CONDITIONAL CASH TRANSFERS

A World Bank Policy Research Report

CONDITIONAL CASH TRANSFERS
REDUCING PRESENT AND FUTURE POVERTY Ariel Fiszbein and Norbert Schady with Francisco H.G. Ferreira, Margaret Grosh, Nial Kelleher, Pedro Olinto, and Emmanuel Skoufias

©2009 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 5 12 11 10 09

This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. Library of Congress Cataloging-in-Publication Data Fiszbein, Ariel, 1960– Conditional cash transfers : reducing present and future poverty / Ariel Fizbein, Norbert Schady. p. cm. Includes bibliographical references and index. ISBN 978-0-8213-7352-1 — ISBN 978-0-8213-7353-8 (electronic) 1. Transfer payments—Latin America—Case studies. 2. Economic assistance, Domestic—Latin America—Case studies. 3. Poverty— Government policy—Latin America—Case studies. I. Schady, Norbert Rüdiger, 1967- II. Title. HC130.P63F564 2009 338.91098—dc22 2008047645 Cover design: Drew Fasick Cover image: Chorale and Landscape by Paul Klee, 1921, 125 (gouache, pencil, and oil on paper, 35 x 31 cm); Zentrum Paul Klee, Bern, private loan. Used with permission. © 2009 Artists Right Society (ARS) New York/ VG Bild-Kunst, Bonn.

Contents

Foreword

xi xv

Acknowledgments Acronyms Overview xvii 1

The CCT Wave 3 The Arguments for CCTs 8 The Impacts of CCT Programs 11 Policy and Design Options 22 Complementary Interventions 24 CCTs in the Context of Social Protection Policies Conclusion 27

26

1.

Introduction

29
40

The CCT Wave 31 Theme and Variations 34 Outline of the Report and Issues Covered

2.

The Economic Rationale for Conditional Cash Transfers
Cash Transfers: Arguments in Support and Against The Microfoundations of Paternalism 51 A Political Economy Argument 59 Social Efficiency Arguments 64 Conclusion 65 46

45

3.

Design and Implementation Features of CCT Programs
Targeting in Practice Benefit Systems 80 67

67

v

CONTENTS

Conditions: Their Definition, Compliance Monitoring, and Enforcement 86 Monitoring and Evaluation 91 Intersectoral and Interinstitutional Challenges 97 Conclusion 100

4.

The Impact of CCTs on Consumption Poverty and Employment 103
Impact of CCTs on Household Consumption and Poverty 104 Analyzing Offsetting Behavioral Responses to CCTs 114 Long-Term Impacts of CCTs on Consumption 123 Conclusion 124

5.

The Impact of CCT Programs on the Accumulation of Human Capital 127
CCT Program Effects on the Use of Education and Health Services 128 Impact of CCTs on “Final” Outcomes in Education and Health 141 Cash, Behavioral Changes, and Outcomes 155 Conclusion 160

6.

CCTs: Policy and Design Options

165

When Is a CCT Program the Right Policy Instrument? 166 Designing an Efficient CCT Program 172 Adapting the Supply of Social Services 186 CCT Programs As Components of Social Protection Systems 195 Conclusion 200

Appendix A: Summary Tables

205 297

Appendix B: Review of CCT Impact Evaluations Notes References Index Boxes
2.1 2.2 2.3 3.1

315 329 351

Efficient Redistribution in the Presence of Market Failure 49 Investing Early in the Life Cycle 55 Fairness, Merit, and the “Deserving Poor” 61 Proxy Means Testing Where Administrative Capacity Is Low: Cambodia’s Scholarship Programs 71

vi

CONTENTS

3.2 3.3 3.4 3.5 3.6 4.1 4.2 4.3 5.1 5.2 5.3 5.4 6.1 6.2 6.3 6.4 6.5 6.6 1 2

Who Benefits from CCT Programs? 72 Analyzing Errors of Exclusion of CCT Programs, Brazil and Ecuador 76 Colombia’s Familias en Acción Sample-Based Site Monitoring, Selected Indicators 93 Evaluation Remains Important in CCTs 96 The International CCT Community of Practice 97 Is Time Spent in School a Perfect Substitute for Time Working? 117 Work Disincentive Effects of Social Assistance Programs in Developed Countries 118 Do Transfers Reduce the Supply of Adult Labor? Evidence from the 120 South African Pension Scheme Monetary Incentives to Students: Evidence from the United States 144 Impact of Oportunidades Transfers on Child Height in the Short Run 149 Do CCTs Help Protect Human Capital Investments during Economic Shocks? 161 Increasing School Enrollment without Improving Learning Outcomes 164 CCTs As an Instrument to Fight Social Exclusion 176 Does It Matter Who Receives the Cash? 183 Using Behavioral Models to Simulate the Effects of CCT Design Changes 184 Cost–Benefit Analysis of a CCT: The Case of Familias en Acción 188 Private Sector Delivery and CCT Programs 193 CCT Programs and the Financial Crisis 196

Figures
CCTs in the World, 1997 and 2008 4 Decision Tree Approach to Identifying CCT Programs as the Right Policy Instrument 12 3 Impact of CCTs on the Distribution of Consumption, Nicaragua and Honduras, 2002 15 4 Heterogeneity of Impacts by Socioeconomic Status, Nicaragua, 2000 21 5 Cognitive Development by Wealth Decile in Ecuador, 2003–04 25 1.1 CCTs in the World, 1997 and 2008 32 2.1 Choice of Investment in Children with Missing 52 Credit Markets 3B.1 Coverage Using per Capita Expenditure Deciles Gross and Net of the CCT Transfer, 2004 72 3.1 Coverage of CCT Programs, by Decile, Various Years 74

vii

CONTENTS

Benefit Incidence of CCT Programs, Various Years 85 Impact of CCTs on the Distribution of Consumption, Nicaragua and Honduras, 2002 109 4.2 Impact of CCTs on Food Shares in Ecuador and Nicaragua 112 5.1 Impact of Transfers of Different Magnitude on School Attendance in Cambodia, 2005–06 133 5.2 Oportunidades Impacts on School Enrollment, by Grade, 1998 135 5.3 Heterogeneity of Impacts by Socioeconomic Status, Nicaragua, 2000 136 5.4 Impact of Transfers Made by the Atención a Crisis Program on Stimulation in Early Childhood, 2005–06 159 6.1 Decision Tree Approach to Identifying CCT Programs as the Right Policy Instrument 167 6.2 Types of Households with Children 170 6.3 Grade Survival Profile, Ages 10–19, Poorest Quintile, Cambodia and Mexico 175 6B.1 Education Attainment, Bulgaria, 2007 176 6.4 Cognitive Development by Wealth Decile in Ecuador, 2003–04 180

3.2 4.1

Tables
Matrix of Program Size and Extent of Conditions 5 Impact of CCTs on per Capita Consumption, Various Years 13 Impact of CCTs on Poverty Measures, Various Years 14 Impact of CCTs on School Enrollment and Attendance, Various Years 17 5 Impact of CCTs on Health Center Visits by Children, Various Years 19 1.1 Matrix of Program Size and Extent of Conditions 33 1.2 Implementation of “Similar” Programs: Contrast between Mexico and Brazil 37 68 3.1 Targeting Methods Used in CCT Programs, by Region 3B.1 Coverage of Poor Households, Brazil and Ecuador 76 3.2 Generosity of CCT Programs, Various Years 84 3.3 Country and Program Variations in Monitoring and Penalties for Noncompliance with Conditions 88 4.1 Impact of CCTs on per Capita Consumption, Various Years 105 4.2 Impact of CCTs on Poverty Measures, Various Years 108 4.3 Impact of CCT Programs on Poverty Indexes at the National Level, Various Years 110 4.4 Impact of CCTs on Food Shares 113 4.5 Impact of Oportunidades on the Probability of Children Working 115 1 2 3 4

viii

CONTENTS

5.1

Impact of CCTs on School Enrollment and Attendance, Various Years 128 5.2 Impact of CCTs on Health Center Visits by Children, Various Years 137 5.3 Impact of CCTs on Vaccination and Immunization Rates, Various Years 140 5.4 Impact of CCTs on Child Growth Indicators 146 5.5 Effect of CCTs on Child Cognitive Development, Ecuador (2004–05) and Nicaragua (2005–06) 154 6.1 Examples of Supply-Side Interventions Complementary to a CCT 189 A.1 Targeting Methods Used in CCT Programs 206 208 A.2 Targeting Structure in CCT Programs A.3 Payment Schedules 211 A.4 CCT Programs-at-a-Glance 214

ix

Foreword

THE POTENTIAL IMPACT OF THE GLOBAL FINANCIAL CRISIS OF 2008

on living standards in the developing world has given renewed emphasis to the importance of social safety net programs. The right policies can be a smart investment in an uncertain world. This report reviews the evidence on conditional cash transfers (CCTs)—safety net programs that have become popular in developing countries over the last decade. It concludes that CCTs generally have been successful in reducing poverty and encouraging parents to invest in the health and education of their children. The CCT programs studied in the report span a range of low- and middle-income countries; large and small programs; and those that work at local, regional, and national levels. Although there are important differences between countries and regions in how CCTs are used, they all share one defining characteristic: they transfer cash while asking beneficiaries to make prespecified investments in child education and health. The largest CCTs, such as Brazil’s Bolsa Família and Mexico’s Oportunidades, cover millions of households. In Chile and Turkey, CCTs are focused more narrowly on extremely poor and socially excluded people, whereas CCTs in Bangladesh and Cambodia have been used to reduce gender disparities in education. Most recently, CCT pilot programs are being implemented in Sub-Saharan Africa to help alleviate the plight of millions of orphans in the wake of the continent’s devastating HIV/AIDS epidemic. CCTs are proven versatile programs, which largely explains why they have become so popular worldwide.

xi

FOREWORD

This report considers the impact that CCTs have had on current poverty, education, health, and nutrition outcomes. It draws heavily on a large number of carefully constructed impact evaluations of CCT programs. As the authors note, it would not have been possible to write this report without the efforts made by the administrators of CCT programs themselves, a number of academics, and staff at international organizations, including the World Bank, to encourage and sustain these evaluations, and to make the results widely available. This clearly is a legacy worth sustaining. By and large, CCTs have increased consumption levels among the poor. As a result, they have resulted in sometimes substantial reductions in poverty among beneficiaries—especially when the transfer has been generous, well targeted, and structured in a way that does not discourage recipients from taking other actions to escape poverty. Because CCTs provide a steady stream of income, they have helped buffer poor households from the worst effects of unemployment, catastrophic illness, and other sudden income shocks. And making cash transfers to women, as virtually all CCTs do, may have increased the bargaining power of women (itself an important goal in many contexts). In country after country, school enrollment has increased among CCT beneficiaries—especially among the poorest children, whose enrollment rates at the outset were the lowest. CCT beneficiaries also are more likely to have visited health providers for preventive checkups, to have had their children weighed and measured, and to have completed a schedule of immunizations. These are important accomplishments. Nevertheless, the report shows that the evidence of CCT impacts on final outcomes in health and education—achievement and cognitive development rather than school enrollment, child height for age rather than growth monitoring—is more mixed. An important challenge for the future is better understanding what complementary actions are necessary to ensure that CCTs have greater impact on these final outcomes. This report argues that these complementary actions broadly fall into two categories: policies that improve the quality of the supply of health and education services, and policies that help promote healthier and more stimulating environments for children in their homes. Even the best-designed CCT program cannot meet all the needs of a social protection system. It is, after all, only one branch of a larger tree that includes workfare, employment, and social pension programs. The

xii

FOREWORD

report therefore considers where CCTs should fit within a country’s social protection strategy. As the world navigates a period of deepening crisis, it has become vital to design and implement social protection systems that help vulnerable households weather shocks, while maximizing the efforts of developing countries to invest in children. CCTs are not the only programs appropriate for this purpose, but as the report argues, they surely can be a compelling part of the solution. Justin Lin Senior Vice President and Chief Economist The World Bank Joy Phumaphi Vice President, Human Development Network The World Bank January 2009

xiii

Acknowledgments

THIS POLICY RESEARCH REPORT WAS MANAGED BY ARIEL FISZBEIN

and Norbert Schady. Contributing members of the report were Francisco H.G. Ferreira, Margaret Grosh, Nial Kelleher, Pedro Olinto, and Emmanuel Skoufias. Substantial contributions were provided by Maria Victoria Fazio, Deon Filmer, Emanuela Galasso, Margaret Koziol, Phillippe Leite, Mette Nielsen, and Christine Weigand. The report benefited from comments from an advisory committee, which included Harold Alderman (World Bank), Orazio Attanasio (University College, London), Jere Behrman (University of Pennsylvania), Timothy Besley (London School of Economics), Santiago Levy (Inter-American Development Bank), Christina Paxson (Princeton University), and Laura Rawlings (World Bank). François Bourguignon, in his role as chief economist of the World Bank, provided strong encouragement and intellectual support to the report. The report was written under the direction and general supervision of Elizabeth King (research manager) and Martin Ravallion (director, Development Research Group). The authors have benefited from comments and useful input from Vivi Alatas, Colin Andrews, Caridad Araujo, Jehan Arulpragasam, Felipe Barrera, Maria Isabel Beltran, Nazmul Chaudhury, David Coady, Dante Contreras, Rafael Cortez, Aline Coudouel, Amit Dar, Gaurav Datt, Damien de Walque, Carlo del Ninno, Gershon Feder, Roberta Gatti, Paul Gertler, Rebekka Grun, Phillip Hay, Budi Hidayat, Jason Hobbs, Robert Holzmann, Emmanuel Jimenez, Theresa Jones, Peter Lanjouw, Benedicte Leroy De la Briere, Dan Levy, Maureen Lewis, Anja Linder, Kathy Lindert, Humberto Lopez, William Maloney, Andrew Mason, Alessandra Marini, Annamaria Milazzo, Amna Mir, xv

ACKNOWLEDGMENTS

Fernando Montenegro, Juan Martin Moreno, Edmundo Murrugarra, Shinsaku Nomura, Berk Ozler, Lucy Payton, Mansoora Rashid, Helena Ribe, Dena Ringold, Manuel Salazar, Tahseen Sayed, Nistha Sinha, Hedy Sladovich, Emma Sorensson, David Steel, Cornelia Tesliuc, Alan Winters, and Elif Yukseker. Financial support from the Development Impact Evaluation Initiative, the Knowledge for Change Program, and the Spanish Impact Evaluation Fund helped greatly in the preparation of this report.

xvi

Acronyms

AFDC AIN-C ATM BDH BEDP BANHCAFE BANSEFI CCT CESSP CSP CT-OVC EMA FFE FISDL FSSAP GDP GNP HIV/AIDS JFPR JPS LATE MEGS NGO OAP OVC

Aid to Families with Dependent Children Atención Integral de la Niñez en la Comunidad automated teller machine Bono de Desarrollo Humano Basic Education Development Project Banco Hondureño del Café Banco del Ahorro Nacional y Servicios Financieros conditional cash transfer Cambodia Education Sector Support Project Child Support Program Cash Transfer for Orphans and Vulnerable Children Education Maintenance Allowance Food for Education Fondo de Inversión Social para el Desarrollo Local Female Secondary School Assistance Program gross domestic product gross national product human immunodeficiency virus/acquired immunodeficiency syndrome Japan Fund for Poverty Reduction Jaring Pengamanan Sosial local average treatment effect Maharashtra Employment Guarantee Scheme nongovernmental organization Old-Age Pension Orphans and Vulnerable Children

xvii

ACRONYMS

PATH PCE PCI PESP PESRP PETI PKH PRAF RDD ROSC RPS SCAE SEDGAP SES SRMP SUF TAE/ILAE TANF TVIP UCT WDI

Program of Advancement through Health and Education per capita expenditure per capita income Primary Education Stipend Program Punjab Education Sector Reform Program Programa de Erradicação do Trabalho Infantil Program Keluarga Harapan Programa de Asignación Familiar regression discontinuity Reaching Out-of-School Children Red de Protección Social Subsidio Condicionado a la Asistencia Escolar–Bogotá Secondary Education Development and Girls Access Program socioeconomic status Social Risk Mitigation Project Subsidio Unitario Familiar Tarjeta de Asistencia Escolar/Incentivo a la Asistencia Escolar Temporary Assistance for Needy Families Test de Vocabulario en Imágenes Peabody unconditional cash transfer World Development Indicators

All dollar amounts are in U.S. dollars, unless otherwise indicated.

xviii

Overview

COnditiOnal Cash transfers (CCts) are prOgrams that

transfer cash, generally to poor households, on the condition that those households make prespecified investments in the human capital of their children. health and nutrition conditions generally require periodic checkups, growth monitoring, and vaccinations for children less than 5 years of age; perinatal care for mothers and attendance by mothers at periodic health information talks. education conditions usually include school enrollment, attendance on 80–85 percent of school days, and occasionally some measure of performance. most CCt programs transfer the money to the mother of the household or to the student in some circumstances. Countries have been adopting or considering adoption of CCt programs at a prodigious rate. Virtually every country in latin america has such a program. elsewhere, there are large-scale programs in Bangladesh, indonesia, and turkey, and pilot programs in Cambodia, malawi, morocco, pakistan, and south africa, among others. interest in programs that seek to use cash to incentivize household investments in child schooling has spread from developing to developed countries— most recently to programs in new York City and Washington, dC. in some countries, CCts have become the largest social assistance program, covering millions of households, as is the case in Brazil and mexico. CCts have been hailed as a way of reducing inequality, especially in the very unequal countries in latin america; helping households break out of a vicious cycle whereby poverty is transmitted from one generation to another; promoting child health, nutrition, and schooling; and helping countries meet the millennium development goals. do those and other claims make sense? are they supported by 1

COnditiOnal Cash transfers: reduCing present and future pOVertY

the available empirical evidence? What does all of this imply for the way in which countries that have CCts should structure or reform the programs? What about countries that do not have CCts but are considering implementing them, often in circumstances very different from those in which the programs were first introduced? this report seeks to answer those and other related questions. specifically, it lays out a conceptual framework that considers the economic and political rationale for CCts; it reviews the very rich evidence that has accumulated on CCts, especially arising from impact evaluations; it discusses how the conceptual framework and the evidence on impacts should inform the design of CCt programs in practice; and it considers where CCts fit in the context of broader social policies. the report shows that there is good evidence that CCts have improved the lives of poor people. transfers generally have been well targeted to poor households, have raised consumption levels, and have reduced poverty—by a substantial amount in some countries. Offsetting adjustments that could have blunted the impact of transfers—such as reductions in the labor market participation of beneficiaries—have been relatively modest. moreover, CCt programs often have provided an entry point to reforming badly targeted subsidies and upgrading the quality of safety nets. the report thus argues that CCts have been an effective way to redistribute income to the poor, while recognizing that even the best-designed and best-managed program cannot fulfill all of the needs of a comprehensive social protection system. CCts therefore need to be complemented with other interventions, such as workfare or employment programs and social pensions. the report also considers the rationale for conditioning the transfers on the use of specific health and education services by program beneficiaries. Conditions can be justified if households are underinvesting in the human capital of their children—for example, if they hold incorrect beliefs about the returns to these investments; if there is “incomplete altruism” between parents and their children; or if there are large externalities to investments in health and education. political economy considerations also may favor conditional over unconditional transfers: taxpayers may be more likely to support transfers to the poor if they are linked to efforts to overcome poverty in the long term, particularly when the efforts involve actions to improve the welfare of children. CCts have led poor households to make more use of health and education services, a key objective for which they were designed. 2

OVerVieW

nevertheless, the evidence on improvements in final outcomes in health and education is more mixed. thus CCts have increased the likelihood that households will take their children for preventive health checkups, but that has not always led to better child nutritional status; school enrollment rates have increased substantially among program beneficiaries, but there is little evidence of improvements in learning outcomes. these findings suggest that to maximize their potential effects on the accumulation of human capital, CCts should be combined with other programs to improve the quality of the supply of health and education services, and should provide other supporting services. they also suggest the need to experiment with conditions that focus on outcomes rather than on the use of services alone.

The CCT Wave interest in and the scope of CCt programs have grown enormously in the last 10 years. the maps shown in figure 1 reveal the expansion between 1997 and 2008. paralleling the rise in the number of countries with CCt programs has been an increase in the size of some programs. mexico’s prOgresa started with approximately 300,000 beneficiary households in 1997, but now covers 5 million households. (this program was renamed Oportunidades in 2001. in this report we will refer to the program as Oportunidades.) Brazil started with municipal Bolsa escola programs in Brasilia and the municipality of Campinas. those programs led to replication by local governments, followed by formulation of sector-specific federal programs, and then their unification and reform. today the federal Bolsa família program serves 11 million families (46 million people). in other countries, the increase in size has been less explosive but still notable. in Colombia, for instance, the program’s initial goal was 400,000 households, but it had expanded to cover 1.5 million beneficiary households by 2007. CCts vary a great deal in scope. some programs are nationwide, others are niche programs that serve a regional or narrow target population, and yet others are small-scale pilot efforts. some programs require that households receiving transfers comply only with schooling conditions; others, especially programs in latin america and the Caribbean, require that households comply with both schooling and 3

COnditiOnal Cash transfers: reduCing present and future pOVertY

Figure 1 CCTs in the World, 1997 and 2008

1997

MEXICO

BANGLADESH BRAZIL

2008

TURKEY PAKISTAN DOMINICAN REP. MEXICO JAMAICA GUATEMALA HONDURAS EL SALVADOR NICARAGUA COSTA RICA COLOMBIA PANAMA ECUADOR BRAZIL PERU BOLIVIA PARAGUAY CHILE ARGENTINA BURKINA FASO NIGERIA INDIA YEMEN KENYA BANGLADESH INDONESIA CAMBODIA PHILIPPINES

Source: World Bank.

health conditions. table 1 presents a partial list of the CCt programs considered in this report. the list is not exhaustive in that it does not cover all existing programs. there are additional programs in operation for which little information was available, and some programs fit the CCt label less well than do others. 4

OVerVieW

Table 1

Matrix of Program Size and Extent of Conditions
Conditions

Program size/Target nationwide

Education and health Bolsa família (Brazil) Oportunidades (mexico) Bono de desarrollo humano (ecuador) familias en acción (Colombia) program of advancement through health and education (Jamaica)

Education only Bolsa escola (Brazil) Jaring pengamanan sosial (indonesia)

niche (regional or narrow target population)

Chile solidario social risk mitigation project (turkey)

female secondary school assistance program (Bangladesh) Japan fund for poverty reduction (Cambodia) education sector support project (Cambodia) Basic education development project (Yemen)

small scale/pilot

programa de asignación familiar (honduras) Cash transfer for Orphans and Vulnerable Children (Kenya) atención a Crisis (nicaragua) red de protección social (nicaragua)

subsidio Condicionado a la asistencia escolar– Bogotá (Colombia) punjab education sector reform program (pakistan)

Source: authors’ compilation.

the role of CCt programs in social policy varies from place to place as a consequence of differences in both program design and the context in which they operate. most obviously, CCt programs vary with respect to pertinent measures of size. in terms of absolute coverage, they range from 11 million families (Brazil) to 215,000 households (Chile) to pilot programs with a few thousand families (Kenya, nicaragua). in terms of relative coverage, they range from approximately 40 percent of the population (ecuador) to about 20 percent (Brazil, mexico) to 1 percent (Cambodia). in terms of budget, the costs range from about 0.50 percent of gross domestic product (gdp) in such countries as Brazil, ecuador, and mexico to 0.08 percent of gdp (Chile). the generosity of benefits ranges from 20 percent of mean household consumption in mexico, to 4 percent in honduras, and to even less for programs in Bangladesh, Cambodia, and pakistan. many of the CCt programs in middle-income countries have pursued an integrated approach to poverty reduction, balancing goals of 5

COnditiOnal Cash transfers: reduCing present and future pOVertY

social assistance and human capital formation. they cover children from birth (or before) through the mid-teens, with conditions on health care use for children from birth to age 5 or 6 and with conditions on school enrollment thereafter. programs usually are administered by ministries of social welfare or freestanding agencies under the presidency. examples of that type of CCt include the programs in Brazil, Colombia, el salvador, Jamaica, mexico, panama, and turkey. mexico’s Oportunidades is one of the iconic cases. the program started early, its evolution has been carried out thoughtfully, and it has been successful. What really makes mexico’s program iconic are the successive waves of data collected to evaluate its impact, the placement of those data in the public domain, and the resulting hundreds of papers and thousands of references that such dissemination has generated. Brazil also is exemplary in its use of CCts. it started early, its programs have evolved enormously, and the current program (Bolsa família) is similar to mexico’s program in coverage and importance. in various respects, Brazil’s Bolsa família program provides something of an interesting contrast to the mexican case—the issue of federalism is more in the forefront; it takes a softer, more gradual tack on conditions; and puts a shade more emphasis on redistribution than on human capital formation. also, unlike Oportunidades, the Brazilian programs did not explicitly incorporate impact evaluations in their design; as a result, much less is known about the effect they have had on consumption, poverty, health, nutrition, and education. Chile solidario works in a very different way to fill a different niche. the program is targeted only to extremely poor people, about 5 percent of Chile’s population. it differs notably from the classic CCt design by customizing conditions. families initially work intensely with social workers to understand actions that could help them get out of extreme poverty. they then commit to action plans that become the householdspecific conditions for receiving the benefit. the cash transfer itself really is intended only to motivate clients to make use of social workers’ services. thus far, Chile solidario is a model unto itself, although other programs are moving to emulate it to a degree. another branch of the CCt program family focuses on education in low-income countries. the programs usually cover a more narrow segment of education—some only secondary (Bangladesh’s female secondary school assistance program [fssap], Cambodia’s Japan fund for poverty reduction [Jfpr], and Cambodia education sector support 6

OVerVieW

project [Cessp]), some only primary (programs in Bolivia and Kenya and proposals in nigeria and tanzania), and occasionally both (indonesia’s Jaring pengamanan sosial [Jps] program). the genesis of these programs is rather varied. in Bangladesh, the fssap was part of a strategy to close a then-significant gender gap in education. in indonesia, the Jps program was instituted following the east asian financial crisis to prevent students from dropping out. in Kenya and tanzania, the programs are geared especially to coping with the crisis of orphans and vulnerable children, a crisis that has burgeoned in the wake of hiV/aids. CCt programs require the same systems as other transfer programs: at minimum, (1) a means to establish the eligibility of clients and enroll them in the program, and (2) a mechanism to pay their benefits. strong monitoring and evaluation systems also are desirable. CCts further require a means to monitor compliance with conditions and to coordinate among the several institutions involved in operating the program. in general, CCt programs have handled these systems rather well and, in some cases, they have been leaders in modernizing social assistance practice. almost all CCts have tried to target their benefits rather narrowly to the poor through a combination of geographic and household targeting (mostly via proxy means testing). moreover, many programs use community-based targeting or community vetting of eligibility lists to increase transparency. in many cases, CCts have been the drivers for developing poverty maps or household targeting systems in their countries, or for upgrades to them. indeed, it would not be an exaggeration to say that CCts have moved forward the state of the art and standards for targeted programs generally. a number of CCt programs have had unusually proactive management based on cutting-edge technical systems, especially with respect to monitoring and evaluation. two features inherent to CCts—the number of actors involved and the need for extensive information management to verify compliance with conditions—may have interacted in ways that have spurred creative development in monitoring and management. this excellence in systems, and the high degree of transparency in documentation and information that characterizes most programs, has contributed to the attraction of CCts, although they are not inherent to them. the evaluation culture around CCts is quite strong, well beyond traditional practice in social policy. many programs have conducted impact evaluations with credible counterfactuals. Of those programs, a large share used experimental methods, 7

COnditiOnal Cash transfers: reduCing present and future pOVertY

at least initially. this culture of evaluation is spreading not only from one CCt program to another, but also from CCts to other programs within the same countries. the role and design of CCt programs is evolving. early successes with the basic model are prompting countries to address second and third rounds of challenges, including the following: should the emphasis on expanding the supply of services be complemented with efforts to improve the quality of those services? should the range or definition of conditions be changed, for example, to reward performance instead of, or in addition to, mere service use? What can be done to ensure that youth who are aging out of the school support provided by the program can get jobs or further training? What should be the balance between targeting younger and older children? in some countries, CCt programs themselves are addressing these challenges through adjustments to their basic design; in other cases, they are catalyzing changes in other programs.

The Arguments for CCTs although market-driven economic growth is likely to be the main driver of poverty reduction in most countries, markets cannot do it alone. public policy plays a central role in providing the institutional foundations within which markets operate, in providing public goods, and in correcting market failures. in addition to laying the foundations for economic growth, policy can supplement the effects of growth on poverty reduction, and one of the instruments that governments can use to that end is direct redistribution of resources to poor households. direct cash transfers have opportunity costs (in terms of forgone alternative public investments) and may have some perverse incentive effects on recipients, but there is a growing body of evidence that in some cases transfers may be both equitable and efficient. Conditional cash transfers make payments to poor households on the condition that those households invest in the human capital of their children in certain prespecified ways. Because attaching a constraint on the behavior of people one is trying to help is an unorthodox approach for economists, this report reviews the conceptual arguments for making cash transfers conditionally. there are two broad sets of arguments for attaching conditions to cash transfers. the first set applies if private investment in children’s 8

OVerVieW

human capital is thought to be too low. the second set applies if political economy conditions show little support for redistribution unless it is seen to be conditioned on “good behavior” by the “deserving poor.” under the first group of arguments, private investment in human capital can be “too low” in two different senses. first, it can be below even the private optimal level for the individual children in question if household decision makers hold persistently misguided beliefs about either the nature of the process of investments in child education and health or the subsequent returns to these investments. for instance, parents may believe that earnings respond to education less elastically than they actually do. in practice, there is some evidence of this from developing countries. among 15- to 25-year-olds in mexico, the expected returns to schooling (calculated from questions asked of respondents) are substantially lower than the realized returns (the mincerian returns calculated from a household survey), especially among children of fathers with low education levels (attanasio and Kaufmann 2008). in the dominican republic, eighth-grade students estimate the rate of return to secondary school to be only one quarter to one third of the rate derived from an income survey (Jensen 2006). parents also may discount the future more heavily than they should, perhaps especially with regard to the returns on investments in their children—a case of “incomplete altruism.” a slightly different but equally plausible version of this problem is a conflict of interest between the parents themselves as opposed to, or in addition to, one between parents and children. mothers’ objectives may be more closely aligned with those of all her children or, perhaps, especially with those of her daughters.1 that alignment often is given as a justification for giving the cash transfer to the mother rather than to the father, as is common practice in most CCt programs. in many countries in south asia, girls’ schooling lags well behind that of boys, even though the returns to female education—both in wages, and in terms of child health—are at least as large as those for males. low levels of investment in girls’ schooling may be rational from the viewpoint of parents who are thinking of their own welfare (either because girls are more costly in terms of dowries or because boys are more likely to take care of their parents than are girls who move to their husbands’ homes upon marriage), but they are prima facie evidence of a socially inefficient outcome. CCts that compel parents to send their daughters to school are one way to address inefficient and inequitable gender disparities. 9

COnditiOnal Cash transfers: reduCing present and future pOVertY

in general, these informational, principal-agent, or behavioral arguments can be seen as providing microfoundations for much older paternalistic arguments for redistribution in-kind or with strings attached. the second sense in which private investments in children’s health and education can be “too low” is that the private optimal level may be below the social optimal level. that situation could occur if there are positive externalities from education and health across households. empirically, many health investments have important external benefits.2 in the case of education, externalities might arise if there are increasing returns to skilled labor in production, at the aggregate level, or if education lowers crime. how large these externalities are and whether (conditioned) cash transfers are the most effective instruments to correct for them, however, remains to be determined. in most countries, education and health services are already heavily subsidized. in many cases, they are publicly provided free of charge. to argue for an additional subsidy that compensates households for some of the indirect or opportunity costs of using these services, on the basis of the externality alone, would require showing that those externalities are quite large. the political economy family of arguments centers around the notion that targeting tends to weaken the support for redistribution because it reduces the number of beneficiaries relative to the number of those who are taxed to finance the program. Whereas the response most commonly considered in the literature is to establish broad-based redistribution that includes the middle class, an alternative is to appeal to the altruistic motive of voters: the same people who object to targeted transfers as “pure handouts” might support them if they are part of a “social contract” that requires recipients to take a number of concrete steps to improve their lives or those of their children. the notion that CCt programs constitute a new form of social contract between the state and beneficiaries is apparent in the use of the term co-responsibilities (instead of conditions) in a majority of programs, at least in latin america. When conditions are seen as co-responsibilities, they appear to treat the recipient more as an adult capable of agency to resolve his or her own problems. the state is seen as a partner in the process, not a nanny. this latter interpretation is particularly plausible when the counterfactual to a CCt is not an automatic, transparent, unconditional cash grant seen as a citizen’s entitlement (which is close to the textbook concept of an unconditional transfer), but is instead a myriad of ad hoc and mostly in-kind transfers intermediated through 10

OVerVieW

various service providers, nongovernmental organizations, and local governments. under those circumstances, conditioning the transfers on “good behavior” may be perceived as less paternalistic than the alternative of conditioning transfers on say, voting for a certain party or belonging to a given social organization. moreover, the fact that the conditions are focused on building the human capital of children (rather than simply supporting parents) adds to CCts’ political acceptability as an instrument to promote opportunities; after all, it is hard to blame children for being poor. in that sense, using public resources to support the human capital development of poor children makes a CCt a poverty reduction program rather than a social assistance one. making payments to mothers also resonates with well-accepted beliefs (mostly supported by evidence, as shown above) that women will tend to put funds to better use than will men. the conclusion is that even in situations where a narrow technical assessment might suggest that an unconditional transfer is more appropriate than a CCt (say, because there is no evidence of imperfect information or incomplete altruism in poor families), conditions might be justified because they lead to a preferable political economy equilibrium. the political process may make significant cash transfers to the poor close to impossible unless those transfers are tied somehow to clear evidence of beneficiaries’ “positive behaviors.” the latin american experience suggests that in the absence of dramatic political shifts, the increasing trend toward cash-based redistribution schemes has been associated with the use of some form of conditioned grants. in sum, when there is a strong rationale to redistribute, a CCt can be justified under two broad sets of conditions: first, when private investment in human capital among the poor is suboptimal from a social point of view and, second, when conditions are necessary for political economy reasons (that is, redistribution is politically feasible only when conditioned on good behavior). this framework can be extended by identifying critical questions that can guide the decision whether to have a CCt program, as depicted in figure 2.

The Impacts of CCT Programs
Beginning with the mexican program Oportunidades, an important feature of CCt programs has been strong emphasis on credible evaluations of their impact on various outcomes. this report draws heavily 11

COnditiOnal Cash transfers: reduCing present and future pOVertY

Figure 2 Decision Tree Approach to Identifying CCT Programs as the Right Policy Instrument
Conditional Cash Transfer Consider trade-offs YES YES

YES Redistribute or not?

Underinvestment in human capital?

NO

Political economy “antipoor”?

NO

Unconditional Cash Transfer

FACTORS TO CONSIDER • Poverty and inequality levels • Availability of resources • Efficiency costs and benefits of redistribution

FACTORS TO CONSIDER • Misinformation (e.g., differences between expected and realized rates of return) • Agency problems (e.g., large gender differences in human capital) • Externalities (e.g., high incidence of crime in poor neighborhoods)

FACTORS TO CONSIDER • Existence of cash transfers targeting the poor • Transparency in eligibility criteria for social assistance • Views on distributive justice

Source: authors’ illustration.

on those evaluations. indeed, it would not have been possible to write the report without the efforts of the program administrators themselves, international donors, and academics around the world to ensure the high quality of many of the evaluations. the accumulating evidence of positive impacts has been instrumental both in sustaining existing programs and in encouraging the establishment of similar programs in other developing countries. most CCts seek both to reduce consumption poverty and to encourage investments in the education and health of children. the report carefully considers the evidence of programs’ impacts on those two dimensions of well-being. The Impact on Consumption, Poverty, and Labor Market Participation By and large, CCts have had positive effects on household consumption and on poverty (as measured by the headcount index, the poverty gap, or the squared poverty gap). tables 2 and 3 summarize the evidence. 12

Table 2
Colombia 2002 0.85 1.19 1.12 1.13 0.79 0.68 0.59 0.58 0.59 0.63 0.53 2006 2003 2005 2000 2002 1998 Jun. 1999 Oct. 1999 2000 2001 Ecuador Honduras Mexico Nicaragua 2002 0.52

Impact of CCTs on per Capita Consumption, Various Years

Consumption 0.83 0.89

Brazil 2002

Cambodia 2007

median daily per capita consumption of control households (current us$) 0.06 0.02 0.12 0.13 0.08 0.08 0.06 0.06 0.12 0.14 0.13 0.16

daily per capita transfer (current us$) 8 2–3 17 13 8 7 9 11 21 20

0.15

0.15

ratio of transfer to consumption (%)a 7.0** B a 10.0** a B a 7.0* B

19

29

31

30

impact on per capita consumption for the median household (%)

7.8**

8.3**

a

29.3** 20.6**

Source: authors’ calculations for all countries in the table except Colombia. for Colombia, see institute for fiscal studies, econometría, and sistemas especializados de información (2006). Note: the estimated impacts presented here are not always equal to the unconditional double difference estimates because some regressions control for other correlates. the impact for honduras was obtained from 2002 regression only. the impacts for mexico are all for single equation cross-sectional regressions for each year. the lack of impact in 1998 is likely the result of the fact that this survey was carried out just a few months after the start of the program. figures are in us$ obtained through the official exchange rates observed at the time of the surveys. in the case of Oportunidades in mexico, the 1998 figures are for a few months after the start of the program. in the case of Bolsa alimentação in Brazil, per capita consumption figures are for more than a year after the start of the program. a. the transfer amounts as a proportion of per capita expenditures (or consumption) are not the same across all tables in the report because of differences in the surveys used, including their coverage and year. a. Baseline, before households in CCt treatment group received transfers. B. no significant impact on consumption. * significant at the 10 percent level. ** significant at the 5 percent level.

OVerVieW

13

14
Colombia 2002 Control impact Control impact Control impact 0.53 a 0.43 –0.02** 0.30 a 0.36 –0.02* 0.28 B 0.35 –0.03** 0.58 a 0.54 –0.07** 0.49 a 0.54 –0.02* 0.47 0.01* 0.55 –0.03** 0.95 a 0.90 –0.03* 0.88 a 0.91 B 0.89 0.02** 0.93 –0.01** 0.94 0.00 0.56 –0.02** 0.36 –0.03** 2006 2000 2002 1998 Jun. 1999 Oct. 1999 Honduras Mexico 2000 0.84 a 0.43 a 0.26 a Nicaragua 2001 0.91 –0.07** 0.50 –0.13** 0.32 –0.12** 2002 0.90 –0.05** 0.50 –0.09** 0.32 –0.09**

Table 3

Impact of CCTs on Poverty Measures, Various Years

Poverty measure

headcount index

poverty gap

COnditiOnal Cash transfers: reduCing present and future pOVertY

squared poverty gap

Source: authors’ calculations. Note: We exclude Cambodia and ecuador from this table because the CCt did not have an effect on median consumption in those countries and so it is not surprising that it did not reduce poverty. We also exclude the Brazilian Bolsa alimentação program because the evaluation sample is not representative of the program’s target population, which makes the analysis of the impact on poverty less informative. for honduras, mexico, and nicaragua, calculations were done via regression of household level foster-greer-thorbecke indicator on treatment dummy and other explanatory variables. using the evaluation sample of each program, we compute P(i,t,a) = (z – y(i,t) / z)a * poor(i,t), for alpha = 0, 1, and 2; and for each household, where y(i,t) is household i’s level of consumption per capita at year t, z is the country-specific poverty line, and poor(i,t) is an indicator function that equals 1 if the household is poor and equals 0 otherwise. for honduras, the poverty line used was lps 24.6 per capita per day in 2000 lempiras. expenditure values for 2002 were deflated to 2000 lempiras. for nicaragua, we used C$13.87 per capita per day in 2000 córdobas. expenditure values for 2001 and 2002 were deflated to 2000 córdobas. for mexico, we used the value of the Canasta Básica of 1997, which was m$320 per capita per month. We inflated this value of the Canasta Básica for 1998 and 1999 using the Canasta Básica price index found at: http://www.banxico .org.mx/polmoneinflacion/estadisticas/indicesprecios/indicespreciosConsumidor.html. therefore, for October 1998, we used m$320 × 1.134. for June 1999, we used m$320 × 1.280. for October 1999, we used m$320 × 1.314. for Colombia (see institute for fiscal studies, econometría, and sistemas especializados de información 2006), the estimated impacts presented here are not equal to the unconditional double difference estimates because regressions control for other correlates. the impact for honduras was obtained from 2002 regression only. the impacts for mexico are all for single equation cross-sectional regressions for each year. a. Baseline, before households in CCt treatment group received transfers. B. no significant impact on poverty measure. * significant at the 10 percent level. ** significant at the 5 percent level.

OVerVieW

table 2 shows that the largest consumption impacts are found when the transfer amount is generous (as with the red de protección social [rps] program in nicaragua). moreover, because transfers generally are well targeted to the poor, the effects on consumption have translated into impacts on poverty, as is shown in table 3. some of the reductions in poverty are quite large. in nicaragua, for example, poverty fell by 5–9 points (using the 2002 data). another way of measuring the impact of CCts on welfare is to compare the cumulative distribution of consumption per capita between those who receive the transfer and those who do not. the advantage of this method is that it does not rely on the selection of a poverty line, which can be somewhat arbitrary. if the cumulative distribution for recipient households lies completely to the right of the distribution for control households—so-called first-order stochastic dominance—current welfare is unambiguously improved by CCts. that is clearly the case for rps beneficiaries in nicaragua, as shown in panel a of figure 3. panel B shows an improvement that is much smaller for honduras—a finding that is not surprising given the smaller magnitude of the transfer. moreover, CCts have affected not only the overall level of consumption, but also the composition of consumption. there is a good deal of

Figure 3

Impact of CCTs on the Distribution of Consumption, Nicaragua and Honduras, 2002
A. Nicaragua B. Honduras
1.0 0.8 0.6 CDF 0.4 0.2 0

1.0 0.8 0.6 CDF 0.4 0.2 0 0

1

2

3

4

0

1

2

3

4

5

Log per capita expenditure (córdobas) Treatment Source: authors’ calculations. Note: Cdf = cumulative distribution function. Control

Log per capita expenditure (lempiras)

15

COnditiOnal Cash transfers: reduCing present and future pOVertY

evidence that households that receive CCts spend more on food and, within the food basket, on higher-quality sources of nutrients than do households that do not receive the transfer but have comparable overall income or consumption levels.3 an important concern when CCts first were launched was that they would result in large reductions in the labor market participation of adults—either because beneficiaries would choose to consume more leisure at higher income levels or because they would cut back on work in order to continue to appear to be “poor enough” to be eligible for transfers. in practice, CCts appear to have had at most modest disincentive effects on adult work. research on Cambodia, ecuador, and mexico shows that adults in households that received transfers did not reduce their work effort. although CCts generally have not resulted in reductions in the labor market participation of adults, they have led to substantial decreases in child labor—as was intended by many of the programs. reduced child work by CCt beneficiaries has been found in Brazil, Cambodia, ecuador, mexico, and nicaragua. in some cases, the reductions are quite large. in Cambodia, for example, the average child receiving the transfer was 10 percentage points less likely to work for pay.4 in addition to possible reductions in labor market participation, a number of behavioral changes by households could have blunted the impact of CCts on consumption and on poverty. in practice, all of these offsetting adjustments to transfers appear to have been small. thus CCts generally have not crowded out remittances and other transfers; they have had only small impacts on fertility, at least in the short run; and they have not had substantial local general equilibrium effects, such as increases in prices or wages. finally, there is some evidence that CCt program beneficiaries invest part of the transfer, that the returns to these investments can result in higher consumption levels in the medium term (in mexico, but not in nicaragua), and that transfers made by CCt programs help households smooth consumption during adverse shocks. 5 The Impact of CCTs on Education and Health Outcomes in country after country, CCts have led to significant and, in some cases, substantial increases in the use of services (tables 4 and 5). school enrollment rates have increased among program beneficiaries, especially 16

OVerVieW

among those who had low enrollment rates at the beginning. these impacts are found in the middle-income countries where CCt programs were first implemented (for example, mexico); in lower-income countries in latin america (for example, honduras and nicaragua); and in low-income countries in other regions (for example, Bangladesh, Cambodia, and pakistan). CCt programs also have had a positive

Table 4

Impact of CCTs on School Enrollment and Attendance, Various Years
Age/Gender/ Grade Baseline enrollment (%) Transfer (% of PCE)b Evaluation method

Country

Program

Impacta

Reference

Latin American and Caribbean countries
Chile Colombia Chile solidario ages 6–15 60.7 91.7 63.2 75.2 66.4 7.5*** (3.0) 2.1** (1.0) 5.6*** (1.8) 10.3** (4.8) 3.3*** (0.3) 0.5** (0.2) 10 9 iV, randomized randomized 7 17 rdd psm, dd galasso (2006) attanasio, fitzsimmons, and gómez (2005) schady and araujo (2008) glewwe and Olinto (2004) levy and Ohls (2007)

familias en acción ages 8–13 ages 14–17

ecuador honduras

Bono de desarrollo ages 6–17 humano programa de asignación familiar program of advancement through health and education Oportunidades ages 6–13

Jamaica

ages 7–17

18 daysc

10

rdd

mexico

grades 0–5 grade 6 grades 7–9

94.0 45.0 42.5 90.5 72.0

1.9 (25.0) 8.7*** (0.4) 0.6 (56.4)

20

randomized

schultz (2004)

nicaragua nicaragua

atención a Crisis red de protección social

ages 7–15 ages 7–13

6.6*** 18 (0.9) 12.8*** 27 (4.3)

randomized randomized

macours and Vakis (2008) maluccio and flores (2005) continued

17

COnditiOnal Cash transfers: reduCing present and future pOVertY

Table 4

continued
Age/Gender/ Grade Baseline enrollment (%) Transfer (% of PCE)b Evaluation method

Country

Program

Impacta

Reference

Non–Latin American and Caribbean countries
Bangladesh female secondary school assistance program Japan fund for poverty reduction Cambodia education sector support project punjab education sector reform program social risk mitigation project ages 11–18 (girls) grades 7–9 (girls) grades 7–9 44.1 12.0** (5.1) 31.3*** (2.3) 21.4*** (4.0) 11.1*** (3.8) –3.0* n.a. 5.2 n.a. 0.6 fe Khandker, pitt, and fuwa (2003) filmer and schady (2008) filmer and schady (2009c) Chaudhury and parajuli (2008) ahmed et al. (2007)

Cambodia Cambodia

65.0 65.0

2–3 2–3

dd rdd

pakistan

ages 10–14 (girls) primary school secondary school

29.0

3

ddd

turkey

87.9 39.2

6

rdd

Source: authors’ compilation. Note: dd = difference-in-differences; ddd = difference-in-difference-in-differences; fe = fixed effects; iV = instrumental variables; n.a. = not available; pCe = per capita expenditure; psm = propensity score matching; rdd = regression discontinuity design. this table contains unweighted means for the coefficients for Colombia ages 8–13 and 14–17, Chile ages 4–5 and 6–15, and mexico grades 0–5 and 7–9. the standard errors in each case are the square roots of the averaged variances of these estimates. a. the column for “impact” reports the coefficient and standard error (in parentheses); the unit is percentage points, with the exception of the Jamaican path program, where the unit is days. b. the transfer amounts as a proportion of per capita expenditures (or consumption) are not the same across all tables in the report because of differences in the surveys used, including their coverage and year. c. impacts were measured in Jamaica only for student attendance over a 20-day reference period. the baseline enrollment rate prior to path was 96 percent. * significant at the 10 percent level. ** significant at the 5 percent level. *** significant at the 1 percent level.

effect on the use of preventive health services, although the evidence is less clear-cut than with school enrollment. moreover, because CCt program effects on utilization are concentrated among households who were least likely to use services in the absence of the intervention, CCts have contributed to substantial

18

OVerVieW

reductions in preexisting disparities in access to education and health. in Bangladesh, pakistan, and turkey, where school enrollment rates among girls were lower than among boys, CCts have helped reduce this gender gap. in Cambodia, the Jfpr program eliminated sharp socioeconomic gradients in enrollment among eligible households— although the coverage of the program was quite small. and in nicaragua, the CCt impact on both school enrollment and growth monitoring was largest among extremely poor households, as shown

Table 5

Impact of CCTs on Health Center Visits by Children, Various Years
Age range (years) 0–6 0–1 2–4 4+ Baseline level (%)a Impactb 17.6 n.a. n.a. n.a. n.a. 2.4 (2.7) 22.8*** (6.7) 33.2*** (11.5) 1.5* (0.8) 2.7 (3.8) 20.2*** (4.7) 10 r paxson and schady (2008) morris, flores, et al. (2004) Transfer (% of Evaluation PCE)c method Reference 7 17 rdd psm, dd galasso (2006) attanasio et al. (2005)

Country Chile

Program Chile solidario

Outcome regular checkups Child taken to growth and development monitoring

Colombia familias en acción

ecuador

Bono de desarrollo humano

Child had growth 3–7 control in last 6 months Child taken to health center at least once in past month number of visits to health center for preventive reasons in past 6 months number of visits to all health facilities in past month 0–3

honduras programa de asignación familiar Jamaica program of advancement through health and education Oportunidades

44.0

9

r

0–6

0.205

0.278*** 10 (0.085)

rdd

levy and Ohls (2007)

mexico

0–2 3–5

0.219 0.221

–0.032 (0.037) 0.027 (0.019)

20

r

gertler (2000)

continued

19

COnditiOnal Cash transfers: reduCing present and future pOVertY

Table 5

continued
Age range (years) 0–6 Baseline level (%)a Impactb 70.5 6.3*** (2.0) 8.4 (5.9) Transfer (% of Evaluation PCE)c method Reference 18 r macours, schady, and Vakis (2008) maluccio and flores (2005)

Country

Program

Outcome Child weighed in last 6 months Child taken to health center at least once in past 6 months Child taken to health center and weighed in past 6 months

nicaragua atención a Crisis nicaragua red de protección social

0–3

69.8

27

r

0–3

55.4

13.1* (7.5)

Source: authors’ calculations. Note: dd = difference-in-differences; n.a. = not available; pCe = per capita expenditure; psm = propensity score matching; r = randomized; rdd = regression discontinuity design. this table contains weighted means for the coefficients for Chile, combining rural and urban estimates. the standard error in this case is the square root of the averaged variances of these estimates. a. the unit for baseline level corresponds to the proportion of children who have been taken to the health center, with the exception of Jamaica and mexico, where the unit corresponds to the number of visits. b. the column for “impact” reports the coefficient and standard error (in parentheses); the units are percentage points, with the exception of Jamaica, where the unit is the number of visits to the health center in the past six months, and mexico, where the unit is the number of visits to the health center in the past month. c. the transfer amounts as a proportion of per capita expenditures (or consumption) are not the same across all tables in the report because of differences in the surveys used, including their coverage and year. * significant at the 10 percent level. *** significant at the 1 percent level.

in figure 4. as amartya sen (1985) and others have noted, poverty takes many forms—including an inability to develop basic “capabilities” in education and health. providing all citizens in a country with an equality of opportunities is an important policy goal, and CCts have helped level the playing field between rich and poor, more and less favored. although there is clear evidence that CCts have increased the use of education and health services, evidence on the impact of CCts on “final” outcomes in education and health is more mixed. some (but by no means all) evaluations have found that CCts contributed to improvements in child height among some population groups; there is also some evidence that program beneficiaries have better health status.6 20

OVerVieW

Figure 4 Heterogeneity of Impacts by Socioeconomic Status, Nicaragua, 2000
School enrollment (children ages 7–13)
Extreme poor Poor Nonpoor

Children weighed in past 6 months (ages 0–3)
Extreme poor Poor Nonpoor

0

5

10 15 20 Impact (percentage points)

25

0

10 20 Impact (percentage points)

30

Source: maluccio and flores 2005.

turning to education outcomes, adults with more exposure to the Oportunidades program in mexico have completed more years of schooling than have those with less exposure; however, the likely increase in wages that can be expected to occur because of this added schooling is small. also, a number of evaluations have concluded that the higher enrollment levels have not resulted in better performance on achievement tests, even after accounting for selection into school.7 this pattern of program effects—increases in enrollment, without more learning—is not particular to CCts. nevertheless, the results are sobering because they suggest that the potential for CCts to improve learning on their own may be limited. the evidence is somewhat more encouraging regarding the impact of CCt programs on cognitive development in early childhood (macours, schady, and Vakis 2008; paxson and schady 2008). this suggests that very early intervention might produce larger payoffs than one would expect, for example, by looking at the pattern of program effects on school enrollment by age or school grade. there are various reasons why CCts may have had only modest effects on “final” outcomes in education and health. One possibility is that some important constraints at the household level are not addressed by CCts as currently designed; these constraints could include poor parenting practices, inadequate information, or other inputs into the production of education and health. another possibility is that the quality of services is so low, perhaps especially for the poor, that increased use alone does not yield large benefits. 21

COnditiOnal Cash transfers: reduCing present and future pOVertY

Policy and Design Options earlier in this overview, we discussed the circumstances under which a CCt is desirable. given that a CCt is put in place, how should it be designed? We now turn to questions of CCt program design, including the selection of beneficiaries, the monitoring of conditions, the size of the transfer, and the complementary interventions that are needed. Defining the Target Population selecting eligible beneficiaries is the first question any policy maker considering a CCt must address. a CCt should be designed to target poor households (for whom there is a stronger rationale to redistribute) that underinvest in the human capital of their children. in practice, selecting the target population for a CCt first implies defining the criteria for eligibility based on poverty. the challenges of selecting the “right” targeting method and setting cut-off points for eligibility (that is, who qualifies as poor) are similar to those faced in the design of any social assistance program. defining the second criterion for targeting (that is, households that underinvest in the human capital of their children) is more complicated. in general, when households have qualified based on poverty criteria, CCt programs continue to make transfers as long as those households have children of the “right” ages and send them to school and/or take them to a health center. in some cases, it may be worthwhile to use a more narrow demographic target to direct transfers to population subgroups that appear to have the largest human capital gaps. this more narrow approach could imply targeting poor households with children transitioning from primary to secondary school in some countries, and poor households with young children in regions with high rates of malnutrition in others. there may be trade-offs between redistributive and human capital goals resulting from alternative targeting approaches. in a setting in which a large share of the poor population experiences significant and similar human capital gaps, trade-offs are likely to be small. On the other hand, when human capital gaps are highly concentrated on a relatively small proportion of the poor, designing a CCt to maximize impact on human capital accumulation may limit its ability to act as a redistributive mechanism. 22

OVerVieW

Selecting the Appropriate Conditions and the Size of the Transfer is the increase in the use of education and health services that results from CCts purely a result of the income effects inherent in the transfer? answering this question has important implications for the extent to which conditions are implemented and monitored, and the degree to which noncomplying households are penalized. as it turns out, evidence from a variety of sources (including comparisons across programs or countries, accidental glitches in program implementation, intentional features of program design, and structural models of household behavior) suggests that the impact of CCt programs on service use cannot be explained by the cash component of the program alone. 8 the conditions are thus important, at least in terms of increasing levels of school enrollment and the use of preventive health care. however, service use is generally a means to an end. thus the first step in selecting the “right” conditions is a review of the evidence on links between service use and the desired outcomes. is getting children into health facilities the most effective way to improve their nutrition and health more broadly? Or is giving mothers nutrition and parenting information and training more effective? Conditioning the cash transfer on the achievement of outcomes themselves is another possibility, particularly when links between such behaviors as service use and outcomes are unknown or complex, but outcomes are judged to be mostly within beneficiaries’ control. in the future, experimentation with alternative incentive schemes (through small-scale pilot programs, for example) should become increasingly important. this could be done by adding performance bonuses to the basic benefits households receive for satisfying attendance conditions. a second question is how to set the appropriate transfer amount. as discussed above, larger transfers generally have produced bigger improvements in consumption (or income) poverty—a result that seems reasonable. in terms of education and health outcomes, the critical questions are (1) how income-elastic are the outcomes? and (2) do larger transfers result in bigger behavioral changes by recipient households? in terms of enrollment in Cambodia, the marginal return to transfers appears to diminish very quickly—even though the “baseline” transfer is quite small (filmer and schady 2009a). more 23

COnditiOnal Cash transfers: reduCing present and future pOVertY

generally, however, the appropriate transfer amount for a CCt is likely to depend on the relative weight given to the program’s redistribution and human capital goals, and is likely to vary across outcomes and settings. structural modeling and small-scale experimentation can help policy makers identify and quantify the trade-offs (Bourguignon, ferreira, and leite 2003; attanasio, meghir, and santiago 2005; todd and Wolpin 2006a). Entry and Exit Rules the design of an effective program also requires careful consideration of rules for entry and exit. this is necessary to avoid confusion among prospective beneficiaries and to minimize the potential for manipulation and abuse. entry and exit rules also are important because they can have unintended incentive effects, particularly related to labor force participation. to date, CCts have used a proxy means rather than an income threshold to target benefits, and so the correspondence between program eligibility and labor supply is weaker than in many welfare programs in developed countries. however, the better a proxy means is at distinguishing “poor” from “nonpoor” households, the more highly it will be correlated with income and consumption—and the more likely it is to provide disincentives for adult labor market participation. potential solutions include the use of time limits on benefits (as in Chile or in the united states under the temporary assistance for needy families [tanf] program), and the adoption of graduated benefits (whereby there is only a partial reduction of benefits after recertification shows households have ceased to be eligible under original criteria) in order to avoid “cliffs” and the associated negative incentive effects on labor supply.

Complementary Interventions in many developing countries, the delivery of education and health services is dysfunctional. poor infrastructure, absenteeism, and lack of adequate supplies are not unusual problems in schools and health centers. achieving the human capital goals of CCt programs will require adaptation of the supply of services. in some countries, this adaptation may require governments or other actors to provide services where none existed before. improving quality is perhaps an even greater challenge, and 24

OVerVieW

some governments have attempted to address this by offering monetary incentives to providers of health and education services for good performance. reforms to increase access and the coverage of services frequently have been undertaken in parallel with or as an integral part of the CCt program. in addition to the poor quality of services, other constraints at the household level may make it difficult for CCts to improve final outcomes in health and education. figure 5 makes this point for ecuador. the figure shows the scores of children on a test of cognitive development in early childhood. at age 3, most children in the ecuador sample are only modestly behind the reference population. By age 6, the age when they enter first grade, children in the two poorest deciles of the national distribution of wealth are almost three standard deviations behind where they should be. the implication is clear: a CCt by itself or even in combination with high-quality schools is unlikely to remedy such disadvantages. this is particularly important because recent theoretical and empirical research suggests that the returns to investments
Figure 5
110 105 100 Standardized TVIP score 95 90 85 80 75 70 65 60 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 Child age (months) First (poorest) decile Second decile Third decile Fourth decile

Cognitive Development by Wealth Decile in Ecuador, 2003–04

Source: paxson and schady 2007. Note: tVip = test de Vocabulario en imágenes peabody. each line corresponds to one decile from the national distribution of wealth, from the first (poorest) decile, to the fourth. the test is coded so that a score of 100 corresponds to the average performance in a reference population, and the standard deviation is 15.

25

COnditiOnal Cash transfers: reduCing present and future pOVertY

later in the life cycle will be limited if children do not have adequate levels of cognitive, social, and emotional development in early childhood (Cunha et al. 2006; Knudsen et al. 2006). under these circumstances, interventions that seek to improve parenting practices and the quality of the home environment are likely to be particularly important. Oportunidades and some other CCts attempt to expose parents to new information and practices by conditioning transfers on participation in talks (known as pláticas). the conditioned cash helps ensure that parents attend and participate in the pláticas. however, the cash-condition package offered by CCt programs may not be enough, and a comprehensive program that relies on more active participation by social workers and others may be needed.

CCTs in the Context of Social Protection Policies
CCt programs are just one option within the arsenal of social protection programs that can be used to redistribute income to poor households. they cannot be the right instrument for all poor households—for example, they cannot serve the elderly poor, childless households, or households whose children are outside the age range covered by the CCt. redistribution to those groups is better handled through other means. in the case of the elderly poor, the potential labor supply disincentives from cash transfers are likely to be low, and the justification for further investments in human capital is questionable. as a result, social (or noncontributive) pensions often are the preferred instrument used by both developed and developing countries to provide assistance to elderly poor people. also, a CCt is unlikely to be the best instrument for social risk management. CCts have been used to help cushion the negative impact of various types of crises on the poor. But their focus on long-term investments in human capital and their reliance on administrative targeting mean that CCt programs generally are not the best instrument to deal with transient poverty. transfer programs that do not involve longterm commitments (such as those implicit in CCt conditions), that are self-targeted (and thus do not involve complex administrative decisions for entry or exit into the program), and that involve beneficiaries in activities that can help address the source of the shock (for example, job-related activities) appear to be better suited than are CCts as instruments for managing risk. 26

OVerVieW

thus in most country contexts, CCt and other cash transfer programs are likely to coexist and should be seen as complements, rather than substitutes, addressing different household characteristics and the nature of the poverty those households experience. it is not surprising that policy makers and program managers for CCts in latin america, the region where such programs have the longest tradition and the most established status, increasingly are casting CCts as part of a broader system of social protection. doing so requires making the basic design features of programs compatible—for example, the transfer size in a CCt must be set in relation to that of other cash transfers to limit distortions, ensure horizontal equity, and make programs politically acceptable. finally, the potential administrative synergies across cash transfer programs are large. perhaps the most obvious examples are common systems for administrative targeting and common systems to make payments to beneficiaries (such as with electronic cards). numerous countries also are considering or experimenting with a common outreach and service platform—one-stop shops that beneficiaries of all social protection programs can use to access benefits and interact with program administrators.

Conclusion
CCt programs often are described in both extremely positive and negative terms. Our review of the CCt experience so far confirms that the programs have been effective in the sense that there is solid evidence of their positive impacts in reducing short-term poverty and increasing the use of education and health services. those achievements should not be minimized because they are powerful proof that well-designed public programs can have significant effects on critical social indicators. CCts also have had positive institutional externalities—most notably, through their emphasis on monitoring and evaluation, whereby they have helped strengthen a results culture within the public sector, at least within social policies. that strengthening is clearly a legacy worth sustaining. at the same time, our review provides ample reasons to be cautious and avoid transforming the obvious virtues of CCts into a blind advocacy campaign in support of them. fifty years ago, albert hirschman (1958) argued that development is a “chain of disequilibria” whereby the expansion of one sector creates 27

COnditiOnal Cash transfers: reduCing present and future pOVertY

backward or forward pressures that can provide the necessary stimulus for the expansion of another sector, which is still underdeveloped. those links operate not only through the standard motivation for profit, but also by building political pressure for government action. CCt programs have increased poor people’s demand for services and have the potential to unleash a broader process to transform health, education, and social protection services. it is still too early to tell whether the current wave of CCt programs will produce those results, but the experience so far provides room for hope.

28

Chapter one

Introduction

CountrIes have been adoptIng or ConsIderIng adoptIng

conditional cash transfer (CCt) programs at a prodigious rate. In some countries, including brazil, ecuador, and Mexico, CCts have become the largest social assistance program, covering millions of households. they have been hailed as a way of reducing inequality, especially in the very unequal countries in Latin america; of helping households break out of a vicious cycle whereby poverty is transmitted from one generation to another; of promoting child health, nutrition, and schooling; and of helping countries meet the Millennium development goals. nancy birdsall, of the Center for global development, calls CCts “as close as you can come to a magic bullet in development” (dugger 2004). Conversely, an article in the Institute of Development Studies Bulletin refers to CCts as “superfluous, pernicious, atrocious and abominable” (Freeland 2007, p. 75), arguing that they represent an impractical way to improve the use of social services (particularly in low-income countries) and are immoral because they may deprive the neediest people of the assistance they deserve. do these and other claims make sense? are they supported by the available empirical evidence? What does all of this imply for the way in which countries that have CCts should structure or reform those programs? What about countries that do not have CCts but are considering implementing them, often in circumstances very different from those in which they were first introduced? this report seeks to answer these and other related questions. specifically, it lays out a conceptual framework for thinking about the economic rationale for CCts; it reviews the very rich evidence that has

29

CondItIonaL Cash transFers: reduCIng present and Future poverty

accumulated on CCts, especially that arising from impact evaluations; it discusses how the conceptual framework and the evidence on impacts should inform the design of CCt programs in practice; and it discusses how CCts fit in the context of broader social policies. the report shows there is considerable evidence that CCts have improved the lives of poor people. transfers generally have been well targeted to poor households, have raised consumption levels, and have reduced poverty—in some countries by a substantial amount. offsetting adjustments that could have blunted the impact of transfers, such as reductions in beneficiaries’ participation in the labor market, appear to have been relatively modest. Moreover, CCt programs often have provided an entry point to reforming badly targeted subsidies and upgrading the quality of safety nets. the report thus argues that CCts have been an effective way of redistributing income to the poor, although it recognizes that even the best-designed and best-managed CCts cannot fulfill all of the needs of a comprehensive social protection system. they need to be complemented with other interventions, such as social pensions and workfare or employment programs. the report also considers the rationale for conditioning transfers on the use of specific health and education services by program beneficiaries. using such conditions (as opposed to making unconditional cash transfers) can be justified as a means to reinforce incentives for households to invest more in the human capital of their children—for example, when there is inadequate information about the returns to these investments, myopia, “incomplete altruism” between parents and their children, or externalities that are not taken into account by households. More generally, political economy considerations sometimes may favor conditional over unconditional transfers. For example, both taxpayers and beneficiaries may be more likely to support transfers to the poor if those transfers are linked to efforts to overcome poverty in the long term, particularly actions to improve the welfare of their children. CCts also have led poor households to make more use of health and education services, a key objective that CCts were intended to accomplish. nevertheless, the evidence on improvements in final outcomes in health and education is more mixed. thus, CCts have increased the likelihood that households take their children for preventive health checkups, but doing so has not always led to better child nutritional status; and school enrollment rates have increased substantially among program beneficiaries, but there is little evidence of improvements in learning outcomes. these findings suggest that to maximize their potential effects 30

IntroduCtIon

on poor households’ accumulation of human capital, CCts should be combined with programs to improve the quality of the supply of health and education services and provide other supporting services. the evidence also suggests the need to experiment with conditions that focus on outcomes rather than on use of services alone.

The CCT Wave the common definition of a conditional cash transfer program is one that transfers cash to poor households if they make prespecified investments in the human capital of their children.1 In general, this has involved attaching “conditions” to transfers. health and nutrition conditions often require periodic checkups or growth monitoring and vaccinations for children less than 5 years of age, perinatal care for mothers, and attendance by mothers at periodic health information talks. education conditions usually include school enrollment and attendance at 80 or 85 percent of school days, and occasionally some measure of performance. Most CCt programs transfer the money to the mother of the household, or occasionally to the student. CCt programs have two clear objectives. First, they seek to provide poor households with a minimum consumption floor. second, in making transfers conditional, they seek to encourage the accumulation of human capital and to break a vicious cycle whereby poverty is transmitted across generations. Interest in and the scope of CCt programs has grown enormously in the last 10 years. the maps shown in figure 1.1 give a feel for this expansion—although they understate the expansion because the whole of brazil and Mexico are shown as active in 1997, when the brazilian bolsa escola programs were run only by a handful of municipalities and Mexico’s oportunidades program was confined to very poor rural areas. those programs did not become national for several years. ten years later, 29 developing countries had some type of CCt program in place (in some cases, more than one) and many other countries were planning one. the range of polities interested covers all continents, although the longest established and most evaluated programs are found predominantly in middle-income countries in Latin america. paralleling the rise in the number of countries with programs has been an increase in the size of some programs. Mexico’s program started with about 300,000 beneficiary households in 1997, but now 31

CondItIonaL Cash transFers: reduCIng present and Future poverty

Figure 1.1

CCTs in the World, 1997 and 2008

1997

MEXICO

BANGLADESH BRAZIL

2008

TURKEY PAKISTAN DOMINICAN REP. MEXICO JAMAICA GUATEMALA HONDURAS EL SALVADOR NICARAGUA COSTA RICA COLOMBIA PANAMA ECUADOR BRAZIL PERU BOLIVIA PARAGUAY CHILE ARGENTINA BURKINA FASO NIGERIA INDIA YEMEN KENYA BANGLADESH INDONESIA CAMBODIA PHILIPPINES

Source: World bank.

covers 5 million households. brazil started with municipal bolsa escola programs in brasilia and the municipality of Campinas. these led to replication by local governments, followed by the formulation of sector-specific federal programs, and then their unification and reform. today, the federal bolsa Família program serves 11 million families or 46 million people. In other countries, the increase in size has been less explosive, but still notable. In Colombia, for instance, the program’s 32

IntroduCtIon

initial goal was 400,000 households, but it had expanded to cover 1.5 million households by 2007. Many social policy analysts also see a parallel in the move to CCts in developing countries and the welfare-to-work agenda in the united states and europe, as embodied by reforms that led to the revenu Minimum d’Insertion in France, the temporary assistance for needy Families (tanF) reforms in the united states, and the new deal in the united Kingdom. Like CCts in the developing world, all of these programs require “desired behavior” in exchange for income support. In settings with highly informal labor markets, the conditions on children’s health and schooling are easier to monitor than job search and work requirements and thus are sensible adaptations of the basic notion of linking social assistance to positive behavioral change. table 1.1 presents a partial list of CCt programs that are considered in this report. the list is not exhaustive: there are additional CCt

Table 1.1

Matrix of Program Size and Extent of Conditions
Conditions

Program size/Target nationwide

Education and health bolsa Família (brazil) oportunidades (Mexico) bono de desarrollo humano (ecuador) Familias en acción (Colombia) program of advancement through health and education (Jamaica)

Education only bolsa escola (brazil) Jaring pengamanan sosial (Indonesia)

niche (regional or narrow target population)

Chile solidario social risk Mitigation project (turkey)

Female secondary school assistance program (bangladesh) Japan Fund for poverty reduction (Cambodia) education sector support project (Cambodia) basic education development project (yemen)

small scale/pilot

programa de asignación Familiar (honduras) Cash transfer for orphans and vulnerable Children (Kenya) atención a Crisis (nicaragua) red de protección social (nicaragua)

subsidio Condicionado a la asistencia escolar– bogotá (Colombia) punjab education sector reform program (pakistan)

Source: authors’ compilation.

33

CondItIonaL Cash transFers: reduCIng present and Future poverty

programs in operation for which little information was available and there are some programs that fit the CCt label less well than others— for example, the bono de desarrollo humano (bdh) program in ecuador never actually has monitored compliance with the education and health conditions, even though a social marketing campaign stressed that beneficiaries were responsible for ensuring that their children were enrolled in school and were taken to health centers for preventive checkups. More information on the programs in table 1.1, as well as on other CCt programs in the developing world, is provided in program-by-program “at-a-glance” tables in appendix a. even without including every program, table 1.1 shows that CCts vary a great deal in scope. some programs are nationwide, others are niche programs, and yet others are small-scale pilot programs. table 1.1 also shows that some programs have required that households receiving transfers comply only with schooling conditions; others, especially programs in Latin america and the Caribbean, have required that households comply both with schooling and with health conditions.

Theme and Variations the role of CCt programs in social policy is different from place to place, as a consequence of differences in both their design and the context in which they operate. Most obviously, CCt programs vary with respect to the pertinent measures of size. In terms of absolute coverage, they range from 11 million families (brazil), to 215,000 (Chile), to pilot programs with a few thousand families (Kenya, nicaragua). In terms of relative coverage, programs cover a range from about 40 percent of the population (ecuador), to approximately 20 percent (brazil, Mexico), to 1 percent (Cambodia). In terms of budget, programs cost from about 0.50 percent of gross domestic product (gdp) in countries such as brazil, ecuador, and Mexico to 0.08 percent of gdp in Chile. the generosity of benefits ranges from 20 percent of mean household consumption in Mexico, to 4 percent of mean household consumption in honduras, and to even less for the programs in bangladesh, Cambodia, and pakistan. CCt programs are expected to fill different niches in social policy. In some countries (brazil, Jamaica, Mexico), the CCt program offers large-scale social assistance and grew out of a reform and expansion of other social assistance programs. 34

IntroduCtIon

In Mexico, oportunidades was established to replace consumption subsidies that were badly targeted and had limited impact on poverty. by the mid-1990s, the Mexican government had 15 food subsidy programs. of those programs, 11 were targeted at urban and rural populations and 4 had no explicit targeting mechanism (see Levy and rodríguez [2004]; Levy [2006]). More than half of social funding was allocated to bread and tortilla subsidies in urban areas—a great deal of which was absorbed by nonpoor urban households. Food subsidies were an inefficient way to redistribute welfare to the rural poor, who often lived in small, hard-to-access communities. approximately 60 percent of poor rural families received no support from the federal government (rodríguez 2003). there was little coordination across programs, administrative tasks were duplicated, there was a noticeable imbalance in spending that favored urban areas, and there was no systematic evaluation to analyze the effectiveness of such programs. oportunidades was an innovation in Mexican social policy. In place of inefficient subsidies and poorly targeted cash transfers, the program made explicit a commitment to give beneficiaries the freedom to choose how they used the transfers as long as they committed to certain behaviors, namely education, health, and nutrition behaviors that were viewed as investments in human capital. In brazil, several states began to experiment with new forms of social assistance in the mid-1990s. In 1995, two programs (bolsa escola and the guaranteed Minimum Family Income program) were initiated in the distrito Federal (brasilia) and Campinas, respectively. the federal government started the programa de erradicação do trabalho Infantil (petI) in 1996. two years later, the government began to provide transfers to municipalities that were running CCts. by 2001, CCts with education conditions expanded to more than 100 municipalities and provided support to approximately 200,000 families (Lindert et al. 2007). In that same year, the federal government decided to create a national version of the bolsa escola program. It also initiated the bolsa alimentação (2001), a conditional cash transfer for pregnant women and lactating women with children; the auxílio gás (2002), an unconditional cash transfer intended to dampen the effects on poor families as cooking gas subsidies were phased out; and the Cartão alimentação (2003), a general cash transfer to the extremely poor population to promote food consumption and prevent hunger. bolsa Família was created in 2003 by merging bolsa escola, bolsa 35

CondItIonaL Cash transFers: reduCIng present and Future poverty

alimentação, Cartão alimentação, and auxílio gás. that consolidation of programs signaled an effort to improve the efficiency of the social safety net and to broaden federal support for poverty-targeted programs. In other countries, such as Chile, programs are smaller and are meant to fill the cracks between and tie together a large number of existing social services. In some countries, CCt programs stand independently (honduras, Jamaica), in others they provide links to a large and sometimes increasing array of other services (Chile, Colombia, Mexico). In several countries, CCt programs are still small pilot efforts (Kenya, nicaragua). In still other countries, the programs’ roots are in the education sector (Cambodia) or are a hybrid of social assistance and education (bangladesh, Kenya). some of the nascent programs will focus more on the nutrition of young children. Many of the programs in middle-income countries have pursued an integrated approach to poverty reduction, balancing goals of social assistance and human capital formation. they cover children from birth (or prenatally) through their mid-teens, with conditions on health care use for children from birth to ages 5 or 6 and with conditions on school enrollment thereafter. targeting usually is done with a proxy means test, sometimes combined with geographic targeting. In most cases, programs are administered by ministries of social welfare or freestanding agencies under the presidency. examples of that type of CCt include argentina, brazil, Colombia, el salvador, Jamaica, Mexico, panama, and turkey. Mexico has one of the iconic programs in this class. the program started early, its evolution has been carried out thoughtfully, and it has been successful. What really makes Mexico’s program iconic is the successive waves of data collected to evaluate its impact, the placement of these data in the public domain, and the hundreds of papers and thousands of references to them that this easy access has generated. brazil’s efforts also have been exemplary. the program started early, has evolved enormously, and is equally large in coverage and importance. brazil’s CCt provides something of an interesting contrast to the Mexican case in various respects—the issue of federalism is more to the fore in the program; it takes a softer, more gradual tack on conditions; and it puts a shade more emphasis on redistribution than on human capital formation. also, unlike the Mexican program, the brazilian programs did not explicitly incorporate impact evaluations in their design; as a result, much less is known about the impact they have had 36

IntroduCtIon

on consumption, poverty, health, nutrition, and education than is true of Mexico’s program. table 1.2 shows some of the more salient similarities and differences between the oportunidades program in Mexico and the bolsa Família program in brazil. there is another branch of the CCt program family that focuses on education in low-income countries. the programs usually cover a
Table 1.2 Implementation of “Similar” Programs: Contrast between Mexico and Brazil
Mexico Brazil

Program feature

More-similar features program size definition of conditions 5 million families 25% of the population education: • chool enrollment and minimum S attendance rate of 85%, both monthly and annually • ompletion of high school (for savings C account) health: • ompliance by all household members C with the required number of health center visits and mother’s attendance at health and nutrition lectures Less-similar features targeting system geographic targeting used to determine which rural areas participated initially proxy means test used for household targeting within localities and in urban areas program itself does targeting and program registration evaluation benefit structure payment mechanism enforcement of conditions explicitly taken into account in program design differentiated by age, grade, gender In cash at program-specific payment points rigorous, reduction in benefits at first round of noncompliance geographic targeting used to assign ration of slots in registry of poor households Means test used as household targeting system Municipalities do program targeting and program registration no systematic attempt to integrate evaluation of program impact into design differentiated by poverty level via debit card usable at banks, atM machines, and lottery points Warning system, noncomplying households seen as in need of additional “care” and problem solving 11 million families 25% of the population education: A • t least 85% school attendance in a 3-month period for children aged 6–15

health: • hildren 0–7: vaccination and follow-up C of nutritional development • regnant women: pre- and postnatal P visits, health and nutrition seminars

Source: authors’ compilation.

37

CondItIonaL Cash transFers: reduCIng present and Future poverty

narrower segment of education—some only secondary (bangladesh’s Female secondary school assistance program [Fssap]; Cambodia’s Japan Fund for poverty reduction [JFpr] and education sector support project [Cessp]), some only primary (bolivia, Kenya, and proposals in nigeria and tanzania), and occasionally both (Indonesia’s Jaring pengamanan sosial [Jps] program). the genesis of these programs is rather varied. In bangladesh, the Fssap was part of a strategy to close the then significant gender gap in education. doing so was seen as an important policy objective: in 1981, the female literacy rate (approximately 13 percent) was about half the literacy rate among men (26 percent). as a result, a series of stipend and tuition waiver programs was made available to girls as long as they attended school regularly, made passing grades, and remained unmarried. In addition, the Food for education (FFe) program was initiated in 1995 to provide in-kind food transfers to poor households as long as they sent their children to primary school (ravallion and Wodon 2000). the FFe in-kind transfer was converted to a cash transfer in 2002, and was renamed as the primary education stipend program. In Indonesia, the Jps program was instituted following the east asian financial crisis in order to prevent children from dropping out. In Kenya and tanzania, the programs are geared especially to coping with the crisis of orphans and vulnerable children that has burgeoned in the wake of hIv/aIds. In many cases, the administrative structure behind these programs is less sophisticated than it is for the big Latin american programs. this is a result of several things. First, because these are primarily education programs and often are run through the education ministry, there are fewer actors to coordinate. second, daily attendance is not always a condition of receipt of the transfer, and that simplifies administration. third, the programs are newer and situated in lower-capacity countries so simpler systems are to be expected. to compensate for the lack of a complex administrative structure, the role of the community in implementing the programs is often greater than it is in the Latin american programs. Chile solidario works in a very different way to fill a different niche. the program is targeted to only the extremely poor, about 5 percent of Chile’s population. It differs notably from classic CCt programs by customizing its conditions. Families initially work intensely with social workers to understand actions that could help them get out of extreme poverty, and then they commit to action plans that become 38

IntroduCtIon

the household-specific conditions of the benefit. the diagnosis covers a total of 53 different so-called minimum conditions grouped along seven dimensions (identification and legal documentation, family dynamics, education, health, housing, employment, and income). households receive a comparatively small cash transfer, with the amount declining periodically during the two years of active participation, and then a still lower amount for an additional three years following the program. however, they receive preferential access to the full range of Chilean social assistance programs from the time they join Chile solidario through the end of the three-year follow-up period. additional, though small, transfers come from those other social programs; the transfer from Chile solidario itself is really intended only to motivate clients to avail themselves of the services of the social worker. Chile solidario is thus far a model unto itself, although other programs are moving to emulate it to a degree.2 the different goals and contexts of the programs suggest that somewhat different benchmarks may be pertinent to judge them, and different weights should be given to results in different dimensions—for example, reduction in consumption poverty versus improvements in human development outcomes. primary school enrollment in Colombia and Mexico already exceeded 90 percent before the CCt programs. enrollment increased slightly due to the CCts, but dramatic gains were not possible because the base was already so high. those programs, however, emphasized their role in social assistance and, with large transfers and well-targeted and extensive coverage, they are successful at it. the bangladesh stipend for girls in secondary school was designed as a gender-targeted education program. It makes small cash payments, however; and given its focus on increasing girls’ enrollments, it did not contemplate poverty targeting. thus, without an understanding of the program and its original goals, one could consider it a failure by the standards of social assistance. variation can occur not only among countries, but even within a single program over time. For example, in the 10 years of its existence, Mexico’s oportunidades has undergone continuous evolution of implementation systems in order to respond to changing needs as the program expanded and as administrative systems were built. the role of geographic targeting was reduced as the program achieved national coverage, the role of community targeting was eliminated, and the role of the proxy means test increased accordingly. Carrying out the proxy 39

CondItIonaL Cash transFers: reduCIng present and Future poverty

means test and eligibility procedures moved from contracted agencies to in-house staff. both eligibility and compliance monitoring were moved from a paper-based system to computerized systems, and most of the data now flows via the Internet. the timeline for bringing people into the program has been reduced from 6–8 months to 4–8 days, and a set of benefits held in a savings account was added to the cash payment. the role and design of CCt programs is evolving in many countries. early successes with the basic model are bringing countries to address a second and third round of challenges, including these:

• • • •

should the emphasis on expanding the supply of services be complemented with efforts to improve the quality of those services, in tandem with the demand-increasing action of the CCt? should the range or definition of conditions be changed, for example, to reward performance instead of, or in addition to, service use? What can be done to ensure that youth who are aging out of the school support provided by the program can attain jobs or receive further training? What should be the balance between targeting younger and older children?

In some countries, CCt programs themselves are addressing those challenges through adjustment to their basic design; in other cases, they are catalyzing changes in other programs.

Outline of the Report and Issues Covered this report seeks to bring together existing knowledge about CCts. starting with the Mexican program oportunidades, an important feature of CCt programs has been the strong emphasis they have placed on credible evaluations of their impact on various outcomes. the report draws heavily on those evaluations, and expands on earlier efforts to assess the performance of CCt programs using evidence from impact evaluations (see das, do, and Özler 2005; rawlings and rubio 2004). Indeed, it would not have been possible to write this report without the efforts of the programs themselves, international donors, and academics around the world to ensure the high quality of many of the evaluations. (see appendix b for further discussion of CCt impact evaluations.) 40

IntroduCtIon

the accumulating evidence of positive impacts has been instrumental both in sustaining existing programs and in encouraging the establishment of similar programs in other developing countries.3 nevertheless, although the initial group of evaluations provided solid evidence of impact along several key dimensions, important policy and operational questions remain. First, much of what is known about CCts is based on evaluations of programs in Latin america, especially Mexico. It is not clear, therefore, whether CCts could be expected to have similar impacts in other settings, especially in countries that are significantly poorer and that tend to have weaker institutions. although much of the evidence discussed in this report is based on studies of Latin american programs, we have made a special effort to discuss the evidence from countries in other regions, especially programs in bangladesh and Cambodia. second, as CCt programs have become larger—in several countries they represent a substantial share of public budgets dedicated to poverty reduction—demand for evidence on their results has grown beyond the initial emphasis on a small number of outcomes. For example, policy makers and academics increasingly are focusing on possible long-term effects of the transfers, as well as on changes in “final” outcomes (say, learning rather than school enrollment, or nutritional status rather than frequency of growth monitoring checkups). this report pays particular attention to these outcomes, which have been discussed less extensively in the literature on CCts. third, the fact that CCt programs are being implemented in very diverse country settings raises many questions regarding their design: the role of conditions, the appropriate means of targeting, the right size of the transfer, and the best way to coordinate CCt programs with the supply of services are just a few of the important questions being asked. again, we make special efforts to analyze the possible importance of these features of program design in explaining changes in outcomes, and to consider carefully the appropriate role of a CCt within a country’s social assistance system (although that is a complex agenda that goes beyond the goals of this report). the rest of the report proceeds as follows. Chapter 2 provides a conceptual framework in which to think about CCts. the chapter focuses particularly on when it makes sense to condition transfers on household investments in child human capital. It discusses three broad sets of circumstances under which CCts are likely to be particularly 41

CondItIonaL Cash transFers: reduCIng present and Future poverty

attractive. the first set of circumstances is a case in which parents invest less in the human capital of their children than is warranted by the private returns to those investments. that situation might happen because parents value their own welfare more than that of their children; are poorly informed about the returns to investments in education, health, and nutrition; or are myopic and discount the future very heavily. In the second set of circumstances, there are externalities to human capital investments, as might be true if there are spillovers from having a bettereducated or more healthy population that are not taken into account by rational individuals when they make decisions about investments. the third set of circumstances is one in which there are political economy considerations that justify imposing conditions on transfers, as might be the case, for example, when it is easier to sustain a budget for a program if transfers are perceived not as a handout but as a quid pro quo whereby a government gives households cash if—and only if—these households act “responsibly” and invest in their children. Following that conceptual discussion, chapter 3 describes in detail how CCt programs work. virtually all CCt programs have attempted to direct their benefits to the poor so the chapter begins with a discussion of the targeting instruments used in different programs. It then describes the benefit systems, including who receives the payment, how payment takes place, and what payment levels are in practice. the chapter continues by describing how programs have monitored conditions and the extent to which households are penalized for noncompliance. the final two sections of the chapter discuss the importance of monitoring and evaluation and how CCts have coordinated with other actors in the social sectors. redistribution of resources to the poor is one of the two fundamental goals of most CCt programs, and chapter 4 presents the evidence of CCt impact on consumption poverty. the chapter discusses the impacts in the short term and, for two countries (Mexico and nicaragua), in the medium term. Many policy makers originally had concerns that the effects of CCts on household consumption would be relatively small as households made offsetting adjustments. as the chapter discusses, however, those adjustments—in terms of reductions in adult labor supply, in remittances, or in household access to other social programs—have been modest. as a result, the impact of CCts on consumption poverty is largely determined (at least in the short run) by the size of the transfer and the extent to which programs effectively can ensure that the cash reaches poor households. 42

IntroduCtIon

the second fundamental goal of CCts is to encourage households to invest in the human capital of their children. Chapter 5 turns to the evidence on the impact that CCts have had on outcomes in education, health, and nutrition. the chapter begins by showing that CCts have had significant effects on the use of education and health services, and that those effects often have been substantial in magnitude. It then discusses the evidence of CCt effects on “final” outcomes in education and health. the chapter shows that the evidence on the impact of CCts on these outcomes is somewhat mixed. thus, CCts appear to have had a modest impact on years of schooling completed by adults; they reduced the incidence of low child height for age in some countries and among some populations but not others; and they had little effect on learning outcomes among either school-age children or adults. addressing those shortcomings is likely to require a combination of efforts: redefining conditions, perhaps including incentives for performance, not only service use; improving the quality of the supply of services; and complementing CCts with interventions that help households overcome other barriers to adequate child nutrition, development, and learning. the chapter closes by discussing whether CCt program impacts that are observed are the result of the “income” effect associated with the transfer or the “price” effect that results from the conditions. Chapter 6, the final chapter of the report, returns to the conceptual framework presented in chapter 2. In particular, with the evidence from chapters 4 and 5 in hand, it discusses when CCt programs are likely to be the right policy instrument. It then turns to a discussion of how CCt programs should be designed—for example, in terms of the population covered, the conditions that are monitored, and the magnitude of the transfer. the chapter closes by considering where CCts fit in the context of social policies. an important message of the chapter is that CCts have shown themselves to be effective and versatile programs. however, they are most likely to be effective in stimulating investments in child human capital and in providing a social safety net when they work closely with other programs. the chapter reviews some of the ongoing efforts by developing countries in this area. It also argues that there are other interventions—workfare or employment programs, pensions—that need to complement even the best-designed and bestmanaged CCt.

43

Chapter two

the economic rationale for Conditional Cash transfers
CCts are Cash transfers that are (1) targeted to the poor

and (2) made conditional on certain behaviors of recipient households. More specifically, these cash transfers are conditioned on minimum levels of use of health and education services, generally by (or for the benefit of) the children in the household. typical requirements include enrollment and actual attendance at schools (for example, minimum attendance rates of 85 percent are required in Brazil’s Bolsa família and a similar restriction applies in Mexico’s oportunidades). programs that have a health component also may require that children make regular visits to a health center and receive immunizations, and that pregnant women and lactating mothers keep a predetermined number of appointments at local clinics or attend informational sessions (pláticas) on hygiene and nutrition. this chapter provides a conceptual understanding of how CCts work. governments have scarce resources and CCts—whether they increase service use or not—compete for funds with other worthwhile projects, such as buying school equipment or upgrading rural roads. why is giving people money to keep their children in (possibly very bad) schools a good use of public funds? wouldn’t it be better to buy more books and supplies or to improve teacher training so as to raise the quality of the education provided in those schools? even if there are good arguments for spending part of the government’s budget on direct cash transfers to households, does it really make sense to attach conditions to the cash? after all, if attending those schools (or walking 5 miles to the local clinic to have the baby weighed) contributed more to expected future well-being than an alternative

45

Conditional Cash transfers: reduCing present and future poverty

use of the child’s time, wouldn’t households already be sending their children there? a condition is a constraint on behavior. what good can come of adding a constraint to the household’s optimization problem? why not just make the transfer unconditionally? economists might think of at least two kinds of disadvantage associated with attaching conditions to cash transfers. first, some of the neediest households might find the conditions too costly to comply with (because the clinics are too far away or because their need for child help in harvesting a living from the land is too pressing), and may thus be deterred from taking up the benefit. Conditions thereby might exclude some of the people the program aims to reach. second, those households that do opt for the benefit may incur a costly distortion to their own behavior for the sake of a little extra cash in the short run. perhaps they know how bad the local school (or clinic) is. perhaps it is wasteful for the children to spend time there, rather than learning how to tend the fields or how to weave a basket with their parents. By pushing poor households to do something that they would otherwise not be doing, CCts might be imposing costly distractions on people who are trying to do the best thing for their families under conditions of severe scarcity. proponents of CCt programs should have good answers to questions and arguments such as those posed above. there are, in fact, a number of good reasons for attaching conditions to targeted cash transfers. this chapter reviews the case for conditioning, and briefly discusses some of the empirical evidence on how likely it is that circumstances that justify conditions are found in practice.

Cash Transfers: Arguments in Support and Against the first question one might ask when considering whether a CCt makes sense is this: are cash transfers in general a good instrument in a particular country? that question is by no means rhetorical. even if everyone accepts poverty reduction as a central policy objective, it does not follow immediately that the government should spend its scarce resources by transferring cash directly to poor people. Broadly, one hears two arguments against such transfers: first, poverty is best reduced by economic growth, particularly in the poorest countries. in those same countries, fiscal efforts and administrative 46

the eConoMiC rationale for Conditional Cash transfers

capacity both tend to be low, and governments should focus on providing basic infrastructure (which could include roads and ports as well as schools and clinics). in this view, cash transfers to a vast poor majority are seen as having a lower future payoff than investment in public capital, and as being harder to target and deliver. the second argument against cash transfers is that they provide the wrong incentives to recipients. for example, they may discourage labor supply or investment in a person’s own human capital for future gainful employment. if the government provides the basic necessities of life, the thinking goes, why would people in low-productivity settings bother with very hard work that pays so little? when combined, those arguments are not to be dismissed immediately. direct investment in public infrastructure is likely to be a serious alternative use of public funds in very poor countries, and handouts of public cash can discourage self-reliance. But there also are many arguments for direct redistribution. first, in most developing countries, public expenditure on infrastructure and public services—of the kind just advocated—often fails to reach the very poor. in nicaragua, only 10 percent of households in the bottom quintile of the expenditure distribution had access to electricity in 1998, compared with more than 90 percent of households in the top quintile (de ferranti et al. 2004, p. 209). the sizable electricity subsidies that were in place in Mexico in 2000 also had a regressive incidence, as documented by scott (2002). in fact, the proponents of the pioneering oportunidades CCt program explicitly couched the initiative as an alternative to the electricity and tortilla subsidies, in a way that would be both more equitable (by reaching the poor) and more efficient (by eliminating the price distortions generated by the subsidies). in this context, if cash transfers can be shown to be targeted to the poor more effectively than other forms of public expenditure, they may contribute to poverty reduction in ways that direct public investment does not. second, markets seldom work perfectly in practice, and sometimes they fail in ways that prevent poor people from being as productive as they might otherwise be. if the root causes of some of these failures are too costly to correct, simple redistribution of current resources may be able to reduce the efficiency costs.1 the classic example is that of creditconstrained families (in an economy with imperfect capital markets) that cannot make profitable investments in their children’s education or in some other business project. a direct transfer of cash to these 47

Conditional Cash transfers: reduCing present and future poverty

families might enable them to undertake an efficient project that would otherwise not have taken place. once again, the transfer would be both equitable (by making a poor person better off) and efficient (by better allocating capital within the economy). similarly, insurance markets often are beyond the reach of many poor families. when incomes are volatile, reflecting a risky economic environment, cash transfers can smooth (some of ) the fluctuations, raising household welfare. fields et al. (2007) review evidence of substantial short-term income volatility in a number of countries in latin america. if these fluctuations are sufficiently severe, they may affect demand for schooling or health investments, potentially with long-term consequences.2 once again, if fixing the insurance markets themselves is too costly or complicated, volatility can provide an additional argument for targeted cash transfer programs. finally, the fact that many of the inequalities one observes in the developing world are inherited from one’s parents may make them ethically objectionable. differences associated with circumstances over which individuals have no control (such as race, gender, or family background) often are regarded as “inequality of opportunity,” which the state has a moral obligation to redress (see roemer [1998] and Bourguignon, ferreira, and walton [2007]). Cash transfers might be suitable instruments for compensating families who suffer from inherited disadvantage. in the face of arguments both for and against, it cannot be stated categorically that every country in the world should have a cash transfer program in place to help reduce poverty. But the case often can be made that some efficient redistribution of the kind just described can be achieved, as discussed in box 2.1. Chapter 5 of the World Development Report 2006 (world Bank 2005) discusses other examples from across the developing world, both of market failures and of the resulting underinvestment. CCt programs still may be justified in the absence of redistributive goals, as a means to improve incentives for households to invest in human capital. in the remainder of this report, however, it is assumed that policy makers have weighed the arguments for and against cashbased redistribution, and have reached a considered decision that there is some room for cash transfers in their policy arsenal against poverty. this chapter next turns to the question of whether it makes sense to attach conditions to these cash transfers. the “theoretical default” 48

the eConoMiC rationale for Conditional Cash transfers

Box 2.1 Efficient Redistribution in the Presence of Market Failure ther e a r e M a n y situations in w hiCh a

market failure opens up the possibility of efficient redistribution. under certain types of credit market imperfection, and if there are economies of scale, the poor may be unable to take advantage of profitable opportunities because they do not have access to the required scale. they thus may be trapped in a low-productivity sector of the economy, even as more productive opportunities go unexploited, because of an inability to commit to repayment in credit markets. some amount of redistribution in cash from the rich to the poor may raise the ability of the latter to take advantage of these more profitable investments, thereby reducing both inequality and inefficiency. this possibility was first modeled by loury (1981), who introduced credit constraints into a model of intergenerational mobility. galor and Zeira (1993) further noted the link between aggregate efficiency and reduced inequality under nonconvex production sets. Banerjee and newman (1993) exploited long-term implications of the same basic type of mechanism by noting the effect of initial levels of inequality on patterns of occupational choice and subsequent inequality trajectories. all those papers demonstrate the theoretical plausibility that some redistribution may increase efficiency.

empirical examples of aggregate underinvestment arising from the inability of the poor to access credit and insurance markets on equal terms now abound. in one striking case from africa, goldstein and udry (1999) document the failure of many farmers to switch from a low-return maize and cassava intercrop to a much more profitable pineapple culture in southern ghana. despite an expected return of 1,200 percent, only 190 out of 1,070 plots in the study sample made the switch. when asked why, the modal answer was “i don’t have the money.” in sri lanka, de Mel, McKenzie, and woodruff (2008) use a randomized experimental design to estimate the return on capital for microenterprises that generally are thought to be credit constrained. they find average monthly real rates of return of 5.7 percent—much higher than the market interest rate. the existence of investment projects (in preexisting firms) that are profitable at the prevailing market rate, but that do not take place (before the intervention) is prima facie evidence that the credit market is imperfect. until the underlying causes of failures in credit and insurance markets can be corrected, this kind of evidence suggests that targeted cash transfers can be useful not only in reducing inequality and current poverty, but also in reducing inefficiencies in the economywide allocation of resources.

position—with informed rational agents, benevolent governments, and functioning markets—should be to favor unconditional cash transfers. standard revealed preference arguments establish that choice sets are larger for an unconditional cash transfer than if the same transfer is given in kind with a no-resale condition. similar arguments also show that a consumer is at least as well-off under a cash transfer as when the same budget is used to subsidize a particular good. a transfer that is conditional on the purchase of a specific good (or the use of a particular service) is, of course, analogous to such a subsidy (even if the posttransfer price is negative). 49

Conditional Cash transfers: reduCing present and future poverty

there are three main conceptual arguments for conditioning a cash transfer. first, agents do not always behave exactly as one would expect fully informed, rational agents to behave. private information about the nature of certain investments, or about their expected returns, may be imperfect and persistent. there is also a body of evidence from recent research in behavioral economics that suggests that people often suffer from self-control problems and excessive procrastination, in the sense that their day-to-day behavior is inconsistent with their own long-term attitude toward the future (for example, see o’donoghue and rabin [1999]). there also may be conflicts of interest within the household, either between the parents (who “pay” for education or health services today) and their children (who benefit tomorrow), or between the father and the mother. these conflicts of interest may result in “incomplete altruism”: parental decisions that are not fully consistent with what the child would have chosen herself, if fully rational. what imperfect information, myopia, and incomplete altruism have in common, for our purposes, is that they may cause a family’s privately chosen level of investment in human capital to be too low, compared with its own “true” private optimal.3 if they are pervasive, then these distortions in private decision making provide some contemporary support to the time-honored notion that governments may “know better” what is privately good for poor people than do the poor themselves, at least in some realms. the following section of this chapter therefore reviews these arguments under the general heading “Microfoundations of paternalism.” the second main conceptual argument for conditioning a cash transfer is that governments typically do not behave like textbook benevolent dictators. policy decisions generally result from decision-making processes that involve voting, lobbying, bureaucratic and interagency bargaining, and a variety of other forms of what one broadly might call political economy. under some circumstances, conditioning cash transfers on “good behavior” may increase public support for them, making the program either feasible or better-endowed. the third section below briefly reviews this political economy argument. a third set of justifications for making cash transfers conditional is that, even if the levels of human capital investment by the poor were privately optimal, they might not be socially optimal because of the presence of market failures, particularly, externalities. these justifications are considered in the fourth section, “social efficiency arguments.” 50

the eConoMiC rationale for Conditional Cash transfers

The Microfoundations of Paternalism the idea that poor people need the push (or nudge) of government “incentives” to behave in ways that are “good for them” is a very old notion. it seems to imply that if left to their own devices, these agents somehow are not capable of choosing what is in their best interests. although it is not a very fashionable notion among most mainstream economists today, paternalism (under different guises) has long been used to promote conditional forms of redistribution. Consider, for instance, the idea that there are some specific goods that society sees as essential, as in richard Musgrave’s (1959) description of merit goods or merit wants. these might be goods that enter the social welfare function directly, implying that “society” derives utility from everyone being educated or from everyone having access to decent housing or health care—in addition to the benefits accruing to each individual from his or her own consumption of those goods. another way to think about merit goods is that a “social planner” places greater welfare weight on consumption levels of certain specific goods than the individual himself would place on them (for example, see Besley [1988]). however it is modeled formally, this old idea of merit goods could be used as an argument in support of today’s CCt programs: if society somehow places a value on every child being schooled or having access to health services that is greater than the value that individuals themselves place, then a CCt will provide an incentive toward that additional consumption of the merit good, as desired.4,5 paternalism well may be justified if the individuals in question hold persistently erroneous beliefs; if they are not unitary agents, but households within which there may be conflicts of interest; or if they behave myopically. recent developments in economic theory and recent empirical evidence both suggest that all three of those phenomena may be at work.6 the basic economics of these sources of private decision-making failure can be illustrated using a simple dynamic model of educational choice, outlined more formally in ferreira (2008). the essence of the model is that individual welfare depends on consumption in two periods—childhood and adulthood. the link between the two periods is that children can contribute to household resources during childhood by working (some of the time) in the first period. But any time spent working comes at the expense of time spent studying (or otherwise 51

Conditional Cash transfers: reduCing present and future poverty

investing in the child’s human capital), and thus at the expense of earnings and consumption during adulthood. this trade-off between present and future welfare is at the heart of educational decisions made for or by children in developing countries, where child earnings (or contributions to family enterprises) often are not negligible.7 this simple framework sheds light on the consequences of each of the three kinds of distortion listed above for educational choice and for child welfare (under different assumptions about how the credit market works). in each case, the model tells us what we might expect from two kinds of policy response: unconditional cash transfers (uCts) and conditional cash transfers.8 when credit markets are missing, many of those insights can be illustrated by figure 2.1, where the household’s choice of child time spent in school is plotted against the market wage rate for child labor. the schooling investment function yields the household’s demand for schooling, given the prevailing child wage rate (wm ) and a set of additional parameters, such as expected returns to schooling, the discount rate, the quality of the education expected from the school, and other incomes available to the family. when plotted against the wage rate, it slopes downward: the higher the opportunity cost of attending

Figure 2.1
High investment 1

Choice of Investment in Children with Missing Credit Markets

Investment in schooling

E D

A Original investment function C Investment function with transfer

Low investment

B

Shifted investment function 0 wm–τ wm

Child wage rate
Source: authors’ illustration. Note: wm = child wage rate.

52

the eConoMiC rationale for Conditional Cash transfers

school, the lower the desired investment in education. Changes in the remaining arguments shift the function up or down in this space. for instance, a decline in the expected returns to (or quality of) education, a rise in the discount rate, or a reduction in the levels of other income sources available to the household would shift the investment function in figure 2.1 from a position such as that denoted “original” to one such as that denoted “shifted.” this simple framework can be used to investigate the effects of the distortions mentioned above. Misinformation or Persistently Misguided Beliefs suppose that, for some reason, potential beneficiaries are poorly informed about the future returns to education. of course, if this is a simple information asymmetry with no mechanism causing the incorrect belief to persist, then the optimal policy intervention is to address the information problem—say through a publicity campaign. But processing information may be costly: Being convinced about the health benefits of greater schooling, for instance, may require time and effort to process the evidence. in addition, certain beliefs may be self-reinforcing so that when agents act on the basis of the beliefs, the outcomes confirm them, even if alternative beliefs would have led to superior outcomes.9 it is possible that large groups of people may then believe that returns to education are lower than they really are. a possible example is that of poorer families believing that effort (perhaps in education) is less important than connections in generating upward mobility, whereas those who are better-off believe the opposite. these beliefs can lead to different actions and thus to different outcomes that appear to confirm the initial beliefs—even though the poor also would have benefited if they had put in greater effort. it also is possible that people hold incorrect beliefs about how human capital itself accumulates (rather than about returns). they may believe that formal schooling requires very high levels of natural talent that are not to be found in their families. they may ignore the existence of links between formal schooling and parenting skills or health and hygiene outcomes. these inaccurate beliefs may result from the insufficient availability of information or from difficulties in processing the information that is available. if parents are poorly educated, it may limit their ability to process the information on education complementarities or on rates of return.10 53

Conditional Cash transfers: reduCing present and future poverty

in fact, there is some evidence that incorrect beliefs about education returns can persist in real populations. a practical way to test for such information problems is to compare expected returns to schooling (for example, by asking students or parents what they think) with the observed Mincerian returns from a household survey (“realized returns”).11 for example, attanasio and Kaufmann (2008) compare expected and realized returns among 15–25-year-olds in Mexico and find that expected returns from additional schooling were lower than the realized returns, especially among children of fathers who have low education levels. similarly, Jensen (2006) finds that, in the dominican republic, eighth-grade students estimated the rate of return to secondary school to be only one quarter to one third of the rate derived from an income survey. when students were told of the real rate, those who underestimated it in the first place increased their secondary school graduation rate by 6 percentage points. Both studies provide evidence consistent with information failures resulting in inefficiently low investments in education. But the evidence is not conclusive, and testing for information failures in other types of investments in human capital is less straightforward.12 also, investments in education may have positive impacts on health, and vice versa—another good reason why individuals may have a hard time adequately estimating returns. for instance, Jalan and ravallion (2003a) show that piped water only improved health status when mothers were educated. along similar lines, de walque (2007) shows that an hiv/aids campaign in uganda was most effective among educated households. Miguel and Kremer (2004) provide an example of how health investments (deworming) improve education outcomes. as previously indicated, if the problem is only missing information, providing the information directly would seem to be the first-best intervention (rather than having a CCt). for example, dupas (2007) shows how informing girls that hiv prevalence was higher among adult males and their partners than among teenage boys led girls to avoid the cross-generational partnerships that are riskier in terms of hiv infection rates. however, there are at least two situations under which simply providing information may not do much. first, incorrect beliefs may be self-reinforcing, in which case merely providing the information will 54

the eConoMiC rationale for Conditional Cash transfers

not help. second, passively providing information may not be enough because, at least initially, people may not think they need the information and so may not respond to it. a transfer conditioned on attending relevant information sessions or activities thus may be a better option. one example is parenting interventions. there is a great deal of evidence that poor outcomes in early childhood can be a result of poor home environments, including inadequate parenting practices (some of this evidence for the united states is discussed in box 2.2). however, most people believe that they are not bad parents and, therefore, are unlikely to respond either to an information campaign or to home-visiting programs in which social workers teach them how to be better parents. oportunidades and some other CCts attempt to expose parents to new information and practices by conditioning transfers on participation in

Box 2.2

Investing Early in the Life Cycle whereas the research by heckman and his coauthors has focused on the united states, some of the conceptual underpinnings are relevant for the design of CCts in developing countries. unlike most other programs, CCts seek to improve outcomes in various dimensions of child well-being, including education, health, and, through the pláticas, parenting practices (although pláticas are required in some but not all CCts). arguably, CCts therefore implicitly attempt to exploit the synergies that heckman and others have identified. in addition, CCts seek to build the human capital of children throughout the life cycle, including at the earliest ages. finally, CCts transfer cash, and this can help alleviate the resource constraints that partly explain the adverse home environments and inadequate investments by parents in their children. that being said, CCt programs would benefit from more experimentation to see what combination of cash, conditions, social marketing, and information provision is most effective at ensuring that children do not fall behind at early ages—one of the main messages of this report.

noBel-winning eConoMist JaMes heCKMan

and many others have argued recently for the importance of investments in early childhood (heckman and Masterov 2007; heckman 2008). this research makes a number of important points: (1) poor outcomes in early childhood have long-lasting implications for functioning in adulthood, including low earnings, increases in the likelihood of criminal activity, and poor parenting practices; (2) poor outcomes in early childhood are often a result of adverse home environments, including the absence of a stable family structure and nurturing relationships for children; (3) interventions in early childhood increase the productivity of interventions later in the life cycle (“learning begets learning”); (4) deficits in early childhood are much more costly to remedy later; and (5) investments in different dimensions of child well-being, such as those that lead to improvements in cognitive skills, behavioral outcomes, and child health, are interlinked in important ways. improving outcomes in one dimension makes it more likely that children will be able to make up deficits in other dimensions.

55

Conditional Cash transfers: reduCing present and future poverty

pláticas; cumulative exposure of this sort may help do the trick, and the conditioned cash helps ensure that parents attend and participate in the talks (schady 2006). what might be the consequences of persistent misinformation? an underestimation of returns to schooling, for instance, could lead to an inefficiently low level of investment in education (or health), even under perfect credit markets. with lower expected returns, the demand for education depicted in figure 2.1 shifts downward and schooling investment falls from a high level in point a, to a low level in a point such as B. since actual returns to schooling are higher than the household’s expectation, point B is a (privately) inefficiently low level of schooling. some form of intervention may be warranted—but what kind? Because credit allows for a separation between investment and consumption decisions, a uCt would have no effect on investment under perfect credit markets. a uCt merely raises the overall level of permanent income. and when capital markets are perfect, investment in one’s children, just like any other investment decision, is independent of one’s income levels and depends only on expected returns and the interest rate.13 a CCt, on the other hand, can help shift the investment level toward the optimal by reducing the opportunity cost of studying. it alters the expected returns to investment by affecting the price associated with the investment good, in addition to raising income levels. a CCt would move the agent along the shifted investment function, from point B upward and to the left. on the other hand, if credit markets are imperfect, the effect of misguided beliefs is likely to be even greater, particularly for the poor. when credit is not available, those who are poor today may find it optimal to use child labor as a (very costly) consumption-smoothing mechanism: children may be sent out to increase the availability of consumption goods today, even at the expense of higher remuneration in the future. in this case, even a uCt would have some effect on present levels of investment in health and education. these effects correspond to the income effect of the transfer and reduce the effects of missing credit markets on educational investment. nevertheless, a CCt generally will have a larger positive effect on investment than will an unconditional transfer of the same amount. this outcome simply follows from the fact that a CCt adds a substitution effect to the income effect of the uCt. if an underestimation of expected returns 56

the eConoMiC rationale for Conditional Cash transfers

to education had shifted the household from point a to point B in figure 2.1, then a small uCt could shift it back upward to point C. a CCt of the same amount will move the household’s choice to point d, entailing a higher level of investment in schooling. By remunerating school attendance, the CCt effectively lowers the opportunity cost of studying, relative to working. note that for the household’s welfare (rather than simply its investment in schooling) at point d to be higher than at point C, it is critical that the household be operating under incorrect beliefs. that is why, as previously discussed, another imperfection (in addition to the credit market failure) is required to justify the condition. Credit constraints are relaxed by cash, not by conditions. if there were no additional imperfection and the only problem were a credit constraint, a uCt should be preferred. a CCt that provided the same income transfer would only inefficiently distort behavior (toward excessive schooling) through the condition. the existence of the substitution effect discussed above has another important implication for program design: it is possible to set a CCt level too high, thus encouraging children to a rate of service use that is greater than optimal. this situation (which corresponds to points northwest of point e in figure 2.1) is evocative of anecdotes about children wasting valuable time in classrooms where they learn nothing instead of helping their parents in the field, or of children taken to unsanitary health facilities that act as disease contamination foci because parents have been bribed to take the risk. the upshot seems to be this: because CCts impose a condition, they are more powerful instruments for inducing behavioral change than are uCts. they are “higher-risk/higher-return” policy instruments. when private behavior is suboptimal, they correct it at a lower cost. when private behavior was fine to begin with, their misuse is likely to be costlier. Principal–Agent Problems within the Household, or “Incomplete Parental Altruism” even if parents have a correct expectation of future returns to education, they may discount the future more heavily than is optimal from the point of view of the child. Basic models of schooling choice usually are written under the simplifying assumption of a unitary household. when that 57

Conditional Cash transfers: reduCing present and future poverty

assumption is relaxed and the objectives of different household members are allowed to differ (say, by having different discount rates), then the ensuing conflict of interest within the household may provide another justification for CCts: parents make the education decision for their children, but discount the future at a higher rate and, therefore, demand less schooling than the child’s optimal. if policy makers take the view that the child is the principal in the matter of her own education, and that parents act as her agents, then a principal–agent problem is characterized. a slightly different but equally plausible version of this problem is a conflict of interest between the parents themselves, as opposed or in addition to one between parents and children. one possibility is that mothers’ objectives are more closely aligned with those of all her children. this closer alignment is mentioned often as a justification for handing the transfer to the mother (when there is one), as is common practice in most CCt programs, rather than to the father. it turns out that differences in the discount rate do not affect the investment decision under perfect credit markets. Changing the discount rate will affect consumption choices—how much is consumed now versus how much is consumed in the future. any adjustment will take place through borrowing or lending, with no effect on schooling or any other investment. as noted previously, investment and consumption decisions are separate.14 But if credit markets are missing, then a higher parental discount rate affects the education decision in a way that is exactly analogous to a lower expected rate of return to education. the results described when discussing the effects of misguided beliefs do hold, with both a uCt and a CCt resulting in higher investments and welfare (for the child), but the conditional transfer does so at lower cost because of the induced substitution effect. what is the empirical evidence on intrahousehold principal–agency problems? it is hard to test conclusively for the presence of “incomplete altruism”.15 perhaps the most compelling evidence in that regard is the presence of gender differences in child human capital. the clearest case is in countries where girls’ education lags significantly behind that of boys, even when the Mincerian rate of return to women’s education is at least as high as that of men.16 this kind of differential is prima facie evidence of inefficient underinvestment in girls’ schooling and is most apparent in south asia. such differentials may be rational from the viewpoint of parents who are thinking of their own welfare (because girls are more costly in terms of dowries or boys are more likely to take 58

the eConoMiC rationale for Conditional Cash transfers

care of their parents than are girls who move to their husbands’ homes upon marrying), but they are most likely socially inefficient. somewhat more indirect evidence from countries outside south asia has suggested that differential bargaining power between men and women affects the level of human capital investment in children. indeed, there is a lengthy body of empirical literature showing that when mothers have greater control over resources, more resources are allocated to food and children’s health and education (thomas 1990; hoddinott and haddad 1995; lundberg, pollak, and wales 1997; Quisumbing and Maluccio 2000; attanasio and lechene 2002; rubalcava, teruel, and thomas 2004; doss 2006; and schady and rosero 2008). that evidence provides a strong justification for making payments to mothers, as CCt programs do. at the same time, in circumstances in which women’s power within the household is limited, attaching strings to the transfers by mandating specific human capital investments could strengthen the mother’s bargaining position and reinforce her ability to shift household spending and time allocation decisions.17 although the extent to which incomplete altruism can provide a blanket justification for the use of conditions is unclear, there is now a substantial body of evidence suggesting that parents (especially fathers) value their own utility more than that of their children. girls, in particular, often are at a disadvantage. the implication is that schooling and health levels chosen on a child’s behalf are likely to be too low relative to the child’s optimal level, and that conditions attached to cash transfers can help drive the actual household choices toward that optimal.

A Political Economy Argument the second class of arguments that may provide a justification for conditioning a cash transfer has to do with the political economy of funding redistribution. transfers, whether conditional or unconditional, need to be financed, and budget allocation decisions are never really the choice of a benign social planner. rather, they are the outcome of a (generally complex) political economy process. Most standard theories of the political determination of redistribution do not distinguish CCts from uCts. voters are assumed to care only about their final welfare level, so they look at how much they receive in transfers and at how much they pay in taxes. if voters are not recipients of a targeted transfer, then, 59

Conditional Cash transfers: reduCing present and future poverty

conditional on their tax bills, they should be indifferent to whether there are specific conditions attached to the transfers. one implication of that kind of analysis is that transfer schemes narrowly targeted to the poor would tend to have limited support because a small share of the population benefit, whereas the costs are dispersed across all taxpayers. gelbach and pritchett (2002) have a model in which an increase in the degree of targeting actually can result in a reduction of both the equilibrium level of the transfer and the welfare of the poor. the implied vulnerability of targeted redistribution schemes to political change enjoys a measure of empirical support (see subbarao et al. [1997] for examples from Colombia and sri lanka). it is conceivable, however, that voters (or other decision makers) are not entirely self-regarding. it is possible, for instance, that taxpayers are more prepared to pay for transfers to those who are seen to be helping themselves than to other equally poor people who are seen to be lazy or careless. some voters who object to unconditional “handouts” may be less averse to “rewards” to “deserving” poor people who are investing in the education or health of their children. (Box 2.3 contains a brief summary of evidence from the recent behavioral literature on fairness, which suggests that many people are routinely prepared to incur real financial losses to reward others whom they think are deserving or to punish those they feel have behaved unfairly.) if this view is commonly held, the introduction of conditions may result in an increase in the overall budget available for redistribution in the political equilibrium. if none of the private inefficiencies discussed in the second section of this chapter existed, then attaching a condition to a cash transfer would be, of itself, suboptimal to beneficiaries (because it adds an additional constraint). But that cost may be offset by an increase in the overall size of the transfers that are funded, in which case the conditions will be justified for political economy reasons. the condition is justified by making redistribution more acceptable to taxpayers and voters—and possibly to many beneficiaries. another way of seeing this is that, unlike a uCt, a CCt can be seen not as plain social assistance, but rather as part of a social contract whereby society (through the state) supports those poor households that are ready to make the effort to “improve their lives”—the deserving poor. the notion that CCt programs constitute a new form of social contract between the state and beneficiaries has been manifested in the use of the term co-responsibilities (instead of conditions) in a majority of 60

the eConoMiC rationale for Conditional Cash transfers

Box 2.3 Fairness, Merit, and the “Deserving Poor” people of ten Beh av e in ways t h at a r e

inconsistent with pure self-regarding preferences. in particular, there is now a substantial body of experimental evidence suggesting that large numbers of people are altruistic rewarders or altruistic punishers in the sense that they are prepared to incur personal losses to reward behavior they regard as socially fair or to punish behavior they regard as unfair (see fehr and schmidt [1999] for the basic theory; fehr and gächter [2000] for a review of the early evidence). some of the main results come from experiments in which subjects were asked to play what is known as an ultimatum game under experimental conditions. in the game, a first-mover proposes a split (of an exogenously given sum) between himself and the second player. if the second player accepts the proposal, the split is implemented. if he rejects it, both players earn zero. if people behaved as standard economic theories used to predict (that is, if preferences were purely self-regarding), then the outcome of this game—the so-called nash equilibrium— would be “as little as possible for you, all the rest for me.” empirically, however, such an outcome is seldom observed. the modal offer is in the 30–60 percent range, depending on the cultural context.

and a sizable fraction of offers below that range is rejected outright even when the sums in play are nontrivial: people appear to be prepared to “pay” for the opportunity to punish a player whom they see as having behaved unfairly. More interesting from the point of view of CCt programs, hoffman et al. (1994) find that players in anonymous ultimatum games tend to be more tolerant of other players in positions of power when those positions are allocated on the basis of “merit” (that is, to those who score higher in a general knowledge quiz) than when they are allocated randomly. the accumulated evidence from the large body of literature on fairness suggests that people take considerations of “justice” into account when making decisions. the evidence from this particular study suggests that people’s perceptions of what is a fair distribution may be affected by the perceived “merits” of the recipient. would a similar line of reasoning imply that taxpayers (or public officials) might be more willing to fund transfers to people who invest in the future of their children than to others who do not? although some tentative suggestions in support of this conjecture are reported in the main text, more research is needed to address the question rigorously.

programs, at least in latin america. this use is illustrated clearly in the words of the architects of the CCt program in Mexico: poor families need help, but this should not suppress or undermine their role as protagonists in transforming their living conditions. shared responsibility and respect are not only prerequisites for effectively combating poverty but are essential elements of a democratic society. shared responsibility and respect inevitably imply a reciprocal effort by the poor families to link the benefits they receive to concrete actions on their part. thus independently of technical considerations, in 1996 61

Conditional Cash transfers: reduCing present and future poverty

it was considered vital that progresa benefits go directly to poor families and be conditioned on direct action by them to improve their own nutrition, health and education, and that such support complement but not substitute for their day-today efforts (levy and rodríguez 2004, p. 48). a recent analysis of the treatment of the Bolsa família program in the Brazilian media (lindert and vincensini 2008) provides additional support to the idea that conditions make transfer programs more politically palatable. there is little question that Bolsa família is a popular program in Brazil. an ipsos opinion poll taken in september 2007 found that the program tops the list of items mentioned in response to a question on what president luiz lula da silva had done well in office. the authors of the study find that most media criticism of the program centered on the possibility that it would “generate dependency” and find that this criticism usually was coupled with reports that the conditions were not being monitored and enforced properly. Conversely, most of those people arguing the program was not assistencialista listed the existence of conditions as one of the top two reasons. lindert and vincensini (2008) conclude that the acceptance of conditions across the political spectrum—where the left sees the conditions as merely restating citizen’s rights, and the right tends to see them as enforceable contracts—played an important role in generating broad-based support for the program in Brazil. this perception of the condition as a mutually agreeable contract leads to an interesting apparent paradox: CCts often are seen as less, not more, paternalistic than uCts. indeed, several authors have argued that CCt programs provide the basis for a less paternalistic (and possibly less clientelistic) form of social assistance (Cohen and franco 2006; de la Brière and rawlings 2006). reconciling this view with the “paternalistic” arguments described in the second section of this chapter requires distinguishing between two very different justifications for conditioning. the first justification, which was discussed in the second section above, relies on imperfections in private decision making by the poor households themselves. they might be poorly informed, parents may not fully internalize the best interests of their children, and so on. Conditions then help, by inducing agents to do what is best for their children, individually. the second view, which is being proposed here, is that when conditions are seen as co-responsibilities, they treat the recipient more as a 62

the eConoMiC rationale for Conditional Cash transfers

“grown-up,” capable of agency to resolve his or her own problems. the state is a partner in the process, not a nanny. this latter interpretation is particularly plausible when the counterfactual to a CCt is not an automatic, transparent, unconditional cash grant seen as a citizen’s entitlement (which is close to the textbook concept of a uCt), but instead a myriad of ad hoc and mostly in-kind transfers, intermediated through various service providers, nongovernmental organizations, and local governments. under those circumstances, conditioning the transfers on “good behavior” may be perceived as less paternalistic than the alternative of conditioning transfers on voting for a certain party or belonging to a given social organization. Moreover, the fact that the conditions are focused on building the human capital of children (rather than simply supporting parents) adds to their political acceptability as an instrument to promote opportunities. after all, it is hard to “blame” children for being poor. in that sense, using public resources to support the human capital development of poor children makes CCt a “poverty reduction” rather than a “social assistance” program. Making payments to mothers also resonates with well-accepted beliefs (mostly supported by evidence, as shown above) that mothers tend to put funds to better use than men do. this view of CCts as an enabling instrument, which creates political viability for targeted redistribution that effectively reaches the poor, resonates in many of the latin american countries that introduced CCt programs over the last decade. social protection systems in the region have been characterized by “truncated welfare states” (de ferranti et al. 2004) that channel significant public resources to subsidize social insurance schemes for the formal labor force and provide little, if any, redistribution (particularly in the form of cash) to the lower segments of the income distribution. as in other spheres (for example, service delivery [fiszbein 2005]), political capture of state institutions and policies by elites meant an historical pattern of low social assistance. from that perspective, the introduction of CCt programs since the late 1990s can be seen as a break with history. take the case of Brazil, a country with extremely high inequality that long has subsidized social insurance programs (with limited reach to the poor). public subsidies for (generally regressive) pension schemes alone represent more than 5 percent of gdp (lindert, skoufias, and shapiro 2006). it was only in the late 1990s, with the introduction of a series of CCt-like programs such as the peti, Bolsa alimentação, and Bolsa 63

Conditional Cash transfers: reduCing present and future poverty

escola, that cash-based social assistance programs became a significant federal public spending item, reaching approximately 0.5 percent of gdp in 2000.18 and it is through the expanded and enhanced Bolsa família program in more recent years that federal spending on social assistance reached the 1 percent of gdp mark (lindert, skoufias, and shapiro 2006). the story is very similar for the case of Mexico: oportunidades represented a major shift from broad price subsidies that benefited the poor only marginally to a cash-based redistribution to more than 5 million poor households. and, as discussed in chapter 1, Mexico’s success appears to have influenced other countries in the region to follow a similar path. the conclusion is that even in situations where a narrow technical assessment might suggest that a uCt is more appropriate than a CCt (say, because there is no evidence of imperfect information or incomplete altruism in poor families), CCts might be justified because they lead to a “superior” political economy equilibrium. the political process may make significant cash transfers to the poor close to impossible unless those transfers are tied somehow to clear evidence of commitment and “positive behaviors” on the part of beneficiaries. once again, the latin american experience suggests that in the absence of dramatic political shifts, the increasing trend toward cash-based redistribution schemes has been associated with the use of some form of conditioned grants.

Social Efficiency Arguments attaching conditions to cash grants might make sense for political economy reasons or because distortions in individual behavior cause decision making in the household to be privately inefficient. we now turn to a third set of reasons for conditioning, namely, human capital externalities. if investments in human capital generate positive externalities that parents do not take into account when making decisions, then the aggregate (market equilibrium) level of human capital in society will be inefficiently low. this is a standard argument for subsidizing provision of education or health care. empirically, health investments have important external benefits. although those benefits are well established in some cases (for example, immunization), the supporting evidence

64

the eConoMiC rationale for Conditional Cash transfers

is rather new in other cases (such as deworming [Miguel and Kremer 2004] or insecticide-treated nets [gimnig et al. 2003]). in the case of education, externalities might arise if there are increasing returns to skilled labor in production, at the aggregate level. there is empirical support for the idea that more education can have spillover effects to other workers in the same plant (Moretti 2004b), in the same village (foster and rosenzweig 1995), or in the same city (Moretti 2004a). possible spillovers also may be present if crime, violence, and related social ills decline with average schooling levels. there is solid evidence for the united states that education lowers crime—with perhaps the best evidence coming from the perry preschool program evaluation, which shows that children randomly assigned to the intervention have much lower incarceration rates as adults (Currie 2001; schweinhart 2004). how large these externalities are and whether (conditioned) cash transfers are the most effective instrument to correct for them, however, remains to be determined. in most countries, education and health services already are heavily subsidized. in many cases, they are publicly provided free of charge. to argue for an additional subsidy that compensates households for some of the indirect or opportunity costs of using these services, on the basis of the externality alone, would require showing that those externalities are quite large. if that were found to be true, then a CCt can be justified on that basis alone: it is effectively an additional component to a pigouvian subsidy, which often already is implicit in the service fees.

Conclusion although market-driven economic growth is likely to be the main driver of poverty reduction in most countries, markets cannot do it alone. public policy plays a central role in providing the institutional foundations within which markets operate, in providing public goods, and in correcting market failures. in addition to laying the groundwork for economic growth, policy also can supplement the effects of growth on poverty reduction, and one of the instruments that governments can use to that end is direct redistribution of resources to poor households. although direct cash transfers have opportunity costs (in terms of forgone alternative public investments) and may have some perverse incen-

65

Conditional Cash transfers: reduCing present and future poverty

tive effects on recipients, there is a growing body of evidence to suggest that some such transfers may be both equitable and efficient. the cash transfer programs that have been growing most rapidly across the developing world over the last decade or so are CCts, by which cash is paid to poor households on condition that they invest (in certain prespecified ways) in the human capital of their children. Because attaching a constraint on the behavior of those you are trying to help is an unorthodox idea for economists, this chapter has reviewed the conceptual arguments for making cash transfers conditionally. essentially, there are two broad sets of arguments for attaching conditions to cash transfers. the first argument applies if private investment in children’s human capital is thought to be too low. the second argument applies if political economy reasons mean that there is little support for redistribution, unless it is seen to be conditioned on “good behavior” by the “deserving poor.” CCts are not a panacea. if there is little evidence to suggest that private levels of investment in human capital are too low (in any of the senses previously discussed), and if the political economy can accommodate the desired levels of redistribution without appealing to coresponsibilities, then uCts (or some completely different kind of public expenditure) may be preferable. there also may be good arguments against conditioning if the same result can be achieved at a lower cost through the social policy equivalent of “moral suasion.” recent research has found evidence of “flypaper” or “labeling” effects, whereby the household expenditure shares of certain goods are higher out of transfers that are notionally earmarked for (but not conditional on) those goods, than out of other incomes.19 flypaper effects constitute a fairly fundamental violation of rationality in that they suggest households do not treat all their income sources as fully fungible. like other departures from full rationality, flypaper effects certainly are plausible and, if ubiquitous, could have serious implications for the design of social protection, tipping the balance in favor of uCts (which then would appear to have some of the benefits of conditions, without the costs). But much more research is needed before the evidence on these effects reaches critical mass.

66

Chapter three

Design and Implementation Features of CCt programs
BeFore DelvIng Deeply Into the ImpaCts oF CCt programs, It

is worth understanding some of the details of how the programs work. CCt programs require the same systems as other transfer programs: at minimum, a means to establish the eligibility of clients and enroll them in the program, a mechanism to pay their benefits, and preferably strong monitoring and evaluation systems. CCt programs further require a means to monitor compliance with co-responsibilities and to coordinate among the several institutions involved in operating the program. In general, CCt programs have handled these systems rather well; in some cases, they have been leaders in modernizing social assistance practices. of course, technical soundness is neither inherent to nor the exclusive domain of CCt programs. that fact should be understood fully by policy makers across the gamut of social policy so that those working on CCt programs make the deliberate choices required to continue the tradition of excellence, and those working on other sorts of programs adopt some of the practices that have led to success in the best of the CCt programs. this chapter describes the nuts and bolts of the operation of CCt programs.1 the chapter is divided into five sections, corresponding to targeting practices, benefit systems, conditions (including their definition, monitoring, and enforcement), monitoring and evaluation, and issues concerned with intersectoral and interinstitutional coordination.

Targeting in Practice almost all CCt programs established to date have tried to target their benefits rather narrowly to the poor.2 table 3.1 shows the targeting mechanisms used for a large number of programs, both established and nascent. 67

Table 3.1

Targeting Methods Used in CCT Programs, by Region
Categorical Household identification Proxy Means Community means test test assessment

Region/Country/Program

Geographic

Other

Africa
Burkina Faso: orphans and vulnerable x Childrena Kenya: Ct-ovCa nigeria: Cope x orphan and vulnerable children incidence x x x x

East Asia and Pacific
Cambodia: Cessp Cambodia: JFpr Indonesia: Jps Indonesia: pKh philippines: 4ps x x xb x x gender and ethnic minority genderc x x x x x x

Europe and Central Asia turkey: srmp x

Latin America and the Caribbean argentina: programa Familias x Beneficiaries of Jefes y Jefas program, with two or more children; head has not completed secondary schoold xe x x x x x not part of social security system x x x x x x xf x x x x x x x x x x

Bolivia: Juancito pinto Brazil: Bolsa alimentação Brazil: Bolsa escola Brazil: Bolsa Família Brazil: petI Chile: Chile solidario Chile: sUF Colombia: Familias en acción Colombia: sCae-Bogotá Dominican republic: solidaridad Dominican republic: tae/Ilae ecuador: BDh el salvador: red solidaria guatemala: mi Familia progresa

68

Categorical Region/Country/Program honduras: praF Jamaica: path mexico: oportunidades nicaragua: atención a nicaragua: rps panama: red de oportunidades paraguay: tekoporã/propaIs peru: Juntos IIh Crisisa x x x x x x Geographic x Other

Household identification Proxy Means Community means test test assessment xg x x x x x x x

Middle East and North Africa yemen: BeDpa x gender

South Asia
Bangladesh: Fssap Bangladesh: pesp Bangladesh: rosC India (haryana): apni Beti apna Dhan pakistan: Cspa pakistan: participation in education through Innovative scheme for the excluded vulnerable pakistan: pesrp/punjab Female school stipend program x x xi x x gender Beneficiary of food support program x x x gender x x

x

gender

Source: program profiles. Note: BDh = Bono de Desarrollo humano; BeDp = Basic education Development project; Cessp = Cambodia education sector support project; Cope = Care of the poor; Csp = Child support program; Ct-ovC = Cash transfer for orphans and vulnerable Children; 4ps = pantawid pamilyang pilipino program; Fssap = Female secondary school assistance program; JFpr = Japan Fund for poverty reduction; Jps = Jaring pengamanan sosial; path = program of advancement through health and education; pesp = primary education stipend program; pesrp = punjab education sector reform program; petI = programa de erradicação do trabalho Infantil; pKh = program Keluarga harapan; praF = programa de asignación Familiar; rosC = reaching out-of-school Children; rps = red de protección social; sCae = subsidio Condicionado a la asistencia escolar; srmp = social risk mitigation project; sUF = subsidio Unitario Familiar; tae/Ilae = tarjeta de asistencia escolar/Incentivo a la asistencia escolar. a. program at the pilot stage. b. at both the national level (to identify the poorer districts) and the district level (to identify the poorer subdistricts/schools). c. at least half of the scholarships at the school level were to be allocated to girls. d. the Jefes y Jefas program started as a workfare program for unemployed heads of household. e. Covers all children in public schools up to fourth grade. f. household targeting is only in the 68 less-poor municipalities. targeting in the poorest 32 municipalities is geographic only. g. only households in the area covered by the Inter-american Development Bank project may participate. h. propaIs II is a project, financed by the Inter-american Development Bank, that builds on the tekoporã program and finances additional beneficiaries using similar procedures. i. only certain types of schools in rural areas may participate.

69

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

about two thirds of countries use geographic targeting; about two thirds use household targeting, mostly via proxy means testing; and many countries use both. moreover, many programs use community-based targeting or community vetting of eligibility lists to increase transparency. the methods of proxy means testing vary in their details. For example, in all cases, the formula for the proxy means test was derived from statistical analysis of a household survey data set; but, of course, there are differences in the quality and detail of that original data set, and differences in the statistical methods used and in the sophistication and rigor thereof. significant variations also exist in how the implementation is done— whether households are visited; whether some variables are verified as part of the application process for all or for a sample of applicants; whether the staff members who help complete applications are permanent or contract workers and to which agency they report; and other such differences. Usually the proxy means testing system is led by a central agency (whether in the CCt program itself, independent, or in the ministry of planning), but the day-to-day staffing for it is delegated, often to municipalities, with considerable variability in independence and quality control. In many cases, CCt programs have been the drivers for developing poverty maps or household targeting systems in their countries or for prompting upgrades to them. Indeed, it would not be an exaggeration to say that CCt programs have moved forward the state of the art and standards for targeted programs generally. many countries first established proxy means tests when designing the CCt program (Cambodia, Jamaica, Kenya, mexico, pakistan, panama, turkey). some countries with older proxy means tests have made significant reforms and improvements in their systems over time—if not because of, then certainly to the advantage of, their CCt programs (Chile, Colombia). some of these are relatively low-income countries with limited administrative capacity, and they have made adaptations to accommodate that situation. Box 3.1 illustrates this for the case of Cambodia’s scholarship program. the household targeting systems used in some of the best-known CCt programs constitute major “institutional capital” for the country. the same system often is used to target many programs, sometimes with different thresholds or ancillary criteria. For example, in Chile (the first country to use proxy means tests extensively), the system is used not just for the recent Chile solidario program, but also for much older child allowance and social pensions, for water price subsidies, for housing subsidies, and for other uses. similarly, in Colombia, the same proxy means test (the sIsBen) used to determine eligibility for subsidized 70

DesIgn anD ImplementatIon FeatUres oF CCt programs

Box 3.1 Proxy Means Testing Where Administrative Capacity Is Low: Cambodia’s Scholarship Programs
B e C aU s e C a m B o D I a h a s r a t h e r l e s s

administrative capacity than the middle-income latin american countries where proxy means testing originated, it has adapted the general practice of proxy means testing in a way that makes rigorous but simplified testing viable. the schools that participate in its scholarship program are subject to a prior round of geographic targeting, and applicants complete a proxy means test that is used to allocate scholarships among each selected school’s students. Cambodia’s Cessp program dispenses with the cadre of field worker/social workers who often administer the instrument. Instead, students fill out the program application/proxy means test form in school. then the teacher reads the information aloud and the classmates help verify/certify that it is correct. a local committee of school and community leaders score the forms by hand.a to assist in manual scoring, the formula uses only integers. the ranking is done only within schools, rather than against a national standard as in most proxy

means tests. In each school, the scored forms are arranged by score and the poorest children, up to the quota for that school, are selected for the scholarship. this process implies that recipients in poorer schools will be poorer, on average, than recipients in less-poor schools. It is thus less accurate than a ranking against a national standard, but eliminates the need for a national database and the information technology and communications networks that would be required to support it. In a previous scholarship program, the formula was not very sound, so the committees were given leeway to deviate when they thought it appropriate; and when they did so, the students selected were, in fact, poor (as judged later by an evaluation survey). subsequently the formula was based on statistical analysis of the same type used elsewhere, and the discretion of the local committees was reduced.
Source: Cessp scholarship team 2005. a. In the first year of the Cessp program, an independent firm scored the forms centrally.

health insurance, hospital fee waivers, the public workfare program, a youth training program, and a social pension has been used to target the CCt program. even in countries with a more recently established proxy means test, those tests can be used in multiple programs. Jamaica established its proxy means test expressly for the CCt program, but now uses it to grant fee waivers in the health system and for secondary education textbook rentals and school lunches. such an investment will pay off sooner for programs that are generous in coverage or benefit levels and for countries that, at least eventually, will use the proxy means tests for multiple programs. What have these procedures accomplished? It is difficult to measure targeting outcomes properly (see box 3.2), but we can approximate 71

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

Box 3.2 Who Benefits from CCT Programs? to KnoW hoW Well a progr am targets, We

need to measure the welfare that a household would have if it did not receive the transfer and then to rank households according to that measure. Doing so would allow for any behavioral responses associated with the receipt (or removal) of the transfer, such as changes in the household’s labor supply, savings, or receipt of remittances. these estimations are undertaken in chapter 4 for those programs for which consumption data are available for beneficiary (treatment) and nonbeneficiary (control) households before and after the start of the CCt program. there are two possible “naïve” ways of determining where transfer recipients fall in the pre-intervention welfare distribution—measured consumption including the transfer and measured consumption minus the transfer. Using consumption including the transfer biases welfare upward, thus making households seem better-off than they would be without

the program. this will give the most conservative estimate of narrow targeting because a beneficiary who is poor even after receiving the transfer surely was poor beforehand. on the other hand, using consumption net of the full transfer value biases the estimate of initial welfare downward if behavioral responses offset the transfer in part. as will be discussed in chapter 4, there is little evidence of significant offsetting behavioral responses through private transfers and labor supply, at least in the early stages of such CCt programs as mexico’s oportunidades and nicaragua’s red de protección social. to investigate how sensitive the targeting assessment is to the welfare measure used, we conducted comparative analysis using both naïve indicators, which will bracket the “true” but imprecisely known counterfactual. Figure 3B.1 shows the results for Jamaica and mexico. as expected, the estimates net of transfer show the programs to be

Figure 3B.1
70 Proportion of households receiving transfer 60 50 40 30 20 10 0 1 2

Coverage Using per Capita Expenditure Deciles Gross and Net of the CCT Transfer, 2004
Jamaica, PATH
70 Proportion of households receiving transfer 60 50 40 30 20 10 0 1 2 3 4 5 6 7 8 9 10 Per capita expenditure Per capita expenditure minus transfer

Mexico, Oportunidades

Per capita expenditure minus transfer

Per capita expenditure

3

4

5

6

7

8

9

10

Decile

Decile

Source: authors’ calculations, using the 2004 Jamaica survey of living Conditions and the 2004 mexico household survey (encuesta nacional de Ingreso-gasto de los hogares).

72

DesIgn anD ImplementatIon FeatUres oF CCt programs

more progressive than do the estimates gross of transfer. the important thing is the magnitude of difference. For Jamaica’s program of advancement through health and education (path), where the transfer is a small share of households’ base consumption, the two curves are rather close together. In the case of mexico’s oportunidades program, where the transfer is among the largest of any CCt program and, therefore, the sensitivity of results to

method is the greatest, the curves diverge more; the estimated participation rate for the poorest decile drops from more than 60 percent (using consumption minus the transfer as the ranking variable) to less than 50 percent (using consumption including the transfer). however, the policy conclusion that the program concentrates resources at the bottom end of the distribution holds for both naïve estimators of the counterfactual.

outcomes for a number of programs, as shown in figure 3.1. those estimates rank households on their observed per capita household consumption (or income, if a measure of consumption is not available) less the value of the transfer received. this is a naïve estimator that will exaggerate the accuracy of targeting if households change their behavior in ways that lower their autonomous, nontransfer income, perhaps through working less or receiving fewer private transfers. as we shall see in chapter 4, these behavioral responses to CCt programs appear to be modest. moreover, the sensitivity analysis illustrated in box 3.2 gives some comfort that biases are not too large and do not affect greatly the conclusion that CCt programs largely have realized their intent to concentrate the benefits on the poorest households. as figure 3.1 reveals, there is significant variation in coverage of the poor, depending on the size and budget of the programs—from about 1 percent of the poorest decile in Cambodia to more than 60 percent in Brazil, ecuador, and mexico.3 the coverage rates in the larger CCt programs seem to compare well with international experience. In the lindert, skoufias, and shapiro (2006) study of 40 targeted programs (including several CCt programs),4 the mean coverage rate of the poorest quintile is 19 percent. In a study of eastern europe and Central asia cash transfers, child allowances, and social pensions (tesliuc et al. 2006), the mean coverage rate of the first quintile is 42 percent. a study reviewing experience in a small number of organisation for economic Co-operation and Development countries was able to model take-up more closely among eligible individuals, and concluded that take-up rates typically are between 40 and 80 percent for social assistance and housing programs (hernanz, malherbet, and pellizzari 2004). 73

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

Figure 3.1
80 70 Proportion of households receiving transfers 60 50 40 30 20 10 0

Coverage of CCT Programs, by Decile, Various Years
Brazil BF 2006 Chile Solidario 2003 Chile SUF 2003 Ecuador BDH 2006 Honduras PRAF 2004/5 Mexico Oportunidades 2004 Jamaica PATH 2004 Nicaragua RPS 2000 Cambodia JFPR 2003 Bangladesh FSSAP 2000 Turkey SRMP Education 2005/6 Turkey SRMP Health 2005/6

1

2

3

4

5

6

7

8

9

10

Per capita consumption minus transfer (deciles) Sources: authors’ calculations, based on the following surveys: ecuador’s encuesta de Condiciones de vida 2006; mexico’s encuesta nacional de Ingreso-gasto de los hogares 2004; Brazil’s pesquisa nacional por amostra de Domicílios 2006; Jamaica’s survey of living Conditions 2004; Chile’s encuesta de Caracterización socioeconómica nacional 2003; honduras’ encuesta de hogares de propósitos multiples 2004; Bangladesh’s household Income and expenditure survey 2000; and the Cambodia Japanese Fund for poverty reduction application form 2003. the numbers for turkey are from ahmed et al. (2007); for nicaragua, the numbers are from the 2002 mesoamerica nutrition program targeting study group. Note: BDh = Bono de Desarrollo humano; BF = Bolsa Família; Fssap = Female secondary school assistance program; JFpr = Japanese Fund for poverty reduction; path = program of advancement through health and education; praF = programa de asignación Familiar; rps = red de protección social; srmp = social risk mitigation project; sUF = subsidio Unitario Familiar. For Brazil and Chile, the deciles are based on per capita income minus transfer per capita. For the remaining countries, the measure of welfare used is per capita expenditure net of the transfer per capita.

In thinking about coverage rates, it is important to distinguish the various reasons for low coverage. some of these are likely to be benign, others problematic. the first and most obvious factor in low coverage is the size of the budget and the role a program is meant to play in broader social policy. Chile solidario is meant to fill a defined and small niche in social policy, with other transfer and subsidy programs providing greater coverage and higher benefits. In ecuador, by contrast, the BDh is meant to be the mainstay of social assistance for families. Its significant size contributes to this goal, as does the wide age range of children 74

DesIgn anD ImplementatIon FeatUres oF CCt programs

covered. naturally, differences in the role of these programs lead to wide variations in the extent to which they cover the poor. several of the smaller programs, however, clearly are constrained by budget and design. Cambodia and honduras are targeted geographically, operating only in defined areas of the country, with the Cambodian program being only a very small pilot at the time of these results. the second obvious factor involved in determining the share of the poorest people that a program can reach has to do with the range of ages and grades that the program covers. a program that covers all families with children aged 0–16 will help more of the poorest quintile than will one that covers families with girls in secondary school grades. (this theme is taken up in more detail in chapter 6.) Box 3.3 provides an in-depth analysis of the extent to which the proxy mean tests and the demographic composition of the households results in errors of exclusion in Brazil and ecuador. a third factor has to do with the requirements for service delivery (nichols and Zeckhauser 1982; Das, Do, and Özler 2005). the essence of a CCt program is that families must make sure their children use health and education services. If those services are not available, then the families will be excluded from the program. some programs, at least in their early years, aimed to cover poor areas as indicated by a poverty map, but they set up operations only in areas where services were deemed accessible and intentionally excluded all those poor people who lived in areas without minimum service capacity. For example, one of the preconditions for rural localities to be covered by mexico’s oportunidades program was that they have a population between 50 and 2,500 individuals as well as a primary school situated within the locality and access to a paved road and a health center within a radius of 5 kilometers. those conditions, by necessity, excluded a small share of very needy villages, although the requirement was relaxed in subsequent years. similarly, for Colombia the program was targeted first by municipality—but only municipalities that offered the necessary services were allowed to join. Initially, 15 percent of municipalities containing 8 percent of the targeted beneficiaries thus were excluded (lafaurie and leiva 2004). more geographically dispersed versions of this problem will occur when households are allowed to participate, but some are more remote from services than others. For example, de Janvry and sadoulet (2005) suggest that oportunidades has little impact on children who live more than 4 kilometers from a secondary school; in 75

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

Box 3.3 Analyzing Errors of Exclusion of CCT Programs, Brazil and Ecuador to get a pICtUre oF the relatIve magnItUDe

of some of the factors that lead to errors of exclusion, we look in detail at the ecuadoran and Brazilian cases, focusing on those two countries because, for them, we are able to infer from the survey more accurately than usual which households are eligible for the programs. In ecuador, the score from the proxy means test used to determine eligibility is available in the data set. In Brazil, the survey captures income, as does the means test. to make the analysis reasonably comparable between countries and pertinent to other countries, we define the target population as the poorest 20 percent of households. For those in the poorest quintile according to our survey-based measure of welfare (consumption net of transfer in ecuador, income net of transfer in Brazil), we parse as best we can the reasons for errors of exclusion. these results are presented in table 3B.1. In ecuador, the proxy means test correctly predicts that 95 percent of households in the poorest quintile are eligible for the benefits, but it erroneously excludes 5 percent of them. among those people in the poorest quintile who are eligible, we find that only 70 percent actually receive program

Table 3B.1 Coverage of Poor Households, Brazil and Ecuador
Share of bottom quintile (%) Reason Classified as eligible because of means or proxy means test eligible, receiving program benefits eligible, receiving program benefits with children ages 0–17 Brazil 100 Ecuador 95

55 54

67 64

Source: authors’ calculations.

benefits—a finding that implies only 67 percent of the poor end up receiving benefits. In the ecuadoran case, the budget covers about 40 percent of households (many more than this exercise considers), so there is no explicit rationing of slots other than the proxy means test. thus one can infer that the lessthan-full coverage of the lowest quintile stems not from a lack of offer by the program but from a lack of take-up by poor households. to understand the factors affecting exclusion from or nonparticipation in the program, we ran a probit analysis on the sample of poor households, looking for predictors of participation. the results show that the BDh has been successful in overcoming some problems endemic in transfer programs—indigenous, less-educated, and female-headed households are less likely than others to be excluded from the program, all else being equal. thus, outreach has been sufficient to include groups who often face barriers to information and access. the role of self-selection would appear to be fairly strong because poorer households participate more often than do the less-poor; the same is true for those in rural areas where the effective value of the transfer is a little higher. there is a caveat: being located in the oriente (the amazonian part of ecuador) does raise the probability of nonparticipation among poor and eligible people, probably a sign that transaction costs deter some of the residents because parts of this region are accessible only by air. that area may be an exception to the conclusion that nonparticipation generally is not too grave an issue in ecuador. In a parallel analysis for Brazil’s Bolsa Família, we see that all households in the poorest quintile have incomes under the eligibility threshold of r$100, a so errors of exclusion from the means test should be zero. that is a simplification because the implementation of the means test may be imperfect. But we would expect that self-declared income on the application would be underreported—if anything, the applicant would

76

DesIgn anD ImplementatIon FeatUres oF CCt programs

have an incentive not to reveal more than she or he had to reveal and the form itself may not fully elicit all income for those with irregular or in-kind income. among those who are eligible by income, enrollment is shown at just 55 percent. In this case, a lack of offer from the program is probably a good deal of the cause. these figures are from 2006, when the program covered about 11 million families. But, of course, there may be other causes for exclusion and an element of self-selection. to understand that possibility, we again ran a probit model to predict the probability of participation based on the sample of poor and eligible households. here, too, there is some good news relative to program outreach: For example, afro-Brazilians were significantly less likely to be excluded, all else being equal. there seems to be an element of self-selection—those with more income and more education and those in urban areas were less likely to participate than were others. (the cost of living differentials are very large in Brazil, so the value of the benefit is implicitly lower in urban areas.) therefore, outreach appeared to be quite good as early as 2006. the government has since expanded

the program and that expansion should reduce any issue of undercoverage substantially. We return to table 3.2 to look at the potential impacts of the demographic restriction on the exclusion of poor households from the program. In ecuador, 95 percent of the eligible participants have children aged 0–17. thus, if the program were restricted to families with children, as is the case with most CCt programs, only a few percent more poor households would be excluded (a reduction of the eligible poor receiving benefits from 67 percent to 64 percent). In Brazil, the poorest families receive a transfer even if they have no children. If the age requirement of having children aged 0–17 were to be enforced, it would exclude only 1 percent of those who were in the program. this finding suggests that demographic restrictions are not an important factor for the exclusion of poor and eligible households from CCt programs with similar designs.
Source: authors’ calculations based on Brazil’s pesquisa nacional por amóstra de Domicílios 2006 and ecuador’s encuesta de Condiciones de vida 2006. a. When this was written, in 2008, the eligibility threshold was r$120, but it was r$100 at the time of the survey in 2006.

turkey, the qualitative evaluation suggests that the transfer is not sufficient for some households to pay for transportation or to compensate households for concerns over students having to travel outside the village (adato et al. 2007). however, the fact that coverage is highest for the poorest decile indicates that the problems are not so widespread that they generally preclude coverage of the poorest people. another possible cause of errors of exclusion is that proxy means tests contain an element of statistical error in making their predictions of household welfare. In panama, the ex ante assessment of the proxy means test is that with the eligibility threshold selected, a quarter of the extreme poor could be excluded from the program (World Bank 2006d). In ecuador, such errors are lower, as we shall see below. errors of exclusion also may occur if outreach to potential beneficiaries to inform them of the benefits of the program and its application 77

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

procedures is lacking and they never apply (see atkinson 1996; grosh et al. 2008, ch. 3). We do not have clear quantitative data on the extent of outreach in all CCt programs, but do have some indications that it has been good in a number of them. Castañeda and lindert (2005) show that in Chile, Colombia, and mexico as early as 2002–04, the proxy means testing systems already had registered more than the number of poor families in the countries. of course, this finding does not mean that all of the poor necessarily were registered, but it does seem that the magnitude of outreach and the administrative mechanism to handle registration were of the right order of magnitude. moreover, we know of a number of innovative or extensive examples of outreach. several countries fielded teams to go door-to-door in poor areas to register households. all countries had information campaigns of one sort or another. In ecuador, the mass media were used. In Cambodia, program rules carefully specify that information posters will be placed in all pertinent schools; on the commune council notice board; and in the health center, market, and pagoda. Furthermore, to ensure that out-of-school students hear of the program, school officials are directed to contact children who finished sixth grade in the last 2 years but did not go on to lower secondary school. even after outreach, some households will decide not to participate because of stigma or because the benefits do not seem worth the transaction costs implied. there is not much systematic evidence on stigma and CCt programs. adato (2004) concludes from qualitative studies in mexico and nicaragua that the issue was not one of stigma for beneficiaries but one of envy of them by nonbeneficiaries. In nicaragua, some communities even went so far as to provide school supplies to nonbeneficiary children because they felt stigmatized for not having what the programs provided. the general impression among many people in the CCt community is that stigma is not much of an issue, or at least is a lesser issue than it is for social assistance in, say, the United states or some european countries. It would seem that the notion of coresponsibilities helps households and the general public feel that the program beneficiaries are behaving in desirable ways and merit support. as is true for other social programs, transaction costs are a concern— and more so where benefits are relatively small. Indeed, it is largely to minimize transaction costs to both participants and program budgets that many programs pay benefits only every two months when the logic of the program would imply that regular small flows of cash each month 78

DesIgn anD ImplementatIon FeatUres oF CCt programs

would be most helpful in supporting food expenditures, defraying the small regular costs of school attendance, and substituting for reduced child earnings. Despite that, we know that transaction costs can be high relative to the transfer for at least a subset of participants. In ecuador, for example, residents of some isolated communities in amazonia, páramo, or remote coastal areas can face very high costs for transportation (as high as $50–$480 by airplane, $10–$50 by motorboat or panga; or they require a walk of one or more days’ duration). however, because beneficiaries allow transfers to accumulate before collecting them, and combine trips to program payment points with other activities they carry out in urban areas, the costs of collecting the transfer are generally much lower—perhaps as low as $0.25–$0.50 per month, on average, including in the amazonia region (Carrillo and ponce 2008).5 In Bangladesh, the benefit level is very low, so many people fail to participate in the program.6 a comparison of the coverage of CCt-like programs in two countries, Bangladesh and Cambodia, helps illustrate the role that targeting practice can have in determining who is reached by a program. Both programs have several elements in common: they are in poor countries where administrative capacity is low, probably lower in Cambodia than in Bangladesh. Both programs have only education conditions and are limited to girls in secondary school. as such, both start with a challenge because base enrollment rates in upper primary grades are low among the poor. nevertheless, because of careful geographic and proxy means targeting, Cambodia was able to concentrate transfers among the poor. simple calculations suggest that approximately 70 percent of benefits reached households in the poorest quintile of per capita consumption (minus transfers), and less than 5 percent reached households in the richest quintile. In Bangladesh, the program operates nationally, except in the four largest cities, and is targeted only by gender. In practice, however, the incidence of the program has been regressive largely because the base enrollment is higher among the less poor. simple calculations suggest that less than 10 percent of benefits reached households in the poorest quintile of per capita consumption (minus transfers), and approximately 35 percent reached households in the richest quintile. It is important to recognize that targeting results seen to date are not inherent to the design of CCts, but reflect the political will and technical effort made in the programs we examined. although it is impossible 79

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

to say whether that will continue over time, we note that a number of countries continue to refine and improve the implementation and use of their proxy means tests and poverty maps, the technical tools that drive the results. several countries have carried out recertification processes to remove from the list of beneficiaries those who have prospered in the interim. We cannot quantify the effect of these changes because measures of comparable targeting outcomes over time are scarce. In many countries, the programs have been rolling out from year to year, covering areas of different inherent poverty characteristics, so measures across time would not be fully comparable. even in countries with nationwide programs over several years, we would need information that straddles recertification periods. Deterioration of targeting outcomes could be expected within a single certification period, such as is observed but not statistically significant for Jamaica. the latin american CCt programs (which are the majority of the programs with known targeting results) have a fairly similar experience in using a combination of geographic targeting and proxy means testing and in devoting considerable effort to implementing these targeting systems well. many of the countries with well-established programs have had time to improve and refine their targeting systems. In some of the countries with newer programs, improvements are likely to be needed. as the range of countries running CCt programs diversifies, we would expect their targeting mechanisms, and possibly the outcomes from them, to diversify as well. some countries may choose universalism over targeting, as Bolivia has done in the Juancito pinto program for all first-grade students. eastern european countries that already have established means testing systems may use those; communitybased targeting may play a larger role in africa and asia than in latin america. moreover, the results that reasonably can be achieved will vary, depending on such context and design features as the range of ages covered by the program.

Benefit Systems
Benefit systems have a number of aspects that can influence outcomes. here we describe some of the main features of CCt programs, especially the structure and level of payments, the payee, and the payment mechanism. 80

DesIgn anD ImplementatIon FeatUres oF CCt programs

Benefit Structures CCt programs often differentiate payments by the number of children in the eligible age range, but otherwise have rather simple benefit structures—only two differentiate by poverty level, none by cost of living,7 and few by age/grade or gender of the student. those simple payment structures streamline administrative systems and greatly facilitate communication and community understanding of the programs. But they represent something of a missed opportunity in terms of fine-tuning the impact on poverty for a given budget and possibly on best leveraging changes in human capital (a theme taken up in chapter 6). In most of the CCt programs, benefits depend directly on the number of children in the household. that is consistent with an underlying logic of the program that recognizes that each child needs to receive health and education services and that there are costs (explicit and implicit) in getting those services. however, a number of programs have capped the number of children who can be covered. Bolsa Família caps the benefit at r$45, equivalent to having three children in the program. mexico caps the benefit at about $153, an amount that roughly corresponds with two children in primary school and one in high school. In the Dominican republic, the maximum benefit amount is $19 for four or more children ($9 for one or two, $14 for three children). a few programs, such as the ones in ecuador, el salvador, panama, and peru, pay a flat benefit per household, irrespective of the number of children. paying such a flat benefit can be done as a way to ration benefits among families when the program budget cannot cover all who are poor, to counter any incentive to increased fertility,8 or if the program logic is that households need an incentive to learn a new behavior but do not need one to practice it with each successive child. Benefits also can be differentiated by grade or by gender. oportunidades in mexico, Familias en acción in Colombia, the social risk mitigation project (srmp) in turkey, and recently Jamaica’s program of advancement through health and education (path) pay higher amounts for children in secondary school than for children in primary school as a way of recognizing that the opportunity cost of the time of older students is higher than of younger students; often, explicit costs of secondary schooling are higher as well because schools are more distant and textbooks more expensive. oportunidades and the srmp pay higher benefits for girls in recognition that they have been disadvantaged in enrollment. the Bangladesh and Cambodia girls’ 81

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

scholarship programs originally were designed to benefit girls only, although Bangladesh is designing a reform to include poor boys and Cambodia now covers both sexes. Jamaica recently decided to make higher payments for boys in secondary school because boys have lower enrollment and schooling outcomes. many programs pay bimonthly or less often to economize on transaction costs for the program and for the beneficiary. sometimes there are no payments for the months when school is not in session; in other cases, payments continue throughout the year; and in yet other instances, a payment is timed before the school year to enable households to pay for uniforms, shoes, textbooks, and any fees. though lately there is much talk of moving to rewarding performance rather than attendance only, Bogotá’s subsidio Condicionado a la asistencia escolar (sCae) program alone gives bonuses at the end of the school year. Payee In most of the programs, the payee is the parent rather than the student; exceptions are mostly in secondary scholarship programs, especially those in asia and in the sCae program in Bogotá. In nearly all programs where the adult is the payee, payments are made to the mother of the children, a feature that may be important, as we shall see in subsequent chapters. Payment Systems the range of payment systems used in CCt programs covers the full gamut of possibilities. In Brazil, payments are made on debit cards and cash can be withdrawn at banks, atm machines, or lottery sales points. In turkey, payments are made through the state bank, in cash, with clients going to tellers to withdraw funds. In mexico, a fairly low-tech “Brinks truck” model is still the main payment modality, though payments through banks are being introduced in urban areas. households are paid in cash at temporary pay points that use available infrastructure (such as community centers), with transportation and payment of the money contracted to the mexican post and telegraph office. In Kenya, payments are made through the post office, but a pilot program being set up is considering paying via cell phone systems. and at the opposite end of the scale, a pilot program in tanzania 82

DesIgn anD ImplementatIon FeatUres oF CCt programs

will disburse funds to community representatives who will make the payments. there is diversity in how countries ensure that payments are made in full and understood by the client. most of the latin american programs work through the banking system, with the full panoply of audits that implies. mexico gives each household pay statements that show details of payments for each different member/set of conditions and for whom any payments were suspended. Cambodia pays in cash, with payments made every quarter at ceremonies that celebrate and encourage the students’ enrollment and academic status and bring an element of transparency and community monitoring to the payment system. Payment Levels one of the most important features of the payment structure is, of course, its level. Capturing this succinctly is complicated because of the differentiation of payments by number of children and other pertinent factors, differences in context, and the targeting of the programs. Data from household surveys enable us to summarize into a single number or two each program’s level of generosity. table 3.2 presents the share of recipient welfare that the transfers represent for the population of recipients. For a subset of the programs we also present the share of the transfer and for the poorer among them (defined in this exercise as those whose pre-transfer per capita expenditure is less than the 25 percentile of national distribution of the pre-transfer per capita expenditure or income). as table 3.2 reveals, there is significant variation in the generosity of CCt programs, from about 1 percent of pretransfer household expenditures in Bangladesh to 29 percent in nicaragua. It is also encouraging that the generosity of the programs is slightly higher for the poorer beneficiaries—especially in the case of mexico where the share of program transfer is estimated to be 33 percent of the pretransfer level of household consumption among households in the bottom quartile. as is documented in the next chapter, the combination of the generosity of the oportunidades transfers and the program’s high coverage of the poor resulted in a significant impact on poverty measures at the national level. as a means of summarizing the extent to which CCt programs concentrate their benefits on the poorer segments of the population, figure 3.2 presents the proportion of various programs’ transfer budgets received by 83

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

Table 3.2

Generosity of CCT Programs, Various Years
Transfer as share of pretransfer consumption among all beneficiaries (%)a 0.6 6.1 17.0 6.0 7.0 8.2 2004b 21.8 29.3 Transfer as share of pretransfer consumption among poord beneficiaries (%) 0.8 11.7 8.3 10.7 33.4

Country/Program/Year Bangladesh: Fssap, 2000b Brazil: BF, 2006b Colombia: Familias en acción, 2002c ecuador: BDh, Jamaica: path, nicaragua: rps, 2006b 2000c 2004b 2000c honduras: praF,

mexico: oportunidades,

Country/Program/Year Bangladesh: Fssap, 2000b Brazil: BF, 2006b 2006b 2004b ecuador: BDh, Jamaica: path,

mexico: oportunidades, 2004b

Source: authors’ calculations. Note: BDh = Bono de Desarrollo humano; BF = Bolsa Familia; Fssap = Female secondary school assistance program; path = program of advancement through health and education; praF = programa de asignacion Familiar; rps = red de protección social. a. the transfer amounts as a proportion of per capita expenditures (or consumption) are not the same across all tables in the report because of differences in the surveys used, including their coverage and year. b. the measure of welfare used for Brazil is pretransfer per capita income (pCI). For the remaining countries, the measure of welfare is pretransfer per capita expenditure (pCe). pretransfer pCI or pCe is constructed by subtracting the value of the transfer per capita received from either pCe or pCI. the numbers reported are the median shares derived by first removing extreme outliers at both ends of the national distribution of pCe or pCI (that is, dropping households with pCe or pCI below the 1st percentile and above the 99th percentile of the national distribution). c. the number reported is the share of consumption for the median control household. d. a poor beneficiary is one whose pretransfer pCe is less than the 25th percentile of national distribution of the pretransfer pCe.

each decile of the welfare distribution. except for Bangladesh, the CCt programs for which we have targeting outcomes have sharply progressive incidence, with much higher shares of benefits going to the poorest households than to the upper end of the distribution. among the big, well-known programs, mexico delivers more than 45 percent of benefits to the poorest decile. next in line are Chile and Jamaica with approximately 35–40 percent of benefits to their poorest deciles. 84

DesIgn anD ImplementatIon FeatUres oF CCt programs

Figure 3.2
80 70 Proportion of households receiving transfers 60 50 40 30 20 10 0

Benefit Incidence of CCT Programs, Various Years
Brazil BF 2006 Chile Solidario 2003 Chile SUF 2003 Ecuador BDH 2006 Honduras PRAF 2004/5 Mexico Oportunidades 2004 Jamaica PATH 2004 Cambodia JFPR 2003 Bangladesh FSSAP 2000

1

2

3

4

5

6

7

8

9

10

Per capita consumption minus transfer (deciles) Source: authors’ calculations based on the following surveys: ecuador’s encuesta de Condiciones de vida 2006; mexico’s encuesta nacional de Ingreso-gasto de los hogares 2004; Brazil’s pesquisa nacional por amostra de Domicílios 2006; Jamaica’s survey of living Conditions 2004; Chile’s encuesta de Caracterización socioeconómica nacional 2003; honduras’s encuesta de hogares de propósitos multiples 2004; Bangladesh’s household Income and expenditure survey 2000; and the Cambodia Japanese Fund for poverty reduction application form 2003. Note: BDh = Bono de Desarrollo humano; BF = Bolsa Família; Fssap = Female secondary school assistance program; JFpr = Japanese Fund for poverty reduction; path = program of advancement through health and education; praF = programa de asignación Familiar; sUF = subsidio Unitario Familiar. For Brazil and Chile, the deciles are based on per capita income minus transfer per capita. For the remaining countries, the measure of welfare used is per capita expenditure net of the transfer per capita.

therefore, it is not surprising that CCt programs have been recognized for their success in both reaching the poor and concentrating benefits among them. although there are serious difficulties in properly measuring the incidence of transfers (see box 3.2), naïve comparisons of the latin american CCt programs with other transfer programs suggest that CCt programs do a better job of concentrating benefits among the poorest (see lindert, skoufias, and shapiro [2006], figure 11, p. 71).9 looking forward, the agenda with respect to benefit systems will be more focused on benefit level and structure issues than on the payment 85

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

mechanisms. establishing payment mechanisms is a key effort, but one that countries generally have found easier to accomplish than developing sound targeting systems or monitoring beneficiaries’ compliance with co-responsibilities. refinements in payment mechanisms to reduce costs, increase convenience, or better ensure accountability will continue, but the basic issues already have been resolved in most countries. Where policy attention is focused is on the issues of how much to pay; whether to customize payments further by age, grade, household composition, poverty, and cost of living; and whether and how to pay for performance as well as attendance.

Conditions: Their Definition, Compliance Monitoring, and Enforcement
CCt programs vary somewhat with respect to the design of their conditions, and more so with respect to their enforcement of those conditions. monitoring compliance with conditions is a complex task. It involves a variety of actors inside and outside the CCt program, requires the collection and processing of large amounts of information, and needs to happen in a timely manner for the conditions to have any meaningful link with the transfer payment the beneficiaries receive. Just as the role of the CCt program in social policy and its basic characteristics (such as targeting method, benefit amount, and payment system) differ from program to program, so do the mechanisms used to monitor beneficiaries’ compliance with the conditions and the degree to which those conditions are enforced. Definition of Conditions many programs condition the transfer both on enrollment and regular attendance of the households’ children in school and on regular health center visits for the younger children and often the pregnant women. the conditions in programs that promote education can be defined for certain age brackets (Dominican republic, Jamaica) or grades (Cambodia, el salvador). almost all CCts require enrollment and attendance on 80 or 85 percent of school days; Bangladesh’s Fssap is the exception, requiring only 75 percent attendance. a few programs condition on some aspect of performance: Cambodia, for example, 86

DesIgn anD ImplementatIon FeatUres oF CCt programs

requires passing grades; turkey allows a grade to be repeated only once; and nicaragua required grade promotion at year’s end. In programs defining conditions by the use of health services, conditions tend to apply to children aged 0 to 5 or 6 years, with the age set to allow continuous eligibility from birth to schooling, assuming “ontime” enrollment. In about half the cases with child health conditions, there also are conditions for pregnant women and/or lactating mothers. less often are there conditions for adult health care more generally, although such conditions are present and well enforced in mexico and are present albeit less enforced in Jamaica. Child health conditions are formulated in various ways, requiring children to have complete immunizations (Brazil) or to adhere to a schedule of regular health center visits for health checkups. In some countries, the kind of health services that mothers and children should receive are defined in great detail (Jamaica), but other countries mandate only that they regularly go to the health center (honduras). growth monitoring is required two to six times a year in most programs with child health conditions. health and nutrition education sessions are a feature of many, but not all, latin american programs and rarely are used elsewhere (for details, see the country at-a-glance tables in appendix a). Indeed, all latin american programs have health conditions of some kind, whereas such conditions are much less common in active and planned programs in africa and south asia. though malnutrition and immunization coverage are more problematic in those regions, services are more limited, so the programs have not focused on health conditions. some programs allow exceptions or exemptions to the conditions they impose. most common is justification for absence during a specified reporting period on grounds of illness. Jamaica waives attendance requirements for children who are disabled and deemed unlikely to benefit from attending school (mont 2006). Kenya waives attendance requirements for children who do not have access to schools or clinics (government of Kenya 2006b). Timing of Compliance Verification the frequency of verifying compliance with conditions varies widely (see table 3.3). Frequency ranges from monthly (turkey) to every four months (honduras) or even once a year (Chile sUF). verification frequency depends, in part, on the type of conditions a program imposes: 87

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

Table 3.3

Country and Program Variations in Monitoring and Penalties for Noncompliance with Conditions
No penalties ecuador pakistan: pesrp Chile: sUF none Light penalties none Bangladesh: Fssap Cambodia: Cessp and JFpr Brazil: Bolsa alimentação, Bolsa escola, Bolsa Família, and petI Chile: Chile solidario honduras: praF Kenya: Ct-ovC pakistan: Csp Colombia: Familias en acción and sCae-Bogotá el salvador: red solidaria Jamaica: path mexico: oportunidades nicaragua: atención a Crisis and rps turkey: srmp Full penalties none none

Degree of monitoring no monitoring light monitoring (annually) Full monitoring (monthly, bimonthly, or quarterly)

Source: authors’ compilation. Note: Cessp = Cambodia education sector support project; Csp = Child support program; Ct-ovC = Cash transfer for orphans and vulnerable Children; Fssap = Female secondary school assistance program; JFpr = Japan Fund for poverty reduction; path = program of advancement through health and education; pesrp = punjab education sector reform program; petI = programa de erradicação do trabalho Infantil; praF = programa de asignación Familiar; rps = red de protección social; sCae = subsidio Condicionado a la asistencia escolar; srmp = social risk mitigation project; sUF = subsidio Unitario Familiar. light penalties = warnings prior to penalties and/or delay or minor reduction in individual benefit and/or irregular application of reductions; full penalties = complete withdrawal of period’s benefit for noncompliant person in payment period immediately following detection of noncompliance.

if beneficiaries are required to attend one training session a year, compliance verification can be determined only annually. Consequently, conditions related to health or awareness raising tend to be verified at longer intervals than do those for school attendance. another factor in determining the frequency of verification is capacity constraints. given the amount of information and number of transactions involved, smaller-scale programs or programs in lowcapacity settings may opt for longer intervals. scholarship programs in Bangladesh and Cambodia have done so. But even if compliance verification is designed to be done more frequently, capacity constraints may yield delays in sanctioning noncompliance. even in a high-capacity environment such as mexico, the benefit amount paid in the period January/February reflects the beneficiary household’s compliance or noncompliance in the period september/october of the preceding year. such long lag times between the noncompliance 88

DesIgn anD ImplementatIon FeatUres oF CCt programs

and the reduction of benefit may weaken the positive quid pro quo effect of the program. Sanctions and Enforcement although all CCt programs specify a schedule of sanctions in the case of noncompliance with the stated conditions, both the type of sanctions and the degree of enforcement vary quite substantially from one program to another. most common is a temporary reduction of all or part of the benefit for the first instances of noncompliance, followed by an eventual termination of the benefit for repeated noncompliance. such is the case, for example, in Colombia, Jamaica, and mexico, where the benefit is reduced immediately for the period for which there was noncompliance and the reduction is reflected in the next payment. programs do not always take a hard line on compliance with the conditions they impose because CCt programs generally are targeted to the poorest and most vulnerable groups of the population—those people most in need of assistance. By design, some programs involve a social worker who, in the case of noncompliance, will reach out to the beneficiaries (el salvador, Jamaica). Brazil even goes a step farther: the conditions fundamentally are viewed as encouraging beneficiaries to take up and exercise their right to free education and free health care, so noncompliance is taken to be a manifestation of some kind of obstacle that the family cannot overcome to access the service rather than an unwillingness to comply. a beneficiary who is noncompliant will receive a warning (written notice) of noncompliance for the first period and may be visited by a social worker to see if there is a noncash-related problem to be solved. only on a third consecutive occasion of noncompliance will the benefit be “blocked” for 30 days, after which the full amount, including the amount accumulated during the blockage, will be paid out. perhaps the ultimate “soft conditions” are those in ecuador, where the program was announced as conditional, but effective systems of monitoring compliance and enforcing sanctions have yet to be developed. In Brazil, el salvador, and mexico, compliance with education conditions has tended to be 90 percent or better among enrolled students; in Jamaica, compliance has improved over the years from 70 percent to 85 percent.10 Compliance with health conditions (among those for whom compliance information is available) is in the same range or better for these countries. 89

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

Compliance Verification Mechanisms the process of verifying beneficiaries’ compliance with the conditions generally involves, at the very least, the providers of the services whose use is mandated, the program, the payment agency, and the beneficiaries themselves. Depending on the country context, not only one but various levels of government may be involved, as may be nongovernmental organizations (ngos) or other community organizations helping with either the delivery of the services or the program itself. Information is collected and processed by these various actors in many different ways. In most programs, the education or health service provider collects data on school enrollment and attendance or on health center visits. the provider gives the data either directly to the program or to a central unit in the provider’s line ministry, which then compiles the data and passes them on to the program. In some programs, such as in Colombia, the beneficiaries are involved more heavily by having to get forms filled out and certified by each service provider and then to submit the forms to the program. practices span anything from paper-and-pencil attendance lists kept by teachers and handed over to the program officials on a regular basis (Bangladesh, Cambodia) to optical-scan forms (mexico) to experiments with different types of smart cards (Brazil). enforcement of sanctions for noncompliance demands the timely availability of reliable information, which may be especially problematic in the beginning phases of a program. although about 93 percent of schools in Brazil reported information in 2006, only 55 percent did so in 2004; by 2006, information on compliance with health conditions was only available for about 33 percent of families. Jamaica was able to reduce the number of schools that did not return the beneficiary lists in time from about 10 percent of schools in 2003 to zero by the end of 2005. mexico’s oportunidades, like some mature programs, now has timely information on compliance with all conditions for 96 percent of its beneficiaries (Castañeda 2006), but initially had problems with delays (IFprI 2000). lack of information clearly precludes establishing a meaningful link between conditions and payments through the enforcement of sanctions. there are few studies that soundly address issues of the accuracy of compliance information. although a qualitative study of turkey suggested that education officials were hesitant to report absences, the bigger issue was getting them to fill in the forms at all (Kudat 2006). a series of operational audits of Jamaica’s compliance monitoring has 90

DesIgn anD ImplementatIon FeatUres oF CCt programs

turned up no big discrepancies between attendance records kept in schools and clinics and the information submitted to the path program (government of Jamaica 2006). Unfortunately, little information is available on the costs of monitoring compliance with conditions, partly because some of the costs are borne by health or education sector employees or subsumed in other administrative costs and so are not easily accounted for. the only piece of comparative analysis of cost structures we are aware of is the study of nicaragua (rps pilot), honduras (praF), and mexico (oportunidades) (Caldés, Coady, and maluccio 2006). this study finds that the cost of verification can range from 2 percent to 24 percent of total program administrative costs (excluding transfers) in any given year. those estimates have to be viewed with caution, however, because they very much depend on a program’s stage of implementation (newly introduced versus mature) and on the associated shift in the relative shares of different project activities in the overall costs of the program. another way to estimate the costs of verifying compliance is to work backward. grosh et al. (2008) compile administrative costs for 10 CCt programs, showing a range from 4 percent to 12 percent of total program costs. those costs include running the systems for targeting and payments as well as monitoring compliance, plus all the support services of management, monitoring, and evaluation. If those functions each took the same share of administrative resources, monitoring compliance would be on the order of 1–3 percent of total program resources. It is interesting to note that the median administrative cost for CCt programs in the study is 8 percent; for other types of cash transfers it is 9 percent; and over the whole range of 54 social assistance programs, the median administrative cost is 10 percent (grosh et al. 2008). apparently the scale and generosity of the (mature) CCt programs has been sufficient to temper the extra administrative requirement imposed by monitoring compliance with conditions.

Monitoring and Evaluation a number of CCt programs have had unusually proactive management based on cutting-edge technical systems, especially monitoring and evaluation systems. two features inherent to CCts programs—the 91

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

number of actors involved and the need for extensive information management to verify compliance with conditions—may have interacted in ways that have spurred creative development of monitoring and management. this excellence in systems, the extensive documentation that has resulted from the amount of information available, and the degree of transparency about the information have contributed to the attractiveness of CCt programs, although such features are not inherent to them. all CCt programs need to know how implementation is being carried out across diverse actors and processes, and they have developed various ways to elicit that information, some of which also add incentives for good performance. Colombia’s Familias en acción has used a system of sample-based site monitoring or “spot checks” as internal process evaluation. Interviews are conducted every six months in a sample of 20 municipalities; for participants, program officials, and local governments, interviewers use defined questionnaires that cover 400 indicators of various program aspects (including inscription processes, verification of compliance with conditions, payment systems, appeals, and quality of the health education component) (box 3.4). the results show which aspects of the program are working well; how much variability there is in program management across locations; and where changes in procedures, training, staffing, or other inputs are needed. the program has been good not only at collecting such information, but also in acting on it. program managers detected problems with long queues for payments (including people waiting outdoors in the rain) and worked with banks to find various ways to reduce the queues. they found that some children were not being served continuously between the preschool and school portions of the program because of when their birthdays fell relative to enrollment, so they changed the specifications. and when they identified a number of areas where staff needed more training to carry out the program efficiently, they provided that training. In Brazil, the Bolsa Família program depends to a great extent on work done by the municipalities. after an initial problem with the quality and timeliness of such work, the Bolsa implemented its “index of decentralized management,” which captures the quality of functions performed by municipalities on the household registry, the monitoring of conditions, and all municipalities’ social controls for the program. municipalities receive support for their administrative 92

DesIgn anD ImplementatIon FeatUres oF CCt programs

Box 3.4 Colombia’s Familias en Acción Sample-Based Site Monitoring, Selected Indicators
Knowledge • other’s knowledge of specific themes M • linic and school staffs’ familiarity with their program guide C Use of materials • o mother leaders use materials? D • o school and clinic staff consult their guides? D Compliance with operational procedures • o mother leaders report complaints? D • ow many days do the banks allow for payments; what is the H waiting time? Infrastructure • o regional offices have necessary equipment? D • o mothers stand in line at the bank, inside or outside? D Organization • ercent of schools that require a written excuse for absence P Procedures • ercent of complaints resolved at the Regional Coordination P Unit • ercent of municipalities that have lists of schools, clinics, and P completed stickers
Source: velásquez 2007.

costs, with the payments adjusted to their performance on the index. From the early stages of program development, oportunidades has put in place three structures to monitor program operations and results. the first structure, operating since 1998, generates a set of 64 monitoring and management indicators every two months (the sistemas de Datos personales de oportunidades). a second structure, a survey of beneficiaries and program providers called sentinel points, was implemented in 2000 and produces information on perception of service quality twice a year. third, external experts use monitoring and management data to make regular assessments of program 93

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

operations. all the information and assessments are available to the public on the program Web site. In mexico, compliance with conditions is certified on schedule in 96 percent of cases, payments are made on time in 98 percent of cases (Castañeda 2006), and overall administrative costs exclusive of payment transactions are 3 percent of total program costs (gomez-hermosillo 2006). of course, not all experiences are so positive. there are real logistic and institutional challenges in running CCt programs: turkey’s qualitative evaluation pointed to problems with the management information system in the first couple of years of operation (ahmed et al. 2007), and honduras sometimes is cited as a case where poor implementation has affected the expected impacts. Compliance monitoring has been a challenge in many countries. But the positive cases get more press and are setting new expectations for the administration of social assistance programs. the evaluation culture around CCt programs is quite strong. It is present in a larger share of programs than is excellent monitoring, and it goes well beyond traditional practice in social policy. many programs either have conducted or have plans to conduct impact evaluations with credible counterfactuals. among those programs, a large share of countries have used experimental methods, at least initially. In several countries, the evaluations are neither simple nor one-off. there have been dozens of studies for mexico’s program, and there is a fairly diverse body of evaluation for Colombia and nicaragua. most evaluations have been conducted by agencies external to the program and often external to the government and international partners as well, a practice that enhances the credibility of the evaluation. In the great majority of cases, evaluations have been made public, often posted on program Web sites and/or published in respected academic journals. a number of those evaluations have had a real impact on policy. evaluations showed, for example, that anemia was not declining as expected among oportunidades beneficiaries. that finding led to a series of investigations of causes and the discovery that the bioavailability of iron in the original food supplement was low. moreover, the supplement was shared among family members and thus the target child got less than the intended amount. the supplement was reformulated and the nutrition education component was strengthened (neufeld 2006). In Jamaica, the increase in secondary enrollment was disappointing, so the government has decided to 94

DesIgn anD ImplementatIon FeatUres oF CCt programs

increase the benefits at the secondary level and to differentiate them by grade and gender. this culture of evaluation is spreading not only from one CCt program to another, but also from CCt programs to other programs within the same countries. mexico’s decision to evaluate the early phases of oportunidades in 1997 was, at the time, an unusually dramatic example of evaluation: it was motivated and paid for wholly by the program designers without external pressure, it used an experimental design, it was contracted to an independent third party, and the data collected were made publicly available so that scholars could replicate and extend the work. since then, the notion of good evaluation has spread. appendix B of this report provides a detailed discussion of the technical aspects of CCt impact evaluations. In mexico, a social development law, passed in 2004, requires that all new programs be evaluated, and established a national Council for the evaluation of social Development policy. a separate transparency law mandates that evaluation results be made public. external evaluations are done and the summaries are passed to program managers, who annually must describe to Congress what they are doing in response to the evaluations (hernández 2006). the mexican ministry of social Development, which is responsible for oportunidades and many other programs, has adopted a system of results-based monitoring. It plans to conduct evaluations of five national programs a year in each of three years and to have installed such a system in half its subnational agencies within six years (rubio 2007). Chile and Colombia also have developed significant evaluation cultures, cited as “good practice” in global reviews (for example, mackay 2007). meanwhile, the evaluation agenda remains vital among CCt programs. the large majority of new programs are planning credible evaluations, and several of those have included interesting dimensions not previously assessed (see box 3.5). For countries with programs that have reached their full intended coverage, a range of issues remains, although experimental design is no longer an option for many aspects of evaluation and so there are additional methodological challenges. a noteworthy aspect of the experience of CCt programs has been the fact that lessons learned in one country often have been shared internationally. that sharing has happened to such an extent that an international community of practice has developed as a result (see box 3.6). 95

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

Box 3.5 Evaluation Remains Important in CCTs
DespIte the UnUsUal qUantIty anD qUalIty

of impact evaluations of CCt programs to date, the agenda remains vital and there are quite a number of evaluations ongoing or planned. those evaluations can be grouped into three types: 1. relatively basic evaluations of new programs: many of these will use experimental design. a significant subset is in new contexts, especially low-income countries (Bolivia, Kenya, pakistan, yemen), or has new twists on service delivery (community-driven development approaches being developed in sierra leone and tanzania). 2. evaluations that will help disentangle the role of different parts of the “classic” program in delivering impacts: • ilot programs in Burkina Faso and P morocco have comparative treatment arms for conditional and unconditional transfers. the transfer will be discounted fully in the payment round that immediately follows noncompliance. experiments already under way in Kenya and pakistan test much softer versions of penalties: initial warnings to households rather than penalties, and final penalties amounting to only a small reduction in the transfer and occurring several months after noncompliance. • o far, virtually all CCT programs have S paid the woman of the household. Doing so imposes no additional direct operational burden because a single person must represent the household in any case. however, the uniform payment to women has made it difficult to understand to what extent impacts have come from the release of the budget constraint via the transfer,

changes in behavior due to conditions, or changes in use of household resources due to the payment to women. pilot programs in Burkina Faso, morocco, and yemen test treatment arms for delivering cash to men/ fathers versus women/mothers. • he role of the health/nutrition education T component in CCt programs has not been studied specifically. In panama, the impact of adding community-based nutrition education (the atención Integral de la niñez en la Comunidad model) will be tested. In Indonesia, a less intense variant on providing health education to leaders of beneficiary groups will be tested. 3. evaluations that propose to go beyond the use of services to look at final outcomes: Burkina Faso, Indonesia, morocco, and tanzania look at examination results or cognitive outcomes for school children. nutritional status is measured in Burkina Faso, el salvador, Indonesia, panama, and tanzania. anemia is measured in Burkina Faso, el salvador, and panama. there is still a large evaluation agenda on secondround and long-run impacts. very little is known about how programs impact savings or investment, long-run autonomous incomes of beneficiary households, or spillover effects at the community level in any setting. evidence on labor supply, on migration, and on remittances beyond the few settings in which these have been measured also would be useful, as would knowing what happens to families or children after they leave the program. evaluations on those issues are under way for Chile and mexico; are being considered in Colombia; and are pertinent in many other countries, especially those with programs that have operated for several years.

96

DesIgn anD ImplementatIon FeatUres oF CCt programs

Box 3.6 The International CCT Community of Practice
CCt progr ams have engageD IntensIvely

in learning from each other—often directly, sometimes facilitated by international agencies—and using a full range of modalities: • A series of three global conferences sponsored by international agencies and hosted by local governments have drawn together almost all countries that have had CCt programs active at the time of a conference: 2002 in puebla, mexico; 2004 in são paulo, Brazil; and 2006 in Istanbul, turkey. a series of smaller regional events has reinforced ties among program managers. • There has been a series of study trips, often to the established latin american programs. • A video conference “learning circle,” facilitated by the World Bank, has been formed by five of the established latin american programs. participants also have organized face-to-face sessions: January 2008 in

Cuernavaca, mexico; september 2008 in Cartagena, Colombia. • Extensive publication of evaluation results in academic journals has enabled a wider range of people to join in the learning. • Several programs have very detailed public Web sites that publicize not only basic program information, but also operational manuals, monitoring statistics, evaluation results, program news, and the like. not only is the learning south-south, but it also has taken on south-north dimensions, with officials in london, new york City, and sydney familiarizing themselves with the international CCt experience. the contacts have gone farther with opportunity nyC: officials from the mayor’s office and program management traveled to mexico to visit oportunidades, joined in some of the activities of the latin american learning circle, and freely cited international experience as a justification for their pilot program.

Intersectoral and Interinstitutional Challenges poverty has been recognized as multidimensional for a long time, but most service provision is organized along unidimensional lines. CCt programs, especially in latin america, often are at the heart of a move toward integrating policy and service delivery to achieve greater synergies among policies. the CCt programs sometimes are catalysts of such a move, and sometimes are forerunners in a larger explicit strategy to improve integration. that move toward coordination or integration has two facets—the increased coordination of actors and the integration of benefits. CCts inherently involve coordination among actors across several sectors (social assistance, health, education, planning, finance, auditing) and levels (federal, state, local, community). many countries with a CCt program have at least national coordinating structures, and countries 97

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

where service provision is local usually have some sort of local coordinating structures as well. For example, mexico’s national coordinating council, composed of senior policy makers, meets twice a year; the technical coordinating committee, comprising senior program managers, meets bimonthly, as do the state coordinating committees with local coordination around the 11,000 service centers (known as the mesas de atención). those structures are concerned principally with operational details of ensuring that program beneficiaries meet conditions, that the information to monitor compliance flows appropriately, and that payments are made on time and correctly. But in handling those concerns, larger issues of the provision of at least health and education services arise. Because most countries with CCt programs also have taken simultaneous action to increase the quantity or quality of education and/ or health services, the points of contact on CCt operations can add a dimension of concrete management interface to such initiatives. (that will be discussed more fully in chapter 6.) some CCt program managers feel that regularly presenting information about service availability (and the shortcomings thereof in specific locations) to local line staff and senior ministerial representatives exercises pressure for improved services, whether or not there is a comprehensive plan to do so. In a number of cases, the CCt program has given the national government a new tool to affect the delivery of decentralized services. In Brazil, for example, the ministry of social Development signed a “joint management agreement” with each municipality that specifies who is responsible for what part of the Bolsa Família’s administration. It also requires that states and municipalities give Bolsa participants priority in the wide range of locally run and funded programs. It thereby induces all these programs to share, at least partly, a targeting system and eligibility threshold. In Colombia, participating municipalities are required to appoint liaison officers who oversee all the functions the municipality is responsible for in the program and who communicate with the national and regional program offices. De facto, however, a good deal of what those municipal liaison officers do is coordinate among different social services supplied by the municipality. moreover, the Familias en acción program only works in municipalities that meet defined service standards for health and education. Initially, some target municipalities did not meet those standards but, over time, all did. thus the national program has resulted in better coordination within municipal governments and has improved service delivery in some. 98

DesIgn anD ImplementatIon FeatUres oF CCt programs

Financing is a potent tool. In el salvador, the red solidaria has a budget to increase health services through contracting with ngos. the national education budget goes to local schools on a capitated basis, so as local enrollments increase in response to the demand-side transfer, the supply-side budget will increase in step. Chile solidario has an even stronger tool than most programs: the portion of the budget needed to provide priority services to its clients passes through the ministry of planning, which releases it only to the ministries in exchange for services. though hard to quantify or document, there is a sense in the CCt community that in at least a number of countries, the avenues of coordination established for the programs not only led to addressing the issues directly involved with the CCt program, but also have facilitated collaborative identification and solution of problems beyond that. among the challenges currently felt in countries with mature and successful CCt programs are the needs to establish more clarity about the role of the CCt program versus other instruments and to pursue greater integration of social policy. In the countries with the most explicit instruments for meeting those challenges (Chile, Colombia, el salvador), the leadership has come from a central agency under the presidency or ministry of planning rather than from within the CCt program itself. one way of integrating benefits is geographic. In panama, when the government developed the red de oportunidades, it also strengthened the social Cabinet in the ministry of economy and Finance and created a multisector committee of the red de oportunidades to ensure that the supply of education and health prioritizes the areas where the red will work. similar approaches to giving priority in health and education improvements to areas in which CCt programs concentrate is not unusual. In some cases, the coordination has gone beyond the minimum level inherent in a CCt program. In el salvador, for example, the red solidaria is focused on the poorest 100 municipalities. It includes the CCt program itself, a program to improve basic infrastructure services in the same municipalities, and a program to provide small-scale productive projects and microcredit. the red then involves coordination on a greater range of subjects and from a wider range of actors, including those in agriculture, electricity, water, and sanitation. another place to integrate benefits is directly at the household level. as mentioned earlier, the household targeting systems used to select CCt beneficiaries often are used to target the benefits of other programs as well. this means that separate programs may reach the same 99

ConDItIonal Cash transFers: reDUCIng present anD FUtUre poverty

households. In Jamaica, the beneficiaries of path are eligible not only for the cash transfer but also for fee waivers for hospital services and pharmaceuticals, for secondary school tuition, and for textbook rentals. In Brazil, Colombia, and ecuador, the scope of programs that use the same household targeting system as the CCt program is much wider, encompassing other cash transfers, housing, school-to-work transition for youth, adult education, and so forth. In some countries, the integration goes beyond a common eligibility threshold in separately administered programs to specifically providing links or referrals. several countries (Chile, Dominican republic, and el salvador, among them) help households get their birth, adoption, marriage, or identification papers in order—and that has spillover benefits in improving access to many other government programs, to voting rights, and sometimes to private banking services. In a decentralized context, Brazil’s Bolsa Família has encouraged municipalities to use social workers to bring additional support and diagnostics to households where children fail to meet their co-responsibilities. Jamaica is beginning to set up a referral system (including a pilot program with one-stop shops) between path beneficiaries and a range of other programs, including skills and job-readiness training, job matching, business development, and caregiving support. the Chile solidario program is focused on linking poor households into all pertinent branches of social policy. Colombia is rolling out a similar program, Juntos, that will link beneficiaries of Familias en acción into a similarly large range of services. one of the basic attractions of CCt programs is the potential synergy of getting health, education, and social assistance to the same families. realizing such synergy at the operational level has been a basic challenge in all countries. the kernels of success often have motivated further waves of ambition toward more fully coordinating the actors in social policy and the benefits they can deliver to households. thus, for the foreseeable future, coordination will remain a core issue of CCt work in the full range of countries, from those whose programs are nascent to those whose programs are mature.

Conclusion
CCt programs have been at the forefront of the modernization of social assistance. many of the CCt programs have been ambitious 100

DesIgn anD ImplementatIon FeatUres oF CCt programs

and in some cases innovative in their mechanics. most have chosen to focus narrowly on the poorest and have used (sometimes creating or refining) geographic and household targeting methods effectively to make possible a good record with respect to incidence. the larger programs also have a good record on coverage. payment systems have been developed, and in most cases the systems function extremely well in getting payments reliably to the right people in the right amount at the right time. also, there has been attention to keeping transaction costs reasonably low for participants. the banking sector or other payment agencies have been used to quite good effect in many countries. the monitoring of compliance with conditionalities has required the development of extensive and rapid information flows among numerous actors. although monitoring compliance is still the least developed aspect of the programs in several countries, the mere fact (even the expectation) of routine exchange of large volumes of data is remarkable and a departure from common practice 10 years ago or in many other kinds of social programs today. CCt programs require the coordinating actions of many parties—the program itself, the providers of health and education services, the payment agency, and often subnational governments at one or more levels. this has reinforced the need for information sharing and led to wider discussions and actions on integrating different parts of social policy. CCt programs have been very self-critical and open to learning. as a whole, they have a remarkable record for the extent and rigor of impact evaluation and public dissemination of findings. In most cases this is matched by extensive continuous monitoring systems that enable program managers to ensure that implementation is going as planned and to make adjustments as needed. there has been a great deal of exchange among programs as they seek to learn from each other how best to handle design issues and operational challenges. the success of CCt programs documented in subsequent chapters could not have been achieved without the reliable implementation that their systems of targeting, payment, and monitoring and evaluation have delivered. however, none of these are inherent in, or limited to, CCt programs. programs of other genres may learn from the implementation experience of CCts. and CCt programs, both new and established, must be cognizant that success, or continued success, is not automatic, but dependent on excellence in their basic systems.

101

Chapter four

the Impact of CCts on Consumption poverty and employment
We saW In Chapter 3 that CCt programs generally have done

well in targeting their transfers to the poor. that does not mean, however, that they necessarily have a large impact on poverty. a number of factors, including behavioral and political economy responses to targeted programs, intervene in determining the ultimate impacts on poverty. for example, a study of China’s di Bao program—the largest cash transfer program in the developing world, though not a conventional CCt—found that the cities of China where the program was better targeted to the poor generally were not the ones where the scheme had the highest impact on poverty or where the program was the most cost effective in reducing poverty (ravallion 2008). this chapter directly assesses the performance of existing CCt programs in reducing consumption poverty. the chapter is divided into three sections and a conclusion. In the first section, we consider the impact of CCts on short-term consumption and consumption poverty. this is done both for the target populations of CCt programs and, for a few countries, for the country population as a whole. We also discuss evidence showing that transfer income is used differently from other sources of income. In principle, the impact of CCts on poverty could be smaller than would be suggested by simple back-of-the-envelope calculations based on the size of the transfer because of both intended and unintended effects of the program. the second section of the chapter discusses the evidence on behavioral changes that could offset the impact of transfers. as will be shown in chapter 5, there is solid evidence that CCts have increased school enrollment levels. If schooling and child work are substitutes, at least in part, then we would expect that CCts might reduce child labor—and therefore reduce the contribution that 103

CondItIonal Cash transfers: reduCIng present and future poverty

children make to household income. We thus begin that section of the chapter with a discussion of CCt impacts on child labor. CCts also could reduce adult labor supply for a variety of reasons: leisure is likely to be a normal good so households will tend to consume more of it as their incomes rise, and households could adjust their labor supply in an attempt to stay “poor enough” to continue being eligible for transfers. for that reason we next review the evidence on program effects on work by adults. finally, we discuss whether CCts appear to have crowded out transfers from other sources, had unintended impacts on fertility, or have had (local-level) general equilibrium effects. If part of the transfer is invested, or if the transfer enables households to better smooth consumption, then CCt programs also can have impacts on consumption in the long run, above and beyond the changes arising from human capital accumulation. the third section of the chapter provides some evidence that this indeed has been the case. (Impacts on human capital accumulation will be discussed in chapter 5.)

Impact of CCTs on Household Consumption and Poverty
Impacts on Consumption among Program Beneficiaries the impact of CCts on immediate consumption is an important determinant of poverty alleviation in the short run, especially because most beneficiaries belong to the poorest part of the population. In this section we assess the impact of CCts on short-term consumption or income for seven programs in which such data were collected as part of their evaluations and in which robust methods can be applied in the estimation of impact, namely, Bolsa alimentação in Brazil,1 familias en acción in Colombia, praf in honduras, oportunidades in mexico, the rps in nicaragua, the Bdh in ecuador, and the Cessp scholarship program in Cambodia.2 In all programs, consumption or income data were obtained through field surveys that interviewed both beneficiary and control households. except for Brazil’s Bolsa alimentação, mexico’s oportunidades, and the Cessp, all evaluations had baseline surveys that can be used to measure averages before the programs were implemented.3 table 4.1 shows that preprogram median per capita consumption levels for the target population were low in all programs. this finding 104

Table 4.1 Impact of CCTs on per Capita Consumption, Various Years
Colombia 2002 0.85 1.19 1.12 1.13 0.79 0.68 0.59 0.58 0.59 0.63 0.53 2006 2003 2005 2000 2002 1998 Jun. 1999 Oct. 1999 2000 2001 Ecuador Honduras Mexico Nicaragua 2002 0.52

Consumption 0.83 0.89

Brazil 2002

Cambodia 2007

median daily per capita consumption of control households (current us$) 0.06 0.02 0.12 0.13 0.08 0.08 0.06 0.06 0.12 0.14 0.13 0.16

daily per capita transfer (current us$) 8 2–3 17 13 8 7 9 11 21 20

0.15

0.15

ratio of transfer to consumption (%)a 7.0** B a 10.0** a B a 7.0* B

19

29

31

30

Impact on per capita consumption for the median household (%)

7.8**

8.3**

a

29.3** 20.6**

the ImpaCt of CCts on ConsumptIon poverty and employment

Source: authors’ calculations for all countries in the table except Colombia. for Colombia, see Institute for fiscal studies, econometría, and sistemas especializados de Información (2006). Note: the estimated impacts presented here are not always equal to the unconditional double difference estimates because some regressions control for other correlates. the impact for honduras was obtained from 2002 regression only. the impacts for mexico are all for single equation cross-sectional regressions for each year. the lack of impact in 1998 is likely the result of the fact that this survey was carried out just a few months after the start of the program. figures are in us$ obtained through the official exchange rates observed at the time of the surveys. In the case of oportunidades in mexico, the 1998 figures are for a few months after the start of the program. In the case of Bolsa alimentação in Brazil, per capita consumption figures are for more than a year after the start of the program. a. the transfer amounts as a proportion of per capita expenditures (or consumption) are not the same across all tables in the report because of differences in the surveys used, including their coverage and year. a. Baseline, before households in CCt treatment group received transfers. B. no significant impact on consumption. * significant at the 10 percent level. ** significant at the 5 percent level.

105

CondItIonal Cash transfers: reduCIng present and future poverty

corroborates the findings of chapter 3 that CCts were well targeted. per capita consumption varied between $0.52 per day in nicaragua and $1.19 per day in Colombia. per capita transfers for the median household varied more widely across countries. they were as low as $0.02 per day in Cambodia, and as high as $0.16 per day in nicaragua. this heterogeneity reflects the different weights that each program assigned to reducing short-term versus long-term poverty. reducing current consumption poverty was a central objective of oportunidades and the rps. By contrast, the Cessp program had no redistributive or poverty alleviation goals. Because the size of the transfer varies a great deal across countries, so does the ratio of the transfer to median consumption. this difference can be seen in the third row of the table: for households in nicaragua, the transfer represented about 30 percent of consumption, whereas in Cambodia that number is only about 2 percent. other programs fall somewhere in between, with familias en acción and oportunidades making relatively large transfers, compared with the smaller transfers for the Bdh program in ecuador, pr af in honduras, Bolsa alimentação in Brazil, and especially the Cessp program in Cambodia. the fourth row of the table summarizes program effects on consumption. the largest impacts are found for the rps, the program that made the largest transfers. other programs included in the table (including familias en acción in Colombia, oportunidades in mexico, praf in honduras, and Bolsa alimentação in Brazil) also had significant impacts on per capita consumption, ranging from 7 to 10 percent.4 By contrast, neither the Bdh program in ecuador nor the Cessp program in Cambodia appears to have increased consumption levels. the results for the Cessp program are not unexpected, given the small size of the transfer and the fact that short-term poverty alleviation was not a program goal. the results for ecuador are more surprising, and they appear to be related to the large reduction in child labor among Bdh program beneficiaries (a point that we discuss in more detail below). the estimated impact on consumption for the median households tells us very little about the potential distributional effects of CCts. therefore we next consider impacts on various poverty measures, including those that are distributionally sensitive. 106

the ImpaCt of CCts on ConsumptIon poverty and employment

The Impact of CCTs on Poverty at the Program Level We estimate program impacts on three poverty measures of the fostergreer-thorbecke (fgt) family: the headcount index, which is the number of people below the poverty line; the poverty gap, which measures the average distance between the consumption of poor people and the poverty line; and the squared poverty gap, which takes into account the distribution of resources among the poor. the analysis in this section focuses on Colombia, honduras, mexico, and nicaragua. We exclude Cambodia and ecuador; in those countries, the CCt did not have an effect on median consumption and, as would be expected, did not reduce poverty. We also exclude the Brazilian Bolsa alimentação program because the evaluation sample is not representative of the program’s target population, which makes the analysis of the impact on poverty less informative. the results from these calculations are summarized in table 4.2. Consistent with table 4.1, programs that had large effects on consumption also had large effects on poverty. In nicaragua, the rps reduced the headcount index among beneficiaries by 5–7 percentage points, the poverty gap by 9–13 points, and the squared poverty gap by 9–12 points. In Colombia, familias en acción also had sizable effects on poverty, especially on the poverty gap, which was reduced by almost 7 percentage points. praf in honduras and oportunidades in mexico had more modest impacts on poverty.5 another way to measure the impact of CCts on welfare is to compare the cumulative distribution of consumption per capita between the treatment and control populations. this method has the advantage of not relying on the selection of a poverty line, which can be somewhat arbitrary. If the cumulative distribution for treated households lies completely to the right of the distribution for control households—socalled first-order stochastic dominance—current welfare is improved unambiguously by CCts. this is clearly the case for rps beneficiaries in nicaragua, as shown in panel a of figure 4.1. panel B shows an improvement that is much smaller for honduras, a result that is not surprising given the smaller magnitude of the transfer. The Impact of CCTs on Poverty at the National Level the welfare effects of CCt programs discussed so far are based on the sample of households in the impact evaluation surveys. that is, we have assessed impacts on those households and individuals directly 107

108
Colombia 2002 Control Impact Control Impact Control Impact 0.53 a 0.43 –0.02** 0.30 a 0.36 –0.02* 0.28 B 0.35 –0.03** 0.58 a 0.54 –0.07** 0.49 a 0.54 –0.02* 0.47 0.01* 0.55 –0.03** 0.95 a 0.90 –0.03* 0.88 a 0.91 B 0.89 0.02** 0.93 –0.01** 0.94 0.00 0.56 –0.02** 0.36 –0.03** 2006 2000 2002 1998 Jun. 1999 Oct. 1999 Honduras Mexico 2000 0.84 a 0.43 a 0.26 a Nicaragua 2001 0.91 –0.07** 0.50 –0.13** 0.32 –0.12** 2002 0.90 –0.05** 0.50 –0.09** 0.32 –0.09**

Table 4.2

Impact of CCTs on Poverty Measures, Various Years

Poverty measure

headcount index

poverty gap

CondItIonal Cash transfers: reduCIng present and future poverty

squared poverty gap

Source: authors’ calculations. Note: We exclude Cambodia and ecuador from this table because the CCt did not have an effect on median consumption in those countries and so it is not surprising that it did not reduce poverty. We also exclude the Brazilian Bolsa alimentação program because the evaluation sample is not representative of the program’s target population, which makes the analysis of the impact on poverty less informative. for honduras, mexico, and nicaragua, calculations were done via regression of household level foster-greer-thorbecke indicator on treatment dummy and other explanatory variables. using the evaluation sample of each program, we compute P(i,t,a) = (z – y(i,t) / z)a * poor(i,t), for alpha = 0, 1, and 2; and for each household, where y(i,t) is household i’s level of consumption per capita at year t, z is the country-specific poverty line, and poor(i,t) is an indicator function that equals 1 if the household is poor and equals 0 otherwise. for honduras, the poverty line used was lps 24.6 per capita per day in 2000 lempiras. expenditure values for 2002 were deflated to 2000 lempiras. for nicaragua, we used C$13.87 per capita per day in 2000 córdobas. expenditure values for 2001 and 2002 were deflated to 2000 córdobas. for mexico, we used the value of the Canasta Básica of 1997, which was m$320 per capita per month. We inflated this value of the Canasta Básica for 1998 and 1999 using the Canasta Básica price Index found at: http://www.banxico .org.mx/polmoneinflacion/estadisticas/indicesprecios/indicespreciosConsumidor.html. therefore, for october 1998, we used m$320 × 1.134. for June 1999, we used m$320 × 1.280. for october 1999, we used m$320 × 1.314. for Colombia (see Institute for fiscal studies, econometría, and sistemas especializados de Información 2006), the estimated impacts presented here are not equal to the unconditional double difference estimates because regressions control for other correlates. the impact for honduras was obtained from 2002 regression only. the impacts for mexico are all for single equation cross-sectional regressions for each year. a. Baseline, before households in CCt treatment group received transfers. B. no significant impact on poverty measure. * significant at the 10 percent level. ** significant at the 5 percent level.

the ImpaCt of CCts on ConsumptIon poverty and employment

Figure 4.1

Impact of CCTs on the Distribution of Consumption, Nicaragua and Honduras, 2002
A. Nicaragua B. Honduras
1.0 0.8 0.6 CDF 0.4 0.2 0

1.0 0.8 0.6 CDF 0.4 0.2 0 0

1

2

3

4

0

1

2

3

4

5

Log per capita expenditure (córdobas) Treatment Source: authors’ calculations. Note: Cdf = cumulative distribution function. Control

Log per capita expenditure (lempiras)

affected by the programs at a given stage of each program’s implementation. given that the evaluation samples were derived either from the pilot stages of a program (as in honduras and nicaragua) or from the early phases of expansion of the program (as in the Bolsa alimentação program in Brazil and the oportunidades program in mexico), these estimates may not be representative of the impacts of CCts on the population of beneficiaries after coverage has been expanded to the national level. In this section, we investigate the poverty impacts of some relatively large CCt programs, using nationally representative household surveys in four countries: Brazil, ecuador, Jamaica, and mexico.6 the welfare measure used is household consumption per capita (except for Brazil, where we use household income per capita because consumption data were not available). the poverty line is set in each country at the 25th percentile of the pretransfer distribution of consumption or income. to approximate pretransfer income or consumption for CCt beneficiaries, we simply subtract the full value of the transfer from income or consumption reported in the survey. this approach has important shortcomings: it amounts to ruling out behavioral changes, such as reductions in labor supply or remittances arising from the receipt of transfers, by assumption. furthermore, and unlike the evaluations 109

CondItIonal Cash transfers: reduCIng present and future poverty

discussed in the previous section, these calculations may be biased by purposeful program placement or self-selection.7 nevertheless, and keeping these important caveats in mind, the results are useful because they enable us to approximate the impact of large-scale CCt programs on measures of poverty at the national level. table 4.3 suggests that CCts generally helped reduce national poverty. In mexico there are large effects on poverty, especially for the poverty gap and squared poverty gap measures. for example, the estimates in table 4.3 suggest that oportunidades decreased the squared poverty gap by approximately 29 percent. In Jamaica, path reduced the squared poverty gap index by 13 percent from its pretransfer value. In Brazil, the impacts of the Bolsa família program on the headcount index and the poverty gap are modest; however, the program reduces the squared poverty gap by a substantial amount, 15 percent. this finding is consistent with the findings of paes de Barros, foguel, and ulyssea (2006), who suggest there is a strong link between the introduction of CCts and the fall in inequality in Brazil. the most puzzling findings correspond to the Bdh program in ecuador because the results in tables 4.2 and 4.3 are very different. It is likely that this difference arises, at least in part, because the estimates in

Table 4.3 Impact of CCT Programs on Poverty Indexes at the National Level, Various Years
Headcount Country Brazil ecuador Jamaica mexico Pre-transfer 0.2421 0.2439 0.2439 0.2406 Post-transfer 0.2369 0.2242 0.2329 0.2222 0.0980 0.0703 0.0659 0.0847 Poverty gap Pre-transfer Post-transfer 0.0901 0.0607 0.0602 0.0683 Squared poverty gap Pre-transfer 0.0553 0.0289 0.0258 0.0422 Post-transfer 0.0471 0.0235 0.0224 0.0298 Size of the transfer (% of PCE)a 11.7 8.3 10.7 33.4

Source: authors’ calculations. Note: pCe = per capita expenditure. the poverty line used in each country is the 25th percentile of the pretransfer national distribution (prior to the symmetrical trimming of the distribution for extreme outliers; that is, values of less than the 1st percentile and above the 99th percentile of the distribution). for Brazil, the measure of welfare used is per capita income (pCI). In the other three countries, the measure of welfare is pCe. pretransfer welfare is derived by subtracting the full value of the per capita cash transfer reported by a beneficiary household in each country from its welfare measure, inclusive of the transfer (pCe or pCI in Brazil). for Brazil, we use the pesquisa nacional por amostra de domicilios 2006. for ecuador, we use the encuesta de Condiciones de vida 2006. for Jamaica, we use the survey of living Conditions 2004. for mexico, we use the encuesta nacional de Ingresos y gastos de los hogares 2004. a. the transfer amounts as a proportion of per capita expenditures (or consumption) are not the same across all tables in the report because of differences in the surveys used, including their coverage and year.

110

the ImpaCt of CCts on ConsumptIon poverty and employment

table 4.3 disregard the very large reduction in child labor, which offsets the impact of the transfer (edmonds and schady 2008). In addition, differences in the coverages of the surveys may be important: the survey used for table 4.3 is nationally representative, whereas the data collected for the impact evaluation of the Bdh was limited to four provinces and to poor households within those provinces.8 Impacts on the Composition of Consumption In addition to impacts on aggregate consumption, CCts may affect disproportionately the consumption of particular items, such as food. this is of interest for a variety of reasons, including the link between food consumption and such measures of nutritional status as height-for-age and weight-for-height in children and body mass index for adults.9 also, in analyzing consumption patterns among CCt recipients, it is possible to test whether households use transfer income differently from other sources of income. this could happen for a variety of reasons. transfers are made to women, and there is a large body of evidence suggesting that women have different preferences over consumption than do men (thomas 1990; hoddinott and haddad 1995; lundberg, pollak, and Wales 1997; doss 2006; Ward-Batts 2008); that the conditions attached to transfers or the social marketing of programs may affect how transfer income is used;10 and that transfer income may be perceived as temporary, in which case households may save rather than consume the bulk of it (as suggested by the permanent income hypothesis). a number of authors have analyzed CCt effects on the food engel curve—the share of consumption that is devoted to food at various levels of total consumption. the intuition behind this is as follows: CCts transfer cash, which increases total consumption, as shown above. If households perceive CCts as any other source of income, we would expect that transfers move them along the food engel curve. on the other hand, if transfer income is treated differently from other sources of income, CCts may result in shifts of the food (and other) engel curves. to see whether that is the case, figure 4.2 graphs food engel curves for treated and control households in ecuador and nicaragua. the figure shows that engel curves in both countries have the familiar downward-sloping shape, with the share of food decreasing as total expenditures rise. this phenomenon is known as “engel’s law.” however, both panels of the figure show that, at the time of follow-up, 111

CondItIonal Cash transfers: reduCIng present and future poverty

Figure 4.2
0.75 0.70 0.65 Food share

Impact of CCTs on Food Shares in Ecuador and Nicaragua
Ecuador, 2005
0.75 0.70 0.65 Food share 0.60 0.55 0.50 0.45 0.40 0.35

Nicaragua, 2006

0.60 0.55 0.50 0.45 0.40 0.35 1.7 1.9 2.1 2.3 2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9 4.1 Log of per capita expenditures Treatment

7.5

7.7

7.9

8.1 8.3 8.5 8.7 8.9 9.1 Log of per capita expenditures

9.3

9.5

Control

Sources: for ecuador, schady and rosero (2008); for nicaragua, macours, schady, and vakis (2008).

the food engel curves of CCt beneficiaries are everywhere above those for control households—clear evidence that transfer income was used differently from other sources of income. similar results are reported elsewhere. using nonexperimental data for the familias en acción program in Colombia and the urban oportunidades program in mexico, attanasio, Battistin, and mesnard (2008) and angelucci and attanasio (2008) report upward shifts of the food engel curves among program beneficiaries. the regression results presented in table 4.4 also corroborate these findings. food share regression results indicate that for a given level of total household expenditure, treated households tend to consume a larger proportion of food. for example, the food share is about 4 percentage points higher among program beneficiaries in Colombia, ecuador, and nicaragua than among non-beneficiaries. moreover, insofar as CCt programs affect total consumption, the effect on the level of food expenditures (as opposed to the share, measured by the engel curve) can be considerable. In mexico, for example, the median value of food consumption was 11 percent higher for beneficiary households than for comparable control households, and the median caloric consumption had increased by 8 percent (hoddinott, skoufias, and Washburn 2000). 112

the ImpaCt of CCts on ConsumptIon poverty and employment

Table 4.4

Impact of CCTs on Food Shares
Brazil 2002 Colombia 2002 0.6 a 74 a 2006 0.65 6** 56 0.04** Ecuador 2005 0.73 B 54 0.04** Honduras 2000 0.53 a 71 a 2002 0.47 B 72 B 73 2000 0.44 Nicaragua 2001 0.35 38** 69 0.04** 2002 0.35 31** 68 0.04**

daily per capita food consumption food shares

Control Impact (%) Control (%) Impact (percentage points)

0.45 12** 60 0.02**

Sources: for Brazil, ecuador, honduras, and nicaragua, authors’ calculations. for Colombia, see Institute for fiscal studies, econometría, and sistemas especializados de Información (2006). Note: daily per capita food consumption is presented in us$ converted by the official current exchange rates at the time of the surveys. food share is the percentage of total per capita consumption dedicated to food. the estimated impacts presented here are not always equal to the unconditional double-difference estimates because some regressions control for other correlates. In the case of honduras, impacts were estimated with 2002 data only (via cross-sectional regression). a. Baseline, before households in CCt treatment group received transfers. B. no significant impact on poverty measure. ** significant at the 5 percent level.

the increase in expenditures on food generally is directed toward increasing quality. households that benefited from familias en acción in Colombia significantly increased items rich in protein, such as milk, meat, and eggs (attanasio and mesnard 2006); and the increases in food expenditures in mexico and nicaragua were driven largely by increased consumption of meat, fruits, and vegetables (hoddinott, skoufias, and Washburn 2000; maluccio and flores 2005). oportunidades also increased caloric diversity as measured by the number of different foodstuffs consumed. at similar overall food expenditure levels in nicaragua, macours, schady, and vakis (2008) show that households that receive transfers from the atención a Crisis program spend significantly less on staples (primarily rice, beans, and tortillas) and significantly more on animal protein (chicken, meat, milk, and eggs), as well as on fruits and vegetables. angelucci and attanasio (2008) report similar results using data for urban oportunidades in mexico. not only did households diversify their diets; they also shifted toward higher-quality sources of calories. What causes these engel curve shifts? schady and rosero (2007, 2008) hypothesize that CCts increase the bargaining power of women within the household, and that this results in increased food expenditures. they use data on the Bdh program in ecuador to test 113

CondItIonal Cash transfers: reduCIng present and future poverty

that hypothesis. specifically, they argue that if changes in bargaining power are important, one would expect to see program effects on the food engel curve among households that included prime-age men and prime-age women at baseline (where bargaining between males and females is an issue), but not among households with only prime-age women (where there is no bargaining of this sort). the results of their analysis are consistent with this prediction.

Analyzing Offsetting Behavioral Responses to CCTs the size of the transfer and the fraction of poor households that receive it are major determinants of CCt impacts on consumption poverty. however, table 4.1 shows that, for most countries, the impact of the transfer is generally somewhat smaller than the magnitude of the transfer (when both are normalized as a fraction of the consumption or income of households in the control group).11 the difference between these two values may be a result of behavioral changes by CCt beneficiaries, which partly offset the value of the transfer itself. We now turn to a discussion of the evidence on these possible offsetting effects, focusing on impacts on child labor, adult labor, remittances, fertility, and spillovers and other general equilibrium effects. Child Labor Whether CCt programs help reduce the prevalence and amount of child work is of interest not only because of the resulting difference between the amount of the transfer and the change in household consumption. rather, a reduction in child work is often seen as a good in its own right: working under poor conditions can adversely affect both the physical and mental health of children, and income-generating activities for children often take place at the cost of reductions in educational attainment and future earnings. there are two main channels through which CCts could reduce the prevalence and amount of work among school-age children. the first channel works though the conditional nature of the programs. given the requirement of school enrollment and regular attendance, children have less time available for participation in income-generating activities. 114

the ImpaCt of CCts on ConsumptIon poverty and employment

Conditions also may increase parents’ awareness of the importance of schooling and thereby decrease child work. the second channel is a pure income effect: households that receive the transfer are less likely to be dependent on the income of their children, and therefore may reduce child work, as suggested by a number of theoretical models (Basu and van 1998; Baland and robinson 2000). several CCts have been successful in reducing child work. frequently, these impacts have been concentrated among older children. table 4.5 shows that oportunidades reduced child work among older children, aged 12–17, especially among boys (for whom baseline levels of child work also were substantially higher). skoufias and parker (2001) also show that domestic work decreased substantially, especially for girls. In ecuador, edmonds and schady (2008) show that the Bono de desarrollo humano program had very large effects on child work among those children most vulnerable to transitioning from schooling to work. those effects are concentrated in work for pay away from the child’s home. on the other hand, Bdh transfers had small effects on child time allocation at peak school attendance ages and among children already out of school at baseline. In Cambodia, the Cessp program, which gives transfers to children in transition from primary to lower-secondary school, reduced work for pay by 11 percentage points (filmer and schady 2009c).

Table 4.5 Impact of Oportunidades on the Probability of Children Working
November 1999 Age group Pre-program level Coefficient t-statistic

Boys
8–11 years 12–17 years 0.0620 0.3775 –0.011 –0.047 –1.3 –2.1

Girls
8–11 years 12–17 years 0.0353 0.1317 0.000 –0.023 –0.5 –1.8

Source: skoufias and parker 2001, table 5.

115

CondItIonal Cash transfers: reduCIng present and future poverty

other CCt programs also appear to have reduced child work. In nicaragua, the rps reduced child work by 3–5 percentage points among children aged 7–13 (maluccio and flores 2005). furthermore, the fraction of children who only studied (as opposed to worked and studied, only worked, or neither worked nor studied) increased significantly (from 59 percent to 84 percent) as a result of the rps (maluccio 2005). yap, sedlacek, and orazem (2008) estimate the effects of the Brazilian petI, another precursor of the Bolsa família program. petI gave out conditional transfers to secondary school-age children enrolled in school. stipends were given directly to students, not to the families, conditional on school attendance and participation in special training workshops. petI beneficiaries reduced substantially their probability of working. attanasio et al. (2006), however, find no effect of the familias en acción program on child work in Colombia (although the program does appear to have reduced the amount of time dedicated to domestic chores); and glewwe and olinto (2004) find no effects of the praf program on child work in honduras. two recent papers consider the impact of CCts on child work when the transfer is conditional on school attendance for only one child in the household, and that child has siblings. potentially, programs of this nature could have positive or negative spillovers for other siblings—positive if the income effect reduces child work for all children, if transfers increase the bargaining capacity of women within the household, or if the social marketing by the program leads parents to reduce child work even for children whose school attendance is not monitored; negative if parents compensate for the reduction in work of one child by increasing the work of other siblings. Barrera-osorio et al. (2008) analyze subsidio Condicionado a la asistencia escolar, a pilot CCt program in Bogotá, Colombia. this program randomized assignment to individual children rather than households, and made transfers directly to students rather than to their parents.12 Barrera-osorio et al. show that, within the same household, a student selected into the program is 2 percentage points more likely to attend school and works about 1 hour less than a sibling who has not been selected. however, the beneficiary’s sibling (particularly if this sibling is a girl) is less likely to attend school than are children in households that received no cash transfer at all. on the other hand, filmer and schady (2009c) find that the Cessp program in Cambodia had no effect on the school enrollment of a beneficiary’s ineligible siblings. more research is needed to understand this difference 116

the ImpaCt of CCts on ConsumptIon poverty and employment

between the two programs, especially if the number of CCts that attempt to target individual children increases. (for a discussion of time spent in school as a substitute for child labor, see box 4.1.) Adult Labor Supply of greater concern than changes in the amount of child labor are any possible reductions in adult work that result from CCts. such reductions could happen for a variety of reasons. If leisure is a normal good, then the income effect associated with the transfer might result in more leisure and less work. there also may be a price effect: beneficiaries of CCts may believe (correctly or incorrectly) that they need to supply less labor to become or continue to be “poor” and eligible for a meanstested program. adults also may have to take time away from work—for instance, to take children to school or health clinics. all of these could result in a reduction in adult work effort. Indeed, concern with possible work disincentives was one of the main reasons for the reform of transfer and other welfare programs in the united states in the 1990s (see box 4.2). In practice, CCts appear to have had, at most, modest disincentives for adult work. two studies (parker and skoufias 2000; skoufias and

Box 4.1

Is Time Spent in School a Perfect Substitute for Time Working? that the decrease in the prevalence of child work for boys was only about a quarter of the increase in the percentage of boys enrolled in school. for girls, it was only about one eighth of the increase in enrollment. on the other hand, using data from oportunidades and a broad definition of work (including market, farm, and domestic work), skoufias and parker (2001) show that the reduction in time spent at work largely equals the increase in schooling for boys, but not for girls. rather, leisure time is reduced significantly for girls. the pattern also varied by gender in other ways, with boys reducing both market and domestic work and girls reducing mainly domestic work.

an Important questIon Is Whether tIme

spent in school and time spent at work are fully substitutable. the answer seems to be rarely so. In Colombia, attanasio et al. (2006) provide evidence of partial substitution between school and work, with at most 25 percent of each extra hour spent on schooling coming from time otherwise spent at work. Because most of the substitution arises from a decrease in hours of domestic work activities, time spent on income-generating activities largely is unaffected and leisure is somewhat reduced. the substitution effects are largest among the children ages 10 to 13 in rural areas and ages 14 to 17 in urban areas. In their analysis of the ffe program in Bangladesh, ravallion and Wodon (2000) show

117

CondItIonal Cash transfers: reduCIng present and future poverty

Box 4.2 Work Disincentive Effects of Social Assistance Programs in Developed Countries t he lIter at ur e on Wor k dIsInCent I v e

effects of social assistance programs in developed countries is vast (for surveys, see atkinson 1987; krueger and meyer 2002; moffitt 2002). moffitt, for instance, finds that because of its implicit tax on income, the u.s. federal assistance program, aid to families with dependent Children (afdC), reduced labor force participation of beneficiaries by 10 to 50 percent, when compared with similar nonbeneficiary households.a to address the issue of a strong built-in disincentive to work, in 1996 the u.s. government replaced afdC with a new program, temporary assistance for needy families (tanf). tanf is different from afdC in many ways. first, there is no entitlement. the fact that a household’s income is below a certain level does not entitle it to a transfer. second, tanf introduced time limits. Individuals cannot receive cash benefits for more than five years (with few exceptions); and after two years in the program, recipients must work at least 30 hours per week to continue to be eligible for the transfers. third, at least 50 percent of single-mother recipients and 90 percent of two-parent families must be working or in a job training program offered by the state. finally, states now may decide on their program’s benefit reduction rates (or implicit tax on income). that is, instead of being required to reduce the benefits by one dollar for each dollar earned, states can decide

whether this implicit tax rate will be zero to one, one to one, or any other rate in between. a study by grogger (2003) indicates that up to 12 percent of the welfare caseload decline observed between 1993 and 1999 (from 33 percent to 15 percent), and up to 7 percent of the observed increase in employment rates of families headed by single mothers (from 69 percent to 83 percent) during the same period were the result of time limits introduced by tanf. the study finds no effects of time limits on hours worked by recipient single mothers. Bloom and michalopoulos (2001) show results from randomized experiments that indicate time limits seem to increase employment of welfare recipients by 4–11 percentage points (from a base that varied from 40 percent to 55 percent employment rates). according to the research surveyed by Blank (2002) and moffitt (2002), however, the marginal tax rate changes embodied in tanf do not seem to have had any effect on work effort. It seems that most of the changes in labor supply induced by tanf came from its time limits and work requirement.
a. Between 1935 and 1996, the main government cash transfer program in the united states was afdC. federal law required that the afdC grant to an individual be reduced by one dollar for each dollar earned as income. this requirement represented a 100 percent implicit tax on income, and many policy makers and academics alike worried that such design features created strong work disincentives.

di maro 2006) examine the effects of oportunidades on adult labor supply; neither finds evidence of disincentive effects. the data used by edmonds and schady (2008) suggest that the Bdh program in ecuador had no effects on adult labor supply; in a similar vein, filmer and schady (2009c) report that adult labor supply was largely unaffected by the Cessp program in Cambodia. only in nicaragua is there some evidence of significant negative effects on adult work: maluccio and flores (2005) show that the rps resulted in a significant reduction 118

the ImpaCt of CCts on ConsumptIon poverty and employment

in hours worked by adult men in the preceding week (by about 6 hours), with no effect among adult women. Why did CCt programs not lead to larger reductions in adult labor supply, as had been a concern of many policy makers and academics? there are various possible explanations. first, the beneficiaries of CCt programs generally are very poor, and the income elasticity of leisure may be quite low for households that are this poor. moreover, for some households the reduction in income from child work and the increase in school expenditures associated with the additional school enrollment offset the amount of the transfer. obviously, that is particularly true for programs that made small transfers but had large effects on school enrollment and child work, as with the Cessp program in Cambodia; but edmonds and schady (2008) show that it was also the case for beneficiaries of the Bdh program in ecuador. under these circumstances, increasing adult labor supply (or at least not reducing it) is one way to keep income and consumption at a level comparable with what it would have been had a household not taken up the program. not coincidentally, perhaps, disincentive effects on adult labor supply are found only for the program that made the most generous transfers, the rps in nicaragua. there are other reasons that might help explain why there have not been large disincentives to adult labor associated with CCts. first, there are issues of timing. If households perceive transfers to be “temporary” rather than a permanent new “entitlement,” they would treat them as a windfall, and generally would not change the labor supply of adults. moreover, the data used to estimate the CCt impacts on labor supply generally reflect household responses shortly after they have become eligible for the program for the first time. In the longer run, as households have more time to adjust their behavior, disincentive effects on adult labor may become more of an issue. nevertheless, recent research on the south african old-age pension (oap) scheme, which makes transfers that dwarf those of even the most generous CCts and which is likely to be seen as “permanent” by beneficiaries, is encouraging. as box 4.3 shows, the oap does not appear to have reduced work effort by prime-age adults. Crowding-Out of Remittances and Transfers the impact of CCts on consumption poverty also could be offset if they crowd out transfers from other sources, such as remittances. that could happen if senders of remittances or other private transfers target 119

CondItIonal Cash transfers: reduCIng present and future poverty

Box 4.3 Do Transfers Reduce the Supply of Adult Labor? Evidence from the South African Pension Scheme the south afrICan oap sCheme provIdes a

generous benefit to retirees in that country. the value of the transfer is more than twice median per capita income for african (black) households. In principle, the program is means tested. In practice, however, all households that do not have a private pension are eligible. the program was made available to black families after the end of apartheid in 1994. By now, it is likely that the program is seen as an “entitlement” by most beneficiary households. early research on the oap suggested that it had substantial negative effects on adult labor supply (Bertrand, mullainathan, and miller 2003). more recent research (ardington, Case, and hosegood 2008) disputes those findings. these new results are

based on better data—specifically, panel data rather than a single cross-section, which allows the authors to control for time-invariant differences between pension recipients and nonrecipients; and data on nonresident (migrant) household members, which are important because migrant status is correlated with pension receipt. the preferred specification in ardington, Case, and hosegood (2008) suggests that the oap had a positive effect on adult labor supply—the probability that prime-age adults are employed is approximately 3 percentage points higher in households with at least one pension recipient. those authors argue that the oap relieves financial and child care constraints, which can be short-run impediments to migrating, even when the medium-run returns to migrating are positive.

a fixed level of income for recipient households or seek to equate marginal utility across donors and recipients. When part of the transfer is crowded out, this fraction will accrue to households outside the target group, and program recipients will benefit less than intended. however, the implied mistargeting of program resources also means that there is a positive effect of the program beyond that measured by evaluation surveys in the treatment areas. empirical evidence on the crowding-out effects of CCts shows mixed results. for mexico, albarran and attanasio (2003) show some indication of crowding out for oportunidades, using one round of ex post evaluation data. however, teruel and davis (2000) reject the crowding-out impact of oportunidades on private transfers, using more rounds of evaluation data. their result holds for both monetary and in-kind transfers. more recently, nielsen and olinto (2008) provide evidence on crowding-out effects of the honduran and nicaraguan CCt programs. they find that both the prevalence and the amount of remittances in the two countries were unaffected by the programs. that finding is comforting because remittances constitute a major source of foreign 120

the ImpaCt of CCts on ConsumptIon poverty and employment

currency for many countries in Central america. however, the evidence in nielsen and olinto (2008) points toward some crowding out of private food transfers and money and food transfers from ngos in nicaragua, which could be a concern if it represents a change in informal insurance schemes. the praf in honduras does not seem to crowd out any of these private transfers, most likely because of the modest size of CCt payments in that program. Fertility and Family Composition transfers made by CCt programs are often a function, in part, of the number of children, sometimes with a cap on the total amount of transfers for which a household can be eligible (see the discussion in chapter 3). one concern is that CCts could provide incentives for increased fertility, which could result in eventual reductions in household (and national) welfare. In practice, any effects on fertility appear to have been modest. a recent paper (stecklov et al. 2006) finds no effects on the total fertility rate among beneficiaries of oportunidades in mexico or the rps in nicaragua; however, it appears that praf increased fertility among eligible households in honduras by 2–4 percentage points. the authors argue that these differences can be explained by differences in program design: In mexico, the transfer to households with preschool children was a lump sum, regardless of the number of children; and (in the first three years of the program) poor, childless households could not become eligible for transfers if they had children after the first wave of inscriptions. In nicaragua, the transfer was also a lump sum, although some households became eligible for transfers once they had children. In honduras, finally, new households could be registered if they gave birth to children, and the amount of the transfer depended on the number of young children. If borne out by results from other countries, the evidence in stecklov et al. (2006) would suggest that the details of program design are important because they can provide incentives that result in unintended outcomes.13 Spillovers and General Equilibrium Effects CCt programs often are targeted geographically to poor and remote rural areas. In some cases, transfers are substantial and a 121

CondItIonal Cash transfers: reduCIng present and future poverty

large proportion of the population in a community receives them. potentially, this could result in general equilibrium and spillover effects in the local economy. for instance, CCts could increase the prices of consumption goods through higher demand, or could increase prevailing local wages because of the reduction in the labor supply of children. to assess whether such spillover effects occurred in mexico, angelucci and de giorgi (2008) analyze the evaluation data of oportunidades for both beneficiaries and ineligible households living in treatment communities. they find that there was no indirect negative effect on labor earnings, prices, and the receipt of other welfare payments. In fact, they observe that the real incomes of ineligible households living in treatment communities seem to have been affected positively by the program. they show that ineligible households in treatment villages consumed more by receiving more private transfers, by borrowing more (almost exclusively from family, friends, or informal moneylenders), and by reducing their stocks of grains and animals. In addition, they show that the indirect program effects on consumption and loans are larger for households hit by a negative shock. the lack of impact on wages and prices of consumer goods is not surprising. In most countries in which CCts have been evaluated, labor and goods markets are sufficiently developed so that both labor and goods are largely tradable. CCts may induce larger local demand for goods and lower local supply of labor, and, in the short run, prices may change to reflect these imbalances; in the long run, however, prices should return to their initial equilibrium. another kind of spillover effect is related to changes in access to and use of the formal banking sector. a number of CCt programs, including Bolsa família in Brazil and the Bdh in ecuador, directly deposit benefits in bank accounts created for beneficiaries, who then can withdraw cash using an automated teller machine (atm) card. that payment system appears to have reduced transaction costs (such as standing in line to receive transfers), and is likely to have reduced any stigma attached to the program. In addition, creating a bank account for CCt beneficiaries and giving them atm cards may make it more likely that they use the formal banking sector in other capacities—potentially, a very important benefit of CCt programs and one that has not been evaluated to date. 122

the ImpaCt of CCts on ConsumptIon poverty and employment

Long-Term Impacts of CCTs on Consumption as we have shown, many CCt programs have had substantial effects on consumption and poverty in the short run. a natural question is whether those positive impacts are likely to remain, at least in part, once households are no longer eligible for the CCt or the program ceases to exist altogether. positive effects could be maintained, for example, if part of the transfer is saved and invested in productive assets, or if the stable income stream of the transfer allowed households to gain access to credit and overcome liquidity constraints. It also is possible that the transfer enables households to smooth consumption when they face negative shocks. If CCt programs do have a long-term impact on household consumption, then the estimates of program impact on short-term consumption and poverty reported above will underestimate the true (medium- and long-run) impact of CCts on poverty. looking first at the investment of transfers, gertler, martínez, and rubio-Codina (2006) provide extensive evidence on the mexican experience. they find that the program had a substantial positive impact on investment in productive activities such as microenterprises and agriculture (animals and land). on average, 12 percent of transfers were invested, and households that received more transfers from oportunidades also invested more. It seems that the CCt helped alleviate two market failures. first, the increased income allowed households to overcome credit constraints. second, the stable stream of income may have made households willing to undertake more risky (and profitable) investments. another study (maluccio 2008) assesses the impact of the rps program in nicaragua on various types of investments. the author finds only limited evidence that the program led to an increase in investment for agricultural equipment. his findings do not imply that the program had no long-term effects—it almost certainly did in terms of investment in child health and education, which should continue to lead to benefits for many years to come. In contrast to mexico, however, there was only weak (albeit positive) evidence that rps improved investment activities, possibly because of an economic downturn during the period, the strong program orientation toward increased food expenditures, and the limited opportunities in the impoverished rural areas where the program operated. CCts also may help households smooth their consumption and protect them from adverse shocks. If that is true, treated households may 123

CondItIonal Cash transfers: reduCIng present and future poverty

be relatively more willing to undertake risky investments and less likely to sell assets or discontinue their children’s school enrollment during an economic downturn. the ability of the rps to function as a social safety net during the so-called coffee crisis provides empirical evidence that CCts can protect households from adverse shocks. for nicaragua during 2000 and 2001, the crisis consisted of a drop in coffee prices to a 30-year low (or a 100-year low, adjusting for inflation) due to a worldwide oversupply of coffee (varangis et al. 2003). the fall in prices hurt farmers and laborers socially and financially. By comparing coffee-growing and non-coffee–growing areas and treatment and control households, one can measure how well the rps performed as a social safety net. maluccio (2005) finds that the rps enabled beneficiary households to maintain per capita expenditures during the crisis and helped reduce labor supply increases in coffee-growing areas. the effect was larger for those who were most affected by the fall in coffee prices. It is important to make clear that the protection of income during a shock was achieved even if the rps in no way was designed to respond to shocks. a similar result holds for mexico, where oportunidades helped beneficiary households smooth their consumption in the face of income fluctuations (skoufias 2002). skoufias also finds that oportunidades provided that protection without replacing existing informal insurance schemes.

Conclusion this chapter has reviewed the impacts of CCts on household consumption, poverty, the composition of consumption, behavioral responses that might offset the effect of transfers, and long-term welfare. We focus on CCt programs for which there are robust evaluation data. policy makers and academics alike long have been concerned with the disincentive and general equilibrium effects of government cash transfers to the poor. that fear stems from hypothesized disincentives to work, crowding out of private transfers, effects on fertility and family composition, and effects on local wages and prices. the evidence reviewed in this chapter, however, suggests that these offsetting effects generally have been modest.

124

the ImpaCt of CCts on ConsumptIon poverty and employment

first, by and large, programs have had positive impacts on consumption, especially when the transfer amount is generous (as with the rps program in nicaragua). In and of itself, those positive impacts on consumption are indirect evidence that the offsetting behavioral responses are unlikely to be large, and that the marginal propensity to consume out of transfer income is high. moreover, because transfers generally are well targeted to the poor, the effects on consumption have translated into impacts on poverty. second, the evidence suggests that CCts generally do not have large disincentive effects on the labor supply of adults. more research is needed to see whether those patterns are maintained as programs mature and beneficiaries have more time to adjust their behavior. however, the results to date indicate that the popular view that cash transfers encourage indolence is not supported by the evidence. third, unlike most social assistance programs in the developed world, CCts do not seem to crowd out private transfers. although there is some evidence that CCts crowd out intracommunity transfers of inkind goods, these usually are transfers among the poor and therefore have very little redistributional impact. also, CCts do not appear to have had large effects on fertility. fourth, CCts seem to have no significant negative effects on local wages, local prices, and the receipt of other welfare payments. In fact, contrary to expectations, there is some evidence that the real incomes of ineligible households living in program communities have been affected positively by CCts. In mexico, nonpoor households in treatment villages received more private transfers, borrowed more, and reduced their stocks of grains and animals; they also consumed more. finally, although the evidence suggests that, as intended, CCts do have significant impacts in reducing child labor, the resulting income losses generally are not large enough to offset the impact of transfers on per capita consumption. (Cambodia and especially ecuador appear to be exceptions to this general pattern.) In sum, the main conclusion of this chapter is that redistribution via direct cash transfers seems to have worked well. most programs, especially those making sizable transfers, have had substantial impacts on consumption and on poverty. the offsetting effects that were a source of concern when CCt programs were created do not appear to have occurred on a scale large enough to offset the bulk of the transfer.

125

CondItIonal Cash transfers: reduCIng present and future poverty

CCts do not seem to reduce the labor supply of adults or to crowd out private transfers. they do reduce the supply of child labor, but this reduction seems to have only a modest impact on household income and consumption. moreover, some CCts seem to increase productive investment, which boosts the impact on poverty even farther.

126

Chapter five

the impact of CCt programs on the accumulation of human Capital in addition to the objeCtive of reduCing Current poverty,

as discussed in the previous chapter, CCt programs try to encourage households to invest in the human capital of their children. this chapter turns to the evidence of program impacts on education, health, and nutrition. as in chapter 4, the evidence provided in this chapter relies on rigorous impact evaluations (described in appendix b) and on calculations done for the purpose of this report. the first section presents evidence of CCt program effects on the use of education and health services. We show that, by and large, CCts have had significant and, in some cases, large effects on school enrollment and attendance. there is also some evidence of increases in the use of preventive health services, although that is not as clear. the second section of the chapter presents evidence of CCt program effects on “final” outcomes in education and health—for example, years of schooling completed, test scores, child height for age, and infant mortality. We show that there are many fewer evaluations to draw on, with a disproportionate amount of the evidence coming from a single country (Mexico). importantly, the evidence on the impact of CCts on these “final” outcomes is somewhat mixed. thus CCts appear to have had a modest impact on years of schooling completed by adults; they reduced the incidence of low child height for age only in some countries and only among some populations; and they resulted in modest improvements in cognitive development among very young children, but had no discernible effect on learning outcomes for children who benefited from CCt programs while they were of school age. the third section of the chapter then considers the extent to which CCt program effects on human capital appear to be a result of the “income” effect associated with the transfer, the “price” effect that results from the condition, or both. 127

Conditional Cash transfers: reduCing present and future poverty

CCT Program Effects on the Use of Education and Health Services
CCts transfer cash and require that households make regular use of education and health services. in this section, we review evidence of the impact of CCt programs on the use of those services, focusing first on education and then on health.

Table 5.1 Impact of CCTs on School Enrollment and Attendance, Various Years
Age/Gender/ Grade Baseline enrollment (%) Transfer (% of PCE)b Evaluation method

Country

Program

Impacta

Reference

Latin American and Caribbean countries
Chile Colombia Chile solidario ages 6–15 60.7 91.7 63.2 75.2 66.4 7.5*** (3.0) 2.1** (1.0) 5.6*** (1.8) 10.3** (4.8) 3.3*** (0.3) 0.5** (0.2) 10 9 iv, randomized randomized 7 17 rdd psM, dd galasso (2006) attanasio, fitzsimmons, and gómez (2005) schady and araujo (2008) glewwe and olinto (2004) levy and ohls (2007)

familias en acción ages 8–13 ages 14–17

ecuador honduras

bono de desarrollo ages 6–17 humano programa de asignación familiar program of advancement through health and education oportunidades ages 6–13

jamaica

ages 7–17

18 daysc

10

rdd

Mexico

grades 0–5 grade 6 grades 7–9

94.0 45.0 42.5 90.5 72.0

1.9 (25.0) 8.7*** (0.4) 0.6 (56.4)

20

randomized

schultz (2004)

nicaragua nicaragua

atención a Crisis red de protección social

ages 7–15 ages 7–13

6.6*** 18 (0.9) 12.8*** 27 (4.3)

randomized randomized

Macours and vakis (2008) Maluccio and flores (2005)

128

the iMpaCt of CCt prograMs on the aCCuMulation of huMan Capital

Effects on School Enrollment and Attendance Overall Program Effects a large number of evaluations estimate the effect of CCts on school enrollment and attendance. table 5.1 shows that virtually every program that has had a credible evaluation has found a positive effect on school enrollment, although those effects sometimes are found among some age groups and not others.

Table 5.1

continued
Age/Gender/ Grade Baseline enrollment (%) Transfer (% of PCE)b Evaluation method

Country

Program

Impacta

Reference

Non–Latin American and Caribbean countries bangladesh female secondary school assistance program japan fund for poverty reduction Cambodia education sector support project punjab education sector reform program social risk Mitigation project ages 11–18 (girls) grades 7–9 (girls) grades 7–9 44.1 12.0** (5.1) 31.3*** (2.3) 21.4*** (4.0) 11.1*** (3.8) –3.0* n.a. 5.2 n.a. 0.6 fe Khandker, pitt, and fuwa (2003) filmer and schady (2008) filmer and schady (2009c) Chaudhury and parajuli (2008) ahmed et al. (2007)

Cambodia Cambodia

65.0 65.0

2–3 2–3

dd rdd

pakistan

ages 10–14 (girls) primary school secondary school

29.0

3

ddd

turkey

87.9 39.2

6

rdd

Source: authors’ compilation. Note: dd = difference-in-differences; ddd = difference-in-difference-in-differences; fe = fixed effects; iv = instrumental variables; n.a. = not available; pCe = per capita expenditure; psM = propensity score matching; rdd = regression discontinuity design. this table contains unweighted means for the coefficients for Colombia ages 8–13 and 14–17, Chile ages 4–5 and 6–15, and Mexico grades 0–5 and 7–9. the standard errors in each case are the square roots of the averaged variances of these estimates. a. the column for “impact” reports the coefficient and standard error (in parentheses); the unit is percentage points, with the exception of the jamaican path program, where the unit is days. b. the transfer amounts as a proportion of per capita expenditures (or consumption) are not the same across all tables in the report because of differences in the surveys used, including their coverage and year. c. impacts were measured in jamaica only for student attendance over a 20-day reference period. the baseline enrollment rate prior to path was 96 percent. * significant at the 10 percent level. ** significant at the 5 percent level. *** significant at the 1 percent level.

129

in latin america and the Caribbean, there are evaluations for programs in Chile, Colombia, ecuador, honduras, jamaica, Mexico, and nicaragua. five of those evaluations identify program effects on the basis of random assignment. in Mexico, the impact of oportunidades in rural areas is significant for children making the transition from primary to secondary school, a point to which we return below (schultz 2004; behrman, sengupta, and todd 2005; de janvry and sadoulet 2006). oportunidades also appears to have had positive spillover effects—school enrollment increased even among children above the cut-off point of the proxy means who were ineligible for transfers. bobonis and finan (2008) argue that the increase was a result of peer effects—barely ineligible children in oportunidades communities were more likely to enroll because their eligible peers were in school. in nicaragua, the rps program was targeted at children aged 7–13 who had not yet completed the fourth grade of primary school. the evaluation results show that the rps had large effects on school enrollment—13 percentage points (Maluccio and flores 2005). in honduras, the praf also had a positive effect on school enrollment, although the impact was much smaller—on the order of 3 percentage points (glewwe and olinto 2004). other evaluations have used quasi-experimental methods. schady and araujo (2008) use instrumental variables to estimate an impact of approximately 10 percentage points on enrollment for the bdh program in ecuador. galasso (2006) analyzes the impact of the Chile solidario program on a variety of outcomes, including preschool and school enrollment. using regression discontinuity techniques, she estimates program effects of 4–5 percentage points on preschool enrollment and of approximately 7 percentage points on the probability that all children aged 6–14 are enrolled in school. in Colombia, fitzsimmons and gomez (2005) use differencesin-differences to compare changes in villages where the familias en acción program was operational with a comparison group of villages. their results suggest that the program had impacts of 2 percentage points for children aged 8–13 at baseline and 6 percentage points for those aged 14–17 at baseline. finally, an evaluation of path in jamaica indicates that the program increased school attendance by approximately 0.5 days per month (levy and ohls 2007). as we showed in chapter 1, CCts have been most popular in latin america, but they have gradually spread to a number of countries in other regions. in those countries, programs often are referred to as 130

the iMpaCt of CCt prograMs on the aCCuMulation of huMan Capital

“scholarship” or “stipend” programs. in practice, however, they work much like CCts. two programs in south asia target girls. Khandker, pitt, and fuwa (2003) assess the impact of the fssap program on enrollment in bangladesh. they use data from a panel of households to show that the probability that a girl enrolls in school increased more in villages that participated in the fssap program earlier than in villages that participated later. on the basis of those comparisons, they estimate that every year of program exposure increased the female enrollment rate by 12 percentage points.1 Chaudhury and parajuli (2008) consider the impact of the punjab education sector reform program in pakistan. they use regression discontinuity and triple-differencing techniques, and conclude that the program increased enrollment by approximately 11 percentage points. in Cambodia, filmer and schady (2008) evaluate the impact of the jfpr program, which is targeted at girls making the transition from elementary to lower-secondary school. their differences-in-differences estimates suggest a very large program impact—approximately 31 percentage points. a follow-on program, the Cessp scholarship, was made available to boys as well as girls. an evaluation of that program using regression discontinuity techniques finds a program effect of 21 percentage points (filmer and schady 2009c). in turkey, finally, the CCt program had an impact on secondary school enrollment, but not on the enrollment of children in primary school (ahmed et al. 2007). Heterogeneity by Baseline Enrollment Whereas table 5.1 shows that virtually all of the programs that have been evaluated have had positive effects on school enrollment, those effects appear to vary considerably across countries and across population groups within countries. one dimension of heterogeneity is by baseline enrollment, with generally higher impacts found in settings where school enrollments prior to the CCt program were low. this pattern is apparent when making comparisons across countries—for example, the impact among children of primary school–age is substantially larger in nicaragua than in Mexico or Colombia. it is also apparent when making comparisons within countries—for example, the results reported in ahmed et al. (2007) suggest that the CCt in turkey had no effect on enrollment in primary school and among boys in secondary school; however, among girls of secondary school–age, for whom baseline enrollment rates were 131

Conditional Cash transfers: reduCing present and future poverty

very low (38.2 percent), the impact of the program on enrollment was approximately 11 percentage points. to some extent, the fact that CCt impacts are larger when enrollment is lower is purely mechanical because net enrollment rates cannot exceed 100 percent. however, this pattern also may be driven at least partly by differences in the expected rate of return to schooling. a number of authors have argued that there may be heterogeneity in these returns (Card 1999; heckman and Carneiro 2003). if children or their parents select into school at least in part on the basis of expected returns (so-called roy selection), the children who can expect to benefit the most from schooling generally will enroll first. in this case, the rate of return for the marginal unenrolled child may be higher when overall schooling levels are low. also, in countries where overall school enrollment rates are low, educated workers will be relatively scarce and generally able to command a higher premium in the labor market. higher rates of return to education should make households more responsive to a transfer that is conditional on schooling. Heterogeneity by Transfer Size and Timing of Payments all else being equal, we would expect programs that make larger transfers to have larger effects on school enrollment (among other outcomes). however, the evidence presented in table 5.1 suggests that, at current transfer sizes, larger transfers are not consistently associated with larger program effects on school enrollment. Within latin america, the program that makes the largest transfers, rps, had the largest effects on enrollment. however, other programs that made large transfers, including oportunidades in Mexico and familias en acción in Colombia, had much smaller impacts on enrollment. Meanwhile, some programs that made more modest transfers, including the bdh in ecuador and Chile solidario, produced substantial impacts on enrollment. furthermore, by far the biggest program effects on enrollment are found among programs in bangladesh, Cambodia, and pakistan, all of which make very modest transfers (between 1 and 3 percent of the expenditures of the median recipient household). the issue of how transfer size affects enrollment is addressed explicitly in a recent paper by filmer and schady (2009a), who exploit the fact that the Cessp scholarship program in Cambodia made payments of different magnitudes: within any school, the poorest 25 students were offered scholarships of $60, and the next-poorest 25 students were 132

the iMpaCt of CCt prograMs on the aCCuMulation of huMan Capital

offered scholarships of $45. figure 5.1 shows a clear jump in the probability that a child who was offered a $45 scholarship was attending school on the day of an unannounced visit, relative to those who were offered no scholarship at all. by contrast, the effect of the additional $15 was small—on average, students who were offered a $60 scholarship were only 4 percentage points more likely to be attending school than those who were offered a $45 scholarship. in per dollar terms, every dollar of the first $45 had more than twice as large an impact on attendance as every dollar of the additional $15. What can we learn from these results? ultimately, the effect of transfer size on enrollment outcomes is likely to be highly context specific and will depend on a variety of other factors. however, one would expect there to be diminishing marginal returns to transfer size. figure 5.1 and the differences across countries in impacts summarized in table 5.1 suggest that, at current transfer levels, the marginal effect of larger transfers on school enrollment may be modest.

Figure 5.1
1.0

Impact of Transfers of Different Magnitude on School Attendance in Cambodia, 2005–06
1.0

0.8 Probability of school enrollment Probability of school enrollment –5 0 5 10 15 20 25

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0 –25 –20 –15 –10

0 –25 –20 –15 –10

–5

0

5

10

15

20

25

Relative ranking (0 = $45 cutoff) Quartic Non-parametric

Relative ranking (0 = $60 cutoff)

Source: filmer and schady 2009a. Note: intent-to-treat effects. the left-hand panel compares the enrollment effect of receiving a $45 scholarship versus no scholarship, whereas the right-hand panel compares the effect of receiving a $60 scholarship versus a $45 scholarship.

133

Conditional Cash transfers: reduCing present and future poverty

timing of payments is another potentially important design feature. CCt programs have adopted different schemes. for example, brazil’s bolsa família pays on a monthly basis, Colombia’s familias en acción pays bimonthly, and Cambodia’s jfpr and Cessp pay quarterly. the driving force behind these design choices has been operational in nature, balancing convenience for beneficiaries with costs for the program. even so, the timing (and frequency) of payments may affect program impacts. that possibility is illustrated in the case of bogotá’s subsidio Condicionado a la asistencia escolar, in which beneficiaries were assigned randomly to different payment structures: a regular bimonthly payment, a smaller bimonthly payment supplemented by an end-of-year bonus, and the same smaller bimonthly payment supplemented by a larger bonus at school graduation. barrera-osorio et al. (2008) find that reducing the monthly payment and adding an end-ofyear bonus does not reduce impacts (a finding the authors interpret as indicative that “short-term” liquidity constrains are low), whereas the lump-sum payment upon graduation has positive effects on attendance. the latter result, in particular, suggests that some payment schedules may augment incentives for behavioral change. to date, however, few programs have experimented in that direction. Heterogeneity by School Grade or Child Age We next turn to a discussion of differences in enrollment effects by school grade. figure 5.2 shows the enrollment trajectories of oportunidades recipients and nonrecipients in Mexico. each line corresponds to the probability of being enrolled in a school grade, conditional on having completed the grade before (the “continuation rate”). the figure shows that in Mexico this probability is highly nonlinear—the bulk of dropouts occur in the transitions from primary school to lower-secondary school (6th to 7th grade) and from lower-secondary to upper-secondary school (9th to 10th grade). the impact of oportunidades transfers on school enrollment is given by the vertical distance between the two lines. that distance is clearly largest for children entering the first grade of secondary school—indeed, the results in schultz (2004) show that oportunidades effects on enrollment are significant only for children enrolled in grade 6 at baseline. schady and araujo (2008) find that the bdh program in ecuador also had the largest program effects among children in transition grades. note also that table 5.1 shows that the largest impacts on enrollment observed for any CCt program are those found among 134

the iMpaCt of CCt prograMs on the aCCuMulation of huMan Capital

Figure 5.2
100

Oportunidades Impacts on School Enrollment, by Grade, 1998

90 Oportunidades recipients

Continuation rate (%)

80

70 Control group children 60

50 Eligibility for Oportunidades 40

2

3

4

5

6 Entering grade

7

8

9

10

Source: de janvry and sadoulet 2006.

recipients of the jfpr and Cessp programs in Cambodia. both of those programs focus on children making the transition from primary to lower-secondary school. Heterogeneity by Socioeconomic Status a final dimension of heterogeneity concerns differences in program effects by the socioeconomic status of households at baseline. A priori, there are several reasons why one might expect that the impact of a CCt program would be larger for the poorest households. those households have worse education outcomes at baseline, so there is more margin for improvement; they may be more credit constrained in ways that affect their schooling choices; the cost of forgoing transfers if children do not comply with the program conditions may be higher for the poor, either because the transfer is a higher fraction of household income or because of the diminishing marginal utility of income; and finally, if there is heterogeneity in the returns to schooling, as discussed above, the expected returns to schooling for the marginal unenrolled child may be higher among the poorest children because their baseline enrollment rates are lower. 135

Conditional Cash transfers: reduCing present and future poverty

in practice, numerous studies have shown larger CCt program effects among households that are poorer at baseline. that point is made in figure 5.3, which presents the results of an impact evaluation of the rps in nicaragua. the left-hand panel of the figure focuses on school enrollment among children aged 7–13, and the right-hand panel focuses on the fraction of children aged newborn to 3 years who have been weighed at least once in the previous 6 months. the figure clearly shows that program effects on both outcomes were largest among extremely poor households. similar results are reported elsewhere. in Cambodia, filmer and schady (2008) show that the impact of the jfpr program on enrollment is approximately 50 percentage points for girls in the poorest two deciles of a composite measure of socioeconomic status, compared with 15 percentage points for girls in the richest two deciles. as a result of the larger program impacts among the poorest households, the jfpr eliminated the “gradient” between poverty and school enrollment among beneficiaries. in honduras, glewwe and olinto (2004) also find significantly larger program effects on enrollment among households with lower per capita expenditures. in Mexico, behrman, sengupta, and todd (2005) argue that oportunidades program effects are largest for children with the lowest propensities to enroll in school at baseline. finally, oosterbeek, ponce, and schady (2008) show that the bdh program in ecuador had a significant effect on enrollment for children around the 20th percentile of the proxy means, but no effect among children around the 40th percentile.2,3
Figure 5.3 Heterogeneity of Impacts by Socioeconomic Status, Nicaragua, 2000
School enrollment (children ages 7–13)
Extreme poor Poor Nonpoor

Children weighed in past 6 months (ages 0–3)
Extreme poor Poor Nonpoor

0

5

10 15 20 Impact (percentage points)

25

0

10 20 Impact (percentage points)

30

Source: Maluccio and flores 2005.

136

the iMpaCt of CCt prograMs on the aCCuMulation of huMan Capital

Effects on the Utilization of Preventive Health Services regular use of preventive health care services is a requirement of many CCt programs, and we next turn to a discussion of program effects on utilization rates. as with education, there is a reasonably large number of evaluations that assess the impact of CCt programs on the use of preventive health services. With one exception (an evaluation of the CCt program in turkey), however, all evaluations are of latin american programs. some evaluations have found that program beneficiaries make more use of health services than they would have made in the absence of the intervention, but that is apparent only for some outcomes (such as growth monitoring for children) and generally not for others (such as immunization rates). table 5.2 summarizes results from various evaluations that estimate the effects of CCt programs on preventive health checkups for children. Most of these evaluations suggest there were positive program effects on growth and development monitoring visits to health centers by children. for nicaragua’s rps, for example, Maluccio and flores (2005) report a borderline-significant impact of 13 percentage points on the probability that a child aged 0–3 had been taken to a health center and weighed in the last 6 months. Macours, schady, and

Table 5.2

Impact of CCTs on Health Center Visits by Children, Various Years
Age range (years) 0–6 0–1 2–4 4+ Baseline level (%)a Impactb 17.6 n.a. n.a. n.a. n.a. 2.4 (2.7) 22.8*** (6.7) 33.2*** (11.5) 1.5* (0.8) 2.7 (3.8) 10 r paxson and schady (2008) Transfer (% of Evaluation PCE)c method Reference 7 17 rdd psM, dd galasso (2006) attanasio et al. (2005)

Country Chile

Program Chile solidario

Outcome regular checkups Child taken to growth and development monitoring

Colombia familias en acción

ecuador

bono de desarrollo humano

Child had growth 3–7 control in last 6 months

continued

137

Conditional Cash transfers: reduCing present and future poverty

Table 5.2 continued
Age range (years) 0–3 Baseline level (%)a Impactb 44.0 20.2*** (4.7) Transfer (% of Evaluation PCE)c method Reference 9 r Morris, flores, et al. (2004)

Country

Program

Outcome Child taken to health center at least once in past month number of visits to health center for preventive reasons in past 6 months number of visits to all health facilities in past month Child weighed in last 6 months Child taken to health center at least once in past 6 months Child taken to health center and weighed in past 6 months

honduras programa de asignación familiar jamaica program of advancement through health and education oportunidades

0–6

0.205

0.278*** 10 (0.085)

rdd

levy and ohls (2007)

Mexico

0–2 3–5 0–6

0.219 0.221 70.5

–0.032 (0.037) 0.027 (0.019) 6.3*** (2.0) 8.4 (5.9)

20

r

gertler (2000)

nicaragua atención a Crisis nicaragua red de protección social

18

r

Macours, schady, and vakis (2008) Maluccio and flores (2005)

0–3

69.8

27

r

0–3

55.4

13.1* (7.5)

Source: authors’ calculations. Note: dd = difference-in-differences; n.a. = not available; pCe = per capita expenditure; psM = propensity score matching; r = randomized; rdd = regression discontinuity design. this table contains weighted means for the coefficients for Chile, combining rural and urban estimates. the standard error in this case is the square root of the averaged variances of these estimates. a. the unit for baseline level corresponds to the proportion of children who have been taken to the health center, with the exception of jamaica and Mexico, where the unit corresponds to the number of visits. b. the column for “impact” reports the coefficient and standard error (in parentheses); the units are percentage points, with the exception of jamaica, where the unit is the number of visits to the health center in the past six months, and Mexico, where the unit is the number of visits to the health center in the past month. c. the transfer amounts as a proportion of per capita expenditures (or consumption) are not the same across all tables in the report because of differences in the surveys used, including their coverage and year. * significant at the 10 percent level. *** significant at the 1 percent level.

138

the iMpaCt of CCt prograMs on the aCCuMulation of huMan Capital

vakis (2008) report a significant increase of 6 percentage points in the probability that a child aged 0–6 had been weighed in the last 6 months (among beneficiaries of the atención a Crisis program). even larger effects are reported by attanasio et al. (2005) for the familias en acción program in Colombia, and by Morris, flores et al. (2004) for praf in honduras. in jamaica, levy and ohls (2007) also report significant effects of path on the number of preventive health care visits by children under the age of 6. on the other hand, there do not appear to be significant effects on preventive health care visits by children for the Chile solidario program (galasso 2006), for the ecuador bdh program (paxson and schady 2008), or for beneficiaries of oportunidades (gertler 2000). table 5.3 presents comparable evidence on the effects of CCt programs on immunization coverage. the effects are mixed. barham (2005b) uses the randomized design in oportunidades to estimate program effects on the coverage of tuberculosis and measles immunization rates. the program effects she estimates are small and not significant—something she attributes to the high immunization rates (around 90 percent) at baseline. in Colombia, attanasio et al. (2005) find positive program effects for familias en acción on immunization rates, although the effects generally are not significant. in honduras, praf appears to have increased coverage of immunization for diphtheria/pertussis/tetanus, but not for measles. barham and Maluccio (2008) find large impacts of the rps on full vaccination coverage in nicaragua. in turkey, too, the CCt program increased vaccination coverage significantly: the fraction of children under the age of 6 who have all required immunizations is 14 percentage points higher among those who participated in the CCt program. to conclude this section, we briefly discuss the evidence on changes in the use of preventive health care services by adults. in honduras, Morris, flores et al. (2004) use a differences-in-differences strategy to show that the fraction of women who reported five or more antenatal visits increased by 19 percentage points more in the randomly assigned praf treatment group than in the control group.4 on the other hand, path in jamaica appears to have had no effect on the use of preventive health care services by the elderly. the oportunidades program effects in Mexico are small and not significant at conventional levels, whereas the impact of the Chile solidario program on checkups by the elderly is not consistent with a positive program impact. 139

Conditional Cash transfers: reduCing present and future poverty

Table 5.3 Impact of CCTs on Vaccination and Immunization Rates, Various Years
Age range 48 months honduras programa de asignación familiar Child vaccinated with dpt Baseline levela n.a. n.a. n.a. Transfer (% of PCE)c 17 Evaluation method psM, dd

Country

Program

Outcome Compliance with dpt vaccination

Impactb 8.9* (4.7) 3.5 (2.6) 3.2 (3.9) 6.9*** (3.0) 4.2 (7.1) –0.2 (4.7) 1.6 (2.4) 2.8 (2.8) 13.0 (9.0) 18.0*** (5.0) 13.6*** (4.2)

Reference attanasio et al. (2005)

Colombia familias en acción