economic growth and human development - welcome to the department

Economic Growth and Human Development

GUSTAV RANISYale University, New Haven, Connecticut, USA

FRANCES STEWARTUniversity of Oxford, UK

and

ALEJANDRO RAMIREZ *

United Nations Development Program, New York, USA

Summary. Ð The connections between economic growth (EG) and human development (HD) formtwo chains. Crosscountry regressions show a signi®cant relationship in both directions, with publicexpenditures on health and education, notably female, especially important in the chain from EG toHD; and the investment rate and income distribution signi®cant in the HD to EG chain. This givesrise to virtuous or vicious cycles, with good or bad performance on HD and EG reinforcing eachother. Evidence over time has strong sequencing implications: countries initially favoring economicgrowth lapse into the vicious category, while those with good HD and poor EG sometimes moveinto the virtuous category. Where choice is necessary human development should be givensequencing priority. Ó 2000 Elsevier Science Ltd. All rights reserved.

Key words Ð human development, economic growth, income distribution, poverty, health,

education

1. INTRODUCTION

Human development has recently beenadvanced as the ultimate objective of humanactivity in place of economic growth. 1 Itsintellectual antecedents may be traced to theearlier basic needs approach of the ILO and theWorld Bank, as well as Sen's concept of capa-bilities. 2 Human development has been de®nedas enlarging people's choices in a way whichenables them to lead longer, healthier and fullerlives. 3 The de®nition of HD as ``enlargingpeople's choices'' is very broad. For thepurpose of exploring the links between HD andEG theoretically, and especially empirically, weneed to narrow it down. We shall consider theHD of a country as consisting of the health andeducation of its people, recognizing that this isvery much a reductionist interpretation.Clearly, there exists a strong connectionbetween economic growth (EG) and humandevelopment (HD). On the one hand, EG

provides the resources to permit sustainedimprovements in HD. On the other, improve-ments in the quality of the labor force are animportant contributor to EG. Yet, while thistwo-way relationship between HD and EG maynow be widely accepted, the speci®c factorslinking them have not been systematicallyexplored. Nor has the question of priorities inthe phasing of policy. The purpose of this paperis to sharpen understanding of the two-waylinks between HD and EG at both theoreticaland empirical levels. This in turn permits us toanalyze priorities in the phasing of policy andto examine the usual assumption that EG mustprecede progress on HD. 4

Section 2 identi®es the major links betweenEG and HD. Section 3 presents some empirical

World Development Vol. 28, No. 2, pp. 197±219, 2000Ó 2000 Elsevier Science Ltd. All rights reserved

Printed in Great Britain0305-750X/00/$ - see front matter

PII: S0305-750X(99)00131-Xwww.elsevier.com/locate/worlddev

* We are grateful for very helpful comments on an

earlier draft from three anonymous referees. Final

revision accepted: 19 July 1999.

197

crosscountry evidence on these links. Section 4develops a typology of country cases, somerepresenting the mutual enhancement of HDand EG and some demonstrating asymmetricperformance. The ®nal section investigates themovement of countries from one category toanother and re¯ects on the implications forpolicy.

2. THE TWO CHAINS

We view HD as the central objective ofhuman activity and economic growth aspotentially a very important instrument foradvancing it. At the same time, achievements inHD themselves can make a critical contributionto economic growth. There are thus twodistinct causal chains to be examined: one runsfrom EG to HD, as the resources from nationalincome are allocated to activities contributingto HD; the other runs from HD to EG, indi-cating how, in addition to being an end in itself,HD helps increase national income. The twochains are pictured in Figure 1.

(a) Chain A: from EG to HD

GNP contributes to HD mainly throughhousehold and government activity; civil soci-ety, e.g., through community organizations andother nongovernmental organizations (NGOs),also plays a role. The same level of GNP canlead to very di�erent performance on HDaccording to the allocation of GNP among andwithin these institutions and variations in theirbehavior.

Households' propensity to spend their after-tax income on items which contribute mostdirectly to the promotion of HD in poor coun-tries, e.g., food, potable water, education andhealth, varies, depending on such factors as thelevel and distribution of income across house-holds as well as on who controls the allocationof expenditure within households. In general,poor households spend a higher proportion oftheir incomes on HD items than those withhigher incomes, and similar results ¯ow fromgreater female control over household income.

When levels of poverty in a country are high,either because per capita income is low or badlydistributed, the expenditure of many house-holds on HD is bound to be low. While evi-dence 5 indicates that, in general, poverty isreduced with economic growth, the extent ofthe reduction varies greatly with the distribu-

tion of income and its change over time. 6 Theway in which growth translates into incomedistribution and poverty reduction depends onthe nature of the growth processÐin particular,the extent to which it is based on the generationof employment and on increasing ruralincomes, e.g., if the output mix is labor inten-sive and rural incomes rise rapidly incomedistribution is more likely to improve andpoverty reduction to occur than if growth isurban biased and capital intensive. 7

Expenditure on HD-related items is stronglya�ected by the rate of poverty reduction. Notsurprisingly, if poor households receive extraincome, they increase their food expenditure andcalorie consumption signi®cantly. 8 EmpiricalevidenceÐfor example, for Bolivia, Brazil,Chile, Cote d'Ivoire, Ghana, India, Indonesia,Pakistan, Philippines, Malaysia, Nicaragua andPeruÐalso indicates the positive e�ects offamily income change on child schooling. 9 Forexample, a crosscountry study of the determi-nants of secondary education found that moreegalitarian countries had higher secondaryenrollment rates. 10 One estimate suggests that ifthe distribution of income in Brazil were as equalas Malaysia's, school enrollments among poorchildren would be 40% higher. 11 While theevidence on the relations between income andhealth is less extensive, studies in Brazil, Chile,Cote d'Ivoire and Nicaragua suggest thathousehold income also has a signi®cant e�ect onthe demand for health, 12 some showing a muchhigher relative response for low than for high-income households. 13

Where women control cash income, itappears that expenditure patterns are gearedrelatively more toward HD inputs, such as foodand education. For example, among Gambianhouseholds, the larger the proportion of foodunder womenÕs control the larger householdcalorie consumption. 14 Similarly, in the Phil-ippines it has been shown that consumption ofcalories and proteins increases with the share ofincome accruing directly to women. 15 In theCote d'Ivoire, an increase in womenÕs share ofcash income was associated with signi®cantlyhigher spending on food and reduced spendingon alcohol and cigarettes. 16

Turning to the government, the allocation ofresources to improving HD is a function oftotal public sector expenditure, of how much ofthis ¯ows to the HD sectors, and of the way inwhich it is allocated within these sectors. Thiscan be expressed in the form of three ratios: 17

the public expenditure ratio, de®ned as the

WORLD DEVELOPMENT198

proportion of GNP spent by the various levelsof government; the HD-allocation ratio,de®ned as the proportion of total governmentexpenditure going to the HD-sectors; and,®nally, the HD priority ratio, de®ned as theproportion of total HD-sector expendituregoing to ``priority areas.'' Within the HD-sec-tors, some expenditures are clearly much moreproductive in terms of achieving advances inHD than others; for example, basic education,especially at an early stage of development, isgenerally recognized to have a larger impact onHD than tertiary education. But the precisede®nition of what constitutes a ``priority area''will inevitably vary according to a country'sstage of development, rendering this third ratiomore arbitrary and di�cult to measure than theother two. There are very large variations

across countries in each of these ratios, whichmeans that the same level of GNP may beassociated with very di�erent levels of govern-ment spending on HD priorities. 18

The underlying determinants of these threeratios are complex, but include the following: (i)the tax capacity of the system; (ii) the strengthof the demand for military expenditure and forother non-HD priorities of the government; (iii)the varying interplay between bureaucraticforces, vested interests and popular pressures. Itshould be noted that all three ratios are a�ectedby the extent of decentralization of government:country evidence suggests that real decentral-ization, i.e. devolution, tends to increase thetotal revenue available; it often raises the HD-allocation ratio; and it almost always improvesthe HD-priority ratios. 19

Figure 1. The HD-GNP cycle.

ECONOMIC GROWTH AND HUMAN DEVELOPMENT 199

The signi®cance of public expenditure choi-ces for improving HD is illustrated by acomparison between Kenya and Malawi. In the1980s, a similar proportion of national incomewent to public expenditure (27% in Kenya; 30%in Malawi) but Kenya had a signi®cantly highersocial allocation ratio (47% compared to 35%)and social priority ratio (34% compared to14%) so that the proportion of GDP goingdirectly to HD-improving priorities in Kenyawas over three times that of Malawi (5.1%compared to 1.5%). 20

The importance of economic growth forraising resources to promote HD is illustratedby a comparison between Botswana andSudan. In 1970 public expenditure on healthand education per person was similar in the twocountries ($96 in 1987 prices in Sudan, and $65in Botswana). During 1970±92 per capitaexpenditure increased more than seven times inBotswana, while remaining practicallyunchanged in Sudan. Hence, by 1992, Sudaneseexpenditure was less than a quarter of that inBotswana ($114 versus $466). This was not dueto di�erences in the public expenditure ratio,which was higher in Sudan during the 1970sand equal to Botswana's during the 1980s, butto the much faster income growth in Botswana.With a similar share of Botswana's GNP goingto health and education, real expenditure rosemuch faster there.

Finally, NGO or other civil society activity,on which information is more scattered, istypically heavily oriented toward HD objectives(e.g., generating incomes for the poor and onschools, nutrition and health projects).Resources are derived from private donationsand government, in each case including bothforeign and domestic sources. There areconsiderable variations in the extent, vitalityand e�ectiveness of NGO activities acrosscountries, depending on their history, culture,tax laws and actual or perceived governmentde®ciencies in providing services. In mostcontexts, NGOs play a supplemental or evenmarginal role, but in a few areasÐe.g., BRACand the Grameen Bank in Bangladesh, theHarambee Schools in Kenya and the``Comedores Populares'' in PeruÐthey appearto represent a major source of HD-enhance-ment. 21

Expenditures on HD inputs are clearly notobjectives in themselves, but rather constituteinstruments for achieving advances in variousdimensions of basic well-being. A furtherimportant link in Chain A, therefore, is the

e�ectiveness of these expenditures in raisingHD levels, i.e. what type of provision is mostproductive at what level of development, andhow di�erent combinations e�ect a change inHD. We shall call this link in the chain the``HD improvement function'' (HDIF); itresembles a production function in that itrelates the inputs into HD, such as publicexpenditure on health services or water, to theHD objective of achieving better health. Anillustration of the type of relationship repre-sented in the HDIF is whether and how far theprovision of safe water is complementary to orsubstitutes for education in contributing toimprovements in health. Another is whatcombination of family planning expenditure,improved nutrition and immunisation levelsmost e�ectively yield improvements in infantmortality. The relationships embodied in theHDIF are complex, depending on individualand community behavior and local knowledgeabout relevant technologies. Moreover, therelevant arguments of the HDIF are bound todi�er at di�erent stages of development.

Some aspects of the HDIF have been greatlyilluminated by detailed empirical work. 22 Anexample of the potential usefulness of animproved understanding of this productionfunction is the abundant evidence that educa-tion, especially female, tends to improve infantsurvival and nutrition. 23 Another study, forBrazil, showed that an increase in the nonlaborincome of women increased the probability ofchild survival by 20 times that of a comparableincrease in the nonlabor income of men. 24

Research in Ghana has shown that in ruralareas the provision of basic health services,including adequate drugs, increases child healthand survival signi®cantly, while the evidence isless clear on urban services. 25

Conditions liable to bring about animprovement in the e�ciency of the HDIFinclude better information about availabletechnologies and the appropriate combinationsof inputs; the generation of new and moree�ective technologies; and the enhancement ofthe motivation to make use of availableoptions, e.g., to send children to schools andclinics. Such motivation appears to be in¯u-enced by education levels, the power structurewithin the family, incentives and the magnitudeof the opportunity costs.

It should be clear from this discussion of thevarious links in the ED-HD chain that, ingeneral, we would expect important causalconnections to exist between the economy and


HD achievements, but these connections arenot automatic: the strength of the links in ChainA varies according to a large range of factors,including the structure of the economy, thedistribution of assets, and the policy choicesmade. The richness of the so-called socialcapital 26 of a society presumably also a�ectsthese choices and the strength of links at eachstage; when people act together to promotetheir well-being, when public morality is high,when the community monitors malfeasance,and when it participates extensively in publiclife, ceteris paribus, we can expect all the linksin Chain A to be stronger.

In summary, we would hypothesize from thisreview of the links in Chain A that theconnection between GNP to HD is likely to bestronger:

A.iÐthe lower the proportion of the popula-tion below the poverty line; for a given levelof GNP per capita, this means the moreequally income is distributed;A.iiÐthe more income households allocateto HD at a given income level; this may berelated both to the level of female educationand to female control over income within thehousehold.A.iiiÐthe higher the proportion of GNP de-voted to priority social expenditure by thegovernment;A.ivÐthe more e�ective the contribution ofsocial capital, including community organi-zations and other NGOs;A.vÐthe more e�cient the HDIF.

We shall test some of these hypotheses; otherswe do not test, either because of conceptualproblems or data constraints, notably A.iv. andA.v.

(b) Chain B: from HD to EG

Turning our attention to the second chain,from HD to EG, ample evidence suggests thatas people become healthier, better nourishedand educated they contribute more to economicgrowth, although some important dimensionsof HD, such as making the lives of the termi-nally ill tolerable, do not directly lead toenhanced productivity. 27 Higher levels of HD,in addition to being an end in themselves, a�ectthe economy through enhancing people'scapabilities and consequently their creativityand productivity. Clearly, the health and edu-cation of a population are among the maindeterminants of the composition and growthof output and exports, and constitute an

important ingredient in a system's capacity toborrow foreign technology e�ectively.

Speci®cally, (i) health, primary and second-ary education and nutrition raise the produc-tivity of workers, rural and urban; (ii)secondary education, including vocational,facilitates the acquisition of skills and mana-gerial capacity; (iii) tertiary education supportsthe development of basic science, the appro-priate selection of technology imports and thedomestic adaptation and development of tech-nologies; (iv) secondary and tertiary educationalso represent critical elements in the develop-ment of key institutions, of government, thelaw, the ®nancial system, among others, allessential for economic growth.

Empirical evidence at both micro and macrolevels lends support to the importance of theserelationships.

At a micro-level, numerous studies indicatethat increases in earnings are associated withadditional years of education, with the rate ofreturn varying with the level of education. 28

While social rates of return (which deduct thepublicly ®nanced costs of education but makeno other adjustments to the private returns) arebelow the private, they are still typically greaterthan returns to most physical investment. 29

These rate of return estimates are sometimesinterpreted as indicating the magnitude of theimpact of education on productivity; but thisview has also been challenged by the argumentthat education performs a signalling function,distinguishing among people of di�erent innateability, rather than itself raising productivity.But, investigations in rural Pakistan, and inurban Kenya and Tanzania, di�erentiatingbetween additional earnings due to cognitiveachievement and those due simply to schooling,showed that cognitive achievement accountedfor a high proportion of the extra earnings. 30

In agriculture, evidence suggests positivee�ects of education on productivity amongfarmers using modern technologies, but lessimpact, as might be expected, among thoseusing traditional methods. 31 In Thailand,farmers with four or more years of schoolingwere three times more likely to adopt fertilizerand other modern inputs than less educatedfarmers. 32 Similarly, in Nepal, the completionof at least seven years of schooling increasedproductivity in wheat by over a quarter, and inrice by 13%. 33

Education is also an important contributorto technological capability and technicalchange in industry. Statistical analysis of the


clothing and engineering industries in SriLanka, to cite just one example, showed thatthe skill and education levels of workers andentrepreneurs were positively related to the rateof technical change of the ®rm. 34 Educationalone, of course, cannot transform an econ-omy. The quantity and quality of investment,domestic and foreign, together with the overallpolicy environment, form other importantdeterminants of economic performance. Yet thelevel of human development has a bearing onthese factors too. The quality of policy-makingand of investment decisions is likely to bein¯uenced by the education of both policy-makers and managers; moreover, the volume ofboth domestic and foreign investment willprobably be larger when a system's humancapital supply is more plentiful.

Improved health and nutrition have beenshown to have direct e�ects on labor produc-tivity, especially among poorer individuals. 35

A range of labor productivity gains has beenobserved associated with calorie increases inpoor countries, 36 including studies of farmersin Sierra Leone, sugar cane workers in Guate-mala, and road construction workers inKenya. 37 In these cases productivity enhance-ment appears to follow fairly immediately ascurrent intakes of calories or micro-nutrientsare increased. In other cases the e�ects aremedium-run (as re¯ected in weight) or long-run(as re¯ected in height), based on evidence fromBangladesh, Brazil, India, Sri Lanka, and thePhilippines. 38 A longitudinal study of a sampleof children in Chile concluded that providingnutritional supplements to children to preventmalnutrition would generate bene®ts six toeight times the cost of the intervention in termsof additional productivity. 39 A similar study ofCali, Colombia found that a health and nutri-tion program increased the lifetime earnings ofindividuals to from 2.5 to 8.9 times those of anilliterate worker. 40

It is di�cult to capture the e�ects of ill-health(other than malnutrition) as there are fewaccurate estimates of the incidence of illness.But studies in both Ghana and Cote d'Ivoireshow the negative impact of morbidity, withmen who reported that their activities had beencurtailed by illness having lower hourly wagerates, reduced hours of work and a smallerprobability of being in the labor force. 41 Insome contexts the evidence even indicateslarger productivity e�ects arising from healthand nutrition than from formal schooling,although the impact of education has been

much more emphasized in the developmentliterature.

From a macro-perspective, the ``new growththeories'' aim to endogenize technical progressby incorporating some of these same e�ects,emphasizing education as well as learning andresearch and development (R&D). Accordingto Lucas (1988), for example, the higher thelevel of education of the workforce the higherthe overall productivity of capital because themore educated are more likely to innovate, andthus a�ect everyone's productivity. In othermodels a similar externality is generated as theincreased education of individuals raises notonly their own productivity but also that ofothers with whom they interact, so that totalproductivity increases as the average level ofeducation rises. 42 A complementary view isthat technical progress depends on the level ofR&D in an economy. By investing labor andcapital in R&D a ®rm is able to improve notonly its own pro®tability but also the produc-tivity of the ®rms which consume its output.Again, education plays a key role, both incontributing to R&D and via interactivelearning. 43 A number of empirical studies haveshown the positive e�ect of education oneconomic growth at a macro level, with its sizevarying according to the measure of educationand the particular macro growth model adop-ted. 44

The impact of education on the nature andgrowth of exports, which, in turn, a�ect theaggregate growth rate, is another way in whichhuman development in¯uences macro perfor-mance. The education and skills of a devel-oping country's labor force in¯uence thenature of its factor endowment and conse-quently the composition of its trade. It hasbeen argued that even ``unskilled'' workers ina modern factory normally need the literacy,numeracy and discipline which are acquired inprimary and lower secondary school. 45

Theoretical models incorporating skills andlearning as important determinants ofcomparative advantage have led to modi®ca-tions of the simple two factor Heckscher±Ohlin model, helping to explain the LeontiefParadox and the dramatic success in themanufactured export growth of some devel-oping countries, notably those of East Asia. 46

Investigations have shown a signi®cant posi-tive correlation between the growth of manu-factured exports and the growth of GDP,although serious identi®cation problemsremain. 47


There is also a positive feedback fromimproved education to greater income equalitywhich, in turn, is likely to favor higher rates ofgrowth. As education becomes more broadlybased, low-income people are better able toseek out economic opportunities. For example,a study of the relation between schooling,income inequality and poverty in 18 countriesof Latin America in the 1980s found that one-quarter of the variation in workers' incomeswas accounted for by variations in schoolingattainment; it concludes that ``clearly educationis the variable with the strongest impact onincome equality.'' 48 Another study suggestedthat a 1% increase in the labor force with atleast secondary education would increase theshare of income of the bottom 40 and 60% bybetween 6 and 15% respectively. 49 An investi-gation of the determinants of income distribu-tion in 36 countries found secondaryenrollment rates to be signi®cant. 50

Finally, education may a�ect per capitaincome growth via its impact on the denomi-nator, i.e. population growth. For example, astudy of 14 African countries for the mid-1980sshowed a negative correlation between femaleschooling and fertility in almost all countries,with primary education having a negativeimpact in about half the countries and nosigni®cant e�ects in the other half, whilesecondary education invariably reduced fertil-ity. 51 The three success countries in terms ofreduced fertility, Kenya, Botswana and Zimba-bwe, had the highest levels of female schoolingas well as the lowest child mortality rates. 52

As in Chain A, the strength of the variouslinks in Chain B varies considerably and thereis no automatic connection between animproved level of HD and increases in percapita GNP. It is not enough to create a largerpool of educated people; there must also beopportunities for them to be productivelyemployed or it might simply increase thenumber of educated unemployed. Relevant tothe demand-side are the savings and investmentrates, technology choice and the overall policysetting.

Although high rates of saving and investmentby themselves, as shown by past East Europeanexperience, do not guarantee high levels ofsustained growth, normally a positive rela-tionship between investment and growthprevails, with its strength depending on suchfactors as the policy environment, the quantityand quality of human resources, the availabilityof technology choices and the ¯exibility of the

institutional framework. As pointed out earlier,both domestic investment and direct foreigninvestment are in¯uenced by a country's HDlevelÐparticularly the education and skill levelsof the workforce. 53 But di�erences ineconomic growth across societies are again notonly due to the level of inputs but also by howe�ectively they are used and for what purpose.One important in¯uence is the incentives set byeconomic policy. Another derives from people'smotives and behavior patterns.

Income distribution again appears to beimportant in Chain B, as it was in Chain A.Recent empirical evidence suggests that thedistribution of assets and income has an e�ecton economic growth, with a more equaldistribution favoring higher rates of growth. 54

Among the explanations put forward for this,one is derived from the relationships in ChainA, i.e., that a more equal distribution of incomeimplies better nutrition and a stronger demandfor education and hence raises labor produc-tivity. Others derive from political economyconsiderations and relate to Chain B, e.g., thatan unequal distribution of income may beassociated with greater political and economicinstability, more likely to interrupt economicprogress. 55

In summary, the connection between HDand GNP, represented by Chain B, is likely tobe stronger:

B.iÐthe higher the investment rate;B.iiÐthe more equal the distribution of in-come;B.iiiÐthe more appropriate the economicpolicy setting.

Because of di�culties in de®ning and measur-ing hypothesis B.iii we shall con®ne our tests toB.i and B.ii.

3. EMPIRICAL FINDINGS

The discussion above led us to a set ofhypotheses about the links between HD andEG for both causal chainsÐfrom EG to HD(Chain A) and from HD to EG (Chain B),some of which we test with crosscountryregressions for 1960±92. Our sample consists of35 to 76 developing countries, according to theavailability of data for particular variables. Wegenerally use lags of the original variables asinstruments to reduce the simultaneity bias thatwould have resulted from applying ordinaryleast squares (OLS). Lagged values arereasonable candidates as instruments since the


correlation between the residuals in the twoperiods analyzed is not substantial.

(a) Chain A

For Chain A the variable chosen as a proxyfor achievement in human development was lifeexpectancy shortfall reduction, 1970±92, from amaximum of 85 years. 56

The explanatory variables selected were:Ðlagged GDP per capita growth rate (for1960±70) as a measure of overall EG;Ðsocial expenditure (de®ned as publicexpenditure on education and health) as apercentage of GDP for the whole period(1970±92), as well as lagged (1970±80). Thehypothesis advanced earlier was that HDimprovement would depend in part on theproportion of GDP devoted to social expen-diture; priority social expenditure data wereviewed as somewhat too arbitrary fordeployment here;Ðseveral measures of income distribution,i.e. income share of the bottom 20% or40%, 1960±92, and the ratio of income shareof the top to the bottom quintile, 1960±92;Ðfemale primary school gross enrollmentrate in 1965; this is a proxy for the changein the stock of educated females. The latteris likely to be associated with greater femalecontrol over household expenditure, thustending to improve the HDIF;Ðregional dummies, with East Asia allo-cated a zero value.

Table 1 summarizes the results for Chain A.GDP per capita growth proved signi®cant andquite strong in all of the equations, with highergrowth of per capita income leading to betterHD performance. According to equation (1), aone percentage point increase in the averagegrowth rate of GDP per capita is estimated toreduce life expectancy shortfall by more thanthree percentage points over the period. Thesocial expenditure ratio also proved signi®-cantly positive in all but one case. For everypercentage point increase in the average shareof GDP invested in health and education, whenlagged, the life expectancy shortfall decreasesby about 1.75 percentage points (equations (1)and (3)).

An interesting ®nding is the mechanismthrough which the social expenditure ratioseems to a�ect human development. Asregressions (4), (5) and (6) indicate, the femaleprimary enrollment rate for 1965 has a signi®-cant but small impact on the rate of improve-

ment of life expectancy. We attribute this to theimpact on household behavior of femaleincome, knowledge and control within thehousehold. Moreover, it should be noted thatwhen this variable is added, social expenditurebecomes less signi®cant, which suggests thatmuch of the impact of social expenditureappears to occur through its e�ect on femaleeducation. According to equations (4), (5) and(6), a 1% increase in the female primary grossenrollment rate is estimated to reduce the lifeexpectancy shortfall by 0.1%.

The income distribution variables run coun-ter to our expectations, i.e. a more equaldistribution does not seem to advance humandevelopment.

Both the African and Latin Americandummies are negative and signi®cant through-out, as we might have expected, given that thecomparator is highly successful East Asia. Ineach case the coe�cient is quite small. Theregional dummies include a number of region-speci®c features, including the level of percapita income.

(b) Chain B

For Chain B, the dependent variable chosenwas GDP per capita growth, 1970±92.

The explanatory variables selected were:Ðlog GDP per capita in 1960, to test forconvergence of income levels as countriesapproach high income levels;Ðinitial levels of HD, using three di�erentmeasures, log life expectancy in 1962, adultliteracy 1970±72, and a combined index oflife expectancy and literacy for 1970(HDI*); 57

Ðchanges in HD over time for two of themeasures of HD, the change in the log of lifeexpectancy, 1962±82; and HDI* shortfallreduction 1960±80; 58

Ðgross domestic investment as a % of GDPfor the period as a whole (1960±92);Ðincome distribution, lagged, (1960±70),using three alternative measures, the ratioof the income share of the top to bottomquintiles, the income share of the bottom40% of the population, and the income shareof the bottom 20%;Ðregional dummies, as in Chain A.

The results are summarized in Table 2. Ourmeasures of the initial level of human devel-opment invariably proved signi®cant, althoughwith low coe�cients; i.e. the initial level of lifeexpectancy in equations (10)±(12), as well as


Ta

ble

1.

Chain

AÐ

ord

inary

least

square

sre

gre

ssio

ns

dep

enden

tva

riable

:li

feex

pec

tancy

short

fall

reduct

ion

1970±92

Va

ria

ble

12

34

56

78

9

GD

P/n

Gro

wth

rate

19

60±

70

3.2

5�

1.6

6�

3.1

9�

1.2

7��

2.6

5�

2.7

0�

2.1

0�

2.1

4��

2.3

1�

(4.0

9)a

(3.6

1)

(4.0

6)

(2.5

6)

(3.1

7)

(3.1

7)

(2.8

2)

(2.5

2)

(2.9

5)

So

cial

Ex

pen

dit

ure

sa

sa

%o

fG

DP

19

70±

92

±1.0

3�

±±

±±

±±

±(2

.84)

So

cial

Ex

pen

dit

ure

sa

sa

%o

fG

DP

19

70±

80

1.7

3��

±1.7

5��

±1.3

61.5

1��

1.2

8��

1.5

4��

1.3

6��

(2.3

7)

(2.5

6)

(1.6

4)

(1.7

4)

(1.9

1)

(1.7

6)

(1.9

1)

Inco

me

share

of

bo

tto

m4

0%

19

60±

92

ÿ0:0

04

±±

ÿ0:0

04

±ÿ0:0

02

±±

±(ÿ

0.9

3)

(ÿ1.2

9)

(ÿ0.4

5)

Inco

me

share

of

bo

tto

m2

0%

19

60±

92

±±

±±

±±

±ÿ1:6

6��

ÿ1:8

8��

(ÿ1.7

6)

(ÿ2.2

6)

Ra

tio

of

inco

me

share

top

tob

ott

om

20

%1

96

0±

92

±±

0.0

02

±0.0

02

±0.0

05�

±±

(1.1

9)

(1.0

1)

(3.0

2)

Fem

ale

pri

mary

gro

ssen

roll

men

tra

te1

96

5±

±±

0.1

2��

0.1

0��

0.1

0��

±0.0

5±

(2.5

9)

(1.9

4)

(1.8

5)

(0.7

1)

La

tin

Am

eric

ad

um

my

±±

±±

±±

ÿ0:0

8��

ÿ0:0

8��

ÿ0:0

7��

(ÿ2.3

5)

(ÿ1.9

8)

(ÿ2.0

2)

Afr

ica

du

mm

y±

±±

±±

±ÿ0

.16�

ÿ0.1

2��

ÿ0.1

3�

(ÿ4.3

4)

(ÿ2.4

9)

(ÿ3.6

3)

So

uth

Asi

ad

um

my

±±

±±

±±

0.0

03

0.0

03

0.0

2(0

.06)

(0.5

0)

(0.3

9)

Mid

dle

East

du

mm

y±

±±

±±

±ÿ0:0

9ÿ0:0

7ÿ0:0

9(ÿ

1.0

1)

(ÿ0.6

7)

(ÿ0.9

7)

Inte

rcep

t0

.15

0.2

8�

0.0

60.2

3�

0.0

21.0

70.1

6�

0.2

5��

0.3

0�

(1.5

1)

(6.0

9)

(1.1

7)

(3.3

6)

(0.5

1)

(0.6

8)

(2.9

2)

(2.0

8)

(3.5

9)

#o

fo

bse

rva

tio

ns

41

76

41

54

38

38

41

38

41

Ad

j.R

-sq

ua

red

0.3

80.2

70.3

80.2

60.4

10.4

00.5

90.5

10.5

5

aT

-Sta

tist

ics

are

giv

enin

pare

nth

eses

.*

Sig

ni®

can

ta

t1

%.

**

Sig

ni®

can

ta

t5

%.

***

Sig

ni®

can

ta

t1

0%

.


Tab

le2

.C

ha

inB

Ðo

rdin

ary

least

square

sre

gre

ssio

ns

dep

enden

tva

riable

:ave

rage

real

GD

P/n

gro

wth

197

0±92

Va

ria

ble

10

11

12

13

14

15

16

17

Lo

gG

DP

/n1

96

0ÿ0

.01�

±ÿ0

.01�

ÿ0.0

1�

ÿ0.0

1��

ÿ0.0

09

ÿ0.0

1��

ÿ0.0

1��

(ÿ3

.01

)a(ÿ

3.0

4)

(ÿ2.8

7)

(ÿ1.8

7)

(ÿ1.4

2)

(ÿ2.3

8)

(ÿ2.5

5)

Ad

ult

lite

racy

rate

19

70

±7

2±

±±

0.0

3��

±±

0.0

3��

0.0

3��

(2.4

5)

(2.1

9)

(2.2

1)

Lo

gli

feex

pec

tan

cy1

96

70

.06�

0.0

3��

0.0

9�

±±

±±

±(3

.79

)(2

.09)

(3.9

0)

Gro

ssd

om

esti

cin

ves

tmen

ta

sa

%o

fG

DP

19

60±

92

0.1

2�

0.1

4��

0.1

2��

0.1

2��

0.1

3��

0.0

60.1

2��

0.1

2��

(3.8

3)

(2.7

2)

(2.3

7)

(1.9

9)

(2.2

6)

(1.0

6)

(2.0

7)

(2.0

9)

Inco

me

sha

reo

fb

ott

om

40

%1

96

0±

70

±±

0.0

01

±0.0

01��

±±

0.0

01��

(1.5

2)

(1.7

0)

(2.2

0)

Inco

me

sha

reo

fb

ott

om

20

%1

96

0±

70

±±

±±

±±

0.4

0��

±(2

.18)

Ra

tio

of

inco

me

sha

reo

fto

pto

bo

tto

mq

uin

tile

19

60

±7

0±

ÿ0:0

007��

±ÿ0:0

007��

±ÿ2:2

1±

±(ÿ

2.1

7)

(ÿ2.0

6)

(ÿ0.0

5)

HD

I*1

97

0±

±±

±0.0

6��

0.0

7��

±±

(2.3

8)

(2.5

8)

La

tin

Am

eric

ad

um

my

±±

±±

±ÿ0:0

2��

±±

(ÿ2.5

7)

Afr

ica

du

mm

y±

±±

±±

ÿ0:0

1��

±±

(ÿ1.9

0)

So

uth

Asi

ad

um

my

±±

±±

±ÿ0:0

03

±±

(ÿ0.3

1)

Mid

dle

Ea

std

um

my

±±

±±

±0.0

1±

±(0

.57)

Inte

rcep

tÿ0:1

7�

ÿ0:1

4��

ÿ0:2

8�

0.1

2��

0.0

30.0

40.0

70.0

7(ÿ

3.7

1)

(ÿ2.3

0)

(ÿ3.9

3)

(2.7

4)

(0.6

4)

(1.0

8)

(1.4

6)

(1.5

4)

#o

fo

bse

rva

tio

ns

73

38

36

24

35

35

24

24

Ad

j.R

-sq

ua

red

0.3

40.2

90.4

50.3

70.3

00.4

10.3

80.3

8C

onti

nued

nex

tpage


Ta

ble

2Ð

Co

nti

nued

Va

ria

ble

18

19

20

21

22

23

Lo

gli

feex

pec

tan

cy1

96

20

.06�

0.0

7�

0.0

5�

0.0

5��

0.0

8�

±(5

.48

)(3

.98)

(3.0

4)

(2.6

5)

(2.8

8)

Ch

an

ge

inth

elo

go

fli

feex

pec

tan

cy1

96

2±

82

0.1

3�

0.1

7��

0.1

4��

0.1

20.1

4��

±(3

.81

)(2

.48)

(2.0

1)

(1.6

5)

(1.7

8)

Gro

ssd

om

esti

cin

ves

tmen

ta

sa

%o

fG

DP

(19

70±

92

av

era

ge)

±±

0.1

2��

0.1

3��

0.0

60.1

3��

(2.4

4)

(2.3

9)

(1.1

7)

(2.4

2)

Inco

me

share

of

bo

tto

m4

0%

19

60±

70

±0.0

02�

0.0

02�

±±

0.0

02�

(2.8

3)

(3.1

6)

(2.9

6)

Ra

tio

of

inco

me

share

of

top

tob

ott

om

qu

inti

le1

96

0±

70

±±

±ÿ0:0

008��

ÿ0:0

002

±(ÿ

2.4

9)

(ÿ0.6

2)

HD

I*1

96

0±

±±

±±

ÿ0:0

007

(ÿ0.0

3)

HD

I*sh

ort

fall

red

uct

ion

19

60±

80

±±

±±

±0.0

9��

(2.0

8)

La

tin

Am

eric

ad

um

my

±±

±±

ÿ0:0

2��

±(ÿ

2.7

1)

Afr

ica

du

mm

y±

±±

±ÿ0:0

05

±(ÿ

0.5

7)

So

uth

Asi

ad

um

my

±±

±±

0.0

03

±(0

.29)

Mid

dle

East

du

mm

y±

±±

±0.0

04

±(0

.34)

Inte

rcep

tÿ0:2

7�

ÿ0:3

3�

ÿ0:2

9�

ÿ0:2

3�

ÿ0:3

3�

ÿ0:0

3�

(ÿ5

.37

)(ÿ

3.9

6)

(ÿ3.5

6)

(ÿ2.8

1)

(ÿ2.8

1)

(ÿ3.8

4)

#o

fo

bse

rva

tio

ns

79

39

38

38

38

37

Ad

j.R

-sq

ua

red

0.2

70.3

00.3

80.3

20.4

40.3

3

aT

-Sta

tist

ics

are

giv

enin

pa

ren

thes

es.

*S

ign

i®ca

nt

at

1%

.**

Sig

ni®

can

ta

t5

%.

***

Sig

ni®

can

ta

t1

0%

.


(18)±(22) were all highly signi®cant. Adultliteracy 1970±72 was signi®cant in equations(13), (16) and (17); and the HDI* for 1970 wassigni®cant in equations (14) and (15), all withlow coe�cients. The change in life expectancy,1962±82, was positive and signi®cant in allcases but one (see equations (18)±(22)). Thechange in HDI* shortfall reduction, 1960±80,was signi®cant in equation (23), but the initiallevel of HDI* was not signi®cant.

The domestic investment rate was alwayssigni®cant except when the regional dummieswere included (equations (15) and (22)). Thelagged income distribution variables virtuallyall gave results with the expected sign (i.e. amore equal income distribution is associatedwith higher economic growth), and were almostalways signi®cantÐif with very low coe�-cientsÐexcept when the regional dummies wereincluded. Moreover, income distribution isapparently more strongly related to GDPgrowth when changes as well as levels of humandevelopment, however measured, are included.The ratio of income shares of the top to bottomquintiles, 1960±70, showed signi®cance inequations (11), (13) and (21); the share of thebottom 40% was signi®cant in equations (14),(17), (19), (20) and (23), but the lagged incomedistribution variable was not signi®cant inequations (12), (15) and (22), although the signswere in the ``right'' direction. Regionaldummies for Latin America were signi®cantlynegative in both instances when deployed and,in one case, also for Africa. In all equations,except when the regional dummies were intro-duced (equation (15)), the initial level of GDPper capita was signi®cant, with a negative sign,indicating weak convergence, i.e. with a lowcoe�cient.

In summary, the two chains, taken as awhole, showed a signi®cantly positive e�ect ofeconomic growth on HD and a signi®cantlypositive e�ect of HD on economic growth.With respect to speci®c links in each of thechains, our ®ndings broadly con®rmed thetested hypotheses, except for income distribu-tion in Chain A. For Chain A, the higher socialexpenditure, the higher adult literacy, and thehigher the female education enrollment for agiven level of GNP per capita, the larger theimprovement in HD. The most surprising®nding, counter to our expectations, was that amore equal distribution of income did notimprove HD performance; indeed, in someequations the opposite result obtained. Oneexplanation of this may well be that we

restricted the de®nition of HD in our Chain Aregressions to life expectancy shortfall reduc-tion which is mainly a�ected by public expen-ditures on health and education, particularlyfemale, and less by household expenditures.For Chain B, the relationship between HD andeconomic growth was stronger the higher theinvestment rate and the more equally distrib-uted the income. The regional dummies weregenerally negative for both Africa and LatinAmerica in both chains but with small coe�-cients. Since, on average, Africa experienced adecline in GDP per capita over the period, thisindicates that the fall in per capita incomes wasnot proportionately translated into a slowdownor reversal in HD improvements. 59

4. VIRTUOUS AND VICIOUS CYCLESAND LOP-SIDED DEVELOPMENT

The existence of two chains linking HD andeconomic growth is thus strongly supported byboth our framework, drawing on micro andmacro studies in the literature, and our empir-ical results. This means that an economy maybe on a mutually reinforcing upward spiral,with high levels of HD leading to high growthand high growth in turn further promoting HD.Conversely, weak HD may result in low growthand consequently poor progress toward HDimprovement. The strength of the links in thetwo chains in¯uences the extent of mutualreinforcement between HD and EG, in eitherdirection.

Country performance can therefore beusefully classi®ed into four categories, virtuous,vicious and two types of lop-sidedness, i.e.lopsided with strong HD/weak growth (calledHD-lopsided); and lopsided with weak HD/strong growth (EG-lopsided). In the virtuouscycle case, good HD enhances growth, which inturn promotes HD, and so on. In the viciouscycle case, poor performance on HD tends tolead to poor growth performance which in turndepresses HD achievements, and so on. Thestronger the linkages in the two chains descri-bed above the more pronounced the cycle ofeconomic growth and HD, either in a positiveor dampening direction. Where linkages areweak, cases of lop-sided development mayoccur. On the one hand, good economic growthmay not bring about good HD, if, for example,there are such weak linkages as a low socialexpenditure ratio; on the other hand, good HDperformance may not generate good EG if


there is a dearth of complementary resourcesbecause of low investment rates. Such cases oflop-sided development are unlikely to persist.Either the weak partner in the cycle eventuallyacts as a brake on the other partner, leading toa vicious cycle case, or, if the linkages arestrengthened, possibly by policy change, avirtuous cycle case results.

One way of classifying countries into the fourcategories is to compare their performance onHD and EG with the average performance ofall developing countries. Figure 2 presents thisclassi®cation for 1960±92 for all developingcountries for which data were available. Thevertical and horizontal grid lines represent theaverage performance for the period, withcountries weighted by their populations in1992. Most developing countries appear aseither virtuous (NE quadrant), or vicious (SWquadrant); a signi®cant number showed anHD-lopsided pattern, and very few an EG-lopsided one. A strong regional pattern emer-ges, with East Asia heavily represented in thevirtuous cycle case (seven out of the eight casesare East or South East Asian, the eighth beingBotswana). Of the 37 countries in the viciouscycle quadrant, 21 are from sub-SaharanAfrica, nine from Latin America. LatinAmerica is also strongly represented in the HD-

lopsided quadrant, i.e. 10 of the 13 are fromthat region. In the EG-lopsided category thereare just four countriesÐEgypt, Pakistan,Mauritius and Lesotho.

The important issue for policy purposes ishow a country may move towards inclusion inthe virtuous cycle category. Much can belearned about this by looking at the ways inwhich countries moved across categories overtime. Taking the movements of countries overthe three decades 1960±70, 1970±80 and 1980±92 (see Table 3 and Figure 3), we ®nd thefollowing:

ÐOver half the countries in the vicious-cyclecategory in 1960±70, i.e. 18 out of 35, re-mained in that category throughout. Mostof these countries were in sub-Saharan Afri-ca, which started with very low levels of HD,handicapping their growth potential; theirlow growth rates and, subsequently, the debtcrisis prevented them from generating the re-sources necessary for improvements in HD.ÐSix countries moved from vicious cycle toEG-lopsided between the 1960s and the1970s, but four of them fell back to the vi-cious cycle category in the 1980s. Threemoved from vicious to HD-lopsided, includ-ing Honduras, Algeria and Madagascar, andonly Madagascar returned to the vicious

Figure 2. Classi®cation of country performance (1960±92). Note: The horizontal and vertical lines de®ning the fourquadrants represent developing country averages weighted by population.


Table 3. Virtuous, vicious and lop-sided performance 1960±92

1960±70 1970±80 1980±92

AfricaBenin Vicious Vicious ViciousBotswana Vicious Virtuous VirtuousBurkina Faso Vicious Vicious ViciousBurundi Vicious Vicious ViciousCameroon Vicious EG lop-sided ViciousCentral African Republic Vicious Vicious ViciousChad Vicious Vicious ViciousCongo Vicious EG lop-sided ViciousCote d'Ivoire EG lop-sided Vicious ViciousGabon EG lop-sided Vicious ViciousGhana Vicious Vicious ViciousKenya Vicious Virtuous ViciousLesotho Virtuous EG lop-sided ViciousMadagascar Vicious HD lop-sided ViciousMalawi Vicious EG lop-sided ViciousMali Vicious Vicious ViciousMauritius HD lop-sided EG lop-sided EG lop-sidedNiger Vicious Vicious ViciousNigeria Vicious Vicious ViciousRwanda Vicious Vicious ViciousSenegal Vicious Vicious ViciousSierra Leone EG lop-sided Vicious ViciousSouth Africa Virtuous Vicious ViciousSudan Vicious Vicious ViciousTanzania Vicious Vicious ViciousZaire Vicious Vicious ViciousZimbabwe Vicious Vicious Vicious

Latin America & CaribbeanArgentina Vicious Vicious HD lop-sidedBarbados Virtuous HD lop-sided HD lop-sidedBolivia Vicious Vicious HD lop-sidedBrazil EG lop-sided EG lop-sided ViciousChile HD lop-sided HD lop-sided VirtuousColombia HD lop-sided Virtuous HD lop-sidedCosta Rica HD lop-sided HD lop-sided HD lop-sidedDominican Republic HD lop-sided EG lop-sided ViciousEl Salvador HD lop-sided Vicious HD lop-sidedGuatemala HD lop-sided EG lop-sided ViciousHaiti Vicious Vicious ViciousHonduras Vicious HD lop-sided HD lop-sidedJamaica Virtuous Vicious ViciousMexico Virtuous Virtuous HD lop-sidedNicaragua Virtuous Vicious HD lop-sidedPanama Virtuous Virtuous HD lop-sided

Latin America & CaribbeanParaguay Vicious EG lop-sided ViciousPeru HD lop-sided Vicious HD lop-sidedTogo EG lop-sided Vicious ViciousTrinidad & Tobago Vicious EG lop-sided HD lop-sidedUruguay Vicious Vicious HD lop-sidedVenezuala HD lop-sided HD lop-sided Vicious

South AsiaIndia Vicious Vicious EG lop-sidedNepal Vicious Vicious ViciousPakistan EG lop-sided Vicious EG lop-sidedSri Lanka Vicious Virtuous VirtuousBangladesh Vicious Vicious Vicious

Continued next page


cycle category. Kenya, which moved fromvicious to virtuous in the 1970s, also subse-quently fell back to vicious. Only two coun-tries managed to move from the vicious tovirtuous category on a sustained basisÐi.e.Sri Lanka and Botswana.

ÐOf the eight countries which were EG-lop-sided in 1960±70, none stayed in that cate-gory throughout, but all moved into thevicious category. OneÐPakistanÐrevertedto EG-lopsided in the 1980s. Brazil andEgypt enjoyed relatively fast growth during

Table 3Ðcontinued

1960±70 1970±80 1980±92

East AsiaChina HD lop-sided Virtuous VirtuousHong Kong Virtuous Virtuous VirtuousIndonesia HD lop-sided Virtuous VirtuousKorea Republic Virtuous Virtuous VirtuousMalaysia Virtuous Virtuous VirtuousMyanmar HD lop-sided Vicious ViciousPhilippines HD lop-sided EG lop-sided ViciousSingapore Virtuous Virtuous VirtuousThailand Virtuous Virtuous Virtuous

Middle EastAlgeria Vicious HD lop-sided HD lop-sidedEgypt EG lop-sided EG lop-sided ViciousMorocco Vicious EG lop-sided HD lop-sidedTurkey Virtuous HD lop-sided HD lop-sided

Figure 3. Country HD-EG quadrant changes over three decades. Note: The country movements indicate the quadrantsin which countries are placed over the three decades but not their actual location relative to the axes.


the 1960s and 1970s (over 3% in the 1960sand about 6% in the 1970s) but did not uti-lize this opportunity to improve the levels ofHD substantially. In the case of Brazil thehighly unequal income distribution (with aGini of 0.634, one of the worst in the world)was one reason why growth did not translateinto HD improvements. In both Pakistanand Egypt, public expenditure on healthand education was low, partly due to heavyexpenditure on the military, while Pakistan'sHD performance su�ered especially fromdiscrimination against females.ÐOf the 13 HD-lopsided countries in the1960s, only Costa Rica stayed in that cate-gory throughout: four moved into a virtuouscycleÐChile, China, Colombia and Indone-sia (Colombia later falling back to HD-lop-sided). In these cases, early progress inhuman development meant that they wereable to take advantage of policy reformsfor generating growth. Egalitarian incomedistribution also assisted the movement to-ward a virtuous cycle. Four moved initiallyfrom HD-lopsided into the vicious category(Venezuela, Myanmar, Peru, and El Salva-dor 60Ðthe latter two moving back intoHD-lopsidedness in the 1980s). 61 ThreeÐDominican Republic, Guatemala and thePhilippinesÐinitially moved to EG-lopsidedand subsequently fell back into the viciouscategory. Among the reasons for the failureto move into high economic growth werethe debt situation, poor economic policiesand internal disturbances. Consequently,they were unable to maintain the rate of pro-gress in HD because of slow economicgrowth.ÐThirteen countries were in the virtuous cy-cle category in the 1960s; ®ve retained thisposition throughout; and ®ve fell back tothe HD-lopsided and three to the viciouscategory. 62 Mostly, the countries that fellback were subject to the depressing e�ectsof the 1980s debt crisis on economic growth.

It is important to note that lopsidedness wasa temporary condition in all cases except CostaRica. 63 Our most signi®cant ®nding is that,while HD-lopsidedness permitted movementtowards a virtuous cycle (occurring in about athird of the cases), in the case of EG-lopsided-ness, all the cases reverted to a vicious cycle.Very few countries managed to go directly fromvicious to virtuous; but some succeeded inmoving to HD-lopsided, from where it waspossible to move into the virtuous category.

Our analysis suggests that it is not possible tomove to virtuous via EG-lopsidedness, as thisproved a dead end.

Hence the process we need to examine moreclosely is that leading to a movement fromvicious to HD-lopsided, from HD-lopsided tovirtuous, or the very unusual case, undoubtedlydi�cult, of taking the direct route from viciousto virtuous.

Our earlier review of the links in the twochains suggests that to move from vicious toHD-lopsided one needs to strengthen the linksin Chain A, which may be achieved by adopt-ing some of the following policies:

Ðthose leading to a shift in resource alloca-tion toward education and health services,especially those serving the majority of thepeople, as apparently occurred in Argentinawith enhanced decentralization.Ðthose generating a more equitable incomedistribution (for example, through land andtax reform or a move toward a moreemployment-intensive pattern of output).Algeria is an example, with land reform inthe 1960s and some large-scale employmentprograms in the 1970s. 64

Ðthose providing extensive opportunitiesfor the unemployed, for example Bolivia'sSocial Fund in the 1980s, which receivedsubstantial support from donors.

Movement from the HD-lopsided to thevirtuous category requires strengthening thelinks in Chain B by, for example:

Ðtaking advantage of an improved HD topromote economic growth through policyreform;Ðincreasing the investment rate, possiblyassisted from the outside;Ðimproving the distribution of income.

Chile is an example where strong economicreforms, combined with a high level of HD,accounted for improved economic growth.Relatively good income distribution in China,along with policy reform and very high invest-ment rates, led to accelerated economic growth.In Indonesia, a similar process was heavilysupported by external capital ¯ows.

We do not wish to argue here that anyparticular set of policies would achieve aparticular movement across categories; ratherwe want to emphasize an important conclusionabout the sequencing of policy change, i.e. thatHD must be strengthened before a virtuouscycle can be attained. Policy reforms whichfocus only on economic growth are unlikely tosucceed. Countries in a virtuous cycle category


may well slip back into HD-lopsidedness, if, forsome reason, growth slows down, but as longas HD stays high such cases have a goodchance of resuming their virtuous cycle pattern.

Whenever either or both chains appear to beweak, leading to lop-sided or vicious cycles, it isimportant to identify where the weak links areand what the appropriate policies might be tostrengthen such links. These policies must,moreover, be viewed in an evolutionarycontext. Even countries initially successful inboth HD and EG will need to change theirpolicies as development proceeds in order tosustain their success. In an early phase, forexample, priority should be given to primaryeducation and some comprehensive healthinterventions, both from the perspective ofimproving HD and that of increasing economicgrowth. At a later stage, the roles of science andtechnology institutions and higher educationbecome more important for strengtheningChain B, while, along Chain A, hospitalsassume a greater role than before.

In summary, we have demonstrated theexistence of an iterative process between theultimate objectiveÐimprovements in HDÐand economic growth as a necessary but notsu�cient condition for achieving suchimprovements. Moreover, by investigating therelative importance of various links connect-ing HD and EG we have identi®ed thedirection policy might take to strengthen suchlinks. One important conclusion concerns thedesirable phasing of policy change. Economicand social policy has tended to focus priorityon getting the economic fundamentals ``right''as a necessary precondition for economicgrowth, while arguing that HD improvementmust await such economic growth. Our®ndings do not deny the importance ofeconomic reforms, but emphasize that afocus on HD must be included from thebeginning of any reform program. Economicgrowth itself will not be sustained unlesspreceded or accompanied by improvements inHD.

NOTES

1. See especially the UNDP's Human Development

Reports, starting in 1990.

2. See, e.g., Sen (1984); Streeten, Burki, ul Haq, Hicks

and Stewart (1981); Fei, Ranis and Stewart (1985).

3. The ®rst UNDP Human Development Report stated

that: ``The basic objective of development is to create an

enabling environment for people to enjoy long, healthy

and creative lives'' (UNDP, 1990, p. 9), and de®ned

human development as ``a process of enlarging people's

choices'' (p. 10).

4. One notable exception is Adelman and Morris

(1973) who argued that educational investment should

be given sequencing priority.

5. E.g. Fields (1989); Deininger and Squire (1996).

6. Bruno, Ravallion and Squire (1995), for example, in

a study of 20 developing countries, show that a 10%

increase in average income per capita, 1984±93, was

associated with a 20% fall in the proportion of people

living on less than $1 a day. See also Ravallion and Datt

(1991).

7. Lipton (1977); Ranis (1979); Stewart (1977); Lipton

and Ravallion (1995).

8. Strauss and Thomas (1995). Of 38 studies in

di�erent countries one-third indicate that at least one-

half of additional income is spent in this way. See also

Bouis and Haddad (1992).

9. E.g., Alderman, Behrman, Khan, Ross and Sabot

(1995, 1996a); Behrman and Wolfe (1987a,b); Birdsall

(1985); Deolalikar (1993); King and Lillard (1987).

10. Williamson (1993).

11. Birdsall, Ross and Sabot (1995).

12. Blau (1986); Harbert and Scandizzo (1982);

Thomas, Strauss and Hanriques (1990, 1991); Thomas,

Lavy and Strauss (1992).

13. Gertler, Locay and Sanderson (1987).

14. von Braun (1988).

15. Garcia (1990).

16. Hoddinott and Haddad (1991).

17. See UNDP (1991).

18. See UNDP (1991) Chapter 3, and UNDP (1996)

Chapter 3.


19. Decentralization of publicly provided services

recently has been introduced in a wide range of

countries. Tentative conclusions about their e�ectiveness

are mixed, with apparent relative success in promoting

e�ciency and contributing to HD in Indonesia, Malay-

sia, Chile and Karnataka in India, but less so in

Argentina, Bangladesh and Brazil. Mostly, local govern-

ments have been severely constrained in their ability to

raise taxes as well as in the freedom of allocative

decision-making, and full democratic devolution has

been rare. See Behrman (1995b); Prud'homme (1995);

Klugman (1995); Ranis and Stewart (1994); Tanzi

(1995).

20. UNDP (1996, p. 71). These calculations adopt a

narrow de®nition of social priority expenditure, includ-

ing pre-primary and ®rst-level education plus primary

health care only.

21. Riddell, Robinson deConinck, Muir and White

(1995).

22. See, e.g., the review by Strauss and Thomas (1995).

23. See e.g., Rosenzweig and Schultz (1982); Wolfe and

Behrman (1987); Barrera (1990).

24. Thomas (1990).

25. Lavy, Strauss, Thomas, deVreyer (1995).

26. A term used broadly to describe civil society

networksÐsee Coleman (1988); North (1990) and Helli-

well and Putnam (1995) for the development of the

concept and some empirical evidence. Tendler and

Freedheim (1994) provide evidence.

27. This clearly does not detract from the intrinsic

value of improving the lives of the unemployable, who

may account for a signi®cant proportion of the popu-

lation.

28. See surveys in Behrman (1990a,b,c, 1995a); Behr-

man and Deolalikar (1988), King and Hill (1993); King

and Bellew (1988); Psacharopolous (1994); Schultz

(1988, 1993a,b); Strauss and Thomas (1995).

29. Following the work of Psacharopolous, the conclu-

sion that the returns to education are highest for primary

education has become part of the accepted wisdom; but

in fact this is by no means always the caseÐsee, e.g.,

Schultz (1993a,b). However, since secondary education

necessarily requires prior attendance at primary school,

the usual measures of returns to primary education

which do not allow for this may understate the total

returns.

30. Boissiere, Knight and Sabot (1985); Alderman,

Behrman, Khan, Ross and Sabot (1996b).

31. Schultz (1975); Welch (1970); Rosenzweig (1995);

Foster and Rosenzweig (1994); Behrman, Rosenzweig

and Vashishtha (1995).

32. Birdsall (1993).

33. Jamison and Moock (1984). A similar level of

education was estimated to increase farm productivity

by 10% or more, according to studies in India and

Pakistan (Azher, 1991; Butt, 1984; Duraisamy, 1992).

Studies in Malaysia, Ghana and Peru show that, on

average, an extra year of schooling of a farmer is

associated with an annual increase in output of 2±5%

(Birdsall, 1993). Examination of the early stages of

India's Green Revolution indicates that the bene®ts

from schooling arose more from conferring on farmers

the capacity for learning from their own experience or

from those of others, rather than from providing them

with an initial information advantage (Foster & Rose-

nzweig, 1995).

34. Deraniyagala (1995).

35. See Behrman and Deolalikar (1987) and surveys in

Behrman (1993, 1996).

36. See Cornia and Stewart (1995).

37. Strauss (1986); Immink and Viteri (1981); Wolge-

muth, Latham, Hall and Crompton (1982).

38. Behrman and Deolalikar (1989); Behrman and

Lavy (1995); Deolalikar (1988); Foster and Rosenzweig

(1993); Haddad and Bouis (1991); Pitt, Rosenzweig and

Hassan (1990); Sahn and Alderman (1988); Strauss and

Thomas (1996).

39. Selowsky and Taylor (1973).

40. Selowsky (1981).

41. Schultz and Tansel (1993).

42. Perotti (1993).

43. See Romer (1990); Grossman and Helpman (1991);

Gemmell (1995).


44. E.g., Barro (1991); Barro and Lee (1993, 1994).

45. Wood (1994); Owens and Wood (1995).

46. See Kenen (1965); Keesing (1966); Pack and

Westphal (1986); Leamer (1993).

47. E.g. Michaely (1977); Krueger (1978); Ram (1985);

Rana (1988); Edwards (1993).

48. Psacharopolous, Morley, Fiszbein, Lee and Wood

(1992, p. 48).

49. Bourgignon and Morrisson (1990).

50. Bourgignon (1995).

51. See Birdsall, Ross and Sabot (1995); Jayaraman

(1995); Strauss and Thomas (1995); Thomas, Strauss

and Henriques (1991); Behrman and Wolfe (1987a).

52. Ainsworth, Beegle and Nyamete (1995).

53. OECD/DAC (1995); World Bank (1995a).

54. Alesina and Rodrik (1994); Alesina and Perotti

(1994); Persson and Tabellini (1994); Birdsall et al.

(1995).

55. Alesina and Rodrik (1994).

56. As incorporated in the Human Development

IndexÐsee UNDP (1995). We also tried regressions

with changes in adult literacy rates and a combined,

equally weighted, measure of the two. The results were

similar but the number of observations more limited.

57. HDI* is a modi®ed version of the Human Devel-

opment Index of UNDP, including only the nonincome

components, i.e. educational attainment as measured by

a combination of adult literacy (2/3 weight), mean years

of schooling (1/3 weight), and longevity as measured by

life expectancy at birth.

58. When changes over time were introduced, the dates

of the initial level of HD were also changed, so that log

life expectancy was for 1962 and HDI* for 1960.

59. One explanation is that there were important

speci®c interventions (notably immunization) that

a�ected HD positively throughout the world in the

1970s and 1980s. In a sense these weakened the links in

Chain A by rendering HD improvements more indepen-

dent of incomes. In Africa, this would translate into a

reduced negative impact on HD of falling incomes.

60. Iraq made the same move between the 1960s and

1970s, but data are not available for the later period, when

con¯ict is likely to have damaged both HD and growth.

61. The Philippines moved from HD-lopsided to EG-

lopsided and then returned to HD-lopsided in the 1980s.

62. Lesotho moved from virtuous to vicious by way of

EG-lopsided.

63. One of the explanations of why Costa Rica was

able to sustain HD achievements despite low economic

growth resides in its early, strong and sustained

commitment to HD, exempli®ed by abolition of its

army in 1948 and its heavy investment (at 10% of GDP)

on health and education during 1970±92.

64. El-Ghonemy (1990).

REFERENCES

Adelman, I., & Morris, C. T. (1973). Economic growthand social equity in developing countries. Stanford,CA: Stanford University Press.

Ainsworth, M., Beegle, K., & Nyamete, A. (1995). Theimpact of female schooling on fertility and contracep-tive use: a study of fourteen sub-Saharan Africancountries. SMS Working Papers, 110, World Bank,Washington, DC.

Alderman, H., Behrman, J., Khan, S., Ross, D., &Sabot, R. (1996b). The returns to endogenous humancapital in Pakistan's rural wage market. OxfordBulletin of Economics and Statistics.

Alderman, H., Behrman, J., Khan, S., Ross, D., &Sabot, R. (1996a). The income gap in cognitive skillsin rural Pakistan. Economic Development and Cultur-al Change, 44.

Alderman, H., Behrman, J. R., Ross, D., & Sabot, R.(1995). Decomposing the gender gap in cognitiveskills in a poor rural economy. Journal of HumanResources, 31.

Alesina, A., & Perotti, R. (1994). The political economyof growth: a critical survey of the recent literature.The World Bank Economic Review, 8.

Alesina, A., & Rodrik, D. (1994). Distributive politicsand economic growth. Quarterly Journal of Econom-ics, 109.

Azher, R. A. (1991). Education and technical e�ciencyduring the green revolution in Pakistan. EconomicDevelopment and Cultural Change, 39.

Barro, R. J., & Lee, J. -W. (1993). Internationalcomparison of educational attainment. Journal ofMonetary Economics, 32.


Barro, R. J., & Lee, J. -W. (1994). Losers and winners ineconomic growth. In Proceedings of the World Bankannual conference on development economics, 1993.Washington, DC: IBRD.

Barro, R. (1991). Economic growth in a cross-section ofcountries. Quarterly Journal of Economics, 106.

Barrera, A. (1990). The role of maternal schooling andits interaction with public health programs in childhealth production. Journal of Development Econom-ics, 32 (1), 69±91.

Behrman, J. R. (1990b). The action of human resourcesand poverty on one another: what we have yet to learn.LSMS Working Paper 74. Washington, DC: WorldBank.

Behrman, J. R. (1990c). Women's schooling and nonmar-ket productivity: a survey and a reappraisal. Mimeo,University of of Pennsylvania, Philadelphia.

Behrman, J. R. (1993a). The economic rationale forinvesting in nutrition in developing countries. WorldDevelopment, 21.

Behrman, J. R. (1995a). The impact of distributivepolicies, governmental expenditure patterns and decen-tralization on human resources. Philadelphia: Mimeo,University of Pennsylvania.

Behrman, J. R. (1995b). Intra-household allocation ofresources: is there a gender bias? In Sex di�erentialsin infant and child mortality. Population Division,United Nations, New York.

Behrman, J. R. (1996). Impact of health and nutrition oneducation. World Bank Research Observer, 11.

Behrman, J. R., & Deolalikar, A. B. (1987). Willdeveloping country nutrition improve with income?a case study for rural south India. Journal of PoliticalEconomy, 95.

Behrman, J. R., & Deolalikar, A. B. (1988). Health andnutrition. In H. B. Chenery & T. N. Srinivasan.Handbook of development economics, vol. 1. Amster-dam: North Holland.

Behrman, J. R., & Deolalikar, A. B. (1989). Wages andlabor supply in rural India: the role of health,nutrition and seasonality. In D.E. Sahn, Causes andimplications of seasonal variability in household foodsecurity. Baltimore, MD: The Johns HopkinsUniversity Press.

Behrman, J.R., & Lavy, V. (1995). Production functions,input allocations and unobservables: the case of childhealth and schooling success. Mimeo, University ofPennsylvania, Philadelphia.

Behrman, J. R., & Wolfe, B. L. (1987a). How doesmotherÕs schooling a�ect the familyÕs health nutritionmedical care usage and household? Journal ofEconometrics, 36.

Behrman, J. R., & Wolfe, B. L. (1987b). Investments inschooling in two generations in pre-revolutionaryNicaragua: the roles of family background andschool supply. Journal of Development Economics,27.

Behrman, J. R. (1990a). Human resource led develop-ment: review of issues and development. ARTEP/ILO,New Delhi.

Behrman, J. R., Rosenzweig, M. R., & Vashishtha, P.(1995). Location-speci®c technical change, humancapital and local economic development: The Indian

Green Revolution experience. In H. Siebert, Locat-ional competition in the world economy. Kiel: KielInstitute of World Economics.

Birdsall, N. (1985). Public inputs and child schooling inBrazil. Journal of development Economics, 18.

Birdsall, N. (1993). Social development is economicdevelopment. World Bank policy research workingpapers, WPS, 1123, World Bank, Washington, DC.

Birdsall, N., Ross, D., & Sabot, R. (1995). Inequalityand growth reconsidered: lessons from East Asia.World Bank Economic Review, 9.

Blau, D. (1986). Fertility child nutrition and childmortality in Nicaragua: an economic analysis ofinterrelationships. Journal of Developing Areas, 20.

Boissiere, M., Knight, J. B., & Sabot, R. H. (1985).Earnings schooling ability and cognitive skills.American Economic Review, 75.

Bouis, H. E., & Haddad, L. J. (1992). Are estimates ofcalorie-income elasticities too high? A recalibrationof the plausible range. Journal of DevelopmentEconomics, 39.

Bourgignon, F. (1995). Equity and economic growth:Permanent questions and changing answers? Back-ground paper prepared for the 1996 Human Devel-opment Report, UNDP, New York.

Bourguignon, F., & Morrisson, C. (1990). Incomedistribution development and foreign trade: a cross-sectional analysis. European Economic Review, 34.

Bruno, M., Ravallion, M., & Squire, L. (1995). Equityand growth in developing countries: old and newperspectives on the policy issues. Paper presented atInternational Monetary Fund Conference on IncomeDistribution and Sustainable Growth, Washington,DC, June 1±2.

Butt, M. S. (1984). Education and farm productivity inPakistan. Pakistan Journal of Applied Economics, 3.

Coleman, J. (1988). Social capital in the creation ofhuman capital. American Journal of Sociology, 94(suppl. S95-S120).

Cornia, G.A., & Stewart, F. (1995). Two errors oftargeting. In F. Stewart, Adjustment and poverty:options and choices. London: Routledge.

Deininger, K., & Squire, L. (1996). A new data setmeasuring income inequality. World Bank EconomicReview, 10.

Deolalikar, A. B. (1988). Nutrition and labor produc-tivity in agriculture: Estimates for rural south India.Review of Economics and Statistics, 70.

Deolalikar, A. B. (1993). Gender di�erences in thereturns to schooling and school enrollment rates inIndonesia. Journal of Human Resources, 28.

Deraniyagala, S. (1995). Technical change and e�ciencyin Sri Lanka's manufacturing industry. Ph.D. disser-tation, University of Oxford, UK.

Duraisamy, P. (1992). E�ects of education and extensioncontacts on agricultural production. Indian Journalof Agricultural Economics, 47.

Edwards, S. (1993). Openness trade liberalization andgrowth in developing countries. Journal of EconomicLiterature, 31.

El-Ghonemy, M. (1990). The political economy of ruralpoverty: the case for land reform. New York: Rout-ledge.


Fei, J. C. H., Ranis, G., & Stewart, F. (1985). Towardsviable balanced growth strategies: A locationalperspective. Paper prepared for UNIDO.

Fields, G. S. (1989). Changes in poverty and inequalityin developing countries. World Bank ResearchObserver, 4.

Foster, A. D., & Rosenzweig, M. R. (1993). Informa-tion, learning, and wage rates in low-income ruralareas. Journal of Human Resources, 28.

Foster, A. D., & Rosenzweig, M. R. (1994). Technicalchange and human resources and investments: Conse-quences of the Green Revolution. Mimeo, Philadel-phia: University of Pennsylvania.

Foster, A. D., & Rosenzweig, M. R. (1995). Learning bydoing and learning from others: Human capital andtechnical change in agriculture. Journal of PoliticalEconomy, 103.

Garcia, M. (1990). Resource allocation and householdwelfare: A study of personal sources of income on foodconsumption, nutrition and health in the Philippines.Ph.D. dissertation, Institute of Social Studies, TheHague.

Gemmell, N. (1995). Evaluating the impacts of humancapital stocks and accumulation on economic growth:Some new evidence. Center for Research in EconomicDevelopment and International Trade, UK: Univer-sity of Nottingham.

Gertler, P., Locay, L., & Sanderson, W. (1987). Are userfees regressive? The welfare implications of healthcare ®nancing proposals in Peru. Journal of Econo-metrics, 36.

Grossman, G. M., & Helpman, E. (1991). Innovation andgrowth in the global economy. Cambridge, MA: MITPress.

Haddad, L., & Bouis, H. (1991). The impact ofnutritional status on agricultural productivity: Wageevidence from the Philippines. Oxford Bulletin ofEconomics and Statistics, 53.

Harbert, L., & Scandizzo, P. L. (1982). Food distributionand nutrition intervention: the case of Chile. Sta�Working Paper 512. Washington, DC: World Bank.

Helliwell, J. F., & Putnam, R. (1995). Economic growthand social capital in Italy. Eastern Economic Journal,21.

Hoddinott, J., & Haddad, L. (1991). Householdexpenditures, child anthropometric status and theintrahousehold division of income: evidence from theCote d'Ivoire. Research Program in DevelopmentStudies, Discussion Paper 155, Woodrow WilsonSchool.

Immink, M., & Viterri, F. (1981). Energy intake andproductivity of Guatemalan sugarcane cutters: Anempirical test of the e�ciency wage hypothesis.Journal of Development Economics, 9.

Jamison, D., & Moock, P. (1984). Farmer education andfarm e�ciency in Nepal: The role of schooling,extension services and cognitive skills. World Devel-opment, 12.

Jayaraman, R. (1995). On the meta-production front:An evidence-gathering exercise. Processed forUNDP, New York.

Keesing, D. B. (1966). Labor skills and comparativeadvantage. American Economic Review, 56.

Kenen, P. B. (1965). Nature capital and trade. Journal ofPolitical Economy, 73.

King, E. M., & Bellew, R. (1988). Education policy andschooling levels in Peru. Mimeo the World Bank,Washington DC.

King, E.M., & Hill, M.A. (1993). Women's education indeveloping countries: barriers, bene®ts, and policies.The Johns Hopkins University Press, published forthe World Bank, Baltimore.

King, E. M., & Lillard, L. A. (1987). Education policyand schooling attainment in Malaysia and thePhilippines. Economics of Education Review, 6.

Klugman, J. (1995). Decentralization: A survey of theliterature from a human development perspective.Human Development Report O�ce Occasional PapersNo. 12.

Krueger, A. (1978). Foreign trade regimes and economicdevelopment: liberalization attempts and conse-quences. Cambridge, MA: Ballinger.

Lavy, V., Strauss, J., Thomas, D., & deVreyer, P. (1995).The impact of the quality of health care on children'snutrition and survival in Ghana. LSMS WorkingPapers, 106. Washington, DC: World Bank.

Leamer, E. (1993). Factor-supply di�erences as a sourceof comparative advantage. American Economic Re-view, 83.

Lipton, M., & Ravallion, M. (1995). Poverty and policy.In J. R. Behrman & T. N. Srinivasan. Handbook ofdevelopment economics, Vol. 3. Amsterdam: NorthHolland.

Lipton, M. (1977). Why poor people stay poor: urban biasin world development. London: Temple Smith.

Lucas, R. E. (1988). On the mechanics of economicdevelopment. Journal of Monetary Economics, 22.

Michaely, M. (1977). Exports and growth: an empiricalinvestigation. Journal of Developing Economies, 4.

North, D. (1990). Institutions and their consequences foreconomic performance. In K. Cook & M. Levi, Thelimits of rationality (pp. 383±401). Chicago: Univer-sity of Chicago Press.

OECD/DAC (1995). Development Cooperation: E�ortsand Policies of the Members of the DevelopmentAssistance Committee, OECD, Paris.

Owens, T., & Wood, A. (1995). Export-oriented indus-trialization through primary processing? IDS WorkingPaper 19, Institute of Development Studies. Univer-sity of Sussex, Brighton.

Pack, H., & Westphal, L. (1986). Industrial strategy andtechnological change: theory versus reality. Journalof Development Economics, 22.

Perotti, R. (1993). Political equilibrium income distri-bution and growth. Review of Economic Studies,60.

Persson, T., & Tabellini, G. (1994). Is inequality harmfulfor growth?. American Economic Review, 84.

Pitt, M. M., Rosenzweig, M. R., & Hassan, M. N.(1990). Productivity health and inequality in theintrahousehold distribution of food in low-incomecountries. American Economic Review, 80.

Prud'homme, R. (1995). The dangers of decentraliza-tion. The World Bank Research Observer, 10.

Psacharopolous, G., Morley, S., Fiszbein, A., Lee, H., &Wood, B. (1992). Poverty and income distribution in


Latin America: the story of the 1980s. Washington,DC: World Bank.

Psacharopolous, G. (1994). Returns to investment ineducation: a global update. World Development, 22(9), 1325±1343.

Ram, R. (1985). Exports and economic growth: someadditional evidence. Economic Development andCultural Change, 33, 415±425.

Rana, P. (1988). Exports policy changes and economicgrowth in developing countries after the 1973oil shock. Journal of Developing Economies, 8,261±264.

Ranis, G. (1979). Appropriate technology in the dualeconomy: re¯ections on Philippine and Taiwaneseexperience. In A. Robinson, Appropriate technologiesfor Third World development. New York: Macmillan.

Ranis, G., & Stewart, F. (1994). Decentralization inIndonesia. Bulletin of Indonesian Economic Studies,30.

Ravallion, M., & Datt, G. (1991). Growth and redistri-bution components of changes in poverty measure: Adecomposition with applications to Brazil and India inthe 1980s. LSMS Working Papers #83. Washington,DC: World Bank.

Riddell, R., Robinson, M., deConinck, J., Muir, A., &White, S. (1995). Nongovernmental organizations andrural poverty alleviation. New York: Oxford Univer-sity Press.

Romer, P. M. (1990). Endogenous technological change.Journal of Political Economy, 98.

Rosenzweig, M. R., & Schultz, T. P. (1982). Childmortality and fertility in Colombia: individual andcommunity e�ects. Health Policy and Education, 2,305±318.

Rosenzweig, M. R. (1995). Why are there returns inschooling? American Economic Review, 85 (2).

Sahn, D. E., & Alderman, H. (1988). The e�ect ofhuman capital on wages, and the determinants oflabor supply in a developing country. Journal ofDevelopment Economics, 29 (2).

Schultz, T. P. (1988). Education investments andreturns. In H. Chenery & T.N. Srinivasan, Handbookof Development Economics. Amsterdam: North Hol-land.

Schultz, T. P. (1993a). Returns to women's education. InE. M. King & M. A. Hill, Women's education indeveloping countries: barriers, bene®ts, and policies.The Johns Hopkins University Press, published forthe World Bank, Baltimore, MD.

Schultz, T. P. (1993b). Investments in the schooling andhealth of women and men: quantities and returns.Journal of Human Resources, 28.

Schultz, T. P., & Tansel, A. (1993). Measurement ofreturns to adult health. LSMS 95, World Bank,Washington, DC.

Schultz, T. W. (1975). The value of the ability to dealwith disequilibria. Journal of Economic Literature,13.

Selowsky, M. (1981). Nutrition health and education:The economic signi®cance of complementarities atan early age. Journal of Development Economics, 9.

Selowsky, M., & Taylor, L. (1973). The economics ofmalnourished children: an example of disinvestment

in human capital. Economic Development and Cultur-al Change, 22.

Sen, A. K. (1984). Rights and capabilities. In A. Sen,Resources, values, and development. Cambridge, MA:Harvard University Press.

Stewart, F. (1977). Technology and underdevelopment.London: Macmillan.

Strauss, J. (1986). Does better nutrition raise farmproductivity?. Journal of Political Economy, 94.

Strauss, J., & Thomas, D. (1995). Human resources:Empirical modeling of household and family deci-sions. In J. R. Behrman & T. N. Srinivasan.Handbook of development economics, Vol. 3. Amster-dam: North Holland.

Strauss, J., & Thomas, D. (1996). Health, wealth andwages of men and women in urban Brazil. Journal ofEconometrics.

Streeten, P., Burki, S. J., ul Haq, M., Hicks, N., &Stewart, F. (1981). First things ®rst: meeting basicneeds in the developing countries. New York: OxfordUniversity Press.

Tanzi, V. (1995). Fiscal federalism and decentralization:A review of some e�ciency and macroeconomicaspects. Mimeo, International Monetary Fund,Washington, DC.

Tendler, J., & Freedheim, S. (1994). Trust in a rent-seeking world: health and government transformedin northeast Brazil. World Development, 22 (12),1771±1791.

Thomas, D., Strauss, J., & Henriques, M. H. (1990).Child survival height-for-age and household charac-teristics in Brazil. Journal of Development Economics,33.

Thomas, D., Strauss, J., & Henriques, M. H. (1991).How does mother's education a�ect child height.Journal of Human Resources, 26.

Thomas, D. (1990). Intrahousehold resource allocation:An inferential approach. Journal of Human Re-sources, 25, 4.

Thomas, D., Lavy, V., & Strauss, J. (1992). Public policyand anthropometric outcomes in the Cote d'Ivoire.LSMS Working Paper 89. Washington, DC: WorldBank.

UNDP (various). Human Development Report. NewYork: Oxford University Press.

von Braun, J. (1988). E�ects of technological change inagriculture on food consumption and nutrition: ricein a West African setting. Economic Development andCultural Change, 37.

Welch, F. (1970). Education in production. Journal ofPolitical Economy, 78.

Williamson, J. (1993). Human capital deepening, incomeinequality and demographic events along the Asia-Paci®c Rim. In N. Ogawa et al. Human resources indevelopment along the Asia-Paci®c rim. Singapore:Oxford University Press.

Wolfe, B. L., & Behrman, J. R. (1987). Determi-nants of women's health status and health-careutilization in a developing country: a latentvariable approach. Review of Economics andStatistics, 56.

Wolgemuth, J. C., Latham, M. C., Hall, A., &Crompton, D. (1982). Worker productivity and


nutritional status of Kenyan road constructionlaborers. American Journal of Clinical Nutrition, 36.

Wood, A. (1994). North-South trade, employment andinequality: Changing fortunes in a skill-driven world.

IDS Development Studies Series. Oxford: OxfordUniversity Press.

World Bank (1995a). Global economic prospects and thedeveloping world. Washington, DC: World Bank.


economic growth and human development - welcome to the department

Documents