5 The consequences of rapid population growth

This chapter shows that rapid populationgrowthat rates above 2 percent, common in mostdeveloping countries todayacts as a brake ondevelopment. Up to a point, population growthcan be accommodated: in the past three decadesmany countries have managed to raise averageincome even as their populations grew rapidly. Inthat strict sense, population growth has beenaccommodated. But the goal of developmentextends beyond accommodation of an ever largerpopulation; it is to improve people's lives. Rapidpopulation growth in developing countries hasresulted in less progress than might have beenlost opportunities for raising living standards, par-ticularly among the large numbers of the world'spoor.

The conclusion that rapid population growth hasslowed development is by no means straightfor-ward or clearcut (see Box 5.1). Under certain condi-tions moderate population growth can be benefi-cial. As Chapter 4 showed, in Europe, Japan, andNorth America economic growth has been accom-panied by moderate population growth, whichmay have stimulated demand, encouraged techno-logical innovation, and reduced investment risks.Moderate labor force growth, combined with extraspending on education, can also mean continuousupgrading of the labor force with better educatedworkers. In sparsely populated countries, fasterpopulation growth shortens the time required toreach the population size that provides economiesof scale in transport, communications, social ser-vices, and production. Some developing countriescould benefit from such economies of scale, espe-cially in rural areas. And a big population canincrease a country's economic as well as politicaland military power; in a world of economic andpolitical uncertainty, countries such as India andChina can seem to benefit from the sheer size oftheir domestic markets.

But these benefits derive from a moderateincrease in population. Most developing countries

are experiencing growth that, by historical stand-ards, is faster than that. Even in uncrowded coun-tries, the long-term benefits of having more peoplemust be weighed against the immediate costs ofcoping with rapid growth. In those few countrieslacking the people to exploit their natural re-sources, immigration from neighboring countries,if politically feasible, would be less costly and moreeffective than a fast natural rate of populationgrowth. And the economic success of many smallcountriesDenmark, Hong Kong, Singapore, andSwitzerlandshows that urbanization and tradeprovide other means to achieve the scale econo-mies of a large population.

There are several reasons why populationgrowth in developing countries is today a greatereconomic burden than it once was in today'sdeveloped countries:

Population growth is now much more rapid.As Chapter 4 showed, in industrializing Europe itseldom exceeded 1.5 percent a year, comparedwith the 2 to 4 percent that most developing coun-tries have averaged since World War II.

Unlike nineteenth century Europe, large-scaleemigration from today's developing countries isnot possible.

Compared with Europe, Japan, and NorthAmerica in their periods of fastest populationgrowth, income in developing countries is stilllow, human and physical capital are less built up,and in some countries political and social institu-tions are less well established.

Many developing countries whose economiesare still largely dependent on agriculture can nolonger draw on large tracts of unused land.

This chapter begins by emphasizing that theimplications of population growth differ consider-ably among countries, depending on their currentsocial, economic, and political conditions. Next itreviews how rapid population growth affects theeconomy as a whole through savings and invest-ment. It then considers the experience of countries

79

in coping with rapidly growing populationstheirefforts to achieve food security, the effects on theirnatural resources, the pressures of internal migra-tion and urban growth, and the options that theinternational economy provides. Throughout this

80

Box 5.1 Consequences of population growth: conflicting viewsThe traditional Malthusian concern isthat population growth will sooner orlater run up against the limits of theearth's finite stock of resources. In hisFirst Essay on Population, Maithus arguedthat the inherent capacity of populationto grow exceeds the earth s capacity toyield increases in food, because of limitsto the supply of cultivable land. Unre-strained population growth eventuallyleads to falling wages and rising foodprices because, as the labor forceexpands, a rising ratio of labor to landleads to smaller and smaller incrementsin output per worker. Population growthis ultimately checked by rising mortality.

In the twentieth century this argumenthas been extended to the availability ofenergy and minerals, the effects of risingenvironmental pollution, and so on. InThe Limits to Growth, Club of Romeresearchers built a simulation model onthe assumption that the pace of techno-logical change would be insufficient toovercome diminishing returns arisingfrom limited supplies of essentialresources. Falling standards of living andincreasing levels of pollution would leadto a population collapse within 100 years.

A related view is that some resourcesland, forests, fisheries.though fixed, arerenewable, but that their sustainableyields do have a maximum limit. Someharvests may exceed this maximum, butthey lead to a permanent reduction in thelong-run productivity of land. A popula-tion whose needs (subsistence and com-mercial) exceed sustainable yields willhave lower per capita incomes in thelong run.

The claim of diminishing returns toresources can easily be criticized for itsfailure to recognize that, as resources aredepleted, rising prices reduce consump-tion and speed the search for substitutes,stimulating technological change. Thiscriticism, extended, leads to the argu-ment that there are no real natural

resource limits, because populationgrowth itself brings the adjustments thatcontinually put off doomsday. To quotefrom Julian Simon's book, The UltimateResource: "The ultimate resource is peo-

pleskilled, spirited, and hopeful peo-plewho will exert their wills and imagi-nations for their own benefit and so,inevitably, for the benefit of us all."Simon argues that natural resources arenot limited; that scarcity is revealed byprices; and that prices of resources arenot rising, at least not as a proportion ofthe income of the United States. Morepeople implies more ideas, more creativetalent, more skills, and thus better tech-nology; in the long run populationgrowth is not a problem but anopportunity.

These different viewpoints each con-tain important truths. Some resourcesare finite; even if prices have notincreased (and they may have done so inrelation to incomes outside the UnitedStates), there have been fundamentalstructural changes in the balancebetween population and resources.Human ingenuity might be a match forthese changes, but it might be able onlyto maintain income, not to lift millions ofpeople out of poverty. Or it may reducepoverty very slowly: even with theassumption of technological change builtinto Simon's model, there are "short-run" difficulties. His short run is thirtyto eighty years, and in that period hefinds even moderate population growthto be detrimental to human welfare. Inthe short run, ideas may be lost andEinsteins go undiscovered if many chil-dren receive little schooling. Policy-makers and poor people live in the shortrun; they do not wish to go through aperiod of greater deprivation to adapteventually to rapid population growth.

At the same time, there is little doubtthat the key to economic growth is peo-ple, and through people the advance of

human knowledge. Per capita measuresof income should not be used to implythat the denominator, people, contrib-utes nothing to the numerator, totalincome. Nor is population growth in andof itself the main cause of naturalresource problemsair pollution, soil

degradation, even food availability.This Report therefore takes a position

that is neither hopeless nor overly opti-mistic. The difficulties caused by rapidpopulation growth are not primarily dueto finite natural resources, at least not forthe world as a whole. But neither doesrapid population growth itself automati-cally trigger technological advance andadaptation. If anything, rapid growthslows the accumulation of skills thatencourage technological advance, andinsofar as there are diminishing returnsto land and capital, is likely to exacerbateincome inequalities. This is most obviousat the family level, where high fertilitycan contribute to a poor start in life forchildren. But it is also true for countriesas a whole.

Moreover, the costs of rapid popula-tion growth differ greatly from countryto country. Those differences are notconfined to differences in naturalresources. In countries heavily reliant onagriculture, a scarcity of naturalresources does matter. But the underly-ing problem is low income and low levelsof education, which are sources of rapidpopulation growth and simultaneouslymake the required adjustments to it moredifficult. Much of the world's populationlives without the benefit of clear signalsto encourage smaller families; yet theseare the families and the nations in theworst position to make the adaptiveresponses that rapid population growthrequires. That is why rapid populationgrowth is, above all, a developmentproblem.

discussion of the effects of population growth oncountries, this chapter will touch on a theme intro-duced in Chapter 4: the implications of high fertil-ity for poor people and for income inequality.Because the poor are usually last in line for jobs,

school places, and public health services, they aremore likely to be penalized by rapid populationgrowth.

The chapter does not treat a reduction in the rateof population growth as a panacea for develop-ment; macroeconomic and sectoral policies matterat least as much. But it does show that within mostcountries, for any given amount of resources, aslower rate of population growth would help topromote economic and social development.

Differences among countries

The implications of population growth differ con-siderably among developing countries. Countrieswhere education levels are already high, wheremuch investment in transport and communica-tions systems is in place, and where political andeconomic systems are relatively stable, are wellequipped to cope with rapid population growth.This is true whether or not their natural resourcesare limited or their countries already "crowded,"as in the fast-growing East Asian economies suchas Hong Kong, Korea, Singapore, and morerecently Malaysia and Thailand. But these tendalso to be countries where population growth isnow slowing.

Countries with untapped natural resourcescould in the long run support more people. Butrapid population growth makes it hard for them todevelop the human skills and administrative struc-tures that are needed to exploit their resources. InBrazil, Ivory Coast, and Zaire, for example, thedevelopment of unused land will require largecomplementary investments in roads, public ser-vices, and drainage and other agricultural infra-structure. Natural resources are not by themselvessufficient (or even necessary) for sustained eco-nomic growth.

Where the amount of new land or other exploit-able resources is limitedas in Bangladesh,Burundi, China, Egypt, India, Java in Indonesia,Kenya, Malawi, Nepal, and Rwandathe short-run difficulties are more obvious. In some areascrop yields are still relatively low, leaving room forrapid growth in agricultural production; in others,the expansion of manufacturing industry couldprovide exports to pay for extra food imports. Butboth solutions require costly investments, devel-opment of new institutions, and numerous eco-nomic and social adjustmentsall easier if popula-tion is growing only slowly.

In any society, change becomes easier if technol-ogy is advancing rapidly. From one point of view,

population growth itself helps to bring about tech-nological change: in agricultural societies it mayhelp spur the development of new farming meth-ods needed to maintain per capita output. In ear-lier centuries it may even have helped provide theminimum population required to support a smallreligious or artistic elite.

But throughout the modern technological era,there is no evidence that a large or rapidly growingpopulation has itself been influential in promotingnew technology. The money and research skillsneeded for important advancesthe Green Revo-lution, for exampleare overwhelmingly in therich countries where population growth is slow. Ifanything, these advances have brought labor-saving, not labor-using, innovations. Althoughadjustment and technical progress can accompanypopulation growth, slower population growthwould permit them to raise average incomes all thefaster.

Macroeconomic effects of rapidpopulation growth

In a crude arithmetical sense, differences in popu-lation growth rates since the 1950s have helped toperpetuate international differences in per capitaincomes. Between 1955 and 1980, GNP grew atabout 4 percent a year in the low-income countries.This growth in general produced modest increasesin income per person (see Table 5.1). However, inmany of the poorest countriesBangladesh andmost of sub-Saharan Africaeconomic activityslowed considerably in the 1970s. Coupled withrapid (and in some cases, accelerating) populationgrowth, this economic slowdown resulted in stag-nating or declining per capita incomes.

In most middle-income countries GNP growthhas been much fasterbetween 5 and 6 percent ayearso that even with rapid population growth,per capita income grew by about 3 percent a year.Industrial countries achieved only sluggish GNPgrowth during the 1970s, but their low populationgrowthi percent a year or lessmeant that theirincreases in per capita income were in generalalmost as large as in the high-growth, middle-income countries. These increases came on top ofmuch higher initial incomes, so that the absolutegulf between them and the rest of the world wid-ened considerably.

The middle-income countries have shown thatrapid population growth can go hand in hand withsubstantial gains in per capita income. But thelong-run relation is more complex than that

81

implied by a simple division of total income bynumbers of people. Indeed, that simple divisionimplies, wrongly, that people are the problem.One question is how population growth affects thedistribution of income within countries, and espe-cially growth in income of poorer groups (see Box5.2). More generally, the question is whether arapid pace of population growth helps or harmseconomic growth. There are several ways popula-tion growth can affect economic growth: throughits influence on savings per person, on the amountof capital invested per person, and on the effi-ciency with which the economy operates.

TABLE 5.1

Growth of population, GNP, and GNP per capita, 1955-80

a. Includes high-income oil exporters and industrial nonmarket economies.

Population growth and private savings

A country's savings are generated by households,businesses, and the public sector. Corporate andgovernment savings do not seem to be related inany systematic way to variations in populationgrowth; governments can, within limits, use fiscaland monetary measures to change a country's sav-ings rate, irrespective of demographic conditions.Theory suggests, however, that household sav-ings-usually the largest component of domesticsavings-should be reduced by the high depen-dency burdens associated with rapid populationgrowth. At any given level of output per worker,greater numbers of dependents cause consump-tion to rise, so savings per capita should fall.

Recent empirical studies find only minor supportfor this view. But many factors account for theweak link between dependency burdens and sav-ings in developing countries; all point to the prob-ability that high fertility is indeed a burden,

82

though its effects on monetized savings are small.First, the bulk of monetized household savings indeveloping countries is produced by relatively fewwealthy families. They tend to have few children,so their savings are little affected by the burden oftheir dependents. In contrast, the majority of fami-lies are poor and save little. Parents have no choicebut to pay for what their children consume byreducing their own consumption or by "dissav-ing"-for example, by farming their land moreintensively than can be sustained in the long run.If parents have more children than they want,their ability to make best use of the resources they

do have is harmed. Whether they are restrictinginvestment in their farm, or in their children's edu-cation, or in security for their old age, their highfertility contributes to their poverty. But even withfewer children, poor parents might not increasetheir savings. Instead, they might simply consumea bit more themselves.

A second reason for the apparently weak linkbetween savings and dependency burdens is thatbanking and credit systems are not well estab-lished in developing countries. Poor families (andeven the not-so-poor) are unlikely to have financialsavings that show up in national accounts; theyare more likely to "save" by accumulating land,tools, or other assets. Even if families wanted tosave in good times (say, before children are born orafter children are old enough to work) and borrowin difficult times, they probably could not withoutpaying a steep price in terms of low real interestrates for saving and high rates for borrowing. Athird reason, as explained in Chapter 4, is that

(average annual percentage change)

Population GNP GNP per capita

Country group 1 955-70 1970-80 1955-70 1970-80 1955-70 1970-80

All developing countries 2.2 2.2 5.4 5.3 3.1 3.1Low-income 2.1 2.1 3.7 4.5 1.6 2.4

China 2.0 1.8 3.3 6.0 1.3 4.1India 2.2 2.1 4.0 3.4 1.8 1.3

Other 2.4 2.7 4.4 2.7 2.0 0.0Middle-income 2.4 2.4 6.0 5.6 3.5 3.1

Industrial market economies 1.1 0.8 4.7 3.2 3.6 2.4Europe 0.7 0.2 4.8 2.6 4.1 2.4Japan 1.0 1.1 10.3 5.4 9.2 4.2United States 1.4 1.0 3.4 3.1 2.0 2.1

World' 1.9 1.9 5.1 3.8 3.1 1.9

Box 5.2 Prospects for poverty and population growth, 1980-2000I-low would a faster decline in popula-tion growth affect the number of poorpeople in the year 2000? Many other eco-

nomic, political, and social factors, inaddition to population, will influencelevels of poverty in the next fifteen years.

But some simple assumptions allow illus-trative estimates. In a World Bank study,the poor were defined as those withannual per capita incomes below $135 (inconstant 1980 dollars). Based on theexperience of many countries, projectedincome growth in each of forty countries(comprising 80 percent of the populationof developing countries) was used tocompute the change in income for thepoorest groups. The findings were com-bined with World Bank country projec-tions of population growth to simulatefuture shifts in income distribution.

The exercise showed that the predictedshare of income going to the poorest 40percent would hardly change, from 14percent in 1980 to 15 percent in 2000. Theestimated number of poor people wouldfall, however, because of income growth.Using the population growth rate basedon a "standard" decline in fertility(described in Chapter 4), the number ofpoor in these forty countries would fallfrom 630 million in 1980 to 410 million in

2000. With a "rapid" decline in fertility,the number could be almost 100 millionfeweralthough at 321 million, it wouldstill exceed the total number of people inBangladesh, Nigeria, and Pakistan today.

Regional differences are worth empha-

sizing. Economic prospects are so limitedin sub-Saharan Africa that all projectionspoint to an increase in poverty. If fertilitydecline occurs only as in the standardprojectionand even that implies con-siderable declinethe number of peopleliving in poverty at the end of the cen-tury will still increase by nearly 70 per-cent. With a rapid fall in fertility, thenumber of poor would increase by lessthan 20 percentin the circumstances, asubstantial achievement.

In South and East Asia, exceptingChina, economic prospects are better, sothat a small reduction in the number ofpoor can be anticipated even assumingthe standard fertility decline. Povertycould be reduced by almost 40 percent,however, with a rapid fall in fertility.Grouping together Latin America, theMiddle East, and North Africawherethe poor are fewer than in Asia and inthe rest of Africarapid fertility declinecould help to reduce the number in pov-erty by 70 percent. As for China, wherefertility is already low, the number ofpoor can be expected to decline bybetween 80 and 90 percent by the year2000.

The exercise probably understates theeffect of rapid fertility decline in reducing

poverty. A faster reduction in fertility islikely to be associated with a narrowingof differences in educational investmentby socioeconomic class, and an increasein wages in relation to rents and profits.These imply a more equal distribution of

income than assumed in the projections,and a more rapid elimination of poverty.

Estimated number of poor in the year 2000under different fertility assumptions, byregion

Sub-Saharan Africa

Index (1980= 100)

150Standard fertility decline

100 Rapid fertility decline

1980 2000

Middle East, North Africa,Latin America

100

50

1980

Asia (excluding China)100

50

"Standard andrapid fertility decline

1980 2000

For explanation of assumptions of populationprojections, see Population Data Supplement.

s;standard fertility decline

Rapid fertility decline2000

Standard fertility decline

50 * Rapid fertility decline

1980 2000

China100

poor people may see children themselves as a wayof "saving" for old age.

These reasons explain why high dependencyburdens reduce household savings rates in indus-trial countries but not in the developing world. Indeveloping countries, though there is no directlink from fertility to household savings, they dobecome indirectly linked as development pro-ceeds. For example, as more women work in themodern sector, family savings tend to rise and fer-tility falls; as urbanization proceeds and financialmarkets improve, monetized savings rise and fer-tility falls.

Capital widening

Although rapid population growth does not seemto influence the supply of financial savings, itclearly affects the demand for savings. To maintainincome, capital per person (including "humancapital," that is, a person's education, health, andskills) must be maintained. And as populationsgrow, "capital widening" is needed to maintaincapital per person. But slower population growthreleases investable resources for "capital-deepen-ing"that is, increasing capital per person. Ofcourse, there may be economies of scale in the pro-vision of schooling, health, and jobs in factories

83

and on farms. But the evidence on education sug-gests that capital-wideningspreading resourcesover more and more peoplecan be counterpro-ductive.

SCHOOLING REQUIREMENTS AND CAPITAL WIDEN-

ING. In industrial countries, school-age popula-tions are expected to grow slowly, if at all, over thenext two decades (see Figure 5.1). The same is trueof those developing countries, such as China,Colombia, and Korea, where fertility has alreadyfallen substantially. In Colombia, the number ofschool-age children doubled between 1950 and1970. But it increased only slightly in the 1970s,

FIGURE 5.1

Index of school-age and working-age populations,selected countries, 1950-2000

Index(1950=100)

700

600

500

400

300

200

100

Index(1950 = 100)

600

500

400

300

200

100

China

Hungary /

1950 1960 1970 1980 1990 2000

Working-agepopulation

Colombia

Korea

1950 1960 1970 1980 1990 2000

84

because fertility had started to fall in the late 1960s.This allowed enrollment rates to rise; as the chil-dren of poor parents were least likely to have beenenrolled before, the poor probably benefited mostfrom the spread of education.

For high-fertility countries, the situation couldnot be more different. Countries such as Kenyaface a doubling or tripling of their school-age pop-ulation by the end of the century. The main impli-cation is clear. More school-age children requiremore spending on education, even if the objectiveis just to maintain current enrollment rates andstandards. As most developing countries want toimprove their schools quantitatively and qualita-tively, they will have to generate more nationalsavings or curtail other investments in, for exam-ple, power and transport. If a country is unwillingor unable to make these sacrifices, spending mustbe spread over a larger group of school children (tothe detriment of the quality of education); other-wise a growing number of children have to beexcluded.

These awkward choices come after a period ofconsiderable progress. Over the past twenty years,enrollment rates have increased at the primary,secondary, and university levels in almost alldeveloping countries. (The enrollment rate is thenumber of students enrolled in schools as a per-centage of the school-age population.) In somecases, progress has been remarkable. Educationtends to spread as per capita income rises, butsome of the lowest-income countriesSri Lanka,Tanzania, Viet Namhave already achieved, or arefast approaching, universal primary education.

Such achievements have substantially raised thefiscal burden of education. For the developingcountries as a group, public spending on educa-tion increased from 2.3 percent of GNP in 1960 to3.9 percent in 1974, and from 11.7 percent to 15.1percent of government budgets. But the propor-tion of GNP allocated to education declinedslightly over the 1970s, as did the share of educa-tion spending in government budgets, especiallyin South Asia, the Middle East and North Africa,and Latin America.

The budgetary downgrading of education, coup-led with slower economic growth, has reduced thequality of education in many developing countries.One study showed that in Latin America publicspending per primary student fell by almost 45percent in real terms between 1970 and 1978. As ashare of educational budgets, spending on non-wage itemschalk, maps, textbooks, and so onfell in eight out of ten Latin American countries. In

Kenya /School-age

\\/ipopulation

Colombia

twenty-five of fifty-four developing countries sur-veyed, student-teacher ratios at primary schoolshave risen; of those twenty-five countries, seven-teen were in Africa. Increases in class size oftenmake sense as they raise the productivity of teach-ers. But in the urban areas of Malawi and Kenya,class size frequently exceeds sixty students. Com-bined with a lack of teaching materials, largeclasses make learning difficult.

Developing countries have little scope to reduceeducational quality any further. The quality gapbetween low- and high-income countries is alreadyenormous. Bolivia, El Salvador, Malawi, and theIvory Coast, for instance, spend less than $2 a yearon classroom materials for each child at primaryschool-compared with more than $300 per stu-dent in Scandinavian countries. This gap seems tobe widening. In 1960, on average, an OECD coun-try spent fourteen times more per primary schoolstudent than did any of the thirty-six countrieswith per capita incomes below $265 (1975 prices).By 1977 the ratio had risen to 50:1.

These differences in educational quality areclearly reflected in student achievement. Researchon twenty-five countries has shown that, afterapproximately the same number of years in school,schoolchildren in low- and middle-income coun-tries have learned significantly less science thanthose in industrial countries. Quality can alsomake a considerable difference within developingcountries. In a study of Brazil, Colombia, India,and Thailand, the quality of schools and teachers-measured by a large number of indicators-explained more than 80 percent of the variance instudent scores on standardized science tests. Thepoor are more likely to attend schools of lowerquality (and to leave school sooner); so rapidexpansion of school systems to accommodategrowing populations often means that the differ-ences in skills between rich and poor, though fall-ing in terms of years of schooling, are persistingbecause of school quality differences.

As lower fertility slows the growth of the school-age population, it can ease the pressures on theeducation system. In Egypt, for example, if fertilitydoes not fall, the number of children of primaryschool age would double by the end of the century.With the standard decline in fertility described inChapter 4, the number would increase by 65 per-cent; with the rapid decline, by only 20 percent.The difference between a standard decline and arapid one would be about 2 million fewer childrena year enrolled in primary schools in the years 2000to 2015. Fewer births in the early 1980s due to a

rapid decline in fertility would decrease the size ofthe age group eligible for Egypt's secondaryschools and universities starting in the late 1990s.

Less rapidly growing enrollment produces con-siderable financial savings; these can be used toimprove school quality. One projection, forMalawi up to 2015, started with the assumptionthat recurrent costs (essentially teachers' salaries)were held constant at their 1980 level of $12.50 perstudent. With unchanged fertility, the budget forprimary education would double about every fif-teen years, even if nothing were done to improvethe coverage and quality of primary schools. Thusthe education budget's share in GDP wouldincrease from 0.7 percent in 1980 to about 1 percentin 1995 if the economy of Malawi were to grow atabout 3 percent a year. The financial savings fromlower fertility would accrue slowly at first, butbuild up considerably (see Table 5.2). Costsexcluded from the projections-for instance, out-lays for teachers' training and school buildings-would also fall and thus boost these savings.

With the money saved by lower fertility, theMalawi government could afford to enroll thecountry's total school-age population in 2005 forless than it would cost to enroll 65 percent if fertil-ity did not fall. If the government chose to main-tain a 65 percent enrollment rate, its spending perpupil could be doubled in real terms by 2015 with-out increasing the share of the primary school bud-

TABLE 5.2

Malawi: projected primary-school costsunder alternative fertility and enrollmentassumptions, 1980-2015(millions of 1980 dollars)

na. Not applicable.Note: Columns 1 and 3 assume a constant enrollment rate of65 percent. Columns 2 and 4 assume the enrollment rateincreases and is 100 percent by the year 2000.a. The percentage cost savings are the same under bothassumptions regarding enrollment rates. Absolute cost savingsare greater under the assumption of universal primaryeducation by the year 2000.

85

Standardfertilitydecline

Rapidfertilitydecline

Savingwithrapid

fertilitydecline

Year (1) (2) (3) (4) (percent)a

1980 9.8 9.8 9.8 9.8 na.1995 19.2 26.9 17.9 25.1 7

2000 22.5 34.6 17.6 27.1 222005 26.6 40.9 17.6 27.0 342010 31.0 47.8 17.3 26.6 442015 35.3 54.3 15.3 23.5 57

get in GDP. Alternatively, all or part of the savingscould be used to increase spending per pupil or toincrease the enrollment rate in Malawi's secondaryschools which in 1980 stood at only 4 percent. Thereturns to using the resources saved on account oflower population growth for improving schoolquality are likely to be higher than the returns toforced rapid expansion of the system if populationgrowth does not slow. But improving quality willbe difficult until a larger share of the populationhas access to basic education, which itself isdelayed if the numbers of school-age children areconstantly increasing.

The potential for cutting educational coststhrough lower fertility is obviously largest forthose countries with the highest fertility rates.Four African countriesBurundi, Ethiopia,Malawi, Zimbabwecould save between 50 and 60percent of their educational spending by 2015 (seeTable 5.3), whereas a rapid fertility decline wouldreduce educational costs by only 5 percent inColombia, by 1 percent in Korea, and by even lessin China, where there is virtually no differencebetween the rapid and standard fertility assump-tions. But these lower-fertility countries havealready gained considerably from slower popula-tion growth. For example, if Korea's fertility ratehad remained at its 1960 level, the number of pri-mary school-age children in 1980 would have beenabout one-third (2 million) larger than it was.Applying actual 1980 costs per student ($300) tothat difference gives a saving in a single year of$600 million, about 1 percent of Korea's GDP.

GROWTH OF LABOR FORCE AND CAPITAL WIDENING.

Keeping up with schooling needs is only one waywhereby rapid population growth contributes to

TABLE 5.3

Potential savings in primary-school costs underrapid fertility decline, selected countries, 2000and 2015

86

a. Compared with standard fertility assumption.

capital widening. For most countries the same istrue of jobs. In contrast to school-age populations,whose rate of growth starts to slow five or six yearsafter a decline in fertility, the growth of working-age populations is more or less fixed for fifteen totwenty years. People born in 1980-84 will be enter-ing the labor force in 2000 and will still be therealmost halfway through the twenty-first century.

High-fertility countries face large increases intheir labor forces. As an example, Nigeria's highfertility in the 1970s guarantees that its working-age population will double by the end of this cen-tury. Kenya can expect an even larger increase.Where fertility has fallen in the past two decades,the increases will be smaller (see Figure 5.1). Chinawill experience a rise of no more than 45 percent.Korea's working-age population has already fallensubstantially and will change little between nowand the year 2000. In all these countries the actuallabor forcepeople who are working or looking forjobswill grow even faster if, for example, morewomen start looking for paid employment.

In countries with growing labor forces, the stockof capital (both human and physical) must contin-ually increase just to maintain capital per workerand current productivity. Unless this happens,each worker will produce less using the reducedland and capital each has to work with. Productiv-ity, and thus incomes, will then stagnate or evenfall. Wages will fall in relation to profits and rents,and thus increase income inequalitiesanotherexample of how rapid population growth harmsthe poor.

For incomes to rise, investment needs to growfaster than the labor force, to ensure capital deep-ening. Capital deepening involves a growingdemand for spending on education, health, roads,energy, farm machinery, ports, factories, and soforth. These requirements have to be traded offagainst extra consumption. Of course, if educa-tional levels are rising quickly, rapid restocking ofthe labor force with young, better-educated peoplecan be an advantage. But, as shown above, it isalso difficult to increase educational spending perchild if population growth is rapid.

Even when developing countries manage toraise investment in line with the growth in theirlabor force, the contrasts with developed countriesare striking. The gap in educational quality hasalready been described. Investment in physicalcapital per new worker is also much larger inindustrial countries because their labor-forcegrowth is slower and their GDP per capita is somuch higher. Even a middle-income country such

CountryTotal fertility

rate (1981)

Cost savings(percent)

2000 2015

Korea, Rep. of 3.0 12 1

Colombia 3.7 23 5

Egypt 4.8 27 23Burundi 6.5 26 56Ethiopia 6.5 25 60Kenya 8.0 22 50Zimbabwe 8.0 19 48

as Korea, with a high investment ratio of 31 per-cent in 1980, could provide only $30,000 of grossinvestment per new worker, compared with$189,000 in the United States, which had an invest-ment ratio of only 18 percent. (The investmentratio is gross domestic investment as a percentageof gross domestic product.) If all investment incountries such as Bangladesh, Ethiopia, Nepal,and Rwanda had been allocated to potential newworkers during 1980, each person would have hadless than $1,700 invested on his or her behalf (seeTable 5.4). At the other extreme, new workers inJapan would have had about $535,000 of grossinvestment available. Countries with the lowestabsolute levels of investment per potential newworker tend to be those also facing the fastestgrowth in their working-age populations. Just tomaintain the current small amount of investmentper potential new worker, they will have toincrease their investment rapidly. In contrast,developed countries can increase the capital avail-able to each potential new worker in 2000 even ifinvestment grows by less than 1 percent a year.

Rapid growth in the labor force has two othereffects.

TABLE 5.4

Gross domestic investment per potential new worker, selected countries, 1980

Note: Countries are listed in ascending order of their GNP per capita in 1982.Gross domestic investment as percentage of gross domestic product.Age cohort 15-64 years.

It is likely to exacerbate income inequalities,particularly if many new young workers have littleeducation. When a large proportion of workers areyoung and inexperienced, their productivity tendsto be lower. Except for those who have more edu-cation than older workers, their starting wages willtend to be lower, and they must compete with eachother. Relatively few will receive employer trainingto upgrade their skills. Over time, the weight ofnumbers of the unskilled will hold down theirwages in relation to those of skilled workers. AWorld Bank study of what determines incomegrowth among countries found that as overallincome rises, the average contribution of individ-ual workers without education falls-uneducatedworkers contribute (and probably earn) relativelyless than they once did.

It increases various forms of unemployment.Although population growth has had a relativelysmall effect on open unemployment in developingcountries, this fact does not demonstrate anydemographic stimulus to job creation. It simplyindicates that unemployment is not a feasibleoption for most people. Open unemployment istypically found most among educated urban

87

Grossdomestic Projected

Gross Increase in investment increase indomestic working-age per potential working-age

Investment investment populationh new worker populationratio (billions 1979-80 (thousands of 1980-2000

Count ry group (percent) of dollars) (millions) 1980 dollars) (percent)

Developing countriesBangladesh 17 1.90 1.70 1.09 74Ethiopia 10 0.37 0.24 1.53 76Nepal 14 0.26 0.21 1.26 78Rwanda 16 0.18 0.11 1.66 99Kenya 22 1.31 0.28 4.70 134Egypt, Arab Rep. 31 7.12 0.80 8.96 68Thailand 27 9.03 0.65 10.66 73Colombia 21 6.21 0.62 10.10 66Korea, Rep. of 31 18.06 0.61 29.85 45Brazil 22 52.35 1.30 40.36 65

Industrialized countriesJapan 32 332.80 0.62 535.04 11Australia 24 35.53 0.16 219.35 19France 23 149.94 0.33 461.34 13Germany 25 204.79 0.43 481.33 IUnited States 18 465.68 2.46 188.99 15

youths, who are presumably able to draw on fam-ily support while seeking work commensuratewith their qualifications or expectations. Manyothers are underemployed: "invisible underem-ployment" (including part-time and low-produc-tivity workers whose skills would permit higherearnings if better jobs were available) is estimatedto range from 20 percent in Latin America to about40 percent in Africa. In urban areas of most poorcountries, occupations that require little or no capi-talhandicraft production, hawking, and personalservices of all sortsare highly visible areas of theso-called informal sector. These occupations havethe advantage of using scarce financial capital effi-ciently, but the incomes they produce are oftenextremely low.

STRUCTURAL TRANSFORMATION OF THE LABOR

FORCE. As shown in Chapter 4, both urban andrural populations will increase rapidly into the nextcentury in the low-income countries of Asia andAfrica. Thus, while the general concern with theprovision of productive employment for urbandwellers is well founded, many countries will alsoface the task of absorbing considerably more work-ers into the rural economy. This double challengediffers from the historical experience of today'sindustrialized countries. Their economic growthwas helped by a massive shift of labor from agri-culture, where the amount of capital per workerand average productivity was relatively low, toindustry and services, where they were relatively

FIGURE 5.2

When will the number of agricultural workersstart to decline?

Annual growth rate of total labor force

4

3

1

Nonagricultural employmentannual growth rate

Kenya"worst case"

Japan,approximately 1890

0

0 50 100 150 200

Years until the agricultural labor force starts to decline

Assumes 70 percent of labor force in agriculture.Sourcn Johnston and Clark, 1982.

88

high. The two principal reasons for this structuraltransformation of the labor force are well known:

As incomes rise, people spend a smaller pro-portion on unprocessed agricultural produce anda larger proportion on industrial products andservices.

Increases in agricultural productivitymadepossible by technological innovations and accumu-lated investmentallow output to grow with aconstant or even declining farm labor force.

As average incomes increase in today's develop-ing countries, and as population growth ratesslow, the number of workers in agriculture shouldeventually decline. In some upper-middle-incomecountries in Latin America, including Argentina,Chile, Uruguay, and Venezuela, already less than20 percent of the labor force is employed in agricul-ture. But the transfer of labor out of agriculture hasproceeded much more slowly in much of low-income South Asia and sub-Saharan Africa. Thereare two reasons: their high rates of growth of thetotal labor force and their low initial shares in mod-ern sector employment.

In 1980 the share of the labor force in agricultureaveraged 73 percent in low-income countries(excluding China and India); in most countries ofsub-Saharan Africa it was between 80 and 90 per-cent. During the 1970s the total labor force in thesecountries grew at 2.3 percent a year. The rate ofgrowth will increase to 3 percent a year between1980 and 2000. The effects on the future growth ofthe agricultural labor force can be illustrated withsome hypothetical calculations.

Figure 5.2 portrays a country in which 70 percentof the labor force is in agriculture and in whichnonagricultural employment is growing at 4 per-cent a year. It shows, for different rates of growthof the total labor force, the time required for thesize of the agricultural labor force to start to declinein absolute numbers. For example, if it is assumedthat the annual growth of the total labor force is 2.5percent (which, combined with a 4 percent growthin nonagricultural employment, is a fairly typicalcombination in low-income countries), the agricul-tural labor force would continue to grow in abso-lute size (though declining slowly as a share of thetotal) for about fifty years (point x). If the totallabor force were to grow by 3 percent a yearinstead, the time required for the agricultural laborforce to start to decline would nearly double toninety-five years (point y). Although this exampleoversimplifiesfor instance, it does not admit thepossibility of massive urban unemploymentitdoes seem clear that the size of the agricultural

labor force in most of today's low-income coun-tries will go on increasing well into the twenty-firstcentury.

In western Europe and Japan, by contrast, thenumber of farm workers began to fall when thelabor force was still largely agrarian, so there werenever any significant increases in the size of theagricultural labor force. In Japan, for example, theshare of agriculture in the labor force in the mid-1880s was about 75 percentmuch the same as intoday's low-income countries, and nonagriculturalemployment grew at between 2 and 3.5 percent ayear in the late 1800s and early 1900s. In these tworespects Japan was similar to many low-incomecountries today. But the total labor force was grow-ing at less than 1 percent a year, much slower thanin developing countries today (see Figure 5.2). Soonly modest rises in nonagricultural employmentwere necessary to absorb the rise in the rural workforce. Between 1883-87 and 1913-17, the share ofthe labor force in agriculture fell by twenty per-centage points and the absolute number of farmworkers fell by some 1 million.

Kenya provides a dramatic contrast with the Jap-anese case. Only about 14 percent of the Kenyanlabor force is in wage employment in the "mod-ern" economy and about half of them are in thepublic sector. Between 1972 and 1980 employmentin the modern sector grew at 4.3 percent a year,higher than in Japan in the late nineteenth centurybut somewhat slower than the growth of GDP (4.9percent). But the rate of growth of the total laborforce was very rapid-3.5 percent. There was someshift of the labor force into the modern economy,since growth in modern sector employment wasfaster than in the total labor force. But the shift wassmall. Nonwage employmentmainly in agricul-

TABLE 5.5

Kenya: projections of employment by sector,under two scenarios, 1976-2050

Unemployment held constant in all years and in both cases(about 1.2 million workers).

Increases at 4 percent a year in both scenarios.Labor force grows at a constant 3.5 percent a year.Growth of labor force slows from 3.5 percent a year in

1976-2000, to 2.5 percent a year in 2001-10, to 1.5 percent a yearin 2011-25, and to 1 percent a year in 2026-50.

tureabsorbed more than 80 percent of theincrease in the labor force.

Agricultural output and jobs must continue togrow rapidly in Kenya: the effective demand forfood is rising at about 4 percent a year, so thatdomestic productionor other agricultural exportsto pay for food importsmust grow at least at thesame pace to avoid draining foreign exchange fromother sectors (if constant terms of trade areassumed). And the rest of the economy has only alimited capacity to absorb labor. The public sectoraccounted for about two-thirds of the growth inwage employment during 1972-80the number ofschoolteachers rose by more than the increase inmanufacturing workersbut its growth is con-strained by fiscal limits. Industry is relatively smalland capital intensive, so its work force is unlikelyto expand much.

These constraints are highlighted by the projec-tions in Table 5.5. In the "worst" caseessentiallya continuation of recent trends, with the laborforce growing at 3.5 percent a year and nonagricul-tural employment at 4 percentKenya's agricul-tural work force would still be increasing in abso-lute size even 100 years from now (see also Figure5.2). In the "best" case, which assumes the samegrowth in nonagricultural employment but slowergrowth in the labor force after 2000 (implying adecline in fertility starting in the mid-1980s), struc-tural transformation proceeds at a faster pace.Even so, agriculture must absorb more than 70 per-cent of the growth in the labor force for the rest ofthis century. It is only after 2025 that the number ofworkers in agriculture starts to decline. In themeantime, how to absorb these extra farm workersproductively is a critical issue in Kenya and inmany other countries in sub-Saharan Africa andSouth Asia.

Efficiency: allocating limited capital

Capital deepening (and associated absorption oflabor into the modern sector) is not the only con-tributor to economic growth. Last year's WorldDevelopment Report highlighted the importance ofmaking better use of existing resources, as well asof innovation and entrepreneurship. Promotingefficiency often requires policy reform. For exam-ple, many developing countries have a history ofsubsidizing capital; subsidies have discouragedlabor-intensive production and led to inefficientuse of scarce capital. Even with reform, efficiencymay not come easily; many technological innova-tions available to developing countries are labor-

89

(millions of workers)

Employment sectora 1976 2000 2025 2050

Nonagricultural employment1' 1.2 3.0 8.0 21.8Agricultural employment

"Worst" casec 3.8 9.9 24.1 56.9"Best" cased 3.8 9.9 12.4 4.5

saving because they come from the capital-richindustrial world. But efficiency is even harder toachieve when population growth is rapid. Forexample, social and political pressure to employyoung people has undoubtedly contributed to thelarge government sector in many developing coun-tries, and in some countries to regulationsdesigned to stop private employers from reducingtheir work force. Selective government concern foreducated young people in urban areas has led topolicies such as Egypt's that guarantee employ-ment to all university graduates. As well as beinginefficient, this policy hurts people who are noteducated because scarce public spending is

TABLE 5.6

Growth rates of food output by region, 1960-80(average annual percentage change)

90

Note: Production data are weighted by world export unit prices.Growth rates for decades are based on midpoints of five-yearaverages except that 1970 is the average for 1969-71.a. Excludes China.Sources: FAQ; World Bank, 1982b.

diverted for the benefit of those who are relativelywell off. Youth unemployment may also contributeto crime and instability and the resulting largeamount of service employment as police and pri-vate guards in some cities of developing countries.None of these, of course, adds to national income.Crime is tied primarily to poverty and social disor-der, but tends to increase wherever there are largecohorts of young people who are unemployed(including in developed countries).

Constraints on agricultural production

Food production in developing countries hasincreased rapidly in recent decades but has stilljust kept pace with population growth (see Table5.6); in the 1970s it failed to do so in many low-income countries, including Bangladesh, Nepal,and twenty-seven of thirty-nine countries in sub-Saharan Africa. Other African countries-includ-ing Kenya, Malawi, Rwanda, and Upper Volta-managed only a slight increase in per capita foodproduction. The output of food in China and Indiahas also exceeded population growth since themid-1960s, but by only a narrow margin.

In the past, increases in food production weremainly due to bringing more land under cultiva-tion: this is still the case in sub-Saharan Africa andin parts of Latin America. About 25 percent of theworld's land-some 3.4 billion hectares-isthought to be of agricultural potential. Of this,only about 1.4 billion hectares (40 percent) is beingcultivated, so there is little evidence of a globalland shortage (see Box 5.3).

For developing countries as a whole, however,increased acreage accounted for less than one-fifthof the growth in agricultural production over thepast two decades. In part this is because land recla-mation is often more costly than intensifying useof existing land; in part it is because further expan-sion of the land frontier is constrained in manyparts of the world. In sub-Saharan Africa, forexample, the development of vast areas is pre-cluded because of such diseases as river blindness(onchocerciasis) and sleeping sickness (trypanoso-miasis). The latter renders livestock productionvirtually impossible on some 10 million squarekilometers of higher rainfall areas, 45 percent of allthe land in sub-Saharan Africa. Major campaignshave been undertaken to free parts of the Suda-nese savanna country from sleeping sickness, butit has not always been possible to prevent a resur-gence of the disease. Moreover, insecticides usedto control tsetse flies, which spread sleeping sick-ness, have had undesirable effects on the environ-ment. For that and other reasons, some countriesin Africa are reaching the limits of their land (seeBox 8.4 in Chapter 8).

In Asia, too, further expansion of agriculturalland does not appear to be an option for severalcountries. For example, in India between 1953-54and 1971-72, a 66 percent increase in the numberof rural households was accompanied by only a 2percent increase in the cultivated area. As a result,the number of marginal holdings of less than one

Region orcount ry group

Total Per capita

1960-70 1970-80 1960-70 1970-80

Developing countries 2.9 2.8 0.4 0.4Low-income 2.6 2.2 0.2 -0.3Middle-income 3.2 3.3 0.7 0.9Africa 2.6 1.6 0.1 -1.1Middle East 2.6 2.9 0.1 0.2Latin

America 3.6 3.3 0.1 0.6Southeast Asia a 2.8 3.8 0.3 1.4South Asia 2.6 2.2 0.1 0.0Southern

Europe 3.2 3.5 1.8 1.9Industrial market

economies 2.3 2.0 1.3 1.1Nonmarket industrial

economies 3.2 1.7 2.2 0.9World 2.7 2.3 0.8 0.5

Box 5.3 Food supplies for a growing world populationThe food crisis of 1972-74 created anatmosphere of impending disaster and arenewed interest in Maithusian pessi-mism. More recent views point to thesuccess of technological change in agri-culture and to the conclusion that theworld as a whole is capable of producingenough food for future generations wellinto the next century. The main issue isnot the worldwide availability of food,but the capacity of nations, groupswithin nations, and individuals to obtainenough food for a healthy diet.

In most countries, particularly low-income ones, the staple food is cereals orcoarse grains; they account for about halfof total food consumption in developingcountries. Over the past thirty yearsglobal grain production has doubledand, according to the FAQ report Agricul-

ture: Toward 2000, could double again bythe year 2000. An American study, TheGlobal 2000 Report to the President, agrees

with this assessment. A doubling ofworld grain production over twentyyears or so amounts to an annual growthrate of about 3.5 percent.

On the demand side, earlier projec-tions indicated that demand for cerealsand grains for both human consumptionand livestock feed would rise at between3.0 and 3.5 percent a year, depending onassumptions for population and incomegrowth. More recent projections suggesta much slower growth of global demand.For example, the International WheatCouncil's recent Long-Tenn Grain Outlook

puts global cereal demand up 50 percentby 2000, equivalent to a rise of about 2.3percent a year; a report published by theUS Department of Agriculture came tosimilar conclusions. Both of these assess-ments included projections of grain thatwould be fed to animals. A World Bankstudy projects an average growth in thedemand for grains of about 2.6 percent ayear.

This optimism for the global Situationstands in sharp contrast to the assess-ments for groups of countries, individualcountries, and households. Various stud-ies suggest that the gap between domes-tic supply and demand is projected to

widen in the developing countries, par-ticularly because of continued rapidgrowth in population and income. Cen-trally planned economies may also con-tinue to have a shortfall. Production inthe industrial world is projected to rise,albeit more slowly than in the past, whilethe growth in its demand is projected tolevel out.

On a national and household basis, theoutlook is even more varied. A numberof industrial countries do not produceenough grain or food to satisfy domesticdemand. But their national food securityis assured because the value of their non-food exports is usually more than ade-quate to finance food imports. Thesecountries also have effective methods ofdistributing food, though their poorestpeople may be vulnerable. For some ofthe developing countries the situation isless secure. Estimates by the FAQ sug-gest that in the year 2000 twenty-ninedeveloping countries may be unable tofeed themselves from their own landwith inputs of fertilizers, seeds, and soon at an "intermediate" level of technol-ogy (a basic package of fertilizers,improved seed, and simple conservationmeasures). Many of these countries arein Africa, where technology is probablybelow the "intermediate" level (see alsoBox 8.4). Outside Africa, the groupincludes Afghanistan, Bangladesh, ElSalvador, Haiti, and Jordan.

Increasing domestic production of foodis not the only solution. Many develop-ing countries with a chronic food deficithave other options, the main one beingto increase exports of nonfood goods soas to finance food imports. For thosecountries with transitory food deficits, acombination of more exports and betterarrangements for storing food may bethe answer. For some countries in Africa,to avert a food crisis will require external

aidto finance food imports in the shortrun and to expand investment in devel-oping long-run potential for food andnonfood production.

Ultimately it is not countries but indi-viduals who suffer from a shortage offoodnot because of fluctuations in

national production but because ofhigher food prices, which they cannotafford, or because of inadequate arrange-ments for marketing food. Their diet willimprove only when their general eco-nomic state does.

Some research on the global food situa-tion has looked well beyond the end ofthis century. Bernard Gilland, for exam-ple, estimates the maximum global out-put of food to be 7,500 million tons grain-equivalent (tge) a year. This figure wasobtained by multiplying a "realistic"maximum yield of 5 tge per hectare (fromthe present average of 2 tge per hectare)by 1.5 billion hectares, allowing for aslight increase over the estimated 1.4 bil-lion hectares of land currently used forfood production. An additional 500 tgewas added for rangeland and marineproduction. Presently, gross consump-tion of plant energy for all purposesfood, seed, and animal feedrangesfrom 3,000 calories per person per day inSouth Asia to 15,000 calories in NorthAmerica, Australia, New Zealand, andFrance. (Consumption of meat raisesconsumption of "plant energy" becausethe conversion of grain to meat throughfeeding of livestock is inefficient com-pared with direct consumption of grain.)Gilland selects a "completely satisfac-tory" average daily per capita allowanceof 9,000 calories of "plant energy"(implying some meat consumption), andconcludes that the earth has the capacityto support about 7.5 billion people. Thispopulation will probably be reached inthe second decade of the next century.On a daily per capita allowance of 6,000calories of plant energythe currentworld averagethe earth would be capa-ble of supporting about 11,4 billion per-sons. That number is roughly equal tothe projected world stationary popula-tion. Cultivated land could be increasedmore, and land-saving technological ad-vances, especially deriving from geneticengineering, would transform the Out-look, allowing for better diets even aspopulation grows. But there are alsodownside risks (new crop diseases, soilerosion, and climate change).

91

acre increased from 15.4 million to 35.6 million andtheir average size fell from 0.27 to 0.14 acres. Totake another example, the average land-man ratioin Bangladesh is estimated to have declined from0.40 acres in 1960-61 to 0.29 acres in 1979-80. Morepeople have been absorbed into agriculture, butincomes have risen little if at all. More people areprobably having to earn a living as landless labor-ers. As their numbers have increased, their wageshave tended to fall in relation to those who own (oreven rent) land. The agricultural system hasadapted, but in ways that have probably increasedincome inequalities in the countryside.

Another constraint on the use of potentialagricultural land is shortage of water. In manydeveloping countries, any large expansion of agri-cultural production would require some form ofirrigation. Worldwide, the area under irrigationexpanded by almost 6 million hectares a year dur-ing the 1960s. India has shown the most dramaticgrowth, with the irrigated area increasing from 28million hectares to 55 million hectares over the pasttwo decades, an average of more than 1 millionhectares a year. In the 1970s, however, worldwideexpansion of irrigation slowed to just over 5 mil-lion hectares a year. This slowdown occurredbecause some countries, such as Pakistan, startedto run out of land that can be irrigated at an accept-able cost.

Shortage of water in many parts of India, in theNile Basin, in Brazil, and in most of the developingcountries is constraining irrigation development,and water transfer projects are being planned onan even bigger scale than those recently built inPakistan. Countries are also putting more empha-sis on groundwater development, on the com-bined use of ground and surface waters, on watereconomy, and on more advanced methods ofwater management. Poor water management isconsidered by many specialists to be the mostimportant single constraint to irrigated crop pro-duction. Bilateral and multinational agencies arenow trying to arrest the decline in managementstandards, and an International Irrigation Manage-ment Institute has recently been established topromote better use of water. The challenge will,however, remain formidable, especially in somecountries of sub-Saharan Africa, where little or noirrigation has been used in the past. This is particu-larly true for the Sahelian countries, where pro-gress has been limited mainly because of high con-struction costs ($10,000 to $15,000 per hectarecompared with $2,000 to $5,000 in Asia), lowfarmer response, and poor project management.

92

Easing constraints

Multiple croppingmore than one crop a yearfrom the same piece of landis a typical way forsocieties to cope with rising populations. In Asia,where the proportion of potential land under culti-vation was an estimated 78 percent in 1975, about 7percent of the cultivated land is cropped more thanonce. For some Asian countries the proportion ismuch greater. In the late 1960s 52 percent of culti-vated land was cropped more than once in Korea.In Bangladesh in the late 1970s, 43 percent of theland was cropped more than once.

Multiple cropping increases production and usesmore labor, so that the chief resource required tofeed the growing populations of developing coun-tries is provided by the people themselves. Farmstudies in Africa and Asia show that, on average, a10 percent increase in farming intensity (defined asthe percentage of time in the rotation cycle that isdevoted to cropping) involves a 3 to 4 percentincrease in the amount of labor per hectare. Laborinput per hectare increases because, under inten-sive farming systems, the extra hours required forland preparation, sowing, weeding, and plant pro-tection more than offset the reduction of hoursessentially for land clearingassociated withshorter fallow periods.

But the combined benefits of more employmentand more food do not come automatically. Withoutmodern technical packagesincluding purchasedinputs such as fertilizers and improved seedsandeffective price incentives, the amount of labor usedcan increase faster than output. Less fertile landmay be brought under cultivation; good land maybe given less time to regain its fertility. Researchinto farming systems and increased use of agricul-tural extension services can help ensure that newfarming methods are compatible with availableresources, including labor. But population pres-sure is likely to continue. In parts of Africa, and inChina, Bangladesh, and Java in Indonesia, popula-tion pressure has already forced people to workharder just to maintain income in traditionalagriculture.

In most developing countries, however, laborproductivity has been maintained. To forestalldiminishing returns to labor, intensification of landuse has usually been accompanied by better farm-ing methods, the use of fertilizers, investments inirrigation and drainage, and mechanization. How-ever, such measures are possible only where rain-fall is favorableor where water is available forirrigationand where topography and soils do not

impose constraints that cannot be eased at accept-able costs.

Higher population density, by permitting econo-mies of scale in the provision of infrastructure andservices, can sometimes help to induce improve-ments in agriculture. In the United States, forinstance, rising population densities stimulatedthe development of the transport system duringthe nineteenth century. An improved transportsystem, in turn, greatly facilitated the growth ofagriculture by lowering transport costs and by rais-ing the farmgate prices of agricultural products.But the potential benefits to be gained from higherpopulation densities are not always realized. In

rural Bangladeshone of the most crowded areasin the worldtransport, marketing and storagefacilities, as well as extension services, are allinadequate.

From one point of view, it is no small achieve-ment to sustain an increase in population on thescale that has occurred, and continues to occur, inmany developing countries. But keeping the pro-duction of food up with (or even ahead of) growthin population is no guarantee that people have ahealthy diet. Where incomes are low and une-qually distributed, and increases in food produc-tion are just barely ahead of increases in popula-tion size, poor people may not be able to afford the

FIGURE 5.3

Foodgrains and population in India, 1950-83

Population(millions)

700

Average annual600 growth rate

2.1 percent

500

400

300

Production of foodgrains(millions of tons)

130

110

90

70

0.15

1950 1955

Production of foodgrains(kg. per capita)

0.20

50 -U'1960 1965

Output trend:average annualgrowth rate of

2.7 percent

I

1970

0 101950 1955 1960 1965 1970 1975 1980

Years shown are end years for the agricultural season. Data for 1983-84 are estimates.

Sources World Bank data; adapted from Cassen, 1978.

93

1950 1955 1960 1965 1970 1975 1980

1975 1980

food they need. Although the amount of laborused in farming has risen in most developingcountries, many families have had little or noincrease in their income. They are particularly vul-nerable when harvests are poor. As Figure 5.3shows, India has managed to expand its food pro-duction faster than its population has grown. Butoutput per person has varied considerably, and inbad years when food prices have risen, the land-less poor probably went hungry.

Most developing countries still have potential foryield increases (see Table 5.7). But in some land-scarce countries-for instance, Egypt, China, andKorea-yields are already high. Paddy yields inEgypt in 1979-81 averaged 5.6 tons per hectare,

TABLE 5.7

Cereal yields and fertilizer use, selected countries, 1969-81

94

Note: Harvested area covers all cropped areas including tree crops.Sources: FAO, 1982; World Bank.

matching what was achieved in Japan. Becausethese countries also use land intensively, theyneed larger inputs of fertilizer to increase food pro-duction. A second group of countries are likely togo on importing food indefinitely because of con-straints on land, irrigation, and so on. For a thirdgroup, although the constraints appear lesssevere, the process of agricultural modernizationmay take several decades. In the meantime, theymay need to increase their food imports.

Population and the environment

Rapid population growth can contribute to envi-ronmental damage, especially when combined

Country group

All cereal yields (tons perhectare of harvested area)

Fertilizer use (kilograms perhectare of harvested area)

1969-71 1979-81 1969-71 1978-80

Industrialized countriesUnited States 3.50 4.20 172.4 192.5Denmark 3.85 4.02 331.0 331.0Netherlands 4.02 5.69 868.0 1,121.5Japan 5.04 5.27 426.2 532.6

Developing countriesAfrica

Burundi 1.04 0.99 0.6 1.1Cameroon 0.89 0.89 7.9 11.6Egypt, Arab Rep. 3.85 4.01 115.0 188.5Kenya 1.47 1.50 18.4 18.4Malawi 1.00 1.18 6.6 16.1Zimbabwe 1.08 1.36 54.3 64.2Tanzania 0.78 0.70 3.3 6.1Zambia 0.76 0.74 25.4 47.5

AsiaBangladesh 1.66 1.96 10.7 31.9Sri Lanka 2.40 2.42 53.3 70.2India 1.11 1.34 12.5 31.7Korea, Rep. of 3.50 4.77 193.1 354.5Malaysia 2.39 2.82 45.3 102.3Pakistan 1.21 1.61 23.1 61.2Philippines 1.30 1.59 23.5 30.7Thailand 2.01 1.94 9.9 20.1

Latin AmericaArgentina 1.71 2.20 3.3 4.4Brazil 1.33 1.50 24.0 72.6Colombia 1.72 2.46 43.0 69.0Chile 1.86 2.12 106.2 84.2Costa Rica 1.55 2.25 141.9 177.9Ecuador 1.04 1.61 17.5 48.5El Salvador 1.57 1.72 104.2 105.3Guatemala 1.12 1.51 25.8 65.2Mexico 1.52 2.11 38.7 79.8

with certain nondemographic factors. For exam-ple, an unequal distribution of farmland, byrestricting access to better soils, can help to pushgrowing numbers of people onto ecologically sen-sitive areaserosion-prone hillsides, semiaridsavannas, and tropical forests. One example is themigration to the Amazon rainforests from ruralareas of northeastern Brazil, where 6 percent of thelandholdings account for more than 70 percent ofthe land area. Social changes can also bring eco-logical threats: in Kenya and Uganda pastoralgroups, whose political power was destroyedunder colonial rule, have seen their closed systemof communal management converted into openaccess to their land. With added populationgrowth, overgrazing and severe environmentaldamage have followed. Population pressure is notalways the main culprit, but it almost always exac-erbates the problem.

Of course, the environmental problems of thedeveloping countries are not confined to the coun-tryside. Industrialization and urbanization havealready led to severe air, water, and noise pollutionin some cities. Although such pollution is a hazardto public health, it does not pose as immediate athreat to the economic life of low-income countriesas does deforestation and desertification. In drycountries, these two threats are closely linked.

Deforestation

Forests are central to the economic and ecologicallife of many developing countries. They help tocontrol floods and thus protect roads in mountain-ous and wet areas. Floods and landslides havebecome serious problems in steep, deforested

areas such as in Nepal (see Box 5.4). Forests alsoprotect power production from hydroelectricschemes. When watersheds are cleared, damsoften start to silt up. Less electricity can be gener-ated (because less water can flow through theturbines); thus the economic life of the investmentis reduced. For example, the useful life of theAmbuklao Dam in the Philippines has been cutfrom sixty to thirty-two years because ofdeforestation.

Satisfying the demand for firewood is a majorcause of deforestation, particularly in the drier andhigher regions where trees grow slowly. To meettheir daily energy needs, an estimated 1.3 billionpeople must cut firewood faster than it can bereplaced by natural growth. Unfortunately, thosewho are exhausting the forest seldom recognizewhat they are doing. The depletion becomesapparent only when obtaining adequate suppliesrequires more physical effort or greater expense. Inthe Gambia and Tanzania population growth hasmade wood so scarce that each household spends250 to 300 worker-days a year gathering the woodit needs. In Addis Ababa, Ethiopia, the price ofwood increased tenfold during the 1970s and nowclaims up to 20 percent of household income.

The scarcity of wood has profound implicationsfor everyday life in developing countries. Whenthere is not enough fuel to heat food and boilwater, diseases spread more rapidly. In Chinamore than 70 million rural householdsabout 350million peoplesuffer serious fuel shortages forup to six months a year when crop residues areexhausted and wood is unavailable in deforestedareas. In much of West Africa, families tradition-ally cooked two meals a day. Now, because wood

Box 5.4 Reclaiming the Himalayan watersheds

The Ganges river, which flows throughIndia and Bangladesh, floods every year,causing millions of dnllars of damageand incalculable human suffering. Butthe floods and the resulting damage aremuch worse than they need be. In themountainous watersheds of northernIndia and Nepal, population growth hasled to severe deforestation, which hascaused the area's heavy rains to run offrather than soak into the soil. In the low-land areas surrounding the Ganges, pop-ulation growth and competition for land

has forced many people to live too closeto the river, in the path of the annualfloods. As testimony to the effectspopulation growth, the severity of flood-ing has increased exponentially over thepast twenty years, even though theannual rainfall has hardly changed.

To help combat the flooding, the WorldBank is funding a pilot project in theIndian state of Uttar Pradesh to developnine small watersheds covering 312,000hectares. By planting trees over a widearea the project will attempt to reclaim

denu . It will also encouragestall feeding 0! stock to help alleviitthe damage dooc by roaming anini.-Farmland will be te'erosion. At the saiments of Banglade'.are pursuing poliupopulation growth,to deforestation as well as high poption density in the flood-prone areas (s'Chapter 8).

-n

95

is so scarce, they can do so only once a day or onceevery other day. A more specific example is that ofsoybeans in Upper Volta. They are a new crop, areexceptionally nutritious, and have grown well, butthey are not popular because they have to becooked a long time. Similar experiences have beenreported in Haiti.

Managed village woodlots, fuelwood planta-tions, or more efficient wood stoves could domuch to ease shortages. For example, a well-man-aged woodlot planted with fast-growing trees canyield as much as twenty cubic meters of wood perhectare annually, six times the yield of an unman-aged natural forest. However, these and othermeasures are not easy to introduce. They requirelocal testing and adaptation, large numbers oftrained staff, and adequate economic and institu-tional incentives. But the returns from forestrydevelopment can be high. In Ethiopia, where fuel-wood shortages have become critical in someregions, estimated rates of return on investmentsin rural forestry are on the order of 23 percent.

Another major cause of deforestation is theexpansion of agriculture. According to the FAO,agricultural growth involves clearing more than 11million hectares of forest a year, primarily inresponse to population pressures. Unless thesemarginal lands are given much commercial atten-tionfertilizers, irrigation, and so onthey soontend to become eroded and infertile. When thishappens the settlers clear more forest, a destruc-tive and unsustainable process. Fertilizers areoften an uneconomic remedy, being expensive andineffective in the soil and rainfall conditions ofmany tropical areas.

Desertification

The effects of gradually spreading desert are oftenconfused with those of drought. But droughts, nomatter how severe, are ephemeral; when the rainsreturn, the land's inherent productivity isrestored. With desertification, even normal rainfallcannot fully restore the land. In extreme cases,land may remain unproductive for many genera-tions unless costly remedies are taken.

While drought can help to turn land into desertand make the effects more obvious to people livingthere, most scientists agree that changes in climateare not responsible for the vast areas of semiaridland going out of production each year. The directcauses of desertification include overcultivation,overgrazing, and deforestation. These practicesstrip vegetation from the topsoil and deprive it of

96

nutrients and organic matter, thereby exposing itto erosion from the sun and wind. These directcauses themselves spring from the pressures ofrapid population growth. In trying to obtain morefood for themselves and their livestock, growingnumbers of people frequently overstretch the car-rying capacity of semiarid areas: keeping produc-tion high during drought reduces the land's natu-ral resilience and sets it on a course to permanentdegradation.

Although some 100 countries are affected bydesertification, the process is most serious in sub-Saharan Africa (particularly the Sahel), northwest-ern Asia, and the Middle East. Every year an addi-tional 200,000 square kilometersan area largerthan Senegalare reduced by desertification to thepoint of yielding nothing. And the process is accel-erating: more than 20 percent of the earth's sur-facenow populated by 80 million peopleisdirectly threatened. The human costs of desertifi-cation often include malnutrition, threat of famine,and dislocation of people who must abandon theirlands to seek employment elsewhere.

Urban population growth and internal migration

Beyond a common concern, perceptions of theproblem of the distribution of population vary con-siderably among developing countries. Some seethe countryside as overpopulated in relation to itsnatural resources. Others complain of labor short-ages in remote but resource rich areas. Most com-monly, however, maldistribution is described interms of "overurbanization" caused by "exces-sive" migration. In some developing countries,rapid urban growth has undoubtedly caused seri-ous administrative difficulties. Urban life requiresa complicated set of serviceshousing, traffic,sewerage, water, and so onthat cannot quicklybe scaled up as population grows. City administra-tions are usually short of money, and may anywaylack the managerial skill to cope with a city thatdoubles its size in a decade. Where this happens,the results are familiar: unemployment, substand-ard housing, deteriorating public services, conges-tion, pollution, crime, and so forth.

An overriding concern with the negative aspectsof urban growth, however, has often led policyma-kers to overlook some of the benefits to be gainedfrom internal migration and urbanization. As aresult, many governments have chosen to carryout costlyand often economically inefficientprograms to redistribute population. They wouldhave done better to have concentrated on rural

development in areas already settled, on improve-ments in urban policies and management, on elim-ination of price distortions (such as keeping foodprices low) that encourage urban populationgrowth, and on development of effective familyplanning programs to reduce rates of natural pop-ulation increase.

Projections of urban growth (which were shownin Table 4.3) are not meant to predict what willactually happenmerely what would happen ifhistorical trends continued. As such, projectionsare sensitive to small changes in trends. There isevidence that the rate of urban growth in develop-ing countries slowed slightly after 1973 in responseto the world economic slowdown. That decreasecould produce a much smaller urban populationthan shown by the projections. Though this wouldmake urban growth easier to cope with, it would(without a compensating decline in the overallpopulation growth rate) imply faster rural growth.

The benefits and costs of urbanization

Urban growth gives rise to economies of scale.Industries benefit from concentrations of suppliersand consumers, which allow savings in communi-cations and transport costs. Large cities also pro-vide big, differentiated labor markets and mayhelp to accelerate the pace of technological innova-tion. They also allow economies of scale for suchservices as water supply and electric power to beexploited. Evidence from India suggests that sub-stantial economies of scale are found in cities of upto 150,000 inhabitants. The point at which dis-economies creep in, because cities are too big, hasnot been clearly demonstrated.

Against these benefits, unemployment tends tobe higher in urban than rural areas. In a survey offourteen developing countries, only one (theIslamic Republic of Iran) had a higher rural unem-ployment than urban unemployment rate; in sixcountries the urban unemployment rate was morethan twice the rural rate. Surveys confirm that airpollution, congestion, social disturbances, crime,and similar problems also increase disproportion-ately with city size. But these problems are oftenaggravated by poor urban management. Typically,governments reduce the absorptive capacity of cit-ies by intervening in labor markets (for instance,through minimum wage legislation, and licensingrequirements and restrictions on small busi-nesses), and by pursuing inappropriate pricingpolicies for public services. National economic pol-icieswhich provide fiscal incentives and low-

interest loans to promote capital-intensive indus-try, for examplemay also exacerbate urbanproblems by encouraging rural-urban migrationwithout creating enough new urban jobs.

Whatever the cause, the drift from countrysideto city is a concern to governments. A 1983 UNsurvey of 126 governments of developing coun-tries found that only 3 considered the distributionof their populations "appropriate." Moreover, allthree were governments of small island nations:Barbados, Malta, and Nauru. Concern was great-est in Africa, the Middle East, and low-incomeAsia: virtually all governments in these regionsconsidered population distribution either "parti-ally appropriate" or "inappropriate." As a rem-edy, more than three-quarters of all respondentsstated that they were pursuing policies to slowdown or reverse internal migration.

Between 1925 and 1950 at least 100 million peo-ple in the developing countriesabout 10 percentof their rural population in 1925migrated fromthe countryside to towns and cities. During thefollowing twenty-five years, the numbers rose toan estimated 330 million, equivalent to almost aquarter of the rural population of the developingcountries in 1950. Population movements withinrural and urban areas, and temporary migration,have undoubtedly involved even more people,although their numbers are not reliably known.

The role of internal migration

Current high rates of urban growth in developingcountries are only partly due to rural-urban migra-tion. Natural population increase is estimated toaccount for 60 percent of the rise in urban popula-tions, according to a UN sample of twenty-ninedeveloping countries. Perhaps another 8 to 15 per-cent is attributable to the reclassification of ruralareas to urban status. Additional evidence fromIndia, Kenya, and several West African countriesconfirms this pattern.

Although fertility rates are on average lower inurban than in rural areas, differences within coun-tries between urban and rural fertility tend to besmall (see Chapter 6). Thus the effect of urbaniza-tion on aggregate fertility is limited in the shortrun, especially because migrants tend to be ofchildbearing age, raising the number of births incities even when the rate of fertility is lower. Natu-ral increase in urban areas is therefore substantial.

Migration then puts even greater strain on thecapacity of cities to cope with rapidly growingnumbers. In broad perspective, the shift of people

97

from rural to urban areas mainly reflects the proc-ess of industrialization and the changes it brings inthe demand for labor. Certain conditions in ruralareasunequal land distribution, landlessness,agricultural mechanization, natural calamities,and, in the past, forced labor migrationshavestrongly influenced population movements inmany countries. But, by and large, people move totowns and cities for higher incomes and better jobopportunities.

For individual families, these attractions can beconsiderable. Once in the city, perhaps three out offour migrants make economic gains. A move fromthe rural Northeast of Brazil to Rio de Janeiro, forexample, can roughly triple the income of anunskilled worker; the family income of a manuallaborer in Sao Paulo is almost five times that of afarm laborer in the Northeast. The higher cost ofurban living may narrow rural-urban wage differ-entials in real terms, but urban dwellers also gen-erally have much better access to basic publicservices. To take one example, in rural areas ofsub-Saharan Africa only about 10 percent of thepopulation has access to a safe water supply, com-pared with 66 percent of the urban population.

Most studies conclude that migrants are assets tothe urban economy. They are mostly between theages of fifteen and twenty-nine and are better edu-cated and more motivated than those who staybehind in the countryside. Evidence from Brazil,Colombia, Kenya, Korea, India, and Malaysiashows that migrants with long urban residencecompare favorably with urban-born people interms of employment and income. A World Bankstudy of Bogota, Colombia, found that migrantsearned more than nonmigrants at all educationallevels. Overall, income and employment levels aremore a function of age, sex, and education than ofwhether a person has migrated or not.

Evidence about the impact on rural areas of emi-gration is mixed. Emigration seldom causes a dropin farm output. In villages of East Kalimantan,Indonesia, for instance, women have adjusted tothe departure of male emigrants by workingharder at rice and vegetable production. Otherreactions include shifts to less labor intensive crop-ping patterns, increased use of wage labor, andagricultural mechanization.

Urban-rural remittances clearly benefit ruralhouseholds. Village studies in India, Malawi, andThailand, however, show that net remittancesmigrants receive as well as send moneyusuallyaccount for only a small proportion of ruralincomes. Returning migrants can be an important

98

source of innovation, but only if opportunitiesexist to exploit their ideas. Studies in Guatemala,Papua New Guinea, Peru, and Tanzania, for exam-ple, have shown that returning migrants can intro-duce new crops and techniques. Other studieshave found that experience gained in modern fac-tories is largely irrelevant to the needs of smallvillages.

Redistribution policies

Governments have employed many differentapproaches to the task of slowing down rural-urban migration, ranging from direct controls onpopulation mobility to efforts to improve economicconditions in the countryside. Few of these poli-cies have achieved their demographic objectives,and their social and financial costs have been high.Moreover, they have often been undermined bynational policies in agriculture, industry, and for-eign trade.

Direct controls on mobility have been most com-mon in centrally planned economies. China, forexample, has employed controls since the early1950s in an attempt to stabilize its urban popula-tion. These controls have taken the form of travelpermits and food ration cards that can be usedonly in specified areas; also, restrictions have beenplaced on labor recruitment in rural areas by urbanindustrial enterprises. In some cases large num-bers of city dwellers have been exhorted to moveto the countryside. The "rustication" program, forinstance, resettled some 10 to 15 million urban sec-ondary school graduates in rural areas between1969 and 1973. Administrative measures haveprobably helped to slow urban population growth:the proportion of the population in urban areashas changed only slightly over the past thirtyyears. But the costs were high to individuals, andthe economy also suffered from misallocations oflabor.

Less stringent controls have been used in Indo-nesia and in the Philippines. Starting in 1970,migrants to Jakarta had to comply with an array ofbureaucratic requirements, including cashdeposits and licenses for various business activi-ties, however informal. To limit the growth ofManila, the city government in 1963 decided tocharge migrants a sizable fee to enter the publicschool system; free education was available only tobona fide residents. In both cases, the controlsproved hard to enforce, gave rise to petty corrup-tion, and failed to slow urban growth significantly.A variant of such controls has been periodic expul-

sions of unemployed migrants from cities, a prac-tice that has been attempted in parts of Africa,notably the Congo, Niger, Tanzania, and Zaire.They too have had little visible impact.

Population redistribution is commonly a majorobjective of land-settlement schemes. The transmi-gration program in Indonesia, for example, aims toease population pressures in rural Javawith onlylimited success, it seems (see Box 5.5). Similarly,Brazil's TransAmazon Program did little to furtherthe goal of reducing population growth in thesemiarid Northeast. Evidence suggests that theFederal Land Development Authority (FELDA)settlement scheme in Malaysia has succeeded inslowing down intrarural and rural-urban migra-

tion. But costs have been high (about $15,000 perfamily in the 1970s), and "second generation"problemsincreasing social differentiation in set-tlement areas, and renewed pressures on land assettler families increase in sizehave begun toappear. Although land settlement may haveimportant political and social objectives, a reviewof World Bank-assisted schemes concluded that, ineconomic terms, it is usually more efficient tointensify production in already settled areas thanto move people elsewhere.

Governments have also tried to modify popula-tion distribution by making small and medium-size towns an attractive alternative to the majorcities. Evidence from India, Peru, Thailand, and

Box 5.5 Indonesia's transmigration programIndonesia's populationestimated at 153million in 1982is unevenly distributedover 13,600 islands, covering about 1.9million square kilometers. A singleisland, Java, accounts for about two-thirds of the country's population butonly 7 percent of the land area. java hasan average of 690 people per square kilo-meter (higher than Bangladesh); in irri-gated areas the density rises to 2,000 peo-ple per square kilometer. In contrast,large areas of the other islands, includingSumatra, Kalimantan, Sulawesi, andIrian Jaya, are sparsely populated.

Java has fertile volcanic soils, whichallow intensive agriculture withoutheavy applications of fertilizer. Some 70percent of the island is cultivated. Theother islands, however, have generallypoor tropical soils. Over the years, muchof the population growth in rural Javahas been absorbed through "agriculturalinvolution," a process through whichland productivity is raised by addingmore and more workers. But growingpopulation pressures have contributed toecologically harmful farming practices,such as the clearing and cultivation ofsteep hillsides. More than 23 millionhectares have already been degraded.Labor productivity and rural incomeshave declined in parts of the island, andlandlessness and rural underemploy-ment are widespread. In 1980 an esti-mated 47 percent of rural Javanese werebelow the absolute poverty line, com-

pared with 28 percent of the rural popu-lation of the other islands.

The big demographic and economicdifferences among the islands of Indone-sia have prompted many programs formoving people from Java to the otherislands. The Dutch began a resettlementprogram in 1905, moving 155 familiesfrom central Java to Lampung provincein Sumatra. By 1932, some 27,000 people

(roughly 1,000 per year) had beenmoved. Between 1932 and 1969 the pro-gramwhich became known as "Trans-migration" in 1950slowly gatheredmomentum. By 1969, about 580,000 morepeople (about 15,000 per year) had beenresettled. But since Java's populationgrew by some 35 million over this period,the transmigration program had only aminor impact.

With the First Five-year Plan (1969-74),the transmigration program became anational priority and was furtherexpanded. The World Bank has sup-ported this expansion with four loansand one credit, totaling about $350 mil-lion. Since 1969, 479,000 families(approximately 2.4 million people) havebeen settled outside of Java at an averagecost per family varying between $4,000and $8,000. The program has alsoencouraged some spontaneous migra-tion, estimated at 1 million people since1969. Although the transmigration pro-gram has in recent years succeeded inresettling the equivalent of a quarter of

Java's natural population increase, theisland's rate of population growth actu-ally increased slightly from 1.9 percent ayear in 1961-71 to 2 percent a year in1971-80.

Of course, the transmigration programshould not be judged solely on its abilityto ease population pressures in Java.Emigrants have been drawn from thepoorest groups in Java and from the mostecologically vulnerable areas. Reviews ofthe program carried out by the WorldBank found that these settlers were bet-ter off in most transmigration sites thanthey had been in Java. Nevertheless,average crop yields and incomes inupland areas have been low and variable.Of 592 farmers surveyed in communitiesdependent on rainfed agriculture, only 9percent reported paddy yields of morethan one ton per family.

The Indonesian government has setambitious targetsto move some 13 mil-lion families from Java over a twenty-year period. For the immediate future,the government intends at least to matchthe target it set in the Third Five-YearPlan (1979-84) of 100,000 families a year.

Costs are likely to increase as more andmore remote areas are opened up, andthis could constrain the program's devel-opment. Nevertheless, transmigrationwill continue to receive a high priorityamong government programs, not leastbecause of what it can do to alleviatepoverty.

99

other developing countries suggests that thisobjective is seldom achieved. One exception isKorea: through the introduction of special tax andcredit incentives in the early 1970s, industrial activ-ity and people were successfully attracted tosmaller cities. One result was that populationgrowth in Seoul slowed from 9.8 percent a year inthe 1960s to 4.5 percent a year in the 1970s. But thisachievement was helped considerably by a combi-nation of circumstances possibly unique to Korea:a rapidly declining rural population, a stable gov-ernment, a wide range of social services, and abooming economy.

Population growth and the international economy

Demographic change is tending to increase eco-nomic disparities between developed and develop-ing countries. Between now and the year 2000, forexample, the number of people aged twenty toforty will increase at about 2.6 percent a year in thedeveloping countries, roughly ten times fasterthan in developed countries. In absolute terms, thedifference is even more striking. Numbers in thetwenty to forty age group will increase by 19 mil-lion in developed countries, less than one-third ofthe increase from 1960 to 1980. In developingcountries the increase will be 600 million, one anda half times the 1960-80 increase. The size of theworking-age population in China and Indiawhich was about 60 percent larger than the totalfor industrial countries in 1960will be more than150 percent larger by 2000. Even if per capitaincome grows faster in developing than in indus-trialized countries, the absolute income gap willnot decrease significantly because the initial differ-ence in per capita income is, for many developingcountries, so large. To what extent can interna-tional migration and trade reduce these disparitiesand alleviate the problem of rapid populationgrowth in developing countries?

International migration

The motivation for most international migration isthe same as for internal migrationhigher wages.Historically, some migration may have beendirectly related to population pressure, but todaywage differences are the main driving force. Forexample, in the late 1970s, an unskilled emigrantworker from Bangladesh earned up to ten timesmore in the Arab gulf states than he did in his owncountry. To the extent that population growthaffects those differences, it is, of course, indirectly

100

a cause of migration.Despite the growing income gap between rich

and poor countries and the widening gap in thesize of the labor force, the scale of present-daymigration is relatively small and unlikely toincrease dramatically (see Chapter 4). The mostimportant reason is the immigration policies ofhost countries. Their policies vary according totheir economic needs, but they generally placesome limits on immigration because of the effectson the wages of natives and because of social andpolitical tensions that are often created by large-scale immigration.

CONSEQUENCES FOR THE RECEIVING COUNTRY. In

general, immigration becomes controversial whennew workers reduce wagesusually because thedemand for urban labor is not rising fast enough toensure that an added supply of labor will not causewages to fall. For example, increased resistance toimmigration in the United States after the 1890swas partly the result of a decline in the growth offarmland, retardation of capital accumulation, andtechnological change that favored capital- andskill-intensive sectorsall of which reduced thegrowth of demand for unskilled labor. Morerecently, restrictions on the use of migrant labor inwestern Europe increased when the 1974-75 reces-sion began.

Host countries generally benefit from immigra-tion; in the Middle East migrants form an indis-pensable part of the labor force. Host countries canalso select immigrants whose skills and qualifica-tions suit their pattern of demand (see Box 5.6).Immigration, often from developing countries,thus provides a flexible source of supply, enablingreceiving countries to adapt more quickly tochanges in demand than they could do withoutimmigration. But the economic gains to host coun-tries must be balanced against social costs. Immi-gration can create social tensions, often concen-trated locally: whole neighborhoods exist inEuropean countries and in the United Stateswhere adults are predominantly first-generationimmigrants. Of France's 4 million foreigners, 40percent live around Paris; in some sections of thecity, more than half the primary-school childrenhave foreign parents.

One response of host countries has been toshorten the stay of immigrants through temporaryrecruitment rather than permanent immigration.These efforts are not always successful: in westernEurope, for example, the same workers returned,and the average length of stay increased.

As immigrants increase as a proportion of thepopulation, they receive increasing attention andpublic resources. In the long run, these factors arelikely to be more important than purely economicfactors in maintaining limits on immigration.

CONSEQUENCES FOR THE SENDING COUNTRY. Asubstantial part of recent migration has involvedunskilled workers. Migrant workers from theYemen Arab Republic (constituting more than 30percent of the national work force in 1981) werepractically all unskilled. So were a large proportionof emigrants from other countries that sent labor tothe booming Middle East in the 1970srangingfrom about 30 percent for Bangladesh and Jordanto about 50 percent for Egypt. About 50 percent ofimmigrants into Ghana and the Ivory Coast areemployed in agriculture, usually as laborers.Unskilled laborers in western Europe and theUnited States may be more skilled than those inthe Middle East, but they comprised 30 percent ofmigrant manual workers in Germany and morethan 40 percent of temporary workers admitted tothe United States in recent years. Illegal workers inthe United States are largely uneducated andunskilled.

In some countries emigration has contributed tosubstantial increases in wages of the unskilled athome. For example, real wages of unskilled con-struction labor in the largest cities of Pakistanincreased at an annual rate of more than 15 percenta year between 1972 and 1978 (faster than the rateof growth of wages of carpenters or masons), afterremaining stagnant for several previous years. Inthe Yemen Arab Republic, which experiencedheavy international as well as rural-urban migra-tion, real wages of agricultural labor increasedalmost sixfold between 1972 and 1978. During1975-79, they rose from 56 to 63 percent of urbanwages; urban wages rose from 45 to 67 percent ofthose in Saudi Arabia. In Egypt the rate of increaseof real wages in construction was about 6 percent ayear during 1974-77, after stagnating in the pre-vious ten years. Considering the low wages thatthe unskilled earn (for example, less than $2 a dayin Pakistan in construction in 1977-78, less than $5a day in Egypt in 1977), these wage increases mustbe considered beneficial, particularly since there isno evidence that output declined.

An additional benefit is the money that emi-grants send back home. It serves not only toincrease the incomes of their families but also tohelp finance their country's trade deficit. Workers'remittances increased from about $3 billion in 1970

to $27 billion in 1980. In 1980, remittances pro-vided almost as much foreign exchange as exportsdid for Pakistan and Upper Volta; they were morethan 60 percent of exports for Egypt, Turkey, andPortugal, and about 40 percent for Bangladesh andYugoslavia.

Many countries have special schemes to attractremittances. India and Yugoslavia allow foreigncurrency accounts, with interest and capital with-drawable in foreign currency. Bangladesh issuesimport permit vouchers, which carry a specialexchange rate and may be freely negotiated.China, Korea, and the Philippines have mandatoryremittance requirements.

Migrant workers tend to save a lot. The averagepropensity to save by Turkish emigrants was 35percent in 1971 (compared to a gross domestic sav-ings rate of 16 percent), and as high as 70 percentfor Pakistanis in 1979 (compared to a gross domes-tic savings rate of less than 10 percent). The aver-age propensity to remit, which may be more rele-vant to the emigrant country, was lower, but still11 percent for workers from Turkey and about 50percent for workers from Pakistan.

With this new source of income, the living stand-ards of many families improve significantly. Alarge part of remittances (about 60 percent, accord-ing to one survey in Pakistan) is spent on food,clothing, rent, and other standard householditems. Many of the consumer durables areimported. Beyond using remittances to increasetheir current spending, families tend to repay debtand invest their extra income, mostly in urban realestate, and in agricultural land and housing. A sur-vey conducted in 1977 in the Indian state of Keralashowed that land and buildings accounted for anaverage 75 percent of the value of assets owned byemigrant households. In Pakistan 63 percent ofinvestment from remittances went into real estate,including agricultural land; in Turkey 58 percent ofmigrants' savings went into housing and land.Investment in equipment and financial assets hasbeen relatively small, although in Mexico and Tur-key some remittances have been invested infamily-owned commercial and manufacturingbusinesses. How remittances are used depends onthe same factors that determine other private con-sumption and investment decisions.

In short, emigration by the unskilled generallyleads to no loss in production, and if there is ascarcity premium on savings and foreign exchange(generated from remittances), then net benefitsfrom emigration are likely to be large. In fact, itmay even be beneficial for countries to facilitate

101

emigration by providing information to potentialemigrants and organizing recruitment on an offi-cial basis. Many countries, including Bangladesh,Korea, and the Philippines are in fact doing this.

However, emigration is not totally costless. Tem-porary migrants and their families often sufferlong periods of separation, although there is evi-dence that women left behind efficiently managethe household and family assets, including agricul-tural land. Emigration has led to a rapid rate ofmechanization in Yemen Arab Republic without asignificant increase in productivity, and neglect ofinfrastructure has led to a collapse of farm terraces.In Oman underground water channels have dete-riorated. Emigrant countries may also lose whenskilled and professional workers emigrate; these

form a large part of both temporary and perma-nent migration (see Box 5.6).

International trade: growth and limits

Trade offers more opportunities for reducing inter-national disparities and absorbing labor in devel-oping countries than does international migration,but the effects of increased trade on labor absorp-tion have, until now, been limited to only a fewcountries.

Unlike international migration, world trade hasgrown rapidly in the past three decades, at 6.7 per-cent a year, compared with less than 4 percent ayear in 1800-1913, and only 1.4 percent in 1913-50.Trade has provided developing countries with

Box 5.6 The brain drain and taxationBetween 1969 and 1979, the UnitedStates admitted nearly 500,000 profes-sional and technical workers. Three-quarters of them were from developingcountries, nearly 50 percent from Asia.During the 1970s they accounted fornearly 30 percent of the rise in theemployment of physicians and relatedpractitioners in the United States, for 12percent of the increase in engineers, andfor 8 percent of the increase in scientists.

Countries that import skilled man-power gain on two counts:

Since professional education is sub-sidized (about 45 percent of revenues ofinstitutions of higher learning in theUnited States, for example, come fromgovernment), receiving countries saveon such public subsidies.

Since countries can select immi-grants, they can adjust more quickly tochanges in demand. For example, theshare of physicians (and related prac-titioners) in professional immigrationinto the United States from developingcountries was 12 percent in 1969, rose tonearly 25 percent in 1973, but droppedto 11 percent in 1979 as the numberof domestically trained physiciansincreased.

Certain developing countries haveexperienced heavy brain drain. Some36 percent of temporarily recruitedmigrants from Sudan, for example, wereprofessionally and technically trained.

They included as much as 44 percent ofSudan's engineers, scientists, and medi-cal practitioners. During the 1970s pro-fessionals from the Philippines who emi-grated to the United States constituted12.3 percent of the increase in their num-bers at home; for Korea the figure was 10

percent. In Bangladesh professional andtechnical personnel constituted 17 per-cent of total emigration during 1976-78,and their departure is believed to havecontributed to a shortage of several typesof professionals. In other countries pro-fessional emigration has been large inabsolute numbers but not necessarily inrelative terms. Professional and technicalworkers who left Egypt for the UnitedStates during 1969-79 were less than 2percent of the increase in their numbersat home (although professional emigra-tion to the Middle East may be larger).Indian professional emigration to theUnited States formed about 1 percent ofthe stock in 1971.

The governments of many sendingcountries feel that emigration is harmfulbecause they subsidize the emigrants'education but lose the opportunity to taxtheir incomes. When skilled workersleave, unskilled workers may becomeunemployed. A country may also put ahigh social value on the services ofprofessional emigrants, such as doctorsand nurses, so that their emigrationinvolves a bigger loss than can be meas-

ured solely by the loss of the moneyvalue of their services. Emigration alsoprevents "internal diffusion' 'skilledpeople moving to backward areas withina country.

These costs are hard to quantify anddepend on each economy's institutionalfeatures. Some of them can be reducedor avoided by a change in the policies ofthe sending countries. There is, forexample, little justification in subsidizinghigher education when the beneficiariesare the richer elite, or when the probabil-ity of their emigrating is high. Govern-ments may also feel that they have aright to tax the incomes of skilled emi-grants, especially if emigrants remain cit-izens of their home country. The UnitedStates and Philippines, for example, taxtheir citizens when they live abroad.

There are few estimates of how muchrevenue would be raised by taxing emi-grants. If it is assumed, however, that 90percent of all professional immigrantsadmitted to the United States during1969-79 were still there in 1979, and thatwithin each major occupation theymatched the average earnings of Ameri-can workers, their total earnings in 1979would have been about $6 billion. A 10percent tax would thus have yielded$600 millionsome 13 percent of Offi-cial Development Assistance from theUnited States in that year.

102

extra jobs, directly in the export sector and indi-rectly as demand for inputs and services hasincreased. In Korea an estimated half a million jobsin 1970 (about 60 percent of them in manufactur-ing) were attributed directly and indirectly toexports. For all developing countries, however,manufacturing exports have added few jobs inrelation to the increases in the size of the laborforce. Most of the increase in manufacturedexports (and thus in total exports, since nonfuelprimary exports have grown less rapidly) has beenin the (now) middle-income countries. Between1965 and 1980 manufactured exports of all devel-oping countries increased by $128 billion, butmiddle-income oil-importing countries, with apopulation of 600 million (out of 3 billion in alldeveloping countries), accounted for 80 percent ofthat increase. Five countries-Brazil, Hong Kong,Korea, Singapore, and Yugoslavia, with 200 mil-lion people-accounted for 55 percent of theincrease. Manufactured exports of low-incomecountries, with a population of 2.2 billion in 1980,increased by only $14 billion, and those of low-income Africa by $0.5 billion (see Table 5.8). Totalexports of low-income countries also grew slowly,reflecting (with a few exceptions such as India)their dependence on primary exports, which grewat only 6.8 percent a year in volume. To the extentthat export revenues determine imports, primaryexporters have gained little, particularly in the faceof large increases in population.

EXPORTS AND EMPLOYMENT. Export success doesnot rely solely, or even necessarily, on a large laborforce and low wages. Of greater importance are anoutward-looking trade policy and a relativelyskilled labor force. As discussed in Chapters 2 and3, exports of many countries have been inhibited

TABLE 5.8

Export structure and human capital

by inward-looking trade policies and price distor-tions. Ironically, employment has suffered as aresult: there is now ample evidence that industriesgeared to import substitution create fewer jobsthan do export industries. Evidence from Brazil,Indonesia, and Thailand, for example, shows thatlabor employed per unit of value added was twiceas high in export industries as in import-substitu-tion industries. In Korea in 1968 manufacturedexports were 33 percent more labor intensive thandomestic manufactures, and 50 percent more laborintensive than import-competing industries. Theunskilled labor component in export industries isalso generally high-50 to 100 percent higher thanin import-competing ones.

The accumulation of human (and physical) capi-tal necessary to expand export capability is, asshown above, made more difficult if population isgrowing rapidly. Even simple manufactures suchas textiles and clothing (the commodities thatdeveloping countries typically export to start with)require skilled workers and versatile managers andentrepreneurs who can keep up with changingfashions and preferences. Modern textile plantstend to use expensive equipment: fixed capital peremployee in Indian firms using nonautomaticpower looms in 1963 was $1,600, more than seventimes the per capita income in that year.

Table 5.8 gives some indication of the differencesin human capital between low-income Africa, theleast successful exporter of manufactures, andmiddle-income oil importers, the most successful.In both groups the labor force has grown at about 2percent a year. But in 1960 the (now) middle-income countries had, on average, a higher adultliteracy rate, a higher index of human skills(defined as the secondary-school enrollment rateplus five times the enrollment rate in higher educa-

Defined as the secondary school-enrollment rate plus five times the enrollment rate in higher education.Based on a limited sample.Secondary-school enrollment rate.Excludes China.

103

Percentage ofinanufactu res in

Value of manu-factu red exports

(billions of dollars)Adult

lit eracg rateIndex of

human skills

Rate ofgrowth oflabor force

Exports GDP

Count rt group 1965 1980 1960 1981 1965 1980 1960 1980 1960 1979 (1960-81)

Low-income Africa 9.8 9.3 6.2 8.7 0.2 0.7 15 39 2.4 19.0 2.0Low-income Asia 37.4 41.8 13.0 170d 1.9 15.4 36 53 30.0' 58.4d 1.8Middle-income non-oil 23.0 51.6 22.0 25.0 4.3 108.9 58 72 38.0 109.0 2.2Industrialized countries 69.6 73.5 30.0 25.0 86.9 902.3 96 99 144.0 274.0 1.3

104

Box 5.7 Coping with rapid fertility decline: supporting the elderly in ChinaSlower population growth can helpdeveloping countries raise living stand-ards more quickly. But a rapid transitionto slow growth does require adjust-mentmost importantly in providingsecurity to the elderly. China's officialdemographic target calls for a populationof 1.2 billion by the year 2000, requiringthat the total fertility rate stay below 2 forthe rest of this century and that manycouples have only one child.

What are the economic implications oflowering fertility well below replacementlevel? During the remainder of this cen-tury there would be some economicadvantages but they are not dramatic. Ifthe total fertility rate is reduced to anaverage of 1.7 between 1985 and 2000(which would keep population to 1.2 bil-lion in 2000, according to World Bankprojections), the school-age populationwould decline to about two-thirds of its1980 level by the year 2000. The declineswould be greater in those regions wheremany couples have already pledged tohave no more than one child (See Box8.9). Savings could be allocated to some-what more rapid expansion of secondaryeducation.

But, in the long run, the decline in theproportion of young people would beoffset by a large increase in the propor-tion of the elderly. Assuming fertilityrises again to replacement level after2000, the proportion of persons in theworking-age group would not fall andthe overall dependency ratio would notincrease in the next century. But thestructure of dependency would be mark-edly differentwith the labor force Sup-porting the dependent old rather thanthe dependent young. In 1980, 60 per-cent of the population was of workingage (fifteen to sixty-four years) and only5 percent (about 45 million people) wasaged sixty-five and over. In the year2050when today's fifteen-year olds areaged eightythe proportion of the work-ing-age group would still be 61 percent,

but the proportion older than sixty-fivewould be 21 percent, by then about 308million people. This level of elderlydependency is unprecedented even inthe developed countries, where the pro-portion aged sixty-five and over rangesfrom about 8 percent in Japan to 18 per-cent in Sweden in 1980, and is about 11percent in the United States (see chart).

The much larger proportion of theelderly would be somewhat offset by adecline in the numbers of young people,from 36 percent in 1980 to 18 percent in2050. Young adults would then havefewer children to support, but, for thosefrom one-child families, no siblings tohelp support their parents. Consump-tion requirements of the elderly areabout double those of children, and atpresent few Chinese workers (only about15 percent of the labor force) are coveredby pension schemes; very few of thosecovered are in rural areas.

Population pyramids, China and United States, 1980 and 2050

The most severe burden will be createdby the large cohorts of the late 1960s andearly 1970s who are now beginning toenter the labor force. Pension funds tocover the retirement of these workers,with opportunities to earn interest andreinvest the substantial net income thatsuch funds would receive in their earlyyears, are urgent if present populationpolicies continue; indeed, they may wellbe necessary to sustain the desired fertil-ity decline. But they will be difficult tofinance at China's still relatively lowincome level. In developed countries,with income per worker ten to thirtytimes greater than in China, each workerin 1980 supported only half as manypensioners as a worker in China wouldhave to in the future if present demo-graphic goals are to be met. Yet even nowin developed countries, there are prob-lems with public financing of old-agesecurity systems.

Male

1980 __.

2050

United States

Female

1

IJ

1111 1111116 42024 6Percentage of total population

230 million, 1980291 million, 2050

Age group Chin

75+ I Male Female

70-74

60-64

50-54

40-44

30-34

20-24

10-14

0-4

I liii 111116 4 2 0 24 6Percentage of total population

980 million, 19801,450 million, 2050

tion), and a larger share of manufactures in outputthan low-income Africa. Low-income Asia occu-pies an intermediate position on most indicators;in 1960 it had a nearly comparable level of humanskills and a higher share of manufactures inexports than middle-income countries. The com-paratively slow increase in exports of manufac-tures from low-income Asia is as much attributableto inward-oriented policies in the two largest coun-tries (China and India), at least until recently, as toany lack of export capabilities.

These simple correlations should not be carriedtoo far. Even among the successful exporters,some countries have fared better and others worsethan their skill index in 1960 might suggest. Never-theless, these comparisons show that factors suchas a literate and educated labor force, accumula-tion of physical capital, and economic diversity areimportant for growth of manufactured exports.

Conclusions

Population and development are intertwined inmany ways, not all of them fully understood.Moreover, the effects of population growth mayvary widely, depending on the institutional, eco-nomic, cultural, and demographic setting. Slowpopulation growth itself requires new adjustmentsto support the growing burden of dependentelderly (see Box 5.7). The complexity of the subjectmakes it tempting to be agnostic about the conse-quences of rapid population growth. Neverthe-less, the evidence discussed above points over-whelmingly to the conclusion that populationgrowth at the rapid rates common in most of thedeveloping world slows development. At the fam-ily level, as Chapter 4 showed, high fertility canreduce the amount of time and money devoted toeach child's development. It makes it harder totackle poverty, because poor people tend to have

large families, and because they benefit less fromgovernment spending on the programs they usemosthealth and education, for examplewhenpublic services cannot keep pace with populationgrowth. At the societal level, as this chapter hasemphasized, it weakens macroeconomic perfor-mance by making it more difficult to finance theinvestments in education and infrastructure thatensure sustained economic growth.

Population growth eventually slows as parentsdecide to have fewer children. The factors behindparents' decisions, discussed in the next chapter,then work their way through to benefit society as awhole. But it does not follow that slower popula-tion growth will be an immediate panacea fordeveloping countries. Declines in fertility, forexample, will cut the growth of the labor force onlyafter fifteen to twenty years.

In the meantime, there are various nondemo-graphic measures by which countries can easethose development problems made more difficultby population growth. The adoption of trade andexchange rate policies that do not penalize laborand the dismantling of institutional barriers to cre-ating jobs would ease the employment problem.Pricing policies in agriculture and more resourcesallocated to rural credit, agricultural research andextension, and so forth, would increase agricul-tural output.

In short, policies to reduce population growthcan make an important contribution to develop-ment (especially in the long run), but their benefi-cial effects will be greatly diminished if they are notsupported by the right macroeconomic and sec-toral policies. At the same time, failure to addressthe population problem will itself reduce the set ofmacroeconomic and sectoral policies that are possi-ble, and permanently foreclose some long-rundevelopment options.

105