the mit press poverty inequality and policy in latin america feb 2009

Prominent examples of this renewed emphasisinclude a number of reports produced recently by the World Bank,such as the recent World Development Reports on Poverty WorldBank 2000 and Equi

Structural Unemployment in Western Europe: Reasons and RemediesMartin Werding, editor

Institutions, Development, and Economic Growth

Theo S Eicher and Cecilia Garcı´a-Pen˜alosa, editors

Competitive Failures in Insurance Markets: Theory and Policy ImplicationsPierre-Andre´ Chiappori and Christian Gollier, editors

Japan’s Great Stagnation: Financial and Monetary Policy Lessons for vanced Economies

Ad-Michael M Hutchison and Frank Westermann, editors

Tax Policy and Labor Market Performance

Jonas Agell and Peter Birch Sørensen, editors

Privatization Experiences in the European Union

Marko Ko¨thenbu¨rger, Hans-Werner Sinn, and John Whalley, editorsRecent Developments in Antitrust: Theory and Evidence

Jay Pil Choi, editor

Schools and the Equal Opportunity Problem

Ludger Woessmann and Paul E Peterson, editors

Economics and Psychology: A Promising New Field

Bruno S Frey and Alois Stutzer, editors

Institutions and Norms in Economic Development

Mark Gradstein and Kai A Konrad, editors

Pension Strategies in Europe and the United States

Robert Fenge, Georges de Me´nil, and Pierre Pestieau, editors

Foreign Direct Investment and the Multinational Enterprise

Steven Brakman and Harry Garretsen, editors

Sustainability of Public Debt

Reinhard Neck and Jan-Egbert Sturm, editors

The Design of Climate Policy

Roger Guesnerie and Henry Tulkens, editors

Poverty, Inequality, and Policy in Latin America

Stephan Klasen and Felicitas Nowak-Lehmann, editors

See http://mitpress.mit.edu for a complete list of titles in this series

edited by Stephan Klasen and Felicitas Nowak-Lehmann

The MIT Press

Cambridge, Massachusetts

London, England

All rights reserved No part of this book may be reproduced in any form by any elec tronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher.

For information about special quantity discounts, please e mail special sales@mitpress mit.edu

This book was set in Palatino on 3B2 by Asco Typesetters, Hong Kong.

Printed and bound in the United States of America.

Library of Congress Cataloging in Publication Data

Poverty, inequality, and policy in Latin America / edited by Stephan Klasen and Felicitas Nowak Lehmann.

p cm (CESifo seminar series)

Includes bibliographical references and index.

Papers from a conference held at the Ibero America Institute for Economic Research in Go¨ttingen, Germany, in July 2005 and cosponsored by the CESifo research network ISBN 978 0 262 11324 3 (hardcover : alk paper)

1 Poverty Latin America Congresses 2 Equality Latin America Congresses.

3 Latin America Economic policy Congresses 4 Latin America Social

conditions Congresses I Klasen, Stephan II Nowak Lehmann D., Felicitas.

III Universita¨t Go¨ttingen Ibero Amerika Institut fu¨r Wirtschaftsforschung IV CESifo HC130.P6P715 2009

Series Foreword vii

Stephan Klasen and Felicitas Nowak-Lehmann

Ewout Frankema

Denis Cogneau and Je´re´mie Gignoux

Phillippe G Leite

of the Effect of Social Services and Community Infrastructures

Philippe De Vreyer, Javier Herrera, and Sandrine Mesple´-Somps

Philippe De Vreyer and Gilles Spielvogel

III Economic Policy, Poverty, and Inequality in Latin

Rainer Schweickert, Rainer Thiele, and Manfred Wiebelt

Jere R Behrman, Susan W Parker, and Petra E Todd

Herwig Immervoll, Horacio Levy, Jose´ Ricardo Nogueira,

Cathal O’Donoghue and Rozane Bezerra de Siqueira

This book is part of the CESifo Seminar Series The series aims tocover topical policy issues in economics from a largely Europeanperspective The books in this series are the products of the papers andintensive debates that took place during the seminars hosted byCESifo, an international research network of renowned economistsorganized jointly by the Center for Economic Studies at Ludwig-Maximilians-Universita¨t, Munich, and the Ifo Institute for EconomicResearch All publications in this series have been carefully selectedand refereed by members of the CESifo research network.

An Introduction

Stephan Klasen and Felicitas Nowak-Lehmann

The causes and consequences of high inequality in incomes, assets, andmany aspects of well-being in Latin America have recently re-emerged

as a central research and policy issue While in previous decades, cern about high inequality in Latin America was, following Kuznets’seminal work in the 1950s, largely focused on the impact of the devel-opment process on inequality, the new emerging literature is consid-ering the reverse causality—that is, the impact of inequality on thedevelopment process Prominent examples of this renewed emphasisinclude a number of reports produced recently by the World Bank,such as the recent World Development Reports on Poverty (WorldBank 2000) and Equity (World Bank 2005), as well as reports focusing

con-on Latin America, including the 2004 report Inequality in Latin America:Breaking with History (World Bank 2003) and the 2006 report PovertyReduction and Growth: Virtuous and Vicious Circles (World Bank 2006).Similarly, academic research has taken on this issue with renewed vig-

or, as shown by works of Eicher and Turnovsky (2003), Deininger andSquire (1998), and Forbes (2000), among many others

There are a number of reasons leading to this re-emergence of equality as a central research and policy issue in Latin America, aswell as the new emphasis on its development impacts First, inequalitywas and is extremely high in Latin America As shown in table I.1,Latin America continues to have the dubious distinction of having thehighest income inequality in the world, as measured by the Gini coeffi-cient Using other measures, or other dimensions, of inequality (e.g.,assets) would yield similar results More disconcerting is the persis-tence of inequality in Latin America across time In contrast to thehope held out by the Kuznets hypothesis that inequality will eventu-ally decline with development, inequality has remained extremelyhigh through the past 30 years and changed little even during episodes

in-of fast economic growth In contrast to the long secular and large cline in inequality in rich countries that took place between about 1930and 1970, we have not experienced a similar secular trend in any LatinAmerican country (or, for that matter, in other developing regions; seetable I.1) In fact, most evidence points to a small but significant rise ininequality in most developing countries since the early to mid-1980s(Gru¨n and Klasen 2003; Cornia and Court 2001) Latin America is noexception, although the extent of the increases vary by country andtime period, and there is some evidence that inequality has declinedslightly in some countries in the last few years (particularly in Brazil;see World Bank 2006) The resistance of high inequality in Latin Amer-ica to vastly different policy regimes and interventions is remarkableand somewhat disconcerting Neither the statist development strat-egies of the 1950s and 1960s nor the liberal market reforms of the1980s and 1990s have greatly affected inequality Great swings frompopulism to orthodox market fundamentalism have also had no effect.Second, inequality is not only persistent in the aggregate, but is alsopersistent across generations of individuals, as mounting research onthe intergenerational transmission of inequality is pointing out (WorldBank 2003, 2005).

de-Third, the detrimental effects of inequality on economic and humandevelopment have become more apparent recently As has been showntheoretically as well empirically, high inequality not only translatesinto higher absolute income poverty at any given level of mean in-comes, but it also lessens the poverty-reducing impact of economicgrowth (see Bourguignon 2003, World Bank 2000, and Klasen 2003).Latin America’s very poor progress in reducing absolute poverty in re-

cent decades is thus not only a consequence of its poor growth mance, but also its high inequality (World Bank 2006).

perfor-Fourth, there is growing evidence that high inequality, particularlyasset inequality, is detrimental to economic growth itself While theempirical evidence is still under some dispute (see, e.g., Deininger andSquire 1998, Forbes 2000, Klasen 2003, and Banerjee and Duflo 2003),the evidence is mounting that such a negative effect exists, particularly

in countries displaying particularly high inequality, with Latin ica often cited as the most prominent example of a high-inequality re-gion with a relatively poor growth performance, particularly in thepast 20 years The channels through which this effect is transmittedrange from capital market failures that prevent the poor in high-inequality countries from investing in human and physical capital orinsuring against risk; to social and political instability that deters in-vestment; to social conflict leading to inefficiencies, economic and polit-ical uncertainties, and growth collapses (e.g., Alesina and Rodrik 1994;Rodrik 1998; Deininger and Squire 1998; World Bank 2003, 2006).Fifth, there is growing debate and awareness about the well-beingcosts of high inequality in Latin America Given inequality aversion,for which there is convincing evidence from Latin America (WorldBank 2003), high inequality carries welfare costs that can be sizable(see Gruen and Klasen 2008) Investigations of subjective well-beingalso point to the welfare costs of high inequality in Latin America.Moreover, since the 1980s inequality has increasingly been associatedwith economic insecurity not only of the poor, but increasing sections

Amer-of the middle class, which has strongly undermined social cohesionand increased social and political conflict (Rodrik 2001)

Sixth, while there was a time where inequality was seen as a sary evil to promote incentives and efficiency, survey evidence fromLatin America clearly points to the fact that inequality is now seen astoo high and as unfair (World Bank 2005, 2006) This is surely related

neces-to the disappointing growth experience of the 1980s and 1990s, wherethe promise that policy reforms would deliver high growth and pov-erty reduction has largely not materialized If the promise of highgrowth and rapid poverty reduction is no longer credible, inequalityreduction will naturally receive more prominence Of particular con-cern is here that much of Latin America’s existing inequality is in factinequality of opportunities; that is, inequality related to one’s origin,race, sex, or parental background, which is seen as particularly repre-hensible (e.g., Roemer 1998, World Bank 2005)

Seventh, the political economy in Latin America has recentlybrought the issue of high inequality to the fore The combination of adisappointing record of growth and poverty reduction in the 1980sand 1990s with a transition to democracy in the same time period hasput considerable pressure on political leaders to find other ways totackle mass poverty Programs and policies to reduce inequality and/

or transfer resources to the poor have been one response to thesechallenges In addition, the recent rise of populism in many LatinAmerican countries (including Venezuela, Peru, Bolivia, and Argen-tina) has partly emerged as a result of the general dissatisfaction withhigh inequality, high poverty, poor growth, and social exclusion ofminorities and marginalized groups Clearly, inequality and persistentpoverty are having a serious impact on political developments and areamong the most prominent political issues across the continent.Lastly, the data and methods of analyzing inequality and povertyand their links to growth and policy interventions have dramaticallyimproved in recent years Regarding data, the implementation of stan-dardized regular representative household income surveys in nearlyall Latin American countries in the past 10 to 15 years has enabledresearchers to study poverty and inequality levels, trends, and deter-minants that were impossible in prior years when all that was avail-able were occasional snapshots from a single household survey Thishas also enabled an analysis of regional poverty and inequality dy-namics, which showed that regional inequality is an important driver

of national inequality (World Bank 2006) In addition to regular hold surveys, specialized surveys, some of which were implementedusing randomized designs, have allowed researchers to study theimpact of particular government programs (e.g., Todd and Wolpin2006; World Bank 2004) Regarding methods, there have been greatimprovements in combining micro and macro data to analyze and sim-ulate the impact of policies on poverty and inequality, as well as signif-icant improvements in studying the spatial dimension of poverty and

Clearly, the high and persistent inequality in Latin America is one

of the central, if not the central, economic policy challenges for LatinAmerican policymakers, and we have new tools and data at hand tostudy this issue It is thus not surprising that there has been such a re-surgence of interest in analyzing the dynamics and determinants ofpoverty and inequality, and the relationship of policies affecting them

in Latin America

While this renewed research interest has already generated manynew insights, analyses, and policy recommendations, many open ques-tions remain I will highlight a few that appear to be of particular rele-vance for understanding inequality and poverty dynamics as well astheir policy drivers First, there is still an incomplete understanding ofthe historical factors that generated these high levels of inequality inLatin America The work by Engermann and Sokoloff (2002) has gener-ated many useful insights in this regard, particularly emphasizing therole of land inequality and of colonial policies in generating high in-equality in Latin America But the historical factors leading to thishigh land inequality remain an open question, as does the persistence

of land inequality over time Related to this is the more general secondquestion regarding the transmission of inequality and poverty acrosstime and across generations What are the key mechanisms of such atransmission? Do they relate largely to the unequal transmission ofphysical, financial, and human capital across generations, to the per-sistence of regional inequality, or to policies that cement inequality?Third, to what extent are observed inequalities driven by inequalities

of opportunities and what are the main factors responsible for ing such inequalities? Fourth, what are the trends in regional inequal-ity, and is there evidence of spatial poverty traps that prevent certainregions from emerging from poverty? If so, what are the key drivers ofsuch poverty traps and are they amenable to policy interventions?Fifth, despite much research in the past on this topic, the impact ofpolicies on poverty and inequality remains poorly understood This ispartly related to the fact that the requisite data and analytical toolshave only recently become available For example, the ability to thor-oughly analyze the impact of tax and spending policies on povertyand inequality requires accurate household-level information as well

generat-as advanced microsimulation tools, both of which have only recentlybecome available Also, new policy issues, both at the macro and atthe micro level have emerged that were not relevant in the past At themacro level, the rise in de jure or de facto dollarization in many LatinAmerican countries has emerged recently and its impact on povertyand inequality (and the policy options to tackle them) is still unclear

At the micro level, a key new development has been the emergence

of conditional cash-transfer programs as a policy instrument to transferfinancial resources to poor households as well as promote the edu-cation and health of children These programs were first introduced

in Mexico and Brazil but are now spreading to many other Latin

American countries While there is some literature on assessing theireffects (e.g., Todd and Wolpin 2006; see World Bank 2004 for a survey),questions regarding their long-term effects can only now be analyzed.The papers in this volume were selected from a conference titled

‘‘Poverty, Inequality, and Policy in Latin America’’ that took place atthe Ibero-America Institute for Economic Research in Go¨ttingen, Ger-many, in July 2005, funded by the German Research Foundation andcosponsored by the CESifo research network They were chosen be-cause they present particularly innovative research addressing themost important research gaps identified earlier In the remainder ofthis introduction, we briefly summarize the contributions and their re-lationship to the open questions raised above, and close by identifyingopen research and policy questions

The contributions are grouped into three parts Part I analyzes theemergence and transmission of poverty and inequality in Latin Amer-ica It centers around, firstly, the relationship between land inequality,institutions, and income inequality and, secondly, the relationship be-tween racial discrimination and/or inequality of opportunities and in-come inequality Chapters 1 to 3 thus try to address the first twoquestions raised above

In chapter 1 Ewout Frankema investigates the colonial origins of equality, comparing the Latin American colonial experience with theWest and Central African colonial past The bipolar distribution ofland, established during three centuries of colonial rule, is consideredone of the crucial underpinnings of persistent high levels of incomeinequality in Latin America It can be explained by the colonial sys-tem and the emergence of plantations of coffee, cocoa, and cotton TheIberian colonial administration deliberately redistributed land fromindigenous peasants to the Creole elite For the Spanish Crown, landwas a convenient resource to reward the early colonists’ efforts of con-quest, conversion, and settlement Additionally, slavery made the pro-duction of tropical cash crops on the large plantations of Brazil and theCaribbean possible and profitable Frankema studies the causes andconsequences of colonial land distribution using new and existing landinequality figures In this study two central questions are addressed:what explains the cross-country variation in land inequality at the end

in-of the colonial period, and how does initial land inequality relate tocurrent income inequality? Frankema argues that postcolonial levels ofland inequality are largely determined by the response of colonists tolocal endowments (land, people, climate, and soil) and settler condi-

tions (disease environment, local resistance).2There are also clear cations that specific Iberian colonial institutions contributed to highland inequality in Latin America.

indi-To study the impact of land inequality on income inequality and velopment, Frankema controls for the level of economic developmentand separates the impact of land inequality in a direct effect (the share

de-of rural inequality in total inequality) from an indirect effect related topath-dependent effects Frankema finds that west and central Africaand Latin America show very different results: in Latin America landinequality is largely driving income inequality, whereas income in-equality in west and central Africa is based on exploitation of the ruralpopulation by a small predatory urban elite But both regions equallyshare the burden of distributive conflicts that are usually characteristic

of countries with very high levels of economic inequality, and he tifies land inequality as one main driver of this process

iden-In chapter 2 Denis Cogneau and Je´re´mie Gignoux study the impact

of educational opportunities on earnings inequality in Brazil, thusaddressing a key transmission channel of inequality across genera-tions The Brazilian economy and society display an extremely high de-gree of dualism, visible both in the education system (private/state)and the labor market (official/unofficial) Brazil is also among thecountries with the lowest intergenerational educational mobility andequality of social and economic opportunities in the world The litera-ture on inequality emphasizes that inequality in Brazil is associatedwith a high degree of intergenerational transmission of education, oc-cupational status, or income Taking up this object of research, Cog-neau and Gignoux study the contribution of education to a reduction

of economic inequality They consider two kinds of earnings inequalityover the 1976–1996 period: first, overall inequality in observed earn-ings, and second, inequality of opportunity computed according toaxiomatics proposed by Roemer and Van de Gaer, Schokkaert, andMartı´nez Cogneau and Gignoux find that overall inequality roseslightly in the period 1976–1996, while inequality of opportunityposted a slight drop Nonparametric weighting techniques (introduced

by Di Nardo, Fortin, and Lemieux) are used to evaluate the effects ofeducational changes on earnings inequality Specifically, Cogneau andGignoux design and implement Blinder-Oaxaca decompositions tocapture the effects of schooling expansion, changes in the structure ofearnings, and changes in the intergenerational educational mobility

on inequality First of all, they find that changes in the distribution of

education contributed to an increase in overall inequality and ity of opportunities among the oldest generations and to a decrease

inequal-in inequal-inequality among the post-WWII cohorts Second, the decrease inequal-inthe returns to education also contributed to equalizing labor marketopportunities in the 1988–1996 subperiod Third, the changes in educa-tional mobility were not large enough to significantly affect earningsinequalities

The issue of inequality of opportunities and its impact on income equality is further elaborated in chapter 3 by Phillippe Leite, whoinvestigates for the Brazilian case how much of the black-white wagedifferences can be attributed to race discrimination in the labor marketand how much can be attributed to inequality of opportunities In-equality of opportunities implies that a great deal of the black-whitewage gap arises from differences in endowments, meaning that blacksmay have lower levels of education and experience than whites due

in-to pre-labor-market inequalities in housing and schooling, which may

be related to discrimination in these markets or to intergenerationaleffects The paper explores the intersection of race and human capitalaccumulation in the labor market, focusing in the intergenerationaltransmission of human capital; that is, the effects of parents’ andgrandparents’ education on individual schooling A two-stage least-squares method is presented to treat individual schooling as a function

of parents’ and/or grandparents’ schooling and/or school quality.Workers with similar characteristics in identical sectors are studiedseparately to reduce omitted variable bias such as preference or ability

of similar workers The analysis is based on data for urban areas and

is conducted separately for southeast and northeast regions of Brazilusing the national representative household survey of 1996 Leite’smain finding is that pre-market human capital difference is the mainfactor responsible for the wage gap between Brazilian workers More-over, the regional profile suggests that the labor market is a moreimportant locus of racial discrimination in the southeast than in thenortheast of Brazil

Chapters 2 and 3 suggest that the intergenerational transmission ofhuman capital is a very powerful driver of existing inequality and pol-icies that merely address existing labor market discrimination will dolittle to reduce overall inequality Breaking this transmission by dras-tically improving educational opportunities for children from pooreducational backgrounds must be accorded top priority if one is tosustainably address income inequality

In part II, the spatial dimensions of poverty and inequality are lyzed for Peru and Brazil using sophisticated panel data techniques.The aim is to understand the regional dynamics of poverty and in-equality in order to discern whether there are spatial poverty trapsand what determines such traps Thus this section is focusing on thefourth question identified earlier In chapter 4, Philippe De Vreyer,Javier Herrera, and Sandrine Mesple-Somps study the effect of socialservices and community infrastructures on living standards in ruralPeru using a four-years unbalanced panel data set Three differentsources of data are used: the ENAHO 1997–2000 household surveys,the population census of 1993, and the district infrastructure census of

ana-1997 The authors look at a multitude of representative householdsfrom seven areas (three rural areas and four urban areas) at four timeperiods (the last quarters of 1997, 1998, 1999, and 2000) to estimate therelationship between the local geographic endowment (pure geographicendowments like ecological conditions, climate, altitude, and latitude)and community-level variables (such as the supply of local publicgoods and infrastructure or the local endowments with private goods)and household growth in living standards They use a quasidiffer-encing method to identify the impact of locally determined geographicand socioeconomic variables, while removing unobserved householdand community fixed effects A two-stage generalized method of mo-ments (GMM) is utilized to estimate the model parameters It is foundthat most socioeconomic variables have a significant impact on con-sumption growth, but not all pure geographic characteristics or localpublic goods do show a significant impact It appears that spatialpoverty traps are linked more strongly to socioeconomic features ofvillages and the provision of public goods rather than to purely geo-graphic attributes Therefore, the authors conclude that it is necessary

to combine policies focused on income transfers and human capitalformation with policies that favor mobility across regional markets

In chapter 5, Philippe De Vreyer and Gilles Spielvogel examinethe growth process of a large sample of Brazilian municipalities for theperiod 1970–1996 and attempt to evaluate not only convergenceamong the municipalities but also the spatial externalities at work.They select Brazil as a case study since it offers a good example of acountry with regions of very different levels of development Per capitaincome in the northeast is only 47 percent of the national average whilethat of the southeast is 34 percent above The northeast is also more un-equally developed than other regions of Brazil The wealthiest state in

northeast has a GDP per capita more than 2.5 times larger than thepoorest, whereas the ratio is only 1.7 in the southeast De Vreyer andSpielvogel find that over the period 1970–1996 inequality increased inthe north and the northeast regions and decreased in other regions ofthe country They use Moran’s indices and Moran scatterplots to ana-lyze the extent and changes of spatial inequalities among Brazilianmunicipios They also find evidence of polarized development and pov-erty traps Relatively low-production municipios tend to be groupedtogether in the north and northeast, and this tendency increases overtime, whereas municipios with a GDP per capita higher than averagetend to be grouped in the south of the country To analyze the process

of growth in per capita GDP at the municipality level, De Vreyer andSpielvogel estimate several versions of a growth model, allowing fordifferent kinds of spatial dependence among neighboring municipios.The paper shows that the presence of spatial externalities in the growthprocess of the Brazilian municipalities can help explain the divergingpattern of inequalities at the local level during the period 1970–1996:while the municipios in the southern part of the country have experi-enced some convergence, this is not the case in the northern regions.Inequalities between municipios have tended to increase in the north-east and north and low-income localities have become more spatiallyclustered, indicating that the polarization of economic activities inthese regions has increased

Both papers thus find that spatial externalities and poverty traps areimportant drivers of poverty and inequality dynamics in Latin Amer-ica Policies to reduce inequality must therefore address these unfavor-able spatial dynamics

Part III deals with the impact of economic policy on poverty and equality, thus addressing selected issues related to the fifth questionraised earlier using innovative data or modeling techniques, or investi-gating recent new policy initiatives In chapter 6 the macroeconomicand distributional effects of exchange rate policy—more specifically

in-of devaluation in a dollarized economy—are studied by RainerSchweickert, Rainer Thiele, and Manfred Wiebelt for the Bolivian case

In the period under study (1990–2005) Bolivia is, by and large, terized by a crawling peg system that helps to avoid excess exchangerate volatility and that provides a visible, short-term anchor for infla-tion Bolivia does stop short of full dollarization, but the degree ofdollarization is extremely high: in 2003, the degree of deposit dollariza-tion in the banking system is at about 92 percent, dollar-denominated

charac-assets account for 77 percent of broad money, and bank credit to theprivate sector in dollars is close to 97 percent Wage indexation is stillsignificant in Bolivia, even though formal wage indexation had beenabolished with the ‘‘New Economic Policy’’ in 1985 The authors use areal-financial computable general equilibrium (CGE) model to simulatethe macroeconomic and distributional effects of exchange rate policy,taking into account that the impact of exchange rate policy is limited

by the potential pass-through to inflation and the impact on the cial system (dollarization, net debtor, and net creditor positions) Sixrepresentative household groups (smallholders, agricultural workers,nonagricultural workers, employees, urban informals, and employers),which are basically characterized by their factor endowments andaccumulated U.S.-dollar–denominated net financial assets, are scruti-nized in the analysis of the macroeconomic and distributional effects

finan-of exchange rate policy The authors find that the potential finan-of nal devaluation to smooth the adjustment path after a negative shockprimarily depends on the absence of wage indexation Financialde-dollarization tends to be contractionary in Bolivia As to the dis-tributional effects, they can show that a nominal devaluation in no cir-cumstance reduces the poverty effect of an external shock Moreover,they find that, even without wage indexation, the strong short-runmacroeconomic expansion has not managed to translate into a notice-able poverty alleviation As a result, they find that a dollarized econ-omy such as Bolivia has few tools to mitigate the impact of externalshocks on the poor, but that measures to de-dollarize carry their ownrisks, leaving small open economies such as Bolivia in a double bind.Chapter 7 turns to the impact of fiscal and social policies on targetedprograms for poverty alleviation This is of particular policy relevance

nomi-as such conditional cnomi-ash-transfer programs have become a principalway through which policymakers across Latin America are trying

to tackle poverty and inequality Jere Behrman, Susan Parker, andPetra Todd evaluate the medium-term impact of social expenditureprograms on rural youth in Mexico, specifically the so-called Oportu-nidades program (formerly called PROGRESA), a conditional cash-transfer program Under this program cash grants are given to poorfamilies in exchange for their children’s regular attendance at schooland for visits in health clinics Currently, five million families partici-pate in the program, which represents about one-fourth of all families

in Mexico The study uses data from a social experiment that was domly carried out in rural Mexican villages The treatment group was

ran-made up of 320 villages randomly selected for early incorporation intothe program, and 186 villages were designated as a control group to beincorporated eighteen months later The study’s objective was to checkwhether differential exposure to the program significantly impactededucational attainment, labor market outcomes, marriage, migration,and cognitive achievement of youth The authors’ estimates reveal sig-nificant positive impacts from medium-term (5.5 years) program expo-sure on school grades completed On average, youth in the treatmentgroup have about 0.2 more years of schooling than youth in the controlgroup, both for boys and girls Their estimates also suggest that boyswith longer exposure progressed significantly faster through school.

As to educational attainment, the results do not reveal any positive nificant impacts on achievement scores (reading, mathematics, writtenlanguage skills) Employment effects are negative for boys and insignif-icant for girls, indicating that boys spend more time in school Finally,the authors find that the Oportunidades program has a statistically sig-nificant impact on the migration rate Male youth are about six percentless likely to migrate out of their household relative to the controlgroup While the positive effects of this program on short-term mea-sures of schooling and health-seeking behavior had been known, thischapter shows that even in the medium term such effects persist, giv-ing rise to the hope that such conditional cash-transfer programs might

sig-be one avenue for combining redistribution with human capital ments in the poor, which could help break the intergenerational trans-mission of poverty

invest-Finally, in chapter 8 the impact of the tax-benefit system on ity and poverty in Brazil is examined by Horacio Levy, Herwig Immer-voll, Jose´ Ricardo Nogueira, Cathal O’Donoghue, and Rozane Bezerra

inequal-de Siqueira While tax-benefit systems in most Organisation for nomic Cooperation and Development (OECD) countries reduce in-come disparities very significantly, the Brazilian government (as well

Eco-as other Latin American governments; see World Bank 2006) hEco-as beenmuch less successful in alleviating inequality and poverty It is theauthors’ objective to quantify the direct impact of tax burdens andbenefit payments on inequality and poverty by using the BrazilianHousehold Microsimulation System (BRAHMS), a newly developedtax-benefit microsimulation model This tax-benefit model calculatestax liabilities and benefit entitlement (rather than relying on tax andbenefit amounts recorded in the data) by combining each household’scharacteristics with detailed institutional information on tax and bene-

fit policies Personal direct taxes and cash transfers are simulated usingmicrodata from the National Household Survey (Pesquisa Nacionalpor Amostra de Domicı´lios, or PNAD) since there is no direct informa-tion in PNAD about the amount of personal income tax or social secu-rity contribution payments Both data and the policy rules relate to theyear 2003 The authors’ study then uses the tax-benefit model to ana-lyze the existing tax-benefit system in Brazil, thus presenting the base-line scenario The authors find that the results produced by BRAHMSclearly reflect the high degree of income inequality and the relativelysmall equalizing effects of the tax-benefit system in Brazil They canshow that 60 percent of the pensions, which constitute 94 percent oftotal cash transfers, flow into the top income quintile, thus favoringthe already rich and the elderly population The bolsa escola benefit

is the only benefit targeted to lower income households However, thebudget devoted to Bolsa Escola and other cash transfer programs isstill a minuscule share of total social spending, and their impact interms of poverty reduction is therefore limited The predominance ofindirect taxes and tax concessions to the most affluent groups makethe tax system a poor redistributive tool Tax avoidance opportu-nities and difficulties in enforcing tax compliance reinforce the ratherregressive character of the current Brazilian tax system Therefore, theauthors argue that Brazil’s tax-benefit system should be as simple andtransparent as possible (with the expenditure side of the budget as thefundamental redistributive instrument), and oriented toward the pro-vision of basic services and well-targeted direct transfers to house-holds

The chapters in this volume add considerably to our understanding

of the dynamics of inequality and poverty in Latin America, as well aspolicy’s scope in addressing this issue In particular, the contributionsgreatly add to our understanding of the evolution and nature of in-equality in Latin America, of its spatial and temporal trends, and ofthe impact of particularly pertinent policy instruments in tackling pov-erty and inequality While all of the studies focus on one country, theissues analyzed are typically paradigmatic for many Latin Americancountries and thus provide lessons beyond the case examined

Nevertheless, open questions remain The extent of inequality ofopportunities and of the intergenerational transmission of inequality

is a subject requiring further scrutiny It needs to be extended tomore Latin American countries, and the impact of specification andmethodological choices on the results needs to be investigated in more

contexts, using different data sets The evidence of spatial povertytraps and spatial externalities is similarly intriguing but calls for moreresearch to solidify the results, examine the determinants of spatialpoverty traps in more detail, and study whether these traps have be-come more or less severe over time On the policy front, there is alsoroom for further research While the positive impact of conditionalcash-transfer programs is now quite well established, the longer-termimpacts on poverty and inequality in a country remain an open ques-tion As these programs are being scaled up and extended, it is inter-esting to see whether they are beginning to seriously affect povertyand inequality levels and have lasting long-term impacts on the inter-generational transmission of economic status Similarly, the scope andlimits of using the tax and expenditure system to sustainably affect in-equality without harming incentives remains an open question Whilethis chapter’s authors suggest that there is little room to make the taxsystem more progressive, recent studies by the World Bank specificallycall for tax changes to increase progressivity in Latin America (includ-ing greater reliance on income and land taxes; see World Bank 2003,2006) Thus this is another area for further research and analysis.Lastly, none of the contributions in this volume offer a single solu-tion for Latin American policymakers to use in tackling the extraordi-narily high levels of inequality and the negative side effects they carry.This remains an area of further research At the same time, the contri-butions in this volume suggest a number of areas of policy interventionthat could have a lasting impact, including more redistributive expen-diture policies, expansion of conditional cash-transfer programs, andpro-poor spatial infrastructure, transfer, and migration policies Theyalso suggest two areas of policy focus: the remaining high land in-equality and the intergenerational transmission of human capital Ifthese issues were addressed in a consistent manner, maybe they wouldtogether help Latin America overcome its legacy of high inequality,high and stagnant poverty, and poor economic growth.

Acemoglu, D., S Johnson, and J Robinson 2001 ‘‘The Colonial Origins of Comparative Development: An Empirical Investigation.’’ American Economic Review 91, no 5: 1369 1401.

Alesina, A., and D Rodrik 1994 ‘‘Distributive Policies and Economic Growth.’’ Quarterly Journal of Economics 109: 465 490.

Banerjee, A., and E Duflo 2003 ‘‘Inequality and Growth: What Can the Data Say?’’ Jour nal of Economic Growth 8: 267 299.

Bourguignon, F 2003 ‘‘The Growth Elasticity of Poverty Reduction: Explaining Hetero geneity across Countries and Time Periods.’’ In Growth and Inequality: Theory and Policy Implications, ed T Eichler and S Turnovsky, 3 26 Cambridge: MIT Press.

Cornia, G A., and J Court 2001 ‘‘Inequality, Growth and Poverty in the Era of Liberal ization and Globalization.’’ Policy Brief No 4, The United Nations University, WIDER, 38 pages.

Deininger, K., and L Squire 1998 ‘‘New Ways of Looking at Old Issues: Inequality and Growth.’’ Journal of Development Economics 57: 259 287.

Di Nardo, J., N Fortin, and T Lemieux 1996 ‘‘Labor Market Institutions and the Distri bution of Wages, 1973 92: A Semiparametric Approach.’’ Econometrica 64, no 5: 1001 1044.

Eicher, T., and S Turnovsky 2003 Growth and Inequality: Theory and Policy Implications Cambridge: MIT Press.

Engermann, S., and K Sokoloff 2002 ‘‘Factor Endowments, Inequality, and Paths of De velopment among New World Economies.’’ Economia 3: 41 109.

Forbes, K 2000 ‘‘A Reassessment of the Relationship between Inequality and Growth.’’ American Economic Review 90: 869 887.

Grimm, M., S Klasen, and A McKay 2007 Determinants of Pro poor Growth: Analytical Issues and Findings from Country Cases Basingstoke, Hampshire: Palgrave Macmillan Gruen, C., and S Klasen 2003 ‘‘Growth, Inequality, and Well Being: Intertemporal and Global Comparisons.’’ CESifo Economic Studies 49: 617 659.

Gruen, C., and S Klasen 2008 ‘‘Growth, Income Distribution, and Welfare: Comparisons across Space and Time.’’ Oxford Economic Papers 60: 212 236.

Klasen, S 2003 ‘‘In Search of the Holy Grail: How to Achieve Pro Poor Growth?’’ In To ward Pro Poor Policies: Aid, Institutions, and Globalization, ed B Tungodden, N Stern, and

I Kolstad, 63 94 New York: Oxford University Press.

Kuznets, S 1955 ‘‘Economic Growth and Income Inequality.’’ American Economic Review 45: 1 28.

Lemieux, T 2002 ‘‘Decomposing Changes in Wage Distributions: A Unified Approach.’’ Canadian Economics Association 35, no 4: 646 688.

Rodrick, D 1998 ‘‘Globalization, Social Conflict, and Economic Growth’’ (Prebisch Lec ture) The World Economy 21, no 2: 143 158.

Rodrick, D 2001 ‘‘Why is There So Much Economic Insecurity in Latin America?’’ CEPAL Review 73: 7 29.

Roemer, J 1996 Theories of Distributive Justice Cambridge: Harvard University Press Roemer, J 1998 Equality of Opportunity Cambridge: Harvard University Press.

Todd, P., and P Wolpin 2006 ‘‘Assessing the Impact of a School Subsidy Program in Mexico.’’ American Economic Review 96: 1384 1417.

Van de Gaer, D., E Schokkaert, and M Martı´nez 2001 ‘‘Three Meanings of Intergenera tional Mobility.’’ Economica 68: 519 537.

World Bank 2000 World Development Report 2000/01: Attacking Poverty New York: Ox ford University Press.

World Bank 2003 Inequality in Latin America and the Caribbean: Breaking with History? Washington, D.C.: The World Bank.

World Bank 2005 World Development Report 2006: Equity and Development New York: Oxford University Press.

World Bank 2006 Poverty Reduction and Growth: Virtuous and Vicious Circles Washington, D.C.: The World Bank.

Latin America

Causes and Consequences of Land Distribution

Ewout Frankema

In order to explain the observed rigidities in the distribution of income,recent inequality literature has paid renewed attention to the historicalevolution of asset inequality (Atkinson and Bourguignon 2000) Thispaper analyzes the causes and consequences of the historical evolution

of land inequality, employing a new dataset of land inequality figures.Land inequality is widely regarded as an important determinant ofpersistent asset inequality and institutional rigidity (Engerman andSokoloff 1997; Galor, Moav, and Vollrath 2003; North, Summerhill,and Weingast 2000)

In no other region of the world was the evolution of land ity so directly intertwined with colonial settlement as in Latin America.For the Spanish Crown, land was a convenient resource to reward theearly colonists’ efforts at conquest and conversion Land was deliber-ately redistributed from indigenous peasants to the Creole elite in ex-change for political loyalty (North, Summerhill, and Weingast 2000).The institution of the encomienda provided the silver mines and the

The redistribution of land was facilitated by the establishment of tinct Spanish and Indian juridical and administrative spheres and bythe demographic disaster resulting from the introduction of ‘‘OldWorld’’ diseases (Bakewell 2004) In Brazil and the Caribbean, perva-sive social dualism was primarily shaped by the creation of plantationeconomies based on the exploitation of African slave labor The Catho-lic Church materialized its position as the supreme religious authority

dis-by acquiring large estates In other words, land inequality formed acore ingredient of the colonial institutional matrix in Latin America(Williamson 1992, Fernandez-Armesto 2003)

With the Latin American experience in mind, this chapter explorestwo questions: what explains the cross-country variation in land in-equality at the end of the colonial age, and how does initial landinequality relate to current income inequality? Exploring these ques-tions may improve our understanding of the path-dependent charac-teristics of inequality and may also indicate to what extent the LatinAmerican experience was a unique experience As this chapter relies

on scarce historical data, the empirical analysis is carried out in a ple ordinary least squares framework where the regression results pri-marily serve to illustrate, rather than demonstrate, the validity of thehypotheses derived from the historical analysis

sim-The paper is structured as follows In section 1.2 the literature on thecauses and consequences of land inequality is discussed Section 1.3introduces the land distribution data (Gini and Theil coefficients) andevaluates these figures in a global comparative perspective Section 1.4presents the two multivariate regression models and the included vari-ables In section 1.5 the results are presented and interpreted In section1.6 the long-run consequences of initial land inequality are evaluated.Section 1.7 concludes

In the literature regarding the development of colonial institutions,specific attention has been paid to the feasibility of settlement and ex-ploitation in various regions of the world (Acemoglu, Johnson, andRobinson 2001; Engerman and Sokoloff 1997, 2005) Local conditionsrelate to the nature of land endowments (soil, climate, location), the rel-ative quantity of land and labor endowments (land-labor ratios, popu-lation density), the local disease environment, and the comparativeresistance of the indigenous population to colonial occupation Colo-nial institutional development is generally regarded as a function ofthe specific policy objectives of the colonial motherland in response tothese local conditions This brief overview of literature intends to dis-cuss these institutional responses

Responding to Local Endowments

Testing the hypothesis of Engerman and Sokoloff (1997) that ‘‘landendowments of Latin America lent themselves to commodities featur-ing economies of scale and the use of slave labor,’’ Easterly concludes

that a natural environment suitable to cash-crop production is ated with high levels of income inequality in the long run (Easterly

associ-2002, 3–4) Cash crops such as sugar, tobacco, coffee, cocoa, rubber,and bananas can be efficiently produced on large estates exploitingcoerced indigenous or slave labor A coexistence of large estates focus-ing on the production of exportable cash crops and small subsistenceholdings concentrating on the production of food crops for the domes-tic market skews the distribution of land (Leamer et al 1999, Easterlyand Levine 2003) Since tropical climates allow a larger variety of cash-crop production than temperate climates, the geographic location of acountry may indirectly affect the distribution of land A specialization

in scale-neutral food crops has a moderating effect on land inequality.Before the introduction of modern agricultural technologies (motor-ized) food crops such as wheat and maize could be efficiently pro-duced on plots of modest size, while rice crops could be grown oneven smaller plots of land (Hayami and Ruttan 1985)

According to Engerman and Sokoloff (1997, 2005) the variety in landendowments produced different paths of colonization in British NorthAmerica, mainland Spanish America, and the sugar plantation econo-mies in Brazil and the Caribbean islands Agriculture in British NorthAmerica became organized around a homogenous group of whiteEuropean farmers producing food crops (mainly wheat) on small tomedium-scale farms Contrary to the slave plantations in the southernstates, the egalitarian distribution of land in the northern states fit into

a strategy to attract British settlers to the land frontier In SpanishAmerica, on the other hand, the main objective was to guarantee asteady supply of mineral resources (silver, gold, diamonds) to the Iber-ian peninsula The redistribution of land was part of a strategy to forceindigenous peasants to work in the mines and on the fields of the largeestate holders The nature of land endowments and the presence ofmineral endowments thus affected the distribution of land

Apart from the nature of land endowments, the relative quantity ofland and labor shapes the potential efficiency of the rural economy.Land abundance invokes labor-saving production methods and favorscrops that use land intensively The land–labor ratio also shapes insti-tutional developments Domar (1970) argues that in premodern agri-cultural societies, elites face the problem of recruiting sufficient labor

to toil their soil In land-abundant countries landless laborers have anopportunity to start farming at the land frontier In response, the land-owning elite tends to develop coercive labor market institutions, such

as serfdom, slavery, or permanent debt peonage (Domar 1970, setz 2000) An alternative strategy is to distribute (virgin) territoriesamong the elite or restrict the access to land of certain groups (indige-nous farmers or landless laborers) In labor-abundant countries on theother hand, elites dispose of more opportunities to extract rents fromtaxes and trade margins without having to intervene directly into theland market All these arguments support the hypothesis that low lev-els of population density create incentives to regressively redistributeland, in particular in the context of the colonial society.

Dem-Acemoglu, Johnson, and Robinson (2001) argue that in areas favorable to colonial settlement—that is, with a high disease incidence

un-or fierce native resistance—colonial institutions will be created in un-order

to maximize the extraction of resources from a distance As a result, thepath of institutional development was characterized by weak propertyrights protection and a political context supporting rent-seeking be-havior In regions favorable to colonial settlement, institutions weremolded according to the motherland with the purpose of accumulatingcapital and skills, enhancing economic growth during the age of inde-pendence Settler mortality rates appear to be significantly correlatedwith present-day risks of expropriation

In sub-Saharan Africa the rates of colonial settlement were ably lower than in the Americas, and also more concentrated in thecoastal regions The colonial powers created institutions to extract rentsvia taxation (head tax) and trade in slaves and high-margin naturalresources (Young 1994, Manning 1988) Since settlers did not directlyinterfere with the daily practices of agricultural production on a largescale, the traditional rural institutions were left intact African slaveswere shipped across the Atlantic in order to work on American planta-tions rather than being put to work on African plantations (Stavrianos

consider-1981, Eltis 2000, Ayittey 2005) High rates of settler mortality may haveprohibited the redistribution of land

The conversion of Native Americans to Catholicism was a specificobjective of Iberian colonial policy As a result of the pact betweenRome and the Iberian monarchies against the backdrop of the Ref-ormation in Europe, the Catholic Church gained omnipresence inIberian-American society The penetration of the regular and secularorders into the daily life of Amerindians went much further than Prot-estant missions were ever capable of Lal (1998) points out, followingGoody (1983), that the Catholic Church devised specific inheritancelaws in order to enlarge its landed estates and traded salvation and

sacraments in return for (generous) gifts from its members The centration of land in the hands of the Catholic Church may, indeed,have had a significant positive impact on land inequality (Van Oss

Several empirical studies have revealed that initial land inequality isbad for growth This result appears to be stronger, more robust, andless contested than tests showing that income inequality is bad forgrowth (Barro 2000, Easterly 2002, Helpman 2004) Using land Ginisaround 1960, Birdsall and Londono (1997) show that the initial distri-bution of land is significantly negatively related to long-run economicgrowth This is confirmed by Deininger and Squire (1998); Li, Squire,and Zou (1998); and Deininger and Olinto (1999)

One of the major theoretical underpinnings of this empirical vation is that land is an important collateral asset in getting capitalmarket access In a context of imperfect capital markets, initial land in-equality can pose barriers to individual entrepreneurship or invest-ments in human capital (Galor and Zeira 1993) If public policy fails toremove capital market imperfections this not only hampers growth,but also leads to persistent asset and income inequality In Hernando

obser-de Soto’s the Mystery of Capital (2000) this argument is obser-developed withspecific attention on the case of Latin America

Initial land inequality also enhances persistent inequality in a context

of concentrated political power If the political and landowning elitesare concentrated and mostly overlapping, policies that suppress demo-cratic accountability and social development in order to preserve the

distributional status quo are likely to prevail (Olson 2000, Bourguignonand Verdier 2000, Acemoglu and Robinson 2006) In comparison to theUnited States and Canada, public education investments and franchiseextensions lagged far behind in Latin America (Engerman, Haber, andSokoloff 2001; Mariscal and Sokoloff 2000) Galor, Moav, and Vollrath(2003) find empirical evidence for a negative effect of land inequality

on public education expenditures in a cross-state U.S analysis andGylfason and Zoega (2002a, 2002b) report a positive relation betweenland and income inequality and a negative relation of both variableswith secondary school enrollment rates in cross-country analyses.Theory clearly predicts a negative relation between initial land in-equality and growth and a positive relationship between land andincome inequality, ceteris paribus It is therefore surprising that Dein-inger and Squire (1998) find a correlation of historical land Ginis(1960s) and current income Ginis (1990s) of just 0.39 (see also Dein-inger and Olinto 1999) Gylfason and Zoega (2002a and 2002b) report

an r of 0.33 The correlation coefficients estimated in section 1.5 areeven lower, ranging from 0.19 to 0.23 If land inequality is an impor-tant determinant of (persistent) income inequality, some conditionalvariables should be included in a multivariate analysis in order to ex-plain the rather modest correlation

Land distribution data are scarce Taylor and Hudson (1972, 267–269)present a dataset consisting of Gini coefficients of land distribution of

54 different countries in some year close to 1960 More recently, inger and Squire (1998) used a dataset of 261 Gini coefficients of 103different countries Of these, 60 observations from around the year

Dein-1960 have been published in a paper by Deininger and Olinto (1999,24) The data are derived from the Food and Agriculture Organization

paper is based on census data from the International Institute of culture (IIA) and the FAO The estimates of land inequality are pre-sented in table A1.1 (appendix, pp 303–308), including the figures ofTaylor and Hudson (1972) and Deininger and Olinto (1999) (T&H andD&O hereafter) The Frankema dataset consists of 186 observations for

Agri-105 different countries, including a considerable amount of prewarestimates (Frankema 2006) The figures are expressed in Gini- and Theilcoefficients, which are almost perfectly correlated (r ¼ 0:98) Table 1.1

presents the correlation between the three datasets and shows that theD&O and Frankema data are more closely related than either are withthe T&H data.

The Frankema figures are compiled according to decile distributions

agricultural land, excluding communal pastures and forests An ple calculus is presented in the appendix table A1.2 The concepts anddefinitions applied in the FAO’s agricultural surveys are rather consis-tent over time and across countries In order to further improve thespatial and temporal comparability of the land inequality figures, sev-eral additional criteria were used: roughly 60 surveys with an incom-plete coverage of agricultural land or an incomplete coverage of land

be-cause the surveys did not make a distinction between communal landholdings and single private land holdings Indeed, the estimated landGinis of socialist Eastern European countries in the 1970s and 1980sdisplay extreme land inequality since private small holders and com-munal holdings are both counted as individual farm holdings In fact,

close to independence: for the majority of Asian and African countriesthis is an observation close to 1960; for noncolonized countries and mostNew World countries it is the earliest observation available As tempo-

the prewar figures precede most of the considerable structural changes

in land distribution during the twentieth century, I regard this sample

as the most reliable proxy of historical land inequality, given the dataavailable Historical evidence for Latin American countries suggeststhat land inequality was very persistent during the nineteenth andtwentieth centuries, which is illustrated by the time-series data for

Table 1.1

Correlation of three datasets of land ginis, circa 1960

Taylor and Hudson

Deininger

Sources: Taylor and Hudson (1972, 267 269), Deininger and Olinto (2001, 24) and appen dix table A1.1.

Argentina, Brazil, and Chile in table A.1 For the United States, ada, Australia, and New Zealand there are early observations available

Table 1.2 presents the descriptive statistics of the sample subdivided

by 13 world regions The descriptive statistics reveal some interestingstylized facts First of all, the extraordinarily high levels of land in-equality observed in Latin American countries appear to be a coherentregional feature Looking at the world’s land inequality distribution,the top twenty countries include 16 in Latin America! The intraregionalvariation is, with the exception of the Caribbean islands, smaller thananywhere else in the world The assertion that there is a ‘‘Latin’’ type

of inequality (World Bank 2004) is indeed supported by a global parison of land inequality In Europe the countries with the most un-equal distribution of land are Spain, Portugal, and Italy It is quiteremarkable that land inequality in the former Iberian colonial mother-lands is as high as in an average Latin American country

com-The four East Asian countries are among the world’s most ian Ranking all land Ginis from low to high, South Korea ranks sec-ond, Taiwan ninth, Japan twelfth and China twentieth Except China,these East Asian countries are known for having realized growth withequity and it is often argued that, by dismantling the power of land-owning elites, land reforms have paved the way for a relatively equita-ble distribution of assets and income The steep drop in the Taiwaneseland Gini (from 53.9 in 1920 to 39.0 in 1960; see appendix table A1.1)illustrates the impact of land reforms carried out in the early 1950s(Fei, Ranis, and Kuo 1979; Frankema and Smits 2005)

egalitar-Perhaps the most remarkable stylized fact is the clear intraregionaldifference in Africa North Africa and the Middle East boast relativelyhigh levels of land inequality The distribution of land in Kenya, Tan-zania, Zambia, and South Africa is highly unequal, while in Mali,Burkina Faso, Cote d’Ivoire, Niger, and Senegal, land Ginis are consid-erably lower than the world’s average of 59.7 In many, mostly westand central African countries, land inequality appears to be confined,and taken as a region it is among the most egalitarian in the world

The hypotheses discussed in section 1.2 are specified in a simple country OLS model The first equation considers the explanation ofcross-country variation in postcolonial land inequality:

cross-Table 1.2

Descriptive statistics of land ginis subdivided into 13 world regions

North Africa and

Middle East

Source: Appendix table A1.1

Notes: CV means Coefficient of Variation; Obs means number of countries observed Countries included: South America: Argentina, Bolivia, Brazil, Chile, Colombia, Ecua dor, Guyana, Paraguay, Peru, Uruguay, Venezuela; Central America: Costa Rica, El Sal vador, Guatemala, Honduras, Mexico, Nicaragua, Panama; Caribbean: Barbados, Dominican Republic, Guadeloupe, Jamaica, Puerto Rico, Trinidad and Tobago, Haiti; East Asia: China, Japan, Korea, Republic of Taiwan; South Asia: Bangladesh, India, Iran, Nepal, Pakistan, Sri Lanka; South East Asia: Indonesia, Laos, Malaysia, Myanmar, Phil ippines, Singapore, Thailand, Vietnam; North Africa and Middle East: Algeria, Cyprus, Egypt, Israel, Iraq, Jordan, Kuwait, Libya, Morocco, Syria, Tunisia, Turkey; East and South sub Saharan Africa: Botswana, Ethiopia, Madagascar, Mauritius, Mozambique, Kenya, Lesotho, Reunion, South Africa, Swaziland, Tanzania, Zambia; West and Central sub Saharan Africa: Burkina Faso, Cameroon, Central African Republic, Cote d’Ivoire, Ghana, Guinea, Liberia, Mali, Niger, Senegal, Sierra Leone, Togo, Uganda; Western Off shoots: Australia, Canada, New Zealand, USA; Western Europe: Austria, Belgium, France, Germany, Greece, Ireland, Italy, Luxembourg, Malta, Netherlands, Portugal, Spain, Switzerland, UK; Eastern Europe: Czechoslovakia, Estonia, Latvia, Lithuania, Po land, Romania, Slovenia, Yugoslavia; Scandinavia: Denmark, Finland, Norway, Sweden.

y ¼ a þ b1x10þ b2x20þ e

where y refers to postcolonial land inequality, a is a constant, and e

of colonial factor endowments and the specific objectives of the land on land inequality The second equation considers the explana-tion of cross-country variation in current national income inequalitylevels:

where z refers to current income inequality, a is a constant, e is an errorterm, y refers to land inequality, and g represents the level of economicdevelopment, capturing the direct effect of land inequality on incomeinequality The m refers to a measure of the dominance of mineral

Institutions

Several proxies for the feasibility of tropical cash-crop production can

be used One rough variable is the mean annual temperature TEMP) (McArthur and Sachs 2001) Easterly (2002) applies variables ofland use (percentage share of land yielding a specific crop) and land suit-ability (percentage share of land suitable to the cultivation of a specificcrop) Since the use of land may be endogenous to the distribution ofland, the land suitability indicator is preferable The FAO providesdata on land suitability for, among others, bananas, cotton, maize, rice,sugarcane, and wheat For coffee, cocoa, rubber, and tobacco, only landuse data is available Land use data are transformed into dummies ofland suitability If a country devotes more than one percent of their ag-ricultural land to one of these crops and has a historical record as a netexporter of one of these crops (Mitchell 2003), the dummy is set at 1.Cash crops (bananas, coffee, cocoa, cotton, rubber, tobacco, and sugar)are expected to yield a positive effect, and food crops (maize, rice, andwheat) are expected to yield a negative effect The crops are alsoincluded as an aggregate measure: the CASHCROP aggregate includesbananas, cotton, and sugar and the FOODCROP aggregate includesmaize, rice, and wheat The land-labor ratio is approximated by thelog of population per square kilometer of agricultural land in the earlycolonial period (LNPOPDENSE) For most New World countries the

(MEAN-year 1700 or 1800 is used For African countries and most Asian tries the year 1900 is used Square kilometers of agricultural area arefrom Taylor and Hudson (1972, 303–305) and population estimates for

coun-1700, 1800, and 1900 are from McEvedy and Jones (1978)

The hypothesis that European colonial rule has interfered in the tribution of land is accounted for by a dummy variable (EURCOLDUMMY) with a value 1 for each former European colony The specialconditions in the Iberian colonies, such as the encomienda, the central-ist administration headed by the Spanish Crown, and the stratifiedsocial order are accounted for by another dummy variable (IBERCOLDUMMY) The log of historical settler mortality rates is included as aproxy for settler conditions of colonial settlement (LNSETMORT) Thedata are taken from Acemoglu, Johnson, and Robinson (2001) Forcountries that have not been subject to European colonial settlement,the mortality rate is set at zero The impact of the Catholic Church onthe distribution of land is covered by inserting the log of the percent-age share of Catholics (LNCATHOLICISM) in a year close to 1965(Taylor and Hudson 1972)

Regression with Fixed Effects

Income inequality figures are derived from the World Income ity Database (UNDP 2000) The sample consists of high-quality Ginicoefficients with national coverage for the latest year available in

(LANDGINI) is represented by the land Ginis discussed in section 1.4.The log of GDP per capita in 1990 (LNGDPPC) (Maddison 2003) isincluded as an interaction term to separate the direct effect from the in-direct effect of land inequality on income inequality

To control for other sources of inequality, a variable is includedthat captures the impact of mineral resource dominance in exports.Isham et al (2003) provide point-source estimates of mineral resourceabundance as the percentage share of one or two minerals dominat-ing national exports Mineral resources include crude petrol, gas, coal,

The variable is specified in log and denoted as LNMINERALS gional fixed effects are accounted for by including dummy variablesfor Latin America, sub-Saharan Africa, Asia, and (former) socialistcountries

Re-1.5 Results and Interpretation

A spatial regression analysis employing scattered historical data, crudeproxy variables, and large time lags without taking stock of the histori-cal dynamics in between cannot do much more than provide someguidance as to the validity of the hypotheses raised in the literature.However, it is better than nothing and helps to sharpen the set of ques-tions we entertain regarding the causes and consequences of landdistribution

In the appendix tables A1.3, A1.4, and A1.5 the descriptive statisticsand pair-wise correlations of the variables included in the regressionare presented Table 1.3 reports the results of the OLS regressions ofland inequality

The hypothesis that tropical climates are related to land inequalitycannot be sustained The mean annual temperature variable does have

a positive sign but is completely insignificant (see regression 1) over, none of the food or cash crop variables yields a significant result(see 2) Although the aggregate cash crop and food crop variable havethe predicted signs (positive and negative respectively), both vari-ables are insignificant The hypothesis that countries with substantialsources of land suitable to the production of cash crops have a moreunequal distribution of land cannot be sustained on the basis of thisanalysis and this does not come as a great surprise Large parts ofwest and central sub-Saharan Africa are suitable to the production ofcash crops but do not reveal high levels of land inequality Land abun-dant countries are clearly characterized by higher levels of land in-equality than densely populated countries The sign is consistent butthe coefficient is not robust The null hypothesis cannot be rejected atthe 90 percent confidence level in most regressions, except 6 and 8.Former European colonies display, on average, significantly higherlevels of land inequality and this result is even stronger for formerIberian colonies The significance of the Iberian colony dummy sup-ports the view that typical Iberian colonial institutions had an inherentenhancing effect on land inequality, apart from the local endowmentsstructure When controlled for the European colony dummy, the for-mer Iberian colonies display land Ginis between 0.08 to 0.18 pointshigher on average

More-Settler conditions also seem to make a substantial difference The log

of settler mortality rates, as a proxy for settler conditions, is consistentlynegative and significant at a 95 to 99 percent confidence level In colo-

nies with favorable settler conditions average land inequality is higher.

It is, moreover, noteworthy that the European colony dummy is onlysignificant when controlled for settler mortality rates (7) This findingsuggests that the feasibility of colonial settlement is an important con-ditional variable affecting the chance and scale of the redistribution ofland in a colonial context

Finally, the presence of the Catholic Church appears to have uted positively to land inequality at a 95 to 99 percent confidence level.This effect is maintained even when the ultimate control variable, theIberian colony dummy, is included The rapid diffusion and sustainedomnipresence of the Catholic Church further supports the view thatthe objectives of the Iberian mother countries have played a decisiverole, apart from the local conditions the Iberian settlers were con-fronted with

contrib-In sum, a typical land-unequal country at the end of the colonial riod is a land-abundant Catholic country, which has comparativelyfavorable settler conditions and specific Iberian institutions Such a de-scription, indeed, comes remarkably close to the average Latin Ameri-can country This impression arises from a global sample (covering allregions) including a dummy for Iberian colonies The nature and struc-ture of factor endowments did not play a decisive role, but provided acontext in which redistribution of land was more or less likely to takeplace The west African endowment conditions to produce cash cropssuch as sugar were probably comparable to those in the Caribbeanand Brazil, but Europeans could not enforce access into this area aseasily as in Latin America The objectives of European mother coun-tries were therefore really constrained by settler conditions in general;local endowments were important in the second instance

pe-In areas with unfavorable settler conditions, colonists adhered to atype of rent-seeking behavior that may have suppressed the rural econ-omy as a whole due to the burden of taxes, but did not affect the exist-ing distribution of land The redistribution of land from indigenouspeople to colonial settlers was part of a colonial strategy, which had ex-tractive as well as developmental objectives The expropriation of landand the exploitation of forced labor served extractive purposes, butinvestments in the local economy in order to raise the profitability ofagricultural production served developmental purposes Current levels

of income inequality in Latin America and sub-Saharan Africa areamong the highest in the world, but the characteristics and sources ofinequality differ distinctively