-age children) or intermediate benefi-ciaries (teachers), who organize themselvesinto powerful political forces. But the institu-tional setup for ECD program delivery—with funds in many instances channeled to myriad small NGOs, community cen-ters, or home-based caregivers—and theabsence of strong central responsibilityinhibit organized political pressure. Thesame institutional setup generates prob-lems of integration with other governmentprograms and of coordination across sev-eral government departments.15

Thinking about the politics and thedesign of ECD from the start is thus impor-tant. Getting information to parents, com-munity leaders, and policymakers about theobjectives and efficacy of ECD services canbuild public awareness and strengthendemand. Monitoring systems build supportby providing timely feedback on a range ofintermediate outputs to policymakers andprogram managers, while proper evaluationsprovide more convincing evidence of impactand broader lessons from interventions.Integrating ECD programs into the broaderdevelopment frameworks and involving par-ents, families, and community membersenhance the sustainability of programs.16

There are two possible approaches toscaling up ECD interventions. The first is toexpand publicly funded preschool pro-grams to all children by making it a statu-tory right, as in several European countries.This would have significant funding impli-cations, but the benefit is potentially wide-spread support from middle-class and poorfamilies with children.

The second approach would target disad-vantaged families. This may be more cost-effective in view of the evidence presentedearlier on larger gains from interventions fordisadvantaged children. To bolster partici-pation, the program could be supplementedby a cash-transfer scheme, with transfersconditioned on various desirable behaviors,including changes in the homecare environ-ment, as well as regular health center visitsfor growth monitoring, immunizations, andnutrition interventions.17 This would con-centrate even more resources on the poor,but the political economy implications areless clear. While targeted programs have asmaller constituency and thus would notbenefit from a broad coalition of support, anational program, with transparent criteriafor eligibility and good monitoring of “con-ditionality,” could mobilize support not onlyfrom the direct beneficiaries but also fromother stakeholders in society.

It is possible to combine a universal pre-school approach with a conditional cashtransfer (CCT) program. This would yieldthe highest benefits in the participation ofthe poor and the productivity gains for all,but it would also be more costly. Theapproach adopted in any country settingwill have to emerge from considerations ofcosts, benefits, and fiscal capacity—andreflect the political economy.

Basic education: expandingopportunities to learnProminent in the Millennium DevelopmentGoals, education is a great equalizer ofopportunities between rich and poor andbetween men and women. But the equaliz-ing promise of education can be realizedonly if children from different backgroundshave equal opportunities to benefit fromquality education. In the previous section,we argued that children’s ability to benefitfrom school is strongly influenced by thecognitive and social skills they acquire intheir early years. Evidence suggests that thegains from early interventions can dissipateif disadvantaged children go on to low-quality primary schools.18

Chapter 2 documented the large inequali-ties in educational attainment within coun-tries by income, region, gender, and ethnicity.


Chapter 5 presented the economic reasonswhy credit-constrained households underin-vest in education, making the efficiency casefor subsidizing education for the poor. Thereare other reasons for parents to choose a levelof education for their children that may belower than what is optimal for the child andfor society. Educational attainment has vari-ous societal benefits that are not fully cap-tured by the individual. For example, it isgenerally associated with enhanced democ-racy and lower crime, while girls’ schooling inparticular has been shown to reduce fertility,empower women, and thereby contribute tothe welfare of children in the family. In addi-tion, education has intrinsic value, enablingpeople to lead fuller lives as informed andactive participants in society.

The case for moving to equalizing accessto education is therefore strong on bothequity and efficiency grounds, especially forbasic education. Beyond basic education,there is an important efficiency rationale forensuring that the most talented and produc-tive people in society have access to highereducation. In today’s globalized world, withcompetition largely on the basis of skills andideas, countries need to cultivate latent talent,wherever it may reside. Motivated and tal-ented children from poorer householdsdeserve the opportunity to excel as much astheir wealthier peers. While we acknowledgethe important equity dimension of policiesfor tertiary education, the discussion here isdevoted primarily to policies that expandaccess to and quality of basic education.

We argue that there is a case for publicaction to enhance equity in learning so thatoutcomes reflect not merely circumstances of luck—parental endowments, socioculturalenvironment, birth place, one dedicatedteacher—but genuine differences in prefer-ences, effort, and talent consistent with thenotion of equal opportunities. This requiresexpanding affordable access and upgradingquality, with a particular focus on excludedgroups, through various interventions thatincrease both the demand for schooling andthe capacity and incentives of the school sys-tem to respond.

There are clear complementarities in thisapproach: quality improvements help only ifchildren are in school, but they also influence

the probability of their attendance. Evenuneducated parents will pull their childrenout of school if they perceive low quality.19

There may be tradeoffs, however, if resourcesdevoted to upgrading quality benefit prima-rily the privileged who are already in schoolat the expense of reaching excluded groupsor areas—or if the rapid expansion of accessreduces the quality of instruction. While thelong-run objective for school systemsaround the world is clear, priorities will varyby country, region, or group.

Expanding access, particularly for excluded groupsExpanding access for all. More than 100million children of primary school age areout of school, either because they neverentered the system or because they droppedout before finishing.20 As a result, some 52countries risk not reaching the goal of uni-versal primary completion.21 In most coun-tries, improving opportunities in educationmeans ensuring affordable access, especiallyfor poor rural children and disadvantagedgroups.

Higher public spending on the supply ofschools is one way to expand access. Analy-sis of the determinants of school enroll-ment in various countries suggests thatproximity to schools is a major factor.22 Acareful evaluation of Indonesia’s schoolconstruction program in the 1970s, thelargest such program on record, finds evi-dence of significant increases in both edu-cation and earnings.23 The program yieldslarge positive returns, but it takes more than30 years to do so because upfront construc-tion costs are high (more than 2 percent ofIndonesia’s GDP in 1973), while the bene-fits are spread over a generation’s lifetime.

But for every success story there are manyothers in which higher spending has nottranslated into better access to infrastructure,inputs, and instruction for children. In manycases, the resources are not used effectively—too much is spent on teacher salaries orreducing class size and not enough is spenton instructional materials.

Incidence studies suggest that the poorstand to benefit more from expansion whenmean levels of access to services are alreadyreasonably high, now the case for primary


schooling, even in many low-income coun-tries. But spending alone is clearly notenough to get the children in school (andeven less effective in ensuring that theylearn). In many countries, the main prob-lem is not facilities but children droppingout or not attending available schools.24

Recent efforts to boost access thus focus ondemand-side interventions: reducing thecost of schooling or providing incentives,even paying for attendance.

In many countries, parents have to pay alot, either for school fees or for other inputs,such as uniforms and textbooks. Eliminat-ing these costs can boost participation. Freeuniforms and textbooks provided by anNGO program in Kenya (along with betterclassrooms) reduced dropout rates consid-erably: after five years, students in the pro-gram completed about 15 percent moreschooling. In addition many students fromnearby schools transferred to programschools to take advantage of the benefits.The result was a 50 percent increase in classsize—an increase that does not appear tohave deterred parents nor has it led to ameasurable negative impact on test scores.This is at least suggestive that a reallocationof the education budget—larger class sizewith the savings used to pay for the inputsunder the program—could raise schoolparticipation at no cost to quality.25

Eliminating user fees for basic schoolshas also been shown to boost studentenrollment, but quality may be compro-mised if reliable alternative sources offinancing are not available to schools (box7.2). In both Tanzania and Uganda, elimi-nating school fees became an importantpolitical issue when the population couldvoice its discontent, helped by the demo-cratic process, an active civil society, and (inTanzania) the Poverty Reduction StrategyPaper process.

In some cases, there may be a need to gobeyond removing the direct financial costsof schooling to induce poor parents to enrolltheir children. This could be accomplishedby providing CCTs and free meals. CCT pro-grams make payments to poor families, typ-ically mothers, on the condition that chil-dren attend school regularly. The programscan be seen as compensating for the oppor-

tunity cost of schooling for poor familiesand represent one approach to addressingfailures in credit markets and the imperfectagency of parents. Many of the cash-for-school-attendance programs are large, rep-resenting significant commitments of publicresources. The biggest are Oportunidades(previously PROGRESA) in Mexico, theBolsa Escola in Brazil, and the Food for Edu-cation Program in Bangladesh.26

The budgets allocated to these programsare between just under 1 percent of totalgovernment current expenditure in Braziland more than 5 percent in Bangladesh.These significant, but not prohibitive, sumscould be generated from savings on otherexpenditures, such as regressive subsidiesfor public services, including tertiary edu-cation. A question remains about how cost-effective the programs are in expandingeducation: the answer depends on how suc-cessful they are in reaching households thatwould not have participated in the schoolsystem without the transfers.

A careful evaluation of PROGRESAfound an average increase in enrollment of3.4 percent for all students in grades onethrough eight, with the largest increase(14.8 percent) for girls who had completedgrade six.27 Morley and Coady (2003) esti-mate an internal rate of return (taking intoaccount the cost of grants) for the programof 8 percent a year and report that thetransfers are 10 times more cost-effective


There are two schools of thought aboutschool fees. Some claim that school feesdeter poor families from sending their kidsto school. Even nominally small amounts canbe a large share of poor households’ income,and these come on top of the forgone bene-fits of children contributing to family busi-ness or household chores. Schooling costsoften figure in parents’ responses about con-straint to enrollment, and eliminating schoolfees appears to have spurred a largeincrease in enrollments in a number ofcountries, including Kenya,Tanzania,Uganda, and Vietnam. Others see user feesas an important accountability tool, a mech-anism for empowering parents to demand

quality services from the schools, and pointto studies that show even poor households’willingness to pay for good quality services.

Sympathetic to the arguments in favorof greater accountability, we argue for elimi-nating user fees when the fiscal impact offorgone revenues can be managed withoutlarge efficiency costs or harmful spendingcuts.The desirable voice and accountabilityaspects of school fees can be harnessedequally or better through contributinglabor for school improvements or workingon parent-teacher advisory committees.Such in-kind fees are cheaper to the parentand engage the parent more fully in schooldecision making.

B O X 7 . 2 School fees—an instrument of exclusion or accountability?

than building schools. But De Janvry andSadoulet (2004) find that most programbenefits are received by those who wouldhave gone to school anyway. They suggestcalibrating transfers to increase programefficiency—for example, through largertransfers to the eldest child, to children withan indigenous father, or to children, espe-cially girls, living in villages without a sec-ondary school.

Reaching excluded groups. Schools withadequate supplies and well-trained andmotivated teachers, who are accountable forthe learning they produce, are good foreveryone. But additional support may benecessary to improve access for excludedgroups, such as disabled children, girls, andindigenous groups.

Including disabled children is possible atrelatively modest costs. In Uruguay, grantsof up to $3,000 are awarded for schools that

put forward proposals for reaching disabledchildren. In the two years since the fund wasset up, 6 percent of all schools in Uruguayhave been awarded grants to cover expensesto adapt school materials, equipment, andinfrastructure and to train teachers inappropriate pedagogical approaches.

Improving gender equity in access toschooling often requires making specialprovisions for girls, especially older girls.28

Specific grants for girls have been effectivein Bangladesh and Mexico. Private latrinesfor girls are essential. Other structuralimprovements including boundary walls,flexible or double sessions when sharing afacility with boys, and perhaps even gender-specific schools may allay parents’ concernsabout girls’ privacy and safety. It is impor-tant for schools to undermine, not under-score, stereotypes and unequal treatment ofwomen—and to be wary of giving boysmore resources, leadership, and attention.Female teachers are good role models forboys and girls, and even young women canbe effective teachers with training, support,and a programmed curriculum. Govern-ments might consider setting national goalsfor hiring women and being flexible withage and education requirements for femaleteachers (while still providing adequate in-service training).29

To expand access for ethnic groups, teach-ers or teacher aides from the target ethnicgroup are particularly helpful in their abilityto connect with the students as powerful rolemodels. Bilingual schools have also beeneffective. In Mali, bilingual programs wereassociated with large declines in dropout andrepetition, and rural students outscoredurban children. In Mexico, geographic target-ing under PROGRESA (now Oportunidades)led to the relatively high participation ofindigenous people (but not those in the mostremote areas without schools).30 An innova-tive approach to encourage the attendance ofRoma children in Vidin, Bulgaria, appears tohave paid off (box 7.3).

Upgrading qualityBetter quality for all. Expanding access tobasic education is necessary but not enough;the quality of education matters for op-portunities. But even children in middle-


In Vidin, the Open Society Institute and theRoma NGO known by the initials DROMhave been collaborating to integrate Romastudents into the mainstream schoolsystem.Vidin is a town of 85,000 in north-west Bulgaria, and 6 percent of its popula-tion was identified as Roma in the 1992 cen-sus. In the 2000–01 school year, 460 Romastudents, or half the school-age students,were enrolled in the mainstream school sys-tem. Students are bused from thesettlement to school and back. And Romamonitors interact with parents and theschool to encourage attendance. Low-income students also receive shoes andschool lunches, with lunches given on thebus to reduce the stigma of receiving it atschool.

While preparing the program, DROMwent door-to-door in the Roma settlementand sought the support of the schools, themayor, and the media.The project eventu-ally gained the support of all the stakehold-ers except the mayor, who neverthelessagreed not to block it. DROM invited the sixmainstream schools in Vidin to present theprogram, philosophy, and teachers on tele-vision. Roma parents then selected a schoolfor their children.This marked the first timethat their views had been solicited by theauthorities.

At the end of the first semester, atten-dance was 100 percent, and first-term final-grade averages were identical to those ofnon-Roma pupils. Parents and teacherswere satisfied, especially with the absenceof reported incidents of anti-Romaprejudice. Education authorities wereencouraged to scale up in other cities. Inaddition, 35 Roma parents of the bused chil-dren returned to school in adult educationprograms, and three teenagers who haddropped out in the third grade asked to jointhe program, prompting teachers to workextra hours with them. On the negative side,24 pupils received failing grades in one ormore subjects, and three left the program.

The success is attributable to threemajor factors. First, parents feel that theirchildren are protected from prejudicebecause they are bused and monitoredthroughout the day by adult Roma. Second,Roma monitors in the schools assure thatthe children are not mistreated, encourageparental engagement and student partici-pation in extracurricular activities, and helpthe teachers ease cultural differences.Third,the children are happy to be in schoolswhere real learning takes place.

Source: Ringold, Orenstein, and Wilkens (2005).

B O X 7 . 3 Desegregating Roma schools in Bulgaria: the Vidin model

income countries do a lot worse than theaverage OECD kid on international tests oflearning achievement, suggesting that muchof the learning in schools does not preparechildren to be productive adults, let alonefor the rigors of competition they will facein the global labor market.31 The qualitydeficit is undoubtedly greater for childrenfrom poorer families, because the better-offchildren can go to better public schools orleave the public system and opt for privateschooling.

Based on the results of a standardizedinternational achievement test—the ThirdInternational Mathematics and Science Study(TIMSS)—Pritchett (2004a) estimates thatthe overwhelming majority of children ages15 to 19 lacks education (not completinggrade nine or performing poorly in theTIMSS) in five middle-income countrieswith data (figure 7.4).32 But the enrollmentproblem remains large only in Morocco.Indonesia and Turkey have difficulty retain-ing kids in secondary school; in Colombia,Morocco, and the Philippines, three of fourchildren who have completed grade ninehave failed to learn enough.

How can countries improve basic learningoutcomes for all? We know broadly from alarge number of studies that have tried toaccount for the “production” of schoolingoutcomes that higher public spending doesnot always translate into better student learn-ing.33 A recent study analyzing the determi-nants of student performance on theTIMSS—using data for more than 260,000students from 6,000 schools in 39 countries—finds that education spending (spending perstudent, class size, student-teacher ratio) ateither the school or country level has no posi-tive impact on student performance. Amongfactors at the school level, the only ones thathave a significant impact on student perform-ance are instructional material and teacherswith an adequate formal education.34

These results are confirmed by severalcareful microlevel studies. Since 1996, agroup of researchers working with a DutchNGO, International Chirstelijk SteunfondsAfrica, has been involved in the design andevaluation of a series of randomized exper-iments to improve learning outcomes in theBusia district in rural Kenya. The results

indicate that increased availability of text-books helps improve test scores, but onlyamong the better-performing students, andthat performance-based prizes for teachersincreased test scores initially, but the gainsdissipated later. What did work in raisingtest scores were merit scholarships for 13-to 15-year-old girls—with positive effectsalso on learning for boys, who were ineligi-ble, and girls with low pretest scores, whowere unlikely to win the scholarships. Thescholarships were the most cost-effective ofall the interventions tested, achieving thesame learning results at less than 20 percentof the cost of textbook provision.35

The results underscore the importance ofcombining additional spending (of the rightkind) with interventions that strengthenincentives to teach and to learn. As the teacherincentive program shows, project design—inthis case, the behavior rewarded—matters.36

Better quality for the most disadvantaged.Many of the programs just discussed focuson improving performance at the school


MoroccoIndonesia Philippines TurkeyColombia0

20

10

50

40

30

60

As a fraction of cohort

80

70

90

100

Completed grade 9 but did poorly on testsDropped out before completing grade 9Dropped out before completing grade 5Never enrolled

Figure 7.4 Boosting enrollments is not enough toovercome the learning gap

Source: Reproduced from Pritchett (2004a).a. Based on TIMSS-R scores for eighth graders on mathematics in1999. To calculate the fraction of students with scores below 400—one standard deviation (100 points) below the OECD median of500)—Pritchett uses the country mean and standard deviation andassumes a normal distribution. This assumes that scores areroughly constant over time so the 1999 test represents the cohortages 15–19 in the survey year, and that eighth- and ninth-gradecompetencies would be roughly similar.

about half the cost of private secondaryschools; vouchers could be renewed as long asstudents maintained satisfactory academicperformance. An evaluation of this random-ized natural experiment (vouchers wereawarded by lottery) found lower repetitionrates and higher test results among voucherwinners.39 But targeted voucher schemes maybe politically difficult to implement—and theColombia program was discontinued.

Strengthening accountabilityDismal learning outcomes in many coun-tries are due to the combination of inade-quate resources and the low responsivenessand accountability of school systems.Efforts to improve school performance willtherefore need to focus on strengtheningaccountability processes: achieving societalconsensus on expanding education, dealingwith capture by vested interests, and tacklingthe weak incentives for service providers toraise the quality of learning.

Achieving societal consensus on expand-ing education helps tackle the pathology ofelite capture, whereby the wealthy opposeincreased spending in public education. His-torically, expansion of voice in a country hasled to wider access and quality improve-ments in basic education, notably in Europeand North America.43 Democratic transi-tions have spurred recent expansions in basiceducation also in Brazil, Guatemala, andUganda.44 But these are long-term processes,and it is essential to make progress nowtoward meeting the urgent needs of millionsof children around the world.

Some progress can be made by counter-ing the stranglehold of interest groups onequity-enhancing reforms, such as whenteachers unions block reforms that wouldstrengthen the link between performanceand accountability.45 Significant payoffs can come from systemic reforms thatstrengthen accountability from clientsdirectly to frontline providers.46 The mostcrucial steps in any such reform are toincrease the schools’ accountability for per-formance and to ensure the availability ofrelevant information to monitor their per-formance. Accountability for performancealso requires autonomy to manage results.This means delegating responsibility and


The Balsakhi program is a large remedialeducation program now implemented in 20Indian cities by an NGO—Pratham—in col-laboration with the government. Prathamhires young women from the community toteach basic literacy and numeracy skills tochildren who reach the third or fourthgrade without having mastered them. Stu-dents are pulled out of regular classes fortwo hours of the school day for the reme-dial education.The program is inexpensive:$5 per child per year. Easily replicable, it hasbeen scaled up rapidly since its inception inMumbai in 1994, now reaching tens ofthousands of students in the 20 cities.

A recent two-year randomized evalua-tion in Mumbai and Vadodara finds that the

program has highly significant positiveresults on student learning. On average, theprogram increased learning by 0.15standard deviations in the first year and0.25 in the second year.The gains werelargest for children at the bottom of the dis-tribution, with those in the bottom thirdgaining 0.20 deviations in the first year, and0.32 in the second year (0.51 for mathalone).The results were similar in the twograde levels and in the two cities. At themargin, extending this program would be12 to 16 times more effective than hiringnew teachers.

Source: Banerjee and others (2004).

B O X 7 . 4 Remedying education: the Balsakhi program in India

level. What about improving learning out-comes for disadvantaged or poorly per-forming students? The merit scholarshipprogram for 13- to 15-year-old girls inrural Kenya mentioned earlier is one suchexample. The Balsakhi program in India—a large remedial education program—rep-resents another highly successful and cost-effective approach to giving poorlyperforming students a leg up (box 7.4).Because children with the lowest abilityregistered the largest gains in test scores,the program had an equalizing effect onstudent achievement.

Many countries group students togetherby similar abilities on efficiency grounds.However, recent findings in 18 to 26 coun-tries show that such tracking increases edu-cation inequality, possibly by reinforcingthe effects of family background, but itdoes not contribute to higher mean levelsof performance.37

Another option to improve learning out-comes for disadvantaged children is to pro-vide school vouchers. There is significantcontroversy around the equity and efficiencyimpacts of generalized voucher schemes (box7.5). Targeted means-tested voucher pro-grams may be more promising.38 Resultsfrom one such scheme in Colombia areencouraging. The PACES program providedmore than 125,000 students from poorneighborhoods with vouchers that covered

power for decision making to the lowestfeasible level consistent with incentives.47

Once the responsibilities of the schoolsystem are well defined, the resources anddecision-making powers of providers areconsistent with their responsibilities, andinformation is available to track perform-ance, various mechanisms become avail-able to pressure schools to deliver betterperformance. School autonomy, commu-nity control, nongovernment providers,voucher programs, and public sectorreforms can strengthen the ability of citi-zens, communities, and public organiza-tions to hold schools accountable for deliv-ering results.48 El Salvador’s experiencewith rebuilding much of its education sys-tem following the destructive civil war ofthe 1980s is a good example of what can beaccomplished through partnership with

local communities. As a result of greaterparental involvement, Educo schools hadrapid enrollment increases without givingup quality, reduced absenteeism amongteachers and students, and increased mathand language scores.

Toward better health for allThe large inequalities in health care useand health outcomes in many developingcountries do not just reflect different pref-erences or needs—they arise from con-straints on the ability of individuals toachieve good health (chapter 2). Income isone important constraint, especially givenincomplete financial markets. Low-incomepeople around the world have worse healthand use fewer health services (chapter 2).Ethnicity, race, and location also influence


School voucher programs increase the power ofparents to choose schools for their children. Par-ents are given a voucher by the government,which (at least in theory) can be applied to theschool of their choice, public or private.Theexpectation is that competition among schoolsand the availability of public resources to accessprivate schools would improve the overall effi-ciency of the school system and studentachievement. But research into the impacts ofvouchers has not produced definitive or gener-alizable results—in large part because ofmethodological challenges and the differencesin the specific design and institutional contextof various reforms. Design can vary according tothe size of the voucher, the pool of eligible stu-dents, whether schools can charge more thanthe value of the voucher, and regulations gov-erning school choice (such as whether or notreligious schools are eligible). Institutional man-agement, bureaucratic control, governance ofpublic schools, and oversight of eligible privateschools also vary and influence the results ofprograms.40

Chile has more than 20 years of experiencewith large voucher programs.Yet detailed analy-sis on the effects of competition on school qual-ity in Chile has not led to a consensus onimpact. In the United States, one study foundthat competition improved achievement in thecity of Milwaukee, while another found noimpact. Similar variance is found in the relatedliterature on the impact of school choice.41

Competition between schools and school

choice imply that weak public schools will losestudents and could be forced to close. Success-ful schools would have to be enlarged, or new—and presumably more effective—schools wouldhave to be built. Such institutional change pres-ents significant political, technical, and adminis-trative hurdles.The hurdles are particularlyacute under a universal voucher program thatenables large student migration.

Solid evidence on productivity differencesbetween public and private schools is also lack-ing. Again consider Chile, whose voucher pro-gram generated a large number of new secularprivate schools that operated alongside moreestablished Catholic schools. An analysis ofChilean fourth-grade achievement data showedthat Catholic schools had higher achievementthan public schools in math and Spanish, whilesecular private schools had lower achievement.Another study found that unrestricted nation-wide school choice in Chile resulted in middle-class flight into private schools, but withoutachievement gains.42

Evidence on peer effects that could influ-ence student achievement is equally inconclu-sive. It is not clear whether peer effects are lin-ear, meaning that gains for students who moveto a higher-quality peer group are offset bylosses for either their new or old classmates—or nonlinear, meaning, for instance, that posi-tive peer effects can disproportionately benefitstudents with low socioeconomic status.

While impacts on efficiency are ambiguous,there are reasons to be cautious about the

equity effects of universal voucher programs.They could lead to increased racial and socioe-conomic stratification of schools as parents seekto improve the quality of their children’s peers(such as middle-class flight in Chile). Such strati-fication could occur if all parents were givenvouchers but low-income families were in a lessfavorable position to exercise choice because oflack of information, prohibitive transportationcosts, or extra fees. Disadvantaged studentswould simply be more concentrated in low-quality schools. Echoing similar concerns, arecent study concludes that, in the UnitedStates,“a large-scale universal voucher programwould not generate substantial gains in overallstudent achievement and ... it could well bedetrimental to many disadvantaged students”(Ladd 2002, 4).

There are ways to make voucher programsmore beneficial for disadvantaged students, butthese may reduce political support for such pro-grams. For instance, vouchers and school choicecan be limited to low-income families. Programdesign can also be enhanced by providingtransportation to school, requiring that schoolsdo not charge extra tuition or fees on top of thevoucher, and requiring oversubscribed schoolsto select students randomly. Irrespective ofdesign specifics, a voucher program needs to beembedded in a larger strategy of educationreform that improves the overall institutionalincentive environment for schools and givesunderperforming schools the instruments andresources to improve.

B O X 7 . 5 School vouchers: efficient and equitable?

outcomes. Infant mortality rates amongblacks in South Africa are 5.5 times higherthan those among whites; life expectancyamong the rural Chinese is almost 6 yearslower than among urban dwellers, while thelife expectancy gap between China’s richestand poorest provinces (Beijing andGuizhou) is 10 years.49

These stark differences in outcomes anduse reflect large group-based inequalities inaccess to information, facilities with rea-sonable standards of care and financialprotection from health risk. A lack ofknowledge about hygiene, nutrition, avail-able services, and treatment options, par-ticularly among the uneducated, lowersdemand for health services. Within thehousehold, some family members have lessvoice (women and children) and this canaffect the level of resources used in theirinterest. Health clinics, especially in poorand remote areas, are often inaccessible,have high rates of absenteeism and low qual-ity and responsiveness to clients. Finally, ill-ness is certainly a burden on poor people,but catastrophic health shocks can alsohave disastrous consequences for the not sopoor, mainly through loss of income butalso through high out-of-pocket paymentsfor health care.

These large group disparities in healthoutcomes are inequitable, because theyimply vastly different opportunities to leadproductive lives. And because they oftenarise from failures in markets and agency,reducing these disparities would have largepayoffs in efficiency and productivity. Wefocus here on ways to level the playing fieldfor attaining good health by boosting peo-ple’s knowledge about basic health practicesand services, expanding their access to

affordable care, and enhancing the account-ability of providers.

Expanding knowledgeUnderinvestments in health by patientsmay reflect a lack of knowledge and agencyand incentive issues within the household,as well as a lack of these resources. Lack ofknowledge can keep people from seekingcare when they need it, even when price isnot an issue. As chapter 5 showed, whendeworming medicine was offered free tochildren in Kenya, the take-up rate was only57 percent. Similarly, in Bolivia, many poorbabies are not delivered by a trained atten-dant even though mothers are eligible forfree care. In India, 60 percent of childrenhave not been fully immunized, althoughimmunization is free; mothers cited igno-rance of the benefits of vaccination and notknowing the clinic locations as the majorreasons for why their children had not beenimmunized.

Lack of knowledge can also lead peopleto pay for inappropriate care. Unqualified orunethical providers can overprescribe treat-ments for patients who do not know what isin their best interest. For instance, instead ofeffective and inexpensive oral rehydrationtherapy, a poor child in Indonesia gets morethan four (often useless) drugs per diarrheaattack.50

Education is a natural way to address thelack of patient knowledge. Elo and Preston(1996) estimate that one year of extra edu-cation nationally reduces mortality rates byabout 8 percent—half directly and halfthrough the effects of additional earnings.Female education is particularly powerful.Better-educated mothers are associated withbetter child-health practices, including handwashing, proper disposal of feces, antenatalcare, delivery assistance by trained person-nel, immunization, and well-baby clinics.

Community health agents also providecost-effective instruction in disease preven-tion and healthy behavior. By employingthese nonspecialized personnel, many coun-tries have increased knowledge among thegeneral public at low cost, as with Brazil’sFamily Health Program and Ethiopia’s“mother coordinators,” supporting home-based malaria treatment (box 7.6). Com-


Malaria kills nearly 1 million children inAfrica each year. Empowering mothers totake actions to treat their children in thehome can be highly effective in reducingmortality.The Tigray region of Ethiopiatrained “mother coordinators,” who wereselected from among the community toeducate other mothers on the symptoms offever and malaria. Mothers were provided

chloroquine and information on how toadminister the drug at a cost of $0.08 perchild treatment dose. By educating mothers,Tigray provided rapid and effectivetreatment without forcing the child to relo-cate, which reduced under-five mortality by40 percent and alleviated the burden ofsevere malaria cases on hospitals.Source: World Bank (2004k).

B O X 7 . 6 Working with mothers to treat malaria

munity health workers have also helpedincrease coverage of poor populations cost-effectively.

Public information campaigns can im-prove health knowledge by working throughexisting health clinics or by directly targetingthe community. It is also possible to collabo-rate with the private sector in marketingsocially valuable products, such as insecticide-treated mosquito nets, water purificationmethods, foods rich in vitamin A, andsoap—as with the Central American Hand-Washing Initiative in Costa Rica, El Sal-vador, and Guatemala.51 Media campaignscan also be effective. For instance, frequentbroadcasts of AIDS messages in Thailand,Uganda, and Brazil were a key element in thecampaign to reduce the spread of the dis-ease. The Thai media campaign is creditedwith reducing the incidence of AIDS to apoint at which the country is now able toconsider a fiscally viable treatment programfor AIDS patients.52

But neither information nor free servicesmay be enough to boost use among the lessempowered or those without voice. Mater-nal and child health is often viewed as merit-ing additional intervention. Through condi-tional transfers, Mexico’s PROGRESA (nowOportunidades) program was designed toencourage women to attend pre- and post-natal clinical visits and bring their childrenfor immunization and growth monitoring.The program saw an 8 percent increase inclinic visits by pregnant women in their firsttrimester, which led to a 25 percent drop inthe incidence of illness in newborns and a 16percent increase in the annual growth rate ofchildren between one and three. An impor-tant design feature of the program is trans-ferring funds to women. Although the pro-gram puts more demands on mothers’ time,participants felt that the benefits were worthit. Women also reported feeling more self-confident and having more control overhousehold resources and their time andtravel. Similar schemes are delivering mater-nal and child care services in Brazil, Colom-bia, and Nicaragua.53

Expanding accessAccess to quality health facilities remains aproblem in many areas, often imposing a

greater burden on rural dwellers throughadditional travel time and hospice costs.City dwellers are within easier reach ofhealth centers. In Burundi, 98 percent of theurban population was within one hour of ahealth center, but only 65 percent of therural population was. Even within ruralareas, there is large variation. Only half ofthe poorest rural Nigerians were within anhour of a clinic, but 84 percent of the rich-est were.

Even when health facilities are accessible,they vary hugely in quality. Some have med-icines and drugs in stock, are run by well-trained and motivated staff, and are wellmaintained. But many are not. They areoften dilapidated, rarely have medicines instock, and are run by poorly trained andrude medical staff, who frequently fail tocome to work. It is often precisely the peo-ple who are materially disadvantaged whoalso have to struggle with poor quality and


New studies from India, Indonesia, Mexico,and Tanzania demonstrate that the poor sys-tematically receive lower-quality care fromprivate and public providers.54 The situation isoften worse for ethnic minorities. Evidencefrom Mexico suggests that, even in poor ruralvillages, there is a difference in the quality ofcare between wealthy and poor and betweenindigenous and nonindigenous groups.Among the poorest fifth of the population in

Mexico, indigenous women receive prenatalcare from doctors who rank only in thetwenty-fifth percentile in quality, whileequally poor nonindigenous women receivecare ranking in the fortieth percentile.Thewealthiest fifth fare much better, but evenamong the wealthy, the indigenous receiveworse care than the nonindigenous, suggest-ing that discriminatory practices or culturalbarriers may be at play (see figure below).

B O X 7 . 7 Poor people and ethnic minorities receive lower-quality care

Doctor quality (percentile)

Poorestquintile

Secondquintile

Average Fourthquintile

Wealthiestquintile

Nonindigenous

Indigenous

0

10

20

30

40

50

60

70

Indigenous Mexicans receive lower-quality care, regardless of income

Source: Barber, Bertozzi, and Gertler (2005).

inaccessible health facilities. Ethnic minori-ties often fare even worse in terms of thequality of health care received (box 7.7).

An important obstacle to the provision ofequitable health services everywhere is thedifficulty of enticing urban-educated doctorsto work in poor areas. Chile, Mexico, andThailand have used financial and otherincentives to encourage qualified staff towork in rural areas.55 In Indonesia, doctorshad to complete compulsory service in healthcenters before they could obtain a lucrativecivil service post. Compulsory service was forfive years, with shorter periods allowed forwork in remote provinces. This systemincreased the number of doctors in healthcenters by an average of 97 percent from 1985to 1994, with gains of more than 200 percentfor the most remote provinces.56

Expanding rural health infrastructureand providing incentives to doctors to work

in poor areas may not be affordable for manypoor countries. But there are otherapproaches to reducing the indirect costs(for transport and time spent in transit) andthe medical isolation of poor communities.Roving extension clinics visit sparsely popu-lated areas in Afghanistan, Somalia, andTunisia to provide care locally and offertransportation to better-equipped facilitieswhen required. Bangladesh, Cuba, Gambia,India, and Madagascar have trained commu-nity health workers to scale up service deliv-ery for a wide range of services, includingmalaria prevention, immunizations, familyplanning, treatment of TB patients, homevisits, and neonatal care. The results haveoften yielded a substantial increase in cover-age and measurable improvements in out-comes at much lower unit costs. Through avillage health worker program that moni-tored infants’ weight and health for the firstmonth of life, rural infant mortality inMaharasthra, India, was cut in half from 75.5to 38.8 per 1,000 live births between 1995and 1998.57

Other outreach programs focus onmaternal health and safe delivery. By mak-ing professional midwives and supervisorynurse-midwives widely available in ruralareas, Malaysia and Sri Lanka dramaticallyreduced maternal mortality rates (box 7.8).In Bolivia, expectant mothers with highobstetric risk are transported to larger clin-ics a few days before their due date; in SriLanka, they are picked up by radio-dispatched four-wheel-drive vehicles.

Financing affordable careFor consumers, health care finance systemshave two goals: affordable access to a basicpackage, and financial protection in theevent of catastrophic illness costs. The clas-sic case for government intervention (pub-lic subsidies) is when the full benefit of a“treatment” accrues not just to the individ-ual but also spills over to the communitymore broadly. Interventions to avoid thespread of malaria fall into this category. Abed-net distribution program—involvingthe Red Cross and national ministries ofhealth—increased use among the poorestquintile from 3 percent to nearly 60 percentin a northern district of Ghana and from 18


Despite huge improvements in health, sur-vival, and fertility around the world in recentdecades, global maternal mortality has notdeclined significantly.Two exceptions are SriLanka and Malaysia. In Sri Lanka the mater-nal mortality ratio—the number of maternaldeaths per 100,000 live births—droppedfrom 2,136 in 1930 to 24 in 1996. In Malaysiait dropped from 1,085 in 1933 to just 19 in1997.What can account for this impressivedecline? Improving access for rural and dis-advantaged communities was an importantpart of the strategy in both countries.

Sri Lanka and Malaysia madecompetent, professional midwives andsupervisory nurse-midwives widelyavailable in rural areas. Midwives assisteddeliveries in homes and small rural hospitalsand performed initial treatment in theevent of complications.They were given asteady supply of appropriate drugs andequipment and supported by improvedcommunication, transportation, and back-up services. Besides reducing financial andgeographic barriers, they also helped over-come cultural obstacles and allegiances totraditional practices. Because midwiveswere available locally and were wellrespected, they developed links with com-munities and partnerships with traditionalbirth attendants.

Malaysia and Sri Lanka pursued othercomplementary strategies.Transportation

(in Malaysia) and transportation subsidies(in Sri Lanka) were provided for emergencyvisits to the hospital. In Malaysia, health pro-grams were part of integrated rural devel-opment efforts that included investment inclinics, rural roads, and rural schools. Simi-larly, in Sri Lanka, the government investedin free primary and secondary education,free health care, and food subsidies for alldistricts.The concept was that basic healthcare acts in synergy with education andother types of infrastructure. For example,better roads make it easier to get to ruralhealth facilities and facilitate transportationof obstetric emergencies. By addressing themultidimensionality of equity, these coun-tries made significant health gains.

Dramatic improvements in maternalmortality are thus possible. Just as impor-tant, the experiences of Malaysia and SriLanka show that these can be attained withonly modest expenditures. Since the 1950s,public expenditures on health services havehovered between 1.4 and 1.8 percent ofGDP in Malaysia and averaged 1.8 percentin Sri Lanka, with spending on maternal andchild health (MCH) services amounting toless than 0.4 percent of GDP in both coun-tries. Countries with similar income levelshave significantly higher healthexpenditures and similar, if not higher,maternal mortality ratios.Source: Pathmanathan and others (2003).

B O X 7 . 8 Better maternal health in Malaysia and Sri Lanka

percent to 82 percent in five rural Zambiandistricts.58 Immunizations, vector control,and interventions for tuberculosis, HIV/AIDS,and other communicable diseases are simi-larly deserving.

But the case for government interven-tion goes beyond these well-accepted publichealth reasons: inequality in access to finan-cial protection from health risk based onwealth, ethnicity, and location providesanother important rationale. Out-of-pocketpayments are the dominant form of healthcare finance in lower-income countries. Butliquidity constraints and imperfect creditmarkets often make out-of-pocket pay-ments more difficult for the poor, reducingtheir use rates, and health and productivity.

In Vietnam in 1998, before the establish-ment of health insurance, 30 percent ofpoor households’ nonfood budget went tomedical costs, while only 15 percent ofspending for the richest fifth of the popula-tion was health related. In Cambodia, a sin-gle hospital stay absorbed 88 percent of anaverage household’s nonfood consumptionin 1997 and, for the poorest among them,the cost was higher than the entire nonfoodbudget. In the transition economies ofEurope and Central Asia, with the collapseof prepayment in the 1990s, out-of-pocketspending skyrocketed, accounting for asmuch as 80 percent of health resources inGeorgia and Azerbaijan. In Armenia, 91percent of patients reported having to makesome payment for service received.59 Whilehealth care use has plummeted in theregion, the collapse of prepayment espe-cially hurts poor people.

The regressive nature of out-of-pocketpayments is well understood, but there areno easy answers, especially in low-incomecountries. Given the small formal sectorand limited administrative capacity, theselow-income countries have limited capacityto mobilize resources to pay for essentialhealth services and to establish largeenough risk pools. So, developing countriesface a difficult tradeoff between providing abasic package of health services and extend-ing financial protection.60 Some evidencesuggests that the poor are better able tocover low-cost, high-frequency healthshocks than low-frequency, high-cost

events.61 If so, poor people may be betterserved by having protection against theselow-risk, high-cost events through sometype of pooling mechanism. It is no easytask, however, to cover catastrophic healthrisks in ways that reach the poor.

Reducing out-of-pocket costs involves acombination of pooling health risks and pre-payment—through contributory insuranceschemes, national health services that arefunded out of general revenues, or a mix ofthe two. In all instances, reaching the poorrequires some means of subsidizing theirhealth care costs, so fiscal room and politicalcommitment are crucial. In very low-incomecountries, community insurance schemes,sometimes supplemented through NGO ordonor funding, can provide some protectionto some people, but generally these servicesdo not reach the poorest.

Contributory schemes—private or social—operate best where the share of the formallabor market is high and administrativecapacity is strong. And because premiumsand copayments can be unaffordably high,purely contributory schemes generally bypassthe poor. Private insurance is a significantpart of health finance systems in Brazil,Chile, Namibia, South Africa, the UnitedStates, Uruguay, and Zimbabwe. But in allseven countries, private insurance is used byformal sector workers, leaving the ministriesof health to provide public funds for pro-grams for the poor and underserved.62

Social insurance is characterized bycompulsory coverage financed by employ-ment taxes. Benefits are often limited to con-tributors, and providers are often from thepublic sector even when private providersare eligible. Social insurance has the appealof generating a large risk-sharing pool andcan, in principle, reach the poor throughcross-subsidization. But, when the formalsector is small this potential is limited,because of the difficulties of enrolling a largeenough share of the population. This canturn the system into a ticket to privilegedaccess to health services for some, whileleaving the bulk of the population under-served. For example, in Mexico, social secu-rity health spending per person is five timeshigher than what the Ministry of Healthspends per person.63 And, the payroll tax


required for social insurance introduceslabor market distortions, especially in set-tings characterized by dual labor markets.

The challenges from both an equity andefficiency perspective are enormous, but ahandful of mainly middle-income countrieshave made important attempts to makesocial insurance systems work. Colombia, forexample, has a cross-subsidization schemefor the poor, topped up by general revenues.The scheme has delivered considerable ben-efits: higher coverage among the poor (48percent, up from 9 percent, in 10 years);lower out-of-pocket costs for ambulatorycare; large increases in physician-assisteddelivery (by 66 percent) and prenatal careamong rural women (by 48 percent); andlower child mortality rates (from 44 per

1,000 births to 15) among the insured. Butthere are questions about the program’ssustainability in the face of mounting fiscalcost—reflecting the difficulty of systemicreforms that threaten the privileges ofestablished interests, in this case, publichospitals and the prereform social securityinstitution.64

Ministries of health in many developingcountries operate essentially as nationalhealth services, with nationally owned healthsector inputs and funding from general taxrevenues. The systems they manage are ofteninefficient and inequitable, reflecting severeresource and institutional capacity con-straints but also a bias in favor of the wealthyand influential. Services are meant to covereveryone, but high out-of-pocket paymentskeep many poor people from participating.Countries have tried various approaches toimprove equity in access to health care pro-vided by the national health system, such aseliminating user fees for all, waiving fees, orgiving vouchers to poor people.

In 2001, Uganda abolished user fees forall. The result was a significant increase inhealth care use, lower probability of sickness,and better anthropometric measures, partic-ularly for the poor.65 But the elimination ofuser fees, if effective, can reduce the resourcesfor the health sector, and thus its quality,unless budgetary funding is topped up tomake up for the shortfall. Uganda appears tohave avoided a fall in quality, thanks to alarge increase in the health budget that morethan compensated for the loss in revenuesfrom eliminating user charges.

Introduced in 2002, Thailand’s “30 baht”or universal coverage scheme aims to guaran-tee health care to every Thai citizen. It com-bines previously existing schemes targeted tothe poor and uninsured, and allocates budg-etary resources to providers on a capitationbasis, with only a small copayment per visit(30 baht). The Ministry of Public Healthremains a strategic manager and central fin-ancier, but the district offices make the deci-sions on choice of providers. The scheme hasmarkedly increased use and coverage, withroughly three-quarters of the country bene-fiting from the scheme and 95 percent of thepopulation insured overall, all at a limitedadditional budgetary cost (box 7.9).


With the introduction of the universal cover-age scheme in 2002, almost the entire Thaipopulation now has health coverage (boxfigure 1).This was possible largely becausethe democratic transition of 1997 ushered ina period of increased voice and opennessand raised the political profile of poor peo-ple’s concerns.Technical preparation—withdesign details that had been under consid-eration and subject to experimentation forsome time—also helped to garner supportfor the reform, while prior investments inhealth care infrastructure, establishing ahealth center in nearly every rural subdis-trict, provided assurance of implementationsuccess.

Years of corruption and politicalinequity in the early 1990s, together withthe intervening period of military rule fuel-ing social discontent, prepared the groundfor democratic reforms and a liberal consti-tution in 1997.Two provisions of the newconstitution were important for the ensuinghealth sector reforms: the principle ofequity in health care access; and the scopefor civil society to propose national legisla-tion affecting citizen rights and the role ofthe state, if the measure had 50,000 signa-tures. In March 2000, nearly at the sametime that a feasibility study for the schemewas completed, a network of 11 NGOs sub-mitted to parliament a draft bill calling foruniversal health care coverage. Between1999 and 2001, the press also captured thegeneral public’s interest and kept the issueon the political agenda by highlighting theshortcomings of the current health system.

This political foment attracted theattention of the opposition Thai Rak Thai(TRT) party.TRT adopted the policy becauseit was broadly supported, administrativelyand technically feasible, and consistent withthe party’s ideology.TRT effectively turnedthe 1997 financial crisis into an opportunityby highlighting the health issuesprecipitated by the crisis. Following its land-slide victory in 2001,TRT introduced its ownuniversal coverage legislation, which passedin November 2002.

Source: Pitayarangsarit (2004).

B O X 7 . 9 Mobilizing support for universal coverage in Thailand

1999 2002 20030%

UninsuredPrivate health insuranceUniversal coverage schemeVoluntary health card schemeMedical welfare schemeSocial SecurityGovernment/state enterprise

25%

50%

75%

100%

Thailand’s increasing coverage

But there is broad agreement that exist-ing capitation rates are too low and that thesystem is underfunded. This provides lim-ited incentives for private providers to par-ticipate and could lead to a financialsqueeze on public providers with adverseimpacts on quality.66 Higher use has alsoput a strain on human resources, withincreased workloads (and low pay) acceler-ating the number of physicians leaving thepublic system. Clearly supply-side measuresneed to be considered in tandem withhealth finance reforms to expand access.Still, the achievements have been consider-able, and the scheme has broad popularsupport. Thailand implemented the reformin large part thanks to the popular supportthat democratic reforms made possible;previous investments in sound design andhealth infrastructure also helped.

Other countries reduce costs for thepoor through targeted programs fundedfrom general revenues. Armenia’s targetedfee-waiver system curbed plummeting useamong the poor. But in many instances,simple legislation of free or reduced-priceservices can be counterproductive withoutgood funding and targeting. Targeted pay-ments, through government vouchers andcivil society partnerships with hospitals, canhelp. Vouchers issued to poor patients, as bythe MCH poverty alleviation fund in Yun-nan Province in China, give providers agreater assurance of payment. In Cambo-dia, a promising partnership has emergedbetween government hospitals, MédecinsSans Frontières (MSF) and a small localNGO, covering the hospital fees of thoseconsidered indigent by the local NGO’ssocial workers. Because the hospital is fullycompensated, poor patients receive thesame care as those who can pay.67

Vietnam has introduced health cards forthe poor. More than 11 million of 14.3 mil-lion eligible people benefited from Viet-nam’s program in its first year of implemen-tation in 2002. The program has alreadysignificantly increased the flow of govern-ment health funding to the poor and to pre-dominantly poor areas of the country. Thefunding per beneficiary, however, is consid-ered inadequate and the cost-sharingarrangements are likely to impose too large a

burden on populations in poorer or ethnicminority regions.68

Many developing countries have a mixedsystem, with ministries of health, privateinsurance, social insurance, and targetedschemes coexisting to serve different seg-ments of the population. These multitracksystems tend to fragment, increasing admin-istrative costs, limiting pool sizes, and under-mining both equity and efficiency objectives.Chile’s two-track universal coverage systemhas caused severe segmentation, with thehealthy and wealthy in the private scheme,leaving the public scheme overburdenedwith the poor and ill. Chile is trying to over-come this by creating a “virtual pool,” man-dating a common basic benefits package,instituting catastrophic insurance, allowingportability of benefits between schemes, andinitiating minimum quality and maximumwait-time standards.

Community-based health insurance (CBHI)schemes have developed in some poor com-munities outside the reach of nationalhealth systems. Communities pool healthrisks through voluntary contributions to alocal fund used when any member incurs ahealth shock. The schemes are reported toreduce out-of-pocket spending and in-crease use by their members, but they gen-erally do not reach the poorest andsocially excluded groups or offer membersenough protection from financial risk.Many are limited by their small risk pools,exposing them to low-frequency but high-cost catastrophic events that can outstripthe community fund. Some communitiesaddress large health risks by increasing themaximum benefit, as in Cameroon. Butthey do so by limiting the number of fam-ily’s claims to one a year and by requiringhigh premiums (which prohibit the poorfrom participating).69

Insurance alone is not enough for equi-table use. Inadequate knowledge of thescheme’s benefits and processes and even thepaperwork for submitting claims to commu-nity insurance schemes can be a deterrent.Hospitals often require payment on orbefore discharge, but insurance claims arenot settled until later, requiring patients topay up front. India’s Self-Employed Women’s


Association (SEWA) has been seeking toremedy similar difficulties encountered in itslarge and well-established CBHI scheme.SEWA is testing door-to-door visits formember education, reimbursement assur-ance with selected hospitals, and reimburse-ment to members while still in hospital.70

Enhancing provider incentivesAddressing knowledge, access and afford-ability constraints are important, but theymay not be enough to raise health use andoutcomes. Hours of operation, waiting time,staff disposition, competence and integrity,and the cultural appropriateness of servicesare all important. Complaints of unprofes-sional treatment, abuse, and corruptionabound worldwide. Public medical staffwho take authorized or unauthorized leavefrom public clinics undermine the credibil-ity of the public health sector, drive up costsfor poor families, and induce the poor to useprivate providers, including traditional heal-ers. In Bangladesh, such absenteeism ratesamount to 40 percent for physicians inlarger clinics and 74 percent in smaller sub-centers with a single physician. More gener-ally, poor service delivery has to do withweak management and incentives within thepublic health system—ineffective technicaland structural backup, lack of professionalcareer structures, and inadequate financialincentives all contribute. But the weakdemand for service provider accountabilityand quality is also a problem.71

If they are organized, poor citizens andcommunities can have more voice andgreater power to influence health providers.Governments can help support organizationby communities and enhance provideraccountability. It helps to have well-definedobjectives for health service delivery withtransparent metrics for monitoring progress.This allows for community oversight ofhealth workers and facilities, and when cou-pled with sufficient management autonomyfor providers to reach the established objec-tives, can lead to improved provider incen-tives and accountability.

There is also a need for governments toengage with nonpublic health care providers:in many countries, NGO and privateproviders make up a large part of the health

network. NGOs are particularly helpful inserving remote areas and hard-to-reach pop-ulations: the Bangladesh Rural AdvancementCommittee (BRAC) trains community work-ers to seek out the extremely poor in need ofurgent medical care. In Jordan, half or moreof outpatient visits are to private providers.72

Many private providers offer excellent serv-ices. But some do not—and misdiagnose,misprescribe, or overprescribe treatment. InMexico, even wealthy women receive worsecare from private providers than from publicproviders (Barber, Bertossi, and Gertler2005). Without unduly discouraging benefi-cial private enterprise in health, governmentsneed to ensure accreditation and appropriateregulation for nonpublic providers.

Social protection: managing riskand providing social assistanceSocial protection policies typically havebeen thought of as a form of redistribution.This certainly is important. But more recenttheoretical and empirical work also high-lights a crucial opportunity-enhancing rolefor social protection.73 As chapter 5 showed,pervasive financial market failures in devel-oping countries lead to widespread unin-sured risks and credit constraints. Unequalcapacity to manage risk means unequalopportunities to engage in risky but high-return activities. Families may deal withcrises in ways that narrow future opportu-nities, such as distress sales and forgoinghealth care, schooling, or food intake. Byhelping poor people manage risks, socialprotection programs expand their opportu-nities and enhance overall efficiency.

Even purely redistributive programs canhave important opportunity-enhancing im-pacts. Take the example of social pensionschemes in Brazil and South Africa. Theseschemes are pure transfers targeted to theelderly, geared strictly to avoiding destitu-tion, but they have important welfareimpacts beyond that. They improve therecipients’ access to credit, thanks to theregularity of pension payments, and lead tohigher investments in the household’s phys-ical capital and in the human capital of itschildren and elderly.74

But social protection systems do morethan help individual households avoid des-


titution and expand their opportunities—they can also help societies embark onreforms that would have insurmountableequity and political costs without them.Reforms desirable for their beneficial im-pacts on efficiency and the government’sfiscal position—such as increasing utilityprices, eliminating general food subsidies,introducing a defined contribution pensionsystem, liberalizing trade—may not bepolitically feasible unless policies are inplace to compensate losers. Importantly,permanent social protection can helpreduce the need for special compensatoryprograms for each and every reform75—allthe more important because such programsare difficult to start and stop and are notalways very efficient.

All of this confirms that there is adynamic efficiency rationale for social pro-tection. But there are also important effi-ciency arguments against transfer policies.Design issues are of particular concern,because poorly designed programs can havelarge negative consequences on efficiency.Taxes or contributions have distortionarycosts, especially when they are not directlylinked to benefits (see focus 5 for a discus-sion of tax policies), while transfers candampen work incentives, reduce private sav-ings, and weaken informal insurance mech-anisms. Europe’s experience in the secondhalf of the 1900s suggests that well-designedsocial (and tax) policies can indeed be con-sistent with strong growth thanks to carefulattention to productivity impacts.76

Program choices vary by countrySocial protection generally encompasses twoclasses of interventions:

• Contributory schemes (social insurance)in which the primary focus is on manag-ing risks through smoothing an individ-ual’s income over time and in the face ofdifficulties These programs often poolrisks across large numbers of individualsand include old-age and disability pen-sions and health and unemploymentinsurance.

• General tax funded transfers (socialassistance) in which the focus is onredistribution from the better off to the

poor. These include a variety of cash orin-kind programs targeted at the poor.

These are complemented by labor mar-ket regulations (for example, on hiring andfiring of workers) that are discussed inchapter 9. There is large variation in theshare of GDP spent on social protection,with more-developed regions devoting con-siderable sums (figure 7.5). Almost allcountries spend more on social insurancethan social assistance programs.

There is no consensus on the appropriatebalance of interventions—even in countriesthat have sufficient resources and capacity toimplement any combination desired. Someobservers argue for the universality of socialinsurance programs over the targeted natureof social assistance programs that are basedon political economy considerations. Theyargue that targeted programs are exclusion-ary, by definition, and divisive as a result.77

But a significant group of OECD countries(notably New Zealand, Australia, the UnitedStates, and the United Kingdom) have optedfor systems with heavier components of tar-geted transfers and less generous or less uni-versal programs.78

Many developing countries face con-straints on the choice of systems because oflimited fiscal and administrative capacity.Many poor and even middle-income coun-


Sub-SaharanAfrica*

Middle Eastand

North Africa

Europeand

CentralAsia

East Asiaand thePacific

South Asia LatinAmericaand the

Caribbean

OECD0

2

4

6

8

10

12

14% of GDP

Social insurancespendingSocial assistancespending

Figure 7.5 Almost all countries spend more on social insurance than on socialassistance (percent of GDP)

Source: Data on 74 countries taken from World Bank Public Expenditure reviews or other similar work.OECD data are from the OECD Social Expenditure database.Note:* The average for Africa is based on data for only two countries. OECD excludes those OECD mem-bers (such as Poland and Mexico) that are already accounted for in the regional averages.

tries lack the administrative sophisticationand levels of urbanization and formalemployment needed to administer a socialinsurance scheme, and high social securitytaxes have segmented the labor market andencouraged informality.

What then are the options for the manydeveloping countries that are far from ableto achieve universal social insurance sys-tems? There is a large range of social assis-tance programs, each different in groupsserved, administrative requirements, com-plementary benefits, incentive effects, andpolitical factors (table 7.1). A judiciousblending of these programs can usuallyresult in a social protection system that cov-ers the appropriate groups with feasibleinstruments. The mix of programs selectedand their specific characteristics will dependon context—that is, the risks faced, the levelof urbanization, the age structure, the size ofthe formal sector, the administrative capac-ity, and the complementary social policiesand sociocultural or political factors.

Next, we discuss programs for four keygroups:

• The working poor

• The nonworking young

• The nonworking elderly

• Special vulnerable groups

In many cases, the second two groups are partof households that could benefit from pro-grams that target the working poor. So themore comprehensive the programs for thefirst group, the less the need for programs forthe latter two, and the smaller and morefocused they might be.

Programs for the working poorMost people, especially poor people, rely onlabor earnings for their livelihoods, many inthe informal sector, through subsistencefarming, or as agricultural laborers for others.Labor market risk can be reduced signifi-cantly by improving the functioning of labormarkets and by pursuing sound macroeco-nomic policies (chapter 9). But even a well-functioning labor market will not fully elimi-nate the risk of unemployment. Moreover, inyears with bad crops or low prices, earningsmay not be enough to stave off poverty.

A range of instruments can help addressthe risk of inadequate incomes—for exam-ple, unemployment insurance, needs-basedsocial assistance, or public works. Evenfood, utility, and housing price subsidies aregeared in part to solving the problems ofinadequate labor incomes, although withnotably poor targeting and sometimes largedistortionary costs.

Unemployment insurance, the obviousinstrument for mitigating the risk of jobloss in the formal sector, will not work wellin countries with large informal sectors.Even so, schemes may be able to cover a use-ful share of workers and take some burdenoff programs more tailored to those in theinformal sector (chapter 9). For example, in1998 in response to the East Asian financialcrisis, Korea expanded its young unemploy-ment insurance program to smaller firms aswell as to temporary and daily workers.79

Needs-based cash transfers, the classicsocial assistance instrument, are commonin high-income countries. Such programsare potentially very efficient. Nontransfercosts can be low, usually 5 to 10 percent of


Table 7.1 Examples of social protection programs

Complementary Group served Social insurance Safety net Labor Market Regulation

Working poor or unemployed Unemployment insurance Transfers Minimum wage laws Public works programs Job security regulations General subsidies to food, utilities, or housing Severance pay

Nonworking young Universal child allowances Means-tested child allowances Child labor lawsMaternity benefits Transfers linked to MCH programs

School feedingConditional cash transfers

Nonworking elderly Contributory pensions Transfers Retirement ageSocial pensions

Special groups Disability insurance for disabled Transfers Affirmative action for minorities

total program costs. The programs need notimpose significant forgone earnings on par-ticipants. And they give cash and, thus fullconsumer sovereignty, to the recipient.

But these programs face two challenges.First, they require a targeting mechanism.The classic mechanism in high-income coun-tries has been a verified means test. Becauseincome in these settings is mostly formal, it ispossible and not too costly to collect accurateinformation on income and assets. EasternEurope also has successful experience withmeans-tested programs, although verifyingincomes and assets is more difficult and lessaccurate there than in high-income coun-tries. Latin America’s proxy means tests(relying on easily observable indicators ofincome) have been shown to be fairly accu-rate and low cost. Low-income countries withlarge shares of income from the informal sec-tor experience greater difficulties in settingup targeting mechanism. Although the evi-dence is less clear-cut, community-based sys-tems have been shown to work well in somecountries around the world, especially infairly homogenous rural communities(Albania, Bangladesh, Ethiopia, Indonesia,Uganda, and Uzbekistan) where elite captureis not a major concern. So the targeting issuecould be surmounted.80

The second challenge is perhaps bigger—both for technical design and political sup-port. Needs-based transfers inherently pres-ent a disincentive to work for those ofworking age because entry into the pro-gram (or the benefit level) depends onincome. Traditional mechanisms to partlymitigate the work disincentives includekeeping benefits substantially lower thanminimum wage, as in Bulgaria or Romania,or lower than the earning of low-skilledagricultural laborers, as in Kyrgyz Republic,or using a sliding withdrawal of benefits asincomes rise, as in much of the industrialworld, or an earned-income tax credit as inthe United Kingdom and United States.81

A newer wave of efforts takes a more activeapproach to encouraging independence or“graduation” from the need for assistancethan under the traditional mechanisms justmentioned. Chile’s Puente program usesextensive social work to diagnose each house-hold’s barriers to independence and to pre-

pare customized contracts with the house-hold to address the most important of thesebarriers over a period of two years.Bangladesh’s program of income generationfor the development of vulnerable groups(IGVGD run by BRAC) gives in-kind assis-tance to destitute rural women for a periodof 18 months. During this time, they arerequired to save some money and participatein business training. At the end of the cycle,women have the opportunity of “graduat-ing” into the regular microfinance program.A few programs, as in Romania and Bulgaria,add a public service requirement (thus blur-ring the line between the means-tested socialassistance and public works programs).82 So,whether through traditional or more inno-vative mechanisms, the disincentive problemcan also be mitigated.

Public works programs that support theworking or unemployed poor have been usedin many countries (box 7.10). By offeringemployment for low wages, these programsself-select the able-bodied poor, avoidingboth the means-testing and work disincen-tives. In good programs, the work is in high-return activities that create assets and serv-ices. The self-targeting aspect is usefulbecause informality is widespread in develop-ing countries and incomes are hard to assess.It is doubly useful as part of a countercyclicalmeasure in fighting poverty during periods ofcrisis—workers leave the program when theirregular source of livelihood picks up againafter the crisis. Public works programs forinfrastructure are especially welcome in low-income countries, postconflict settings, andsometimes post–natural disaster settings.

Public works programs have some disad-vantages too. The administrative capacity toselect and run the programs is significant.Indeed the often-cited good programs are aminority of all the public works programsimplemented around the world. Many havefailed, often over the inability to line up anddeliver useful public works, to provide suffi-cient nonlabor inputs, or to set the wage right.

Even when programs are well run, the netbenefits transferred to participants are oftena small share of total program costs. First,management, materials, equipment, andskilled labor requirements can run up to 40to 60 percent of program costs. Second,


workers have to forgo some income to par-ticipate: in the absence of the program, theyusually would have been able to pick upsome part-time work or engage in low-returnself-employment activities. In Argentina’sTrabajar, forgone earnings were half of grossearnings; in Maharashtra’s EmploymentGuarantee scheme, 53 percent; and inBolivia’s ESF, 60 percent.83 Theoretically, for-gone earnings can be minimized by allowingflexible working hours or part-time work,but this can complicate the supervision ofthe public works and the workers.

Programs for the nonworking youngEarlier, we focused on interventions geared toovercoming the disadvantaged family circum-stances of young children through a variety ofservices that improve their cognitive and socialskills and prepare them for learning in school.Here we focus on ways of augmenting thefamily’s income to alleviate poverty amongchildren and improve their chances in life.

Programs aimed at children can be uni-versal or means-tested, free-standing orlinked to the use of health and educationservices. In both Eastern and WesternEurope, the traditional approach to incomesupport for children is through childallowances, independent from but comple-mented by extensive public education andhealth care. Most Western European pro-grams are universal, although a few aremeans-tested (Italy and Spain). A higherfraction of Eastern European programs andthose in other middle-income countries aremeans-tested (Bulgaria, Belarus, the CzechRepublic, the Kyrgyz Republic, Poland,Romania, the Russian Federation, Serbiaand Montenegro, the Slovak Republic,Argentina, Chile). In Africa, Latin America,and Asia, there is a long and extensive his-tory of school feeding programs and mater-nal and child health programs that distrib-ute food (or occasionally food stamps).Many of these programs rely on existing


Public works programs have beendemonstrated to work in some middle-incomecountries (Chile, Argentina, and South Africa)and low-income countries (Senegal, Kenya,India, and Bangladesh)—and not to work inmany others.This international experienceoffers several lessons in the design and imple-mentation of public works programs.

Wage rate. The key to self-targeting is settingthe wage rate low enough—no higher than themarket wage for unskilled manual labor in agri-culture or the informal sector during a normalyear.While determining the precise level of thewage rate may not be easy, it is better to startwith a wage rate that is too low—if there is nodemand at the offered wage rate, it can beraised. Setting a low wage rate level does morethan ensure that the workfare scheme will bewell self-targeted. It also maintains the incentiveto take up regular work when it becomes avail-able, and it helps ensure that the program canreach as many of those in need as possible.

Conditions of eligibility. Rationing should beavoided; ideally the only requirement should bethe willingness to work at the offered wage. Ifrationing is unavoidable (for example, if thedemand for employment at the wage setexceeds the available budget), explicit secondarycriteria should be used—the program may tar-get poor areas, work may be offered only in sea-sons of greatest need, the length of employmentof any individual may be limited, or such addi-

tional options as community-based selection ofthe neediest or a lottery may be implemented.Least desirable is rationing with entrydetermined by foremen or political figures.Women’s participation can be enhanced throughnondiscriminatory wages, the provision of onsitechild care, and adequately private latrines.

Employment guarantee. A workfare programthat guarantees employment can reduce thelonger-term risk the poor face.While highlydesirable, guarantees have not been a feature ofmost public workfare schemes. One exception isthe Employment Guarantee Scheme in Maha-rasthra, India, which guarantees unskilled man-ual work within the district within 15 days ofregistering for employment with the scheme.While this does not necessarily mean locallyaccessible employment, the scheme comes clos-est to offering a guarantee of any kind. Indiarecently announced its intention to extend theguarantee by providing 100 days ofemployment on rural public works projects at aminimum wage.The scheme is not far enoughinto implementation to draw lessons. Murgaiand Ravallion (2005) simulate some possibleoutcomes for a range of design parameters: thetargeting could be good and impacts of povertylarge, but the costs could also be substantial—1to 2 percent of GDP for the 100-day scheme.

Labor intensity. The labor intensity—that is,the share of the wage bill in total costs—shouldbe higher than normal for similar projects in the

same setting.There is a tradeoff between imme-diate income gains through employment of thepoor, and gains to the poor from the quality anddurability of the assets created. In a crisis situa-tion, in which current transfers to the poor havehigh weight, a high labor intensity is desirable.Illustrative average labor intensities range from0.5 to 0.65 percent in low-income countries andare somewhat lower (0.4) in middle-incomecountries, although labor intensity often variessignificantly by subprojects.

Administration and implementation. Admin-istering and implementing an effective schemeis hard—requiring the selection and manage-ment of a plethora of small projects over a widegeographic area and many administrative enti-ties. Ideally, public works schemes require amenu of works well-integrated into the localplanning process yet elastic in size and timing.This can be difficult in low-capacity settingsbecause of the forward planning andinteragency coordination needed. In high-capacity settings, fitting many small labor-intensive projects into the sophisticated andoften capital-intensive infrastructure plans oflarge- and middle-income cities can be difficult.Moreover, ensuring that the workfare program ispoverty focused is not easy because of conflict-ing pressures from alternative target groups,such as the skilled unemployed.

Sources: Subbarao (2003) and Murgai and Ravallion(2005).

B O X 7 . 1 0 Public works programs: key issues

service delivery mechanisms to make theiradministration feasible or cheaper.

A new wave of CCT recognizes thatimperfect markets can lead to underinvest-ment in human capital (chapter 5) andexplicitly seeks to enhance the opportunity-generating potential of income supportthrough links to the use of services. CCTprograms are now being implemented inabout two dozen countries, mainly in LatinAmerica—but they are being discussed bymany other countries and in all regions.These programs transfer income in cash orin kind to poor households with children.They grant benefits only if children complywith standards for attendance in school orparticipation in a health care program. In theCCT programs with good data, the targetingoutcomes have been quite good at generallyreasonable administrative costs. All five pro-grams reviewed by Morley and Coady (2003)distribute far more than a proportional shareto the bottom quintiles (table 7.2). On aver-age, the share of program benefits going tothe bottom 40 percent of the population isan impressive 81 percent. The evidence onpoverty impact is more limited, but PRO-GRESA (now Oportunidades) had a power-ful effect: program communities experienceddeclines of 17.4 percent in the incidence ofpoverty compared with the control group.84

The conditioning of benefits on use ofhealth and education services serves thedual objectives of avoiding severe depriva-tion and enhancing opportunities forhuman development. But there is a tensionbetween these goals. A simulation of theresults that might be expected from the fed-eral Bolsa Escola program in Brazil showsonly a small reduction (1 percentage point)in the poverty index because of the (simu-lated) loss of labor income of children whodrop out of the labor force to attend school.Mexico’s PROGRESA (now Oportunidades)had impressive poverty impacts butincreased primary enrollment rates by onlyabout 1 percentage point because they werealready above 90 percent. Cambodia’s pro-gram, which focuses on grades seventhrough nine, may well help with the transi-tion to secondary school, but it misses someof the poorest households because so manyhave dropped out by then.85

In settings with low access to health andeducation services, this tension means thatconditional transfer programs may not beappropriate vehicles for social assistance.The conditions would keep the programfrom serving the poorest. The opposite maybe true as well: when the use of services isalready satisfactory, it may not be worthusing administrative resources to verifycompliance with service use conditions.

Programs for the nonworking elderlyMost countries have public pensions pro-grams for the elderly. Two arguments pro-vide a rationale for governments to mandatea pension system to provide for old-agesecurity: imperfect financial markets limitthe scope for redistribution over one’s life,and human “failures” to see far enough intothe future may lead to undersaving for oldage. The need for old-age security will grow.The population of 60 year olds, about 10percent of the world population today, isprojected to reach about 21 percent by 2050.Within this group, the fraction of peopleover the age of 80, about 12 percent today, isexpected to reach 19 percent by 2050.86

Contributory pension programs havenot solved the problem of old-age security.Coverage is low—only 20 percent of theglobal workforce. Even in pension systemswith extensive coverage, the lifetime poorcannot contribute enough to have a pensionat old age that would keep them out ofpoverty. Elderly women who have notworked outside the home are particularlyvulnerable. Moreover, in some countries,such as in Kenya, Uganda, Sri Lanka, andZambia, poorly governed schemes gave


Table 7.2 Targeting performance of conditional transfer schemesa

Quintile PRAF RPS PROGRESA SUF FFE (Honduras) (Nicaragua) (now Oportunidades) (Chile) (Bangladesh)

(Mexico)

1 43 55 40 67 —

2 80 81 62 89 48

3 94 94 81 97 —

4 98 99 93 100 —

5 100 100 100 100 100

Source: Morley and Coady (2003), table 5-3.Note: PRAF = Programa de Asignación Familiar; RPS = Red de Protección Social; SUF = Subsido Unitario Familiar; FFE = Food For Education; — = not available. a. Cumulative share (percent) of benefit captured, by income quintile.

lower-income workers returns less thanbank deposits and the alternatives of invest-ing in land, tools, or a vehicle.87

Options for assisting the elderly poorinclude the following: broadening pensionsystems to include more people, adding aredistributive element as part of an existingcontributory pension scheme, or coveringthem through a separate “social pension”financed by general revenue. A fourthoption is a general needs-based social assis-tance program.

Broadening the coverage of contributorypensions has been attempted, generally withlittle success. In the Republic of Korea, man-dating the expansion of coverage to farmers,fishermen, and self-employed was met withmassive protests; in the end, the governmenthad to subsidize in full or in part the contribu-tions of almost two-thirds of the target popu-lation. Adding a redistributive element iscommon, but as pension reforms strengthenthe link between contributions and benefitsfor efficiency reasons, that redistributive ele-ment is becoming smaller.

Social pensions provide transfers to theelderly without requiring prior contributionsor withdrawing from the labor force.88 Theycan be universal, as in Botswana, Mauritius,Namibia, or Bolivia. Or they can be means-tested, as in South Africa, Senegal, India,Bangladesh, a number of Latin Americancountries, Australia, Italy, and New Zealand.Many of them complement contributory sys-tems that cover higher-income groups. Whenthe transfers are means-tested, the programsare really a special case of needs-based cashtransfers limited to the elderly. The targetingchallenges discussed earlier and the potentialsolutions are similar. Labor disincentives arelessened, however, because societies expectlower work efforts from the elderly.89

Evidence from various countries imple-menting large social pension schemes indi-cates that the costs are 1 to 2 percent of GDP,not negligible for low-income developingcountries. Schwartz (2003) simulates thecosts in six African countries of providingsocial pensions, limiting the benefit to 40percent of GDP per capita and eligibility tothose age 75 and above. The costs wouldrange from 0.2 percent of GDP in Kenya to0.7 percent of GDP in Ghana, still not insub-

stantial. Kakwani and Subbarao (2005) sim-ulate various options for 15 African coun-tries, and conclude that the best—taking intoaccount poverty impact, fiscal cost, andincentive effects—is to keep the benefit low(about one-third of the poverty threshold),the eligible age limit at 65 or older, and totarget only the elderly poor, thus sacrificingthe administrative simplicity and politicaladvantage of universalism. There is enoughvariation across countries to warrant country-specific efforts to determine benefit and eli-gibility levels and targeting methods ratherthan relying on rules of thumb.

How should we think about the balancebetween social pensions for the elderly andother programs, such as those targetingfamilies with children? Are the elderly poormore deserving than other poor? Brazilspends 1 percent of GDP to transfer $70 amonth to 5.3 million elderly poor and only0.15 percent of GDP to transfer $6 to $19per month to 5 million families to supportschool attendance through the Bolsa Escolaprogram.90 When considering whether thisthe right balance, one can argue generallythat young families with children, who havetheir entire lives ahead of them, should havehigher priority. Indeed some argue for shift-ing public spending away from pensionsand toward families with children in Braziland others suggest that a focus on unem-ployment may be more appropriate toreduce poverty in South Africa.91

There may be important political econ-omy reasons why programs for the elderlygarner such political support. There is directevidence from attitude surveys, across soci-eties and age groups, that concerns about oldage poverty are strong and widely shared—perhaps because most people expect to be oldone day (but not necessarily unemployed, ora single parent, or disabled) and also becauseold age is more easily verifiable and less sub-ject to moral hazard, for example, when com-pared with unemployment insurance.92

Programs for special vulnerable groupsSome groups are vulnerable regardless ofage—the disabled, HIV infected, ethnicminorities, certain castes, internally dis-placed households, refugees, and orphans


(box 7.11).93 One of the key issues in provid-ing transfers for these groups is whether toset up specific programs or to include themin a more general program. There is no uni-versal answer to this, and a complex set ofissues must be diagnosed in each case. One istargeting. Not all orphans, widows, or dis-abled are poor, so universal programs willinclude some non-poor. A second issuerelates to the special needs of the groups. Willincome support alone be sufficient, and ifnot (as is usually the case) does it make senseto link the income support to other pro-grams for the group? For example, whenlarge groups of internally displaced people orrefugees emerge suddenly, their needs forhousing, food, and health care may strainlocal availability. In such cases in-kind provi-sion to the group is usually the first response.Only after the groups become long-standingor somewhat smaller does the question ofwhether to switch to a needs-based cashassistance or workfare program arise. Specialprograms for groups viewed positively or asdeserving, such as veterans or the disabled,may have adequate political support, but ifthe group is excluded, as ethnic minorities orthe HIV positive are in many cases, such pro-grams may not garner sufficient support.

While standard transfers may protectthese groups, a broader set of policies canhelp expand their opportunities and facili-tate their integration into society. Somecountries have used affirmative action (seechapter 8). Others have used regulationsand awareness campaigns that sanctionharmful local practices to help prevent dis-crimination. Policy responses include creat-ing a framework to hear advocacy groupsand mainstreaming such concerns into gov-ernment practice, often building on infor-mal or private arrangements, such as thoseof faith-based organizations.

SummaryEquity in the acquisition of human capaci-ties—through early childhood develop-ment, formal education, health services,and social protection—is at the core of astrategy to equalize the opportunities forpeople to lead productive, fulfilling lives.Broad provisioning of these services is alsogood for development and poverty reduc-

tion through impacts on innovation, pro-ductivity, and social cohesion. But there arebig challenges to equitable provisioning—getting the relevant issues on the policyagenda, fighting political capture of institu-tions so that they do not only serve thepowerful and the influential, and managingefficiency-equity tradeoffs, especially in theshort term. There are also good prospectsfor incremental change through advocacyto point out long-run benefits even whenthere are short-run costs, through soundprogram and tax design to minimize effi-ciency costs and build accountability struc-tures, and through political coalitions thatcan thwart elite holdups.

The power of greater equity in humancapacities to unhinge inequality traps istremendous—through directly contribut-ing to leveling the economic, political, andsociocultural playing fields. But achievinggreater equity in human capacity is notenough to break the inequality trap. Itneeds to be complemented with fairness inthe returns to those capacities and in theaccess to complementary assets, topics dis-cussed in the next chapter.


Conflict and the HIV/AIDS pandemic aregenerating a major humanitarian crisis forfamilies in Sub-Saharan Africa.There are asmany as 43 million orphaned children in theregion today, 10 percent of whom have lostboth parents. Orphans make up more than15 percent of all children in 11 countries,and the numbers are rising .

The death of an earning family memberis most likely to drive a family into penurybecause of the costs of funeral, the loss ofregular income, and the risk of losing one’sproperty. Erosion of human capital isanother major risk: microstudies and analy-sis of household surveys suggest that, rela-tive to other children in the household, fos-tered children are underenrolled in schools,work longer hours doing household chores,and have lower immunization coverage—and the disadvantage is stronger forfostered girls than for fostered boys. Psycho-logical risks are also high because the deathof a parent often leaves the child in a stateof trauma, lacking nurturance andguidance, and impeding socialization.

The main coping strategy in Africa is fos-tering by the extended family.When possible,

interventions should first try to strengthengrassroots responses to orphan care, and turnto supplementary interventions only whenthe extended family is no longer sufficient orcapable.When no other living arrangement ispossible, experience and research show thatorphanages must be the “last resort.”Recog-nizing the scope for exploitation of vulnera-ble children under all arrangements, appro-priate checks and balances must be in place,including oversight by NGOs or community-based organizations.

When access to basic education andhealth services is generally low, waivingschool fees and uniform obligations wouldhelp increase enrollment rates of allchildren including orphans, as in Uganda.When average access to services is high, butthe difference in access between the poorand the non-poor, and between orphansand non-orphan children, is large, cashtransfers conditional on children attendingschool seem appropriate. Innovativeprograms along these lines are just begin-ning (as in Swaziland).

Sources: Subbarao and Coury (2004); USAID,UNAIDS, and UNICEF (2004).

B O X 7 . 1 1 Africa’s orphans and public action

It takes more than building human capacitiesto broaden people’s opportunities. Peoplealso need complementary assets, access to themarketplace, and security of person andproperty. This chapter starts with a descrip-tion of justice systems, showing how criticalthey are in ensuring a level playing field andfair returns. It then turns to policies forexpanding access to the complementary assetsof land and infrastructure. Promoting fair-ness in markets is addressed in chapter 9.

Building equitable justice systems Society’s rules, and the institutions thatestablish, maintain, and transform them,govern market and nonmarket interactions.They determine people’s endowments, theirrights and obligations, and their ability togenerate fair returns. Reflecting and pro-ducing the distribution of power amonggroups, good institutions (so necessary forprosperity) emerge only when the distribu-tion of political power and enforceablerights is equitable.

Legal institutions play a key role in the dis-tribution of power and rights. They alsounderpin the forms and functions of otherinstitutions that deliver public services andregulate market practices. Justice systems canprovide a vehicle to mediate conflict, resolvedisputes, and sustain social order. Butinequitable justice systems may perpetuateinequality traps by maintaining or reproduc-ing elite interests and discriminatory prac-tices. Equitable justice systems are thus cru-cial to sustained equitable development.

Building more equitable justice systemsruns into three main challenges—ofteninterrelated and reinforcing. First, legalinstitutions may be open to capture by eliteinterests or may discriminate against certaingroups. Second, these institutions are often

inaccessible, because they are incompatiblewith local norms and customs and they arephysically or economically inaccessible, orbecause people lack the knowledge or capac-ity to navigate the system. Third, elitecapture and the inaccessibility of the legalsystem may mean that policies relating tocrime and personal security are inequitableand perpetuate crime-related inequality traps.

Combating elite capture and discrimination Political and economic elite interests oftencoincide at the expense of a disempoweredmajority. When power is in the hands of anarrow elite, the rights of most citizens areunstable. A century of banking in Mexico,outlined in chapter 6, illustrates how dealsbetween the political and economic elite ledto the establishment of banking monopo-lies and laws that maintained a system ofrent-sharing between banks and govern-ments.1 Another striking example of elitecapture comes from the transition economiesand the rise of oligarchs who manipulatedpoliticians and shaped institutions to getrich.2 Legal systems that cater to narrowinterests also tend to discriminate againstother groups through inequitable laws andpractices.

Ensuring equality before the law andsecuring both personal and property rightsfor a broad section of the community giveindividuals the incentive and the opportu-nity to take part in economic and politicallife. This requires an independent andaccountable judiciary and laws and prac-tices that protect citizen’s rights in a nondis-criminatory way.

Enhancing judicial independence andaccountability. In many countries, a rule of

Justice, land, and infrastructure

8c h a p t e r

156

Justice, land, and infrastructure 157

law system—administered by multiple armsof government—constrains political power.3

In this system, an independent judiciary actsas a safeguard against abuses of state andnonstate power. Because judges are also opento elite capture and corruption, accountabil-ity mechanisms are a key aspect of legitimatejudicial independence.4

In many developing countries, shiftstoward an accountable and independentjudiciary require a change in culture andinstitutional practice. Ethiopia established anindependent judiciary for the first time in1995.5 In Vietnam “telephone justice” wascommon, with party elites habitually contact-ing judges to direct decisions.6 Changingingrained institutional practices in bothcountries has been a slow process. Poor con-ditions of judicial service in many countriescan increase corruption.7 For example, lowremuneration for magistrates in Kenya madethem open to alternative funding for theirservices; Kenya removed almost one-third ofjudicial staff for corruption in 2004.8

Promoting judicial independence with-out establishing accountability mecha-nisms can further entrench elite interests.Institutional safeguards, transparency, andthe existence of a civic constituency are keyto both accountability mechanisms andjudicial independence. Institutional safe-guards include providing for security oftenure and improving conditions of serv-ice for judges; rigorous and transparentappointment and disciplinary processes;transparent mechanisms of case allocationand case management; transparent andopen hearings; appeal rights and the publi-cation of judicial decisions; and publicinformation about the courts.9 Manycountries have enshrined judicial inde-pendence in the constitution or statelaws.10 Bolivia has established open com-petitions for judgeships and ethical stan-dards for judges. Courts in the Philippineshave a performance management systemfor judicial and nonjudicial personnel.

Public information campaigns can en-hance the independence and accountabilityof the courts, increasing public confidence inand commitment to the system, and enhanc-ing people’s capacity to demand better gov-ernance and hold those in positions of power

accountable. In Colombia public informa-tion centers in major courts disseminateinformation and help people use the court.In Venezuela information is provided to thepublic through an Internet-based judicialportal for the Supreme Tribunal.11

Strengthening the relationship amongcivil society, the media, and the courts hasalso improved public awareness and scrutinyof the judicial system. Bad judges haveresigned because of high-pressure mediacampaigns, such as the recent mediascrutiny in the Philippines.12 The media canalso disseminate information, such as the“My Rights” television show in Armenia(box 8.1). Similar shows have been devel-oped in other parts of Eastern Europe. InGeorgia, an NGO disseminating informa-tion about the courts increased public satis-faction with the courts.

The existence of an independent andaccountable judiciary is not enough to pro-tect citizens against abuses of state power.Adequate laws and institutional mecha-nisms are also needed. In Thailand, forexample, separate administrative courtswere established for the first time in 2001 toprotect citizens against arbitrary uses ofstate power. The courts aim to ensure thatstate authorities act in accordance with statelaws and regulations. They also aim toenhance citizen participation in public pol-icy formulation and oversight. In the firstthree years, the courts processed almost

Many people in Armenia have no understand-ing of the legal system or the rights affordedthem under the law.And distrust of the courtsis widespread. In a recent public awarenesscampaign,the government funded a televi-sion show to provide citizens with examples,advice,and information on their legal rights.

“My Rights”uses mock trials to depictreal-life disputes in Armenian courts.The tel-evision judge is a deputy minister of justice,and the parties are often those in the realdispute.The topics—such as rental and prop-erty disputes, customs issues, and family lawmatters—are timely and of broad interest. Alive studio audience of judges, lawyers, legalofficials, and others discusses the trials on air.

The show airs once a week on Armenia’sstate television channel. After only five or sixshows,“My Rights” became the number oneshow in Armenia.There have been numer-ous reports of viewers requesting legal doc-uments and decisions from notaries, judges,and other legal officials based on what theylearned from the show. And when thepower went out in one village a fewminutes before “My Rights” was going onthe air, the people in the town marched onthe mayor’s office and accused the localofficials of intentionally cutting the powerso that people could not watch the show!

Source: Decker and others (2005).

B O X 8 . 1 Increasing legal literacy and public awareness:“My Rights” on Armenian public television

17,000 cases, most concerning corruptionor other unlawful acts by public officials.Many of the cases made front-page news inThailand because of their social impact.

Combating discriminatory norms and prac-tices. Laws that reinforce exclusionary prac-tices in norm-based institutions perpetuateunequal power relations. Some laws maydiscriminate against particular groups, suchas laws affecting indigenous people or thelaws in apartheid South Africa. The absenceof laws can also reinforce unequal powerrelations as for domestic violence, often rele-gated to the nonlegal private realm.

In many countries, antidiscriminationand equal opportunity laws have reduced dis-criminatory practices. Historical disadvan-

tage may mean, however, that legal equality isnot enough. Some countries have passed lawsthat discriminate in favor of certain groups,creating affirmative action programs on thebasis of race, ethnicity, and gender or for peo-ple with disabilities. An assessment of two ofthe most widely implemented affirmativeaction programs, in India and the UnitedStates, suggests mixed impacts (box 8.2).

The mere existence of “equitable laws”for affirmative action does not guaranteetheir equitable implementation or enforce-ment. For example, in Peru and Honduras,gender discrimination in judicial decisionsand treatment by police and judges discour-age women from using the system to resolvedisputes.13 Such disadvantaged groups aremore likely to experience the law-and-orderside of the law than the protection of theirrights (as discussed below under crime andpersonal security).

Making justice accessiblePeople’s legal rights remain theoretical if theinstitutions charged with enforcing them areinaccessible. Accessibility depends on howcompatible laws are with the norms andunderstandings that shape people’s lives.Legal institutions need to be physically andeconomically accessible and people need tohave the knowledge and capacity to claimtheir rights.

Addressing the compatibility of state andcustomary justice systems. Forms of cus-tomary or nonstate law operate in amajority of countries.14 Yet they are oftenneglected in justice sector reform policies.Engaging with customary systems is animportant part of equitable reform strate-gies for two main reasons. First, customarylaw is often a fundamental part of a com-munity’s identity and belief system; thus, alack of recognition can be intrinsically dis-criminatory and serve to exclude communi-ties from the wider state system. Second, afailure to engage with customary systemsmay leave inequitable and inefficient prac-tices at the local level unchecked.

Where state and nonstate systems havedeveloped in tandem, they often comple-ment each other and reinforce sociallyaccepted codes and rules. But in communi-


The affirmative action program in India isbased primarily on caste and gender andthat in the United States primarily on race.Before independence in India, the Britishgovernment introduced affirmative actionto address discrimination against“untouchable” castes (now known as Dal-its) and “tribals” (now known as Adivasis).After independence in 1947 the policy ofreserving 22.5 percent of seats in educa-tion institutions, government jobs andelectoral seats was written into the consti-tution. Since 1991, a further 27 percentquota has been introduced for other lowcastes (called Other Backward Castes), butwith no constitutional guarantee. Andsince 1993, 33 percent of the seats in localgovernments have been reserved forwomen, Dalits, and Adivasis (Deshpande2005).

In the United States, slavery was perva-sive for more than two centuries, and notuntil 1866 were blacks granted citizenshiprights.The system that replaced slavery wasonly marginally better, with several featuressimilar to the Indian caste system: segrega-tion, denial of education, restrictions to low-paid, menial jobs, social and economic dis-crimination, negative stereotyping, andviolence.The Civil Rights Act of 1964 andsubsequent legislation, Supreme Court rul-ings, and executive orders in the 1970s intro-duced affirmative action into the political,judicial, administrative, and economicspheres of American society. Starting withthe label “equal opportunity,”selection pro-cedures incorporate compensatory correc-tion to ensure adequate representation of

minorities in education and employment(Deshpande 2005).

The programs in both countries havebecome centerpieces of political battlesover race and caste. Critics argue that theytend to benefit the upper echelon of minor-ity groups, and they are difficult to end. InIndia, the programs are said to apply to sub-castes that have not traditionally faced dis-crimination (Sowell 2004).They may alsoreinforce negative stereotypes by placingminorities in positions they are not qualifiedfor (Coate and Loury 1993). Despite theseweaknesses, India’s program has providedformal sector employment and higher edu-cation for many Dalit and Adivasi families,freeing them from subservient roles.Withthe reservations in local government,elected women leaders make decisions inline with women’s needs (Chattopadhyayand Duflo 2004). Low-caste representativesin state assemblies increase the allocation ofquota-based jobs to low-caste constituents(Pande 2003). And Dalit representatives invillage government improve the targeting ofbenefits to Dalits (Besley and others 2004).

In the United States, disparities betweenblacks and whites continue to be significanton all economic indicators, and there is evi-dence of discriminatory gaps in earnings.But affirmative action in jobs has increasedblack employment and enrollment inhigher education (Holzer and Neumark2000, Bowen and Bok 1998). But the U.S.program’s quasi-voluntary element meansthat litigation can dilute the program, andblack representation in government bodiescontinues to be very low.

B O X 8 . 2 Affirmative action in India and the United States

ties where the state systems lack legitimacyand political reach, customary systemsoften act independently from the state legalsystem, which may be rejected, ignored, ornot understood. Real difficulties arise whenlocal customary systems are at odds withthe rights and responsibilities articulated instate law.

In many developing countries, custom-ary systems are the dominant form of regu-lation and dispute resolution. In SierraLeone, about 85 percent of the populationfell under customary law as of 2003.15 Cus-tomary tenure, discussed below, affects 90percent of land transactions in Mozam-bique and Ghana.16 Customary justicedepends on local traditions, as well as thepolitical history of a country or region.Ethiopia officially recognizes more than 100distinct “nations or peoples” and more than75 languages.

Customary systems can be incompatiblewith economic, social, and civil rights.Many forms of customary law are seen todiscriminate against marginal groups. Inmuch of Sub-Saharan Africa, for example,customary systems systematically denywomen’s rights to land, assets, or opportu-nities.17 Customary practices are also seenas archaic and rigid—not amenable tomodernization, efficient market relations,or broader development goals. They areoften seen as overly localized and complex,making more generalized reform initiativesdifficult. They can lack legitimacy at thelocal level. For example, many systems inSub-Saharan Africa have been substantiallydistorted by colonial rule, which often usedlocal chiefs to maintain control and estab-lished more authoritarian and ethnic-basedstructures than previously existed.18

However, it is wrong to presume that allcustomary law discriminates against mar-ginalized groups—or that western law doesnot. For example, in the AmaHlubi commu-nity of KwaZulu Natal Province in SouthAfrica, women and men are consideredequal, with both entitled to own property.19

Furthermore, there are often good reasonsfor people to choose to use customary sys-tems. The state systems may lack legitimacyor be seen as mechanisms of control used byoppressive regimes. Or the state systems may

lack capacity, be inaccessible, or dramati-cally increase transaction costs.20 In ruralTanzania, a perception that state institutionscan not supply law and order has led to theemergence of “new” forms of organized vil-lage defense groups called sungusungu.While technically illegal, the sungusungu areoften informally supported by the state,given their success in reducing crime.21

Ignoring or trying to stamp out custom-ary practices can also have serious negativeimplications. Top-down reform can under-mine informal institutions without provid-ing viable alternatives, and the vacuum canlead to power grabbing, lawlessness, or evenviolent conflict. When neither formal norinformal mechanisms are functioning,human rights abuses and serious conflictare more likely. For example, a study inrural Columbia found the incidence of vigi-lantism, “mob justice,” or lynching to be fiveand a half times greater in communities inwhich informal mechanisms are no longerfunctioning effectively and the state pres-ence remains limited.22

A failure to engage with customarysystems may mean that discriminatorypractices go unchallenged. While state lawofficially protects women’s rights in manycountries, local norms and power structurescontinue to make it almost impossible forwomen to claim these rights.

Considered attention to customary sys-tems in broader institutional reform is fairlynew. But many governments, such as SouthAfrica, have begun working toward inte-grating customary institutions into widerstate frameworks (box 8.3).23 Many coun-tries have attempted to integrate customaryland systems into formal land law systems(box 8.7). Local NGOs and communitygroups have also helped empower marginalgroups to challenge discriminatory normsat the local level.

Establishing adequate and open legal insti-tutions. Even when formal systems doexist, they often lack adequate infrastruc-ture or are so institutionally weak that citi-zens cannot claim their rights. Formal insti-tutions may exist only in large cities, andeven then excessive delays, unfair proce-dures, or unreasonable costs may leave


them inaccessible to much of the commu-nity. 26 Institutions can also be inaccessibleif people do not know their rights and can-not navigate the systems charged with pro-tecting them.

A large array of information campaignshas informed citizens about their rights(discussed above). But even if people knowtheir rights, they may have limited capacityto navigate the system. Access to legal serv-ices is often restricted or costly. In Hon-duras, legal fees to obtain a monthlyalimony of 100 lempiras (US$5.30) in a

child-support case could amount to asmuch as 2,000 lempiras (US$106.00), oralmost two years of alimony. Adding to thecosts are requirements that parties be repre-sented by lawyers. For example, most LatinAmerican countries do not permit self-representation, effectively denying accessbased on economic status. Legal aid canincrease people’s access to basic legal servicesand the courts (box 8.4). So can communitymediation centers, lay judges, and mobilecourts. The mobile courts introduced inremote areas of Brazil in 1999 have beenreplicated in the Philippines and Mexico. InGuatemala, 24 mediation centers have beencreated, employing mediators fluent inSpanish and local Mayan languages.27

In some situations, social movementsprovide the support for people to use thecourts and claim their formal rights, as withthe landless peasant movements in Braziland Mexico. In Argentina, too, unemployedfactory workers have occupied closed fac-tory sites and pursued their rights throughthe courts and the legislative process. Inother cases, civil society organizations haveassisted groups in claiming their rights—asin the “right to health” cases discussed inchapter 10.

Civil unrest and conflict may furtherweaken legal institutions. In Sierra Leone, adecade of civil war left the justice system inshambles: courthouses were destroyed, andjudges, lawyers, and police officers werekilled or forced to flee. During the conflictin Liberia, more than three-quarters of thepopulation left their homes.28 In Bosnia, 2.3million people—more than half the coun-try’s population—fled their homes duringand immediately after the war. To preventthe return of minorities, many propertyrecords were destroyed or tampered with.

Reestablishing legitimate legal institu-tions is crucial to restoring such people’spersonal and property rights and enhancingconfidence in newly established governancestructures. In postwar Bosnia, the DaytonAgreement established the Commission forReal Property Claims of Displaced Personsand Refugees, which collected claims for318,780 properties. As of June 2003, it hadissued about 290,000 final decisions onproperty titles.29


The coexistence of various official state lawsin South Africa began as early as the 1830swhen chiefs in Cape Colony were grantedauthority to enforce indigenous law (sub-ject to review by a colonial official).24 At theend of the apartheid era there were approx-imately 800 officially recognized traditionalcommunities and traditional leaders, 12,000headmen, and 12 kings.

Since 1994, South Africa has workedtoward bringing traditional systems into thestate framework.Traditional institutions andlaws are all officially recognized in the 1996constitution. After a long political process,the national Traditional Leadership and Gov-ernance Framework Act was promulgated in2004, setting out the roles and responsibili-ties of different levels of traditional leadersand institutions, and their relationships tothe different levels of government.

Many celebrated the constitutional andadministrative recognition of customarylaw, but there clearly are difficulties.Customary practices have been criticized asincompatible with rights in the constitutionand the new South African Bill of Rights.25

Of 800 traditional leaders recognized bythe state in South Africa, only one is female.In an attempt to deal with this, the stateissued a regulation in early 2005 thatfemale participation must be at 30 percentby the end of the year, but there is no con-sensus on how this might be achieved. Rec-ognizing the difficult task of effectivelyintegrating the different systems, the SouthAfrican model aims at “progressive align-ment” with the constitution.

Source: Adapted from Chirayath and others(2005).

B O X 8 . 3 State frameworks and customary institutions in South Africa

As in much of the rest of the world,Ecuador’s poor face numerous barriers inusing the legal system.Women consideringclaims against their former spouses mayface an added obstacle: physical violence.

As part of a larger judicial reform effort,three local NGOs—Centro Ecuatoriano para laPromoción y Acción de la Mujer,CorporaciónMujer a Mujer,and Fundación María Guare—provide legal information and representationas well as psychological counseling and refer-rals to shelters.A survey in 2002 revealed thatwomen’s use of legal aid clinics reduced theprobability of severe physical violence afterseparation by 17 percent.Legal aid clients also

attained better legal and economic resultsthan nonclients, raising their chances ofobtaining a child-support award by 20percent and their chances of receiving a child-support payment by 10 percent.

Receiving assistance from the legal aidclinics also had intergenerational impacts.Child-support payments increased the prob-ability of the child attending school (by 4.8percent) as did the lower incidence of vio-lence. Anecdotal evidence also suggests thatthe payments, a small but important sourceof family income, were used to pay for food.

Source: World Bank (2003g).

B O X 8 . 4 The impact of legal aid in Ecuador

Improving crime and personalsecurity policies—breaking crime-related inequality trapsThe cycle of inequality, crime, victimization,and discrimination exemplifies the processesby which inequality traps, outlined in chapter2, are perpetuated. Marginal groups are notonly more likely to move into criminalbehavior, they are also more likely to be vic-tims of crime. Furthermore, discriminatorypractices in the justice sector mean that mar-ginal groups are more likely to experience thelaw-and-order side of the legal system—blackmen in the United States are incarcerated ondrug charges at a rate 13.4 times that of whitemen, bearing little relation to differences inoffenses30—and are less likely to have accessto institutions charged with protecting them.At the same time, given their role in shapingthese processes, legal institutions are alsoplaces for change—that is, vehicles for chal-lenging inequality traps.

Breaking the cycle of inequality, crime, andviolence. Traditional approaches to reduc-ing crime and violence based on increasingmechanisms of control and harsher sanc-tions have failed to reduce crime. By con-trast, promoting protective strategies andminimizing the risk of crime have moresuccess at lower cost.31

To target the many risk factors of in-creased crime, crime prevention programsoften require support from the judicialservices, social services, health, education,media, police, local government, civil soci-ety organizations, and the private sector(box 8.5). Local governments and policeservices often coordinate such programs.32

Many more effective interventions tar-get children and adolescents, who are seenas particularly at risk of falling into crimi-nal behavior.33 Preventative interventionsinclude family support and parenting skillsprograms, early childhood developmentprograms, special needs programs, after-school care, antibullying programs, lifeskills and cultural programs, and commu-nity participation programs.34

School-based interventions have targetedyouth crime and have kept young people inschools in many countries.35 Tilsa Thuto is aschool crime prevention program imple-

mented in 42 schools in Soweto, South Africa,in 2000, in areas known for high levels ofcrime, unemployment, and poverty.36 Theprogram—established in partnership withthe department of education, communityorganizations, and local police—creates saferschools through the active participation ofstudents, teachers, parents, school adminis-trators, and the local community in differenttraining modules. Both teacher and studentattendance at school increased by some 70percent, acts of violence and aggression fell by67 percent, and the pass rate increased by anaverage of 78 percent.

Alternative ways of dealing with youngoffenders have also been effective in break-ing the cycle of crime. Interventions includediversion programs, restorative justice, alter-native sentencing, and reintegration proj-ects. Diversion programs aim to moveyoung offenders into welfare-based pro-grams, as with attempts in Africa to keepstreet children out of prison. Restorative jus-tice programs, such as community confer-encing, mediate between offenders and thoseaffected by the crime, helping to reintegrateyoung offenders into their communities.Noncustodial sentencing, such as commu-nity service orders, is used in different partsof Africa today to promote the reintegrationand rehabilitation of offenders.37

Increasing personal safety. Appropriate andaccessible police and support services for all


Unlike most Colombian cities, Bogota—a citypreviously considered unsafe and violent—has seen a huge reduction in crime since theearly 1990s and a substantial increase in citi-zen perception of safety.

The city’s administration targeted civicculture and education, urban planning andsafety, and the regeneration of public spaces.Individual and community behavior werechanged by establishing a citizen disarma-ment program; restricting alcohol consump-tion and use of fireworks; increasing thenumber of social service centers; enhancingawareness through media campaigns andeducational programs; preventing domesticviolence and child abuse; strengthening thecapacity of the police and the judiciary todeal with crime, violence, and victimization;

enhancing neighborhood watch programs;and revitalizing urban public spaces. Employ-ment and educational programs were intro-duced to support populations most at risk.

No longer one of the most dangerouscities in the western hemisphere, Bogota hasseen remarkable results.The homicide ratefell from 80 per 100,000 inhabitants in 1993to 22 in 2004, with homicides related tointoxication falling dramatically. By 2001,6,500 weapons had been turned in and gunconfiscations fell from 6,000 in 1995 to 1,600in 2003. Arrests for homicide, assault, and cartheft rose by 500 percent between 1994 and2003 (with no increase in police personnel).

Sources: Llorente and Rivas (2005), World Bank(2003a).

B O X 8 . 5 Bogota, Colombia: civic culture program

are crucial. Violence against women, a hugeproblem in many parts of the world, is exac-erbated by underreporting, inadequate sup-port systems for victims, discriminatorypractices within justice sector institutions,and the lack of adequate sanctions for perpe-trators. Many governments have attemptedto address the problem by introducing moresevere sentences for perpetrators, establish-ing remedial programs for offenders, andrunning gender-sensitive training programsfor police and the judiciary.

More recently, some countries, includ-ing Argentina, Brazil, Colombia, Peru, andUruguay, have set up women’s police stations.Other countries have set up police cells forwomen in regular police stations. These serv-ices have shown mixed results.38 Women’spolice stations have increased reporting ofabuse and the likelihood that women willreceive medical and social services. But criticsargue that services encourage regular policeto abdicate responsibility for crimes againstwomen and that women officers have notnecessarily demonstrated better attitudestoward victims of violence. Where stationsare working fairly well, their efforts are oftenundermined by other parts of the justice sys-tem, as prosecution rates remain unchanged.

The notion that the physical environmentcan increase personal safety has been an inte-gral part of many recent crime preventionstrategies,39 and it has been applied to city-planning, public transport systems, parks andrecreational spaces, low-income housing, anddowntown areas where people feel most vul-nerable to violence and crime.40

Toward greater equity in access to landLand is a key asset for poor people. Owningit provides a means of livelihood to many,facilitates access to credit markets, has aninsurance value, determines influence inlocal politics, permits participation in socialnetworks, and influences intrahouseholddynamics. That is why inequality in the own-ership of land has such far-reaching conse-quences for the distribution of well-beingand the organization of society for genera-tions to come. Yet landownership in manycountries is highly unequal, substantiallymore so than income or consumption.

Building on chapters 5 and 6, we argue herethat there are strong equity and efficiencyreasons for addressing inequalities in landdistribution—both rural and urban—andthen discuss the experience with land reformand options for broadening access to land:providing security of tenure, improving thefunctioning of land markets, and imple-menting cost-effective land redistribution.

Equity and efficiency reasons to address inequalities in land distributionInequalities in landownership in dozens ofcountries can be traced to interventions overthe past 500 years to establish and supportlarge farms at the expense of indigenous peo-ples and the local peasantry. This historicaldiscrimination against certain groups—ormore generally a lack of legitimacy for theprevailing pattern of landownership—offersa rationale for equity-enhancing reforms.Additional motivation comes from the factthat the landless are among the poorest indeveloping countries.41

Access to land can give the poor morevoice in the political arena and can lead tohigher investments in children’s education,arresting the intergenerational transmissionof poverty. Galasso and Ravallion (2005), intheir study of the Food for Education pro-gram in Bangladesh, find that villages withmore unequal distribution of land wereworse at targeting the poor. This is consis-tent with the view that land inequality isassociated with less power for the poor invillage decision making. Land inequality hasalso been found to impair the ability ofcommunities to engage in socially optimalcollective action, resulting in the underpro-visioning of public goods. It also contributesto social tensions that can lead to consider-able upheaval, as in Southern Africa.42

Inequalities in landownership can weighparticularly heavily on women. Land rights(and control of other assets) often residewith the head of household, which hasimplications for intrafamily bargainingpower and control of resources. Womenwith secure land rights (including inheri-tance on the death of a husband) are morelikely to engage in independent economicactivity, a result that has positive economic


and equity implications for the household.Inheritance rights that disadvantage womenare of particular concern in Africa, wherethey are often based on customary institu-tions that conflict with constitutionalnorms and international conventions onwomen’s rights. In fact, insecure inheritancerights pose an additional burden on widowswho lose their husbands, often throughHIV/AIDS.43

There are strong efficiency reasons toaddress inequalities in land distribution.Pervasive imperfections in land and finan-cial markets in developing countries reduceinvestment in land and keep countries fromefficient land allocations (chapter 5). Theseeffects—together with lower human capitalinvestment, reduced social cohesion, anddistorted political power—are consistentwith a positive association between moreunequal land distribution and lower GDPgrowth (figure 8.1).44

Experience with land reformThe discussion here implies that redistrib-uting land could enhance equity and effi-ciency. This is likely to be true, but there aresignificant hurdles in practice. For instance,the specter of land redistribution can alsoworsen efficiency, because farmers arereluctant to invest in land that they mightlose. Or political imperatives can overridesound program design. Successful landreforms—such as in Japan, the Republic ofKorea, and Taiwan, China—are rare andoften associated with exceptional events,such as war or political upheaval. Indeed,the history of land reforms is littered withpartial successes and failures.

In India, abolishing the land rights ofrent-collecting intermediaries45 has beenhighly successful, whereas the implementa-tion of landownership ceilings and laws toprotect tenants was, with few exceptions,half-hearted. The absence of political leader-ship has been identified as the “prime reasonfor the poor implementation of land reformsin India.”46 Still, where tenant protection wasimplemented seriously—primarily in WestBengal—it helped improve productivity.47

But restrictions on subleasing land by benefi-ciaries (or their children) tend to reduce thescope for productivity-enhancing land trans-

fers. And the fact that both tenants and land-lords have rights to the same plot of landseverely undermines investment incentives.48

In Latin America, where the potential forland reform should have been highest giventhe high inequality of landownership,reforms have generally been “incomplete.”Beneficiaries have often lacked the tools tobecome competitive and, as a result, theimpact on poverty has been disappointing.49

In Kenya and Zimbabwe, postindependencereforms were quite effective but short livedfor political reasons.50 In South Africa, thegovernment’s land redistribution programin the late 1990s fell way short of its targetsbut recently gathered steam (box 8.6).Clearly, both the politics of land reform andits implementation are complex.

Why the rather disappointing resultsfrom attempts at land reform? First, giventhat the motivation for land reform is oftento address political grievances, efficiencyand poverty reduction tend to be secondary.Guided by short-term political objectives,bureaucrats often targeted high-productiv-ity areas rather than high-potential areas,resulting in costly land acquisition and lim-ited scope for sustained productivityimpacts. Central administration of the pro-grams also often meant that a large portionof land reform budgets was spent on the


Initial land distribution (Gini coefficient, inverted scale)

Nicaragua

El Salvador

R.B. de Venezuela

PeruKenya South Africa

MexicoIndia

Indonesia

Japan

ChinaThailandVietnam

EgyptSri Lanka

Costa Rica

Malaysia

Dominican Rep.Brazil

ColombiaParaguay

ArgentinaGuatemala

Rep. of Korea

Taiwan, China

0.30.40.50.60.70.80.91.0

Average GDP per capita growth, 1960–2000, %8

6

4

2

0

–2

Honduras

Figure 8.1 Unequal initial land distributions go together with slower economic growth

Source: World Bank (2003i).Note: Land distribution is measured using the Gini coefficient.

wages and salaries of civil servants ratherthan used to benefit the poor. The prospectof gaining access to valuable real estate,instead of having to put in sweat equity todevelop a piece of land, undermined the self-targeting properties of reform programs,often politicizing beneficiary selection.

Second, the relationship between farmsize and productivity depends on the type ofcrop, quality of land, degree of mechaniza-tion, and such associated factors as market-ing and credit. For most crops, under normalavailability of mechanical services, produc-tion is neutral to farm size. But whenmanagement requirements are substantial(labor-intensive crops, erratic weather con-ditions, frequent pest incidence) familyfarms—as opposed to larger, wage-laborfarms—can be more efficient because ofadvantages in supervising labor. By contrast,large farms often have better access to inputand output markets, financing, and technical

assistance. Such advantages can be counteredif small farmers coordinate their effortsthrough cooperatives.51 If policymakers donot properly account for all these conditionsin land reform schemes, efficiency can suffer.

Third, many traditional land reformefforts failed to provide beneficiaries withsecure long-term rights backed by a well-functioning and equitable legal system.Affordable channels to adjudicate landaccess and ownership claims must be opento everyone. Without such channels, disad-vantaged groups cannot take full advantageof tenure security, land market, and distrib-utive reforms. Even with full property rights,underdeveloped credit and insurance mar-kets limit the use of land for collateral.

Fourth, the full productivity benefits ofland reform cannot be realized withoutcomplementary inputs and training; put-ting land in the hands of inexperiencedfarmers without the needed support oftenled to high rates of desertion.52 More gener-ally, a broader rural development strategy is required to complement land reformbecause rural households get their liveli-hoods from several different sources. Thishas implications for the design of landreform (for example, determining viablefarm size) and highlights the importance ofinvestments that can facilitate off-farmemployment, such as education.

Broadening access by improving the security of tenureThe benefits of secure tenure for ruralhouseholds are well known: higher produc-tivity, greater access to credit, higher propen-sity to invest in physical assets (figure 8.2)and the education of children, and time andeffort saved in securing land rights.53 Fur-ther benefits arise from removing the dis-cretionary power of bureaucrats to decideon the allocation of land (improved localgovernance was mentioned as a benefit ofproperty rights reforms introduced after1992 in Mexico).54

These benefits are observed also in urbancontexts. Capitalizing on a natural experi-ment that allocated land titles to somesquatters but not others in a poor suburbanarea of Buenos Aires, Galiani and Schar-grodsky (2004) found significant effects of


To redress apartheid-era asset inequality,South Africa embarked on land reform in1994 with a program that rested on redistri-bution, restitution, and tenure.Targets forredistribution were ambitious: the govern-ment aimed to transfer 24 million hectaresof agricultural land (30 percent of the total)to about 3 million people between 1994and 1999. Under the program, self-selectedgroups use grants to purchase land fromwilling sellers and to invest in the land’sdevelopment. But by February 2005, onlyabout 3.5 million hectares had been redis-tributed to 168,000 households. Restitutionproceeded at a snail’s pace—only 41 of79,000 claims were settled between 1995and 1999. And progress on tenure in the for-mer “homelands” was equally slow.

After this sluggish start, some keychanges accelerated the restitution andredistribution programs.The Restitution Actwas amended to allow for negotiated settle-ments, speeding the process considerably;previously all claims had to be settled incourt. By March 2005, more than 58,000claims were settled, and all claims are sched-uled to be resolved by March 2008.

The redistribution program wasimproved in 2001, making it more flexibleand decentralized.The grants for landpurchase and farm development now followa sliding scale, depending on the contribu-tion by the beneficiaries, and can be obtained

by individuals as well as groups. Approvalauthority is now delegated from the ministerto the provincial directors of land affairs. As aresult, redistribution has significantlyincreased. For the first time since 1994, landdelivery is now constrained only by thebudget for it.

But some big challenges remain. Pressureis growing from civil society on governmentto meet the revised target of redistributing 30percent of agricultural land by 2014. Restitu-tion settlements can be complex when ruralclaims target highly productive, capital-inten-sive farms and the claimants refuse, as is theirright, to accept financial compensation,instead of the physical restoration of theclaimed land.The agricultural impact of theredistribution scheme has been stunted byan inappropriate emphasis on collectivefarming and a lack of beneficiary power indecision making.The land market continuesto be biased against family farming throughcostly restrictions on subdividing agriculturalland and a regressive land tax dating from1939. A new land tax based on the value ofunimproved agricultural land could provideincentives for large farms to sell unused orunderused parcels. New legislation that trans-fers communal lands from state to commu-nity ownership is now in place, but it stillneeds to be implemented.

Sources: World Bank (2003i), van den Brink, deKlerk, and Binswanger (1996).

B O X 8 . 6 Land reform in South Africa: picking up steam

titling on housing investment, householdsize, and school achievement. The quality ofhouses in titled parcels was higher. Titledhouseholds had fewer members (eventhough their houses were larger), and theyseemed to invest more in their children’seducation. In India, unclear land titles com-bined with unreliable courts were found tolimit the supply of land and discourageinvestments. Southern states tend to havehigher tenure security, which increases theshare of modern retailers. Evidence from amassive urban squatter titling program inPeru suggests that titling resulted in morework done outside the home and substitu-tion of adult for child labor.55

Formal land titling is one way to pro-vide for secure tenure, but titling takestime and can be expensive. Thailand, thefirst country with a national program,completed the program this year, 20 yearsafter its inception. One solution is to allowalternatives to conventional private landtitles, especially in urban areas.56 InTrinidad and Tobago, a 1998 law author-ized three incremental levels of statutorysecurity, each requiring additional docu-mentation and commitment from the set-tler and the government. In one year, anestimated 80 percent of informal settlerson state land had applied for the lowestlevel.57 Because many of these instrumentsdo not require prior physical planning,infrastructure servicing, and surveying ofsettlements, they can offer widespreadcoverage at lower costs. The limitations ontransfer associated with many of theseinstruments also check the tendency ofsome informal dwellers to capitalize landsubsidies immediately through land sales.

Several countries have taken steps torequire joint titling of land in the names ofhusband and wife, bolstering women’seffective right to land, particularly duringtheir husbands’ absences. Vietnam has tar-gets for the joint titling of land as part of theVietnam Development Goals, incorporatedin its Poverty Reduction Strategy. Attentionto women’s land rights is particularlyimportant when women are the main culti-vators, when out-migration is high, whencontrol of productive activities is differenti-ated by gender, or when high levels of adult

mortality and unclear regulation couldundermine a woman’s livelihoods in case ofher husband’s death.58

Despite potentially large benefits fromtitling, there are challenges in urban andrural contexts. In urban areas, access tocredit may not increase if banks are unwill-ing to accept titled shanties in marginalareas as collateral. And where squatters’land is valuable, titling programs can besubverted by powerful interests who use theopportunity to relocate squatters to mar-ginal areas—for example, in Phnom Penh,Cambodia. This does not reflect a problemwith titling per se, but suggests that whenthe urban poor lack voice and governance isweak, titling programs can backfire.59 Oneway to protect squatters from predatoryurban developers would be to grant themgroup land rights as a first step toward indi-vidual titles. 60

Some studies indicate that formal landtitles in several African countries did notbring the expected benefits in higherincomes and investment. This may reflectweaknesses in the institutions responsiblefor registration and recordkeeping and forthe adjudication of rights and resolution ofconflicts. In some cases, it appears thatindigenous tenure was already sufficientlysecure.61 It may thus be more appropriateand more cost-effective to strengthen thesecurity of tenure through institutions thatcombine legality with social legitimacy. This


Brazil1996

Thailand1988

Honduras1996

Untitledland

Titledland

Relative increase in access to credit(untitled land = 100)

0

100

200

300

400

500

Relative increase in investment(untitled land = 100)

0

50

100

150

200

250

Thailand1988

Honduras1996

Untitledland

Titledland

Figure 8.2 Title to land increases investment and access to credit

Source: Feder (2002).

approach is particularly pertinent wherecustomary tenure practices predominate.Recognizing customary rights and institu-tions in the law helps protect large popula-tions that are governed by them and buildsa bridge to the formal system (box 8.7).Clearly, the complexities of engaging withcustomary justice systems outlined earlierin the chapter need to be considered—andwhen formalizing customary rights, careshould be taken not to simply codify exist-ing inequities, particularly for women.

Broadening access by improving the func-tioning of land markets. Land markets,both in sales and rentals, can in theory do

much to equalize land access. In practice,however, land sales/purchases are generallynot an avenue to expand access by the poorand may actually reduce it. High transac-tion costs, undeveloped credit markets, andhigh prices for land, reflecting its collateralvalue and any government subsidies tocrops, lead to thin sales markets that keeppoor farmers out altogether. This suggeststhat land sales markets will not contributeto greater equity in landholdings, especiallyin settings characterized by long-standingdiscrimination against specific groups,unless the government relieves the savingsconstraints of the poor through subsidies(as we will see in the next section).62

Distress sales of land by the poor canoccur in risky environments where smalllandowners do not have access to insurance.Thus, they are unable to smooth consump-tion through mechanisms other than landsales, such as safety nets.63 In a comparison ofland transactions in Indian and Bangladeshivillages during 1960–80, Cain (1981) foundthat poor farmers who had access to safetynet programs used the land market toaugment their landholdings and undertookproductivity-enhancing investment. Distresssales to obtain food and medicine predomi-nated when safety nets were absent.

While government interventions to im-pose restrictions on the transferability ofland can undermine investment incentivesand depress off-farm activity, they can playa role, especially during periods of transi-tion. In many Commonwealth of Indepen-dent States (CIS) countries the unrestrictedtransferability of land led to a concentra-tion of landholdings in the hands of a smallnumber of farm bosses, as poorer ruralhouseholds were enticed to sell their land inconditions of uncertainty and incompletemarkets and information.64 Initially limit-ing the transferability of land to lease trans-actions rather than outright sales wouldhave been better.

A gradual transition may be preferable incases in which an equitable distribution ofland has been the primary instrument ofsocial protection, as in China and Vietnam.Governments in both countries rightly per-ceive the tensions in active land sales mar-kets, which could lead to efficiency-enhancing


Customary land tenure is created,maintained, and protected by norm-basedcustomary systems at the local level. Rightsto land, whether communal or individual,are generally based on family lineage ormembership in a particular cultural group-ing. Exchanges through sales or rentals arelimited to members of the community. Cus-tomary systems are dominant in mostAfrican countries and in indigenous areas ofmany Latin American and Asian countries.Having evolved over long periods inresponse to local conditions, they are oftenquite flexible. Problems arise when transfersto outsiders become widespread or internalland dispute resolution mechanismsbecome inadequate.

There have been many efforts to makethe transition from collective or customarysystems to more individual landholdings. Butthe transition, if not properly managed, canend in disaster, such as in Kenya. Because theelimination of lineage rights and the legaliza-tion of land sales were not accepted by ruralpopulations, conflicts over land, sometimesviolent, ensued between those claiming landunder customary norms and those seekingto enforce the new rules.

Botswana has had more success. Since1970, the authorities have graduallystrengthened individual rights, starting withthe right to exclude other people’s animalsand to fence arable lands. Common law resi-dential leases for commercially valuableland have been introduced, as have lawsallowing the allocation of land to all adultcitizens, whether male or female, married orsingle (Adams 2000 and Toulmin and Quan2000 in World Bank 2003i). South Africa is

just starting down the road to communalland reform, having recently passed legisla-tion that will transfer state-owned land tocommunities and create democratic, trans-parent, and secure property rights regimesin these areas.

Mexico made a hybrid transition fromcollective land—known as the ejidosector—toward more individual landhold-ings. Ejidos are rural communities modeledafter a mixture of soviet-style collectivesand precolonial indigenous socialstructures. Reforms in 1992 strengthenedthe self-governance of ejidos, allowing themto choose a property rights regime. Eachejido could choose whether land would beheld under communal or individual owner-ship and issue property rights certificatesaccordingly. By 2001 the program hadissued property certificates to more than 3million households and given secure landrights to more than 1 million householdsthat previously had no formal recognitionof occupancy rights (World Bank 2001d).

Clarifying how customary rights—inrural and urban areas—relate to the formalsystem of property rights protectedthrough modern law is important for mil-lions of people. It is also important forexpanding the scope for outsideinvestment, particularly in urban areas.Experience with managing the intersectionbetween the two systems suggests theimportance of effective dispute resolutionmechanisms and a transparent and well-defined trajectory for transition, with theextension of secure landownership rights toan entire group as an effective and low-costfirst step (World Bank 2003i).

B O X 8 . 7 Clarifying how customary rights fit with formal systems

consolidation but also run the risk ofincreasing the number of landless poor. InChina, the government is considering mov-ing on several fronts, including developingsafety nets and rural finance and eliminat-ing residency-based restrictions (hukou) onlabor mobility.

While land sales markets have ambiguousimpacts on equity, the equity case for broad-ening access to land rentals appears moreclear-cut. When farmers do not have access tocredit that would enable them to purchaseland outright, rental markets are an impor-tant avenue for enhancing productivity andequity by facilitating low-cost transfers ofland to more productive producers (chapter5). Rental markets also enable landholderswith low agricultural skills (or no desire tofarm) to seek employment in the nonfarmsector while still earning a return on theirland.65 Evidence from Sudan suggests thatland rental markets do transfer land tosmaller producers. And in Colombia, rentalmarkets have been more effective than gov-ernment-sponsored land reforms in bringingland to productive and poor producers.66

If land rental markets have so muchpotential to improve equity and efficiency,why is there such a large variation in theincidence of rentals across countries? Onereason is a lack of tenure security—ortrust that the security will last. Without it,landlords are unwilling to rent out theirland for fear that they will not be able toreclaim it. Other reasons include currentor past government interventions torestrict tenancy, the availability of reliableconflict resolution mechanisms, and im-perfections in information.67 While con-cern about exploitation in sharecroppingarrangements may be justified,68 tenantprotection and rent ceilings can backfire.Such restrictions on land rentals often pushtransactions into informality or lock poorfarmers into less efficient and less equitablewage-labor arrangements. Estimates indi-cate that new tenancy legislation in Indiawas associated with the eviction of morethan 100 million tenants, causing the ruralpoor to lose access to about 30 percent ofthe total operated area.69 Interventions toenhance the bargaining power of thepoor—including better access to financial

markets, off-farm employment opportuni-ties, and equitable contract enforcementmechanisms—are likely to be preferable.

Well-functioning rental markets can be arung in the ownership ladder, but they proba-bly are not in circumstances of extremeinequality in landownership and power. Inthese instances, options for directly redistrib-utive policies need to be exploited.

Options for cost-effective landredistribution to broaden accessImproving tenure security and promotingland rental markets are good for bothequity and efficiency. Analysis and experi-ence indicate that land redistribution is notnearly as straightforward. It can be costly inprogram resources and reduced productiv-ity, and it can be an instrument of politicalpatronage. Substantial personnel and finan-cial resources are necessary to assess andpurchase (or expropriate) land, select bene-ficiaries, and supply training and credit.

When does land redistribution makesense? In some countries, redistributioncould be a necessary political step toaddress historical inequities and stave offviolence. In others, it could simply be a toolto shift underused land to more productiveuses while enhancing equity. In countrieswhere state landownership is high, landredistribution could involve limited budg-etary costs if bundled into a one-time trans-fer of state land to private ownership.Conversely, in countries with strong tradi-tions of tenure security, just the threat ofredistribution could undermine investment.

The feasibility of land redistribution alsodepends on the instrument. Expropriation islikely to be the most disruptive. Divestingstate lands and recuperating illegal settle-ments may be two cost-effective alternatives.For example, the mayor of Brasilia (Brazil)identified lands with uncertain titles andnegotiated the surrender of part of thoselands. In exchange, official titles were grantedto the remaining areas. Expropriating withcompensation and assisting land purchasesand rentals—for example, through commu-nity-driven land reform—are also feasiblealternatives. Subsidizing land purchases canbe costly, however, because land is often over-priced relative to its income generation


potential from productive use (reflecting itsspeculative, insurance, and status value).Also, subsidies are difficult to justify if thecurrent pattern of landownership is not con-sidered legitimate.70

Market- or community-driven land re-forms are a potential option. The reformstend to be decentralized and transparent,allowing community members to obtainresources for land access. They can be flexi-ble, allowing for land rental or purchaseaccording to the willing-buyer-willing-seller principle. Community-driven landreforms often give beneficiaries full propertyrights and involve coordination betweenlocal government and NGOs to provideaccess to training, technology, and credit. Acommunity-driven approach has been triedin several countries, including Brazil,Columbia, Guatemala, Honduras, India,Malawi, and South Africa. But the programsare relatively new, so rigorous impact evalu-ations are not yet available.

A land tax, possibly combined with anoutput tax, can be an important comple-ment. It can generate revenues to purchaseland to redistribute or encourage redistrib-ution by disproportionately taxing largelandholders or owners of unused or under-used land, both rural and urban (box 8.8).

Regardless of the instrument, commonlessons can be drawn from previous attemptsat land redistribution.

• Complementary investments—trainingand credit. Evidence from Latin America71

and Africa indicates that simply redistrib-

uting land does little. Beneficiaries mustbe provided with a package of assistanceto ensure self-sufficiency and maximizeproductivity. The right package will varyby country but could include training andcredit. Technical assistance, such as helpelaborating farm plans and crop budgetsor instruction on new technologies, canbe a success factor for subsistence farmersand those who lack commercial farmingexpertise. Credit can allow beneficiariesto make productivity-enhancing invest-ments, for example in irrigation, fencing,tools, or draft animals.

• Beneficiary selection and targeting. Whenbeneficiary selection is politicized, redis-tributed land will not always go to themost needy and capable farmers. Thesolution is to have transparent rules thatallow communities to understand exactlyhow and why each beneficiary was cho-sen. Self-targeting—where potential ben-eficiaries seek out land for sale at low costand then apply for grants and/or loans—can also be effective.

• Tenure security. Those who gain access toredistributed land should be given clearownership rights. In some instances, itmay be enough to have less than fullownership (such as a certificate of con-trol or long-term lease) to reduce un-certainty, encourage investment, andpromote the benefits discussed above.

Providing infrastructureequitablyInfrastructure in most developing countriesis characterized by low and unequal provi-sion—about 2 people of every 10 in thedeveloping world were without access tosafe water in 2000, 5 of 10 lived withoutadequate sanitation, and 9 of 10 lived with-out their wastewater being treated72—withmany families suffering from inadequateaccess especially in rural areas.

Economic opportunities are stronglyshaped by access to infrastructure. Muchinfrastructure is traditionally governmentprovided and so is driven by the politicalprocess. Financing constraints and technicaldesign challenges are very real but perhapseasier to overcome. When more equal voice


Taxes on land can be an effective, nondistor-tionary tool for collecting local revenue andfacilitating land redistribution.They can alsoencourage productive land usage by taxingunderused land at higher rates.This can beattractive when large unproductive plots ofland (often held for speculative reasons) cre-ate artificially high land prices and limit landaccess for poor farmers. But administering atax on land requires data on the size, value,ownership status, productive capacity, andoutput of each plot. Because it is hard for thegovernment to measure the actual degreeof land use, especially for large landholders,there is a strong incentive for tax evasion.

One way to limit evasion by large land-holders is to use a mix of land taxes and avalue-added tax (VAT). A VAT can have muchlower rates of evasion and facilitate accu-rate reporting of cultivation levels. Knowl-edge of true cultivation levels would thenlimit the scope for overreporting the degreeof land use to evade higher tax rates forunderused land. In the absence of insurancemarkets, a mix of VAT and land taxes canalso reduce the risk facing smallholdersbecause tax burdens would be correlatedwith output fluctuations.

Sources: Assunção and Moreira (2001), WorldBank (2003i).

B O X 8 . 8 Land and output tax combinations

or the political interests of the governingregime make for more broadly accountablepolicies, infrastructure is provided in waysthat are supportive of the economic interestsof poorer groups—for example, in EastAsia’s intensification of irrigation and trans-port (see focus 4 on Indonesia). In moreunequal societies, those without influencereceive less and lower-quality access to pub-lic services—this often means the poor. Thisinequitable access also applies to remoteregions and excluded groups, and some-times it has a gender dimension. Even worse,some infrastructure services (utilities espe-cially) often become instruments of patron-age and riddled with problems of inefficientprovision and corrupt practices.

As in the case for land, more equal accessto infrastructure would be good for equityand will often be good for growth. Thisrequires addressing difficult financing issues,constraints that limit the poor’s ability toaccess infrastructure, and major accountabil-ity issues through institutional designs thatsupport more equitable response to needs.

More equitable access to infrastructure is good for growth and equityThere is solid evidence that infrastructureinvestments broaden opportunities for peo-ple and communities by integrating theminto regional and national systems of pro-duction and commerce, and by improvingtheir access to public services. Locationstrongly influences household market partic-ipation in Vietnam. And households with thesame characteristics and endowments yielddifferent returns in different geographic set-tings in parts of rural China.73 Leipziger andothers (2003), based on a sample of 73 coun-tries, find that a 10 percent improvement in acountry’s infrastructure index is associatedwith a 5 percent reduction in child mortality,a 3.5 percent reduction in infant mortality,and a 7.8 percent reduction in maternalmortality, controlling for incomes and theavailability of health services. Microevidencefrom rural India lends support to thesecross-country findings: the prevalence andduration of diarrhea among children underfive are significantly lower for families withpiped water.74

Investments in basic water and energyinfrastructure can improve gender equity.Around the world, the burden of gatheringand transporting fuelwood and water tra-ditionally falls on women and girls. InGhana, Tanzania, and Zambia womenaccount for two-thirds of all householdtime devoted to water and fuel collection,while children—mostly girls—account forbetween 5 and 28 percent of time spent onthese activities. In rural Morocco, havingwells or piped water increases the probabil-ity that both girls and boys will enroll inschool, with larger impacts for girls, whoare responsible for collecting water. Studiesin Pakistan show that poor access to fire-wood and water in rural areas means thatwomen work longer hours and have lesstime for income-generating activities, withimpacts on the intrahousehold balance ofpower. Women and children are also moresubject to health risks from indoor air pol-lution, given the disproportionate amountof time spent inside the home. Electricityin the home can reduce the need to burnpolluting fuels for light and cooking. Andimprovements in electricity and gas distri-bution can eliminate time spent collectingtraditional fuels.75

Improving rural transport infrastruc-ture can reduce transactions costs, expandaccess to markets, and improve ruralincomes. It is estimated that nearly two-thirds of African farmers are effectivelyinsulated from national and world marketsbecause of poor market access.76 In con-trast, substantial investments in Indone-sian roads over the last three decades haveallowed poor households to successfullyenter the market economy.77 And many ofthe roads were built as labor intensive pub-lic works, making jobs available tounskilled labor. Similarly, investments inrural infrastructure (roads, bridges, cul-verts, and marketplaces) in Bangladeshhave deepened the vibrancy of the ruraleconomy for both agriculture and non-agriculture. Investing in rural roads is anexample of how expanding access to infra-structure can benefit equity and efficiencyin the long run, especially in areas withlarge numbers of poor people and agro-climatic potential.78


Is privatization the answer?What accounts for the failure of infrastruc-ture services to serve poor people, especiallyin Africa (box 8.9)? Beyond the importantrole played by historical and geographic con-texts, there are major financing constraintsand governance challenges. In most develop-ing countries, the public sector is fiscallystrapped—public investment in infrastruc-ture in Latin America dropped from 3 per-cent of GDP in 1980 to less than 1 percent in200179—and public spending requires taxes,which can exert a drag on efficiency and canmean forgone investments in other areas.Local financial markets are generally not suf-ficiently developed to intermediate privatesavings into long-term, risky infrastructureinvestments (and, in any case, private savingsare often not large). Foreign private capital isinterested generally in large markets and eventhen only when risks (including policy andexchange rate risks) are acceptable. State-owned infrastructure companies in manycountries—especially when political inequal-

ities are large—are often inefficient andbecome instruments of patronage.

The 1990s were characterized by a massivepolicy redirection toward private participa-tion in infrastructure—reflecting the disap-pointment with ineffective state-operatedutilities, the promise of private funding, andthe greater flexibility offered by technologicalchange and regulatory innovation.80 But theprivatization wave bypassed many develop-ing countries—Sub-Saharan Africa receivedonly 3 percent of total private infrastructureinvestment in developing countries between1995 and 200081—and even where privatecapital became the dominant source ofinvestment (as in Latin America in the mid-1990s when private investment was 2.5 per-cent of GDP), the results for equity weremixed. There were many cases of privatiza-tion in which access for the poor improved,especially when competition reduced politi-cal capture, but evidence suggests that privateoperators also focused on wealthier segmentsof the population (box 8.10).


Infrastructure development in Africa is abysmal,lagging behind the rest of the world in terms ofquantity, quality, cost, and equality of access.Only 16 percent of roads are paved, the averagewaiting time for a telephone connection is threeand half years, transport costs are the highest ofany region, and fewer than one in five Africanshas access to electricity.What can explain thistremendous underdevelopment?

• Difficult geography and complex history. Dis-tance from major markets, the Sahara Desert,a shortage of natural ports, and vast tracts oflandlocked areas all increase transport costs.Infrastructure development during colonialtimes focused on building transportationfrom resource sites to a port. And thepostcolonial division of Africa into manysmall states drives up transport and energycosts, with cumbersome border crossings, lit-tle regional cooperation on water and powerprojects, and incompatible rail systems,among other factors.

• Financing constraints. Lack of investment hasled to a deterioration in infrastructure, espe-cially road transport. For instance, in nine EastAfrican countries, maintenance spendingcould cover only 20 percent of currentnetworks. Lack of disposable income meanslow demand for infrastructure, and small and

scattered populations make economies ofscale difficult and require higher investments.Widespread subsidies that go to the relativelywealthy (the poor generally do not have con-nections to networked utilities) underminegovernments’ ability to expand access.

• Bad policies and poor accountability. Licens-ing, competition barriers, and corruption alsoimpede affordable infrastructure provision.

Despite the difficult geographic and struc-tural factors, there are opportunities to improveAfrica’s infrastructure. One necessary step is toboost investment, especially in rehabilitationand maintenance. Higher public investment willbe needed. Private participation can helpfinance some investments and increaseefficiency, but it is unlikely to solve Africa’s infra-structure problem. A sound institutional andpolicy environment is required to attract freshinvestment and use it effectively.

Macroeconomic stability, freedom to repatri-ate capital, competitive taxes, contract enforce-ment, low corruption, and adherence to trans-parent rules are all important to privateinvestors, especially given the long paybackperiods of many infrastructure investments. Incontexts in which policies are sensible but for-eign private investment is hard to attract,foreign aid can provide both financing for pub-

lic provision and guarantees to help foster pri-vate participation. Prudent oversight, regulation,and contract design can also ensure that publicand private financing are equitably andefficiently used.

Increasing cross-border and regional coop-eration is one way to make Africaninfrastructure more affordable. Streamlined bor-der crossings and improved road and rail linkswould reduce transport costs.Trade in powerand water resources could significantly cut costsas well. For example, it is estimated that SouthAfrica could save $80 million a year in operatingcosts by exchanging electricity with its neigh-bors (Masters, Sparrow, and Bowen 1999). Seek-ing innovative ways to broaden access is alsoneeded. Mozambique has tested a promisingapproach: the government set up utility compa-nies using diesel generators in rural areas andthen sold them to private investors below costfor continuing commercial operation. Grantsfrom government, NGOs, or donors for commu-nity-driven infrastructure projects are anotherpossibility. Regardless of the approach, commu-nity and user involvement in infrastructure con-struction, maintenance, and management is oneof the most effective ways to expand access inrural areas.

Source: World Bank (2000a).

B O X 8 . 9 Lagging infrastructure in Africa

In the end, experience suggests that pri-vatization alone is not the answer. Whetherinfrastructure services are provided by pri-vate operators or public utilities seems lessimportant for equity and efficiency thanspecific measures to improve access for thepoor, the structure of incentives facingproviders, and provider accountability tothe general public.

Expanding access and makingservices affordableWhether expanding general access benefitsthe poor depends on initial levels of cover-age. In many African countries, overall accessrates improved over the last decade, but thebottom 40 percent of the population regis-tered no gains at all (figure 8.3). This is not

surprising. Given the low initial coverage inmany of these countries, the expansionfavored wealthier households. This does notmean, however, that expanding access whenoverall levels of service provision are low isbad for equity. On the contrary—better toexpand access in this case than to focus onupgrading quality, which would benefit onlythe few who already have access.

To expand access to the poor, policymak-ers can set service obligations or createincentives for providers. One way is to spec-ify universal service obligations, which iscommon in the telecommunications sec-tor.83 While this is a worthy social objective,it may not be practical in the short run whenstarting from low access rates. That is whyservice obligations should include details on


Private participation in infrastructure increaseddramatically in Latin America during the late1990s. It went from $21 billion in 1995 to a peak of$80 billion in 1998, dropping back to $20 billionby 2002 (World Bank 2004f).The distributionalimpact of private investment depends on howefficiency gains are allocated between public andprivate interests. In the best cases, privatizationcan solve patronage problems and lead to greaterefficiency and equity. In worse cases, efficiencygains can be shared between government andprivate operators, or go primarily to the privatesector and lead to the consolidation of privatemonopoly power (as in Mexican telecommunica-tions).The outcome depends on the market andaccountability structure, including the effective-ness of regulation. Evidence from Latin Americashows that privatizations fall into each of thesecategories, yielding mixed results for affordableaccess by the poor.

Research from Argentina, Bolivia, Mexico,and Nicaragua shows that utility privatizationincreased access and enhanced service qualityfor poor consumers in some cases (McKenzieand Mookherjee 2003). In Chile, access to powerservices increased greatly for low-incomegroups during the first 10 years of private opera-tion (Estache, Gómez-Lobo, and Leipziger 2001).In Colombia, private utilities have connectedmore of the poorest consumers than their pub-lic counterparts (World Bank (2001a)). Otherresearch shows that child mortality caused bywaterborne diseases fell by 5 to 9 percent in the30 Argentine locales where water services wereprivatized, with the strongest benefit—a morethan 25 percent decline in mortality—occurringin the poorest neighborhoods (Galiani, Gertler,and Schargrodsky 2002).

Despite the increases in access, there aretwo reasons why privatization may have hadadverse effects on the poor through highertariffs and connection costs. First, privatizationcan reduce the scope for cross-regional subsi-dies. One study (Campos and others 2003)showed that the fiscal cost of utilitiesincreased a few years after privatization. Thiswas the result of “cream-skimming,” asobserved in Argentina. In some of theprovinces of Argentina, the water concessionswere for the large cities only, leaving theresponsibility for the small cities and ruralareas to the governments. Because the bigcities were cross-subsidizing the other regionsunder public provision, privatization reducedthis source of funding and increased the netfiscal costs once the transaction payoffs fromprivatization had disappeared.

Second, connection costs and tariffs wereadjusted to cost-recovery levels following priva-tization, leading in many instances to higherprices. In the early 1990s, public utilities indeveloping countries subsidized an average of20 percent of gas and 70 percent of water costs(World Bank 1994). So when subsidies were cut,services often became too expensive for poorconsumers. For the water concession in BuenosAires, the initial connection charge was set sohigh that many users could not afford it(Estache, Foster, and Wodon 2001), which wasan issue at the center of one of the first majoradjustments to the contract (Ugaz and Price2003). The telecommunications sector inArgentina also saw price increases followingprivatization, largely to rebalance local and longdistance charges. But price increases are not thenorm—competition can drive them down. In

Chile, the liberalization of the telecommunica-tions market in 1994 reduced call prices bymore than 50 percent. In Argentina, thanks tothe entry of 21 new operators in the generationsector, residential electricity customers enjoyeda 40 percent drop in tariffs in the five years afterprivatization (1992–97) (World Bank 2002b).Ultimately, price changes depend on initial con-ditions, quality improvements, and regulatoryand institutional frameworks that determineprofits.

Episodes of privatization can be opportuni-ties to strengthen accountability.They can gen-erate public discussion about the current stateof service delivery, the options for reform, theterms of the contract, and the tradeoffs underconsideration. Such occasions can be powerfulin overcoming collective action problems andmobilizing consumers to express their interests.But there is also the danger that, in the absenceof the voice of consumers, the process of privati-zation may be captured by narrow interests withpolitical connections and better information.Indeed, that might be one reason why the pub-lic perception of privatization in Latin America isso negative.82 There is evidence that privatiza-tion has been associated with increased powerfor conglomerates and their foreign partners—and with higher profits in noncompetitive sec-tors. Accusations of corruption during the priva-tization process, concentrated gains by a fewactors (whether made legally or illegally) con-trasted with worker layoffs, and unrealistic cus-tomer expectations regarding service levels(driven by politicians overselling the promise ofprivatization) are also probably to blame for pri-vatization’s negative public image (De Ferrantiand others 2004).

B O X 8 . 1 0 The distributional impact of infrastructure privatization in Latin America: a mixed bag

time frames and ways to finance the obliga-tion when customers are unable to pay.Defining connection targets is another wayto promote access. Targets are easy to moni-tor and can be enforced by financial penal-ties. Of course, connection targets can bemet only if customers are able and willingto take up the service—and this depends ontheir ability to overcome impediments,including title requirements and incomeand liquidity constraints (given the lumpi-ness of connection charges).

In many countries, new connections aresubsidized to meet access objectives andkeep providers solvent. New connectionscan be subsidized from charges to existingusers, particularly if the group of existingusers is much larger and wealthier thanpotential new users. The water and sanita-tion concessionaire in Buenos Aires adoptedthis type of cross-subsidy after renegotiat-ing an initial contract that charged onerousconnection fees to the poor. Governmentfinancing for connection subsidies is also anoption, as is offering credit to consumersfor connection purchases. In Colombia, thelaw requires that connection charges forpoor customers be spread over at least threeyears.84

A complement to connection subsidiesare consumption subsidies, either throughmeans-tested transfers financed out of gen-eral tax revenues or through lifeline tariffs.Subsidized lifeline tariffs require a transferfrom those with high levels of consumptionto those with low levels.85 When consideringlifeline subsidies, care must be taken to set athreshold that is high enough to garnerpolitical support, yet low enough that thepoor are the primary beneficiaries. Forinstance, evidence from Honduras suggeststhat their electricity subsidy is too high—83.5 percent of residential customers benefitfrom the subsidy (those consuming under300 kWh monthly). Means-tested vouchersfor purchasing services are another subsidyoption. They are similar to means-tested tar-iff subsidies with added flexibility for theuser to select a service provider.86

Given the liquidity constraints of thepoor and the possible seasonality of use andincome, introducing flexibility in paymentis likely to help expand access. Increasingthe frequency of billing is one such option.Prepayment devices, which facilitate budg-eting for low-income households, areanother. On the downside, prepaymentcould lead to frequent “self-disconnection.”


1st period

Access to electricity for 1st and 2nd quintile (average)

CameroonGhana

Nigeria

Nigeria

Senegal

ZambiaBenin

BurkinaFaso

Tanzania

Malawi

Rwanda

Niger

Namibia

Mali

Madagascar

Madagascar

Zimbabwe

Côte d’Ivoire

100%80%60%40%20%

2nd period100%

75%

50%

25%

0%0%

1st period

Access to electricity for 4th and 5th quintile (average)

100%

45° 45°

80%60%40%20%

2nd period100%

75%

50%

25%

0%0%

Uganda

Kenya

Figure 8.3 Poor families did not benefit from an expansion of access in Africa

Source: Diallo and Wodon (2005). Note: First and last observation dates vary for each country. The first observation is in the early to mid-1990s and the last in the late 1990s or early 2000s. The average time between observa-tions is seven years.

Utilities could also allow customers tochoose from a menu of tariffs with differentcombinations of fixed (standing) and vari-able charges. Allowing lower fixed chargesin exchange for slightly higher variable onescould benefit smaller consumers.87

Enabling consumers to make certainquality-price tradeoffs by encouraginglower-quality services is also likely to bebeneficial. This can be done by allowinglower standards of formal provisioning incertain poor areas or by encouraging avibrant network of informal providers thatcan either operate independently or throughsubcontracting arrangements with the for-mal provider.

Water service provides an example. Astudy using data from 47 countries showsthat informal providers such as point-source vendors (kiosks) and mobile distrib-utors (such as tanker trucks and carters)systematically charge more than networkedproviders, both public and private.88 Micro-economic evidence supports this. In Niger,for example, wealthier households are morelikely to use networked rather than infor-mal providers and pay less per unit con-sumed (figure 8.4). But informal providerscan offer a valuable service, because manypoor users cannot afford connectioncharges or monthly lump-sum bills or livein areas inaccessible to utilities for legal ortechnical reasons. Recognizing that privateconnections for all households may not be afeasible goal in the near term, governmentsand utilities can work with informalproviders. For instance, kiosk services couldbe improved by subsidizing kiosk connec-tions, increasing competition among kiosks,and introducing performance measures andquality standards.89 In Senegal, one studyargues for directing subsidies away fromprivate connections toward water providedat standposts or by other informalproviders (box 8.11).

High tariffs for the consumer reflect thepricing decisions of the utilities concernedand the taxation decisions of local and cen-tral governments. Avoiding exclusivity incontracts and liberalizing entry, includingthrough the participation of independentprivate providers or communities in non-network services, can help reduce costs.

Moreover, many governments view utilitiesand telecommunications as cash cows,imposing indirect consumption taxes thattend to be regressive where connectivity tothe formal system is high. In Argentina,utilities generate about 1 percent of tax rev-enue; in addition to the income tax, there is


In 1995, the government of Senegallaunched sweeping reforms in the urbanwater sector.The bankrupt public sectorutility was dissolved. A new state asset-holding company was created to managethe sector. And a private operator was con-tracted to run the system based on an inter-national competitive bidding process.Thereform had positive outcomes for the poor,thanks to strong government commitment,connection subsidies in low-income neigh-borhoods, and well-designed operatorincentives. But tariff inequities and poor tar-geting of subsidies remain.

Subsidies targeted at the poor take threeforms:consumption subsidies,connectionsubsidies,and construction of standposts inareas lacking private connections. In Senegal,consumption subsidies are delivered throughan increasing block tariff with a low “social tar-iff”for household consumption under 10 m3

per month.The problem with consumptionsubsidies is that many poor families do not

have a connection at all or share a connectionwith several other families,which bumpsthem into a higher consumption block.

Connection subsidies have benefitedthe poor, but those who could benefit mostare likely to be ineligible because they lacktitle to their land and an established house.The construction of public standposts hasexpanded access but does not necessarilyprovide the lowest-cost water.Tariffs onwater sold to licensed standpost vendorsare considerably higher than the subsidizedsocial tariff and vendors also charge anoverhead fee. Analysis of one NGO-runstandpost showed that users were paying350 percent more than the social tariff rate.Given these shortcomings, it would makemore sense to direct consumption subsidiesaway from private connections and intowater provided at standposts or by otherinformal providers who serve the poorest.

Source: Brocklehurst and Janssens (2004).

B O X 8 . 1 1 The pro-poor agenda for urban water in Senegal

Average price CFA francper cubic meter

600

400

200

003%

77%

20%

30%

62%

9%

43%50%

7%

61%

33%

7%

89%

9%3%

25

50

Percent of quintileusing each source

75

100

3 4 521Wealth quintile

Piped waterWater sources

Neighbors, wells, or rivers

Averageprice

Fountains or vendors

Figure 8.4 Poorer households have lower-quality water and pay more in Niger

Source: Bardasi and Wodon (2004).

a 21 percent VAT plus municipal and provin-cial taxes. So indirect taxes can be as high as55 percent in some municipalities. Reducingsuch taxes would reduce tariffs.90

Naturally, benefits can be lower and costshigher for investments in poor, remoteareas that are short on economic potential.This analysis is not blind to those consider-ations, but also it recognizes the argumentthat there can be long-run benefits fromgreater inclusion of groups that are margin-alized because of location or poverty (seefocus 6 on regional inequality).

Strengthening governance, voice,and accountabilityInfrastructure provision suffers from severeproblems of corruption and lack of account-ability. Because infrastructure investmentsare typically large and lumpy, and oftenexhibit increasing returns to scale or net-work externalities, they are less amenable tocompetition in financing and provision andgenerate significant scope for corruptionand patronage. Politicians bribe public util-ity officials and pursue political goals bytransferring resources to politically influen-tial groups, rather than encouraging utili-ties to expand service and cut costs.

One possible solution is to subject utili-ties to performance pressures and reducepoliticians’ willingness or ability to use themfor political purposes.91 The prudent use of

regulation is another way to strengthenaccountability and reduce political captureand corruption. Regulators have a key rolein ensuring that the public interest is beingserved. This includes safeguarding the valueof public assets, upholding the sectoralnorms relating to health and safety, provid-ing information about the performance ofthe service provider, and enforcing compli-ance with contractual obligations.

If effective, regulation will have animportant impact on efficiency and equity.Evidence comes from a study of the energy,telecommunications, and water sectors inArgentina that separated the benefits ofprivatization from those of effective reg-ulation. The study found that effectiveregulation yielded operational gains morethan one-third higher than the gains fromprivatization alone and equivalent to 0.35percent of GDP, or 16 percent of the averageexpenditure on utility services. Gains forthe lowest quintiles were proportionatelyhigher. Another study of 1,000 concessionsin Latin America found that even a moder-ately well-functioning regulator can temperopportunistic renegotiation of contracts. Itconcluded that the probability of renegotia-tion goes from 17 percent in a setting inwhich a regulatory body exists to 60 percentin settings it does not.92

Positive impacts of regulation are condi-tioned on the ability to insulate regulatorssomewhat from pressures coming frompoliticians and providers. Measures tostrengthen the independence of the regula-tor are paramount, and this may require aseparate agency with reliable and ring-fenced funding and staffing. When assetsare decentralized, the regulatory agency canbe at the level of the central or regional gov-ernment. In short, regulators should besubject to substantive and proceduralrequirements that ensure integrity, inde-pendence, transparency, and accountability(box 8.12).

In situations in which local provision ismore susceptible to capture by local elites,the central government can influence localoutcomes by using fiscal incentives to nudgelocal governments toward broader accessand by reducing costs for the poor throughcontingent intergovernmental transfers. This


Brazil’s National Telecommunications Regula-tory Agency has a Web site that providesinformation on service price comparisons,laws, and operator compliance. Its AdvisoryCouncil (with civil society representatives)assesses the agency’s annual reports andpublishes the findings in the official gazetteand on the Web. And an ombudsperson eval-uates the agency’s performance every twoyears. In 2000, it became the world’s firsttelecommunications regulator to receive ISO-9001 certification, an international standardfor meeting customers’technical needs.

Peru has made similar progress inimproving regulation by increasingtransparency.The Supervisory Authority forPrivate Investment in Telecommunications

sets prices, ensures a competitive market, andmonitors compliance with concession con-tracts and quality standards.The agency setsnorms through a transparent process. Regula-tory proposals must be supported by assess-ments of welfare benefits and best practices,published in the official gazette, and undergoa 30-day consultation period. Some proposalsalso are subject to public hearings. Inaddition, the agency has multiple dispute res-olution mechanisms. Independent commit-tees, supported by experts, resolve disputesamong service providers and an internal tri-bunal handles consumer complaints not sat-isfactorily managed by phone companies.

Source: World Bank (2004l).

B O X 8 . 1 2 Addressing accountability and transparency intelecommunications in Brazil and Peru

requires setting performance targets, elicit-ing competition among local governments,and benchmarking to monitor performance.It also requires sufficient policy autonomy atthe local level to meet the specified targets.

Monitoring the performance of providersrequires reliable information and perform-ance benchmarks. This is more easily donewhen there is a management contract or aconcession agreement with clear service obli-gations or when there are performance con-tracts with similar features for public utilities.Community involvement can help monitorcompliance. A national regulator can bench-mark local government performance wheninfrastructure assets and policy decisions aredecentralized and provide useful informationfor intergovernmental fiscal transfers.

SummaryJustice systems can do much to level theplaying field in the political, economic, andsociocultural domains, especially whensocieties press for equitable laws and fortransparency and accountability in theirimplementation. Legal institutions canuphold the political rights of citizens andcurb the capture of the state by the elite.They can equalize economic opportunitiesby protecting property rights for all andensuring nondiscrimination in the market.They can force change in the social domainby challenging inequitable practices. But

justice systems and legal institutions,embedded as they are in the political andsocioeconomic structure of societies, can behijacked by special interests.

Broadening access to land can enhancepeople’s opportunities to engage in produc-tive activities. The distribution of land rights,especially ownership rights, is skewed inmany countries, and challenging this patternis difficult. Past land reform efforts show theneed for a broader menu of policies that gobeyond redistribution through expropria-tion and include improvements to tenuresecurity and the functioning of rental mar-kets. There is also scope for redistributionthrough channels other than expropriation.

More equitable access to infrastructurealso has equity and efficiency benefits.Broadening access to infrastructure bringspeople closer to markets and services and tothe power and water they need for produc-tive activities and daily existence, expandingtheir economic opportunities. Expandingaffordable access for poor people and poorareas requires tackling difficult financingissues, designing subsidies effectively, work-ing with informal providers, as well asmaking providers more accountable andstrengthening the voice of beneficiaries. Thechallenges of expanding access to justice,land, and infrastructure and rooting outcorruption and elite capture remain signifi-cant for many developing countries.


taxation f o c u s 5 o n

Raising revenues for equitablepoliciesThe publicly-provided, equity-enhancing pro-grams outlined in this report, from earlychildhood development interventions to watersupply, have to be financed. The resources tofund them must generally be raised throughtaxation. In fact, a key ingredient for a well-functioning public sector is a societal under-standing that the quality of public servicesdepends on everyone paying their share of thetax burden. Where this perception fails, thesocial compact breaks down, and tax avoid-ance and evasion become widespread. Thisleads to a vicious circle of free-riding and taxrate increases, with adverse consequences forthe public finances, the quality of service deliv-ery, and social cohesion.

It follows that the same institutionsthat influence the quality and breadth ofservice delivery also affect the overall taxeffort. Revenues, like spending, increasewith a country’s level of income, but thequality of institutions—notably voice andaccountability—also matters, even whencontrolling for income1 (see figure S5.1).Voice and accountability can strengthen thetax effort, as the services provided become areflection of the desires of the broader elec-torate rather than of a privileged few. Lin-dert (2004) argues that the expansion ofvoice in twentieth-century Europe was thedriving force behind fashioning the socialcompact that delivered high and equitablegrowth alongside extensive public provi-

sion of services during 1950–80. In Chileand the Republic of Korea, too, the emer-gence of representative institutions (andhigher incomes and administrative capac-ity) led to higher taxation and spending.

For similar reasons, high inequality inthe distribution of political power andwealth may be prejudicial to the tax effort.The low tax revenues in most of CentralAmerica may reflect the low solidarity ofthe elite with middle- and lower-incomegroups: the small, wealthiest segment of thepopulation is unwilling to pay more in taxesto provide public services, because the elitecan procure many private substitutes forpublicly-provided services, ranging frompublic safety to education and roads.2

Resource rents can relieve the fiscal con-straint on spending and in principle pro-vide resources for equitable provision,though they raise a host of governancechallenges. The ability of resource-richstates to rely on “unearned” revenues canundermine the accountability embedded inthe social compact underpinning soundpublic finance.3 If not properly addressed,the result may be wasted natural resources,corrupt state institutions, and poorerprospects for long-run growth and equity.A few recent efforts to harness resourcerents for broad-based development incountries with poor institutions aim tointroduce greater transparency andaccountability (see box on next page).

Foreign aid can also weaken the social

compact, in much the same way as resourcerevenues, by making governments lessbeholden to civic interests.4 Some evidencesuggests that higher levels of aid are associ-ated with lower revenue collection, espe-cially among poorly governed countries.5

Donors should thus consider ways of sup-porting accountable institutions in recipientcountries—both for spending and taxation(see chapter 10 for discussion of aid).

While institutional transparency andaccountability and linking good publicservices to the taxes that fund them areprobably the first-order determinants ofsuccessful revenue-raising, technical aspectsof public finance are important to reduceefficiency costs. Lindert (2004) argues thatamong industrial countries, high socialbenefit/high tax societies—most notablyScandinavia—have paid particular atten-tion to tax design to reduce adverse incen-tive effects for labor effort and capitalinvestment to sustain growth. Tax design isan immense area. From the perspective ofequity, the primary contribution of taxa-tion is in providing the resources to fundequity-enhancing spending. For this, theprincipal criteria are minimizing efficiencycosts, administrative feasibility, and politi-cal supportability. For specific tax instru-ments, there can also be potential forpositive direct effects on equity. Here we sug-gest seven basic principles for mobilizing taxrevenues in ways that minimize efficiencycosts, while not undermining equity.6

1. Tax bases should be as broad as possible.A broad-based consumption tax, forexample, will still discourage labor sup-ply on the margin, but choices betweentradable and nontradable goods andservices will not be distorted, if all aretaxed at the same rate. A few items (suchas gasoline, tobacco products, and alco-hol) may be chosen for higher tax rates,because of their negative spillovereffects or because the demand for theseproducts is relatively unresponsive totaxation. As a result, at any given taxrate, efficiency costs will be relativelylow and revenues relatively high.Income tax bases should also be broad,treating all incomes, from every source,as uniformly as possible.

2. Tax rates should be as low as possible (aslong as they raise sufficient revenue tofinance the appropriate expenditures ofgovernment). Of course, the broader the

Rule of Law2 31

y = 0.0515x + 0.2049R2 = 0.2799

0–1–2

Tax revenue as a percent of GDP Tax revenue as a percent of GDP45

40

35

30

25

20

15

10

5

0

GDP per capita (US$ current)40,000 50,00030,000

y = 5E-06x + 0.1848R2 = 0.2227

20,00010,0000

45

35

40

30

25

20

15

10

5

0

Fiscal effort increases with income and the quality of institutions

Source: Authors’ calculations based on Kaufmann, Kraay, and Mastruzzi (2005).

low in developing countries. But in thelight of the second point, higher rev-enues should be sought first by closingloopholes and enforcing greater compli-ance, and only later through higher mar-ginal rates. Income taxes need to apply topersons and to corporations. To keep taxavoidance in check, the top marginal rateof the personal income tax should befairly close to the rate of the corporateincome tax, which means that the rate isnot likely to be all that high.

5. Use property taxes more. Property taxesaccount for only a small share of taxes indeveloping countries. Their coverage istypically not comprehensive, and assess-ments and collection rates are low.Although nominal rates are also low,governments usually find rate increasesin this very visible tax difficult to sellpolitically. Simply raising the tax rateusually would burden only the few actu-ally paying taxes. Higher nominal ratesare likely to be acceptable only with bet-ter tax administration, such as morecomprehensive coverage, better andmore frequently evaluated assessments,and enforced penalties for late payment.7

6. Consider inheritance taxes. Because heirshave not earned the wealth, taxing gifts,

base, the lower the rate needed to gener-ate a given revenue level. Lower rateslower the efficiency costs. The generalrule is that the distorting effect of taxesincreases proportionally to the square ofthe tax rate, so halving the tax rateimplies a fourfold increase in efficiency.From an efficiency perspective, it is bet-ter to impose a single rate on a broadbase of taxpayers, rather than dividingthat base into segments and imposingdifferent rates on each one. This needsto be balanced against the distributionalargument for graduated rate schedules.

3. Keep indirect taxes from being regressive.With a few key exemptions, value-addedtaxes (VAT) can be made less regressive.Bird and Miller (1989) show in Jamaicathat exempting just five specific items fromVAT halves the burden on the poorest 40percent of the population. Reducingregressive excise taxes (as well as importduties), on food or kerosene is also desir-able. To make up for lost revenue from anyof these measures, there are often good rea-sons for higher taxes on private transport.Export taxes are generally best avoided,both on efficiency and on equity grounds.

4. Raise personal income tax collections.Collection from personal income taxes is

Focus on taxation 177

High overall standards of transparency andaccountability are essential if the revenuesfrom extractive industries are to be used well.However, as a second best solution in weakinstitutions of accountability, many countries,developed and developing, do not pool therevenues from extractive industries with otherresources in a unified planning and budgetprocess. Instead, they are channeled to a dedi-cated fund, with special arrangements toearmark the revenues for specific purposes;define reporting and disclosure requirements;and establish oversight bodies to ensure thatthe arrangements are properly implemented.

One common objective of such revenuemanagement arrangements is to save part ofthe revenue stream.The savings may be used forshort-term budgetary stabilization, so thatexpenditures are protected from fluctuations inprices and output.They may also be used tobuild up financial assets, which can generaterevenues over an extended period, in somecases constituting a perpetual fund, so thatfuture generations can benefit from therevenues generated from the depletion ofresources. Azerbaijan, Chad, and Kazakhstan

have legislated savings to deflect pressures tospend revenues quickly and unproductively.

Revenue management arrangements can alsoserve distributional objectives.In Chad,the Oil Rev-enue Management Law assigns a share of revenuesto a “future generation fund”and specifies alloca-tions for poverty reduction.Such earmarked alloca-tions may be anathema to financial managers,butthey communicate and underline the government’scommitment to prudent and redistributive spend-ing.Nigeria’s experience in the Niger Delta duringthe 1980s and 1990s highlights the resentment andpolitical instability that can result when local com-munities see no demonstrable benefits from theextractive industries in their midst.The democraticgovernment has responded by creating a NigerDelta Development Commission to fund localdevelopment,with statutory contributions from thefederal government and the oil companies.

High standards of transparency and account-ability are essential if the revenues from extractiveindustries are to be used for developmentpurposes.To reduce the risk of diversion ofrevenues, the Extractive Industries TransparencyInitiative (EITI), championed by the United King-dom, calls on governments and extractive indus-

tries to report and reconcile payments and rev-enues. Some companies have taken the initiativeof publishing payments to government in someof the countries in which they operate—for exam-ple, Shell in Nigeria and British Petroleum in Azer-baijan.The “Publish What You Pay”coalition advo-cates more systematic disclosure of companies’annual reports and for home-country legislationto make such declarations mandatory. Equallyimportant are improvements in governmentreporting arrangements to ensure transparencyin the application of revenues generated by largeprojects, such as the Chad-Cameroon pipelineand the Nam Theun 2 hydropower project in theLao People’s Democratic Republic.

Revenue management arrangements arelikely to be more successful when they are theproduct of a broad consultative process and therationale is widely understood.Timor-Leste hasfacilitated broad civil society involvement in thedesign of the revenue management arrangementsfor its offshore oil and gas industry.Draft legisla-tion on the industry’s commercial and tax regimeand legislation on a proposed petroleum fund,have been published and subjected to numerouspublic consultations.

Managing resource rents transparently and equitably

estates, and inheritances are consistentwith this report’s notion that predeter-mined circumstances should not affect aperson’s life chances.8 The efficiency argu-ments and the evidence are mixed: parentsmay work more or less to avoid inheritancetaxes, and so may save more or less.Although inheritance taxes may be diffi-cult to collect and are likely to representonly a modest reduction in wealth concen-tration, they may help prevent “extremeconcentrations of wealth from beingpassed from generation to generation.”9 Inaddition, a design that restricts transfers ofcontrol rights on corporations can be goodfor both equity and efficiency.10

7. Avoid implicit taxes. In many instances,the most important taxes affecting thepoor are not formal ones levied throughthe tax code, but implicit taxes, includingbribes11 and inflation. Other implicit taxesto be avoided include many instances of“regulation as taxation” such as quasi-taxes imposed through controls ontrade, prices, credit, foreign exchange, orcapital markets.12

Achieving more equal access to markets isfundamental to greater equity within soci-eties as well as to moving countries ontodynamic growth paths, thus enhancingglobal equity. And both equity and growthare best served by prudent macropolicy(allowing for its countercyclical role). Thischapter is organized around the three mar-kets for capital, labor, and goods (land wascovered in chapter 8) and the macroecon-omy, exploring in each domain the poten-tial and options for leveling the economicplaying field and strengthening voice andaccountability.

How markets relate to equityIssues of design of market-related reformsand macroeconomic policy are often allo-cated to ministries of finance, macro- andtrade economists, financial specialists, andthe like. By contrast, policies for equity,including those for managing the conse-quences of market conditions and macro-conditions, are typically considered thedomain of the providers of schools, healthcenters, rural roads, safety nets, and justicesystems. This division of labor is profoundlyincorrect. The first set of policy domains isas important for equity as the second.

The main issue is access. The playing fieldis typically far from level in the workings ofmarkets. Barriers are intrinsically inequitablewhen they privilege insiders’ access to capital,good jobs, and favored product markets. Butthey are also bad for the innovation andinvestment that lie at the heart of moderneconomic growth. That is why leveling theplaying field has the potential to be bothmore equitable and more efficient. It is alsowhy broadening access typically requiresmore economic competition and more polit-ical accountability.

There are two broad categories ofpathology that make the playing fielduneven (see table 9.1). The first pathologyarises when the influence of powerfulpolitical and economic elites is bad forequity and typically bad for growth—whether this takes the form of outrightpredation by political elites or excessiveinfluence of economic elites in the shapingof policies and institutions, as under “oli-garchic capitalism.” As we read in chapter6, Mexico’s financial system through muchof the country’s history provides an exam-ple of elite capture—a protected, relational,incumbent-oriented financial system.

The second pathology arises when policyefforts to control or manage markets aredirected, at least notionally, to improveequity, but with high costs for efficiency,and are often captured by middle groups(or indeed elites) in ways that harm thepoor. This had its extreme manifestation incommunist economic policy, but it is alsoprevalent in societies in which markets playa large role. Another example is when pro-tection for formal sector workers, whilebringing valued benefits to some, slowsprocesses of restructuring and job creationfor other workers. Both pathologies, butespecially the first, reflect Adam Smith’sconcern that the influential may shape mar-kets to serve the interests of incumbents. Ashe said, “People of the same trade seldommeet together, even for merriment anddiversion, but the conversation ends in aconspiracy against the public, or in somecontrivance to raise prices.”1

The main purpose of this chapter is notto diagnose from where such pathologiescame but to explore possibilities for changethat are feasible within the prevailing politi-cal, sociocultural, and economic context.Casual observation suggests that change is

Markets and themacroeconomy

9c h a p t e r

178

Markets and the macroeconomy 179

possible. The Mexican financial system hasbeen reformed after the 1995 Tequila Crisis.Moroccan industry was (partially) dispro-tected. And many countries no longer pur-sue populist macropolicy, especially afterthey experienced its ill effects.

Reducing barriers to market access willoften bring initial benefits, or redistribu-tions, not to the poorest but to middlegroups. There may even be temporary risesin inequalities in parts of the income distri-bution, for example when there are increasesin returns to skills. This is desirable for effi-ciency and consistent with equity if institu-tions allow households and individuals torespond to the new incentives, and there aresafety nets to manage those who are hurt. Itis undesirable, however, if it creates newpossibilities for middle groups to “hoardopportunities” at the cost of future growthand gains for the poor.2

Design choices will depend on the mar-ket and local context, including the politicalcontext, but two cross-cutting issues areworth highlighting. First, there is often anapparent paradox: when systems are shapedto favor those with influence and connec-tion, economic liberalization should begood for equity; however, that is not alwaysthe case. Liberalization can also be capturedby the powerful, perpetuating inequitableand inefficient economic structures—andrisking political and social backlash to mar-ket-oriented reforms. This is why liberaliza-tion needs to be designed in ways thatpromote genuine competition and effectiveaccountability structures, whether in theform of regulation, transparency, or otherforms of societal control, and sequencedproperly.

Second, while greater equity can be com-plementary to long-run prosperity, the sec-ond pathology vividly shows the potentialfor inefficient choices in the name of equity.Tradeoffs between equity and efficiencyexist. And even when there are aggregategains, protected incumbents will be losers,at least in the short to medium term (ineconomists’ jargon, changes will often notbe Pareto-improving). Whether societieschoose to compensate losers is a matter ofsocial welfare (especially if they are poor)and political economy (especially if they arenot poor).

We now turn to the three markets forcapital, labor, and goods and the macro-economy, exploring in each domain thepotential and options for broadening accessand strengthening accountability. Box 9.1illustrates some of the interactions amongequity, inequality, and growth in China.

Achieving equity and efficiencyin financial marketsThe performance of enterprises drives eco-nomic growth. An innovative and dynamicenterprise sector requires low barriers toentry, effective guarantees for propertyrights, and access to finance. We focus onthe latter here. Unequal access to finance isassociated with reduced productive oppor-tunities and reflects unequal influence.Financial market liberalization can in-crease access, but it can also be captured.Sound technical design and strengthenedaccountability can help expand accesswhile reducing both the risk of captureand disincentives to broader lending, thusenhancing opportunities.

Table 9.1 Two pathologies in the interaction between equity and growth

Domain Policy capture by powerful elites Ill-designed attempts at equity with large inefficiencies

Financial markets Protected, relational, incumbent-oriented financial system in Directed subsidized credit in India and elsewhere, with Mexico for much of its history dismal repayment largely from better-off farmers

Labor markets Repressive labor market conditions in the predemocratic Excessive protection of formal sector insiders in the middle Republic of Korea of the distribution—India, South Africa

Product markets Clove and timber monopolies in Indonesia Protection for inefficient agricultural food production (Philippines) and industry (Morocco) before trade liberalization

Macroeconomic management Regressive resolution of macroeconomic crises in Latin Populist macroeconomic policies (Peru under García) that America and Indonesia fuel future (regressive) crises

Unequal access to finance is associated with unequalproductive opportunities and reflects unequal influenceAs discussed in chapter 5, the access tofinancial services and their cost are un-equally distributed, especially in developingcountries. Many firms and householdscomplain that the right financial servicesare not provided, that procedures for open-ing an account or getting a loan are toocumbersome and costly (with high rejec-

tion rates), and that financial institutionsdemand collateral, which (poorer) borrow-ers typically lack.

Financial institutions respond that theycannot provide services profitably for tech-nical and economic reasons. Poorer groupshave small savings, and seek small loans andinsurance (life, health, crop), which arehard to provide. Smaller clients borrow fre-quently and repay in small installments,making serving their needs very costly. Andthe underserved are new and not experi-enced in business, making them poor creditrisks. But such reasons form only part of thestory. The microcredit movement and largebanks, such as Bank Rakyat Indonesia andICICI Bank in India, show that it is possibleto provide financial services profitably topoorer customers and small firms. Access isalso unequal in areas of finance in whichenforcement is not as much of a concern—such as deposit-taking.

Inequalities in access are also, in part, aproduct of unequal influence. Incumbentswho benefit from restricted patterns offinance may lobby to limit access to finance,or erect other barriers to protect the rents ofestablished firms. Barriers to entrepreneurialactivity are indeed generally more onerous inpoorer, more corrupt, and more unequalcountries.3 Weak property rights are part ofthe problem. But weak property rights can bethe outcome of political economy forces—economic elites have an interest in the selec-tive protection of property rights, becausethey stand to gain more when security ofcontracts and property depends on theirposition, connections, and wealth.4

We are primarily concerned here withthe first pathology mentioned earlier,namely the influence of economic elites onthe shaping of financial systems. In manycountries, a small number of wealthy fami-lies or groups exert extensive control overthe corporate sector, notably through con-trol pyramids in which cross-shareholdingslead to dominant control rights in extensiveparts of the corporate sector, often substan-tially in excess of the share of capital owned.These wealthy families are typically linkedto political elites through economic deals,family connections, and shared social and


China looms large in any attempt to inter-pret development, especially the role ofequity in development processes. Hasn’t itmoved from a highly equitable form ofcommunism to extensive use of domesticand international markets? And didn’t thislead to both rising income inequality andthe most extraordinary pace and scale ofreduction in poverty and expansion insocial welfare in history? This looks, at firstglance, like a brute fact refutation of a cen-tral message of this report: that equity canlay the basis for prosperous development.

An account along these lines misreadschange in China in important ways: many ofthe changes were equity-enhancing in thesense of leading to expansion of opportuni-ties of the bulk of the population (chapter 6discussed the institutional underpinnings ofthese changes). Consider some of the majorshifts in Chinese policy that the literatureinterprets as having driven growth andincome poverty: the institutional shift to thehousehold responsibility system allowingpeasants to produce for themselves (1979and early 1980s), the expansion of townshipand village enterprises (TVEs) and the mas-sive indirect effects of opening to interna-tional trade (whole period), the opening toinward foreign direct investment (especiallyin the 1990s), and the huge internal migra-tion flows. All these led to major expansionsin opportunity (and were major sources ofgrowth) for large segments of the popula-tion. Rural reform quickly expanded oppor-tunities for most peasants.TVEs werebroadly dispersed.While the effects of inter-national opening were concentratedinitially in coastal areas, they havebroadened in scope, both through the indi-rect effects of migration and the relocationof industries inland.

By contrast, where investment has beenlinked to connections and corruption, it isclearly inequitable, in the sense of lackingfair process and equality of opportunity toall potential investors. In the long term, intro-ducing fairer and more transparent processwill be important to sustained growth.

There were periods in which policy-related shifts (such as selective opening, thepricing of foodgrains, tighter controls oninternal migration, and access to urban jobs)were associated with biases either againstinner provinces of China or against ruralareas.These were factors behind the rise ininequality in outcomes and stagnatingincome poverty between the late 1980s andearly 1990s.This was probably inequitable interms of rising inequality of opportunity. Forthese cases there may have been sometradeoff, but few observers argue that suchpolicy-induced biases were essential to Chi-nese growth, in contrast to the overall insti-tutional change and opening.

Moreover, even using the muchnarrower lens of income inequality, the peri-ods when inequality fell (notably the early1980s and the mid-1990s) actually had thehighest growth rates, not the lowest. Andthe provinces where rural inequality rosethe least had the highest growth rates.

There are rising concerns in China overthe adverse consequences for developmentof growing inequality, including in some areasof social provisioning (in health, for example),and concentrations of wealth through con-nections.But there is no evidence that thesebrought benefits in income growth,the issueof concern here,and most observers (and theChinese government) would see this as anarea in which policy could be improved.

Sources: Ravallion and Chen (2004), World Bank(1997b).

B O X 9 . 1 Markets and development: policy, equity, and social welfare in China

cultural capital. Examples include the Mex-ican banking system, until the reforms ofthe second half of the 1990s, and the con-centrated wealth and close connectionsbetween economic and political elites inEast Asia (figure 2.8).5

Concentrated corporate control andwealth should be a concern if it leads tobiases in access toward the influential and,even more, if it is associated with reducedinnovation and dynamism. Reduced inno-vation can occur through direct restrictionson opportunity and the indirect effects ofweaker property rights. Cross-country evi-dence suggests that there is a relationship.Countries with more self-made billionairestend to grow faster, whereas those withmore hereditary billionaires grow moreslowly, suggesting costs of dynastic familycontrol over economies. Societies withgreater family control over the corporatesector also grow more slowly.6

More compelling evidence on the linksbetween unequal power and financial sectordistortions comes from case study material,from the historical experience of now-developed societies, and from contemporarydeveloping countries.7 Table 9.2 provides aselective list of results from recent studies ofdeveloping and transition countries. Thesecase studies illustrate that unequal wealthand influence and low political accountabil-ity can be bad for entry and bad for thebroad protection of property rights—and socan hurt the efficiency, growth, and thehealth of the financial system.

Such adverse long-term effects can bemagnified through crises, further dampen-ing growth and reducing equity. Connectedlending lowers asset quality and makesfinancial systems more vulnerable. Andwell-connected interests do disproportion-ately well in crises, through looting orsecuring greater protection and bailouts, as


Table 9.2 Financial policy and institutions are often captured by the few: case study evidence

Country Evidence

Brazil Public financial institutions in Brazil appear to have served larger firms more than private banks have (Kumar 2005).

Chile Following liberalization in the late 1970s, many privatizations of state-owned banks went to groups of insiders (Larrain 1989).

Czech Republic Mass privatization in the Czech Republic delayed the establishment of a securities and exchange commission, facilitating tunneling(stealing assets by channeling to another firm owned by insiders) (Cull, Matesova, and Shirley 2002).

Indonesia Market attributes large financial value for political connections, suggesting politics rather than economics determined access orrents (Fisman 2001b).

France, pre-1985 Banks, protected and dependent on government support, lend to less productive firms (Bertrand, Schoar, and Thesmar 2004).

Korea, Rep. of The opening up of new segments of financial services provision was dominated by insiders. Increasing openness primarily expandedand strengthened the politically most connected firms (Haggard, Lim, and Kim 2003, Siegel 2003).

Malaysia The imposition of capital controls in September 1998 primarily benefited firms with ties to Prime Minister Mahathir (Johnson andMitton 2003).

Mexico late 1800s There was capture of the financial sector in Mexico in the late 1800s blocking entry in emerging industries (Haber, Noel, and Razo2003).

Mexico early 1990s Lending to connected interests in the early 1990s was prevalent (20 percent of commercial loans) and took place on better termsthan arms’ length lending (annual interest rates were 4 percentage points lower). Related loans were 33 percent more likely todefault and had lower recovery rates (30 percent less) than unrelated ones (La Porta, López-de-Silanes, and Zamarripa 2002).

Pakistan Insider activities have significant economic costs. Politically connected firms borrowed twice as much from government banks andhad 50 percent higher default rates between 1996 and 2002, with economywide costs of rent-seeking estimated to be 0.3 percent to1.9 percent of GDP per year. Brokers trading on their own behalf earned annual rates of return 50–90 percentage points higher thanthose earned by outside investors. Hence, price manipulation by intermediaries helps keep equity markets marginal with fewoutsiders investing and little capital raised (Khwaja and Mian 2004, 2004b).

Russian Federation Russia’s free-for-all banking entry, combined with its choice of a universal banking system, gave great discretion to insiders toconduct asset stripping through the loan-for-shares scheme. The weak political accountability could not stop the capture of stateresources or protected rents (Perotti 2002, Black, Kraakman, and Tassarova 2000).

Thailand Connected lending was large before the 1997 crisis and firms with connections to banks and politicians had greater access to long-term debt (Wiwattanakantang, Kali, and Charumilind forthcoming).

United States, early 1800s New bank licenses went largely to insiders in New York state (Haber 2004).

Ghana, Kenya, Tanzania, Firms that have an owner of Asian or European descent have a 0.34 higher probability of obtaining credit from suppliers Zambia, and Zimbabwe (Fisman 2003).

discussed in the section on macroeconomicmanagement.8

The liberalization paradox: Rapid and premature liberalizationcan also be capturedA seemingly obvious implication of thepathology sketched above is that a moreopen and liberalized financial systemshould be good for access, innovation, andgrowth. Yet overly rapid liberalization canbring new perils. Table 9.2 includes exam-ples of liberalization and privatization offinancial systems leading to highly concen-trated benefits. Rapid privatizations ofstate-owned banks often meant that bankswent to powerful insiders or corporategroups, as in Chile in the 1970s, Mexico inthe 1980s, and the Russian Federation in the1990s.9 In Chile and Mexico, the largestbanks were sold to a small number ofwealthy families in dubious auctions, withforeigners stopped from bidding. The buy-ers were allowed to fund the purchases withloans from the banks themselves, leading toextremely poor incentives for solvency. Inboth countries, the owners used the banksto grant themselves large loans, aimed in

part at accumulating control over otherfirms. In Mexico, this was associated with amajor consumer credit boom.

In the Republic of Korea, chaebols (fam-ily-run conglomerates) came to dominatenonbank financial institutions. This causedserious conflicts of interest and “producednumerous incidents of illegal and unfairactivities, where funds from affiliated finan-cial institutions were exploited for the bene-fit of chaebol’s ailing subsidiaries.”10 And inthe particularly dramatic case of Russia, thefree-for-all liberalization was a source ofboth rapid concentration of assets andincreased financial vulnerabilities (box 9.2).

Rapid or premature liberalization in acontext of low political accountability canincrease financial fragility, and the risk ofopportunistic default.11 Reckless lendingstrategies in Chile and Mexico createdextreme vulnerability for the financial sys-tem, which collapsed when there wereshocks to interest and exchange rates. Agreater percentage of large loans defaultedand subsequent losses on large loans werelarger than losses on smaller loans; losseswere particularly large on loans to partiesconnected with the owners, who escapedsignificant sanctions. In both countries, thebanking system had to be bailed out atenormous public cost, while much of thelent capital probably disappeared in capitalflight, as in Russia. These cases suggest thatrash liberalization with limited scrutiny canresult in concentrated control of the bank-ing system, with owners putting up limitedamounts of their own capital, and weak orcorrupt supervision and public guaranteesto depositors. In all these cases the rise inlow-quality liabilities (and the associatedmoral hazard) became a major factor in thesubsequent financial crises.

In countries with greater accountability,financial liberalizations went a differentway. In France, before the 1985 financialreforms, government subsidies and limitson competition led banks to support lessproductive firms and provide poor-qualityloans. After 1985, the loan allocationsimproved and employment increased.12

Although the reforms exposed some prob-lematic lending patterns, there was no


Within a couple of years of liberalization, thenumber of banks in Russia had risen fromfour to around 3,000. One could argue thatthis was prime evidence that no elite wasblocking entry. But such rapid entry in a reg-ulatory power vacuum precluded anychance of regulatory oversight. It compro-mised the public perception of what a bankis and how it operates, undermining the veryfoundation required for the development ofthe domestic banking sector. In practice,many of these “banks”were not banks butprivate fund management entities used tochannel capital flight.Those raising depositsfrom the general public lent the cash toinsiders, gambled it irresponsibly, or simplyshipped it abroad, leaving the banks asempty shells full of liabilities.

Banks could get away with such behaviornot just because rapid entry overwhelmedthe (rather unprepared) regulators, but alsobecause the banking lobby further promoted

laws that granted banks an extraordinaryfreedom to operate and dispose of otherpeople’s money. Russia endorsed the “univer-sal bank”model, for example, hardly a struc-ture suited to a legal and regulatory vacuum(although there is debate as to whether thiswas a key factor in itself ). Bank lobbyists alsoensured that banks were exonerated fromthe new commercial bankruptcy code (thebankruptcy code established before the 1998crisis vaguely stated that banks would besubject to a specific bankruptcy legislation,which was not even tabled before 1998).

The universal banking structure andlack of bankruptcy system contributed tothe severity of the financial crisis of August1998, resulting in massive losses to deposi-tors, foreign investors, and cost to the statebudget (as many liabilities were transferredto the state-owned Sberbank).

Source: Claessens and Perotti (2005).

B O X 9 . 2 Too much and too little regulation: Russia before and after the transition

financial crisis or capture by the few. Thesystem was less concentrated to begin with,and the reform process received more pub-lic scrutiny.

Premature or ill-designed liberalizationcan also lead to reform backlash, if the ben-efits are perceived to be concentrated in afew powerful groups while the losses arebroadly socialized. There is less specific evi-dence on this for the financial system, but itforms part of a broader pattern of reactionagainst liberalizing processes. Witness thedramatic fall in support for privatization inLatin America between the mid-1990s andearly 2000s (as documented by the Latino-barómetro surveys). Such a backlash islikely to be particularly sharp when associ-ated with gains for particular groups—”economically dominant minorities” in theinterpretation of a series of case studies byChua (2004)—that can heighten the senseof horizontal inequities for other groups. Itcan undercut support for the very reformsthat are critical for equity and growth. Thisis why policy designs need to consider bothtechnical and political economy concerns.

Increasing access to financial services:Technical design, accountability,and competitionIf both financial systems and financial liber-alizations are often captured, what does thisimply for the design of reform? The answer iscomplex and to some degree specific to theinitial financial and legal institutions andpolitical context of a country. But we canoutline some general principles. Options toexpand access entail moving financial insti-tutions closer to the country’s “access possi-bilities frontier.” This will not necessarilyimply finance for all: for all the success ofmicrocredit for the poor, the major benefici-aries of greater access will be small andmedium entrepreneurs from the middleclass. But this is good for the broad-basedgrowth that will benefit all groups. Thisinvolves both issues of technical design anddeveloping the political and social account-abilities that will support and sustain change.

Technical design issues. For financial institu-tions, expanding the client base has much to

do with scale, which is often too small. Recentexperiences like those of ICICI bank in Indiashow that the high transaction costs for smallvolumes and the large cost of expandingreach can be overcome. One option is theinnovative use of existing networks. Postalsystems, with their broad coverage, can beused to deliver new services by many privatefinancial services providers. Many technolog-ical solutions now exist for small-scale bank-ing, from mobile banking to broadening therange of delivery points—through kiosks,small branches, and joint ventures with non-banks. Simpler banking products, like the“Mzansi” account in South Africa, and pre-paid cards for small transactions can lowercosts. Handheld computers have been usedfor quick approval of microfinance loans.Reverse factoring (lending on the basis ofreceivables from a creditworthy institution)using an Internet platform has allowedNacional Financiera (NAFIN) in Mexico toextend trade finance. There has also beenmuch innovation in the market for interna-tional remittances, which many banks haveentered.

Some of these innovations need regula-tory changes—for example, customer iden-tification, anti-money-laundering rules andother rules can hinder access to a bankaccount, as when individuals do not have afixed address or formal job. Better regula-tory approaches for consumers can involveadopting “truth in lending” requirementsfor small borrowers and educating peopleon the risks of (new) financial services.However, a general lesson is to be wary ofregulation in weak environments: all toooften regulation ostensibly designed to pro-tect savers and borrowers is ineffective inprotection but still hinders access.

Are there shortcuts to enhancing access,especially when overall institutional improve-ments will take time? Too often there isemphasis on the more complicated andsophisticated aspects of financial systems,while, some of the basics—broadening accessto financial services, including deposit-takinginstitutions—can be more important froman equity viewpoint. Information sharing canhelp improve competition in banking sys-tems and can be encouraged more quickly in


some segments, including allowing nonbankfinancial institutions access to existing net-works (such as payments systems). There canalso be some scope for specific governmentinterventions. But government interventionsto broaden access through directed credithave typically been unsuccessful, causing inef-ficient distortions with few access benefits—the second type of pathology discussed inthis chapter. Many governments, especiallyin the 1960s and 1970s, used various formsof subsidized directed credit, typicallythrough state banks, to try to channelfinance to poor farmers and small enter-prises. Directed credit undermines institu-tional development because banks have noreason to develop credit analysis skills, asplentiful examples of defunct developmentbanks show. By one account, default ratesranged from 40 to 95 percent throughoutthe developing world.13

Moreover, subsidies for housing, lendingfor small and medium enterprises, and agri-cultural finance are often captured by thewell connected. Some schemes do reach thepoor: India’s social banking program did so,but at a high cost.14 Such schemes thenbecome, at best, an inefficient means ofsupport for the poor, fostering unsustain-able models of financial sector develop-ment. For example, India’s Integrated RuralDevelopment Program provided loans tosocially excluded groups (certain castes andtribes, and women) with high levels of sub-sidy (25 to 50 percent of the loan volume tosuch weak sectors). By 2000, loan recoverywas only 31 percent, and there was little evi-dence of repeat borrowing.

Microfinance clearly has a role in ex-panding access. It is best viewed as a com-plement, not a substitute, for more equi-table financial reform and core financialsystem development. In most countries,microcredit and similar microfinance insti-tutions reach less than 2 percent of the pop-ulation. Only in a few countries is accessreally extensive—Bangladesh, Indonesia,and Sri Lanka stand out with coverageratios in the order of 8 percent or more.15

Subsidies are often used to encourage thesetup of microfinance institutions, but theyneed to be designed carefully because they

can increase final costs, by encouraging theformation of institutions that are too smalland that are forced to raise prices to recoverfixed costs. Co-sharing costs and risks withthe private sector is a key market test. Start-up subsidies and other support can fosterexploration of alternative business models,phasing out support over time. Maintaininga segmented system makes sense until themicrofinance sector matures, with strongermicrofinance institutions coming into thecore financial system as they evolve.

Accountability and competition. Technicaldesign is important, but the core of a reli-able reform is building political and regula-tory accountability. Public scrutiny has akey role, given the risks of capturing reformprocess and institutions. Potential actors tohelp oversee the process include associa-tions of small firms, consumer groups,NGOs, media, and labor unions. But giventhe technical and complex nature of finan-cial sector functioning, societal accounta-bility is likely to be most effective if groupswith an interest in a more open financialsystem are empowered, engaging independ-ent nongovernmental technical bodies withthe capacity to analyze financial sector con-ditions. The shadow regulatory commis-sions established in almost all regions, andsuch research centers as the recently formedCenter for Financial Stability in Argentina,can be sources of education and avenues forinterest groups to express their voice. It willremain important to design these mecha-nisms with care to avoid the creation of vetopowers to reform (which can lead tocounter-reform capture!)

It is possible to promote reforms thatbuild more reliable and inclusive financialsystems in a context of unequal influence.Formal regulatory structures can comple-ment societal accountability. The develop-ment of the stock markets in Poland and theCzech Republic show what regulation anddisclosure can do. At the transition fromcommunism in 1989, these two countrieswere quite similar in economic structure andhistory. But the design of financial reformswas very different, driven primarily by differ-ences in philosophy toward markets.16


The Czech Republic went for a radicalvoucher-based privatization of state-ownedassets, convinced of the power of the mar-ket to organize itself: with property rightstransferred to the private sector, it wasexpected that private actors would effi-ciently contract with one another. Polandpursued a more gradual approach, based oncase-by-case privatization and a measuredinstitutional development effort to buildregulatory and supervisory capacity. Com-pany and securities laws in the two coun-tries reflected these differences in approach,with much greater requirements in Polandthan in the Czech Republic for disclosure tothe public, more protection of minorityshareholders, and more power for the inde-pendent regulator.

The results were quite different. Thestock market in the Czech Republic startedbigger but was quickly dominated by cor-porate insiders, who captured “58 percentof the values of companies over and abovetheir legitimate shareholding, comparedwith an insignificant 1 percent in theUnited States.”17 There was widespread“tunneling,” a form of asset stripping byinsiders through transfers to other institu-tions they controlled. The Polish stock mar-ket, by contrast, started more slowly, butthen overtook the Czech market (figure 9.1)Public scandals led to the regulator effec-tively pursuing violations, using the greaterlegal protections, and laying the basis formore broad-based property rights, greaterconfidence, and openness. By the late 1990s,there were already several initial publicofferings in the market.

Segmentation provides another exampleof the need for appropriate accountabilitymechanisms. Financial sector regulation inmany countries, including developed ones,enforced segmentation for long periods—both on a geographic basis and by type offinancial services, as among commercialbanking, investment banking, and insurance.The Italian experience with dispersed localbanking suggests that mutual and coopera-tive banks performed an important functionin supporting local activities, which wasmuch better than state-owned banks orbanks dominated by politicians.18 But regu-

latory capacity has been eroded by techno-logical change. Given that segmentationoften resulted in capture by local elites, theerosion of barriers has likely improved accessas often as not. Yet there is room for smaller,locally managed intermediaries to promoteaccess. Such locally focused intermediariesneed explicit disclosure and accountabilityrequirements to local users (as opposed tolocal politicians), which has been the tradi-tion in cooperative or mutual banks, to limitthe undue political influence of the few.

Opening the financial sector to competi-tion from foreign financial institutions canalso spur financial broadening. Foreignbank entry can help by improving efficiencyand stability, reducing protected profits,and forcing (local) financial institutions tofocus more on providing financial servicesto all. Financing obstacles are perceived tobe much lower by borrowers in countrieswith high levels of foreign bank penetra-tion, with evidence that even small enter-prises benefit. But note that allowing theentry of foreign banks is not synonymouswith capital account liberalization. Rapidlyopening the capital account before ade-quate domestic regulatory and supervisorystructures are in place can be dangerous,especially in a world of large, and oftenherdlike, international capital flows, cou-pled with politically connected lending ofpoor quality. This increased the vulnerabil-ity of Indonesia, the Republic of Korea, andThailand in the East Asian crisis.

Finally there is the potential for externalcommitment. The relative success of Cen-tral European countries in strengtheningaccountability has been attributed to theconstraint on abuse induced by the need toprepare for accession to the EuropeanUnion. In Slovakia, after a decade of influ-ence-peddling and slow reform in thefinancial sector, the pendulum swungtoward reform only as the date of possibleEU accession approached.19

Achieving equity and efficiencyin labor marketsFor most of the world’s people, economicopportunities are primarily determined, or


01991 1992 1993 1994 1995 1996 1997 1998

20,000

15,000

10,000

5,000Poland

Czech Rep.

25,000US$ millions, end of year

Figure 9.1 Poland’s stock marketstarted slowly but then surpassed theCzech Republic’s

Source: Glaeser, Johnson, and Shleifer (2001).

at least mediated, by the labor market—informal and informal work. The wages andemployment conditions in the labor marketaffect the quality of life of workers and theirfamilies, sometimes in ways that may seemharsh or unfair. The functioning of thelabor market has a profound effect onequity—across workers, in patterns ofaccess to work, and between workers andemployers. Government intervention toachieve greater equity is frequent in labormarkets, but often with costs in terms of effi-ciency—exemplifying the second pathology.While this is an area in which genuine trade-offs exist between protection of weaker work-ers (which is good for equity) and flexibility(which is good for growth), elite capture andgross inefficiencies can be addressed throughbetter design and broader accountability.

Equity and efficiency reasons to intervene in the labor marketLabor markets are different from other mar-kets. Unlike the markets for many com-modities, labor markets generally are notcompetitive. They may be characterized byuneven market power (between employersand workers), by imperfect mobility ofworkers, by insufficient information, or bydiscrimination. These imperfections gener-ate rents in the employment relationship,which both sides can try to capture. Thiscan lead to unfair and inefficient outcomeswhen the bargaining position of the work-ers is weak. For example, employers mayunderpay workers who are not mobile,force workers to work in hazardous condi-tions, or discriminate against vulnerablegroups. Private markets, left to themselves,also do a poor job of protecting workersagainst the risk of unemployment. In theabsence of perfect access to financial mar-kets, or complete insurance markets, work-ers may not be able to smooth consumptionin response to labor income shocks. If theycannot gain access to financial markets,they may also be prevented from movingfrom bad jobs to good.

All governments, irrespective of income,intervene heavily in the labor market. Gov-ernments typically intervene to correct these

failures: to protect workers and endow themwith rights and “voice” in the employmentrelationship, to empower unions to repre-sent workers in negotiations with employ-ers, to ensure compliance with labor lawsand regulations, and to provide insuranceagainst income shocks. Public interventioncan improve market outcomes and lead tosignificant equity gains: more equal oppor-tunities for workers, better working condi-tions, and less discrimination. It can alsolead to large gains in efficiency: for example,by allowing full use of the labor of discrimi-nated groups, by increasing labor mobility,or by better managing income risk.20

The problem is that poorly designed orinappropriate government intervention canalso make things worse, with results that arebad for equity and efficiency. For example,excessive protection of formal sector insid-ers can lead to “rationing” jobs in the formalsector, pushing surplus labor into eitherinformal employment (as in India) 21 orunemployment (as in South Africa). 22

The problem is particularly acute indeveloping countries, because labor marketregulations and standards typically applyonly to formal sector workers, leaving themajority of the workforce uncovered.23 Pro-tecting workers through legislation and reg-ulation that is enforced only in the formalsector, without other measures to improveworking conditions in informal employ-ment, can reinforce the segmentation betweenformal and informal employment in waysthat are inherently unfair. In Colombia,workers are legally entitled to severance pay-ments for dismissals deemed unjust, butthese entitlements are not enforced in theinformal sector, which employs more thanhalf the workforce. Not only do Colombianinformal sector workers not benefit from thelegislation, but arguably they are harmed byit, because the resulting higher cost of laborin the formal sector limits formal employ-ment opportunities for “outsiders” (mainlywomen and youth).24

In reality, the distinction between formaland informal employment is often blurred.Some authors argue that the informal econ-omy functions partly as an unregulatedentrepreneurial sector, often voluntarily


entered even at the expense of lowerincome.25 It is clear that the informal sectoris heterogeneous, and includes both thosewho choose to work there and those whowork there out of necessity. Those in the topstrata—microentrepreneurs who hire oth-ers and many of the self-employed—do rel-atively well in average earnings. Those at thebottom—intermittent casual laborers andindustrial outworkers—do not. Womentend to be underrepresented in the topstrata and overrepresented in the bottom.26

They also often earn less than men withineach strata—although some of these differ-ences may reflect voluntary choices formore flexible, part-time work. In a recentstudy, the International Labour Office(ILO) argued that the formal and informalsectors are part of a continuum of workingconditions, earnings, and rights.27 A signifi-cant share of formal sector employees hadsome of the (poor) working conditionsassociated with informality, while a fractionof informal sector workers enjoyed condi-tions more typically associated with formalsector jobs. The challenge for governmentsis to shift more jobs along this continuumtoward better working conditions andhigher wages, and to do so in ways that donot come at the expense of efficiency.

Indeed, poorly designed or inappropri-ate government intervention can also beinefficient and bad for long-term growth.Recent research on India, for example, sug-gests that starting from a common legalframework (the Industries Disputes Act of1947), the states that amended the legisla-tion in the direction of reinforcing securityrights of workers and other prolabor meas-ures had lower output and productivitygrowth in formal manufacturing than thosethat did not change it or that made laborregulations more flexible.28 Relatively pro-tective legislation may have reduced oppor-tunities for workers—especially the major-ity without a formal sector job.

A look at some African labor marketsillustrates the impact of government poli-cies (figure 9.2). Many countries—includ-ing Ghana, Uganda, and Tanzania—havelarge self-employment sectors, which absorbincreases in the labor supply and help keep

unemployment low. South Africa stands insharp contrast, with a small informalsector—absorbing only about 19 percent ofthe total workforce in 2002, much lowerthan the share of nonagricultural employ-ment in other African countries—and highunemployment (42 percent in 2003).29 Partof the story lies in much larger wage differ-ences between formal and informal sectorwork in South Africa than in the othercases. But it also appears to be caused by theunusually small size of the informal sector(compared with Latin America, for exam-ple). Some suggest that the legacy ofapartheid that inhibited the development oftraditions of small-scale entrepreneurialactivity, and labor regulations that areenforced for firms of all sizes (depending onthe industry and region), may explain thelack of entrepreneurs and small firms inSouth Africa.

In Ethiopia, on the other hand, the major-ity of the urban unemployed are well edu-cated and from middle-class households.30

They also tend to be young, have never heldpaid work, and have a median duration ofunemployment of nearly four years. Indeed,about half of young unemployed males aresearching for public sector jobs, which pay onaverage a 125 percent premium over self-employed work.

Addressing links with unequal powerGovernment intervention in the labor mar-ket is often a reflection of the underlying dis-tribution of political agency. Governmentsmay (and often do) intervene in the labormarket in pursuit of goals other thanaddressing market failure. They may inter-vene to buy off the support of certain groups(for example, urban formal sector workers)or to suppress social dissent under anauthoritarian regime or to serve the interestsof those with greater political influence. Oli-garchic capitalist societies can be associatedeither with labor repression or with union-ized and (relatively) advantaged formal sec-tor workers who share in the rents.

Interventions aimed at shifting aggregatewelfare toward politically powerful middlegroups, often in the name of equity, at theexpense of others (an illustration of our


SouthAfrica

Ethiopia(urban)

Ghana0

25

50

Percentage of total labor force

75

100

Employment by sector for selectedAfrican countries

UnemployedSelf-employment(including agriculture)Private wage jobPublic wage job

Figure 9.2 Patterns of employment andunemployment vary widely acrossAfrican countries

Source: Kingdon, Sandefur, and Teal (2005).Note: Ghanaian data are for 1998/99 and publicwage employment includes government andstate enterprise workers. South African data arefor 2003 (Labor Force Survey). Ethiopian data arefor urban areas in 1997 (Labour Force Survey)and because of definitional issues may not befully comparable.

second pathology) are inherently bad forequity and usually bad for efficiency. Politi-cally influential energy sector and teacherunions in Mexico, for example, have pro-tected their employment and wage entitle-ments by blocking reforms that would leadto, respectively, energy sector reform andhigher-quality and more equitable school-ing. Public sector workers in France haveused their political force, with the aid of mas-sive strikes, to curtail attempts to bring theirnonwage benefits and other entitlements inline with those of the private sector.

Stronger civil and political rights andbroader mechanisms for voice can reducethe likelihood that the government’s laborpolicy agenda will be hijacked by politicallypowerful groups. There is a strong associa-tion between democracy and the level ofwages, both across countries and withinsocieties that have experienced a politicaltransition, such as the Republic of Korea,and Taiwan, China.31 Stronger respect forcivic rights in Latin America has also beenassociated with greater formalization ofemployment and a higher wage share.32 InSpain, the transition to democracy in themid-1970s led to demands for greaterequity that were associated with the legal-ization of free trade unions,33 the rapid

introduction of extensive social transfers(including pensions and unemploymentinsurance), and the implementation of pro-gressive income taxation. The result was ashift from a regulated state, in which pro-tection for workers came mainly throughpermanent jobs and controlled housingrentals, to a liberal economy with flexiblemarkets, and more extensive public provi-sioning of services (see focus 3 on Spain).

Better design: Can labor marketinstitutions be designed to be pro-growth and pro-equity?The challenge for governments is to designinterventions that balance equity and effi-ciency goals in ways that are within a coun-try’s institutional capacity. History suggeststhat this is a complex task, and there are realtradeoffs that need to be assessed. Differentsocieties are likely to make different choices.

Scandinavia and the United States havevery different sets of labor market institu-tions, yet they share a good track record ofsolid growth and high employment-to-population ratios (figure 9.3). The Nordiccountries have mandated generous benefitsand protection for workers, financed by ahigh tax effort. But they also have a highlycoordinated and centralized approach towage-setting and policymaking, whichallowed all parties to internalize the conse-quences of their actions, with the unionmovement historically an advocate foropenness and competition. The UnitedStates leaves the setting of wages and workconditions, including benefits, much moreto discretion and employer-worker negotia-tion. This fits well within its tradition ofdecentralized bargaining, which gives free-dom to individual firms to bargain withtheir workers in response to their varyingeconomic and financial conditions. It leadsto greater wage inequality and more work-ers without health and other forms ofinsurance. But it is consistent with lowertaxes and high levels of flexibility.34 TheNordic countries and the United Statesopted for different labor market models (inline with their history, legal tradition, andsocietal preferences), but all succeeded indelivering to their workforces a large pool


0

10

20

30

40

50

2000199519901985

Finland

Norway

United States

Denmark

Sweden

198019751970

60Labor productivity (GDP per hour worked, US$)

Figure 9.3 Different labor market institutional setupscan yield equally good productivity growth paths:Scandinavia versus the United States

Source: Underlying series for OECD (2005).Note: Measures used are GDP volumes, in U.S. dollars, at constantprices, constant purchasing power parity (PPPs) prices with abase year of 2000 and total hours worked for total employment.

of job opportunities, productivity growth,and rising incomes over the long term.

What doesn’t work? There are also manyways of getting labor market institutions“wrong.” And when countries get it wrong,one segment of the labor market (typicallypublic and formal sector workers who rep-resent the middle and high end of the usagedistribution) benefits from extra protectionat the expense of others. Outcomes arehighly unequal, and the costs to efficiencyand growth are usually severe. The experi-ences of India, South Africa, and Colombiacited earlier are vivid examples.

Governments also get it wrong if theyintervene extensively in the labor marketwhen product markets are not competitive.A typical example is the unionization ofpublic sector workers. In Mexico, before thereforms of the 1990s, workers in public util-ities and in the public oil sector securedhigh wages by capturing a share of the rentsgenerated in these monopolistic sectors. Buttheir gain came at a cost for employmentand competitiveness in Mexico’s privatesector. Mexico’s experience is not unique: itapplied equally to state enterprise workersin Turkey until the 1990s and to public sec-tor workers in India and many other coun-tries.

Things can also go greatly wrong whengovernments mandate protection with noattention to incentives. In much of Europe,governments implemented generous unem-ployment benefits with little connection toworkers’ actual job search behavior. Theresult was an increase in the duration ofunemployment and the emergence of long-term unemployment, with its destructiveimpact on human capital, loss of employa-bility, weakened ties to economic and sociallife, and for many, high degrees of povertyand social exclusion.35

What works? The key to avoiding these pit-falls is achieving a balance between protec-tion and flexibility. Design specifics mattera lot, as does the underlying structural con-text. Collective organization of workers isone of the main channels for securing bet-ter and more equitable working conditions.Free trade unions are the cornerstone of any

effective system of industrial relations.Unions act as agents for labor, coordinatingthe demands of workers and organizingthem into a single entity whose collectivebargaining power matches that of theemployer. Trade unions can help provide apositive work environment by reducinglabor turnover and by promoting workertraining and higher productivity. Unionshave also been found to reduce inequalityand wage discrimination in countries asdiverse as Ghana, the Republic of Korea,Mexico, and Spain.36 Unions also can havean important noneconomic role. They havebeen a force for progressive political andsocial change in many countries (Poland,the Republic of Korea, and South Africa).

But the involvement of unions in wage-setting can also have significant negativeeconomic effects. Evidence from industrialsocieties suggests that union involvementreduces the employment of young andolder workers (“outsiders”) and benefitsprime-age males and females.37 Unionsoften act as monopolists, improving wagesand work conditions for their members atthe expense of consumers and nonunionlabor. For example, recent studies of manu-facturing wages in Africa show substantialunion wage premiums (30 to 40 percent insome countries).38

Formal trade unions are most effective atimproving conditions for workers withouthuge efficiency costs when product marketsare competitive, so that unions cannot raisewages for their members by capturing rentsat the expense of other parts of society;when collective bargaining arrangementsand institutions have enough flexibility toaccommodate different demand and supplyconditions for different types of workers;and when unions operate in a context thatallows them to internalize and absorb thecost of their actions. On the other hand,when unions are co-opted by political elitesor by the state, their actions can have signif-icant costs for efficiency.

An illustration of the potentially highlypositive role of unions in improving workingconditions while supporting productivitygrowth comes from a study of worker-firmrelationships in high-value export crops in


worker associations in India, South Africa,and Thailand suggest that organizing infor-mal workers decreases the workers’ invisi-bility to policymakers and legislators, helpsthem gain access to information, gives themvoice and self-identity, and in some caseshelps to provide them with a range of socialprotection services (box 9.3).

Providing income security is anotherarea in which the structural context anddesign specifics, which pay attention toincentives and reward desirable behaviors,are critical to policy outcomes (box 9.4).This is also true of minimum wage policiesfor which the key to avoiding large effi-ciency costs is to get the level right and toallow for enough flexibility across types ofworkers to accommodate different demandand supply elasticities for their labor.40

Design specifics and the broader struc-tural context are equally critical to the suc-cess of legislation on other work standards(health and safety) or protection for specificvulnerable groups (such as child laborers,ethnic minorities, or the disabled). There isan international consensus that core laborstandards—freedom from forced and childlabor, freedom from discrimination at work,freedom of association, and the right tocollective bargaining—have such intrinsicvalue that they should always be pursued.But even for these core standards there arequestions about how to achieve them mosteffectively and with minimum cost.41

An example of government interven-tion to protect workers from abuse comesfrom Cambodia’s successful experience withimplementing core labor standards in thegarment industry. Starting in 1999 Cambodiacould earn a higher quota for exports to theUnited States by demonstrating improve-ments in working conditions. A monitoringsystem—developed and implemented by theILO, with support from the U.S. Departmentof Labor, the Government of Cambodia, andthe Garment Manufacturers Association ofCambodia—has virtually eliminated theworst labor abuses, such as child labor andsexual harassment. A recent survey showedenforcement of core labor standards in thegarment sector has boosted Cambodianexports to Europe and North America.42 But


Developing countries have trade unionsand large informal economies, twophenomena commonly thought to beincompatible. But recent work by Women inInformal Employment: Globalizing andOrganizing (WIEGO) has uncovered a sur-prising amount of organizing in the infor-mal economy. Formal trade unions thatextend their coverage to informal workersand trade unions of informal workers aretwo organizational forms that haveemerged. Organizing can also be done bycooperatives, savings-and-credit groups,producer groups, and neighborhood andtrade associations.

Organizing in the informal economydiffers from organizing in the formal econ-omy along several dimensions. Collectivebargaining can take many forms. Bargain-ing partners are not just employers but canalso include municipal authorities, police,wholesalers, and other interest groups.Activities of informal workers’ trade unionscan also extend beyond collective bargain-ing and include a range of services, such assavings, credit, social security, andadvocacy. Because members do not work ina standard workplace or for a singleemployer, membership in trade unions cantake unusual forms.

• Registered as a trade union, the Self-Employed Women’s Association (SEWA)in India is both an organization of poorself-employed women workers and amovement combining elements of thelabor, cooperative, and women’s move-ments. It offers a broad range of servicesto its nearly 700,000 members, includingbanking, health care, child care,insurance, legal aid, housing assistance,and capacity building (www.sewa.org).

• StreetNet International, launched in 2002in South Africa, has 15 affiliates (unions,cooperatives, or associations) in Asia,Africa, and Latin America that organizestreet vendors, informal market vendors,and hawkers. As of early 2004, these affili-ates represented 128,000 members(UNRISD Gender Policy Report).

• HomeNet Thailand is a network thathelps organize informal home-basedsubcontracted and own-account workers(mainly women). It drew attention to theplight of these workers following thefinancial crisis in 1998.

Sources: United Nations Research Institute forSocial Development (UNRISD) (2005); Womenin Informal Employment Globalizing and Orga-nizing (2005).

B O X 9 . 3 Organizing in the informal economy

northeastern Brazil. Union activity amonglandless agricultural laborers in this case wasan important factor behind improvementsin work practices that led to higher quality(critical for export crops), higher productiv-ity, and better working conditions for land-less workers. To effect this change, however,the union had to shift from defending theinterests of its traditional base of support(small farmers) to representing the interestsof a larger group of landless laborers. Successwas also facilitated by the fact that, on theemployer side, the union was negotiatingwith large firms from southeastern Brazil.These southeastern firms had experience incollective bargaining, in contrast to themore traditional large farmers of the north-east, which were accustomed to morerepressive and conflictive labor relations.39

Collective organization of workers canalso secure greater bargaining power andthus better working conditions for informaleconomy workers. Studies of informal

whether the system will survive the end of theU.S. quota incentives remains to be seen (seechapter 10).

How to reform a “bad” set of labor market institutionsReforming labor market institutions istechnically and politically difficult. It istechnically difficult because reforms needto be coordinated across a variety of labormarket institutions, and often also withreforms outside the labor market. It ispolitically difficult because there usuallyare vested interests in maintaining the sta-tus quo. Moreover, the short-term costs ofreform can be large and unevenly distrib-uted. Take reforms to reduce employmentprotection: those currently protected seethemselves as having much more to losefrom reform than to gain from such areduction. And if they are also politicallyinfluential—represented by unions andwith political voice—their power to block

reforms may be an insurmountable bar-rier.

Several countries have implementedsubstantial labor market reforms more orless successfully: Ireland, the Republic ofKorea, the Netherlands, New Zealand, andthe Slovak Republic among OECD coun-tries; Chile and Colombia among develop-ing countries (see box 9.5 for Colombia andthe Slovak Republic). China is in the midstof a large labor market transition, and theBalkans are struggling through dramaticlabor market reforms.

Experience suggests that effective changeinvolves a combination of factors: designingand implementing a consistent and com-prehensive policy package; tackling vestedinterests; broadening societal accountabil-ity, including increasing the voice of poorergroups; and, in some cases, compensatinglosers. Macroeconomic and financial shockscan facilitate change, although not always apositive one. Reforming institutions in the


Left to itself, the labor market does a poor job ofprotecting workers against a sharp loss in incomeassociated with unemployment. As a result, mostsocieties have developed ways to cope with thethreat of job loss. Often this involves some com-bination of informal support mechanisms, privatesavings, and obligations on employers.Whenthese mechanisms break down—as they dowhen the shocks are large, sudden, protracted, oraffect an entire community—the governmentneeds to step in. Government intervention typi-cally involves one or several of the followinginstruments: job security regulations, mandatedseverance pay, unemployment insurance, ormandatory self-insurance mechanisms.

Job security legislation is typically aimed atprotecting jobs and preventing job destruction.Most evidence suggests that it is quite effectiveat doing so. Across countries, more stringentemployment protection legislation (EPL) appearsstrongly associated with more stableemployment. But EPL reduces job destruction ata significant cost, as the expectation of high sep-aration costs makes firms more reluctant toexpand employment, and makes it lessprofitable to start new ventures or create newfirms. So, employment protection also reducesjob creation. For example, researchers found thatstrict new job-security laws enacted in the 1980sreduced employment in many industries in Zim-

babwe (Fallon and Lucas 1993). Overall, the neteffect of job security legislation on employmentis ambiguous (Bertola 1990; OECD 1999; Bertola,Blau, and Kahn 2001; Kugler 2004).

What is clear is that EPL changes the natureof unemployment. Lower job destructionreduces the incidence of unemployment. Butlower job creation increases the duration ofunemployment and can lead to the emergenceof long-term unemployment.

Not surprisingly, EPL appears to have a dif-ferent impact on different groups of workers. Inboth Colombia and Spain the reduction of dis-missal costs and job security provisions wasassociated with moderate increases in theemployment of young men and women (Kugler2004; Kugler, Jimeno, and Hernanz 2003). ForChile, Montenegro and Pagés 2004 find that jobsecurity regulations reduce the employmentrate of youth and the unskilled to the benefit ofolder and more skilled workers. A study acrossthe countries of the Organisation for EconomicCo-operation and Development arrives at a sim-ilar conclusion: consistent with the story thatEPL protects “insiders,” stricter EPL increases theemployment of adult men and reduces that ofyoung workers and women (OECD 1999).

Given the complex effects of EPL, how cangovernments best intervene to help protectworkers against temporary drops in income

associated with unemployment? Some EPL maybe efficient, reducing excessive volatility inturnover. But too strong EPL—as is typical ofmany formal sectors in the developing world—slows the pace of creative destruction central toinnovation and growth, with disproportionateadverse effects on those without “good” jobs.Yetreducing EPL needs to be complemented bygreater worker security that is not linked to spe-cific jobs, both on social welfare and politicaleconomy grounds.

The design of the optimal solution dependson the institutional and administrative capacityof government and on the structural characteris-tics of the labor market (Blanchard 2004). Coun-tries with significant administrative capacity andmedium to high incomes can implement unem-ployment insurance systems with incentives forjob search (declining benefits with duration andprovision of benefits conditional on acceptanceof acceptable job offer). Middle-incomecountries reluctant to implement a full-blownunemployment insurance system can supportself-insurance mechanisms, such as mandatorysavings accounts (but because of the limits toself-insurance, these are not as effective). Low-income countries can opt for public worksschemes, which if effectively designed are self-targeting and can be implemented even whenlevels of informality are high (chapter 7).

B O X 9 . 4 Employment protection legislation

midst of widespread job loss and highunemployment (as in the Republic of Koreafollowing the 1998 financial crisis) is partic-ularly difficult, even though it may be easierat such times to achieve societal consensusfor reform.

Designing a consistent and coherent policypackage. One of the strongest lessons fromcountry experience is that piecemeal reformdoes not work (tinkering at the margin usu-ally has perverse distributional effects).Moreover, reforms need to cover a range oflabor market policies and to be linked withreforms in social protection systems.Reform is more effective and more equi-table when different labor market instru-ments are coordinated: measures to reduceinsider power and increase flexibility bylowering the restrictions and costs of firingcan be linked to setting up of unemploy-ment insurance mechanisms and eliminat-ing dual status contracts. Reforms in othermarkets and the public sector are often keyto the success of labor market reform. The

depth and competitiveness of other markets(including product and financial markets)are critical.

Tackling vested interests. Reforms are oftenheld hostage by politically more powerfulgroups. For example, policies to reduceemployment protection, allow for submini-mum wages, or streamline and improve pub-lic sector employment typically encountersharp resistance from unions. Building broadsocietal consensus for reform is often theonly way to tackle these vested interests. As afirst step, this may require documenting thehigh costs of bad labor market policiesthrough good data collection, analysis, anddissemination (as in the Slovak Republic).

Broadening social accountability. Buildingsocietal consensus in support of labor marketreform may require specific measures toempower the groups of so-called outsiders ordisenfranchised workers who bear the costsof nonreform. It helps to have political partiesand societal organizations with broad bases


Comprehensive reform in the SlovakRepublicIn 2000 unemployment in the Slovak Republicreached 19 percent of the labor force—the high-est in the OECD at that time.The main factor wasthe substantial job reallocation generated by thetransition to a market economy, compounded bylow labor mobility because of skill and regionalmismatches. But the impact of the transitionshock on the labor market was worsened by aninadequate set of institutions: high rates of taxa-tion of labor and overly generous unemploymentbenefit and social assistance systems, which dis-couraged job search and encouraged informality.

Reforming these institutions in the midst ofhigh unemployment was extremely difficult,particularly for a reformist government with asmall majority in parliament. But in early 2003,bolstered by its reelection, the reformist Slovakgovernment undertook a comprehensive andambitious reform of social and labor market pol-icy.The government’s multipronged strategycombined measures to reduce the taxation oflabor, increase the incentives to work throughreform of unemployment insurance and socialassistance, invest in the skills of labor andemployability, improve the matching of workers

and jobs through higher labor flexibility andmobility, and strengthen state administration inlabor and social policy.

The new strategy represented a markedchange in philosophy in labor and social policyin the Slovak Republic: from a system that mixedingredients of insurance and redistributiontoward one that separates social insurance fromequity objectives; and from a tradition of entitle-ments based on subjective or “moral” norms toone that guarantees a certain living standard toall citizens irrespective of the reason they mayhave for being poor, but that rewards individualinitiative and motivation.

The key ingredients in pushing the reformthrough were a reformist government with astrong popular mandate; strong leadership andtechnical competency from the Ministry ofLabor and Social Affairs; accession into the Euro-pean Union as a disciplining device; and wide-spread public perception, built on analysis anddissemination of this analysis, that institutionalreforms were needed. Moreover, labor marketreform was carried out not in isolation but aspart of a broader policy package to make theSlovak economy more competitive in light of EUaccession.This package included a substantial

overhaul of personal income taxes, as well asreform of the education system.

Partial labor market reform in ColombiaIn 1990 Colombia introduced a labor marketreform that substantially reduced the costs ofdismissing workers.The reform reduced sever-ance payments, widened the definition of “just”dismissals, extended the use of temporary con-tracts, and speeded up the process of mass dis-missals.The joint effects of these reforms wereto reduce the costs associated with firing work-ers in firms covered by the legislation. But thereforms did not affect informal sector firms,which did not comply with the legislation.

An analysis of the effects of the reforms sug-gests that they did increase the dynamism ofthe Colombian labor market by increasing exitrates into and out of unemployment (greaterchurning).There was an increase in workerturnover for formal sector worker, greatestamong young workers, more educated workers,and workers employed in larger firms.Thereforms may have also contributed to increasingcompliance with labor legislation by reducingthe costs of formality.

Source: Kugler (2004).

B O X 9 . 5 Two cases of labor market reform: one comprehensive, one partial

of representation and support. When this isnot the case, it may be necessary to look forways to open up institutions and give greatervoice in bargaining (at all levels) to represen-tatives of disenfranchised groups. This iseasier when there are democratic local gov-ernments and strong autonomous associa-tions—independent private business associa-tions, worker associations that represent theinterests of specific groups, and so on. Theindependent private sector is also a naturalally when it comes to reforming public sectoremployment and wage practices.

Compensating losers. The short-term costsof reforms can be high for certain groups ofworkers: unemployment insurance and socialassistance reform in the Slovak Republic dis-proportionately hurt Roma workers andthose living in high unemployment regions.So it may be necessary to compensate the los-ers. It is best to do this in ways that addressthe obstacles that losers face in reentering thelabor market (support for education or train-ing) or that facilitate labor mobility andreward work incentives (transport vouchersfor workers moving from social assistance towork). Such compensatory measures havebeen introduced as part of the Slovak Repub-lic’s labor market reform package.

Product markets and trade reformProduct markets are intimately related toequity, with two-way patterns of causation:product markets shape the distribution ofeconomic opportunities, and inequalities ininfluence shape the functioning of productmarkets. Both the design of external tradepolicy and the workings of internal productmarkets reflect patterns of influence. Re-moving barriers and excessive regulationneeds to be complemented by measures toexpand skills, infrastructure, and safety netsto achieve genuine access and help losers.

Market broadening and deepening arecentral to the expansion of opportunity:directly for firms and the self-employed,indirectly for workers. How equitable thisexpansion of opportunity is depends oninteractions among external trade opening,domestic markets, patterns of infrastruc-ture, labor markets, safety nets, and other

measures to improve the business climate.Product market and trade reforms havegreat potential to bring expansion inopportunity, but there can be costs in theshort to medium term, and these can hitparticular groups, from relatively powerfulprotected incumbents to middle and poorergroups. The costs are associated with howmarkets and investment processes work:labor is typically not fully mobile, new skillstake time to acquire, and new investmentsare often lumpy and can take time, espe-cially when firms face imperfect credit mar-kets (see earlier section) and an uncertaininvestment environment.

The functioning of product markets isembedded in political and social structures.Elite capture ranges from the apparent andegregious—as in the granting of the Indone-sian clove monopoly to a son of PresidentSuharto (a monopoly since disbanded)—tothe less transparent shaping of trade policyto protect the profits of the influential. It isalso true that policies with (genuine orrhetorical) equitable purpose can lead tooutcomes that are bad for growth and mixedfor equity. This is evident in the characteris-tically high levels of protection for relativelylabor-intensive manufacturing and for foodproduction (such as maize in Mexico, rice inthe Republic of Korea, and the infamousagricultural subsidies in the EuropeanUnion, Japan, and the United States). Whilepoorer groups sometimes gain, it is morecommon that middle and elite groups arethe main beneficiaries, while food con-sumers lose out.

While policies that reduce the power ofincumbents in product markets will typi-cally be good for efficiency and equity, aversion of the “liberalization paradox” dis-cussed earlier may often apply. Groups andindividuals with economic capacity andpolitical influence are best positioned totake advantage of market opening. Undersome conditions this can lead to marketbacklash. Chua (2004) documents cases of“market-dominant minorities” who benefitfrom free market reforms, including tradeliberalization. Traditionally dominant eth-nic minorities, such as the Chinese inSoutheast Asia, the Lebanese in West


Africa, and whites in Latin America andSouth Africa, seem to be the primary bene-ficiaries of the marketization of theireconomies. Such outcomes can fuel deep-seated resentments and lead to violence.These political economy and reform back-lash considerations are additional reasonsfor integrating attention to equity into thedesign of product market reforms andtrade liberalization.

Trade liberalizationTrade liberalization changes relative pricesin an economy, causing shifts in output,wages, and employment. Analyses of tradeliberalization are primarily about out-comes, providing only indirect evidence onopportunities. They show that trade open-ness is positively associated with growthand, on average, there are no strong corre-lations with income distribution. Morley(2001), using data from Latin America,found slightly negative effects of trade lib-eralization on income distribution, whileBehrman, Birdsall, and Székely (2003)found positive influences of trade liberal-ization on wage inequality. Another studyusing panel data for 41 countries foundthat trade openness is associated withincreases in inequality, after controlling fora set of other structural and policy influ-ences.43

These average effects mask a great deal ofdiversity in impacts across groups, espe-cially over the short to medium term. Theimpact of trade-induced price changesdepends not only on average pass-throughbut also on exactly which prices change andhow producers and consumers respond.44

For example, the effects of removing pro-tection for agriculture will depend onwhether agricultural prices subsequentlyrise or fall and whether the poor are netproducers or consumers of disprotectedproducts. Normally, it is assumed that tradeliberalization in agriculture will benefitpoor small-scale farmers and be good forequity. After all, “developing countries havetraditionally taxed the agricultural sectorwhile developed countries have protectedit.”45 But the impacts must be analyzed caseby case at the microlevel.

Cereal protection in Morocco offers anillustration. In a simulation analysis ofremoval of wheat tariffs, Ravallion (2004b)finds that, contrary to expectations, ruralfamilies would tend to lose while urbanones would gain. Although the results pre-dict that there would be more gainers thanlosers among the rural poor, aggregatelosses outweigh aggregate gains. Further-more, expected effects would be enor-mously varied, with significant horizontalinequalities: households with the sameincomes were predicted to experiencewidely differing outcomes, depending ontheir specific structure of production andconsumption. Using a similar simulationanalysis, China’s accession to the WorldTrade Organization (WTO) was found tohave a small aggregate poverty-reducingeffect, but this masked considerable varia-tion in impacts across households in ruralversus urban areas and across differentregions.46

While aggregate effects of trade reformon poverty and equity are not alwaysclear—whether diverse impacts translateinto inequalities in opportunities dependson how new activities open up, and whetherlabor can move into them—we do knowthat there will be winners and losers. Out-comes depend on the ability and willing-ness of governments to mitigate losses toparticularly hard-hit sectors, possibly byredistributing some of the gains accruing towinners.

Domestic product markets and equityLack of competition among traders, remotegeography, poor infrastructure, and hightransport costs can all prevent the transmis-sion of border price changes to intendedreform beneficiaries. Addressing such prob-lems can improve the impact of tradereforms on equity.

The case of public or private marketingagencies for export crops is a typical exam-ple. Small farmers in many countries tradi-tionally have had no option but to sell to amarketing agency at prices substantiallylower than the free on board (f.o.b.) exportprice. Legitimate transport and marketingcosts account for some of the price differ-


ence, but monopsonistic profits often do aswell. Marketing agencies can thus preventtrade-induced price changes from reachingfarmers at all.47 A 1998 study found that theVietnamese rice marketing system wascontrolled by a small number of state enter-prises. These enterprises limited the trans-mission of border price changes to farmersand pushed up transaction costs.48 A moreextreme example was when marketingboards became instruments for extractionof surplus from agricultural exporters, suchas for cocoa farmers in the post-independ-ence period in Ghana. Malawi has facedsimilar issues: cartels by companies buyingtea, sugar, and tobacco have forced downthe returns to farmers.49

Abolishing marketing boards does notguarantee efficient marketing, however,they may play a useful role when marketsare thin, or a tradition of trading has notdeveloped, as in the raw cashew exportmarket in Mozambique. Although the statetrading company was privatized in the late1980s, there was insufficient competitionamong private marketers when raw cashewexport restrictions were lifted in the early1990s. Indeed, raw cashews had to movethrough three tiers of intermediaries withnear monopsony power before reachingworld markets. As a result, the trading mar-gin from the farm to the factory was 50 per-cent and the expected liberalization priceincreases never reached farmers.50

Trade reforms in Mexico show how infra-structure and transportation costs can shapeopportunities through their impact on pricetransmission. Mexico’s trade liberalization inthe 1980s and entry into North AmericanFree Trade Agreement (NAFTA) in 1994appear to have led to wage increases in statesbordering the United States relative to the restof the country.51 A different study on Mexicofound that tariff reductions translate intodomestic price reductions less and less as dis-tance from the main port of entry increases.This effect can be substantial (figure 9.4).Studies of Rwanda and Indonesia have alsodocumented the isolation of remote house-holds from border price changes.52

Even when trade liberalization is intendedto be pro-poor, incomplete transmission of

border price changes for both exports andimports means that benefits are not evenlydistributed. In other words, ease of access tointernational markets matters. Unfortu-nately, those who could benefit most fromfavorable price changes—the poor in remoterural areas—are those least likely to beaffected by them.

Many of the factors responsible for trans-mitting trade-related price changes are alsoimportant to the functioning of domesticproduct markets. For instance, competitionamong traders, infrastructure quality, andtransportation costs all influence how prof-its are allocated across a product’s life cycleor value chain. And the percentage of agood’s final price that goes to the primaryproducer, intermediary producer, distribu-tor, and retailer can vary tremendously.

The cashmere sector in Mongolia showshow product market reforms could be bothequity- and efficiency-enhancing. If properlydeveloped, cashmere could be a pillar ofMongolia’s successful transition from a com-mand to a market economy. It is the country’ssingle largest employer and a principal sourceof livelihood for the poor. It provided jobs formore than 16 percent of the workforce andaccounted for more than 6.3 percent of GDP


Changes inhousehold welfare

> 5 percent4–5 percent2–4 percent0–2 percent

Figure 9.4 It’s better for household welfare to be close to economic opportunities

Source: Nicita (2004).Note: Welfare changes were calculated from the effects of trade liberalization–related price changes that affectedboth the purchasing power and incomes of households.

during 1993–2002. But unsatisfactory publicsector policies have meant that the industryhas not lived up to its potential.

One significant bottleneck is in market-ing and distribution. Mongolia’s major cash-mere market is in Ulaanbataar, 600–1,000km from most regional production centers.Herders generally have to sell to traders atthe farmgate or at informal provincial mar-ketplaces at discounts of 10 to 45 percentfrom prices in Ulaanbataar. With littleknowledge of market demand, individualherders can incur costs traveling to marketswith no certainty of a sale. Policies thatencourage regional market centers andherder cooperatives, as well as infrastructureimprovements, could reduce marketingcosts and increase herder’s margins.53

Soybean farming in India offers anotherillustration of how product marketingchannels can improve equity. Ninety per-cent of the soybean crop is sold by smallfarmers to traders, who act as purchasingagents for buyers at a local, government-mandated marketplace, called a mandi.Farmers have only general informationabout price trends and no choice but toaccept prices offered to them by traders orthe auction price on the day they bring theirgoods to the mandi. As a result, traders canexploit farmers and buyers using practicesthat create systemwide inefficiencies.

Under the e-Choupal Initiative, ITC, oneof India’s leading private companies, placedcomputers with Internet access in ruralfarming villages. Each computer—placed ina farmer’s home and serving about 10villages—becomes a social gathering placefor the exchange of information and an e-commerce hub. Farmers can use the com-puters to check prices, learn about farmingtechniques, purchase inputs, and sell theirsoybean crops at the previous day’s marketprice. Farmers then have to transport theircrop to an ITC processing center, where it iselectronically weighed and the farmer isimmediately paid. Farmers selling to ITCthrough an e-Choupal receive, on average,2.5 percent more for their crops, and ITCsaves an additional 2.5 percent on procure-ment costs by cutting traders out of the loop.Farmers also benefit from accurate weighing,

prompt payment, and information aboutprices and price trends that allows them tooptimize their sales decisions. The e-Choupal system continues to grow rapidly,reaching more than 3.1 million farmers bylate 2004.54 The initiative illustrates howimprovements in technology and communi-cations infrastructure can be good for bothequity and efficiency in product markets.

In addition to better marketing channelsand new technology, there are other ways toimprove product market competition thatcan be good for equity. Measures that facili-tate the entry of new firms often mean thatsmall and medium enterprises benefit at theexpense of large, politically connectedincumbents. Product market competitioncan also drive consumer prices down andmake goods more affordable for the poor.Of course, measures to improve competi-tion benefit efficiency and growth as well,improving the welfare of the poor.

Licensing restrictions—even when de-signed in the name of equity—are one wayto hamper competition. India reserves theproduction of more than 600 manufacturedproducts, including apparel and textiles, tosmall companies. This licensing regimecould cost the country jobs by preventingsmall producers from growing and compet-ing with larger manufacturers in, for exam-ple, China. High regulatory, administrative,and fiscal burdens can also harm productmarkets by keeping firms in the informalsector. Informal firms face a number of con-straints, including limited access to financ-ing, which tend to leave them significantlyless productive than their formal sectorcounterparts. For example, informal Turkishbrake manufacturers achieve only 22 percentof U.S. productivity, while their formal sectorcompetitors achieve 89 percent. Affordableaccess to titled land and reliable infrastruc-ture (chapter 8) can also enhance firm andproduct market competitiveness.55

As discussed, improving transportationand logistical infrastructure can reduce thecost of moving goods. Better transportationlinks with other regions can also provideinsurance against regional price fluctuations.For example, if there is a drought or foodshortage in one area, efficient regional con-


nections would allow consumers to importreasonably priced food from other parts ofthe country. Finally, better transport andlogistics systems reduce inventory costs bymaking the timing of delivery more reliable,again benefiting producers and consumers.56

Interactions between product and labor marketsChanges in product markets, whetherinduced by internal developments or exter-nal trade-related changes, can have power-ful influences on the opportunities facingworkers. Standard trade theory predictsthat countries should export products thatuse relatively abundant factors intensively.Labor-abundant countries that open upshould see relative gains in unskilled wages,as indeed occurred among the East Asiantigers in the 1960s and 1970s.57

The Latin American experience stands insharp contrast. Many countries, includingArgentina, Chile, Colombia, Costa Rica, Mex-ico, and Uruguay, saw wider wage differen-tials with increasing trade openness duringthe late 1980s and 1990s. Some argue that thiswas due to the massive insertion of low-income Asian countries into global markets.Others interpret the evidence as supportinggeneralized skill-biased technical change, inwhich trade opening facilitated processes ofrestructuring, including the destruction ofjobs in inefficient industries, and risingdemand and relative wages for skilled work-ers.58 Whatever the reason, the question iswhether this was a source of rising inequalityin opportunity. Over the short to mediumterm, this was almost certainly the case,because unskilled workers cannot increaseskills quickly. Over the longer term, risingwage differentials provide incentives forinvestment in education, if education systemsprovide equal opportunities (chapter 7).

Effects of economic restructuring onworkers also depend on the extent of labormobility. One study from India shows thatthe effects of trade liberalization in the early1990s on poverty varied by state, dependingon the flexibility of labor laws. In states withless flexible laws, where liberalization did notproduce any measurable effect on the alloca-tion of labor across sectors, the adverse

impact of trade opening on poverty was feltthe most. In states with more flexible laborlaws, movements of labor across sectorseased the shock of relative price changes.59

While greater mobility is desirable, thedesign of measures that increase flexibilityneeds to be balanced with the levels ofworker protection that are appropriate forthe institutional setting (see the earlier dis-cussion on labor markets).

Safety nets and opportunitySafety nets complement product marketdeepening and are often an essential ele-ment of a strategy to ensure that marketexpansion leads to more equal opportuni-ties. General questions of the design ofsafety nets were discussed in chapter 7; herewe highlight the links with product marketchange. As discussed above, trade openingcreates winners and losers. How this affectsequity depends partly on how governmentscan offer support to the losers.

Rodrik (1998) finds that openness isassociated with higher government spend-ing. The argument is that open economiesare more subject to external shocks andspend more on social insurance to mitigateexternal risk. In more advanced economieswith the capacity to manage social welfaresystems, exposure to external risk is stronglycorrelated with spending on social securityand welfare. In less-developed economies,governments rely on a broader set of tools,such as public employment, to reduce risk.

The specific design of safety nets canexpand opportunities to those who sufferadverse effects. For instance, trade adjust-ment assistance programs in the UnitedStates extended unemployment benefits,training, and relocation subsidies to displacedworkers. While the United States was offeringthe programs in response to the NAFTA, theMexican government established Procampo, acash transfer program for grain farmers toease the pain of NAFTA-induced competitionfrom the United States. It was designed toprovide consumption support to compensatefor price declines and to allow farmers todiversify into other activities. While the size ofthe transfers was hurt by the 1995 TequilaCrisis, there is evidence of gains to farmers


and of the use of proceeds for investmentpurposes.60

While the ideal policy mix may be one thatcombines reduced barriers and extensivesafety nets, in practice, this is not always feasi-ble. Many countries phase liberalization toseek to ensure that processes of job creationprecede or accompany job destruction—thishas been a central feature of the East Asianexperience (see discussion of China in chap-ter 6). This carries risks of slowing restructur-ing and extending protection beyond periodsjustified by equity concerns because of cap-ture by influential beneficiaries.

Credibility, political supportability,and the design of product market reform While technocrats can design trade andother product market reforms that appearto be good for growth and equity, theexpected gains will never materialize ifthere is not enough political support.Because of the nature of trade policy—aconcentrated set of winners from trade bar-riers versus a diffuse set of winners fromliberalization (consumers in general)—it iseasy for vested interests to capture policy.Steel tariffs in the United States and agricul-tural subsidies in the United States, Japan,and Europe are obvious examples.

Even after trade liberalization laws havebeen passed, they are still not immune tocapture. If economic actors do not believethat reforms are credible—that is, thatpoliticians will recant at the behest of vestedinterests—the anticipated adjustments willnever take place. Cashews in Mozambiqueagain provide an apt example. In the early1990s, the Mozambican government (work-ing with the World Bank) implemented anew pricing regime that liberalized theexport of raw cashews. But there was nocredible political commitment to the newpricing regime, so neither cashew farmersnor cashew processors adjusted to the newprice signals. Efficiency gains from the real-location of resources never materialized.61

While this section has emphasized the het-erogeneity of effects of product marketchanges, the policy message is not necessarilyone of detailed fine-tuning of reforms. That

brings risks of greater capture. The idealbalance is a combination of gradual but com-mitted liberalization with extensive engage-ment in the complementary measures thatbroaden opportunities for all: education,infrastructure, competition, and safety nets.Societal debate and information can ensurethat governments remain accountable to allgroups, not just those with access and connec-tions. Of particular importance for externalopening is the role of external commitments.Entry into international agreements, such asthe WTO, the European Union, or NAFTA,can effectively lock politicians into tradereforms. When trade regulations are boundby international agreements, reform commit-ments are more credible and less susceptibleto capture by domestic special interests(asymmetries in power among the interna-tional parties to such agreements remain, aswe will read in chapter 10).

Macroeconomic managementand equityMacroeconomic instability is both acause and consequence of inequityMacroeconomic stability is a public good andmight be expected to equally affect all. Thereis a well-established association betweenmacroeconomic stability and long-termgrowth, and growth typically brings expan-sion in opportunities to everyone. But thefact that stability is a public good does notmean that the incidence of benefits is equal.As discussed in chapter 4, the distribution ofincome gains from economic growth is typ-ically as unequal as the initial income distri-butions. Moreover, macroeconomic insta-bility, whether in the form of volatility orhigh inflation, can have differential andpotentially inequitable effects, because thepattern of power and wealth can influencethe distribution of losses—and differentgroups have differential capacities to man-age the consequent shocks.

As in many other areas, there are two-way patterns of causation between macro-economic conditions and equity. Unequalpatterns of power and associated institu-tional structures are at the center ofcausative influences from inequity to insta-


bility and in regressive effects of crises. Byemphasizing these links, we are not arguingagainst the large body of literature on eco-nomic causes of crises. Depending on thetype of crisis, this literature sees the causesas fiscal imbalances, herdlike movements ofinvestors behind exchange rate crises, andinteractions among external liabilities,exchange rates, and financial-corporateconditions, especially under “crony capital-ism.”62 Some of the processes in this litera-ture complement the diagnosis here; othersare manifestations of underlying distribu-tional and institutional conditions.

Figure 9.5 shows the bivariate correla-tion between macroeconomic volatility anda measure of “constraints on the executive”branch of government, which would beexpected to be closely linked to restraints onelite power. Weaker constraints are associ-ated with greater volatility (and a higherpropensity for macroeconomic crises).

The correlation says nothing about cau-sation. But there is evidence supporting theview that “weak and unequal” institutionshave a causative influence on economicinstability. A tradition of work interpretsinstability as a consequence of distributionalstruggles that are ineffectively managed byinstitutions.63 As discussed in chapter 6, theseminal work by Bates (1981) on Ghanainterprets exchange rate overvaluation andinternal pricing policies as mechanisms forgovernments to severely tax cocoa farmersin the early postindependence period toprovide resources to buy off urban groups. Acombination of predatory governments andweak or absent balancing institutions cre-ated the preconditions for fights over rentsand systematic political instability until theearly 1980s. Analyses of hyperinflation, insettings as diverse as Bolivia and Israel,interpret macroeconomic instability as aconsequence of failures to manage societalconflicts.64

Macroeconomic instability caninteract with unequal influence inthe fallout from crisesCrises, whatever the causes, are systemati-cally bad for growth, more so in the pres-ence of distributional struggles. Rodrik

(1999a) argues in a cross-country empiricalanalysis that the effects of external shocks inthe 1970s were significantly worse for sub-sequent growth in societies in which latentdistributional conflicts (proxied by incomeinequality or ethnolinguistic fragmenta-tion) were more severe and conflict-management mechanisms (proxied byinstitutional strength and indicators ofdemocracy) were weaker.

High inflation and macroeconomiccrises can be particularly harmful to thepoor, who are least equipped to manageadverse shocks. For the impacts on distribu-tional outcomes, household survey evidencedoes not display systematic disequalizing orequalizing biases across countries: the Mexi-can 1994–95 crisis was slightly equalizing(although strongly poverty-increasing); the2001 Argentine crisis was disequalizing.There is some evidence that high inflation isworse for poorer groups, for example, in thePhilippines65 and Brazil.66

Mechanisms for crisis resolution tend tobe inequitable as a result of unequal influ-ence. Many of the results are not fullyreflected in household income and spendingsurveys. The reason is that the big actionusually takes place elsewhere, notably inchanges in capital income and fiscal posi-tions that the surveys typically fail to capture.


Constraints on the executive index80 2 4 6

Standard deviation of GDP growth20

15

10

5

0

Figure 9.5 Weaker institutions are associated withmacroeconomic volatility and crises

Source: Authors’ calculations, using World Bank statistics and thePolity IV database for constraints on the executive index.Note: A higher value for the constraints on the executive indexdenotes greater accountability.

There is evidence that crises lead to reduc-tions in measured labor share (strongly influ-enced by formal sector earnings andemployment). Diwan (2001) finds thatlabor shares systematically fall during crisesand don’t fully recover afterward, a cross-country result illustrated for Mexico andPeru in figure 9.6.67 The flip side of this pat-tern is that the shares of corporate andfinancial sector capital income rise relativeto wages. There are also significant interac-tions with structural variables. In particu-lar, closed trade, capital controls, and fiscaldeficits are associated with higher laborshares in normal times, but with larger fallsin labor shares when crises occur. Crises aremechanisms for the resolution of distribu-tional conflicts that are not tackled duringgood economic times. Labor is relativelyimmobile and so typically bears a higherproportion of the cost. Pre-crisis laborshares may, in some cases, have been toohigh for competitiveness and stability, butthe point is that crises are a high-cost formof conflict resolution. And the interactionbetween shocks and weak conflict resolu-tion mechanisms is associated with weakerlong-run growth.68

In addition to any effects through thedistribution of labor and capital income,important mechanisms work through thefinancial sector and associated fiscal action.

Major crises lead to large financial losses,typically financed by both explicit andimplicit fiscal outlays. Case study evidenceindicates that these are highly regressive,through gainers and losers from capitalflights, transfers from those outside towithin the financial system, and the pat-terns of bailout among financial sector par-ticipants.

The fiscal costs of crises are large (table9.3). For example, the post–Tequila CrisisMexican bailout is estimated at $112 bil-lion, with a large additional amount spenttrying to prevent crises through liquiditysupport, sovereign bond swaps, and thefinancing of large investors who withdrawmoney from projects.69 Halac and Schmuk-ler (2003) use the $23 billion decline in thecentral bank’s reserves between Februaryand December 1994 as a proxy, calculating atotal fiscal and quasi-fiscal cost of the crisisof $135 billion. This represents about one-quarter of Mexico’s GDP in 2000 and somefour times the $33 billion in capital receiptsfrom privatization during the 1990s.

What is the pattern of gainers and losers?Some wealthy individuals undoubtedly losetheir shirts. But there are strong tendenciesfor the poor to lose, sometimes to lose a lot.First, the wealthy with information andaccess to international banking systems gettheir money out first. And they may actuallyexperience capital gains when domestic assetprices tumble and the exchange rate goesagainst the currency. In Argentina, the ratioof foreign assets to domestic GDP rose fromabout a quarter to more than 90 percentbetween 2001 and early 2002, because of acombination of capital flight and currencydepreciation (figure 9.7).

Second, the recipients of fiscal bailoutsare those within the financial system—depositors, creditors, and equity owners,who are systematically better off than thoseoutside. (There are of course small middle-income depositors but, as noted below, it ispossible to protect them without providingblanket protection for all.) Case study evi-dence finds biases toward wealthy and moreinfluential individuals and groups withinfinancial systems. Owners of large depositsenjoyed the greatest compensation (and


Table 9.3 Fiscal costs of selectedbanking crises

Country Fiscal cost and episode (percent of GDP)

Argentina, 1980–82 55.1

Brazil, 1994–96 13.2

Chile, 1981–83 41.2

Ecuador 1996– 13.0

México, 1994– 19.3

Venezuela, 1994–97 22.0

Korea, Rep. of, 1997– 26.5

Indonesia, 1997– 50.0

United States, 1981–91 3.2

Source: Honohan and Klingebiel (2000). Note: Costs refer to both fiscal and quasi-fiscaloutlays and the present value of the futurestream of costs. Banking crises in Ecuador,Mexico, the Republic of Korea, and Indonesiawere ongoing at the time of the study.

1973

1970

1976

1979

1982

1985

1988

1991

1994

1997

1973

1970

1976

1979

1982

1985

1988

1991

1994

1997

MexicoLabor share

45

40

35

30

25

20

50

15

PeruLabor share

45

40

35

30

25

20

50

15

Crisis year Crisis year

Figure 9.6 Labor shares fall during crises and don’t fully recover afterward

Source: Authors’ calculations, based on national accounts.Note: Crisis years are defined as years in which at least two out of three of the following occur: a 25 percent (orhigher) nominal devaluation, negative growth, and 50 percent (or higher) rate of inflation.

sometimes capital gains) in crises inArgentina, Ecuador, and Uruguay, oftenthrough getting their money out of thecountry, while small depositors sufferedcapital losses. In addition, there is evidencethat large borrowers with close connectionsto banks were especially favored in crises inChile, Ecuador, and Mexico.70

Crisis costs are paid for by some combi-nation of higher taxes and lower spending.Who pays depends on the marginal patternof taxation and spending. As a first approxi-mation on the tax side, many developing-country tax systems are roughly propor-tional (everyone pays the same proportionof their income, primarily through indirecttaxation). On the spending side, work onLatin America and Asia in the 1990s findsthat marginal spending was generally pro-gressive, as the expansion of social andinfrastructure programs “crowded in”poorer groups.71 Thus, the forgone spend-ing hurt poorer groups the most. The neteffect is a regressive workout financedlargely by regressive fiscal adjustment.

While case study evidence indicates astrong pattern of regressive consequences of

crises, this is contingent on initial struc-tures, patterns of influence, and policyspecifics. For example, the Russian crisis,while undoubtedly costly in social terms,may have led to a surprisingly positive shiftto more equitable structures of resourcemanagement (box 9.6).

Policy directions need to take accountof policy design and accountabilitystructuresMacroeconomic instability is thus bothproduct and cause of underlying inequali-ties and associated weak institutions. Thecosts are large to equity and growth. Whatcan be done? As in other areas, it helps toanswer this question in terms of the com-


Dec.2000

20.0 22.4 21.024.4

28.6

110.0

107.3

102.2

97.295.494.4

94.4

Mar.2001

Jun.2001

Sep.2001

Dec.2001

Mar.2002

Net private foreign assets (US$ billions)Net private foreign assets (percent of GDP)

Capital flight = $12.9 billio

n

Figure 9.7 In Argentina, the wealthy had a way outduring the crisis

Source: Ministry of the Economy, Argentina.Note: The sharp increase between December 2001 and March2002 was because of the exchange rate devaluation.

The Russian crisis was at first glance typical ofcrises of the 1990s—driven by interactionsbetween private capital movements anddomestic institutional structures that encour-aged moral hazard,and with large adverseeffects for welfare.The social costs weresevere,with a fall in GDP of 5 percent in 1998.Between 1996 and 1998 household per capitaexpenditures fell 25 percent,expenditurepoverty rose from 22 to 33 percent and gov-ernment transfers fell by 18 percent,albeitwith better targeting (Lokshin and Ravallion2000).There was also major capital flight,withmany of the rich getting their assets out fast,leaving the rest of the society to share thecosts.This fact alone is a sign of great inequity:the rich had more opportunity to protect theirassets by shifting money out of the country.This (individually rational) behavior imposedcosts on less wealthy groups.

But the transition appears to have hadadditional effects that were, at least to somedegree, more equitable. Russia before thecrisis was in a particularly inefficient andinequitable equilibrium, in which firms didnot pay their energy or tax bills, and theenergy sector didn’t pay its taxes. Firmswere not allowed to go under for fear ofemployment effects, but this was a highlyineffective safety net. Subsidies were esti-mated at 15 to 20 percent of GDP in thethree years before the crisis, fueling corrup-tion and delaying enterprise restructuring.There was extensive asset-stripping at thecost of the broader society.

The crisis triggered major changes ineconomic relationships.The strategy of“devalue and default” led to large relativeprice movements and cut Russia off frominternational capital markets.The cutoff,combined with the authorities’ recognitionof the unpopularity of hyperinflation, finallyforced hard budget constraints on the sys-tem, which had powerful ripple effects ininducing the demise of the nonpaymentssystem, making economic transactionsmore transparent and laying the basis for arecovery in taxes. Real exchange rate depre-ciation made many firms competitive again,and this fed through into employment. Fur-thermore, the big Moscow banks wereallowed to fail (with total costs, largelyincurred before the meltdown, a relativelylow 2 percent of GDP). And the defaultmeant that external holders of Russianpaper took an immediate loss, effectivelyallowing a degree of international burden-sharing.While there were undoubtedly rep-utational costs, there may have been anadvantage in taking this upfront, ratherthan after protracted negotiations. Overall,while a serious analysis of impacts on equal-ity of opportunity is not feasible, the effec-tive shift in regime from a system in whichinfluence played a dominant role inresource allocation to one of hard budgetconstraints and greater transparency wasprobably good for efficiency and equity.

Source: Pinto and others (forthcoming).

B O X 9 . 6 Did the Russian 1998 crisis have equitableconsequences?

plementary role of specific policy designand the deepening of accountability struc-tures and societal mechanisms to manageconflict. The Israeli hyperinflation providesan illustration. Its resolution involved botha full set of financial and macroeconomicpolicies and an intensive process of societalinteraction to manage underlying conflictsbetween organized labor, the corporate sec-tor, and other interest groups.72

There is an extensive body of literatureon design specifics. Thus, we conclude withsome comments on the principles of macro-economic management that emerge from afocus on equity. Some of the principles arefairly familiar, especially the need to buildstronger regulatory and supervisory struc-tures for the financial system and compre-hensive insurance mechanisms while out ofcrisis. Once a crisis hits, it is particularly dif-ficult to design and implement such meas-ures and politically difficult to implementmore equitable outcomes. By contrast exante insurance design—whether for deposi-tors, bankruptcy, or unemployment—ismore likely to be broad based and, if alreadyin place, reduces the case for ex post dealstailored to the influential.

Less obvious is a heightened emphasison fiscal prudence. In public debates,adopting a less stringent macroeconomicstance is often portrayed as a distribution-ally progressive approach, whether in goodtimes or bad. While there will always bespecific judgments about the distribu-tional impacts of a range of fiscal andmonetary policy options, the analysis heresuggests that taking a “superprudent” posi-tion over the course of the cycle providesgreater hope for supporting a more equaldevelopment pattern. On one level, thissharply reinforces the common prescrip-tion to break procyclical policy positions.Macroeconomic restraint in good timeswill facilitate automatic stabilizers and asensible easing of policies to be applied ina disciplined fashion when adverse shocksoccur. Thus, the priority is to build fiscalrules and institutions that help overcomethe political pressures to deplete potentialsurpluses in good times, as well as infor-mational asymmetry problems. These insi-

tutions should also improve the credibilityof countercyclical fiscal policies duringdownturns.73 This strategy would, in par-ticular, provide the macroeconomic foun-dation for broad-based, self-expandingsafety nets.

Finally, there is a case for different formsof accountability. Over the long term, themost effective approach is to move to newfiscal and social contracts built on deeperaccountability structures—to a better polit-ical equilibrium.74 There is then a need foran appropriate mix of stronger regulatoryaccountability shielded from short-runpolitical pressures from all sides—moreindependent central banks and strongerfinancial sector supervision—and greatertransparency and debate about the overalldesign of macropolicy and the incidence ofworkouts. There are important interactionswith external actors and rules for crisismanagement that will be taken up in chap-ter 10.

In sum, we have explored policies that,by leveling the playing field in the marketsfor capital, labor, and goods and by manag-ing the macroeconomy, can lead to greaterequity and prosperity. Financial marketsare typically biased toward incumbents,reflecting the historical political influenceof the powerful. Yet rapid and ill-designedliberalizations can lead to further concen-tration of influence. Greater societal con-trols are needed as well as a more measuredtackling of barriers—especially to smalland medium firms—backed by regulatorystructures and more information to reducethe power of connections.

Labor market outcomes may reflect theweak bargaining position of workers, butlabor policies often lead to patterns of jobprotection that create economic rigiditiesand help those in good jobs, to the detri-ment of those in the informal economy.Support for unions and security for workersare important objectives, but designs needto be adapted to economic conditions inways that reach poorer, informal workersand minimize impediments to economicrestructuring.

Both the design of external trade policyand the workings of internal product mar-


kets reflect patterns of influence. Remov-ing biases and ensuring access to all needto be complemented by measures to expandskills, infrastructure, and safety nets toachieve genuine access and to manage losses(especially horizontal inequities). Impru-dent macroeconomic policy is typicallyinequitable: high inflation hurts those leastcapable of managing its consequences, and

financial crises are particularly pernicious,because the powerful can benefit or bebailed out at the expense of the rest of soci-ety. Prudent macroeconomic management,backed by strong countercyclical policy andindependence in policy design is an ally,not a foe, of greater equity. We now turn topolicies that can help level the global play-ing field.


that have weaker fiscal capacity, such as inArgentina’s experience with decentralizingreforms in education.5 In poor regions whereregional elites have particularly concentratedpower, decentralization may also deepenboth intra- and inter-regional inequalities.6

Trends in inter-regional inequality havevaried considerably across countries. TheUnited States has experienced convergenceand lower interregional income disparities.Indonesia shows convergence of provincialincomes since the 1970s. Brazil has seendivergence over many decades, but recentlyhas shown convergence. Evidence on Indiaalso suggests divergence. China’s pattern ofgrowth has reduced gaps in the 1970s and1980s, which widened in the 1990s. And inMexico a long-run trend of slow convergencein incomes shifted to one of slow divergenceafter an opening that started in the late 1980s.

Characteristics of lagging regionsThe reasons for regions to lag varies, and wepresent a simple taxonomy.

Low poverty density, low market access.These regions are sparsely populated,remote, and face particular geographic chal-lenges. Distance and poor resource endow-ment—often with weak social indicators,generally poor infrastructure, and weakregional voice—place these regions at theperiphery of national economic activity andopportunity. Supporting development ofthese regions may be desirable on povertygrounds, but it is likely to be expensive.

Low poverty density, high market access.These regions typically have been boomingat one point in history, and were well inte-

The average income in Brazil’s north-east is less than half the nationalaverage. Poverty rates are far higher

than the national average in India’s denselypopulated states of Bihar, Uttar Pradesh, andOrissa, and the regional income gap appearsto be widening. In 1990, children in thenorthwest region of Nigeria were four timesless likely to receive any immunizations and50 percent more likely to die by age five thanthose near the capital in the southwest; by1999 they were five times less likely to receiveany immunizations and 85 percent morelikely to die before age five. Chronic regionalunderperformance can give rise to manyconcerns and threaten national unity—losteconomic potential, unfairness in regionalopportunities, potential instability, loss ofsocial cohesion, and adverse social conse-quences, including higher crime and disease.

The geographic and historical factorsunderlying interregional inequality arecomplex and overlapping. Weak resourceendowments and distance from marketscan constrain development in laggingregions. In many cases, economic differ-ences are linked with long-standing,unequal relations of power between advan-taged and lagging regions, and institutionalweaknesses within the latter.1 When actorsin advantaged regions control the assets,decision-making and policy formationprocesses, and the terms of the policydebates on which lagging regions depend,regional “catch up” is much more difficult.2

When historically disadvantaged ethnic,racial, and social groups are concentrated inparticular regions, group-based inequitiesbecome reflected in regional inequalities.This is the case in parts of Latin America,where indigenous groups are both poorerand concentrated in poorer regions,3 and inVietnam and in India where tribal groups(adivasis) are spatially concentrated.4

In the absence of redistributive fiscaltransfers, recent reforms in many countriestoward greater decentralization may aggra-vate regional disparities. The positive effectsof decentralization may be lost in regions

grated with the national economy. Butchanging demand patterns or resourceexhaustion became sources of decline, eventhough political influence may have per-sisted. For such “rustbelt” regions, there is acase for public support for movement ofpeople and resources out of decliningindustries, backed by social safety nets forthe affected workforce.

High poverty density, high or low marketaccess. These regions are most often consid-ered for targeted interventions: poverty isconcentrated in them, population density isrelatively high, and the lack of market inte-gration is due to history rather than geogra-phy. Possible culprits include weak gover-nance, poor institutional capacity andhuman capital, a history of socioculturalconflict and domination, a poor investmentclimate, and security problems. Suchregions are often home to socially, racially,and ethnically disadvantaged groups.Where such groups are dispersed or patron-client relationships dominate, the challengeof fostering organization, agency, and polit-ical influence is especially great.7

Regional development policiesand tradeoffsRegional development policies involveinterventions to facilitate inward invest-ment, enhance income opportunities andwell-being in lagging regions, help house-holds move to opportunities elsewhere, andshift interregional power relations. Policiesare context specific and involve tradeoffs. Iflagging regional performance reflects geo-graphic disadvantages or an absence of

The role of public policy in addressing spatial inequalitiesThe persistence of regional disparities within countries is a major policy concern confronting many governments in richand poor countries alike. Clarity on the causal factors of weak regional performance and careful consideration ofpotential tradeoffs are needed to guide policy choice over regional interventions.

regional inequalityf o c u s 6 o n

• Chile’s “zonas extremas”• Russian North (state-sponsored settlements)• Northern Canada

LowPoverty density(poor or disadvantaged

people per squarekilometer)

Low

Market access (population density, transport costs)

High

High

• N.E. China’s “rustbelt” region• Developed country “coal towns” (France, U.K., U.S.)

• Thailand’s northeast• Mexico’s southern states

• India’s “Hindu Belt” poor and populous states• Italy’s southern Mezzogiorno

Focus on regional inequality 205

to sell to lagging regions. This is one factorthat has slowed development of the rela-tively poor Mezzogiorno region of southernItaly, despite large investments in nationalnorth-south infrastructure that has reducedtransport costs.9

Facilitating labor mobility Facilitating voluntary labor movement tohigher opportunity areas is another strat-egy. In contrast to fiscal incentives and pub-lic investments that focus on bringing jobsto poor areas, this strategy focuses onbringing poor people to areas with morepotential. Relocation assistance can includetransport, housing, training, resettlementallowances, and portable safety nets. Exam-ples included incentives in Russia for fami-lies to relocate from their northern settle-ments—developed at huge state expense forresource extraction and security purposesduring the Cold War—and incentives tosupport movement of labor out of declin-ing industries, such as the moribund coalsectors in Western Europe and the formerSoviet Union since the 1960s. While theprograms have helped ease the impact ofunemployment, there are questions aboutcost effectiveness and long-term impact.

There is also a long history of efforts todirect settlement of remote regions orencourage migration to frontier lands.Early settlers to the Americas, including thewest and midwestern regions of the UnitedStates, were beneficiaries of legal landgrants to clear and use new land. Morerecent programs include Indonesia’s trans-migration program that shifted Javanese tosparsely populated outer islands in the1970s and 1980s, or early Ethiopian reset-tlement programs to fertile areas in thesouth and southwest regions of the country.However, these and other resettlement pro-grams have been criticized for their coerciveor ethnic dimensions, raising questions

agglomeration and scale economies, publicinterventions may be particularly expen-sive. But when public policy is designed tocorrect market failures (such as underde-veloped insurance or credit markets),address specific social or historical factorshandicapping regional performance, orcapture externalities intrinsic to nationalwelfare (cultural, environmental, security),there may be few or no efficiency tradeoffs.

Fiscal incentivesA popular approach involves fiscal incen-tives to induce industry to locate and investin lagging regions: tax advantages, insuranceor risk-sharing arrangements, direct subsi-dies, or indirect subsidies through provisionof low-cost public services. But evaluationsof fiscal incentives generally indicate thatthey can be costly and ineffective. Brazil’sefforts to develop the manufacturing centerof Manaus in the north have been a successby some measures, but costs per job createdare high.8 Interregional “fiscal wars” can alsooccur as regions compete to attract busi-nesses. If uncoordinated or unconstrained,these can have adverse consequences forlocal tax bases and public services in com-peting jurisdictions. Compared to the alter-natives listed below, this tends to be a high-distortion strategy.

Public investmentTargeted public investment, particularly incore infrastructure, is another policyresponse aimed at reducing geographic dis-incentives to firm location, whether forexisting or new firms. China has followedthis strategy, first in the coastal special eco-nomic zones, and now in western regions(see box below).

Investment in regional infrastructurelinks may enhance productivity of existingfirms and attract new firms. But, it alsoallows more efficient firms in richer regions

about the adverse impact on indigenouspopulation groups and settlers.

Enhancing agencyWhere intergroup inequalities in agencyunderlie regional disadvantage, nationaland regional policies addressing discrimi-nation, racism, and citizenship deficits canbe important instruments for dealing withspatial inequality. Enhancing voice and par-ticipation of excluded groups is also impor-tant for national peace and cohesion. Whileethnic discrimination and regional disad-vantage do not necessarily lead to conflict,researchers and truth and reconciliationcommissions alike have identified them ascontributing factors.10 In Aceh, Indonesia,oil rents have been transferred back to theregion since 1976, yet regional conflict anddemands for autonomy have increasedrather than abated.11 This suggests thattransfers alone are not sufficient to addressregionally concentrated grievances—theymust be accompanied by meaningful politi-cal participation and dialogue.

ConclusionThe specific nature of the constraints toregional growth and investment perform-ance in lagging regions needs to be identi-fied and prioritized. Policies that providefiscal incentives to investors are likely to failif the main factors that adversely influenceregional investment climate—quality oflocal institutions, skilled labor availability,proximity to key markets, functioning capi-tal and land markets, security risks—stillpose binding constraints.

Public investment in infrastructure thatreduces transport costs for both people andgoods has often proved an effective strategyfor integration. And, as with other policies,well designed technical solutions are morelikely to be implemented if those living inpoorer regions are empowered.

Unprecedented economic growth and povertyreduction in China have been accompanied bysignificant increases in regional disparities sincethe economic reform in the late 1970s.Thesocioeconomic costs of a sustained divergencein income between leading and lagging regionshas become a major concern of the government.

In 1999, the government initiated the “GoWest” strategy to develop the lagging westernregion. Through targeted public investmentsand fiscal subsidies, the central government

spent some 1,000 billion yuan (US$120billion) in the past five years, focusing oninfrastructure, education, health, and the envi-ronment. A variety of investment incentivesand low interest loans aimed to attractdomestic and foreign firms to areas in whichthe western region has some comparativeadvantage, such as energy, agriculture, andagroprocessing.

The relative decline of the historicallyadvantaged northeastern region has also

attracted concern. China’s northeast currentlysuffers from slow growth and high unemploy-ment in declining industries, along with manyseverely distressed towns and cities.The govern-ment started the “Revitalize Northeast” strategyin 2003.This involves new initiatives, includingstrengthening the investment climate, develop-ing greater flexibility in factor markets, usingpublic funds to support rather than postponeadjustment, and mitigating social costs throughimproved and portable safety nets.

Development of lagging regions in China

Focus on regional inequality 205

206

We read in chapter 2 that there are hugeinequities in the world. Even better-off citi-zens in most of the developing world faceworse opportunities than the poor in richcountries. The fact that country of birth is akey determinant of people’s opportunitiesruns counter to our view of equity—that is,that people should enjoy the same opportu-nities regardless of their background,including where they are born.

Greater global equity is desirable foritself to all those who find equity intrinsi-cally valuable. The international humanrights regime testifies to the shared beliefthat all should have equal rights and bespared extreme deprivation. Some evenargue that there is a powerful moral case forrich countries to take action, because of thehuge disparities and (arguably) becausethey partly created and perpetuate globalinequities.1 Greater equity is also desirablebecause it would likely be beneficial toglobal prosperity in the long run. Greaterequity in access to health and health reme-dies, especially for transmittable diseases,would reduce global health inequalities andbe beneficial to poor and rich countriesalike. Greater equity in access to and controlover natural resources and the global com-mons may lead to more sustainable use.Some argue that greater equity could alsolead to greater international stability: fragileand failed states pose a threat to local andglobal stability.2

What can be done to reduce the hugeinequities we experience today? The debateabout what causes global inequities andhow to address them is highly contentious.Some see globalization—greater globalintegration—as a source of equalization,others a source of widening inequalities,with richer countries and corporations

making rules that benefit themselves at thecost of the weak, poor, and voiceless. Thereis some truth on all sides of the debate. Interms of trends, we saw in chapter 2 that thepicture is mixed: convergence in health and(probably) education for many, conver-gence in incomes for some, but divergencein incomes and health for others. In termsof causes, just as some of the major sourcesof convergence have been associated withglobalization of markets and knowledge—the East Asian tigers, China, India makinguse of global markets, the spread of thegreen revolution and health-related tech-nology—so unequal rules and unequalinfluence profoundly shape opportunity.

Domestic action is clearly central toreducing inequities. Developing countrieshold the keys to their prosperity; globalaction cannot substitute for equitable andefficient domestic policies and institutions.But global conditions powerfully affect thescope for and impact of domestic policies.Global action—by governments, people,and organizations in developed countriesand by international institutions—can deter-mine whether the globalization processbrings about greater equity, peace, and pros-perity, or fuels tensions and conflicts thatwill lead to backlash and violence.

Current disparities are products of inter-actions between two factors: the endowmentsof different countries, and the rules shapingthe options for deploying these endowmentson domestic and global markets. Endow-ments are greatly unequal due to history andgeography—although some of the historyand aspects of geography are a product ofunequal development patterns. Infrastruc-ture underdevelopment in Africa, for exam-ple, is partly a legacy of colonial political andeconomic patterns. Institutional weaknesses

Achieving greater global equity

10c h a p t e r

Achieving greater global equity 207

of poorer societies—now part of theirendowment—also reflect historical pat-terns, as discussed in chapter 6. Differencesin endowments are often exacerbated by theinequitable functioning of markets. As inthe domestic realm, market imperfectionscan be either a product of policy (as in bar-riers to labor mobility or agricultural pro-tection) or of intrinsic market failures (as inweak protection of global commons andlack of incentives for knowledge creation).

Achieving greater global equity thusrequires global policies that improveendowments and address market imperfec-tions and more representative global insitu-tions. We first discuss the global markets forlabor, goods, ideas, and capital—all func-tioning within the context of internationallaw (box 10.1). For each market, we high-light existing inequities and their impact,discuss the processes that lead to suchinequities, and explore some options forchange. We next turn to rectifying past and

present inequities in the use of naturalresources. Then we look at whether aid—the traditional response to globalinequity—can be used effectively to acceler-ate domestic efforts to build endowments.The current state of international relationsmay cause some to wonder whether anychange is possible. So we close the chapterby examining factors that have facilitatedtransitions to more equitable policies andinstitutions in the past. We conclude thatchange may be difficult but not impossible.3

Making global markets workmore equitablyGlobal markets have many faces: Filipinonurses, Sri Lankan domestic workers, Polishcare providers, Indian engineers, Ugandancoffee growers, Bangladeshi women workingin garment factories, Moroccan craftsmen,employers of migrants, and the consumers ofdeveloping-country products in Australia,

Globalization takes place (mostly) in the contextof international law, which governs relationsamong states, and other international legal sub-jects, such as international organizations. Moreequitable development, application, monitoring,and enforcement of international law is essen-tial to make globalization more equitable.

The meaning of equity in international law. Equityconsiderations inform the development of interna-tional law,confirming that greater global equity is ashared value.The principle of equity has accompa-nied the development of international law over thecenturies (chapter 4).Equity in international lawencompasses notions of corrective justice and dis-tributive justice—that the strict application of thelaw should be tempered by considerations ofequity or fairness to achieve a just result,and thatinternational law should promote a more even dis-tribution of resources among states.Equitable prin-ciples have been applied to many areas of interna-tional law,from the sharing of scientific benefits,technology,and natural resources to laws govern-ing the sea,international waterways,outer space,and carbon emissions.As highlighted in chapter 4,the most pertinent example of the application ofprinciples of equity in international law is the inter-national human rights regime.In today’sinternational law,equity has not only an interstatedimension; it also has an intergenerational dimen-sion,in the preservation of the environment andother global commons,as we will see below.

Rule-setting processes. International laws areformed through complex negotiating processes.The degree to which these processes areperceived to be equitable affects their adoptionand implementation—so processes mattergreatly. Generally, a state remains free to decidewhether to become a party to a convention orcovenant. And a state’s satisfaction with theprocess leading to the adoption of a conventionmay facilitate signing and subsequent adoption.For example, the Universal Declaration ofHuman Rights, seen by many as the basis of sub-sequent human rights instruments, wasadopted by the U.N. General Assembly, where allcountries are represented and have one vote.While only a declaration, and not intended tobind states at the time it was adopted, theprocess leading to its adoption was perceived tobe equitable.The body of standards set by theILO is another example of rules set through aninternational process that is broadly consulta-tive, encompassing not just governments butunions and private sector representatives. Onthe other hand, the rule-setting processes of theWorld Trade Organization (and its predecessor,the General Agreement on Tariffs and Trade) areperceived by some as inequitable, and this ispartly responsible for the current stalemate.

Application and enforcement mechanisms.The processes that interpret, apply, and enforceinternational laws are crucial to realizing

greater equity. In general, the ability of statesto pursue and enforce rights underinternational law depends on appropriateadjudication processes or complaintmechanisms and their effectiveness. A numberof international courts and other adjudicativebodies often have voluntary jurisdiction, butthere is a trend toward judicialization and com-pulsory jurisdiction. For example, dispute set-tlement arrangements established under the1982 U.N. Convention on the Law of the Seaand the 1994 World Trade OrganizationDispute Settlement Understanding mark a sig-nificant move toward compulsory jurisdictionand binding decision making.

The ability of citizens and other nonstateactors to pursue their rights and seek redressunder international law depends on whethertheir state has become a party to theinstruments that allow the use of compliancemechanism. For example, for citizens to make acomplaint against their state under the Interna-tional Covenant of Civil and Political Rights, thestate must have signed and ratified the FirstOptional Protocol, which allows a complaint tobe heard by the Human Rights Committeeestablished by the covenant. As the discussionindicates and in parallel to what happens on thedomestic arena, rules often block access, evenbefore expenses, knowledge, and capacity limiteffective recourse.

B O X 1 0 . 1 International law, globalization, and equity

the European countries, Japan, the UnitedStates, and the richer middle-income coun-tries. Global markets create valuable eco-nomic opportunities for millions of people,who develop ideas, raise capital, and sell theirproducts and their labor.

But unequal endowments and unfairprocesses mean that opportunities and rulesare not the same for all. Inequities exist inthe functioning of these markets. Unskilledworkers from poor countries, who couldearn higher returns in rich countries, facegreat hurdles in migrating. Developing-country producers face obstacles in sellingagricultural products, manufactured items,and services in developed countries. Foreigninvestors often get better deals in debt crises.

In most cases, more equitable ruleswould bring benefits to both developed anddeveloping countries, but the extent of ben-efits varies by market. Barriers are massivelygreater in the market for labor—the factorof production that the poor own in relativeabundance—than in the markets for goodsand capital, and factor price equalizationclearly does not work through trade alone.So removing barriers to migration couldhave a significant impact on expanding

people’s opportunities (of course, migra-tion raises complex issues that are politi-cally and socially difficult to tackle insending and receiving countries).

Benefits also vary greatly depending oncountry context. The fast-growing develop-ing countries, including China and Indiathat are home to half the world’s poorestpeople, stand to benefit significantly frommore equitable global markets. Leveling theglobal playing field can help them sustainfast growth, while equitable domestic poli-cies help ensure that this growth is shared.Countries with more limited endowments,such as many African countries, that are leftbehind in the global economy, stand to ben-efit less in the short to medium run frommore equitable global markets.

Greater international labor mobilityReturns to capital, and to some extentskilled labor, tend to equalize across coun-tries, but returns to unskilled labor, ownedby poor people and in abundant supply inpoor countries, generally do not converge.Wage differentials across countries for jobsrequiring similar skills are large, and sub-stantially larger than the wage gap betweenthe United States and migrant-sendingcountries in the late nineteenth century (fig-ure 10.1). Developed countries severely limitin-migration of unskilled and semi-skilledworkers, which contributes to the lack ofequalization in returns to unskilled labor.

Greater migration of unskilled laborwould tend to equalize returns, with win-ners and losers, but with potentially benefi-cial effects on efficiency. History teaches usthat migration has, at various times, allevi-ated human suffering and promoted cul-tural and technological exchanges. The massmigration from Europe to the Americas inthe nineteenth and early twentieth centuriesenabled 60 million people to escape povertyand persecution, creating some of today’swealthiest societies (although Native Ameri-cans faced enormous losses in the process).4

Economic analyses indicate that gainsfrom expanding migration could be verysignificant. Hamilton and Whalley (1984)use a highly simplified economic model ofthe world to suggest that the benefits fromreallocation of labor could be huge (on the


0.0

USA/Ireland1870

USA/Norway1870

USA/Sweden1870

USA/Italy1870

Spain/Morocco1990s

USA/Guatemala1990s

UK/Kenya1990s

Italy/Ethiopia1990s

NLD/Indonesia1990s

Japan/Vietnam1990s

1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0Ratio of wages in PPP

Figure 10.1 Wage differentials are substantiallylarger today than at the end of the nineteenth centuryRatios of purchasing power parity adjusted wages ofthe United States and its migration partners in 1870and pairs of countries in the 1990s

Source: Pritchett (2003).

order of doubling GDP). This, of course,depends on the specific assumptions usedand ignores a host of adjustment issues, butit does serve to illustrate that the gainscould be large and probably much largerthan the gains from the, by comparison,already greatly liberalized trade in goods.Indeed, using an approach similar to that of analyses of trade impacts, Walmsley and Winters (2003) estimated that increas-ing temporary migration into industrialcountries by 3 percent of host countries’current skilled and unskilled work force—equivalent to permitting an extra 8 millionskilled and 8.4 million unskilled workers tobe employed at any time, roughly a dou-bling of current net migration into high-income countries—would generate an esti-mated increase in world welfare of morethan $150 billion a year. This increasewould be shared fairly equally betweendeveloping- and developed-country citi-zens. Much of the gain would come fromthe migration of unskilled workers. Coun-try studies confirm that migration couldhave a significant impact. Annabi and oth-ers (forthcoming) found that a 50 percentincrease in the flow of remittances toBangladesh would reduce the incidence of$1 per day income poverty by 0.8 percent inthe short run and by 4 percent by 2020.5

Doesn’t migration raise income inequal-ity in sending countries? As a high-risk,high-return activity, migration is morelikely to be undertaken first by members ofwealthier, less credit-constrained, better-educated households. Successful migrantslater provide information and assistance topotential migrants through social networks,thus lowering risks and costs and making itpossible for members of households inlower parts of the income distribution tomigrate.6 In the first stages of the migrationprocess, remittances sent to wealthierhouseholds can increase inequality, if theyare higher than forgone income.7 As migra-tion expands, remittances begin to arrive toless well-off households and income distri-bution improves.8 Remittances also haveindirect effects through greater spending,risk diversification, and easing of creditconstraints, which are generally inequality-reducing.9 On balance, the evidence does

not support the view that migration leadsunequivocally to higher inequality in send-ing countries.

In receiving countries, migration relieveslabor shortages in labor-intensive sectors,such as health care, hotels and restaurants,and construction. As developed-countrypopulations age and their levels of educa-tion and training rise, these shortages arelikely to become more severe. Demographictrends are another powerful force behindmigration. Current population projectionsimply that the labor forces of Europe andJapan will decline over the next century, andthat the ratio of people of working age topeople of retirement age (the support ratio)will grow to levels that would make currentpension and social transfer schemes unvi-able. Meanwhile, the population of theNorth Africa countries south of Europe isgrowing rapidly.

Despite its large benefits, migration isfiercely opposed in receiving countries.Migration involves complex issues of nationaland individual identity exacerbated by con-cerns over security. Cultural and socialintegration appears more difficult in somecountries than it was earlier thought.Moreover, unskilled workers experiencewage erosion and unemployment. Forindustrial workers, however, this is no dif-ferent than if goods produced in countrieswith lower labor costs displace domesticproduction.

In sending countries, there are concernsabout the human and social costs of migra-tion, for instance, on how migration ofnurses and doctors hinders progress towardthe Millennium Development Goals (MDGs)and migration of women creates majordeficits in child rearing, family support,and care for the elderly.10 Licensing restric-tions (as for doctors) often force skilledmigrants to work in lower-skilled jobs inhost countries—the “brain waste,” andhigher returns to education, do not appearto spur human capital accumulation or“brain gain.”11

Going against the political tide—withthe partial exceptions of some currents inthe United States, Canada and Spain—weargue that greater migration would be goodfor both equity and efficiency. But what are


the prospects for greater migration in thecurrent political climate? Multilateral nego-tiations in the World Trade Organization(WTO) offer a framework to addressmigration under Mode IV of the GeneralAgreement in Trade and Services (GATS),part of the treaty establishing the WTO.12

But progress toward greater liberalization oftemporary migration under GATS Mode IVis unlikely in the near future, given thatcontentious issues on agricultural and mer-chandise trade are dominating negotiationson the Doha Round.

In this context, progress is more likely tocome from bilateral and regional negotia-tions. Receiving countries could bilaterallyexpand temporary migration (box 10.2 dis-cusses some features of “development-friendly” temporary migration schemes).These countries could also extend greaterprotection to migrants. One way to do thiscould be to ratify the 1990 U.N. Conventionon the Rights of All Migrant Workers andTheir Families. If a significant number ofhost countries were to ratify the conven-tion, none would risk being considered ahaven for undocumented migrants, andfears about ratification leading to greaterinflows might be allayed.13 Facilitatingremittance flows is another action withpotentially high payoffs, and governmentsshould work together with the private sec-tor and NGOs to achieve this.14

Sending countries should take action toreduce the likelihood that their migrantsbecome victims of exploitation, with afocus on combating trafficking of girls and

women.15 Two possible areas for action areto better regulate recruitment agencies toensure greater respect for workers’ rightsand to enter agreements that regulatemigrant flows and conditions with key des-tination countries, as the Philippines hasdone. Sending countries should also helpmigrants use remittances properly, investback home, and reintegrate upon return.

It is unclear whether an internationalorganization in which poor countries havean equal seat at the table could help makeprogress toward freer migration. Bhagwati(2003) argued that a new World MigrationOrganization—or even a stronger Interna-tional Organization for Migration in theU.N. system—might help increase thedevelopmental impact of migration by pro-tecting migrants’ rights, providing a forumto set rules on migration, and monitoringand enforcing compliance. But migrant-receiving developed countries resist propos-als to give up even some control over immi-gration policies, which they view as part ofthe domestic policy agenda.

Freer and fairer tradeInequities in the trade arena are wellknown: rich countries protect their marketswith tariff and nontariff barriers on thegoods that poor countries produce moreadvantageously (such as agricultural pro-duce and textiles). They provide handsomesubsidies to their farmers, subsidize theirexports, and discourage value-added pro-cessing in developing countries. Reducingsuch protection and subsidies would have abeneficial impact on world trade, growth,and poverty reduction.

Potential benefits from liberalization. Sev-eral recent studies have estimated the poten-tial impact of various trade liberalizationmeasures, including those being consideredduring the Doha Round of negotiationsunder the WTO. Estimates vary, dependingon the reforms considered (various packagesof partial reforms up to full liberalization)and on whether dynamic productivity gainsare taken into account. At the lower end ofthe range, Hertel and Winters (forthcoming)estimated that the measures being discussedin the Doha Round would have a modest


Attention to the design of temporary migrantworker schemes and complementary policiescould make them more developmentfriendly.Temporary migration schemes usu-ally allow workers into a country from severalweeks to up to three to five years.

Recent research for the United Kingdomidentified two main policy interventions:centralized recruitment with governmentscreening in sending countries, and manda-tory saving schemes, possibly coupled withcredit schemes in home countries. Central-ized recruitment and government screening

could help reduce exploitation of potentialmigrants by local recruitment agencies andensure that temporary migrants do notoverstay their visas, reducing resistance inreceiving countries. Saving schemes wouldencourage migrants to send remittanceshome, increase incentives to return, andfacilitate the start of productive activities onreturn. Good examples are the Canada-Mex-ico program for agricultural workers and theagreement between France and Sri Lankaon sharing information about migrants.Sources: Barber (2003); Schiff (2005).

B O X 1 0 . 2 Making migrant worker schemes moredevelopment friendly

impact on world prices, welfare gains, andpoverty, with the number of people livingbelow $2 a day declining by 9 million in2015 over a baseline estimate of around 2billion. According to this study, even full lib-eralization would not bring huge gains, as itwould help lift 80 million people out of $2 aday poverty. At the higher end of the range,Cline (2004) estimated that full trade liber-alization would lift up to 440 million peopleout of $2 a day poverty by 2015.

Whatever the size of the overall impact,researchers agree that it would be heteroge-neous across countries and regions. In bothpartial and full reform scenarios, the gainswould accrue mostly to large countriesalready significantly integrated in globalmarkets, such as Brazil, China, India, andIndonesia. Parts of many Sub-Saharan coun-tries and remote areas in Asia and elsewhereare simply not connected to global markets,and farmers eke out a living on subsistenceagriculture, far from roads, markets, technol-ogy, and information. Many countries areunable to make full use of improved marketaccess because of significant supply-side andinstitutional constraints. Detailed studies onCambodia, Ethiopia, Madagascar, and Zam-bia showed that the potential impact of thetrade reforms likely to be included in theDoha Round would be small for such coun-tries. Some countries would even lose in theshort run: Bangladesh and Mozambique, forinstance, would experience a decline inincomes, as existing preferences are erodedand the prices of key food imports rise. Sim-ilarly, Bourguignon, Levin, and Rosenblatt(2004b) found that countries in the bottomtwo deciles of the international distributionof income would benefit more from a dou-bling of aid over current levels than from fulltrade reform. The estimated impact of tradeliberalization varies greatly within countriesas well (chapter 9).

Specific liberalization measures wouldalso have differential impacts. Andersonand Martin (2004) found that the removalof OECD agricultural subsidies would hurtnet food-importing least developed coun-tries, such as those in the Middle East andNorth Africa, and countries that now enjoyspecial preferences, such as the Philippines,because of the consequent increase in

prices.16 But net food-producing countries,and farmers within them, would benefit.There is, indeed, some evidence that risingagricultural world prices were partly respon-sible for the fact that rural incomes inChina grew more rapidly than urbanincomes in 2004.

The phasing out of the Multi-Fiber Agree-ment, which set quotas on exports of textilesfrom developing countries, also has heteroge-neous effects. Chinese textile exports havemade significant gains in markets not pro-tected by tariffs—for instance, their share ofAustralian and Japanese markets, where therewere no quota restrictions, is 70 percent.Their share of the U.S. baby clothes segment,where quotas were removed in 2002, jumpedfrom 11 to 55 percent in two years. Exportsfrom Cambodia and Nepal are reported tohave declined significantly. The impact ofthese shifts on global income inequality is notclear and it depends on the relative positionof garment workers and of those benefitingfrom indirect effects in the global distribu-tion. Changes in the existing tariff structure,whereby producers from the poorest coun-tries have duty-free access to markets in theUnited States and Europe while others face a16 percent tariff on average, would also havean unclear impact on inequality.17 Con-versely, renewed protectionism in developedcountries is likely to have negative effects.

Currently, no global assistance programexists to compensate losers from trade lib-eralization. However, international assis-tance to help meet adjustment costs is animportant focus, along with addressingsupply-side constraints, of current effortsby a range of donors, recipients, and inter-national organizations, including the WorldBank, to increase aid for trade in the contextof the WTO Doha round.

Setting trade rules. Where do the rules thatgovern trade come from, and what are thechances of changes? Trade rules, includingthe most egregiously inequitable, are part ofcomplex multilateral, regional, and bilateralagreements. As mentioned in box 10.1, thereare significant concerns about the fairness ofWTO decision-making processes, and theseprocesses are partly responsible for the cur-rent stalemate in negotiations.



But the reality of WTO negotiations iscomplex. In the WTO, each country has onevote, and the practice of decision making byconsensus means that each country canveto decisions (although the practice of“single undertaking,” or voting on all mat-ters together, in practice weakens vetopower). Countries choose to sign on to theWTO following not only extensive externalnegotiations but also domestic decision-making processes. So this is not a primafacie example of unfair rule-setting. Inpractice, however, poor countries find it dif-ficult to follow negotiations, to understandthe implications of proposals to them, andto develop alternative proposals—we saw inchapter 3 that even their capacity to be pres-ent in Geneva is limited.

So, in the end, the rules may at times beunfair not because the formal processes areunfair but because of the underlying powerimbalance between rich countries withstrong commercial interests and poor coun-tries with weak capacity.18 The balance iseven tilted against taxpayers and consumers

in rich countries, who often stand to losefrom the protection of vested commercialinterests. Consider, for example, cotton sub-sidies (see box 10.3) and international car-tels.19 Poor countries are in an even weakerposition when negotiating bilaterally withstronger trading partners than they arewhen negotiating multilaterally. Paradoxi-cally, in light of the intense antiglobalizationprotests, multilateral negotiations in thecontext of the WTO hold the greatest prom-ise to reduce inequities that harm poorcountries. Although even an ambitiousDoha Round would bring limited benefits, itremains an important goal to pursuebecause failure would further undermineconfidence in multilateral negotiations.

The WTO has another advantage: it pro-vides for a mechanism to adjudicate dis-putes. This is important, as seen earlier, toensure that international law is applied andenforced. The WTO dispute settlementmechanism provides a forum for poor coun-tries to bring complaints and possibly winthem. Unfortunately, winning a case does

The International Cotton Advisory Committeeestimated that, in 2001/02, direct productionassistance by the eight countries that providedsubsidies (United States, China, the EuropeanUnion, and to a much smaller extent Turkey,Egypt, Mexico, Brazil, Cote d’Ivoire, in that order)was around $5.8 billion. Direct assistance to U.S.cotton producers reached $3.3 billion, China’s sup-port totaled $1.2 billion (although some questionthis estimate), and the European Union’s supportwas $979 million (for Greece and Spain) (Interna-tional Cotton Advisory Committee 2003).Themain impact of U.S. and European subsidies is tomake cotton produced in the United States andEurope competitive and depress world prices. It isestimated that in 2001/02 prices would have been71 percent higher without subsidies.

Subsidies benefit large rich farmers in theUnited States and not-so-rich but relatively well-off farmers in Europe, and harm poor, smallfarmers in Africa. Cotton is a crucial commodityfor a number of poor African and Central Asiancountries, contributing up to 40 percent of mer-chandise exports and 5 to 10 percent of GDP.Most growers are smallholders, so the impact ofcotton prices on poverty is significant. A studyon Benin found that a 40 percent reduction infarmgate cotton prices—equivalent to the pricedecline from December 2000 to May 2002—

implied an 8 percent reduction in rural percapita income in the short run and a 6 to 7 per-cent reduction in the long run, with theincidence of poverty among cotton growers ris-ing in the short run from 37 percent to 59 per-cent (Minot and Daniels 2002).

Estimates of the impact of subsidy removalon cotton prices are in the range of 8 to 12 per-cent. Increases of this magnitude would nothurt consumers—the price of raw cotton is asmall component of the price of textiles andgarments. Full subsidy removal and the conse-quent rise in prices would help Africancountries, although the distribution of in-coun-try benefits would depend on domestic reforms.A recent study of the impact of subsidy removalon three cotton-producing provinces of Zambia,for instance, indicates that the direct impact ofthe subsequent cotton price increase would besmall: about 1 percent of income on average.Greater gains would require farmers switchingfrom subsistence crops to cotton, which in turnrequires complementary domestic reforms inextension services and robust growth ofdemand for cotton exports (Balat and Portoforthcoming).

Benefits to African countries would increaseif they were to expand their clothing productionand exports.The U.S. African Growth and Oppor-

tunity Act provides an opening, but underrather restrictive conditions: apparel from 14African countries gets duty-free and quota-freeaccess to U.S. markets, but only if made fromU.S. fabric, yarn, and thread. So to takeadvantage of this provision, countries need toestablish an effective input visa system toensure compliance with rules of origin (Baffes2004), which seems exceedingly complex.

Within the WTO, poor cotton-producingWest African countries took the unusual step ofissuing a joint statement calling for full subsidyremoval and for cotton to be treated separately.But the July 2004 Framework Agreement of theDoha Development Agenda does not includeseparate treatment of cotton, stating only thatcotton will receive “adequate priority” in agricul-tural negotiations. Subsidy removal is politicallyunlikely.

In the current climate, a second-best optionwould be to implement well-designed decou-pled support, in which subsidies do not dependon production and thus do not encourage over-production and consequent “dumping,” as is thecase with the current schemes. Existing mecha-nisms would need to be reformed, because theystill depend on acreage and thus create incen-tives for overproduction. Less overproductionmay help lift prices a bit.

B O X 1 0 . 3 Cotton subsidies are huge—and tenacious

not automatically bring redress: the loser inthe case may not necessarily change its action.The existing mechanisms to enforce decisionsrely on voluntary compensation of the loserand, when this is not satisfactory, the possibil-ity of retaliatory action (such as suspension oftariff and other concessions) on the part ofthe winner. Clearly, poor countries’ retalia-tion against powerful trading partners isunlikely to provide much of an incentive forrich countries to comply with unfavorablerulings, because of their typically smaller vol-ume of trade with a developed-countrydefendant. Even so, developing countrieshave in recent years brought forward, andwon, an increasing number of cases.

The fair and ethical trade movements.Interestingly, some NGOs and civil societyorganizations in both developed and devel-oping countries have acted directly to estab-lish more equitable trade relations. Onesuch example is “fair trade.” Fair trade ini-tiatives, led by consumer groups, NGOs,trade unions, and other civil society organi-zations, aim to control the supply chainfrom production to market to improve thewell-being of developing-country produc-ers by ensuring a stable price for their com-modities, linking them more directly withmarkets in rich countries, and strengthen-ing their organizations. The approach isworking: sales of fair trade bananas, cocoa,coffee, brown sugar, tea, and a few otherproducts have seen phenomenal growth inrecent years and now represent a significantshare of exports for some countries (forinstance, 11 percent of Ecuadorian bananasand 20 percent of Ghanaian coffee are nowsold through fair trade).

The few impact studies that exist showthat fair trade initiatives have indeed made adifference to producers, not only through thepremiums paid over world prices but alsothanks to the services and assistance pro-vided to farmers by producer cooperativessupported by fair trade organizations. Wheninequities arise from unequal access to mar-kets and lack of information, credit, and risk-mitigation mechanisms, strengtheningproducer associations can lead to more equi-table outcomes, even without paying a pre-mium, in the context of existing trade rules.20

The reach of fair trade initiatives, whilegrowing, remains small. In Switzerland,where consumer support is strong, fairtrade bananas still represented only 25 per-cent of overall banana purchases and con-sumer spending on all fair trade productswas a mere $10 per person in 2002 (Swissagricultural subsidies amounted to roughly$750 per person in the same year). Fairtrade coffee accounts for, at most, 3 percentof world sales, and only about 20 percent ofthe capacity of certified fair trade producersis absorbed by the fair trade circuit.21

Another example of organizations act-ing directly to establish more equitabletrade relations is the growing number ofinitiatives for corporate social responsibil-ity and ethical trade. Companies that joinan ethical trade organization, such as theEthical Trading Initiative in the UnitedKingdom or the Fair Labor Association inthe United States, pledge to respect a codeof conduct in return for favorable consid-eration by consumers and investors whocare about equitable development.22 Codesof conduct generally cover fair laborpractices (usually those set out in ILO con-ventions), environmental standards, andmonitoring mechanisms—and apply notjust to a firm’s direct production facilitiesbut also to those of all its suppliers alongthe supply chain.

Are consumers in rich countries willingto pay a bit more to ensure that the goodsthey buy are produced in fair and safe con-ditions? Proponents of codes of conductbelieve they are. Researchers found thatalmost 90 percent of Americans said theywould pay at least an extra $1 on a $20 itemif they could be sure it had not been pro-duced by exploited workers.23 Skepticspoint to the fact that prices dominate thedecisions of the major corporate buyers.

Codes of conduct inspired by ethicalconsiderations might have a positive impacton equity, but are they applied? Impactstudies conducted by the Ethical TradingInitiative found mixed evidence. Consumersmay not be willing to pay higher prices inexchange for an uncertain (and oftenunmonitored) positive impact. Consumerpressure may thus not be enough (box 10.4).So, these initiatives, while important, are no



substitute for more equitable trade rulesunder the WTO and other arrangements.

Intellectual property rights and the global market for ideasProtection of intellectual property rights(IPR) is another area in which market failureand power structures shape unequal pro-cesses and outcomes; the interests of a fewpowerful actors impose costs on the generalpublic, particularly the poor. The require-ment set forth in the Trade-Related Aspects ofIntellectual Property Rights agreement(TRIPS)24—that all member countries offer20-year patent protection—is perceived bymany to be grossly inequitable. Becausepatent protection was adopted in OECDcountries before the 1990s, the main result ofthis requirement is to strengthen patent pro-tection in poor countries that become WTOmembers. Countries adopting patent protec-tion today are doing so at levels of GDPbetween $500 and $8,000 per capita, whileOECD countries did so when their GDP percapita was around $20,000 in 1995 prices.25

Patents stem from a legitimate desire toprovide incentives for the generation ofknowledge and cover the cost of developingnew knowledge. A drug or other patentedinnovation cannot be copied while a patentis in force, so developers enjoy a monopoly

position and can charge higher prices.Extending patent protection to developingcountries can thus increase total profits byallowing companies to earn them in poorcountries—and changes the distribution ofR&D financing, with a greater share borneby poorer countries. But protection of IPRmust be balanced by the concern that itrestricts access to new technologies. Patentsrestrict access to innovations by makingthem more expensive and more difficult tocopy. There is great concern in developingcountries on the availability of variousinnovations, including patented seeds anddrugs. Antiretroviral drugs to fight AIDSare a case in point (box 10.5).

We look in more detail at pharmaceuticalpatents as an illustration of the broaderissues. Chaudhuri, Goldberg, and Jia (2004)estimate that the gains to the Indian econ-omy from not following international patentprotection standards were around $450 mil-lion, of which $400 million were a gain toconsumers and the rest profits of domesticproducers. Profit losses to foreign producerswere only around $53 million a year. Thisstudy illustrates the important point that theprofits pharmaceutical companies could gainin poor countries are not very large. Lanjouwand Jack (2004) estimate that extendingpatent protection to developing countries to20 years would be equivalent, for firm prof-its, to extending patents in developed coun-tries by two weeks.

A solution exists that would lead to moreequitable provision without underminingefficiency: wherever rich country marketsalready support the cost of research, poorcountries could be allowed to produce orimport cheaper generic substitutes, at nosignificant cost to either rich countries orthe firms that carry out research (see focus 7on drug access at the end of this chapter).

As with all international law, the existingIPR protection rules are the result of com-plex negotiations. TRIPS—which was basi-cally written by industry lawyers26—is partof the agreement establishing the WTO, amultifaceted deal that included the Multi-Fiber Agreement and other provisions thatdeveloping countries deemed beneficial tothem. Many bilateral free trade agreements(such as recent agreements between the

As mentioned in chapter 9, the 1999 bilat-eral trade agreement between Cambodiaand the United States included a provisionwhereby Cambodia’s clothing exportswould increase each year if labor standardsimproved.The ILO was mandated to preparea report twice a year based on factory visitsand interviews with workers and unionsand to make it widely available.

The provision helped bring about a grad-ual improvement in working conditions inclothing factories, but this progress is underthreat with the end of the quota system.Thegovernment agreed to continue ILO inspec-tions until 2008, but employers can no longercount on increases in exports to the UnitedStates if they uphold labor standards. Someare aware that labor standards compliance istheir only real competitive advantage, but

there are reports of union leaders beingfired, lack of adherence to minimum andovertime pay rules, and repressed demon-strations. Employers are allegedly using thethreat of tough competition from China tocut salaries and benefits. But the employersare being watched—an independent unionmovement has grown in the industry andILO monitoring is increasingly sophisticated.Monitors are now using hand-held comput-ers to transmit findings from their factoryvisits, allowing timely reporting. If workingconditions deteriorate, activists, researchers,unions, and, most important of all, consumerswill know.Whether their pressure will beenough to ensure adherence to labor stan-dards is an open question.

Sources: International Confederation of FreeTrade Unions (2005), Washington Post (2004).

B O X 1 0 . 4 Will improved working conditions in Cambodia’s textile industry survive the end of the quota system?

United States and Chile, Jordan, Morocco,Singapore, Vietnam, and others) includeeven stronger IPR protection rules thanTRIPS, such as granting patent extensionson pharmaceuticals and specific types ofprotection on clinical trial data submittedto obtain marketing approval. Signatoriesto these agreements agreed on these rulesgenerally in exchange for preferential accessto U.S. markets for their products.

But it is hard to argue that the parties tothese various bilateral and multilateralagreements were on a level playing field.Poor countries are in a weaker bargainingposition overall. For example, the preferen-tial access they gain through bilateral tradeagreements is eroded whenever the UnitedStates reduces remaining tariffs and quotasin bilateral or multilateral negotiations,while IPR protection does not weaken overtime.27 Moreover, the issues involved in IPRprotection are complex and require skillsand capacity that rich countries can betterafford—often with input from pharmaceu-tical firms. Some capacity-building effortsfor developing countries are under way, but

at least some of the agencies responsible(such as WIPO and developed-countrypatent offices) are perceived as biased.Inequitable as TRIPS may be, it still providesan internationally agreed standard subjectto intense scrutiny and study, which doesmake it harder for rich countries to get morefavorable deals in bilateral agreements.

An additional advantage of negotiatingsessions under the WTO is that they pro-vide focal events for mobilizing publicopinion. An example of how positive resultscan be achieved within the WTO process isthe Declaration on the TRIPS Agreementand Public Health adopted at Doha in 2001,which affirms the primacy of public healthconcerns over IPR protection. Three subse-quent U.S. bilateral agreements include sideletters on public health that affirm the sig-natories’ understanding that IPR protectiondoes not affect their ability to “protect pub-lic health by promoting medicines for all.”28

When negotiations are shifted away fromthe spotlight, as drug companies managedto do with drug licensing under the July2004 Doha Development Agenda Frame-

In response to the rising AIDS crisis, the govern-ment of South Africa in 1997 amended the Med-icines and Related Substances Control Act of1965 in an attempt to ensure the supply of moreaffordable drugs to all South Africans.Theamendment encouraged pharmacists to substi-tute costly patented drugs with cheaper genericequivalents, allowed for the importation ofcheaper drugs available on the marketelsewhere (parallel imports), and introduced acompulsory licensing system allowing competi-tors to produce patented drugs.

The Pharmaceutical Manufacturers Associ-ation and 39 drug companies challenged thegovernment’s legislation in the Pretoria HighCourt on several grounds, including that itviolated South Africa’s obligations under TRIPS. The Treatment Action Campaign (TAC)and a labor union, COSATU, supported the gov-ernment defense in the case, asserting that thelegislation was valid in that it constituted thegovernment’s positive duty to fulfill the rightto health. Arguably as a result of public pres-sure and attention, the Pharmaceutical Manu-facturers Association and the drug companieswithdrew their case. An indirect result was tobring down the price of antiretroviral medicine

from about 4,000 rand a month to 1,000 rand amonth.

Other legal cases (not involving TRIPS)helped expand access to antiretroviral drugs. In2002, a group of complainants, including TAC,brought a case against GlaxoSmithKline andBoehringer Ingelheim at the South African Com-petition Commission. In its October 2004 ruling,the commission found that the two firms hadengaged in excessive pricing of patented anti-retrovirals and refused to allow generic produc-tion of the drugs in return for royalty payments,actions that the commission ruled were in viola-tion of the South Africa Competition Act.Tokeep the case from moving to a higher tribunal,the firms came to a settlement agreement thatincluded licensing generic production.

TAC also attempted to compel the nationaland provincial governments to provideantiretroviral drugs to all pregnant women toprevent the transmission of HIV from mothers totheir children; the impact of the existing govern-ment policy was to make the drug Nevirapineunavailable in public health facilities other thanthe 10 or so pilot sites.The governmentappealed to the Constitutional Court after TACsecured a successful decision.

The Constitutional Court declared that theSouth African Constitution required the SouthAfrican government to devise and implementwithin its available resources a comprehensiveand coordinated program to realizeprogressively the rights of pregnant women andtheir newborn children to have access to healthservices to combat mother-to-child transmissionof HIV.The Court found the state policy ofrestricting the availability of antiretroviral drugsand related services for preventing mother-to-child transmission of HIV to a few pilot test sitesunreasonable, and ordered the government torectify the situation by taking reasonable stepsto facilitate the availability and use of antiretrovi-ral drugs in all public health facilities.

In 1999,TAC had also been part of a success-ful constitutional challenge relating to discrimi-nation of South African Airways cabinattendants with HIV.The judgment reinforcedthe right to equality for people with HIV.Theselegal challenges had important indirect impacts,setting groundbreaking precedents, increasingjudicial awareness of human rights obligations,and heightening public awareness of rights.

Sources: Decker and others (2005), South AfricaCompetition Commission (2003).

B O X 1 0 . 5 Expanding access to antiretroviral drugs in South Africa


work Agreement, monitoring progress andcampaigning become more difficult. Somultilateral negotiations within the WTO,which are held under the spotlight, proba-bly hold the most promise in terms ofadopting more equitable rules.

Financial market liberalizationCapital flows to developing countries havegrown tremendously in the 1990s, bringingboth advantages and challenges. Short-termcapital flows are at times accused of con-tributing to financial instability while notenhancing growth in countries with imma-ture financial systems. Most countries thatreceived high volumes of short-term capitalinflows in the 1990s—Argentina, Brazil,Indonesia, Korea, Mexico, Russia, Thailand,and Turkey—have been hit by financialcrises, triggered or deepened by the flight offoreign short-term capital.

Domestic factors play a key role in finan-cial instability, but global rules also play arole. For instance, debt workout mecha-nisms follow informal processes; the IMF’sproposal for a Sovereign Debt WorkoutMechanism was not adopted. The result isthat deals tend to benefit internationallenders at the expense of domestic investorsand taxpayers.29

In contrast to short-term capital flows,foreign direct investment (FDI) is generallyregarded as having a positive impact onreceiving countries, but it goes to only a fewcountries. In 2002, 84 percent of FDI todeveloping counties went to 12 mostlymiddle-income countries (including Chinaand India), with the other 150-odd develop-ing countries receiving almost nothing.Only 5.3 percent of FDI went to Sub-SaharanAfrica.30 Domestic factors play a key rolealso in determining the location of FDI, butagain global rules contribute to inequitableoutcomes. The Basel II Capital Accord, thatsets capital adequacy standards for banks,may overestimate the risk of bank lending todeveloping countries (in part because itignores the benefits of diversifying portfo-lios across countries), thus raising the costand reducing access to external capital, inaddition to increasing the procyclicality ofloans and possibly contributing to increasedvolatility.31 Emerging global standards—

including those assessed under the Reportson the Observance of Standards and Codes(ROSC), international accounting stan-dards, and the Core 25 Principles for Bank-ing Supervision—are also costly for devel-oping countries and may not be appropriateto their level of development.

Rules-setting in global financial markets.Some of the key rules governing globalfinancial markets are developed by institu-tions to which developing countries do notbelong. The Financial Stability Forum,established in 1999 to promote globalfinancial stability, brings together seniorrepresentatives of central banks, supervi-sory authorities and treasury departmentsof nine OECD countries, internationalfinancial institutions, international regula-tory and supervisory groupings, com-mittees of central bank experts, and theEuropean Central Bank. The only emergingmarket economies that are members areHong Kong (China) and Singapore.

The Basel Committee on Banking Super-vision, which developed the Basel II CapitalAccord, comprises representatives of thecentral banks and banking supervisionauthorities of Belgium, Canada, France,Germany, Italy, Japan, Luxembourg, theNetherlands, Spain, Sweden, Switzerland,the United Kingdom, and the United States.Its main interlocutor in the development ofthe Accord was the Institute for Interna-tional Finance, a Washington-based consul-tative group of major international banks.Neither the Financial Stability Forum northe Basel Committee can legitimately repre-sent the interests of developing countries.32

Various other standards, often developed bysemiprivate agencies (such as the Interna-tional Accounting Standards Board), arebased on practices in the United States andEuropean Union. Greater participation andvoice in rule-setting bodies would helpensure that outcomes are more favorable todeveloping countries.

Rectifying past and present inequitiesin the use of natural resourcesThe use of natural resources is another majorarena in which market failures and unequalpower conjure to create major inequities.


This is greatly skewed in favor of devel-oped countries and impacts are grosslyinequitable. Without major technologicalinnovations several key resources, such asoil, could be exhausted before the world’spoor get a chance to attain standards ofliving comparable to those of today’sdeveloped-country citizens. Moreover,global warming threatens to destroy thelivelihoods of people living in low-lyingcoastal areas, small islands, and semiaridregions. Yet the people potentially affectedby these changes (tomorrow’s citizens andmany of today’s poor) have virtually novoice in setting rules.

The international community has takensome steps to manage natural resources in amore equitable way. Some internationallegal instruments, such as the Conventionon the Law of the Sea of 1982, reflect theconcept of distributive justice discussedearlier by taking an approach whereby theseabed and ocean floor, beyond nationaljurisdiction, are classified as global com-mons and subject to a system of equitablesharing of the economic benefits derivedfrom activities in these areas.

Key steps toward redressing inequitiesin the use of global resources are the 1992U.N. Framework Convention on ClimateChange and the 1997 Kyoto Protocol. Theprotocol is structured to reflect the prin-ciple of “common but differentiatedresponsibilities” between developed anddeveloping countries. It recognizes thatindustrial nations have emitted the major-ity of greenhouse gases in the atmosphere,causing the majority of the harm, andplaces greater demands on them. It setsbinding quantified commitments forindustrial countries to reduce their green-house gas emissions by 2008–12, with theunderstanding that the agreement wouldinclude emission reduction efforts bydeveloping nations some time after 2012.

One important aspect of the Kyoto Pro-tocol is the unique set of provisions thatallow industrial nations to meet their com-mitments through actions not only withintheir borders but also outside. One of theseprovisions, the Clean Development Mecha-nism, helps address the perceived inequalityof obligations and the costs of compliance.

It allows industrial countries to purchaseemission reduction “credits” generatedfrom activities that reduce greenhouse gasemissions in developing countries and toapply these credits against their obligationsunder the Protocol. It thus assists industrialcountries in meeting their commitmentsunder the Kyoto Protocol more cost effec-tively and promotes sustainable develop-ment in developing countries, throughsupporting greater investments in cleaner,more efficient technologies as well asforestry projects.

Fairness in processes is also an issue. Innegotiating the Kyoto Protocol, as in mostglobal treaty negotiations, industrial nationshad greater power at the negotiating table.An imbalance of technical expertise, a lack ofadequate public support for the issues, andproblems forming coalitions because ofdiverse interests have attenuated the bargain-ing power for many developing countries.

The United States, the single largest emit-ter of greenhouse gases, has announced thatit is not becoming a party to the Kyoto Pro-tocol, significantly reducing the protocol’sefficacy. With the protocol having come intoeffect in February 2005, the United Stateswill be a mere observer at the Meetings of theParties to the Kyoto Protocol, but because ofthe size of its emissions, the other parties willnot want to ignore U.S. concerns.33

Equitable access to information is animportant ingredient for more equitableuse of global resources. The UN/ECEConvention on Access to Information, Pub-lic Participation in Decision-making andAccess to Justice in Environmental Matters(Aarhus Convention) deals with public par-ticipation in environmental managementand access to information on environmen-tal issues. The Convention, adopted in 1998and in force among 35 parties since 2001,grants citizens the right to impose obliga-tions on public authorities and parties tointernational environmental conventions,including information disclosure, access toinformation, public participation in envi-ronmental decision making, and access tojustice. A Convention Compliance Com-mittee has been established, to which citi-zens and NGOs can bring allegations ofnoncompliance.


Providing development assistanceto help build endowmentsIn shaping global inequities, rules andprocesses interact with unequal endow-ments. Even if all the reforms suggested inthe previous sections were implemented,many poor countries would still not be ableto participate in global markets because oftheir limited endowments of skills, capital,infrastructure, knowledge, and ideas. Actionto build endowments is primarily domestic,through private and public investments ininfrastructure and other areas. Can domes-tic action be supported by aid?

Better development assistanceFrom an equity perspective, the main rolesof aid are to help countries build theendowments of those who are resource-poor, generally through no fault of theirown, and avoid extreme deprivation (whichjustifies to some extent the use of aid tosupport current consumption). The focuson building endowments implies that boththe level of aid and its effectiveness matter.

Enhancing aid effectiveness. If the goal is toequalize opportunities for the poor, aideffectiveness is crucial. Aid that sustainscorruption or marginal projects, or is usedto increase the resources at the disposal ofthe rich, does not help. Aid effectivenesshinges crucially on aid delivery modalitiesand on the fairness and transparency ofdomestic political processes. Birdsall (2004)cites seven “deadly sins”: impatience withinstitution building, failure to exit, failureto evaluate, pretending that participationequals ownership, failure to collaborate,stingy and unreliable financing, and under-funding of regional and global programs—in addition to tying aid to the use of con-sultants and firms from the donor countryand allocating it according to political pri-orities. Existing aid planning and deliverypractices are rooted in political and incen-tive constraints that the donors face, sochange is difficult and slow. But some cur-rent directions are promising: emphasizingresults (including through tracking indica-tors of intermediate actions and final out-comes related to the MDGs), moving awayfrom ex ante conditionality, and progres-

sively shifting design and managementfrom donors to countries. The United King-dom’s Commission for Africa (2005) rec-ommended a major shift away from ex anteconditionality toward a new partnership inwhich African countries continue to workto improve governance and accountability,and donors deliver more, cheaper, morepredictable aid. High levels of aid reducethe need for domestic tax efforts, whichhave historically helped strengthen overallaccountability of governments and citizendemand for quality services, so particularattention should be paid to revenue col-lection.34 The preparation of povertyreduction strategies is a key, if imperfect,instrument to shift to country-led processeswith greater participation and monitoringof how public resources are spent.

Fragile states pose a special challenge.Stabilization and peacekeeping need to becomplemented with efforts to build stateinstitutions and legitimacy. The sequencingof interventions matters—there is some evi-dence that ring-fenced, long-term invest-ment in human capital development andworking with NGOs and the private sectorcan be useful first steps. Technical assistanceappears more effective after reforms take offand can help lay the basis for capital invest-ment and service delivery interventions.35

When domestic political processes aremanifestly inequitable and corrupt, donorscan try to support moves toward a more equi-table revenue collection and allocation; decen-tralization to lower levels of government,which can challenge central control; and thestrengthening of community-based organiza-tions, the media, and domestic entrepreneur-ship, which can help create a middle class witha voice and a stake in better governance.

Improving the allocation of aid. The distri-bution of aid matters as well. A lively debatehas taken place in recent years on aid alloca-tion criteria. Burnside and Dollar (2000) andCollier and Dollar (2001, 2002) found thataid was more effective in reducing poverty ifit was allocated to countries that followedgood policies and had good institutions. Theycalculated that reallocating actual aid pro-vided in 1996 across countries to maximizepoverty reduction according to their formula


would have led to directing aid to roughly 20instead of the 60 countries considered, andlifted twice as many people out of poverty.36

Their findings have been questioned byHansen and Tarp (2001) and others, whoargued that their analysis does not takecountry conditions into account and is notrobust to different specifications. If aid effec-tiveness varies across countries not becauseof policies but as the result of different coun-try circumstances, such as climate, a differentaid allocation rule would maximize thepoverty impact of foreign aid.37 Cogneauand Naudet (2004) suggested an alternativerule for aid allocation and showed that gainsin poverty reduction similar to those foundby Collier and Dollar could be obtained if aidwas directed to countries that have greaterstructural disadvantages (geographic, histor-ical, or economic, as discussed in chapter 3).The resulting allocation would spread therisk of poverty more evenly across theworld’s population, while reducing globalpoverty almost as much as the allocationproposed by Collier and Dollar.

In sum, an equity perspective suggeststhat an approach that does not take a coun-try’s circumstances into account is likely toignore important information about need.But an approach that ignores aid effective-ness does not lead to expanded opportuni-ties. To contribute toward an equalization ofopportunities across the world’s individuals,aid should be targeted where the probabilityis greatest that it effectively reaches thosewith the most limited opportunities—thepoorest of the poor, in opportunity terms.That clearly depends on the poverty anddeprivation levels in each country and on itsgovernment’s ability and political commit-ment to deliver the aid where and how it isintended. But more research is needed tofully understand the causal mechanisms.

In practice, recent research showed thatmany donors indeed seem to rely on bothgood policies and poor initial conditions. Astudy of 40 donor agencies by Dollar andLevin (2004) found that aid was positivelycorrelated with a measure of good policiesand with per capita GDP, and the agenciesthat focused the most on good policies alsodirected their aid to poor countries. However,some fragile states (“aid orphans”) receive less

aid than predicted by their policy and institu-tional strength, mostly because of dispropor-tionately low flows from bilateral donors,while others (“aid darlings”) received more.38

Increasing aid levels. Conditional on effec-tiveness and distribution, levels of aid domatter. Aid levels fell between 1990 and 2001both as a share of rich countries’ grossnational income (GNI) and in nominalterms. Calls for more aid to help countriesachieve the MDGs have resonated loudly inrecent international gatherings. At the 2002International Conference on Financing forDevelopment in Monterrey, rich countriescommitted to increasing their aid flows sig-nificantly. Net aid flows indeed increased sig-nificantly in 2002–04 in nominal and realterms, reaching $78 billion.39 Three majorfactors were behind these increases: continu-ing growth in bilateral grants (but with alarge share going to technical cooperation,debt forgiveness, emergency and disasterrelief, and administrative costs); the provisionof reconstruction aid to Afghanistan and Iraqby the United States (in 2004, $0.9 billion toAfghanistan and $2.9 billion to Iraq); and thedepreciation of the U.S. dollar. While therewas a small increase in new developmentassistance to Sub-Saharan Africa in 2003,even after accounting for debt relief andemergency assistance, Highly Indebted PoorCountries (HIPC) received less in real termsin 2004 than the year before. On the positiveside, the International Development Associa-tion, the soft-lending arm of the World Bank,recently received a replenishment for 2006–8,which is at least 25 percent higher than theprevious one and represents the largest fund-ing increase in two decades.

These recent increases notwithstanding,aid flows remain small not just in relation toneed but also in comparison to domestichuman development and safety net programsthat aim to equalize opportunities and ensureagainst deprivation. Such programs generallyaccount for more than 10 percent of GDP indonor countries. Official development assis-tance (ODA), by contrast, was only 0.25 per-cent of donor countries’ GNI in 2003. OnlyDenmark, Luxembourg, the Netherlands,Norway, and Sweden meet the U.N. target ofproviding ODA equal to or greater than 0.7



percent of GNI. Many countries are not ontrack to meet their Monterrey commitments(table 10.1).

Aid also is low in comparison with otheruses of public resources. Agricultural subsi-dies, for instance, were almost five timeslarger than aid in 2002. Japan, the EuropeanUnion, and the United States had subsidiesequal to 1.4, 1.3, and 0.9 percent of GDP andaid of 0.23, 0.35, and 0.13 percent respectively(figure 10.2). Rich countries should deliveron their Monterrey commitments; this alonewould add around $18 billion to develop-ment assistance by 2006. To make furtherprogress toward the 0.7 percent goal, coun-tries could establish intermediate targets for2010. But again, higher aid that is poorlyspent, supports corrupt regimes, or under-mines domestic accountability can hinder,rather than support, greater equity.

Table 10.1 ODA as a share of GNI, 2002, 2003, and simulation for 2006

ODA as % ODA as % ODA as % of GNI Net ODA 2003 Net ODA 2004 of GNI of GNI Simulation

Country ($ millions) ($ millions) 2003 2004 2006

Austria 505 691 0.20 0.24 0.33

Belgium 1,853 1,452 0.60 0.41 0.64

Denmark 1,748 2,025 0.84 0.84 0.83

Finland 558 655 0.35 0.35 0.41

France 7,253 8,475 0.41 0.42 0.47

Germany 6,784 7,497 0.28 0.28 0.33

Greece 362 464 0.21 0.23 0.33

Ireland 504 586 0.39 0.39 0.61

Italy 2,433 2,484 0.17 0.15 0.33

Luxembourg 194 241 0.81 0.85 0.87

Netherlands 3,981 4,235 0.80 0.74 0.80

Portugal 320 1,028 0.22 0.63 0.33

Spain 1,961 2,547 0.23 0.26 0.33

Sweden 2,400 2,704 0.79 0.77 1.00

United Kingdom 6,282 7,836 0.34 0.36 0.42

EU members, total 37,139 42,920 0.35 0.36 0.44

Australia 1,219 1,465 0.25 0.25 0.26

Canada 2,031 2,537 0.24 0.26 0.27

Japan 8,880 8,859 0.20 0.19 0.22

New Zealand 165 210 0.23 0.23 0.26

Norway 2,042 2,200 0.92 0.87 1.00

Switzerland 1,299 1,379 0.39 0.37 0.38

United States 16,254 18,999 0.15 0.16 0.19

DAC members, total 69,029 78,569 0.25 0.25 0.30

Source: OECD-DAC (2004).Note: DAC = Development Assistance Committee; EU = European Union; GNI = gross national income; ODA = official developmentassistance.

Japan EuropeanUnion

UnitedStates

All OECD-DACcountries

Aid

Agriculturalsubsidies

Percent of GDP, 2002

Aid and agricultural subsidiesrelative to GDP in OECD-DAC countries

0

1.6

1.2

0.8

0.4

Figure 10.2 More subsidies than aid

Sources: OECD-DAC (2004) and OECD (2003).

Additional debt relief. Aid should not beundermined by debt payments. Multilateraldebt, the largest share of debt for the HIPC,is the result of loans received in the 1980s,and new loans, while generally on moreconcessional terms, continue to add to thedebt burden. Supporters of debt relief arguethat debt payments divert scarce resourcesfrom health and education and other pro-poor spending.

There has been progress in the last decade.In 1995, debt relief was not on the agenda ofinternational organizations, partly because offinancing issues and partly because of con-cerns about creating a moral hazard (if debtsare forgiven, governments of borrowing coun-tries may think they are really not expected torepay). Over the following five years, thanks toa strong grassroots mobilization in rich coun-tries, effective research on the impact of debtand committed leadership in some rich coun-tries and the World Bank, the HIPC Initiativewas launched and then expanded. As of March2005, 27 countries had received debt reliefexpected to amount to about $54 billion overtime, up from $34.5 billion at the end of 2000.The ratio of debt service to exports for HIPChas declined roughly by half, to 15 percent.Poverty-reducing expenditures in the 27 coun-tries that receive HIPC assistance are estimatedto have increased from 6.4 percent of GDP in1999 to 7.9 percent of GDP in 2003.

Even so, many countries continue to bearan unsustainable debt burden, and moreneeds to be done. The agreements reached inOctober 2004 to extend the HIPC Initiativeand in June 2005 to grant 100 percent debtcancellation of the debt owed to the AfricanDevelopment Bank, IMF, and World Bank to18 countries are important steps.40 This andany further debt relief should truly be addi-tional rather than substitute fresh aid. Fur-ther debt relief should also be accompaniedby careful consideration of debt sustainabilityissues, including increasing grants for verylow-income countries, to avoid the buildupof unsustainable debt in the future.

Innovative mechanisms to fund developmentassistance. Several innovative mechanismsto expand development assistance are underdiscussion, including the InternationalFinancing Facility (IFF), global taxes, and vol-

untary contributions. The IFF would makefuture aid available for immediate use (front-load aid) and possibly reduce volatility. It isan option for some donors, such as Franceand the United Kingdom, given theiraccounting and legislative frameworks, butnot for others, who would not be able tomake long-term commitments or considerthem off-budget. Even when feasible, the IFFwould move aid off-budget in the short term,but it would expand financing for develop-ment only if it increased overall aid levelsrather than simply shift future aid forward.

Proposals involving global tax instrumentshave also been advanced, including a “Tobin”tax on short-term capital movements; taxesrelated to pollution, such as a global carbontax, an international aviation fuel tax, and amaritime pollution tax; taxes on arms sales;and surcharges on multinational profits andon value-added or income taxes. These pro-posals would need to be assessed on the basisof the revenues they could generate, their effi-ciency, collectability, feasibility, and not leasttheir impact on equity.

Voluntary contributions from individuals,corporations,private foundations,and NGOs—another source of development assistancealongside public aid—are increasing. Buteffectiveness is an issue for private assistancetoo. As seen for the December 2004 Asiantsunami, private charity can be mobilizedfaster than public resources. But private con-tributions are influenced by press coveragemore than actual need; contributors weremuch less generous for the Iranian earth-quake that hit in February 2005, which wasvirtually ignored in the news. Moreover, lackof coordination, fragmentation, and infra-structure bottlenecks—such as bad roads anda lack of electricity and telecommunications,which cannot generally be alleviated throughprivate charity—can hinder its effectiveness.Moreover, alignment with recipient countrystrategies needs to be ensured.

Transitions to greater equityEquity-enhancing changes in global policiesand institutions come about through action bygovernments and coalitions of governments—often within international fora, informedleadership and grassroots mobilization,analysis and policy research to inform alter-


natives, and networks that disseminate thosealternatives. This section looks at some exam-ples illustrating the change processes; it doesnot attempt to be comprehensive or to assessthe weight of individual factors.

Examples include developed-country gov-ernments that take initiative unilaterally—such as the countries that have alreadyreached the target of 0.7 percent of GNI fordeveloping assistance or that cancelled a largeportion of the debts owed to them by thepoorest countries—as well as governmentsacting jointly to form coalitions for change.The latter are becoming more frequent intrade negotiations, in which a group of largedeveloping countries (including Brazil, China,and India) is spearheading proposals forgreater trade liberalization.

A way to spur equity-enhancing policychanges by developed countries is to accom-pany calls for change with tracking mecha-nisms. The eighth MDG relates to greaterprovision of aid and debt relief and moreequitable trade policies. Progress toward thisgoal has been reviewed in September 2005 aspart of the Millennium Summit+5.

Another exercise to monitor rich coun-try policies, conducted by the Center forGlobal Development and Foreign Policymagazine, is the Commitment to Develop-ment Index. The index examines more indi-cators than the eighth MDG, includingenvironment, security, investment, andtechnology (Center for Global Develop-ment 2004). While there are questionsabout the methodology, particularly onaggregating scores in various areas, theindex exposes how some countries do betterin some areas than in others—Norway, forinstance, does well on aid but poorly ontrade; Switzerland scores poorly on tradebut does better on environment; the UnitedStates scores poorly on environment buthas, with Canada, the most favorablemigration policies—and how all countrieshave considerable opportunities to improvetheir policies.

Citizen mobilization. Citizen mobilization,combining both grassroots and middle-classinterest groups across countries, has grown inrecent years. In some cases, an internationalsocial movement, network, or alliance has

emerged to try to influence the global agenda.An example is the launching of the EnhancedHIPC Initiative in 2000. The original HIPCInitiative benefited some countries, butprogress was slow and there were severalproblems. By 1999 these were largely recog-nized, but an expanded initiative needed togarner support in creditor countries and inthe World Bank and IMF governing commit-tees, because it required additional funding.The Jubilee 2000 campaign, which combinedawareness of the pernicious effects of exces-sive debt with a call to debt forgivenessinspired by the Christian Jubilee idea, mobi-lized hundreds of thousands of people incountries such as Germany, Italy, the UnitedStates, and United Kingdom. The govern-ments of these countries took notice andfinally agreed to various actions, expandingthe HIPC Initiative and canceling bilateraldebt. Other examples of pressure by civilsociety organizations leading to changes inrules are the campaigns to reform WorldBank policies on indigenous peoples, resettle-ment, and other safeguards.

In a second set of cases, internationalrules already exist on paper, and socialmovements bring them into effect by mak-ing them visible and insisting that they beimplemented. In many cases, this processhappens at the country level, but it involvesan interaction with global rule and policychanges. The ethical trade initiatives dis-cussed earlier are citizen mobilizations toenforce global and local laws. Similarly,efforts by indigenous movements, NGOs, andother activists ensured that ILO Covenant169 on indigenous peoples was recognizedas having legal weight (in practice) in vari-ous countries. Experience shows that citizenmobilization is most effective when itbuilds broad-based coalitions for changeacross countries and groups.

But citizen mobilization also poses risks.Civil society movements may partly counterunequal formal channels, but they arehighly imperfect mechanisms of aggregat-ing voice, and their accountability is oftenunclear. In recent years, there have beeninstances of NGO campaigns that have ledto perverse outcomes, such as donors with-drawing from infrastructure and resettle-ment projects only to see governments



move ahead anyway without internationalmonitoring of social and environmentalconsequences.

Analysis and research. Socioeconomic analy-sis and policy research also contribute tomaking particular domains of inequityobjects of public debate and action. Globalanalysis of gender discrimination and miss-ing girls and women (box 2.9) has fosteredpublic action to redress gender inequities. Exante analysis and policy research are also vitalingredients for informing the design of policyproposals. A vast body of recent research,including serious impact evaluations, hasfocused on efficient and effective ways toachieve the MDGs. The more research is con-ducted by and with developing-countryresearchers, the more likely it is that its resultswill inform policymaking.

Some of these key elements have beenmissing in failed attempts at change. Analysisand policy research is carried out and techni-cal solutions are proposed; the political will toimplement them is missing, however, becausepolitical leaders do not think the issue isimportant or coalitions are not formed thatensure sufficient support. In other cases,grassroots mobilization is strong, but it lackswell-developed, implementable proposals forreform. Indeed, some NGO campaigns haveled to perverse outcomes, as when interna-tional organizations have withdrawn supportfor projects under international criticismonly to see governments proceed withoutinternational monitoring of social and envi-ronmental safeguards.

International organizations. Internationalfinancial institutions can help bring aboutglobal equity-enhancing action throughsetting agendas and providing a focal pointfor international negotiations. Their dis-pute settlement and enforcement mecha-nisms help ensure that their policies areimplemented. But the governance struc-tures of the World Bank and IMF have notevolved in line with the increased size androle of emerging market, developing, and

transition countries in the world economy.Moreover, small and low-income countrieshave a limited role in their decision-makingprocesses. Developed-country governmentshave a majority of the votes on the boardsof the IMF and World Bank, and two execu-tive directors represent more than 40African countries.

Several options to enhance voice in theIMF and World Bank have been explored,but limited progress has been made. InApril 2005, the ministers of the intergov-ernmental Group of Twenty-Four urgedthat a new quota formula be developed(voting rights depend on quotas), whichwould give greater weight to measures ofgross domestic product measured in pur-chasing power parity terms. They also sug-gested that, in order to strengthen the voiceof small and low-income countries, basicvotes should be increased to restore theiroriginal share of total voting power.41 Mak-ing progress on enhancing the voice andparticipation of developing countries in thedecision-making processes of the WorldBank and IMF is of fundamental impor-tance to increase the legitimacy of interna-tional financial institutions and enhancetheir effectiveness in fostering greater globalequity. The 13th General Review of IMFQuotas provides an important opportunityto make progress on issues of quotas, voice,and participation.

SummaryIn sum, global actions can play a key role inredressing inequitable rules and helpingequalize endowments. The rules that governmarkets for labor, goods, ideas, capital, andthe use of natural resources need to becomemore equitable. Domestic action to buildthe endowments of the poor can be sup-ported through aid, but not if aid is poorlyspent, supports corrupt regimes, or under-mines domestic accountability. Changeswill require, above all, greater accounta-bility at the global level, with greater repre-sentation of poor people’s interests in rule-setting bodies.

curiosity (similar to what happens withopen-source software). Leads thatemerge could then feed into any of theother schemes for the next stages ofdevelopment and clinical testing.4

Providing patent protection in poorcountries for drugs primarily for their mar-kets would not by itself provide sufficientincentives, because their purchasing poweris very low. But, even a small increase inmarket-based research incentives could be auseful part of a larger strategy to addressthe treatment of diseases specific to thedeveloping world.

Other diseases have global incidenceand worldwide markets, and they are animportant cause of death and disabilityamong the poor. In the high-mortalityregions of the world, cardiovascular diseaseis estimated to cause a greater share of thetotal disease burden than malaria and othertropical diseases combined.5

Although many people in poor coun-tries suffer from global diseases, they are aninsignificant part of the commercial mar-ket. Estimates suggest that currently almosthalf of the world’s people live in countriesthat together represent less than 2 percentof global spending on drugs for cardiovas-cular disease.6 Because of the great asym-metries in markets, many of the poor couldbe allowed generic access to importantclasses of drugs without damaging researchincentives. The foreign filing license approachdescribed below is a feasible way to attainthis outcome.

Legally binding commitments notto enforce patent rightsThe proposal considered here7 would haveinventors in developed countries makelegally binding commitments to their owngovernments not to enforce patent rights incertain pharmaceutical markets. Thesemarkets would be defined as those togetherrepresenting the bottom, say, 2 percent of

Some diseases, such as malaria, mainlyaffect poor countries, but they havereceived little R&D investment, and

few treatments are available. Much atten-tion has been devoted in recent years to cre-ating the right incentives and structuringfinancing to increase R&D investment indrugs for diseases that affect poor coun-tries, for which commercial potential doesnot provide enough stimulus.1 Policy initia-tives include the following:

• Increasing research through public sec-tor institutions—for example, programscoordinated by the U.S. National Insti-tute of Allergy and Infectious Diseases,the World Health Organization’s SpecialProgramme for Research and Trainingin Tropical Diseases, and the nonprofitDrugs for Neglected Diseases Initiativefounded by Medécins Sans Frontiérès.

• Establishing public-private partnerships,such as the Medicines for Malaria Ven-ture and the Malaria Vaccine Initiative,the International AIDS Vaccine Initia-tive, and the Global Alliance for Tuber-culosis Drug Development.

• Designing a purchase commitment fornew vaccines (“AdvancedMarkets”). Spon-sored by the Bill and Melinda GatesFoundation, the groundwork has beenlaid to create markets by having donorscommit, in advance, to paying part ofthe cost at a guaranteed price for a new,as yet undiscovered, vaccine. This wouldgive firms an incentive to invest in thisarea.2 (A similar initiative has been pro-posed for agricultural research relevantto developing countries.3)

• Developing an open-source approach toearly-stage tropical diseases research.The idea is to harness the expertise andresources of academic scientists, stu-dents, public sector researchers, andothers who may be happy to spend someof their time doing research on tropicaldiseases either for altruism or scientific

global drug sales in each disease class (seefigure below).

Along the horizontal axis of the figure aredisease classes, listed with those concentratedin poor countries toward the left, and thosewith worldwide incidence toward the right.Along the vertical axis are countries orderedby per capita income. The white area showsthe “generic region” that would be created bythe policy. Within the generic region, firmswould be able to manufacture and trade ingeneric products without any political or pro-cedural complexity arising from the patentsystem. The generic region would be recalcu-lated each year to accommodate changes inincome and the evolution of markets.

Because diseases to the left are moreconcentrated in developing countries, the 2percent of global markets cutoff is reachedat lower levels of real GDP per capita. Itmay seem counterintuitive to propose dif-ferentiating in this way, but it is preciselyfor diseases that are concentrated in devel-oping countries that some incentive forproduct development may need to comefrom sales in the developing world.

Balancing access to medicines for the world’s poor with incentives for pharmaceutical innovationWhat’s the best way to expand access to drugs in developing countries, while preserving incentives for pharmaceuticalresearch? A possible solution entails recognizing that medicine markets are far from uniform and that both the nature ofdiseases and the income of countries matter for access and incentives.

drug accessf o c u s 7 o n

$5,000

Countries,ordered by

real GDPper capita

0Concentratedin LDCs Global

Generic region

Rest of the world(no change)

Disease classes

The proposed generic region

Source: Lanjouw (2004).

that relate to the policy—about where toenforce patent rights—would arise onlyafter products have reached the market. Toensure compliance, the patent holder wouldlose the right to enforce the domestic patenton the same product if the holder were tobreak the commitment and begin anenforcement action in one of the pro-scribed markets.

Implementation would need to becoordinated across the developed countriesthat have pharmaceutical research activity,including, at least, Canada, Europe, Japan,and the United States. The policy wouldrequire legislation to amend the patentcode in each country. In the United Statesand the United Kingdom this wouldinclude adding an inventor declaration toan existing foreign filing license process;other countries would need to put a for-eign filing license provision into theircodes.8 The classification of countries anddisease classes could be carried out by aninternational organization and reviewedannually.

Developing countries would not berequired to take any action to implement

The poorest countries falling below thedashed line would be allowed generics onall pharmaceuticals. Countries higher up,such as India, would have a mixed situa-tion. They would be in the TRIPS envi-ronment for diseases concentrated in thedeveloping world, while in the genericregion for more global diseases. For mar-kets in the gray area above the curved line,the policy as no effect. Both the responsi-bilities and the flexibilities of TRIPSremain unchanged.

The size of the generic region dependson two parameters: the ceiling income level(here $5,000) and, more important, theglobal sales cutoff (here 2 percent).

The proposal would be implemented byhaving inventors in developed countriesmake a legally binding commitment totheir own governments not to enforcepatent rights in the generic region, as partof obtaining a license to make foreignpatent filings (the foreign filing license).Firms would continue to obtain patentswherever they like, and no decisions relatedto the policy would be needed at the time ofapplying for a patent. Instead, decisions

this policy. They would continue to takesteps to comply with TRIPS and any bilat-eral treaty obligations in accord with theircurrent plans. The countries in the genericregion could be treated as one country fromthe viewpoint of patents. Production couldbe based in any one country and drugscould be exported to all other countries inthat group, without any costs associatedwith patents and compulsory licensing.

So, if any country in the generic regionhad the ability to produce a given drug, thenall other countries in the region could takeadvantage of its production capacity. Thiswould help get around the problem thatmost small countries do not have the capac-ity or market size to make domestic produc-tion of generic drugs a viable activity.

Firms have been willing to make a vol-untarily commitment not to exercisepatent rights in the poorest countries. Theforeign filing license proposal discussedhere would take that commitment andconvert it into a reliable part of the globalrules-based system.

Source: Lanjouw (2002, 2004).

Focus on drug access 225

This report has argued that equity has acentral place in the interpretation of devel-opment experience and in the design ofdevelopment policy—and that this placehas been inadequately understood andundervalued in much current thinking. Wedo not propose, however, yet another newframework for development. Instead, recog-nizing the importance of equity (that is,equality of opportunity and the avoidanceof absolute deprivation) implies the need tointegrate and extend existing approaches. Inthis epilogue, we seek to place the analysisand messages of the report in the context ofsome of the major contemporary strands ofthinking and action in development.

Four broad strands of thinking havebeen at the core of development discourseand practice over the last three decades orso: the central role of markets as resourceallocation mechanisms, the importance ofhuman development, the role of institu-tions, and a focus on empowerment.

The first strand emphasizes the superior-ity of markets over central planning asbroad mechanisms to allocate resources anddetermine the evolution of economic activ-ity. This has long been understood in eco-nomics, but there was a time when it was aminority view among development econo-mists.1 The situation had definitely changedby the 1980s, as first India and then Chinamoved away from planning, and the impor-tance of incentives in determining individ-ual behavior (as consumers, producers, andregulators) became more widely under-stood. The rapid and sustained growth thatfollowed in those two countries underlinedthe point. In the 1990s, the economic tran-sition away from planning in the formerlycommunist states of Eastern Europe andCentral Asia dispelled any serious view that

development was possible without marketsand the private sector.

Although the resulting “Washington Con-sensus” is sometimes interpreted as anti-state, this is not the main message thatsurvives after more measured consideration.Instead, just as events in the 1990s confirmedthat markets were essential for development,they also showed that good governments areessential for well-functioning markets. Mar-kets operate within a framework determinedby institutions, and they work only as well asthose institutions do. They work best, there-fore, when a capable state maintains orderwithin the rule of law, provides effectiveregulation, macroeconomic stability andother public goods, and corrects other mar-ket failures.

The second strand sees human develop-ment as central to the development process,through the expansion of the skills, health, andcapacity of all people to engage in social andeconomic activities and to manage the risksthey face. Although the World DevelopmentReport 1980 was on Human Development,2

U.N. agencies—notably the UNDP in theirseries of Human Development Reports(United Nations 2003)—later took the lead inputting these concerns at the center of thedevelopment agenda. In this, they have beenfollowed (rightly) by the whole developmentcommunity.

For the World Bank, the 1990 WorldDevelopment Report on Poverty3 markedthe beginning of a multiyear process ofmaking poverty reduction the overarchingobjective of the institution, building onthese first two strands of developmentthought. The 1990 Report argued thatpoverty reduction required a two-partstrategy—employment creation throughmarket-based growth; and expansion of

226

Epilogue

human capital, especially through broad-based provisioning of social services.

During the 1990s, the third and fourthstrands of thought rose to prominence. Thethird strand emphasized the role of “institu-tions” in development, building both ontrends in academic thought and on devel-opment practice across many organiza-tions. It reflected the acknowledgment thatmarkets, however important, do not workin a vacuum. They need rules and the insti-tutional enforcement of those rules. Theemphasis on institutions took a variety offorms: a focus on the costs of corruption; abroader concern with governance; supportfor judicial reform; and a greater practicalunderstanding of the need for well-designed, accountable, and effective publicregulation of privatized monopolies.

The fourth strand sought the empower-ment of the people for whom developmentwas supposedly taking place. If the centralgoal of development is poverty reduction,the poor should have a great deal of voiceover its directions. If development needsmarkets, and markets need institutions, itshould clearly matter how those institutionsare governed. If power helps determine theoutcomes in market and government pro-cesses alike, the distribution of that powerover the population must be important fordevelopment. In practical terms, the em-phasis on empowerment has sought greaterparticipation of the poor in projects affect-ing them, a greater preoccupation with thepolitical economy of support for reforms,and explorations of the role of culture indevelopment.

Several World Development Reportshave sought to integrate the third and fourthstrands of thought: the 1997 Report on the“State in a Changing World,”4 the 2002Report on “Building Institutions for Mar-kets,”5 and, emblematically, the millennial2000/01 Report on “Attacking Poverty.”6 The2001 Report argued that poverty reductionrequired an expansion in the opportunitiesof the poor (notably through market-oriented growth), the empowerment of thepoor, and measures that provide security forthe poor. At the World Bank, this synthesiswas crystallized in a Strategic Framework fordevelopment, which consisted of two pillars:

building a good climate for investment andempowering the poor.7

The first pillar combined the strands ofthinking on the primacy of markets and onthe centrality of institutions. It argued thatonly by having governance institutions thatwere at once effective and accountablecould markets generate the best possibleresults for investment and growth. Thistheme was explored in the 2005 Report, “ABetter Investment Climate for Everyone.”8

The second pillar was also a merger ofsorts: in seeking the empowerment of poorpeople—who should be seen as the driving sub-jects, not as passive objects of development—it combined thinking on human develop-ment, institutions, and empowerment. The2004 Report, “Making Services Work forPoor People,” explored these themes in thedelivery of basic services.9

Although the various elements of think-ing and policy are complementary—andindeed have been considered elements of a“comprehensive” or “holistic” developmentprocess—the narratives associated with thedifferent strands have often suffered, inpractice, from two important limitations.One is a tendency to compartmentalizepoverty. The second is to treat actions in thevarious realms as separate. There is a ten-dency to assign market-related and macro-policies to macroeconomic managers andtrade ministries, as though the “investmentclimate” concerned only rich people or asthough the poor stood to benefit only indi-rectly from the trickle-down effects of theinvestments by the rich today.

Conversely, it sometimes seems asthough empowerment would have noimpact on the quality of institutions, on theinvestment opportunities of the poor, or onthe economy’s growth process. According tothis view, empowerment should be the pre-serve of well-meaning NGOs and socialdevelopment folk, of little import for eco-nomic performance.

Such a separation of the two pillars—for theinvestment climate and for empowerment—is profoundly misguided. The analysis inthis report suggests that the root causes ofpoverty are to be found in the combineddeprivations of power and investmentopportunities. Lack of incomes, lack of

Epilogue 227

access to services, lack of assets—thesedeprivations go together with lack of voice,lack of power, lack of status. Public actioncould enhance the investment capabilitiesof those who have limited opportunities byinvesting in their human capital and in theinfrastructure they use and by ensuringfairness and security in the markets inwhich they transact. And if public actionfails to do that, it is because it somehow hasbeen decided otherwise. In that case, thegovernment will invest in expensiveschools or universities, rather than in thoseused by the poor, for example. It will notenforce tax collection, rather than buildrural roads. It will allow banks to retain adegree of market power and lend to thefriends of the government, rather thanallow for entry and promote competitionthat forces intermediaries to seek the great-est returns on capital. Observed policiesthat fail to address inefficient inequities arethe result of political choices, implicitly orexplicitly.

Such failures in public action, which arisefrom and perpetuate inequity, are alsoinimical to prosperity. Those who have noopportunities cannot contribute to thedevelopment of their countries. Their poten-tial talents are wasted, and capital, land, andother resources are used in inferior ways.Unequal control over resources reinforcesthe unequal concentration of power, and thisis reflected in worse governance institutions:public service delivery agencies are notpushed to become more accountable. If allthe power brought to bear on regulators isthat of the friends of the regulated, the qual-ity of regulatory agencies is not likely toimprove much. Police forces and judicial sys-tems will not treat everyone the same way.And so on. These institutional failures onlyadd to the negative effects of inequity ondevelopment.

Government policies are what they are—from Mali to Chile—because someone ismaking them. No group is powerless, unlesssome other group is powerful. If aninequitable distribution of opportunitiesmeans that the investment climate for largegroups is poor, this is intimately linked tothe lack of power in those groups to affectthe decision-making processes that could

lead to changes in that distribution overtime. And if power is imbalanced, it isbecause wealth and economic opportunitiesare uneven. Inequality traps are vicious cir-cles, with economic and political inequali-ties mutually reinforcing one another.

This report has argued that policy andinstitutional reforms can help break theseinequality traps, and turn the vicious circlesinto a virtuous process of greater equality ineconomic opportunities reinforcing greaterpolitical equality, and vice versa. Reformscan do this in many ways, which are closelyrelated to the four strands of thinking dis-cussed above. Interventions that buildgreater human capacities for those with themost limited opportunities (generally thepoor) will prepare them to be more eco-nomically productive and more politicallyeffective. Processes that redistribute accessto land, or to infrastructure services, orindeed to justice, can add both to empower-ment and to the investment opportunitiesof the poor. And promoting fairness inmarkets is all about improvements in thequality of institutions that support andcomplement markets in ways that broadenaccess and ensure equitable rules.

This is consistent with the twin pillars ofa better investment climate and greaterempowerment for the poor. It makes it clearthat—for most people in the developingworld, and certainly for the poor—it is notpossible to have one without the other. Agood investment climate is about real eco-nomic opportunities. Equity is about level-ing the playing field so that opportunitiesare available on the basis of talent andefforts, rather than on the basis of gender,race, family background, or other predeter-mined circumstances. A level economicplaying field is not sustainable without alevel political playing field, and vice versa. Ifwe want a better investment climate foreveryone, we want empowerment. Thecombination of both implies equity.

These issues apply with equal force at theglobal level. The extraordinary inequalitiesin opportunity faced by individuals born indifferent countries reflect different politicaland economic histories across nations.While domestic policy is undoubtedly fun-damental, global interactions help shape


Epilogue 229

the context for national economic andsocial advance. The Monterrey Consensusexplicitly emphasizes the need for a com-pact between rich and poor societies if theMillennium Development Goals are to beachieved. It recognizes the role of action byrich countries, especially in the areas of aidand trade. For aid, this is reflected in thequest to change the donor-recipient rela-tionship from one of giving to one of part-nership, with developing countries clearlyin the lead in designing their policies andinstitutions.

This report underlines the importance ofthat compact and a more equal interna-

tional partnership. But it also highlightsinequalities in the processes forming therules of the game in the international play-ing field. Inequalities in economic andpolitical power in the global arena influencethe design of rules in ways that oftenrestrict, rather than expand, the opportuni-ties of poorer countries—and, even more,of poorer groups within these countries.Just as in the domestic context, therefore,equity and efficiency in the internationalarena are more likely to be attained by re-forms that enhance the power and broadenthe economic access of the countries wherethe world’s poor live.

This Report draws on a wide range of World Bank docu-ments and on numerous outside sources. Backgroundpapers were prepared by Martin Andersson, ArmandoBarrientos, Carles Boix, Leila Chirayath, Stijn Claessens,Klaus Decker, Ashwini Deshpande, Leopoldo Fergusson, JoséFernández-Albertos, Christer Gunnarsson, EmmanuelGymi-Boadi, Karla Hoff, Markus Jäntti, José AntonioLucero, Marco Manacorda, Siobhan McInerney-Lankford,Joy Moncrieffe, Enrico Perotti, Vibha Pinglé, Pablo Querubi,Martin Ravallion, Michael Ross, Juho Saari, Rachel Sabates-Wheeler, Carolina Sánchez-Páramo, Norbert Schady,Andrew Shepherd, Milena Stefanova, and Juhana Vartiainen.Background papers for the Report are available either on theWorld Wide Web http://www.worldbank.org/wdr2006,under Background Papers or through the World Develop-ment Report office. The views expressed in these papers arenot necessarily those of the World Bank or of this Report.

Many people inside and outside the World Bank gave com-ments to the team. Valuable comments and contributions wereprovided by Nisha Agrawal, Asad Alam, Sabina Alkire, SudhirAnand, Cristian Baeza, Gianpaolo Baiocchi, Catherine Baker,Judy L. Baker, Giorgio Barba Navaretti, Catherine Barber,Jacques Baudouy, Gordon Betcherman, Lisa Bhansali, Vinay K.Bhargava, Amar Bhattacharya, Nancy Birdsall, Andrea Bran-dolini, John Bruce, Barbara Bruns, Donald Bundy, Luis FelipeLópez Calva, Shubham Chaudhuri, Martha Chen, ShaohuaChen, Aimee Christensen, Denis Cogneau, Giovanni AndreaCornia, Anis Dani, Roberto Dañino, Jishnu Das, Klaus Decker,Arjan de Haan, Klaus Deininger, Asli Demirguc-Kunt, KemalDervis, Jean-Jacques Dethier, Shanta Devarajan, Peter A.Dewees, Charles Di Leva, Mala Escobar, Antonio Estache, JoanMaria Estebán, Shahrokh Fardoust, Massimo Florio, DavidFreestone, Adrian Fozzard, Teresa Genta Fons, Vivien Foster, M.Louise Fox, Sebastián Galiani, Alan Gelb, Alec Gershberg, ElenaGlinskaya, Delfin Go, Carol Graham, Maurizio Guadagni,Susana Cordeiro Guerra, Isabel Guerrero, David Gwatkin, JeffHammer, Patrick Heller, Amy Jill Heyman, Bert Hofman,Patrick Honohan, R. Mukami Kariuki, Christine Kessides,Homi Kharas, Elizabeth King, Larry Kohler, Jacob Kolster,Somik Lall, Ruben Lamdany, Danny M. Leipziger, VictoriaLevin, Sandy Lieberman, Peter Lindert, Amy Luinstra, XubeiLuo, Bill Maloney, Katherine Marshall, Siobhan McInerney-

Lankford, Stephen Mink, Pradeep Mitra, Ed Mountfield, ChrisMurray, Edmundo Murrugarra, Mamta Murthi, Ijaz Nabi,Mustapha K. Nabli, Julia Nielson, Pedro Olinto, Robert O’Sulli-van, Çaglar Özden, John Page, Sheila Page, Guillermo Perry,Lant Pritchett, Agnes Quisumbing, Maurizio Ragazzi, RobinMichael Rajack, Raghuram G. Rajan, Dena Reingold, KasparRichter, Maria Estela Rivero-Fuentes, Peter Roberts, John Roe-mer, Fabio Sánchez, Milena Sánchez de Boado, Stefano Scar-petta, George Schieber, Maurice Schiff, Jesica Seacor, BinayakSen, Shekhar Shah, Tony Shorrocks, Ricardo Silveira, NisthaSinha, Milena Stefanova, Nicholas Stern, Kalanidhi Subbarao,Mark Sundberg, Rosa Alonso i Terme, Vinod Thomas, PeterTimmer, Bernice K. Van Bronkhorst, Rogier J.E. van den Brink,Rudolf Van Puymbroeck, Tara Vishwanath, Adam Wagstaff,L. Alan Winters, Ruslan Yemtsov, Nobuo Yoshida, Mary EmingYoung, Hassan Zaman, and Heng-Fu Zou.

Other valuable assistance was provided by Jean-Pierre S.Djomalieu, Gytis Kanchas, Polly Means, Nacer MohamedMegherbi, and Kavita Watsa. Christopher Neal and StephenCommins assisted the team with consultations and dissemina-tion.

Despite efforts to compile a comprehensive list, some whocontributed may have been inadvertently omitted. The teamapologizes for any oversights and reiterates its gratitude to allwho contributed to this Report.

Background PapersAndersson, Martin, and Christer Gunnarsson. “Egalitarianism in

the Process of Modern Economic Growth: The Case of Sweden.”

Barrientos, Armando. “Cash Transfers for Older People ReducePoverty and Inequality.”

Boix, Carles. “Spain: Development, Democracy and Equity.”

Chirayath, Leila, Caroline Sage, and Michael Woolcock. “Custom-ary Law and Policy Reform: Engaging with the Plurality of Jus-tice Systems.”

Claessens, Stijn, and Enrico Perotti. “The Links Between Financeand Inequality: Channels and Evidence.”

de Haan, Arjan. “Disparities Within India’s Poorest Regions: WhyDo The Same Institutions Work Differently In Different Places?”

Decker, Klaus, Siobhan McInerney-Lankford, and Caroline Sage.“Human Rights and Equitable Development: ‘Ideals’, Iissues andImplications.”

Bibliographical Note

Decker, Klaus, Caroline Sage, and Milena Stefanova. “Law or Jus-tice: Building Equitable Legal Institutions.”

Deshpande, Ashwini. “Affirmative Action in India and the UnitedStates.”

Hoff, Karla.“What Can Economists Explain by Taking into AccountPeople’s Perceptions of Fairness? Punishing Cheats, BargainingImpasse, and Self-Perpetuating Inequalities.”

Jäntti, Markus, Juho Saari, and Juhana Vartiainen. “Country CaseStudy: Finland-Combining Growth with Equity.”

Lucero, José Antonio. “Indigenous Political Voice and the Strugglefor Recognition in Ecuador and Bolivia.”

Moncrieffe, Joy. “Beyond Categories: Power, Recognition, and theConditions for Equity.”

Pinglé, Vibha. “Faith, Equity, and Development.”

Rao, Vijayendra. “Symbolic Public Goods and the Coordination ofCollective Action: A Comparison of Local Development in Indiaand Indonesia.”

Ravallion, Martin. “Why Should Poor People Care about Inequal-ity?”

Ross, Michael. “Mineral Wealth and Equitable Development.”

Sabates-Wheeler, Rachel. “Asset Inequality and AgriculturalGrowth: How Are Patterns of Asset Inequality Established andReproduced?”

Shepherd, Andrew and Emmanuel Gyimah-Boadi, with SulleyGariba, Sophie Plagerson, and Abdul Wahab Musa. “Bridging theNorth-south Divide in Ghana.”


Overview

1. The infant mortality rates are computed separately at theprovince level only, and do not take into account racial, gender, orother social differences. The life expectancy statistics are for raceand gender groups, and do not take regional or income differencesinto account. The real differences between typical individuals withthe listed characteristics are therefore likely to be understated. Inaddition, Nthabiseng’s life expectancy could be much lower if shewere to become infected with HIV/AIDS, as are many young SouthAfrican women. Data come from Day and Hedberg (2004). Pre-dicted years of schooling rely on disaggregated information—byprovince, sex, race, rural-urban location, consumption expenditurequintile, and mother’s education—from the Labor Force Surveyand the Income and Expenditure Survey for 2000, which were car-ried out by the South Africa Statistics Office.

2. Predicted monthly consumption expenditures in 2000 forpersons with those characteristics were Rand 119 ($45 adjusted forpurchasing power parity) for Nthabiseng and Rand 3,662 ($1,370)for Pieter. The average white male with a highly educated mother,who lives in Cape Town and is in the top 20 percent of the distribu-tion, happens to be in the ninety-ninth percentile of the overallincome distribution. Data come from the Labor Force Survey andthe Income and Expenditure Survey for 2000.

3. There are also income, consumption, and other differences:Sven can expect to earn $833 a month compared with the SouthAfrican mean of $207 (Nthabiseng will have $44 a month). If Svenhad been even luckier, and had been born at the same rank in Swe-den’s income distribution as Pieter occupied in South Africa, hisexpected monthly earnings would climb to $2,203. Sven will be ableto visit any country on a whim, whereas Nthabiseng and Pieter canexpect to spend hours waiting for a visa, which they may or maynot obtain.

4. In some cases, such as China’s decollectivization of agricul-ture in the late 1970s, a reform may lead to more efficiency, broaderopportunities, and yet greater (intrarural) income inequality.China’s experience—and the decompression of wages in a numberof transition economies in Europe and Central Asia—is a goodillustration of a more general point: because equity refers to fairprocesses and equal opportunities, it cannot be inferred fromincome distributions alone. Greater fairness will generally lead tolower income inequality, but not always. And not all policies thatreduce inequality increase equity.

5. Mazumder (2005).6. Other interactions between unequal opportunities and social

conditions are also of societal concern, including links betweeninequality and crime, and between inequality and health. We

review them briefly in the report, but focus on the channels of moredirect significance to equity.

7. There may be many sound economic reasons for these risk-adjusted interest rates to vary, including fixed administrative loancosts, greater informational asymmetries, and the like. The point isthat this affects poorer groups more, in ways unrelated to theirinvestment opportunities, thus leading to both greater inefficiencyand the perpetuation of inequalities.

8. These averages are based on actual episodes and refer to thetotal growth elasticity of poverty reduction, inclusive of anychanges in inequality. “Low” and “high” inequality refers to a Ginicoefficient of 0.3 and 0.6 respectively. The partial elasticity ofpoverty to growth, assuming no change in the Lorenz curve, showsa similar decline, but not to zero (see chapter 4).

9. While equity-enhancing redistribution will usually be fromricher groups to poorer groups, it could happen that “good” redis-tributions will be to groups that are not poor, especially those in themiddle. This will depend on the nature of the market failure. Forexample, the initial beneficiaries of a less-captured financial systemmay be small and medium entrepreneurs. Benefits to the poor willoccur when enhanced access to financial services among middle-class entrepreneurs translates into faster growth and job creation.

Chapter 1

1. This formulation draws loosely on Roemer (1998), but it isalso related to the works of Dworkin (1981b), Dworkin (1981a),and Sen (1985).

2. See Rawls (1971).3. See Acemoglu and Robinson (2000), Bourguignon and

Verdier (2000), and Ferreira (2001).4. This theme is one that recurs across many disciplines—in

sociology and anthropology see Bourdieu (1986), Bourdieu (1990),and Tilly (1998), in economics see Engerman and Sokoloff (2001),Bénabou (2000), and Piketty (1995).

5. See Acemoglu and Robinson (2000) for various perspectiveson the interconnections among cultural, social, and economicinequality, and Stewart (2001) for a general discussion of group-based inequalities.

6. Appadurai (2004).7. See Bourdieu (1990) on symbolic violence for more on this.8. Steele (1999).9. Appadurai (2004).10. See Rao and Walton (2004) for more on such “inequalities of

agency.”11. See also Birdsall, Graham, and Sabbot (1998) and Thorbecke

(2005).

233

Endnotes

12. “Too much” and “too little” are used here with respect to thesocial optimum. The choices are usually privately rational.

13. The role of high inequality of political power and wealth as abarrier to economic development in the modern era is a rapidlygrowing area of research. Striking results are reported in Sokoloffand Khan (1990), Engerman and Sokoloff (1997, 2002), Acemoglu,Johnson, and Robinson (2001, 2002), and Banerjee and others(2001). Overviews of the broad set of research questions are Enger-man and Sokoloff (2002), Hoff (2003), and Acemoglu, Johnson,and Robinson (2004).

14. This is so if the distribution of talents and informationalasymmetries between employers and employees are similarbetween men and women. Notice that the efficiency gain does notsatisfy the Pareto criterion. Some individual men might lose out.The efficiency criterion here is first-order dominance, under theanonymity axiom.

15. Of course, the intrinsic valuation of equity may bereflected in the appropriate distributional weights underpinningthese benefits.

16. This has been a recurrent theme in development economicsand at the World Bank. For an influential early treatment, seeChenery and others (1994).

17. World Bank (1990).18. World Bank (2001h).

Focus 1 on Palanpur

1. The study was launched out of the Agricultural EconomicsResearch Centre of the University of Delhi and Oxford University,and then continued from the London School of Economics.

2. Drèze, Lanjouw, and Sharma (1998), 51.3. Drèze, Lanjouw, and Sharma (1998), 211.

Chapter 2

1. Filmer (2004).2. See, for instance, Wilkinson (1992) and Wilkinson (2000).3. High rank, on the other hand, can be protective of health,

including enhancing one’s resistance to infectious disease.4. Fujii (2005).5. Interestingly, for each of the three population groupings, the

relative gap in infant mortality between groups rose in approxi-mately half of the countries and fell in the other half.

6. Chaudhury and others (2005).7. Araujo, Ferreira, and Schady (2004) use a new data set, con-

structed from individual record data in household surveys for 124countries. Qualitatively similar results were also obtained byCastello and Domenech (2002) and Thomas, Wang, and Fan(2002), both of which rely on the Barro and Lee (2001) interna-tional education data set.

8. See, for example, Kanbur (2000).9. See Kanbur (2000), Elbers and others (2005).10. Wan, Lu, and Chen (2004).11. These correlations hold whether or not between-group

inequality is measured on the basis of the conventional or alterna-tive methodology. They also hold after controlling for outliers(although South Africa is a particularly influential observation inthe regression of inequality on social group differences). The resultsalso hold when all the between-group shares are combined into asingle-count indicator that reflects whether a particular country

has a higher-than-median value of the between-group share for aparticular population breakdown.

12. Bourguignon and Morrisson (1998).13. Li, Squire, and Zou (1998).14. The consumption module in India’s National Sample Survey

was slightly modified between the 1993/4 round and the 1999/0,compromising both poverty and inequality comparisons over time.Deaton and Kozel (2004) provide a clear overview of the issues andtheir implications.

15. Deaton and Drèze (2002), Sen and Himanshu (2004) andBanerjee and Piketty (2003) all document increases in inequality,but by varying amounts. These estimates are all predicated on cer-tain assumptions given the fundamental problems of data compa-rability referred to above.

16. Khan and Sen (2001).17. Narayan and Yoshida (2004).18. Nepal National Planning Commission (1995–6).19. For a sociological perspective on the inheritance of inequal-

ity, see Erikson and Goldthrope (2002).20. Solon (1999).21. Mazumder (2005).22. Hertz (2001) for South Africa; Dunn (2003) for Brazil.23. Solon (2002).24. World Bank (2001h).25. Jha, Rao, and Woolcock (forthcoming).26. Appadurai (2004).27. See, for example, Fernández-Kelly (1995) on life in the U.S.

inner city.28. Appadurai (2004).29. This term comes from Nussbaum (2000) and her work on

gender discrimination and development, although the core idea hasa long intellectual history in social science.

30. It is also important to note that the appearance of “internaliz-ing disadvantage”(for example, working slowly, being tardy) can actu-ally be a covert strategy used by marginalized groups to subvert sys-tems over which they otherwise have little influence. See Scott (1986).

31. If income inequality, a static measure, is a good proxy forsocial mobility, a dynamic concept, then these theories find supportin the large literature on income inequality and crime. See, forexample, Demombynes and Özler (2005) for a discussion of thesubject at the country level, and Fajnzylber, Lederman, and Loayza(2000), among others, on the cross-country level relationshipbetween income inequality and crime.

32. See Narayan (2002) for a categorization of approaches andcase studies.

33. See Petesch, Smulovitz, and Walton (2005).34. Gibson and Woolcock (2005); for methodological details on

the broader research project of which this study is a part (and theevidence base on which the assertions made here rest), see Barron,Smith, and Woolcock (2004).

35. See also Rao (2005) on the related idea of such spaces as“symbolic public goods.”

36. Maynard (1966); Casagrande and Piper (1969).37. Oster (2005) has recently argued that hepatitis B affects off-

spring ratios and has suggested that geographic patterns of hepati-tis B prevalence can explain about 45 percent of the missingwomen. The magnitude of the hepatitis B effect is currently still thesubject of controversy (see e.g., Klasen 2005).


38. Das Gupta and others (2003).39. Lloyd (forthcoming), with original data from The Center for

Reproductive Law and Policy.40. Lloyd (2005)(forthcoming).41. Agarwal (1994); Deere and León (2003).42. Mason and Carlsson (forthcoming).43. Rahman and Rao (2004).44. Kabeer (1999).45. Kabeer (1997).46. This disparity is less apparent in enrollments in primary

school, which has become more equitable over the last decade.47. Browning and Chiappori (1998).48. See Strauss, Mwabu, and Beegle (2000) for an excellent sur-

vey of economic models of intrahousehold allocation.49. Thomas (1990).50. Lundberg, Pollak, and Wales (1997).51. Quisumbing and Maluccio (2003).52. Heise, Ellsberg, and Gottemoeller (1999).53. Kabeer (1997).54. Rahman and Rao (2004).55. Munshi and Rosenzweig (forthcoming).56. Quisumbing, Estudillo, and Otsuka (2004).57. Das Gupta and others (2004).

Chapter 3

1. Deaton (2004), 31. The large differences in mortality do notjust exist between rich and poor countries, or between differentgroups of citizens in poor countries only. In 2002, the median age atdeath in Australia for indigenous people was roughly 20 years lessthan that for nonindigenous males, which was approximately 76years (Australian Bureau of Statistics 2003).

2. Botswana, which has one of the highest HIV/AIDS infectionrates in the world, has seen its life expectancy drop from its peak of60 years in the mid-1980s to 39 years in 2000.

3. Bourguignon, Levin, and Rosenblatt (2004a), 3.4. See the following quotes that nicely demonstrate the debate

on globalization and the inequality of incomes: “Globalizationraises incomes and the poor participate fully,” The Economist, May27, 2000, 94. “There is plenty of evidence that current patterns ofgrowth and globalization are widening income disparities.” PolicyDirector of Oxfam, letter to The Economist, June 20, 2000, 6.

5. Doing this in practice requires household survey data and var-ious adjustments and assumptions—all far from simple—which iswhy so many people discuss international or intercountry inequalityusing National Accounts data instead of global inequality.

6. Pritchett (1997).7. It is important to note that all the evidence presented here

refer to intercountry or international inequality in life expectancy atbirth, and not to global inequalities, because we do not use any evi-dence on within-country differences in life expectancy.

8. Goesling and Firebaugh (2004).9. Note that the declines in life expectancy for the mainly Sub-

Saharan African countries do not seem to be a direct result ofchanges in infant mortality, which in many cases continued todecline. However, as Cornia and Menchini (2005) argue, the rate ofdecline in infant mortality also slowed down significantly in the lastdecade, which cannot be explained solely by the changes in Europeand Central Asia and Sub-Saharan Africa.

10. Deaton (2004).11. Pritchett (forthcoming). The richest quintile is defined

based on an asset index created by Filmer and Pritchett (1999).12. One example of such decomposition is Pradhan, Sahn, and

Younger (2003), who claim that less than one-third of the worldinequality in child health is attributable to differences betweencountries.

13. Gwatkin (2002). Again, the richest decile here is definedbased on the asset index by Filmer and Pritchett (1999).

14. Preston (1980) as cited in Deaton (2004).15. Also note that all of the countries that are significantly below

the curve, that is, those with much lower life expectancy at birththan their GDP per capita would predict, such as South Africa, arein Sub-Saharan Africa.

16. See Deaton (2004). In fact, Cornia and Menchini (2005) citethe rise in parents literacy and female education as the main reasonbehind the remarkable declines in infant mortality rates in LatinAmerica and the Caribbean and Middle East and North Africa inthe 1980s.

17. The increase in inequality between countries in lifeexpectancy at birth in the 1990s is also confirmed by Goesling andFirebaugh (2004).

18. A recent paper by Brainerd and Cuttler (2004) cites increasedalcohol use and psychosocial stress brought on by an unpredictablefuture as the two main contributors to the decline in life expectancyin Russia and other countries of the former Soviet Union.

19. This figure in Morocco compares with roughly three-quarters of the same cohort in Indonesia and Turkey, and about 85 percent in Colombia and Philippines.

20. While making these comparisons between countries, thereader should bear in mind the discussion on the comparability ofthese data from box 2.5 and not interpret these differences literally.

21. The urban and rural figures for China (and India) have notbeen corrected for cost-of-living differences. Given that rural-urbanprice differentials have been increasing over time and are significanttoday (Chen and Ravallion 2004), the figures here exaggerate the dif-ferences in living standards between the rural and urban Chinese.

22. Kolm (1976) proposed the following class of absolute meas-ures of inequality:

,

where κ>0 is a parameter that captures inequality aversion.23. Atkinson and Brandolini (2004) discuss changes in interna-

tional as well as global inequality. We use only their results on inter-national inequality to make our point on the trend differenceswhen one switches from a relative concept of inequality to anabsolute one.

24. Such as Milanovic (2005).25. See Firebaugh and Goesling (2004), Sala-i-Martin (2002).26. Our analysis shows that we can estimate GE(0), that is, mean

log deviation, from grouped data, especially if the number ofgroups is larger than 10, without any significant bias using oursmoothing techniques.

27. The difficulties in assembling a database to calculate globalinequality over such a long time period are detailed in Bour-guignon and Morrisson (2002), 729–30.

28. Pritchett (1997), 14.

K =1

κln

1

ne

κ ( µ−yi )

i=1

n

∑

Endnotes 235

29. Bourguignon, Levin, and Rosenblatt (2004a).30. Ravallion (2004a) as cited in Bourguignon, Levin, and

Rosenblatt (2004a).31. See, for example, Chen and Ravallion (2004).32. Vietnam and Thailand practically eliminated $1 a day

poverty during this period.33. The projections for meeting some of the other Millennium

Development Goals do not give one cause for greater optimism.East Asia and Europe and Central Asia are close to the target on pri-mary education, while others are falling behind. A handful of coun-tries are on track to attain the child mortality goal, mainly in LatinAmerica and the Caribbean. Only 21 percent of the developingworld’s population is expected to achieve the maternal mortalitygoal. Halving the proportion of people without access to safe waterand sanitation will require providing about 1.5 billion people withaccess to safe water and 2 billion with basic sanitation facilitiesbetween 2000 and 2015. At current rates of service expansion, onlyabout one-fifth of countries are on track. See http://www.un.org/millenniumgoals/ for more details on the 8 goals, 18 targets, andnearly 50 indicators that comprise the Millennium DevelopmentGoals.

34. Deaton (2004). The rules of the game at the global level, andespecially the process of setting those rules, will be discussed ingreat detail in chapter 10.

35. Voting shares in the soft-lending arm of the World Bank, theInternational Development Association, are similar: Part I coun-tries hold 61 percent of the vote and Part II countries 39 percent.Developing countries are somewhat better represented in regionaldevelopment banks, but developed countries hold large votingshares there too.

36. In all other cases, the delegates in charge of economic mat-ters may be covering, in addition to the WTO, also UNCTAD, ITU,ILO and WIPO.

37. Each country belongs to one of the following four categoriesbased on its GDP per capita in 2002: low income, lower-middleincome, higher-middle income, and high income.

38. López (2000) as cited in Deaton (2004).39. Pritchett (2001) and Pritchett (2004b).40. Firebaugh and Goesling (2004).41. Sala-i-Martin (2002).42. Pritchett attributes this phrase to Gerschenkron (1962).43. Pritchett (1997), 15.44. As cited in Goesling and Firebaugh (2004).45. As cited in Sala-i-Martin (2002).46. Goesling and Firebaugh (2004).

Focus 2 on empowerment

1. Evans (2004).2. See Petesch, Smulovitz, and Walton (2005) for the interac-

tions between opportunity structure and the agency of subordinategroups.

3. Chaudhuri and Heller (2003).4. Baiocchi, Chaudhuri, and Heller (2005).5. Chaudhuri, Harilal, and Heller (2004).6. The concepts of “terms of recognition” and “capacity to

aspire” come from Appadurai (2004); “capacity to engage” is fromGibson and Woolcock (2005).

7. Rao and Walton (2004).

8. On the origins, structure and goals of KDP, see Guggenheim(forthcoming).

9. Gibson and Woolcock (2005).10. Barron, Diprose, and Woolcock (2005).

Part II

1. In his second inaugural address, Roosevelt also suggested thatavoiding deprivation in outcomes was a legitimate policy aim. As heput it, “The test of our progress is not whether we add more to theabundance of those who have much; it is whether we provideenough for those who have too little” (Washington, DC, January 20,1937). There tends to be less agreement on the importance ofinequality in specific outcome spaces. Writing about incomeinequality, Feldstein (1998) argues that increases in inequalitycaused by larger incomes at the very top warrant no policy atten-tion, and represent “pure Pareto improvements,” to which only“spiteful egalitarians” would object.

2. Michel Ferry, online consultation message, dated October 26,2004, 10:56 a.m.

3. As chapter 1 also indicated, this is not equivalent to sayingthat implementing redistributions is free of tradeoffs, particularlyin the short run, or that terrible inefficiencies cannot arise if redis-tributions are implemented in ways that pay no heed to incentives.We return to these difficult policy issues in chapters 7, 8, and 9.

Chapter 4

1. See Pinglé (2005).2. Plato, The Laws, 745, quoted in Cowell (1995), 21.3. The reason for this reputation is the combination of a utilitar-

ian objective with a set of (rather restrictive) assumptions aboutindividual utilities: (i) individual preferences can be represented bya utility function; (ii) cardinal utility levels are meaningful indica-tors of the individual’s well-being; (iii) utilities can be comparedacross different people; (iv) the utility functions are increasing butconcave (that is, increase at a declining rate) in incomes; and (v) allindividuals have identical utility functions. If they all hold, then, fora fixed level of aggregate income, the greatest sum of utilities isachieved by an equal division of incomes across all individuals.

4. Sen (2000), 69.5. For example, within German Courts, the standard of what is

equitable is based on “the opinion of all those people who think inequitable and just terms”(Palandt 2004, article 242).

6. For example, within common law jurisdictions, key principlesthat guide the application of equity include the principles of“unconscionability,” “undue influence,” “duress,” and “unjustenrichment.” A transaction has been deemed unconscionablewhere “one party to a transaction is at a special disadvantage indealing with the other party because illness, ignorance, inexperi-ence, impaired faculties, financial need, or other circumstancesaffect his ability to conserve his own interests, and the other partyunconscientiously takes advantage of the opportunity thus placedin his hands” (High Court of Australia, Blomley v. Ryan 1956, 99CLR 362 at p. 415, per Kitto J.).

7. Aristotle, The Nicomachean Ethics, book 5, chapter 10, 350 BC.8. Kritzer (2002), p. 495. For a more detailed discussion of the

philosophical evolution of the concept, see Alland and Rials (2003).9. The Chancellory Courts have their origin in the residual dis-

cretionary power of the King, generally exercised by the Chancellor,


to correct or overturn cases in which justice could not be obtainedby a common law court.

10. In the common law tradition, equitable jurisdiction is alsoassociated with a number of well-established principles, sometimescalled the “maxims of equity.” Examples of these include: He whoseeks equity must do equity—that is, a claimant who seeks equitablerelief must be prepared to act fairly toward the defendant (futureconduct); He who comes to equity must come with clean hands—similar to the previous principle, but related to the claimant’s pastconduct; Equity looks at intent rather than form—equity is moreconcerned with substance than form.

11. Cadiet (2004), 425.12. See Arnaud (1993) and Draï (1991).13. Preamble to the Universal Declaration of Human Rights

(1948).14. Note also Article 1(3)—purposes of Charter—“promoting

and encouraging respect for human rights and for fundamentalfreedoms for all without distinction as to race, sex, language or reli-gion” Article 2(1)—“sovereign equality of all members”—indicatesbalance between the principles of sovereignty and noninterferenceand of respect for human rights; Article 55—“universal respect for,and observance of human rights.” Other references to human rightsinclude Articles 13(b)(1), 62(2), and 68.

15. Vienna Declaration and Programme of Action Vienna Dec-laration (UNGA) (A/CONF. 157/23), July 12, 1993. This was issuedby the U.N. World Conference on Human Rights, Vienna, Austria,June 14–25, 1993.

16. Rights identified in the UDHR include the right “withoutany discrimination to equal protection before the law” (Article 7),the right to “take part in the government of his country, directly orthrough freely chosen representatives” (Article 21), and the right toeducation (Article 26).

17. There exist several different characterizations of the “corehuman rights treaties.” One is the “International Bill of Rights,”which comprises the Universal Declaration on Human Rights, theInternational Covenant on Civil and Political Rights, and the Inter-national Covenant on Economic, Social and Cultural Rights. Anotherconceptualization is that there are six major human rights treaties:the ICCPR (1966), the ICESCR (1966), the Convention on the Elim-ination of All Forms of Racial Discrimination (ICERD) (1964), theConvention on the Elimination of All Forms of DiscriminationAgainst Women (CEDAW) (1989), the Convention Against Torture(CAT) (1984) and the Convention on the Rights of the Child (CRC)(1989). None of these characterizations should be taken as exhaustiveor exclusive; there are several other extremely important humanrights treaties beyond these, both within the U.N. system and outsideit in regional systems. The OHCHR lists “seven core internationalhuman rights instruments”: the six above, plus the InternationalConvention on the Protection of the Rights of Migrant Workers andthe Members of their Families (1990). In addition to this, mentionshould also be made of the Genocide Convention (1948).

18. Henrich and others (2004) report on Ultimatum Gamesplayed in 15 small-scale societies around the world. In an Indone-sian experiment, Cameron (1999) found that high offers persistedin games played with (locally) very high monetary stakes.

19. See, for instance, Fehr and Schmidt (1999) on a theory of“inequity aversion,” and Rabin (1993) on a model of behavior thattakes other people’s intentions directly into account.

20. Ibid.21. Terry, Carey, and Callan (2001).22. This is the question in the United States General Social Sur-

vey, which was fielded every year between 1972 and 1997. A similarquestion is included in the Euro-Barometer surveys, which are alsoused in the study.

23. Subjective well-being—or “happiness”—surveys have alsobeen used to test the relative income hypothesis in a number of dif-ferent settings. Following on original work by Easterlin (1974),which suggested that, despite sustained economic growth, people insome rich countries were not growing any happier, a number ofrecent studies have found that reported well-being rises with per-sonal income, but declines with the income of other people in aperson’s reference group. This is usually interpreted to mean thatwell-being is driven at least in part by relative incomes, rather thanpurely by their absolute levels. See Graham and Felton (2005) forLatin America, Ravallion and Lokshin (2002) for the Russian Fed-eration, and Luttmer (2004) for the United States. The relativeincome literature is of tangential interest for this report, because itprovides additional backing for the idea that one’s position in soci-ety in relation to others matters, both for individual behavior andfor welfare.

24. See De Ferranti and others (2004) for an earlier discussion ofthis data.

25. The professions included skilled factory worker, doctor inprivate practice, chairman of a large national company, lawyer,shop assistant, federal cabinet minister, judge of the nation’s high-est court, owner/manager of a large factory, and unskilled worker.Respondents were also asked about their own income.

26. Countries in the ISSP sample include the following: Austria,Australia, Canada, France, Germany, Italy, Israel, Japan, the Nether-lands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland,the United Kingdom, and the United States.

27. This figure is an update, based on more recent data, of anoriginal figure that appeared in Ravallion and Chen (1997).

28. See, once again, Ravallion and Chen (1997) and Bour-guignon (2003).

29. Using a slightly different dataset, and a decomposition tech-nique that also accounts for the differential sensitivity of differentpoverty measures to changes in mean incomes, Kraay (forthcom-ing) finds that the growth share in this variance is closer to 70 per-cent. In a sample of long spells only, that share rises to 94 percent.

30. Whereas the total growth elasticity of poverty reductionsimply relates total change in poverty to the growth in meanincomes, without controlling for changes in the relative distribu-tion and is therefore given by

,

the partial growth elasticity of poverty reduction controls forchanges in inequality. That means that it is calculated for a givenLorenz curve:

.εµP =

∆P

P

µ∆µ

Lt

=P Lt ,

µt +1

z

− P Lt ,

µt

z

P Lt,µt

z

µ∆µ

εµT =

∆P

P

µ∆µ

Endnotes 237

This definition, like the decomposition on which it is based, is path-dependent, and the elasticity would be somewhat different if it wascalculated on the basis of the final-period Lorenz curve, rather thanthe initial one. In this report, we always use the initial period rela-tive distribution when computing partial growth elasticities. Noregression method is used, so no functional form of any kind isimposed on the relationship.

31. This finding is qualitatively consistent with alternative esti-mates of the relationship between partial and total growth elastici-ties, on the one hand, and income inequality, on the other. SeeBourguignon (2003) and Ravallion (2005).

32. The other four combinations of poverty lines and measureswere also computed, and the upward-sloping pattern is qualita-tively robust.

33. See also Bourguignon, Ferreira, and Lustig (2004) for adetailed discussion of seven country studies.

34. In fact, World Bank (2005b) finds that, on average, the poor-est 20 percent of the population in each of 14 countries in theirstudy grow 0.7 percent for each 1 percentage point growth in meanincomes.

Chapter 5

1. See, for example, Fisman (2001a).2. Even in a hypothetical world with perfectly functioning mar-

kets, there may still be an indirect influence, coming from the effectof wealth or income on savings decisions. It has been suggested thatthe poor are less inclined to save than the rich, and as a result,aggregate savings as a proportion of aggregate income may go up ifthe rich gain at the expense of the poor. This could affect invest-ment decisions through the effect of the supply of savings on theprice of capital. Inequality, in this Kaldorian view of the world(after Nicholas Kaldor, the Cambridge economist), would enhancegrowth, although it might yet be a Pyrrhic victory. Kaldor worriedabout the inevitability of crises under capitalism, and he saw fastergrowth accompanied by burgeoning inequality as a recipe forongoing crises.

3. Note that such redistributions from wealthier to poorer peo-ple do not necessarily have to be to the poorest people in society,but rather to those with good growth opportunities.

4. Unless there are two different investment opportunities thathave the exact same payoff, net of interest.

5. Dasgupta, Nayar, and Associates (1989).6. See also Gill and Singh (1977) and Swaminathan (1991).7. See Djankov and others (2003).8. See Deaton (1997) for more details.9. See Townsend (1995).10. Fafchamps and Lund (2003) find that, in the Philippines,

households are much better insured against some shocks thanagainst others. In particular, they seem to be poorly insured againsthealth risk, a finding corroborated by Gertler and Gruber (2002) inIndonesia.

11. See Banerjee (2000) for a discussion of the alternativeviews.

12. Although, there are instances of government scholarshipsfor tertiary education in many countries that require beneficiariesto provide a specified number of years of service in governmentafter graduation.

13. Blair, Judd, and Chapleau (2004). See also Loury (2002) for awide-ranging study.

14. See the survey in Steele and Aronson (1995).15. From McKenzie and Woodruff (2003), figure 1.16. Olley and Pakes (1996).17. The ICOR measures the increase in output predicted by a

one-unit increase in capital stock. It is calculated by extrapolatingfrom the past experience of the country and assumes that the nextunit of capital will be used exactly as efficiently (or inefficiently) asthe previous one. The inverse of the ICOR therefore gives an upperbound for the average marginal product for the economy—it is anupper bound because the calculation of the ICOR does not controlfor the effect of the increases in the other factors of production,which also contributes to the increase in output. The implicitassumption that the other factors of production are growing isprobably reasonable for most developing countries, except perhapsin Africa.

18. Banerjee and Munshi (2004); Banerjee, Duflo, and Munshi(2003).

19. This is not because capital and talent happen to be substi-tutes. In these data, as is generally assumed, capital and abilityappear to be complements.

20. From Goldstein and Udry (1999), figure 4.21. From Goldstein and Udry (1999), 38.22. Based on Berry and Cline (1979).23. Some of the effects of lack of insurance may be quite subtle.

Banerjee and Newman (1991) argue, for example, that the avail-ability of insurance in one location (the village) and its unavailabil-ity in another (the city) may lead to inefficient migration decisions,because some individuals with high potential in the city may preferto stay in the village to remain insured.

24. The fact that there is underinvestment on average, and notonly a set of people with too many bullocks and a set of people withtoo few, is probably due to the fact that bullocks are a lumpy invest-ment, and owning more than two is inefficient for production(there is no small adjustment possible at the margin).

25. Another piece of relevant evidence comes from the effects oftitling nonagricultural land. Field (2003) shows evidence from a land-titling program in the slums of urban Peru that suggests that the lackof a clear title to the land where you have built your home reduces theability of the household members to work outside. Field hypothesizesthat this is because someone needs to be home to defend the untitledproperty from expropriation by others. But she does not find any evi-dence that land titling improves access to credit.

26. This number takes into account the fact that only 20 percentof the Indonesian population is iron deficient. The private returnsof iron supplementation for someone who knew they were irondeficient—which they can find out using a simple finger prick—would be $200.

27. Kremer (1993).28. Li and others (1994).29. Banerjee and Duflo (2004b), 11.30. See Bénabou (1996) for a survey.31. Forbes (2000) also corrects for the bias introduced by intro-

ducing a lagged variable in a fixed-effect specification by using theGeneralized Method of Moments (GMM) estimator developed byArellano and Bond (1991).


32. Barro (2000) estimates a cross-sectional relationshipbetween inequality and short-term growth and finds a negativerelation in poor countries but a positive relation in rich ones.

33. Banerjee, Gertler, and Ghatak (2002).

Chapter 6

1. Engerman and Sokoloff (1997).2. Acemoglu, Johnson, and Robinson (2001), 1387.3. Klarén (2000), 44–48.4. See Hemming (1970), 264 on Pizzaro and Melo (1996), 222

on Colombia.5. Bakewell (1984), Cole (1985), Lockhart (1969) and Mörner

(1973).6. Lockhart and Schwartz (1983), 34.7. Parry (1948).8. Cardoso (1991), Mahoney (2001), and Lang (1975), 28.9. An analogous story applies to the parts of the region suitable

for plantation agriculture, especially sugar. Colonists in theCaribbean, present-day Brazil, and the southern United States tookadvantage of the international market in slaves and developed dis-tinct but also highly coercive regimes.

10. See Crosby (1986), 143–44.11. See Craven (1932) and Morgan (1975).12. See Acemoglu, Johnson, and Robinson (2002), Acemoglu,

Johnson, and Robinson (2004), and Rodrik, Subramanian, andTrebbi (2002).

13. In addition to arguments noted above, see the econometricanalysis arguing that governance causes growth, but not the otherway, in Kaufmann and Kraay (2002).

14. Haggard (1990) and Kang (2002).15. Liddle (1991) and MacIntyre (2001b), 259.16. Geertz (1963), Elson (2001), 194–201, 280–81; Rock (2003), 14.17. World Bank (1993).18. Bates (1981).19. Bates (1981), 122.20. Bates (1989).21. Bowman (1991).22. Fisman (2001b), MacIntyre (2001a), Stern (2003).23. Lipset (1959).24. This conjecture is substantiated by the standard account of

the evolution of institutions in Britain, culminating in the GloriousRevolution of 1688 (North and Thomas 1973, North and Weingast1989, and O’Brien 1993).

25. See Brenner (1976), Tawney (1941), Moore (1966), and Ace-moglu, Johnson, and Robinson (2002b).

26. See Thompson (1963), Tilly (1995), and Tarrow (1998).27. Acemoglu and Robinson (2000), Acemoglu and Robinson

(2005).28. Lindert (2003), Lindert (2004).29. Other data on real wages and real rental rates on land sup-

ports this claim; see O’Rourke and Williamson (2002).30. Li, Squire, and Zou (1998), Rodrik (1999b).31. This section is based on Andersson and Gunnarsson (2005)

and Jäntti and others (2005).32. Wade (1990).33. Andersson and Gunnarsson (2005), 4.34. Chen and Ravallion (2004).35. Montinola, Qian, and Weingast (1995), 51–52.

36. Qian (2003), 305.37. In China revenues were collected at the local level and shared

up with higher levels of government

Focus 4 on Indonesia

1. World Bank (1993).2. Using data from van der Eng (1993a, 1993b, 2002), it is possi-

ble to construct an index of pro-poor growth from 1880 to 1990(Timmer 2005). Using this index, the rates of pro-poor growth foreach historical era were as follows: 1880–1905 = 0.05 percent peryear; 1905–1925 = 4.57 percent per year; 1925–1950 = –2.57 per-cent per year; 1950–1965 = 2.37 percent per year; and 1965–1990 =6.56 percent per year.

3. All Indonesian distributional data are subject to the usualcaveats about the difficulty of obtaining adequate responses fromboth the very poor and the very rich. In particular, SUSENAS doesnot capture the extremely uneven accumulation of wealth or theconspicuous consumption of the ultra rich.

4. There is a “standard” economic model behind this story, inwhich the capital of individuals—formal and informal skills,land, location, and even accumulated savings—is deployed toearn incomes, with the possibilities determined by technology,transactions costs, and market prices. See Alatas and Bour-guignon (2004).

5. Temple (2001). These institutions, for better and worse, livedwell beyond the Suharto era.

6. Timmer (2005).

Chapter 7

1. The study finds that it is the quality of parent-child interac-tions, especially talking to the child, that matters the most (Hartand Risley 1995).

2. Pollitt, Watkins, and Husaini (1997), Grantham-McGregorand Ani (2001), Stoltzfus, Kvalsvig, and Chwaya (2001), Black(2003), Dickson and others (2000), Smith, Brooks-Gunn, and Kle-banov (1997).

3. Shore (1997).4. Rutter, Giller, and Hagell (2000), Karr-Morse and Wiley

(1977).5. Paxson and Schady (2005).6. Pritchett (2004a).7. Carneiro and Heckman (2003).8. Three of the best-documented studies for the United States

are on the Chicago Child-Parent Centers, a half-day program on alarge scale in the Chicago public schools; the Abecedarian program,a full-day year-round education program in Chapel Hill, NorthCarolina, with followup to age 21; and the High/Scope PerryPreschool, a half-day program on a small scale in the Ypsilanti,Michigan, public schools with follow up to age 40. The last two aresmall randomized experiments. Studies on the Chicago programuse statistical techniques to control for selection bias and other fac-tors that might confound interpretation of results.

9. Slide 5 of presentation on Heckman and Masterov (2004), athttp://www.ced.org/docs/presentation_heckman1.pdf.

10. Young (2002). In the United States, Currie (2000) notes that,while all the children in the North Carolina Abecedarian projectwere at risk of mental retardation, positive effects were twice aslarge for the most disadvantaged children among them. Currie and

Endnotes 239

Thomas (1999) find that gains in test scores associated with HeadStart were greater for Hispanic children than for non-Hispanicwhites.

11. Grantham-McGregor and others (1991).12. Deutsch (1998), Attanasio and Vera-Hernandez (2004).13. Paxson and Schady (2005).14. Scott-McDonald (2004).15. Doryan, Gautman, and Foege (2002).16. Kirpal (2002).17. Many conditional cash transfers in Latin America already

have some of these features.18. Currie and Thomas (1995) find that gains in vocabulary and

reading test scores faded out among black Head Start childrenwhile they were still in elementary grades but not among whiteseven though initial gains in test scores were the same for bothgroups. Currie and Thomas (2000) attribute this to their findingthat black children who attended Head Start go on to attend lower-quality schools than other black children but that the same is nottrue for whites.

19. King, Orazem, and Paterno (1999).20. UNESCO (2005).21. Thirty-two more countries have too little data to assess

(Bruns, Mingat, and Rakotomalala 2003 and FTI Secretariat 2004).22. See Wodon (2005) for Niger and Senegal—in Senegal,

when the school is located less than 15 minutes away from home,the probability of enrollment for boys goes up by 30 percentagepoints as compared with having the school more than one houraway. Burney and Irfan (1995) found negligible impacts of prox-imity.

23. The study finds that children age two to four, in 1974, had anadditional 0.12 to 0.19 years of education and 1.5 to 2.7 percenthigher wages for each new school built per 1,000 children (Duflo2001). Other studies that find positive impacts of expanding schoolinfrastructure include Birdsall (1985) in urban Brazil, DeTray(1988) and Lillard and Willis (1994) in Malaysia, Lavy (1996) inGhana, and Case and Deaton (1999) in South Africa.

24. Pritchett (2004a).25. Kremer, Moulin, and Namunyu (2002).26. In Brazil in 2001, the Bolsa Escola program was spending

almost $700 million a year and reaching 8.6 million school children,one-third of all primary school children in Brazil. In Bangladesh, theFood for Education budget in 2000 was $77 million, benefiting 2.2million children, 13 percent of total school enrollment (Morley andCoady 2003). In Mexico, according to government figures, by 2004,Opportunidades was spending approximately $2.3 billion annually,or about 1.5 percent of total expenditture, and reaching 5 millionhouseholds—almost 20 percent of all families in Mexico.

27. Morley and Coady (2003) and Schultz (2004).28. There are some countries in which gender disparities require

attention to boys even for primary schooling (Mongolia) and quitea few for higher levels of attainment (Brazil, Philippines).

29. Herz and Sperling (2004).30. World Bank (2004d).31. Pritchett (2004a).32. A score of more than one standard deviation below the

OECD mean is considered poor performance.33. See Hanushek (1986) and Hanushek (1996).34. Wößmann (2000).

35. See Glewwe, Kremer, and Moulin (2002), Glewwe, Ilias, andKremer (2003), Glewwe and others (2004), Kremer, Miguel, andThornton (2004), Miguel and Kremer (2004).

36. It is also important to take account of household behavior inresponse to changes in school inputs. Das, Dercon, Habrayimana,and Krishnan (2004) find that household spending and (nonsalary)cash grants to schools are close substitutes with no impact on learn-ing from anticipated increases in school funds. Unanticipatedfunds have significant learning impacts.

37. Hanushek and Wößman (2005).38. Ladd (2002).39. The analysis is not able to separate the effects of private

schools from the effects of the incentives for greater effort gener-ated by the voucher policy. The distinction is important becauseeffort can be encouraged also by awarding scholarships to studentsattending public schools, as we saw with the Balshaki program. SeeAngrist and others (2002).

40. Chubb and Moe (1990).41. See McEwan (2000) on Chile. Hoxby (2002) finds improve-

ments from school choice from charter schools in Michigan andArizona; Holmes, DeSimone, and Rupp (2003) find that competi-tion from school choice in North Carolina improved test scores sig-nificantly. Cullen, Jacob, and Levitt (2000) and Cullen, Jacob, andLevitt (2003) found no impact from school choice in Chicago. Seealso Witte (2000). Hoxby (2000) finds significant productivity gainsfor a representative sample across the United States, while Roth-stein (2005) uses the same data to find far more modest results.

42. McEwan and Carnoy (2000) and Hsieh and Urquiola(2003).

43. Lindert (2004).44. In Brazil, a 1997 reform set a minimum floor on education

spending in all regions, which led to an increase in primary schoolenrollments from 55 to 85 percent in only six years (World Bank2005f).

45. See Grindle (2004) and Fiszbein (2005) for examples inLatin America.

46. World Bank (2003j).47. Wößmann (2004) argues for giving decision-making

authority to those with the most relevant information and the low-est prospects for personal gain. He finds that student performanceimproves with central control over the curriculum, budget, andexams; delegation of process and personnel decisions to the school;greater say for teachers in the selection of appropriate teachingmethods; regular scrutiny of students’ education performance; andparental interest in education content. Competition from privateeducation institutions and the presence of teacher unions, on theother hand, have adverse effects.

48. See World Bank (2003j) and Pritchett (2004a).49. World Bank (2003j) for South Africa; Duan (2005).50. World Bank (2004k).51. World Bank (2004k).52. Thailand’s Ministry of Public Health and World Bank

(2005).53. Rawlings (2004).54. See Barber, Bertozzi, and Gertler (2005) for Mexico; Das and

Hammer (2004) and Das and Hammer (2005) for India; Barber,Gertler, and Harimurti (2005) for Indonesia; and Leonard andMasatu (2005) for Tanzania.


55. World Bank (2003j).56. Gertler and Barber (2004).57. World Bank (2005f).58. World Bank (2004h).59. See World Bank (2001c) for Vietnam, World Bank (2005f)

for Cambodia, and Lewis (2002) on ECA countries.60. Schieber (2005).61. Gertler and Gruber (2002) find that while Indonesian fami-

lies were able to self-insure reasonably well the costs of minor ill-ness (up to 70 percent of the income loss associated with these),they were able to insure only 38 percent of the income loss associ-ated with illness that severely limits physical functioning.

62. Sekhri and Savedoff (2005).63. Savedoff (2004).64. Escobar (2005); Escobar and Panopoulou (2002); Castaneda

(2003).65. Uganda’s use rates had dipped with the introduction of fees

in the 1990s, and although Uganda’s poor were formally exempt,they often paid as much as other users. Deininger and Mpuga(2004) estimate that the revenues lost to the health system from theelimination of user fees in 2001 were smaller than the value of ben-efits from the reduction in foregone earnings, thanks to a healthierpopulation.

66. Thailand’s Ministry of Public Health and World Bank(2005).

67. See Chaudhury, Hammer, and Murrugarra (2003) for Arme-nia, World Bank (2005f) for China, and Meesen and Van Damme(2004) for Cambodia.

68. Socialist Republic of Vietnam and World Bank (2005).69. Preker and Carrin (2004), Atim (1999).70. Ranson and others (2005).71. World Bank (2004k), World Bank (2003j), Banerjee, Deaton,

and Duflo (2004), Chaudhury and Hammer (2004) on Bangladesh.72. World Bank (1997a).73. The equity- and opportunity-enhancing role of social pro-

tection is the conceptual underpinning of the Bank’s socialprotection sector strategy (World Bank 2001e, Holzmann andJorgensen 2001). Other sources are World Bank (2003f), Devereux(2001), and Ravallion (2003).

74. Improved access to credit is reported for South Africa(Ardington and Lund 1995) and Brazil (Schwarzer and Querino2002), where the electronic banking card issued by the programis often used as proof of creditworthiness. A significant share ofrural pension beneficiaries in Brazil report using part of the pen-sion to purchase seeds and agricultural tools (Delgado and Car-doso Jr. 2000); others report investing in improvements in theirhousing. Children and older people in beneficiary householdshad better health in South Africa (Case 2001) and enrollmentrates of school-age children were higher among pension benefi-ciary households in both South Africa (Duflo 2003) and Brazil(de Carvalho Filho 2000).

75. Kanbur (2005).76. Lindert (2004).77. Esping-Andersen (1990) argues, for example, that the suc-

cess of the welfare system in continental Europe is based on its uni-versality.

78. Alesina and Glaeser (2004) suggest that the different choicesdepend on political and institutional factors, including the form of

democracy, the degree of ethnic homogeneity, and societal beliefsabout the causes of poverty.

79. This scheme covered about 10 percent of the unemployedlabor force, so it was usefully supplemented by workfare (WorldBank 2000b).

80. Coady, Grosh, and Hoddinott (2004).81. See World Bank (2002a), World Bank (2003d), and World

Bank (2003e).82. Social workers assess 52 separate factors grouped in 7

dimensions (identification, health, education, family dynamics,housing, work, and income). See Chile’s Ministry of Planning(2004); on Bangladesh, see Hashemi (2000).

83. On Argentina, see Jalan and Ravallion (1999); on the Maha-rashtra scheme, Ravallion (1991); on Bolivia, Newman, Jorgensen,and Pradhan (1992).

84. The analysis showed a decline in poverty of 11 percent in thePROGRESA communities and an increase in poverty in the controlgroup.

85. See Bourguignon, Ferreira, and Leite (2004) on Brazil’s BolsaEscola; see Rawlings (2004) and Morley and Coady (2003) on Mex-ico’s PROGRESA; and see World Bank (2005d) on Cambodia’slower secondary school program.

86. United Nations (2002).87. See Holzmann and Hinz (2005) and World Bank (2001b).88. Except for the implicit constraint imposed by a means-tested

scheme (Barrientos 2005).89. Case and Deaton (1998).90. Barrientos (2005), Rawlings (2004), De Ferranti and others

(2004).91. See James (2000) for the general argument, van der Berg and

Brendenkamp (2002) on South Africa, and Paes de Barros and deCarvalho (2004) on Brazil.

92. Lund (1999) and Atkinson (1995).93. Hoogeveen and others (2004).

Chapter 8

1. See Haber (2001).2. Hellman, Jones, and Kaufmann (2003).3. The separation of powers between the executive, legislative,

and judicial arms of government aims to combat the dangers ofinvesting state power in one person or group, with each branchholding the others accountable through differing “checks and bal-ances.” This system also arguably maintains competition withinpolitical institutions. See Haber (2001).

4. Asian Development Bank (2003), 24–25; Garapon (2003),Russel and O’Brien (2001).

5. The Constitution of the Federal Democratic Republic ofEthiopia (1995).

6. Tien Dung (2003), 8.7. See Buscaglia and Dakolias (1999).8. Ringera and others (2003).9. See Langseth and Stolpe (2001), Dakolias and Thatchuk (2000).10. For East Asia, see Asian Development Bank (2003).11. See www.tsj.gov.ve for an explanation of the judicial portal.12. The important distinction here is between campaigns that

impede judicial independence (targeting unpopular decisions) andthose that serve as a check on judicial misbehavior (exposing judi-cial corruption). See Asian Development Bank (2003), 8–9.

Endnotes 241

13. See World Bank (2004g), World Bank (2004c), and Ham-mergreen (2004).

14. It is important to note that a vast array of practices, systems,and traditions have been defined as informal, traditional, or cus-tomary law, all existing within vastly differing contexts. The use of“informal” is used in contrast to “formal” state systems and is notmeant to imply that such institutions are procedurally informal.

15. World Bank (2004j), 12.16. Augustinus (2003).17. See Centre for Housing Rights and Evictions (2004).18. Mamdani (1996).19. Fortman (1998).20. See Buscaglia (1997) and Mattei (1998). See also Kranton

and Swamy (1999) and Pistor and Wellons (1999).21. The sungusungu have avoided cooption by the state, and

their lack of knowledge of the law and people’s rights can lead toabuse. See Mwaikusa (1995), 166–78, and Bukurura (1994).

22. Buscaglia (1997).23. See Bush (1979) discussion on different African nations’

attempts at dynamic integrations.24. Interestingly, a hundred years later, the 1927 Black Adminis-

tration Act, which recognized a dual system of official law, formedthe basis for separating whites and Africans under apartheid. SeeVan Niekerk (2001).

25. See Bennett (1999) for a discussion of this issue.26. More than two-thirds of the European Court of Human

Rights’ case load between 1999 and 2003 was about violations ofthe right to a fair trial, in particular, the excessive length of courtproceedings.

27. Between 2002 and 2004, 14,992 mediation cases were han-dled. See Malik (2005).

28. Feierstein and Moreira (2005).29. Das (2004).30. See Human Rights Watch (2000).31. See Greenwood and others (1998), Gottfredson (1998),

Tremblay and Craig (1995), Waller, Welsh, and Sansfaçon (1999),Waller and Sansfaçon (2000), World Bank (2003a).

32. World Bank (2003a), Council for Scientific and IndustrialResearch (2000).

33. Graham and Bowling (1995), Shaw (2001).34. Sloth-Nielsen and Gallinetti (2004), Bottoms (1990), Shaw

(2001).35. See Harber (1999), Shaw (2004).36. For further details see the project Web site at http://www

.bac.co.za/Web%20Content/Projects/Tiisa%20Thuto/Intro%20Template%20for%20Tiisa%20Thuto.htm.

37. Penal Reform International assisted in establishing commu-nity service orders as alternative sentencing in Zimbabwe, Kenya,Malawi, Uganda, Zambia, Burkina Faso, Congo, the Central AfricanRepublic, and Mozambique.

38. Morrison, Ellsberg, and Bott (2004).39. Council for Scientific and Industrial Research (2000).40. Mtani (2002).41. See Binswanger, Deininger, and Feder (1995). In India,

employment in agricultural wage labor (as opposed to self-employed cultivation) is strongly correlated with poverty (Kijimaand Lanjouw 2004). Similarly, in Thailand, controlling for other

factors, tenancy increases the risk of poverty by nearly 30 percentcompared with being a landowner (World Bank 2001f).

42. Cardenas (2003), Conning and Robinson (2002).43. Deininger (forthcoming).44. Deininger and Olinto (2000) go further and suggest that

asset inequality is a causal determinant of growth performance.45. Intermediaries were essentially tax collectors who received

land rights (in different forms, depending on the state) to “interme-diate” the flow of resources between cultivators and the colonialadministration.

46. Appu (1997), 196.47. Banerjee, Gertler, and Ghatak (2002). In West Bengal, efforts

in the late 1970s and 1980s to register sharecroppers and codifytheir rights were backed by the ruling left parties (Appu 1997).

48. Building on spontaneous transactions in which either thetenant or landlord buys out the other party’s interest, the govern-ment is considering ways to facilitate such buyouts in a more sys-tematic and equitable fashion, thereby allowing one party toobtain full ownership rights to the land (Nielsen and Hanstad2004).

49. De Janvry and Sadoulet (2002).50. Scott (1976), Gunning and others (2000), Deininger,

Hoogeveen, and Kinsey (2004), and Kinsey and Binswansger(1993).

51. De Ferranti and others (2004).52. See De Janvry and Sadoulet (1989), Jonakin (1996), Alston,

Libecap, and Mueller (1999).53. See, for instance, Feder (1988) and Jacoby, Li, and Rozelle

(2002). Some studies report a doubling of investment and largegains in land values (30–80 percent) with more secure tenure(World Bank 2003i). To the extent that highly visible investments,such as trees or fences, may be a means to establish landownershiprights, causality might run the other way (Brasselle, Gaspart, andPlatteau 2002). A recent study from Ethiopia demonstrates thattenure insecurity indeed encourages investment in trees that havelittle impact on productivity, while higher levels of tenure securityare unambiguously associated with productivity-enhancing invest-ments (Deininger and others 2003).

54. De Ferranti and others (2004).55. See Palmade (2005) on India. See Field (2003) on Peru.56. Payne (2002), Durand-Lasserve (2003).57. The three levels are as follows: (1) the right to somewhere to

live (either on the occupied plot or an alternative should relocationbe necessary); (2) a 30-year lease transferable in the event of death;and (3) a 199-year lease effectively conferring ownership rightsupon survey and payment for the land.

58. World Bank (2003i).59. Land regulation can also work against the urban poor.

When formal land development parameters (such as minimumplot sizes, setbacks, and infrastructure servicing standards) are notbenchmarked against affordability levels for the majority of theurban population, the poor are excluded from access to formallandownership. Bertaud and Malpezzi (2001), Payne and Majale(2004).

60. Gravois (2005)61. Brasselle, Gaspart, and Platteau (2002).62. World Bank (2003i)


63. Historically, distress sales have played a major role in theaccumulation of land by large manorial estates in China (Shih1992) and in early Japan (Takeoshi 1967) and by large landlordestates in Punjab (Hamid 1983). The abolition of communal tenureand the associated loss of mechanisms for diversifying risk areamong the factors underlying the emergence of large estates inCentral America (Brockett 1984).

64. World Bank (2003i).65. World Bank (2003i).66. For Sudan, see Kevane (1996); for Colombia, Deininger,

Castagnini, and González (2004).67. World Bank (2003i).68. Long-term fixed rental contracts are the most efficient from

an economic perspective because they are compatible with incen-tives for effort and investment but various imperfections in mar-kets make these rare (chapter 5).

69. In most states, lack of political support for the legislationallowed landlords to subvert the intent of the laws. West Bengal isan exception—a tenant registration campaign had strong politicalsupport and succeeded in protecting tenants (Appu 1997).

70. De Janvry and Sadoulet (2002).71. De Ferranti and others (2005).72. World Bank (2003j).73. See Van de Walle and Cratty (2004) on Vietnam. See Jalan

and Ravallion (2002) on China.74. See Jalan and Ravallion (2003). The findings also indicate

that the health gains largely bypass children in poor families withpoorly educated mothers.

75. See Malmberg Calvo (1994) on African countries; Ilahi andGrimard (2000) and Ilahi and Jafarey (1999) on Pakistan; WorldBank (2001g) on electricity and gas; and Khandker, Lavy, andFilmer (1994) on Morocco.

76. Some 60 percent of the rural population in Africa lives inareas of good agricultural potential, but with poor marketaccess, while only 23 percent live in areas of good agriculturalpotential and good market access. The remainder have bothpoor market access and poor agricultural potential (Byerlee andKelley 2004).

77. Between 1970 and 1998, roads (in km) increased by 8.3 per-cent per year (World Bank 2005c).

78. World Bank (2005c).79. De Ferranti and others (2004).80. Economies from large production and delivery have dimin-

ished in some activities, especially telecommunications and powergeneration. And regulatory innovation made unbundling possible.Unbundling promoted competition by separating activities inwhich economies of scale are important (for example, electricitytransmission and distribution) from activities in which it is less so(electricity generation). See World Bank (1994) and World Bank(2004l).

81. World Bank (2004f) and chapter 10.82. According to the Latinobarómetro, the proportion of

respondents who said they thought privatization had benefitedtheir country dropped steadily from 46 percent in 1998 to 21 per-cent in 2003 (Lagos 2005).

83. Estache, Foster, and Wodon (2001).84. Estache, Foster, and Wodon (2001).

85. Lifeline tariffs mean that rates increase after surpassing acertain consumption threshold (the level deemed necessary to meetbasic household needs).

86. Wodon, Ajwad, and Siaens (2005) and Estache, Foster, andWodon (2001).

87. Estache, Foster, and Wodon (2001).88. Kariuki and Schwartz (2005).89. Gulyani, Talukdar, and Kariuki (2005).90. Estache (2003).91. Irwin and Yamamoto (2004).92. See Chisari, Estache, and Romero (1999) and Guasch

(2003).

Focus 5 on taxation

1. Bird, Martínez-Vazquez, and Torgler (2004).2. Bird, Martínez-Vazquez, and Torgler (2004).3. Moore (2001); Davis, Ossowski, and Fedelino (2003).4. Moore (2001).5. Gupta and others (2003).6. De Ferranti and others (2004).7. Bird and Slack (2002).8. Rudnik and Gordon (1996).9. Boskin (1977).10. Rajan and Zingales (2003).11. Prud’Homme (1990).12. Bird (1991).

Chapter 9

1. Smith (1776), 128.2. See Tilly (1998) on opportunity hoarding by middle groups.3. World Bank (2004b), Perotti and Volpin (2004).4. See De Soto (2000), Glaeser, Sheinkman, and Shleifer (2003),

and Haber, Noel, and Razo (2003).5. See Morck, Wolfenzon, and Yeung (2004) and Claessens,

Djankov, and Lang (2000).6. Morck, Stangeland, and Yeung (2000), Morck and Yeung

(2004).7. On developed countries, see Rajan and Zingales (2003).8. Halac and Schmukler (2003) and Perotti and Feijen (2005).9. See Velasco (1988), Valdés-Prieto (1992), Haber and Kantor

(2004), Claessens and Pohl (1994), and Perotti (2002). Also see box9.2 on Russia.

10. Haggard, Lim, and Kim (2003), 87. See also Siegel (2003) fora discussion of the importance of political connectedness in theRepublic of Korea.

11. Feijen and Perotti (2005); Claessens and Perotti (2005).12. Bertrand, Schoar, and Thesmar (2004).13. Braverman and Guasch (1986).14. See discussion in Armendáriz de Aghion and Morduch

(2005), Pulley (1989), and Meyer and Nagarajan (2000) for thepoor repayment performance, and Burgess and Pande (2004) forpositive impacts on the poor of social banking.

15. Honohan (2004).16. The discussion of the Czech and Polish stock markets builds

on Glaeser, Johnson, and Shleifer (2001).17. Rajan and Zingales (2003), 159.18. Guiso, Sapienza, and Zingales (2004).

Endnotes 243

19. Roland and Verdier (2000) and Claessens and Perotti (2005).20. See, for example, Diamond (1981), Blanchard (2004),

Bertola (2003), Agell (2002), among others.21. For discussions of India’s labor market regulation and its

adverse effects, see Stern (2002), Hasan, Mitra, and Ramaswamy(2003), and Besley and Burgess (2004). For example, industrial sec-tor legislation requires employers with more than 100 workers toseek prior approval of the government before dismissing workers.This has earned the country a 90 out of 100 on the World Bank’sDoing Business in 2005 “difficulty in firing index” (World Bank(2005e).

22. See Kingdon, Sandefur, and Teal (2005). Strict enforcementof labor market regulations in South Africa—even for smallfirms—can also contribute to unemployment by inhibiting thedevelopment of informal sector firms.

23. In 1997–98, 83 percent of nonagricultural employment inIndia was informal (ILO 2002).

24. Kugler (2004).25. Maloney (1999), Maloney and Nuñez Mendez (2004).26. Chen, Vanek, and Carr (2004).27. ILO (2002).28. Besley and Burgess (2004).29. The unemployment figure relies on the broad definition of the

labor force (employed + searching and nonsearching unemployed).Using the strict labor force measure (employed + searching unem-ployed), the unemployment rate was 32 percent in 2003. Data isfrom the 2003 Labor Force Survey cited by Kingdon, Sandefur, andTeal (2005). “The ratio of non-agricultural informal sector em-ployment to unemployment is 0.7 in South Africa but 4.7 in Sub-Saharan Africa, 7.0 in Latin America and ll.9 in Asia.” (Kingdon,Sandefur, and Teal 2005).

30. Estimates suggest high levels of unemployment (20 to 30percent for urban males), but there are measurement problems thatbias the estimates upward.

31. Rodrik (1999b).32. Galli and Kucera (2004).33. The unions then played a critical role pushing the social and

economic agenda toward more welfare-oriented policies (Fishman1990, Boix 1998, and Boix 2005).

34. Some might argue that the Scandinavian systems are noteasily replicable, on the grounds that they rely on extraordinary lev-els of trust and social capital. In other words, the political condi-tions necessary to obtain Scandinavian-type social democracy and“solidaristic bargaining” might be difficult to achieve in developingcountries, despite their potential economic benefits. See Moene andWallerstein (2002). But this was not always the case; high levels oftrust emerged from widespread conflict in the mid-1930s.

35. Nickell (1997).36. Boeri (2002); Blunch and Verner (2004); Bover, Bentolila,

and Arellano (2002); Chaykowsky and Slotsve (2002); Panagidesand Patrinos (1994).

37. Bertola, Blau, and Kahn (2001).38. Kingdon, Sandefur, and Teal (2005).39. Damiani (2003).40. See Neumark, Cunningham, and Siga (forthcoming) for a

discussion of how Brazil’s minimum wage does not seem to havelifted family incomes at the lower points of the income distri-bution. In Colombia, Arango and Pachón (2004) find that “the

minimum wage ends up being regressive, improving the livingconditions of families in the middle and the upper part of theincome distribution with net losses for those at the bottom.”

41. See the ILO Web site (www.ilo.org) for a description of laborstandards. As an example of the range of policy choices for the spe-cific case of child labor, see ILO (2003) and Burra (1995).

42. See World Bank (2004a). Almost 80 percent of Cambodia’soverseas buyers rated labor standards as one of their top prioritiesin sourcing decisions. They also stated that standards have had pos-itive effects on accident rates, workplace productivity, productquality, turnover, and absenteeism.

43. López (2004).44. There are two ways to measure outcomes: ex ante simula-

tion using CGE (Computable General Equilibrium) analysis andex post econometric analysis relying on household survey data. Inthe absence of such analysis, it is hard to predict the impact oftrade reform on poverty or equity. A number of studies related tothe themes of this section can be found in Hertel and Winters(2005).

45. Schiff and Valdés (1998), 30.46. The CGE analysis generates price and wage changes that

include the direct price effect of trade policy changes and“second-round” indirect effects on factor returns and nontradedgoods prices. Certain dynamic gains from greater trade opennessare not captured. For instance, trade could bring new technolo-gies and innovations that boost long-term productivity (Raval-lion 2004b).

47. Winters, McCulloch, and McKay (2004); Mundlak and Lar-son (1992); Lloyd and others (1999); McKay, Morrissey, and Vail-lant (1997).

48. Minot and Goletti (1998).49. See Bates (1981) and chapter 6 for Ghana, and CUTS (Con-

sumer Unity and Trust Society) (2003) for Malawi.50. McMillan, Rodrik, and Horn Welsh (2002).51. See Hanson (2003). Other research attributes some of the

benefits of trade reform in border states to higher levels of humanand industrial capital and better communications and transporta-tion infrastructure (Chiquiar 2005).

52. See Nicita (2004) on Mexico and Goetz (1992); Interna-tional Fund for Agricultural Development (2001); Minot (1998);and Thomas and others (1999) on Indonesia.

53. Arulpragasam and others (2004), World Bank (2003c).54. Annamalai and Rao (2003), India Today (2004).55. Palmade (2005).56. Carruthers, Bajpai, and Hummels (2004).57. Wood (1997).58. See Wood (1997), Sánchez-Páramo and Schady (2003), and

De Ferranti and others (2004).59. Topalova (2004).60. World Bank (2004i), Cord and Wodon (2001).61. McMillan, Rodrik, and Horn Welsh (2002).62. For a brief survey of the different generations of models of

crises, see Krugman (1999). For selected references see Aghion,Bacchetta, and Banerjee (2001), Chang and Velasco (2001), Krug-man (1979), Obstfeld (1996), and Velasco (1996).

63. Acemoglu and others (2003).64. On Bolivia, see Morales and Sachs (1998); on Israel, see

Bruno (1993).


65. Blejer and Guerrero (1990).66. Ferreira and Litchfield (2001).67. See also Diwan (2002).68. Rodrik (1999a).69. Honohan and Klingebiel (2000).70. Halac and Schmukler (2003).71. See De Ferranti and others (2004), Lanjouw and Ravallion

(2005).72. Bruno (1993).73. Perry (2003).74. Robinson (2003); see focus 5 on taxation for discussion of

social-fiscal contracts.

Focus 6 on regional inequality

1. Massey (2001).2. Shepherd and others (2005); Massey (2001).3. Psacharopoulos and Patrinos (1994); Hall and Patrinos (2005);

Vakis (2003).4. de Haan (2005).5. Galiani and Schargrodsky (2002).6. Manor (1999).7. Fox (1990).8. Ferreira (2004).9. Funck and Pizzati (2003).10. Shepherd and others (2005); Willanakuy (2004).11. Ross (2005).

Chapter 10

1. Pogge (2004).2. Chauvet and Collier (2004) estimated that the cost, to a coun-

try and its neighbors, of a country descending into the status oflow-income country under stress (LICUS), as defined by the WorldBank, is in present value terms approximately $80 billion, and thatthis cost was mostly borne by neighbors.

3. These conclusions are consistent with the findings of theWorld Commission on the Social Dimensions of Globalization(2004).

4. Goldin and others (forthcoming).5. That is, 4 percent not 4 percentage points.6. Stark and Bloom (1985); Cox, Eser, and Jimenez (1998).7. The assumption that income after migration is higher than

income forgone does not always hold. Adams finds that incomelevels would have been higher without migration in three Egyptianvillages and would have been the same in four rural districts inPakistan (Adams 1989, 1992). Barham and Boucher (1998) findessentially similar results for three neighborhoods in Nicaragua.On the potential impact on income inequality, Mendola (2004)finds that remittances from members of households with largeland holdings in rural Bangladesh are correlated with adoption ofnew agricultural technologies, thus leading to higher productivityfor already better-off households and a possible increase ininequality.

8. McKenzie and Rapoport (2004).9. A number of studies by the World Bank Research Program on

International Migration and by others document the positiveimpact of remittances. See Taylor (1992), Taylor and Wyatt (1996),and Yang (2004) for examples of easing of credit constraints.Adams (2005) finds that Guatemalan households receiving remit-

tances spend significantly more on housing, which has positiveindirect effects on wages, business, and employment opportunities.Yang (2004) finds that Filipino households receiving higher remit-tances, because of positive exchange rate shocks, had greater childschooling, reduced child labor, more hours worked in self-employ-ment, and greater entry into relatively capital-intensive enterprises.

10. Ehrenreich and Hochschild (2003).11. On the phenomenon of brain waste, see Mattoo, Neagu, and

Ozden (2005); on brain drain, see Faini (2003) and Schiff (2005).12. Mode IV is one of four interrelated modes of supply of serv-

ices across borders considered in the GATS.13. As of May 2004, only 25 countries, mostly migrant-sending,

had ratified the convention.14. An interesting example of innovative partnerships to facili-

tate remittance flows is the New Alliance Task Force launched bythe Federal Deposit Insurance Corporation (FDIC) and MexicanConsulate of Chicago, which aims to improve access to the U.S.banking system, provide financial education, and develop productswith remittance features.

15. Smuggling and trafficking are a multibillion dollar industryincreasingly run by criminal networks. The United Nations Inter-national Organization for Migration (2000) estimated that up to 2million women and children are trafficked globally every year.

16. One option to address these losses is to lock developmentassistance to poor countries hurt by specific measures into theWTO agreements on these measures; see Ricupero (2005).

17. The current tariff structure could change through bothbilateral and multilateral negotiations. For instance, if the CentralAmerican Free Trade Agreement is approved by the U.S. Congress,most textiles from countries such as Honduras and El Salvadorwould benefit from tariff-free entry into the United States, now thecase for some items under existing unilateral preferences.

18. See for example Birdsall (2002).19. Price-fixing by international cartels brings significant losses

to developed- and developing-country consumers alike. Analysis ofsix high-profile cartels uncovered in the 1990s (vitamins, citricacid, bromine, seamless steel tubes, graphite electrodes, lysine)indicated that the estimated price increases ranged from 10 percentfor stainless steel tubes to 45 percent for graphite electrodes, andthat cumulative overcharges to developing countries over the life ofthe six cartels ranged from $3 billion to $7 billion, depending onthe calculation method. See Connor (2001), OECD (2000), andWorld Bank (2003b).

20. Ronchi (2001) and Ronchi (2002).21. Lewin, Giovannucci, and Varangis (2004).22. For more information, see www.ethicaltrade.org and

www.fairlabor.org.23. Fung, O’Rourke, and Sabel (2001).24. “Trade-Related Aspects of Intellectual Property Rights,”

Annex 1C of the Agreement Establishing the World Trade Organi-zation.

25. Lanjouw and Jack (2004).26. Sell (2003).27. Fink and Reichenmuller (2005).28. These are the agreements between the United States and

Morocco (2004), Bahrain (2004), and the Dominican Republic-CAFTA (signed but not yet approved by the U.S. Congress).

29. Claessens and Underhill (2004).

Endnotes 245

30. World Bank (2005g) and UNCTAD (2004).31. See Claessens, Underhill, and Zhang (2003) and Bhat-

tacharya and Griffith-Jones (2004).32. We discuss representation on the World Bank and IMF

Boards below.33. O’Sullivan and Christensen (2005).34. Moore (2004).35. McGillivray (2005).36. Updating the results for 2001 led to similar results (Levin

2005).37. The findings of Burnside and Dollar (2000) and Collier and

Dollar (2001), Collier and Dollar (2002) are not robust to changesin functional form, the specification of the interaction terms, andsample selection; see Hansen and Tarp (2001), Easterly, Levine, andRoodman (2004), and Dalgaard, Hansen, and Tarp (2004). Thesestudies also point to the importance of climate, among other fac-tors.

38. Levin and Dollar (2005).39. These figures include official development assistance

(ODA) provided by countries that are members of the Develop-ment Assistance Committee (DAC) of the Organisation forEconomic Co-operation and Development (OECD). Non-DACcountries also provide ODA; the largest non-DAC donors areSaudi Arabia (providing $2.4 billion in 2003 out of a total of $3.4billion), the Republic of Korea ($366 million), the United ArabEmirates ($188 million), and Kuwait ($133 million).

40. Eighteen countries qualify immediately, and as many asanother 20 could qualify over time. See the G-8 Finance Ministers’Conclusions on Development, London, June 10–11, 2005 (G-8Finance Ministers, 2005) at http://www.hm-treasury.gov.uk/oth-erhmtsites/g7/news/conclusions_on_development_110605.cfm.The G8 Finance Ministers’ agreement was confirmed at the July2005 G8 Gleneagles Summit.

41. See the G-24 Communique, available at http://web.worldbank.org/WBSITE/EXTERNAL/DEVCOMMEXT/0,,menuPK:60001663~pagePK:64001141~piPK:64034162~theSitePK:277473,00.html.

Focus 7 on drug access

1. See Lanjouw and MacLeod (2005) and World Health Organi-zation (2004).

2. See Kremer and Glennester (2004) and Barder (2004).3. See Masters (2005).4. See an interesting article by Maurer, Sali, and Rai (2004) on

how an open-source approach could work.5. World Health Organization (2004).6. Based on data from IMS HEALTH Global Services at http://

www.ims-global.com.7. This proposal is detailed in Lanjouw (2002).8. For legal details see Lanjouw (2002).

Epilogue

1. This is evident from the very title of Bauer (1971).2. World Bank (1980).3. World Bank (1990).4. World Bank (1997c).5. World Bank (2002b).6. World Bank (2001h).7. See Stern, Dethier, and Rogers (2005) for a detailed treatment

of this synthesis.8. World Bank (2005h).9. World Bank (2003j).


The word “Processed” describes informally reproduced works thatmay not be commonly available through libraries.

Acemoglu, Daron, Simon Johnson, and James Robinson. 2001.“The Colonial Origins of Comparative Development: An Empir-ical Investigation.” American Economic Review 91(5):1369–401.

————. 2002a.“Reversal of Fortune: Geography and Institutionsin the Making of the Modern World Income Distribution.”Quarterly Journal of Economics 117(4):1231–94.

————. 2002b. “The Rise of Europe: Atlantic Trade, InstitutionalChange, and Economic Growth.” Cambridge, MA: NationalBureau of Economic Research Working Paper Series 9378.

————. 2004. “Institutions as a Fundamental Cause of Long-runGrowth.” In Philippe Aghion and Steven Durlauf, (ed.), Hand-book of Economic Growth. Amsterdam: North Holland.

Acemoglu, Daron, Simon Johnson, James Robinson, and YunyongThaicharoen. 2003. “Institutional Causes, MacroeconomicsSymptoms: Volatility, Crises and Growth.” Journal of MonetaryEconomics 50(1):49–131.

Acemoglu, Daron, and James Robinson. 2000. “Why Did the WestExtend the Franchise? Growth, Inequality and Democracy in His-torical Perspective.” Quarterly Journal of Economics 115(4):1167–99.

————. 2005. Economic Origins of Dictatorship and Democracy.Cambridge, U.K.: Cambridge University Press.

Adams, Martin. 2000. Breaking Ground: Development Aid for LandReform. London: Overseas Development Institute.

Adams, Richard. 1989. “Workers’ Remittances and Inequalities inRural Egypt.” Economic Development and Cultural Change38(1):45–71.

————. 1992. “The Impact of Migration and Remittances onInequality in Rural Pakistan.” Pakistan Development Review31(4):1189–203.

Adams, Richard H. Jr. 2005.“Remittances, Household Expendituresand Investment in Guatemala.” Washington, DC: World BankPolicy Research Working Paper Series 3532. Available on line athttp://econ.worldbank.org/resource.php?type=5.

Agarwal, Bina. 1994. A Field of One’s Own: Gender and Land Rightsin South Asia. Cambridge, UK: Cambridge University Press.

Agell, Jonas. 2002. “On the Determinants of Labor Market Institu-tions: Rent Seeking vs. Social Insurance.” German EconomicReview 3(2):107–35.

Aghion, Philippe, Philippe Bacchetta, and Abhijit Banerjee. 2001.“A Corporate Balance Sheet Approach to Currency Crises.” Lon-don: Centre for Economic Policy Research 3092.

Alatas, Vivi, and François Bourguignon. 2004. “The Evolution ofIncome Distribution during Indonesia’s Fast Growth, 1980–96.”In François Bourguignon, Francisco Ferreira, and Nora Lustig,(ed.), The Microeconomics of Income Distribution Dynamics: InEast Asia and Latin America. New York: Oxford University Pressfor the World Bank.

Aleem, Irfan. 1990. “Imperfect Information, Screening, and theCosts of Informal Lending: A Study of Rural Credit Market inPakistan.” World Bank Economic Review 4(3):329–49.

Alesina, Alberto, Rafael Di Tella, and Robert MacCulloch. 2004.“Inequality and Happiness: Are Europeans and Americans Dif-ferent?” Journal of Public Economics 88(n9-10):2009–42.

Alesina, Alberto, and Edward Glaeser. 2004. Fighting Poverty in theUS and Europe: A World of Difference. Oxford: Oxford UniversityPress.

Alland, Denis, and Stéphanie Rials. 2003. Dictionnaire de la CultureJuridique. Paris: Presses Universitaires de France.

Alston, Lee J., Gary D. Libecap, and Bernardo Mueller. 1999. “AModel of Rural Conflict: Violence and Land Reform Policy inBrazil.” Environment and Development Economics 4(2):135–60.

Andersson, Kim, and Will Martin. 2004. “Agricultural TradeReform and the Doha Development Agenda.” Paper presented atthe International Trade Brown Bag Lunch Seminar. December15. Washington, DC.

Andersson, Martin, and Christer Gunnarsson. 2005. “Egalitarian-ism in the Process of Modern Economic Growth: The Case ofSweden.” Background paper for the WDR 2006.

Angrist, Joshua D., Eric Bettinger, Eric Bloom, Elizabeth M. King,and Michael Kremer. 2002. “Vouchers for Private Schooling inColombia: Evidence from a Randomized Natural Experiment.”American Economic Review 92(5):1535–58.

Annabi, Nabil, Bazlul Khandkher, Selim Raiham, John Cockburn,and Bernard Decaluwe. Forthcoming. “The Impact of TradeReforms on Bangladesh.” In Thomas W. Hertel and L. Alan Win-ters (ed.) Putting Development Back into the Doha Agenda: PovertyImpacts of a WTO Agreement. Washington, DC: World Bank.

Annamalai, Kuttayan, and Sachin Rao. 2003. What Works: ITC’s E-Choupal and Profitable Rural Transformation. University ofMichigan: World Resource Institute Digital Dividend. Availableon line at http://povertyprofit.wri.org/.

Appadurai, Arjun. 2004. “The Capacity to Aspire: Culture and theTerms of Recognition.” In Vijayendra Rao and Michael Walton,(ed.), Culture and Public Action. Stanford, CA: Stanford Univer-sity Press.

247

References

Appu, P. S. 1997. Land Reforms in India: A Survey of Policy, Legisla-tion and Implementation. New Delhi: Vikas Publishing House.

Arango, Carlos A., and Angelica Pachón. 2004. “Minimum Wages inColombia: Holding the Middle with a Bite on the Poor.” Colombia:Banco de la República (The Colombian Central Bank) 280. Avail-able on line at http://www.banrep.gov.co/docum/ftp/borra280.pdf.

Araujo, Caridad, Francisco Ferreira, and Norbert Schady. 2004. “Isthe World Becoming More Unequal? Changes in the WorldDistribution of Schooling.” World Bank. Washington, DC.Processed.

Ardington, E., and F. Lund. 1995. “Pensions and Development:Social Security as Complementary to Programmes of Recon-struction and Development.” Development Southern Africa12(4):557–77.

Arellano, Manuel, and Stephen Bond. 1991. “Some Tests of Speci-fication for Panel Data: Monte Carlo Evidence and an Applica-tion to Employment Equations.” Review of Economic Studies58(2):277–97.

Armendáriz de Aghion, Beatriz, and Jonathan Morduch. 2005. TheEconomics of Microfinance. Cambridge, MA: MIT Press.

Arnaud, André-Jean, eds. 1993. Dictionnaire Encyclopédique deThéory et de Sociologie du Droit (2nd edition). Paris: LibrairieGénérale de Droit et de Jurisprudence.

Arulpragasam, Jehan, Francesco Goletti, Tamar Manuelyan Atinc,and Vera Songwe. 2004. “Trade in Sectors Important to the Poor:Rice in Cambodia and Vietnam and Cashmere in Mongolia.” InKathie Krumm and Homi Kharas, (ed.), East Asia Integrates.Washington, DC: Oxford University Press for the World Bank.

Asian Development Bank. 2003. Judicial Independence Overviewand Country-Level Summaries. Manila, Philippines: Asian Development Bank. Available on line at http://www.adb.org/Documents/Events/2003/RETA5987/Final_Overview_Report.pdf.

Assunção, Juliano Junqueira, and Humberto Moreira. 2001.“Towards a Truthful Land Taxation Mechanism in Brazil.” Paperpresented at the LACEA. October 18. Montevideo, Uruguay.

Atim, C. 1999. “Social Movements and Health Insurance: A CriticalEvaluation of Voluntary, Non-profit Insurance Schemes withCase Studies from Ghana and Cameroon.” Social Science andMedicine 48(7):881–896.

Atkinson, A. B. 1995. Incomes and the Welfare State: Essays onBritain and Europe. Cambridge: Cambridge University Press.

Atkinson, Anthony B. 2003. “Income Inequality in OECD Coun-tries: Data and Explanations.” Munich: CESifo Working PaperSeries 881.

Atkinson, Anthony B., and Andrea Brandolini. 2001. “Promise andPitfalls in the Use of ‘Secondary’ Data-sets: Income Inequality inOECD Countries as a Case Study.” Journal of Economic Literature39(3):771–99.

————. 2004. “Global World Inequality: Absolute, Relative orIntermediate?” Paper presented at the 28th General Conferenceof the International Association for Research on Income andWealth. August 22. Cork, Ireland.

Atkinson, Anthony B., A. Maynard, and Christopher Trinder. 1983.Parents and Children: Incomes in Two Generations. London:Heinemann.

Atkinson, Anthony B., and Joseph E. Stiglitz. 1980. Lectures on Pub-lic Economics. London: McGraw-Hill International Editions.

Attanasio, Orazio P., and A. Marcos Vera-Hernandez. 2004.“Medium and Long Run Effects of Nutrition and Child Care:Evaluation of a Community Nursery Programme in RuralColombia.” London: Institute for Fiscal Studies Working PaperEWP04/06. Available on line at http://www.ifs.org.uk/publications.php?publication_id=3146.

Augustinus, Clarisa. 2003.“Comparative Analysis of Land Adminis-tration Systems: African Review with Special Reference toMozambique, Uganda, Namibia, Ghana, South Africa.” WorldBank. Washington, DC. Processed.

Australian Bureau of Statistics. 2003. Deaths in Australia 2002. Can-berra: Australian Bureau of Statistics.

Ayadi, Mohamed, Ghazi Boulila, Mohamed Lahouel, and PhilippeMontigny. 2004. Pro-Poor Growth in Tunisia. Paris: InternationalDevelopment and Strategies. Available on line at http://www.kfw-entwicklungsbank.de/EN/Fachinformationen/Pro-PoorGr77/oppgtunisia.pdf.

Azam, Jean-Paul, Magueye Dia, Clarence Tsimpo, and QuentinWodon. 2005. “Has Growth in Senegal after the 1994 Devalua-tion Been Pro-Poor?” World Bank. Washington, DC. Processed.

Baffes, John. 2004. “Cotton: Market Setting, Trade Policies andIssues.” Washington, DC: World Bank Policy Research WorkingPaper Series 3218.

Baiocchi, Gianpolo, Shubham Chaudhuri, and Patrick Heller. 2005.“Evaluating Empowerment: Participatory Budgeting in Brazil.”World Bank. Washington, DC. Available on line at http://siteresources.worldbank.org/INTEMPOWERMENT/Resources/Brazilpres.pdf. Processed.

Bakewell, Peter J. 1984. Miners of the Red Mountain. Albuquerque:University of New Mexico Press.

Balat, Jorge, and Guido Porto. Forthcoming. “The WTO DohaRound, Cotton Sector Dynamics and Poverty Trends in Zambia.”In Thomas W. Hertel and L. Alan Winters (ed.) Putting Develop-ment Back into the Doha Agenda: Poverty Impacts of a WTOAgreement. Washington, DC: World Bank.

Banerjee, Abhijit. 2000. “Land Reforms: Prospects and Incentives.”In Boris Pleskovic and Joseph E. Stiglitz, (ed.), Annual WorldBank Conference on Development Economics, 1999. Washington,DC: World Bank.

Banerjee, Abhijit, Shawn Cole, Esther Duflo, and Leigh Linden.2004. “Remedying Education: Evidence from Two RandomizedExperiments in India.” Massachusetts Institute of Technology.Cambridge, MA. Available on line at http://econ-www.mit.edu/faculty/download_pdf.php?id=677. Processed.

Banerjee, Abhijit, Angus Deaton, and Esther Duflo. 2004. “HealthCare Delivery in Rural Rajasthan.” Cambridge, MA: PovertyAction Lab Papers 7.

Banerjee, Abhijit, and Esther Duflo. 2003. “Inequality and Growth:What Can the Data Say?” Journal of Economic Growth8(3):267–99.

————. 2004a. “Do Firms Want to Borrow More? Testing CreditConstraints using a Direct Lending Program.” London: CEPRWorking Paper Series 4681.


————. 2004b. “Growth Theory Through the Lens of Develop-ment Economics.” Cambridge, MA: MIT Department of Eco-nomic Working Papers 05-01.

Banerjee, Abhijit, Esther Duflo, and Kaivan Munshi. 2003. “TheMiss(allocation) of Capital.” Journal of the European EconomicAssociation 1(2-3):484–94.

Banerjee, Abhijit, Paul Gertler, and Maitreesh Ghatak. 2002.“Empowerment and Efficiency: Tenancy Reform in West Ben-gal.” Journal of Political Economy 110(2):239–80.

Banerjee, Abhijit, Dilip Mookherjee, Kaivan Munshi, and DebrajRay. 2001. “Inequality, Control Rights, and Rent Seeking: SugarCooperatives in Maharashtra.” Journal of Political Economy109(1):138–90.

Banerjee, Abhijit, and Kaivan Munshi. 2004. “How Efficiently isCapital Allocated? Evidence from the Knitted Garment Industryin Tirupur.” Review of Economic Studies 71(1):19–42.

Banerjee, Abhijit, and Andrew Newman. 1991. “Risk-Bearing andthe Theory of Income Distribution.” Review of Economic Studies58(2):211–35.

Banerjee, Abhijit, and Thomas Piketty. 2003. “Top Indian Incomes:1956–2000.” Cambridge, MA: MIT Department of EconomicsWorking Paper 03-32.

Barber, Catherine. 2003. “Making Migration ‘Development-Friendly’.” Kennedy School of Government Master’s thesis. Har-vard University.

Barber, Sarah, Stefan Bertozzi, and Paul Gertler. 2005. “Variationsin Prenatal Care Quality in Mexico Mirror Health Inequalities.”World Bank. Washington, DC. Processed.

Barber, Sarah, Paul J. Gertler, and Pandu Harimurti. 2005. “Pro-moting High Quality Care in Indonesia: Roles for Public and Pri-vate Ambulatory Care Providers.” World Bank. Washington, DC.Processed.

Bardasi, Elena, and Quentin Wodon. 2004. “Comparing Subsidiesfor Access or Consumption in Basic Infrastructure: A SimpleApproach.” World Bank. Washington, DC. Processed.

Barder, Owen. 2004. “Making Markets for Vaccines: A PracticalPlan.” Washington, DC: Center for Global Development Brief 1. Available on line at http://www.cgdev.org/Publications/?PubID=173.

Barham, Bradford, and Stephen Boucher. 1998. “Migration, Remit-tances and Inequality: Estimating the Net Effect of Migration on Income Distribution.” Journal of Development Economics55(2):307–31.

Barrientos, Armando. 2005. “Cash Transfers for Older PeopleReduce Poverty and Inequality.” Background paper for the WDR2006.

Barro, Robert J. 2000. “Inequality and Growth in a Panel of Coun-tries.” Journal of Economic Growth 5(1):5–32.

Barro, Robert J., and Jong-Wha Lee. 2001. “International Data onEducational Attainment: Updates and Implications.” Oxford Eco-nomic Papers 53(3):541–63.

Barron, Patrick, Rachel Diprose, and Michael Woolcock. 2005.“Local Conflict and Community Development in Indonesia:Assessing the Impact of the Kecamatan Development Program.”World Bank Jakarta Regional Office. Jakarta. Processed.

Barron, Patrick, Claire Q. Smith, and Michael Woolcock. 2004.“Understanding Local Level Conflict in Developing Countries:Theory, Evidence and Implications for Indonesia.” Washington,DC: World Bank, Social Development Papers, Conflict Preven-tion & Reconstruction 19.

Basta, S., D. Soekirman, and N. Scrimshaw. 1979. “Iron DeficiencyAnemia and the Productivity of Adult Males in Indonesia.”American Journal of Clinical Nutrition 32(4):916–25.

Bates, Robert H. 1981. Markets and States in Tropical Africa: ThePolitical Basis of Agricultural Policies. Berkeley, CA: University ofCalifornia Press.

————. 1989. Beyond the Miracle of the Market. Cambridge, NY:Cambridge University Press.

Bauer, Peter. 1971. Dissent on Development: Studies and Debates inDevelopment Economics. London: Weidenfeld and Nicolson.

Behrman, Jere, Nancy Birdsall, and Miquel Székely. 2003.“Economic Policy and Wage Differentials in Latin America.”Washington, DC: Center for Global Development WorkingPaper 29. Available on line at http://www.cgdev.org/Publications/?PubID=29.

Bénabou, Roland. 1996. “Inequality and Growth.” In Ben Bernankeand Julio J. Rotemberg, (ed.), National Bureau of EconomicResearch Macroeconomics Annual 1996. Cambridge, MA: MITPress.

————. 2000. “Unequal Societies: Income Distribution and theSocial Contract.” American Economic Review 90(1):96–129.

Bennett, T. W. 1999. Human Rights and African Customary Lawunder the South African Constitution. Cape Town: Juta and Co.

Bentham [1781], Jeremy. 2000. The Principles of Morals and Legisla-tion. Kitchener, Ontario Canada: Batoche Books.

Berry, R. Albert, and William Cline. 1979. Agrarian Structure andProductivity in Developing Countries: A Study Prepared for theInternational Labour Office within the Framework of the WorldEmployment Program. Baltimore: Johns Hopkins UniversityPress.

Bertaud, Alain, and Stephen Malpezzi. 2001. “Measuring the Costsand Benefits of Urban Land Use Regulation: A Simple Modelwith an Application to Malaysia.” Journal of Housing Economics10(3):393–418.

Bertola, Giuseppe. 1990. “Job Security, Employment and Wages.”European Economic Review 34(4):851–86.

————. 2003. “Distribution, Efficiency, and Labor Market Regu-lation in Theory.” Paper presented at the Séptima ConferenciaAnual del Banco Central de Chile: Mercado Laboral e Institu-ciones. November 6. Santiago de Chile.

Bertola, Giuseppe, Francine D. Blau, and Lawrence M. Kahn. 2001.“Comparative Analysis of Labor Market Outcomes: Lessons forthe US from International Long-run Evidence.” Cambridge, MA:National Bureau of Economic Research Working Paper Series8526.

Bertrand, Marianne, and Sendhil Mullanaithan. 2003. “Are Emilyand Greg more Employable than Lakisha and Jamal?: A FieldExperiment on Labor Market Discrimination.” Cambridge, MA:National Bureau of Economic Research Working Paper Series9873.

References 249

Bertrand, Marianne, Antoinette S. Schoar, and David Thesmar.2004. “Banking Deregulation and Industry Structure: Evidencefrom the French Banking Reforms of 1985.” London: CEPR Dis-cussion Papers 4488.

Besley, Timothy, and Robin Burgess. 2000. “Land Reform, PovertyReduction, and Growth: Evidence from India.” Quarterly Journalof Economics 115(2):389–430.

————. 2004. “Can Labor Regulation Hinder Economic Perfor-mance? Evidence from India.” Quarterly Journal of Economics119(1):91–134.

Besley, Timothy, Rohini Pande, Lupin Rahman, and VijayendraRao. 2004. “The Politics of Public Good Provision: EvidenceFrom Indian Local Governments.” Journal of the European Eco-nomic Association 2(2-3):416–26.

Bhagwati, Jagdish. 2003. “Borders Beyond Control.” Foreign Affairs82(1):98–104.

Bhattacharya, Amar, and Stephany Griffith-Jones. 2004. “TheSearch for a Stable and Equitable Global Financial System.” InJan Joost Teunissen and Age Akkerman, (ed.), Diversity in Devel-opment: Reconsidering the Washington Consensus. Washington,DC: Fondad.

Binswanger, Hans, Klaus Deininger, and Gershon Feder. 1995.“Power Distortions, Revolt and Reform in Agricultural LandRelations.” In Jere Behrman and T. N. Srinivasan, (ed.), Handbookof Development Economics, vol. 3B. Amsterdam: North Holland.

Binswanger, Hans, and Mark Rosenzweig. 1986. “Behavioural andMaterial Determinants of Production Relations in Agriculture.”Journal of Development Studies 22(3):503–39.

Bird, Richard M. 1991. More Taxing than Taxes? The Tax-like Effectsof Non-tax Policies in LDC’s. San Francisco: ICS Press.

Bird, Richard M., Jorge Martínez-Vazquez, and Benno Torgler. 2004.“Societal Institutions and Tax Effort in Developing Countries.”Toronto:University of Toronto, International Tax Program Paper 04011.

Bird, Richard M., and Barbara Diane Miller. 1989. “The Incidenceof Indirect Taxes on Low-income Households in Jamaica.” Eco-nomic Development and Cultural Change 37(2):393–409.

Bird, Richard M., and Enid Slack. 2002. “Land and Property Taxa-tion: A Review.” Paper presented at the Workshop on Land Issuesin Latin American and the Caribbean. May 19. Pachuca, México.

Birdsall, Nancy. 1985. “Public Inputs and Child Schooling inBrazil.” Journal of Development Economics 18(1):67–86.

————. 2002. “Asymmetric Globalization. Global MarketsRequire Good Global Politics.” Washington, DC: Center forGlobal Development Working Paper 12.

————. 2004. “Seven Deadly Sins: Reflections on Donor Fail-ings.” Washington, DC: Center for Global Development WorkingPaper Series 50.

Birdsall, Nancy, Carol Graham, and Richard Sabot. 1998. BeyondTradeoffs: Market Reforms and Equitable Growth in Latin Amer-ica. Washington, DC: Brookings Instittution Press.

Black, Bernard, Reinier Kraakman, and Anna Tassarova. 2000.“Russian Privatization and Corporate Governance: What WentWrong?” Stanford Law Review 52(6):1731–808.

Black, Maureen M. 2003.“Micronutrient Deficiencies and CognitiveFunctioning.” Journal of Nutrition 133(11 Suppl 2):3927S–3931S.

Blackhurst, Richard, Bill Lyakurwa, and Ademola Oyejide. 2000.“Options for Improving Africa’s Participation in the WTO.”World Economy 23(4):491–510.

Blair, Irene V., Charles M. Judd, and Kristine M. Chapleau. 2004.“The Influence of Afrocentric Facial Features in Criminal Sen-tencing.” Pscyhological Science 15(10):674–79.

Blanchard, Olivier. 2004. “Reforming Labor Market Institutions:Unemployed Insurance and Employment Protection.” Cam-bridge, MA: MIT Department of Economics Working PaperSeries 04-38.

Blejer, Mario, and Isabel Guerrero. 1990. “The Impact of Macro-economic Policies on Income Distribution: An Empirical Studyof the Philippines.” Review of Economics and Statistics72(3):414–23.

Bloch, Francis, and Vijayendra Rao. 2002. “Terror as a BargainingInstrument: A Case-study of Dowry Violence in Rural India.”American Economic Review 92(4):1029–43.

Blunch, Niels-Hugo, and Dorte Verner. 2004. “Asymmetries in theUnion Wage Premium in Ghana.” World Bank Economic Review18(2):237–52.

Boeke, Julius Herman. 1946. The Evolution of the Netherlands IndiesEconomy. New York: Netherlands and Netherlands Indies Coun-cil, Institute of Pacific Relations.

Boeri, Tito. 2002. “Increasing the Size of the European Labor Force:The Relevant Trade-offs.” Economic Survey of Europe 2:99–108.

Boix, Carles. 1998. Political Parties, Growth and Equality: Conserva-tive and Social Democratic Economic Strategies in the World Econ-omy. Cambridge, UK: Cambridge University Press.

————. 2005. “Spain: Development, Democracy and Equity.”Background paper for the WDR 2006.

Boskin, Michael J. 1977. “An Economist’s Perspective on Estate Tax-ation.” In Edward C. Halbach Jr., (ed.), Death, Taxes and FamilyProperty. St. Paul, MN: West Publishing Co.

Bottoms, A. E. 1990. “Crime Prevention Facing the 1990s.” Policyand Society 1(1):3–22.

Bourdieu, Pierre. 1986. “The Forms of Capital.” In John G. Richard-son, (ed.), Handbook of Theory and Research for the Sociology ofEducation. Westport, CT: Greenwood Press.

————. 1990. The Logic of Practice. Stanford, CA: Stanford Uni-versity Press.

Bourguignon, François. 2003. “The Growth Elasticity of PovertyReduction: Explaining Heterogeneity across Countries andTime-periods.” In T. Eichler and S. Turnovsky, (ed.), Growth andInequality. Cambridge, MA: MIT Press.

Bourguignon, François, Francisco Ferreira, and Phillippe G. Leite.2004. “Conditional Cash Transfers, Schooling, and Child Labor:Microsimulating Brazil's Bolsa Escola Program,” World BankEconomic Review 17(2):229–54.

Bourguignon, François, Francisco Ferreira, and Nora Lustig. 2004.The Microeconomics of Income Distribution Dynamics in East Asiaand Latin America. New York: Oxford University Press for theWorld Bank.

Bourguignon, François, Francisco Ferreira, and Marta Menendez.2005. “Inequality of Opportunity in Brazil?” World Bank. Wash-ington, DC. Processed.


Bourguignon, François, Victoria Levin, and David Rosenblatt.2004a. “Declining Economic Inequality and Economic Diver-gence: Reviewing the Evidence through Different Lenses.”Economie Internationale 100(4).

————. 2004b. “Global Redistribution: The Role of Aid,Market Access, and Remittances.” World Bank. Washington, DC.Processed.

Bourguignon, François, and Christian Morrison. 1990. “IncomeDistribution, Development and Foreign Trade: A Cross-sectionalAnalysis21`.” European Economic Review 34(6):1113–32.

Bourguignon, François, and Christian Morrisson. 1993. “ExternalTrade and Income Distribution.” Journal of Development Eco-nomics 41(1):207–9.

————. 1998. “Inequality and Development: The Role of Dual-ism.” Journal of Development Economics 57(2):233–57.

————. 2002. “Inequality among World Citizens: 1820–1992.”American Economic Review 92(4):727–44.

Bourguignon, François, and Thierry Verdier. 2000. “Oligarchy,Democracy, Inequality and Growth.” Journal of DevelopmentEconomics 62(2):285–313.

Bover, Olimpia, Samuel Bentolila, and Manuel Arellano. 2002. “TheDistribution of Earnings in Spain During the 1980s: The Effectof Skill, Unemployment and Union Power.” Madrid: Banco DeEspana, Servicio de Estudios, Documento de Trabajo 015. Avail-able on line at http://www.bde.es/informes/be/docs/dt0015e.pdf.

Bowen, William G., and Derek Bok. 1998. The Shape of the River:Long Term Consequences of Considering Race in College andUniversity Admissions. Princeton, NJ: Princeton UniversityPress.

Bowman, Larry W. 1991. Mauritius: Democracy and Development inthe Indian Ocean. Boulder, CO: Westview Press.

Brainerd, Elizabeth, and David M. Cuttler. 2004. “Autopsy on anEmpire: Understanding Mortality in Russia and the FormerSoviet Union.” Cambridge, MA: National Bureau of EconomicResearch Working Paper Series 10868.

Brasselle, Anne-Sophie, Frederic Gaspart, and Jean-Philippe Plat-teau. 2002. “Land Tenure Security and Investment Incentives:Puzzling Evidence from Burkina Faso.” Journal of DevelopmentEconomics 67(2):373–418.

Braverman, Avishay, and J. Luis Guasch. 1986. “Rural Credit Mar-kets and Institutions in Developing Countries: Lessons for PolicyAnalysis from Practice and Modern Theory.” World Development14(10/11):1253–67.

Brenner, Robert. 1976. “Agrarian Class Structure and EconomicDevelopment in Preindustrial Europe.” Past and Present70(1976):30–75.

Brockett, C. D. 1984. “Malnutrition, Public Policy, and AgrarianChange in Guatemala.” Journal of Interamerican Studies andWorld Affairs 26(4):477–97.

Brocklehurst, Clarissa, and Jan G. Janssens. 2004. “Innovative Con-tracts, Sound Relationships: Urban Water Sector Reform inSenegal.” Washington, DC: World Bank Water Supply and Sani-tation Sector Board Discussion Paper 1.

Brosnan, Sarah F., and Frans B. M. De Waal. 2003. “Monkeys RejectUnequal Pay.” Nature 425(6955):297–99.

Browning, Martin, and Pierre-Andre Chiappori. 1998. “EfficientIntra-household Allocation: A General Characterization andEmpirical Tests.” Econometrica 66(6):1241–78.

Bruno, Michael. 1993. Crisis, Stabilization, and Economic Reform:Therapy by Consensus. New York: Clarendon Press.

Bruno, Michael, Martin Ravallion, and Lyn Squire. 1998. “Equity andGrowth in Developing Countries: Old and New Perspectives on thePolicy Issues.” In Vito Tanzi and Ke-young Chu, (ed.), Income Dis-tribution and High-quality Growth. Cambridge, MA: MIT Press.

Bruns, Barbara, Alain Mingat, and Ramahatra Rakotomalala. 2003.Achieving Universal Primary Education by 2015: A Chance forEvery Child. Washington, DC: World Bank.

Buchanan, James M. 1976. “A Hobbesian Interpretation of theRawlsian Difference Principle.” Kyklos 29(1):5–25.

Bukurura, Sufian Hemed. 1994. “ The Maintenance of Order inRural Tanzania: The Case of the Sungusungu.” Journal of LegalPluralism 34:1–29.

Burgess, Robin, and Rohini Pande. 2004. “Do Rural Banks Matter?Evidence from the Indian Social Banking Experiment.” London:Center for Economic Policy Research Discussion Papers 4211.

Burney, Nadeem A., and Mohammad Irfan. 1995. “Determinants ofChild School Enrollment: Evidence from LDCs Using Choice-theoretic Approach.” International Journal of Social Economics22(1):24–40.

Burnside, Craig, and David Dollar. 2000. “Aid, Policies andGrowth.” American Economic Review 90(4):847–68.

Burra, Nera. 1995. Born to Work: Child Labour in India. New York:Oxford University Press.

Buscaglia, Edgardo. 1997. “Introduction.” In Edgardo Buscaglia,Louise Cord, and W. Ratliff, (ed.), Law and Economics of Develop-ment. Greenwich, CT: JAI Press Inc.

Buscaglia, Edgardo, and Maria Dakolias. 1999. An Analysis of theCauses of Corruption in the Judiciary. Washington, DC: WorldBank, Legal and Judicial Reform Unit.

Bush, Robert A. 1979. “Access to Justice and Societal Pluralism.” InMauro Cappelletti and Bryant Garth, (ed.), Access to Justice, Vol.3. Milan: Guiffre Editore.

Byerlee, D., and T. Kelley. 2004. “Surviving on the Margin: Agricul-tural Research and Development Strategies for Poverty Reduc-tion in Marginal Areas.” World Bank, Agriculture and RuralDevelopment. Washington, DC. Processed.

Cadiet, Loïc. 2004. Dictionnaire de la Justice. Paris: Presses Universi-taires de France.

Cain, Mead. 1981. “Risk and Insurance: Perspective on Fertility andAgrarian Change in India and Bangladesh.” Population andDevelopment Review 7(3):435–74.

Cameron, Lisa. 1999. “Raising the Stakes in the Ultimatum Game:Experimental Evidence from Indonesia.” Economic Inquiry37(1):47–59.

Campos, Javier, Antonio Estache, Noelia Martin, and Lourdes Tru-jillo. 2003. “Macroeconomic Effects of Private Sector Participa-tion in Infrastructure.” In William Easterly and Luis Servén,(ed.), The Limits of Stabilization. Washington, DC: World Bankand Stanford Social Sciences, an imprint of Stanford UniversityPress.

References 251

Cardenas, Juan-Camillo. 2003. “Real Wealth and ExperimentalCooperation: Experiments in the Field Lab.” Journal of Develop-ment Economics 70(2):263–89.

Cardoso, Ciro F. S. 1991. “The Liberal Era, 1870-1930.” In LeslieBethell, (ed.), Central America Since Independence. Cambridge,UK: Cambridge University Press.

Carneiro, Pedro, and James Heckman. 2003. “Human Capital Pol-icy.” Cambridge, MA: National Bureau of Economic ResearchWorking Paper Series 9495.

Carruthers, Robin, Jitendra N. Bajpai, and David Hummels. 2004.“Trade and Logistics: An East Asia Perspective.” In KathieKrumm and Homi Kharas, (ed.), East Asia Integrates: A TradePolicy Agenda for Shared Growth. Washington, DC: Oxford Uni-versity Press for the World Bank.

Casagrande, Joseph B., and Arthur R. Piper. 1969. “La Transforma-ción Estructural de una Parroquia Rural en las Tierras Altas delEcuador.” América Indígena 29:1029–64.

Case, Anne. 2001. “Does Money Protect Health Status? Evidencefrom South African Pensions.” Cambridge, MA: National Bureauof Economic Research Working Paper Series 8495.

Case, Anne, and Angus Deaton. 1998. “Large Cash Transfers to theElderly in South Africa.” Economic Journal 108(450):1330–61.

————. 1999. “School Quality and Educational Outcomes inSouth Africa.” Quarterly Journal of Economics 114(3):1047–84.

Castaneda, Tarsicio. 2003. “Targeting Social Spending to the Poorwith Proxy-Means Testing: Columbia’s SISBEN System.” WorldBank. Washington, DC. Processed.

Castello, Amparo, and Rafael Domenech. 2002. “Human CapitalInequality and Economic Growth: Some New Evidence.” Eco-nomic Journal 112(478):C187–C200.

Center for Global Development. 2004. Ranking the Rich: The 2004CGD/FP Commitment to Development Index. Washington, DC:Center for Global Development. Available on line at http://www.cgdev.org/rankingtherich/home.html.

Centre for Housing Rights and Evictions. 2004. Bringing EqualityHome: Promoting and Protecting the Inheritance Rights of Women.Geneva: Centre for Housing Rights and Evictions. Available online at http://www.cohre.org/downloads/womens-inheritance-rights-africa.pdf.

Chang, Roberto, and Andrés Velasco. 2001. “A Model of FinancialCrises in Emerging Markets.” Quarterly Journal of Economics116(2):489–517.

Chattopadhyay, Raghabendra, and Esther Duflo. 2004. “Women asPolicy Makers: Evidence from a Randomized Policy Experimentin India.” Econometrica 72(5):1409–43.

Chaturvedi, S., B. C. Srivastava, J. V. Singh, and M. Prasad. 1987.“Impact of Six Years Exposure to ICDS Scheme on Psycho-socialDevelopment.” Indian Pediatrics 24:153–64.

Chaudhuri, Shubham, Pinelopi K. Goldberg, and Panle Jia. 2004.“Estimating the Effects of Global Patent Protection in Pharma-ceuticals: A Case Study of Quinolones in India.” World Bank.Washington, DC. Available on line at http://www.econ.yale.edu/~pg87/TRIPS.pdf. Processed.

Chaudhuri, Shubham, K. N. Harilal, and Patrick Heller. 2004. DoesDecentralization Make a Difference? A Study of the Peoples Cam-

paign for Decentralized Planning in the Indian State of Kerala.New Delhi: Ford Foundation.

Chaudhuri, Shubham, and Patrick Heller. 2003. “The Plasticity ofParticipation: Evidence from a Participatory Governance Experi-ment.” New York: Columbia University ISERP Working Paper03-01.

Chaudhury, Nazmul, and Jeffery Hammer. 2004. “Ghost Doctors:Absenteeism in Rural Bangladeshi Health Clinics.” World BankEconomic Review 18(3):423–41.

Chaudhury, Nazmul, Jeffery Hammer, K. Muralidharan, and F. H.Rogers. 2005. “Missing in Action: Teacher and Health WorkerAbsence in Developing Countries.” World Bank. Washington,DC. Processed.

Chaudhury, Nazmul, Jeffery Hammer, and Edmundo Murru-garra. 2003. “The Effects of a Fee-Waiver Program on HealthCare Utilization among the Poor: Evidence from Armenia.”Washington, DC: World Bank Policy Research Working PaperSeries 2952.

Chauvet, Lisa, and Paul Collier. 2004. “Development Effectivenessin Fragile States: Spillovers and Turnarounds.” Oxford Univer-sity, Centre for the Study of African Economies. Oxford, UK.Available on line at http://www.oecd.org/dataoecd/32/59/34255628.pdf. Processed.

Chaykowsky, Richard P., and Richard A. Slotsve. 2002. “EarningsInequality and Unions in Canada.” British Journal of IndustrialRelations 40(3):493–519.

Chen, Martha, Joann Vanek, and Marilyn Carr. 2004. Mainstream-ing Informal Employment and Gender in Poverty Reduction. Lon-don: Commonwealth Secretariat.

Chen, Shaohua, and Martin Ravallion. 2004. “How Have TheWorld’s Poorest Fared since the Early 1980’s?” World BankResearch Observer 19(2):141–69.

Chenery, Hollis, Clive Bell, J. Duloy, and Richard Jolly. 1974. Redis-tribution with Growth. Oxford: Oxford University Press.

Chile’s Ministry of Planning. 2004. Sistema de Proteccion Social:Chile Solidario. Santiago de Chile: Ministry of Planning, Govern-ment of Chile.

Chiquiar, Daniel. 2005.“Why Mexico’s Regional Income ConvergenceBroke Down.” Journal of Development Economics 77(1):257–75.

Chirayath, Leila, Caroline Sage, and Michael Woolcock. 2005. “Cus-tomary Law and Policy Reform: Engaging with the Plurality ofJustice Systems.” Background paper for the WDR 2006.

Chisari, Omar, Antonio Estache, and Carlos Romero. 1999. “Win-ners and Losers from the Privatization and Regulation Utilities:Lessons from a General Equilibrium Model of Argentina.” WorldBank Economic Review 13(2):357–78.

Chua, Amy. 2004. World on Fire: How Exporting Free MarketDemocracy Breeds Ethnic Hatred and Global Instability. New York:Anchor Books.

Chubb, John E., and Terry M. Moe. 1990. Politics, Markets, andAmerica’s Schools. Washington, DC: The Brookings Institution.

Claessens, Stijn, Simeon Djankov, and Larry H. P. Lang. 2000. “TheSeparation of Ownership and Control in East Asian Corpora-tions.” Journal of Financial Economics 58(1-2):81–112.


Claessens, Stijn, and Enrico Perotti. 2005. “The Links BetweenFinance and Inequality: Channels and Evidence.” Backgroundpaper for the WDR 2006.

Claessens, Stijn, and Gerhard Pohl. 1994. “Banks, Capital Markets,and Corporate Governance: Lessons from Russia for EasternEurope.” Washington, DC: World Bank Policy Research WorkingPaper Series 1326.

Claessens, Stijn, and Geoffrey R. D. Underhill. 2004. “The Need forInstitutional Changes in the Global Financial System: An Analyt-ical Framework.” Paper presented at the Developing Countries,Global Finance, and the Role of the IMF: Towards a New Rela-tionship? November 12. The Hague.

Claessens, Stijn, Geoffrey R. D. Underhill, and Xiaoke Zhang. 2003.“Basle II Capital Requirements and Developing Countries: APolitical Economy Perspective.” Paper presented at the Quantify-ing the Impact of Rich Countries’ Policies on Poor Countries.October 23. Washington, DC.

Cline, William R. 2004. Trade Policy and Global Poverty. Washing-ton, DC: Institute for International Economics.

Coady, David, Margaret Grosh, and John Hoddinott. 2004. Target-ing Transfers in Developing Countries: Review of Lessons andExperience. Washington, DC: World Bank.

Coate, Stephen, and Glenn Loury. 1993. “AntidiscriminationEnforcement and the Problem of Patronization.” American Eco-nomic Review 83(2):92–8.

Coatsworth, John H. 1993. “Notes on the Comparative EconomicHistory of Latin America and the United States.” In W. L. Ber-necker and H. W Tobler, (ed.), Development and Underdevelop-ment in America: Contrasts of Economic Growth in North andLatin America in Historical Perspective. Berlin: de Gruyter.

Cogneau, Denis. 2005. “Equality of Opportunity and Other EquityPrinciples in the Context of Developing Countries.” Paris,France: Développement, Institutions & Analyses de Long terme(DIAL) Working Paper DT/2005/01.

Cogneau, Denis, and Jean-David Naudet. 2004. “Who DeservesAid? Equality of Opportunity, International Aid, and PovertyReduction.” Paper presented at the Equity and DevelopmentWorkshop. September 6. Berlin.

Cole, Jeffrey A. 1985. The Potosi Mita, 1573-1700: CompulsoryIndian Labor in the Andes. Palo Alto, CA: Stanford UniversityPress.

Collier, Paul, and David Dollar. 2001. “Can the World CutPoverty in Half? How Policy Reform and Effective Aid CanMeet International Development Goals.” World Development29(11):1787–802.

————. 2002. “Aid Allocation and Poverty Reduction.” EuropeanEconomic Review 46(8):1475–500.

Commission on Macroeconomics and Health. 2001. Macroeconom-ics and Health: Investing in Health for Economic Development.Geneva: World Health Organization.

Conning, Jonathan H., and James A. Robinson. 2002. “LandReform and the Political Organization of Agriculture.” London:CEPR Working Paper 3204.

Connor, John M. 2001. Global Price-Fixing: Our Customers Are theEnemy. Boston, MA: Kluwer Academic Publishing.

Cord, Louise, and Quentin Wodon. 2001. Do Mexico’s AgriculturalPrograms Alleviate Poverty: Evidence from the Ejido Sector. Wash-ington, DC: World Bank.

Cornia, Andrea Giovanni, and Leonardo Menchini. 2005. “ThePace and Distribution of Health Improvements during the Last40 Years: Some Preliminary Results.” Paper presented at theForum on Human Development. January 17. Paris.

Council for Scientific and Industrial Research. 2000. Making SouthAfrica Safe: A Manual for Community Based Crime Prevention.Pretoria: Deparment of Safety and Security, South African PoliceService.

Cowell, Frank A. 1995. Measuring Inequality (second edition).Wheatsheaf: Prentice Hall.

Cox, Donald, Zekeriya Eser, and Emmanuel Jimenez. 1998.“Motives for Private Transfers Over the Life Cycle: An AnalyticalFramework and Evidence for Peru.” Journal of Development Eco-nomics 55(1):57–80.

Craven, Wesley F. 1932. Dissolution of the Virginia Company. NewYork: Oxford University Press.

Crosby, Alfred. 1986. Ecological Imperialism: The Biological Expan-sion of Europe 900-1900. Cambridge, UK: Cambridge UniversityPress.

Cull, Robert, Jana Matesova, and Mary Shirley. 2002. “Ownershipand the Temptation to Loot: Evidence from Privatized Firms inthe Czech Republic.” Journal of Comparative Economics30(1):1–24.

Cullen, Julie Berry, Brian A. Jacob, and Steven Levitt. 2000. “TheImpact of School Choice on Student Outcomes: An Analysis ofthe Chicago Public Schools.” Cambridge, MA: National Bureauof Economic Research Working Paper Series 7888.

————. 2003. “The Effect of School Choice on Student Out-comes: Evidence from Randomized Lotteries.” Cambridge, MA:National Bureau of Economic Research Working Paper Series10113.

Currie, Janet. 2000.“Early Childhood Intervention Programs: Whatdo we Know?” Chicago: JCPR Working Paper 169.

Currie, Janet, and Duncan Thomas. 1995. “Does Head Start Make aDifference?” American Economic Review 85(3):341–64.

————. 1999. “Does Head Start Help Hispanic Children?” Inter-national Journal of Social Economics 74(2):235–62.

————. 2000. “School Quality and the Longer-Term Effects ofHead Start.” Journal of Human Resources 35(4):755–74.

CUTS (Consumer Unity and Trust Society). 2003. Spine ChillingExperiences of Anti-Competitive Practices in Malawi. India:CUTS.

Dakolias, Maria, and Kim Thatchuk. 2000. “The Problem of Eradi-cating Corruption from the Judiciary.” In Marco Fabri and PhilipM. Langbroek, (ed.), The Challenge of Change for Judicial Sys-tems. Amsterdam: IOS Press.

Dalgaard, Carl-Johan., Henrik Hansen, and Finn Tarp. 2004. “Onthe Empirics of Foreign Aid and Growth.” Economic Journal114(496):F191–F216.

Damiani, Octavio. 2003.“Effects on Employment, Wages and LaborStandards of Nontraditional Export Crops in Northeast Brazil.”Latin American Research Review 38(1):83–112.

References 253

Das Gupta, Monica, Sunhwa Lee, Patricia Uberoi, Danning Wang,Lihong Wang, and Xiaodan Zhang. 2004. “State Policies andWomen’s Agency in China, The Republic of Korea, and India,1950–2000: Lessons from Contrasting Experiences.” In Vijayen-dra Rao and Michael Walton, (ed.), Culture and Public Action.Stanford: Stanford University Press.

Das Gupta, Monica, Jian Zhenghua, Li Bohua, Xie Zhenming, Woo-jin Chung, and Bae Hwa-Ok. 2003. “Why is Son Preference soPersistent in East and South Asia? A Cross-country Study ofIndia, China and the Republic of Korea.” Journal of DevelopmentStudies 40(2):153–87.

Das, Hans. 2004. “Restoring Property Rights in the Aftermath ofWar.” International and Comparative Law Quarterly53(2):429–44.

Das, Jishnu, and Jeffery Hammer. 2005. “Poverty and the Access toQuality Health-Care: Evidence from Delhi.” World Bank. Wash-ington, DC. Processed.

Das, Jishnu, and Jeffrey Hammer. 2-28-2004. “Strained Mercy;Quality of Medical Care in Delhi.” Economic and Political Weekly.

Dasgupta, A., C. P. S. Nayar, and Associates. 1989. Urban InformalCredit Markets in India. New Delhi: National Institute of PublicFinance and Policy.

Datt, Gaurav, and Martin Ravallion. 1992. “Growth and Redistribu-tion Components of Changes in Poverty Measures: A Decompo-sition with Applications to Brazil and India in the 1980s.” Journalof Development Economics 38(2):275–95.

Davies, James B., and Anthony F. Shorrocks. 2005. “Wealth Hold-ings in Developing and Transition Countries.” Paper presented atthe Luxembourg Wealth Study Conference. January 27. Luxem-bourg.

Davis, Jeffrey, Rolando Ossowski, and Annalisa Fedelino, eds. 2003.Fiscal Policy Formulation and Implementation in Oil-ProducingCountries. Washington, DC: International Monetary Fund.

Day, Candy, and Calle Hedberg. 2004. “Health Indicators.” SouthAfrican Health Review 2003(4):349–420.

de Carvalho Filho, Irineue Evangelista. 2000. “Household Incomeas a Determinant of Child Labour and School Enrollment inBrazil: Evidence from a Social Security Reform.” Boston Univer-sity. Boston. Processed.

De Ferranti, David, Guillermo Perry, William Foster, Daniel Leder-man, and Alberto Valdés. 2005. Beyond the City: The Rural Con-tribution to Development. Washington, DC: World Bank.

De Ferranti, David, Guillermo E. Perry, Francisco H. G. Ferreira,and Michael Walton. 2004. Inequality in Latin America: Breakingwith History? Washington, DC: World Bank.

de Haan, Arjan. 2005. “Disparities Within India’s Poorest Regions:Why Do The Same Institutions Work Differently In DifferentPlaces?” Background paper for the WDR 2006.

De Janvry, Alain, and Elisabeth Sadoulet. 1989. “A Study in Resis-tance to Institutional Change: The Lost Game of Latin AmericanLand Reform.” World Development 17(9):1397–407.

————. 2002. “Land Reforms in Latin America: Ten Lessonstoward a Contemporary Agenda.” Paper presented at the WorldBank’s Latin American Land Policy Workshop. June 14. Pachuca,Mexico.

————. 2004. “Conditional Cash Transfer Programs: Are theyReally Magic Bullets?” University of California. Berkeley, CA.Available on line at http://are.berkeley.edu/~sadoulet/papers/ARE-CCTPrograms.pdf. Processed.

De Soto, Hernando. 2000. The Mystery of Capital: Why CapitalismTriumphs in the West and Fails Everywhere Else. New York: BasicBooks.

Dearden, Lorraine, Stephen Machin, and Howard Reed. 1997.“Intergenerational Mobility in Britain.” Economic Journal107(440):47–66.

Deaton, Angus. 1997. The Analysis of Household Surveys. Washing-ton, DC: World Bank.

————. 2003. “Health, Inequality and Economic Development.”Journal of Economic Literature 41(1):113–58.

————. 2004. “Health in an Age of Globalization.” Paper pre-sented at the Brookings Trade Forum. May 13. Brookings Institu-tion, Washington, DC.

Deaton, Angus, and Jean Drèze. 2002. “Poverty and Inequality inIndia: A Reexamination.” Economic and Political Weekly Septem-ber 7:3729–48.

Deaton, Angus, and Valerie Kozel. 2004. Data and Dogma: TheGreat Indian Poverty Debate. New Delhi: McMillan.

Decker, Klaus, Siobhan McInerney-Lankford, and Caroline Sage.2005. “Human Rights and Equitable Development: ‘Ideals’, Iis-sues and Implications.” Background paper for the WDR 2006.

Decker, Klaus, Caroline Sage, and Milena Stefanova. 2005. “Law orJustice: Building Equitable Legal Institutions.” Background paperfor the WDR 2006.

Deere, Carmen Diana, and Magdalena León. 2003.“The Gender AssetGap: Land in Latin America.” World Development 31(6):925–47.

Deininger, Klaus. Forthcoming. “Land Policy Reforms.” In AlineCoudouel and Stefano Paternostro (ed.) The DistributionalImpact of Reforms: A Practitioners’ Guide, Volume I, Trade, Mone-tary and Exchange Rate Policy, Utility Provision, Agricultural Mar-kets, Land, and Education. Washington, DC: World Bank.

Deininger, Klaus, Raffaella Castagnini, and María A. González.2004. “Comparing Land Reform and Land Markets in Colombia:Impacts on Equity and Efficiency.” Washington, DC: World BankPolicy Research Working Paper Series 3258.

Deininger, Klaus, J. Hoogeveen, and B. Kinsey. 2004. “EconomicBenefits and Costs of Land Redistribution in Zimbabwe in theearly 1980s.” World Development 32(10):1697–709.

Deininger, Klaus, Songquing Jin, Berhanu Adenew, Samuel Gebre-Selassie, and Berhanu Nega. 2003. “Tenure Security and Land-Related Investment: Evidence from Ethiopia.” Washington, DC:World Bank Policy Research Working Paper Series 2991.

Deininger, Klaus, and Paul Mpuga. 2004. “Economic and WelfareEffects of the Abolition of Health User Fees: Evidence fromUganda.” Washington, DC: World Bank Policy Research WorkingPaper Series 3276.

Deininger, Klaus, and Pedro Olinto. 2000. “Asset Distribution,Inequality and Growth.” World Bank: World Bank PolicyResearch Working Paper Series 2375.

Deininger, Klaus, and Lyn Squire. 1996. “A New Data Set MeasuringIncome Inequality.” World Bank Economic Review 10(3):565–91.


————. 1998. “New Ways of Looking at Old Issues: Inequalityand Growth.” Journal of Development Economics 57(2):259–87.

Delgado, Guilherme, and José Celso Cardoso Jr., eds. 2000. A Uni-versalização de Direitos Sociais no Brasil: A Previdência Rural nosAnos 90. Brasília: IPEA.

Demombynes, Gabriel, and Berk Özler. 2005. “Crime and LocalInequality in South Africa.” Journal of Development Economics76(2):265–92.

Deshpande, Ashwini. 2005. “Affirmative Action in India and theUnited States.” Background paper for the WDR 2006.

DeTray, Dennis. 1988. “Government Policy, Household Behaviorand the Distribution of Schooling: A Case Study in Malaysia.” InT. P. Schultz, (ed.), Research in Population Economics, Vol. 6.Greenwich, CT: JAI Press.

Deutsch, Ruthanne. 1998. “Does Child Care Pay? Labor ForceParticipation and Earnings Effects of Access to Child Care inthe Favelas of Rio.” Washington, DC: Inter-American Develop-ment Bank, Office of the Chief Economist Working PaperSeries 384.

Devereux, Stephen. 2001. “Livelihood Insecurity and Social Protec-tion: A Re-emerging Issue in Rural Development.” DevelopmentPolicy Review 19(4):507–19.

Diallo, Amadou Bassirou, and Quentin Wodon. 2005. “A Note onAccess to Network-based Infrastructure Services in Africa: Bene-fit and Marginal Benefit Incidence Analysis.” World Bank. Wash-ington, DC. Processed.

Diamond, Peter A. 1981. “Mobility Costs, Frictional Unemploy-ment, and Efficiency.” Journal of Political Economy 89(4):798–812.

Dickson, Rumona, Shally Awasthi, Paula Williamson, ColinDemellweek, and Paul Garner. 2000. “Effects of Treatment forIntestinal Helminth Infection on Growth and Cognitive Perfor-mance in Children: Systematic Review of Randomized Trials.”British Medical Journal 320(7251):1697–701.

Diwan, Ishac. 2001. “Debt as Sweat: Labor, Financial Crisis, and theGlobalization of Capital.” World Bank. Washington, DC.Available on line at http://info.worldbank.org/etools/docs/voddocs/150/332/diwan.pdf. Processed.

————. 2002. “The Labor Share during Financial Crisis: NewResults.” World Bank. Washington, DC. Processed.

Djankov, Simeon, Rafael La Porta, Florencio López-de-Silanes, andAndrei Shleifer. 2003. “Courts.” Quarterly Journal of Economics118(2):453–517.

Do, Toan, and Laksmi Iyer. 2003. “Land Rights and EconomicDevelopment: Evidence from Vietnam.” Washington, DC: WorldBank Policy Research Working Paper Series 3120.

Dollar, David, and Aart Kraay. 2002. “Growth is Good for the Poor.”Journal of Economic Growth 7(3):195–225.

————. 2004. “Trade, Growth and Poverty.” Economic Journal114(493):F22–F49.

Dollar, David, and Victoria Levin. 2004. “The Increasing Selectivityof Foreign Aid.” Washington DC: World Bank Policy ResearchWorking Paper Series 3299.

Doryan, Eduardo A., Kul C. Gautman, and William H. Foege. 2002.“The Political Challenge: Commitment and Cooperation.” InMary Eming Young, (ed.), From Early Childhood Development to

Human Development: Investing in our Children’s Future. Wash-ington, DC: World Bank.

Draï, Raphaël. 1991. Le Mythe de la Loi du Talion. Paris: Alinea.

Drèze, Jean, Peter Lanjouw, and Naresh Sharma. 1998. “EconomicDevelopment in Palanpur, 1957–93.” In Peter Lanjouw andNicholas Stern, (ed.), Economic Development in Palanpur overFive Decades. New York: Clarendon Press.

Duan, Chengrong. 2005. “China: Gender Inequality Status Report.”Renmin University. Beijing. Processed.

Duflo, Esther. 2001. “Schooling and Labor Market Consequences ofSchool Construction in Indonesia: Evidence from an UnusualPolicy Experiment.” American Economic Review 91(4):795–813.

————. 2003. “Grandmothers and Granddaughters: Old-agePensions and Intrahousehold Allocation in South Africa.” WorldBank Economic Review 17(1):1–25.

Duflo, Esther, Michael Kremer, and Jonathan Robinson. 2004.“Understanding Technology Adoption: Fertilizer in WesternKenya, Preliminary Results from Field Experiments.” Paper pre-sented at the LSE Conference ‘Behavioral Economics, PublicEconomics and Development Economics. May 28. London.

Duflo, Esther, and Christopher Udry. 2004. “IntrahouseholdResource Allocation in Côte d’Ivoire: Social Norms, SeparateAccounts and Consumption Choices.” Cambridge, MA:National Bureau of Economic Research Working Paper Series10498.

Dunn, Christopher. 2003. “Assortative Matching and Intergenera-tional Mobility in Family Earnings: Evidence from Brazil.” Uni-versity of Michigan. Ann Harbor, MI. Processed.

Durand-Lasserve, A. 2003. “Land Issues and Security of Tenure.Background Report for the UN Millennium Project Task Forceon Improving the Lives of Slum Dwellers.” National Centre forScientific Research. France. Processed.

Dworkin, Ronald. 1981a. “What is Equality? Part 2: Equality ofResources.” Philosophy and Public Affairs 10(3):283–345.

————. 1981b. “What is Equality? Part 1: Equality of Welfare.”Philosophy and Public Affairs 10(3):185–246.

Easterlin, Richard. 1974. “Does Economic Growth Improve theHuman Lot?: Some Empirical Evidence.” In David A. Paul andMelvin W. Reder, (ed.), Nations and Households in EconomicGrowth: Essays in Honor of Moses Abramowitz. New York: Acade-mic Press.

Easterly, William, and Ross Levine. 1997. “Africa’s Growth Tragedy:Policies and Ethnic Divisions.” Quarterly Journal of Economics112(4):1203–50.

Easterly, William, Ross Levine, and David Roodman. 2004. “Aid,Policies, and Growth: Comment.” American Economic Review94(3):774–80.

Ehrenreich, Barbara, and Arlie Russell Hochschild, eds. 2003.Global Woman. Nannies, Maids, and Sex Workers in the NewEconomy. New York: Metropolitan Books.

Elbers, Chris, Peter Lanjouw, Joan Mistiaen, Berk Özler, and K.Simler. 2004.“On the Unequal Inequality of Poor Communities.”World Bank Economic Review 18(3):401–21.

Elbers, Chris, Peter Lanjouw, Johan Mistiaen, and Berk Özler. 2005.“Re-interpreting Sub-group Inequality Decompositions.” World Bank.

References 255

Washington, DC. Available on line at http://globetrotter.berkeley.edu/macarthur/inequality/papers/OzlerReinterpretingDecomp.pdf.Processed.

Elo, Irma, and Samuel H. Preston. 1996. “Educational Differences inMortality: United States.” Social Science and Medicine 42(1):47–57.

Elson, Robert E. 2001. Suharto: A Political Biography. Cambridge,UK: Cambridge University Press.

Engerman, Stanley, and Kenneth Sokoloff. 1997. “Factor Endow-ments, Institutions, and Differential Paths of Growth AmongNew World Economies: A View from Economic Historians of theUnited States.” In Stephen Haber, (ed.), How Latin America FellBehind. Stanford, CA: Stanford University Press.

Engerman, Stanley L., and Kenneth Sokoloff. 2002. “Factor Endow-ments, Inequality, and Paths of Development among New WorldEconomies.” Economia 3(1):41–88.

Engerman, Stanley L., and Kenneth L. Sokoloff. 2001. “The Evolu-tion of Suffrage Institutions in the New World.” Cambridge, MA:National Bureau of Economic Research Working Paper Series8512.

Erikson, Robert, and John Goldthrope. 2002. “IntergenerationalInequality: A Sociological Perspective.” Journal of Economic Per-spectives 16(3):31–44.

Escobal, Javier, and Máximo Torero. 2003. “Adverse Geography andDifferences in Welfare in Peru.” Helsinki: United Nations Univer-sity, World Institute for Development Economics Research, Dis-cussion Paper 2003/73.

Escobar, María-Luisa. 2005. “The Columbia Health Sector Reformand the Poor.” World Bank. Washington, DC. Processed.

Escobar, María-Luisa, and Panagiota Panopoulou. 2002. “Chapter6: Health.” In Giugale, Lafourcade, and Luff, (ed.), Columbia: TheEconomic Foundations for Peace. Washington, DC: World Bank.

Esping-Andersen, Gosta. 1990. The Three Worlds of Welfare Capital-ism. Cambridge, UK: Polity Press.

Estache, Antonio. 2003. “On Latin America’s Infrastructure Privati-zation and its Distributional Effects.” Paper presented at the Dis-tributional Consequences of Privatization Conference. February24. Washington, DC.

Estache, Antonio, Vivien Foster, and Quentin Wodon. 2001.Accounting for Poverty in Infrastructure Reform: Learning fromLatin America’s Experience. Washington, DC: World Bank.

Estache, Antonio, Andrés Gómez-Lobo, and Danny Leipziger. 2001.“Utilities Privatization and the Poor: Lessons and Evidence fromLatin America.” World Development 29(7):1179–98.

Esteban, Joan, and Debraj Ray. 1994. “On the Measurement ofPolarization.” Econometrica 62(4):819–51.

Evans, Peter. 2004. “Development as Institutional Change: The Pit-falls of Monocropping and Potentials of Deliberation.” Studies inComparative International Development 38(4):30–53.

Fafchamps, Marcel. 2000. “Ethnicity and Credit in African Manu-facturing.” Journal of Development Economics 61(1):205–35.

Fafchamps, Marcel, and Susan Lund. 2003. “Risk-sharing Networksin Rural Philippines.” Review of Economic Studies 71(2):261–87.

Faini, Riccardo. 2003.“The Brain Drain: An Unmitigated Blessing?”Milan: Centro Studi Luca d’Agliano Development Studies 173.Available on line at http://ssrn.com/abstract=463021.

Fajnzylber, Pablo, Daniel Lederman, and Norman Loayza. 2000.“Crime and Victimization: An Economic Perspective.” Economía1(1):219–78.

Fallon, Peter R., and Robert E. B. Lucas. 1993. “Job Security Reg-ulations and the Dynamic Demand for Industrial Labor in India and Zimbabwe.” Journal of Development Economics40(2):241–75.

Feder, Gershon. 1988. Land Policies and Farm Productivity in Thai-land. Baltimore, MD: Johns Hopkins University.

————. 2002. “The Intricacies of Land Markets: Why the WorldBank Succeeds in Economic Reforms through Land Registrationand Tenure Security.” Paper presented at the International Feder-ation of Surveyors Conference. April 19. Washington, DC.

Fehr, Ernst, and Urs Fischbacher. 2003. “The Nature of HumanAltruism.” Nature 425(October):785–91.

Fehr, Ernst, and Simon Gachter. 2000. “Cooperation and Punish-ment in Public Goods Experiments.” American Economic Review90(4):980–94.

Fehr, Ernst, and Klaus M. Schmidt. 1999. “A Theory of Fairness,Competition and Cooperation.” Quarterly Journal of Economics114(3):817–68.

Feierstein, Mark, and John Moreira. 2005. Liberians Have NewOutlook On Their Future. Washington, DC: Greenberg Quin-land Rosner Research Inc. Available on line at http://www.greenbergresearch.com.

Feijen, Erik, and Enrico Perotti. 2005. “Lobbying for StrategicDefault.” University of Amsterdam. Amsterdam. Processed.

Feldstein, Martin. 1998. “Income Inequality and Poverty.” Cam-bridge, MA: National Bureau of Economic Research WorkingPaper Series 6770.

Fernández-Kelly, Patricia. 1995. “Social and Cultural Capital in theUrban Ghetto: Implications for the Economic Sociology ofImmigration.” In Alejandro Portes, (ed.), The Economic Sociologyof Immigration: Essays in Network, Ethnicity, and Entrepreneur-ship. New York: Rusell Sage Foundation.

Ferreira, Francisco H. G. 2001. “Education for the Masses? TheInteraction between Wealth, Educational and Political Inequali-ties.” Economics of Transition 9(2):533–52.

Ferreira, Francisco H. G., and Julie A. Litchfield. 2001. “Educationor Inflation? The Micro and Macroeconomics of the BrazilianIncome Distribution during 1981–95.” Cuadernos de Economía38(114):209–38.

Ferreira, Pedro Cavalcanti. 2004. Regional Policy in Brazil: A Review.Rio de Janeiro: Fundacao Getulio Vargas.

Field, Erica. 2003. “Entitled to Work: Urban Tenure Security andLabor Supply in Peru.” Princeton, NJ: Princeton University,Princeton Law & Public Affairs Working Paper 02-1.

Fields, Gary S. 1989. “Changes in Poverty and Inequality in Devel-oping Countries.” World Bank Research Observer 4(2):167–85.

Fields, Gary S., and George H. Jakubson. 1994. “New Evidence onthe Kuznets Curve.” Cornell University. Ithaca, NY. Processed.

Filmer, Deon. 2004. “If you Build it, Will They Come? SchoolAvailability and School Enrollment in 21 Poor Countries.”Washington, DC: World Bank Policy Research Working PaperSeries 3340.


Filmer, Deon, Margaret Grosh, Elizabeth M. King, and DominiqueVan de Walle. 1998. “Pay and Grade Differentials at the WorldBank.” Washington, DC: World Bank Policy Research WorkingPaper Series 1912.

Filmer, Deon, and Lant Pritchett. 1998. The Effect of HouseholdWealth on Educational Attainment. Policy Research WorkingPaper #1980, The World Bank.

Filmer, Deon, and Lant Pritchett. 1999. “The Effect of HouseholdWealth on Educational Attainment: Evidence from 35 Coun-tries.” Population and Development Review 25(1):85–120.

Fink, Carsten, and Patrick Reichenmuller. 2005. “Tightening TRIPs:The Intellectual Property Provisions of Recent US Freed TradeAgreements.” Washington, DC: World Bank, Trade Note 20.

Firebaugh, Glenn, and Brian Goesling. 2004. “Accounting for theRecent Decline in Global Income Inequality.” The American Jour-nal of Sociology 110(2):283–312.

Fishman, Robert M. 1990. Working Class Organization and the Returnto Democracy in Spain. Ithaca, NY: Cornell University Press.

Fisman, Raymond. 2001a. “Trade Credit and Productive Efficiencyin Developing Countries.” World Development 29(2):7311–21.

————. 2001b. “Estimating the Value of Political Connections.”American Economic Review 91(4):1095–102.

————. 2003.“Ethnic Ties and the Provision of Credit: Relation-ship-Level Evidence from African Firms.” Advances in EconomicAnalysis and Policy 3(1):1211–1211.

Fiszbein, Ariel. 2005. Citizens, Politicians and Providers: The LatinAmerican Experience with Service Delivery Reform. Washington,DC: World Bank.

Fleshman, Michael. 2001. “AIDS Orphans: Facing Africa’s SilentCrisis.” Africa Recovery 15(3):1–1.

Forbes, Kristin J. 2000. “A Reassessment of the RelationshipBetween Inequality and Growth.” American Economic Review90(4):869–87.

Fortman, Louise. 1998. “Why Women’s Property Rights Matter.”Paper presented at the International Conference and Workshopon Land Tenure in the Developing World. Cape Town.

Foster, Andrew D., and Mark R. Rosenzweig. 1995.“Learning by Doingand Learning from Others: Human Capital and Technical Change inAgriculture.” Journal of Political Economy 103(6):1176–209.

Fox, Jonathan, eds. 1990. The Challenge of Rural Democratization:Perspectives from Latin America and the Philippines. London:Frank Cass and Company.

FTI Secretariat. 2004. Education for All (EFA)-Fast Track Initiative(FTI); Status Report. Washington, DC: World Bank. Available online at http://www1.worldbank.org/education/efafti/documents/Brasilia/status_report_dec6.pdf.

Fujii, Tomoki. 2005. “Micro-level Estimation of Child MalnutritionIndicators and its Application in Cambodia.” Washington, DC:World Bank Policy Research Working Paper Series 3662.

Funck, Bernard, and Lodovico Pizzati, eds. 2003. European Inte-gration, Regional Policy, and Growth. Washington, DC: WorldBank.

Fung, Archon, Dara O’Rourke, and Charles Sabel. 2001. Can we Putan End to Sweatshops? A New Democracy Forum on Raising GlobalLabor Standards. Boston, MA: Beacon Press.

Galasso, Emanuela, and Martin Ravallion. 2005. “DecentralizedTargeting of an Anti-poverty Program.” Journal of Public Eco-nomics 89(4):705–27.

Galiani, Sebastián, Paul Gertler, and Ernesto Schargrodsky. 2002.“Water for Life: The Impact of the Privatization of Water Ser-vices on Child Mortality.” Stanford, CA: Stanford University,Center for Research on Economic Development and PolicyReform Working Paper 154.

Galiani, Sebastián, and Ernesto Schargrodsky. 2002. “Evaluating theImpact of School Decentralization on Ecucational Quality.”Economía 2(2):275–314.

————. 2004. “Effects of Land Titling.” Paper presented at theWorld Bank Poverty and Applied Micro Seminar Series. October6. Washington, DC.

Galli, Rossana, and David Kucera. 2004. “Labor Standards andInformal Employment in Latin America.” World Development32(5):809–828.

Garapon, Antoine, eds. 2003. Les Juges Un Pouvoir Irresponsable?Paris: Éditions Nicolas Philippe.

García-Montalvo, José, and Marta Reynal-Querol. Forthcoming.“Why Ethnic Fractionalization? Polarization, Conflict andGrowth.” American Economic Review.

Geertz, Clifford. 1963. Peddlers and Princes; Social Change and Eco-nomic Modernization in Two Indonesian Towns. Chicago: Univer-sity of Chicago Press.

Gerschenkron, Alexander. 1962. Economic Backwardness in Histori-cal Perspective: A Books of Essays. Cambridge: Belknap Press ofHarvard University Press.

Gertler, Paul, and Sarah Barber. 2004. “The Returns to Investing inQuality Healthcare.” Institute of Business and EconomicResearch, University of California. Berkeley, CA. Processed.

Gertler, Paul, and Jonathan Gruber. 2002. “Insuring Consumptionagainst Illness.” American Economic Review 92(1):51–76.

Gibson, Christopher, and Michael Woolcock. 2005. “Empowermentand Local Level Conflict Mediation in Indonesia: A ComparativeAnalysis of Concepts, Measures, and Project Efficacy.” WorldBank. Washington, DC. Processed.

Gill,A., and U. C. Singh. 1977.“Financial Sector Reforms, Rate of Inter-est and the Rural Credit Markets: The Role of Informal Lenders inPunjab.” Indian Journal of Applied Economics 6(4):37–65.

Glaeser, Edward L., Simon Johnson, and Andrei Shleifer. 2001. “Coaseversus the Coasians.” Quarterly Journal of Economics 116(3):853–99.

Glaeser, Edward L., José Sheinkman, and Andrei Shleifer. 2003.“The Injustice of Inequality.” Journal of Monetary Economics50(1):199–222.

Glewwe, Paul, Nauman Ilias, and Michael Kremer. 2003. “TeacherIncentives.” Cambridge, MA: National Bureau of EconomicResearch Working Paper Series 9671.

Glewwe, Paul, Michael Kremer, and Sylvie Moulin. 2002. “Text-books and Test Scores: Evidence from a Prospective Evaluation inKenya.” Harvard University. Cambridge, MA. Available on line athttp://post.economics.harvard.edu/faculty/kremer/webpapers/Textbooks_Test_Scores.pdf. Processed.

Glewwe, Paul, Michael Kremer, Sylvie Moulin, and Eric Zitzewitz.2004. “Retrospective vs. Prospective Analyses of School Inputs:

References 257

The Case of Flip Charts in Kenya.” Journal of Development Eco-nomics 74(1):251–68.

Goesling, Brian, and Glenn Firebaugh. 2004.“The Trend in InternationalHealth Inequality.”Population and Development Review 30(1):131–46.

Goetz, Stephan J. 1992. “A Selectivity Model of Household FoodMarketing Behavior in Sub-Saharan Africa.” American Journal ofAgricultural Economics 74(2):444–52.

Goldin, Ian, Kenneth A. Reinert, and Andrew Beath. Forthcoming.“Migration.” In Ian Goldin and Kenneth A. Reinert (ed.) Global-ization and Poverty. Washington, DC: World Bank.

Goldstein, Markus, and Christopher Udry. 1999. “AgriculturalInnovation and Resource Management in Ghana.” Yale Univer-sity. New Haven, CT. Processed.

————. 2002. “Gender, Land Rights and Agriculture in Ghana.”Yale University. New Haven, CT. Processed.

Gottfredson, Denise. 1998. “School-based Crime Prevention.” InLawrence W. Sherman, Denise Gottfredson, Doris MacKenzie,John Eck, Peter Reuter, and Shawn Bushway, (ed.), PreventingCrime. What Works. What Doesn’t. What’s Promising. Washing-ton, DC: U.S. Department of Justice.

Graham, Carol, and Andrew Felton. 2005. “Does Inequality Matterto Individual Welfare?: An Initial Exploration based on Happi-ness Surveys from Latin America.” Washington, DC: BrookingInstitution, CSED Working Paper 138. Available on line athttp://www.brookings.edu/es/dynamics/papers/csed_wp38.pdf.

Graham, Jean, and Benjamin Bowling. 1995. “Young People andCrime.” London: Home Office Research Study 145.

Grantham-McGregor, S, and C. Ani. 2001. “A Review of Studies onthe Effect of Iron Deficiency on Cognitive Development in Chil-dren.” Journal of Nutrition 131(2S-2):649S–668S.

Grantham-McGregor, S., C. Powell, S. P. Walker, and J. H. Himes.1991. “Nutritional Suplementation, Pshychosocial Stimulation,and Mental Development of Stunted Children: The JamaicanStudy.” The Lancet 338(8758):1–5.

Gravois, John. 1-28-2005. “The De Soto Delusion.” Slate.

Greenwood, Peter W., Karyn T. Model, C. Peter Rydell, and JamesChiesa. 1998. Diverting Children from a Life of Crime: MeasuringCost and Benefits. Santa Monica, CA: Rand Corporation.

Grindle, Merilee S. 2004. Despite The Odds: Contentious Politics ofEducation Reform. Princeton, NJ: Princeton University Press.

Guasch, J. Luis. 2003. Concessions of Infrastructure Services: Inci-dence and Determinants of Renegotiation: An Empirical Evalua-tion and Guidelines for Optimal Concession Design. Washington,DC: World Bank Institute.

Guggenheim, Scott. Forthcoming. “Crises and Contradictions:Understanding the Origins of a Community Development Pro-ject in Indonesia.” In Anthony Bebbington, Scott Guggenheim,Elizabeth Olson, and Michael Woolcock (ed.) The Search forEmpowerment: Social Capital as Idea and Practice at the WorldBank. Bloomfield, CT: Kumarian Press.

Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2004. “Does LocalFinancial Development Matter?” Quarterly Journal of Economics119(3):929–69.

Gulyani, Sumila, Debabrata Talukdar, and R. Mukami Kariuki.2005. “Water for the Urban Poor: Water Markets, Household

Demand, and Service Preferences in Kenya.” Washington, DC:World Bank Water Supply and Sanitation Sector Board Discus-sion Paper Series 5.

Gunning, Jan Willem, John Hoddinott, Bill Kinsey, and TrudyOwens. 2000. “Revisiting Forever Gained: Income Dynamics inthe Resettlement Areas of Zimbabwe, 1983–96.” Journal of Devel-opment Studies 36(6):131–54.

Gunther, Richard, José Ramón Montero, and Joan Botella. 2004.Democracy in Modern Spain. New Haven, CT: Yale University Press.

Gupta, Sanjeev, Benedict Clements, Alexander Pivovarsky, andErwin R. Tiongson. 2003. “Foreign Aid and Revenue Response:Does the Composition of Aid Matter?” Washington, DC: Inter-national Monetary Fund Working Paper Series WP/03/176.

Gurr, Ted Robert. 1997. “Polity II: Political Structures and Regime Change, 1800–1986.” Ann Arbor, MI: ICSPR Study No.9263. Available on line at http://webapp.icpsr.umich.edu/cocoon/ICPSR-STUDY/09263.xml.

Gwatkin, Davidson R. 2002. “The Poor Come Last: Socio-Economic Inequalities in Use of Maternal and Child Health Ser-vices in Developing Countries.” Paper presented at the Presenta-tion at the Meeting of the Fogarty International Center. July 24.National Institutes of Health, Bethesda, MD.

Gwatkin, Davidson R., Shea Rutstein, Kiersten Johnson, EldawAbdalla Suliman, and Adam Wagstaff. 2004. Socio-economic dif-ferences in health, nutrition, and population, volumes I–III. Wash-ington, DC: World Bank.

Haber, Stephen. 2004. “Political Institutions and Economic Devel-opment: Evidence from the Banking Systems of the United Statesand Mexico.” Paper presented at the Economics, Political Institu-tions, and Financial Markets II: Institutional Theory and Evi-dence from Europe, the United States, and Latin America Con-ference. February 5. Palo Alto, CA.

Haber, Stephen, and Shawn Kantor. 2004. “Getting PrivatizationWrong: The Mexican Banking System, 1991–2003.” Paper pre-sented at the World Bank Conference on Bank Privatizacion inLow and Middle-Income Countries. November 23. Washington,DC.

Haber, Stephen, and Noel Maurer. 2004. “Related Lending and Eco-nomic Performance: Evidence from Mexico.” Paper presented atthe Sixty Fifth Annual Meeting of the American Finance Associ-ation. January 7. Philadelphia.

Haber, Stephen, Maurer Noel, and Armando Razo. 2003. The Poli-tics of Property Rights: Political Instability, Credible Commitments,and Economic Growth in Mexico: 1876–1929. Cambridge, UK:Cambridge University Press.

Haber, Stephen H. 2001. “Political Institutions and Banking Sys-tems: Lessons from the Economic Histories of Mexico and theUnited States, 1790–1914.” Department of Political Science,Stanford University. Stanford, CA. Processed.

Haggard, Stephan. 1990. Pathways from the Periphery: The Politics ofGrowth in the Newly Industrializing Countries. Ithaca, NY: Cor-nell University Press.

Haggard, Stephan, Wonhyuk Lim, and Euysung Kim, eds. 2003.Economic Crisis and Corporate Restructuring in Korea: Reformingthe Chaebol. New York: Cambridge University Press.


Halac, Marina, and Sergio L. Schmukler. 2003. “Distribution Effectsof Crises: The Role of Financial Transfers.” Washington, DC:World Bank Policy Research Working Paper Series 3173.

Hall, Gillette, and Harry Anthony Patrinos, eds. 2005. IndigenousPeoples, Poverty and Human Development in Latin America:1994–2004. New York: Palgrave MacMillan.

Hall, Robert E., and Charles I. Jones. 1999. “Why Do Some Coun-tries Produce so Much More Output per Worker than Others?”Quarterly Journal of Economics 114(1):83–116.

Hamid, N. 1983. “Growth of Small-scale Industry in Pakistan.” Pak-istan Economic and Social Review 21(1-2):37–76.

Hamilton, Bob, and John Whalley. 1984. “Efficiency and Distribu-tional Implications of Global Restrictions on Labor Mobility.”Journal of Development Economics 14(1-2):61–75.

Hammergreen, Linn. 2004. “Use and Users Study of the Justice Sys-tem in Peru.” World Bank. Washington, DC. Processed.

Hansen, Henrik, and Finn Tarp. 2001. “Aid and Growth Regres-sions.” Journal of Development Economics 64(2):547–70.

Hanson, Gordon H. 2003. “What Has Happened to Wages in Mex-ico since NAFTA? Implications for Hemispheric Free Trade.”Cambridge, MA: National Bureau of Economic Research Work-ing Paper Series 9563.

Hanushek, Eric A. 1986. “The Economics of Schooling: Productionand Efficiency in Public Schools.” Journal of Economic Literature24(3):1141–77.

————. 1996. “School Resources and Student Performance.” InGary Burtless, (ed.), Does Money Matter? The Effects of SchoolResources on Student Achievement and Adult Success. Washington,DC: Brookings Institution.

Hanushek, Eric A., and Ludger Wößman. 2005. “Does EducationalTracking Affect Performance and Inequality? Differences-in-Differences Evidence Across Countries.” Cambridge, MA: NationalBureau of Economic Research Working Paper Series 11124.

Harber, Clive. 1999. Protecting Your School from Violence and Crime:An Evaluation of One-Year Programme. KwaZulu-Natal, SouthAfrica: Independent Project Trust.

Hargreaves, James R., and Judith R. Glynn. 2002. “EducationalAttainment and HIV Infection in Developing Countries: A Sys-tematic Review.” Tropical Medicine and International Health7(6):489–98.

Harsanyi, John C. 1955. “Cardinal Welfare, Individualistic Ethics,and Interpersonal Comparisons of Utility.” Journal of PoliticalEconomy 63(4):309–21.

Hart, Betty, and Todd R. Risley. 1995. Meaningful Differences inEveryday Experiences of Young American Children. Baltimore,MD: Paul H. Brookes Publishing Co.

Hasan, Rana, Devashish Mitra, and K. V. Ramaswamy. 2003. “TradeReforms, Labor Regulations and Labor-demand Elasticities:Empirical Evidence from India.” Cambridge, MA: NationalBureau of Economic Research Working Paper Series 9879.

Hashemi, Syed. 2000. “Linking Microfinance and Safety Net Pro-grams to Include the Poorest: The Case of IGVGD inBangladesh.” Washington, DC: World Bank, Focus Notes 21.

Haslam, Alexander S. 2001. Psychology in Organizations: The Social-identity Approach. Thousand Oaks, CA: Sage Publications Ltd.

Heckman, James, and Dimitri V. Masterov. 2004. “The ProductivityArgument for Investing in Young Children.” Washington, DC:Committee for Economic Development Working Paper 5.

Heise, L., M. Ellsberg, and M. Gottemoeller. 1999. “Ending ViolenceAgainst Women.” Baltimore, MD: Johns Hopkins UniversitySchool of Public Health, Population Information Program Pop-ulation Report Series L, number 11.

Hellman, Joel S., Geraint Jones, and Daniel Kaufmann. 2003. “Seizethe State, Seize the Day: State Capture and Influence in TransitionEconomies.” Journal of Comparative Economics 31(4):751–73.

Hemming, John. 1970. The Conquest of the Incas. London: Paper-mac.

Henrich, Joseph, Robert Boyd, Samuel Bowles, Colin Camerer,Ernst Fehr, and Herbert Gintis, eds. 2004. Foundations ofHuman Society: Economic Experiments and Ethnographic Evi-dence from Fifteen Small-Scale Societies. New York: OxfordUniversity Press.

Hertel, Thomas W., and L. Alan Winters. Forthcoming. “PuttingDevelopment Back into the Doha Agenda: Poverty Impacts of aWTO Agreement.” World Bank. Washington, DC. Processed.

Hertz, Thomas. 2001. “Education, Inequality and Economic Mobil-ity in South Africa.” PhD thesis. University of Massachussetts.

————. 2005. “Rags, Riches and Race: The IntergenerationalEconomic Mobility of Black and White Families in the U.S.” InSam Bowles, Herb Gintis, and Melissa Osborne Grove, (ed.),Unequal Chances. Family Background and Economic Success.Princeton, NJ: Princeton University Press.

Herz, Barbara, and Gene B. Sperling. 2004. What Works in Girls’Education. New York: Council on Foreign Relations.

Hoff, Karla. 2003. “Paths of Institutional Development: A View fromEconomic History.” World Bank Research Observer 18(2):2205–26.

Hoff, Karla, and Priyanka Pandey. 2004. “Belief Systems andDurable Inequalities: An Experimental Investigation of IndianCaste.” Washington, DC: World Bank Policy Research WorkingPaper Series 3351.

Hoffman, Elizabeth, Kevin A. McCabe, and Vernon L. Smith. 1996.“On Expectations and Monetary Stakes in Ultimatum Games.”International Journal of Game Theory 25(3):289–301.

Holmes, George M., Jeff DeSimone, and Nicholas G. Rupp. 2003.“Does School Choice Increase School Quality?” Cambridge, MA:National Bureau of Economic Research Working Paper Series9683.

Holzer, Harry, and David Neumark. 2000. “Assessing AffirmativeAction.” Journal of Economic Literature 38(3):483–568.

Holzmann, Robert, and Richard Hinz. 2005. Old-Age Income Sup-port in the Twenty-First Century: An International Perspective onPension Systems and Reform. Washington, DC: World Bank. Avail-able on line at http://www1.worldbank.org/sp/incomesupport.asp.

Holzmann, Robert, and Steen Jorgensen. 2001.“Social Risk Manage-ment: A New Conceptual Framework for Social Protection andBeyond.” Washington, DC: World Bank, Social Protection Discus-sion Paper 0006.

Honohan, Patrick. 2004. Financial Sector Policy and the Poor:Selected Findings and Issues. Washington, DC: World Bank.

References 259

Honohan, Patrick, and Daniela Klingebiel. 2000. “Controlling theFiscal Costs of Banking Crises.” Washington, DC: World BankPolicy Research Working Paper Series 2441.

Hoogeveen, Hans. 2003.“Census-Based Welfare Estimates for SmallPopulations: Poverty and Disability in Uganda.” World Bank.Washington, DC. Processed.

Hoogeveen, Johannes, Emil Tesliuc, Renos Vakis, and Stefan Der-con. 2004. “A Guide to the Analysis of Risk, Vulnerability andVulnerable Groups.” World Bank. Washington, DC. Available online at http://siteresources.worldbank.org/INTSRM/Publications/20316319/RVA.pdf. Processed.

Hoxby, Caroline M. 2000. “Does Competition Among PublicSchools Benefit Students and Taxpayers?” American EconomicReview 90(5):1209–38.

————. 2002. “School Choice and School Productivity (orCould School Choice be a Tide that Lifts All Boats?).” Cam-bridge, MA: National Bureau of Economic Research WorkingPaper Series 8873.

Hsieh, Chang-Tai, and Miguel Urquiola. 2003. “When SchoolsCompete, How Do They Compete? An Assessment ofChile’s Nationwide School Voucher Program.” Cambridge, MA:National Bureau of Economic Research Working Paper Series10008.

Human Rights Watch. 2000. United States Punishment and Preju-dice: The Racial Disparities in the War on Drugs. New York:Human Rights Watch. Available on line at http://www.hrw.org/reports/2000/usa/.

Ilahi, Nadeem, and Franque Grimard. 2000. “Public Infrastructureand Private Costs: Water Supply and Time Allocation of Womenin Rural Pakistan.” Economic Development and Cultural Change49(1):45–75.

Ilahi, Nadeem, and Saqib Jafarey. 1999. “Guestworker Migration,Remittances, and the Extended Family: Evidence from Pakistan.”Journal of Development Economics 58(2):485–512.

ILO. 2002. Women and Men in the Informal Economy: A StatisticalPicture. Geneva: International Labour Organization. Availableon line at http://www.ilo.org/public/english/employment/gems/download/women.pdf.

————. 2003. Investing in Every Child—An Economic Study of theCosts and Benefits of Eliminating Child Labor. Geneva: Interna-tional Labour Organization. Available on line at http://www.ilo.org/public/english/standards/ipec/publ/download/2003_12_investingchild.pdf.

India Today. 12-13-2004. “Rural Markets: Call of the Countryside.”India Today.

Inglehart, Ronald, Miguel Basáñez, Jaime Díez-Medrano, Loek Hal-man, and Ruud Luijkx, eds. 2004. Human Beliefs and Values: ACross-cultural Sourcebook based on the 1999–2002 Values Surveys.Mexico, DF: Siglo Veinteiuno.

International Confederation of Free Trade Unions. 2005. TheImpact of the Ending of Quotas on the Textile Industry. Cambodia:Increasing Pressure on Trade Union Rights. Brussels: InternationalConfederation of Free Trade Unions. Available on line athttp://www.icftu.org/displaydocument.asp?Index=991221551&Language=EN.

International Cotton Advisory Committee. 2003. Production andTrade Policies Affecting the Cotton Industry. A Report by the Secre-tariat. Washington, DC: International Cotton Advisory Com-mittee. Available on line at http://www.icac.org/cotton_info/ publications/statistics/stats_wtd/prod_trade_policies_03.pdf.

International Fund for Agricultural Development. 2001. RuralPoverty Report 2001: The Challenge of Ending Rural Poverty.Oxford: Oxford University Press. Available on line athttp://www.ifad.org.

Irwin, Timothy, and Chiaki Yamamoto. 2004. “Improving the Gover-nance of State-owned Electricity Utilities.” Washington, DC: WorldBank Energy and Mining Sector Board Discussion Paper 11.

Jacoby, Hanan G., Guo Li, and Scott Rozelle. 2002. “Hazards ofExpropriation: Tenure Insecurity and Investment in RuralChina.” American Economic Review 92(5):1420–47.

Jalan, Jyostna, and Martin Ravallion. 1997. “Spatial Poverty Traps?”Washington, DC: World Bank Policy Research Working PaperSeries 1862.

————. 2002. “Geographic Poverty Traps? A Micro Model ofConsumption Growth in Rural China.” Journal of Applied Econo-metrics 17(4):329–46.

————. 2003. “Does Piped Water Reduce Diarrhea for Childrenin Rural India?” Journal of Econometrics 112(1):153–73.

Jalan, Jyotsna, and Martin Ravallion. 1999. “Income Gains to thePoor from Workfare: Estimates for Argentina’s Trabajar Pro-gram.” Washington, DC: World Bank Policy Research WorkingPaper Series 2149.

James, Estelle. 2000. “Old-age Protection for the Uninsured: Whatare the Issues?” In Nora Lustig, (ed.), Shielding the Poor: SocialProtection in the Developing World. Washington, DC: BrookingsInstitutiond and Inter-american Development Bank.

Jäntti, Markus, Juho Saari, and Juhana Vartiainen. 2005. “CountryCase Study: Finland-Combining Growth with Equity.” Back-ground paper for the WDR 2006.

Jha, Saumitra, Vijayendra Rao, and Michael Woolcock. Forth-coming. “Governance in the Gullies: Democratic Responsive-ness and Community Leadership in Delhi’s Slums.” WorldDevelopment.

Johnson, Simon, and Todd Mitton. 2003. “Cronyism and CapitalControls: Evidence from Malaysia.” Journal of Financial Econom-ics 67(2):351–82.

Jonakin, Jon. 1996. “The Impact of Structural Adjustment andProperty Rights Conflicts on Nicaraguan Agrarian Reform Bene-ficiaries.” World Development 24(7):1179–91.

Kabeer, Naila. 1997. “Women, Wages and Intra-household Power Rela-tions in Urban Bangladesh.”Development and Change 28(2):261–302.

————. 1999. “Resources, Agency, Achievements: Reflections onthe Measurement of Women’s Empowerment.” Development andChange 30(3):435–64.

Kakwani, Nanak, and Kalanidhi Subbarao. 2005. Ageing and Povertyin Africa and the Role of Social Pensions. Washington, DC: WorldBank. Available on line at http://www-wds.worldbank.org/.

Kanbur, Rabi. 2000. “Income Distribution and Development.” InAnthony B. Atkinson and François Bourguignon, (ed.), Hand-book of Income Distribution. Amsterdam: North Holland.


Kanbur, Rabi, and Xiaobo Zhang. 2001. “Fifty Years of RegionalInequality in China: A Journey through Revolution, Reform andOpenness.” London: Centre for Economic Policy Research Dis-cussion Paper 2887.

Kanbur, Ravi. 2005. “Pareto’s Revenge.” Cornell University. Ithaca,NY. Available on line at http://www.arts.cornell.edu/poverty/kanbur/ParRev.pdf. Processed.

Kang, David C. 2002. Crony Capitalism: Corruption and Develop-ment in South Korea and the Philippines. Cambridge, UK: Cam-bridge University Press.

Kariuki, Mukami, and Jordan Schwartz. 2005. “Small-Scale PrivateService Providers of Water Supply and Electricity: A Review ofIncidence, Structure, Pricing and Operating Characteristics.”World Bank—Energy and Water Department, Bank NetherlandsWater Partnership, Public-Private Infrastructure Advisory Facil-ity. Washington, DC. Processed.

Karr-Morse, Robin, and Meredith S. Wiley. 1977. Ghost from theNursery: Tracing the Roots of Violence. New York: AtlanticMonthly Press.

Kaufmann, Daniel, and Aart Kraay. 2002. “Growth without Gover-nance.” Economia 3(1):169–215.

Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2004.“Governance Matters III: Governance Indicators for 1996, 1998,2000, and 2002.” World Bank Economic Review 18(2):253–87.

————. 2005. “Governance Matters IV: Governance Indicatorsfor 1996–2004.” World Bank. Washington, DC. Available on lineat http://www.worldbank.org/wbi/governance/pubs/govmatters4.html. Processed.

Kevane, Michael. 1996. “Agrarian Structure and Agricultural Prac-tice: Typology and Application to Western Sudan.” AmericanJournal of Agricultural Economics 78(1):236–245.

Khan, Azizur Rahman, and Binayak Sen. 2001. “Inequality and ItsSources in Bangladesh: 1991/92 to 1995/96: An Analysis Basedon Household Expenditure Surveys.” Bangladesh DevelopmentStudies 27(1):1–50.

Khandker, Shahidur, Victor Lavy, and Deon Filmer. 1994. “School-ing and Cognitive Achievements of Children in Morocco: Canthe Government Improve Outcomes?” Washington, DC: WorldBank Discussion Paper 264.

Khwaja, Asim Ijaz, and Atif Mian. 2004. “Unchecked Intermedi-aries: Price Manipulation in an Emerging Stock Market.”Boston: Bureau for Research in Economic Analysis of Develop-ment Working Paper 061. Available on line at http://www.cid.harvard.edu/bread/papers/working/061.pdf.

Kijima, Yoko, and Peter Lanjouw. 2004. “Agricultural Wages, Non-Farm Employment and Poverty in Rural India.” World Bank.Washington, DC. Processed.

King, Elizabeth M., Peter F. Orazem, and Elizabeth M. Paterno.1999. “Promotion with and without Learning: Effects on Stu-dents Dropouts.” Washington, DC: World Bank Working PaperSeries on Impact Evaluation of Education Reforms No. 18.

King, Steve. 1997.“Poor Relief and English Economic DevelopmentReappraised.” Economic History Review 50(2):360–68.

————. 2000. Poverty and Welfare in England, 1700-1850: ARegional Perspective. Manchester, UK: Manchester University Press.

Kingdon, Geeta, Justin Sandefur, and Francis Teal. 2005. “Patternsof Labor Demand in Sub-Saharan Africa: A Review paper.” Cen-tre for the Study of African Economies, Department of Econom-ics, University of Oxford. Oxford, UK. Processed.

Kinsey, Bill H., and Hans P. Binswansger. 1993. “Characteristics andPerformance of Resettlement Programs: A Review.” World Devel-opment 21(9):1477–94.

Kirpal, Simone. 2002. “Communities Can Make a Difference: FiveCases Across Continents.” In Mary Eming Young, (ed.), FromEarly Childhood Development to Human Development: Investingin our Children’s Future. Washington, DC: World Bank.

Klarén, Peter Flindell. 2000. Peru: Society and Nationhood in theAndes. New York: Oxford University Press.

Klasen, Stephen. 2005. Comments on “Hepatitis B and the Case ofthe Missing Women” by Emily Oster. University of Göttingen.Processed.

Knack, Steven, and Philip Keefer. 1995. “Institutions and EconomicPerformance: Cross-Country Tests using Alternative Measures.”Economics and Politics 7(3):207–27.

Kolm, Serge-Christophe. 1976. “Unequal Inequalities I.” Journal ofEconomic Theory 12(3):416–42.

Korinek, Anton, Johan Mistiaen, and Martin Ravallion. Forthcom-ing. “Survey Non-response and the Distribution of Income.”Journal of Economic Inequality.

Kraay, Aart. Forthcoming. “When is Growth Pro-Poor? Evidencefrom a Panel of Countries.” Journal of Development Economics.

Kranton, Rachel E., and Anand V. Swamy. 1999. “The Hazards ofPiecemeal Reform: British Civil Courts and the Credit Market inColonial India.” Journal of Development Economics 58(1):1–24.

Kremer, Michael. 1993. “The O-Ring Theory of Economic Devel-opment.” Quarterly Journal of Economics 108(3):551–75.

Kremer, Michael, and Rache Glennester. 2004. Strong Medicine -Creating Incentives for Pharmaceutical Research on Neglected Dis-eases. Princeton, NJ: Princeton University Press.

Kremer, Michael, Edward Miguel, and Rebecca Thornton. 2004.“Incentives to Learn.” Cambridge, MA: National Bureau of Eco-nomic Research Working Paper Series 10971.

Kremer, Michael, Sylvie Moulin, and Robert Namunyu. 2002.“Unbalanced Decentralization.” Harvard University. Cam-bridge, MA. Available on line at http://econ.bu.edu/dilipm/40-kremounam.pdf. Processed.

Kritzer, Herbert M. 2002. Legal Systems of the World: A Political,Social and Cultural Encyclopedia. Santa Barbara, CA: ABC-CLIO.

Krueger, Alan B., and Alexandre Mas. 2004. “Strikes, Scabs, andTread Separations: Labor Strife and the Production of Defec-tive Bridgestone/Firestone Tires.” Journal of Political Economy112(2):253–89.

Krugman, Paul. 1979. “A Model of Balance-of-Payments Crises.”Journal of Money, Credit, and Banking 11(3):311–25.

————. 1999. “Balance Sheets, the Transfer Problem andFinancial Crises.” In P. Isard, A. Razin, A. K. Rose, and KluwerDordrecht, (ed.), International Finance and Financial Crises,Essays in Honor of Robert P. Flood. Netherlands and Washing-ton, DC: Kluwer Academic Publishers and International Mone-tary Fund.

References 261

Kugler, Adriana. 2004. “The Effect of Job Security Regulations onLabor Market Flexibility: Evidence from the Colombian LaborMarket Reform.” Cambridge, MA: National Bureau of EconomicResearch Working Paper Series 10215.

Kugler, Adriana, Juan F. Jimeno, and Virginia Hernanz. 2003.“Employ-ment Consequences of Restrictive Permanent Contracts.” London:Centre for Economic Policy Research Discussion Paper 3724.

Kumar, Anjali. 2005. Assessing Financial Access In Brazil. Washing-ton, DC: World Bank.

La Porta, Rafael, Florencio López-de-Silanes, and GuillermoZamarripa. 2002. “Related Lending.” Cambridge, MA: NationalBureau of Economic Research Working Paper Series 8848.

Ladd, Hellen F. 2002. “School Vouchers: A Critical View.” Journal ofEconomic Perspectives 16(4):3–24.

Laffont, Jean-Jacques, and Mohamed Salah Matoussi. 1995. “MoralHazard, Financial Constraints and Sharecropping in El Oulja.”Review of Economic Studies 62(3):381–99.

Lagos, Marta (2005). Personal communication with Marta Lagos,Director, Latinobarómetro.

Lang, James. 1975. Conquest and Commerce: Spain and England inthe Americas. New York: Academic Press.

Langseth, Petter, and Oliver Stolpe. 2001. Strengthening JudicialCorruption against Corruption. Vienna: United Nations Office forDrug Control and Crime Prevention, Global Programme AgainstCorruption. Available on line at http://www.unodc.org/pdf/crime/gpacpublications/cicp10.pdf.

Lanjouw, Jean O. 2002. “A New Global Patent Regime for Diseases:U.S. and International Legal Issues.” Harvard Journal of Law andTechnology 16(1):85–124.

————. 2004. Outline of the Foreign Filing License Approach.Washington, DC: Center for Global Development.

Lanjouw, Jean O., and William Jack. 2004. “Trading Up: How MuchShould Poor Countries Pay to Support Pharmaceutical Innova-tion?” CGD Brief 4(3):1–8.

Lanjouw, Jean O., and Margaret MacLeod. 2005. “Statistics Trendsin Research on Tropical Diseases.” Berkeley University. Berkeley,CA. Available on line at http://www.who.int/intellectualproperty/studies/Lanjouw_Statistical%20Trends.pdf. Processed.

Lanjouw, Peter, and Martin Ravallion. 2005. “Progresiveness ofSocial and Infrastructure Spending in India.” World Bank. Wash-ington, DC. Processed.

Larrain, Mauricio. 1989.“How the 1981–1983 Chilean Banking Crisiswas Handled.” Washington, DC: World Bank Working Paper 300.

Lavy, Victor. 1996. “School Supply Constraints and Children’s Edu-cational Outcomes in Rural Ghana.” Journal of Development Eco-nomics 51(2):291–314.

Lees, Lynn Hollen. 1998. The Solidarities of Strangers: The English PoorLaws and the People. Cambridge, UK: Cambridge University Press.

Leipziger, Danny, Marianne Fay, Quentin Wodon, and Tito Yepes.2003. “Achieving the Millennium Development Goals: The Roleof Infrastructure.” Washington, DC: World Bank Policy ResearchWorking Paper Series 3163.

Leonard, Kenneth L., and Melkiory C. Masatu. 2005. “Variation inthe Quality of Care Accessible to Rural Communities in Tanza-nia.” University of Maryland. College Park, MD. Processed.

Levin, Victoria. 2005a. “Updating Aid Allocation Estimates.” Back-ground Note for the WDR Note 2006.

———. 2005b. “Updating the Poverty-efficient Aid Allocation to2001.” World Bank. Washington, DC. Processed.

Levin, Victoria, and David Dollar. 2005. “The Forgotten States: AidVolumes and Volatility in Difficult Partnership Countries(1992–2002).” OECD, DAC Learning and Advisory Process onDifficult Partnerships. Paris. Available on line at http://www.oecd.org/dataoecd/32/44/34687926.pdf. Processed.

Lewin, Bryan, Daniele Giovannucci, and Panos Varangis. 2004.“Coffee Markets: New Paradigms in Global Supply andDemand.” Washington, DC: World Bank, Agriculture and RuralDevelopment Discussion Paper 3.

Lewis, Arthur. 1954. “Economic Development with Unlimited Sup-plies of Labor.” Manchester School of Economic and Social Studies22(2):139–91.

Lewis, M. 2002. “Informal Health Payments in Central and EasternEurope and the Former Soviet Union: Issues, Trends and PolicyImplications.” In Elias Mossialos, Ana Dixon, Josep Figueras, andJoe Kutzin, (ed.), Funding Health Care: Options for Europe. Buck-ingham: Open University Press.

Lewis, Stephen. 2003. “Opening Address to the 13th InternationalConference on AIDS and STIs in Africa (ICASA).” Paper pre-sented at the 13th International Conference on AIDS and STIs inAfrica (ICASA). September 21. Nairobi.

Li, Hongyi, Lyn Squire, and Heng-fu Zou. 1998. “Explaining Inter-national and Intertemporal Variations in Income Inequality.”Economic Journal 108(446):26–43.

Li, Hongyi, and Heng-fu Zou. 1998. “Income Inequality is notHarmful for Growth: Theory and Evidence.” Review of Develop-ment Economics 2(3):318–34.

Li, R., X. Chen, H. Yan, P. Deurenberg, L. Garby, and J. G. Hautvast.1994. “Functional Consequences of Iron Supplementation inIron-deficient female Cotton Workers in Beijing China.” Ameri-can Journal of Clinical Nutrition 59(4):908–13.

Liddle, R. William. 1991. “The Relative Autonomy of the ThirdWorld Politician: Soeharto and Indonesian Economic Develop-ment in Comparative Perspective.” International Studies Quar-terly 35(4):403–27.

Lillard, Lee A., and Robert J. Willis. 1994. “Intergenerational Educa-tional Mobility: Effects of Family and State in Malaysia.” Journalof Human Resources 29(4):1126–66.

Lindert, Peter H. 2003. “Voice and Growth: Was Churchill Right?”Journal of Economic History 63(2):315–50.

————. 2004. Growing Public: Social Spending and EconomicsGrowth since the Eighteenth Century. Cambridge, UK: CambridgeUniversity Press.

Lindert, Peter H., and Jeffrey G. Williamson. 1982. “Revising Eng-land’s Social Tables, 1688–1812.” Explorations in Economic His-tory 19(4):385–408.

————. 1983.“Reinterpreting Britain’s Social Tables, 1688–1913.”Explorations in Economic History 20(1):94–109.

Lipset, Seymour M. 1959. “Some Social Requisites of Democracy:Economic Development and Political Legitimacy.” AmericanPolitical Science Review 53(1):69–105.


Llorente, María, and Angela Rivas. 2005. “Reduction of Crime inBogota: A Decade of Citizen’s Security Policies.” World Bank,Department of Finance, Private Secton and Infrastructure.Washington, DC. Processed.

Lloyd, Cynthia B., eds. 2005. Growing up Global: The ChangingTransitions to Adulthood in Developing Countries. Washington,DC: The National Academies Press.

Lloyd, Tim, Wyn Morgan, Tony Rayner, and Charlotte Vaillant.1999. “The Transmission of World Agricultural Prices in Côted’Ivoire.” Journal of International Trade and Economic Develop-ment 8(1):125–41.

Lockhart, James B. 1969. “Encomienda and Hacienda: The Evolu-tion of the Great Estate in the Spanish Indies.” Hispanic AmericanHistorical Review 49(3):411–29.

Lockhart, James B., and Stuart B. Schwartz. 1983. Early Latin Amer-ica: A History of Colonial Spanish and America and Brazil. Cam-bridge, UK: Cambridge University Press.

Lokshin, Michael, and Martin Ravallion. 2000. “Welfare Impacts ofthe 1998 Financial Crisis in Russia and the Response of the Pub-lic Safety Net.” Economics of Transition 8(2):269–95.

López, Alan D. 2000. “Reducing Chile’s Mortality.” Bulletin of theWorld Health Organization 78(10):1173–73.

López, J. Humberto. 2004. “Pro-growth, Pro-poor: Is there a Trade-off?” Washington, DC: World Bank Policy Research WorkingPaper Series 3378.

Loury, Glenn C. 2002. The Anatomy of Racial Inequality. Cam-bridge, MA: Harvard University Press.

Lucas, Robert. 2003. The Industrial Revolution: Past and Future.Minneapolis: Federal Reserve Bank of Minneapolis.

Lund, Frances. 1999. “Understanding South African Social Securitythrough Recent Household Surveys: New Opportunities andContinuing Gaps.” Development Southern Africa 16(1):55–67.

Lundberg, Mattias, and Lyn Squire. 2003. “The SimultaneousEvolution of Growth and Inequality.” Economic Journal113(487):326–44.

Lundberg, Shelley J., Robert A. Pollak, and Terence J. Wales. 1997.“Do Husbands and Wives Pool their Resources?” Journal ofHuman Resources 32(3):463–80.

Luttmer, Erzo F. P. 2004. “Neighbors as Negatives: Relative Earningsand Well-Being.” Cambridge, MA: Faculty Research WorkingPaper Series 4-029.

MacIntyre, Andrew. 2001a. “Institutions and Investors: The Politicsof the Economic Crisis in Southeast Asia.” International Organi-zation 55(1):81–122.

————. 2001b. “Rethinking the Politics of Agricultural PolicyMaking: The Importance of Institutions.” In J. Edgard Campos,(ed.), The Evolving Roles of State, Private, and Local Actors inAsian Rural Development. Cambridge, UK: Oxford UniversityPress.

Maddison, Angus. 1995. Monitoring the World Economy:1820–1992.Paris: Development Centre of the Organisation for EconomicCo-operation and Development.

Mahoney, James. 2001. The Legacies of Liberalism: Path Dependenceand Political Regimes in Central America. Baltimore, MD: JohnsHopkins University Press.

Malik, Waleed Haider. 2005.“Guatemala: The Role of Judicial Mod-ernization in Post Conflict Reconciliation.” Washington, DC:Social Development Notes, World Bank 99.

Malmberg Calvo, Christina. 1994. “Case Study on the Role ofWomen in Rural Transport: Access of Women to DomesticFacilities.” World Bank and Economic Commission for Africa,Washington, DC: Sub-Saharan Africa Transport Policy ProgramWorking Paper 11.

Maloney, William F. 1999. “Does Informality Imply Segmentationin Urban Labor Markets? Evidence from Sectoral Transitions inMexico.” World Bank Economic Review 13(2):275–302.

Maloney, William F., and Jairo Nuñez Mendez. 2004. “Measuringthe Impact of Minimum Wages, Evidence from Latin America.”In James Heckman and Carmen Pagés, (ed.), Law and LaborMarkets. Chicago: University of Chicago Press.

Mamdani, Mahmood. 1996. Citizen and Subject: ContemporaryAfrica and the Legacy of Late Colonialism. Princeton, NJ: Prince-ton University Press.

Manor, James. 1999. The Political Economy of Democratic Decentral-ization. Washington, DC: World Bank.

Marmot, Michael G. 2004. The Status Syndrome: How Social Stand-ing Affects our Health and Longevity. New York: Times Books.

Mason, Karen O., and Helene M. Carlsson. Forthcoming. “TheDevelopment Impact of Gender Equality in Land Rights.” InPhilip Alston and Mary Robinson (ed.) Human Rights andDevelopment: Towards Mutual Reinforcement. New York: OxfordUniversity Press.

Massey, D. 2001. “The Progress in Human Geography Lecture,Geography on the Agenda.” Progress in Human Geography25(1):5–17.

Masters, Williams A. 2005. “Paying for Prosperity: How and Why toInvest in Agricultural R&D for Development in Africa.” Journalof International Affairs 58(2):35–64.

Masters, Williams A., F. T. Sparrow, and Brian H. Bowen. 1999.Modeling Electricity Trade in Southern Africa 1999–2000. WestLafayette, Indiana: Purdue University. Available on line athttps://engineering.purdue.edu/IE/Research/PEMRG/PPDG/SAPP/1999proposal.pdf.

Mattei, U. 1998. “Legal Pluralism, Legal Change and EconomicDevelopment.” In Lyda Favali, E. Grande, and M. Guadagni,(ed.), New Law for New States. Torino, Italy: L’Harmattan Italia.

Mattoo, Aaditya, Ileana Cristina Neagu, and Caglar Ozden. 2005.“Brain Waste? Educated Immigrants in the U.S. Labor Market.”Washington, DC: World Bank Policy Research Working PaperSeries 3581.

Maurer, Stephen M., Andrej Sali, and Arti Rai. 2004.“Finding Curesfor Tropical Disease: Is Open Source the Answer?” Public Libraryof Science: Medicine 1(3):33–7.

Maynard, E., eds. 1966. The Indians of Colta. Essays on the ColtaLake Zone, Chimborazo (Ecuador). Ithaca, NY: Cornell Uni-versity.

Mazumder, Bhakshar. 2005.“The Apple Falls Even closer to the Treethan We Thought: New and Revised Estimates of the Intergener-ational Inheritance of Earnings.” In Samuel Bowles, Herbert Gin-tis, and Melissa Osborne Groves, (ed.), Unequal Chances: Family

References 263

Background and Economic Success. Princeton, NJ: Princeton Uni-versity Press.

McEwan, Patrick J. 2000. “The Impact of Vouchers on School Effi-ciency: Empirical Evidence from Chile.” Ph.D. thesis. StanfordUniversity.

McEwan, Patrick J., and Martin Carnoy. 2000. “The Effectivenessand Efficiency of Private Schools in Chile’s Voucher System.”Educational Evaluation and Policy Analysis 22(3):213–39.

McGillivray, Mark. 2005. “Aid Allocation and Fragile States.” Paperpresented at the Senior Level Forum on Development Effective-ness in Fragile States. January 13. London.

McKay, Andrew, Oliver Morrissey, and Charlotte Vaillant. 1997.“Trade Liberalization and Agricultural Supply Response: A Studyof Rural Asia.” European Journal of Development 9(2):129–47.

McKenzie, David, and Dilip Mookherjee. 2003. “DistributiveImpact of Privatization in Latin America: An Overview of Evi-dence from Four Countries.” Economia 3(2):161–218.

McKenzie, David, and Hillel Rapoport. 2004. “Network Effects andthe Dynamics of Migration and Inequality: Theory and Evidencefrom Mexico.” Washington DC: BREAD Working Paper Series 63.

McKenzie, David, and Christopher Woodruff. 2003. “Do EntryCosts Provide an Empirical Basis for Poverty Traps? Evidencefrom Mexican Microenterprises.” Cambridge, MA: Harvard Uni-versity, Bureau for Research in Economic Analysis of Develop-ment (BREAD) Working Paper 20. Available on line athttp://www.cid.harvard.edu/bread/papers/working/020.pdf.

McMillan, Margaret, Dani Rodrik, and Karen Horn Welch. 2002.“When Economic Reform Goes Wrong: Cashews in Mozam-bique.” Cambridge, MA: National Bureau of Economic ResearchWorking Paper Series 9117.

Meesen, Bruno, and Wim Van Damme. 2004. “Health Equity inCambodia, A New Approach to Fee Exemptions.” Paper pre-sented at the Reaching the Poor with Effective Health, Nutrition,and Population Services: What Works, What Doesn’t, and Why.February 18. Washington, DC.

Melo, Jorge Orlando. 1996. Historia de Colombia: La DominaciónEspañola. Bogotá: Biblioteca Familiar.

Mendola, Mariapia. 2004. “Migration and Technological Change inRural Households: Complements or Substitutes?” Milano, Italy:Centro Study Luca d’Agliano, Development Studies WorkingPaper 195.

Mesnard, Alice, and Martin Ravallion. 2004. “The Wealth Effects onNew Business Startups in a Developing Economy.” Instituteof Fiscal Studies and World Bank. London, U.K. and Wash-ington, DC. Available on line at http://www.cefims.ac.uk/pdfs/mesnard_paper.pdf. Processed.

Meyer, Richard, and Geetha Nagarajan. 2000.“Rural Financial Mar-kets in Asia: Policies, Paradigms, and Performance.” In AsianDevelopment Bank, (ed.), A Study of Rural Asia vol. 3. HongKong, China: Oxford University Press for the Asian DevelopmentBank.

Miguel, Edward, and Michael Kremer. 2004. “Worms: IdentifyingImpacts on Education and Health in the Presence of TreatmentExternalities.” Econometrica 72(1):159–217.

Milanovic, Branko. 2002. “Can We Discern the Effect of Globaliza-tion on Income Distribution? Evidence from Household Sur-veys.” Washington, DC: World Bank Policy Research ReportWorking Paper Series 2876.

————. 2005. Worlds Apart: International and Global Inequality1950-2000. Princeton, NJ: Princeton University Press.

Minot, Nicholas W. 1998. “Distributional and Nutritional Impactof Devaluation in Rwanda.” Economic Development and CulturalChange 46(2):379–402.

Minot, Nicholas W., and Lisa Daniels. 2002. “Impact of Global Cot-ton Markets on Rural Poverty in Benin.” Washington, DC: Inter-national Food Policy Research Institute Discussion Paper 48.

Minot, Nicholas W., and Francesco Goletti. 1998. “Rice Market Lib-eralization and Poverty in Vietnam.” Washington, DC: Interna-tional Food Policy Research Institute Research Report 114.

Moene, Karl Ove, and Michael Wallerstein. 2002. “Social Democ-racy as a Development Strategy.” University of Oslo, Departmentof Economics: Memorandum 35/2003.

Moncrieffe, Joy. 2005. “Beyond Categories: Power, Recognition, andthe Conditions for Equity.” Background paper for the WDR2006.

Montenegro, Claudio E., and Carmen Pagés. 2004. “Who Benefitsfrom Labor Market Regulations?” In James Heckman and Car-men Pagés, (ed.), Law and Employment: Lessons from the LatinAmerican and the Caribbean. Chicago: University of ChicagoPress. Reprinted in Jorge Enrique Restrepo and Andrea TokmanR. (ed.) “Labor Markets and Institutions,” (2005), Santiago deChile: Banco Central de Chile.

Montinola, Gabriela, Yingyi Qian, and Barry Weingast. 1995. “Fed-eralism Chinese-Style: The Political Basis for Economic Successin China.” World Politics 48(1):50–81.

Moore, Barrington Jr. 1966. The Social Origins of Dictatorship andDemocracy: Lord and Peasant in the Making of the Modern World.Boston, MA: Beacon Press.

Moore, Mick. 2001. “Political Underdevelopment: What Causes‘Bad Governance’?” Public Management Review 3(3):385–418.

————. 2004. “Revenues, State Formation, and the Quality ofGovernance in Developing Countries.” International Political Sci-ence Review 25(3):297–319.

Morales, Juan Antonio, and Jeffrey D. Sachs. 1998. “Bolivia’s Eco-nomic Crisis.” Cambridge, MA: National Bureau of EconomicResearch Working Paper Series 2620.

Morck, Randall, Daniel Wolfenzon, and Bernard Yeung. 2004.“Cor-porate Governance, Economic Entrenchment and Growth.”Cambridge, MA: National Bureau of Economic Research Work-ing Paper Series 10692.

Morck, Randall, and Bernard Yeung. 2004. “Special Issues Relatingto Corporate Governance and Family Control.” Washington, DC:World Bank Policy Research Working Paper Series 3406.

Morck, Randall K., David A. Stangeland, and Bernard Yeung. 2000.“Inherited Wealth, Corporate Control, and Economic Growth:The Canadian Disease?” In Randall Morck, (ed.), ConcentratedCorporate Ownership. National Bureau of Economic ResearchConference Volume. Chicago: University of Chicago Press.


Morduch, Jonathan. 1993. “Risk Production and Saving: Theoryand Evidence from Indian Households.” Cambridge, MA. Har-vard University. Processed.

Morgan, Edmund S. 1975. American Slavery, American Freedom:The Ordeal of Colonial Virginia. New York: W. W. Norton & Co.

Morley, Samuel. 2001. The Income Distribution Problem in LatinAmerica and the Caribbean. Santiago de Chile: Economic Com-mission for Latin American and the Caribbean.

Morley, Samuel, and David Coady. 2003. From Social Assistance toSocial Development: A Review of Targeted Education Subsidies inDeveloping Countries. Washington, DC: International Food Pol-icy Research Institute.

Mörner, Magnus. 1973. “The Spanish American Hacienda: A Sur-vey of Recent Research and Debate.” Hispanic American Histori-cal Review 53(2):183–216.

Morrison, Andrew, Mary Ellsberg, and Sarah Bott. 2004. “Address-ing Gender-Based Violence in the Latin American and CaribbeanRegion: A Critical Review of Interventions.” Washington, DC:World Bank Policy Research Report Working Paper Series 3438.

Mtani, Anna. 2002.“Safety Audits and Beyond.”Paper presented at theFirst International Seminar on Women’s Safety. May 9. Montreal.

Mundlak,Yair,and Donald F.Larson.1992.“On the Transmission of WorldAgricultural Prices.”World Bank Economic Review 6(3):399–422.

Munshi, Kaivan. 2003. “Networks in the Modern Economy: Mexi-can Migrants in the U.S. Labor Markets.” Quarterly Journal ofEconomics 118(2):549–97.

Munshi, Kaivan, and Mark Rosenzweig. Forthcoming. “TraditionalInstitutions Meet the Modern World: Caste, Gender and School-ing in a Globalizing Economy.” American Economic Review.

Murgai, Rinku, and Martin Ravallion. 2005. “A Guaranteed LivingWage in Rural India: Who Would Gain and at What Cost?”World Bank. Washington, DC. Processed.

Mwaikusa, Jwani T. 1995. “Maintaining Law and Order in Tanza-nia: The Role of Sungusungu Defence Groups.” In Joseph Sem-boja and Ole Therkildsen, (ed.), Service Provision Under Stress inEast Africa: The State, NGOs, and People’s Organizations in Kenya,Tanzania, and Uganda. Copenhagen: Center for DevelopmentResearch.

Myers, Robert. 1995. The Twelve Who Survive: Strengthening Pro-grams of Early Childhood Development in the Third World. Ypsi-lanti, MI: High Scope Press.

Narayan, A., and N. Yoshida. 2004. “Poverty in Sri Lanka: TheImpact of Growth with Rising Inequality.” Paper presented at theWorld Development Consultation Meeting. December 12. Delhi.

Narayan, Deepa, eds. 2002. Empowerment and Poverty Reduction: ASourcebook. Washington, DC: World Bank.

Neumark, David, Wendy Cunningham, and Lucas Siga. Forthcoming.“The Effects of Minimum Wages in Brazil on the Distribution ofFamily Incomes: 1996–2001.” Journal of Development Economics.

Newman, John, Steen Jorgensen, and Menno Pradhan. 1992. “HowDid Workers Benefit?” In Steen Jorgensen, Margaret Grosh, andMark Schacter, (ed.), Bolivia’s Answer to Poverty, Economic Crisisand Adjustment: The Emergency Social Fund. Washington, DC:World Bank.

Nicita, Alessandro. 2004. “Who Benefited from Trade Liberalizationin Mexico? Measuring the Effects on Household Welfare.” Wash-ington, DC: World Bank Policy Research Working Paper Series3265.

Nickell, Stephen. 1997. “Unemployment and Labor Market Rigidi-ties: Europe versus North America.” Journal of Economic Perspec-tives 11(3):55–74.

Nielsen, Robin, and Tim Hanstad. 2004. “From Sharecroppers toLandowners: Paving the Way for West Bengal’s Bargadars.” Seat-tle, WA: Rural Development Institute, Reports on Foreign Aid121. Available on line at http://www.rdiland.org/PDF/PDF_Reports/RDI_121.pdf.

North, Douglas C. 1981. Structure and Change in Economic History.New York: W. W. Norton & Co.

North, Douglas C., and Robert P. Thomas. 1973. The Rise of theWestern World: A New Economic History. Cambridge, U.K.: Cam-bridge University Press.

North, Douglas C., and Barry R. Weingast. 1989. “Constitutionsand Commitment: The Evolution of Institutions GoverningPublic Choice in Seventeenth-Century England.” Journal of Eco-nomic History 49(4):803–32.

Nozick, Robert. 1974. Anarchy, State And Utopia. New York: BasicBooks.

Nussbaum, Martha C. 2000. Women and Human Development: TheCapabilities Approach. New York: Cambridge University Press.

O’Brien, Patrick K. 1993. “Political Preconditions for the IndustrialRevolution.” In Patrick O’Brien and Roland Quinault, (ed.), TheIndustrial Revolution and British Society. Cambridge, UK: Cam-bridge University Press.

O’Rourke, Kevin H., and Jeffrey G. Williamson. 2002. “FromMalthus to Ohlin: Trade, Growth and Distribution Since 1500.”Cambridge, MA: National Bureau of Economic Research Work-ing Paper Series 8955.

O’Sullivan, Robert, and Amiee Christensen. 2005. “Equity andInequity in the Context of Climate Change.” Background notefor the WDR 2006.

Obstfeld, Maurice. 1996. “Models of Currency Crises with Self-fulfilling Features.” European Economic Review 40(3-5):1037–47.

OECD. 1999. Employment Outlook. Paris: Organisation for Eco-nomic Co-operation and Development.

————. 2000. Hard Core Cartels. Paris: Organisation for Eco-nomic Co-operation and Development. Available on line athttp://www.oecd.org/dataoecd/36/24/2367816.pdf.

————. 2003. Agricultural Policies in OECD Countries: Monitor-ing and Evaluation 2003. Paris: Organisation for Economic Co-operation and Development. Available on line at www1.oecd.org/publications/e-book/5103081E.PDF.

————. 2005. “OECD Productivity Database (February 2005version)”. Paris, Organisation for Economic Co-operation andDevelopment.

OECD-DAC. 2004. Development Cooperation Report StatisticalAnnex 2004. Paris: Organization for Economic Co-operation andDevelopment. Available on line at http://www.oecd.org/document/9/0,2340,en_2649_33721_1893129_1_1_1_1,00.html.

References 265

Olley, G. Steven, and Ariel Pakes. 1996. “The Dynamics of Produc-tivity in the Telecommunications Equipment Industry.” Econo-metrica 64(6):1263–97.

Osberg, Lars, and Timothy Smeeding. 2004. “Fair’ Inequality?: AnInternational Comparison of Attitudes to Pay Differentials.” Dal-housie University. Dalhousie. Available on line at http://wwwcpr.maxwell.syr.edu/faculty/smeeding/selectedpapers/Economicaversion27October2004.pdf. Processed.

Oster, Emily. 2005. “Hepatitis B and the Case of the MissingWomen.” Cambridge, MA: Harvard University, CID GraduateStudent and Postdoctoral Fellow Working Paper 7. Available online at http://www.cid.harvard.edu/cidwp/graduate.html.

Paes de Barros, Ricardo, and Mirela de Carvalho. 2004. “Targetingas an Instrument for a more Effective Social Policy.” Inter-American Development Bank. Washington, DC. Available on line at http://www.iadb.org/sds/doc/POV-SEFVI-Barros_Carvalho.pdf. Processed.

Palandt, Otto. 2004. Bürgerliches Gesetzbuch. Munich: C. H. Beck.

Palmade, Vincent. 2005. “Industry Level Analysis: The Way to Iden-tify the Binding Constraints to Economic Growth.” Washington,DC: World Bank Policy Research Working Paper Series 3551.

Panagides, Alexis, and Harry Anthony Patrinos. 1994. “Union-nonunion Wage Differentials in the Developing World: A CaseStudy of Mexico.” Washington, DC: World Bank Policy ResearchReport Working Paper Series 1269.

Pande, Rohini. 2003. “Can Mandated Political RepresentationIncrease Policy Influence for the Disadvantaged Minorities? The-ory and Evidence from India.” American Economic Review93(4):1132–51.

Parry, John H. 1948. The Audiencia of New Galicia in the SixteenthCentury: A Study in Spanish Colonial Government. Cambridge,UK: Cambridge University Press.

Pathmanathan, Indra, Jerke Liljestrand, Jo M. Martins, Lalini C.Rajapaksa, Craig Lissner, Amalia de Silva, Swarna Selvajuru, andPrabha Joginder Singh. 2003. Investing in Maternal Health:Learning from Malaysia and Sri Lanka. Washington, DC: WorldBank.

Paxson, Christina H., and Norbert Schady. 2004. “Child Health andEconomic Crisis in Peru.” Washington, DC: World Bank PolicyResearch Working Paper Series 3260.

————. 2005. “Cognitive Development among Young Children inEcuador: The Roles of Wealth, Health and Parenting.”Washington,DC: World Bank Policy Research Working Paper Series 3605.

Payne, Geoffrey, eds. 2002. Land, Rights, and Innovation: ImprovingTenure for the Urban Poor. London, UK: ITDG Publishing.

Payne, Geoffrey, and Michael Majale. 2004. The Urban HousingManual: Making Regulatory Frameworks Work for the Poor. Lon-don, U.K.: Earthscan Publications.

Perotti, Enrico. 2002. “Lessons from the Russian Meltdown: TheEconomics of Soft Legal Constraints.” International Finance5(3):359–99.

Perotti, Enrico, and Erik Feijen. 2005.“Lobbying for Exit: The Polit-ical Economy of Financial Fragility.” University of Amsterdam.Amsterdam. Available on line at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=686661. Processed.

Perotti, Enrico, and Paolo Volpin. 2004. “Lobbying on Entry.” Lon-don, UK: Centre for Economic Policy Research Discussion PaperSeries 4519.

Perry, Guillermo. 2003. “Can Fiscal Rules Reduce MacroeconomicVolatility in the Latin American and Caribbean Region?”Washing-ton, DC: World Bank Policy Research Working Paper Series 3080.

Petesch, Patti, Catalina Smulovitz, and Michael Walton. 2005.“Evaluating Empowerment: A Framework with Cases from LatinAmerica.” In Deepa Narayan, (ed.), Measuring Empowerment:Cross-disciplinary Perspectives. Washington, DC: World Bank.

Piketty, Thomas. 1995. “Social Mobility and Redistributive Poli-tics.” Quarterly Journal of Economics 110(3):551–84.

Pinglé, Vibha. 2005. “Faith, Equity, and Development.” Backgroundpaper for the WDR 2006.

Pinto, Brian, Evsey Gurvich, and Sergei Ulatov. Forthcoming.“Lessons from the Russia Crisis of 1998 and Recovery.” In JoshuaAizenman and Brian Pinto (ed.) Managing Volatility and Crisis:A Practitioner’s Guide. Washington, DC: World Bank.

Pistor, Katharina, and Phillip A. Wellons. 1999. The Role of Law andLegal Institutions in Asian Economic Development: 1960-1995.Hong Kong, China: Oxford University Press for the Asian Devel-opment Bank.

Pitayarangsarit, Siriwan. 2004. “Agenda Setting Process.” In VirojTangcharoensathien and Pongpisut Jongudomsuk, (ed.), FromPolicy to Implementation: Historical Events During 2001–2004 ofUniversal Coverage in Thailand. Nonthaburi, Thailand: NationalHealth Security Office.

Pogge, Thomas W. 2004. “Assisting the Global Poor.” In Deen K.Chatterjee, (ed.), The Ethics of Assistance: Morality and the Dis-tant Needy. Cambridge, UK: Cambridge University Press.

Pollitt, E., W. E. Watkins, and M. A. Husaini. 1997. “Three-MonthNutritional Supplementation in Indonesian Infants and Tod-dlers Benefits Memory Function 8 Years Later.” American Journalof Clinical Nutrition 66(6):1357–363.

Pradhan, Menno, David E. Sahn, and Stephen D. Younger. 2003.“Decomposing World Health Inequality.” Journal of Health Eco-nomics 22(2):271–93.

Preker, Alexander, and Guy Carrin, eds. 2004. Health Financing forPoor People: Resource Mobilization and Risk Sharing. Washington,DC: World Bank.

Preston, Samuel H. 1980. “Causes and Consequences of MortalityDeclines in Less Developed Countries During the Twentieth Cen-tury.” In Richard Easterlin, (ed.), Population and Economic Changein Developing Countries. Chicago: University of Chicago Press.

Pritchett, Lant. 1997. “Divergence: Big Time.” Journal of EconomicPerspectives 11(3):3–17.

————. 2001. “Where Has All the Education Gone?” World BankEconomic Review 15(3):367–91.

————. 2003. “The Future of Migration: Irresistible Forces MeetImmovable Ideas.” Paper presented at the The Future of Global-ization: Explorations in Ligth of the Recent Turbulence Confer-ence. October 11. New Haven, CT.

————. 2004a. “Towards a New Consensus for Addressing theGlobal Challenge of the Lack of Education.”Kennedy School of Gov-ernment, Harvard University. Cambridge, MA. Available on line at


http://www.copenhagenconsensus.com/Files/Filer/CC/Papers/Education_230404.pdf. Processed.

————. 2004b. “Does Learning to Add Up Add Up? The Returnsto Schooling in Aggregate Data.” Cambridge, MA: Bureau forResearch in Economic Analysis of Development (BREAD)Working Paper 053.

————. Forthcoming. “Who is Not Poor? Dreaming of a WorldTruly Free of Poverty.” World Bank Research Observer.

Prud’Homme, Rémy. 1990. “Decentralization of Expenditures orTaxes: The Case of France.” In Robert J. Bennett, (ed.), Decentral-ization, Local Governments, and Markets. Oxford, UK: ClarendonPress.

Psacharopoulos, George, and Harry Anthony Patrinos. 1994.Indigenous People and Poverty in Latin America: An EmpiricalAnalysis. Washington, DC: World Bank.

Pulley, Robert V. 1989. “Making the Poor Creditworthy: A CaseStudy of the Integrated Rural Development Program in India.”Washington, DC: World Bank Discussion Paper 58.

Qian, Yingyi. 2003. “How Reform Worked in China.” In DaniRodrik, (ed.), In Search of Prosperity: Analytic Narratives of Eco-nomic Growth. Princeton, NJ: Princeton University Press.

Quisumbing, Agnes R., Jona P. Estudillo, and Keijiro Otsuka. 2004.Land and Schooling: Transferring Wealth Across Generations. Bal-timore, MD: Johns Hopkins University Press.

Quisumbing, Agnes R., and John A. Maluccio. 2003. “Resources atMarriage and Intrahousehold Allocation: Evidence fromBangladesh, Ethiopia, Indonesia, and South Africa.” Oxford Bul-letin of Economics and Statistics 65(3):283–327.

Rabin, Mathew. 1993. “Incorporating Fairness Into Game Theoryand Economics.” American Economic Review 83(5):1281–302.

Rahman, Lupin, and Vijayendra Rao. 2004. “The Determinants of Gender Equity in India: Examining Dyson and Moore’s Thesis with New Data.” Population and Development Review30(2):239–68.

Rajan, Ragurham G., and Luigi Zingales. 2003. Saving Capitalismfrom the Capitalists: Unleashing the Power of Financial Markets tocreate Wealth and Spread Opportunity. New York: Crown Business.

Ranson, Kent, Tara Sinha, Mirai Chatterjee, Akash Adharya, AmiBhavsar, Saul Morris, and Anne J. Mills. 2005. “Making HealthInsurance Work for the Poor: Learning from SEWA’s Community-based Health Insurance Scheme.” Processed.

Rao, Vijayendra. 2005. “Symbolic Public Goods and the Coordi-nation of Collective Action: A Comparison of Local Develop-ment in India and Indonesia.” Background paper for the WDR2006.

Rao, Vijayendra, and Michael Walton. 2004. Culture and PublicAction. Stanford, CA: Stanford University Press.

Ravallion, Martin. 1991. “Reaching the Rural Poor through PublicEmployment: Arguments, Evidence and Lessons from SouthAsia.” World Bank Research Observer 6(2):153–75.

————. 2001. “Growth, Inequality and Poverty: Looking BeyondAverages.” World Development 29(11):1803–15.

————. 2003. “Targeted Transfers in Poor Countries: Revisitingthe Trade-Offs and Policy Options.” Washington, DC: WorldBank Social Protection Discussion Paper Series 0314.

————. 2004a. “Competing Concepts of Inequality in the Glob-alization Debate.” In Susan Margaret Collins and Carol Graham,(ed.), Brookings Trade Forum 2004. Washington, DC: BrookingsInstitution.

————. 2004b. “Looking Beyond Averages in the Trade andPoverty Debate.” Washington, DC: World Bank Policy ResearchWorking Paper Series 3461.

————. 2005. “Why Should Poor People Care about Inequality?”Background paper for the WDR 2006.

Ravallion, Martin, and Shaohua Chen. 1997. “What Can New Sur-vey Data Tell us about Recent Changes in Distribution andPoverty?” World Bank Economic Review 11(2):357–82.

————. 2004. “China’s (Uneven) Progress Against Poverty.”Washington, DC: World Bank Policy Research Working PaperSeries 3408.

Ravallion, Martin, and Michael Lokshin. 2002. “Self-rated Eco-nomic Welfare in Russia.” European Economic Review46(8):1453–73.

Ravallion, Martin, and Quentin Wodon. 1999. “Poor Areas, or onlyPoor People?” Journal of Regional Science 39(4):689–711.

Rawlings, Laura B. 2004. “A New Approach to Social Assistance:Latin America’s Experience with Conditional Cash Transfer Pro-grams.” Washington, DC: World Bank, Social Protection Discus-sion Paper Series 0416.

Rawls, John. 1971. A Theory of Justice. Cambridge, MA: HarvardUniversity Press.

Revenga, Ana. 1991. “La Liberalización Económica y la Distribu-ción de la Renta: La Experiencia Española.” Moneda y Crédito193:179–224.

Ricupero, Rubens. 2005. “Why Should Small Developing CountriesEngage in the Global Trading System? Three Points of View on aHot Topic in the Doha Round: Overcoming Fear First.” Financeand Development 42(1):10–11.

Ringera, A., J. W. Onyango-Otieno, W. Karanja, and M. W. Muigai.2003. An Anatomy of Corruption in the Kenyan Judiciary: TheReport of the Integrity and Anti-Corruption Committee of the Judi-ciary. Nairobi, Kenya: The Judiciary.

Ringold, Dena, Mitchell A. Orenstein, and Erika Wilkens. 2005.Roma in an Expanding Europe: Breaking the Poverty Cycle. Wash-ington, DC: World Bank.

Robinson, James. 2003. “Politician-Proof Policy?” Washington, DC:World Bank, Working Paper 26945. Available on line athttp://www.people.fas.harvard.edu/~jrobins/researchpapers/unpublishedpapers/jr_WDR2004.pdf.

Rock, Michael T. 2003. “The Politics of Development Policy andDevelopment Policy Reform in New Order Indonesia.” AnnArbor, MI: University of Michigan, William Davidson InstituteWorking Paper 632.

Rodrik, Dani. 1998. “Why Do More Open Economies Have BiggerGovernments?” Journal of Political Economy 106(5):997–1032.

————. 1999a. “Where Did All the Growth Go? External Shocks,Social Conflict, and Growth Collapses.” Journal of EconomicGrowth 4(4):385–412.

————. 1999b. “Democracies Pay Higher Wages.” QuarterlyJournal of Economics 114(3):707–38.

References 267

Rodrik, Dani, Arvind Subramanian, and Francesco Trebbi. 2002.“Institutions Rule: The Primacy of Institutions over Geographyand Integration in Economic Development.” Cambridge, MA:National Bureau of Economic Research Working Paper Series 9305.

Roemer, John E. 1998. Equality of Opportunity. Cambridge, MA:Harvard University Press.

Roland, Gérard, and Thierry Verdier. 2000. “Law Enforcement andTransition.” London, UK: Centre for Economic Policy ResearchDiscussion Paper Series 2501.

Ronchi, Loraine. 2001. “The Impact of Fair Trade on Producers andTheir Organizations: A Case Study with Coocafe in Costa Rica.”Brighton: University of Sussex, Poverty Research Unit WorkingPaper 11. Available on line at http://www.sussex.ac.uk/Units/PRU/wps/wp11.pdf.

————. 2002. Monitoring Impact of Fair Trade Initiatives: A CaseStudy of Kuapa Kokoo and the Day Chocolate Company. London:Twinsight. TWIN. Available on line at http://www.divinechocolate.com/ImpactAssessmentLeaflet.pdf.

Rosenzweig, Mark R., and Hans P. Binswansger. 1993. “Wealth,Weather Risk and the Composition and Profitability of Agricul-tural Investments.” Economic Journal 103(416):56–78.

Rosenzweig, Mark R., and Kenneth I. Wolpin. 1993. “Credit MarketConstraints, Consumption Smoothing, and the Accumulation ofDurable Production Assets in Low-income Countries: Invest-ments in Bullocks in India.” Journal of Political Economy101(2):223–44.

Ross, Michael. 2005. “Mineral Wealth and Equitable Development.”Background paper for the WDR 2006.

Rothstein, Jesse. 2005. “Does Competition Among Public SchoolsBenefit Students and Taxpayers? A Comment on Hoxby (2000).”Cambridge, MA: National Bureau of Economic Research Work-ing Paper Series 11215.

Rudnik, Rebecca S., and Richard K. Gordon. 1996. “Taxation ofWealth.” In Victor Thuronyi, (ed.), Tax Law Design and Drafting.Washington, DC: International Monetary Fund.

Russel, Peter H., and David M. O’Brien, eds. 2001. Judicial Indepen-dence in the Age of Democracy: Critical Perspectives from Aroundthe World. Charlottesville, VA: University of Virginia Press.

Rutter, Michael, Henri Giller, and Ann Hagell. 2000. Antisocial Behav-ior by Young People. Cambridge, UK: Cambridge University Press.

Sala-i-Martin, Xavier. 2002. “Unhealthy People are Poor People . . .And vice-versa.” Paper presented at the European Conference onHealth Economics of the International Health Economics Asso-ciation. July 7. Paris.

Samuels, David, and Richard Snyder. 2001. “Devaluing the Vote inLatin America.” Journal of Democracy 12(1):146–59.

Sánchez-Páramo, Carolina, and Norbert Schady. 2003. “Off andRunning? Technology, Trade, and the Rising Demand for SkilledWorkers in Latin America.” Washington, DC: World Bank PolicyResearch Working Paper Series 3015.

Savedoff, William D. 2004. “Is there a Case for Social Insurance?”Health Policy and Planning 19(3):183–84.

Schady, Norbert. 2005. “Changes in the Global Distribution of LifeExpectancy and Education.” World Bank. Washington, DC.Processed.

Schieber, George. 2005. “Financing Health Systems in the 21st Cen-tury.” World Bank. Washington, DC. Processed.

Schiff, Maurice. 2005. “Brain Gain: Claims About its Size andImpact on Welfare and Growth are (Greatly) Exaggerated.”Bonn, Germany: IZA Discussion Paper 1599.

Schiff, Maurice, and Alberto Valdés. 1998. “Agriculture and theMacroeconomy.” Washington, DC: World Bank Policy ResearchWorking Paper Series 1967.

Schultz, Paul T. 1998. “Inequality in the Distribution of PersonalIncome in the World: How is it Changing and Why?” Journal ofPopulation Economics 11(3):307–44.

————. 2004. “School Subsidies for the Poor: Evaluating theMexican Progresa Poverty Program.” Journal of DevelopmentEconomics 74(1):199–250.

Schwartz, Anita. 2003. Old Age Security and Social Pensions. Wash-ington, DC: World Bank.

Schwarzer, H., and A. C. Querino. 2002. “Non-contributory Pen-sions in Brazil: The Impact on Poverty Reduction.” Geneva: ILO,Social Security Policy and Development Branch, ESS Paper 11.

Scott, James C. 1976. The Moral Economy of the Peasant: Rebellion andSubsistence in Southeast Asia. New Haven, CT: Yale University Press.

————. 1986. “Everyday Forms of Peasant Resistance.” In JamesC. Scott and Benedict J. Tria Kerkvilet, (ed.), Everyday Forms ofPeasant Resistance in Southeast Asia. London: Frank Cass. & Co.

Scott-McDonald, Kerida. 2004. “Elements of Quality in Home Vis-iting Programs: Three Jamaican Models.” In Mary Eming Young,(ed.), From Early Child Development to Human Development:Investing in our Chidren’s Future. Washington, DC: World Bank.

Sekhri, Neelam, and William Savedoff. 2005. “Private Health Insur-ance: Implications for Developing Countries.” Bulletin of theWorld Health Organization 83(2):127–134.

Sell, Susan K. 2003. Private Power, Public Law: The Globalization ofIntellectual Property Rights. Cambridge, UK: Cambridge Univer-sity Press.

Sen, Abhijit, and Himanshu. 2004. “Poverty and Inequality in India:Getting Closer to the Truth.”Centre for Economic Studies and Plan-ning, Jawaharlal Nehru University. New Delhi. Available on line athttp://www.networkideas.org/featart/may2004/Poverty_WC.pdf.Processed.

Sen, Amartya. 1985. Commodities and Capabilities. Amsterdam:North-Holland.

————. 2000. “Social Justice and the Distribution of Income.” InAnthony B. Atkinson and François Bourguignon, (ed.), Hand-book of Income Distribution. Amsterdam: North-Holland.

————. 2001. “10 Theses on Globalization.” Global Viewpoint,July 12.Available on line at http://www.digitalnpq.org/global_services/global%20viewpoint/07-12-01.html.

————. 2004. “Disability and Justice.” Disability and InclusiveDevelopment Conference, keynote speech, World Bank. Wash-ington, DC. Processed.

Shaban, Radwan. 1987. “Testing between Competing Models ofSharecropping.” Journal of Political Economy 95(5):893–920.

Shaw, M. 2001. Investing in Youth: International Approaches to Pre-venting Crime and Victimization. Montreal: International Centerfor the Prevention of Crime.


Shaw, Margaret. 2004. Police, Schools and Crime Prevention: A Pre-liminary Review of Current Practices. Montreal: InternationalCenter for the Prevention of Crime. Available on line athttp://www.crime-prevention-intl.org/publications/pub_110_1.pdf.

Shepherd, Andrew, Emmanuel Gyimah-Boadi, with Sulley Gariba,Sophie Plagerson, and Abdul Wahab Musa. 2005. “Bridging theNorth-south Divide in Ghana.”Background paper for the WDR 2006.

Shih, James C. 1992. Chinese Rural Society in Transition: A CaseStudy of the Lake Tai Area. Berkeley, CA: University of California,Inst.of East Asian Studies.

Shore, Rima. 1997. Rethinking the Brain: New Insights into EarlyDevelopment. New York: Family and Work Institute.

Shorrocks, A. F. 1980. “The Class of Additively DecomposableInequality Measures.” Econometrica 48(3):613–25.

Siegel, Jordan. 2003. “Is Political Connectedness a ParamountInvestment after Liberalization? The Successful Leveraging ofContingent Social Capital and the Formation of Cross BorderStrategic Alliances Involving Korean Firms and their GlobalParners (1987–2000).” Cambridge, MA: Harvard NOM WorkingPaper No. 03-45 03-45.

Slack, Paul. 1990. The English Poor Law, 1531–1782. London: TheMacmillan Press.

Sloth-Nielsen, Julia, and Jackie Gallinetti. 2004. Child Justice inAfrica: A Guide to Good Practice, Community Law Center.Bellville, South Africa: University of the Western Cape Commu-nity Law Centre.

Smith [1776], Adam. 1937. An Inquiry into the Nature and Causes ofthe Wealth of Nations. New York: The Modern Library.

Smith, J. R., J. Brooks-Gunn, and P. K. Klebanov. 1997. “Conse-quences of Living in Poverty for Young Children’s Cognitive andVerbal Ability and Early School Achievement.” In J. Duncan andJ. Brooks-Gunn, (ed.), Consequences of Growing up Poor. NewYork: Russell Sage Foundation.

Smith, Peter H. 1978. “The Breakdown of Democracy in Argentina:1916-1930.” In Juan J. Linz and Alfred Stepan, (ed.), The Break-down of Democratic Regimes: Latin America. Baltimore, MD:Johns Hopkins University Press.

Socialist Republic of Vietnam, and World Bank. 2005. Vietnam:Managing Public Expenditure for Poverty Reduction and Growth;Public Expenditure Review and Integrated Fiduciary Assessment2004 (Vols. I and II). Washington, DC: World Bank.

Sokoloff, Kenneth L., and B. Zorina Khan. 1990. “The Democrati-zation of Invention During Early Industrialization: Evidencefrom the United States, 1790–1846.” Journal of Economic History50(2):363–78.

Solar, Peter. 1995. “Poor Relief and English Economic Developmentbefore the Industrial Revolution.” Economic History Review48(2):1–22.

————. 1997. “Poor Relief and English Economic Development:A Renewed Plea for Comparative History.” Economic HistoryReview 50(2):369–74.

Solon, Gary. 1999. “Intergenerational Mobility in the Labor Mar-ket.” In Orley Ashenfelter and David Card, (ed.), Handbook ofLabor Economics. Amsterdam: North-Holland.

————. 2002. “Cross-country Differences in IntergenerationalEarnings Mobility.” Journal of Economic Perspectives 16(3):59–66.

South Africa Competition Commission. 2003. Competition CommissionMedia Release 33. Johannesburg: South Africa Competition Com-mission. Available on line at http://www.compcom.co.za/resources/Media%20Releases/MediaReleases%202003/Jul/Med%20Rel%2034%200f16%20Dec%202003.asp.

Sowell, Thomas. 2004. Affirmative Action Around the World: AnEmpirical Study. New Haven: Yale University Press.

Stark, Oded, and David E. Bloom. 1985. “The New Economics ofLabor Migration.” American Economic Review 75(2):173–78.

Steele, Claude M. 1999. “Thin Ice: ‘Stereotype Threat’ and BlackCollege Students.” Atlantic Monthly 284(2):44–54.

Steele, Claude M., and J. Aronson. 1995. “Stereotype Threat and theIntellectual Test Performance of African Americans.” Journal ofPersonality and Social Psychology 69(5):797–811.

Stern, Joseph J. 2003. “The Rise and Fall of the Indonesian Econ-omy.” Cambridge, MA: Harvard University, KSG Working PaperSeries RWP03-030.

Stern, Nicholas. 2002. A Strategy for Development. Washington, DC:World Bank.

Stern, Nicholas, Jean-Jacques Dethier, and F. Halsey Rogers. 2005.Growth and Empowerment: Making Development Happen. Cam-bridge, MA: MIT Press.

Stewart, Frances. 2001. “Horizontal Inequalities: A NeglectedDimension of Development.” Oxford, UK: Queen ElizabethHouse, University of Oxford, Working Paper 1.

Stoltzfus, R. J., J. D. Kvalsvig, and H. M. Chwaya. 2001. “Effects ofIron Supplementation and Anthelmintic Treatment on Motorand Language Development of Preschool Children in Zanzibar:Double Blind, Placebo Controlled Study.” British Medical Journal323(7326):1–8.

Stone, Jeff, Christian I. Lynch, Mike Sjomeling, and John M. Darley.1999. “Stereotype Threat Effects on Black and White AthleticPerformance.” Journal of Personality and Social Psychology77(6):1213–27.

Stone, Jeff, Zachary W. Perry, and John M. Darley. 1997. “WhiteMen Can’t Jump: Evidence for the Perceptual Confirmation ofRacial Stereotypes Following a Basketball Game.” Basic andApplied Social Psychology 19(3):291–306.

Strauss, John, Germano Mwabu, and Katheleen Beegle. 2000.“Intrahousehold Allocations: A Review of Theories and Empiri-cal Evidence.” Journal of African Economies 9(0):83–143.

Subbarao, Kalanidhi. 2003. “Systemic Shocks and Social Protection:Role and Efectivenss of Public Works Programs.” Washington,DC: World Bank, Social Protection Discussion Paper 0302.

Subbarao, Kalanidhi, and Diane Coury. 2004. Reaching Out toAfrica’s Orphans: A Framework for Public Action. Washington,DC: World Bank.

Sudjana, Brasukra G., and Satish Mishra. 2004. “Growth andInequality in Indonedia Today: Implications for Future Devel-opment Policy.” Jakarta: UNSFIR Discussion Paper Series04/05.

Swaminathan, Madhura. 1991. “Segmentation, Collateral Under-valuation, and the Rate of Interest in Agrarian Credit Markets:

References 269

Some Evidence from Two Villages in South India.” CambridgeJournal of Economics 15(2):161–78.

Szreter, Simon. 2005. “Public Health and Security in an Age ofGlobalizing Economic Growth: The Awkward Lessons of His-tory.” In Simon Szreter, (ed.), Health and Wealth: Studies in Policyand History. Rochester, NY: University of Rochester Press.

Takeoshi, Yosoburo. 1967. The Economic Aspects of the History of theCivilization of Japan. London: Taylor & Francis.

Tarrow, Sidney. 1998. Power in Movement: Social Movement andContentious Politics. New York: Cambridge University Press.

Tawney, R. H. 1941. “The Rise of Gentry: 1558–1640.” EconomicHistory Review 11(1):1–38.

Taylor, J. Edward. 1992. “Remittances and Inequality Reconsidered:Direct, Indirect, and Intertemporal Effects.” Journal of PolicyModeling 14(2):187–208.

Taylor, J. Edward, and T. J. Wyatt. 1996. “The Shadow Value ofMigrant Remittances, Income and Inequality in a Household-farm Economy.” Journal of Development Studies 32(6):899–912.

Temple, Jonathan. 2001. “Growing into Trouble: Indonesia After1966.” London: C.E.P.R. Discussion Papers 2932.

Terry, Deborah J., Craig J. Carey, and Victor J. Callan. 2001.“Employee Adjustment to an Organizational Merger: An Inter-group Perspective.” Personality and Social Psychology Bulletin27(3):269–90.

Thailand’s Ministry of Public Health, and World Bank. 2005.“Expanding Access to ART in Thailand: Achieving TreatmentBenefits while Promoting Effective Prevention.” Thailand’s Min-istry of Health and World Bank. Bangkok and Washington, DC.Processed.

Thomas, Duncan. 1990. “Intra-household Resource Allocation: AnInfluential Approach.” Journal of Human Resources 29(4):635–64.

Thomas, Duncan, Elizabeth Frankenberg, Jed Friedman, Jean-Pierre Habitch, Mohamed Hakimi, Nathan Jones, Gretel Pelto,Bondan Sikoki, Teresa Seeman, James P. Smith, Cecep Sumantri,Wayan Suriastini, and Siswanto Wilopo. 2005. “Causal Effect ofHealth on Labor Market Outcomes: Experimental Evidence.”University of California. Los Angeles, CA. Available on line athttp://csde.washington.edu/downloads/thomas05.05.20.pdf.Processed.

Thomas, Duncan, Elizabeth Frankenburg, Katheleen Beegle, andGraciela Teruel. 1999. “Household Budgets, Household Compo-sition and the Crisis in Indonesia: Evidence from LongitudinalHousehold Survey Data.” Paper presented at the PopulationAssociation of America Meetings. March 25. New York.

Thomas, Vinod. 1987. “Differences in Income and Poverty withinBrazil.” World Development 15(2):263–73.

Thomas, Vinod, Yan Wang, and Xibo Fan. 2002. “A New Dataset on Inequality in Education: Gini and Theil Indices of Schoolingfor 140 Countries, 1960–2000.” World Bank. Washington, DC.Processed.

Thompson, Edward P. 1963. The Making of the English WorkingClass. New York: Pantheon Books.

Thorbecke, Erik. 2005. “Economic Development, Income Distribu-tion and Ethics,” Ithaca, Cornell University. Processed.

Tien Dung, Luu. 2003. Judicial Independence in TransitionalCountries. New York: United Nations Development Program.Available on line at http://www.undp.org/oslocentre/docsjuly03/DungTienLuu-v2.pdf.

Tilly, Charles. 1995. Popular Contention in Britain. Cambridge, MA:Harvard University Press.

————. 1998. Durable Inequality. Berkeley, CA: University ofCalifornia Press.

Timmer, Peter. 2005. “Operationalizing Pro-poor Growth: Indo-nesia Country Study.” Center for Global Development.Washington, DC. Available on line at http://siteresources.worldbank.org/INTPGI/Resources/342674-1115051237044/oppgindonesiaMay2005.pdf. Processed.

Topalova, Petia. 2004. “Factor Immobility and Regional Impacts ofTrade Liberalization: Evidence on Poverty and Inequality fromIndia.” Ph.D thesis. Massachusetts Institute of Technology.

Toulmin, Camilla, and Julian Quan. 2000. Evolving Land Rights,Policy and Tenure in Africa. London, UK: International Institutefor Environment and Development and Natural Resources Insti-tute.

Townsend, Robert M. 1994. “Risk and Insurance in Village India.”Econometrica 62(3):539–91.

————. 1995. “Financial Systems in Northern Thai Villages.”Quarterly Journal of Economics 110(4):1011–46.

Tremblay, Richard E., and Wendy M. Craig. 1995. “DevelopmentCrime Prevention.” In Michael Tonry and David P. Farrington,(ed.), Building a Safer Society: Strategic Approaches to Crime Pre-vention. Chicago, IL: Chicago University Press.

Udry, Christopher. 1996. “Gender, Agricultural Production and theTheory of Household.” Journal of Political Economy 104(5):1010–46.

Ugaz, Cecilia, and Catherine Waddams Price, eds. 2003. Utility Pri-vatization and Regulation: A Fair Deal for Consumers?Northampton, MA: Edward Elgar.

UNAIDS. 2002. Report on the Global HIV/AIDS Epidemic 2002.Geneva: UNAIDS.

UNCTAD. 2004. World Investment Report 2004. New York: UnitedNations Conference on Trade and Development. Available on line athttp://www.unctad.org/Templates/WebFlyer.asp?intItemID=3235&lang=1.

UNESCO. 2005. EFA Global Monitoring Report. New York: UNESCO.Available on line at http://portal.unesco.org/education/.

UNICEF. 2003. UNICEF Programs: Support for Families and Com-munities. Paris: UNICEF.

United Kingdom’s Commission for Africa. 2005. Our CommonInterest. London: United Kingdom’s Commission for Africa.

United Nations. 2002. Population Ageing Report. New York: UnitedNations.

————. 2003. Human Security Now: Protecting and Empowering Peo-ple. New York: United Nations, Commission on Human Security.

United Nations Development Programme. 1995. Human Develop-ment Report 1995. New York: Oxford University Press.

————. 2003. Human Development Report 2003: MillenniumDevelopment Goals: A Compact among Nations to End HumanPoverty. New York: Oxford University Press.


United Nations International Organization for Migration. 2000.World Migration Report 2000. Geneva: International Organiza-tion for Migration.

United Nations Research Institute for Social Development(UNRISD). 2005. Policy Report on Gender and Development: 10Years after Beijing. New York: United Nations.

USAID, UNAIDS, and UNICEF. 2004. Children on the Brink 2004: AJoint Report on Orphan Estimates and Program Strategies. Paris:UNICEF. Available on line at http://www.unicef.org/publications/index_22212.html.

Vakis, Renos. 2003. “Livelihoods, Labor Markets and RuralPoverty.” Washington, DC: World Bank, Guatemala Poverty Ass-esment (GUAPA) Program Technical Paper 1.

Valdés-Prieto, Salvador. 1992. “Financial Liberalization and theCapital Account: Chile 1974–1984.” In Gerard Caprio, IzacAtiyas, and James A. Hanson, (ed.), Financial Reform: Theory andExperience. Cambridge, MA: Cambridge University Press.

Van de Walle, Dominique, and Dorothyjean Cratty. 2004. “Is theEmerging Non-Market Economy the Route out of Poverty inVietnam?” Economics of Transition 12(2):237–75.

van den Brink, Rogier, Mike de Klerk, and Hans Binswanger. 1996.“Rural Livelihoods, Fiscal Costs and Financing Options: A FirstAttempt at Quantifying the Implications of Redistributive LandReform.” In Johan van Zyl, Johann Kirsten, and Hans Bin-swanger, (ed.), Agricultural Land Reform in South Africa: Policies,Markets and Mechanisms. Cape Town: Oxford University Press.

van der Berg, Servaas, and Caryn Brendenkamp. 2002. “DevisingSocial Security Interventions for Maximum Poverty Impact.”Cape Town: CSSR Working Paper 13.

Van Niekerk, Gardiol. 2001. “State Initiatives to Incorporate Non-stateLaws into the Official Legal Order: A Denial of Legal Pluralism?”Comparative and International Law of Southern Africa 34(3):349–61.

Velasco, Andrés. 1988. “Liberalization, Crisis, Intervention: TheChilean Financial System.” Washington, DC: International Mon-etary Fund Working Paper Series 1988/66.

————. 1996. “Fixed Exchange Rates: Credibility, Flexibility andMultiplicity.” European Economic Review 40(3-5):1023–35.

Voitchovsky, Sarah. 2004. “The Effect of Inequality on Growth: AReview of the Recent Empirical Literature.” Oxford University.Oxford. Processed.

Wade, Robert. 1990. Governing the Market: Economic Theory andthe Role of Government in East Asian Industrialization. Princeton,NJ: Princeton University Press.

Walder, Andrew G., and Jean C. Oi. 1999. “Property Rights in theChinese Economy: Contours of the Process of Change.” InAndrew G. Walder and Jean C. Oi, (ed.), Property Rights and Eco-nomic Reform in China. Stanford, CA: Stanford University Press.

Waller, Irving, and Daniel Sansfaçon. 2000. Investing Wisely inCrime Prevention: International Experiences. Washington, DC:Office of Justice Programs, Bureau of Justice Assistance. Availableon line at http://www.ncjrs.org/pdffiles1/bja/182412.pdf.

Waller, Irwin, Brandon Welsh, and Daniel Sansfaçon. 1999. CrimePrevention Digest II: Comparative Analysis of Successful Commu-nity Safety. Quebec: International Center for the Prevention ofCrime.

Walmsley, Terrie L., and L. Alan Winters. 2003. “Relaxing theRestrictions on the Temporary Movement of Natural Persons: ASimulation Analysis.” London: CEPR Discussion Paper 3719.

Wan, Guanghua, Ming Lu, and Zhao Chen. 2004. “Globalizationand Regional Income Inequality.” Helsinki: United Nations Uni-versity, World Institute for Development Economics Research,Discussion Paper 2004/10.

Washington Post. 2004. “Cambodia: Pinning Hope on Fair LaborStandards.” Washington Post, November 17. Page: A19.

Widner, Jennifer A. 1993.“The Origins of Agricultural Policy in IvoryCoast: 1960-1986.” Journal of Development Studies 29(4):25–60.

Wilkinson, Richard. 1992. “Income Distribution and LifeExpectancy.” British Medical Journal 304(6820):165–68.

————. 2000. Mind the Gap: Hierarchies, Health and HumanEvolution. New Haven, CT: Yale University Press.

Willanakuy, Hatun. 2004. Versión Abreviada del Informe Final de laComisión de la Verdad y Reconciliación. Lima, Peru: Comisión dela Verdad y Reconciliación.

Williamson, Jeffrey G. 1985. Did British Capitalist Breed Inequality?Boston, MA: Allen and Unwin.

Winters, L. Alan, Neil McCulloch, and Andrew McKay. 2004.“TradeLiberalization and Poverty: The Evidence so Far.” Journal of Eco-nomic Literature 42(1):72–115.

Witte, John F. 2000. The Market Approach to Education: An Analysisof America’s First Voucher Program. Princeton, NJ: Princeton Uni-versity Press.

Wiwattanakantang, Yupana, Raja Kali, and Chutatong Charu-milind. Forthcoming. “Connected Lending: Thailand Before theFinancial Crisis.” Journal of Business.

Wodon, Quentin. 2005. “Access to Basic Facilities in Africa.” WorldBank. Washington, DC. Processed.

Wodon, Quentin, Mohamed Ihsan Ajwad, and Corinne Siaens.2005. “Targeting Utility Subsidies: Lifeline or Means-Testing?”World Bank. Washington, DC. Processed.

Women in Informal Employment Globalizing and Organizing.2005. The Importance of Organizing Informal and CasualizedWomen Workers: Findings of HomeNet Thailand’s Recent Studies.Cambridge, MA: Women in Informal Employment Globalizingand Organizing (WIEGO). Available on line at http://www.wiego.org/textonly/news.shtml.

Wood, Adrian. 1997.“Openness and Wage Inequality in DevelopingCountries: The Latin American Challenge to East Asian Conven-tional Wisdom.” World Bank Economic Review 11(1):33–57.

World Bank. 1980. World Development Report 1980. New York:Oxford University Press.

————. 1990. World Development Report 1990: Poverty. NewYork: Oxford University Press.

————. 1993. World Bank Policy Research Report 1993. The EastAsian Miracle: Economic Growth and Public Policy. New York:Oxford University Press.

————. 1994. World Development Report 1994: Infrastructure forDevelopment. New York: Oxford University Press.

————. 1997a. Hashemite Kingdom of Jordan: Health Sector Study(World Bank Country Study). Washington, DC: World Bank.

References 271

————. 1997b. Sharing Rising Incomes: Disparities in China.Washington, DC: World Bank.

————. 1997c. World Development Report 1997: The State in aChanging World. New York: Oxford University Press.

————. 2000a. Can Africa Claim the 21st Century? Washington,DC: World Bank.

————. 2000b. East Asia: Recovery and Beyond. Washington, DC:World Bank.

————. 2000c. Making Transition Work for Everyone: Poverty andInequality in Europe and Central Asia. Washington, DC: World Bank.

————. 2001a. Colombia: Water Sector Reform Assistance Project.Washington, DC: Project Appraisal Document Report No: 21868.

————. 2001b. Coverage: The Scope of Protection in RetirementIncomes. Washington, DC: World Bank Pension Reform PrimeNote, Social Protection Unit.

————. 2001c. Growing Health: A Review of Vietnam’s HealthSector. Washington, DC: World Bank.

————. 2001d. Mexico—Land Policy: A Decade After the EjidoReform. Washington, DC: World Bank.

————. 2001e. Social Protection Sector Strategy Paper—FromSafety Nets to Spring Board. Washington, DC: World Bank.

————. 2001f. Thailand Social Monitor: Poverty and Public Pol-icy. Washington, DC: World Bank.

————. 2001g. World Bank Policy Research Report 2001: Engen-dering Development Through Gender Equality In Rights, ResourcesAnd Voice. New York: Oxford University Press.

————. 2001h. World Development Report 2000/01: AttackingPoverty. New York: Oxford University Press.

————. 2002a. Bulgaria: Poverty Assessment Report 24516-BUL.Washington, DC. World Bank.

————. 2002b. World Development Report 2002: Building Institu-tions for Markets. New York: Oxford University Press.

————. 2003a. A Resource for Municipalities: Community BasedCrime and Violence Prevention in Urban Latin America. Washing-ton, DC: World Bank, Department of Finance, Private Sector andInfrastructure, Latin American Region.

————. 2003b. Global Economic Prospects. Washington, DC:World Bank.

————. 2003c. From Goats to Coats: Institutional Reform in Mon-golia’s Cashmere Sector. Washington, DC: World Bank.

————. 2003d. Kyrgyz Republic: Enhancing Pro-poor GrowthReport 24638-KG. Washington, DC: World Bank.

————. 2003e. Romania: Poverty Assessment Report 26169-RO.Washington, DC: World Bank.

————. 2003f. The Contribution of Social Protection to the Mil-lennium Development Goals. Washington, DC: World Bank,Social Protection Unit.

————. 2003g. The Impact of Legal Aid: Ecuador. Washington,DC: World Bank, Legal Vice Presidency.

————. 2003h. World Bank Policy Research Report 2003. Breakingthe Conflict Trap: Civil War and Development Policy. New York:Oxford University Press.

————. 2003i. World Bank Policy Research Report 2003. LandPolicies for Growth and Poverty Reduction. New York: OxfordUniversity Press.

————. 2003j. World Development Report 2004: Making ServicesWork for Poor People. New York: Oxford University Press.

————. 2004a. A Fair Share for Women: Cambodia GenderAssessment. Phnom Penh: UNIFEM, The World Bank, ADB,UNDP and DFID/UK in cooperation with the Ministry ofWomen’s and Veteran’s Affair.

————. 2004b. Doing Business in 2004: Understanding Regulation.Washington, DC: World Bank, IFC and Oxford University Press.

————. 2004c. “Gender Assessment of the Honduras JudicialSector.” World Bank. Washington, DC. Processed.

————. 2004d. Implementation Completition Report: ImprovingLearning in Primary School Project (Report No. 27345). Washing-ton, DC: World Bank.

————. 2004e. Kingdom of Morocco Poverty Report: Strengthen-ing Policy by Identifying the Geographic Dimensions of Poverty.Washington, DC: World Bank.

————. 2004f. “Private Participation in Infrastructure ProjectDatabase”. Washington, DC, World Bank. Available on line athttp://ppi.worldbank.org/.

————. 2004g. Project Appraisal Document, Justice ServicesImprovement Project, Annexes 11 and 12. Washington, DC: WorldBank.

————. 2004h. Reaching the Poor with Effective Health, Nutri-tion, and Population Services. Washington, DC: World Bank.

————. 2004i. “Rural Poverty in Mexico.” World Bank. Washing-ton, DC. Processed.

————. 2004j. Sierra Leone: Legal and Judicial Sector Assessment.Washington, DC: World Bank, Legal Vice Presidency.

————. 2004k. The Millennium Development Goals for Health:Rising to the Challenges. Washington, DC: World Bank.

————. 2004l. World Bank Policy Research Report 2004. Reform-ing Infrastructure: Privatization, Regulation, and Competition.New York: Oxford University Press.

————. 2005a. “Growth, Poverty, and Inequality in Europe andCentral Asia. Past, Present, and Future.” World Bank. Washing-ton, DC. Processed.

————. 2005b. Pro-poor Growth in the 1990s: Lessons and Insightfrom 14 Countries. Washington, DC: World Bank.

————. 2005c. Pro-Poor Growth: Country Experiences in the1990s. Washington, DC: World Bank, Poverty Reduction Group.

————. 2005d. “Cambodia—Quality Basic Education for All.”Washington, DC: World Bank.

————. 2005e. Doing Business in 2005: Removing Obstacles toGrowth. Washington, DC: Oxford University Press for the Inter-national Finance Corporation (World Bank).

————. 2005f. Global Monitoring Report. Millennium Develop-ment Goals: From Consensus to Momentum. Washington, DC:World Bank.

————. 2005g. World Development Indicators. Washington, DC:World Bank.

————. 2005h. World Development Report 2005: A Better Invest-ment Climate for Everyone. New York: Oxford University Press.

World Commission on the Social Dimensions of Globalization.2004. A Fair Globalization: Creating Opportunities for All.Geneva. ILO.


World Health Organization. 2004. World Report on Knowledge forBetter Health. Geneva: World Health Organization. Available online at www.who.int/rpc/meetings/wr2004/en/.

Wößmann, Ludger. 2000. “Schooling Resources, Educational Insti-tutions, and Student Performance: The International Evidence.”Kiel Institute of Work Economics: Kiel Working Paper 983.

————. 2004. “How Equal are Educational Opportunities? Fam-ily Background and Student Achievement in the Europe and theUS.” Munich, Germany: Center for Economic Studies and IfoInstitute for Economic Research 1162.

Wrigley, E. A. 1998. Continuity, Chance and Change: The Characterof the Industrial Revolution in England. Cambridge, UK: Cam-bridge University Press.

Yang, Dean. 2004. “International Migration, Human Capital, andEntrepreneurship: Evidence from Philippine Migrants’ ExchangeRate Shocks.” University of Michigan: Ford School of Public Pol-icy Working Paper Series 02-011. Available on line at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=546483.

Young, Mary Eming. 2002. “Early Child Development: A SteppingStone to Success in School and Lifelong Learning.” World Bank.Washington, DC. Processed.

Zaninka, Penninah. 2003.“Uganda.” In J. Nelson and L. Hossack, (ed.),Indigenous People and Protected Areas in Africa: From Principles toPractice. Moreton-in-Marsh, UK: Forest People Programme.

Zeldes, Stephen. 1989. “Consumption and Liquidity Constraints: AnEmpirical Investigation.” Journal of Political Economy 97(2):305–46.

References 273

Measuring equity 277Introduction

Table A1. Poverty

Table A2. Income/consumption inequality measures

Table A3. Health

Table A4. Education

Technical notes

Selected world development indicators 289Introduction

Classification of economies by region and income

Table 1. Key indicators of development

Table 2. Millennium Development Goals: eradicating poverty andimproving lives

Table 3. Economic activity

Table 4. Trade, aid, and finance

Table 5. Key indicators for other economies

Technical notes

Selected Indicators

277

This report has concerned itself with equity, a concept whichwe have defined to mean equality of opportunities and theavoidance of absolute deprivation. As we made clear in thereport, the focus is mostly on inequality of opportunity, andmuch less on overall inequality in a one-dimensional space,such as income or education. We might be quite sanguineabout certain types of inequality across people if, for exam-ple, their outcomes varied for reasons that had to do mainlywith their own efforts. But we are concerned with systematicdifferences in opportunities between individuals and groupsof people due to different “circumstances” not under theircontrol, i.e. when these groups are distinguishable from oneanother only in characteristics that in some sense can beargued to be “morally irrelevant” to their opportunities andoutcomes in life.

To give the audience a sense of inequality of opportuni-ties, we chose to present data on income/consumption, infant

mortality, and years of education in a slightly different formatthan what we are normally used to seeing for as many coun-tries in the world as possible. Table A1 presents poverty ratesusing national poverty lines as well as those using the interna-tional $1 and $2 per day poverty lines; information for thenational poverty line is also presented for urban and ruralbreakdowns of the population. Table A2 presents alternativeinequality measures to the commonly used Gini Index, fol-lowed by evidence on inequality of land. In Table A3, we pre-sent infant mortality rates for more than 50 countries by thesex of the infant, the education level of the mother, the loca-tion (urban or rural) of the family, and the ranking of thehousehold by an asset index. Finally, in Table A4, we presenteducational attainment by location and gender, but also pre-sent measures of inequality in years of schooling, along withthe share of this inequality that is attributable to gender andlocation.

Measuring equity


Table A1. Poverty National poverty line International poverty line

Population below Population below Population Poverty Population Povertythe poverty line the poverty line below gap at below gap at

Survey Rural Urban National Survey Rural Urban National Survey $1 a day $1 a day $2 a day $2 a dayyear % % % year % % % year % % % %

Albania 2002 29.6 19.8 25.4 .. .. .. .. 2002 a <2 <0.5 11.8 2.0Algeria 1995 30.3 14.7 22.6 1998 16.6 7.3 12.2 1995 a <2 <0.5 15.1 3.8Angola .. .. .. .. .. .. .. .. .. .. .. .. ..Argentina 1995 .. 28.4 .. 1998 .. 29.9 .. 2001 b 3.3 0.5 14.3 4.7Armenia 1998–99 50.8 58.3 55.1 2001 48.7 51.9 50.9 2003 a, c <2 <0.5 31.1 7.1Australia .. .. .. .. .. .. .. .. .. .. .. .. ..Austria .. .. .. .. .. .. .. .. .. .. .. .. ..Azerbaijan 1995 .. .. 68.1 2001 42.0 55.0 49.0 2001 a 3.7 0.6 33.4 9.1Bangladesh 1995–96 55.2 29.4 51.0 2000 53.0 36.6 49.8 2000 a 36.0 8.1 82.8 36.3Belarus 2000 .. .. 41.9 .. .. .. .. 2000 a <2 <0.5 <2 <0.5Belgium .. .. .. .. .. .. .. .. .. .. .. .. ..Benin 1995 25.2 28.5 26.5 1999 33.0 23.3 29.0 .. .. .. .. ..Bolivia 1997 77.3 53.8 63.2 1999 81.7 50.6 62.7 1999 a 14.4 5.4 34.3 14.9Bosnia & Herzegovina 2001–02 19.9 13.8 19.5 .. .. .. .. .. .. .. .. ..Brazil 1996 54.0 15.4 23.9 1998 51.4 14.7 22.0 2001 b 8.2 2.1 22.4 8.8Bulgaria 1997 .. .. 36.0 2001 .. .. 12.8 2003 a, c <2 <0.5 6.1 1.5Burkina Faso 1994 51.0 10.4 44.5 1998 51.0 16.5 45.3 1998 a 44.9 14.4 81.0 40.6Burundi 1990 36.0 43.0 36.4 .. .. .. .. 1998 a 54.6 22.7 87.6 48.9Cambodia 1997 40.1 21.1 36.1 1999 40.1 13.9 35.9 1997 a 34.1 9.7 77.7 34.5Cameroon 1996 59.6 41.4 53.3 2001 49.9 22.1 40.2 2001 a 17.1 4.1 50.6 19.3Canada .. .. .. .. .. .. .. .. .. .. .. .. ..Central African Rep. .. .. .. .. .. .. .. .. 1993 a 66.6 38.1 84.0 58.4Chad 1995–96 67.0 63.0 64.0 .. .. .. .. .. .. .. .. ..Chile 1996 .. .. 19.9 1998 .. .. 17.0 2000 b <2 <0.5 9.6 2.5China 1996 7.9 <2 6.0 1998 4.6 <2 4.6 2001 a 16.6 3.9 46.7 18.4

Hong Kong, China .. .. .. .. .. .. .. .. .. .. .. .. ..Colombia 1995 79.0 48.0 60.0 1999 79.0 55.0 64.0 1999 b 8.2 2.2 22.6 8.8Congo, Dem. Rep. .. .. .. .. .. .. .. .. .. .. .. .. ..Congo, Rep. .. .. .. .. .. .. .. .. .. .. .. .. ..Costa Rica 1992 25.5 19.2 22.0 .. .. .. .. 2000 b 2.0 0.7 9.5 3.0Côte d’Ivoire .. .. .. .. .. .. .. 2002 a, c 14.8 4.1 48.8 18.4Croatia .. .. .. .. .. .. .. .. 2001 a <2 <0.5 <2 <0.5Czech Rep. .. .. .. .. .. .. .. .. 1996 b <2 <0.5 <2 <0.5Denmark .. .. .. .. .. .. .. .. .. .. .. .. ..Dominican Rep. 1992 49.0 19.3 33.9 1998 42.1 20.5 28.6 1998 b <2 <0.5 <2 <0.5Ecuador 1994 47.0 25.0 35.0 .. .. .. .. 1998 b 17.7 7.1 40.8 17.7Egypt, Arab Rep. 1995–96 23.3 22.5 22.9 1999–00 .. .. 16.7 1999–2000 a 3.1 <0.5 43.9 11.3El Salvador 1992 55.7 43.1 48.3 .. .. .. .. 2000 b 31.1 14.1 58.0 29.7Eritrea 1993–94 .. .. 53.0 .. .. .. .. .. .. .. .. ..Ethiopia 1995–96 47.0 33.3 45.5 1999–00 45.0 37.0 44.2 1999–2000 a 23.0 4.8 77.8 29.6Finland .. .. .. .. .. .. .. .. .. .. .. .. ..France .. .. .. .. .. .. .. .. .. .. .. .. ..Georgia 1997 9.9 12.1 11.1 .. .. .. .. 2001 a 2.7 0.9 15.7 4.6Germany .. .. .. .. .. .. .. .. .. .. .. .. ..Ghana 1992 .. .. 50.0 1998–99 49.9 18.6 39.5 1998–99 a 44.8 17.3 78.5 40.8Greece .. .. .. .. .. .. .. .. .. .. .. .. ..Guatemala 1989 71.9 33.7 57.9 2000 74.5 27.1 56.2 2000 b 16.0 4.6 37.4 16.0Guinea 1994 .. .. 40.0 .. .. .. .. .. .. .. .. ..Haiti 1987 .. .. 65.0 1995 66.0 .. .. 2001 a, c 67.0 40.0 83.3 58.5Honduras 1992 46.0 56.0 50.0 1993 51.0 57.0 53.0 1999 b 20.7 7.5 44.0 20.2Hungary 1993 .. .. 14.5 1997 .. .. 17.3 2002 a <2 <0.5 <2 <0.5India 1993–94 37.3 32.4 36.0 1999–00 30.2 24.7 28.6 1999–2000 a 35.3 7.2 80.6 34.9Indonesia 1996 .. .. 15.7 1999 27.1 2002 a 7.5 0.9 52.4 15.7Iran, Islamic Rep. .. .. .. .. .. .. .. .. 1998 a <2 <0.5 7.3 1.5Ireland .. .. .. .. .. .. .. .. .. .. .. .. ..Israel .. .. .. .. .. .. .. .. .. .. .. .. ..Italy .. .. .. .. .. .. .. .. .. .. .. .. ..Jamaica 1995 37.0 18.7 27.5 2000 25.1 12.8 18.7 2000 a <2 <0.5 13.3 2.7Japan .. .. .. .. .. .. .. .. .. .. .. .. ..Jordan 1991 .. .. 15.0 1997 .. .. 11.7 2002 a, c <2 <0.5 6.5 1.4Kazakhstan 1996 39.0 30.0 34.6 .. .. .. .. 2003 a <2 <0.5 24.9 6.3Kenya 1994 47.0 29.0 40.0 1997 53.0 49.0 52.0 1997 a 22.8 5.9 58.3 23.9Korea, Rep. .. .. .. .. .. .. .. .. 1998 b <2 <0.5 <2 <0.5Kuwait .. .. .. .. .. .. .. .. .. .. .. ..Kyrgyz Rep. 2000 56.4 43.9 52.0 2001 51.0 41.2 47.6 2002 a <2 <0.5 24.7 5.8Lao PDR 1993 48.7 33.1 45.0 1997–98 41.0 26.9 38.6 1997–98 a 26.3 6.3 73.2 29.6

Measuring equity 279

Table A1. Poverty—continuedNational poverty line International poverty line

Population below Population below Population Poverty Population Povertythe poverty line the poverty line below gap at below gap at

Survey Rural Urban National Survey Rural Urban National Survey $1 a day $1 a day $2 a day $2 a dayyear % % % year % % % year % % % %

Latvia .. .. .. .. .. .. .. .. 1998 a <2 <0.5 11.5 2.6Lebanon .. .. .. .. .. .. .. .. .. .. .. .. ..Lithuania .. .. .. .. .. .. .. .. 2000 a <2 <0.5 6.9 1.5Macedonia, FYR .. .. .. .. .. .. .. .. 2003 a, c <2 <0.5 <2 <0.5Madagascar 1997 76.0 63.2 73.3 1999 76.7 52.1 71.3 2001 a 61.0 27.9 85.1 51.8Malawi 1990–91 .. .. 54.0 1997–98 66.5 54.9 65.3 1997–98 a 41.7 14.8 76.1 38.3Malaysia 1989 .. .. 15.5 .. .. .. .. 1997 b <2 <0.5 9.3 2.0Mali 1998 75.9 30.1 63.8 .. .. .. .. 1994 a 72.3 37.4 90.6 60.5Mauritania 1996 65.5 30.1 50.0 2000 61.2 25.4 46.3 2000 a 25.9 7.6 63.1 26.8Mexico 1988 .. .. 10.1 .. .. .. .. 2000 a 9.9 3.7 26.3 10.9Moldova 1997 26.7 19.3 23.3 .. .. .. .. 2001 a 21.8 5.7 64.1 25.2Mongolia 1995 33.1 38.5 36.3 1998 32.6 39.4 35.6 1998 a 27.0 8.1 74.9 30.6Morocco 1990–91 18.0 7.6 13.1 1998–99 27.2 12.0 19.0 1999 a <2 <0.5 14.3 3.1Mozambique 1996–97 71.3 62.0 69.4 .. .. .. .. 1996 a 37.9 12.0 78.4 36.8Namibia .. .. .. .. .. .. .. .. 1993 b 34.9 14.0 55.8 30.4Nepal 1995–96 44.0 23.0 42.0 .. .. .. .. 1995–96 a 39.1 11.0 80.9 37.6Netherlands .. .. .. .. .. .. .. .. .. .. .. .. ..New Zealand .. .. .. .. .. .. .. .. .. .. .. .. ..Nicaragua 1993 76.1 31.9 50.3 1998 68.5 30.5 47.9 2001 a 45.1 16.7 79.9 41.2Niger 1989–93 66.0 52.0 63.0 .. .. .. .. 1995 a 60.6 34.0 85.8 54.6Nigeria 1985 49.5 31.7 43.0 1992–93 36.4 30.4 34.1 2003 a, c 70.8 34.5 92.4 59.5Norway .. .. .. .. .. .. .. .. .. .. .. .. ..Oman .. .. .. .. .. .. .. .. .. .. .. .. ..Pakistan 1993 33.4 17.2 28.6 1998–99 35.9 24.2 32.6 2001 a, c 17.0 3.1 73.6 26.1Panama 1997 64.9 15.3 37.3 .. .. .. .. 2000 b 7.2 2.3 17.6 7.4Papua New Guinea 1996 41.3 16.1 37.5 .. .. .. .. .. .. .. .. ..Paraguay 1991 28.5 19.7 21.8 .. .. .. 2002 b 16.4 7.4 33.2 16.2Peru 1994 67.0 46.1 53.5 1997 64.7 40.4 49.0 2000 b 18.1 9.1 37.7 18.5Philippines 1994 53.1 28.0 40.6 1997 50.7 21.5 36.8 2000 a 15.5 3.0 47.5 17.8Poland 1993 .. .. 23.8 .. .. .. .. 2002 a, c <2 <0.5 <2 <0.5Portugal .. .. .. .. .. .. .. .. 1994 b <2 <0.5 <2 <0.5Romania 1994 27.9 20.4 21.5 .. .. .. .. 2002 a <2 0.5 14.0 3.4Russian Federation 1994 .. .. 30.9 .. .. .. .. 2002 a <2 <0.5 7.5 1.3Rwanda 1993 .. .. 51.2 1999–00 65.7 14.3 60.3 1999–2000 a 51.7 20.0 83.7 45.5Saudi Arabia .. .. .. .. .. .. .. .. .. .. .. .. ..Senegal 1992 40.4 23.7 33.4 .. .. .. .. 1995 a 22.3 5.7 63.0 25.2Serbia & Montenegro .. .. .. .. .. .. .. .. .. .. .. .. ..Sierra Leone 1989 .. .. 82.8 2003–04 79.0 56.4 70.2 1989 a 57.0 39.5 74.5 51.8Singapore .. .. .. .. .. .. .. .. .. .. .. .. ..Slovak Rep. .. .. .. .. .. .. .. 1996 b <2 <0.5 2.9 0.8Slovenia .. .. .. .. .. .. .. .. 1998 a <2 <0.5 <2 <0.5South Africa .. .. .. .. .. .. .. .. 2000 a 10.7 1.7 34.1 12.6Spain .. .. .. .. .. .. .. .. .. .. .. .. ..Sri Lanka 1990–91 22.0 15.0 20.0 1995–96 27.0 15.0 25.0 2002 a, c 5.6 <0.5 41.6 11.9Sudan .. .. .. .. .. .. .. .. .. .. .. .. ..Sweden .. .. .. .. .. .. .. .. .. .. .. .. ..Switzerland .. .. .. .. .. .. .. .. .. .. .. ..Syrian Arab Rep. .. .. .. .. .. .. .. .. .. .. .. .. ..Tajikistan .. .. .. .. .. .. .. .. 2003 a 7.4 1.3 42.8 13.0Tanzania 1991 40.8 31.2 38.6 2000–01 38.7 29.5 35.7 1991 a 48.5 24.4 72.5 43.3Thailand 1990 .. .. 18.0 1992 15.5 10.2 13.1 2000 a, c <2 <0.5 32.5 9.0Togo 1987–89 .. .. 32.3 .. .. .. .. .. .. .. .. ..Tunisia 1990 13.1 3.5 7.4 1995 13.9 3.6 7.6 2000 a <2 <0.5 6.6 1.3Turkey .. .. .. .. .. .. .. .. 2002 a, c 4.8 1.0 24.7 7.5Turkmenistan .. .. .. .. .. .. .. .. 1998 a 12.1 2.6 44.0 15.4Uganda 1993 .. .. 55.0 1997 .. .. 44.0 .. .. .. .. ..Ukraine 1995 .. .. 31.7 .. .. .. .. 1999 b 2.9 0.6 45.7 16.3United Kingdom .. .. .. .. .. .. .. .. .. .. .. .. ..United States .. .. .. .. .. .. .. .. .. .. .. .. ..Uruguay .. .. .. .. .. .. .. .. 2000 b <2 <0.5 3.9 0.8Uzbekistan 2000 30.5 22.5 27.5 .. .. .. .. 2000 a 17.3 4.3 71.7 25.2Venezuela, RB de 1989 .. .. 31.3 .. .. .. .. 2000 b, c 9.9 3.6 32.1 12.2Vietnam 1998 45.5 9.2 37.4 2002 35.6 6.6 28.9 .. .. .. .. ..West Bank & Gaza .. .. .. .. .. .. .. .. .. .. .. .. ..Yemen, Rep. 1998 45.0 30.8 41.8 .. .. .. .. 1998 a 15.7 4.5 45.2 15.0Zambia 1996 82.8 46.0 69.2 1998 83.1 56.0 72.9 1998 a 63.7 32.7 87.4 55.4Zimbabwe 1990–91 35.8 3.4 25.8 1995–96 48.0 7.9 34.9 1995–96 a 56.1 24.2 83.0 48.2

a = expenditure base; b = income base; c = preliminary data; .. denotes no data.


Table A2. Income/consumption inequality measuresIncome/consumption inequality Land inequality

Survey year y/c* Gini index GE (0) 90th/10th percentile ratio Survey year Gini index

Albania 2002 c 0.31 0.15 3.95 1998 0.84Algeria 1995 c 0.35 .. .. .. ..Argentina—urban 2001 y 0.51 0.49 13.71 1988 0.83Armenia 2003 c 0.26 0.11 3.17 .. ..Australia 1994 y 0.32 0.20 4.88 .. ..Austria 1997 y 0.28 0.14 3.58 1999/2000 0.59Azerbaijan 2001 c 0.36 0.22 4.62 .. ..Bangladesh 2000 c 0.31 0.16 3.85 1996 0.62Belarus 2000 c 0.30 .. .. .. ..Belgium 2000 y 0.26 0.12 3.22 1999/2000 0.56Benin 2003 c 0.36 0.22 4.93 .. ..Bolivia 2002 y 0.58 0.76 29.65 .. ..Bosnia & Herzegovina 2001 c 0.25 0.10 3.25 .. ..Botswana 1993.5 c 0.63 .. .. .. ..Brazil 2001 y 0.59 0.65 16.25 1996 0.85Bulgaria 2003 c 0.28 0.12 3.56 .. ..Burkina Faso 2003 c 0.38 0.23 4.91 1993 0.42Burundi 1998 c 0.42 0.31 6.49 .. ..Cambodia 1997 c 0.40 0.28 4.80 .. ..Cameroon 2001 c 0.45 .. .. .. ..Canada 2000 y 0.33 0.18 4.52 1991 0.64Central African Rep. 1993 c 0.61 .. .. .. ..Chile 2000 y 0.51 0.47 10.72 .. ..China 2001 c 0.45 .. .. .. ..Colombia 1999 y 0.54 0.57 15.00 2001 0.8Costa Rica 2000 y 0.46 0.39 9.65 .. ..Côte d’Ivoire 2002 c 0.45 0.33 6.75 .. ..Croatia 2001 c 0.29 0.17 .. .. ..Czech Rep. 1996 y 0.25 0.12 .. 2000 0.92Denmark 1997 y 0.27 0.14 .. 1999/2000 0.51Dominican Rep. 1997 y 0.47 0.40 9.17 .. ..East Timor 2001 c 0.37 0.22 5.42 .. ..Ecuador 1998 y 0.54 0.61 16.09 .. ..Egypt, Arab Rep. 2000 c 0.34 0.20 .. 1990 0.65El Salvador 2002 y 0.50 0.52 15.88 .. ..Estonia 1998 c 0.32 0.17 4.73 2001 0.79Ethiopia 2000 c 0.30 0.15 3.34 2001 0.47Finland 2000 y 0.25 0.10 3.12 1999/2000 0.27France 1994 y 0.31 0.15 .. 1999/2000 0.58Gambia, The 1998 c 0.48 0.44 .. .. ..Georgia 2002 c 0.38 0.25 6.11 .. ..Germany 2000 y 0.28 0.12 3.58 1999/2000 0.63Ghana 1999 c 0.41 0.28 7.30 .. ..Greece 1998 c 0.36 0.22 .. 1999/2000 0.58Guatemala 2000 y 0.58 0.66 16.81 .. ..Guinea 2003 c 0.39 0.24 5.09 .. ..Guinea-Bissau 1993 c 0.40 .. .. 1988 0.62Guyana 1998 y 0.45 .. .. .. ..Haiti 2001 y 0.68 0.98 45.43 .. ..Honduras 1999 y 0.52 0.51 11.72 1993 0.66Hungary 2002 c 0.24 0.09 2.96 .. ..India 1999/2000 c 0.33 .. .. .. ..Indonesia 2000 c 0.34 .. .. 1993 0.46Iran 1998 c 0.43 0.33 .. .. ..Ireland 2000 y 0.31 0.16 4.27 1999/2000 0.44Israel 2001 c 0.35 0.20 4.90 .. ..Italy 2000 c 0.31 0.16 4.26 1999/2000 0.73Jamaica 2001 c 0.42 0.28 5.90 .. ..Japan 1993 y 0.25 0.10 .. 1995 0.59Jordan 2002 c 0.39 0.25 5.46 1997 0.78Kazakhstan 2003 c 0.30 0.14 3.88 .. ..Kenya 1997 c 0.44 0.32 6.56 .. ..Korea, Rep. 1998 y 0.32 0.15 .. 1990 0.34Kyrgyzstan 2002 c 0.29 0.13 3.63 .. ..Lao PDR 1997/1998 c 0.35 0.20 4.10 1999 0.39


Table A2. Income/consumption inequality measures—continuedIncome/consumption inequality Land inequality

Survey year y/c* Gini index GE (0) 90th/10th percentile ratio Survey year Gini index

Latvia 1998 c 0.34 0.19 .. 2001 0.58Lesotho 1995 c 0.63 .. .. 1989/1990 0.49Lithuania 2000 c 0.29 0.14 3.94 .. ..Luxembourg 2000 y 0.29 0.13 3.92 1999/2000 0.48Macedonia, FDR 2003 c 0.36 0.21 5.60 .. ..Madagascar 2001 c 0.46 0.36 8.05 .. ..Malawi 1997/1998 c 0.50 0.44 .. 1993 0.52Malaysia 1997 y 0.49 0.43 .. .. ..Mali 2001 c 0.39 0.25 5.81 .. ..Mauritania 2000 c 0.38 0.24 5.92 .. ..Mexico 2002 y 0.49 0.47 11.87 .. ..Moldova 2001 c 0.36 .. .. .. ..Mongolia 1998 c 0.30 0.16 .. .. ..Morocco 1998 c 0.38 0.23 5.33 1996 0.62Mozambique 1996/1997 c 39.60 0.27 .. .. ..Namibia 1993 c 70.70 .. .. 1997 0.36Nepal 1996 c 0.36 0.21 4.54 1992 0.45Netherlands 1999 y 0.29 0.16 3.87 1999/2000 0.57New Zealand 1997 y 0.37 0.23 .. .. ..Nicaragua 2001 c 0.40 0.27 6.52 2001 0.72Niger 1995 c 0.51 .. .. .. ..Nigeria 2003 c 0.41 0.29 7.26 .. ..Norway 2000 y 0.27 0.14 2.95 1999 0.18Pakistan 2001 c 0.27 0.12 3.09 1990 0.57Panama 2000 c 0.55 0.60 18.65 2001 0.52Paraguay 2001 y 0.55 0.61 18.26 1991 0.93Peru 2000 c 0.48 0.51 14.60 1994 0.86Philippines 2000 c 0.46 .. .. 1991 0.55Poland 2002 c 0.31 0.15 4.03 2002 0.69Portugal 1997 y 0.39 0.27 .. 1999/2000 0.74Romania 2002 c 0.28 0.12 3.63 .. ..Russian Federation 2002 c 0.32 0.17 4.67 .. ..Senegal 1995 c 0.40 0.26 5.18 1998 0.5Serbia & Montenegro 2003 c 0.28 0.12 3.60 .. ..Singapore 1998 y 0.43 0.33 .. .. ..Slovak Rep. 1996 y 0.26 0.12 .. .. ..Slovenia 1998 c 0.28 0.13 .. 1991 0.62South Africa 2000 c 0.58 0.61 16.91 .. ..Spain 2000 y 0.35 0.21 4.74 1999/2000 0.77Sri Lanka 2002 c 0.38 0.23 4.98 .. ..St. Lucia 1995 c 0.44 0.37 9.38 .. ..Sweden 2000 y 0.25 0.11 3.18 1999/2000 0.32Switzerland 1992 y 0.31 0.17 .. 1990 0.5Taiwan, China 2000 c 0.24 0.09 2.86 .. ..Tajikistan 2003 c 0.32 0.16 4.08 .. ..Tanzania 2001 c 0.35 0.20 4.89 .. ..Thailand 2002 c 0.40 0.25 5.56 1993 0.47Trinidad & Tobago 1992 c 0.39 0.26 6.24 .. ..Tunisia 2000 c 0.40 0.28 .. 1993 0.7Turkey 2002 c 0.37 0.23 5.73 1991 0.61Turkmenistan 1998 c 0.41 0.28 .. .. ..Uganda .. .. .. 1991 0.59Ukraine 1999 y 0.29 .. .. .. ..United Kingdom 1999 y 0.34 0.20 5.00 1999/2000 0.66United States 2000 y 0.38 0.26 6.30 1997 0.76Uruguay—urban 2000 y 0.43 0.32 7.73 2000 0.79Uzbekistan 2000 c 0.27 0.12 .. .. ..Venezuela, RB de 2000 y 0.42 0.33 7.94 1996/1997 0.88Vietnam 2002 c 0.35 0.20 4.73 1994 0.53Yemen, Rep. 1998 c 0.33 0.19 4.56 .. ..Zambia 1998 c 0.53 0.51 .. .. ..Zimbabwe 1995 c 0.57 .. .. .. ..

Note: * c in this column indicates that the inequality measures refer to a distribution of consumption expenditures; y indicates that the inequality measures refer to a distributionof incomes; .. denotes no data.


Table A3. HealthInfant mortality rate (deaths under age 12 months per 1,000 live births)

By gender By asset quintiles By location By mother’s education level

Survey year Overall Male Female Lowest Second Middle Fourth Highest Urban Rural No Primary Secondary education or higher

Armenia 2000 44.1 46.1 41.9 52.3 50.0 36.8 49.6 27.3 35.9 52.7 .. .. 44.3Bangladesh 1993 100.5 107.3 93.4 .. .. .. .. .. 80.9 102.6 113.3 89.0 57.5Bangladesh 1996/1997 89.6 94.9 84.3 96.5 98.8 96.7 88.8 56.6 73.0 91.2 98.1 82.3 64.8Bangladesh 1999/2000 79.7 82.3 76.9 92.9 93.6 78.1 62.8 57.9 74.2 80.7 91.9 74.5 54.7Benin 1996 103.5 109.3 97.6 119.4 111.1 105.8 103.8 63.3 84.4 112.3 108.4 94.0 49.9Benin 2001 94.8 97.6 92.0 111.5 108.2 106.3 78.1 50.0 72.9 104.5 100.2 87.5 53.1Bolivia 1989 90.6 98.9 82.0 .. .. .. .. .. 73.9 106.6 116.1 98.7 50.2Bolivia 1994 86.6 90.8 82.3 .. .. .. .. .. 68.8 105.8 122.2 99.5 48.2Bolivia 1998 73.5 77.6 69.2 106.5 85.0 75.5 38.6 25.5 53.0 99.9 112.5 86.6 41.3Brazil 1986 84.0 97.3 70.1 .. .. .. .. .. 72.9 106.0 113.2 89.1 23.1Brazil 1996 48.1 51.6 44.4 83.2 46.7 32.9 24.7 28.6 42.4 65.3 93.2 58.1 32.0Botswana 1988 38.6 46.4 31.0 .. .. .. .. .. 38.5 38.7 43.7 35.6 37.3Burkina Faso 1992/1993 107.6 114.5 100.3 .. .. .. .. .. 76.4 113.0 111.3 84.0 52.8Burundi 1987 85.8 97.1 74.2 .. .. .. .. .. 84.5 85.9 87.8 82.2 33.4Cambodia 2000 92.7 102.8 82.2 109.7 108.2 88.2 88.7 50.3 72.3 95.7 102.5 93.6 59.7Cameroon 1991 80.3 86.4 74.3 103.9 101.0 78.8 65.1 51.2 71.7 85.8 112.7 51.6 50.6Cameroon 1998 79.8 85.1 74.6 108.4 86.3 72.6 58.7 55.8 61.0 86.9 103.9 74.1 49.9Central African Rep. 1994/1995 101.8 109.2 94.1 132.3 116.8 99.2 97.6 53.7 79.9 116.3 114.2 100.2 52.0Chad 1996/1997 109.8 119.6 100.0 79.8 136.7 120.2 115.0 89.3 99.3 112.8 112.7 101.6 74.9Colombia 1986 38.7 40.8 36.4 .. .. .. .. .. 37.5 40.7 49.3 42.0 28.6Colombia 1990 27.0 27.6 26.4 .. .. .. .. .. 28.9 23.4 60.5 27.3 20.4Colombia 1995 30.8 34.9 26.5 40.8 31.4 27.0 31.5 16.2 28.3 35.2 26.9 36.5 25.6Colombia 2000 24.4 28.5 20.1 32.0 31.6 22.0 11.9 17.6 21.3 31.1 42.3 28.2 19.6Comoros 1996 83.7 92.5 74.8 87.2 108.5 83.7 62.6 64.6 63.8 90.0 87.4 78.5 67.1Côte d’Ivoire 1994 91.3 99.1 83.2 117.2 97.3 88.9 78.8 63.3 74.7 99.7 98.8 78.1 61.0Côte d’Ivoire 1998 111.5 130.3 92.5 .. .. .. .. .. 84.7 123.9 123.5 94.7 61.8Dominican Rep. 1986 70.1 79.0 61.0 .. .. .. .. .. 71.9 67.9 96.1 73.7 47.5Dominican Rep. 1991 44.4 53.3 34.9 .. .. .. .. .. 37.2 54.4 46.8 54.1 25.9Dominican Rep. 1996 48.6 51.0 46.1 66.7 54.5 52.3 33.5 23.4 45.8 52.6 84.7 53.8 29.3Dominican Rep. 1999 36.8 38.8 34.9 .. .. .. .. .. 35.3 39.1 34.7 50.6 17.9Ecuador 1987 65.2 70.4 59.7 .. .. .. .. .. 51.6 77.7 104.5 68.7 39.4Egypt, Arab Rep. 1988 93.1 93.7 92.4 .. .. .. .. .. 64.2 113.8 112.3 82.8 37.8Egypt, Arab Rep. 1992 79.9 84.4 75.3 .. .. .. .. .. 54.4 96.2 97.8 73.0 42.4Egypt, Arab Rep. 1995 72.9 72.5 73.3 109.7 88.7 64.6 50.6 31.8 51.1 86.8 93.4 70.0 37.5Egypt, Arab Rep. 2000 54.7 55.0 54.5 75.6 63.9 53.9 43.9 29.6 43.1 61.8 68.3 58.8 35.9El Salvador 1985 70.9 81.1 59.7 .. .. .. .. .. 57.6 82.4 99.7 64.2 24.9Eritrea 1995 75.6 81.9 69.0 74.0 66.2 87.0 85.8 67.5 79.8 74.4 76.0 77.0 67.2Ethiopia 2000 112.9 124.4 100.6 92.8 114.9 141.5 118.1 95.1 96.5 114.7 119.1 85.0 63.5Gabon 2000 61.1 73.6 48.9 57.0 68.1 66.6 72.7 35.9 60.7 62.2 65.5 58.7 62.5Ghana 1988 80.9 88.9 72.5 .. .. .. .. .. 66.0 86.6 87.2 74.5 80.2Ghana 1993 74.7 79.2 70.1 77.5 94.6 82.8 64.2 45.8 54.9 82.2 87.1 66.7 44.9Ghana 1998 61.2 64.4 57.9 72.7 58.0 82.1 52.5 26.0 42.6 67.5 66.1 70.3 51.3Guatemala 1987 79.2 89.5 68.5 .. .. .. .. .. 66.6 84.2 82.9 80.1 41.8Guatemala 1995 57.2 62.7 51.5 56.9 79.7 55.7 46.7 35.0 45.4 62.9 69.8 53.6 26.1Guatemala 1998/1999 49.1 50.0 48.1 58.0 50.8 52.1 39.6 39.2 49.0 49.1 55.7 46.5 41.1Guinea 1999 106.6 112.3 100.6 118.9 127.9 113.5 91.4 70.2 79.2 115.8 112.0 78.4 60.6Haiti 1994 87.1 97.7 76.2 93.7 93.6 85.6 81.7 74.3 83.2 88.9 95.2 78.4 75.6Haiti 2000 89.4 96.5 82.6 99.5 70.0 93.4 88.4 97.2 87.0 90.5 90.9 97.5 55.9India 1992/1993 86.3 88.6 83.9 109.2 106.3 89.7 65.6 44.0 59.4 94.3 100.6 68.2 46.3India 1998/1999 73.0 74.8 71.1 96.5 80.7 76.3 55.3 38.1 49.2 79.7 87.0 66.9 42.2Indonesia 1987 .. 84.1 63.8 .. .. .. .. .. 49.9 83.3 100.9 75.0 36.2Indonesia 1991 .. 79.9 67.9 .. .. .. .. .. 57.2 81.0 89.0 81.1 34.6Indonesia 1994 .. 73.5 58.8 .. .. .. .. .. 43.1 75.2 90.5 70.4 39.5Indonesia 1997 52.2 59.1 44.9 78.1 57.3 51.4 39.4 23.3 35.7 58.0 77.5 58.8 28.0Jordan 1990 .. 36.4 37.3 .. .. .. .. .. 35.8 39.2 38.7 41.1 33.8Jordan 1997 29.0 34.3 23.4 35.4 28.8 30.1 25.9 23.4 26.7 39.1 54.2 31.9 25.5Kazakhstan 1995 40.7 46.7 34.6 39.2 43.1 36.6 48.9 35.1 39.2 42.1 .. .. 40.9Kazakhstan 1999 54.9 62.0 47.3 67.6 65.3 65.8 27.3 42.3 43.7 63.8 .. .. 55.2Kenya 1989 .. 63.4 54.3 .. .. .. .. .. 56.7 59.2 72.1 55.4 42.3Kenya 1993 .. 66.6 58.6 .. .. .. .. .. 45.5 64.9 66.3 70.6 34.8Kenya 1998 70.7 74.5 66.8 95.8 82.9 58.5 61.0 40.2 55.4 73.8 82.2 79.7 40.0Kyrgyz Rep. 1997 66.2 71.9 60.2 83.3 73.3 67.5 49.6 45.8 54.3 70.4 .. 255.6 66.0Liberia 1986 .. 168.9 135.4 .. .. .. .. .. 140.4 160.7 162.7 146.3 112.5Madagascar 1992 .. 103.2 101.8 .. .. .. .. .. 74.7 106.8 137.9 97.6 72.9Madagascar 1997 99.3 108.7 89.5 119.1 118.3 103.2 76.2 57.5 77.9 105.0 124.2 102.0 63.5


Table A3. Health—continuedInfant mortality rate (deaths under age 12 months per 1,000 live births)

By gender By asset quintiles By location By mother’s education level

Survey year Overall Male Female Lowest Second Middle Fourth Highest Urban Rural No Primary Secondary education or higher

Malawi 1992 136.1 141.7 130.4 141.2 133.7 154.1 139.2 106.1 118.1 138.4 143.4 129.6 96.3Malawi 2000 112.5 117.1 107.9 131.5 110.7 117.4 109.1 86.4 82.5 116.7 116.6 114.3 65.4Mali 1987 .. 136.6 125.5 .. .. .. .. .. 89.8 144.1 139.1 74.6 74.1Mali 1995 133.5 140.5 126.5 151.4 146.9 138.9 129.0 93.2 98.7 145.0 139.6 112.7 59.6Mali 2001 126.2 136.4 115.6 137.2 125.2 140.6 128.7 89.9 105.9 131.9 130.0 122.4 51.7Mauritania 2000/2001 66.8 .. .. 60.8 59.4 78.0 72.8 62.3 .. .. .. .. ..Mexico 1987 .. 60.4 52.4 .. .. .. .. .. 41.6 79.2 27.6 .. 83.9Morocco 1987 .. 82.8 80.6 .. .. .. .. .. 64.1 90.9 85.6 52.8 62.5Morocco 1992 63.1 68.6 57.4 79.7 67.7 62.4 58.5 35.1 51.9 69.3 67.7 53.2 20.9Mozambique 1997 147.4 153.0 141.9 187.7 136.2 144.3 134.2 94.7 100.8 159.7 155.6 143.9 72.5Namibia 1992 61.5 66.6 56.5 63.6 63.0 48.4 72.2 57.3 63.1 60.7 57.9 65.5 57.0Nepal 1996 93.0 101.9 83.7 96.3 107.2 103.6 84.7 63.9 61.1 95.3 97.5 80.0 53.4Nepal 2001 77.2 79.2 75.2 85.5 87.7 76.6 72.8 53.2 50.1 79.3 84.6 61.0 39.1Nicaragua 1997/1998 45.2 50.2 40.2 50.7 53.7 45.7 40.2 25.8 40.0 51.1 62.1 45.3 31.0Niger 1990 .. 135.8 133.0 .. .. .. .. .. 89.0 142.6 137.0 114.9 48.8Niger 1998 135.8 140.9 130.5 131.1 152.3 157.2 142.0 85.8 79.9 146.7 140.9 99.6 70.1Nigeria 1990 91.6 93.9 89.3 102.2 102.3 93.1 85.8 68.6 75.6 95.9 96.1 87.2 69.9Nigeria 1999 .. 73.3 68.0 .. .. .. .. .. 59.3 74.9 76.9 70.8 55.7Pakistan 1990 94.0 102.1 85.5 88.7 108.7 109.3 95.7 62.5 74.6 102.2 98.6 90.4 59.5Paraguay 1990 35.9 39.0 32.6 42.9 36.5 46.1 33.5 15.7 32.6 38.7 52.2 39.1 22.9Peru 1986 .. 83.2 74.8 .. .. .. .. .. 55.8 106.1 118.8 88.3 41.5Peru 1992 .. 68.1 59.2 .. .. .. .. .. 47.5 89.9 100.0 83.2 33.9Peru 1996 49.9 56.1 43.5 78.3 53.6 34.4 36.0 19.5 34.9 71.0 78.9 61.7 30.6Peru 2000 43.2 46.0 40.2 63.5 53.9 32.6 26.5 13.9 28.4 60.3 73.4 53.5 27.4Philippines 1993 .. 43.5 32.9 .. .. .. .. .. 31.9 44.3 76.7 46.6 28.9Philippines 1998 36.0 39.4 32.3 48.8 39.2 33.7 24.9 20.9 30.9 40.2 78.5 45.1 28.3Rwanda 1992 90.2 98.4 82.1 .. .. .. .. .. 87.5 90.4 97.3 84.9 65.3Rwanda 2000 117.4 123.2 111.6 138.7 120.2 123.4 118.9 87.9 77.9 123.5 134.8 113.9 59.5Senegal 1986 90.9 98.6 82.9 .. .. .. .. .. 70.1 101.9 96.2 67.2 51.4Senegal 1992/1993 76.1 83.6 68.7 .. .. .. .. .. 54.5 86.8 81.2 58.5 32.1Senegal 1997 69.4 73.6 65.0 84.5 81.6 69.6 58.8 44.9 50.2 79.1 76.1 52.1 28.7South Africa 1998 42.2 49.0 35.3 61.6 51.6 35.8 34.0 17.0 32.6 52.2 58.8 47.6 36.1Sri Lanka 1987 .. 39.6 24.9 .. .. .. .. .. 34.4 32.2 52.2 34.0 27.9Sudan 1990 77.1 83.7 70.3 .. .. .. .. .. 74.0 78.6 82.4 70.1 62.5Tanzania 1992 99.4 103.6 95.1 .. .. .. .. .. 108.3 97.1 103.1 97.9 71.8Tanzania 1996 94.1 100.8 87.1 87.3 118.0 95.6 102.1 64.8 81.7 96.8 105.9 89.3 63.9Tanzania 1999 107.8 .. .. 114.8 107.5 115.4 106.8 91.9 .. .. .. .. ..Thailand 1987 38.5 45.6 30.9 .. .. .. .. .. 25.9 40.8 55.5 38.7 18.5Togo 1988 84.0 88.5 79.3 .. .. .. .. .. 74.7 87.3 88.2 79.3 54.3Togo 1998 80.3 89.1 71.4 84.1 81.7 90.0 73.9 65.8 65.3 85.0 87.4 72.1 54.4Trinidad & Tobago 1987 30.5 28.4 32.8 .. .. .. .. .. 34.2 27.9 69.0 24.2 38.5Tunisia 1988 55.5 56.3 54.7 .. .. .. .. .. 49.6 61.8 61.8 49.6 34.4Turkey 1993 68.3 70.5 66.0 99.9 72.7 72.1 54.4 25.4 58.1 82.6 92.2 63.4 25.4Turkey 1998 48.4 51.0 45.5 68.3 54.6 42.1 37.5 29.8 42.2 58.6 66.3 46.2 27.9Turkmenistan 2000 71.6 83.0 59.7 89.3 78.6 68.2 62.4 58.4 60.1 79.9 113.8 29.7 71.4Uganda 1988 106.0 111.3 100.6 .. .. .. .. .. 103.8 106.2 114.9 101.1 85.8Uganda 1995 86.1 87.4 84.9 109.0 79.5 90.4 84.5 63.2 74.4 87.6 94.0 87.9 48.0Uganda 2000/2001 89.4 93.3 85.5 105.7 98.3 94.5 81.0 60.2 54.5 93.7 106.7 88.4 52.6Uzbekistan 1996 43.5 50.2 36.7 54.4 39.8 36.0 39.0 45.9 42.9 43.8 .. .. 43.6Vietnam 1997 34.8 42.0 26.9 42.8 43.2 35.2 27.2 16.9 23.2 36.6 48.8 43.3 29.0Yemen, Rep. 1991/1992 100.3 108.1 92.1 .. .. .. .. .. 90.9 102.2 102.4 77.5 43.7Yemen, Rep. 1997 89.5 98.4 80.0 108.5 102.0 88.9 80.9 60.0 75.4 93.6 92.6 71.6 66.9Zambia 1992 98.3 106.2 90.5 .. .. .. .. .. 78.0 116.0 114.9 98.9 79.4Zambia 1996 107.7 116.3 99.3 123.6 131.5 105.1 104.1 69.8 91.9 117.9 132.9 110.2 81.7Zambia 2001/2002 93.9 95.1 92.7 115.2 93.1 113.8 80.8 56.7 76.7 102.6 108.1 98.8 70.3Zimbabwe 1988 56.4 63.2 49.5 .. .. .. .. .. 37.0 63.4 77.1 53.9 38.2Zimbabwe 1994 51.2 56.9 45.5 52.0 49.5 47.4 64.2 41.6 44.3 53.6 61.6 53.9 38.6Zimbabwe 1999 59.7 63.1 56.2 59.1 63.9 67.1 63.1 44.3 47.2 65.3 81.1 60.6 54.0

Note: Only countries for which some data are available are included in this table; .. denotes no data.


Table A4. EducationShare of total population Education Share of inequality

by years of schooling Mean years of schooling inequality measures attributable

By location By gender To location To gender

Survey year 0 1–6 7–12 13 or Total Urban Rural Male Female Gini index GE (0.5)years years above

Afghanistan .. .. .. .. .. .. .. .. .. .. .. .. .. ..Albania 2002 0.04 0.10 0.76 0.10 9.19 10.55 8.09 9.58 8.83 0.21 0.14 0.06 0.01Angola 2000 0.33 0.47 0.20 0.00 3.65 .. .. .. .. .. .. .. ..Argentina 2001 0.01 0.08 0.65 0.26 10.33 .. .. 10.26 10.40 0.22 0.09 .. 0.00Armenia 2000 0.01 0.02 0.61 0.36 11.44 11.98 10.60 11.50 11.38 0.13 0.04 0.05 0.00Australia 1994 0.00 0.00 0.58 0.42 12.50 12.64 12.21 13.07 11.89 0.15 0.04 0.00 0.03Austria 1995 0.00 0.01 0.89 0.10 10.64 11.10 10.35 10.97 10.30 0.14 0.03 0.02 0.02Azerbaijan 1995 0.02 0.03 0.65 0.30 10.99 11.61 10.23 11.62 10.43 0.15 0.07 0.03 0.02Bangladesh 1999/2000 0.46 0.26 0.24 0.04 3.92 6.31 3.29 4.94 2.90 0.62 1.18 0.04 0.03Belarus 2002 0.02 0.28 0.27 0.44 11.27 .. .. 10.84 11.61 0.25 0.13 .. 0.00Belgium 1997 0.03 0.12 0.47 0.38 11.52 .. .. 11.65 11.39 0.22 0.12 .. 0.00Benin 2001 0.63 0.23 0.12 0.03 2.47 4.28 1.34 3.63 1.53 0.75 1.69 0.10 0.05Bolivia 1998 0.10 0.35 0.35 0.20 7.63 9.28 4.07 8.48 6.84 0.38 0.38 0.16 0.02Bosnia & Herzegovina 2001 0.06 0.16 0.69 0.10 9.32 11.03 8.71 10.29 8.39 0.24 0.18 0.03 0.03Brazil 2001 0.20 0.21 0.23 0.36 8.38 8.67 6.61 8.44 8.32 0.39 0.53 0.01 0.00Bulgaria 2003 0.06 0.22 0.24 0.48 10.85 12.22 7.50 10.76 10.94 0.19 0.26 0.05 0.04Burkina Faso 1998/1999 0.86 0.08 0.04 0.02 1.00 4.28 0.33 1.48 0.63 0.90 2.63 0.27 0.03Burundi 2000 0.61 0.32 0.05 0.02 2.13 .. .. .. .. .. .. .. ..Cambodia 1999 0.00 0.63 0.36 0.01 5.70 7.12 5.52 6.35 5.15 0.28 0.12 0.04 0.04Cameroon 1998 0.32 0.29 0.30 0.08 5.32 7.58 4.14 6.54 4.27 0.50 0.84 0.05 0.03Canada 2000 0.00 0.01 0.34 0.65 14.27 14.39 13.30 14.34 14.20 0.13 0.03 0.01 0.00Central African Rep. 1994/1995 0.48 0.35 0.14 0.02 2.95 4.53 1.82 4.22 1.79 0.66 1.32 0.08 0.07Chad 1996/1997 0.76 0.16 0.06 0.01 1.30 3.09 0.69 2.20 0.53 0.86 2.23 0.12 0.10Chile 2000 0.02 0.19 0.54 0.24 10.27 10.83 6.77 10.42 10.14 0.23 0.13 0.08 0.00China 2000 0.07 0.33 0.55 0.05 6.54 8.53 5.18 7.22 5.82 0.37 0.35 0.08 0.02Colombia 2000 0.07 0.44 0.36 0.13 7.19 8.29 4.08 7.19 7.19 0.36 0.31 0.13 0.00Comoros 1996 0.64 0.17 0.16 0.04 2.76 4.41 2.06 3.69 1.96 0.71 1.82 0.04 0.00Congo, Dem. Rep. 2000 0.25 0.35 0.36 0.04 5.39 .. .. .. .. .. .. .. ..Costa Rica 2000 0.05 0.48 0.31 0.16 7.90 9.02 6.15 7.91 7.89 0.30 0.22 .. 0.00Côte d’Ivoire 1998/1999 0.56 0.24 0.14 0.06 3.43 5.30 2.16 4.40 2.52 0.68 1.48 0.07 0.03Czech Rep. 1996 0.00 0.16 0.74 0.10 9.14 .. .. 9.31 8.98 0.19 0.06 .. 0.00Denmark 1992 0.00 0.00 0.83 0.17 11.62 11.82 11.23 11.78 11.47 0.11 0.02 0.01 0.00Dominican Rep. 2002 0.10 0.35 0.40 0.15 7.47 8.32 5.85 7.38 7.56 0.38 0.36 0.04 0.00East Timor 2001 0.60 0.19 0.20 0.02 3.19 5.40 2.52 3.43 0.86 0.69 1.68 0.04 0.03Ecuador 1998/1999 0.08 0.42 0.33 0.18 8.12 9.67 5.49 8.26 7.98 0.33 0.28 0.12 0.00Egypt, Arab Rep. 2000 0.35 0.19 0.28 0.17 6.60 8.60 4.83 7.90 5.28 0.51 0.90 0.05 0.02El Salvador 2000 0.18 0.38 0.32 0.12 6.56 8.32 3.53 6.98 6.22 0.45 0.56 0.13 0.00Estonia 2000 0.00 0.02 0.58 0.40 12.49 .. .. 12.13 12.80 0.16 0.04 .. 0.01Ethiopia 2000 0.74 0.16 0.09 0.01 1.56 5.16 0.88 2.21 0.94 0.83 2.14 0.15 0.04Finland 2000 0.00 0.00 0.70 0.30 12.03 12.24 11.26 11.88 12.17 0.15 0.03 0.02 0.00France 1994 0.20 0.12 0.48 0.20 8.26 8.58 7.36 8.24 8.28 0.37 0.49 0.00 0.00Gabon 2000 0.19 0.32 0.38 0.11 6.71 7.45 4.55 7.62 5.78 0.39 0.52 0.04 0.02Gambia, The 2000 0.58 0.14 0.19 0.09 3.82 .. .. .. .. .. .. .. ..Germany 2000 0.02 0.36 0.39 0.23 10.07 10.39 9.57 10.07 10.07 0.25 0.13 0.01 0.00Ghana 1998/1999 0.31 0.14 0.41 0.14 6.62 8.79 5.39 8.31 5.22 0.46 0.78 0.04 0.04Guatemala 1998/1999 0.29 0.45 0.21 0.05 4.58 6.28 3.10 5.14 4.07 0.54 0.83 0.07 0.01Guinea 1999 0.77 0.09 0.09 0.05 1.97 4.44 0.86 3.06 1.08 0.84 2.22 0.14 0.06Guinea-Bissau 2000 0.72 0.14 0.05 0.09 2.34 .. .. .. .. .. .. .. ..Guyana 2000 0.00 0.28 0.62 0.10 8.89 10.00 8.37 8.94 8.84 0.20 0.07 0.05 0.00Haiti 2000 0.40 0.33 0.22 0.05 3.93 6.70 2.14 4.75 3.19 0.61 1.12 0.15 0.02Honduras 2001 0.19 0.55 0.20 0.06 5.55 7.41 3.64 5.57 5.52 0.45 0.56 0.11 0.00Hungary 1999 0.00 0.10 0.75 0.14 10.01 .. .. 10.09 9.94 0.18 0.06 .. 0.00India 1998/2000 0.41 0.20 0.31 0.08 5.03 7.78 3.93 6.50 3.57 0.56 1.02 0.05 0.04Indonesia 2002 0.09 0.50 0.34 0.07 7.38 9.04 5.85 7.99 6.77 0.32 0.29 0.08 0.01Iraq 2000 0.26 0.33 0.27 0.14 6.36 .. .. .. .. .. .. .. ..Ireland 1996 0.00 0.03 0.79 0.18 11.00 .. .. 11.14 10.86 0.11 0.04 .. 0.00Israel 2001 0.02 0.03 0.51 0.44 12.63 12.55 13.08 12.75 12.52 0.14 0.07 0.00 0.00Italy 2000 0.03 0.19 0.68 0.10 9.05 9.49 8.56 9.32 8.79 0.23 0.12 0.01 0.00Jamaica 2000 0.01 0.15 0.71 0.12 9.31 .. .. 9.17 9.43 0.19 0.08 .. 0.00Japan 2000 0.00 0.11 0.53 0.36 11.74 11.99 10.79 11.95 11.52 0.17 0.08 0.01 0.00Jordan 2002 0.00 0.19 0.54 0.26 10.42 10.70 9.55 10.74 10.13 0.21 0.07 0.01 0.01Kazakhstan 1999 0.01 0.03 0.79 0.17 10.69 11.15 10.23 10.75 10.64 0.12 0.04 0.02 0.00Kenya 1999 0.20 0.26 0.52 0.02 6.26 8.05 5.48 7.01 5.56 0.38 0.51 0.03 0.01Kosovo 2000 0.07 0.12 0.63 0.18 9.35 10.46 8.61 10.85 7.93 0.46 0.21 0.03 0.15Kyrgyz Rep. 1997 0.01 0.03 0.79 0.17 10.58 11.35 10.16 10.76 10.41 0.12 0.05 0.03 0.00


Table A4. Education—continuedShare of total population Education Share of inequality

by years of schooling Mean years of schooling inequality measures attributable

By location By gender To location To gender

Survey year 0 1–6 7–12 13 or Total Urban Rural Male Female Gini index GE (0.5)years years above

Lao PDR 1997 0.32 0.44 0.20 0.04 4.08 7.32 3.36 5.38 2.85 0.53 0.84 0.07 0.06Lesotho 2000 0.15 0.39 0.42 0.03 5.82 .. .. .. .. .. .. .. ..Luxembourg 2000 0.01 0.17 0.34 0.49 12.31 .. .. 12.96 11.65 0.21 0.08 .. 0.02Madagascar 2001 0.00 0.65 0.26 0.09 6.34 8.02 5.65 6.38 6.30 0.31 0.16 0.08 0.00Malawi 2000 0.30 0.40 0.30 0.01 4.23 7.67 3.60 5.46 3.08 0.52 0.80 0.06 0.05Mali 2001 0.81 0.10 0.06 0.03 1.45 3.80 0.56 2.03 0.94 0.87 2.36 0.18 0.03Mexico 1999 0.08 0.41 0.37 0.14 7.78 8.63 4.67 8.10 7.49 0.34 0.30 0.09 0.00Moldova 2000 0.01 0.07 0.55 0.37 11.75 .. .. 11.71 11.77 0.20 0.07 .. 0.00Mongolia 2000 0.02 0.08 0.63 0.27 10.05 .. .. .. .. .. .. .. ..Morocco 1992 0.63 0.18 0.15 0.04 2.84 4.70 0.94 3.92 1.90 0.74 1.72 0.03 0.04Mozambique 1997 0.48 0.43 0.08 0.00 2.24 4.65 1.54 3.20 1.45 0.65 1.27 0.11 0.06Myanmar 2000 0.26 0.47 0.27 0.00 4.32 .. .. .. .. .. .. .. ..Namibia 2000 0.20 0.23 0.53 0.04 6.65 8.29 5.35 6.73 6.57 0.38 0.52 0.05 0.00Nepal 2001 0.64 0.17 0.17 0.02 2.46 5.38 2.09 3.88 1.22 0.74 1.76 0.04 0.09Netherlands 1999 0.00 0.01 0.71 0.28 12.36 .. .. 12.67 12.03 0.13 0.03 .. 0.01Nicaragua 2001 0.23 0.41 0.26 0.10 5.57 7.28 2.91 5.54 5.59 0.49 0.67 0.13 0.00Niger 1998 0.85 0.09 0.05 0.02 1.12 3.49 0.52 1.57 0.75 0.88 2.56 0.16 0.03Nigeria 1999 0.39 0.23 0.28 0.11 5.77 8.06 4.77 7.06 4.61 0.53 0.97 0.03 0.02Norway 2000 0.00 0.00 0.70 0.30 12.70 12.97 12.36 12.75 12.65 0.11 0.03 0.01 0.00Pakistan 2001 0.59 0.15 0.21 0.05 3.51 5.95 2.43 5.05 2.02 0.70 1.55 0.06 0.06Panama 2000 0.04 0.32 0.43 0.21 9.52 10.84 7.04 9.29 9.74 0.27 0.17 0.11 0.00Papua New Guinea 1996 0.48 0.33 0.11 0.08 3.90 .. .. 4.98 2.79 0.62 1.25 .. 0.03Paraguay 2000 0.06 0.53 0.29 0.12 7.26 8.77 5.15 7.36 7.16 0.35 0.26 0.12 0.00Peru 2000 0.08 0.32 0.39 0.21 8.76 10.24 5.56 9.51 8.03 0.30 0.26 0.14 0.01Philippines 1998 0.03 0.32 0.46 0.19 8.77 9.94 7.41 8.71 8.84 0.24 0.14 0.07 0.00Poland 1999 0.00 0.21 0.67 0.11 9.27 .. .. 9.05 9.47 0.19 0.06 .. 0.00Romania 2002 0.01 0.14 0.70 0.15 9.73 .. .. 10.14 9.33 0.21 0.09 .. 0.01Russian Federation 2000 0.00 0.01 0.40 0.59 13.70 .. .. 13.60 13.79 0.14 0.04 .. 0.00Rwanda 2000 0.38 0.41 0.20 0.01 3.59 6.67 2.96 4.19 3.14 0.55 0.99 0.06 0.01São Tomé & Principe 2000 0.17 0.42 0.27 0.15 6.54 .. .. .. .. .. .. .. ..Senegal 1992/1993 0.77 0.13 0.07 0.03 1.80 3.73 0.51 2.60 1.19 0.83 2.18 0.19 0.03Sierra Leone 2000 0.74 0.04 0.19 0.03 2.44 .. .. .. .. .. .. .. ..Slovak Rep. 1992 0.01 0.14 0.74 0.11 10.36 .. .. 10.74 9.99 0.15 0.05 .. 0.01Slovenia 1999 0.01 0.00 0.86 0.14 11.32 .. .. 11.37 11.27 0.10 0.03 .. 0.00South Africa 1998 0.74 0.14 0.09 0.03 1.95 3.93 0.58 2.72 1.33 0.79 2.10 0.19 0.11Spain 1990 0.13 0.22 0.43 0.22 9.12 .. .. 9.48 8.77 0.31 0.33 .. 0.00Sri Lanka 2002 0.00 0.25 0.57 0.18 9.22 .. .. 8.94 9.47 0.23 0.10 .. 0.00Sudan 2000 0.51 0.20 0.24 0.05 4.01 .. .. .. .. .. .. .. ..Suriname 2000 0.01 0.38 0.52 0.09 7.96 .. .. 7.95 7.98 0.24 0.11 .. 0.00Swaziland 2000 0.20 0.24 0.52 0.04 6.78 .. .. .. .. .. .. .. ..Sweden 2000 0.00 0.09 0.61 0.30 12.00 12.53 11.46 11.84 12.15 0.16 0.04 0.02 0.00Switzerland 1992 0.00 0.00 0.79 0.21 11.64 11.57 11.71 12.24 11.04 0.13 0.03 0.00 0.05Taiwan, China 2000 0.05 0.22 0.47 0.26 9.48 9.74 7.03 10.15 8.84 0.30 0.24 0.02 0.01Tajikistan 1999 0.00 0.05 0.63 0.32 11.96 11.33 12.18 11.94 11.97 0.20 0.07 0.01 0.00Tanzania 1999 0.30 0.19 0.50 0.01 4.58 6.03 4.05 5.36 3.93 0.41 0.74 0.02 0.02Thailand 2000 0.05 0.47 0.34 0.15 6.89 8.97 5.79 7.19 6.62 0.33 0.21 0.10 0.00Togo 1998 0.47 0.32 0.19 0.02 3.15 5.03 2.12 4.57 1.98 0.62 1.25 0.08 0.07Trinidad & Tobago 2000 0.01 0.12 0.78 0.09 9.17 .. .. 9.19 9.14 0.19 0.09 .. 0.00Turkey 1998 0.17 0.50 0.23 0.09 6.14 6.93 4.61 7.23 5.08 0.38 0.47 0.04 0.03Turkmenistan 1998 0.00 0.02 0.77 0.20 10.60 10.98 10.27 10.96 10.28 0.12 0.03 0.02 0.00United Kingdom 1999 0.00 0.00 0.68 0.31 12.16 12.31 11.98 12.21 12.11 0.11 0.02 0.00 0.00United States 2000 0.00 0.02 0.42 0.55 13.83 13.96 13.37 13.85 13.80 0.13 0.04 0.00 0.00Uganda 1995 0.32 0.39 0.27 0.03 4.23 7.53 3.71 5.46 3.12 0.50 0.82 0.05 0.05Uruguay 2000 0.01 0.34 0.45 0.20 9.41 .. .. 9.32 9.49 0.24 0.10 .. 0.00Uzbekistan 1996 0.01 0.02 0.81 0.17 10.66 11.06 10.37 11.00 10.33 0.11 0.03 0.01 0.01Venezuela, RB de 2000 0.08 0.34 0.42 0.17 8.29 9.92 7.96 8.08 8.51 0.30 0.26 0.01 0.00Vietnam 2000 0.06 0.34 0.57 0.02 6.96 8.48 6.44 7.43 6.53 0.28 0.22 0.04 0.01Yemen, Rep. 1999 0.65 0.11 0.17 0.07 3.34 5.95 2.27 5.35 1.54 0.73 1.81 0.06 0.10Zambia 1992 0.16 0.30 0.49 0.06 6.26 8.45 4.91 7.41 5.14 0.37 0.44 0.08 0.04Zimbabwe 1999 0.10 0.21 0.62 0.07 7.57 9.52 6.22 8.41 6.81 0.30 0.30 0.08 0.02

Technical notes

Table A1 Poverty National poverty rate is the percentage of the populationliving below the national poverty line. Rural (urban)poverty rate is the percentage of the rural (urban) popula-tion living below the national rural (urban) poverty line.Population below $1 a day and population below $2 a dayare the percentages of the population living on less than$1.08 a day and $2.15 a day at 1993 international prices. Allabove poverty indicators are also called headcount ratios.Poverty gap is the mean shortfall from the poverty line(counting the nonpoor as having zero shortfall), expressedas a percentage of the poverty line.

To measure poverty, one needs to define the relevant wel-fare measure, to select a poverty line, and to select a povertyindicator. The two most commonly used poverty indicatorsare the headcount ratio and the poverty gap, part of the FGTclass of indexes from Foster, Greer, and Thorbecke (1984).The indexes are defined by

where i is a subgroup of individuals with income below thepoverty line Z; n is the total number of individuals in thesample; Yi is the income of individual i; and α is a distin-guishing parameter between FGT indexes. When α equals 0,the expression simplifies to J/n, or the headcount ratio. Thepoverty gap is given by α equal to 1.

The welfare measure can be income or consumption.Income is generally more difficult to measure accurately, andconsumption comes closer to the notion of standard of liv-ing. And income can vary over time even if the standard ofliving does not. So whenever possible, consumption data areused to estimate poverty. But when consumption data arenot available, income data are used.

Poverty line is a threshold below which a given householdor individual will be regarded as poor. National poverty linesare established according to countries’ own judgment ofminimum acceptable living standards. Because countrieshave different definitions of poverty, consistent comparisonsbetween countries can be difficult. Local poverty lines tendto have higher purchasing power in rich countries, wheremore generous standards are used than in poor countries. Isit reasonable to treat two people with the same standard ofliving—in terms of their command over commodities—dif-ferently because one happens to live in a better-off country?

Poverty measures based on an international poverty lineattempt to hold the real value of the poverty line constantacross countries, as is done when making comparisons overtime. The commonly used $1 a day standard, an interna-tional poverty line measured in 1985 international prices

Pn

Z Y

Zi

i

J

α

α

=1

= 1

( )−

∑

and adjusted to local currency using purchasing power pari-ties (PPP), was chosen because it is typical of the nationalpoverty lines in low-income countries. Recalculated in 1993consumption PPP terms in 1993 prices, the original $1 a dayin 1985 PPP is now about $1.08 a day. PPP exchange rates areused because they take into account the local prices of goodsand services not traded internationally. But PPP rates weredesigned for comparing aggregates from national accounts,not for making international poverty comparisons. As aresult, there is no certainty that an international poverty linemeasures the same degree of need or deprivation acrosscountries. Furthermore, any revisions in the PPP of a coun-try to incorporate better price indexes can produce dramati-cally different poverty lines in local currency.

Since the World Bank produced its first global povertyestimates for World Development Report 1990 using house-hold survey data, the database has expanded considerablyand now includes 440 surveys representing almost 100developing countries. Some 1.1 million randomly sampledhouseholds were interviewed in these surveys, representing93 percent of the population of developing countries. Alongwith improvements in data coverage and quality, the under-lying methodology has also improved, resulting in better andmore comprehensive estimates.

Data availability. Since 1979 there has been considerableexpansion in the number of countries that field such sur-veys, in the frequency of the surveys, and in the quality oftheir data. The number of data sets rose dramatically from amere 13 between 1979 and 1981, to 100 between 1997 and1999. Sub-Saharan Africa continues to lag behind all otherregions, with only 28 countries out of 48 having at least onedata set available.

Data quality. A number of issues arise in measuringhousehold living standards from survey data. As indicatedabove, one relates to the choice of income or consumption asa welfare indicator. Another issue is that household surveyscan differ widely, for example, in the number of consumergoods they identify. And even similar surveys may not bestrictly comparable because of differences in timing or thequality and training of survey enumerators.

Comparisons of countries at different levels of develop-ment pose a potential problem because of differences in therelative importance of consumption of nonmarket goods.The local market value of all consumption in kind (includ-ing own production, particularly important in underdevel-oped rural economies) should be included in total con-sumption expenditure. Similarly, imputed profit from theproduction of nonmarket goods should be included inincome. This is not always done, though such omissionswere a far bigger problem in surveys before the 1980s. Mostsurvey data now include valuations for consumption orincome from own production. Nonetheless, valuation meth-ods vary. For example, some surveys use the price in the


nearest market, while others use the average farm-gate sell-ing price.

In all cases the measures of poverty have been calculatedfrom primary data sources (tabulations or household data)rather than existing estimates. Estimation from tabulationsuses an interpolation method based on Lorenz curves withflexible functional forms, which have proved reliable in pastwork. Empirical Lorenz curves were weighted by householdsize, so they are based on percentiles of population, nothouseholds.

The Poverty Monitoring team in the World Bank’s Devel-opment Research Group calculates the number of people liv-ing below various international poverty lines, as well as otherpoverty and inequality measures that are published in WorldDevelopment Indicators. That database is updated annually asnew survey data become available, and a major reassessmentof progress against poverty is made about every three years.

Table A2 Income DistributionThe Gini index measures the extent to which the distribu-tion of income/consumption (or land) among individualsor households within an economy deviates from a perfectlyequal distribution. A Lorenz curve plots the cumulative per-centage of total income received against the cumulative pro-portion of recipients, starting with the poorest individual orhousehold. The Gini index measures the area between theLorenz curve and a hypothetical line of absolute equality,expressed as the share of the maximum area under the line.Thus a Gini index of zero represents perfect equality, whilean index of 1 implies perfect inequality. The Gini coefficienttakes on values between 0 and 1 with zero interpreted as noinequality.

Generalized Entropy (or GE) indexes provide us with analternative class of income/consumption (or other) inequal-ity measures, given by

.

The value of the measure GE ranges from 0 to infinity,with zero representing an equal distribution (all incomesidentical) and higher values represent higher levels ofinequality. The parameter c in the GE class represents theweight given to distances between incomes at different partsof the income distribution, and can take any real value. Forlower values of c, GE is more sensitive to changes in thelower tail of the distribution, and for higher values GE ismore sensitive to changes that affect the upper tail. The most

GEci

c

i

n

c c n

y

y=

−

−

=∑1 1

12

1

Gini = −==∑∑1

2 211n y

y yi jj

n

i

n

common values of c used are 0, 1 and 2: hence a value of c =0gives more weight to distances between incomes in the lowertail; c = 1 applies equal weights across the distribution; and avalue of c = 2 give proportionately more weight to gaps inthe upper tail. The GE measures with parameters 0 and 1become, with l’Hopital’s rule, two of Theil’s measures ofinequality (Theil, 1967), the mean log deviation and theTheil-T index respectively, as follows:

90th/10th percentile ratio is constructed by dividing theincome (consumption) in the 90th percentile by the income(consumption) in the 10th percentile. A 90th/10th ratio of 5means that the household in the 90th percentile earns (spends)five times as much as the household in the 10th percentile.

Survey year gives the year in which the country surveyused to generate the reported data was completed.

Table A3 HealthTo measure equity in health, we have only used data from 123Demographic Health Surveys (DHS) collected in 67 coun-tries between 1985 and 2002. In addition to breaking downinfant mortality rates by “asset indices” created by Filmer andPritchett (1998) (as documented in Gwatkin and others(2003, 2004), we compiled population breakdowns for differ-ent groupings using the interactive “STATcompiler” featurein the DHS website.

Infant mortality rate is the number of deaths to childrenunder 12 months of age per 1,000 live births. Figures used inthe table are based on births in the 10 years preceding thesurvey.

Asset quintiles are constructed using the Filmer-Pritch-ett method to create an index of wealth based on 20–30household attributes—type of flooring and/or roof, sourceof water, availability of electricity, possession of such itemsas watches, radios, etc. Once the index is created it is appliedto the country’s household surveys to construct a distribu-tion of assets that is then divided into fifths, each householdbelonging to one of these quintiles. Indicators for the quin-tiles are then formed as the average result for all families forthat indicator (e.g. infant mortality rate) within each assetquintile.

Education is the number of years (or level) of formaleducation the child’s mother has completed at the time ofthe survey.

Gender is the sex of the child (male or female) asreported by the child’s mother (or household head if motheris not present).

GE log( )11

1

==∑n

y

y

y

yi i

i

n

GE log( )01

1

==∑n

y

yii

n


Location is listed as urban if the surveyed household livesin a recognized city or surroundings, and listed as rural ifotherwise.

Table A4 EducationFor measuring educational attainment, we used a databasethat was put together by Araujo, Ferreira, and Schady. Thedata come from the individual-level records of varioushousehold surveys for 124 countries. The selection criteriawere to choose a survey instrument that: (a) was nationallyrepresentative; (b) was collected on 2000 or on the closestyear; and (c) included information on the actual number ofyears of education completed by the interviewees. The five-year cohorts group adults who were likely to have completedtheir education at the time of the survey. The cohorts areconstructed based on one survey per country.

The measure share of total population by years ofschooling gives the percent of the population having com-pleted the reported number of years of schooling at the timethe survey was taken.

Mean years of schooling gives the arithmetic mean foryears of formal schooling for the total population, those liv-ing in urban areas, and those living in rural areas, as well asfor males and females.

The Gini index and the Generalized Entropy indexesreported in this table are the same as those described in tableA2, except for the fact that y now denotes years of schooling.

We report the share of inequality in education which isdue to differences between urban and rural dwellers (loca-tion) and between males and females (gender).

The GE class of inequality measures can be decomposedinto a between- and within-group component along the fol-lowing lines:

if c =/ 0,1

if c = 0

if c = 1

where µ is average per capita consumption, j refers to sub-groups, g j refers to the population share of group j, and GE j

refers to inequality in group j. The between-group compo-nent of inequality is captured by the first term to the right ofthe equal sign. It can be interpreted as measuring whatwould be the level of inequality in the population if everyonewithin the group had the same (group average) consump-tion level µ j. The second term on the right reflects the withingroup inequality GE j. Ratios of the respective componentswith the overall inequality level provide a measure of per-centage contribution of between-group and within-groupinequality to total inequality.

GE log GEc jj

j j

jj

g=

+∑ ∑

µ

µ

µ

µgg j

jµ

µ

GE log GEc jj j

jj

jg g=

+∑ ∑µ

µ

GE GEc jj

j

c

jj

c cg=

−−

+∑ ∑1

11

( )

µ

µgg j

j

cµ

µ


289

In this year’s edition, development data are presented in fourtables presenting comparative socioeconomic data for morethan 130 economies for the most recent year for which dataare available and, for some indicators, for an earlier year. Anadditional table presents basic indicators for 75 economieswith sparse data or with populations of less than 2 million.

The indicators presented here are a selection from morethan 800 included in World Development Indicators 2005.Published annually, World Development Indicators reflects acomprehensive view of the development process. Its openingchapter reports on the Millennium Development Goals,which grew out of agreements and resolutions of world con-ferences organized by the United Nations (U.N.) in the pastdecade, and reaffirmed at the Millennium Summit in Sep-tember 2000 by member countries of the U.N. The other fivemain sections recognize the contribution of a wide range offactors: human capital development, environmental sustain-ability, macroeconomic performance, private sector devel-opment and the investment climate, and the global links thatinfluence the external environment for development. WorldDevelopment Indicators is complemented by a separatelypublished database that gives access to over 1,000 data tablesand 800 time-series indicators for 222 economies andregions. This database is available through an electronic sub-scription (WDI Online) or as a CD-ROM.

Data sources and methodologySocioeconomic and environmental data presented here aredrawn from several sources: primary data collected by theWorld Bank, member country statistical publications,research institutes, and international organizations such asthe U.N. and its specialized agencies, the International Mon-etary Fund (IMF), and the Organisation for Economic Co-operation and Development (OECD). Although interna-tional standards of coverage, definition, and classificationapply to most statistics reported by countries and interna-tional agencies, there are inevitably differences in timeliness

and reliability arising from differences in the capabilities andresources devoted to basic data collection and compilation.For some topics, competing sources of data require reviewby World Bank staff to ensure that the most reliable dataavailable are presented. In some instances, where availabledata are deemed too weak to provide reliable measures oflevels and trends or do not adequately adhere to interna-tional standards, the data are not shown.

The data presented are generally consistent with those inWorld Development Indicators 2005. However, data have beenrevised and updated wherever new information has becomeavailable. Differences may also reflect revisions to historicalseries and changes in methodology. Thus data of differentvintages may be published in different editions of WorldBank publications. Readers are advised not to compile dataseries from different publications or different editions of thesame publication. Consistent time-series data are availableon World Development Indicators 2005 CD-ROM andthrough WDI Online.

All dollar figures are in current U.S. dollars unless other-wise stated. The various methods used to convert fromnational currency figures are described in the Technical notes.

Because the World Bank’s primary business is providinglending and policy advice to its low- and middle-incomemembers, the issues covered in these tables focus mainly onthese economies. Where available, information on the high-income economies is also provided for comparison. Readersmay wish to refer to national statistical publications andpublications of the OECD and the European Union for moreinformation on the high-income economies.

Classification of economies and summary measuresThe summary measures at the bottom of each table includeeconomies classified by income per capita and by region.GNI per capita is used to determine the following incomeclassifications: low-income, $825 or less in 2004; middle-

Selected world development indicators


income, $826 to $10,065; and high-income, $10,066 andabove. A further division at GNI per capita $3,255 is madebetween lower-middle-income and upper-middle-incomeeconomies. See the table on classification of economies onthe next page for a list of economies in each group (includ-ing those with populations of less than 2 million).

Summary measures are either totals (indicated by t if theaggregates include estimates for missing data and nonre-porting countries, or by an s for simple sums of the dataavailable), weighted averages (w), or median values (m) cal-culated for groups of economies. Data for the countriesexcluded from the main tables (those presented in Table 5)have been included in the summary measures, where dataare available, or by assuming that they follow the trend ofreporting countries. This gives a more consistent aggregatedmeasure by standardizing country coverage for each periodshown. Where missing information accounts for a third ormore of the overall estimate, however, the group measure isreported as not available. The section on Statistical methodsin the Technical notes provides further information on aggre-gation methods. Weights used to construct the aggregates arelisted in the technical notes for each table.

From time to time an economy’s classification is revisedbecause of changes in the above cutoff values or in the econ-omy’s measured level of GNI per capita. When such changesoccur, aggregates based on those classifications are recalcu-lated for the past period so that a consistent time series ismaintained.

Terminology and country coverageThe term country does not imply political independence butmay refer to any territory for which authorities report separatesocial or economic statistics. Data are shown for economies asthey were constituted in 2003, and historical data are revised toreflect current political arrangements. Throughout the tables,exceptions are noted.

Technical notesBecause data quality and intercountry comparisons are oftenproblematic, readers are encouraged to consult the Technicalnotes, the table on Classification of Economies by Region andIncome (next page), and the footnotes to the tables. For moreextensive documentation see World Development Indicators2005.

Readers may find more information on the WDI 2005, andorders can be made online, by phone, or fax as follows:

For more information and to order online: http://www.worldbank.org/data/wdi2005/index.htm.

To order by phone or fax: 1-800-645-7247 or 703-661-1580; Fax 703-661-1501

To order by mail: The World Bank, P.O. Box 960, Herndon,VA 20172-0960, U.S.A.

Classification of economies by region and income, FY2006

East Asia and the Pacific Latin America and the Caribbean South Asia High income OECDAmerican Samoa UMC Antigua & Barbuda UMC Afghanistan LIC AustraliaCambodia LIC Argentina UMC Bangladesh LIC AustriaChina LMC Barbados UMC Bhutan LIC BelgiumFiji LMC Belize UMC India LIC CanadaIndonesia LMC Bolivia LMC Maldives LMC DenmarkKiribati LMC Brazil LMC Nepal LIC FinlandKorea, Dem. Rep. LIC Chile UMC Pakistan LIC FranceLao PDR LIC Colombia LMC Sri Lanka LMC GermanyMalaysia UMC Costa Rica UMC GreeceMarshall Islands LMC Cuba LMC Sub-Saharan Africa IcelandMicronesia, Fed. Sts. LMC Dominica UMC Angola LMC IrelandMongolia LIC Dominican Republic LMC Benin LIC ItalyMyanmar LIC Ecuador LMC Botswana UMC JapanNorthern Mariana Islands UMC El Salvador LMC Burkina Faso LIC Korea, Rep.Palau UMC Grenada UMC Burundi LIC LuxembourgPapua New Guinea LIC Guatemala LMC Cameroon LIC NetherlandsPhilippines LMC Guyana LMC Cape Verde LMC New ZealandSamoa LMC Haiti LIC Central African Rep. LIC NorwaySolomon Islands LIC Honduras LMC Chad LIC PortugalThailand LMC Jamaica LMC Comoros LIC SpainTimor-Leste LIC Mexico UMC Congo, Dem. Rep. LIC SwedenTonga LMC Nicaragua LIC Congo, Rep. LIC SwitzerlandVanuatu LMC Panama UMC Côte d’Ivoire LIC United KingdomVietnam LIC Paraguay LMC Equatorial Guinea UMC United States

Peru LMC Eritrea LICEurope and Central Asia St. Kitts and Nevis UMC Ethiopia LIC Other high-income Albania LMC St. Lucia UMC Gabon UMC AndorraArmenia LMC St. Vincent & the Gambia, The LIC ArubaAzerbaijan LMC Grenadines UMC Ghana LIC Bahamas, TheBelarus LMC Suriname LMC Guinea LIC BahrainBosnia & Herzegovina LMC Trinidad & Tobago UMC Guinea-Bissau LIC BermudaBulgaria LMC Uruguay UMC Kenya LIC BruneiCroatia UMC Venezuela, RB UMC Lesotho LIC Cayman IslandsCzech Rep. UMC Liberia LIC Channel IslandsEstonia UMC Middle East and North Africa Madagascar LIC CyprusGeorgia LMC Algeria LMC Malawi LIC Faeroe IslandsHungary UMC Djibouti LMC Mali LIC French PolynesiaKazakhstan LMC Egypt, Arab Rep. LMC Mauritania LIC GreenlandKyrgyz Rep. LIC Iran, Islamic Rep. LMC Mauritius UMC GuamLatvia UMC Iraq LMC Mayotte UMC Hong Kong, ChinaLithuania UMC Jordan LMC Mozambique LIC Isle of ManMacedonia, FYR LMC Lebanon UMC Namibia LMC IsraelMoldova LIC Libya UMC Niger LIC KuwaitPoland UMC Morocco LMC Nigeria LIC LiechtensteinRomania LMC Oman UMC Rwanda LIC Macao, ChinaRussian Federation UMC Syrian Arab Rep. LMC São Tomé & Principe LIC MaltaSerbia & Montenegro LMC Tunisia LMC Senegal LIC MonacoSlovak Rep. UMC West Bank & Gaza LMC Seychelles UMC Netherlands AntillesTajikistan LIC Yemen, Rep. LIC Sierra Leone LIC New CaledoniaTurkey UMC Somalia LIC Puerto RicoTurkmenistan LMC South Africa UMC QatarUkraine LMC Sudan LIC San MarinoUzbekistan LIC Swaziland LMC Saudi Arabia

Tanzania LIC SingaporeTogo LIC SloveniaUganda LIC Taiwan, ChinaZambia LIC United Arab EmiratesZimbabwe LIC Virgin Islands (U.S.)

Note: This table classifies all World Bank member economies, and all other economies with populations of more than 30,000. Economies are divided among income groups according to 2004GNI per capita, calculated using the World Bank Atlas method. The groups are: low-income economies (LIC), $825 or less; lower-middle-income economies (LMC), $826–3,255; upper-middle-income economies (UMC), $3,256–10,065; and high-income economies, $10,066 or more.

Source: World Bank data.

291

292

Table 1. Key indicators of developmentPopulation Gross national PPP Life

income (GNI)a gross national expectancy income (GNI)b at birth

Millions Average Density $ $ $ $ Gross Adult Carbon annual people per billions per capita billions per capita domestic literacy dioxide

% growth sq. km product (GDP) rate emissionsper capita Male Female % ages 15 per capita % growth years years and older metric

tons 2004 2000–4 2004 2004 2004 2004 2004 2003–4 2003 2003 1998–2004 2000

Albania 3.2 0.6 116 6.6 2,080 16 5,070 5.6 72 77 99 c 0.9Algeria 32.4 1.6 14 73.7 2,280 203 d 6,260 d 3.4 70 72 70 e 2.9Angola 14.0 3.0 11 14.4 1,030 28 d 2,030 d 7.7 45 48 67 e 0.5Argentina 38.2 1.0 14 142.3 3,720 476 12,460 8.0 71 78 97 c 3.8Armenia 3.0 –0.5 108 3.4 1,120 13 4,270 10.3 71 79 99 c 1.1Australia 20.1 1.2 3 541.2 26,900 588 29,200 1.8 77 83 .. 18.0Austria 8.1 0.3 98 262.1 32,300 258 31,790 1.9 76 82 .. 7.6Azerbaijan 8.3 0.7 100 7.8 950 32 3,830 10.6 .. .. 99 c 3.6Bangladesh 140.5 1.7 1,079 61.2 440 278 1,980 3.7 62 63 41 0.2Belarus 9.8 –0.4 47 20.9 2,120 68 6,900 11.5 62 74 100 c 5.9Belgium 10.4 0.4 344 322.8 31,030 326 31,360 2.6 75 81 .. 10.0Benin 6.9 2.6 62 3.7 530 8 1,120 0.2 51 55 34 c 0.3Bolivia 9.0 1.9 8 8.7 960 23 2,590 1.6 62 66 87 c 1.3Bosnia & Herzegovina 3.8 0.4 75 7.8 2,040 29 7,430 4.6 71 77 95 e 5.1Brazil 178.7 1.2 21 552.1 3,090 1,433 8,020 3.9 65 73 88 e 1.8Bulgaria 7.8 –0.9 70 21.3 2,740 61 7,870 6.1 69 76 98 c 5.3Burkina Faso 12.4 2.4 45 4.4 360 15 d 1,220 d 1.6 42 43 .. 0.1Burundi 7.3 1.9 286 0.7 90 5 d 660 d 3.5 41 42 59 e 0.0Cambodia 13.6 1.8 77 4.4 320 30 d 2,180 d 4.2 53 56 74 e 0.0Cameroon 16.4 2.0 35 13.1 800 34 2,090 2.8 47 49 68 e 0.4Canada 31.9 0.9 3 905.6 28,390 978 30,660 2.0 76 83 .. 14.2Central African Rep. 3.9 1.5 6 1.2 310 4 d 1,110 d –0.8 41 42 49 e 0.1Chad 8.8 2.9 7 2.3 260 13 1,420 27.4 47 50 26 e 0.0Chile 16.0 1.2 21 78.4 4,910 168 10,500 4.9 73 80 96 c 3.9China 1,296.5 0.7 139 1,676.8 1,290 7,170 f 5,530 f 8.8 69 73 91 c 2.2

Hong Kong, China 6.8 0.7 6,569 183.5 26,810 216 31,510 7.7 78 83 .. 5.0Colombia 45.3 1.7 44 90.6 2,000 309 d 6,820 d 2.3 69 75 94 e 1.4Congo, Dem. Rep. 54.8 3.0 24 6.4 120 37 d 680 d 3.2 45 46 65 e 0.1Congo, Rep. 3.9 2.8 11 3.0 770 3 750 1.4 50 54 83 0.5Costa Rica 4.1 1.6 80 19.0 4,670 39 d 9,530 d 2.7 76 81 96 1.4Côte d’Ivoire 17.1 2.0 54 13.3 770 24 1,390 –4.0 45 46 48 e 0.7Croatia 4.5 0.7 81 29.7 6,590 53 11,670 2.2 70 78 98 c 4.5Czech Republic 10.2 –0.2 132 93.2 9,150 187 18,400 4.2 72 79 .. 11.6Denmark 5.4 0.3 127 219.4 40,650 170 31,550 2.2 75 80 .. 8.4Dominican Rep. 8.9 1.5 183 18.4 2,080 60 d 6,750 d 0.6 64 70 88 e 3.0Ecuador 13.2 1.5 48 28.8 2,180 49 3,690 5.0 69 74 91 c 2.0Egypt, Arab Rep. 68.7 1.8 69 90.1 1,310 283 4,120 2.5 68 71 .. 2.2El Salvador 6.7 1.7 321 15.6 2,350 33 d 4,980 d –0.2 67 74 80 1.1Eritrea 4.5 2.2 44 0.8 180 5 d 1,050 d –0.2 50 52 .. 0.1Ethiopia 70.0 2.1 70 7.7 110 57 d 810 d 11.2 41 43 42 0.1Finland 5.2 0.2 17 171.0 32,790 154 29,560 3.6 75 82 .. 10.3France 60.0 0.5 109 1,858.7 30,090 g 1,759 29,320 1.9 76 83 .. 6.2Georgia 4.5 –1.1 65 4.7 1,040 13 d 2,930 d 9.6 69 78 .. 1.3Germany 82.6 0.1 237 2,489.0 30,120 2,310 27,950 1.5 76 81 .. 9.6Ghana 21.1 1.8 93 8.1 380 48 d 2,280 d 3.3 54 55 54 c 0.3Greece 11.1 0.4 86 183.9 16,610 244 22,000 3.8 75 81 91 e 8.2Guatemala 12.6 2.6 116 26.9 2,130 52 d 4,140 d 0.1 63 69 69 c 0.9Guinea 8.1 2.1 33 3.7 460 17 2,130 0.5 46 47 .. 0.2Haiti 8.6 1.9 312 3.4 390 14 d 1,680 d –5.5 50 54 52 0.2Honduras 7.1 2.5 64 7.3 1,030 19 d 2,710 d 2.1 63 69 80 c 0.7Hungary 10.1 0.1 109 83.3 8,270 157 15,620 4.6 69 77 99 c 5.4India 1,079.7 1.5 363 674.6 620 3,347 d 3,100 d 5.4 63 64 61 c 1.1Indonesia 217.6 1.3 120 248.0 1,140 753 3,460 3.7 65 69 88 1.3Iran, Islamic Rep. 66.9 1.2 41 154.0 2,300 505 7,550 5.7 68 71 77 e 4.9Ireland 4.0 1.3 58 137.8 34,280 133 33,170 4.2 75 80 .. 11.1Israel 6.8 1.9 313 118.1 17,380 160 23,510 2.6 77 81 97 e 10.0Italy 57.6 –0.1 196 1,503.6 26,120 1,604 27,860 1.3 77 83 .. 7.4Jamaica 2.7 0.8 246 7.7 2,900 10 3,630 1.2 74 78 88 4.2Japan 127.8 0.2 351 4,749.9 37,180 3,838 30,040 2.5 78 85 .. 9.3Jordan 5.4 2.7 61 11.6 2,140 25 4,640 4.9 71 74 90 e 3.2Kazakhstan 15.0 –0.2 6 33.8 2,260 104 6,980 8.8 56 67 100 c 8.1Kenya 32.4 1.9 57 15.0 460 34 1,050 0.4 45 46 74 e 0.3Korea, Rep. 48.1 0.6 488 673.0 13,980 982 20,400 4.1 71 78 .. 9.1Kuwait 2.5 2.9 138 43.1 17,970 47 d 19,510 d 7.1 75 79 83 21.9Kyrgyz Rep. 5.1 0.9 27 2.1 400 9 1,840 6.1 61 69 99 c 0.9Lao PDR 5.8 2.3 25 2.2 390 11 1,850 3.6 54 56 69 e 0.1Latvia 2.3 –0.7 37 12.6 5,460 27 11,850 9.4 66 76 100 c 2.5Lebanon 4.6 1.3 445 22.7 4,980 25 5,380 5.0 69 73 .. 3.5Lithuania 3.4 –0.5 55 19.7 5,740 43 12,610 7.1 66 78 100 c 3.4Macedonia, FYR 2.1 0.4 81 4.9 2,350 13 6,480 1.9 71 76 96 c 5.5Madagascar 17.3 2.8 30 5.2 300 14 830 2.6 54 57 71 e 0.1Malawi 11.2 2.0 119 1.9 170 7 620 1.8 37 38 64 c 0.1Malaysia 25.2 2.0 77 117.1 4,650 243 9,630 5.2 71 76 89 c 6.2Mali 11.9 2.4 10 4.3 360 12 980 –0.3 40 42 19 c 0.1Mauritania 2.9 2.4 3 1.2 420 6 d 2,050 d 4.5 49 53 51 c 1.2

Note: For data comparability and coverage, see the technical notes. Figures in italics are for years other than those specified.

293

Table 1. Key indicators of development—continuedPopulation Gross national PPP Life

income (GNI)a gross national expectancy income (GNI)b at birth

Millions Average Density $ $ $ $ Gross Adult Carbon annual people per billions per capita billions per capita domestic literacy dioxide

% growth sq. km product (GDP) rate emissions$ per capita Male Female % ages 15 per capita

% growth years years and older metric tons

2004 2000–4 2004 2004 2004 2004 2004 2003–4 2003 2003 1998–2004 2000

Mexico 103.8 1.4 54 703.1 6,770 995 9,590 2.9 71 77 90 e 4.3Moldova 4.2 –0.4 128 2.6 710 h 8 1,930 7.8 63 71 96 e 1.5Mongolia 2.5 1.2 2 1.5 590 5 2,020 9.1 64 68 98 c 3.1Morocco 30.6 1.6 69 46.5 1,520 125 4,100 1.9 67 71 51 1.3Mozambique 19.1 2.0 24 4.7 250 22 d 1,160 d 5.9 40 42 46 0.1Namibia 2.0 1.8 2 4.8 2,370 14 d 6,960 d 3.2 41 40 85 c 1.0Nepal 25.2 2.2 176 6.5 260 37 1,470 1.6 60 60 49 c 0.1Netherlands 16.3 0.5 480 515.1 31,700 507 31,220 1.2 76 81 .. 8.7New Zealand 4.1 1.3 15 82.5 20,310 90 22,130 3.1 77 81 .. 8.3Nicaragua 5.6 2.5 46 4.5 790 18 3,300 1.4 67 71 77 e 0.7Niger 12.1 3.0 10 2.8 230 10 d 830 d –1.9 46 47 14 c 0.1Nigeria 139.8 2.4 154 54.0 390 130 d 930 d 1.1 44 45 67 0.3Norway 4.6 0.5 15 238.4 52,030 177 38,550 2.5 77 82 .. 11.1Oman 2.7 2.5 9 20.5 7,890 34 13,250 0.1 73 76 74 8.2Pakistan 152.1 2.4 197 90.7 600 328 2,160 3.9 63 65 49 e 0.8Panama 3.0 1.5 41 13.5 4,450 21 d 6,870 d 4.7 73 77 92 c 2.2Papua New Guinea 5.6 2.3 12 3.3 580 13 d 2,300 d 0.5 56 58 57 c 0.5Paraguay 5.8 2.3 15 6.8 1,170 28 d 4,870 d 0.4 69 73 92 e 0.7Peru 27.5 1.5 22 65.0 2,360 148 5,370 3.5 68 72 88 e 1.1Philippines 83.0 2.0 278 96.9 1,170 406 4,890 4.3 68 72 93 c 1.0Poland 38.2 –0.3 125 232.4 6,090 482 12,640 5.4 71 79 .. 7.8Portugal 10.4 0.5 114 149.8 14,350 201 19,250 1.1 73 80 .. 5.8Romania 21.9 –0.7 95 63.9 2,920 179 8,190 7.7 66 74 97 c 3.8Russian Federation 142.8 –0.5 8 487.3 3,410 1,374 9,620 7.7 60 72 99 c 9.9Rwanda 8.4 2.2 341 1.9 220 11 d 1,300 d 3.5 39 40 64 e 0.1Saudi Arabia 23.2 2.8 11 242.2 10,430 325 d 14,010 d 2.1 72 75 79 e 18.1Senegal 10.5 2.3 54 7.0 670 18 d 1,720 d 3.8 51 54 39 e 0.4Serbia & Montenegro 8.2 .. 80 21.7 2,620 i .. .. 7.0 70 75 96 c 3.7Sierra Leone 5.4 1.9 76 1.1 200 4 790 5.4 36 39 30 e 0.1Singapore 4.3 1.9 6,470 105.0 24,220 115 26,590 6.3 76 80 93 c 14.7Slovak Rep. 5.4 0.0 110 34.9 6,480 77 14,370 5.5 69 78 100 c 6.6Slovenia 2.0 0.1 99 29.6 14,810 41 20,730 4.6 72 80 100 7.3South Africa 45.6 0.9 38 165.3 3,630 500 d 10,960 d 4.3 45 46 .. 7.4Spain 41.3 0.5 83 875.8 21,210 1,035 25,070 2.6 76 84 .. 7.0Sri Lanka 19.4 1.3 301 19.6 1,010 78 4,000 4.8 72 76 90 c 0.6Sudan 34.4 2.2 14 18.2 530 64 1,870 3.5 57 60 59 e 0.2Sweden 9.0 0.3 22 321.4 35,770 267 29,770 3.3 78 82 .. 5.3Switzerland 7.4 0.7 187 356.1 48,230 261 35,370 1.3 78 83 .. 5.4Syrian Arab Rep. 17.8 2.3 97 21.1 1,190 63 3,550 1.3 68 73 83 e 3.3Tajikistan 6.4 1.1 46 1.8 280 7 1,150 9.4 63 69 99 c 0.6Tanzania 36.6 2.0 41 11.6 j 330 j 24 660 4.3 42 43 69 c 0.1Thailand 62.4 0.7 122 158.7 2,540 500 8,020 5.4 67 72 93 c 3.3Togo 5.0 2.1 91 1.9 380 8 d 1,690 d 0.8 49 51 53 e 0.4Tunisia 10.0 1.1 64 26.3 2,630 73 7,310 4.5 71 75 74 c 1.9Turkey 71.7 1.5 93 268.7 3,750 551 7,680 7.4 66 71 88 e 3.3Turkmenistan 4.9 1.5 10 6.6 1,340 34 6,910 15.4 61 68 .. 7.5Uganda 25.9 2.7 132 6.9 270 39 d 1,520 d 3.1 43 44 69 0.1Ukraine 48.0 –0.8 83 60.3 1,260 300 6,250 12.9 63 74 99 c 6.9United Kingdom 59.4 0.2 247 2,016.4 33,940 1,869 31,460 3.0 75 80 .. 9.6United States 293.5 1.0 32 12,150.9 41,400 11,655 39,710 3.4 75 80 .. 19.8Uruguay 3.4 0.6 19 13.4 3,950 31 9,070 11.6 72 79 98 1.6Uzbekistan 25.9 1.3 63 11.9 460 48 1,860 6.3 64 70 99 4.8Venezuela, RB 26.1 1.8 30 105.0 4,020 150 5,760 15.3 71 77 93 c 6.5Vietnam 82.2 1.1 252 45.1 550 222 2,700 6.4 68 72 90 c 0.7West Bank & Gaza 3.5 4.2 564 3.8 1,120 .. .. –5.6 71 75 92 e ..Yemen, Rep. 19.8 3.0 37 11.2 570 16 820 –0.4 57 58 49 0.5Zambia 10.5 1.6 14 4.7 450 9 890 3.2 36 37 68 e 0.2Zimbabwe 13.2 1.0 34 .. .. k 28 2,180 –6.7 39 38 90 1.2World 6,345.1 s 1.2 w 49 w 39,833.6 t 6,280 w 55,584 t 8,760 w 2.9 w 65 w 69 w 82 w 3.8 wLow income 2,338.1 1.8 80 1,184.3 510 5,279 2,260 4.4 57 59 64 0.8Middle income 3,006.2 0.9 44 6,594.2 2,190 19,483 6,480 6.0 67 72 90 3.2

Lower middle income 2,430.3 0.9 63 3,846.9 1,580 13,709 5,640 6.2 68 72 89 2.9Upper middle income 575.9 0.7 20 2,747.8 4,770 5,814 10,090 5.9 65 73 93 6.3

Low & middle income 5,344.3 1.3 55 7,777.5 1,460 24,753 4,630 5.5 63 66 81 2.2East Asia & Pacific 1,870.2 0.9 118 2,389.4 1,280 9,488 5,070 7.6 68 71 85 2.1Europe & Central Asia 472.1 –0.1 20 1,553.3 3,290 3,947 8,360 7.0 64 73 98 6.7Latin America & Carib. 541.3 1.4 27 1,948.1 3,600 4,146 7,660 4.5 68 74 88 2.7Middle East & N. Africa 294.0 1.8 33 588.6 2,000 1,693 5,760 3.3 67 70 74 4.2South Asia 1,447.7 1.7 303 860.3 590 4,103 2,830 5.0 62 64 64 0.9Sub-Saharan Africa 719.0 2.2 30 432.0 600 1,331 1,850 2.4 45 46 61 0.7

High income 1,000.8 0.7 30 32,064.0 32,040 31,000 30,970 2.8 75 81 91 12.4

Note: a. Calculated using the World Bank Atlas method. b. PPP is purchasing power parity; see definitions. c. National estimates based on census data. d. The estimate is based on regression;others are extrapolated from the latest International Comparison Programme benchmark estimates. e. National estimates based on survey data. f. Estimate based on bilateral comparisonbetween China and the United States (Ruoen and Kai 1995). g. GNI and GNI per capita estimates include the French overseas departments of French Guiana, Guadeloupe, Martinique, andRéunion. h. Excludes data for Transnistria. i. Excludes data for Kosovo. j. Data refers to mainland Tanzania only. k. Estimated to be low income ($825 or less).

294

Table 2. Millennium Development Goals: eradicating poverty and improving livesAchieve Combat universal Promote Reduce HIV/AIDS

Eradicate extreme primary gender child and other poverty and hunger education equality mortality diseases Improve maternal health

Proportion of Prevalence of Primary Gender parity Under-five HIV Maternal Births attended population child malnutrition completion ratio in mortality prevalence mortality by skilled below $1 % of children rate (%) primary and rate per % of rate per health staff

(PPP) a day under 5 secondary 1,000 population 100,000 % of total% school ages 15–49 live births

(%) Modeled estimates

1988/89– 2000/01–Survey year 1989–94a 2000–3a 1993/94b 2003/04a 1990/91 2002/03 1990 2003 2003 2000 1990–2a 2000–3a

Albania 2002 c <2 .. 13.6 .. 101 96 100 45 21 .. 55 .. 94Algeria 1995 c <2 9.2 6.0 80 96 83 99 69 41 0.1 140 77 92Angola .. 20.0 30.5 39 .. .. .. 260 260 3.9 1,700 .. 45Argentina 2001 d 3.3 1.9 .. 100 103 .. 102 28 20 0.7 82 96 99Armenia 2003 c, e <2 .. 2.6 91 110 .. 101 60 33 .. 55 .. 97Australia .. .. .. .. .. 101 98 10 6 0.1 8 100 ..Austria .. .. .. .. 101 95 97 10 6 0.3 4 .. ..Azerbaijan 2001 c 3.7 .. 6.8 .. 106 100 97 105 91 <0.1 94 .. 84Bangladesh 2000 c 36.0 68.3 52.2 46 73 77 107 144 69 .. 380 .. 14Belarus 2000 c <2 .. .. 94 99 .. 102 17 17 .. 35 .. 100Belgium .. .. .. .. .. 101 106 9 5 0.2 10 .. ..Benin .. 35.0 22.9 22 51 48 66 185 154 1.9 850 .. 66Bolivia 1999 c 14.4 14.9 .. 71 101 90 98 120 66 0.1 420 .. 65Bosnia & Herzegovina .. .. 4.1 .. .. .. .. 22 17 <0.1 31 97 100Brazil 2001 d 8.2 7.0 .. 97 112 .. 103 60 35 0.7 260 72 ..Bulgaria 2003 c, e <2 .. .. 90 97 99 97 19 17 0.1 32 .. ..Burkina Faso 1998 c 44.9 32.7 37.7 19 29 61 72 210 207 1.8 f 1,000 .. ..Burundi 1998 c 54.6 37.5 45.1 47 31 82 79 190 190 6.0 1,000 .. 25Cambodia 1997 c 34.1 .. 45.2 .. 81 73 85 115 140 2.6 450 .. 32Cameroon 2001 c 17.1 15.1 .. 56 70 83 85 139 166 5.5 g 730 58 60Canada .. .. .. .. .. 99 100 8 7 0.3 6 .. ..Central African Rep. 1993 c 66.6 .. .. 27 .. 60 .. 180 180 13.5 1,100 .. 44Chad .. .. 28.0 19 25 41 59 203 200 4.8 1,100 .. 16Chile 2000 d <2 1.0 0.8 .. 104 101 100 19 9 0.3 31 .. 100China 2001 c 16.6 17.4 10.0 105 98 87 98 49 37 0.1 56 .. 97

Hong Kong, China .. .. .. 102 101 103 101 .. .. 0.1 .. .. ..Colombia 1999 d 8.2 10.1 6.7 71 88 114 104 36 21 0.7 130 82 86Congo, Dem. Rep. .. .. 31.0 47 32 .. .. 205 205 4.2 990 .. 61Congo, Rep. .. 23.9 .. 54 59 85 87 110 108 4.9 510 .. ..Costa Rica 2000 d 2.0 2.2 .. 72 94 100 101 17 10 0.6 43 98 98Côte d’Ivoire 2002 c, e 14.8 23.8 .. 46 51 66 69 157 192 7.0 690 .. 63Croatia 2001 c <2 0.7 .. 83 96 102 101 13 7 <0.1 8 .. ..Czech Rep. 1996 d <2 1.0 .. .. 106 98 101 13 5 0.1 9 .. ..Denmark .. .. .. 98 107 101 103 9 6 0.2 5 .. ..Dominican Rep. 1998 d <2 10.3 5.3 62 93 .. 108 65 35 1.7 150 93 98Ecuador 1998 d 17.7 16.5 .. 92 100 .. 100 57 27 0.3 130 .. ..Egypt, Arab Rep. 1999–2000 c 3.1 9.9 8.6 .. 91 81 94 104 39 <0.1 84 37 69El Salvador 2000 d 31.1 11.2 10.3 59 89 101 96 60 36 0.7 150 .. 69Eritrea .. 41.0 39.6 19 40 .. 76 147 85 2.7 630 .. 28Ethiopia 1999–2000 c 23.0 47.7 47.2 22 39 68 69 204 169 4.4 850 .. 6Finland .. .. .. 97 101 109 106 7 4 0.1 6 .. ..France .. .. .. 104 98 102 100 9 6 0.4 17 .. ..Georgia 2001 c 2.7 .. .. 81 82 98 100 47 45 0.1 32 .. ..Germany .. .. .. 101 101 99 99 9 5 0.1 8 .. ..Ghana 1998–99 c 44.8 27.3 22.1 61 62 77 91 125 95 3.1 540 .. ..Greece .. .. .. 100 .. 99 100 11 5 0.2 9 .. ..Guatemala 2000 d 16.0 33.2 22.7 .. 66 .. 93 82 47 1.1 240 .. 41Guinea .. .. .. 17 41 44 69 240 160 3.2 740 31 ..Haiti 2001 c, e 67.0 26.8 17.2 29 .. 95 .. 150 118 5.6 680 .. 24Honduras 1999 d 20.7 18.3 16.6 65 79 .. .. 59 41 1.8 110 45 56Hungary 2002 c <2 2.2 .. 82 102 100 100 17 7 0.1 16 .. ..India 1999–2000 c 35.3 53.2 .. 78 81 70 88 123 87 0.9 540 .. 43Indonesia 2002 c 7.5 39.9 27.3 93 95 93 98 91 41 0.1 230 32 68Iran, Islamic Rep. 1998 c <2 .. .. 101 107 85 95 72 39 0.1 76 .. 90Ireland .. .. .. .. .. 104 104 9 7 0.1 5 .. ..Israel .. .. .. .. .. 105 99 12 6 0.1 17 .. ..Italy .. .. .. 104 101 100 99 9 6 0.5 5 .. ..Jamaica 2000 c <2 4.6 .. 89 85 102 101 20 20 1.2 87 .. ..Japan .. .. .. 101 .. 101 100 6 5 <0.1 10 100 ..Jordan 2002 c <2 6.4 4.4 104 98 101 101 40 28 <0.1 41 87 100Kazakhstan 2003 c <2 .. .. .. 110 102 100 63 73 0.2 210 .. ..Kenya 1997 c 22.8 22.5 19.9 86 73 92 94 97 123 6.7 f 1,000 .. 41Korea, Rep. 1998 d <2 .. .. 98 97 99 100 9 5 <0.1 20 98 ..Kuwait .. .. .. 53 96 97 104 16 9 .. 5 .. ..Kyrgyz Rep. 2002 c <2 .. 5.8 .. 93 .. 100 80 68 0.1 110 .. ..Lao PDR 1997–98 c 26.3 40.0 40.0 46 74 75 83 163 91 0.1 650 .. 19Latvia 1998 c <2 .. .. 73 101 100 99 18 12 0.6 42 .. ..Lebanon .. .. .. .. 68 .. 102 37 31 0.1 150 .. ..Lithuania 2000 c <2 .. .. 89 102 .. 99 14 11 0.1 13 .. ..Macedonia, FYR 2003 c <2 .. .. 99 100 99 99 33 11 <0.1 23 .. 98Madagascar 2001 c 61.0 45.2 33.1 35 47 98 .. 168 126 1.7 550 57 46Malawi 1997–98 c 41.7 27.6 25.4 36 71 81 92 241 178 14.2 1,800 55 61Malaysia 1997 d <2 22.4 .. 88 92 102 105 21 7 0.4 41 .. 97Mali 1994 c 72.3 30.6 33.2 12 40 58 71 250 220 1.7 h 1,200 .. 41Mauritania 2000 c 25.9 47.6 31.8 33 43 67 94 162 107 0.6 1,000 40 57


295

Table 2. Millennium Development Goals: eradicating poverty and improving lives—continuedAchieve Combat universal Promote Reduce HIV/AIDS

Eradicate extreme primary gender child and other poverty and hunger education equality mortality diseases Improve maternal health

Proportion of Prevalence of Primary Gender parity Under-five HIV Maternal Births attended population child malnutrition completion ratio in mortality prevalence mortality by skilled below $1 % of children rate (%) primary and rate per % of rate per health staff

(PPP) a day under 5 secondary 1,000 population 100,000 % of total% school ages 15–49 live births

(%) Modeled estimates

1988/89– 2000/01–Survey year 1989–94a 2000–3a 1993/94b 2003/04a 1990/91 2002/03 1990 2003 2003 2000 1990–2a 2000–3a

Mexico 2000 c 9.9 16.6 .. 88 99 98 103 46 28 0.3 83 .. ..Moldova 2001 c 21.8 .. .. 95 83 105 102 37 32 0.2 36 .. ..Mongolia 1998 c 27.0 12.3 12.7 .. 108 109 110 104 68 <0.1 110 .. 99Morocco 1999 c <2 9.5 .. 47 75 70 88 85 39 0.1 220 31 ..Mozambique 1996 c 37.9 .. .. 28 52 73 79 242 147 12.2 1,000 .. 48Namibia 1993 d 34.9 26.2 24.0 77 92 111 104 86 65 21.3 300 68 78Nepal 1995–96 c 39.1 .. 48.3 55 78 57 85 145 82 0.5 740 7 11Netherlands .. .. .. .. 98 97 98 9 6 0.2 16 .. ..New Zealand .. .. .. 98 96 100 103 11 6 0.1 7 .. ..Nicaragua 2001 c 45.1 11.0 9.6 44 75 112 104 68 38 0.2 230 .. 67Niger 1995 c 60.6 42.6 40.1 18 26 56 69 320 262 1.2 1,600 15 16Nigeria 2003 c, e 70.8 39.1 28.7 63 82 78 81 235 198 5.4 800 31 35Norway .. .. .. .. .. 102 101 9 5 0.1 16 .. ..Oman .. 24.3 .. 73 73 89 97 30 12 0.1 87 .. 95Pakistan 2001 c, e 17.0 40.0 35.0 .. .. .. 71 138 98 0.1 500 19 23Panama 2000 d 7.2 6.1 .. 86 98 99 100 34 24 0.9 160 .. ..Papua New Guinea .. .. .. 51 53 79 87 101 93 0.6 300 .. ..Paraguay 2002 d 16.4 3.7 .. 66 93 98 98 37 29 0.5 170 67 ..Peru 2000 d 18.1 10.7 7.1 .. 102 .. 97 80 34 0.5 410 .. 59Philippines 2000 c 15.5 29.6 .. 87 95 100 102 63 36 <0.1 200 .. 60Poland 2001 c <2 .. .. 96 98 101 97 19 7 0.1 13 .. ..Portugal 1994 d <2 .. .. 98 .. 103 102 15 5 0.4 5 .. ..Romania 2002 c <2 5.7 3.2 96 89 99 100 32 20 <0.1 49 .. ..Russian Federation 2002 c <2 4.2 5.5 95 93 104 .. 21 21 1.1 67 .. 99Rwanda 1999–2000 c 51.7 29.4 24.3 44 37 96 95 173 203 5.1 1,400 26 31Saudi Arabia .. .. .. 57 61 84 93 44 26 .. 23 .. ..Senegal 1995 c 22.3 22.2 22.7 45 48 68 87 148 137 0.8 690 .. 58Serbia & Montenegro .. .. 1.9 71 96 103 101 26 14 0.2 11 .. 99Sierra Leone 1989 c 57.0 28.7 27.2 .. 56 67 70 302 284 .. 2,000 .. 42Singapore .. .. 3.4 .. .. 95 .. 8 5 0.2 30 .. ..Slovak Rep. 1996 d <2 .. .. 96 99 .. 100 15 8 <0.1 3 .. ..Slovenia 1998 c <2 .. .. 97 95 .. 99 9 4 <0.1 17 100 ..South Africa 2000 c 10.7 .. .. 81 99 103 100 60 66 15.6 i 230 .. ..Spain .. .. .. .. .. 104 102 9 4 0.7 4 .. ..Sri Lanka 2002 c, e 5.6 37.7 .. 103 113 102 103 32 15 <0.1 92 .. 97Sudan .. 33.9 40.7 44 49 77 86 120 93 2.3 590 69 ..Sweden .. .. .. 96 101 102 111 7 4 0.1 2 .. ..Switzerland .. .. .. .. 99 97 96 9 6 0.4 7 .. ..Syrian Arab Rep. .. 12.1 6.9 99 88 85 93 44 18 <0.1 160 .. ..Tajikistan 2003 c 7.4 .. .. 100 100 .. 88 119 95 <0.1 100 .. 71Tanzania 1991 c 48.5 28.9 .. 46 58 96 .. 163 165 8.8 1,500 44 ..Thailand 2000 c <2 18.6 .. .. 86 95 97 40 26 1.5 44 .. 99Togo .. 24.6 .. 40 78 59 .. 152 140 4.1 570 .. 49Tunisia 2000 c <2 10.3 4.0 75 101 86 102 52 24 <0.1 120 .. 90Turkey 2002 c, e 4.8 10.4 .. .. 95 81 85 78 39 .. 70 .. ..Turkmenistan 1998 c 12.1 .. 12.0 .. .. .. .. 97 102 <0.1 31 .. 97Uganda .. 23.0 22.9 .. 63 77 96 160 140 4.1 880 .. 39Ukraine 1999 d 2.9 .. 3.2 93 98 .. 99 22 20 1.4 35 .. ..United Kingdom .. .. .. .. .. 98 116 10 7 0.2 13 .. ..United States .. 1.4 .. .. .. 100 100 11 8 0.6 17 .. ..Uruguay 2000 d <2 4.4 .. 95 92 .. 105 24 14 0.3 27 .. ..Uzbekistan 2000 c 17.3 .. 7.9 .. 103 94 98 79 69 0.1 24 .. 96Venezuela, RB 2000 d, e 9.9 4.5 4.4 81 90 105 104 27 21 0.7 96 .. 94Vietnam 2000 c .. 44.9 33.8 .. 95 .. 93 53 23 0.4 130 .. 85West Bank & Gaza .. .. .. .. 106 .. 1 .. .. .. .. .. 97Yemen, Rep. 1998 c 15.7 39.0 .. .. 66 .. 61 142 113 0.1 570 16 ..Zambia 1998 c 63.7 25.2 28.1 .. 69 .. 91 180 182 15.6 j 750 51 43Zimbabwe 1995–96 c 56.1 15.5 .. 96 81 96 95 80 126 24.6 1,100 .. ..World 29.3 t .. w .. w .. w 87 w 95 w 95 w 84 w 1.1 w 407 w .. w 57 wLow income 46.8 .. 65 71 74 87 148 119 2.1 689 .. 38Middle income 14.7 11.8 94 96 91 99 56 40 0.7 115 .. 86

Lower middle income 15.9 12.3 95 96 89 99 60 42 0.7 121 .. 85Upper middle income 9.2 .. 90 96 99 99 40 30 0.6 67 .. ..

Low & middle income 30.6 .. 81 84 84 94 103 85 1.2 444 .. 57East Asia & Pacific 20.6 15.3 97 97 89 98 59 41 0.2 116 .. 87Europe & Central Asia .. .. 94 k 95 k 98 .. 46 36 0.7 58 97 ..Latin America & Carib. 9.7 .. 88 96 .. 102 53 33 0.7 193 .. ..Middle East & N. Africa 13.0 .. 82 87 82 92 80 56 0.1 162 .. 80South Asia 53.2 .. 74 80 71 89 130 86 0.8 567 .. 36Sub-Saharan Africa 33.2 31.4 50 59 79 83 187 171 7.2 916 .. 39

High income .. .. .. .. 100 101 11 7 0.4 13 .. ..

a. Data are for the most recent year available. b. Data are for 1990 or closest year. c. Expenditure base. d. income base. e. Preliminary data. f. Survey data, 2003. g. Survey data, 2004. h. Surveydata, 2001. i. Survey data, 2002. j. Survey data, 2001/2002. k. Represent only 61% of the population.

296

Table 3. Economic activityValue added as % of GDP

Agricultural productivityAgr. Value added per

Gross domestic agricultural worker product 2000 dollars Agriculture Industry Services

External GDP Household General gov’t. Gross balance of implicit

Avg. final cons. final cons. capital goods and deflator $ annual expenditure expenditure formation services Avg. annual

millions % growth % of GDP % of GDP % of GDP % of GDP % growth2004 2000–4 1989–91 2001–3 2004 2004 2004 2004 2004 2004 2004 2000–4

Albania 7,590 6.0 770 1,354 25 19 56 88 10 25 –23 3.6Algeria 84,649 4.8 1,801 1,964 13 74 14 49 8 29 14 6.0Angola 20,108 8.1 207 161 9 65 27 71 .. a 12 17 95.3Argentina 151,501 –0.1 6,507 9,272 10 32 59 70 8 18 5 13.3Armenia 3,549 12.0 .. 2,646 25 39 36 83 10 25 –18 4.1Australia 631,256 3.3 20,601 26,957 3 26 71 60 18 25 –2 2.9Austria 290,109 1.2 11,153 24,456 2 32 66 57 19 23 2 1.8Azerbaijan 8,523 10.7 .. 1,026 13 54 32 63 12 49 –24 4.1Bangladesh 56,844 5.1 239 309 21 27 53 78 5 23 –7 4.5Belarus 22,849 6.7 .. 2,259 16 38 46 77 11 14 –2 41.7Belgium 349,830 1.2 19,687 38,431 1 26 72 55 23 20 3 1.9Benin 4,075 4.5 360 583 36 14 50 77 14 20 –11 3.2Bolivia 8,773 2.6 662 739 15 30 55 72 14 13 1 4.8Bosnia & Herzegovina 8,121 3.8 .. .. 15 32 53 91 23 21 –35 3.0Brazil 604,855 2.0 1,658 3,004 5 17 78 61 14 19 5 10.6Bulgaria 24,131 4.7 2,434 6,310 10 27 63 69 19 23 –11 4.0Burkina Faso 4,824 5.2 140 163 31 20 49 82 13 19 –14 3.0Burundi 657 2.7 119 104 51 20 29 98 8 11 –16 6.6Cambodia 4,597 5.6 .. 292 36 28 37 88 .. a 23 –11 1.5Cameroon 14,733 4.6 725 1,143 44 16 40 71 11 18 0 2.7Canada 979,764 2.5 27,739 36,702 .. .. .. 56 19 20 4 2.2Central African Rep. 1,331 –1.4 291 407 61 25 14 87 10 7 –4 2.0Chad 4,285 14.3 164 220 61 9 30 53 5 25 18 5.7Chile 94,105 3.4 4,775 6,177 9 34 57 58 12 23 7 5.5China 1,649,329 8.7 242 357 15 51 35 42 12 45 1 2.0

Hong Kong, China 163,005 3.2 .. .. 0 12 88 59 10 22 9 –3.9Colombia 97,384 2.9 3,315 2,900 13 0 87 67 21 15 –2 7.0Congo, Dem. Rep. 6,571 3.5 230 196 58 19 23 92 4 7 –3 55.5Congo, Rep. 4,384 3.4 319 329 6 56 39 36 16 23 26 –2.9Costa Rica 18,395 3.9 3,039 4,306 9 29 63 71 10 21 –2 9.0Côte d’Ivoire 15,286 –1.5 610 806 25 19 55 73 9 8 10 3.3Croatia 34,200 4.5 .. 8,956 8 29 63 57 20 28 –5 3.3Czech Rep. 107,047 2.9 .. 4,300 3 39 57 50 23 28 0 3.1Denmark 243,043 1.2 18,564 36,320 2 26 71 47 26 20 7 1.9Dominican Rep. 18,673 2.4 2,273 4,076 11 31 58 73 5 21 1 20.6Ecuador 30,282 4.2 1,969 1,441 7 30 63 64 11 22 3 12.0Egypt, Arab Rep. 75,148 3.5 1,497 1,952 15 32 52 75 10 17 –2 4.3El Salvador 15,824 1.9 1,571 1,613 9 33 58 86 12 17 –14 2.7Eritrea 925 3.3 .. 64 15 24 61 97 54 22 –73 15.8Ethiopia 8,077 3.7 .. 123 46 10 44 77 22 20 –19 2.3Finland 186,597 2.2 16,056 30,391 3 31 66 52 22 18 7 1.3France 2,002,582 1.4 20,265 38,647 3 24 73 55 24 19 1 1.7Georgia 5,091 7.6 .. 1,374 20 25 54 81 9 24 –15 5.0Germany 2,714,418 0.5 10,963 22,127 1 29 69 59 19 18 4 1.2Ghana 8,620 4.8 315 338 35 22 43 80 12 27 –19 24.4Greece 203,401 4.1 7,579 9,226 7 24 69 67 15 26 –8 3.5Guatemala 27,451 2.3 2,121 2,261 22 19 59 90 5 17 –12 7.2Guinea 3,508 2.9 171 225 25 37 38 86 6 11 –2 8.6Haiti 3,535 –1.0 802 469 28 17 55 98 5 23 –27 17.5Honduras 7,371 3.3 950 1,133 14 31 55 74 14 29 –17 7.3Hungary 99,712 3.5 2,247 4,041 4 31 65 69 11 24 –4 7.6India 691,876 6.2 341 397 22 26 52 67 11 23 –1 3.9Indonesia 257,641 4.6 477 556 17 46 38 65 8 23 4 7.9Iran, Islamic Rep. 162,709 6.2 1,799 2,354 11 41 48 49 14 36 1 19.3Ireland 183,560 5.4 .. .. 3 42 55 44 15 22 19 3.8Israel 117,548 0.9 .. .. .. .. .. 59 29 18 –6 1.8Italy 1,672,302 0.8 11,411 21,436 3 28 70 60 19 20 1 2.8Jamaica 8,030 1.7 1,910 1,937 5 29 66 71 16 32 –20 9.9Japan 4,623,398 1.3 19,163 25,339 1 30 68 57 18 24 2 –1.9Jordan 11,196 5.1 1,456 960 2 25 73 81 20 21 –22 1.7Kazakhstan 40,743 10.3 .. 1,385 7 39 53 58 11 25 6 9.2Kenya 15,600 1.5 184 148 16 19 65 79 17 12 –8 9.9Korea, Rep. 679,674 4.7 5,312 9,888 3 35 62 55 13 29 3 2.9Kuwait 41,748 2.4 .. .. .. .. .. 50 26 9 16 0.6Kyrgyz Rep. 2,205 4.5 .. 929 39 23 38 71 17 16 –4 4.2Lao PDR 2,412 5.7 351 459 49 26 25 84 5 19 –8 11.3Latvia 13,629 7.5 .. 2,385 4 25 71 63 21 29 –13 3.9Lebanon 21,768 4.4 .. 24,371 13 19 68 82 17 21 –20 2.7Lithuania 22,263 7.5 .. 4,071 7 33 60 67 16 23 –7 0.4Macedonia, FYR 5,246 0.8 .. 2,935 12 28 60 83 11 22 –17 2.3Madagascar 4,364 0.9 187 176 29 16 55 81 9 24 –15 9.6Malawi 1,813 1.8 77 130 39 15 46 88 15 11 –15 15.0Malaysia 117,776 4.3 3,694 4,571 10 48 42 45 14 21 21 2.8Mali 4,863 6.3 203 227 38 26 36 78 10 20 –8 4.7Mauritania 1,357 5.3 244 278 19 30 51 85 18 17 –20 6.8


297

Table 3. Economic activity—continuedValue added as % of GDP

Agricultural productivityAgr. Value added per

Gross domestic agricultural worker product 2000 dollars Agriculture Industry Services

External GDP Household General gov’t. Gross balance of implicit

Avg. final cons. final cons. capital goods and deflator Millions annual expenditure expenditure formation services Avg. annual

of dollars % growth % of GDP % of GDP % of GDP % of GDP % growth2004 2000–4 1989–91 2001–3 2004 2004 2004 2004 2004 2004 2004 2000–4

Mexico 676,497 1.5 2,224 2,708 4 25 71 68 12 22 –2 7.0Moldova 2,595 6.9 .. 726 23 21 55 97 15 21 –32 11.5Mongolia 1,525 5.2 1,003 694 26 14 60 53 19 38 –10 9.5Morocco 50,055 4.5 1,580 1,515 17 30 53 65 18 24 –6 1.0Mozambique 5,548 8.5 117 136 26 31 43 76 11 22 –9 12.0Namibia 5,456 3.2 792 1,003 11 26 64 56 29 23 –7 6.6Nepal 6,707 2.6 196 208 40 23 37 76 10 26 –12 3.9Netherlands 577,260 0.3 23,496 38,085 3 26 72 50 25 21 5 3.2New Zealand 99,687 3.9 19,930 26,526 .. .. .. 60 18 21 1 2.6Nicaragua 4,353 2.3 1,167 1,934 18 25 57 74 16 36 –26 5.8Niger 3,081 4.1 174 172 40 17 43 82 12 16 –10 1.9Nigeria 72,106 4.9 576 836 26 49 24 40 22 21 17 15.7Norway 250,168 1.7 19,055 30,854 1 38 61 46 23 18 14 1.4Oman 21,698 3.5 .. .. .. .. .. 44 22 16 18 –0.6Pakistan 96,115 4.1 563 690 23 24 54 73 9 18 0 5.1Panama 13,793 3.3 2,320 3,470 7 17 76 64 6 27 3 1.0Papua New Guinea 3,909 0.6 390 434 26 39 35 56 14 18 12 7.3Paraguay 7,127 1.2 2,201 2,380 27 24 49 79 7 18 –4 11.9Peru 68,395 3.6 1,196 1,734 10 30 60 70 10 19 2 2.2Philippines 86,429 4.2 910 1,016 14 32 54 73 10 17 0 5.0Poland 241,833 2.8 .. 1,358 3 31 66 64 18 20 –2 1.9Portugal 168,281 0.3 3,807 5,444 4 29 68 61 21 25 –7 3.6Romania 73,167 5.5 2,079 3,430 13 40 47 67 11 22 –1 23.7Russian Federation 582,395 6.1 .. 2,204 5 34 61 51 19 21 9 15.8Rwanda 1,845 5.1 179 222 42 22 36 84 13 21 –18 5.1Saudi Arabia 250,557 3.4 7,270 13,964 5 55 40 30 23 19 28 3.9Senegal 7,665 4.6 270 260 17 21 62 76 14 21 –11 1.9Serbia & Montenegro 23,996 4.5 .. .. .. .. .. 92 18 18 –29 29.6Sierra Leone 1,075 15.8 .. .. 53 30 17 83 13 20 –16 4.7Singapore 106,818 2.8 25,523 32,980 0 35 65 41 11 18 30 0.5Slovak Rep. 41,092 4.6 .. .. 3 29 68 56 20 26 –3 4.3Slovenia 32,182 3.2 .. 30,243 3 36 61 54 20 27 –1 6.4South Africa 212,777 3.2 1,992 2,359 4 31 65 63 20 18 0 7.1Spain 991,442 2.5 8,740 14,852 3 30 67 58 18 26 –2 4.3Sri Lanka 20,055 3.8 696 737 17 25 58 76 8 25 –9 8.4Sudan 19,559 6.0 308 613 39 18 43 71 12 20 –3 8.3Sweden 346,404 2.0 20,416 30,469 2 28 70 49 28 16 7 1.7Switzerland 359,465 0.5 .. .. .. .. .. 61 12 20 7 1.2Syrian Arab Rep. 23,133 3.1 2,065 2,799 24 28 47 60 10 23 7 3.2Tajikistan 2,078 9.9 .. 412 24 21 55 101 .. a 9 –10 23.8Tanzania b 10,851 6.8 246 283 45 16 39 78 13 19 –10 5.9Thailand 163,491 5.3 493 588 10 44 46 57 11 27 5 2.1Togo 2,061 2.6 356 404 41 23 36 86 10 18 –13 0.7Tunisia 28,185 4.3 2,144 2,438 13 28 60 65 14 25 –4 2.5Turkey 301,950 4.2 1,749 1,764 12 27 61 65 13 26 –4 31.9Turkmenistan 6,167 18.5 .. 1,253 .. .. .. 51 14 27 8 7.0Uganda 6,833 5.8 187 230 32 21 47 76 16 22 –14 4.0Ukraine 65,149 8.6 .. 1,442 14 40 46 55 19 19 7 9.0United Kingdom 2,140,898 2.2 21,655 25,609 1 27 72 66 21 16 –3 3.0United States 11,667,515 2.6 26,105 47,566 .. .. .. 71 15 18 –4 1.9Uruguay 13,138 –1.2 5,346 6,632 13 27 60 71 12 15 2 13.4Uzbekistan 11,960 4.8 .. 1,520 35 22 43 55 18 18 8 33.1Venezuela, RB 109,322 –1.3 5,016 6,153 4 41 54 50 13 21 16 27.7Vietnam 45,210 7.2 212 290 22 40 38 66 7 35 –8 5.1West Bank & Gaza 3,454 –13.3 .. .. 6 12 82 84 53 3 –39 10.9Yemen, Rep. 12,834 3.6 361 504 15 40 45 78 13 17 –8 8.0Zambia 5,389 4.4 188 205 21 35 44 68 13 25 –6 20.8Zimbabwe 17,750 –7.0 260 277 17 24 59 72 17 8 2 87.9World 40,887,837 t 2.5 w .. w 817 w .. w .. w .. w 62 w 17 w 21 w 0 wLow income 1,253,353 5.4 320 375 23 25 52 69 12 22 –3Middle income 6,930,704 4.4 .. 699 10 34 56 58 13 27 2

Lower middle income 3,941,575 5.7 413 567 12 37 51 55 13 31 1Upper middle income 2,988,438 2.7 .. 2,664 7 30 64 62 14 21 3

Low & middle income 8,183,030 4.6 434 556 12 33 55 59 13 26 1East Asia & Pacific 2,367,508 7.5 .. 398 15 49 36 47 12 39 2Europe & Central Asia 1,768,088 5.0 .. 1,856 8 31 61 60 17 23 –1Latin America & Carib. 2,018,715 1.5 2,174 2,837 7 23 70 65 12 20 4Middle East & N. Africa 600,256 4.5 .. .. 14 39 47 62 12 26 –1South Asia 878,785 5.8 344 406 22 26 52 69 10 22 –3Sub-Saharan Africa 543,990 3.9 312 326 13 28 58 65 18 19 0

High income 32,715,777 2.0 .. .. .. .. .. 63 18 20 0

Note: a. Data on general government final consumption expenditure are not available; they are included in household final consumption expenditure. b. Data covers mainland Tanzania only.

298

Table 4. Trade, aid, and financeMerchandise trade

Exports Imports External debt

High Official Manufactured technology Current Net private Foreign development

exports exports account capital direct assistance Domestic credit % of total % of balance flows investment or official aida Total Present provided by Net

$ $ merchandise manufactured $ $ $ $ $ value banking sector migration millions millions exports exports millions millions millions per capita millions % of GNI % of GDP thousands

2004 2004 2003 2003 2004 2003 2003 2003 2003 2003 2004 1995–2000

Albania 580 2,150 84 1 –407 176 178 108 1,482 21 45.7 –267Algeria 31,713 18,199 2 2 .. 593 634 7 23,386 40 24.8 –185Angola 14,440 4,960 .. .. 1,178 1,903 1,415 37 9,698 100 4.5 –120Argentina 34,320 22,309 27 9 3,029 1,169 1,020 3 166,207 115 45.4 –100Armenia 715 1,351 62 1 –167 115 121 81 1,127 29 6.6 –225Australia 86,582 107,763 30 14 –39,542 .. 7,032 .. .. 110.0 510Austria 115,657 115,072 78 13 988 .. 7,276 .. .. 122.7 45Azerbaijan 3,600 3,500 6 5 –2,021 3,235 3,285 36 1,680 23 11.2 –128Bangladesh 8,150 12,100 89 0 132 86 102 10 18,778 25 40.7 –300Belarus 11,093 16,343 62 4 –1,043 127 172 3 2,692 18 21.2 14Belgium 308,854 287,236 80 b 8 b .. .. 125,060 b .. .. .. 112.2 99Benin 600 770 8 2 –143 51 51 44 1,828 28 c 9.9 –29Bolivia 2,092 1,772 17 8 36 295 167 105 5,684 37 c 52.2 –100Bosnia & Herzegovina 1,784 5,890 .. .. –1,917 400 382 141 2,920 37 45.7 350Brazil 96,474 65,904 52 12 11,669 13,432 10,144 2 235,431 54 80.9 –130Bulgaria 9,888 14,378 66 4 –1,813 1,655 1,419 53 13,289 86 36.2 –50Burkina Faso 380 1,150 17 2 –449 11 11 37 1,844 19 c 13.5 –121Burundi 42 180 2 22 –100 8 0 31 1,310 150 36.6 –400Cambodia 2,455 2,985 1 .. –125 87 87 38 3,139 71 8.7 100Cameroon 2,630 2,100 7 2 .. 154 215 55 9,189 52 c 14.9 0Canada 321,967 275,799 61 14 25,870 .. 6,273 .. .. 96.8 733Central African Rep. 115 145 37 0 .. 4 4 13 1,328 155 16.4 11Chad 1,820 780 .. .. .. 837 837 29 1,499 45 c 7.7 99Chile 32,000 24,823 16 3 1,390 3,844 2,982 5 43,231 67 70.2 60China 593,369 561,423 91 27 45,875 59,455 53,505 1 193,567 15 166.9 –1,950

Hong Kong, China 265,670 d 273,010 93 d 13 16,039 .. 13,624 1 .. .. 149.3 300Colombia 16,090 16,530 36 7 –1,110 –1,185 1,746 18 32,979 46 34.2 –200Congo, Dem. Rep. 1,600 1,940 10 .. .. 187 158 101 11,170 149 1.3 –1,410Congo, Rep. 3,150 1,570 .. .. –3 201 201 19 5,516 368 11.8 42Costa Rica 6,301 8,268 66 45 –967 842 577 7 5,424 36 42.5 128Côte d’Ivoire 5,500 3,650 20 8 –305 69 180 15 12,187 89 18.8 150Croatia 8,022 16,583 72 12 –1,668 8,031 1,998 27 23,452 102 68.4 –150Czech Rep. 66,008 67,876 90 13 –5,661 5,342 2,514 26 34,630 48 45.7 52Denmark 75,565 67,200 66 20 6,963 .. 1,185 .. .. 165.9 84Dominican Rep. 5,660 7,660 34 1 867 1,112 310 8 6,291 33 36.2 –180Ecuador 7,538 7,861 12 6 –455 2,143 1,555 14 16,864 82 20.1 –300Egypt, Arab Rep. 7,682 12,831 31 0 3,743 –361 237 13 31,383 31 116.2 –500El Salvador 3,295 6,269 57 5 –612 406 89 29 7,080 56 49.2 –38Eritrea 20 670 .. .. –78 22 22 70 635 57 148.2 –9Ethiopia 650 3,300 11 0 –65 54 60 22 7,151 24 c 4.0 –77Finland 61,144 51,043 84 24 7,810 .. 3,436 .. .. 69.5 20France 451,034 464,090 81 19 –4,833 .. 43,068 .. .. 107.2 219Georgia 649 1,847 31 24 –349 320 338 48 1,935 44 18.8 –350Germany 914,839 717,491 84 16 104,301 .. 25,568 .. .. 142.9 1,134Ghana 2,830 3,910 16 3 352 –166 137 44 7,957 38 c 31.4 –51Greece 14,760 53,082 58 12 –11,225 .. 717 .. .. 105.1 300Guatemala 2,792 7,420 40 7 –1,051 68 116 20 4,981 21 15.1 –390Guinea 640 700 25 0 –245 79 79 30 3,457 59 c 15.5 –227Haiti 362 1,301 .. .. –13 8 8 24 1,308 29 31.7 –105Honduras 1,560 3,890 21 0 –279 140 198 56 5,641 55 37.4 –20Hungary 54,175 59,216 87 26 –8,819 5,149 2,506 25 45,785 70 59.6 100India 72,530 95,156 77 5 6,853 10,651 4,269 1 113,467 19 59.9 –1,400Indonesia 69,710 46,180 52 14 .. –3,685 –597 8 134,389 71 48.8 –900Iran, Islamic Rep. 42,450 32,700 8 2 .. 1,151 120 2 11,601 8 9.7 –456Ireland 104,100 60,118 86 34 –748 .. 26,599 .. .. 118.4 89Israel 36,874 43,425 93 18 504 .. 3,880 66 .. .. 82.8 276Italy 346,060 349,049 87 8 –20,556 .. 16,538 .. .. 105.3 600Jamaica 1,385 3,641 64 0 –761 513 721 1 5,584 86 31.3 –100Japan 565,490 454,530 93 24 172,059 .. 6,238 .. .. 154.8 280Jordan 3,970 7,892 69 2 –44 –161 376 233 8,337 82 94.1 35Kazakhstan 20,251 13,300 18 9 533 5,674 2,088 18 22,835 94 18.5 –1,320Kenya 2,650 4,660 24 4 –847 195 82 15 6,766 43 40.8 –21Korea, Rep. 253,910 224,440 93 32 27,613 .. 3,222 –10 .. .. 100.8 –80Kuwait 27,390 11,630 7 1 18,884 .. –67 2 .. .. 106.0 347Kyrgyz Rep. 719 941 39 2 –95 –12 46 39 2,021 98 8.4 –27Lao PDR 455 655 .. .. .. 19 19 53 2,846 91 9.6 –7Latvia 3,882 6,898 60 4 –1,673 570 300 49 8,803 92 54.5 –56Lebanon 1,749 9,338 68 2 –4,109 394 358 51 18,598 104 179.0 –30Lithuania 9,111 12,362 63 5 –1,590 –141 179 108 8,342 58 30.0 –109Macedonia, FYR 1,637 2,856 72 1 –279 90 95 114 1,837 40 22.1 –5Madagascar 990 1,260 38 0 –309 13 13 32 4,958 31 c 15.0 –3Malawi 470 745 12 1 –185 23 23 45 3,134 109 c 23.2 –50Malaysia 126,497 105,176 77 58 13,381 2,207 2,473 4 49,074 56 134.3 390Mali 1,140 1,200 40 8 –271 129 129 45 3,129 42 c 17.7 –284Mauritania 365 400 21 .. .. 218 214 85 2,360 73 c –6.7 10


299

Table 4. Trade, aid, and finance—continuedMerchandise trade

Exports Imports External debt

High Official Manufactured technology Current Net private Foreign development

exports exports account capital direct assistance Domestic credit % of total % of balance flows investment or official aida Total Present provided by Net

$ $ merchandise manufactured $ $ $ $ $ value banking sector migration millions millions exports exports millions millions millions per capita millions % of GNI % of GDP thousands

2004 2004 2003 2003 2004 2003 2003 2003 2003 2003 2004 1995–2000

Mexico 188,627 206,423 81 21 –7,798 9,541 10,783 1 140,004 25 34.9 –2,000Moldova 986 1,774 32 3 –132 84 58 28 1,901 95 32.0 –70Mongolia 858 988 38 0 –105 131 132 100 1,472 97 36.8 –90Morocco 9,661 17,514 69 11 1,434 2,395 2,279 17 18,795 47 82.5 –300Mozambique 1,390 1,765 8 3 –516 313 337 55 4,930 38 c 5.9 75Namibia 1,830 2,450 41 3 337 .. .. 73 .. .. 55.8 20Nepal 756 1,877 .. .. 171 14 15 19 3,253 38 .. –99Netherlands 358,781 319,864 71 31 16,403 .. 15,695 .. .. 166.9 161New Zealand 20,358 23,186 29 10 –6,232 .. 2,438 .. .. 120.6 20Nicaragua 771 1,884 13 4 –780 230 201 152 6,915 40 c 88.4 –155Niger 430 560 8 3 .. 23 31 39 2,116 26 c 11.4 –6Nigeria 31,148 14,164 .. .. .. 952 1,200 2 34,963 76 13.2 –95Norway 82,018 48,203 21 19 34,445 .. 2,055 .. .. 11.1 67Oman 14,236 7,865 14 2 1,446 –557 138 17 3,886 19 38.1 –40Pakistan 13,326 17,908 85 1 –808 132 534 7 36,345 41 40.1 –41Panama 950 3,466 11 1 –1,104 1,077 792 10 8,770 92 90.4 11Papua New Guinea 2,460 1,670 6 39 .. 2 101 40 2,463 81 23.4 0Paraguay 1,626 2,916 14 6 76 121 91 9 3,210 51 18.5 –25Peru 12,467 9,880 22 2 –72 2,562 1,377 18 29,857 60 17.4 –350Philippines 39,598 42,635 90 74 3,347 1,350 319 9 62,663 81 54.0 –900Poland 74,094 87,849 81 3 –3,585 7,118 4,123 31 95,219 48 34.6 –71Portugal 34,983 53,776 86 9 –12,682 .. 6,610 .. .. 151.1 175Romania 23,553 32,691 83 4 –3,311 3,880 1,844 28 21,280 46 15.3 –350Russian Federation 183,185 94,834 21 19 60,109 15,784 7,958 9 175,257 52 26.0 2,300Rwanda 80 250 10 25 –76 5 5 39 1,540 58 c 13.5 1,977Saudi Arabia 119,550 42,954 10 0 51,488 .. –587 1 .. .. 64.2 75Senegal 1,530 2,680 34 9 –507 79 78 44 4,419 36 c 21.7 –100Serbia & Montenegro 3,408 11,194 .. .. –3,148 1,462 1,360 162 14,885 e 84 .. –100Sierra Leone 140 285 7 31 –65 3 3 56 1,612 100 c 30.3 –110Singapore 179,547 d 163,820 85 d 59 28,183 .. 11,431 2 .. .. 80.2 368Slovak Rep. 27,660 29,448 88 4 –282 1,525 571 30 18,379 69 44.0 9Slovenia 15,805 17,297 90 6 –275 .. 337 33 .. .. 55.7 8South Africa 45,929 f 55,200 f 58 f 5 –6,982 4,148 820 14 27,807 22 84.5 364Spain 178,960 249,813 77 7 –49,225 .. 25,513 .. .. 138.7 676Sri Lanka 5,800 7,950 74 1 –131 236 229 35 10,238 50 44.6 –160Sudan 3,777 4,075 3 7 –818 1,349 1,349 19 17,496 120 11.5 –207Sweden 121,012 97,644 81 15 22,844 .. 3,268 .. .. 113.1 60Switzerland 118,384 111,468 93 22 50,568 .. 17,547 .. .. 175.2 80Syrian Arab Rep. 6,435 5,320 11 1 752 146 150 9 21,566 113 30.1 –30Tajikistan 915 1,375 .. .. –40 6 32 23 1,166 77 16.5 –345Tanzania 1,440 2,535 18 2 –1,062 264 248 47 7,516 22 c,g 9.2 –206Thailand 97,701 95,384 75 30 7,281 1,155 1,949 –16 51,793 41 105.4 –88Togo 720 930 58 1 –140 20 20 9 1,707 91 16.7 128Tunisia 9,685 12,738 81 4 –715 1,326 541 31 15,502 75 71.0 –20Turkey 62,774 97,161 84 2 –15,451 2,849 1,562 2 145,662 81 0.0 135Turkmenistan 3,870 3,320 .. .. 444 .. 100 6 .. 0 .. –50Uganda 705 1,480 9 8 –250 202 194 38 4,553 33 c 11.0 –66Ukraine 32,672 28,996 67 5 2,891 1,550 1,424 7 16,309 37 30.7 –700United Kingdom 345,610 461,983 78 26 –46,879 .. 20,696 .. .. 157.9 574United States 819,026 1,526,380 80 31 –665,939 .. 39,889 .. .. 270.8 6,200Uruguay 2,905 3,072 34 2 103 37 275 5 11,764 91 53.3 –16Uzbekistan 4,238 3,310 .. .. 1,134 79 70 8 5,006 46 .. –400Venezuela, RB 31,360 17,300 13 4 14,575 3,539 2,520 3 34,851 43 10.8 40Vietnam 26,229 31,029 50 2 –604 1,192 1,450 22 15,817 39 61.0 –200West Bank & Gaza .. .. .. .. .. .. .. 289 .. .. .. 11Yemen, Rep. 4,555 3,790 .. .. –296 –89 –89 13 5,377 40 5.2 –50Zambia 1,410 1,670 14 2 .. 91 100 54 6,425 121 35.3 86Zimbabwe 1,250 2,990 38 3 .. –5 20 14 4,445 50 58.7 –125World 9,122,837 t 9,338,667 t 77 w 18 w .. s 572,774 s 12 w .. s 171.1 w .. w,iLow income 215,695 251,818 60 4 18,208 13,283 14 414,454 47.1 –4,422Middle income 2,244,720 2,138,024 64 20 181,237 138,493 9 2,139,684 76.4 –9,689

Lower middle income 1,223,079 1,170,291 68 22 103,824 90,627 8 1,053,736 104.4 –10,646Upper middle income 1,021,641 967,734 61 19 77,412 47,867 10 1,085,948 40.7 957

Low & middle income 2,460,424 2,389,837 64 19 199,444 151,776 14 2,554,138 72.1 –14,111East Asia & Pacific 964,989 895,174 81 33 62,049 59,612 4 525,535 140.7 –3,859Europe & Central Asia 615,333 j 626,097 j 57 12 67,110 35,614 22 675,998 27.2 –1,858Latin America & Carib. 458,500 437,379 57 14 41,087 36,533 12 779,632 49.7 –4,156Middle East & N. Africa 170,996 153,367 20 3 4,848 4,756 26 158,827 49.0 –1,396South Asia 101,332 138,464 79 4 11,143 5,163 4 182,785 56.1 –2,401Sub-Saharan Africa 149,265 139,357 .. .. 13,208 10,099 34 231,360 45.4 –439

High income 6,662,445 6,948,809 80 18 .. 420,998 .. 205.5 14,104

Note: a. Regional aggregates include data for economies that are not specified elsewhere. World and income group totals include aid not allocated by country or region. b. IncludesLuxembourg. c. Data are from debt sustainability analysis undertaken as part of the Heavily Indebted Poor Countries (HIPC) initiative. d. Includes re-exports. e. Data are estimates and reflectborrowing by the former Socialist Federal Republic of Yugoslavia that are not yet allocated to the successor republics. f. Data on total exports and imports refer to South Africa only. Data onexport commodity shares refer to the South African Customs Union (Botswana, Lesotho, Namibia, South Africa, and Swaziland). g. GNI refers to mainaland Tanzania only. i. World totals com-puted by the UN sum to zero, but because the aggregates shown here refer to World Bank definitions, regional and income group totals do not equal zero. j. Data include the intratrade of theBaltic states and the Commonwealth of Independent States.

300

Table 5. Key indicators for other economiesPPP Life

Gross national gross national expectancy Population income (GNI)a income (GNI)b at birth

Gross Adult Carbon Density domestic literacy dioxide

Avg. people $ $ product rate emissionsannual per $ per $ per per capita Male Female % ages 15 per capita

Thousands % growth sq. km millions capita millions capita % growth years years and older metric tons 2004 2000–4 2004 2004 2004 2004 2004 2003–4 2003 2003 1998–2004 2000

Afghanistan .. .. .. 5,543 .. c .. .. .. .. .. .. ..American Samoa 57 .. 285 .. .. d .. .. .. .. .. .. ..Andorra 66 .. 136 .. .. e .. .. .. .. .. .. ..Antigua and Barbuda 80 2.5 182 800 10,000 829 10,360 2.3 73 78 .. 4.9Aruba 99 .. 521 .. .. e .. .. .. .. .. .. ..Bahamas, The 320 1.2 32 4,684 14,920 5,068 16,140 –0.6 66 74 .. 5.9Bahrain 725 2.0 1,022 8,834 12,410 12,860 18,070 4.7 71 76 88 f 29.1Barbados 272 0.4 632 2,507 9,270 4,075 15,060 0.9 72 77 100 4.4Belize 283 3.1 12 1,115 3,940 1,840 6,510 0.9 70 73 77 f 3.1Bermuda 64 0.0 1,280 .. .. e .. .. .. 75 80 .. 7.2Bhutan 896 2.7 19 677 760 .. .. 2.3 62 65 .. 0.5Botswana 1,727 0.8 3 7,490 4,340 15,405 8,920 4.3 38 38 79 2.3Brunei 361 1.7 69 .. .. e .. .. .. 74 79 93 f 14.2Cape Verde 481 2.5 119 852 1,770 2,720 g 5,650 g 2.9 66 72 76 0.3Cayman Islands 44 .. 745 .. .. e .. .. .. .. .. .. ..Channel Islands 149 0.0 745 .. .. e .. .. .. 75 84 .. ..Comoros 614 2.4 276 328 530 1,131 g 1,840 g –0.5 60 63 56 0.1Cuba 11,365 0.4 103 .. .. h .. .. 0.9 75 79 97 2.8Cyprus 776 0.6 84 13,633 17,580 17,320 g 22,330 g 2.9 76 81 97 f 8.5Djibouti 716 1.8 31 739 1,030 1,624 g 2,270 g 1.6 43 43 .. 0.6Dominica 71 0.0 95 261 3,650 375 5,250 1.6 75 79 .. 1.4Equatorial Guinea 506 2.5 18 .. .. d 3,745 7,400 7.4 50 54 84 i 0.4Estonia 1,345 –0.5 32 9,435 7,010 17,741 13,190 6.8 65 77 100 f 11.7Faeroe Islands 48 .. 34 .. .. e .. .. .. .. .. .. ..Fiji 848 1.1 46 2,281 2,690 4,893 g 5,770 g 2.2 68 71 .. 0.9French Polynesia 246 1.1 67 .. .. e .. .. .. 71 77 .. 2.3Gabon 1,374 2.2 5 5,415 3,940 7,692 5,600 –0.2 52 54 .. 2.8Gambia, The 1,449 2.5 145 414 290 2,753 g 1,900 g 6.2 52 55 .. 0.2Greenland 57 0.4 0 .. .. e .. .. .. 65 73 .. 9.9Grenada 106 1.0 311 397 3,760 740 7,000 –3.8 70 76 .. 2.1Guam 164 1.5 298 .. .. e .. .. .. 76 80 .. 26.3Guinea-Bissau 1,533 2.9 55 250 160 1,058 690 1.3 44 47 .. 0.2Guyana 772 0.4 4 765 990 3,173 g 4,110 g 1.1 58 67 .. 2.1Iceland 290 0.8 125 11,199 38,620 9,384 32,360 4.8 78 82 .. 7.7Iraq 25,261 2.1 58 .. .. h .. .. .. 62 64 .. 3.3Isle of Man 77 .. 135 .. .. e .. .. .. .. .. .. ..Kiribati 98 1.9 134 95 970 .. .. 0.3 60 66 .. 0.3Korea, Dem. Rep. 22,745 0.5 189 .. .. c .. .. .. 61 65 .. 8.5Lesotho 1,809 0.9 60 1,336 740 5,806 3,210 2.1 36 38 81 i ..Liberia 3,449 2.4 171 391 110 .. .. –0.2 46 48 56 0.1Libya 5,674 2.0 3 25,257 4,450 .. .. 2.4 70 75 82 10.9Liechtenstein 34 .. 213 .. .. e .. .. .. .. .. .. ..Luxembourg 450 0.7 174 25,302 56,230 27,549 61,220 4.0 75 82 .. 19.4Macao, China 449 1.0 265 .. .. e 9,605 g 21,880 g 8.9 77 82 91 f 3.8Maldives 300 2.2 998 752 2,510 .. .. 6.5 68 71 97 1.8Malta 401 0.7 400 4,913 12,250 7,507 18,720 0.9 76 81 .. 7.2Marshall Islands 60 3.7 174 142 2,370 .. .. –3.6 .. .. .. ..Mauritius 1,234 1.0 16,842 5,730 4,640 14,650 11,870 3.2 69 76 84 f 2.4Mayotte 172 .. 460 .. .. d .. .. .. .. .. .. ..Micronesia, Fed. Sts. 127 1.8 181 252 1,990 .. .. –5.5 67 71 .. ..Monaco 33 .. 159 .. .. e .. .. .. .. .. .. ..Myanmar 49,910 1.2 76 .. .. c .. .. .. 55 60 90 i 0.2Northern Mariana Islands 77 .. 161 .. .. d .. .. .. .. .. .. ..Netherlands Antilles 222 0.8 277 .. .. e .. .. .. 73 79 97 46.2New Caledonia 229 1.8 13 .. .. e .. .. .. 70 78 .. 7.8Palau 20 1.2 43 137 6,870 .. .. 0.5 .. .. .. 12.7Puerto Rico 3,929 0.7 277 .. .. e .. .. .. 72 82 94 2.3Qatar 637 2.1 58 .. .. e .. .. .. 75 75 89 i 69.6Samoa 179 1.0 63 333 1,860 1,015 g 5,670 g 2.6 67 73 99 0.8San Marino 28 .. 463 653 .. e .. .. .. .. .. .. ..São Tomé & Principe 161 2.0 167 60 370 .. .. 2.4 63 69 .. 0.6Seychelles 85 1.1 188 685 8,090 1,320 15,590 –3.2 69 77 92 f 2.8Solomon Islands 471 2.9 17 260 550 829 g 1,760 g 0.7 68 71 .. 0.4Somalia 9,938 3.3 16 .. .. c .. .. .. 46 49 .. ..St. Kitts and Nevis 47 1.5 131 357 7,600 526 11,190 3.3 69 74 .. 2.4St. Lucia 164 1.2 268 706 4,310 910 5,560 1.6 72 76 90 f 2.1St. Vincent & the Grenadines 108 –0.8 278 396 3,650 677 6,250 4.8 70 76 .. 1.4Suriname 443 1.0 3 997 2,250 .. .. 3.5 68 73 88 i 5.0Swaziland 1,120 1.7 65 1,859 1,660 5,566 4,970 0.8 42 43 79 i 0.4Timor-Leste 925 4.3 62 506 550 .. .. –3.5 60 64 .. ..Tonga 102 0.4 141 186 1,830 735 g 7,220 g 1.3 69 74 .. 1.2Trinidad & Tobago 1,323 0.7 258 11,360 8,580 14,795 11,180 5.3 70 74 98 20.5United Arab Emirates 4,284 6.9 51 .. .. e 78,834 g 21,000 g –5.4 74 77 77 18.1Vanuatu 215 2.2 18 287 1,340 600 2,790 0.7 67 70 74 f 0.4Virgin Islands (U.S.) 113 1.0 333 .. .. e .. .. .. 77 80 .. 121.2

Note: For data comparability and coverage, see the technical notes. Figures in italics are for years other than those specified.a. Calculated using the World Bank Atlas method. b. PPP is purchasing power parity; see Definitions. c. Estimated to be low income ($825 or less). d. Estimated to be upper middle income($3,256–$10,065 ). e. Estimated to be high income ($10,066 or more). f. National estimate based on census data. g. The estimate is based on regression; others are extrapolated from the latestInternational Comparison Programme benchmark estimates. h. Estimated to be lower middle income ($826–$3,255 ). i. National estimates based on survey data.


Technical notesThese technical notes discuss the sources and methods usedto compile the indicators included in this edition of SelectedWorld Development Indicators. The notes follow the orderin which the indicators appear in the tables. Note that theSelected World Development Indicators uses terminology inline with the 1993 System of National Accounts (SNA). Forexample, in the 1993 SNA gross national income replacesgross national product.

SourcesThe data published in the Selected World DevelopmentIndicators are taken from World Development Indicators2005. Where possible, however, revisions reported since theclosing date of that edition have been incorporated. In addi-tion, newly released estimates of population and grossnational income (GNI) per capita for 2004 are included intable 1.

The World Bank draws on a variety of sources for the sta-tistics published in the World Development Indicators. Dataon external debt for developing countries are reporteddirectly to the World Bank by developing member countriesthrough the Debtor Reporting System. Other data are drawnmainly from the U.N. and its specialized agencies, from theIMF, and from country reports to the World Bank. Bank staffestimates are also used to improve currentness or consis-tency. For most countries, national accounts estimates areobtained from member governments through World Bankeconomic missions. In some instances these are adjusted bystaff to ensure conformity with international definitions andconcepts. Most social data from national sources are drawnfrom regular administrative files, special surveys, or periodiccensuses.

For more detailed notes about the data, please refer to theWorld Bank’s World Development Indicators 2005.

Data consistency and reliability Considerable effort has been made to standardize the data,but full comparability cannot be assured, and care must betaken in interpreting the indicators. Many factors affect dataavailability, comparability, and reliability: statistical systemsin many developing economies are still weak; statisticalmethods, coverage, practices, and definitions differ widely;and cross-country and intertemporal comparisons involvecomplex technical and conceptual problems that cannot beunequivocally resolved. Data coverage may not be completebecause of special circumstances or for economies experi-encing problems (such as those stemming from conflicts)affecting the collection and reporting of data. For these rea-sons, although the data are drawn from the sources thoughtto be most authoritative, they should be construed only asindicating trends and characterizing major differences

among economies rather than offering precise quantitativemeasures of those differences. Discrepancies in data pre-sented in different editions reflect updates by countries aswell as revisions to historical series and changes in method-ology. Thus readers are advised not to compare data seriesbetween editions or between different editions of WorldBank publications. Consistent time series are available fromthe World Development Indicators 2005 CD-ROM.

Ratios and growth rates For ease of reference, the tables usually show ratios and ratesof growth rather than the simple underlying values. Valuesin their original form are available from the World Develop-ment Indicators 2005 CD-ROM. Unless otherwise noted,growth rates are computed using the least-squares regressionmethod (see statistical methods on page 305). Because thismethod takes into account all available observations duringa period, the resulting growth rates reflect general trendsthat are not unduly influenced by exceptional values. Toexclude the effects of inflation, constant price economicindicators are used in calculating growth rates. Data in italicsare for a year or period other than that specified in the col-umn heading—up to two years before or after for economicindicators and up to three years for social indicators, becausethe latter tend to be collected less regularly and change lessdramatically over short periods.

Constant price series An economy’s growth is measured by the increase in valueadded produced by the individuals and enterprises operatingin that economy. Thus measuring real growth requires esti-mates of GDP and its components valued in constant prices.The World Bank collects constant price national accountsseries in national currencies and recorded in the country’soriginal base year. To obtain comparable series of constantprice data, it rescales GDP and value added by industrial ori-gin to a common reference year, currently 2000. This processgives rise to a discrepancy between the rescaled GDP and thesum of the rescaled components. Because allocating the dis-crepancy would give rise to distortions in the growth rate, itis left unallocated.

Summary measures The summary measures for regions and income groups, pre-sented at the end of most tables, are calculated by simpleaddition when they are expressed in levels. Aggregate growthrates and ratios are usually computed as weighted averages.The summary measures for social indicators are weighted bypopulation or subgroups of population, except for infantmortality, which is weighted by the number of births. See thenotes on specific indicators for more information.


For summary measures that cover many years, calcula-tions are based on a uniform group of economies so that thecomposition of the aggregate does not change over time.Group measures are compiled only if the data available for agiven year account for at least two-thirds of the full group, asdefined for the 2000 benchmark year. As long as this crite-rion is met, economies for which data are missing areassumed to behave like those that provide estimates. Readersshould keep in mind that the summary measures are esti-mates of representative aggregates for each topic and thatnothing meaningful can be deduced about behavior at thecountry level by working back from group indicators. Inaddition, the estimation process may result in discrepanciesbetween subgroup and overall totals.

Table 1. Key indicators of developmentPopulation is based on the de facto definition, which countsall residents, regardless of legal status or citizenship, exceptfor refugees not permanently settled in the country of asy-lum, who are generally considered part of the population ofthe country of origin.

Average annual population growth rate is the exponen-tial rate of change for the period (see the section on statisti-cal methods on page 305).

Population density is midyear population divided by landarea. Land area is a country’s total area excluding areas underinland bodies of water and coastal waterways. Density is cal-culated using the most recently available data on land area.

Gross national income (GNI—formerly gross nationalproduct or GNP), the broadest measure of national income,measures total value added from domestic and foreignsources claimed by residents. GNI comprises gross domesticproduct (GDP) plus net receipts of primary income fromforeign sources. Data are converted from national currencyto current U.S. dollars using the World Bank Atlas method.This involves using a three-year average of exchange rates tosmooth the effects of transitory exchange rate fluctuations.See the section on statistical methods for discussion of theAtlas method.

GNI per capita is GNI divided by midyear population. Itis converted into current U.S. dollars by the Atlas method.The World Bank uses GNI per capita in U.S dollars to classifyeconomies for analytical purposes and to determine borrow-ing eligibility.

PPP gross national income, which is GNI converted intointernational dollars using purchasing power parity (PPP)conversion factors, is included because nominal exchangerates do not always reflect international differences in rela-tive prices. At the PPP rate, one international dollar has thesame purchasing power over domestic GNI that the U.S. dol-lar has over U.S. GNI. PPP rates allow a standard compari-son of real price levels between countries, just as conven-tional price indexes allow comparison of real values over

time. The PPP conversion factors used here are derived fromprice surveys covering 118 countries conducted by the Inter-national Comparison Program. For Organisation for Eco-nomic Co-operation and Development (OECD) countriesdata come from the most recent round of surveys, com-pleted in 1999; the rest are either from the 1996 survey, ordata from the 1993 or earlier round and extrapolated to the1996 benchmark. Estimates for countries not included in thesurveys are derived from statistical models using availabledata.

PPP GNI per capita is PPP GNI divided by midyear pop-ulation.

Gross domestic product (GDP) per capita growth isbased on GDP measured in constant prices. Growth in GDPis considered a broad measure of the growth of an economy.GDP in constant prices can be estimated by measuring thetotal quantity of goods and services produced in a period,valuing them at an agreed set of base year prices, and sub-tracting the cost of intermediate inputs, also in constantprices. See the section on statistical methods for details ofthe least-squares growth rate.

Life expectancy at birth is the number of years a new-born infant would live if patterns of mortality prevailing atits birth were to stay the same throughout its life.

Adult literacy rate is the percentage of persons aged 15and above who can, with understanding, read and write ashort, simple statement about their everyday life.

Carbon dioxide (CO2) emissions measures those emis-sions stemming from the burning of fossil fuels and themanufacture of cement. These include carbon dioxide pro-duced during consumption of solid, liquid, and gas fuels andfrom gas flaring.

The Carbon Dioxide Information Analysis Center(CDIAC), sponsored by the U.S. Department of Energy, cal-culates annual anthropogenic emissions of CO2. These calcu-lations are derived from data on fossil fuel consumption,based on the World Energy Data Set maintained by theUNSD, and from data on world cement manufacturing,based on the Cement Manufacturing Data Set maintained bythe U.S. Bureau of Mines. Each year the CDIAC recalculatesthe entire time series from 1950 to the present, incorporatingits most recent findings and the latest corrections to its data-base. Fuels supplied to ships and aircraft engaged in interna-tional transportation are excluded in these estimates becauseof the difficulty of apportioning these fuels among the coun-tries benefiting from that transport.

Table 2. Millennium Development Goals:eradicating poverty and improving livesProportion of population below $1 a day (PPP$) is the per-centage of the population living on less than $1.08 a day at1993 international prices. For further information onpoverty data, see the technical note for Table A1.


Prevalence of child malnutrition is the percentage ofchildren under five whose weight for age is less than minustwo standard deviations from the median for the interna-tional reference population ages 0–59 months. The referencepopulation, adopted by the World Health Organization(WHO) in 1983, is based on children from the United States,who are assumed to be well nourished. Estimates of childmalnutrition are from national survey data. The proportionof children who are underweight is the most common indi-cator of malnutrition. Being underweight, even mildly,increases the risk of death and inhibits cognitive develop-ment in children. Moreover, it perpetuates the problem fromone generation to the next, as malnourished women aremore likely to have low-birth-weight babies.

Primary completion rate is the percentage of studentscompleting the last year of primary school. It is calculated bytaking the total number of students in the last grade of pri-mary school, minus the number of repeaters in that grade,divided by the total number of children of official graduationage. The primary completion rate reflects the primary cycle asdefined by the International Standard Classification of Educa-tion (ISCED), ranging from three or four years of primaryeducation (in a very small number of countries) to five or sixyears (in most countries), and seven (in a small number ofcountries). Because curricula and standards for school com-pletion vary across countries, a high rate of primary comple-tion does not necessarily mean high levels of student learning.

Gender parity ratio in primary and secondary school isthe ratio of female gross enrollment rate in primary and sec-ondary school to the gross enrollment rate of males. Eliminat-ing gender disparities in education would help to increase thestatus and capabilities of women. This indicator is an imper-fect measure of the relative accessibility of schooling for girls.With a target date of 2005, this is the first of the targets to falldue. School enrollment data are reported to the UNESCOInstitute for Statistics by national education authorities. Pri-mary education provides children with basic reading, writing,and mathematics skills along with an elementary understand-ing of such subjects as history, geography, natural science,social science, art, and music. Secondary education completesthe provision of basic education that began at the primarylevel, and aims at laying foundations for lifelong learning andhuman development, by offering more subject-or skill-oriented instruction using more specialized teachers.

Under-five mortality rate is the probability that a new-born baby will die before reaching age five, if subject to cur-rent age-specific mortality rates. The probability is expressedas a rate per 1,000. The main sources of mortality data arevital registration systems and direct or indirect estimatesbased on sample surveys or censuses. To produce harmo-nized estimates of under-five mortality rates that make useof all available information in a transparent way, a method-ology that fits a regression line to the relationship between

mortality rates and their reference dates using weighted leastsquares was developed and adopted by both UNICEF andthe World Bank.

Prevalence of HIV is the percentage of people ages 15–49who are infected with HIV. Adult HIV prevalence ratesreflect the rate of HIV infection in each country’s popula-tion. Low national prevalence rates can be very misleading,however. They often disguise serious epidemics that are ini-tially concentrated in certain localities or among specificpopulation groups and threaten to spill over into the widerpopulation. In many parts of the developing world most newinfections occur in young adults, with young women espe-cially vulnerable. The estimates of HIV prevalence are basedon extrapolations from data collected through surveys andfrom surveillance of small, nonrepresentative groups.

Maternal mortality rate is the number of women who diefrom pregnancy-related causes during pregnancy and child-birth, per 100,000 live births. The data shown here have beencollected in various years and adjusted to a common 1995base year. The values are modeled estimates based on an exer-cise carried out by the World Health Organization (WHO)and United Nations Children’s Fund(UNICEF). In this exer-cise maternal mortality was estimated with a regression modelusing information on fertility, birth attendants, and HIVprevalence. This cannot be assumed to provide an accurateestimate of maternal mortality in any country in the table.

Births attended by skilled health staff are the percentageof deliveries attended by personnel trained to give the neces-sary supervision, care, and advice to women during preg-nancy, labor, and the postpartum period, to conduct deliverieson their own, and to care for newborns. The share of birthsattended by skilled health staff is an indicator of a health sys-tem’s ability to provide adequate care for a pregnant women.Good antenatal and postnatal care improves maternal healthand reduces maternal and infant mortality. But data may notreflect such improvements because health information systemare often weak, material deaths are underreported, and ratesof maternal mortality are difficult to measure.

Table 3. Economic activityGross domestic product is gross value added, at purchasers’prices, by all resident producers in the economy plus anytaxes and minus any subsidies not included in the value of theproducts. It is calculated without deducting for depreciationof fabricated assets or for depletion or degradation of naturalresources. Value added is the net output of an industry afteradding up all outputs and subtracting intermediate inputs.The industrial origin of value added is determined by theInternational Standard Industrial Classification (ISIC) revi-sion 3. The World Bank conventionally uses the U.S. dollarand applies the average official exchange rate reported by theInternational Monetary Fund for the year shown. An alterna-tive conversion factor is applied if the official exchange rate is


judged to diverge by an exceptionally large margin from therate effectively applied to transactions in foreign currenciesand traded products.

Gross domestic product average annual growth rate iscalculated from constant price GDP data in local currency.

Agricultural productivity refers to the ratio of agricul-tural value added, measured in constant 2000 U.S. dollars, tothe number of workers in agriculture.

Value added is the net output of an industry after addingup all out-puts and subtracting intermediate inputs. Theindustrial origin of value added is determined by the Inter-national Standard Industrial Classification (ISIC) revision 3.

Agriculture value added corresponds to ISIC divisions1–5 and includes forestry and fishing.

Industry value added comprises mining, manufacturing,construction, electricity, water, and gas (ISIC divisions 10–45).

Services value added correspond to ISIC divisions 50–99.Household final consumption expenditure (private

consumption in previous editions) is the market value of allgoods and services, including durable products (such as cars,washing machines, and home computers), purchased byhouseholds. It excludes purchases of dwellings but includesimputed rent for owner-occupied dwellings. It also includespayments and fees to governments to obtain permits andlicenses. Here, household consumption expenditure includesthe expenditures of nonprofit institutions serving house-holds, even when reported separately by the country. Inpractice, household consumption expenditure may includeany statistical discrepancy in the use of resources relative tothe supply of resources.

General government final consumption expenditure(general government consumption in previous editions)includes all government current expenditures for purchasesof goods and services (including compensation of employ-ees). It also includes most expenditures on national defenseand security, but excludes government military expendituresthat are part of government capital formation.

Gross capital formation (gross domestic investment inprevious editions) consists of outlays on additions to the fixedassets of the economy plus net changes in the level of invento-ries and valuables. Fixed assets include land improvements(fences, ditches, drains, and so on); plant, machinery, andequipment purchases; and the construction of buildings,roads, railways, and the like, including commercial and indus-trial buildings, offices, schools, hospitals, and private dwellings.Inventories are stocks of goods held by firms to meet tempo-rary or unexpected fluctuations in production or sales, and“work in progress”. According to the 1993 SNA net acquisitionsof valuables are also considered capital formation.

External balance of goods and services is exports ofgoods and services less imports of goods and services. Tradein goods and services comprise all transactions between res-

idents of a country and the rest of the world involving achange in ownership of general merchandise, goods sent forprocessing and repairs, nonmonetary gold, and services.

The GDP implicit deflator reflects changes in prices for allfinal demand categories, such as government consumption,capital formation, and international trade, as well as the maincomponent, private final consumption. It is derived as theratio of current to constant price GDP. The GDP deflator mayalso be calculated explicitly as a Paasche price index in whichthe weights are the current period quantities of output.

National accounts indicators for most developing coun-tries are collected from national statistical organizations andcentral banks by visiting and resident World Bank missions.Data for high-income economies come from the OECD datafiles.

Table 4. Trade, aid, and financeMerchandise exports show the f.o.b. (free on board) value ofgoods provided to the rest of the world valued in U.S. dollars.

Merchandise imports show the c.i.f. value of goods (thecost of the goods including insurance and freight) purchasedfrom the rest of the world valued in U.S. dollars. Data onmerchandise trade come from the World Trade Organization(WTO) in its annual report.

Manufactured exports comprise the commodities inStandard Industrial Trade Classification (SITC) sections 5(chemicals), 6 (basic manufactures), 7 (machinery andtransport equipment), and 8 (miscellaneous manufacturedgoods), excluding division 68.

High technology exports are products with high R&Dintensity. They include high-technology products such as inaerospace, computers, pharmaceuticals, scientific instru-ments, and electrical machinery.

Current account balance is the sum of net exports ofgoods and services, net income, and net current transfers.

Net private capital flows consist of private debt and non-debt flows. Private debt flows include commercial banklending, bonds, and other private credits; nondebt privateflows are foreign direct investment and portfolio equityinvestment.

Foreign direct investment is net inflows of investment toacquire a lasting management interest (10 percent or moreof voting stock) in an enterprise operating in an economyother than that of the investor. It is the sum of equity capital,re-investment of earnings, other long-term capital, andshort-term capital, as shown in the balance of payments.Data on the current account balance, private capital flows,and foreign direct investment are drawn from the IMF’s Bal-ance of Payments Statistics Yearbook and International Finan-cial Statistics.

Official development assistance or official aid from thehigh-income members of the OECD are the main source of


official external finance for developing countries, but officialdevelopment assistance (ODA) is also disbursed by someimportant donor countries that are not members of OECD’sDevelopment Assistance Committee (DAC). DAC has threecriteria for ODA: it is undertaken by the official sector; itpromotes economic development or welfare as a main objec-tive; and it is provided on concessional terms, with a grantelement of at least 25 percent on loans.

Official development assistance comprises grants andloans, net of repayments, that meet the DAC definition ofODA and are made to countries and territories in part I ofthe DAC list of aid recipients. Official aid comprises grantsand ODA-like loans, net of repayments, to countries and ter-ritories in part II of the DAC list of aid recipients. Bilateralgrants are transfers in money or in kind for which no repay-ment is required. Bilateral loans are loans extended by gov-ernments or official agencies that have a grant element of atleast 25 percent and for which repayment is required in con-vertible currencies or in kind.

Total external debt is debt owed to nonresidentsrepayable in foreign currency, goods, or services. It is thesum of public, publicly guaranteed, and private non-guaran-teed long-term debt, use of IMF credit, and short-term debt.Short-term debt includes all debt having an original matu-rity of one year or less and interest in arrears on long-termdebt.

Present value of debt is the sum of short-term externaldebt plus the discounted sum of total debt service paymentsdue on public, publicly guaranteed, and private nonguaran-teed long-term external debt over the life of existing loans.

The main sources of external debt information arereports to the World Bank through its Debtor Reporting Sys-tem from member countries that have received World Bankloans. Additional information has been drawn from the filesof the World Bank and the IMF. Summary tables of theexternal debt of developing countries are published annuallyin the World Bank’s Global Development Finance.

Net migration is the total number of migrants during theperiod, that is, the number of immigrants less the number ofemigrants, including both citizens and noncitizens. Datashown in the table are five-year estimates. Data are from theUnited Nations Population Division’s World PopulationProspects: The 2004 Revision.

Domestic credit provided by banking sector includes allcredit to various sectors on a gross basis, with the exceptionof credit to the central government, which is net. The bank-ing sector includes monetary authorities, deposit moneybanks, and other banking institutions for which data areavailable (including institutions that do not accept transfer-able deposits but do incur such liabilities as time and savingsdeposits). Examples of other banking institutions includesavings and mortgage loan institutions and building and

loan associations. Data are from the IMF’s InternationalFinance Statistics.

Statistical methods This section describes the calculation of the least-squaresgrowth rate, the exponential (endpoint) growth rate, and theWorld Bank’s Atlas methodology for calculating the conver-sion factor used to estimate GNI and GNI per capita in U.S.dollars.

Least-squares growth rate Least-squares growth rates are used wherever there is a suffi-ciently long time series to permit a reliable calculation. Nogrowth rate is calculated if more than half the observationsin a period are missing.

The least-squares growth rate, r, is estimated by fitting alinear regression trendline to the logarithmic annual valuesof the variable in the relevant period. The regression equa-tion takes the form

ln Xt = a + bt,

which is equivalent to the logarithmic transformation of thecompound growth equation,

Xt = Xo (1 + r)t.

In this equation, X is the variable, t is time, and a = log Xo

and b = ln (1 + r) are the parameters to be estimated. If b* isthe least-squares estimate of b, the average annual growthrate, r, is obtained as [exp(b*) – 1] and is multiplied by 100to express it as a percentage.

The calculated growth rate is an average rate that is repre-sentative of the available observations over the entire period.It does not necessarily match the actual growth rate betweenany two periods.

Exponential growth rate The growth rate between two points in time for certaindemographic data, notably labor force and population, iscalculated from the equation

r = ln (pn /p1)/n,

where pn and p1 are the last and first observations in theperiod, n is the number of years in the period, and ln is thenatural logarithm operator. This growth rate is based on amodel of continuous, exponential growth between twopoints in time. It does not take into account the intermediatevalues of the series. Note also that the exponential growthrate does not correspond to the annual rate of change mea-sured at a one-year interval which is given by

(pn – pn – 1)/pn – 1.


World Bank Atlas method In calculating GNI and GNI per capita in U.S. dollars for cer-tain operational purposes, the World Bank uses the Atlasconversion factor. The purpose of the Atlas conversion factoris to reduce the impact of exchange rate fluctuations in thecross-country comparison of national incomes. The Atlasconversion factor for any year is the average of a country’sexchange rate (or alternative conversion factor) for that yearand its exchange rates for the two preceding years, adjustedfor the difference between the rate of inflation in the countryand that in Japan, the United Kingdom, the United States,and the Euro Zone. A country’s inflation rate is measured bythe change in its GDP deflator. The inflation rate for Japan,the United Kingdom, the United States, and the Euro Zone,representing international inflation, is measured by thechange in the SDR deflator. (Special drawing rights, or SDRs,are the IMF’s unit of account.) The SDR deflator is calcu-lated as a weighted average of these countries’ GDP deflatorsin SDR terms, the weights being the amount of each coun-try’s currency in one SDR unit. Weights vary over timebecause both the composition of the SDR and the relativeexchange rates for each currency change. The SDR deflator iscalculated in SDR terms first and then converted to U.S. dol-lars using the SDR to dollar Atlas conversion factor. TheAtlas conversion factor is then applied to a country’s GNI.The resulting GNI in U.S. dollars is divided by the midyearpopulation to derive GNI per capita.

When official exchange rates are deemed to be unreliableor unrepresentative of the effective exchange rate during aperiod, an alternative estimate of the exchange rate is used inthe Atlas formula (see below).

The following formulas describe the calculation of theAtlas conversion factor for year t:

and the calculation of GNI per capita in U.S. dollars for year t:

Y$t = (Yt/Nt)/et

*

where et* is the Atlas conversion factor (national currency tothe U.S. dollar) for year t, et is the average annual exchangerate (national currency to the U.S. dollar) for year t, pt is theGDP deflator for year t, pt

S$ is the SDR deflator in U.S. dollarterms for year t, Y t

$ is the Atlas GNI per capita in U.S. dollarsin year t, Yt is current GNI (local currency) for year t, and Nt

is the midyear population for year t.

Alternative conversion factorsThe World Bank systematically assesses the appropriatenessof official exchange rates as conversion factors. An alterna-tive conversion factor is used when the official exchange rate

t tt

t

t

S

t

S tt

e ep

p

p

pe

p* [ /

$

$=

+−

− −

−1

3 22 2

1tt

t

S

t

S tp

p

pe

− −

+

1 1

/ ]

$

$

is judged to diverge by an exceptionally large margin fromthe rate effectively applied to domestic transactions of for-eign currencies and traded products. This applies to only asmall number of countries, as shown in Primary data docu-mentation table in World Development Indicators 2005.Alternative conversion factors are used in the Atlas method-ology and elsewhere in the Selected World DevelopmentIndicators as single-year conversion factors.

Table 5. Key indicators for other economiesPopulation is based on the de facto definition, which countsall residents, regardless of legal status or citizenship, exceptfor refugees not permanently settled in the country of asy-lum, who are generally considered part of the population ofthe country of origin.

Average annual population growth rate is the exponen-tial rate of change for the period (see the section on statisti-cal methods below).

Population density is midyear population divided by landarea. Land area is a country’s total area excluding areas underinland bodies of water and coastal waterways. Density is cal-culated using the most recently available data on land area.

Gross national income (GNI—formerly gross nationalproduct or GNP), the broadest measure of national income,measures total value added from domestic and foreignsources claimed by residents. GNI comprises gross domesticproduct (GDP) plus net receipts of primary income fromforeign sources. Data are converted from national currencyto current U.S. dollars using the World Bank Atlas method.This involves using a three-year average of exchange rates tosmooth the effects of transitory exchange rate fluctuations.(See the section on statistical methods below for further dis-cussion of the Atlas method.)

GNI per capita is GNI divided by midyear population. Itis converted into current U.S. dollars by the Atlas method.The World Bank uses GNI per capita in U.S dollars to classifyeconomies for analytical purposes and to determine borrow-ing eligibility.

PPP Gross national income, which is GNI converted intointernational dollars using purchasing power parity (PPP)conversion factors, is included because nominal exchangerates do not always reflect international differences in relativeprices. At the PPP rate, one international dollar has the samepurchasing power over domestic GNI that the U.S. dollar hasover U.S. GNI. PPP rates allow a standard comparison of realprice levels between countries, just as conventional priceindexes allow comparison of real values over time. The PPPconversion factors used here are derived from price surveyscovering 118 countries conducted by the International Com-parison Program. For Organisation for Economic Co-opera-tion and Development (OECD) countries data come fromthe most recent round of surveys, completed in 1999; the restare either from the 1996 survey, or data from the 1993 or ear-


lier round and extrapolated to the 1996 benchmark. Esti-mates for countries not included in the surveys are derivedfrom statistical models using available data.

PPP GNI per capita is PPP GNI divided by midyear pop-ulation.

Gross domestic product (GDP) per capita growth isbased on GDP measured in constant prices. Growth in GDPis considered a broad measure of the growth of an economy.GDP in constant prices can be estimated by measuring thetotal quantity of goods and services produced in a period,valuing them at an agreed set of base year prices, and sub-tracting the cost of intermediate inputs, also in constantprices. See the section on statistical methods for details ofthe least-squares growth rate.

Life expectancy at birth is the number of years a new-born infant would live if patterns of mortality prevailing atits birth were to stay the same throughout its life.

Adult literacy rate is the percentage of persons aged 15and above who can, with understanding, read and write ashort, simple statement about their everyday life.

Carbon dioxide (CO2) emissions measures those emis-sions stemming from the burning of fossil fuels and themanufacture of cement. These include carbon dioxide pro-duced during consumption of solid, liquid, and gas fuels andfrom gas flaring.

The Carbon Dioxide Information Analysis Center(CDIAC), sponsored by the U.S. Department of Energy, cal-culates annual anthropogenic emissions of CO2. These calcu-lations are derived from data on fossil fuel consumption,based on the World Energy Data Set maintained by theUNSD, and from data on world cement manufacturing,based on the Cement Manufacturing Data Set maintained bythe U.S. Bureau of Mines. Each year the CDIAC recalculatesthe entire time series from 1950 to the present, incorporatingits most recent findings and the latest corrections to its data-base. Estimates exclude fuels supplied to ships and aircraftengaged in international transportation because of the diffi-culty of apportioning these fuels among the countries bene-fiting from that transport.

Aarhus Convention, 217absolute deprivation, avoidance of, 19accountability. See voice and accountabilityacquired immunodeficiency syndrome. See HIV/AIDSaffluence and wealth

business investment and, 102inequalities in, 55social distinctions and distribution of wealth, 2–3, 74, 89

Afghanistanhealth care, 144reconstruction aid, 219

Africaaid to, 218, 219capital markets, 8Commission for Africa, 218customary systems, 159development experiences, 115–16education trends, 35–36, 68government policy effect on labor markets, 187health services, 32human rights regimes, 79inequality trends, 46, 68infant mortality, 6infrastructure, 169, 170, 171intercountry inequality, 38land relations, 93life expectancy, 56, 58, 69overall inequality levels, 65programs for nonworking young, 152schooling years, 60–61trade credit, 97(see also specific countries)

agency, 5, 48–49, 49–50enhancing, 205inequalities of, 28, 49–51internalization of disadvantage and, 49–50

agriculturesubsidies, 220underinvestment, 98–100

aid, 17“aid orphans/darlings,” 219debt cancellation, 221debt forgiveness, 222debt relief, 221designing, 17efficient and effective, 17, 218, 229

improving allocation of, 218–19increasing levels of, 219–20innovative funding mechanisms, 221official development assistance, 219–20providing assistance for endowments, 218–21tax issues and, 176voluntary contributions/private assistance, 221

altruistic rewarding/altruistic punishment, 80Argentina

Center for Financial Stability, 184educational attainment/achievement, 61infrastructure, 172, 173–74legal approach to equity, 79poor institutional quality, 113, 119squatters’ rights, 164–65

Aristotle, 78–79Armenia

health care payments, 147“My Rights” television show, 157stunting in children, 29

Asiacrisis behavior and costs, 201inequality trends, 45infant mortality, 6poverty trends, 66programs for nonworking young, 152schooling years, 60–61(see also specific countries and regions)

Asian Development Bank, credit market report, 90Australia

educational attainment/achievement, 61means-tested pensions, 154

Bangladeshabsent health care providers, 148Food for Education Program, 137health care, 144income inequality, 64income information sharing within household, 54inequality trends, 45infrastructure, 169land and safety net programs, 166legal approach to equity, 79means-tested pensions, 154microfinance, 184

309

Index

Bangladesh (continued)poverty trends, 66teacher absenteeism, 35women and household assets, 53

Basel Committee on Banking Supervision, 216biases, institutional and judicial, 5Bolivia

obstetric care, 144pensions, 154

Bosnia, legal institutions, 160Botswana

HIV/AIDS, 69pensions, 154

Brazilbasic education, 140Bolsa Escola, 137, 153educational attainment/achievement, 61health education, 142, 143health insurance, 145inequality trends, 46infant mortality, 29land access, 13legal approach to equity, 79local government democratization, 70–71pensions, 148–49productivity gaps in agriculture, 99stunting in children, 29trade unions, 189–90

Bulgaria, education, 12Burkino Faso, head of household education, 35

Cambodiahealth care payments, 147labor standards, 190programs for nonworking young, 153spatial differences, 42squatters’ rights, 165stunting/underweight in children, 32voice rank, 68

Canadaintergenerational mobility, 47intergenerational transmission of inequality, 47

capacity to aspire/capacity to engage, 21, 71capital markets, 7–8, 16, 21–22

(see also financial markets; human capital markets)Caribbean. See Latin America and the Caribbeancaste systems, 5, 8, 26, 96CBHI. See community-based health insurance (CBHI)Center for Global Development, 222Central America

elites and taxes, 176health education, 143human rights regimes, 79(see also specific countries)

Central Asiaeducation trends, 36health care payments, 145

HIV/AIDS, 59inequality trends, 46life expectancy, 58

Centre for Development Studies, Trivandrum, credit survey,90

childrendisadvantaged, 2health issues. See health and health care

Chilebank privatization, 182educational attainment/achievement, 61health insurance, 145, 147labor market reforms, 191physician availability, 144

Chinaaccession to WTO and trade reforms, 194economic growth, 7gender equity, 54health care payments, 147income and expenditure, 63, 64, 65inequality trends, 45, 68infrastructure, 169institutional transitions and economic development, 119,

122–24, 125labor market reforms, 191land distribution, 166–67life expectancy, 56, 142poverty trends, 66voice rank, 68(see also specific provinces)

circumstances of birth and life. See predeterminedcharacteristics/circumstances of human beings

CIS. See Commonwealth of Independent States (CIS)civil law, 79collective action, Palanpur example, 27Colombia

education, 11health insurance, 12, 146infant mortality, 29infrastructure, 172labor market reforms, 191land rental markets, 167PACES voucher program, 140student achievement, 139

Commission for Real Property Claims of Displaced Persons andRefugees, 160

Commission on Macroeconomics and Health report, 100–101Commitment to Development Index, 222common law, 79Commonwealth of Independent States (CIS), land transferability,

166community-based health insurance (CBHI), 147–48complementaries between equity and efficiency, 21–22, 74

equity–efficiency tradeoffs, 22–23, 130, 179, 204–5complementaries between equity and prosperity, 2, 23, 74,

130Core 25 Principles for Banking Supervision, 216corruption, 130, 227


Côte d’IvoireCôte d’Ivoire Living Standards Measurement Surveys, 91–92trade credit, 97

credit markets, 89–91crime/violence

domestic violence, 54improved crime policies, 161–62inequality traps and, 49–50against women, 162

crises and crises resolution, 199–202Cuba, health care, 144culture, social networks and, 21Czech Republic

educational attainment/achievement, 61effects of regulation and disclosure in, 184–85financial markets, 14

Dayton Agreement, 160debt cancellation, 221Declaration on the TRIPS Agreement and Public Health,

215–16democracy

in governance institutions, 70–71Spanish transition to, 105–6

Demographic and Health Survey (DHS), 29–34, 59Denmark, voice rank, 68developing countries

“advantage to backwardness,” 69catch up countries, 68–69customary systems, 159

developmentlong-term, 8–9, 22regional inequality and lagging regions, 204–5(see also equity–development relationship)

Development Agenda Framework Agreement, 216development assistance. See aid; equity–development

relationshipdevelopment community, recent thinking and agreements of, 131,

226–29DHS. See Demographic and Health Survey (DHS)Dictator Game, 81directed credit, 130, 184disability and disabled persons, 32, 154–55discrimination and stereotyping, 7, 21, 74, 94–96, 131

analysis of missing women and girls/gender discrimination,223

explicit discrimination, 21in justice systems, 158

displaced persons and refugees, 160distributive justice, 77–78, 217distributive policies, 3, 23, 228

investment and growth using, 102–4poverty reduction and redistributive policies, 9–10, 74

Doha Round, 16, 210–11, 212Declaration on the TRIPS Agreement and Public Health,

215–16Development Agenda Framework Agreement, 216

early childhood development (ECD), 11, 132–33benefits of early intervention, 133–34designing programs, 134–35

earnings. See incomes, assets, and liabilitiesEast Asia

billionaire wealth, 38development experiences, 114–15East Asian crisis, 185elite capture of financial policy and institutions, 181income distribution dynamics, 46inequality trends, 45life expectancy, 56poverty trends, 66schooling years, 60–61(see also specific countries)

Eastern Europeeducation trends, 36inequality trends, 46(see also specific countries)

ECD. See early childhood development (ECD)economic inequalities, 28, 36–43, 229

between-group shares of total inequality, 40–48relationship between group differences and inequality,

43–48spatial differences/inequalities, 42–43, 204–5

economic liberalization, 179economic policies

pathologies arising in pursuit of equity, 129–30populist macropolicy, 16, 130

Ecuadoreducation test results, 34Program for the Development of Indigenous and Afro-

Ecuadorian Peoples, 51Quichua people, 5, 50–51spatial differences, 42teacher absenteeism, 35

education and educational attainmentaccess to schooling, 6–7, 136–38access to teachers, 35differences in, 6, 56disadvantaged students, 139–40excluded groups, 138global inequality, 55, 60–62, 68improving gender equity, 138inequalities in, 27inequalities within countries, 34–36male and female household heads, 34–35reaching ethnic groups, 138rural and urban household heads, 35schooling and basic education, 6–7, 11–12, 27, 135–41strengthening accountability, 140–41test results, 34TIMSS data, 139trends, 35–36upgrading quality, 138–40women and girls, 52–53, 138

efficiency, increased, 10, 74, 101–4elderly, social programs for nonworking elderly, 153–54

Index 311

elites, 130, 131capture and discrimination by, 156–58, 175, 178, 180–82,

194El Salvador, education system, 141employment and occupations

discrimination and stereotypes, 95–96inequalities in, 27women, 52–53

empowerment, 50–51, 70–71, 227endowments, 19–20

development assistance, 218–21unequal, 17

Enhanced HICP Initiative, 222environment, Aarhus Convention, 217equal opportunity, 2–3, 18–19, 74“equal opportunity policy,” 77equity

concepts of, 78defined, 18–19, 74versus equality, 74ethical and philosophical approaches to, 76–78importance/significance considerations, 73–75legal institutions/justice and, 78–81

equity and fairness concerns, 73–75, 129income inequality and poverty reduction, 84–88income inequality and subjective well-being, 82–83intrinsic human behavior/shared concern, 7, 76, 80–84well-being and, 76, 82–83(see also complementaries between equity and prosperity;

equity–development relationship)equity–development relationship, 2–4, 7, 10, 17, 129, 226–29

contemporary evidence, 113–18historical evidence, 109–13Indonesian example, 126–27institutional quality, 124–25, 227–28power distribution and institutional quality, 107–9, 129, 228transitions to equitable institutions, 118–24

ethicscodes of conduct, 213–14ethical approaches to equity, 76–78fair and ethical trade, 213–14, 222

Ethiopiahealth education, 142land transactions, 93voice rank, 68

Europebasic education, 140health care payments, 145HIV/AIDS, 59human rights regimes, 79income inequality and subjective well-being, 82income support for children, 152inequality trends, 46legal approach to equity, 79life expectancy, 58overall inequality levels, 65programs for nonworking young, 152(see also specific countries)

European Union, human rights norms and regimes, 80extreme poverty, 19extreme predation, 130

fair and ethical trade, 213–14, 222fairness

equity and. See equity and fairness concernsin markets, 11

financial markets, 14, 131accountability and, 184–85achieving equity and efficiency in, 179–85Basel II Capital Accord, 215bias in, 202competition in, 184–85Core 25 Principles for Banking Supervision, 215elite influence/capture, 180–81Financial Stability Forum, 216foreign direct investment, 216global standards, 216liberalization, 216–17liberalization paradox, 14, 179, 182–83, 203ROSC, 216segmentation, 185technical design issues, 183–84unequal access considerations, 180–81

Financial Stability Forum, 216Finland

educational attainment/achievement, 61institutional transitions, 119, 121–23intergenerational mobility, 47intergenerational transmission of inequality, 47

Foreign Policy magazine, 222Francefinancial liberalization, 182reading levels, 62

Gambia, health care, 144GATS. See General Agreement in Trade and Services (GATS)gender differences/inequalities, 4–5, 26, 51–54

gender discrimination, 158, 223improving equity in education, 138infrastructure investments and gender equity, 138redressing, 223

General Agreement in Trade and Services (GATS), Mode IV, 210

Ghanainfrastructure, 169opportunities for women, 54trade unions, 189

global equity achievement, 206–7analysis and research, 223citizen mobilization, 222–23greater global equity policies, 11, 16–17, 22international organizations and, 223making global markets work more equitably, 207–17, 227providing development assistance for endowments, 218–21


summary, 223transitions to greater equity, 221–23

global governance, 131global inequity and well-being, 6–7, 129

catch up countries and, 68–69education, 60–62examples and concepts, 55–56future of, 68–69global income inequality, 7, 55health, 56, 58–59three competing concepts of, 56trends, 6

global markets, 17, 207–17, 227(see also markets; specific markets)

governance, 11, 227deepening and extending democracy in institutions, 70–71global, 131inequalities of power and, 48–51, 129

Greece, educational attainment/achievement, 61Group of Twenty-Four, 223groups and group differences

individual behavior/performance and group identity, 81–82local group participation and empowerment, 23, 70–71marginal groups, 70–71, 161membership and initial circumstances, 19political inequality between groups, 21relationship between group differences and inequality,

43–48social programs for special vulnerable groups, 154–55trends, 44–48

Guatemalabasic education, 140between-group differences in inequality, 41stunting in children, 31

Guyana, development experiences, 116–18

health and health care, 12access to immunization, 31cognitive development in children, 11, 133community-based health insurance, 147–48differences in, 6disability, 32enhancing provider incentives, 148expanding access to, 143–44expanding knowledge of, 142–43financing affordable care, 144–48high-impact health services, 31–32inequalities in, 141–42inequalities within countries, 29–34infant and child mortality, 6, 29, 56, 59, 68life expectancy, 7, 55, 56, 58–59, 69social inequalities damaging health, 32stunting, 29–31trends, 32–34

Highly Indebted Poor Countries (HIPC), 219, 221Enhanced HICP Initiative, 222

HIPC. See Highly Indebted Poor Countries (HIPC)

history and historical perspective, 23evidence regarding development and political power, 109–13

HIV/AIDS, 6, 18, 32–33, 33, 58, 59, 68, 69, 143, 145–55, 154Honduras, gender discrimination, 158Hong Kong, China, inequality trends, 45households

domestic violence, 54gender inequality and function of, 51–54

human behaviorconditioned by group identity, 81–82discrimination/ stereotype effects on, 94–96equity and fairness concerns, 7, 80–82

human capacities, 10–11, 11–13, 130, 226–27considerations, 132early childhood development, 11, 132–35health and health care, 12, 141–48schooling and basic education, 11–12, 135–41social protection systems/risk management, 12, 148–55summary, 155

human capital markets, 8, 94–96, 130underinvestment, 100–101

Human Development Report of the United Nations DevelopmentProgramme, 17

human immunodeficiency virus. See HIV/AIDShuman rights protection, 17

IBRD. See International Bank for Reconstruction andDevelopment (IBRD)

ICOR. See incremental capital–output ratio (ICOR)ICRISAT. See International Crop Research Institute in the Semi-

Arid Tropics (ICRISAT)IFF. See International Financing Facility (IFF)ILO. See International Labour Office (ILO)immunization, access to, 31imperfect markets. See marketsincomes, assets, and liabilities

income inequality and incentives, 83–84income inequality and poverty reduction, 84–88income inequality and subjective well-being, 82–83inequalities in, 23, 27, 55, 62–66, 68redistribution of, 102–4women, 52–53

incremental capital–output ratio (ICOR), 97–98India

affirmative action programs, 158billionaire wealth, 38capital markets, 8caste system, 5, 8, 26, 96, 184credit markets, 90economic growth, 7education, 12effects of government intervention in labor markets, 187gender equity, 54Gounders, 98, 102health care, 144health care payments, 147–48ICOR estimates, 98

Index 313

India (continued)ICRISAT household insurance data, 92improved marketing channels and technology, 196income and expenditure, 63, 64inequality trends, 45, 68infant and child mortality, 144infrastructure, 169Integrated Rural Development Program loans, 184lack of voice, 48land and safety net programs, 166land rights of rent-collecting intermediaries, 163legal approach to equity, 79local government democratization, 70–71means-tested pensions, 154misallocation of capital, 98, 102Operation Barga, 103–4poverty trends, 66“priority sector” lending, 97product licensing restrictions, 196product market reforms, 195–96remedial education program, 140Self-Employed Women’s Association, 147–48stereotypes and behavior, 96teacher absenteeism, 35voice rank, 68worker organizations/trade unions, 190

indigenous peoplesILO Covenant 169, 222Quichua people, 5, 50–51

individualsdifferences in opportunities for, 131responsibility for own welfare, 77

Indonesiadevelopment experiences, 114–15, 118, 125elite influence/capture, 193financial markets, 14growth, equity, and poverty reduction in, 126–27health insurance, 12infant mortality, 29Kecamatan Development Project, 50, 71microfinance, 184physician availability, 144reading levels, 62school enrollment problems, 139teacher absenteeism, 35wealth inequality, 37

industry and trade, underinvestment, 96–98inefficient inequalities, 89inequalities

interaction of various, 20–21macroeconomic instability as cause and consequence of,

198–99mobility and, 2–3, 50three competing concepts of, 56views on, 84(see also global inequity and well-being; inefficient

inequalities; inequity within countries; specific areas ofinequality)

inequality traps, 2–3, 20–23, 28, 129, 228between-group differences, 43crime/violence-related, 49–50, 161–62Palanpur example, 26–27for women, within countries, 51–54

inequity within countries, 4–6, 23, 28–29, 129economic inequalities, 28, 36–43in education, 34–36examples in various countries, 45–46group differences and, 43–48in health, 29–34inequalities of power, 48–51, 229. (see also agency; power and

influence)inequality traps for women, 51–54intergenerational transmission of inequality, 28, 46–48regional disparities/inequalities, 204–5

infrastructureaccess to, 6, 10–11, 169, 171–74, 175accountability and, 174–75corruption and, 14, 174, 175equitable provision of, 14, 168–75privatization, 14, 170–71service affordability, 171–74strengthening governance, voice, accountability of, 174–75

inheritance and inheritance laws, gender inequalities and, 52,162–63, 165

institutional quality, 124–25power distribution and, 107–9, 129, 228

institutions and institutional development, 8–9, 107–9, 130,227–28

colonial origins of contemporary institutions, 9, 109–13contemporary evidence/perspective, 113–18democratic governance institutions, 70–71historical evidence/perspective, 74–75, 109–13inequalities of power and, 48–51, 109–18, 129, 228inequitable/exclusionary, 2, 74transitions to more equitable, 118–24

insurance markets, 91–93, 130intellectual property rights protection, 214–16intergenerational transmission of inequality, 5–6, 28, 46–48, 73International Bank for Reconstruction and Development (IBRD),

membership and voting shares, 66–67International Chirstelijk Steunfonds Africa, learning experiments,

139International Conference on Financing for Development, 218–19International Crop Research Institute in the Semi-Arid Tropics

(ICRISAT), 92, 99International Development Association, 219International Financing Facility (IFF), 221international inequity. See global inequity and well-beingInternational Labour Office (ILO), 187, 190

Covenant 169 on indigenous peoples, 222international law, 79

global markets and, 17human rights, 79–80intellectual property rights protection, 214–16

International Monetary Fund, 223ICOR estimates, 98


International Organization for Migration, 210International Social Survey Program, 83Inter-university Consortium for Political and Social Research,

83investment and investment opportunity, 7–8, 227–28

inequalities and, 101–4microcredit, 102underinvestment, 96–101(see also markets; specific markets)

Iraq, reconstruction aid, 219Ireland

educational attainment/achievement, 61labor market reforms, 191voice rank, 68

Islamic law, justice and equity in, 79Israel

hyperinflation, 202voice rank, 68

Italybanking segmentation, 185means-tested pensions, 154

Jamaicachild development study, 11value-added tax exemptions, 177

Japaneducational attainment/achievement, 61life expectancy, 56

Jewish law, justice and equity in, 79Joint United Nations Programme on HIV/AIDS, 69Jordan

infant mortality, 29stunting in children, 29

Jubilee 2000 campaign, 222justice and justice systems

access to, 10–11, 130–31, 158–60building equitable systems, 13, 130–31, 156–62combating discriminatory norms and practices, 158combating elite capture and discrimination, 156–58, 175compatibility of state and customary systems, 158–59enhancing judicial independence and accountability,

156–58equity and legal institutions, 78–81improved crime policies, 161–62improved personal safety policies, 161–62legal institution establishment, 159–60to level the playing field, 175theories of, 76–78

Kazakhstan, stunting in children, 29Kecamatan Development Project (KDP), 50, 71Kenya

land reform, 163learning experiments, 139trade credit, 97

Kyoto Protocol, 217

labor markets, 15, 16, 21–22, 131, 202achieving equity and efficiency in, 185–94addressing links with unequal power, 187–88improved design of, 188–90interaction between product markets and, 1997international labor mobility, 208–10reasons to intervene, 186–87reforming poor institutions, 191–93worker organizations/trade unions, 189–90

lagging regions, 204–5land

access to, 10–11, 13, 162–68, 175formal land titling, 165–66sharecropping, 93squatters’ rights, 164–65transferability, 166women’s inheritance rights, 52, 162–63, 165

land distribution, 162–63cost-effective, 167–68

land markets, 8, 21–22, 93–94, 166–67land reform, 13, 175

experience with, 163–64Laos People’s Democratic Republic, head of household education, 35Latin America and the Caribbean

billionaire wealth, 38colonization example, 9, 111–12crises behavior and costs, s201education trends, 68financial liberalization, 183income distribution dynamics, 46income inequality and subjective well-being, 82–83, 84inequality trends, 45–46infant mortality, 6infrastructure investments, 170land reform, 163legal approach to equity, 79means-tested pensions, 154means tests, 151overall inequality levels, 65product and labor market interaction, 197programs for nonworking young, 152schooling years, 60–61trade liberalization, 194(see also specific countries)

legal institutionsequity and, 78–81establishment, 159–60

leveling the economic and political playing field, 3, 4, 9–17, 23, 131justice and justice systems and, 175pathologies making uneven playing field, 178–79role of public action, 10–11, 228. (see also specific focal points,

e.g., human capacities, markets, etc.)life chances, 1–2, 19, 55

macroeconomy, 14–17accountability structures and, 201–3macroeconomic policies, 131

Index 315

macroeconomy (continued)macroeconomic stability/instability, 15–17, 198–203policy design considerations, 201–3

Madagascarhealth care, 144spatial differences, 42

Malaysiafinancial markets, 14inequality trends, 45obstetric care, 144productivity gaps in agriculture, 99wealth inequality, 37

Malihead of household education, 35inequality contrasted to United States, 56infant mortality, 29voice rank, 68

marginal products, 97–98markets, 7–8, 14–17, 226

access to, 11, 178–79barriers to, 178, 179, 202-3fairness in, 11imperfect markets and market failures, 7–8, 17, 21–22, 74, 89,

101–2, 130inequality traps and market failures, 21making global markets work more equitably, 207–17, 227possibilities for change, 178–79relating to equity, 178–79(see also specific markets, e.g., credit markets, financial markets,

etc.)marriage and kinship systems, gender inequalities and, 52Mauritius

development experiences, 116–18, 119, 125economic inequality, 38pensions, 154

medicine markets, drug access example, 224–25Mexico

banking system evolution, 109bank privatization, 182capital markets, 8crises, 200educational attainment/achievement, 61financial markets, 14financial system reform, 179inequality trends, 46legal approach to equity, 79Oportuniades (PROGRESA) program, 12, 137–38, 143, 153physician availability, 144Procampo, 197property rights reforms, 164safety nets and, 197-98social security spending, 145–46trade liberalization, 15, 195trade unions, 189

Middle Easteducation trends, 36inequality trends, 46life expectancy, 56

schooling years, 60–61(see also specific countries)

Millennium Declaration, 131Millennium Development Goals, 17, 66Millennium Summit+5, 222mobility

anomie and, 50facilitating, 205inequalities and, 2–3international labor mobility, 208–10women, 52

Mongolia, product market reforms, 195-96Monterrey Consensus, 131Morocco

academic achievement, 62education, 11immunization, 31school enrollment problems, 139spatial differences, 42student achievement, 139trade liberalization, 194

Mozambiqueinfant mortality, 29product markets and trade reform, 194–95, 198spatial differences, 42

Multi-Fiber Agreement phaseout, 211

NAFTA. See North American Free Trade Agreement (NAFTA)Namibia

health insurance, 145pensions, 154

Nash equilibrium, equity and fairness and, 81National Institute of Public Finance and Policy, credit market

report, 90natural resources, rectifying inequities in use of, 216–17Nepal, inequality trends, 45the Netherlands, labor market reforms, 191New Zealand

labor market reforms, 191means-tested pensions, 154

Nigerinfant mortality, 29infrastructure, 173legal approach to equity, 79

Nigeria, poverty trends, 66North America

basic education, 140colonization example, 9, 112–13human rights regimes, 79(see also Canada; Mexico; United States)

North American Free Trade Agreement (NAFTA), 195safety nets and, 197

Norway, educational attainment/achievement, 61

official development assistance, 219–20opportunity, equal. See equal opportunity


opportunity disparities/inequalities, 1, 2–3, 18–19, 73–74, 131(see also specific aspects and scope, e.g., global inequity and

well-being, inequity within countries, etc.)outcomes, significance and importance of, 3

Pakistancredit market, 89–90financial markets, 14inequality trends, 45infrastructure, 169poverty trends, 66productivity gaps in agriculture, 99stunting in children, 29voice rank, 68

Palanpur, India, inequality traps example, 26–27Paraguay, between-group differences in inequality, 41pensions

programs for nonworking elderly, 148–49, 153–54South Africa, 104

personal safety, protection, 131, 161–62Peru

gender discrimination, 158labor shares and crises, 200teacher absenteeism, 35

pharmaceutical patents, intellectual property rights issue,214–16

the Philippineseducation, 11opportunities for women, 54student achievement, 139wealth inequality, 38

philosophical approaches to equity, 76–78Plato, 76Poland

effects of regulation and disclosure in, 184–85trade unions, 189voice rank, 68

policy design, 3–4, 10, 131costs of policies, 130growth and equity policy dichotomy, 10oligarchic dominance and, 130two types of pathologies in, 129–30(see also leveling the economic and political playing field)

policy implementation, 10policy tradeoffs, 19–20, 130, 204–5

short-term versus long-term, 3–4, 22–23political inequality, 2–3, 20, 74–75, 109–18

between groups, 21political power

contemporary evidence regarding development and, 113–18government intervention in labor markets and, 187–88institutional quality and distribution of, 107–9, 129

poor people“negative terms of recognition,” 49–50social programs for working poor, 150–52

Portugal, educational attainment/achievement, 61poverty, 227–28

group-based inequalities and, 49–51. (see also agency; poorpeople)

poverty reduction, 3, 23, 226–27income inequality and, 84–88Indonesian example, 126–27redistributive policies and, 9–10, 23, 74

power and influenceglobal inequalities, 55–56, 66–68institutional quality and distribution of, 107–9, 129(see also agency; political power)

predetermined characteristics/circumstances of human beings, 19,28, 55, 131, 205

product markets, 15, 16, 131design of reform, 198, 202-03domestic product markets and equity, 194–97drug access example, 224–25elite influence/capture, 193-94interaction between labor markets and, 196–97safety nets, 197-98trade reform and, 193–98

Program for the Development of Indigenous and Afro-EcuadorianPeoples, 51

property rightsaccess to finance and weak rights, 180gender inequalities and, 52protection and security, 100, 131, 175, 214–16

prosperity. See complementaries between equity and prosperityprotection of rights, 13, 17

(see also specific rights)

Quichua people, 5, 50–51

reforms, equity-enhancing, 70–71, 228religion and religious faiths/beliefs, 21

equity and fairness concerns, 7, 76relocation assistance, 205Repeated Public Good Game with Punishment, 81Reports on the Observance of Standards and Codes (ROSC),

216Republic of Korea

chaebols, 182development experiences, 114–15, 119, 125educational attainment/achievement, 61financial markets, 14gender equity, 54inequality trends, 45labor market reforms, 192trade unions, 189

Republic of Yemenstunting in children, 29voice rank, 68

resource allocation/distributionarguments and principles of, 77–78rectifying natural resource use inequities, 216–17resource flows, 11

Roosevelt, Franklin D., 74

Index 317

ROSC. See Reports on the Observance of Standards and Codes(ROSC)

Russian Federationbank privatization, 182crisis of 1990s, 201financial markets, 14voice rank, 68

safety netsproduct markets and, 197-98risk management and, 12

Scandinavialabor markets, 188tax design, 176(see also specific countries)

schooling. See education and educational attainment; specificcountries and regions

secular philosophical traditions, equity and, 76Senegal

growth incidence curve and, 85–86infrastructure, 173means-tested pensions, 154

Serbian Poverty Reduction Strategy, disability report, 32Singapore

inequality trends, 45wealth inequality, 37

Slovak Republic, labor market reforms, 191Smith, Adam, 178social distinctions, distribution of wealth and, 2–3social inequalities, damaging health, 32social justice, 76–78social networks, 21social protection systems/risk management, 12, 148–49

contributory schemes, 149general tax funded transfers, 149program choice variance by country, 149–50programs for nonworking elderly, 153–54programs for nonworking youth, 152–53programs for special vulnerable groups, 154–55programs for working poor, 150–52social assistance programs, 149–55

Somalia, health care, 144South Africa

billionaire wealth, 38customary systems, 159development experiences, 1health insurance, 145HIV/AIDS, 69infant and child mortality, 142land access, 13land reform, 163means-tested pensions, 154pensions, 104, 148–49voice rank, 68women and household assets, 53worker organizations/trade unions, 189, 190

South Americahuman rights regimes, 79(see also specific countries)

South Asiaeducation trends, 35–36inequality trends, 45, 68life expectancy, 56, 69poverty trends, 66schooling years, 60–61(see also specific countries)

Soviet Union (former), inequality trends, 46Spain

equity and development, 105–6Latin America colonization, 9, 111–12trade unions, 189

Sri Lankainequality trends, 45microfinance, 184obstetric care, 144

Sudan, land rental markets, 167Sumatra, opportunities for women, 54“Summary Report on Informal Credit Markets in India,” 90sustainable development, 3, 74–75Sweden

institutional transitions, 119, 120–22intergenerational mobility, 47intergenerational transmission of inequality, 47

Taiwan, Chinadevelopment experiences, 114–15inequality trends, 45

Tanzaniainfrastructure, 169trade credit, 97

taxes and tax policies, 12–13, 130, 176–77tenure, land, 13, 164–67Thailand

health education, 143health insurance, 12, 146–47insurance, 92land titling, 165physician availability, 144worker organizations/trade unions, 190

Third International Mathematics and Science Study (TIMSS),139

13th General Review of IMF Quotas, 223TIMSS. See Third International Mathematics and Science Study

(TIMSS)trade liberalization, 15, 194, 210–14

capture by vested interests and, 197–98trade reform

codes of conduct in, 213–14fair and ethical trade, 213–14, 222product markets and, 193–98

Trade-Related Aspects of Intellectual Property Rights (TRIPS),215–16


Trinidad and Tobagoland security, 165stunting in children, 29

TRIPS. See Trade-Related Aspects of Intellectual Property Rights(TRIPS)

Tunisiagrowth incidence curve and, 85–86health care, 144

Turkeyeducational attainment/achievement, 61school enrollment problems, 139

Ugandabasic education, 140health care user fees, 146health education, 143ICOR estimates, 98teacher absenteeism, 35

Ultimatum Game, 80–81Ultimatum Game with Multiple Proposers, 81U.N. Charter, 79underachievement, perpetuating, 2underinvestment

agriculture, 98–100human capital, 100–101industry and trade, 96–98

UN/ECE Convention on Access to Information, PublicParticipation in Decision-making and Access to Justice inEnvironmental Matters (Aarhus Convention), 217

unequal power, perpetuation of inequality and, 8–9, 229United Kingdom

Britain’s institutional transitions, 119–20, 125Ethical Trading Initiative, 213income inequality, 64inequality trends, 45intergenerational mobility/elasticity, 47, 48intergenerational transmission of inequality, 47women and household assets, 53

United NationsCommission for Africa, 218Kyoto Protocol, 216membership, 66U.N. Charter, 79U.N. Convention on the Rights of All Migrant Workers and

Their Families, 210United States

affirmative action programs, 158banking system evolution, 109educational attainment/achievement, 61Fair Labor Association, 213health insurance, 145income inequality, 64income inequality and subjective well-being, 82inequality contrasted to Mali, 56inequality trends, 45intergenerational mobility/elasticity, 6, 47–48intergenerational transmission of inequality, 47

Kyoto Protocol and, 217labor markets, 188migrant labor, 95reconstruction aid, 219safety nets, 197sharecropping, 93voice rank, 68

Uruguaydisabled children education grants, 138health insurance, 145

utilitarianism, 76–77

values and normsanomie and, 50World Value Survey, 83–84

Vietnamhealth insurance, 12, 145, 147inequality trends, 45infrastructure, 169land distribution, 166land titling program, 165Poverty Reduction Strategy, 165women’s land rights, 165

violence. See crime/violencevoice and accountability, 2, 22–23, 67–68, 178, 227, 228

enhancing judicial independence and accountability, 156–58macroeconomic policies and accountability structures, 201–3political accountability, 124, 130strengthening accountability in education, 140–41

voluntary contributions/private assistance, 221vulnerable groups

social programs for, 154–55(see also specific groups, e.g., caste systems, disability and

disabled persons, etc.)

wealth. See affluence and wealthwelfare

distributive justice and, 77intergenerational transmission of inequality, 46–48

well-being. See equity and fairness concerns; global inequity andwell-being

WHO. See World Health Organization (WHO)women

benefits from infrastructure investments, 169crime/violence against, 162discrimination/missing women and girls, 223inequality traps, 51–54land/inheritance rights, 52, 162–63, 165status and inequality traps, 20–21

World Development Report 1990, 23, 226–27World Development Report 2000, 23World Development Report 2004, 12World Development Report 2006, 2, 17World Development Reports, various, 227World Health Organization (WHO), violence data, 54World Migration Organization, 210

Index 319

World Trade Organization (WTO)fairness of decision-making process, 211–13GATS Mode IV, 210membership and representation, 66–67migration issues, 210

World Value Survey, 83–84WTO. See World Trade Organization (WTO)

youth, social programs for nonworking youth, 152–53

Zambiainfrastructure, 169trade credit, 97

Zimbabwehealth insurance, 145land reform, 163trade credit, 97


E C O - A U D I T

Environmental Benefits StatementThe World Bank is committed to pre-serving endangered forests and nat-ural resources. We printed the WorldDevelopment Report 2006 on 15 per-cent post-consumer recycled paper,processed chlorine free. The WorldBank has formally agreed to followthe recommended standards forpaper usage set by Green Press Ini-tiative—a nonprofit program sup-porting publishers in using fiber thatis not sourced from endangeredforests. For more information, visitwww.greenpressinitiative.org.

Saved:

• 148 trees

• 6,913 pounds ofsolid waste

• 62,699 gallons ofwater

• 13,579 pounds ofnet greenhousegases

• 25,215 kw hoursof electricity

Inequalities in incomes, in health, and in educational outcomes have long been a stark fact of lifein many developing countries. When such inequalities in outcomes arise from unequal opportu-nities, there are both intrinsic and instrumental grounds for concern. Because inequalities in

opportunity are often accompanied by profound differences in influence, power, and social status—whether at the level of individuals or groups—they have a tendency to persist. And because it leadsto an inefficient use of resources and to less effective institutions, inequity is inimical to long-termdevelopment. It follows that there is a legitimate role for public action in the promotion of fairnessand in the pursuit of equity, provided such action is cognizant of the primacy of individual freedoms,and of the role of markets in allocating resources.

World Development Report 2006 presents evidence on the inequality of opportunity, within and acrosscountries, and illustrates the mechanisms through which it impairs development. The Report advocatestaking explicit account of equity in determining development priorities: public action should aim toexpand the opportunities of those who, in the absence of policy interventions, have the least resources,voice, and capabilities. Domestically, it makes the case for investing in people, expanding access to jus-tice, land, and infrastructure, and promoting fairness in markets. Internationally, the report considersthe functioning of global markets and the rules that govern them, as well as the complementary pro-vision of aid to help poor countries and poor people build greater endowments. Drawing on the WorldBank’s 60 years of development experience, the World Development Report 2006 is essential readingfor understanding how greater equity can reduce poverty, enhance economic growth, advance devel-opment, and deliver increased opportunities to the poorest groups in our societies.

ISBN 0-8213-6249-6