Economic Growth, Inequality and Technology Adoption in ...at various stages of my PhD journey. In particular, I appreciate the suggestions he made before my confirmation of candidature

i

Economic Growth, Inequality and Technology

Adoption in Transitional Economies

Nawaratne Gedara Shyama Chandani Ratnasiri B.Sc (Agriculture) Hons., University of Peradeniya, Sri Lanka

M.Sc (Agricultural Economics), University of Peradeniya, Sri Lanka

School of Economics and Finance

Faculty of Business

Queensland University of Technology

Gardens Point Campus

Brisbane, Australia

This dissertation is submitted to the

Faculty of Business, Queensland University of Technology

for the degree of Doctor of Philosophy.

July 2009

ii

Statement of Original Authorship

This work has not previously been submitted for a degree or diploma at any other

educational institution. To the best of my knowledge, this thesis contains no

material from any other source, except where due reference is made.

Shyama Ratnasiri

25th

July, 2009

iii

This thesis is dedicated to my dear son, Nevindu

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my supervisory team, Dr. Radhika Lahiri and

Professor Tim Robinson (School of Economics and Finance, Faculty of Business,

Queensland University of Technology), for their supervision, guidance and advice. In

particular, my special thanks go to Dr. Radhika Lahiri, my principal supervisor, for

introducing me to macroeconomics. I also appreciate her efforts in teaching me

MATLAB as a macroeconomic modeling tool. The reading group led by Dr. Lahiri also

significantly helped me with learning advanced technical tools in macroeconomics.

Special thanks to Professor Tim Robinson for his support as well as for suggestions made

at various stages of my PhD journey. In particular, I appreciate the suggestions he made

before my confirmation of candidature seminar, which helped me to improve this study.

I also would like to thank the panel members of my PhD confirmation seminar, especially

Professor Paul Frijters, for his constructive comments. I thank Professor Greg Huffman

(Department of Economics, Vanderbilt University, USA) and Dr. Kam Ki Tang (School

of Economics, University of Queensland, Australia) for reading a preliminary working

paper of this study and making detailed comments. As well, I thank participants of local

and international conferences for providing useful feedback and facilitating discussion on

this work that I have presented to them.

I am indebted to my loving son, Nevindu, as I spent a lot of time on this study he missed

his mum quite a lot in the very first year of his life! He was so kindly with me during all

the hard times that I faced! I also thank specially my husband Sudath, without his

significant encouragement during my PhD, particularly when I was disappointed, I would

not have completed this study. I also thank my parents, sisters and my grandmother, who

helped me in numerous ways along my education journey. Their continuous support

helped me to complete this study at this time.

I also thank Elanor Adamson for proof reading my thesis. It is only with the help of QUT

International Postgraduate Research Scholarship (QIDS), I was able to undertake this

study. I therefore acknowledge and thank QUT for awarding me this scholarship.

iv

TABLE OF CONTENTS

ACKNOWLEDGEMENTS iii

TABLE OF CONTENTS iv

LIST OF TABLES vi

LIST OF FIGURES vi

ABSTRACT ix

CHAPTER 1

Introduction

CHAPTER 2

Related Literature and Motivation

2.1 Technological Progress and Economic Growth

2.2 Economic Growth and Inequality

Impact of technology adoption on inequality in the process of economic

growth

2.3 Technology Adoption, Inequality and Economic Growth: Political-

Economy Issues

CHAPTER 3

Growth Patterns and Inequality in the Presence of Costly Technology

Adoption

3.1 Introduction

3.2 The Economic Environment

3.2.1 Model 1

3.2.2 Model 2

3.2.3 Model 3

3.3 Results of Numerical Experiments and Discussion

3.3.1 Results of Experiments Conducted Using Model 1

1

8

12

26

35

37

42

48

48

59

61

63

64

v

(i) Adoption Cost Fixed Across Households and Time ( it ≡ )

(a) Experiments with the adoption-cost parameter

(b) Experiments with varying initial inequality levels

(c) Experiments with θ

(d) Experiments with α

(ii) Time-varying Adoption Costs ( it ≡ t )

(iii) Household Specific Adoption Costs

3.3.2 Experiments Conducted Using Model 2 and Model 3

(i) Experiments Conducted Using Model 2

(ii) Experiments Conducted Using Model 3

3.4 Empirical Study and Results

3.4.1. Construction of the Technology Adoption Index

3.5 Concluding Remarks

CHAPTER 4

Growth Patterns and Inequality in The Presence of Costly Technology

Adoption: A Political Economy Perspective

4.1 Introduction


4.3 Numerical Experiments

4.3.1 Political Outcome

4.3.2 Policy Choice Under Welfare Maximization and Under Political

Process

4.3.3 Experiments That Vary Income and Wealth Tax Rates

4.3.4 Experiments That Vary Initial Inequality Levels


CHAPTER 5

Concluding Remarks and Discussion

APPENDIX FOR CHAPTER 3

Appendix 3.1: Proof of Proposition 3.1

64

71

74

76

79

82

84

87

88

90

92

93

97

101

106

113

114

119

122

125

127

130

141

vi

Appendix 3.2: The optimal plans for consumption, bequests and capital

accumulation in the model 2



Appendix 3.4: Sensitivity analysis for the case of “poverty trap”

Appendix 3.5: Sensitivity analysis for the case of “dual economy”

Appendix 3.6: Implications of increasing adoption costs for technology

adoption process

Appendix 3.7: Implications of adoption costs that vary over time for the case of

“poverty trap”

Appendix 3.8: Implications of adoption costs that vary over time for the case of

“dual economy”

Appendix 3.9: Implications of household specific adoption costs for the case of

“poverty trap”

Appendix 3.10: Implications of household specific adoption costs for the case

of “dual economy”

Appendix 3.11: Sign of

*

itW

Appendix 3.12: Sensitivity analysis of the parameters in Model 2


Appendix 3.14: Data Set

APPENDIX FOR CHAPTER 4


Appendix 4.2. Proof of Proposition 4.2

Appendix 4.3. Analysing a vote on the tax rate (τ)

Appendix 4.4. Experiments that vary altruism parameter

REFERENCES

143

145

147

155

163

164

165

166

167

168

169

173

181

182

184

185

186

187

vii

LIST OF TABLES

Table 2.1:Overall technological progress in different income cohorts in the

world.

Table 3.1: Parameter Values

Table 3.2: Regression Results

18

65

97

LIST OF FIGURES

Figure 2.1

Figure 2.2.

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.6

Figure 2.7

Figure 3.1:

Figure 3.2:

Figure 3.3:

Figure 3.4:

Figure 3.5:

Figures in Chapter 2

Recent growth patterns of per capita income in different

regions in the world.

Recent growth patterns of per capita income in selected

countries in Asia

High technology exports (as a % of manufactured exports) and

per capita GDP growth.

Education attainments of different countries during 1960-2000.

World income inequality: 1970-2000.

Income inequality of some selected Asian countries: 1950-

2000.

Income shared by the richest 20% and the poorest 20% of the

population in 2007

Percentage of the population living on less than $2/day


The values of productivity parameters that determine various

properties of the model.

Poverty trap.

Dual economy

Balanced growth

Technology adoption, inequality and economic growth in the

9

10

15

16

28

30

31

33

56

57

58

58

viii

Figure 3.6:

Figure 3.7:

Figure 3.8:

Figure 3.9:

Figure 3.10:

Figure 3.11:

Figure 3.12:

Figure 3.13

Figure 3.14:

Figure 3.15

Figure 3.16:

Figure 3.17:

Figure 3.18:

Figure 3.19:

Figure 3.20:

Figure 3.21:

Figure 3.22:

Figure 3.23:

case of poverty trap.


case of dual economy.


case of balanced growth.

Number of households in Technology A and B over time for

different adoption costs.

Inequality over time for different adoption costs.

Growth rate for median agent with different adoption costs.

Number of households in Technology A and B for different

initial inequality levels.

Evolution of income over time for different initial inequality

levels.

Rate of growth of median household for varying initial

inequality levels.

Number of households in Technology A and B for varying

levels of altruism parameter.

Gini coefficient for varying levels of altruism parameter

Growth rate of median household for different altruism

parameter.

Number of households in Technology A and B for varying

levels of α parameter.

Gini coefficient for varying levels of altruism parameter.

Growth rate of poor household for different altruism

parameter.

Adoption costs vary randomly over time for the case of

“balanced growth”.

Households adopting Technology A and B in different time

periods.

Growth rates of households in different cohorts of income

distribution.

Bequests as a proportion of income during transition and at

66

67

68

72

73

73

74

75

75

77

78

78

80

80

81

83

85

86

ix

Figure 3.24

Figure 3.25:

Figure 4.1:

Figure 4.2:

Figure 4.3:

Figure 4.4:

Figure 4.5:

Figure 4.6:

Figure 4.7:

Figure 4.8:

Figure 4.9:

Figure 4.10:

Figure 4.11:

Figure 4.12:

Figure 4.13:

Figure A 1:

steady state for different altruism parameter (θ).

Evolution of inequality over time.

Growth rate of an average household.


Number of households adopting Technology A or B in

different time periods.

Gini coefficient in different time periods.

Winning in different time periods.

Proportion of households vote in favour of winning in

different time periods.

Growth rates experienced by the various cohorts of

households.

Winning value of under welfare maximization path and

political process.

Number of households adopts technology B under welfare

maximization path and political process.

Evolution of Gini coefficient over time under welfare


Growth rates experienced by the different cohorts of

households under welfare maximization path and political

process.

Evolution of inequality with and without taxes.

Growth rates experienced by the different cohorts of

households with and without taxes.

Gini coefficient in different time periods with varying levels of

initial inequality.

Growth rates experienced by the poor cohort of households

with varying levels of initial inequality.

Figures in Appendix for Chapter 3.


89

90

91

116

117

117

118

118

120

121

121

122

124

125

126

126

x

Figure A 2:

Figure A 3:

Figure A 4:

Figure A 5:

Figure A 6:

Figure A 7:

Figure A 8:

Figure A 9:

Figure A 10:

Figure A 11:

Figure A 12:

Figure A 13:

Figure A 14:

Figure A 15:

Figure A 16:

Figure A 17:

Figure A 18:

Figure A 19:

Figure A 20:

levels of adoption costs.

Inequality over time for different levels of adoption costs.

Growth rate for rich household with different levels of

adoption costs.



Inequality over time for different initial inequality levels.

Growth rate for rich household with different initial inequality

levels.



Inequality over time for different levels of altruism parameter.

Growth rate for rich household with different levels of altruism

parameter.



Inequality over time for different levels of α parameter.

Growth rate for rich household with different levels of α

parameter.




Growth rate for median household with different levels of

adoption costs.




Growth rate for poor household with different initial inequality

levels.




149

150

150

151

152

152

153

154

154

155

156

156

157

158

158

159

160

160

161

162

xi

Figure A 21:

Figure A 22:

Figure A 23:

Figure A 24:

Figure A 25:

Figure A 26:

Figure A 27:

Figure A 28:

Figure A 29:

Figure A 30:

Figure A 31:

Figure A 32:

Figure A 33:

Figure A 34:

Figure A 35:

Figure A 36:

Figure A 37:

Figure A 38:


altruism parameter.



Inequality over time for different levels of α parameter.

Growth rate for median household with different levels of α

parameter.

Adoption costs increases over time

Technology adoption, evolution of inequality and growth

patterns for the case of “poverty trap”


patterns for the case of “dual economy”


patterns for the case of “poverty trap”


patterns for the case of “dual economy”


different time periods with varying adoption costs.


different time periods with varying levels of education

expenditure parameter (α).

Gini coefficient over time for different education expenditure

parameter (α).

Growth rates experienced by the various cohorts of households





adoption costs.




162

163

164

164

165

166

167

168

169

171

172

173

174

175

176

176

177

178

xii

Figure A 39:

Figure A 40:

Figure A 41:

Figure A 42:

Figure A 43:

Figure A 44:

Figure A 45:

Growth rate for poor household with different initial inequality

levels.

Growth rate for rich household with different initial inequality

levels.




Growth rate for rich household with different levels of altruism

parameter.

Growth rate for poor household with different levels of

altruism parameter.

Figures in Appendix for Chapter 4

Growth rates experienced by the median household with

varying levels of altruism parameter

178

179

180

181

181

182

186

xiii

ABSTRACT

The stylized facts that motivate this thesis include the diversity in growth

patterns that are observed across countries during the process of economic

development, and the divergence over time in income distributions both within

and across countries. This thesis constructs a dynamic general equilibrium model

in which technology adoption is costly and agents are heterogeneous in their

initial holdings of resources. Given the households‟ resource level, this study

examines how adoption costs influence the evolution of household income over

time and the timing of transition to more productive technologies.

The analytical results of the model constructed here characterize three growth

outcomes associated with the technology adoption process depending on

productivity differences between the technologies. These are appropriately labeled

as „poverty trap‟, „dual economy‟ and „balanced growth‟. The model is then

capable of explaining the observed diversity in growth patterns across countries,

as well as divergence of incomes over time.

Numerical simulations of the model furthermore illustrate features of this

transition. They suggest that that differences in adoption costs account for the

timing of households‟ decision to switch technology which leads to a disparity in

incomes across households in the technology adoption process. Since this

determines the timing of complete adoption of the technology within a country,

the implications for cross-country income differences are obvious. Moreover, the

timing of technology adoption appears to be impacts on patterns of growth of

households, which are different across various income groups.

The findings also show that, in the presence of costs associated with the

adoption of more productive technologies, inequalities of income and wealth may

increase over time tending to delay the convergence in income levels. Initial levels

of inequalities in the resources also have an impact on the date of complete

adoption of more productive technologies.

The issue of increasing income inequality in the process of technology

adoption opens up another direction for research. Specifically increasing

inequality implies that distributive conflicts may emerge during the transitional

process with political- economy consequences. The model is therefore extended to

include such issues. Without any political considerations, taxes would leads to a

reduction in inequality and convergence of incomes across agents. However this

process is delayed if politico-economic influences are taken into account.

Moreover, the political outcome is sub optimal. This is essentially due to the fact

that there is a resistance associated with the complete adoption of the advanced

technology.

Introduction

1

CHAPTER 1

Introduction

This thesis studies a positive theory of economic growth aimed at

understanding the variation in growth outcomes in transitional economies.

Specifically, it focuses on the technological progress that is associated with

modern economic growth. The thesis addresses two facets of this issue and is

organized into two essays. The first essay looks at the economic facet of

technological progress while the second essay focuses on the socio-political

aspects associated with technological progress. In both cases, the issues are

addressed within the framework of dynamic general equilibrium models.

The motivation for this study relates to the large variation that is observed

in economic outcomes within and across countries that fall into the category of

transitional economies. In particular, the patterns of growth in these countries

exhibit a great deal of diversity. Typically, some studies in the literature on

economic growth conclude that advances in technologies lead to sustained

economic growth. In contrast, some studies suggest that at the times of rapid

technological advancements/technological revolutions, a short-run slow-down in

productivity growth is also a possibility as the economy invests in knowledge

needed to operate the new technologies. However, the question of why some of

these transitional economies grow rapidly, while others stagnate, or even

experience reversals and declines in their growth processes is as yet far from

being well understood.

Introduction

2

Another motivation for this study relates to the issues regarding the dynamics of

income distributions in transitional economies. Empirical studies in relation to

this suggest that there has been a divergence in incomes over time both within and

across these economies. For example, Quah (1996) shows that world income

distribution is shifting towards a “bi-modal” pattern which he describes as the

emergence of “twin peaks”. This pattern is also observed within countries. (See

for example Sala-i-Martin, 2006). Typically, the studies that explore this issue

suggest “barriers” or “resistance” to the adoption of advanced technology as one

of the proximate causes of differences in incomes across countries (Mokyr, 1993;

Greenwood and Jovanovic, 1990; Parente and Prescott, 1994; Galor and Tsiddon,

1997). Some of the studies that examine this issue assume that these barriers take

the form of exogenous “costs” associated with adopting an advanced technology

(See for example Greenwood and Yorukoglu, 1997; Khan and Ravikumar, 2002).

In addition, there are disparities or variations in the barriers. Generally these

disparities are caused by exogenous stochastic shocks such as institutional

variations, policy or legislation changes, socio-political issues etc. However, the

issue of the variability associated with the costs is not explicitly examined in

previous studies. Therefore, how such shocks produce differences in the incomes

within and across countries is another question that is explored in this thesis. To

that end, this thesis conjectures that a model in which adoption costs are allowed

to embody household-specific stochastic shocks has the potential to explain the

observed income differences across and within countries.

Introduction

3

The first essay presented in Chapter 2 of this thesis addresses these issues using

a simple growth model. This essay is in the spirit of studies by Greenwood and

Yorukoglu (1997), and Khan and Ravikumar (2002), which focus on exploring

the effects of barriers in the form of costs of technology adoption. They examine

the impact of technological advances on the growth of output and income

inequality, based on the assumption that adopting a new technology involves a

one-time startup cost. This cost is exogenous to their models and includes

learning new skills that are needed to use the new technology. The agents in the

economy pay this fixed start-up cost and the dynasty to which an agent belongs

continues to use the new technology. This thesis develops a framework that

relaxes the above mentioned assumptions. Firstly, this study allows for household

specific stochastic shocks. Secondly, an overlapping-generations structure is

imposed. This enables us to consider a situation in which the adoption decision is

no longer of an irreversible “one-time” nature. That is, in every time period the

new generation undertakes the technology adoption decision irrespective of

whether the previous generation switched to the better technology or not. The

latter assumption is even more important in the context of adoption costs that are

time-varying and household-specific, a phenomenon that is not analyzed in the

previous literature.

In terms of the theoretical contribution to the literature, the model developed

here has several features. First, the model predicts that differences in adoption

costs account for the timing of households‟ decision to switch the technologies.

For example, poor (rich) households who face relatively higher (lower) adoption

Introduction

4

costs may significantly defer (advance) their switch to the technology with high

productivity. It is then obvious that this would lead to a disparity in incomes

across households in the technology adoption process. Since this determines the

timing of complete adoption of the technology within a country, the implications

for cross-country income differences are obvious.

Second, an important contribution of the model constructed here is its ability to

characterize three types of growth outcomes depending on productivity

differences between the technologies. In the first situation the productivity

differences are such that all households in the economy enter into a poverty trap.

The second situation is characterized by a dual economy – a situation in which

growth rate of the two different groups of households remain distinct. In the third

situation the economy experiences a balanced growth. In extant literature there

are no studies that develop a single model that is able to characterize all of these

scenarios.1

The numerical experiments conducted using the model also generate

several novel outcomes. Firstly, it is found that households in different income-

cohorts show a significant diversity in terms of the patterns of growth over time.

For example, in the technology adoption process, while the “poor” and the

“median” households initially experience reversals or declines in their growth

rates, households positioned at the “rich” end of the income distribution show a

1 A notable exception is found in Iwaisako, (2002). This study described the cases of “permanent

growth”, “convergence to steady state”, “permanent cyclical fluctuations” and “poverty traps” in

his model. The model in this thesis however, is unique in the sense that it is also able to

characterize a situation that is referred to as dual economy.

Introduction

5

relatively smooth transition to the sustained growth. The timing of these reversals

appears to be related to the timing of technology adoption, which is, of course,

different across various income groups. It is, in fact, possible to infer that this

characteristic would translate into a corresponding diversity in the experiences of

countries that are in different positions in the world distribution of income.

Furthermore, these simulations predict that income inequality widens in

the process of technology adoption. As the households in the “rich” end of the

income distribution adopt the technology with higher productivity sooner they

receive the benefit of it earlier than the poorer households. Since the post-

adoption growth rate of output of all households is the same, income inequality

widens during the process of technology adoption.

Moreover, these experiments are also capable of providing a rationale for

the empirically observed fact that countries similar in other features but differing

initial distributions of income converge to steady state growth at different dates.

The experiments suggest that initial distribution of income has a direct relevance

to the date in which all households in the economy adopt the better technology.

Ceteris paribus, higher initial levels of inequality translate into a delay in the date

of complete adoption of the better technology, consequently delaying the process

of development.

As mentioned above, if distributional implications such as widening the

rich and poor gap are associated with the technology adoption process,

distributive conflicts among agents cannot be ignored. This issue therefore opens

Introduction

6

up another direction for further research.2 That is, it is of interest to investigate

whether such distributive conflicts may have different implications on growth

outcomes. To that end, Chapter 3 of this thesis extends the previous model to

include politico-economic determination of policies. This extension involves

endogenizing the costs associated with the adoption of the advanced technology.

In particular, the proportion of government revenue that is allocated towards

adoption-cost-reducing expenditure is allowed to be determined by a political

process.

The extension presented in Chapter 3 in fact suggests different growth outcomes

relative to the previous model. In this model, a role for the government is

introduced by incorporating taxation as a mechanism of redistribution. Without

any political considerations, this would lead to a reduction in inequality and

convergence of incomes across agents. However, in light of the historical

experience of countries, it has been argued that redistributive policies that have

been determined politically often lead to a slow-down in income convergence

(Alesina and Perotti, 1994). The outcomes that arise from the model in this thesis

also form similar implications. In particular, the outcomes of this model suggest

that the political outcome does not ensure maximum welfare of the society.

Furthermore, the results appear to support the fact that the political outcome is

more likely to represent a kind of resistance associated with adoption of advanced

technology. This resistance can be reflected in the alternate policy choice, i.e,

2 There is limited work offered in the extant literature in relation to this issue. However a notable

exception is found in Krusell and Rios-Rull (1996).

Introduction

7

whether government tax revenue should be channeled towards lump-sum transfer

or towards adoption-cost-reducing expenditure. Typically, the latter will promote

faster adoption of advanced technologies, and reduce the income inequality. In

this model, the agents at the lower end of the income distribution resist adoption

of advanced technologies through the political process. Moreover, the model also

supports the idea that income inequality is not harmful for economic growth, as

predicted by Li and Zou (1998) among several others. Essentially, initial levels of

inequality in income and wealth in this model promote quicker adoption of more

productive technologies. In the presence of high inequality, a redistributive

mechanism with proportional tax enables relatively poor individuals to switch to

more productive technology quicker, consequently leads to a faster economic

growth.

The remaining chapters of this thesis are organized as follows. The next chapter

will review some related literature. Chapter 3 presents the benchmark model of

this thesis. This chapter details the economic environment of the model as well as

important implications of the model for technology adoption and economic

growth. The political-economy extension of the benchmark model is presented in

Chapter 4. Chapter 5 presents the concluding remarks.


8


9

CHAPTER 2


The process of economic growth was first described by Malthus (1798) as

the “perpetual struggle for room and food” and has been increasingly researched

and explored along various dimensions in the subsequent literature on economic

growth. In contemporary literature, economic growth has been defined as a

“sustained increase in per capita or per worker product”. This process is often

viewed as a transition process that is characterized by two distinct phases. The

first is “Malthusian stagnation” where the economic growth of a nation stagnates

for a long period of time, and technological progress is negligible. This phase is

followed by a “Modern growth” regime, the second phase, which is characterized

by a steady growth in income per capita and by increases in the level of

technology used in a country. The transition between these phases is often

referred to as take-off as economic growth „takes off” from stagnation to sustained

growth. This is observed in the data as well. However, the timing of the take-off

and its magnitude are different across countries/regions. For example, Western

European countries have taken off to sustained growth at the beginning of the 19th

century, whereas in Latin American and Asian countries this take-off took place at

the end of the 19th

century (Maddison, 2005).

The process of transition is always associated with certain structural

changes, such as a shift of employment and production across sectors as

emphasized by Clark (1940), Nurkse (1952), Solow (1956) and Kuznets, (1957)


10

and further studied by others such as Laitner (2000) and Kongsamut et al. (2001).

These structural changes are evident from historical data, for example historical

statistics for the USA and UK document a massive decline in the relative share of

employment in agriculture over the last two centuries. The employment shares in

agriculture in these two countries were well above 75% of the total labour force at

the beginning of 19th

century. Over time this share has continued to decline and

now it is reported to be less than 5% in both countries. On the other hand, the

employment share of manufacturing had increased by the mid 1800s but has since

been decreasing. The share in labour force in services has been increasing

throughout this period, and more that 70% of the work force in these countries is

now reported to be working in the services sector. Similar phenomenon is

observed in the production shares over the same time frame.

Figure 2.1 Recent growth patterns of per capita income in different regions in the

world. (Drawn by using data in Maddison, 2009).3

3 The classification of regions follows Maddison, 2001. Purchasing Power Parity (PPP) conversion

in this study adopts Geary–Khamis (GK) method invented by Geary and Khamis to measure GDP

per capita. The benchmark year is 1990 therefore the GDP per capita estimates are in 1990 GK$.


11

Another possibly more important phenomenon accompanied by the above

transition process is the evolution of income. The income evolution has exhibited

a significant diversity within and across countries/regions, in the process of

economic growth. For example, according to Maddison‟s (2009) estimates, GDP

per capita in Western Europe4 before 1800 increased very slowly, however, later

in the early 19th

century, there was a steady and rapid increase in per capita

output. The Western offshoots (i.e. USA, Canada, Australia and New Zealand)

grew at an even smaller rate before 1800, but, in the second half of 18th

century

these countries exhibit a dramatic increase in per capita income. In fact, this

growth is even faster than for Western European countries. This type of

divergence in evolution of incomes seems to be an important element associated

with modern economic growth. In fact, there is a significant difference in per

capita income within and across countries/regions today. Figure 2.1 illustrates the

most recent growth patterns of per capita income in different regions in the world

from 1950 to 2006.

Figure 2.2. Recent growth patterns of per capita income in selected countries in

Asia (Drawn by using data in Maddison, 2009).

4 The countries in the Western European region refer to Maddison‟s (2009) classification


12

Moreover, a closer look at evolution of incomes in individual countries

shows considerable divergence as well as diversity in patterns of growth. For

example, in Asia, which is the most populous and invariably one of the important

regions for attention in the study of modern growth experiences. Figure 2.2 looks

at a few countries in Asia that had approximately the same level of per capita

income in 1950. The figure illustrates that, the growth in South Korea was very

rapid. In particular, there was a dramatic boom in South Korea, a country that had

experienced so called “Miracle Growth” during the last three to four decades.

Similar growth miracles were observed in countries including Hong Kong,

Singapore, Taiwan etc. On the other hand, China and India had experienced

extended periods of stagnation before taking-off to rapid growth in late 1990s. In

fact, these two economies were among the poorest countries in the world in the

1960s but are now experiencing the transition process, and have risen to represent

newly emerging economies in the region. The economic growth of Nepal appears

to have experienced ongoing stagnation throughout the period. For instance,

Nepal‟s current per capita income is roughly equal to the 3 international dollars

per day and Nepal still remains among the poorest countries in the world.

Overall, it is apparent that the patterns of growth of different economies vary

significantly, and consequently current levels of per capita income differ across

countries.

Understanding such divergence in incomes and international income

differences is one of the most important, and perhaps the most challenging task

for researchers. This is because causes of cross-country differences in income per


13

capita are difficult to identify. To that end, Baumol‟s (1986) suggestion of

convergence clubs can be considered an innovative idea. This hypothesis enables

economists to interpret structural characteristics specific to countries as possible

factors contributing to divergence across regions (See Quah, 1996 for supporting

empirical evidence). Moreover, the idea of conditional convergence; the

narrowing of income gap between countries which are similar in observable

structural features, also appears to provide a similar theory (See Barro (1991) and

Mankiw et al. (1992) for cross-country evidence in support of this view). These

ideas emphasize that country-specific characteristics might be important in

explaining the observed diversity in economic growth as well as explaining cross-

country income differences. In the literature on economic growth, growth

decomposition exercises, in particular, suggest that the most important sources of

growth that appear to have an impact on income differences are technological

progress, accumulation of physical capital, and improvement in human capital. Of

course, both these types of capital accumulation typically embody technological

progress. The following section therefore reviews issues pertaining to

technological progress and economic growth, as this line of literature has an

immediate bearing on the benchmark model developed in this thesis.

2.1 Technological Progress and Economic Growth

The improvement in technologies involved in production is considered a

driving force behind economic growth. In particular, during the period 1760-1830,

the emergence and continuous improvement of new techniques of production in


14

Western Europe and especially in Britain had an enormous impact on productivity

growth. Within the context of the literature on economic growth, this is often

discussed in relation to the industrial revolution (1760-1830) that triggered the

transition from stagnation to sustained growth.5 Empirical evidence in this regard

suggests that acceleration in technological progress associated with the industrial

revolution generated a significant increase in average per capita output. For

example, in the Western Europe, the average level of income per capita remained

below $1204 per year until 1820, and during the period of 1820-1992, per capita

product increased thirteen-fold (Maddison, 1995). The average rate of growth of

per capita output in the same region increased from the pre-industrial level of

0.15% to 0.95% per year after the industrial revolution (Galor, 2005).

This type of invention and adoption of improved technologies often takes

place all over the world continuously. Can this account for differences in

evolution of incomes in modern economic growth? In order to answer this

question, this section presents some empirical evidence of recent trends of growth

in many economies in relation to technological progress. Much of the following

discussion concerns the level of technology that helps to determine income levels

but not the level of income that affects the ability to gain access to technology.

Technological progress is not easy to measure in a summarising statistic.

As manifested in the empirical literature it has been measured in various indirect

ways and has often been compared with a country that plays a “leading” role in

5.This issue has been comprehensively discussed in Mokyr (1990) and Mokyr (2005).


15

technology adoption. For example, just after industrial revolution, the UK

reported the highest level of productivity in the world and was supposed to be a

world “leader” in the invention of advanced technologies. Therefore, comparisons

with the UK provided information on technological progress relative to the

leading economy.

However, turning to measures of technology adoption, some measures

focus on the impact of technological change rather than technological

advancement itself. For example, some measures compute technological changes

as some sort of „residual‟ that emerges from a production function when all input

measurements have been taken into account. In particular, within the growth

accounting framework, changing pace of technical advance can be measured by

performance in the economy in terms of Total Factor Productivity (TFP). It is

interesting to look at TFP gains in Newly Industrialized Economies (NIEs; i.e

Singapore, Hong Kong, South Korea, and Taiwan) in which rapid growth started

in 1960s, as well as in India and China where striking growth performances are

exhibited more recently. During the period 1970 to 2005, TFP has contributed

approximately 1.5% to growth in NIEs (Jaumotte et al. 2006), while

corresponding figures for China and India are 3.6% and 1.6% respectively during

the period of 1978 to 2004 (Bosworth and Collins, 2008). Moreover, in the recent

past, the average level of TFP among the poorest nations in the world is about 5%

of that of the USA, while the same figure for high-income OECD (Organization

for Economic Co-operation and Development) countries is reported as high as

77% (World Bank, 2008). Nevertheless, how can it be suggested that TFP growth


16

results directly from technology advancement and adoption? Wirz (2008)

develops a theoretical framework that finds the TFP growth resulting from

technology adoption and calibrated this model to Chinese economic performance

during 1978-2005. This suggests that 80% of TFP growth in China can be

explained by technology adoption. It is then possible to suggest that TFP is a

reasonable proxy to explain the technology adoption in transitional economies.

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

4.50

5.00

China India South

Korea

Singapore Malaysia United

Kingdom

High

income:

OECD

High

income

Sub-

Saharan

Africa

Incr

ease

rel

etiv

e to

19

88

val

ue

High technology exports (as a % of

manufactured exports)

Per capita GDP

Figure 2.3 High technology exports (as a % of manufactured exports) and per

capita GDP growth. (Data source World Bank, 2007)

Another measure that focuses on the impact of technological progress is

the share of high-tech activities in manufacturing exports. The figure 2.3 shows

the increase in share of high-tech imports and the increase in per capita income

during the period 1988 - 2006 for some selected countries and groups of


17

countries.6 Again, this figure particularly looks at NIEs, China and India as these

countries show striking growth performance during the recent past. The figure

illustrates that there is a four-fold increase in Chinese high-tech exports while

there are significant increases in NIEs and India. There is no change in the share

of high-tech exports in the Sub-Saharan African countries during 1988-2006.

Chinese GDP has risen by approximately 4.5 times during this period while the

same figure for Sub-Saharan Africa is as small as 0.12. It follows that the share of

high tech exports provides, if not ample evidence, at least approximate inference

on the relationship between the extent of technological achievement of a country

and its economic performance.

0

10

20

30

40

50

60

Hong

Kong

Indonesia South

Korea

MalaysiaSingapore India China Bangladesh Nepal

Ed

uca

tio

n a

ttai

nm

ent

(%

of

tota

l p

op

ula

tio

n a

ged

25

an

d a

bo

ve)

1960 1970 1980

1990 2000

Figure 2.4 Education attainments of different countries during 1960-2000.

(Data source: Barro and Lee, 2000)

6 For China the period under consideration is 1992-2006 while for Sub-Saharan Africa it is 1995-

2003. This categorization follows the World Bank classification of countries based on income.


18

In contrast to the above measures, some studies emphasize inputs into

technological advancement, such as education levels, numbers of scientists and

engineers, and expenditures on research and development (R&D) or R&D

personnel as proxy measures of technological achievement (Archibugi and Coco,

2005). In fact, these variables are assumed to be determinants of the technological

progress of countries. Many economists speculate that the rapid growth

performance observed recently in various countries including China and India is

associated with technological advancement resulting from increased educational

attainment, as this education facilitates rapid diffusion of advanced technologies

(Caselli and Coleman 2001). Figure 2.4 looks at education attainment during the

period of 1960 and 2000 in some countries that had rapid economic growth during

the recent past, including China and India.7 This figure illustrates the percent of

population aged 25 and above who attained a secondary or above level of

education. The educational attainment in countries like Hong Kong, South Korea,

Singapore and Malaysia has risen during the last four decades, however India and

China have not yet made it to the same level. On the other hand, the countries that

have not taken off from stagnation or that have taken off very recently, for

example Nepal, are far behind in their level of educational attainment. Overall, it

appears that the extent to which the educational attainment of a country is raised

has important ramifications on the growth outcomes of the country. If the degree

to which a technology is used within a country depends on the level of educational

achievement of the population (because such education helps individuals to learn

new techniques or adopt available technology), then the data presented in Figure

7 This exercise employs the Barro and Lee (2000) database of education.


19

2.4 provides substantial evidence in favour of a positive link between

technological progress and economic growth.

Table 2.1.Overall technological progress in different income cohorts in the world.

Change in the overall technological achievement

Income Groups Absolute change Change in the index

(%)

High-income

Upper-middle-income

Lower-middle-income

Low-income

0.068

0.046

0.028

0.022

77

109

103

161

(Source: World Bank, 2008)

All the measures of technology discussed above have their own strengths

and weaknesses. In particular, most of the above measures are proxy measures of

technology achievement and they are limited in their ability to provide realistic

information on technological progress. Aiming to overcome such weakness in

existing measures, the World Bank has developed a summary index. This index

encloses a range of different dimensions including the extent of scientific

innovation and invention, the diffusion of older technologies, and the diffusion of

newer technologies (See Table 2.1). This table reports the absolute and relative

changes in the overall technological achievement over a decade starting from

early 1990 for different income cohorts in the world. It is apparent that in absolute

terms, technological progress over this decade is larger among higher-income

countries than lower-income countries. Meanwhile the relative improvement of

technology as a percentage of its level at the beginning of the period is lowest in

high income countries while highest for low income countries (for which data is

available) followed by upper middle income countries. Generally speaking, in


20

terms of average growth rates during the same decade, the developing countries

have outpaced the developed countries, implying that the pace of technological

improvement is related to the rapid growth experiences of most of the developing

countries.

The above mentioned issues that attempt to examine the accountability of

technological progress to economic growth have been theoretically explored using

various approaches. Typically, they stem from the neoclassical growth models due

to Solow (1956) and its variants such as Cass (1965), Koopmans (1965), Lucas

(1986). These models basically treat technology as an exogenously given factor,

hence long-run sustained growth rate is solely determined by the exogenous rate

of technical change. For example, in the Solow model, returns to capital diminish

while labour efficiency is constant. It is for this reason that the economy

eventually converges to a steady-state growth rate in the which capital-labour ratio

is constant. Hence, the only source of growth in per capita output is technological

progress. Even though these models are handy to explain the income

convergence, the technological progress and technological choices are not

explicitly handled, so these models are limited in their ability to explain

differences in levels of per capita output across countries.

As an alternative, Romer (1986) developed a model that rules out the

assumption of exogenous technological change. Later on, a considerable number

of studies with this type of endogenous growth model appear in the literature

including Romer (1987, 1990), Rebelo (1991), Grossman and Helpman (1991),


21

Aghion and Howitt (1992). These theoretical works presume growth through the

“accumulation of knowledge”, either through learning by doing (Romer, 1986;

Young, 1991) or through technological innovation as a result of Research and

Development (R&D) (Romer, 1990; Grossman and Helpman, 1991; Aghion and

Howitt, 1992). Furthermore, Romer‟s (1986) pioneering work assumes that the

aggregate production function exhibits increasing returns to scale, which is a

common feature of models of endogenous growth.8 This feature obviously forces

the output per capita over time to increase, thus economic growth is unbounded

unlike in the case of neoclassical models. Overall, from the point of view of

modern growth experiences of countries these types of models seem to perform

more plausibly as they can better account for the non-convergence in incomes

across countries/regions.

While these endogenous models emphasize role of increasing returns to

scale in the development path, literature acknowledges the fact that capital

deepening process that is seen in such model can occur in the case of constant

returns to scale technologies also (See for example, Glomm and Ravikumar,

1992). This process has been analysed, using a simple stylised linear model of

endogenous growth. This type of models is known as AK models, and is widely

employed after Barro‟s (1990) work to explain divergence of incomes. The

assumption of constant returns to capital is more plausible here as capital is

broadly defined to encompass both human and physical capital. Sustained growth

is possible in this type of model as continuous capital accumulation can act as an

8 In fact, in Romer‟s (1986) model K and L in the familiar production technology, Y=F(K,L,A)

(with K, L and A denoting capital labour and technology) will exhibits constant returns to scale,

while endogenized A will exhibit increasing returns to scale.


22

engine of growth. In fact, AK formulation emphasizes differences in capital

deepening as a source of non-convergence in incomes. See Acemoglu and

Guerrieri (2008), for a model with capital deepening as one of the sources of

economic growth. (See also Chongvilaivan (2008) for some empirics).

Besides the simplicity, the AK framework appears to be quite useful as it is

typically known to imply divergence in international incomes. Moreover, the

presence of linearity in this framework allows the economy always to grow at a

constant rate, therefore the model generates sustained growth. The capital

deepening process combined with the technological improvements that increase

the relative output of the sector in question is another useful feature of the AK

formulation. Most importantly, it is a tractable framework for the analysis of

aggregate technological change, which is of course, the main theoretical concern

of the model developed in the thesis.

Turning to look at the empirical validity of AK models, Jones (1995) first

argues that implications of AK-type models are inconsistent with data. He tests the

key prediction of AK theory which implies a positive link between investment

rates and GDP growth. The study compares investment as a share of GDP and the

growth rate of GDP for 15 countries that belong to the Organisation for Economic

Co-operation and Development (OECD). Jones‟ (1995) critique was defended by

McGrattan (1998). They have presented data on the investment share and GDP

growth to show that the key prediction of AK theory is consistent with the data

when versions of the model and the data are compared appropriately. However the


23

literature in relation to the empirical validity of AK model is inconclusive. (See

for example Ejarque and Reis, 2003, 2005 and Romero-Avila, 2006).

The discussion so far has emphasized that the diversity in evolution of

incomes associated with modern growth experiences of various countries reflects

technological advancement in these countries. A set of literature that deals with

the above issues suggests that countries or regions with access to similar

technologies tend to converge to an identical rate of growth in income levels in

the balanced growth path (Baumol, 1986; Barro and Sala-i- Martin, 1992).

According to empirical evidence presented in these papers, countries converge to

the steady state level of income at a rate of approximately 2 to 3 percent a year

(See also Barro, 1991; Barro and Lee, 1994a; 1994b). Furthermore, Caselli et al.

(1996) suggest that substantial differences in the state of technology across

countries play a major role in the observed differences in per capita income levels

in the steady state. He finds that countries converge to the steady state level of

income at a relatively higher rate, approximately 10 percent a year, when he

accounts for country specific effects that represent differences in the use of

technology.

The above strand of theoretical literature also has some implications in

relation to the persistent differences in the levels of production technologies used

across countries.9 For example, depending on economic circumstances, some

9 There is also some empirical literature that explores the causes of differences in the levels of

production technologies used across countries. For example, among several others, Comin and

Hobijin (2004) show that important determinants of the state of technology used across countries


24

countries may use technologies with low productivity, while others use an

advanced technology for the same production sector. Several explanations have

been proposed for the differences in the state of technologies across countries.

Most of the studies discussed above view “factor endowments”, in particular

physical and human capital endowments, as a useful explanation for differences in

the state of technologies across countries. Basu and Weil (1998) therefore suggest

that new technologies can only be implemented if countries have appropriate

factor endowments. Moreover, Mokyr (1990) and Rosenberg and Birdzell (1986)

document that the adoption of new technologies often experiences a severe

“resistance” which leads to differences in the state of technologies across

countries. This resistance may be derived from sources such as government

regulations, political pressure, violence, unionism etc. Parente and Prescott

(1994) suggest that countries differ in terms of the “barriers” to technology

adoption that they place on the process of transition. These barriers can take

different forms, such as institutional barriers, policy induced barriers, legal

constraints, monopoly rights etc.

From a theoretical point of view, there are different ways of modelling

barriers to technology adoption. For example using a variant of the neoclassical

growth model, Parente and Prescott (1994) modelled these barriers as additional

investment that a firm must make to adopt a more advanced technology. This

study refers to such investments as “technology adoption investments”. Allowing

relates to country‟s human capital endowment, type of government, degree of openness to trade and

adoption of predecessor technologies. See also Bernard and Jones (1994).


25

these barriers to appear in the legal or policy parameters associated with

technology adoption is another way of modelling barriers to technology adoption.

For example, Ngai (2004) modelled such barriers in terms of policy parameters

that discourage investments within the context of a standard growth model. A

similar method is employed in Holmes and Schimitz (1995). In this study barriers

appear in legal parameters that hinder consumption of goods produced using

advanced technology.

Another way of modelling these barriers is to assume the existence of

“cost” associated with adopting an advanced technology. There are several studies

that examine growth in the presence of costs associated with adoption of advanced

technology.10

For example, the adoption cost in Easterly et al. (1994) relates to

how workers learn to use new technologies sequentially in the process of

adoption. In their model accumulation of human capital means learning how to

work in the new technology. Hornstein and Krusell (1996) examine whether

investment-specific technological change can slow down productivity growth. In

their model, they summarize all costs of adoption that relate to the new

technology in a single variable which, in part, accounts for learning. Their

empirical work suggests that the productivity slow-down in USA and elsewhere in

the early 1970‟s was due to increases in the rate of investment specific

technological change. Moreover, Greenwood and Yorukoglu (1997) suggest that

when a leap in the state of technologies occurs, successful implementation of a

new technology requires skilled labour. Therefore adoption of new technologies

10

See Bessen (2002) for a comprehensive review of literature on costs associated with technology

adoption.


26

involves a significant cost in terms of learning such skills. This learning process,

as well as skill formation, is endogenous in their model. Their model suggests that

slow-down in productivity growth is possible in periods of technological

innovation. Furthermore, in a variant of the standard endogenous growth model,

Leung and Tse (2001) explore the technology choice in the presence of one-time

fixed cost of technology adoption. In this model, this cost is represented in terms

of the units of capital that individuals must pay in order to adopt the advanced

technology. As it is a sunk cost, agents will never switch back to the old

technology. Moreover, Canton et al. (2002) examine the costs in terms of the

forgone benefits that account for the productivity slow-down. In particular, the

adoption cost in their model represents the forgone leisure time of the workers that

they utilize to acquire training to operate new technologies.

Before proceeding, to sum up, the discussion so far has investigated the

modern growth experiences of countries, and in particular very large differences

in income per capita across countries with the accountability of technological

progress for these growth outcomes. The literature reviewed suggests that

technological progress is perhaps the most important cause that leads an economy

towards a sustained growth. This review therefore implies that technology

adoption is a potential candidate for explaining the differences in growth patterns

observed within and across countries. In addition to technological progress, the

empirical patterns discussed in the cross sectional variation in income and wealth

which manifests itself in the form of inequality is another aspect that plays a

major role in economic growth. The following section therefore reviews the


27

effects of inequality within and across countries on the growth performance of

nations.

2.2 Economic Growth and Inequality

The first part of this section will look at the recent trends observed in

income inequality while the next part will investigate the theories linking income

distribution and economic growth with their empirical relevance. Finally, the

impact of technological progress on the evolution of inequality over time will be

discussed.

When the evolution of income inequality is considered, two facets deserve

particular attention. The first relates to inequality in the world income distribution.

What has happened to the world income inequality during the past few decades or

so remains a debate among scholars. Central to this debate are differences in the

concept of world income inequality as employed in the literature. One definition

refers to the inequality in the distribution of mean incomes of different countries,

while another refers to the inequality in the country means weighted by their

population sizes. A third definition refers to global inequality in the distribution of

individual incomes. Some empirical reports suggest that as income rises over

time, the world distribution of income shifts to the right with a relatively larger

spread. The implication here is that there is a tendency to increase the gap

between per capita levels of incomes of the rich and the poor throughout the

world. For example, Bourguignon and Morrisson, (2002) suggest that during the

period of 1950 to 1992, world income inequality has increased from 0.64 to 0.657


28

and 0.805 to 0.855 measured in terms of Gini coefficient and Theil Index

respectively. In essence, this study suggests that inter country inequality is

increasing.

0.6

0.65

0.7

0.75

0.8

0.85

0.9

1970 1975 1980 1985 1990 1995 2000Year

Gini Index

Theil Index

Figure 2.5 World income inequality: 1970-2000. (Drawn by using data in Sala-i-

Martin, 2006)

A contrasting outcome can be seen in Figure 2.5 which reports the

evolution of inequality measured in terms of population-weighted Gini index and

Theil index in the world from 1970 to 2000, as documented by Sala-i- Martin

(2006). This study reports that there is 2.4% decrease in the global income

inequality according to the Gini index while according to the Theil index a

decrease of 3.7% occurred over the same period. However, this study further

suggests that rapid growth particularly in China and India which accounts for

approximately 40% of the world population strongly influenced the observed


29

decline in inequality. Excluding China and India global income inequality would

instead exhibit a roughly increasing trend during these four decades. Moreover,

Park (2001) considers the trend in global inequality of individual incomes and

shows that global income distribution was getting more unequal during 1960s, but

later inequality was on a secular decline between 1976 to 1992.

Secondly, we look at the evolution of income inequality within individual

countries. Usually the empirical measures of within-country inequality generally

encompass the problems relate to inconsistency and quality. For this reason, cross

country comparability of inequality is difficult. However, this section attempts to

present some empirics of recent trends in inequality in some countries and regions

in the world.

A recent investigation of income inequality in the USA, UK and OECD

countries by Atkinson, (2003) suggests that inequality increased in USA and UK

during the second half of the 20th

century. Contrary to this, inequality has

decreased in OECD countries overall in the same period though the trend is not

remarkable in the last decade (See, Zartaloudis (2007) for further discussion of the

levels, trends and causes of income inequality in Europe and the USA). The WIID

(World Income Inequality Database) version 2.0c (2008) published by the World

Institute for Development Economics Research of United Nations Universities

also reports statistics in support of this phenomenon in the USA, the UK and

OECD countries. Using the same data source, Figure 2.6 illustrates the pattern of

inequality over the same period for some emerging economies in Asia. The figure


30

reports that the inequality levels in China and India are remarkably higher at the

end of 1990s than during the previous couple of decades. Taiwan and South

Korea, representatives from NIEs have declining trends generally with some

volatility. Inequality level in Bangladesh was approximately 41% in 1960 and has

moved with some fluctuation to approximately 33% by 2000. Countries in other

regions of the world also exhibit varying trends. For example, in Latin America

inequality increased in Argentina, a country which has had a considerable growth,

and also increased in Bolivia. In contrast, inequality declined slightly in Brazil

and Mexico. Most of the countries in the African sub continent, for example,

Nigeria, Sudan, and Uganda had increased inequality in the latter part of the

twentieth century, while countries like Kenya and Malawi had a decreasing trend.

20

25

30

35

40

45

50

55

60

1950

1960

1970

1980

1990

2000

Year

Ineq

ual

ity

lev

el

T aiwan India China Bangladesh S.Korea

Figure 2.6 Income inequality of some selected Asian countries: 1950-2000. Data

(Source: World Institute for Development Economics Research of United Nations

Universities -World Income Inequality Database, 2008)

Moreover, the current level of inequality remains the highest in Latin

American and Caribbean regions while surprisingly it is at its lowest in the South


31

Asian region (World Bank, 2005). The Figure 2.7 illustrates the income shared by

richest 20% and poorest 20% of the population in different geographical regions.

In fact, the poorest income holders who live on less than 2 US dollars a day

currently account for about half of the world‟s population (Azariadis and

Stachurski, 2005). According to Madisson (2009) Zaire was the poorest country

in 2006 in terms of PPP adjusted per capita income which was $230 a year. Since

record keeping began the per capita income of Zaire has stagnated or declined

significantly, experiencing a poverty trap. This type of poverty traps is also

observed in neighbouring Uganda, Tanzania, Rwanda and Burundi whose annual

per capita incomes are strictly less than $1000. Figure 2.8 illustrates the

percentage of population living on less that $2 a day. According to the figure

highest poverty level is reported in South Asia, particularly in India in 2004.

0

10

20

30

40

50

60

70

East Asia

and

Pacific

Europe

and

Central

Asia

Latin

America

and

Carrebean

Middle

East and

North

Afrfica

South Asia Aub

Saharan

Africa

High

Income

Percen

tag

e o

f in

co

me

Poorest 20%

Richest 20%

Figure 2.7 Income shared by the richest 20% and the poorest 20% of the

population in 2007 (Source: World Bank, 2005).


32

It has become apparent that the temporal movement of inequality within a

country is unique, thus country specific investigations are required to

systematically conclude what causes such diversity across countries. The

discussion above highlights to a certain extent whether these movements in

inequality have a connection to the growth performances of these economies.

There is a large body of empirical literature that explores this link. Some studies

show a positive relationship between income inequality and economic growth,

while other studies show a negative relationship or suggest inconclusive

relationships. Early literature suggests that the link between inequality and

growth, at least in the early stages of the growth process, would be positive. For

instance, since Kuznets (1955) and Kaldor (1957), studies on inequality have

shown that income inequality within a country may lead to higher economic

growth. In particular, Kuznets (1963) suggests that there is an inverted-U shaped

relationship between inequality and growth. This is interpreted as follows: higher

initial levels of inequality encourage economic growth which eventually reduces

the inequality through a “trickle down” mechanism. While Acemoglu and

Robinson (2002a) further analysed this hypothesis and argue that development

does not necessarily induce a Kuznets type phenomenon, recent work of Borissov

and Lambrecht (2009) finds a relationship that resembles the inverted-U shaped

curve hypothesized in Kuznets (1963) under reasonable assumptions.

On the other hand, work of Benabou (1996) suggests that countries with

higher initial inequality grow slowly relative to countries with low inequality. He

compares two countries (South Korea and Philippines) that were similar in


33

macroeconomic features except for the initial inequality levels in the early

1960‟s.11

Alesina and Rodrik (1994) also reveal that greater inequality in the

distribution of income and wealth seems to slow down the economic growth.

Their measure of inequality is the Gini coefficient of income and land

distributions. In a similar empirical setting, Persson and Tabellini (1994) use a

different measure for equality which is represented by income share of the third

quintile of the income distribution and suggest that equality appears to promote

economic growth. Supporting studies include Perotti (1992, 1996) and Figini

(1999) among several others. However, this line of research generally reveals that

the level of inequality, particularly initial inequality, is unfavourable for long-run

economic growth.

0

10

20

30

40

50

60

70

80

90

100

East Asia

and the

Pacific

China Rest of

East Asia

and the

Pacific

South

Asia

India Rest of

South

Asia

Europe

and

Central

Asia

Middle

East and

North

Africa

Sub-

Saharan

Africa

Latin

America

and the

Caribbean

World

(%) 1990 2004

Figure 2.8 Percentage of the population living on less than $2/day (Source: World

Bank, 2005).

11

Parente and Prescott (1994) discover that the observed differences in these two countries relate

to the extent of the technology adoption barriers present in these countries.


34

In contrast, there are studies that support the earlier hypothesis that

inequality appears to promote long-run growth. For example, using panel data to

control for time-invariant country specific effects, Forbes (2000) finds that in the

short and medium term, inequality affects growth positively and significantly. Li

and Zou (1998) also find positive association between income inequality and

growth. Overall, however, empirical evidence on the link between inequality and

growth is considered inconclusive (See the survey by Zweimuller, 2000).

Moreover, theoretical models in the literature suggest that this link should

be inconclusive. Specifically, it is suggested that the nature of the link would

depend on underlying conditions (Banerjee and Newman, 1991; Galor and Zeira,

1993 and Aghion and Bolton, 1992, 1997). However the early classical view

suggests that inequality stimulates physical capital accumulation and thus promotes

economic growth (Bourguignon, 1981). On the other hand, some studies, for

example Galor and Zeira (1993), propose that growth is affected by the initial

distribution of wealth through investment in human capital. Specifically, they

suggest that wealth distributions with wealthier households who can invest in

human capital have a positive impact on the growth of output. Moreover,

Greenwood and Jovanovic (1990) suggest financial-structure development while

Aghion and Bolton (1997) suggest imperfect capital markets as possible

explanations for a Kuznets curve-type link between inequality and economic

growth. In particular, Greenwood and Jovanovic (1990) suggest that in the

intermediate stage of development income levels go up due to the formation of

financial structure and redistribution of incomes. As a result, growth rates increase


35

rapidly and this leads to increased income inequality. Eventually, the rate of

economic growth and the income distribution stabilize with a fully developed

financial structure.

In addition to the above underlying conditions that determine inequality

over time, adoption of advanced technology also contributes to the inequality and

economic growth. The following section therefore reviews this hypothesis in detail,

because this has direct relevance to the models developed in this thesis.

Impact of technology adoption on inequality in the process of economic growth

This line of literature deals with the importance of the role of technology

adoption in the determination of the impact of inequality in the process of

economic growth. For example, Galor and Tsiddon (1997) suggest that the

technology adoption process is associated with widening income inequality. The

reason is that technology adoption is often associated with skilled labour that is

necessary to operate the advanced technology. Due to higher returns earned by

skilled labour relative to unskilled labour, income inequality increases. Similar

examples are given in Gollin et al. (2002) and (2004). Moreover, if the

technological advancements are skill-biased or ability-biased they also appear to

be widening the wage inequality among workers (Caselli, 1999; Galor and Moav,

2000).

The studies that examine the technology adoption process suggest that

adoption of new technologies involves skilled labour and acquiring such skills


36

involves significant costs associated with learning a new method of production.

For example, cost component in Caselli (1999) includes the costs of acquiring

skills that are needed to operate the advanced technology. Khan and Ravikumar

(2002) identified this cost of adopting a new technology as an exogenous fixed

cost while Greenwood and Yorukoglu (1997) suggest that there is a one-time

start-up cost associated with adoption of a new technology.12

These models

suggest that, in the presence of technological revolutions, the agents who are

capable of paying the fixed or start-up cost can upgrade their skills and adopt the

new technology. As the return to skill increases with advances in the technology,

increasing income inequality is inevitable. Moreover, Khan and Ravikumar (2002)

suggest that higher productivity of new technology induces households to adopt

the new technology sooner, and a higher fixed cost postpones the adoption date.

The motivation for the study in this thesis relates to issues that are

explored to a limited degree in the above literature. In particular, a common

feature of the models discussed above is that the households face a one-time

adoption decision which is irreversible (See for example Greenwood and

Jovanovic, 1990 and Khan and Ravikumar, 2002). The implication here is that if a

household makes the decision to switch to a new technology, the dynasty to which

the household belongs to should continue with the use of the new technology.

Moreover, the contemporary literature assumes that this adoption cost is a one-

time start-up cost or an exogenous fixed cost incurred by households.

Furthermore, heterogeneity in costs, if present in these models, essentially reflects

12

The models in Leung and Tse (2001), Jones (1979) and Marjit (1988, 1992 and 1994) also

involve a fixed adoption cost.


37

the cognitive ability of the agents or agents‟ access to credit (See Caselli, 1999).

This thesis develops a simple growth model that considers the issues discussed

above. This model has interesting implications for technology adoption process

and economic growth. They are presented in Chapter 3.

The discussion so far has highlighted the role of technology adoption on

economic growth and evolution of inequality over time from an economic point of

view. However, in recent research on economic growth it is argued that

economics alone cannot explain the diverse growth outcomes. This line of

literature has investigated other causes of such diversity. Studies suggest that

different institutions (for example, whether property rights are properly enforced)

and/or policies (such as tax, subsidies, distortions) also have an influence on the

economic performance of countries (Acemoglu, 2008). In fact, economic outcome

and policy choices together may provide a wider scope for useful understanding

of variance in growth outcomes within and across countries. Therefore, the

political economy concerns that lead to different policy choice are important to

investigate. The following section therefore reviews literature that discusses

political issues entailed in the technology adoption process that is associated with

modern growth. These issues are directly relevant to the political economy model

developed in this thesis.


38

2.3 Technology Adoption, Inequality and Economic Growth: Political-

Economy Issues

As mentioned before, this section reviews a strand of literature that

particularly involves political considerations behind policy determination in the

process of economic growth. In particular, this discussion focuses on the politico-

economic issues associated with the technology adoption process and the

evolution of inequality over time.

The literature that falls in this category describes a political process, where

households with conflicting interests vote to choose a best policy option which

determines economic growth. It is observed that the most fundamental causes of

conflicting interests are the issues related to distribution of income or resource

endowments. Therefore, these studies typically deal with the distributional

conflicts among the agents. In the voting process then, the main determinant in

these models is income or resource endowments of agents (Galor and Zeira, 1993;

Bertola, 1993; Alesina and Rodrik, 1994).

The political mechanism of this line of literature focuses on redistribution

of resources/income through a political process. The agents can either vote over a

preferred tax rate or a preferred level of government expenditure to redistribute

resources. For example, agents in Alesina and Rodrik (1994), Persson and

Tabellini (1994) vote on a preferred tax rate while agents in Saint-Paul and

Verdier (1993) vote on a preferred level of government expenditure on education.

Some of these models suggest that the income or resource endowment of the


39

median voter determines the level of tax rate, which in turn decides investments

and economic growth. It is then obvious that initial income distribution is vitally

important to economic growth.

Though the link between inequality and growth has been discussed in

detail previously, it is worth looking at how political economy literature in

particular, describes this link. In introductory political economy literature that

includes work such as Bertola (1993) and Alesina and Rodrik (1994) it is

suggested that inequality and growth are negatively related. According to some of

these models, the negative impact of inequality is likely to be caused by the fact

that, in a society with more unequal distribution of income, the poor will vote for

a high level of taxation, which impedes investments and economic growth. These

models further propose that relatively equal income distribution will weaken the

tendency of distortions associated with redistribution, thus stimulating investment and

economic growth (Persson and Tabellini, 1994). 13

In a re-examination of the above

hypothesis, Bao and Guo (2004) also suggest that inequality can be negatively

related to economic growth. While this proposition in the conventional political

economy literature is questioned by others such as Josten and Truger (2003), there

are studies that suggest income inequality may either relate positively or

ambiguously to economic growth. For example, Li and Zou (1998) argue as

follows: with more equal income distribution, people will vote for higher income

tax to allocate more resources to public consumption than production. Thus

economic growth is lower.

13

For comprehensive discussions regarding this issue, see Alesina and Perotti (1994, 1996),

Aghion, et al. (1999), Perotti (1992, 1996), Galor and Zeira (1993), Lindert (1996), Benabou

(1996) and Forbes (2000) among many others.


40

Moreover, in this literature, there are also studies which attempt to explore

politico-economic issues underlying the technology adoption associated with

economic growth. In particular, these studies examine why determination of

policies leads to a situation in which technologies that promote economic growth

are adopted by some countries while not by others. For example, Krusell and

Rios-Rull (1996) investigate the role played by “vested interests” in the process of

technology adoption.14

They suggest that, in the process of policy determination,

political elites may block the adoption of superior technology to protect their

economic returns. This would lead to diversity in growth outcomes observed

across and within countries.15

This idea was developed further in Acemoglu and

Robinson (2000, 2002b, 2006). These studies add that political elites may block

technological development, as well as institutional developments that enhance

growth. However, political elites are unlikely to block economic development

when there is a high degree of political competition.

Bellettini and Ottaviano (2005) examine a political economy model in

which organized interest groups of skilled workers lobby a government regulator

for a ban on the adoption of new technology. Their political equilibrium is

characterized either by perpetual innovation or by alternating periods of

technological change and stagnation depending on the demographic structure of

the population as well as technological and preference parameters. They also

suggest that international income differences may be linked to national

14

Two variants of this model; Krusell et al. (1997), Krusell and Rios-Rull (2002), discuss slightly

different aspects of technology adoption. 15

A similar idea was developed in Kuznets (1968) and Mokyr (1990) within a non political

economy framework.


41

propensities such as flexibility and abundance of human capital, patience of the

agents, nature of credit markets etc. Bridgeman et al. (2007) investigate the role of

government corruption in the process of technology adoption. This study suggests

if governments value interest group support (in terms of bribes) over social

welfare (measured in terms of GDP) barriers are erected to block the adoption of

technologies.

Finally, in summary, the above discussion surveys two central issues that

determine the pace of economic development within the context of a political

process, namely income inequality and technology adoption. In particular, the

discussion focuses on the roles played by these issues in choosing policies for

economic growth. However, in this developing body of literature, there still

remains room for further research to explore (i) whether policies determined

through a political process that lead to the adoption of advanced technology

guarantee a social optimum (ii) are there situations in which a country would

choose a particular level of inequality in order to achieve sustained growth?. To

that end, the fourth chapter of this thesis extends the benchmark model to address

these issues. This extension of the model has interesting implications on the

evolution of inequality and economic growth.

Costly Technology Adoption

42


43

CHAPTER 3

Growth Patterns and Inequality in the Presence of Costly

Technology Adoption

3.1 Introduction

The effect of macroeconomic variables on the evolution of income inequality

across and within countries has long been discussed in the literature on economic

growth and development. As mentioned in the previous chapter, the stylised facts

about income inequality across countries relate to Kuznets‟ (1955) hypothesis,

which suggests an inverted U-shaped relationship between economic growth and

income inequality. There are intense investigations of this hypothesis in both

theoretical and empirical literature, and the evidence is somewhat inconclusive. In

particular, recent evidence from the growth experience of East Asian economies

shows a decreasing pattern of inequality over time while many Latin American

countries experience an increasing pattern of inequality (See Forbes, 2000 and

Acemoglu and Robinson, 2002b, for a similar discussion).

Moreover, empirical evidence suggests that there has been a divergence over

time in income distributions across countries and within countries. For example,

based on the work of Quah (1996, 1997), there is strong evidence to suggest an

emergence of “twin-peaks” in cross-sectional world income distributions. There

is also substantial evidence to suggest that this type of polarization is present in

income distributions within countries. (See, for example, Schluter, 1998; Jappelli

and Pistaferri, 2000; Piketty and Saez, 2003 and Sala-i-Martin, 2006, among


44

others). Typically the empirical evidence of economic growth supports Baumol‟s

(1986) idea of “convergence clubs” emerging across and within countries. In

addition, Galor and Zeira (1993), Aghion and Bolton (1997), and Benabou (1996)

suggest that inequalities in initial income distributions also have a bearing on the

issue of divergence in incomes across and within countries. In particular,

differences in initial income distributions matter even in the case of countries

characterized by similar structural features and per-capita income levels. See

Galor (1996) for a detailed survey of literature in this regard.

Furthermore, Pritchett (1997) observes that the growth patterns of countries that

fall into the “developing economies” category exhibit a great deal of diversity.

For example, the economic growth of some of these countries converges rapidly

on the leaders, while others stagnate, or even experience reversals and declines in

their growth processes. Pritchett cites the experience of Mozambique (-2.2 percent

per annum), and Guyana (-0.7 percent per annum), as examples from a group of

16 developing economies which experienced negative growth rates in the period

1960 – 1992.

There is a large volume of theoretical literature that seeks to explain impacts

of economic factors associated with growth on the issues discussed above. An

interesting strand within this literature looks at the implications of technology

adoption and consequent income disparities within and across countries. Among

many others, prominent examples in this strand of literature include Hansen and

Prescott, (2002) Parente and Prescott (2004), and Mokyr (1990, 1993). These


45

authors suggest that the rates of technological progress may have significant

implications for cross country income differences. In particular, Greenwood and

Yorukoglu (1997) suggest that differences in the rate of technological progress

play a major role in widening inequality. Recent efforts in this direction (e.g.

Greenwood and Jovanovic, 1990; Parente and Prescott, 1994 and Ngai, 2004)

suggest barriers to adopting more productive technologies as an explanation for

cross-country income differences. The idea here is that the state of technology

changes over time and adoption of new technologies involves a significant cost in

terms of acquiring skills as well as reforming the institutional and structural

setting of the economy. There are several studies including Greenwood and

Jovanovic (1990), Hornstein and Krusell (1996), and Bessen (2002), which

support the view that „adoption costs‟ of this type account for a productivity slow-

down. This in turn produces cross country income differences. In particular, with

an exogenous fixed cost of adopting technology, Khan and Ravikumar (2002)

show that income inequality within a country increases over time.

The model developed in this chapter is similar in spirit to the literature on

technology adoption discussed above. However the motivation for this study

relates to issues that are explored to a limited degree in the above literature. In

particular, a common feature in the models discussed above is that the households

face a one-time adoption decision which is irreversible (See for example Khan

and Ravikumar, 2002, and Greenwood and Jovanovic, 1990). Moreover, this

adoption cost is assumed to be exogenous and fixed across heterogeneous agents.

The aim here is to develop a framework that relaxes the above mentioned


46

assumptions. Firstly, the model allows for household specific stochastic shocks.

Secondly we impose an overlapping-generations structure which enables us to

consider a situation in which the adoption decision is no longer of an “one-time”

irreversible nature - in every time period the new generation undertakes the

technology adoption decision irrespective of whether the previous generation

switched to the better technology or not. The latter assumption is even more

important in the context of adoption costs that are time-varying and household-

specific.

A further motivation for introducing an adoption decision that takes place every

period relates to the idea that this feature may be associated with the diversity of

growth patterns observed in the data. In particular, extant models of technology

adoption are unable to explain reversals in the process of economic growth that

are experienced by some countries. As will become clear later, this feature of our

model produces reversals in the output growth of dynasties that are part of the

lower end of the initial distribution of income and wealth in the economy. For

example, a poor household undertaking the switching decision may leave a

smaller amount in bequests than would otherwise have been the case.

Consequently the next generation may not have enough resources to be able to

adopt the better technology leading to a reversal of the growth process. It is also

obvious that the household-specific and time-varying nature of adoption will

exacerbate the possibility of reversals.


47

The model constructed here is very appealing in terms of its ability to

characterize growth outcomes of a very diverse nature. Specifically, depending on

initial productivity differences there are three possibilities. In one situation the

model produces a poverty trap. In the second situation a dual economy type

scenario results, while in the third situation there is balanced growth for all agents

in the economy. To date, no single model that is able to characterize all of these

scenarios has been developed in related literature.

The numerical experiments conducted in Section 2.3 of this chapter unearth

some interesting and empirically testable implications for the transitional process

of economies, most of which have been explored only to a limited degree by

previous studies. The model present below suggests that there is a negative link

between the size of adoption costs and the extent of contemporary technology

adoption. It also finds that assumptions about the initial distribution of wealth and

capital can have very different implications for the date at which all households in

the economy adopt better technology. Specifically, the higher is the initial level of

inequality the later is the date of complete adoption of the better technology.

Inequality can therefore increase and remain persistent for very long periods of

time, consequently delaying the process of structural transformation that is

associated with development.

It also appears that in contrast to previous literature, preference parameters do

matter. Specifically, a higher degree of altruism enables complete adoption to take

place sooner. Typically, more altruistic households leave larger bequests for the


48

next generation, such that the extent of adoption is positively related to the

altruism-parameters in the model. For example, if parents spend a large share of

their income on children‟s education complete adoption to better technology takes

place sooner. Post transitional inequality is then decreasing in the degree of

altruism, as poorer households tend to leave a larger proportion of their income in

the form of bequests, as previously suggested by the empirical findings of Tomes

(1981).

Another interesting feature revealed by our experiments using the model is the

diversity of growth patterns observed for different cohorts of households in the

economy. Household dynasties positioned at the “rich”, “poor”, or median levels

of the income distribution are all capable of experiencing reversals in the growth

of income over time. The timing of these reversals, which are temporary, appears

to be related to the timing of technology adoption, which is, of course, different

across various income groups.

A brief empirical analysis that involves looking at the impact of initial

inequality on the extent of technology adoption appears to loosely support some

of the predictions of the analytical and numerical work presented in this chapter.

Specifically, there is evidence of a negative correlation between inequality and the

extent of adoption. In addition, the extent of technology adoption is negatively

related to adoption costs and positively related to altruism. Note that due to

limited availability of appropriate data it is difficult to find proxies of variables

that correspond to exact measures of the variables modelled here. The results of


49

the empirical section are therefore subject to this caveat, and emphasis must be

placed on the need for future empirical research of a more rigorous nature directed

towards addressing these issues.

The following section describes the economic environment used to address the

issues described above. In particular, section 3.2.1 presents the general version of

the model. To develop some intuition in relation to the general model some

special cases are presented in sections 3.2.3 and 3.2.4. Section 3.3 details the

results of the numerical experiments conducted using the models. Some of the

empirically testable implications of the results unearthed in previous sections are

examined using a cross-country data set in section 3.4. Section 3.5 concludes this

chapter. Appendix of the chapter presents proofs of various propositions presented

in section 3.2, results of some numerical experiments and description of data used

for the empirical work.


3.2.1 Model 1

The model presented here consists of two-period lived overlapping generations

of agents. There are N agents in the economy and they are heterogeneous in their

holdings of wealth and capital. An agent born in period t inherits a certain amount

of capital and wealth. The initial distributions of capital and wealth are described

by F( . ), and G( . ) respectively. Time is discrete, with t = 0, 1, 2, … The

preferences of ith

agent born in period t are described as follows:

)1()ln()ln()ln()ln(),,,( 12111111 itititititititit sxccsxccU


50

Here, itc and 1itc denote the agents‟ consumption in the first and second period

of life respectively. Each agent is born with a unit of unskilled labour endowment

that may be used to earn a subsistence wage w . They also receive resources in the

form of bequests from their parents. Part of this bequest is given by 1itx , which

represents the resources left to the next generation after the death of the parents.

Parents also provide children with a share α of their second period income. This

second component of bequests received by the children of the agents born in

period t is represented by the variable 1its . The parameter is the subjective

discount factor in this model and θ1 and θ2 are parameters representing the extent

of imperfect intergenerational altruism in the model.

In order to produce output individuals have to decide to adopt one of two

technologies. These two technologies are referred to as Technology A and

Technology B. Technology A is associated with lower productivity but does not

involve any adoption costs. Technology B is associated with higher productivity

and involves a household specific adoption cost )( it , incurred in the agent‟s

youth. The economy produces output (Y) using composite human and physical

capital (K) and the production relationships F(K) assume simple “AK”

specifications. Here, the total factor productivities associated with the two

technologies are denoted by parameters A and B where AB . The rate of total

factor productivity in this model is assumed to be a time-invariant constant. The

adoption costs in the model are interpreted as exogenous costs that may result

from institutional or structural features of countries in addition to the present


51

value of “learning by doing” costs of acquiring skills that are needed to operate

the advanced technology.

The agents born in period t use their wage-income and resource endowment for

consumption and capital accumulation in the first period. In the second period,

they use returns on their capital holdings to finance consumption and bequests.

Households adopting Technology A face the following budget constraints:

)2(1 it

a

it

a

it WwKc

)3()1( 111

a

it

a

it

a

it xAKc

Here a

itc , a

itc 1 and a

itK 1 refer to first period consumption, second period

consumption and second period capital holding of ith

individual adopting

Technology A. The variable itW represents the resource endowment of the ith

agent in period t. In this model, the resource endowment of an agent depends on

the technology that was adopted by the agent‟s parents. This means that

a

it

a

it

a

itit sxWW if the agent‟s parent adopted Technology A and

b

it

b

it

b

itit sxWW if the agent‟s parent adopted Technology B. Here, the

bequests that arise from agents‟ second period income t

a

it AKs )( if the agents

adopted Technology A, and t

b

it BKs )( if the agents adopted Technology B. As

is evident from the budget constraints, these resources may be converted by the

young for the purpose of consumption as well as capital accumulation. To that

end, the interpretation of “capital” as consisting of both human and physical

components is critical in our model. Note that the “AK” structure of production

functions assumed here is typically known to generate non-convergence in


52

incomes across countries. See for example Mankiw, Romer, and Weil, (1992) and

references therein.

Households adopting Technology B, on the other hand, face the constraints:

)4(1 itit

b

it

b

it WwKc

)5(.)1( 111

b

it

b

it

b

it xBKc

As mentioned above, a household specific adoption cost (δit) of adopting

Technology B is experienced by the agents in period t. The section on numerical

experiments will also focus on a special case of the model in which the adoption

cost is a fixed, economy-wide cost ( it = ) rather than a household specific

variable cost. The special case where adoption costs vary over time but are fixed

across households will also be examined in this chapter.

Note also that the model here has a structure similar to that of Khan and

Ravikumar (2002), but the key difference is that here there is a two-period

overlapping-generations structure. Khan and Ravikumar consider an infinite

horizon model with non-overlapping generations and a one-time adoption cost,

after which the old technology is never used. In the model here, each generation

faces a technology adoption problem, even if the previous generation belonging to

the same cohort had adopted Technology B. It appears that the overlapping-

generations structure imposed here has very different implications for the

outcomes of the model. A further innovation is the household specific nature of

adoption costs, which has interesting implications for the growth patterns of the

household dynasties in the model.


53

For agents adopting Technology A, the optimal plans for consumption, bequests

and capital accumulation are described by following equations.

)7()1(

1it

a

a

it Wwc

)8()1(

11 it

a

a

a

it Wwc

)9()1(

111 it

a

a

a

it Wwx

)10()1(

1

)1(

)1( 1

1 it

a

a

a

it WwA

K

Likewise the agents who adopt B will have:

)11()1(

1itit

b

b

it Wwc

)12()1(

11 itit

b

b

b

it Wwc

)13()1(

111 itit

b

b

b

it Wwx

)14()1(

1

)1(

)1( 11 itit

b

b

b

it WwB

K

where,

)1()1(

1

2Aa and

)1()1(

1

2Bb

aaA)1(

)1( 1 and bb

B)1(

)1( 1 .

The ith

agent will adopt technology B iff

)15(),,(),,( ititit

A

ititit

B sxKUsxKU


54

where AU and BU represent the indirect utility functions for agents adopting

the A and B technologies respectively. It is then easy to show that this implies the

following:

Proposition 3.1: Let ;)( 21

1* wW it

it

where b

a

B

A

A

B

)1()1(

)1()1(

1

1

1 and )1(1

)1(

21

212

21

21

)1(

)1(

B

A.

A household will adopt technology B iff *

itit WW .

The above proposition defines a threshold level of resources required for a

household to find it worthwhile to adopt the more productive technology B. (See

Appendix 3.1 for a proof of this proposition).

In earlier work, Khan and Ravikumar (2002) derive a unique threshold level of

capital above which households will adopt the more productive technology and

show that this threshold level is independent of preference parameters. In contrast

to their analysis the threshold level of resources in this model depends on

technology parameters, preference parameters and adoption costs associated with

Technology B.

The dynamics of this model are described by the following system of first

order difference equations.


55

)16(

)1(

1

)1(

1

)1(

)1(

*

11

11

itit

it

a

a

a

it

it

a

a

a

it

WWfor

Wwx

WwA

K

)17(

)1(

1

)1(

1

)1(

)1(

*

11

11

itit

itit

b

b

b

it

itit

b

b

b

it

WWfor

Wwx

WwB

K

where ;)( 21

1* wW itit with 1 and 2 defined as in Proposition 3.1. Note

that if adoption costs are household specific stochastic shocks the threshold level

of resources varies over time and across households.

The following section shows analytical explanation of some predictions of this

model in relation to process of technology adoption and growth. For a given

time-invariant economy-wide fixed adoption cost (i.e. δit=δ), a steady state level

of resources ( s

itW ) is defined below which corresponds to the two technologies in

this economy if it exists.

)18(

;1)/1(

;1)/1(

)(

*

*

itit

a

itit

bs

it

WWifw

WWifw

W

Here2

2

1

1

)1(

)(

)1(

)(

A

AAa and

2

2

1

1

)1(

)(

)1(

)(

B

BBb


56

In the above equation, the steady state level of resources corresponding to

Technology A is determined by productivity parameters, preference parameters

and subsistence wage rate. In addition to these three variables, the steady state

level of resources corresponding to Technology B also depends on adoption costs

associated with Technology B. Note that the parameter embodies productivity

levels of technologies and agents‟ preferences in this economy.16

Figure 3.1 illustrates relationship between parameter and productivities of

technologies represented as P. The notation P* refers to the productivity level at

which =1 for the technology in question. Depending on the productivity

difference between the two technologies, the model has diverse implications for

the process of technology adoption and economic growth. There are three

different cases namely (i) A<B<P*, (ii) A<P*<B, (iii) P*<A<B. The

implications of these cases are summarized below.

(i) Poverty trap: If A<B<P* this economy converges to a unique steady state

under Technology A. This can be further explained by looking at the equilibrium

dynamics illustrated in Figure 3.2. In this figure, the vertical axis refers to Wit+1

while horizontal axis refers to Wit. The equilibrium dynamics for technology A are

given by )(1 wWW ita

a

it while the same for Technology B is given

by )(1 ititb

b

it wWW . Therefore the resource accumulation under technology

A is given by the line A-A‟ while resource accumulation under technology B is

16

The subscript a and b of parameter respectively represent Technology A and Technology B.


57

given by the line B-B‟. As shown in the figure sA

itW represents the unique steady

state of this economy which corresponds to Technology A. The inequality within

this economy increases in the transition process to the steady state. Once all

agents have adopted Technology A, inequality starts to decrease, eventually

converging to zero. Of course, when the adoption costs are household-specific

stochastic shocks, the economy fluctuates around the steady state level sA

itW . As

all households in this economy adopt Technology A, the rate of growth of the

economy also eventually converges to zero.

Figure 3.1: The values of productivity parameters that determine various

properties of the model.

(ii) Dual economy: If A<P*<B the economy uses both technologies. As

illustrated in Figure 3.3, the households with resources above sB

itW (i.e. the

Productivity (P) P*

1

ba or

)(Pfp


58

unstable steady state level of resources corresponding to Technology B) adopt

Technology B and experience continuous growth. The households who hold

resources below sB

itW reach a stable steady state under Technology A. The

inequality within this economy increases in the transition process and is persistent.

The growth rates of the two different groups of households do not converge and

remain distinct. This situation can be considered representative of a dual economy

in which one group of agents falls into stagnation while others experience growth.

The models in the literature that produce such features are numerous. See for

example Bourguignon (1990) and references therein.

Figure 3.2: Poverty trap.

(iii) Balanced growth: If P*<A<B this economy is characterized by balanced

growth. As illustrated in Figure 3.4 all households in the economy eventually

adopt Technology B. The inequality within this economy increases over time and

Wit+1

Wit *

itW sA

itW

A

A’

B’

B

450


59

is persistent. As all households in this economy adopt Technology B the rate of

growth of the economy corresponds to the rate of growth of Technology B.

Figure 3.3: Dual economy

Figure 3.4: Balanced growth

Wit+1

Wit *

itW

A

A’

B’

B

450

Wit+1

Wit *

itW sA

itW

A

A’

B’

B

450

sB

itW


60

Note that, the above mentioned implications are numerically illustrated in

Section 3.3 of this chapter.

3.2.2 Model 2

This section considers a special case of the model discussed above. The key

difference between the previous model and the model presented here is that, a part

of the bequests that young agents receive from their parents now is regarded as

non-altruistic in nature. More specifically, the parents provide children with a

share ( ) of their second period income but they do not derive any utility from

leaving that part to the children. Specifically now 02 in equation 1 so that the

lifetime utility of ith

agent born in period t is described as follows:

)19()ln()ln()ln(),,( 1111 itititititit xccxccU

This incorporates the idea that parents leave this type of bequests to their

children specifically for the purpose of nurturing and educating them representing

the commonly accepted social norm of devoting resources to children‟s

upbringing and education.

Agents using technology A maximize (19) subject to (2) and (3) while agents

using technology B maximize (19) subject to (4) and (5). The implied optimal

plans for consumption, bequests and capital accumulation as well as the dynamics

of this model are given in Appendix 3.2.


61

Applying the same reasoning as before, or simply setting 02 in

proposition 1, it is easy to derive the proposition below.

Proposition 3.2: Let ;1

* wW itit where

)1(1

)1(

B

A. A household will

adopt technology B iff *

itit WW .

The above proposition defines a threshold level of resources required for a

household to find it worthwhile to adopt the more productive technology B. Note

that resources here are defined as the capital and wealth endowment of agents and

thus the threshold level constitutes both capital and wealth.

Parallel to the analysis conducted earlier, again for the case of (δit=δ) if a steady

state exists, it can be defined as,

)20(

;1)/1(

;1

1)(

*

*

itit

itit

s

it

WWifA

w

WWifB

w

W

where )1(1

)(.

Similar to the analytical work conducted in Model 1, we can derive the same

results characterizing the growth outcomes based on the differences in

productivities of the two technologies. In particular, here the eventual rates of


62

growth that are experienced by agents in this economy can be easily evaluated

analytically.

Proposition 3.3: For a given, economy-wide fixed adoption cost (i.e. δit=δ),

(i) If 1

BA , this economy converges to a unique steady state under

Technology A. All households eventually adopt Technology A. A poverty trap

exists. The economic growth rate is given by A .

(ii) if BA1

, the households who hold resources below the unstable steady

state level of resources that corresponds to technology B, reach a stable steady

state under Technology A. This is a dual economy that uses both technologies.

The growth rates of the two sectors remain distinct.

(iii) if BA1

, this economy is characterized by balanced growth. All

households in the economy eventually adopt Technology B. The economic growth

rate is given by B .

3.2.3 Model 3

The distinct feature of the model presented in this section is that the parents do

not provide children with a share ( ) of their second period income for the

purpose of nurturing and educating them. Specifically, here 0 . It is then

obvious that 02 also. This means that preferences of the agents in this model

are also identical to the preferences given by equation 19 however the budget

constraints should be re-defined. These budget constraints are then defined as


63

follows. Agents born in period t, who choose Technology A, face the following

budget constraints:

)21(1 it

a

it

a

it WwKc

)22(111

a

it

a

it

a

it xAKc

On the other hand, those adopting Technology B face the constraints:

)23(1 itit

b

it

b

it WwKc

)24(111

b

it

b

it

b

it xBKc

Note here that the variables xit and xit+1 refer to the bequests that an agent born

in period t receives at period t and the bequest that an agent leaves to his child at

period t+1 respectively.

Agents using technology A maximize (19) subject to (21) and (22) and agents

using technology B maximize (19) subject to (23) and (24).17

Using this condition

or simply setting 0 and 02 , a threshold level of resources (xit*) that is

required for a household to find it worthwhile to adopt Technology B can be

defined as follows.

Proposition 3.4: Let ;1

* wx itit where

)1(1

)1(

B

A. A household will

adopt technology B iff itit xx*

In this version of the model, it is apparent that the agents‟ switching decision

simply depends on their wealth endowment. However, long run economic growth

is determined by both capital and wealth accumulation.

17

Again the optimal plans for consumption, bequests and capital accumulation are presented in

Appendix 3.3.


64

As before, the following section presents the implications of productivity

differences on the dynamics of the model. Here, if a steady state level of resources

( s

itx ) exist, it is defined for two technologies in this economy for the case of (δit=δ)

as follows.

)25(

;1)/1(

;1

1)(

*

*

itit

ititit

s

it

xxifA

w

xxifB

w

x

where )1(1

.

Note that here, except for the parameter η, the remaining part of the equation is

identical to the equation 18 in the previous section. Therefore, the implications of

this model in relation to the steady-state behavior are similar to the previous

model. However, although no qualitative differences in the results are expected

there may be quantitative differences because of changes associated with

parameters of altruism. Such quantitative differences will be explored in the

section on numerical experiments.

The next section reports numerical experiments associated with the models

presented above.

3.3 Results of Numerical Experiments and Discussion

Since there are three models as well as several different cases associated with

each model, a brief summary of the organization of this section is provided here to


65

assist the reader. Basically, in this section the main focus is on the results of the

more general version of our model (Model 1), presented in section 3.3.1 below.

The organization of section 3.3.1 is as follows. Sub-section (i) focuses on a

special case of the model in which the adoption cost is a time- invariant,

economy-wide fixed cost ( it ≡ ). In sub-section (ii) the adoption cost is fixed

across households but allowed to vary over time ( it ≡ t ). Finally, these

assumptions are relaxed and allow the adoption costs to vary across households

and time in sub-section (iii).

As discussed in section 3.2 of this chapter, the combination of productivity

parameters A and B can lead to three distinct implications for growth, which are

labeled as “poverty trap”, “dual economy” and “balanced growth”. In the special

case where the fixed adoption costs are analyzed (i.e sub section (i) below) we

will refer to the above three cases which also form a benchmark for interpreting

the case in which adoption costs are allowed to vary. This section also includes

results of numerical experiments conducted using the other two models, only if

the results provide insights that are easier to interpret relative to the general

version of our model in section 3.2.

3.3.1 Results of Experiments Conducted Using Model 1

(i) Adoption Costs Fixed Across Households and Time ( it ≡ )

Firstly, this section provides a numerical analysis of the three cases (i.e.

“poverty trap”, “dual economy” and “balanced growth”) mentioned above. The

parameter values associated with these cases are presented in summary form in


66

Table 3.1 below. Recall that P* is the value of the productivity level at which γ in

equation 18 equals to 1.

Table 3.1: Parameter Values

Figure 3.5 shows the implications for technology adoption and economic growth

in the case A<B< P* which was labeled as “poverty trap”. The panel (a) of this

figure shows the number of households adopting Technology A or B in different

time periods, while panel (b) shows the evolution of inequality in this economy

over time. The average rate of growth in this economy is given in panel (c) while

the growth rates experienced by various cohorts of households in the income

distribution are presented in panel (d). In particular, this panel presents the rate of

growth of the output for the median, richest 20% and poorest 20% of the

households of the income distribution. In the three figures above In this case all

agents in the economy eventually adopt technology A as illustrated in Figure 3.2

of the previous section.

Parameter (P*=2.9) A B

Poverty trap (A<B< P*) 1 2

Dual economy (A<P*<B) 2 3

Balanced growth (P*<A<B) 3 5


67

Figure 3.5: Technology adoption, inequality and economic growth in the case of

poverty trap.

Furthermore, as mentioned in the previous section, the inequality within this

economy initially increases in the transition process, before the eventual decline in

inequality. Basically, the initial distribution of income and wealth implies that

there are two sets of households. One set adopts Technology A while the other set

adopts Technology B. Over time the set A increases as those households at the

lower end of the set B do not leave sufficient resources for their offspring due to

the high cost of adopting Technology B. The resources of each subsequent

generation decline over time to the point that only adopting Technology A is

feasible. Once all agents have adopted Technology A, inequality starts to


68

decrease, and eventually converges to zero. The rate of growth for each agent in

this economy eventually increases from a negative growth rate in the transition

process and converges to zero. This has an obvious implication for the sectoral

growth rates of households in Technology A and B respectively. Since all

households are adopting Technology A eventually the economy stagnates.

Figure 3.6 looks at similar implications for the case that is labeled as “dual

economy”. Here, the economy uses both technologies in the steady state as

illustrated previously in Figure 3.3. It is apparent that the steady state of this

economy involves two distinct sectors growing at different rates. An obvious

implication of this characteristic is that inequality in this economy will increase in

the transition process and remain persistent.


dual economy.


69


balanced growth.

Figure 3.7 above explores the implications for the case that is labeled as

“balanced growth”. In this case, both technologies have productivities that are

large enough to allow households to grow at a relatively higher rate. Therefore all

agents in the economy eventually adopt Technology B. See again the section that

analyzes Figure 3.4 and discussion associated with it. The inequality in this

economy is increasing in the transition process and remains persistent. Intuitively

it is apparent that, households who switch to the more productive technology

sooner accumulate more resources than those who delay the switch to the

productive technology. As all agents adopt Technology B the growth rates of the

output of agents in this economy eventually converge.


70

As mentioned before, all three cases reported above, look at the rate of growth

of output for the median, richest 20% and poorest 20% of the households of the

income distribution. In the three figures above, it appears that the growth pattern

for different cohorts of households is very diverse.

For example, in the case that is labeled as “poverty trap” the growth rates of all

households start with a negative value. (See panel (d) of Figure 3.5). Over time

however growth rates of the poorer and the median households monotonically

increase while rich agents experience a reversal. Intuitively, as all agents in the

richer end of the income distribution jump from Technology B to A, their growth

rate is characterized by a reversal at the beginning. Eventually growth rates of all

households converge, but the economy experiences stagnation.

From the panel (d) of Figure 3.6, which corresponds to the case of “dual

economy”, it is clear that growth patterns of both the richer the poorer agents are

smooth and monotonic while median agents experience reversals. This can be

explained as follows. Initially, there are two sets of households in the income

distribution of this economy. The poorer end consists of one set that adopts

Technology A while the richer end consists of the other set that adopts

Technology B. Again, as agents in the lower end of set B jump from Technology

B to A, their growth rate is characterized by a reversal at the beginning. Then over

time growth rates of these households converge to the growth rate of poorer

households. As a result, in the steady state, the economy is characterized by two

distinct rates of growth that correspond to the two technologies.


71

The “balanced growth” case shows a monotonically increasing growth rate

experienced by richer agents while the poorer the median agents have a variable

pattern. (See panel (d) of Figure 3.7). In the case of the poorer and the median

households however, a very high savings rate is required prior to reaching the

stage when the household is able to make the switching decision. When the

households make the switching decision they incur a heavy cost of adoption, so

the amount invested in the new technology is relatively low. Therefore these

agents experience a temporary reversal in the growth rate of output. In the steady

state however, the rates of growth of the different groups of households converge

and the economy is characterized by a permanent and stable growth rate.

The models‟ ability to generate a diverse set of patterns for growth rates of

households at different positions in the income distribution is worthy of comment

here. A critique of contemporary growth models has often been related to their

inability to explain the patterns of reversals in the growth process that have been

experienced by several countries even after they embarked on modern economic

growth (Pritchett, 2000). If agents in our model are interpreted as countries that

occupy various positions in the world income distribution, the model here can be

regarded as explanatory step in the direction of such phenomena.

The remainder of this sub-section will present some numerical experiments with

varying (a) adoption cost parameter (δ), (b) initial inequality levels, (c) altruism

parameters (θ1, θ2) and (d) parents‟ income share on children‟s education (α). In

fact, these are sensitivity analyses of the above parameters, and the main


72

presentation of results resumes in sub section (ii). Note that the experiments

presented below assume the “balanced growth” case that is discussed previously.

However the sensitivity analyses for the transition process under the remaining

two cases (i.e “poverty trap” and “dual economy”) also imply similar

interpretations. For this reason results are not presented here, but they are

presented in Appendix 3.4 and 3.5 of this chapter respectively.

(a) Experiments with the adoption-cost parameter

Figure 3.8 examines the effect of increasing cost of adoption on the date at

which all households shift to using Technology B. Values of are set to equal

10, 15, 20, and 25. As illustrated in the figure the corresponding dates of

transition *T are equal to 5, 6, 7, 8 respectively. In terms of the model this means

that the households completely adopt Technology B, after 175, 210, 245, and 280

years respectively.18

Figure 3.9 illustrates the effect of increasing adoption costs

on the evolution of income inequality in countries. It appears that higher adoption

costs increase the income gap between rich and poor more than lower adoption

costs do. Furthermore post-adoption inequality is higher for higher adoption costs.

If the adoption cost in the model is interpreted to be a cost that represents

institutional or structural features of countries, the implication of these results for

cross country differences in income is obvious. Also considered here are the

growth patterns of median households in this economy for different adoption

costs.19

Figure 3.10 illustrates that higher adoption costs imply an increasing

18

Recall that this is an overlapping-generations model, so each period represents about 35 years

(See Hansen and Prescott, 2002). 19

Similar outcomes are observed in the growth patterns of rich and poor cohorts of households as

in the case of median households.


73

length of time before steady state growth. Note that here the eventual steady-state

rate of growth of output does not vary as adoption costs change, but the time to

reach steady-state growth is delayed in the presence of higher adoption costs. This

feature of the model also exhibits strength in explaining diverse patterns of growth

experienced by transitional economies unlike the previous models found in related

literature.

Figure 3.8: Number of households in Technology A and B over time for different

adoption costs.

δ=10

δ=20

δ=15

δ=25


74

Figure 3.9: Inequality over time for different adoption costs.

Figure 3.10: Growth rate for median agent with different adoption costs.


75

Figure 3.11: Number of households in Technology A and B for different initial

inequality levels.

(b) Experiments with varying initial inequality levels

As heterogeneous agents are introduced in to the model, it is capable of making

implications for initial income distribution on technology adoption process,

evolution of inequality and growth patterns of countries. To explore such

implications, an experiment with varying initial inequality levels with four

different initial distributions of resources is carried out here. The initial inequality

is measured in terms of Gini coefficient and given by 0.34, 0.40, 0.52, and 0.60.

in panels (a), (b), (c) and (d) respectively in Figure 3.11. The figure illustrates

that complete adoption to Technology B takes place sooner with low initial

inequality. The post-adoption inequality is also low for smaller values of Gini

coefficient (Figure 3.12).

Initial Gini=0.34 Initial Gini=0.40

Initial Gini=0.52 Initial Gini=0.60


76

Figure 3.12: Evolution of income over time for different initial inequality levels.

Figure 3.13: Rate of growth of median household for varying initial inequality

levels.


77

Next, the patterns of growth experienced by different income-cohorts of

households for various initial inequality levels (See Figure 3.13) are examined

here.20

It appears that the same group of households follow similar growth

pattern, however their length of time before reaching steady state growth is

different for different initial inequality levels. For example, if a country is

characterized by higher initial inequality, this delays the date of transition to

steady state growth. This feature of the model can be used to explain the widely

discussed fact in growth literature that countries with similar structural

characteristics and institutional features, but differing in their initial distribution of

income, converge to steady state growth at different dates (See Galor, 1996 and

references there in).

(c) Experiments with θ

As stated in the economic environment-section of this chapter, agents born in

period t receive two types of bequests. The altruism parameters of these two types

are given by θ1 and θ2. In terms of the model θ1 represents parents‟ desire to leave

wealth for their children and θ2 represents parents‟ desire to spend resources on

upbringing and educating their children. This section presents the effect of

varying altruism parameter θ1 on the technology adoption process and on

evolution of inequality and growth patterns.21

The four panels in Figure 3.14

report the date of transition to technology B for different values of altruism

parameter (i.e.0.7, 0.8, 0.9, 1.0). This result suggests that if the parents are more

altruistic, their children are more likely to adopt the productive technology sooner

20

We present here, the growth patterns of poor cohort of households, however we observe similar

outcome as in the cases of rich and median households. 21

The implications are similar for the case of θ2 parameter also.


78

as they receive more resources in terms of bequests. Moreover the post-adoption

inequality in this economy is negatively linked with the altruism parameter. As

illustrated in Figure 3.15, if an economy consists of more altruistic households

their quick adoption to Technology B reduces post transitional inequality.

Moreover, over time, growth rates of the agents in this type of economy converge

to a higher in the steady-state, when compared with an economy with more non-

altruistic households (see Figure 3.16).

Figure 3.14: Number of households in Technology A and B for varying levels of

altruism parameter.

θ1 =0.7 θ1 =0.8

θ1 =0.9 θ1 =1.0

θ1 =0.7 θ1 =0.8


79

Figure 3.15: Gini coefficient for varying levels of altruism parameter.

Figure 3.16: Growth rate of median household for different altruism parameter.


80

As the implications of the experiment with varying altruism parameter (θ2) on the

technology adoption process and on evolution of inequality and growth patterns

are similar to the implications reported above, they are not reported here. In

addition, an experiment with different α parameter is performed and the

implications are reported below.

(d) Experiments with α

This section examines the effect of changing parameter α on the technology

adoption process, evolution of income disparity and growth patterns of different

cohorts of households in the income distribution.

Figure 3.17 illustrates that if agents‟ parents devote more resources for the

upbringing and education of their children, the children are more likely to adopt

the more productive technology sooner. In terms of the model, the children utilize

the resources provided to them by their parents to acquire the necessary skills and

know-how to adopt the better technology. It is obvious that quicker adoption to

better technology increases the inequality in the adoption process, however, post-

adoption inequality for higher values of α is low (Figure 3.18). This means that

inequality at steady state is low for a country with parents who devote more

resources to their young. In an economy with higher α a higher steady state

growth rate can be observed in comparison to an economy with lower α. (see

Figure 3.19).


81

3.17: Number of households in Technology A and B for varying levels of α

parameter.

Figure 3.18: Gini coefficient for varying levels of α parameter.

α=0.10 α=0.15

α=0.20 α=0.25


82

.

Figure 3.19: Growth rate of poor household for different α parameter.

.

As mentioned before, the following sub sections will look at two other cases of

the model that allow variation in the assumption of adoption costs (i.e “time

varying” case and “household specific” case). Recall that the “time varying” case

refers to a situation in which all households face the same adoption cost in a given

period. In contrast, in the “household specific” case adoption costs differ for each

household in the income distribution. It follows, then, that they are automatically

time varying adoption costs as well.


83

(ii) Time-varying Adoption Costs ( it ≡ t )

This section presents an experiment in which the adoption cost is fixed across

households, but allowed to vary over time. The experiment presented here is

motivated by the idea that the exogenous shocks such as institutional or structural

changes that vary over time may have important implications for the adoption of

advanced technologies. To explore such implications, an experiment is conducted

in which adoption costs are allowed to vary randomly over time but are fixed

across households. 22

In order to represent periodical variation in the adoption

costs, random values from a uniform distribution are chosen where U(20,60). The

following section presents the results of this experiment for the case analogous to

the earlier mentioned “balanced growth” scenario. Analogous experiments are

also conducted to consider the remaining two cases (i.e. “poverty trap” and “dual

economy”). The findings lead to similar interpretations in relation to the growth

process. For details see Appendix 3.7 and 3.8 of this chapter.

Figure 3.20 illustrates implications of this experiment for technology adoption,

the evolution of inequality, and convergence in growth rates of different cohorts

of households in the income distribution. It appears that the technology adoption

process significantly reflects the temporal variation in the adoption costs.

Essentially, unlike in the corresponding case of fixed adoption costs that we

reported earlier, reversals in the technology adoption process are possible in this

case, as evident from panel (b) of Figure 3.20. This means that during the

22

In addition, an experiment is conducted which adoption costs increase over time. As results

found have analogous implications for the technology adoption process, they are not presented

here. But they are presented in Appendix 3.6 for details.


84

transition process, some agents shift back to using Technology A before complete

adoption takes place. Overall however, in the presence of exogenous shocks, the

technology adoption process experiences significant delays particularly because

the poorest cohort of agents takes a longer time to adopt the advanced technology.

Figure 3.20: Adoption costs vary randomly over time for the case of “balanced

growth”.

Moreover in the transition process, the growth rates of the households in this

economy are also characterized by a significant diversity which again reflects the

variation in adoption costs over time. Of course, such diversity was observed in

the corresponding case of fixed adoption costs as reported earlier in section

3.3.1.(i). However in this case, there are more reversals in the growth rates of the


85

households. In particular the growth rates of those households who belong to the

poor cohorts of the income distribution are characterized by significant overturns.

(See panel (d) of Figure 3.20). Typically, the burden of exogenous shocks for

relatively poorer households is significant, and leads to more reversals in their

adoption process.

The panel (c) in the same figure looks at the evolution of inequality in income

and wealth over time. Overall, it appears that inequality increases in the transition

process, and remains persistent at the steady state. Relative to the fixed adoption

costs case, however, inequality increases at a slower rate during the transition

period. It is likely that the reversals in the technology adoption discussed earlier

lead to this pattern.

As mentioned above, what is considered here is the case of “balanced growth” in

which productivity parameters of the two technologies are such that the economy

experiences permanent and stable growth eventually. It is then obvious that the

eventual growth outcomes in this case do not qualitatively differ from the results

presented in the section 3.3.1.(i). That is, individuals switch to better technology

eventually and inequality is persistent in this economy.

(iii) Household Specific Adoption Costs

In this section adoption costs are allowed to vary across time and across

households. These adoption costs are considered to be household specific

exogenous stochastic shock. As interpreted previously, this type of stochastic


86

shocks may represent sudden changes in structural or institutional features of a

country that may vary across households, but are not explicitly modeled here. In

these experiments, values have been randomly drawn for adoption cost from a

uniform distribution. Again, experiments conducted here are analogous to the

three cases referred to as “poverty trap”, “dual economy” and “balanced growth”

earlier.23

Figure 3.21: Households adopting Technology A and B in different time periods.

As before the results in the three cases are similar in essence to the case of fixed

adoption costs. Therefore, this section will be limited to a presentation of the case

of balanced growth, and the other two cases will be dealt with in the Appendix 3.9

23

Recall that these refer to the fixed adoption cost case and correspond to the situations in which

(i) A<B<P* (ii) A<P*<B (iii) P*<A<B, labeled respectively as poverty trap, dual economy and

balanced growth. Note that these cases only for easy reference as they apply in a literal sense for

the variable adoption costs case are labeled here.


87

and 3.10. Figure 3.21 shows the number of households adopting technology A and

B in different time periods. From this figure it is apparent that when adoption

costs are allowed to vary across households and time, reversals in the adoption

process are more frequent. However the magnitude of these reversals is not

particularly prominent. Also, the variability in adoption costs is reflected in the

additional variability of the growth patterns of households in the various cohorts

of income distribution (See Figure 3.22). Nevertheless, as in the case of “balanced

growth”, individuals switch to better technology eventually. Inequality remains

persistent.

Figure 3.22: Growth rates of households in different cohorts of income

distribution.


88

Also in this section, a comparison of growth outcomes is presented for these

two cases (i.e. economy-wide fixed adoption cost case vs. household specific

adoption costs case). To summarize results briefly: the experiment presented in

this section looks at the implications in relation to the complete adoption of

Technology B, and the evolution of inequality over time. When the adoption

costs are allowed to vary over time, the mean of the adoption costs is normalized

to be the same as that used for the corresponding fixed cost experiment. The

results suggest that in the case of fixed adoption costs across households,

complete adoption of Technology B is faster. Intuitively, if adoption costs vary

across households, poor households who face a higher adoption cost may

significantly delay their switching decision to better technology. On the other

hand, if the adoption costs are fixed across households, all households in the

economy are appear to affect equally by adoption cost.

Moreover, the gap between the rich and the poor widens during technology

adoption process and follows a similar pattern in both cases as observed earlier.

Interestingly, our model also shows that the economy eventually converges to the

same steady state growth rate in the case of household specific adoption costs, as

in the case of an economy- wide fixed adoption cost.

3.3.2 Experiments Conducted Using Model 2 and Model 3

Recall that Model 2 and Model 3 are special cases of Model 1. Therefore this

section will focus on features of the Model 2 and Model 3 that are easier to


89

interpret relative to the more general version of the model. Overall, results of

these two models are similar in the sprit to that of Model 1.

The sub section (i) below, explore the role of altruism in the technology

adoption process in Model 2. In sub section (ii), as overall results of the three

models are fairly similar, the outcomes of these models are compared to show

how the degree of altruism matters in the models. Essentially, this allows for

quantitative investigation how weakening of altruism matters in the processes of

technology adoption and economic growth. 24

(i) Experiments Conducted Using Model 2

In a previous section, it was observed that when adoption costs are fixed, a more

altruistic household is likely to adopt better technology sooner as it enables the

household to leave larger bequests for the next generation. This is analyzed

further using Model 2 and the results are presented in the Appendix 3.11. This

section, however, explore whether a household leaves a higher proportion of their

income in the form of bequests prior to adoption of the more productive

technology. For this purpose an experiment is conducted using model 2 presented

in the previous section as it provides relatively easier analysis and interpretation

compared to the general version of the model. Figure 3.23 presents a transitional

period in which all households have not yet adopted technology B for two cases:

θ=0.9 and θ=1. Bequest as a proportion of income is higher in the case of θ=1.

Eventually after complete adoption the percentage of bequests left is constant, and

24

Recall that in the Model 2 and Model 3, the degree of altruism is weaker relative to Model 1. In

Model 2, 02 and in Model 3, 0 as well as 02 .


90

lower in the case of θ =0.9. This feature of the model is consistent with empirical

evidence. Based on panel data consisting of 659 estates in Ohio, U.S.A., Tomes

(1981) finds that inheritance received from parents is inversely related to

children‟s income.25

Figure 3.23: Bequests as a proportion of income during transition and at steady

state for different altruism parameter (θ).

Note that we present results of the sensitivity analysis for the parameters in

model 2 in Appendix 3.12.

25

Please see Owen and Weil (1997) and Borjas (1992) for further discussion.


91

(ii) Experiments Conducted Using Model 3

As mentioned before, the qualitative results of Model 3 provide parallel

interpretations as in the cases of Model 2 and the general version of our model.

Therefore results of the sensitivity analysis for the parameters in model 3 are

present in Appendix 3.13. However this section presents a comparison of the

results of the three models for the purpose of illustrating the quantitative

differences between these three cases.

Figure 3.24: Evolution of inequality over time.

In the experiments conducted for this purpose, initial distributions of income

and wealth as well as all other parameters are held fixed across the three models.

First, the implications of these three models for timing of complete adoption of

Technology B are examined. The results suggest that, in terms of rapidness of

complete adoption of Technology B, Model 1 is fastest followed by Model 2 and


92

Model 3 respectively. In fact, the weakening of altruism also follows a similar

pattern in these three models. Recall that in Model 2, 02 and in Model 3

0 as well as 02 . It is then obvious that degree of altruism in these models

significantly matters in respect to timing of complete adoption of Technology B.

Figure 3.25: Growth rate of an average household.

The same comment applies for the implications of inequality for these three

Models. In fact, as illustrated in Figure 3.24, the post-adoption inequality

increases with weakening altruism. Moreover, the rate of growth in the steady

state also relates to the degree of altruism in these models. For example, in the

case of Model 1, the economy experiences a higher growth rate relative to the


93

other two models, where as the economy in Model 3 experiences the lowest rate

of growth in the steady state (Figure 3.25).

The next section will present the results of empirical work conducted to support

some of the model‟s predictions.

3.4 Empirical Study and Results

This section presents some empirical evidence in support of the numerical

implications of the model. In particular, this empirical examination focuses on the

model‟s implication of (i) a negative link between initial income and wealth

inequality within a country and the extent of technology adoption, (ii) a negative

link between adoption costs and extent of technology adoption and (iii) a positive

link between the degree of altruism and the extent of technology adoption.

In order to test these implications, this empirical exercise attempt to construct

appropriate measures for these variables which mirrors the rationale of theoretical

model of this thesis. This type of exercise is subject to a caveat that proxy

variables such as “technology adoption” are very hard to construct given limited

data availability. However, the exercise here attempts to construct a proxy

variable for “extent of technology adoption” based on the model, and is discussed

in detail in section 3.4.1 below.


94

3.4.1. Construction of the Technology Adoption Index

First, an “Index of technology adoption” (ITA) is constructed here for the model

as well as for the data.26

This index, which may be considered a measure of the

“extent of adoption” of a particular technology, is defined as follows.

BB

BB

i

iNN

NNITA

minmax

min

Where, B

iN is the number of households in country i which have adopted a certain

technology.

i=1,…,n.

B

n

BB NNN ,.....min 1min

B

n

BB NNN ,.....max 1max

Here, this index is calculated for three different measures that are considered to

be representing the extent of adoption of Technology B. The first case B

iN , for

example, represents the number of households per 1000 in country i that have

adopted telephones. The other two proxy variables are (a) households per 1000

using cellular telephones and (b) households per 1000 using internet facilities.

These three indices are denoted as ITA-I, ITA-II and ITA-III respectively. By

averaging these three indices a fourth index is constructed, which is referred to as

an “aggregate index of technology adoption” (AITA). The variable in the

theoretical model which is the counterpart to the adoption indices constructed, is

simply given by,

N

NITA

B

t

t

26

To gather data the “Technology: creation and diffusion” data base of Human Development

Report (2006) was utilized enabling compilation of a dataset consisting of 104 countries.


95

Here the subscript t is interpreted as the “stage of development” so that it

corresponds to the interpretation of the subscript i for the data set. In that sense, it

is possible to look at a cross section of data that is interpreted as countries at

different stages of development as measured by the index of adoption. The model

counterpart of the index however corresponds to “different time periods” which is

also analogous to the idea of different stages of growth.

First, this section aims at exploring whether a country with more initial

inequality in income and wealth distributions has a smaller extent of technology

adoption. For evidence of a negative correlation between the constructs of the

“degree of technology adoption” and the initial income inequality of a country, a

cross country data set is examined here. The initial levels of inequality is

measured using the Gini coefficient (GN) which is the main independent variable

in our analysis in addition to variables measuring the cost of adoption (AC) and

the degree of altruism (AT). Specifically, the estimate of “initial inequality” here

is based on the measure of Gini coefficients dated approximately around the date

of the country‟s transition to “modern economic growth in the sense characterized

by Kuznets (1955). (See also Hansen and Prescott, 2002). However, since it is not

always possible to get the relevant estimate of inequality for all of the countries in

the sample, the nearest possible consistently measured estimate of initial

inequality for some of the countries in our sample is used in this analysis.27

27

For some of the developing economies in the sample, the closest available estimate is

approximately around 1964. However countries for which a reasonable measure of “initial

inequality” is not available were excluded from the sample.


96

Similar caveats as discussed above in relation to data availability and quality

also apply for the variables of altruism and adoption cost. However, Tomes

(1981), states that investment on human capital development is a frequently used

measure for degree of altruism. Therefore, in accordance with the theoretical

model, degree of parental altruism is measured here by educational expenditure as

a proportion of GDP. (See Data appendix for more details.) Moreover, as adoption

costs in the model imply barriers to technology adoption, a proxy variable that

measures barriers in terms of required procedures in countries that govern entry of

entrepreneurs to a business is used as suggested by Djankov et.al (2002). This

proxy measure includes the direct-cost estimates for entrepreneurs associated with

meeting government requirements in relation to starting businesses, plus the

monetized value of the entrepreneur‟s time spent in activities that require to going

through such „red tape‟ (both measured as a fraction of GDP per capita in 1999).

The data source for this measure of adoption costs is also Djankov et.al (2002).

The other explanatory variables in this analysis are selected based on the

following arguments. The stylised facts of growth and development suggest that a

primary factor determining the degree of technology adoption of any transitional

economy is its level of output (Kuznets, 1955). Furthermore various institutional,

structural, social and political characteristics may have implications for the

adoption levels of various technologies. For example, the levels of educational

attainment of the population, longevity and health of the population, or the

country‟s openness to trade with the rest of the world may have implications for


97

the process of technology adoption. To that end, the set of other independent

variables included in our regression analysis are;

HDi :level of human development of country i. Please see Data Appendix for a

description of this variable.

TRi: degree of openness to trade with the rest of the world, measured as the ratio

of total imports to a country‟s total trade following Caselli and Coleman (2001).

In addition, this variable is regarded as an indirect measure of the costs of

adoption, i.e. a variable that inversely represents the cost of adoption.

It is therefore possible to estimate a model of the form,

)26(54321 iiiiiii TRACATHDGNAITA

where, AITAi is the aggregate index of technology adoption of country i, and

HD,TR, AT and AC are the explanatory variables discussed above for ith

country.

The error component is εi and it has usual properties ))1,0(~( Ni . To check the

robustness of the results of this analysis, equation 14 is estimated for three other

measures of technology adoption (i.e. ITA-I, ITA-II, and ITA-III) discussed in

subsection 4.1. For details of data set please see Appendix 3.14.

Table 3.2 below presents the results of this exercise. The second column of the

table represents the regression results for the model with degree of technology

adoption measured in terms of the aggregate index of technology adoption (AITA).

The next three columns show the regression results based on three other measures

of technology adoption discussed previously.


98

Table 3.2: Regression Results.

AITA ITA- I ITA- II ITA- III

0.496 0.669 0.510 0.217

GN -0.008***

(0.003)

-0.009***

(0.003)

-0.007***

(0.003)

-0.004

(0.003)

HD 0.006***

(0.001)

0.006***

(0.001)

0.006***

(0.001)

0.004***

(0.001)

AT 0.039**

(0.015)

0.037***

(0.018)

0.051**

(0.017)

0.032*

(0.018)

AC -0.245***

(0.061)

-0.262***

(0.069)

-0.119*

(0.060)

-0.291***

(0.064)

TR 0.145

(0.501)

0.059

(0.615)

0.141

(0.557)

0.193

(0.595)

Adjusted R2 0.722 0.653 0.631 0.628

Number of countries 42 42 46 47

*** Significant at 1 percent, ** significant at 5 percent, * significant at 10 percent (Figures within

parenthesis are standard errors).

The results appear to support the fact that there is a negative link between initial

income inequality and country‟s degree of technology adoption. Furthermore, the

sign of coefficients of other variables are mostly consistent with the hypothesized

impact on technology adoption. Specifically, there is evidence in support of a

positive link between altruism and the extent of adoption, and a negative link

between adoption costs and the extent of adoption.


Empirical evidence suggests that there has been a divergence over time in

income distributions across countries and within countries. For example, there is

strong evidence to suggest an emergence of “twin-peaks” in cross-sectional world


99

income distributions (Quah, 1996, 1997). This type of polarization is also present

in income distributions within countries. (Sala-i-Martin, 2006). Moreover, the

growth patterns of countries that follow the take-off process, exhibit a great deal

of diversity (Pritchett, 1997). For example, as discussed in the previous chapter,

some countries have taken off to sustained growth at the beginning of the 19th

century, while others have remained stagnant for an extended period of time.

According to Maddison‟s (2009) estimates, GDP per capita in countries that have

taken off to modern growth increased very rapidly and steadily, while countries

with stagnant growth lagged behind. Moreover, the countries in the former

category have adopted new technologies and have industrialized rapidly, while a

setback was experienced by countries in the second category.

This chapter studies a simple dynamic general equilibrium model of technology

adoption which is consistent with these stylized facts. In the model developed

here growth is endogenous, and agents are assumed to be heterogeneous in their

initial holdings of wealth and capital. The model here is very appealing in terms

of its ability to characterize growth outcomes of a very diverse nature.

Specifically, depending on initial productivity differences of the technologies, our

model is capable of characterizing three different growth outcomes that are

labelled as “poverty trap”, “dual economy” and “balanced growth”. As discussed

in the chapter 2, these types of growth performance are observed in different

countries/regions around the world. For example, it can be suggested that poverty

trap type of phenomenon may be observed in countries such as Zaire, Uganda,

Rwanda, and Nepal among many others because they are well behind the world


100

technology frontier, they experience extended periods of stagnation and their per

capita incomes are strictly less than the international poverty line (i.e. $2 a day).

The model constructed here also has the potential to explain diversity in the

growth patterns of transitional economies. For example, recall diversity of the

type observed in Figure 2.2 of Chapter 2. There were growth miracles in some

countries like South Korea, Taiwan, and Singapore etc., while there were

extended periods of stagnation in countries like Nepal. What causes such

diversity? According to the model developed here, this diversity may be the result

of productivity differences of technologies, however the variability in technology

adoption costs also adds to this.

Further findings indicate that in the presence of barriers or costs associated with

the adoption of more productive technologies, inequalities in wealth and income

may increase over time, tending to delay convergence in international income

differences. As discussed in chapter 2, it is accepted that during the early stages of

the British industrial revolution, per capita incomes of most workers either fell or

remained stagnant, which lead to increased inequality. According to the model,

this type of phenomenon may be explained as follows: poor households who

entail higher adoption costs relative to others may significantly defer their switch

to technology with higher productivity. Thus inequality increases. This idea is

consistent with the „skill-biased technological change hypothesis‟, discussed

earlier, which claims that technological change biased towards skilled labour leads


101

to increase (reduce) demand for skilled (unskilled) labour, and rise (drop) the

premium for skilled (unskilled) labour. Thereby inequality increases.

The results of the empirical study presented in this chapter appear to support

some of the implications of the model in this thesis. In particular, studies

presented support the model‟s prediction that initial inequality has a negative

impact on technology adoption. Due to constraints on data availability, testing all

of the implications of the model was not a possibility, but this is an interesting

direction for future research.

Some of the quantitative experiments suggest some interesting directions for

future research. Ideally, the variability in adoption costs should be modeled as a

process that is endogenous in the sense that it arises due to some institutional or

structural features characteristic of developing economies, and that is explicitly

modeled into the framework. Risks associated with the variability of adoption

costs may also be of importance for further research. Furthermore, the inequalities

that result from the process of transition indicate that political economy issues

would also have a bearing on economic growth. This issue will be addressed in

the next chapter of this thesis.

Political Economy of Costly Technology Adoption

102


103

CHAPTER 4

Growth Patterns and Inequality in The Presence of Costly Technology

Adoption: A Political Economy Perspective

4.1 Introduction

The subject of this chapter relates to the growing literature on the political

economy of development. Contemporary theoretical literature in this area

recognizes that policies and institutions are essentially endogenous, in the sense

that they are determined by what agents in the economy prefer (Krusell et al.

1997). In the context of technology adoption, one therefore has to consider

whether redistributive revenues of the government may, in fact, be allocated

towards reducing the fixed costs associated with productive technologies. To that

end, this chapter modifies the model of Chapter 3 by making the costs of

technology adoption endogenous. Specifically, the adoption cost is assumed to be

a decreasing function of the amount of government revenue allocated towards

cost-reducing research and development expenditures. Agents in the model vote

on the proportion of revenues allocated towards such expenditures.

Various strands of literature have motivational relevance for this study. Firstly,

the early political economy literature involving voting by agents includes the work

of Alesina and Rodrik (1994), in which inequality and growth are negatively

related, suggesting that the political economy mechanism does not necessarily

ensure that the best policies are chosen.28

The conventional explanation in this

28

See Alesina and Perotti (1994) also for a comprehensive discussion regarding this issue.


104

type of models is that the negative impact of inequality is likely to be caused by

the fact that, in a society with more unequal distribution of income, the poor will

vote for a high level of taxation, which impedes investments and economic

growth. In contrast to this idea, Li and Zou (1998) suggest that in the presence of

income inequality, the choice of income taxation through political process may

lead to higher economic growth. The rationale behind this is that, when

government revenue is used to finance public consumption instead of production,

poor agents in a more unequal society will vote for higher income taxation. More

precisely, they suggested that income inequality is not harmful, rather desired for

choosing policies that promote economic growth. However, depending upon the

framework in question, diverse conclusions are possible in relation to these issues.

Secondly, the stylized facts that motivate this study are linked to the ongoing

debate that was initially documented in Lucas (1993) and further discussed in

Benabou (1996). This debate relates to the fact that in a very egalitarian society,

the distribution of income plays a significant role in the take-off to modern

economic growth. This phenomenon is pertinent to some countries or regions

while not for others as apparent from the recent evidence. For example, China, a

developing nation had a very low inequality level before the policy reforms took

place in 1978, and was characterised by an egalitarian society. As evident in the

Figure 2.5 of Chapter 2, China experienced a dramatic increase in inequality since

1980s, after taking-off to rapid growth. In the case of China, it is then argued that

low level of inequality in the initial distribution played a significant role in

choosing policies that leads the take-off to modern economic growth (See Wan et


105

al. 2006 for more details in support of this evidence). On the other hand, the

Indian economy was characterised by high levels of inequality before taking off to

rapid growth in 1990s.

However, the presence of a correlation between inequality and choice of policies

that eventually leads to take-off is not necessarily a direct proof of the existence of

such a link. There is other type of additional evidence that lends support to this

hypothesis. For example, policies of this type are documented in Besley and

Burgess (2000) who suggest that the regulations implemented in relation to

wealth, in particular land redistribution in India were motivated by the presence of

high inequality. Moreover the inequality in the UK, a nation that has experienced

modern growth, has risen sharply compared to other developed nations in the

1980s. According to Zartaloudis (2007), a more plausible explanation for this

increase is based on political preferences, policy choices and reforms of the

government. Moreover, Massey (2009) suggests that over the past 30 years

inequality in the USA has increased dramatically, as the institutional arrangements

specific to the USA have failed to redistribute income to the same extent as other

industrial nations. Overall, therefore empirical evidence to substantiate whether

inequality directly influences growth oriented policies is inconclusive.

Another issue that has been explored to a very limited degree in this literature

relates to the implications for technology adoption in the presence of politico-

economic determination of policies. A notable exception is the model developed

by Krusell and Rios-Rull (1996). In a model with three-period lived agents they


106

study the technology adoption process and how vested interests of agents account

for policies that imply poor growth outcomes. Vested interests in their model arise

due to the presence of different trade-offs faced by heterogeneous agents in

relation to the technology adoption process. Agents operating the old technology

benefit more from preventing the adoption of a new technology since they have

not fully reaped the rewards from “learning by doing” that are associated with the

old technology. While their model has a very rich technological structure, this

complexity entails a simplification of agents‟ preferences which are assumed to be

linear.

The model constructed here, on the other hand, has more general preferences but

a simpler technological structure. Interestingly, results here indicate that even in

the absence of the type of technological trade-offs present in Krusell and Rios-

Rull (1996) there can be a delay in the adoption of more productive technologies.

The trade-offs in this model relate to the choice of alternative mechanisms of

redistribution. Tax revenues may be allocated towards two forms of redistribution:

one form of redistribution is adoption-cost reducing expenditure, while the other

form of redistributive expenditure is a lump-sum transfer. The political

equilibrium is characterized by situations in which the agents at the lower end of

the distribution may influence the outcome. In fact, the agents at the bottom end

of the distribution prefer redistribution in the form of the lump-sum transfer rather

than cost reducing research and development expenditure. This leads to an

outcome in which a less than optimal amount of government revenue is allocated

towards expenditure aimed at reducing costs of adoption.


107

In terms of the model developed here this government expenditure can be

viewed as government revenue channelled towards a variety of investments that

reduce the costs associated with adopting more productive technologies. For

example, during the green revolution (in 1970s), agricultural research investments

that were directed towards invention of high yielding varieties (HYV) of crops in

India and elsewhere significantly reduced the production cost, which led to wide-

spread adoption of planting HYVs (Fan, 2002). Moreover, the proportion of

R&D expenditure relative to GDP appears to increase in most developing regions

of the world, while East Asia and Pacific region records a highest level since 1997

(World Bank, 2008).

The results of numerical simulations indicate that technology with low

productivity is used by the majority of the individuals in the early stages of the

development. During this stage they prefer a very low proportion of government

revenue to be used to finance adoption-cost reducing expenditures. At this stage

the income distribution is characterized by a relatively higher level of inequality.

As capital deepening and redistribution of income and wealth takes place, the

inequality among individuals tends to decrease. Once this happens individuals

prefer a relatively larger proportion of government revenue to be allocated

towards cost-reducing Research and Development (R&D) expenditures.

Eventually all individuals make the switch to the better technology and

consequently their incomes converge. The economy is characterized by balanced

growth.


108

Another interesting outcome of the model is that there is a positive relationship

between inequality and economic growth. This result differs from Alesina and

Rodrik (1994) but is consistent with the studies of Li and Zou (1998). This result

is also consistent with empirical observations of Perotti (1992, 1993, 1996), and

Lindert (1996). Specifically, higher initial inequality in income and wealth in the

model of this thesis, promotes quicker adoption of more productive technologies.

The agents at the lower end of income distribution in the model prefer

redistribution in the form of the lump-sum transfer. Once they have accumulated

sufficient resources due to capital deepening, they make the switch to the

technology with higher productivity. More specifically, presence of a

redistributive mechanism with a proportional tax combined with capital deepening

enables relatively poor individuals to switch to more productive technology

quicker when initial inequality is high.

The section that follows describes the economic environment of the model.

Section 4.3 reports results of various numerical experiments that involve varying

some of the parameters of the model and the initial distributions of capital and

wealth. Section 4.4 concludes this chapter.


This section details a political–economy extension of the benchmark model

presented in the previous chapter. This extension assumes that the adoption cost

associated with the better technology is endogenous and dependent on cost

reducing public expenditures on Research and Development (R&D).


109

This modification also entails introducing a role for the government in this

economy. Here, a proportion ( ) of government tax revenue is used to finance

expenditure aimed at reducing adoption costs associated with the advanced

technology. The government raises revenue by means of an income and wealth

tax. This tax rate (τ) is levied on the heterogeneous resource endowments Wit of

each young agent and remains constant over time. This resource endowment

constitutes both capital and wealth of the agents and the distribution of this

endowment is described by a density function f(Wit) with support (0, υ). The

government tax revenue raised in any period is then given by

titititt WdWWfWGR0

)( where itW represents the resource endowment of

ith

agent in period t. The variable gt, which refers to the amount of government

tax revenue that is used to finance the adoption cost associated with technology B

is then given by tt Wg . The remainder of the government revenue is given

to the young agents as lump-sum transfers (trt), which are given by

tt Wtr )1( . At the “first stage” of each period, the agents vote over desired

value of and the political outcome is determined by majority rule.

In the “second stage” of period t, considering the political outcome, individuals

have to decide to adopt one of two technologies in order to produce output. As in

the previous models, the two technologies are referred to as Technology A and

Technology B. Here too, Technology A is associated lower productivity but does

not involve any adoption costs while Technology B is associated higher

productivity and involves an adoption cost. In contrast to the previous model, the


110

adoption cost associated with Technology B in this model is specified as a

decreasing function of the amount of government tax revenue that is used to

finance the adoption cost associated with technology B viz. (gt). In the related

literature, to our knowledge an example of a “reasonable” functional form for this

( )( tg ) is not available. However, to be consistent with empirical observations

this study looks at a functional form that fulfils the following conditions.

(i) 0)(' g ; 0)(" g .

(ii) 0)(lim gg .

Therefore the adoption costs function is specified as )1( t

tg

, where

)0( .

As in the previous model, the economy produces output (Y) using composite

human and physical capital (K) and the production relationships F(K) assume

simple “AK” specifications. Here also, the total factor productivities associated

with the two technologies are denoted by parameters A and B where AB .

Also as before, agents live for two periods with a new generation born in each

period. There are N agents in the economy and time is discrete, with t = 0, 1, 2….

The agents born in period t maximize following lifetime utility function, taking

into account what has occurred in the previous two stages.

)1()ln()ln()ln()ln(),,,( 12111111 itititititititit sxccsxccU


111

Here, as before itc and 1itc denote the agents‟ consumption in the first and

second period of life respectively. Each agent is born with a unit of unskilled

labour endowment that may be supplied inelastically to earn a subsistence

wage w . They also receive resources in the form of bequests from their parents.

Part of this bequest is given by, xit+1 which represents the wealth left to the next

generation. Parents also provide children with a share (α) of their second period

income. This component of bequests received by the young is represented by the

variable 1its . The parameter is the subjective discount factor in this model and

θ1 and θ2 are parameters representing the extent of intergenerational altruism in

the model.

The agents born in period t use their net wage-income plus resource

endowment and government transfer payments for consumption and capital

accumulation in the first period. In the second period, they use returns to their

capital holdings to finance consumption and bequests.

Households adopting Technology A face the following budget constraints:

)2())1(())(1(1 tit

a

it

a

it WWwKc

)3()1( 111

a

it

a

it

a

it xAKc

Here a

itc , a

itc 1 and a

itK 1 refer to first period consumption, second period

consumption and second period capital holding of the ith

individual who has

adopted Technology A. The variable itW represents the resource endowment of ith

agent in period t. In this model, the resource endowment that an agent can earn


112

depends on the technology that has been adopted by agent‟s parents. This means

that a

it

a

it

a

itit sxWW if the agent‟s parent adopted Technology A and

b

it

b

it

b

itit sxWW if the agent‟s parent adopted Technology B. Here, the

bequests that arise from parents‟ second period income t

a

it AKs )( if the agent‟s

parent adopted Technology A, and t

b

it BKs )( if the agent‟s parent adopted

Technology B. Households adopting Technology B, on the other hand, face the

constraints:

)4())1(()())(1(1 tttit

b

it

b

it WgWwKc

)5(.)1( 111

b

it

b

it

b

it xBKc

Note that here a household specific adoption cost (δit) of adopting Technology B

is experienced by the agents in period t.

The optimal plans for consumption, bequests and capital accumulation that takes

place in the third stage are described by the following equations. Agents adopting

Technology A will have:

)7())1(())(1()(

1

2

tit

a

a

it WWwc

)8())1(())(1(2

1

1 tit

a

aa

it WWwc

)9())1(())(1(2

1

11 tit

a

aa

it WWwx

)10())1(())(1()()1(

)1(

2

111 tit

a

aa

it WWwA

K

Likewise, agents who adopt Technology B will have:


113

)11()())1(())(1()(

1

2

tttit

b

b

it WWWwc

)12()())1(())(1()( 2

1

1 tttit

b

bb

it WWWwc

)13()())1(())(1()( 2

1

11 tttit

b

bb

it WWWwx

)14()())1(())(1()()1(

)1(

2

111 tttit

b

bb

it WWWwB

K

where,

)1()1(

1

21 Aa and

)1()1(

1

21 Bb

)})1(({1 212

Aa and. )})1(({1 2

12B

b

Applying the same reasoning as in the previous models, for a given value

of it is possible to write the following proposition that defines a threshold level

of resources required for a household to find it worthwhile to adopt the more

productive technology B. (See Appendix 4.1 for a proof of this proposition).

Proposition 4.1: Let ;)1)((

)())1)(((

12

2

*

21* wgW

W t

it

where )1(1

)1(

1

1

1

21

21

b

a and b

a

2

22 . A household will adopt

technology B iff *

itit WW .

Note that threshold level of resources ( *

itW ) here is decreasing in the proportion of

the government revenue that is used to finance the expenditure aimed at reducing

the costs of adopting technology with higher productivity ( ).


114

For this reason, it is hard to explicitly analyse the political outcome in the first

stage. However, in order to look at how agents will vote for desired , the effect

of changes in on an agent‟s indirect utility functions ),(itV is considered here.

This specifically involves examination of ),('

itV for each individual. This type

of analysis does not offer an explicit solution for the political outcome, however

some benchmarks can be identified that allow characterization of the political

outcome. Therefore, this exercise attempts to identify conditions under which

agents prefer extreme values of ( ) (i.e. a value of ( ) equal to 0 or 1). If

),('

itV is decreasing or increasing over the entire range of )1,0( the political

outcome is characterized by a “corner solution”. Otherwise, the political outcome

is characterized by an “interior solution”- a situation in which agents prefer

(0< <1).29

In order to interpret these conditions, two sets of individuals in this economy are

identified: (i) agents who are in the lower end of the income distribution- whose

resource endowments are strictly less than the threshold level of resources

( *

itit WW ), and (ii) agents in the upper end of the income distribution- whose

resource endowments are above the threshold level of resources ( *

itit WW ). Note

again that the critical level of resources ( *

itW ) is a decreasing function of ( ).

Therefore changes associated with ( ) also change the number of agents in these

two sets. This means that some agents at the top end of the first set are likely to

29

We look at the case in which agents vote on (τ) in the Appendix 3.3. The results are not

presented here as it is a relatively uninteresting problem given the structure of the model. It is

obvious that in the presence of inequality the majority outcome would entail (τ) =1.


115

switch to the second set as ( ) changes. The conditions for the two sets of

individuals are summarized in the following proposition (See Appendix 4.2 for

proof of this proposition).

Proposition 4.2:

(i) For agents, *

itit WW , ,0),('

itV ; all agents in this group vote for

0

(ii) For agents, *

itit WW , ),('

itV is ambiguous; the agents in this group prefer

a value of )1,0( iff 1)]('[ tW

Overall, this proposition implies that the poorer agents prefer redistribution in

the form of the lump-sum transfer, while richer individuals prefer redistribution in

the form cost reducing research and development expenditure. These issues are

also analyzed numerically in the numerical experiments section below.

4.3 Numerical Experiments

This section presents results of numerical experiments conducted using the

model developed in this chapter. Firstly, this section focuses on how voting on

takes place in the political process and consequences of this political outcome on

the process of technology adoption and economic growth. Secondly an

examination of the extent to which political outcomes differ from welfare

maximizing ones is presented. Thirdly, there will be an exploration of the

implications of varying levels of tax rate for the technology adoption process. In

particular, these implications are compared with the implications for the


116

technology adoption process of the previous model presented in Chapter 3.

Finally, the cases of how initial levels of inequality matter for the political

outcome and the technology adoption process are addressed.

4.3.1 Political Outcome

To examine how voting on takes place in the political process the experiment

conducted here look at the proportion of individuals that vote for different values

of .30

The implications of these experiments, in fact, are consistent with the

Proposition 2 presented in section 2 of this chapter. This means that at early

stages, the majority of agents wish to allocate entire government tax revenue in

the form of lump-sum transfers. The political outcome at the early stage is then

characterized by = 0 and the majority of the agents adopt Technology A at this

stage. However, in the latter stages political outcome is characterized by a

relatively lower value of , and eventually the winning value of reaches zero.

The underlying reason for this outcome can be explained as follows. As

discussed above, there are two sets of agents who vote for different values of .

The first set (i.e. agents at the lower end of income distribution) prefer

redistribution in the form of the lump-sum transfer while the second set (i.e.

agents in the top end of the income distribution) prefer that a positive fraction of

the government tax revenue is used in the cost reducing expenditure associated

with adoption of Technology B. Figure 4.1 shows that, at early stages of

technology adoption process, the first set consists of approximately 80% of agents

30

The values of other parameters in this experiment are same as that of Table 2 of Chapter 3.


117

while the second set consists of the rest of the households. The first set votes for a

value of = 0 while the second set votes for a higher value (0.05) of . (See

Figures 4.2, and 4.3, for illustration of these facts). Over time however, as

redistribution takes place, agents who are at the top end of the first set also wish to

allocate tax revenue on cost reducing R&D expenditure. At this stage these agents

have accumulated sufficient resources as capital deepening takes place, to allow

them to make the switch from Technology A to B. Therefore the proportion of

agents who vote for lower value of decreases and the political outcome is now

characterized by a relatively higher value of . Once all agents adopt

Technology B, as illustrate in Figure 4.1, all agents vote for a relatively lower

value of , and eventually the winning value of reaches zero.

Next, the implication of the above process for the evolution of inequality

within the economy over time is analyzed. At early stages the income distribution

is characterized by a relatively higher level of inequality. In contrast to the

evolutionary pattern of inequality in the benchmark model, here, in the presence

of a redistributive mechanism, inequality decreases over time and after complete

adoption to Technology B the level of inequality converges to zero (See Figure

4.4). This outcome of the model appears to support the idea that downward

segment of the Kuznets curve is driven by issues related to political reforms and

its consequences- a fact described in contemporary political economy literature

(For example, see Lindert, 1994).


118

This segment of the discussion examines the pattern of growth rates of

output over time for this economy. As before, this exercise looks at patterns of

growth for households that are in the lowest 20%, the highest 20%, and the mean

and median positions in the income distribution. These patterns show a significant

amount of diversity across different cohorts of households. As illustrated in Figure

4.5, richer households in the model show a monotonic pattern of growth while

poor and median households‟ growth patterns are characterized by rapid growth

and reversals. However, eventually outputs of all individuals in this economy

converge to a unique steady state growth rate. It is apparent from this figure that

the outputs of agents who are at the bottom end of income distribution grow at a

rapid rate relative to the growth rates of outputs of agents that are at the top end of

income distribution.

Figure 4.1: Number of households adopting Technology A or B in different time

periods.


119

Figure 4.2: Winning in different time periods.

Figure 4.3: Proportion of households vote in favour of winning in different

time periods.


120

Figure 4.4: Gini coefficient in different time periods.

Figure 4.5: Growth rates experienced by the various cohorts of households.


121

4.3.2 Policy Choice under Welfare Maximization and under Political Process

The experiment presented here, firstly, compares the political economy outcome

for , (above section) with the welfare maximizing outcomes. Here, welfare is

measured using utilitarian concepts, especially the value of that maximizes the

sum of utilities of all agents in the economy is considered here. It is interesting to

observe that the policy choices in these two cases are significantly different,

particularly in the transition period before complete adoption of advanced

technology. As illustrated in Figure 4.6, the individuals always vote for a smaller

while a welfare maximization point of view suggests that relatively large is

efficient. This bears out the fact that aggregate outcome of public choice is more

likely to be conservative in nature, as it represents the majority choice among

conflicting preferences of households. This is an issue often discussed in public

choice literature (Besley and Coate, 2003).

Secondly, how the implications of these two cases differ for technology

adoption, evolution of inequality, and growth are examined here. It is clear from

Figure 4.7 that the political process slows down the process of technology

adoption relative to the optimum welfare policy. This, in part, appears to support

idea that technology adoption always involves some kind of resistance as

explained by Mokyr (1993) and Krusell and Rios-Rull (1996). However,

reduction in inequality does not differ noticeably across these two cases, as in

Figure 4.8. Furthermore, the diversity in the patterns of growth in these two paths

shows significant differences. (See Figure 4.9) In the case of political economy

outcomes, the rate of growth in the transition period before technology adoption is


122

characterized by drastic rises and falls relative to the case that involves a welfare

maximization path. However eventually, the economy converges to the same

steady-state growth rate. In terms of this feature of the model, it is clear that if

fundamental characteristics are similar, economies eventually converge to an

identical rate of growth regardless of the alternate policy choices.

Figure 4.6: Winning value of under welfare maximization path and political

process.


123

Figure 4.7: Number of households adopts technology B under welfare


Figure 4.8: Evolution of Gini coefficient over time under welfare maximization

path and political process.


124

Figure 4.9: Growth rates experienced by the different cohorts of households

under welfare maximization path and political process.

4.3.3 Experiments That Vary Income and Wealth Tax Rates

This section details the implications for varying levels of the tax rate for the date

of complete adoption, the evolution of inequality and the diversity of growth

patterns for various cohorts of households are addressed in this section. In the

model, taxes enter in a very simple way, given that labour supply in this economy

is inelastic. An obvious consequence of this is that higher taxes have positive

implications on technology adoption and economic growth. Intuitively, higher

taxes imply a faster redistribution of income and wealth which enable poorer

households to pay for the adoption costs associated with Technology B. This in

turn reduces the income and wealth inequality among the agents in this economy.


125

Another obvious feature of the model relates to the implications for varying tax

levels on diverse growth patterns of different cohorts of households in the income

distribution. High income and wealth taxes in the model therefore accelerate the

starting point of steady state growth. Moreover, the rate at which the economy

grows at the steady state increases with higher tax rates.

The results presented above can be compared with the results of a case that does

not involve a redistribution process. It is obvious that, if τ = 0, the implications of

this for technology adoption, evolution of inequality and growth are the same as

that of the benchmark model of this thesis. In order to illustrate the beneficial

nature of the taxation however, we set income and wealth tax rate equal to a value

which is very close to zero (τ = 0.001) and analyze the outcome of the model.

Results suggest that very low rate of taxation increases inequality in the process of

technology adoption. In the very long run however, inequality tends to decrease.

(See Figure 4.10). The implication of this feature on cross country differences in

the evolution of income is obvious. In the process of technology adoption,

countries in which effective taxation system exists are likely to reduce inequality

sooner.


126

Figure 4.10: evolution of inequality with and without taxes.

Furthermore, results here suggest that in contrast to the benchmark model the

growth patterns of households are relatively smooth and monotonic and are less

likely to be characterized by reversals. A significantly low tax rate which moves

the economy here closer to the benchmark case, on the other hand characterizes

reversals in the growth rates particularly in the cases of poorer and median

households. However, as illustrated in Figure 4.11, output of poor and median

agents grows rapidly relative to the richer households. This is a fact that has been

observed empirically as well (For example see Quah, 1996).


127

Figure 4.11: Growth rates experienced by the different cohorts of households

with and without taxes.

4.3.4 Experiments That Vary Initial Inequality Levels.

The implications for varying initial inequality levels on the extent of technology

adoption, evolution of inequality and patterns of growth across different cohorts

of households in the income distribution are examined here. The results suggest

that if the initial inequality is relatively high, the rate at which technology is

adopted, as well as rate at which inequality decreases over time is also high

(Figure 4.12). Moreover, as illustrated by Figure 4.13, our model suggests that

higher levels of initial inequality have a positive impact on economic growth- a

fact that can be interpreted as supportive of the idea that countries with identical

technological and structural features but differing in initial inequality levels grow

at different rates before they converge to steady state growth.


128

Figure 4.12: Gini coefficient in different time periods with varying levels of

initial inequality.

Figure 4.13: Growth rates experienced by the poor cohort of households with

varying levels of initial inequality.


129


Contemporary literature on the political economy of development suggests that,

to some extent, political considerations behind policy determination provide a

potential explanation for uneven growth records within and across countries. This

chapter extends the benchmark model of this thesis to accommodate such political

considerations. The assumption here is that the adoption cost associated with the

better technology depends on cost reducing public expenditures on R&D. The

proportion of government tax revenue used to finance this expenditure is

determined by a political process.

The model constructed here suggests that agents at the bottom end of the income

distribution prefer redistribution in the form of the lump-sum transfer while agents

at the top end of the distribution prefer redistribution in the form of cost reducing

R&D expenditure. The political outcome depends on the majority of votes. Over

time however, as capital deepening and redistribution takes place, complete

adoption to Technology B is inevitable and the economy converges to a steady

state.

Furthermore, the results appear to support the fact that the policies chosen

through the political economy mechanism do not necessarily ensure maximum

welfare of the society. In particular, in the transition period before complete

adoption of advanced technology, public choice of policy is different from that of

the social planner. This bears out the fact that aggregate outcome of public choice

is more likely to be conservative in nature, as it represents the majority choice


130

among conflicting preferences of households. This is, in part, consistent with

Mokyr‟s (1990) idea that adoption of technologies often faces severe „resistance‟

of various interest groups. This is also consistent with the model of Krusell and

Rios-Rull (1996) which suggests the fact that „vested interests‟ of the political

elite leads to a slower pace of technological change.

Furthermore, in common with the some of the previous literature including

Alesina and Rodrik (1994), the model of this chapter suggests that higher initial

inequality in income and wealth is positively linked to economic growth.

However, the political mechanism involved in this process is different from

previous work. According to the model, the positive impact is likely to be caused

by the fact that, in a society with more unequal distribution of income, the poor

will vote for a high level of lump-sum transfers. After reaching a certain wealth

point, these transfers allow individuals to switch the technologies. This eventually

facilitates the capital deepening process which leads to economic growth.

Moreover, the model further suggests that the above positive impact is likely to be

decelerated by policy outcomes of a political process relative to those of a welfare

maximization process. This again emphasizes the fact described previously,

related to the „resistance‟ of various interest groups associated with adoption of

advanced technologies.

Some of the implications of the model developed here suggest several useful

directions for further research. In particular, empirical analysis to test the

implications of the model is of importance. Furthermore, alternative mechanism


131

of redistribution could introduce different types of trade-offs that have not been

explicitly analyzed here. For example government revenue could be used to

finance other public goods such as health care, environment etc. Depending on the

menu of choices available one could then have different outcomes for the

proportion of revenue allocated to cost reducing R&D expenditure. This point has

been considered for example in Lahiri and Magnani (2008)


132


133

CHAPTER 5


This chapter contains a brief summary and the main outcomes of the study

presented in previous chapters of this thesis. The main outcomes are discussed in

relation to the stylized facts and policy issues discussed in the previous chapters

followed by insights for potential policy recommendations.

This thesis attempts to explore various macroeconomic issues associated with

technology adoption in the process of economic growth. The motivation for the

study generally relates to income differences within and across countries, and the

diversity of growth patterns observed in countries. As discussed in the previous

chapters, empirical evidence suggests that there has been a divergence over time

in income distributions across countries and within countries (Quah, 1996).

Moreover, the growth patterns of countries that follow the take-off process, exhibit

a great deal of diversity. Some countries have adopted new technologies and have

industrialized rapidly, while others have experienced setbacks.

This study is further motivated by the issues related to the politico-economic

determination of policies for technology adoption and growth. In particular, the

increased inequality associated with the technology adoption process may entail

social conflicts. The study aims at exploring how conflicting interests of agents in

the population influence the growth-related economic policies of countries.


134

In order to address these issues a simple dynamic general equilibrium model of

costly technology adoption has been developed. In the benchmark model of this

thesis growth is endogenous, and agents are assumed to be heterogeneous in their

initial holdings of wealth and capital. In order to produce output individuals have

to decide to adopt one of two technologies available in this economy where one

technology is associated with higher productivity relative to the other. The

adoption of the advanced technology is associated with costs incurred by each

agent. Unlike the models in the related literature that deal with similar processes

of technology adoption, the adoption costs in this model are allowed to represent

household specific stochastic shocks that also vary over time. An overlapping-

generations structure is imposed here in order to consider a situation in which the

adoption decision is no longer of a “one-time” irreversible nature - in every time

period the new generation undertakes the technology adoption decision

irrespective of whether the previous generation had switched to the advanced

technology or not. As mentioned before, some examples of these types of

technologies involve high yielding varieties (HYVs) of agricultural crops,

Genetically Modified (GM) crops etc. In fact, HYVs represent higher

productivity over the traditional agricultural crop varieties which have been

adopted by countries during the initial periods of the growth processes. However,

today GM crops have suppressed HYVs and now appear to play the role of an

advanced technology relative to established HYVs. The overlapping generations-

structure imposed in the model here is therefore more plausible interms of its

relevance to these types of technologies as it reflects the fact that while a

generation is faced with a one-time cost of adoption, the dynasty to which the


135

household belongs faces ongoing adoption decisions in successive time periods.

In the literature, other examples of advancing technology of the type discussed

here include the energy sector- for example steam-power which arrived in the

early 18th

century was superseded in many applications by electricity and the

internal combustion engine which arrived at the turn of the century. Moreover,

information technology (IT) had experienced ongoing technological advancement

since its introduction in 1970s (See, Jovanovich and Rousseau, 2005 for further

discussion). More examples include communication technologies such as mobile

phones and internet facilities. As adoption of such technologies often takes place

all over the world all the time, the variable and repeating nature of adoption costs

included in the new model therefore provide better insight towards understanding

how these type of advanced technologies are adopted in practice.

The benchmark model presented here is very appealing in terms of its ability to

characterize diverse growth outcomes of countries. Specifically, depending on

initial productivity differences, the model is capable of characterizing three

different outcomes which are labeled as “poverty trap”, “dual economy” and

“balanced growth”. These types of stylized features have not been dealt with the

help of a single model in the existing literature.

The above outcomes are very commonly observed among the growth

experiences of various countries and regions in the world. Recall the economic

performance of poverty trap-countries discussed earlier in Chapter 2. The per

capita incomes of several less developed countries are less than $2 a day (World


136

Bank, 2005) and per capita incomes of most of these countries have stagnated or

declined significantly since record keeping began (Madisson, 2009). For example,

annual per capita incomes of countries like Zaire, Uganda, Tanzania, Rwanda and

Burundi are still strictly less than $1000 - that is approximately $3 per day

(Madisson, 2009). In contrast, it is observed that countries in the Western region

and Western off-shoots appear to have experienced sustained growth during the

last several decades (Galor, 2005). Most of the newly emerging economies

including China, India and perhaps Vietnam are more likely to be characterized by

a dual structure representing elements of sustained growth in modern sectors

while stagnation or limited growth in traditional sectors (Jaumotte et al. 2006).

Moreover, the model also has the potential to explain the observed diversity in

the growth pattern of transitional economies. In the case that is labeled as

“poverty trap”, growth rates of the poorer and median households monotonically

increase while rich agents experience a reversal before eventual stagnation. In the

case of “dual economy”, the economy is characterized by two distinct rates of

growth that correspond to the two technologies at the steady state. In the

“balanced growth” case, the rates of growth of the different groups of households

converge in the steady state, and the economy is characterized by a permanent and

stable growth. As discussed in the previous chapters, growth experiences of most

countries resemble these phenomena. For example, Pritchett (2000) reveals that

all OECD countries have steady growth (hills) and most Sub-Saharan African

countries experience continuous stagnation (plains) or rapid growth followed by

decline (mountain) during the period 1960 to1992. He further suggests, though


137

this describes the general trend, there is enormous volatility of growth around this

trend in these countries, which is dominated by shocks and recovery. Recall also

that the growth patterns of several countries observed in Chapter 2 (Figure 2.2)

exhibit this type of volatile pattern. The policy-relevant insights provided by the

above features of the model illuminate some of the causes of such diversity.

According to the model developed here, this type of diversity may be the result of

productivity differences of technologies; however the variability in technology

adoption costs also contributes. It is thus evident that, in the episodes of policy

reforms which result in technological changes of any form, growth performance

of countries are likely to exhibit wide-ranging patterns.

Furthermore, Khan and Ravikumar, (2002) and other technology adoption

models in general, suggest that income inequality widens in the process of

technology adoption, given the fact that costs are “fixed for all” and incurred by

agents only “once”. However, in terms of the model constructed here, it is found

that in the presence of either fixed or variable barriers / costs associated with the

adoption of more productive technologies, convergence in international income

differences may be delayed over time. In particular, situations in which variability

is associated with barriers to technology adoption, the shocks and recoveries in the

adoption process are more frequent, leading to additional variability in the growth

patterns. This further delays the convergence in international income differences.

A similar type of phenomenon is not uncommon in the developing world; for

example, the World Bank (2008) reports that though most of the developing

countries have converged with the level of technological achievement in the high-


138

income countries over the past 15 years, macroeconomic turmoil/shocks

experienced by many countries in Latin America, in part, contributed to weak

technological achievement and low growth performance in this region.

In terms of policy relevant implications, the above outcome of the model is able

to provide some insights which are essentially interrelated with variability of

adoption costs. As the variability associated with adoption of advanced

technology in the model represents external macroeconomic shocks, the policies

addressing the external shocks may be more effective in assisting the process of

technology adoption. In particular, it is important to have effective institutional

arrangements that may shield the adverse effects of such shocks in the transition

process in order to attain positive growth.

Moreover the model in this thesis suggests that in the presence of either fixed

or variable barriers, the inequalities in wealth and income may increase over time.

If such distributional impacts are associated with the technology adoption process,

emergence of social conflicts is inevitable. This further suggests that the process

of technology adoption invariably requires promising mechanisms or policy

instruments aimed at redistribution of wealth and income.

The above outcomes of the benchmark model suggest that consideration of

distributive conflicts among agents is of importance in terms of further research.

To that end, the benchmark model was extended in Chapter 4 to explore politico-

economic considerations for determination of policies regarding technology


139

adoption. This extension assumes that the adoption cost associated with the better

technology is endogenous and depends on cost-reducing public expenditures on

Research and Development (R&D). Here, the proportion of government tax

revenue used to finance expenditure aimed at reducing adoption costs is

determined by a political process. The trade offs in this process therefore involve

two forms of redistribution: one form of redistribution is adoption-cost reducing

expenditure, while the other form of redistributive expenditure is a lump-sum

transfer. The majority of votes determine the political outcome.

Results of this extension appear to support the idea that the policies chosen

through the political economy mechanism do not necessarily ensure maximum

welfare of the society. In particular, in the transition period before complete

adoption of advanced technology, policy outcomes of a political mechanism are

not as efficient as those of a social planner. This bears out the fact that the

aggregate outcome of public choice is more likely to be conservative in nature, as

it represents the majority choice among conflicting preferences of households

(Besley and Coate, 2003). As emphasized previously, this is, in part, consistent

with Mokyr‟s (1990) idea that adoption of technologies often faces severe

„resistance‟ of various interest groups. This is also consistent with the model of

Krusell and Rios-Rull (1996) which suggest the fact that „vested interests‟ of the

political elite lead to a slower pace of technological change.

Another interesting outcome of the model of this thesis is that there is a

positive relationship between inequality and economic growth. However, the


140

political mechanism of the model is different from the explanations proposed in

the previous literature including Alesina and Rodrik (1994) and Li and Zou

(1998). The model of this thesis suggests that the positive impact of inequality is

likely to be caused by the fact that, in a society with more unequal distribution of

income, the poor will vote for a high level of lump-sum transfers. At certain

wealth points, these transfers can be used to pay the costs involved in adoption of

advanced technologies. This will eventually facilitate the capital deepening

process and leads to economic growth. However, the positive impact of inequality

is likely to be decelerated by policy outcomes of a political process relative to

those of a welfare maximization process. Though it is hard to provide robust

support to advocate growth-oriented policies using the outcomes of the new

model, it can be suggested that political equilibrium is relatively restricted in

devising such policies compared to strict welfare maximization processes.

Furthermore, in the model developed in this thesis choosing the level of inequality

does not directly fall within the political process. However, it can be further

suggested that there are particular situations in which endogenously determined

inequality levels can lead to political outcomes which ensure rapid take-off to

modern economic growth.

In addition to the political-economy extension, the benchmark model of this

thesis also provides some important directions for further research, and they are

discussed below. In particular, the variability in adoption costs in the benchmark

model should in future be allowed to capture other stochastic shocks, essentially

in an endogenous form. These shocks that account for variability may include


141

changes associated with institutional or structural setting, and in particular policies

that produce such variability. For example, one interesting direction would be to

consider the variability in adoption costs caused by trade policies. The hypothesis

in relation to this is that countries that import goods from technologically

advanced countries are more likely to adopt advanced technologies as there is a

knowledge spillover through trade (Some evidence in support of this hypothesis

can be found in Coe and Helpman, 1995; Coe et al. 1997).

Moreover, the economy in the model of this thesis has an inelastic labor

supply. If an endogenous labour-leisure choice is introduced to the model, the

political outcome of the model in Chapter 4 will be a different one. That is, in the

voting process, agents in the model will consider work effort in addition to

redistributive conflicts when choosing suitable policies. Furthermore, the issue of

a politically determined tax rate remains a non-trivial one. This means that

individuals are allowed to vote over a desired tax rate in addition to the proportion

of government expenditure on R&D. Moreover, the trade-offs of the current

model are two-fold. It would be interesting to consider a set / package of policies;

for example, government revenue could be used to finance other public goods

such as health care, environment, etc. Such alternative set of choices may leads to

different outcomes for the proportion of revenue allocated to cost reducing R&D

expenditure.

Moreover, the resource endowments in the model are not allowed to transfer

across agents through a financial market. However, in the literature that discusses


142

technology adoption process, the role of financial intermediation has been

investigated by Greenwood and Jovanovic (1990). In a similar vein however,

introduction of financial markets into this model will be another important

direction for further research.

The benchmark model here abstracts from risk associated with variability in

adoption costs. An interesting direction of research would therefore take into

consideration the role of risky technology adoption.

Finally, to sum up, it is demonstrated that by using a model of costly

technology adoption with simple linear technologies, barriers to technology

adoption can be shown to hinder economic growth by delaying the technology

adoption process. While variability associated with such barriers exacerbates this

effect, the productivity differences of technologies have a significant impact on

overall growth outcomes. In particular, depending on productivity differences

between the technologies, the model constructed here characterizes three growth

outcomes labeled as „poverty trap‟, „dual economy‟ and „balanced growth‟. The

model is then capable of explaining the observed diversity in growth patterns

across countries, as well as divergence of incomes over time. Furthermore, the

inequalities increase in the process of technology adoption which entails the

probability of social conflict emerging in the growth process. To that end, one of

the chapters of this thesis considers a political economy extension which allows

agents to vote over alternative mechanisms of redistribution. This extension

suggests that the outcomes of the political process lead to complete adoption of


143

the better technology. In the transition process however, the policies chosen

through the political process do not ensure the maximum welfare of the society.

In terms of the growth rates, the poor grow at a relatively rapid rate and catch-up

the growth rates of the rich and their incomes eventually converge.

Appendix for Chapter 3

144


145



Households adopt Technology B iff indirect utility of Technology B is greater

than indirect utility of Technology A. This implies

)ln()ln()ln()ln(

)ln()ln()ln()ln(

),,,(),,,(

12111

12111

111111

a

it

a

it

a

it

a

it

b

it

b

it

b

it

b

it

a

it

a

it

a

it

a

it

ab

it

b

it

b

it

b

it

b

sxcc

sxcc

sxccUsxccU

Recall that, here 11)( t

a

it AKs and 11)( t

a

it BKs

Also re-write 6-8 in terms )( a

itc of and equations 10-13 in terms of )( b

itc . Then

substitute all of these into the above inequality. After simplifying,

)1(

)1(ln.ln).ln()ln(

)1(

)1(ln.ln).ln()ln(

1

211

1

211

a

a

ita

a

ita

a

it

a

it

b

b

itb

b

itb

b

it

b

it

cccc

cccc

We can further simply the above inequality condition as follows.

2

1

2

1

)1(

)1(ln.ln)ln()ln(

)1(

)1(ln.ln)ln()ln(

1

1

1

1

a

a

ita

a

ita

a

it

a

it

b

b

itb

b

itb

b

it

b

it

cccc

cccc

2

121

2

121

)1(

)1(ln)ln()ln()ln(

)1(

)1(ln)ln()ln()ln(

1

1

)1(

1

1

)1(

aaa

a

it

bbb

b

it

c

c


146

2

12121

2

12121

)1(

)1(ln)ln()ln()ln(

)1(

)1(ln)ln()ln()ln(

1

1

)1(

1

1

)1(

a

a

it

b

b

it

c

c

21211

21212121 )ln()ln()ln()ln(11

b

a

a

it

b

it

ba

a

it

b

it

c

c

cc

Substituting for a

itc , b

itc (in equations 7 and 11 in the Chapter 3) a and b as

defined in the same chapter, it is possible to define threshold level of resources

required for a household to adopt more productive technology )( *

itW as follows.

;)( 21

1* wW itit where 1 and 2 are defined as in Chapter 3.


147



For agents adopting Technology A, optimal plans are given by following

equations.

)1())1(1(

1it

A

it Wwc

)2())1(1(

)1(1 it

A

it WwA

c

)3())1(1(

)1(1 it

A

it WwA

x

)4())1(1(

)1(1 it

A

it WwK

Likewise it can be shown that agents who adopt Technology B will have:

)5())1(1(

1itit

B

it Wwc

)6())1(1(

)1(1 itit

B

it WwB

c

)7())1(1(

)1(1 itit

B

it WwB

x

)8())1(1(

)1(1 itit

B

it WwK

The dynamics of this model are described by the following system of first order

difference equations.

)9(

)1(1

)1(

)1(1

)1(

*

1

1

itit

it

A

it

it

A

it

WWfor

WwAx

WwK


148

)10(

)1(1

)1(

)1(1

)1(

*

1

1

itit

itit

B

it

itit

B

it

WWfor

WwBx

WwK

where wW itit

1

* with defined as in Proposition 2


149



For agents adopting Technology A, optimal plans are given by following

equations.

)11())1(1(

1it

A

it xwc

)12())1(1(

1 it

A

it xwA

c

)13())1(1(

1 it

A

it xwA

x

)14())1(1(

)1(1 it

A

it xwK

Likewise it can be shown that agents who adopt Technology B will have:

)15())1(1(

1itit

B

it xwc

)16())1(1(

1 itit

B

it xwB

c

)17())1(1(

1 itit

B

it xwB

x

)18())1(1(

)1(1 itit

B

it xwK

The dynamics of this model are then described by the following system of first

order difference equations.


150

)20(

)1(1

)1(1

)1(

)19(

)1(1

)1(1

)1(

*

1

1

*

1

1

itit

B

it

B

it

B

it

B

it

itit

A

it

A

it

A

it

A

it

xxfor

xwAx

xwK

xxfor

xwAx

xwK

where wx itit

1

* with defined as in Proposition 3.


151

Appendix 3.4: Sensitivity analysis for the case of “poverty trap”

This section presents the results of the experiments that were conducted

assuming the “poverty trap” case of Model 1. That is, productivity parameters are

given as A<B<P*. Typically, in this case all agents in the economy eventually

adopt Technology A. The inequality initially increases and then falls to zero.

Growth rates of the various cohorts of households in the income distribution

converge and economy experiences stagnation. In the section below, the impact of

changes in other parameters are examined (i.e. δ, θ1, θ2, α and initial inequality

levels) for the outcomes of the model. The results suggest that the variabilities of

these parameters do not affect the qualitative implications of the model although

they matter on a quantitative sense. The results are presented below. The figures

are self explanatory.

(i) Experiments with the adoption-cost parameter

Figure A 1: Number of households in Technology A and B for different levels of

adoption costs.


152

Figure A 2: Inequality over time for different levels of adoption costs.

Figure A 3: Growth rate for rich household with different levels of adoption

costs.


153

(ii) Experiments with varying initial inequality levels

Figure A 4: Number of households in Technology A and B for different initial

inequality levels.


154

Figure A 5: Inequality over time for different initial inequality levels.

Figure A 6: Growth rate for rich household with different initial inequality levels.


155

(iii) Experiments with θ

Figure A 7: Number of households in Technology A and B for different levels of

altruism parameter.


156

Figure A 8: Inequality over time for different levels of altruism parameter.

Figure A 9: Growth rate for rich household with different levels of altruism

parameter.


157

(iv) Experiments with α

Figure A 10: Number of households in Technology A and B for different levels

of α parameter.


158

Figure A 11: Inequality over time for different levels of α parameter.

.

Figure A 12: Growth rate for rich household with different levels of α parameter.


159

Appendix 3.5: Sensitivity analysis for the case of “dual economy”

This section presents the results of the experiments that were conducted

assuming the case of “dual economy” of Model 1. That is productivity parameters

are given as A<P*<B. These experiments investigate the impact of changes in

other parameters (i.e. δ, θ1 ,θ2, α and initial inequality levels) on the outcomes of

the model. Note that the overall outcomes of these experiments are typical to

“dual economy” case, as discussed in section 3.3.1 (i). Again, the variabilities of

these parameters do not affect the qualitative implications of the model. In the

section below, these results are presented.



of adoption costs.


160


Figure A 15: Growth rate for median household with different levels of adoption

costs.


161

(ii) Experiments with varying initial inequality levels


inequality levels.


162


Figure A 18: Growth rate for poor household with different initial inequality

levels.


163

(iii) Experiments with θ


of altruism parameter.


164


Figure A 21: Growth rate for median household with different levels of altruism

parameter.


165

(iv) Experiments with α


of α parameter.


166

Figure A 23: Inequality over time for different levels of α parameter.

Figure A 24: Growth rate for median household with different levels of α

parameter.


167

Appendix 3.6: Implications of increasing adoption costs for technology

adoption process

Figure A 24 looks at increases in adoption costs over time. Consider an

experiment in which adoption costs grow at a rate of 1%, over time, starting at a

minimum value of 15. It is important to emphasize that this is simply a thought

experiment based on a somewhat “ad-hoc” process for adoption costs. The results

reported in Figure A 24 suggest that increasing adoption costs appears to delay the

date of complete adoption significantly.

Figure A 25: Adoption costs increases over time


168

Appendix 3.7: Implications of adoption costs that vary over time for the case

of “poverty trap”

This section presents an experiment that is analogous to the experiment

presented in section 3.3.1(ii). This section deals with the case of “poverty trap”.

Results produce similar interpretations in relation to the growth process, as

presented below.

Figure A 26: Technology adoption, evolution of inequality and growth patterns

for the case of “poverty trap”


169

Appendix 3.8: Implications of adoption costs that vary over time for the case


This section presents an experiment that is analogous to the experiment

presented in section 3.3.1(ii). This section deals with the case of “dual economy”.

Results produce similar interpretations in relation to the growth process, as shown

below.


for the case of “dual economy”


170


of “poverty trap”

Below experiments are conducted which are analogous to the experiment

presented in section 3.3.1 (iii). This section considers the case that of “poverty

trap”. The overall results of this experiment also resemble the typical outcomes of

the “poverty trap” case.


for the case of “poverty trap”


171



Here experiments were conducted analogous to the experiment presented in

section 3.3.1 (iii). This section considers the case that of “dual economy”. The

overall results of this experiment also resemble the typical outcomes of the “dual

economy” case.


for the case of “dual economy”


172

Appendix 3.11: Sign of *

itW

As stated in proposition 1, the threshold level of resources in the model depends

on altruism parameter (θ). The effect of changing altruism parameter (θ) on the

threshold level of resources (Wit*) therefore can be written as,

))1(1()1(

)/ln(2

* BAW itit where defined as in Proposition 2. If adoption

costs are fixed across households and time, and as (A/B)> 0, it can be seen from

this equation that ;0*

itW which implies that more altruistic households are

likely to reach the threshold level of resources sooner, and they are capable of

making the switching decision sooner.


173


Only the results for the case of balanced growth and adoption costs fixed across

households are presented here.


Figure A 30: Number of households adopting Technology A or B in different

time periods with varying adoption costs. (T* in this figure refers to the date of

complete adoption).


174

(ii)Experiments with α

Figure A 31: Number of households adopting Technology A or B in different

time periods with varying levels of education expenditure parameter (α). (T* in

this figure refers to the date of complete adoption)


175

Figure A 32: Gini coefficient over time for different education expenditure

parameter (α).


176

(iii) Growth patterns of different cohorts of households

Figure A 33: Growth rates experienced by the various cohorts of households


177


Below are the results for the case of “balanced growth” and adoption costs fixed

across households only.



of adoption costs.


178


Figure A 36: Growth rate for median household with different levels of adoption

costs.


179

(ii) Experiments with the varying initial inequality levels


inequality levels.


180


Figure A 39: Growth rate for poor household with different initial inequality

levels.


181

Figure A 40: Growth rate for rich household with different initial inequality

levels.


182

(ii) Experiments with the different levels of altruism parameter


of altruism parameter.


183


Figure A 43: Growth rate for rich household with different levels of altruism

parameter.


184

Figure A 43: Growth rate for poor household with different levels of altruism

parameter.


185

Appendix 3.14: Data Set

The “Technology: creation and diffusion” data base of the Human Development

Report (2006) was used to construct the “Index of technology adoption” (ITA)

and the data set consists of 104 countries. To gather data for TR and AT variables

in the data set the same report was again used. Specifically, the data for 2004 was

used and the countries with missing data were excluded from analysis in each

regression. As a result the number of countries included varies across the four

models estimated. Estimates of “initial inequality” are taken from version 2.b of

the World Income Inequality Database (WIID) which is an updated version of the

Deininger and Squire (1996) database. Since it is not always possible to get the

relevant estimate of “initial inequality” for all of the countries in the sample, the

nearest possible consistently measured estimate of initial inequality for some of

the countries was used. The variable HD represents the difference between the

Human Development Index (HDI) and the Human Poverty Index (HPI-1) as

measured by the Human Development Report 2006. For some countries in the

data set the HPI-1 measure is not available; the report uses and alternative

measure (HPI-2) for the OECD, Central and Eastern Europe and Commonwealth

of Independent States (CIS) countries. In order to construct a consistent measure

of our HD variable the HPI-1 was computed for these countries based on the

formula presented on page 342 of the report.


186


187



Households adopt Technology B iff indirect utility of Technology B is greater that

indirect utility of Technology A. This implies

)ln()ln()ln()ln(

)ln()ln()ln()ln(

),,,(),,,(

12111

12111

111111

a

it

a

it

a

it

a

it

b

it

b

it

b

it

b

it

a

it

a

it

a

it

a

it

ab

it

b

it

b

it

b

it

b

sxcc

sxcc

sxccUsxccU

Recall that, here 11)( t

a

it AKs and 11)( t

a

it BKs

Also re-write 7-10 in terms )( a

itc of and equations 11-14 in terms of )( b

itc . Then

substitute all of these to above inequality. After simplifying we can write,

)1(

)1(ln.ln).ln()ln(

)1(

)1(ln.ln).ln()ln(

1121111

1121111

a

a

ita

a

ita

a

it

a

it

b

b

itb

b

itb

b

it

b

it

cccc

cccc

We can further simply the above inequality condition as follows.

2

1

2

1

)1(

)1(ln.ln)ln()ln(

)1(

)1(ln.ln)ln()ln(

11111

11111

a

a

ita

a

ita

a

it

a

it

b

b

itb

b

itb

b

it

b

it

cccc

cccc

2

121

2

121

)1(

)1(ln)ln()ln()ln(

)1(

)1(ln)ln()ln()ln(

11111

)1(

11111

)1(

aaa

a

it

bbb

b

it

c

c

2

12121

2

12121

)1(

)1(ln)ln()ln()ln(

)1(

)1(ln)ln()ln()ln(

1

1

)1(

1

1

)1(

a

a

it

b

b

it

c

c


188

21211

21212121

1

1

11

11)ln()ln()ln()ln(

b

a

a

it

b

it

ba

a

it

b

it

c

c

cc

Substituting for b

itc , and a

itc , as in the Chapter 4, we can define threshold level of

resources required for a households to adopt more productive technology )( *

itW as

follows.

1

21

1

1

1

2

2

))1(())(1(

)())1(())(1(

b

a

titb

tttita

WWw

WWWw

Simplifying the above and we can define threshold level of resources required for

a household to adopt more productive technology )( *

itW as follows.

;)1)((

))1)(((

12

221* wW

W t

it where 1 and 2 are defined as in

Chapter 4.


189

Appendix 4.2. Proof of Proposition 4.2

For agents *

itit WW , derive utility from using Technology A. Therefore their

indirect utility function (given in equation 1) can be re-written in terms of

proportion ( ) as follows by substitution of optimal plans for consumption,

bequests and capital accumulation.

a

a

aa

a

aa

a

aa

a

A CA

CCCIUF2

112

2

1

11

2

1

2 )1(

)1(lnlnln

1ln)(

where ))1(())(1( tit

a WWwC ;

Simplifying this, we can rewrite as follows.

a

a

a

a

a

a

a

aaaaA

ACCCCIUF

2

11

2

1

1

2

1

2 )1(

)1(1lnlnlnlnln)(

21

a

a

a

a

a

a

a

aA

ACIUF

2

11

2

1

1

2

1

2

1

)1(

)1(1lnln)(

21

Substitute Ca with the ))1(())(1( tit WWw and differentiate IUF

A with

respect to ( );

a

a

a

a

a

a

a

tit

A

AWWw

IUF

2

11

2

1

1

2

1

2

1

)1(

)1(1ln))1(())(1(ln

))(( 21

After simplifying, we can write the FOC for IUFA as follows.

0)())1(())(1(

)1( 21t

tit

A WWWw

IUF

Similarly for agents *

itit WW , we can derive FOC for IUFB

and we state it

below.

))('()())1(())(1(

)1( 21ttt

tttit

B WgWWWWw

IUF


190

Appendix 4.3. Analysing a vote on the tax rate (τ)

In order to look at how agents will vote for desired tax rate (τ), we look at how

their indirect utility functions ),(itU are affected by changes in (τ). That is we

look at ),('

itU of each individual. Analogous to the case in which we analyse

how agents vote over ( ), here also we end up with two possible solutions. The

first is the “corner solution” when ),('

itU is decreasing or increasing over the

entire range of )1,0( . Otherwise, we end up with an “interior solution”- a

situation in which agents prefer (0<τ<1). We summarize these outcomes below.

(i) For agents, *

itit WW , ,0),('

itU ; all agents in this group vote for

1

(ii) For agents, *

itit WW , 0),(' orU it ; the agents in this group prefer a

value of )1,0( iff 1)]('[ tW


191

Appendix 4.4. Experiments that vary altruism parameter

The experiment that we present here, allows the altruism parameters in the model

to vary with the values of θ1=θ2=0.8, and θ1=θ2=0.9. Our results indicate that

complete adoption to technology B is quicker with higher altruism parameter. As

discussed several times in this thesis, the reason for this is obvious. More altruistic

households leave larger bequest which allows their next generation to adopt the

advanced technology sooner. For the same reason high altruism parameter leads to

a higher steady state growth rate (See Figure A 42).

Figure A 45: Growth rates experienced by the median household with varying


References

192

References

193

References

Acemoglu, D. and Robinson, J.A. (2000). Political Losers as a Barrier to

Economic Development. American Economic Review Papers and

Proceedings. 90. 126-130.

Acemoglu, D. and Robinson, J.A. (2002a). The Political Economy of the Kuznets

Curve. Review of Development Economics. 6 (2). 183-203.

Acemoglu, D. and Robinson, J.A. (2002b). Economic Backwardness in Political

Perspective. National Bureau of Economic Research. NBER Working Papers

No. 8831.

Acemoglu, D. and Robinson, J.A. (2006). Economic Backwardness in Political

Perspective. American Political Science Review. 100. 115-131.

Acemoglu, D. (2008). Introduction to Modern Economic Growth. Princeton

University Press, Princeton, New Jersey, USA.

Acemoglu, D. and Guerrieri, V. (2008). Capital Deepening and Nonbalanced

Economic Growth. Journal of Political Economy. 116 (3).467-498.

Aghion, P. and Bolton, P. (1992). Distribution and Growth in Models of Imperfect

Capital Markets. European Economic Review. 36. 603-611.

Aghion, P. and Bolton, P. (1997). A Theory of Trickle Down Effect and

Development. Review of Economic Studies. 64 (2). 151-172.

References

194

Aghion, P. and Howitt, P. (1992). A Model of Growth Through Creative

Destruction. Econometrica. 60 (2). 323-352.

Aghion, P., Caroli, E. and Garcia-Penalosa, C. (1999). Inequality and Economic

Growth: The Perspective of the New Growth Theories. Journal of Economic

Literature. 37. 1615-1660.

Alesina, A. and Perotti, R. (1994). The Political Economy of Growth: A Critical

Survey of the Recent Literature. The World Bank Economic Review. 8 (3).

351-371.

Alesina, A. and Perotti, R. (1996). Income Distribution, Political Instability and

Investment. European Economic Review. 40 (6). 1203-1228.

Alesina, A. and Rodrik, D. (1994). Distributive Politics and Economic Growth.

The Quarterly Journal of Economics. 109 (2). 465-490.

Archibugi, D. and Coco, A. (2005). Measuring Technological Capabilities at the

Country Level: A Survey and a Menu for Choice. Research Policy. 34 (2).

175-194.

Atkinson, T. (2003). Income Inequality in OECD Countries: Data and

Explanations. CESIFO working paper No. 881. category 10: empirical and

theoretical methods.

Azariadis, C. and Stachurski, J. (2005). Poverty Traps. In Handbook of Economic

Growth, Vol. 1B. Eds. Aghion, P. and Durlauf, S. Elsevier Publishers.

Amsterdam, The Netherlands.

References

195

Banerjee, A. and Newman, A. (1991). Risk Bearing and the Theory of Income

Distribution. Review of Economic Studies. 58. 211-235.

Bao, Y. and Guo, J. (2004). Re Examination of Economic Growth, Tax Policy and

Distributive Politics. Review of Development Economics. 8 (3). 474-482.

Barro, R. J. (1990). Government Spending in a Simple Model of Endogenous

Growth. Journal of Political Economy. 98 (S5). S103-S125.

Barro, R. and Lee, J.W. (1994a). Losers and Winners in Economic Growth. In

Proceedings of the World Bank Annual Conference on Development

Economics. World Bank. Washington D.C. 267-297.

Barro, R. and Lee, J.W. (1994b). Sources of Economic Growth. Carnegie-

Rochester Conference on Public Policy. 40. 1-46.

Barro, R.J. and Lee, J.W. (2000). International Data on Educational Attainment

Updates and Implications. National Bureau of Economic Research, NBER

Working Papers No. 7911. (web database: http://go.worldbank.org/).

Barro, R.J. (1991). Economic Growth in Cross Section of Countries. Quarterly

Journal of Economics. 106. 407-443.

Barro, R.J. and Lee, J. (2000). International Data on Educational Attainment:

Updates and Implications. Manuscript and research datasets. Centre for

International Development. Harvard University.

Barro, R.J. and Sala-i-Martin, X. (1992). Convergence. Journal of Political

Economics. 100 (2). 223-251.

References

196

Basu, S. and Weil, D.N. (1998). Appropriate Technology and Growth. Quarterly

Journal of Economics. 113 (4). 1025-1054.

Baumol, W. (1986). Productivity Growth, Convergence, and Welfare: What the

Long-Run Data Show? American Economic Review. 76 (5). 1072-1085.

Bellettini, G. and Ottaviano, G.I.P. (2005). Special Interests and Technological

Change. Review of Economic Studies. 72. 43-56.

Benabou, R. (1996). Inequality and Growth, In NBER Macroeconomic Annual

1996 (Eds. Bernakne B.S. and Rotemberg, J.J.). 11-74. MIT Press,

Cambridge.

Bernard, A.B. and Jones, C.I. (1994). Comparing Apples to Oranges:

Productivity, Convergence and Measurement Across Industries and Countries.

American Economic Review. 86 (5). 1216-1238.

Bernard, A.B. and Jones, C.I. (1996). Technology and Convergence. Economic

Journal. 106. 1037-1044.

Bertola, G. (1993). Market Structure and Income Distribution in Endogenous

Growth Models. American Economic Review. 83. 1184-1199.

Besley, T. and Burgess, R. (2000). Land Reform, Poverty Reduction, and Growth:

Evidence from India. Quarterly Journal of Economics. 115 (2). 389-430.

Besley, T. and Coate, S. (2003). On the Public Choice Critique of Welfare

Economics. Public Choice. 114. 253-273.

References

197

Bessen, J. (2002). Technology Adoption Costs and Productivity Growth: The

Transition to Information Technology. Review of Economic Dynamics. 5. 443-

469.

Borissov, K. and Lambrecht, S. (2009). Growth and Distribution in an AK-Model

with Endogenous Impatience. Economic Theory. 39(1). 93-112.

Borjas, G.J. (1992). Ethnic Capital and Intergenerational Mobility. Quarterly

Journal of Economics. 108. 619-652.

Bosworth, B. and Collins, S.M. (2008). Accounting for Growth: Comparing China

and India. Journal of Economic Perspectives. 22 (1). 45-66.

Bourguignon, F. (1981). Pareto Superiority of Unegalitarian Equilibria in Stiglitz‟

Model of Wealth Distribution with Convex Saving Function. Econometrica.

49. 1469-1475.

Bourguignon, F. (1990). Growth and Inequality in the Dual Model of

Development: The Role of Demand Factors. Review of Economic Studies. 57.

215-228.

Bourguignon, F. and Morrisson, C. (2002). Inequality among World Citizens:

1820-1992. American Economic Review. 92 (4). 727-744.

Bridgeman, B., Igor R., Livshits, D. and MacGee. J.C. (2007). Vested Interests

and Technology Adoption. Journal of Monetary Economics. 54. 649-666.

References

198

Canton, E.J.F., Groot, H.L.F., Nahuis, R. (2002). Vested Interests, Population

Ageing and Technology Adoption. European Journal of Political Economics.

18. 631-652.

Caselli, F. (1999). Technological Revolutions. American Economic Review. 89

(1). 78-102.

Caselli, F. and Coleman, W.J. (2001). Cross-Country Technology Diffusion: the

Case of Computers. American Economic Review. 91 (2). 328-335.

Caselli, F., Esquival, G. Lefort, F. (1996). Reopening the Convergence Debate: A

New Look at Cross-Country Growth Empirics. Journal of Economic Growth.

1 (3). 363-369.

Cass, D. (1965). Optimum Growth in and Aggregative Model of Capital

Accumulation. Review of Economic Studies. 32. 233-240.

Chongvilaivan, A. (2008). Learning by Exporting and High-tech Capital

Deepening in Singapore Manufacturing Industries: 1974-2006. National

University of Singapore, Department of Economics, SCAPE Policy Research

Working Paper No. 804.

Clark, C. (1940). The Conditions of Economic Progress. MacMillan, London.

Coe, D.T., Helpman, E. (1995). International R & D Spillovers. European

Economic Review. 39. 859-887.

References

199

Coe D.T., Helpman, E., Hoffmaister, A.W. (1997). North-South R & D

Spillovers. The Economic Journal. 107 (440). 134-149.

Comin, D. and Hobijin, B. (2004). Cross Country Technology Adoption: Making

the Theories Face the Facts. Journal of Monetary Economics. 51. 39-83.

Deininger, K. and Squire, L. (1996). A New Dataset Measuring Income

Inequality. The World Bank Economic Review. 10 (3). 565-591.

Djankov, S., La Porta, R., Lopez-De-Silanes, F. and Shleifer, A. (2002). The

Regulation of Entry. Quarterly Journal of Economics. 65 (1). 1-37.

Easterly, W., King, R., Levine, R. and Rebelo, S. (1994). Policy, Technology

Adoption and Growth. National Bureau of Economic Research. Working

Paper No. 4681.

Ejarque, J and Reis, A.B. (2003) Lessons from Taking an AK Model to the Data.

University of Copenhagen. Department of Economics. Discussion Papers

series no 03-37.

Ejarque, J. and Reis, A.B. (2005). Relative Price: Lessons from Taking an AK

Model to the Data. Society for Economic Dynamics. Meeting Papers series

no 312.

Fan, S. (2002). Agricultural Research and Urban Poverty in India. Environment

and Production Technology Division Discussion Paper no 94, International

Food Policy Research Institute.

References

200

Figini, P. (1999). Inequality and Growth Revisited. Dept. of Economics, Trinity

College. Trinity Economic Paper Series, Paper No. 99 /2, Dublin 2, Ireland.

Forbes, K.J. (2000). A Reassessment of the Relationship Between Inequality and

Growth. American Economic Review. 90 (4). 869-887.

Galor, O. (1996). Convergence? Inferences from Theoretical Models. Economic

Journal. 106 (437). 1056-1069.

Galor, O. (2005). From Stagnation to Growth: Unified Growth Theory. In

Handbook of Economic Growth, Vol. 1B. Eds. Aghion, P. and Durlauf, S.

Elsevier Publishers. Amsterdam, The Netherlands.

Galor, O. and Moav, O. (2000). Ability-Bias Technological Transition, Wage

Inequality, and Economic Growth. Quarterly Journal of Economics. 115 (2).

469-497.

Galor, O. and Tsiddon, D. (1997). Technological Progress, Mobility and

Economic Growth. American Economic Review. 87 (3). 363-382.

Galor, O. and Weil, D.N. (2000). Population, Technology and Growth: From

Malthusian Stagnation to the Demographic Transition and Beyond. American

Economic Review. 90 (4). 806-828.

Galor, O. and Zeira, J. (1993). Income Distribution and Macroeconomics. Review

of Economic Studies. 60. 35-52.

References

201

Glomm, G. and Ravikumar, B. (1992). Public Versus Private Investment in

Human Capital: Endogenous Growth and Income Inequality. Journal of

Political Economy. 100 (4) 818- 834.

Gollin, D., Parente S. and Rogerson, R. (2002). The Role of Agriculture in

Development. American Economic Review. 92(2). 160-164.

Gollin, D., Parente, S., Rogerson, R. (2004). Farm work, Home Work and

International Productivity Differences. Review of Economic Dynamics. 7. 827-

850.

Greenwood, J. and Yorukoglu, M. (1997). 1974. Carnegie-Rochester Conference

Series on Public Policy. 46. 49-95.

Greenwood, J. and Jovanovic, B. (1990). Financial Development, Growth and the

Distribution of Income. Journal of Political Economy. 98 (5). 1076-1106.

Grossman, G.M. and Helpman, E. (1991). Innovation and Growth in the Global

Economy. MIT Press. Cambridge, MA.

Hansen, G. and Prescott, E. (2002). Malthus to Solow. American Economic

Review. 92. 1205-1217.

Holmes, T.J. and Schmitz, J.A. (1995). Resistance to New Technology and Trade

Between Areas. Federal Reserve Bank of Minneapolis Quarterly Review. 19

(1). 2-16.

References

202

Hornstein, A. and Krusell, P. (1996). Can Technology Improvements Cause

Productivity Slowdown? In NBER Macroeconomics Annual. Eds. Bernanke,

B. and Rotemberg, J. MIT press, Cambridge, MA.

Iwaisako, T. (2002). Technology Choice and Pattern of Growth in an Overlapping

Generations Model. Journal of Macroeconomics. 24. 211-231.

Jappelli, T. and Pistaferri, L. (2000). The Dynamics of Household Wealth

Accumulation in Italy. Fiscal Studies. 21 (2). 269-295.

Jaumotte, F., Poirson, H., Spatafora, N., Vu, K., de Guzman, C., and Hettinger, P.

(2006) Asia rising: Patterns of Economic Development and Growth, in World

Economic Outlook. International Monetary Fund ,Washington, D.C.

Jones, C.I. (1995). Time Series Tests of Endogenous Growth Models. The

Quarterly Journal of Economics. 110 (2). 495-525.

Jones, R. (1979). International Trade: Essays in Theory, Amsterdam: North

Holland.

Josten, S.D and Truger, A. (2003). Inequality, Politics, and Economic Growth.

Three Critical Questions on Politico-Economic Models on Growth and

Distribution. Helmut Schmidt University, Working Paper No. 3/2003.

Hamburg.

Jovanovich, B. and Rousseau, P. (2005). General Purpose Technologies. In

Handbook of Economic Growth, Vol. 1B. Eds. Aghion, P. and Durlauf, S.


References

203

Kaldor, N. (1957). A Model of Economic Growth. The Economic Journal. 67

(268). 591-624.

Khan, A. and Ravikumar, B. (2002). Costly Technology Adoption and Capital

Accumulation. Review of Economic Dynamics. 5. 489-502.

Kongsamut, P., Rebelo, S., Xie, D. (2001). Beyond Balanced Growth. Review of

Economic Studies. 68. 869-882.

Koopmans, T.C. (1965). On the Concept of Optimal Economic Growth. In The

Econometric Approach to Development Planning. Amersterdam, North

Holland.

Krusell, P. and Rios-Rull, J.V. (1996). Vested Interests in a Positive Theory of

Stagnation and Growth. Review of Economic Studies. 63. 301-329.

Krusell, P. and Rios-Rull, J.V. (2002). Politico-Economic Transition. Review of

Economic Design. 7. 309-329.

Krusell, P., Quadrini, V. and Rios-Rull, J.V. (1997). Politico-Economic

Equilibrium and Economic Growth. Journal of Economic Dynamics and

Control. 21. 243-272.

Kuznets, S. (1955). Economic Growth and Income Inequality. American

Economic Review. 65. 1-28.

Kuznets, S. (1957). Quantitative Aspects of the Economic Growth of Nations.

Economic Development and Cultural Change Supplement. 5 (4). 3-111.

References

204

Kuznets, S. (1963). Quantitative Aspects of the Economic Growth of Nations:

Distribution of Income by Size. Economic Development and Cultural Change.

11 (1).

Kuznets, S. (1968). Towards a Theory of Economic Growth. Yale University

Press, New Haven.

Lahiri, R. and Magnani, E. (2008). On Inequality and the Allocation of Public

Spending. Economics Bulletin. 8. 1-8.

Laitner, J. (2000). Structural Change and Economic Growth. Review of Economic

Studies. 67. 545-561.

Leung, C.K.Y. and Tse, C.Y. (2001). Technology Choice and Saving in the

Presence of a Fixed Adoption Cost. Review of Development Economics. 5 (1).

40-48.

Li, H. and Zou, H. (1998). Income Inequality is not Harmful for Growth: Theory

and Evidence. Review of Development Economics. 2 (3). 318-334.

Lindert, P.H. (1994). The Rise in Social Spending 1880-1930. Exploration of

Economic History. 31 (1). 1-37.

Lindert, P.H. (1996). What Limits Social Spending? Exploration of Economic

History. 33 (1). 1-34.

Lucas, R. (1986). Principles of Fiscal and Monetary Policy. Journal of Monetary

Economics. 17 (1). 117-134.

Lucas, R. (1993). Making a Miracle. Econometrica, 61, 251-272.

References

205

Maddison, A. (1995). Monitoring the World Economy: 1820-1992. Organization

for Economic Cooperation and Development (OECD). Washington, D.C.

Maddison, A. (2005). Growth and Interaction in the World Economy: The Roots

of Modernity. The AEI Press. Washington, D.C.

Maddison, A. (2009). Historical Statistics of the World Economy: 1-2006AD.

(http://www.ggdc.net/maddison/ last updated 2009 march).

Malthus, T.R. (1798). An Essay on the Principle of Population. St. Paul‟s Church,

London.

Mankiw, N.G., Romer, D. and Weil, D.N. (1992). A Contribution to the Empirics

of Economic Growth. Quarterly Journal of Economics. 107 (2). 407-437.

Marjit, S. (1988). A Simple Model of Technology Transfer. Economics Letters.

26. 63–67.

Marjit, S. (1992). Technology and Obsolescence: A Reinterpretation of the

Specific-Factor Model of Production. Journal of Economics. 55.193–207.

Marjit, S. (1994). A Competitive General Equilibrium Model of Technology

Transfer, Innovation, and Obsolescence. Journal of Economics. 59. 133–148.

Massey, D.S. (2009). Globalization and Inequality: Explaining American

Exceptionalism. European Sociological Review. 25 (1). 9-23.

References

206

McGrattan, E.R. (1998). A Defense of AK Growth Models. Federal Reserve Bank

of Minneapolis Quarterly Review. 22 (4) 13–27.

Mokyr, J. (1990). The Lever of Riches: Technological Creativity and Economic

Progress. Oxford University Press. New York.

Mokyr, J. (1993). The New Economic History and the Industrial Revolution. In

The British Industrial Revolution: An Economic Perspective. Eds. Mokyr, J.

Westview Press, Boulder.

Mokyr, J. (2005). Long Term Economic Growth and the History of Technology.

In Handbook of Economic Growth. Vol. 1B. Eds. Aghion, P. and Durlauf, S.


Ngai, L.R. (2004). Barriers and the Transition to Modern Growth. Journal of

Monetary Economics. 51 (7). 1353-1383.

Nurkse, R. (1952). Problems of Capital Formation in Underdeveloped Countries.

Oxford University Press. New York.

Owen, A.L. and Weil D.N. (1997). Intergenerational Earnings Mobility,

Inequality and Growth. National Bureau of Economic Research. Working

paper No. 6070.

Parente, S. and Prescott, E. (1994). Barriers to Technology Adoption and

Development. Journal of Political Economy. 102 (2). 298-301.

References

207

Parente, S. and Prescott, E. (2004). A Unified Theory of the Evolution of

International Income Levels. Federal Reserve Bank of Minneapolis, Research

Department Staff Report-333.

Park, D. (2001). Recent Trends in the Global Distribution of Income. Journal of

Policy Modeling. 23 (5).

Perotti, R. (1992). Income Distribution, Politics and Growth. American Economic

Review. 82 (2). 311-317.

Perotti, R. (1993). Political Equilibrium, Income Distribution, and Growth.

Review of Economic Studies. 60 (4). 755-776.

Perotti, R. (1996). Growth, Income Distribution and Democracy: What the Data

Say? Journal of Economic Growth. 1 (2). 149-187.

Persson, T. and Tabellini, G. (1994). Is Inequality Harmful for Growth? American

Economic Review. 84 (3). 600-621.

Piketty, T. and Saez, E. (2003). Income Inequality in the United States. Quarterly

Journal of Economics. 118 (1). 1-39.

Pritchet, L. (2000). Understanding Patterns of Economic Growth: Searching for

Hills among Plateaus, Mountains, and Plains. The World Bank Economic

Review. 14. 221-250.

Pritchett, L. (1997). Divergence, Big Time. Journal of Economic Perspectives. 11

(3). 3-17.

References

208

Quah, D.T. (1996). Twin Peaks: Growth and Convergence in Models of

Distribution Dynamics. Economic Journal. 106 (437). 1045-1055.

Quah, D.T. (1997). Empirics for Economic Growth and Distribution: Polarization,

Stratification and Convergence Clubs. Journal of Economic Growth. 2 (1). 27-

59.

Rebelo, S. (1991). Long Run Policy Analysis and Long Run Growth. Journal of

Political Economy. 99. 500-521.

Romer, P. (1986). Increasing Returns and Long Run Growth. Journal of Political

Economy. 94. 1002-1037.

Romer, P.M. (1987). Crazy explanations for the productivity slowdown. NBER

Macroeconomics Annual (Eds. Fisher, S.) 2. 163-210. MIT Press, Cambridge

MA.

Romer, P.M. (1990). Endogenous Technological Change. Journal of Political

Economy. 98 (5). S71-S102.

Romero-Avila, D. (2006). Can the AK Model Be Rescued? New Evidence from

Unit Root Tests with Good Size and Power. B.E. Journal of Macroeconomics

-topics. 6(1). article 3.

Rosenberg, N. and Birdzell, L.E. (1986). How the West Grew Rich: The

Economic Transformation of the Industrial World. Basic Books. New York.

References

209

Saint-Paul, G. and Verdier, T. (1993). Education, Democracy and Growth.

Journal of Development Economics. 42. 399-407.

Sala-i-Martin, X. (2006). The World Distribution of Income: Falling Poverty and

… Convergence, Period. Quarterly Journal of Economics. 121 (2). 351-397.

Schluter, C. (1998). Income Dynamics in Germany, the USA and the UK:

Evidence from Panel Data. Centre for Analysis of Social Exclusion – CASE

papers.

Solow, R.M. (1956). A Contribution to the Theory of Economic Growth.

Quarterly Journal of Economics. 70. 65-94.

World Bank. (2005). World Development Report 2006: Equity and Development.

The World Bank. Washington D.C.

World Bank. (2007).World Development Indicators (WDI) online database. The

World Bank. Washington D.C

World Bank. (2008). Global Economic Prospects: Technology Diffusion in the

Developing World. The World Bank. Washington D.C.

Tomes, N. (1981). The Family, Inheritance and Intergenerational Transmission of

Inequality. The Journal of Political Economy. 89 (5). 928-957.

United Nations Development Program (2006). Human Development Report 2006.

Beyond Scarcity: Power, Poverty and the Water Crisis.

United Nations University World Institute for Development Economics Research

(WIDER) (2008).(Web database: http://www.wider.unu.edu/

research/Database/ en GB/database/) last updated May, 2008.

References

210

Wan, G. Lu, M. Chen, Z. (2006). The inequality–growth nexus in the short and

long run: Empirical evidence from China. Journal of Comparative

Economics. 34(4). 654-667.

Wirz, N. (2008). Assessing the Role of Technology Adoption in China's Growth

Performance. University of Copenhagen. Department of Economics.

Economic Policy Research Unit (EPRU). Working Paper Series No 6.

Young, A. (1991). Learning by Doing and the Dynamic Effects of International

Trade. The Quarterly Journal of Economics. 106 (2). 369-405.

Zartaloudis, S. (2007). Equality: A Political Choice-Income Inequality in Europe

and the US: Trends, Causes and Solutions, Policy Network Paper, Policy

Network. United Kingdom.

Zweimuller, J. (2000). Inequality, Redistribution, and Economic Growth?

Empirica. 27. 1–20.

Documents

Economic Growth, Inequality and Technology Adoption in ...at various stages of my PhD journey. In particular, I appreciate the suggestions he made before my confirmation of candidature