WORLD DEVELOPMENT REPORT 2 ATTACKING POVERTY 22684 September 2000 -7-7 ~ 7 I WORLD DDEVELOPMENT REPORT /2ool ATTACKING POVERTY Cover Art Voyage to the New World, 1999. Mixed media on wood, by Manuel Cunjama, Mexico. The images used to open the chapters are also taken from this painting and from three others by Cunjami: Dialogue with the Universe, Magic Kite, and Looking for the Cosmic Balance. Manuel Cunjama was born in 1971 in Tuxtla Gutierrez, Chiapas. He began exhibiting his work in Chiapas in 1992. Currently, Ana Quijano Gallery in Mexico City is repre- senting Cunjamr in traveling exhibits in the United States. Cunjama's work is included in the collection of the World Bank art program. In my work I seek and try to incorporate native symbols and elements that would take me to prehispanic issues and all that I consider useflu to this effect I show it in my work. I use the sun, the moon, the night, and the universe as symbols representing the worry and anxiety of the human beingefor the wholeness surrounding him: The infinite. -Manuel Cunjama The World Bank Art Program makes particular efforts to identify artists from develop- ing nations and make their work available to a wider audience. The art program orga- nizes exhibits, educational and cultural partnerships, competitions, artists' projects, and site-specific installations. WORLD DEVELOPMENT REPORT 2a/2001 a _~~~~~~~~~~~~~~~~~~~~~E ATTACKING POVERTY Published for the WORLD BANK OXFORD UNIVERSITY PRESS Oxford University Press OXFORD NEX' YORK ATHENS AUCKLAND BANGKOK BOGOTA BUENOS AIRES CALCUTTA CAPE TOWN CHENNAI DAR ES SALA,AM DELHI FLORENCE HONG KONG ISTANBUL KARACHI KUALA LUMPUR MADRID MELBOURNE MEXICO CITY MUMBAI NAIROBI PARIS SAO PAULO SINGAPORE TAIPEI TOKYO TORONTO WVARSAW and associated companies in BERLIN IBADAN ©D 2001 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, N.W., Washington, D.C. 20433, U.S.A. Cover design and chapter openers by Tomoko Hirata. Published by Oxford University Press, Inc. 200 Madison Avenue, New York, N.Y. 10016 Oxford is a registered trademark of Oxford University Press. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Manufactured in the United States of America First printing September 2000 This volume is a product of the staff of the World Bank, and the judgments made herein do not necessarily reflect the views of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. ISBN 0- 19-521598-2 clothbound ISBN 0-19-521129-4 paperback ISSN 0163-5085 Text printed on recycled paper that conforms to the American Standard for Permanence of Paper for Printed Library Material Z39.48- 1984. Foreword IlPoverty amid plenty is the world's greatest chal- to them, it expands this definition to include power- lenge. We at the Bank have made it our mission to fight lessness and voicelessness, and vulnerability and fear. poverty with passion and professionalism, putting it at These dimensions of human deprivation emerged the center of all the work we do. And we have recog- forcefully from our Voices ofthe Poor study, conducted nized that successful development requires a compre- as background for the report, which systematically hensive, multifaceted, and properly integrated mandate. sought the views of more than 60,000 men and women This report seeks to expand the understanding of living in poverty in 60 countries. poverty and its causes and sets out actions to create a These different dimensions of poverty interact in world free of poverty in all its dimensions. It both builds important ways. So do interventions to improve the on our past thinking and strategy and substantially well-being of poor people. Increasing education leads broadens and deepens what we judge to be necessary to better health outcomes. Improving health increases to meet the challenge of reducing poverty. It argues that income-earning potential. Providing safety nets allows major reductions in human deprivation are indeed poor people to engage in higher-risk, higher-return ac- possible, and that the forces of global integration and tivities. And eliminating discrimination against women, technological advance can and must be harnessed to ethnic minorities, and other disadvantaged groups serve the interests of poor people. Whether this occurs both directly improves their well-being and enhances will depend on how markets, institutions, and societies their ability to increase their incomes. function-and on the choices for public action, glob- The 20th century saw great progress in reducing ally, nationally, and locally. poverty and improving well-being. In the past four The report accepts the now established view of decades life expectancy in the developing world in- poverty as encompassing not only low income and con- creased 20 years on average, the infant mortality rate fell sumption but also low achievement in education, more than half, and fertility rates declined by almost half health, nutrition, and other areas of human develop- In the past two decades net primary school enrollment ment. And based on what people say poverty means in developing countries increased by 13 percent. Between v VI WORLD DEVELOPMENT REPORT 2000/2001 1965 and 1998 average incomes more than doubled in de- ucation) and increasing the returns on these assets, veloping countries, and in 1990-98 alone the number of through a combination of market and nonmarket people in extreme poverty fell by 78 million. actions. But at the start of a new century, poverty remains a X Facilitating empowerment: Making state institutions global problem of huge proportions. Of the world's 6 bil- more accountable and responsive to poor people, lion people, 2.8 billion live on less than $2 a day, and 1.2 strengthening the participation of poor people in po- billion on less than $1 a day. Six infants of every 100 do litical processes and local decisionmaking, and re- not see their first birthday, and 8 do not survive to their moving the social barriers that result from distinctions fifth. Of those who do reach school age, 9 boys in 100, of gender, ethnicity, race, religion, and social status. and 14 girls, do not go to primary school. Enhancingsecurity: Reducing poor people's vulnerability These broad trends conceal extraordinary diversity in to ill health, economic shocks, crop failure, policy- experience in different parts of the world-and large induced dislocations, natural disasters, and violence, variations among regions, with some seeing advances, and as well as helping them cope with adverse shocks others setbacks, in crucial nonincome measures of poverty. when they occur. A big part of this is ensuring that ef- Widening global disparities have increased the sense of fective safety nets are in place to mitigate the impact deprivation and injustice for many. And social mobility of personal and national calamities. and equal opportunity remain alien concepts for far too Advances in the three areas are fundamentally many people. complementary-each is important in its own right and Future demographic changes will add to the chal- each enhances the others. Drawing on this framework, lenge we face in further reducing poverty. In the next 25 countries need to develop their own poverty reduction years roughly 2 billion people will be added to the world's strategies, in a manner consistent with preservation of population-almost all of them (an estimated 97 percent) culture. Decisions on priorities must be made at the na- in developing countries, putting tremendous pressure tional level, reflecting national priorities. But action on these societies. Clearly, much must be done to reduce must also take place with local leadership and ownership, poverty in its multiple dimensions and to promote human reflecting local realities. There is no simple, universal freedom, today and in the years ahead. blueprint. While current and future challenges remain daunt- Action at the local and national levels is not enough, ing, we enter the new millennium with a better under- however. The evidence of the past decade vividly reveals standing of development. We have learned that traditional the importance of global action, both to ensure that the elements of strategies to foster growth-macroeconomic opportunities from global integration and technological stability and market-friendly reforms-are essential for advance benefit poor people and to manage the risks of reducing poverty. But we now also recognize the need insecurity and exclusion that may result from global for much more emphasis on laying the institutional and change. Five actions are key: social foundations for the development process and on Promoting global financial stability and opening the managing vulnerability and encouraging participation to markets of rich countries to the agricultural goods, ensure inclusive growth. And while domestic action is manufactures, and services of poor countries. critical, we have also learned that global developments Bridging the digital and knowledge divides, thus exert a potent influence on processes of change at national bringing technology and information to people and local levels-and that global action is central to throughout the world. poverty reduction. We have taken a fresh look at our work t Providing financial and nonfinancial resources for in- through the Comprehensive Development Framework, ternational public goods, especially medical and agri- which converges with the views and findings of this re- cultural research. port. v Increasing aid and debt relief to help countries take Based on its analysis of ideas and experience, this re- actions to end poverty, within a comprehensive frame- port recommends actions in three areas: work that puts countries themselves-not external Promoting opportunity. Expanding economic oppor- aid agencies-at the center of the design of develop- tunity for poor people by stimulating overall growth ment strategy and ensures that external resources are and by building up their assets (such as land and ed- used effectively to support the reduction of poverty. FOREWORD VI:[ * Giving a voice to poor countries and poor people in viewed in this report show that there is now sufficient un- global forums, including through international links derstanding to make actions to reduce poverty truly effective. with organizations of poor people. We are living in a time in which the efforts and issues sur- Public action must be driven by a commitment to rounding poverty reduction are subject to great scrutiny; poverty reduction. The public and private sectors must work In the aftermath of protests and in the midst of contro- together-along with civil society-both within and be- versy, this report offers real substance to the public debate tween countries. While we have much to learn, and while and brings the dialogue to the foreground, where indeed the world continues to change rapidly, the experiences re- the goal of a world without poverty belongs. James D. Wolfensohn President The World Bank August 2000 This report has been prepared by a team led by Ravi Kanbur (director of the team until May 2000) and Nora Lustig (deputy di- rector until May 2000 and director since). Monica Das Gupta, Christiaan Grootaert, Victoria Kwakwa, Christina Malmberg Calvo, and Kevn Morrison served as full-time team members. Other core team members included Alice Sindzingre, Michael Woolcock, and Zainal Yusof Major contributions to chapters were made by Homi Kharas, Aart Kraay, Peter Lanjouw, and Giovanna Pren- nushi as well as by Benu Bidani, William Easterly, Enrique Flores, Hlene Grandvoinnet, Richard Newfarmer, Gi-Taik Oh, and Mattia Romani. Michael Walton, in his capacity as director, Poverty Reduction, worked closely with the team throughout the process. Shanka Chakraborty and Shahin Yaqub assisted the team. Interns from the Washington Center provided other valuable assistance. The work was carried our under the general direction of Jozef Ritzen and Joseph E. Stiglitz and, in the final stages, Nicholas Stern. The World Bank Development Data Group was responsible for the Selected World Development Indicators. Bruce Ross-Larson and Meta de Coquereaumont were the principal editors of the report. The team was advised by Anthony Atkinson, Anthony Bebbington, Nancy Birdsall, Francois Bourguignon, Angus Deaton, Alain de janvry, Yujiro Hayami, Emmanuel Jimenez, Grzegorz Kolodko, Michael Lipton, Lant Pritchett, Mar- tin Ravallion, Amarrya Sen, Lyn Squire, T. N. Srinivasan, and Mariano Tommasi. Deepa Narayan led the Voices ofthe Poor study. Vinod Thomas led the Quality of Growth study, which complements the work of this report. Ariel Fiszbein led the organization of the consultation process. Much insight was gained from the papers and discussions at the Sum- mer Workshop held in Washington, D.C., on 6-1 0 July 1999, as well as from the background papers prepared for this report. The authors, commentators, and participants, as well as the background papers, are listed in the bibliographic note. Many others inside and outside the World Bank provided helpful comments and other contributions; their names are also listed in the bibliographic note. A wide range of consultations with academics, grassroots leaders, NGOs, private sector representatives, and policymakers were undertaken for this report, from initial outline to final draft. Thanks go to all the public institutions and civil society organizations that helped make these consultations possible, as well as the organizers and participants, whose contributions greatly enriched this report. Particular thanks go to the Bretton Woods Project and the New Policy Institute in the United Kingdom, which moderated the electronic discussion on the Web draft, and to all who submitted comments: 424 contri- butions were posted from 44 countries (44 percent from developing countries). Consultations were held in Argentina (Buenos Aires), Bangladesh (Dhaka), Canada (Ottawa), Chile (Santiago), Denmark (Copenhagen), Egypt (Cairo), Ethiopia (Addis Ababa), France (Paris), Germany (Berlin), Guatemala (Antigua), Hungary (Budapest), India (Ahmedabad, New Delhi), Japan (Tokyo), Malaysia (Kuala Lumpur), Morocco (Marrakech), Russia (Moscow), Senegal (Dakar), South Africa Jo- hannesburg), Sweden (Stockholm), the United Kingdom (Brighton, London), the United States (Boston, New York, Wash- Vill WORLD DEVELOPMENT REPORT 2000/2001 ington, D.C.), and Vietnam (Hanoi). Consultations were also held at and in conjlunction with the International Monetary Fund, the regional development banks, and several United Nations organizations. The preparation of background papers and the convening of several workshops were supported by the governments of Canada, Denmark, France, Germany, Japan, the Netherlands, Sweden, Switzerland, the United Kingdom, and the United States; the MacArthur Foundation; Cornell University; and the Development Policy Forum of the German Foundation for International Development. The Voices of the Poor study was supported by the MacArthur Foundation, Cornell University, the U.K. Department for International Development, and the Swedish International Development Coopera- tion Agency; and several of the country studies were partially or fully financed by the NGOs that conducted the research. Rebecca Sugui served as executive assistant to the team, and Maribel Flewitt, Shannon Hendrickson, Khin-U Khine, Rudeewan Laohakittikul, Jimena Luna, Nelly Obias, Gracie Ochieng, Leila Search, and Robert Simms as team assistants. Maria D. Ameal served as administrative officer. Nacer Megherbi and Edith Thomas provided technical support. Book copyediting and production were by Fiona Blackshaw, Garrett Cruce, Terry Fischer, Wendy Guyette, Daphne Levitas, Molly Lohman, Megan Klose, Jessica Saval, and Alison Strong of Communications Development. Book design, editing, and production were managed byjamila Abdelghani, Catherine Hudson, Brett Kravitz, Nancy Lammers, Brenda Mejia, Randi Park and Betty Sun of the of the World Bank's Office of the Publisher. Artemis Zenetou, manager and cu- rator of the World Bank Art Program, and her staff were instrumental in the use of the artwork in the book. Tomoko Hi- rata designed the cover and the chapter openers. Ravi Kanbur is T. H. Lee Professor of World Affairs in the Department ofAgricultural, Resource, and Managerial Economics, Cornell University, Ithaca, New York. Nora Lustig is senior advisor and chief of the Poverty and Inequality Advisory Unit of the Inter-American Development Bank. Kevin Morrison is a fellow at the Overseas Development Council. Alice Sitidz- ingre is a researcher at the Centre National de la Recherche Scientifique (CNRS) in Paris and an associate researcher at the Centre d'Etude d'Afrique Noire (CEAN) in Bordeaux. Zainal Yusof is deputy director general of the Institute of Strategic and International Studies (ISIS) in Kuala Lumpur, Malaysia. The World Bank is grateful to the six institutions for mak- ing possible their participation in the preparation of the report. The many contributors to the report may not endorse every viewpoint or assertion in it, and they bear no responsibil- ity for any errors that remain. Contents Overview Attacking Poverty: Opportunity, Em powverment, arid Secuirity I Basrabai's story 2 Poverty in an unequal world 3 A strategy for poverty reduction 6 From strategy to action 7 Working together to fight poverty 12 Part I Framework Chapter 1 The Nature and Evolution of Poverty 15 Poverty's many dimensions 15 Measuring poverty in its multiple dimensions 16 The evolution of poverty 21 Chapter 2 Causes of Poverty and a Framework for Action 31 The causes of poverty 34 A framework for action 37 Part 11 Opportunity Chapter 3 Growth, Inequality, and Poverty 4 Economic growth and poverty reduction 46 What drives economic growth? 49 Why are similar rates of growth associated with different rates of poverty reduction? 52 Economic growth and nonincome poverty 57 Chapter 4 Making Markets Work Better for Poor Pople 6o Have market reforms delivered growth? 62 Have market reforms delivered benefits to poor people? 64 Making markets do more for poor people 72 Chapter 5 Expanding Poor People's Assets aride TackSiq Inequaboes 77 Assets and their synergies 77 Public action to facilitate the accumulation of assets 78 Redistributing public spending 80 Institutional reforms for effective delivery: governance, markets, and competition 85 Participation: choice, monitoring, and accountability 88 Complementarities in public action 93 lx X WORLD DEVELOPMENT REPORT 2000/2001 Part 1II Empowerment Chapter 6 Making State institutions More Responsiv/e to Poor People 99 Pubtic administration and poverty reduction 99 Poor people and the rule of law 102 How can decentralization be made pro-poor? 106 The politics of poverty reduction: pro-poor coalitions 108 Political regimes and poverty 112 Chapter 7 Removing Social Barriers and BuiiId'ng Sci_ai Institutions 117 Gender discrimination and poverty 117 Social stratification and poverty 123 Social fragmentation and conflict 126 Building social institutions and social capital 128 Part IV Security Chapter 8 Helpirg Poor PeopIe Manage Risk 135 A typology of risks 136 The nature and magnitude of vulnerability 138 Responses to risk by households and communities 140 Policy responses for improving risk management 146 C'haptexi 9 Managing Eco,nomic Crises anda ?Jaturai Disstefs 161 Preventing and coping with economic crises 161 Reducing vulnerability to natural disasters 170 Part V International Actions ' 1 if',aessng Global Forces for Poor PeoplF 179 Expanding market access in high-income countries 179 Reducing the risk of economic crises 180 Producing pro-poor international public goods 181 Ensuring a voice for poor people in global forums 185 Chapler 1R Ieforrwng i Development Cooperation to Attack Poverty 189 Making aid more effective in reducing poverty 192 Relieving the debt burden of poor countries 200 Bibfinqraphic Note 205 Se-ecl eri VVorid Development Indicators 267 5 oxe$ I The voices of the poor 3 2 A better world for all: international development goals 5 1.1 Poverty in the voices of poor people 16 1.2 Measuring income poverty: 1899 and 1998 17 CONTENTS Xi 1.3 Measuring vulnerability 20 1.4 Measuring voice and power using participatory methods 21 1.5 Measuring governance: participatory methods and cross-country surveys 21 1.6 Multidimensionality: dealing with aggregation 22 1.7 Uganda's poverty reduction goals 22 1.8 Tracking poverty in India during the 1990s 26 2.1 On interacting with state institutions: the voices of the poor 36 2.2 Preventing famines: the local press matters 40 2.3 Attacking poverty in Vietnam 41 3.1 Population, growth, and poverty 49 3.2 How war devastates poor people 50 3.3 Divergence and worldwide income inequality 51 3.4 Inequality trends and poverty reduction 53 3.5 What makes growth pro-poor in India? 54 3.6 Complex patterns of distributional change in three economies 55 3.7 Diversification and migration in rural China 56 3.8 Redistribution can be good for efficiency 57 4.1 The Washington consensus (3 4.2 Why do reforms sometimes fail? 65 4.3 Agricultural reforms in Chile and China help small farmers 68 4.4 Land markets and poor peasants in Mexico 68 4.5 Listening to farmers in Zambia 69 4.6 Attacking poverty with information 73 5.1 Interactions between human and physical capital 78 5.2 Links between the environment and health 78 5.3 Win-win policies in the health sector 79 5.4 Locked out by health and education fees 83 5.5 Mexico's Progresa: paying parents to send children to school 84 5.6 Some general principles on how to design subsidies 85 5.7 Effective public-private partnership in immunization 87 5.8 Local entrepreneurs increase access to telecommunications services 87 5.9 Local participation in Nicaragua's decentralized education system 89 5.10 Single-sector and multisector arrangements for improving rural roads in Zambia 91 5.11 Rejuvenating India's forests through joint action 92 5.12 A new approach to land reform in Brazil 94 5.13 West African businesses pioneer water and sanitation services for the urban poor 95 5.14 Sustainable water and sanitation for Brazil's urban poor 96 6.1 Poor people are often harassed by public officials 100 6.2 Lawlessness contributes to poverty 103 6.3 Legal service organizations help poor people gain access to the protections of the legal system 105 6.4 Community monitoring can reduce environmental pollution 107 6.5 National coalitions against communicable diseases in the West 109 6.6 NGOs can help mobilize and empower communities 110 6.7 Politics and poverty in OECD countries 1 14 6.8 The evolution of civil society and state reform in Mexico 1.15 7.1 Making land titling less gender biased in Latin America 121 7.2 Using subsidies to close gender gaps in education 122 Xii WORLD DEVELOPMENT REPORT 2000/200I Boxes {continued) 7.3 Toward stronger female voice in policymaking: women's budget initiatives in southern Africa 122 7.4 Using development programs to break the power of agrarian elites: a case study from eastern Uttar Pradesh, India 123 7.5 Discrimination is psychologically devastating 125 7.6 Using affirmative action against caste-based discrimination in India 126 7.7 Ethnic divisions and civil conflict 127 7.8 How does social capital affect development? 129 7.9 The federation of comedores in Peru: the creation of linking social capital 130 7.10 Mobilizing and creating social capital in development projects 131 8.1 Poor people's exposure to risk 137 8.2 AIDS and poverty 139 8.3 Some key terms: risk, risk exposure, and vulnerability 139 8.4 Managing risk: the modular approach to social safety nets 147 8.5 Is targeting by the community a good idea? 150 8.6 Insurance options for the informal sector 151 8.7 Two universal health insurance systems: Costa Rica and Singapore 153 8.8 Social pensions in Chile and Namibia 154 8.9 Principles of successful workfare programs 156 8.10 The Eritrean Community Development Fund 157 8.11 The Self-Employed Women's Association of India 157 9.1 Providing social protection in response to crisis in the Republic of Korea 167 9.2 Public expenditure reviews to assess the impact of fiscal retrenchment on poor people 168 9.3 Protecting poor people during fiscal adjustment: Peru's Fiscal Prudence and Transparency Law 169 9.4 Mitigation is the cornerstone of emergency management in the United States 173 9.5 Mitigating the risks of natural catastrophes: lessons from the 1999 earthquakes in Turkey 173 9.6 Turning reconstruction into risk mitigation with the help of a local NGO 174 9.7 7 Mitigating risk with catastrophe bonds 174 9.8 Sharing the costs of catastrophes: the Mexican fund for natural disasters 175 9.9 Involving communitities in postdisaster reconstruction: lessons from the Maharashtra Emergency Earthquake Rehabilitation Program 176 10.1 A success story: the fight against river blindness in Africa 182 10.2 Research, maize, and pigs in rural Guizhou 184 10.3 Most biotechnology patents are private 185 10.4 The Global Environment Facility: a model for developing country participation 187 11.1 Learning about the consultative process through the Comprehensive Development Framework 195 11.2 The new poverty reduction strategy initiative 195 11.3 Sectorwide development cooperation 197 11.4 The common pool for development cooperation 197 11.5 Assessing country policies and institutions 198 11.6 How aid can help in countries with a weak policy environment 199 11.7 The Enhanced Heavily Indebted Poor Countries Debt Relief Initiative 201 11.8 How debt relief fits into a poverty reduction strategy: Uganda's Poverty Action Fund 204 Figures 1 Where the developing world's poor live 4 2 Where poverty has fallen, and where it has not 4 3 Infant mortality rates vary widely across the world 4 CONTENTS XIII 1.1 Poverty in the developing world is shifting toward South Asia and Sub-Saharan Africa 24 2.1 Poverty shows a strong link with economic contractions and expansions 35 3.1 In general, the wealthier a country, the lower the incidence of poverty 46 3.2 Health indicators improve as incomes rise 47 3.3 Poverty trends tracked growth trends in the 1980s and 1990s 47 3.4 Economic growth was a force for poverty reduction in the 1980s and 1990s ... 48 3.5 Inequality varied widely in the 1980s and 1990s but showed no systematic association with growth 53 3.6 Initial inequalities influence the pace of poverty reduction 5,4 3.7 Across countries, the ratio of female to male literacy rises with per capita income 5l8 4.1 Indications of successful policy reforms in the developing world 63 4.2 Reforms delivered growth in Latin America, although the gains varied 64 4.3 The gap between skilled and unskilled wages widened in Mexico 71 6.1 Merit-based recruitment in government is associated with less corruption and bureaucratic delay 101 6.2 Corruption is a regressive tax 102 6.3 Better rule of law is associated with higher per capita income 103 6.4 Decentralization lowers the cost of raising the income of poor people in South Africa 106 6.5 State-community coalitions can foster rapid development and better service delivery 112 6.6 Good political and administrative institutions go hand in hand with economic growth 112 7.1 Closing the gender gap in schooling more rapidly would boost economic growth 119 7.2 Trends in female education and life expectancy reflect increasing equality between women and men 120 7.3 Minority groups in Vietnam have less access to services than nonminorities 124 7.4 Ethnic diversity is associated with violence where institutional quality is low 128 8.1 Private transfers represent a large share of the income of the poor 144 8.2 Central government spending on social security varied greatly in 1995 148 9.1 Developing countries bore the brunt of natural disasters in 1990-98 171 10.1 High-income countries protect manufacturing and agriculture 180 10.2 The burden of HIV/AIDS is heavily concentrated in Sub-Saharan Africa 132 11.1 While donor countries' economies grew after 1992, their development assistance shrank 190 11.2 With the exception of Europe and Central Asia, the regional distribution of official development assistance remained roughly constant ... 190 11.3 . . . while receipts fell after 1992-93 in all regions but Latin America and the Caribbean 191 11.4 Aid does not go only to poor countries 196 11.5 NGOs channeled some $10 billion to developing countries in 1998, about half of it from the official sector 200 11.6 As per capita income in the heavily indebted poor countries has gone down, debt has gone up-and vice versa 201 11.7 Concessional transfers largely compensate for negative net transfers of nonconcessional resources 202 Tables 1.1 Income poverty by region, selected years, 1987-98 23 1.2 Relative income poverty by region, selected years, 1987-98 24 1.3 Income poverty in seven African countries, various years 25 4.1 Impact of reforms on agricultural prices, output, and productivity in seven countries 67 5.1 Public spending on education by income quintile in selected developing countries, various years 80 5.2 Public spending on health by income quintile in selected developing countries, various years 81 8.1 Main sources of risk 136 8.2 Shocks faced by rural households in Ethiopia 140 8.3 Mechanisms for managing risk 141 XiV WORLD DEVELOPMENT REPORT 2000/2001 8.4 Income diversification among African farmers 142 8.5 Private cash and in-kind transfers for poor households 144 9.1 Effect of economic crises on incidence of poverty in selected countries 163 9.2 Social impacts of economic crises in selected countries 164 Definitions and data notes The countries included in regional groupings in this report tries may be used as a matter of convenience to denote the are listed in the Classification of Economies table at the end high-income economies. of the Selected World Development Indicators. The income Aggregate poverty measures in the report are often based groupings in the main text of the report are based on 1998 on the "$1 a day" poverty line. This line is equal to $1.08 a GNP per capita estimates, while those in the Selected World day in 1993 purchasing power parity terms (for further ex- Development Indicators are based on 1999 GNP per capita planation see box 1.2 in chapter 1). estimates (see the Classification of Economies table). The Dollar figures are current U.S. dollars, unless otherwise thresholds for the income classifications and the differences specified. Billion means 1,000 million; trillion means 1,000 between the 1998 and 1999 classifications of countries may billion. be found in the introduction to the Selected World Devel- opment Indicators. Group averages reported in the figures The following abbreviations are used: and tables of the main text are unweighted averages of the countries in the group unless noted to the contrary. AIDS Acquired immune deficiency syndrome In addition to the changes in income classifications, re- GATT General Agreement on Tariffs and Trade cent revisions to data in the Selected World Development In- GDP Gross domestic product dicators may result in differences with the data in the main GNP Gross national product text (see the introduction to the Selected World Development HIPC Heavily indebted poor country Indicators). HIV Human immunodeficiency virus The use of the word countries to refer to economies im- NGO Nongovernmental organization plies no judgment by the World Bank about the legal or other OECD Organisation for Economic Co-operation status of a territory. The term developing countries includes and Development low- and middle-income economies and thus may include PPP Purchasing power parity economies in transition from central planning, as a matter SEWA Self-Employed Women's Association of convenience. The terms developed, industri4 or rich coun- WTO World Trade Organization OVERVIEW Attacking Poverty: Opportunity, Empowerment, and Security il. oor people live without fundamental freedoms Basrabai's story serves as a backdrop to the exploration of action and choice that the better-off take for of the nature and causes of poverty and of what can be granted.' They often lack adequate food and shelter, done. Poverty is the result of economic, political, and education and health, deprivations that keep them social processes that interact with each other and fre- from leading the kind of life that everyone values. They quently reinforce each other in ways that exacerbate the also face extreme vulnerability to ill health, economic deprivation in which poor people live. Meager assets, in- dislocation, and natural disasters. And they are often accessible markets, and scarce job opportunities lock peo- exposed to ill treatment by institutions of the state and ple in material poverty. That is why promoting society and are powerless to influence key decisions opportunity-by stimulating economic growth, mak- affecting their lives. These are all dimensions of ing markets work better for poor people, and building poverty. up their assets-is key to reducing poverty. The experience of multiple deprivations is intense But this is only part of the story. In a world where and painful. Poor people's description of what living political power is unequally distributed and ofien in poverty means bears eloquent testimony to their pain mimics the distribution of economic power, the way (box 1). For those who live in poverty, escaping it can state institutions operate may be particularly unfa- seem impossible. But it is not impossible. The story vorable to poor people. For example, poor people fre- of Basrabai-the chair of a local council in an Indian quendy do not receive the benefits of public investment village-illustrates both the many facets of poverty and in education and health. And they are often the vic- the potential for action (see page 2). tims of corruption and arbitrariness on the part of the I 2 WORLD DEVELOPMENT REPORT 2000/200I Basrabai lives in Mohadi, a village 500 kilometers from Ahmed- fishers seemed to be paying the price. The big trawlers could con- abad, in the Indian state of Gujarat, on the shores of the Ara- tinue to fish as long as they paid the right officials. bian Sea 2 She is the first woman to be sarpanch of the lnthemiddleofthemeetingacommotionoccurredattheside. panchayat-chair of the ocal council-as a result of constitutional Basrabai's brother had been gored in the face while trying to sep- amendments that reserve a third of local council seats and a third arate two fighting cows. Without immediate treatment the wound of headships for women. was bound to become infected. But it was late at night, and the Arriving in her village after a long drive, we crossed a small nearest doctor was in the next big settlement, 10 kilometers sea inlet on a road impassable at high tide. The first building away. Normally, this would have made immediate treatment im- we saw was a recently built concrete structure-the primary possible. As it happened, however, our Jeep was there and could school. In last year's cyclone, the worst in living memory, as take Basrabai's brother to the doctor. the villagers' straw huts were blown away, they took shelter During our stay we also saw the craft work that the village in the only stabie structure in the village-the school. When women have been doing for generations. Demand for their tra- the cyclone relief operation arrived, the villagers asked for ditional embroidered and tie-dyed products is high, thanks to the more concrete buildings, and the village now has about a international love affair with things Indian and the rediscovery dozen of them. by the growing Indian middle class of its roots. But the traders We arrived at Basrabai's one-room concrete house, next to get away with offering very low prices because of the women's a straw hut. After the usual greetings, talk turned to the school. isolation. Since it was a weekday, we wondered if we could sit in on a The national and state governments have countless schemes class. Basrabai informed us that the master (the teacher) was to support traditional crafts, none very effective. So SEWA is step- not there and had not been there for a while. In fact, he came ping in to organize the home-based craft workers and to provide only once a month, if that. Protected by the district education direct access to international markets. One piece of embroidery officer, he did pretty much what he pleased. we looked at would fetch 150 rupees in the international market, The master came the next day. Word had gotten to him that 60 rupees in government outlets, and 20 rupees from traders. the village had visitors. He came into Basrabai' s house, and we On the last day of our stay we went to Basrabai's field, an hour's began talking about the school and the children. Believing the walk from her house. The risks of agriculture were plainly visible. educated guests to be kindred spirits, he launched into a litany The lack of rain had left the ground hard and dry. If it didn't rain of his troubles and the difficulties of teaching the children. He in the next few days, her millet crop would be lost, and with it referred to them as jong/ee-from the jungle." her outlay to a hired tractor driver to till her field, an investment This was too much for Meeraiben, a member of the Self- made possible by the sale of her crafts. When we met her in Employed Women's Association (SEWA), who had arranged our Ahmedabad days later, it still had not rained. visit. She pointed out that his salary was 6,000 rupees a month The interactions with Basrabai and the many thousands of (more than six times the Indian poverty Ine) in a secure job and poor people consulted in preparing this report bring to the fore that his responsibility was at least to show up for work. The par- recurrent-and familiar-themes. Poor people mention the lack ents wanted their children to learn to read and write, even if at- of income-earning opportunities, the poor links with markets, and tending school meant that the boys could not help their fathers the failure of state institutions to respond to their needs. They with fishing and the girls could not help their mothers fetch water mention insecurity, such as health risks, the risk of being out and wood and work in the fields. of work, and the agricultural risks that make any gains always Later in the even ng Basrabai conducted the village meeting. fragile. Everywhere-from the villages in India to the favelas of There were two main topics. The first was compensation for the Rio de Janeiro, the shantytowns outside Johannesburg, and the cyclone: despite the great fanfare with which relief schemes had farms in Uzbekistan-the stories bring forward similar issues. been announced in the state capital, local delivery left much to be But talking to Basrabai and other poor people also reveals what desired and local officials were unresponsive. SEWA organizers is possible. Although local officials and state structures are still took down the names of those who had not yet received the com- not accountable to Basrabai and her village, an explicit affirma- pensation to which they were entitled, and it was agreed that they tive action policy allowed Basrabai's election as sarpanch, show- and Basrabai would meet with local officals the following week. ing what can be done through state action. And SEWA shows The second issue was a fishing ban that the government had how poor people can make a difference if they organize them- imposed on coastal waters to protect fish stocks. It was the big selves to defend their rights, take advantage of market oppor- trawlers that were responsible for the overfishing, but the small tunities, and protect themselves from risks. ATTACKING POVERTY: OPPORTUNITY, EMPOWERMENT, AND SECURITY 3 Box 1 The voices of the poor The Voices of the Poor study, based on realities of more than scarce, there is no money and we feel poor. If there were 60,000 poor women and men in 60 countries, was conducted as money ... background for World Development Report 2000/2001. It consists -From a discussion group of poor men and women, Ecuador of two parts: a review of recent participatory poverty studies in 50 countries involving about 40,000 poor people, and a new compar- We face a calamity when my husband falls ill. Our life comes ative study in 1999 in 23 countries engaging about 20,000 poor peo- to a halt until he recovers and goes back to work. ple. The study shows that poor people are active agents in their -Poor woman, Zawyet Sultan, Egypt lives, but are often powerless to influence the social and eco- nomic factors that determine their well-being. Poverty is humiliation, the sense of being dependent on them, The following quotations are an illustration of what living in and of being forced to accept rudeness, insults, and indiffer- poverty means. ence when we seek help. -Poor woman, Latvia Don't ask me what poverty is because you have met it outside my house. Look at the house and count the number of holes. At first I was afraid ofeveryone and everything: my husband, Look at the utensils and the clothes I am wearing. Look at the village sarpanch, the police. Today I fear no one. I have my everything and write what you see. What you see is poverty, own bank account, I am the leader of my village's savings -Poor man, Kenya group ... I tell my sisters about our movement. And we have Certainly our farming is little; all the products, things bought a 40,000-strong union in the district from stores, are expensive; it is hard to live, we work and -From a discusswon group earn little money, buy few things or products; products are of poor men and women, India Source: Narayan, Chambers, Shah, and Petesch 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 2000. state. Poverty outcomes are also gready affected by social birthday, while in the poorest countries as many as a fifth norms, values, and customary practices that, within the of children do not. And while in rich countries fewer than family, the community, or the market, lead to exclusion 5 percent of all children under five are malnourished, in of women, ethnic and racial groups, or the socially dis- poor countries as many as 50 percent are. advantaged. That is why facilitating the empowerment This destitution persists even though human concii- of poor people-by making state and social institutions tions have improved more in the past century than in the more responsive to them-is also key to reducing poverty. rest of history-global wealth, global connections, and Vulnerability to external and largely uncontrollable technological capabilities have never been greater. But the events-illness, violence, economic shocks, bad weather, distribution of these global gains is extraordinarily un- natural disasters-reinforces poor people's sense of ill- equal. The average income in the richest 20 countries is being, exacerbates their material poverty, and weakens their 37 times the average in the poorest 20-a gap that has bargaining position. That is why enhancing security- doubled in the past 40 years. And the experience in dif- by reducing the risk of such events as wars, disease, eco- ferent parts of the world has been very diverse (figure 2; nomic crises, and natural disasters-is key to reducing see also table 1.1 in chapter 1). In East Asia the number poverty. And so is reducing poor people's vulnerability to of people living on less than $1 a day fell from around risks and putting in place mechanisms to help them cope 420 million to around 280 million between 1987 and with adverse shocks. 1998-even after the setbacks of the financial crisis.3 Yet in Latin America, South Asia, and Sub-Saharan AfrSica Poverty in an unequal world the numbers ofpoor people have been rising. And in the countries of Europe and Central Asia in transition to mar- The world has deep poverty amid plenty. Of the world's ket economies, the number of people living on less than 6 billion people, 2.8 billion-almost half-live on less than $1 a day rose more than twentyfold.4 $2 a day, and 1.2 billion-a fifth-live on less than $1 a There have also been major advances and serious set- day, with 44 percent living in South Asia (figure 1). In rich backs in crucial nonincome measures of poverty. India lhas countries fewer than 1 child in 100 does not reach its fifth seen marked progress in girls attending school, and in the 4 WORLD DEVELOPMENT REPORT 2000/2001 Figure 1 Figure 2 Where the developing world's poor live Where poverty has fallen, and where it has not Distribution of population living on less than S1 a day, Change in number of people living on less than S1 a day, 199811.2 billion) 1987-98 Europe and Central Asia Middle East and North Africa 100 Millions 2.0%/ 0.5% Latin America and 75 the Caribbean 50 25 East Asia and Pacific South Asia 0 23.2% 43.5% -25 -50 -75 -100 Su -Saharan -125 Africa 24.3% -150 East Asia Europe Latin Middle South Sub- Source: World Bank 2000s. and and America East and Asia Saharan Pacific Central and the North Africa Asia Caribbean Africa most advanced state, Kerala, life expectancy is greater than Source: World Bank 2000s. in other places with many times the level of income (such as Washington, D.C.). Yet in countries at the cen- ter of the HIV/AIDS epidemic in Africa, such as Botswana and Zimbabwe, one in four adults is infected, AIDS or- Figure 3 phans are becoming an overwhelming burden on both Infant mortality rates vary widely across the traditional and formal support mechanisms, and all the world gains in life expectancy since the middle of the 20th will soon be wiped out. The varying infant mor- Infant mortality rate, 1998 century e u.Te ayn nfn o-Per 1,000 live births taltty rates across the world-Sub-Saharan Africa's is 15 100 times that of high-income countries-give an idea of 90 this widely differing experience (figure 3). 80 Experiences are also vastly different at subnational 70 levels and for ethnic minorities and women. Different re- gions in countries benefit to very different extents from 60 growth. In Mexico, for example, total poverty fell- 50 though modestly-in the early 1990s, but rose in the 40 poorer Southeast. Inequalities also exist across different 30 ethnic groups in many countries. In some African coun- 20 tries infant mortality rates are lower among politically pow- 10 10 erful ethnic groups, and in Latin American countries 0 indigenous groups often have less than three-quarters East Europe Latin Middle South Sub- High- the schooling on average of nonindigenous groups. And Asia and America East and Asia Saharan income be disadvantaged thn In Iand Central and the North Africa countries women continue to be more disadvantaged than men. In Pacific Asia Caribbean Africa South Asia women have only about half as many years of education as men, and female enrollment rates at the Source: World Bank 2000s. secondary level are only two-thirds the male rates. ATTACKING POVERTY: OPPORTUNITY, EMPOWERMENT, AND SECURITY '; Box 2 A better world for all: international development goals Reduce the proportion of People living on less than The goals for international development people living in extreme $t a day (%) 30 address that most compelling of human W poverty by half between Poress 1990-98 desires-a world free of poverty and free 1990 and 2015 20 of the misery that poverty breeds. Each of the seven goals addresses an Average path to goal''*- ., aspect of poverty. They should be viewed 1990 2015 together because they are mutually re- inforcing. Higher school enrollments, es- Enroll all children in primary Net primary enrollment rate (%) pecially for girls, reduce poverty and school by 2015 100 mortality. Better basic health care in- _ Average path to goal --' creases enrollment and reduces poverty. ''i P'rogress 1990-98 75 Many poor people earn their living from the environment. So progress is needed 50 on each of the seven goals. 1990 2015 In the past decade on average the world has not been on track to achieve Make progress toward Ratio of girls to boys in primary the goals. But progress in some countries gender equality and and secondary school (%) 1 and regions shows what can be done. _ empowering women ,__________ °__ by eliminating gender Average path to go,al .-.-.-- - China reduced its number in poverty from disparities in primary and Progres 1990 98 75 360 million in 1990 to about 210 million secondary education by 2005 in 1998. Mauritius cut its military budget 50 and invested heavily in health and edu- 19902005 cation. Today all Mauritians have access to sanitation, 98 percent to safe water, Reduce infant and child Under-5 mortality rate and 97 percent of births are attended by mortality rates by two-third (per 1,000 live births) loO skilled health staff. And many Latin Amer- w between 1990 and 201 5 _ ________ Progress 1990-98 ican countries moved much closer to " gender equality in education. The message: if some countries can Avrg ah to goa - - 0 make great progress toward reducing 1990 2015 poverty in its many forms, others can as well. But conflict is reversing gains in so- O Reduce maternal mortality Births attended by skilled health cial development in many countries in ratios by three-quarters personnel a%) 10d Sub-Saharan Africa. The spread of between 1990 and 201 5 Averagepathtogoal - - HIV/AIDS is impoverishing individuals, ,.--''- - - -- - 50 families, and communities on all conti- Progress 1988-98 nents. And sustained economic growth- 0 that vital component for long-run 1988 2015 reductions in poverty-still eludes half the world's countries. For more than 30 ~ Provide access for all who Contraceptive prevalencerate of them, real per capita incomes have ET;_ need reproductive health _ services by 2015 80 fallen over the past 35 years. And where Progress 1993-98 there is growth, it needs to be spread more equally. 6 00 The goals can be met-with a combi- 50 nation of effective domestic and interna- 1993 1998 tional actions. O Implement national Countries with environmental strategies for sustainable strategies 1Ml _ development by 2005 so as so to reverse the loss of _ _8_ environmental resources by Progres 25 2015 1984 1997 Note: Data are for low- and middle-income countries except for those on environmental strategies, which refer to all countries. Source: IMF, OECD, United Nations, and World Bank 2000 (www.paris2l .org/betterworldh. 6 WORLD DEVELOPMENT REPORT 2000/2001 Faced with this picture of global poverty and in- liver more health services, and ensuring that technolog- equality, the international community has set itself sev- ical progress in the medical field spills over to benefit the eral goals for the opening years of the century, based on developing world.5 And meeting the gender equality discussions at various United Nations conferences in the goals in education will require specific policy measures 1990s (box 2). These international development goals, to address the cultural, social, and economic barriers most for 2015, include reducing income poverty and that prevent girls from attending school.6 Furthermore, human deprivation in many dimensions (the bench- actions to ensure greater environmental sustainability marks are figures for 1990): will be crucial in augmenting the assets available to poor * Reduce by half the proportion of people living in ex- people and in reducing the long-term incidence of treme income poverty (living on less than $1 a day). poverty.7 These actions will all interact to push toward * Ensure universal primary education. the achievement of the goals. Hence the need for a * Eliminate gender disparity in primary and secondary broader, more comprehensive strategy to fight poverty. education (by 2005). * Reduce infant and child mortality by two-thirds. A strategy for poverty reduction • Reduce maternal mortality by three-quarters. * Ensure universal access to reproductive health services. The approach to reducing poverty has evolved over the * Implement national strategies for sustainable devel- past 50 years in response to deepening understanding of opment in every country by 2005, so as to reverse the the complexity of development. In the 1950s and 1960s loss of environmental resources by 2015. many viewed large investments in physical capital and in- These will have to be achieved in a world whose popu- frastructure as the primary means of development. lation will grow by some 2 billion in the next 25 years, with In the 1970s awareness grew that physical capital was 97 percent of that increase in developing countries. Stud- not enough, and that at least as important were health and ies of what must be done to achieve these goals reveal the education. WorldDevelopmentReport 1980 articulated this magnitude of the challenge. For example, cutting income understanding and argued that improvements in health and poverty by half between 1990 and 2015 would require a education were important not only in their own right but compound rate of decline of 2.7 percent a year over those also to promote growth in the incomes of poor people. 25 years. The World Bank's latest estimates indicate a re- The 1980s saw another shift of emphasis following the duction of approximately 1.7 percent a year between 1990 debt crisis and global recession and the contrasting ex- and 1998. Much of the slow progress observed in some re- periences of East Asia and Latin America, South Asia, and gions is due to low or negative growth. In some cases ris- Sub-Saharan Africa. Emphasis was placed on improving ing inequality compounded this effect; this was particularly economic management and allowing greater play for so in some countries in the former Soviet Union. The cur- market forces. World Development Report 1990: Poverty rent pace of educational enrollment is unlikely to bring uni- proposed a two-part strategy: promoting labor-intensive versal primary education, especially in Sub-Saharan Africa. growth through economic openness and investment in Reducing infant mortality rates by two-thirds between infrastructure and providing basic services to poor peo- 1990 and 2015 would have required a 30 percent decline ple in health and education. between 1990 and 1998, far greater than the 10 percent de- In the 1990s governance and institutions moved to- veloping countries experienced. In some parts of Sub-Sa- ward center stage-as did issues of vulnerability at the local haran Africa infant mortality is actually on the rise, partly and national levels. This report builds on the earlier as a result of the AIDS epidemic. And maternal mortality strategies in the light of the cumulative evidence and ex- ratios are declining too slowly to meet the goals. perience of the past decade-and in the light of the Attaining the international development goals will changed global context. It proposes a strategy for at- require actions to spur economic growth and reduce in- tacking poverty in three ways: promoting opportunity, come inequality, but even equitable growth will not be facilitating empowerment, and enhancing security. enough to achieve the goals for health and education. Re- * Promoting opportunity. Poor people consistently em- ducing infant and child mortality rates by two-thirds phasize the centrality of material opportunities. This depends on halting the spread of HIV/AIDS, increasing means jobs, credit, roads, electricity, markets for their the capacity of developing countries' health systems to de- produce, and the schools, water, sanitation, and health ATTACKING POVERTY: OPPORTUNITY, EMPOWERMENT, AND SECURITY 7 services that underpin the health and skills essential for There is no hierarchy of importance. The elements are work. Overall economic growth is crucial for generat- deeply complementary. Each part of the strategy affects urn- ing opportunity. So is the pattern or quality of growth. derlying causes of poverty addressed by the other two. For Market reforms can be central in expanding opportu- example, promoting opportunity through assets and mar- nities for poor people, but reforms need to reflect local ket access increases the independence of poor people and institutional and structural conditions. And mechanisms thus empowers them by strengthening their bargaining need to be in place to create new opportunities and com- position relative to state and society. It also enhances secu- pensate the potential losers in transitions. In societies rity, since an adequate stock of assets is a buffer against ad- with high inequality, greater equity is particularly im- verse shocks. Similarly, strengthening democratic institutions portant for rapid progress in reducing poverty. This re- and empowering women and disadvantaged ethnic and racial quires action by the state to support the buildup of groups-say, by eliminating legal discrimination against human, land, and infrastructure assets that poor peo- them-expand the economic opportunities for the poor and ple own or to which they have access. socially excluded. Strengthening organizations of poor peo- F Facilitating empowerment. The choice and implemen- ple can help to ensure service delivery and policy choices tation of public actions that are responsive to the needs responsive to the needs of poor people and can reduce cor- of poor people depend on the interaction of political, ruption and arbitrariness in state actions as well. And if poor social, and other institutional processes. Access to mar- people do more in monitoring and controlling the local de- ket opportunities and to public sector services is often livery of social services, public spending is more likely ito strongly influenced by state and social institutions, help them during crises. Finally, helping poor people cope which must be responsive and accountable to poor peo- with shocks and manage risks puts them in a better posi- ple. Achieving access, responsibility, and accountability tion to take advantage of emerging market opportunities. is intrinsically political and requires active collabora- That is why this report advocates a comprehensive ap- tion among poor people, the middle class, and other proach to attacking poverty. groups in society. Active collaboration can be greatly fa- cilitated by changes in governance that make public ad- From strategy to action ministration, legal institutions, and public service delivery more efficient and accountable to all citizens-and by There is no simple, universal blueprint for implementing strengthening the participation of poor people in political this strategy. Developing countries need to prepare their processes and local decisionmaking. Also important is own mix of policies to reduce poverty, reflecting national removing the social and institutional barriers that result priorities and local realities. Choices will depend on tlhe from distinctions of gender, ethnicity, and social status. economic, sociopolitical, structural, and cultural context Sound and responsive institutions are not only impor- of individual countries-indeed, individual communities. tant to benefit the poor but are also fundamental to the While this report proposes a more comprehensive ap- overall growth process. proach, priorities will have to be set in individual cases based * Enhancing security. Reducing vulnerability-to on resources and what is institutionally feasible. Progress economic shocks, natural disasters, ill health, dis- in reducing some aspects of deprivation is possible even ability, and personal violence-is an intrinsic part of if other aspects remain unchanged. For example, inex- enhancing well-being and encourages investment in pensive oral rehydration campaigns can significantly re- human capital and in higher-risk, higher-return ac- duce infant mortality, even if incomes of poor people do tivities. This requires effective national action to man- not change.8 But actions will generally be necessary in all age the risk of economywide shocks and effective three clusters-opportunity, empowerment, and security-- mechanisms to reduce the risks faced by poor people, because of the complementarities among the three. including health- and weather-related risks. It also re- The actions of developed countries and multilateral or- quires building the assets of poor people, diversify- ganizations will be crucial. Many forces affecting poor peo- ing household activities, and providing a range of ple's lives are beyond their influence or control. Developing insurance mechanisms to cope with adverse shocks- countries cannot on their own produce such things as iin- from public work to stay-in-school programs and ternational financial stability, major advances in health and health insurance. agricultural research, and international trading opportu- 8 WORLD DEVELOPMENT REPORT 2000/2001 nities. Actions by the international community and de- had major reductions in income poverty have made use of velopment cooperation will continue to be essential. international trade. But opening to trade can create losers Here are the suggested areas for action, first national as well as winners, and it will yield substantial benefits only and then international. when countries have the infrastructure and institutions to underpin a strong supply response. Thus the opening Opportunity needs to be well designed, with special attention to coun- The core policies and institutions for creating more op- try specifics and to institutional and other bottlenecks. The portunities involve complementary actions to stimulate sequencing of policies should encourage job creation and overall growth, make markets work for poor people, and manage job destruction. A more pro-poor liberalization is build their assets-including addressing deep-seated in- not necessarily a slower one; moving fast can create more equalities in the distribution of such endowments as opportunities for the poor. And explicit policies should off- education. set transitory costs for poor people, as the grants for small Encouraging effectiveprivate investment. Investment and Mexican maize producers did in the wake of the North technological innovation are the main drivers of growth American Free Trade Agreement (NAFTA). in jobs and labor incomes. Fostering private investment re- The opening of the capital account has to be managed quires reducing risk for private investors-through stable prudently-in step with domestic financial sector fiscal and monetary policy, stable investment regimes, development-to reduce the risk of high volatility in sound financial systems, and a clear and transparent busi- capital flows. Long-term direct investment can bring ness environment. But it also involves ensuring the rule of positive externalities, such as knowledge transfer, but law and taking measures to fight corruption-tackling short-term flows can bring negative externalities, partic- business environments based on kickbacks, subsidies for ularly volatility. Policies need to address them separately. large investors, special deals, and favored monopolies. Building the assets of poor people. Creating human, Special measures are frequently essential to ensure physical, natural, and financial assets that poor people own that microenterprises and small businesses, which are or can use requires actions on three fronts. First, increase often particularly vulnerable to bureaucratic harassment the focus of public spending on poor people in particu- and the buying of privilege by the well-connected, can lar, expanding the supply of basic social and economic participate effectively in markets. Such measures include services and relaxing constraints on the demand side ensuring access to credit by promoting financial deepening (through, for example, scholarships for poor children). and reducing the sources of market failure; lowering the Second, ensure good quality service delivery through in- transactions costs of reaching export markets by expanding stitutional action involving sound governance and the use access to Internet technology, organizing export fairs, of markets and multiple agents. This can imply both re- and providing training in modern business practices; forming public delivery, as in education, or privatizing and building feeder roads to reduce physical barriers. in a fashion that ensures expansion of services to poor peo- Creating a sound business environment for poor house- ple, as often makes sense in urban water and sanitation. holds and small firms may also involve deregulation and Third, ensure the participation of poor communities complementary institutional reform, for example, reducing and households in choosing and implementing services restrictions on the informal sector, especially those affecting and monitoring them to keep providers accountable. women, and tackling land tenure or registry inadequacies This has been tried in projects in El Salvador, Tunisia, and that discourage small investments. Uganda. Programs to build the assets of poor people Private investment will have to be complemented by include broad-based expansion of schooling with parental public investment to enhance competitiveness and cre- and community involvement, stay-in-school programs ate new market opportunities. Particularly important is (such as those in Bangladesh, Brazil, Mexico, and Poland), complementary public investment in expanding infra- nutrition programs, mother and child health programs, structure and communications and upgrading the skills vaccinations and other health interventions, and of the labor force. community-based schemes to protect water resources Expanding into international narkets. International mar- and other elements of the natural environment. kets offer a huge opportunity for job and income growth- There are powerful complementarities between actions in agriculture, industry, and services. All countries that have in different areas. Because of close linkages between ATTACKING POVERTY: OPPORTUNITY, EMPOWERMENT, AND SECURITY 9 human and physical assets, for example, improving poor etal benefits of pro-poor public action and build politi- people's access to energy or transport can increase their cal support for such action. access and returns to education. And improving the en- Laying the political and legal basis for inclusive devel- vironment can have significant effects on poverty. This opment. State institutions need to be open and account- is well documented in terms of the substantial gains in able to all. This means having transparent institutions, health from reduced air and water pollution-which with democratic and participatory mechanisms for mak- have a major influence on some of the most important ing decisions and monitoring their implementation, diseases of poor people, including diarrheal problems of backed up by legal systems that foster economic growth children and respiratory infections. and promote legal equity. Since poor people lack the re- Addressing asset inequalities across gender, ethnic, racial, sources and the information to access the legal system, and social divides. Special action is required in many so- measures such as legal aid and dissemination of infor- cieties to tackle socially based asset inequalities. Although mation on legal procedures-for example, by the Ain-O- political and social difficulties often obstruct change, Salish Kendra (ASK) organization in Bangladesh-are there are many examples of mechanisms that work, using especially powerful instruments for creating more inclu- a mix of public spending, institutional change, and par- sive and accountable legal systems. ticipation. One is negotiated land reform, backed by Creatingpublic administrations thatfoster growth and public action to support small farmers, as in Northeast equity. Public administrations that implement policies ef- Brazil and the Philippines. Another is getting girls into ficiently and without corruption or harassment improve school, such as by offering cash or food for schooling, as service delivery by the public sector and facilitate growth in Bangladesh, Brazil, and Mexico, and hiring more fe- of the private sector. Appropriate performance incentives male teachers, as in Pakistan. A third is support for mi- are needed to make public administrations accountable crocredit schemes for poor women. and responsive to users. Access to information such as bud- Gettinginfiastructureand knowledge topoor areas-rural gets, participatory budget mechanisms, and performance and urban. Special action is also needed in poor areas, where rating of public services all enhance citizens' capacity to a combination of asset deprivations-including at the shape and monitor public sector performance while re- community or regional level-can diminish the material ducing opportunities and scope for corruption. Reform- prospects for poor people. Tackling this again requires pub- ing public administrations and other agencies such as the lic support and a range of institutional and participatory police to increase their accountability and responsiveness approaches. It requires providing social and economic to poor people can have a major impact on their daily lives. infrastructure in poor, remote areas, including transport, Promoting inclusive decentralization and community de- telecommunications, schools, health services, and electricity, velopment. Decentralization can bring service agencies closer as in China's poor areas programs. It also requires broad- to poor communities and poor people, potentially en- based provision of basic urban services in slums, within hancing people's control of the services to which they are an overall urban strategy. Also important is expanding ac- entitled. This will require the strengthening of local capacity cess to information for poor villages, to allow them to par- and devolution of financial resources. It is also necessary to ticipate in markets and to monitor local government. have measures to avoid capture by local elites. Decentral- ization needs to be combined with effective mechanisms for Empowerment popular participation and citizen monitoring of government The potential for economic growth and poverty reduc- agencies. Examples include decentralization that fosters tion is heavily influenced by state and social institutions. community-driven choices for resource use and project Action to improve the functioning of state and social in- implementation. There is also a range of options for involving stitutions improves both growth and equity by reducing communities and households in sectoral activities-sach bureaucratic and social constraints to economic action and as parental involvement in schooling and users associa- upward mobility. However, devising and implementing tions in water supply and irrigation. these changes require strong political will, especially Promoting gender equity. Unequal gender relations are when the changes fundamentally challenge social values part of the broader issue of social inequities based on so- or entrenched interests. Governments can do much to in- cietal norms and values. But gender equality is of such per- fluence public debate to increase awareness of the soci- vasive significance that it deserves extra emphasis. While 10 WORLD DEVELOPMENT REPORT 2000/200I patterns of gender inequity vary greatly across societies, in Supportingpoorpeoples social capital. Social norms and almost all countries a majority of women and girls are dis- networks are a key form of capital that people can use to advantaged in terms of their relative power and control over move out of poverty. Thus it is important to work with material resources (in most countries land titles are vested and support networks of poor people and to enhance in men), and they often face more severe insecurities (for their potential by linking them to intermediary organi- example, after the death of their husband). Poor women zations, broader markets, and public institutions. Doing are thus doubly disadvantaged. Moreover, the lack of au- this also requires improving the legal, regulatory, and in- tonomy of women has significant negative consequences stitutional environments for groups representing poor for the education and health of children. people. Since poor people usually organize at the local level, Greater gender equity is desirable in its own right actions will also be needed to strengthen their capacity to and for its instrumental social and economic benefits for influence policy at the state and national levels, such as poverty reduction. There has been progress-for exam- by linking local organizations to wider organizations. ple, in education and health-but much more needs to be done. Experience indicates that a mix of political, Security legal, and direct public action is required. Thirty-two coun- Achieving greater security requires a heightened focus on tries, from Argentina to India, have measures to promote how insecurity affects the lives and prospects of poor peo- women's representation in local and national assemblies, ple. It also takes a mix of measures to deal with econo- and this is already transforming women's ability to par- mywide or regionwide risks and to help poor people ticipate in public life and decisionmaking. Some coun- cope with individual adverse shocks. tries are correcting gender biases in the law, as in the 1994 Formulating a modular approach to helping poor peo- Colombian Agrarian Law. Use of public resources to ple manage risk. Different interventions-at the com- subsidize girls' education has been shown to pay off in munity, market, and state levels-are needed to address Bangladesh and Pakistan. A range of measures in pro- different risks and different segments of the population. ductive activities, notably microfinance and farming in- A mix of interventions may be needed to support the man- puts, have produced documented benefits in terms of agement of risks for communities and households, de- increased yields (in Kenya, for example) and increased au- pending on the type of risk and the institutional capacity tonomy for women and better nutritional status of chil- of the country. Microinsurance programs can complement dren (in Bangladesh and in virtually every setting where microcredit programs for poor women, built around this issue has been examined). their organizations, as in the schemes SEWA runs in Tackling social barriers. Social structures and institutions India for women in the informal sector. Public work form the framework for economic and political relations schemes can expand in response to local or national and shape many of the dynamics that create and sustain shocks. Food transfer programs and social funds to help poverty-or alleviate it. Social structures that are exclu- finance projects identified by communities can also be sionary and inequitable, such as class stratification or gen- effective in coping with disaster. der divisions, are major obstacles to the upward mobility Developing nationalprograms toprevent, preparefor, and of poor people. Governments can help by fostering debate respond to macro shocks-financial and natural. Economy- over exclusionary practices or areas of stigma and by sup- wide shocks are often the hardest for poor communities porting the engagement and participation of groups rep- and households to cope with, especially when the shocks resenting the socially excluded. Groups facing active are repeated, deep, or persistent. To manage the risk of fi- discrimination can be helped by selective affirmative ac- nancial and terms of trade shocks, sound macroeconomic tion policies. Social fragmentation can be mitigated by policy and robust financial systems are fundamental. But bringing groups together in formal and informal forums they have to be complemented by prudent management and channeling their energies into political processes in- of the opening of the capital account, to reduce the risk stead of open conflict. Other actions could include removing of volatile short-run flows. Special measures are also needed ethnic, racial, and gender bias in legislation and the oper- to ensure that spending on programs important to poor ation of legal systems and encouraging the representation people-social programs and targeted transfers-does not and voice of women and disadvantaged ethnic and racial fall during a recession, especially relative to the rising need. groups in community and national organizations. Equally important, countercyclical safety nets should be ATTACKING POVERTY: OPPORTUNITY, EMPOWERMENT, AND SECURITY II permanent and ready to be deployed when countries are Tackling the HIVIAIDS epidemic. HIV/AIDS is already hit by a shock. These and other actions can also help in one of the most important sources of insecurity in severely coping with natural shocks. "Calamity funds" can finance afflicted countries in Africa. While the immediate, dev- relief efforts following natural disasters and support new astating effects are at the individual and household level, technology and training for better risk assessment. Mak- the consequences are much broader, from intolerable ing investments and insurance arrangements in normal strains on traditional child fostering mechanisms to times can reduce personal costs when a disaster occurs. extreme pressures on health systems and loss of produc- Designing national systems ofsocial risk management that tive labor affecting whole communities and nations. are also pro-growth. There is demand across the world for More than 34 million people are infected with HIV (90 national systems of social risk management. The challenge percent in the developing world), with 5 million more is to design them so that they do not undercut compet- infected each year. More than 18 million people have al- itiveness and so that poor people benefit. Some examples: ready died of AIDS-related illness. Action at the inter- systems that both provide insurance for the nonpoor national level to develop an AIDS vaccine is crucial for and include social pensions for the poor, as in Chile; health the future, but differing experiences show that what will insurance that protects against catastrophic illness that really make a difference now is effective leadership and could wipe out a family's assets, as in Costa Rica; and un- societal change to prevent the spread of HIV and care for employment insurance and assistance that do not com- those already infected. This can involve confronting promise the incentive to work. To gain the full benefits taboos about sexuality, targeting information and support of such schemes, however, economies need the institu- to high-risk groups such as prostitutes, and providing tional capacity to manage them effectively. compassionate care for AIDS sufferers. Brazil, Senegal, Addressing civil conflict. Civil conflict is devastating Thailand, and Uganda all illustrate what can be done when for poor people: the bulk of conflicts are in poor coun- there is a will to act decisively. tries and most are civil wars-more than 85 percent of all conflicts were fought within country borders between International actions 1987 and 1997. In addition to the direct loss of life, they Action at national and local levels will often not be enougJh wreak social and economic havoc and create a terrible legacy for rapid poverty reduction. There are many areas that r e- of psychological and social trauma. Child soldiers are quire international action-especially by industrial coun- often recruited to fight-as in Sierra Leone-and many tries-to ensure gains to poor countries and to poor more children suffer the loss of family, disruption of people within the developing world. An increased focus schooling, and psychological scars that permanently di- on debt relief and the associated move to make develop- minsh their prospects. ment cooperation through aid more effective are part of While it is immensely important to sustain the focus on the story. Of equal importance are actions in other areas-- rebuilding societies after conflict, such as in Cambodia trade, vaccines, closing of the digital and knowledge di- and Rwanda, it is equally urgent to take measures to pre- vides-that can enhance the opportunity, empowerment, vent conflict. There is some evidence that strengthening plu- and security of poor people. ralist institutions-supporting minority rights and providing Opportunity. Within a rule-based trading system, in- the institutional basis for peaceful conflict resolution- dustrial countries could expand opportunities by open- has a significant influence. Also important for averting ing their markets more completely to imports from poor conflict are efforts to get different groups to interact through countries, especially in agriculture, labor-intensive man- more inclusive and participatory political institutions and ufactures, and services. It has been estimated that OECD through civil institutions. As noted below, international ac- tariffs and subsidies cause annual losses in welfare of tion to reduce access to the resources to finance conflict and almost $20 billion in developing countries, equivalent to to reduce international trade in armaments is also neces- about 40 percent of aid in 1998. Many developing coun- sary. If countries can get onto a path of inclusive economic tries feel that while they are liberalizing their trade regimes, development, they have the potential to shift from a vicious key dimensions of the trade regimes of rich countries are to a virtuous cycle. Violent conflict constitutes one of the putting them at a disadvantage. Furthermore, donor most urgent and intractable areas for action affecting some countries could strengthen developing countries' ability of the poorest people in the world. to pursue poverty reduction, by increasing aid flows to I2 WORLD DEVELOPMENT REPORT ZOOO/200I countries with a sound policy environment supportive of Working together to fight poverty poverty reduction and by financing the Enhanced Heavily Indebted Poor Countries Debt Relief Initiative with The strategy in this report recognizes that poverty is funds additional to aid budgets. more than inadequate income or human development- Empowerment. Global action can empower poor it is also vulnerability and a lack of voice, power, and rep- people and poor countries in national and global forums. resentation. With this multidimensional view of poverty Aid should be delivered in ways that ensure greater own- comes greater complexity in poverty reduction strategies, ership by recipient countries, and it should go increas- because more factors-such as social and cultural forces- ingly to country-driven, results-oriented poverty need to be taken into account. reduction programs, developed with the effective en- The way to deal with this complexity is through gagement of civil society and private sector agents. Poor empowerment and participation-local, national, and people and poor countries should have greater voice in international. National governments should be fully international forums, to ensure that international accountable to their citizenry for the development path priorities, agreements, and standards-such as in trade they pursue. Participatory mechanisms can provide voice and intellectual property rights-reflect their needs and to women and men, especially those from poor and interests. excluded segments of society. The design of decentralized The international financial institutions and other in- agencies and services needs to reflect local conditions, ternational organizations should continue their efforts to social structures, and cultural norms and heritage. And ensure full transparency in their strategies and actions- international institutions should listen to-and promote- and open, regular dialogue with civil society organizations, the interests of poor people. The poor are the main ac- particularly those representing poor people. International tors in the fight against poverty. And they must be brought organizations should also support the ongoing global center stage in designing, implementing, and monitoring coalitions of poor people so that they may inform global antipoverty strategies. debates. Actions by multinational corporations, such as There is an important role in this for rich countries adhering to ethical investment practices and adopting and international organizations. If a developing coun- labor codes, can also empower poor groups. try has a coherent and effective homegrown program of Security. Actions are also needed to reduce risks from poverty reduction, it should receive strong support-to adverse international forces. Jointly with governments and bring health and education to its people, to remove the private sector, the international financial institu- want and vulnerability. At the same time global forces tions must strengthen the international financial archi- need to be harnessed for poor people and poor countries, tecture and improve its management to lessen economic so that they are not left behind by scientific and med- volatility, which can be devastating for poor people. In- ical advances. Promoting global financial and environ- dustrial country governments, often in cooperation with mental stability-and lowering market barriers to the the private sector, should also provide more support for products and services of poor countries-should be a core international public goods-for developing and dis- part of the strategy. tributing vaccines for HIV/AIDS, tuberculosis, and A divergent world? Or an inclusive one? A world with malaria and for producing and disseminating agricultural poverty? Or a world free of poverty? Simultaneous actions advances for tropical and semiarid conditions. to expand opportunity, empowerment, and security can International action to protect the environment can re- create a new dynamic for change that will make it pos- duce the harmful effects of environmental degradation, sible to tackle human deprivation and create just societies which can be severe in some poor countries. And the that are also competitive and productive. If the developing international community should seek to stem armed world and the international community work together conflict-which affects poor people the most-by taking to combine this insight with real resources, both finan- measures to reduce the international arms trade, promote cial and those embodied in people and institutions-their peace, and support physical and social reconstruction after experience, knowledge, and imagination-the 21st conflicts end. century will see rapid progress in the fight to end poverty. PART I Framework CHAPTER 1 The Nature and Evolution of Poverty I overty is pronounced deprivation in well-being. strict what Amartya Sen calls the "capabilities that a per- But what precisely is deprivation? The voices of poor son has, that is, the substantive freedoms he or she en- people bear eloquent testimony to its meaning (box joys to lead the kind of life he or she values. "1 1.1). To be poor is to be hungry, to lack shelter and This broader approach to deprivation, by giving a clothing, to be sick and not cared for, to be illiterate better characterization of the experience of poverty, in- and not schooled. But for poor people, living in poverty creases our understanding of its causes. This deeper tin- is more than this. Poor people are particularly vulnerable derstanding brings to the fore more areas of action and to adverse events outside their control. They are often policy on the poverty reduction agenda (chapter 2). treated badly by the institutions of state and society and Another important reason for considering a broader excluded from voice and power in those institutions. range of dimensions-and hence a broader range of policies-is that the different aspects of poverty interact Poverty's many dimensions and reinforce one another in important ways (chap- ter 2). This means that policies do more than simply This report accepts the now traditional view of poverty add up. Improving health outcomes not only im- (reflected, for example, in World Development Report proves well-being but also increases income-earning po- 1990) as encompassing not only material deprivation tential. Increasing education not only improves (measured by an appropriate concept of income or con- well-being-it also leads to better health outcomes and sumption) but also low achievements in education and to higher incomes. Providing protection for poor peo- health. Low levels of education and health are of con- ple (reducing vulnerability in dealing with risk) not only cern in their own right, but they merit special attention makes them feel less vulnerable-it also allows them when they accompany material deprivation. This re- to take advantage of higher-risk, higher-return op- port also broadens the notion of poverty to include vul- portunities. Increasing poor people's voice and par- nerabiity and exposure to risk-and voicelessness and ticipation not only addresses their sense of exdusion--it powerlessness. All these forms of deprivation severely re- also leads to better targeting of health and education 15 I6 WORLD DEVELOPMENT REPORT 2000/2001 Poverty in the voices of poor people Poor people in 60 countries were asked to analyze and share Alongside the material, physical well-being features promi- their ideas of well-being (a good experience of life) and "ill-being" nently in the characterizations of poverty. And the two meld to- (a bad experience of life). gether when lack of food leads to ill health-or when ill health leads Well-being was variously described as happiness, harmony, to an inability to earn income. People speak about the importance peace, freedom from anxiety, and peace of mind. In Russia peo- of looking well fed. In Ethiopia poor people say, "We are skinny," ple say, "Well-being is a life free from daily worries about lack of "We are deprived and pale," and speak of life that "makes you money." In Bangladesh, "to have a life free from anxiety." In Brazil, older than your age." "not having to go through so many rough spots." Security of income is also closely tied to health. But insecu- People describe ill-being as lack of material things, as bad ex- rity extends beyond ill health. Crime and violence are often men- periences, and as bad feelings about oneself. A group of young tioned by poor people. In Ethiopia women say, "We live hour to men in Jamaica ranks lack of self-confidence as the second biggest hour, " worrying about whether it will rain. An Argentine says, "You impact of poverty: "Poverty means we don't believe in self, we have work, and you are fine. If not, you starve. That's how it is." hardly travel out of the community-so frustrated, just locked up Two social aspects of ill-being and poverty also emerged. For many in a house all day." poor people, well-being means the freedom of choice and action and Although the nature of ill-being and poverty varies among lo- the power to control one's life. A young woman in Jamaica says that cations and people-something that policy responses must take poverty is "like living in jail, living in bondage, waiting to be free." into account-there is a striking commonality across countries. Not Linked to these feelings are definitions of well-being as social surprising, material well-being turns out to be very important. well-being and comments on the stigma of poverty. As an old Lack of food, shelter, and clothing is mentioned everywhere as woman in Bulgaria says, "To be well means to see your grand- critical. In Kenya a man says: "Don't ask me what poverty is be- children happy and well dressed and to know that your children cause you have met it outside my house. Look at the house and have settled down; to be able to give them food and money count the number of holes. Look at my utensils and the clothes I whenever they come to see you, and not to ask them for help and am wearing. Look at everything and write what you see. What you money." A Somali proverb captures the other side: "Prolonged sick- see is poverty." ness and persistent poverty cause people to hate you." Source. Narayan, Chambers, Shah, and Petesch 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 2000. services to their needs. Understanding these comple- by a century, Seebohm Rowntree's classic study of poverty mentarities is essential for designing and implement- in the English city ofYork in 1899 and the World Bank's ing programs and projects that help people escape current estimates of global income poverty share a com- poverty. mon approach and a common method (box 1.2). Based on household income and expenditure surveys, the ap- r vt'S frmffi (A tXi119nproach has become the workhorse of quantitative poverty - a.r1i.rc,,', 'l.~ analysis and policy discourse. It has several strengths. Because it is based on nationally representative samples, Measuring poverty permits an overview of poverty that it allows inferences about the conditions and evolution goes beyond individual experiences. It aids the formula- of poverty at the national level. Moreover, since house- tion and testing of hypotheses on the causes of poverty. hold surveys collect information beyond monetary income It presents an aggregate view of poverty over time. And or consumption, the approach makes it possible to ob- it enables a government, or the international community, tain a broader picture of well-being and poverty, inves- to set itself measurable targets for judging actions. In what tigate the relationships among different dimensions of follows, the chapter discusses the measurement of income poverty, and test hypotheses on the likely impact of pol- poverty and the indicators of education and health-and icy interventions. then turns to vulnerability and voicelessness. Poverty measures based on income or consumption are not problem free. Survey design varies between countries and over time, often making comparisons difficult. For Using monetary income or consumption to identify and example, some countries ask respondents about their measure poverty has a long tradition. Though separated food spending over the past month, while others do so THE NATURE AND EVOLUTION OF POVERTY ]7 Box 1.2 Measuring income poverty: 1899 and 1998 In a classic study first published in 1901, Seebohm Rowntree cal- tional income based on national accounts data. This procedure, un- culated that 10 percent of the population of the English city of York changed from past exercises, scales back income to obtain con- in 1899 was living in poverty (below minimum needed expendi- sumption but leaves the distribution unchanged. tures). As we enter the next century, the World Bank calculates Prices. To compare consumption levels across countries, es- that a fourth of the population of the developing world-about 1.2 timates of price levels are needed, and the World Bank's purchasing billion people-is living in poverty (below $1 a day). These two cal- power parity (PPP) estimates for 1993 were used. These estimates culations of income poverty are separated by a century and have are based on new price data generated by the International Com- very different coverage. Nevertheless, the basic concepts and meth- parison Program (ICP), which now covers 110 countries, up from ods they embody have strong similarities. 64 in 1985, and a more comprehensive set of commodities. Poverty lines. The 1990 calculations of the international poverty Rowntree's approach lines had to be updated using 1993 price data and the 1993 PPP Rowntree's method was to conduct a survey covering nearly estimates. In 1990 national poverty lines for 33 countries were con- every working-class family in York to collect information on earn- verted into 1985 PPP prices, and the most typical line among the ings and expenditures. He then defined poverty as a level of total low-income countries for which poverty lines were available was earnings insufficient to obtain the minimum necessities for the main- selected. In 1999 the same lines were converted into 1993 PPP tenance of " merely physical efficiency, " including food, rent, and prices, and the new line was obtained as the median of the 10 low- other items. He calculated that for a family of five-a father, est poverty lines. That line is equal to $1.08 a day in 1993 PPP terms mother, and three children-the minimum weekly expenditure to (referred to as "$1 a day" in the text). This line has a similar pur- maintain physical efficiency was 21 shillings, 8 pence; he proposed chasing power to the $1 a day line in 1985 PPP prices, in terms other amounts for families of different size and composition. of the command over domestic goods. The upper poverty line (re- Comparing these poverty lines with family earnings, he arrived at ferred to as "$2 a day") was calculated by doubling the amount his poverty estimate. of the lower poverty line, as in 1990, reflecting poverty lines more commonly used in lower-middle-income countries. The World Bank's approach Estimates for 7998. To obtain consumption levels for 1998 The World Bank has been estimating global income poverty figures where survey data were not yet available, estimated growth rates since 1990. The latest round of estimation, in October 1999, used of per capita private consumption from national accounts statistics new sample survey data and price information to obtain comparable were used to update consumption data from the latest survey year figures for 1987, 1990,1993, 1996, and 1998 (the figures for 1998 to 1998. This meant assuming that the distribution of consumption are preliminary estimates). The method is the same as in past es- did not change from the time of the last survey to 1998. The per timates (World Bank 1990,1996d). capita private consumption growth rates came from estimates Consumption. Poverty estimates are based on consumption based on the model used for other World Bank forecasts (World or income data collected through household surveys. Data for 96 Bank 1999j). Surveys were available for 1997 or 1998 only for Be- countries, from a total of 265 nationally representative surveys, larus, China, India, Jordan, Latvia, Nigeria, Pakistan, Panama, Rus- corresponding to 88 percent of the developing world's people, are sia, Thailand, and Yemen. So the 1998 figures should be considered now available, up from only 22 countries in 1990. Of particular note tentative, and trends should be interpreted cautiously, particularly is the increase in the share of people covered in Africa from 66 in light of the controversy surrounding Indian data (see box 1.8 later to 73 percent, a result of extensive efforts to improve household in the chapter). data in the region. Country-specificpovertylines. The $1 and $2 a day poverty es- Consumption is conventionally viewed as the preferred wel- timates described here are useful only as indicators of global fare indicator, for practical reasons of reliability and because con- progress, not to assess progress at the country level or to guide sumption is thought to better capture long-run welfare levels than country policy and program formulation. Country-specific poverty current income. Where survey data were available on incomes but lines, reflecting what it means to be poor in each country's situ- not on consumption, consumption was estimated by multiplying ation and not affected by international price comparisons, are all incomes by the share of aggregate private consumption in na- used in country-level analysis. Source: Chen and Ravallion 2000. for the past week. One-month recall data tend to result hold size and composition in converting household data in higher poverty estimates than one-week recall data. into measures for individuals. Poverty estimates are very Converting the information on income or consumption sensitive to these assumptions (see, for example, the dis- collected in household surveys into measures of well- cussion in box 1.8, later in the chapter).2 being requires many assumptions, such as in deciding how Moreover, income or consumption data collected at the to treat measurement errors and how to allow for house- household level have a basic shortcoming: they cannot re- i8 WORLD DEVELOPMENT REPORT 2000/2001 veal inequality within the household, so they can under- In comparing poverty estimates across countries or over state overall inequality and poverty. One study that dis- time, it is important to check the extent to which con- aggregated household consumption by individual members clusions vary with the selection of poverty measure.5 found that relying only on household information could lead to an understatement of inequality and poverty by Health and education more than 25 percent.3 In particular, the conventional Measuring deprivation in the dimensions of health and household survey approach does not allow direct mea- education has a tradition that can be traced back to such surement of income or consumption poverty among classical economists as Malthus, Ricardo, and Marx. De- women. That is one reason why data on education and spite Rowntree's primarily income-based approach to health, which can be collected at the individual level, are measuring poverty, he devoted an entire chapter of his so valuable-they allow a gender-disaggregated perspec- study to the relation of poverty to health and went on to tive on key dimensions of poverty. argue that the death rate is the best instrument for mea- A key building block in developing income and con- suring the variations in the physical well-being of peo- sumption measures of poverty is the poverty line-the ple.6 Classifying his sample into three groups ranging from critical cutoff in income or consumption below which an poorest to richest, he found that the mortality rate was individual or household is determined to be poor. The in- more than twice as high among the very poor as among ternationally comparable lines are useful for producing the best paid sections of the working classes. Calculating global aggregates of poverty (see box 1.2). In principle, they infant mortality, he found that in the poorest areas one test for the ability to purchase a basket of commodities child out of every four born dies before the age of 12 that is roughly similar across the world. But such a uni- months. According to this argument, mortality could be versal line is generally not suitable for the analysis of used as an indicator both of consumption poverty and poverty within a country. For that purpose, a country- of ill-being in a broader sense. specific poverty line needs to be constructed, reflecting the The tradition of measuring deprivation in health and country's economic and social circumstances. Similarly, the education is well reflected in the international develop- poverty line may need to be adjusted for different areas ment goals (see box 2 in the overview). But data on these (such as urban and rural) within the country if prices or nonincome indicators have their own problems. For ex- access to goods and services differs.4 The construction of ample, infant and under-five mortality rates derived country profiles based on these country-specific poverty mostly from census and survey information are available lines is now common practice. for most countries only at periodic intervals.7 A complete Once a poverty line has been specified, it remains to vital registration system would be the best source for be decided how to assess the extent of poverty in a par- mortality data, but such a system exists in only a few de- ticular setting. The most straightforward way to measure veloping countries. For the period between censuses or poverty is to calculate the percentage of the population surveys, estimates of vital rates are derived by interpola- with income or consumption levels below the poverty line. tion and extrapolation based on observed trends and This "headcount" measure is by far the most commonly models, such as life tables that estimate survival from one calculated measure of poverty. But it has decided disad- year to the next. Infant mortality rates are available for vantages. It fails to reflect the fact that among poor peo- most developing countries for only one year since 1990, ple there may be wide differences in income levels, with and the year differs because surveys are conducted at some people located just below the poverty line and oth- different times. The data situation is even worse for life ers experiencing far greater shortfalls. Policymakers seek- expectancy, because it is often not measured directly ing to make the largest possible impact on the headcount Education data are also far from satisfactory. The measure might be tempted to direct their poverty allevi- most commonly available indicator, the gross primary en- ation resources to those closest to the poverty line (and rollment rate, suffers from serious conceptual short- therefore least poor). comings. The greatest is that school enrollment is only Other poverty measures, which take into account the a proxy for actual school attendance. Moreover, the gross distance of poor people from the poverty line (the poverty primary enrollment rate can rise if grade repetitions in- gap) and the degree of income inequality among poor peo- crease. The much-preferred net primary enrollment rate ple (the squared poverty gap), can be readily calculated. (showing the ratio of enrolled primary-school-age children THE NATURE AND EVOLUTION OF POVERTY 19 to all primary-school-age children) is available for only Multidimensionality and measuring progress around 50 developing countries for 1990-97-not Defining poverty as multidimensional raises the ques- enough to make reliable aggregations by region. A num- tion of how to measure overall poverty and how to corn- ber of ongoing survey initiatives, however, are improv- pare achievements in the different dimensions. One ing the quantity and quality of data on health and dimension might move in a different direction from an- education. other. Health could improve while income worsens. Or an individual might be "income poor" but not "health Vulnerability poor." Or one country might show greater improvement In the dimensions of income and health, vulnerability is in health than in vulnerability-while another shows the the risk that a household or individual will experience an converse. episode of income or health poverty over time. But vul- This brings to the fore the relative value of the differ- nerability also means the probability of being exposed to ent dimensions: how much income are people willing to a number of other risks (violence, crime, natural disas- give up for, say, a unit of improvement in health or in voice? ters, being pulled out of school). In other words, what weights can be assigned to the dif- Measuring vulnerability is especially difficult: since the ferent dimensions to allow comparisons across countries, concept is dynamic, it cannot be measured merely by ob- households, or individuals and over time? There are no easy serving households once. Only with household panel answers. data-that is, household surveys that follow the same One approach to addressing comparability is to define households over several years-can the basic information a multidimensional welfare function or a composite be gathered to capture and quantify the volatility and vul- index. An alternative is to define as poor anybody who nerability that poor households say is so important. More- is poor in any one of the dimensions-without attempt- over, people's movements in and out of poverty are ing to estimate tradeoffs among the dimensions-or informative about vulnerability only after the fact. The chal- anybody who is poor in all dimensions, and to define the lenge is to find indicators of vulnerability that can iden- intensity of poverty accordingly (box 1.6). This report does tify at-risk households and populations beforehand. not try to define a composite index or to measure trade- Many indicators of vulnerability have been proposed offs among dimensions. Instead, it focuses on depriva- over the years, but there is now a growing consensus tion in different dimensions and, in particular, on the that it is neither feasible nor desirable to capture vul- multiple deprivations experienced by the income-poor. nerability in a single indicator. If the government provides This is a necessary first step in developing a comprehensive an effective workfare program, for example, households multidimensional framework. may do less than they otherwise would to diversify their How should indicators be selected to monitor progress? income or build up their assets. Similarly, a household The international development goals are a good starting that is part of a reliable network of mutual support may point. But in practice, these goals will have to be adapted see less need for large buffer stocks of food or cattle. So (by lengthening or shortening the time span, for exam- a vulnerability measure based solely on household assets- ple) and modified (increasing the number of dimen- or on income and its sources-may not reflect the house- sions), depending on context. The specific goals wil hold's true exposure to risk (box 1.3). have to emerge from a participatory process in which gov- ernments and civil society agree on priorities. This process Voicelessness and powerlessness is already under way in many countries, and multilateral Voicelessness and powerlessness can be measured using organizations are helping with resources and technical as- a combination of participatory methods (box 1.4), polls, sistance (box 1.7).8 and national surveys on qualitative variables such as the extent of civil and political liberties (box 1.5). However, Investing in measurement and monitoring measuring these dimensions of poverty in an accurate, ro- Measurements of poverty thus must cover many dimen- bust, and consistent way so that comparisons can be sions. So far, the income and consumption dimension has made across countries and over time will require con- received most attention. Thanks to efforts over the past siderable additional efforts on both the methodological 20 years by such international agencies as the United Na- and data-gathering fronts. tions, the World Bank, and the regional development 20 WORLD DEVELOPMENT REPORT 2000/2001 Box 1.3 Measuring vulnerability Since vulnerability is a dynamic concept, its measurement centers sessment of vulnerability should be based not only on the ob- on the variability of income or consumption or on the variability of served transfers but also on the household's expectation other dimensions of well-being, such as health or housing. In about the assistance it will receive in a crisis. It is this expec- much of the literature on risk, this variability is measured by the tation that determines the household's decisions about en- standard deviation or coefficient of variation of income or con- gaging in other risk management activities. Unfortunately, sumption. From the perspective of poor people, this measure is household surveys rarely include direct information on networks flawed in several ways: or on expectations of assistance. * It gives equal weight to upward and downward fluctuations, * Participation in the formal safety net. A household's vulnera- Yet poor people are concerned primarily with downward bility is reduced if it is entitled to social assistance, unem- fluctuations. ployment insurance, pensions, and other publicly provided * It has no time dimension. Given 10 fluctuations, the coefficient transfers-and if it can benefit from workfare programs, so- of variation is the same whether good and bad years alternate cial funds, and similar mechanisms. So information on such pro- or five bad years are followed by five good ones. Yet bunched grams and their rules of eligibility is also important in assessing downward fluctuations are more difficult for poor people to cope vulnerability and risk exposure. with. * Access to credit markets. Similarly, a household's vulnerabil- * A scenario with many small and one large fluctuation may yield ity is reduced if it has access to credit for consumption the same coefficient of variation as a scenario with equal mod- smoothing. erate fluctuations. Yet poor people are likely to be hurt more Clearly, assessing vulnerability is more complex than mea- by the first scenario. suring poverty at a point in time. The length of time over which The coefficient of variation is, moreover, a measure after the vulnerability is to be assessed is of great importance and may fact. Needed are indicators that make it possible to assess a well differ across people and circumstances. Conventional an- household's risk exposure beforehand-information both on the nual measures of income or consumption may often be too household and on its links to informal networks and formal safety long. Furthermore, measuring vulnerability requires data on nets: household assets (physical, human, and social capital) in com- * Physical assets. A household's physical assets-those that bination with data on formal safety nets, the functioning of mar- can be sold to compensate for temporary loss of income- kets, and the economic policies that determine a household's are a measure of its capacity to self-insure. What matters opportunity set and the range of activities it can pursue to man- is not just the total value of the assets, but also their liquidity. age risk. Many of today's household surveys do not provide the Thus knowledge of the functioning of asset markets is needed information. needed to determine the usefulness of the assets as Cross-sectional surveys need to expand their standard ex- insurance. penditure modules by adding questions on assets, links with net- * Human capital. Households with limited education tend to be works, perceptions of sources of emergency assistance, and more subject to income fluctuations and less able to manage participation in formal safety nets. One World Bank survey has taken risk-for example, through access to credit or multiple in- a step in this direction: the recent Local-Level Institutions Surveys come sources. combine asset data with detailed questions on households' links * Income diversification. The extent of diversification of income with local associations. Some Living Standards Measurement sources has often been used to assess vulnerability. In rural Surveys have also begun to incorporate modules on social capi- settings analysts might look at nonfarm income, which tends tal. Ultimately, such enriched cross-sectional surveys need to be to fluctuate less than farm income, thus providing a measure combined with panel surveys, monitoring the same households of protection against weather-related risks. But income diver- over time, to allow direct observation of how households deal with sification can be a misleading indicator of risk exposure. A sin- shocks. gle low-risk activity could be preferable to multiple high-risk Vulnerability to nonincome risks can be measured by the preva- activities that are strongly covariant. So more diversification is lence of these risks (crime, natural disasters, and so on) in special not necessarily less risky. Diversification needs to be evaluated modules of household surveys. A program sponsored jointly by the in the context of the household's overall risk strategy. Inter-American Development Bank, World Bank, and Economic * Links to networks. Family-based networks, occupation-based Commission for Latin America and the Caribbean, the Program for groups of mutual help, rotating savings and credit groups, and the Improvement of Surveys and the Measurement of Living Con- other groups or associations to which a household belongs- ditions (known as Mecovi for its Spanish acronym) is incorporat- all part of the household's social capital-can be a source of ing such modules in specific countries in Latin America (the transfers in cash or kind in the event of a calamity. An as- questionnaire can be found in IDB 2000). Source: Dercon 1999; Grosh and Glewwe 2000; Holzmann and Jorgensen 1999; IDB 2000; Sinha and Lipton 1999; World Bank 1998t. THE NATURE AND EVOLUTION OF POVERTY 21 Box 1.4 need to be supported. But the efforts need to go much Measuring voice and power using participatory farther than this, focusing on improving information on methods education and health indicators. The Demographic and Health Surveys need to be continued and expanded. As In the Voices of the Poor study, in small group discussions, poor important are efforts to expand and improve the very sniall people discussed the range of institutions important in their daily . . . of vulnerability and on voicel.s. lives and then identified the criteria that were important in rat- ing institutions. Once criteria were identified and agreed on, ness and powerlessness. groups rated institutions on these criteria using pebbles, beans, or other local material. Characteristics included trust, participa- The evolution of poverty tion, accountability, ability to build unity, responsiveness, respect, fairness and caring, and listening and loving. Poor people defined these criteria in clear and simple terms before scoring institutions. What are the magnitudes and patterns of poverty in the developing world? How has poverty evolved over the past decade? The answers to these questions are impor- Box 1.5 tant in framing the challenge of attacking poverty. Measuring governance: participatory methods The rest of this chapter describes global trends in the and cross-country surveys income (consumption), education, and health dimensions of poverty and shows the large diversity of outcomes- Can countrywide information on voice and participation be across dimensions, regions, countries, communities, obtained systematically to assess their role in development g and to compare countries? A recent study brought together h a a database covering 178 countries to assess the wider issue mance reflect differences in growth, in the distribution of governance, with voice and accountability measured by of assets, in the quality and responsiveness of state insti- indicators of civil liberties, political rights, the transparency tutions, in the degree of inclusiveness in societies (lower of the legal system, and the existence of independent media. social barriers for women, ethnic minorities, and the so- The data came from two types of sources: polls of experts on the country or region (including agencies specializing in risk cially disadvantaged more generally), and in how coln- rating, opinion surveys, and political analysis) and cross-country tries and people manage risks. surveys of residents by international organizations and NGOs. Highlighting the diversity in outcomes is important Indicators from the two types of data tend to corTelate strongly, for at least two reasons. It allows the identification of increasing confidence in the results. The studyfound a strong positive association between voice and accountability and successes and failures in poverty reduction and thereby five other clusters of governance indicators and three devel- enhances the understanding of what causes poverty opment outcomes: per capita income, infant mortality, and adult and how best to reduce it. And it brings to the fore the literacy (Kaufmann, Kraay, and Zoido-Lobaton 19991. fact that aggregate trends can hide significant differences The study also highlighted major weaknesses in existing databases on voice, empowerment, and governance. Mar- in poverty outcomes-for different ethnic groups, re- gins of error in the results are wide. Significant investment gions, and sectors within a country, for example. Aware- is needed in developing and undertaking surveys, with com- ness of these differences will help policymakers set parable methods across countries, to collect data on this im- priorities, concentrating actions where they are most portant dimension of poverty and well-being. National surveys on voice and empowerment would complement participatory needed. assessments. In designing the surveys, care would have to be taken to ensure that they are capable of capturing differ- Global and regional patterns: income poverty ences by region, gender, ethnicity, and so on. Such differences and social indicators are important not just in material poverty but in voice and em- Between 1987 and 1998 the share of the population in powerment as well. Bten18 n 98tesaeo h ouao developing and transition economies living on less than $1 a day fell from 28 percent to 24 percent (table 1.1). banks, 85 percent of the developing world's population This decline is below the rate needed to meet the inter- lives in countries with at least two household income or national development goal of reducing extreme income expenditure surveys. These surveys need to be improved poverty by half by 2015 (see box 2 in the overview). greatly and made more accessible to the public. Efforts Because of population growth, the number of people such as the Living Standards Measurement Surveys at the in poverty hardly changed. But there are large regional World Bank and Mecovi9 in Latin America (see box 1.3) variations in performance. East Asia and the Middle East 22 WORLD DEVELOPMENT REPORT 2000/2001 Box 1.6 Multidimensionality: dealing with aggregation There are several possible approaches to aggregating measures areas in figure). This method adds value because it goes beyond of the different dimensions of poverty and well-being. income. But it can be criticized because it would imply, for ex- Welfare function. A welfare function approach includes various ample, that a person who has very high income but is uneducated dimensions of well-being and defines poor people as all individuals is poor. An alternative is to count as poor everybody who is poor below a specified minimum level of total welfare (Tsui 1995,1997; in all dimensions (see dark shaded area in figure). In both cases Bourguignon and Chakravarty 1998). Thewelfarefunction approach the complications of making comparisons remain when one allows for tradeoffs, using individuals' own choices for comparing wants to measure not only the extent but also the intensity of situations and for assessing how much improvement is needed in poverty of individuals with multiple deprivations or with depriva- one dimension to maintain welfare if another dimension worsens. tions in different dimensions. The difficulty is finding a suitable welfare function for comparisons between nonmarket elements of individual welfare. While using a money metric and total expenditure is appropriate for assessing how many additional eggs or apples a person would have to consume Alternative aggregation rules to measure to accept less rice, it is less reliable for such important dimensions the multiple dimensions of poverty of welfare as social exclusion and political voicelessness. Moreover, choosing appropriate "weights" to form a single aggregate of these H nonmarket elements of individual welfare from existing data has so far proved to be an insurmountable challenge. Composite index. An alternative to using weights estimated from people's observed choices is to simply impose weights, as a simplistic, special-case application of the welfare function ap- proach. There have been several well-known efforts, such as the physical quality of life index Icombining the literacy rate, the in- fant mortality rate, and life expectancy; Morris 1979) and the H human development index (UNDP 1999a). While easy to use, these indexes do not really solve the intractable weighting problem be- cause they assign arbitrary (usually equal) weights to each com- ponent (Ravallion 1997b). Y Income Altemative aggregation rules. If the objective is to measure Note: H is the threshold defining the health-poor, and Ythat the number of poor people, another possibility is to count as poor defining the income-poor. everybody who is poor in any one of the dimensions (see all shaded Box 1.7 Uganda's poverty reduction goals The recent poverty reduction strategy paper for Uganda presents * Cutting HIV prevalence by 35 percent. a clear statement of the poverty reduction goals that the govern- * Reducing the incidence of stunting to 28 percent. ment has set. The goals focus on reducing absolute income * Reducing total fertility to 5.4 births per woman. poverty to 10 percent by 2017 and achieving universal primary en- The poverty reduction strategy paper outlines the govern- rollment (along with higher primary completion rates and educa- ment's approach to achieving these goals, with well-developed in- tional achievement) by 2004-05. The government also set a series terventions in four broad areas: creating a framework for economic of other human development goals for 2004-05: growth and transformation, ensuring good governance and security, * Reducing the under-five mortality rate to no more than 103 per directly increasing the ability of poor people to raise their incomes, 1,000 live births. and directly improving the quality of life of poor people. Source: IDA 2000. and North Africa have reduced their numbers in poverty- over the decade, from 474 million to 522 million, even East Asia dramatically so. But in all other regions the num- though the share of people in poverty fell from 45 per- ber of people living on less than $1 a day has risen. In cent to 40 percent. In Latin America and the Caribbean South Asia, for example, the number of poor people rose the number of poor people rose by about 20 percent. THE NATURE AND EVOLUTION OF POVERTY 23 Table 1.1 Income poverty by region, selected years, 1987-98 Population covered by at least People living on less than $1 a day one survey (millions) Region (percent) 1987 1990 1993 1996 19988 East Asia and Pacific 90.8 417.5 452.4 431.9 265.1 278.3 Excluding China 71.1 114.1 92.0 83.5 55.1 65.1 Europe and Central Asia 81.7 1.1 7.1 18.3 23.8 24.0 Latin America and the Caribbean 88.0 63.7 73.8 70.8 76.0 78.2 Middle East and North Africa 52.5 9.3 5.7 5.0 5.0 5.5 South Asia 97.9 474.4 495.1 505.1 531.7 522.0 Sub-Saharan Africa 72.9 217.2 242.3 273.3 289.0 290.9 Total 88.1 1.183.2 1,276.4 1,304.3 1,190.6 1,198.9 Excluding China 84.2 879.8 915.9 955.9 980.5 985.7 Share of population living on less than $1 a day (percent) Region 1987 1990 1993 1996 1998a East Asia and Pacific 26.6 27.6 25.2 14.9 15.3 Excluding China 23.9 18.5 15.9 10.0 11.3 Europe and Central Asia 0.2 1.6 4.0 5.1 5.1 Latin America and the Caribbean 15.3 16.8 15.3 15.6 15.16 Middle East and North Africa 4.3 2.4 1.9 1.8 1.9 South Asia 44.9 44.0 42.4 42.3 40.0 Sub-Saharan Africa 46.6 47.7 49.7 48.5 46.:3 Total 28.3 29.0 28.1 24.5 24.0 Excluding China 28.5 28.1 27.7 27.0 26.2 Note: The poverty line is $1.08 a day at 1993 PPP. Poverty estimates are based on income or consumption data from the countries in each region for which at least one survey was available during 198S-98. Where survey years do not coincide with the years in the table, the estimates were adjusted using the closest available survey and applying the consumption growth rate from national accounts. Using the assumption that the sample of countries covered by surveys is representative of the region as a whole, the number of poor people was then estimated by region. This assumption is obviously less robust in the regions with the lowest survey coverage. For further details on data and methodology see Chen and Ravallion (2000). a. Preliminary. Source: World Bank 2000s. Two regions fared particularly badly. In Europe and figures for 1998 are tentative because of the limited num- Central Asia the number in poverty soared from 1.1 mil- ber of surveys available (see box 1.2). lion to 24 million. In Sub-Saharan Africa the number of Relativepoverty. The poverty estimates in table 1.1 are poor people increased from an already high 217 million based on a poverty line that reflects what it means to be to 291 million over the same period, leaving almost half poor in the world's poorest countries (see box 1.2). f his the residents of that continent poor. definition judges poverty by standards common in South These variations in regional performance are leading Asia and much of Sub-Saharan Africa, regardless of the to a shift in the geographical distribution of poverty. In region for which poverty is being measured. An alterna- 1998 South Asia and Sub-Saharan Africa accounted tive definition of poverty-expounded by the British so- for around 70 percent of the population living on less ciologist Peter Townsend, among others-is a lack ofthe than $1 a day, up 10 percentage points from 1987 (fig- resources required to participate in activities and to enjoy ure1.1). living standards that are customary or widely accepted in While these numbers provide a sense of broad trends, the society in which poverty is being measured."0 they should be treated with caution in light of the short- Table 1.2 presents estimates of poverty based on a com- comings of the data mentioned above and the fact that bination of absolute and relative poverty concepts. The 24 WORLD DEVELOPMENT REPORT 2000/2001 Figure 1.1 Poverty in the developing world is shifting toward South Asia and Sub-Saharan Africa Distribution of population living on less than S1 a day Sub-Saharan Africa ~ - Sub-Saharan East Asia and 18.4% Africa Pacific East Asia and Pacific _ 35.3% Europe and Central Asia 2.0% Latin America and Europe and the Caribbean 6.5% Central Asia 0.1% South Asia Middle East and 40.1% Latin America North Africa and the Caribbean 5.4% South Asia 0.5% Middle East and 43.5% North Africa 1987 0.8% 1998 Source: Chen and Ravailion 2000. poverty estimates are based on the same data and proce- 15 percent of the population was below the $1 a day dures as those in table 1.1, but a different poverty line. poverty line, more than 50 percent of the population was A country-specific poverty line was used, equal to one- under the relative poverty line. Similarly, in the Middle third of a country's average consumption level in 1993 East and North Africa and in Europe and Central Asia at 1993 purchasing power parity (PPP), if that figure is poverty estimates are much higher by the relative poverty higher than the $1 a day poverty line. Otherwise, the $1 criterion. But the time trends remain unchanged."l a day line was used. The effect of using a relative poverty Social indicators. Social indicators in developing coun- line-instead of the $1 a day poverty line-is that poverty tries have improved on average over the past three decades. is now much higher in regions with higher average con- For example, infant mortality rates fell from 107 per sumption. It is also higher in regions with greater in- 1,000 live births in 1970 to 59 in 1998. But the decline equality. In Latin America, for example, where roughly between 1990 and 1998 was only 10 percent, while Table 1.2 Relative income poverty by region, selected years, 1987-98 Regional average Share of population living on less than poverty line one-third of average national consumption for 1993 (1993 PPP (percent) Region dollars a day) 1987 1990 1993 1996 1998a East Asia and Pacific 1.3 33.0 33.7 29.8 19.0 19.6 Excluding China 1.9 45.1 38.7 30.8 23.2 24.6 Europe and Central Asia 2.7 7.5 16.2 25.3 26.1 25.6 Latin America and the Caribbean 3.3 50.2 51.5 51.1 52.0 51.4 Middle East and North Africa 1.8 18.9 14.5 13.6 11.4 10.8 South Asia 1.1 45.2 44.2 42.5 42.5 40.2 Sub-Saharan Africa 1.3 51.1 52.1 54.0 52.8 50.5 Total 1.6 36.3 37.4 36.7 32.8 32.1 Excluding China 1.8 39.3 39.5 39.3 38.1 37.0 Note: See text for a definition of the poverty line. a. Preliminary. Source: Chen and Ravallion 2000. THE NATURE AND EVOLUTION OF POVERTY 25 meeting the international development goal would have ica show that between 1989 and 1996 the incidence required 30 percent. of poverty fell in Brazil, Chile, the Dominican Re- Theseaggregatefiguresmaskwideregionaldisparities. public, and Honduras-and rose in Mexico and Life expectancy in Sub-Saharan Africa in 1997 was still Republica Bolivariana de Venezuela.15 In another only 52 years-13 years less than the developing world group of countries for which only urban surveys were average-and 25 years-a full generation-less than the available, poverty rose in Ecuador, stayed nearly un- OECD average. One of the main causes is the still un- changed in Uruguay, and fell in Argentina, Bolivia, acceptably high infant mortality rate in Sub-Saharan Colombia,'6 and Paraguay. Africa, 90 per 1,000 live births. The rate is also very In East Asia poverty trends in the 1990s were influ- high in South Asia (77). Those levels are a far cry from enced by the impact of the recent economic crisis. In- the OECD average of 6 per 1,000. The AIDS crisis has donesia, the Republic of Korea, and Thailand all suffered aggravated the situation, leading to rising infant mortality increases in poverty, though to differing degrees (see in several African countries. Between 1990 and 1997 the chapter 9).17 In Indonesia one recent study estimated that infant mortality rate rose from 62 to 74 in Kenya and from poverty increased from around 11 percent in February 52 to 69 in Zimbabwe. Maternal mortality also remains 1996 to 18-20 percent in February 1999. Since then, exceptionally high in the region: of the 12 countries in poverty appears to have declined considerably, though the world with rates exceeding 1,000 deaths per 100,000 live births, 10 are in Sub-Saharan Africa. Table 1.3 Regional differences are equally obvious in education Income poverty in seven African countries, indicators. South Asia improved its gross primary en- various years rollment rate from 77 percent to more than 100 percent Share of population in 1982-96. But Sub-Saharan Africa's rate remained un- below the national changed at 74 percent (between 1982 and 1993 it actu- poverty lineb ally declined). Other education indicators confirm the Country (percent) importance of regional differences. Almost the entire de- and perioda Area Year 1 Year 2 cline in the illiteracy rate in the developing world has been Burkina Faso Rural 51.1 50.7 1994, 1998 Urban 10.4 15.8 in East Asia. By contrast, the number of illiterate people Total 44.5 45.3 increased by 17 million in South Asia and by 3 million Ghana Rural 45.8 36.2 in Sub-Saharan Africa. 12 Sub-Saharan Africa also has the 1991/92, Urban 15.3 14.5 lowest net primary enrollment rate. 1998/99 Total 35.7 29.4 Mauritania Rural 72.1 58.9 Variations in poverty across countries 1987, 1996 Urban 43.5 19.0 Detailed studies using national income poverty lines and Total 59.5 41.3 national-level social indicators show equally large varia- Nigeria Rural 45.1 67.8 tionsipoveryperfrmance across countris w1992, 1996 Urban 29.6 57.5 tions in poverty performance across countries within Total 42.8 65.6 each region. Uganda Rural 59.4 48.2 In Europe and Central Asia the proportion of the 1992, 1997 Urban 29.4 16.3 population living on less than $2 a day (at 1996 PPP) Total 55.6 44.0 ranges from less than 5 percent in Belarus, Bulgaria, Zambia Rural 79.6 74.9 Estonia, Hungary, Lithuania, Poland, and Ukraine to 1991, 1996 Urban 31.0 34.0 19 percent in Russia, 49 percent in the Kyrgyz Re- Total 57.0 60.0 public, and 68 percent in Tajikistan.13 Among seven Zimbabwe Rural 51.5 62.8 African countries with data spanning the 1990s, four 1991, 1996 Urban 6.2 14.9 (Burkina Fasou Nigeriat Zambias and Zimbabwe) ex- Total 37.5 47.2 (Burkina F'so, Nigeria,'Zambia, and Zimbabwe) ex- a. The dates in this column correspond to year 1 and year 2. perienced an increase in poverty, matching the re- b. Nutrition-based poverty lines. Comparisons between gional pattern for the decade, while three (Ghana, countries are not valid. gional patern for te decade,while thre (Ghana, Source: Demery 1999; Ghana Statistical Service 1998. Mauritania, and Uganda) had a decline (table 1.3).14S Available national poverty estimates for Latin Amer- 26 WORLD DEVELOPMENT REPORT 2000/200I it is still substantially higher than precrisis levels.'8 no progress in poverty reduction in the 1990s.21 For Trends in China in 1996-98 are sensitive to the choice India, there is an ongoing debate on the accuracy of the of welfare measure. Income poverty measures based on statistics. It provides a telling example of how difficult it the $1 a day or national poverty line show continued de- is to track poverty over time, even within countries cline. But a consumption-based poverty measure shows (box 1.8). a stalling in poverty reduction between 1996 and 1998, suggesting that poor households, especially in rural areas, Variations in poverty within countries have been saving an increasingly large share of their in- Country aggregates of different dimensions of poverty pro- comes.19 The most recent data for Vietnam show that vide a useful overview of performance. But they hide as between 1993 and 1998 the incidence of poverty, based much as they reveal. There are distinct patterns of poverty on a national poverty line, fell from 58 percent to 37 within countries, and different groups within a country percent.20 can become better or worse off. Poverty reduction also varied in South Asia in the Poverty in different areas within a country can-and 1990s. Bangladesh turned in a good performance despite does-move in different directions. In Burkina Faso and its worst floods in living memory, with GDP growth of Zambia rural poverty fell and urban poverty rose, but the 4.5 percent in 1998-99, thanks to a bumper rice crop after urban rise dominated and overall poverty rose (see table the floods. The concerted relief efforts by the government, 1 .3).22 In Mexico, while overall poverty declined- NGOs, and donors-and the ongoing food-for-work though modestly-between 1989 and 1994, there were programs-limited the loss of life and the impact of the large variations across regions within the country.23 In floods on poverty. Pakistan and Sri Lanka made little or China rapid income growth has been accompanied by ris- Box 1.8 Tracking poverty in India during the 1990s Recent data from India's National Sample Surveys (NSS) suggest sumption made when everybody's consumption is adjusted pro- that the pace of poverty reduction slowed in the 1 990s, particu- portionately. For example, it cannot be ruled out a priori that non- larly in rural areas. This occurred against a backdrop of strong eco- response and nonsampling errors in measuring consumption may nomic growth (GDP growth of 6.1 percent a year during 1990-98), differ among income groups. Also, Visania 12000)finds the differences according to the national accounts (NAS). There are signs of ris- betweentheNSSandNAStobeconsiderablylessifoneweekrather ing inequality nationally in the NSS data, due in large part to ris- than one month is used in the NSS as the reference period for con- ing average consumption in urban areas relative to rural areas, sumption. Srinivasan (2000) presents a detailed discussion of these though with some signs of higher inequality in urban areas. How- issues (Srinivasan and Bardhan 1974 present earlier discussions of ever, an important factor in the slow rate of poverty reduction was these issues.) slow growth in average consumption, as measured by the NSS. There is also evidence that part of the observed trend in rural Closer examination shows that NSS consumption is an increas- poverty in the earlier part of the 1990s may result from using in- ingly smaller fraction of private consumption as estimated in the NAS. adequate price deflators for rural areas. As a result, "it is likely that NSS consumption has declined relative to NAS consumption during the decline in rural poverty rates has been understated in the of- the past three decades; the two were much closer in the 1950s and ficial poverty counts. Indeed, we are led to suggest as a working 1 960s (Mukherjee and Chatterjee 1974). If the average consumption hypothesis that, between 1987-88 and 1993-94, there was no great figures from the NSS are replaced by the average consumption fig- difference in the rate of decline of urban and rural poverty, at least ures from the NAS, and everybody's consumption is adjusted pro- according to the headcount measure" (Deaton and Tarozzi 1999, portionately, poverty would show a downward trend during the pp. 34-35). 1990s (as found by Bhalla 2000). It is plausible that the NSS-based poverty numbers are un- But comparing NSS and NAS data is a complex matter, involv- derestimating the rate of poverty reduction in India. The issues in- ing differences in coverage, recall biases in the NSS, price imputa- volved are important not only because of the Indian poverty tions (for example, for home-produced consumption and in-kind figures' weight in global poverty trends, but aiso because similar wages in the NSS and for nonmarketed output in the NAS), and sam- problems are likely to arise elsewhere. India has a stronger sta- pling and nonsampling errors in both. Thus, without examining why tistical tradition than most poor countries. And it is not simply a the differences between the two have widened, adjusting the NSS matter of getting accurate estimates of poverty. Such surveys are mean upward to equal the NAS mean would be an arguable proce- a key resource for identifying the characteristics of poor people dure. For one thing, it is not clear why the average consumption data and thus are a vital input for focusing policy. Research in this area from the NSS would be wrong but not the inequality data, the as- is a high priority. THE NATURE AND EVOLUTION OF POVERTY 27 ing inequality between urban and rural areas and between Similar disparities show up in access to schooling provinces.24 and in educational achievement. In some poor countries Poverty tends to be associated with the distance from most children from the poorest households have no cities and the coast, as in China, Vietnam, and Latin Amer- schooling at all. A study of Demographic and Health Sur- ica.25 In China many of the poor reside in mountainous vey data found 12 countries in which more than half the counties and townships. In Peru two-thirds of rural 15- to 19-year-olds in the poorest 40 percent of house- households in the poorest quintile are in the mountain holds had zero years of schooling: Bangladesh, India, Mto- region, while fewer than a tenth are in the coastal region.26 rocco, Pakistan, and eight countries in Sub-Sahatan In Thailand the incidence of poverty in the rural north- Africa. In contrast, the median number of years corm- east was almost twice the national average in 1992, and pleted by 15- to 19-year-olds in the richest 20 percent although only a third of the population lives there, it ac- of households was 10 in India and 8 in Morocco. In other counted for 56 percent of all poor. countries the gap in educational achievement was much smaller: one year in Kenya, two in Ghana and Tanzania, Differences in health and education between and three in Indonesia and Uganda.32 In Mexico aver- low- and high-income households age schooling was less than 3 years for the poorest 20 per- Social indicators in many countries remain much worse cent in rural areas and 12 years for the richest 20 percent for the income-poor than for the income-nonpoor- in urban areas. often by huge margins. In Mali the difference in child mor- Primary enrollment rates show similar gaps. The en- tality rates between the richest and poorest households rollment rate for 6- to 14-year-olds is 52 percentage is equal to the average gain in child mortality rates points lower for the poorest households than for lthe recorded over the past 30 years.27 In South Africa the richest households in Senegal, 36 percentage points lower under-five mortality rate for the poorest 20 percent is twice in Zambia, and 19 percentage points lower in Ghana. The as high as the rate for the richest 20 percent, and in gaps are also large in North Africa (63 percentage points Northeast and Southeast Brazil, three times as high. in Morocco) and South Asia (49 percentage points in The picture is the same for malnutrition. A study of Pakistan).33 19 countries found that stunting (low height for age-an Within-country differences in social indicators also indicator of long-term malnutrition), wasting (low weight exist between urban and rural areas, across regions, and for height-an indicator of short-term malnutrition), across socioeconomic classes. In China there has been and being underweight (low weight for age) are higher a widening rural-urban gap in health status and health among poor people in almost all countries.2" But the dif- care use. While the rural population's use of hospital ser- ferences between poor and nonpoor tend to be smaller in vices declined 10 percent between 1985 and 1993, the countries with high average rates of malnutrition.29 urban population's increased by 13 percent.34 In Rus- The incidence of many illnesses, especially commu- sia the increase in mortality during the transition has nicable diseases, is higher for poor people, while their ac- been concentrated among younger males, and stunting cess to health care is typically less. In India the prevalence of children, relatively high for an industrialized coun- of tuberculosis is more than four times as high in the poor- try, has been most prevalent in rural areas and among est fifth of the population as in the richest, and the preva- poor people.35 lence of malaria more than three times as high.30 In 10 developing countries between 1992 and 1997, only 41 Gender disparities percent of poor people suffering from acute respiratory One of the key variations within a country is the differ- infections were treated in a health facility, compared ent achievement of women and men. The allocation of with 59 percent of the nonpoor. In the same period only resources within households varies depending on the age 22 percent of births among the poorest 20 percent of peo- and gender of the household member. But estimating the ple were attended by medically trained staff, compared number of poor men and women independently is dif- with 76 percent among the richest 20 percent.31 Al- ficult, if not impossible, because consumption data are though HIV/AIDS initially affected the poor and the rich collected at the household level.36 Even so, available almost equally, recent evidence indicates that new infec- health and education data indicate that women are often tions occur disproportionately among poor people. disadvantaged. 28 WORLD DEVELOPMENT REPORT 2000/2001 A recent study of 41 countries shows that female dis- poorest states, only 8 percent of rural scheduled caste advantage, defined as the gap between male and female women are literate, a third the rate for rural women in primary enrollment rates, varies enormously. In Benin, Uttar Pradesh. But new research suggests that literacy Nepal, and Pakistan the male-female gap in the primary rates of rural scheduled caste women are on the rise enrollment rate is more than 20 percentage points, and across India. Although only 31 percent of rural sched- in Morocco, 18. But in Brazil, Indonesia, Kenya, Mada- uled caste or scheduled tribe girls in the primary school gascar, the Philippines, and Zambia the enrollment rates age group were enrolled in school in 1986-87, 53 per- of boys and girls are almost the same.37 The gender gap cent were by 1995-96.48 in education is often lower for the richest housholds and highest for the poorest households. In India the gender Volatility at the household level gap in enrollment rates is 4.7 percentage points for chil- Studies of income poverty changes for the same house- dren from the wealthiest 20 percent of households, com- holds over time show significant movement in and out pared with 11 percentage points for children from the of poverty. While some groups are chronically below the poorest 20 percent of households.38 poverty line, other groups face a high risk of falling into poverty some of the time. Studies for China, Ethiopia, Disparities by caste, ethnicity, Russia, and Zimbabwe find that the "always poor" group and indigenous status is smaller than the "sometimes poor" group.49 However, There may also be groups that face particular social bar- these results should be treated with caution because ob- riers. Disadvantaged in many developing and developed served changes reflect measurement errors as well as real countries and transition economies, ethnic minorities changes.50 and racial groups often face higher poverty.39 The in- One immediate question is whether some types of digenous populations have a much higher incidence of in- households are more likely to suffer from chronic (rather come poverty in a sample of Latin American countries for than transitory) poverty. The answer differs from coun- which data are available.40 Schooling attainments for try to country, but asset holdings often play a key role. these disadvantaged groups are also lower than for other In China a lack of physical capital is a determinant of both groups. The indigenous groups in Guatemala have 1.8 years chronic and transitory poverty, but household size and of schooling, and the nonindigenous 4.9 years.4' In Peru education of the head of household determine the like- indigenous people were 40 percent more likely to be poor lihood of chronic but not of transitory poverty.5' than nonindigenous groups in 1994 and 50 percent more In the transition economies of Europe and Central Asia likely in 199742 In rural Guatemala children of indige- economic mobility has increased, but chronic poverty is nous mothers are more likely than those of nonindigenous emerging as a key issue.52 Whether a household joins the mothers to be stunted.43 In the inner cities of the United ranks of the new poor or the new rich depends very States white married couples have an incidence of poverty much on its characteristics, especially its links with the of 5.3 percent, while black or Hispanic single-mother labor market. The transition has increased the disad- households have an incidence of more than 45 percent.44 vantage of "old poor" (pensioners, families with large num- Evidence for India shows that scheduled castes and bers of children, and single-parent families) and given rise scheduled tribes face a higher risk of poverty.45 These to "new poor" (long-term unemployed, agricultural work- are among the structural poor who not only lack eco- ers, young people in search of their first job, and refugees nomic resources but whose poverty is strongly linked displaced by civil conflict).53 In Poland the chronically to social identity, as determined mainly by caste.46 poor constitute a distinct segment of the population. They also have worse social indicators. Among rural Larger households, those working on farms, and house- scheduled caste women in India the literacy rate was holds dependent on social welfare are most at risk of stay- 19 percent in 1991, half that for the country, and ing poor.54 Russia has seen the emergence of new poor among scheduled caste men, 46 percent, compared during the transition. In the early 1990s new groups of with 64 percent for the country.47 When several dis- poor formed as a result of the erosion of real wages and advantages are combined-being a woman from a so- pensions and the impact of unemployment,55 and poverty cially excluded group in a backward region-the is becoming longer in term and more resistant to economic situation is worse. In Uttar Pradesh, one of India's recovery.56 THE NATURE AND EVOLUTION OF POVERTY 2') * * * for example, also reflect the extent to which state in- stitutions are responsive and accountable to poor peo- This chapter has shown that progress in income ple (chapter 6). In other cases social barriers linked poverty reduction and human development varies widely with gender, ethnicity, race, and social status help per- across regions, countries, and areas within countries. It petuate income poverty and low levels of health and edL- has also shown the existence of significant gaps in per- ucation among the socially disadvantaged (chapter 7). formance by gender, ethnicity, race, and social status. Policy biases against labor-intensive sectors such as agri- Much of the difference in performance across regions culture and light manufacturing at the national (chap- and countries can be attributed to differences in economic ter 4) or international (chapter 10) level and skill-biased growth (chapter 3). The growth collapses in many coun- technological change (chapter 4) can result in lower re- tries in Africa and the former Soviet Union had a ductions in income poverty at similar growth rates. devastating impact on poverty. The economywide crises This chapter has also noted that there can be large and natural disasters in East Asia, Latin America, Sub- volatility in incomes of households. This brings to the Saharan Africa, and Europe and Central Asia also led to fore the importance of understanding the sources of risk important setbacks in poverty reduction (chapter 9). By that households face and the mechanisms best suited to contrast, the spectacular growth performance in China managing those risks (chapters 8 and 9). resulted in a sharp drop in income poverty. In the rest of Finally, this chapter has argued that the experience of East Asia, despite the financial crisis, steady growth rates poverty goes beyond material deprivation and low levels also translated into significantly lower poverty over the of health and education. The inability to influence tlhe 1990s. decisions that affect one's life, ill treatment by state in- But the initial inequalities and the pattern of growth stitutions, and the impediments created by social barri- also account for the differences in performance in ers and norms are also dimensions of ill-being. Another poverty reduction in its multiple dimensions as some ge- is vulnerability to adverse shocks, natural disasters, dis- ographic areas and social groups are left behind. In ease, and personal violence. This broader conception of some cases initial differences include unequal access to poverty leads to a deeper understanding of its causes and assets, markets, and infrastructure and an uneven dis- a broader range of actions for attacking it. These are tribution of skills (chapters 3, 4, and 5). The differences outlined in chapter 2 and developed in more detail in sub- in health and education among and within countries, sequent chapters. CHAPTER 2 Causes of Poverty and a Framework for Action That report also noted that these efforts had to be com- From World Development Report 1990 ... plemented by safety nets for people exposed to shocks - Labor-intensive growth and unable to benefit from the strategy. But safety nets * Broad provision of social services were clearly seen as playing a supporting role for the ... to World Development Report 2000/2001 two main parts of the strategy. Opportunity The 1990 report's framework for action was derived * Empowerment from its concept of poverty, its analysis of the causes * Security of poverty, the experience of the 1970s and 1980s, and the state of the world economy at the end of the 1980s. It viewed poverty as low consumption and low achievement in education and health. Economic A decade ago World Development Report 1990 pre- development-brought about essentially by liberaliz- sented a two-part strategy for poverty reduction: ing trade and markets, investing in infrastructure, and providing basic social services to poor people, to in- Countries that have been most successful in attacking crease their human capital-was seen as key to re- poverty have encouraged a pattern of growth that makes ducing poverty. efficient use of labor and have invested in the human The experience that defined the 1990 report, from capital of the poor. Both elements are essential. The first its vantage of 1989, was the contrasting experience in provides the poor with opportunities to use their most the 1970s and 1980s of East Asia, where poverty had abundant asset-labor. The second improves their fallen sharply, and of Africa, Latin America, and South immediate well-being and increases their capacity to take Asia, where poverty had declined less or even risen. Why advantage of the newly created possibilities. Together, did Indonesia outperform Brazil in the 1970s and they can improve the lives of most of the world's poor. 1980s in reducing income and nonincome (education -World Bank 1990 (p. 51) and health) poverty? The answer was labor-intensive 31 32 WORLD DEVELOPMENT REPORT 2000/200I growth and broad provision of social services-the report's of growth in developing countries is not necessarily two-part strategv. intensive in unskilled labor. This report uses new evidence and multidisciplinary * The emphasis on social services for building human thinking that together broaden the choices for devel- capital was perhaps too optimistic about the institu- opment action to reduce poverty in its multiple di- tional, social, and political realities of public action. mensions. The evidence confirms that economywide Public investment in basic education and health in de- growth improves the incomes of poor people-and in veloping countries has been rising-though in some the longer run reduces nonincome poverty. And ex- countries more slowly than GDP, suggesting a pos- panding the human capabilities of poor people re- sible lack of commitment to expanding social services. mains central in any poverty reduction strategy, both In many countries social spending is regressive. More- for the intrinsic value of such capabilities as health over, such investment has been less effective than ex- and education and for their instrumental contribu- pected, in part because of serious problems in quality tion to other dimensions of well-being, including and in responsiveness to poor people's needs- income. institutional failures highlighted in the Voices of the But the experiences of the 1990s show that: Poor study and other research. But there have been * Growth cannot be switched on or off at will. Mar- successes even in seemingly difficult conditions of low ket reforms can indeed boost growth and help poor resources. Experience and research show that effec- people, but they can also be a source of dislocation. tiveness in service delivery is highly dependent on local The effects of market reforms are complex, deeply institutional capabilities, market structure, and pat- linked to institutions and to political and social terns of political influence. structures. The experience of transition, especially in * There is a powerful case for bringing vulnerability the countries of the former Soviet Union, vividly il- and its management to center stage. Participatory lustrates that market reforms in the absence of ef- poverty work underlines the importance of vulnera- fective domestic institutions can fail to deliver growth bility to economic, health, and personal shocks. So do and poverty reduction. Furthermore, there is evi- the financial crises of the 1990s-not least in East Asia, dence that technological change in the past decade the shining example of success in development and has been increasingly biased toward skills. So in con- poverty reduction-and the sequence of devastating trast to what was expected and needed, the pattern natural disasters. BRAZIL Brazil has made impressive improvements still leave school with skills inadequate for ties are tackled effectively, the gains for in social indicators. Net enrollment in pri- a middle-income country integrated with the poor people will be modest. To reduce mary education increased from 88.2 percent global economy. And reducing income structural inequalities, a large land reform in 1992 to 97.1 percent in 1997. Infant mor- poverty has proved difficult. Indeed, in the program is under way, and there have been tality fell from 62 per 1,000 live births in the unstable macroeconomic environment of promising experiments in negotiated land mid-1980s to 38 in the mid-1990s. And the 1980s and early 1990s, poverty rose. reform in the Northeast. In the ongoing much urban infrastructure helps poor peo- Two recent events confirm that the groups education effort the next steps will proba- ple. New programs guarantee minimum most vulnerable to economic insecurity are bly require even broader, deeper, and more per capita spending for basic health care and those with the highest incidence of poverty. participatory reforms-many of these are minimum per student spending in primary Drought in the Northeast hit poor rural now under discussion. The government is schools. Innovative action to get children workers severely, and the ripple effects of also continuing to ease the constraints of into school includes the Bolsa Escola, which the East Asian crisis, though more benign constitutional entitlements, which limit the gives poor families grants if their children than expected, reduced the income of room for maneuver on public spending. Fi- go to school. workers with the least education. nally, successful community-driven devel- Despite the advances, the inequalities Some illustrative priorities for action: opment approaches-in urban upgrading, in health and education remain great, with job growth through productive investment small-farm investments, and community the poorest fifth of the population having and prudent macroeconomic management health agents-show what is possible three years of education, and the top fifth is clearly central to increasing income op- when there is an empowering mobilization more than nine years. The income-poor portunities. But unless structural inequali- of citizens. CAUSES OF POVERTY AND A FRAMEWORK FOR ACTION 3 * Inequality is back on the agenda-in the realm of turns on these assets, through a combination of mar- ideas and experience and in the political discourse of ket and nonmarket actions (part II). many developing (and developed) countries. New * Facilitating empowerment: making state institutions work shows the importance of gender, ethnic, and more accountable and responsive to poor people, racial inequality as a dimension-and a cause-of strengthening the participation of poor people in po- poverty. Social, economic, and ethnic divisions are litical processes and local decisionmaking, and re- often sources of weak or failed development. In the moving the social barriers that result from distinctions extreme, vicious cycles of social division and failed of gender, ethnicity, race, and social status (part III). development erupt into internal conflict, as in Bosnia * Enhancingsecurity: reducing poor people's vulnerabiliry and Herzegovina and Sierra Leone, with devastating to ill health, economic shocks, policy-induced consequences for people. dislocations, natural disasters, and violence, as well as * The global forces of integration, communication, and helping them cope with adverse shocks when they technological advance have proceeded apace, bringing occur (part IV). significant advances to some. But they have passed oth- Opportunity, empowerment, and security have in- ers by. Private capital flows now dominate official trinsic value for poor people. And given the important flows in the world, but they reinforce positive economic complementarities among them, an effective poverty re- developments, either neglecting or punishing coun- duction strategy will require action on all three fronts, by tries with weak economic conditions. the full range of agents in society-government, civil The new evidence and broader thinking do not negate society, the private sector, and poor people themselves. earlier strategies-such as that of World DevelopmentRe- Actions cannot be confined to individual countries port 1990. But they do show the need to broaden the in the developing world. Harnessing global forces in favor agenda. Attacking poverty requires actions beyond the eco- of poor countries and poor people will be essential. Ac- nomic domain. And public action has to go beyond in- tions are needed to promote global financial stability-- vesting in social services and removing antilabor biases and to ensure that poor countries are not left behind by in government interventions in the economy. advances in technology and in scientific and medical re- Acknowledging the need for a broader agenda, this re- search. The markets of rich countries must be opened port proposes a general framework for action in three to the products of poor countries, and aid and debt re- equally important areas: lief must be increased to help poor people help them- * Promoting opportunity: expanding economic oppor- selves. And poor countries and poor people need to be tunity for poor people by stimulating overall growth given a voice and influence in international forums and by building up their assets and increasing the re- (part V). CHINA China stands out for its extraordinary decline areas, but micro evidence suggests that vii- slowdown, insecurity in areas dependent on in income poverty and its high levels of lage mechanisms continue to provide high now inefficient state production could be education and health. But it has also had sig- levels of food security through land alloca- acute. Second, the smooth integration of nificant increases in inequality-between tion processes-confirmed as politically China into the global trading system will be town and country, and between the coastal popular in villages by democratic votes. key in locking in reforms and ensuring eco- areas and inland China, with the poor, semi- Ensuring the voice of the new poor in urban nomic stability and steady job growth. But arid areas of inland China participating lit- areas and those left behind in inland China if new opportunities are not to widen dis- tie in growth. will be important in guiding action. parities, this will have to be accompanied by Formal structures of security are in tran- Three areas for action are illustrative. greater emphasis on building the assets of sition, and there are deep concerns about First, maintaining rapid growth through high people living in the poorer areas. Third, con- the less dynamic parts of urban China, nonstate investment is crucial if there isto tinuing area-based integrated rural devel- which are experiencing the beginnings of be a smooth process of job destruction in opment activities in poor areas of inland a major shakeout in state enterprise and inefficient state activities and smooth reform China and, more generally, balancing in- government employment. Formal provi- of the social protection arrangements for vestment across geographic areas need to sion of security was always weaker in rural state employees. If there is a sustained be important parts of any overall strategy. 34 WORLD DEVELOPMENT REPORT 2000/200I The causes of poverty depends on a legal structure that defines and enforces pri- vate property rights or on customary norms that define One route for investigating the causes of poverty is to ex- common property resources. Access may also be affected amine the dimensions highlighted by poor people: by implicit or explicit discrimination on the basis of gen- * Lack of income and assets to attain basic necessities- der, ethnicity, race, or social status. And both access to food, shelter, clothing, and acceptable levels of health assets and returns to assets are affected by public policy and education. and state interventions, which are shaped by the politi- * Sense of voicelessness and powerlessness in the insti- cal influence of different groups. tutions of state and society. Also important is the volatility of returns. Volatility * Vulnerability to adverse shocks, linked to an inabil- results from market fluctuations, weather conditions, ity to cope with them. and, in some societies, turbulent political conditions. To understand the determinants of poverty in all its Volatility affects not only returns, but also the value of dimensions, it helps to think in terms of people's assets, assets, as shocks undermine health, destroy natural and the returns to (or productivity of) these assets, and the physical assets, or deplete savings. volatility of returns. These assets are of several kinds: * Human assets, such as the capacity for basic labor, skills, Lack of income and assets and good health. * Natural assets, such as land. If you have a job at all now, you're overworked and * Physical assets, such as access to infrastructure. underpaid. * Financial assets, such as savings and access to credit. -- VUUng I'omain fi7om Diffi-rozgra6d Bkdgarzia * Social assets, such as networks of contacts and recip- rocal obligations that can be called on in time of need, Some have land, but they can't buy fertilizer; if some and political influence over resources. work as weavers, they aren't wellpaid; if some workfor The returns to these assets depend on access to mar- daily wages, they aren't paid a just wage. kets and all the global, national, and local influences on --(.ackchiquel hIdian, (GuatemlaL returns in these markets. But returns depend not just on the behavior of markets, but also on the performance of Poor people consistently emphasize the centrality of institutions of state and society. Underlying asset own- work to improving their lives. A countty's overall wealth ership and returns to assets are not only economic but also is an important influence on this: as countries grow fundamental political and social forces. Access to assets richer, so on average do poor people in those countries, INDIA India suffers severe deprivations in edu- sial because of measurement problems in most of India. In areas with deep depri- cation and health-especially in the North, (see box 1.8). vation in health and education, the devel- where caste, class, and gender inequities There are also marked differences within opment of social infrastructure is critical. are particularly strong. In studies in Bihar India-with the South, particularly the state Expanding education and health services and Uttar Pradesh poor women and men of Kerala, having sharply better education and will require that state governments reverse emphasized their extreme vulnerability and health. Kerala has life expectancies greater the deterioration in their fiscal positions, as the ineffectiveness of state institutions, than those in Washington, D.C., despite subsidies to the loss-making power sector from schools to police. vastly lower income levels. The effectiveness crowd out spending in the social sectors. In the past, poverty reduction in India of public action in Kerala has been attributed The higher spending will need to be lagged behind that of East Asia because of to its strong tradition of political and social matched by better service provision. This will slower growth and significantly less mobilization. require deep improvements in governance, progress in promoting mass education and What are the priorities for action in India? often weakest in India's poorest regions, and basic health. More recently, however, Accelerated poverty reduction will require in combating teacher absenteeism. Also growth has accelerated and poverty has faster growth, which in turn demands lib- needed is more equitable service provision, fallen, although the actual impact of growth eralization, especially in agriculture, and bet- which will require empowering women and on poverty reduction remains controver- ter provision of infrastructure, sorely lacking members of lower castes. CAUSES OF POVERTY AND A FRAMEWORK FOR ACTION 35 with the main mechanism being better-paid work. Figure 2.1 With economic growth, income poverty falls; with Poverty shows a strong link with economic economic contraction, income poverty rises (figure contractions and expansions 2. 1). Some countries in East Asia sustained per capita Annual change in poverty rate GDP growth rates of 4-5 percent over four decades, Percent with massive improvements in living standards and in 15 health and education for poor people and for everyone else. Other countries, most in Africa, registered nega- 10 tive growth or no growth at all over the same period, delivering no improvements even in average living standards. While economic growth is systematically associated . with poverty reduction, the rate at which growth trans- 0 lates into lower poverty depends on the initial level of in- equality in the distribution of income and how that -5 distribution changes over time. Growth-and its effec- tiveness in reducing poverty-also depends on sound, sta- 10 ble governance. So confronting socioeconomic inequalities -10 - o 5 10 and building sound institutions can be important both Percent for providing a socially sustainable basis for overall growth Note: The data refer to 150 country-level spells of poverty change and for ensuring that poor people gain substantially from and mean income change from poverty surveys in 1980-98. Each that growth. point represents one-fourth of the sample, ordered from strongest contraction to strongest expansion. Source: World Bank staff estimates based on Chen and Ravallion (20001. Voicelessness and powerlessness- the institutional basis ofpoverty Those materially deprived feel acutely their lack of voice, hands of the institutions of state and society (box 2.1). power, and independence (see box 1.1 in chapter 1). Absence of the rule of law, lack of protection against vi- This helplessness subjects them to rudeness, humilia- olence, extortion and intimidation, and lack of civility and tion, shame, inhumane treatment, and exploitation at the predictability in interactions with public officials-all these JORDAN Poverty and inequality in Jordan increased disabled, microcredit, and health insurance ries. Its 1998 service delivery survey reflects at the end of the 1980s as a result of a benefits. But by focusing on the always dissatisfaction among beneficiaries, who macroeconomic shock. But between 1992 poor rather than the sometimes poor, gov- complain of procedural difficulties and ob- and 1997 Jordan reduced income poverty, ernment programs fail to address the shal- stacles, benefits canceled without verifi- despite lagging-at times even negative- lowness of poverty in Jordan. So the cation, and inadequate assistance. The fund per capita GDP growth rates. The expla- vulnerability of the poor and the near-poor is acting on some of these findings, mov- nation lies in a decline in inequality due in to outside shocks is high, though tempered ing it closer to true accountability and to em- part to the phaseout of regressive food by substantial nongovernmental and reli- powering poor people. subsidies coupled with expansion of the gious charitable activity that complements Jordanians stand to gain much from government safety net (World Bank 1999q). strong family and community networks. the long-awaited Arab-Israeli peace divi- To sustain these gains, it is important to im- That vulnerability can be addressed by com- dend. But for the gains to materialize, prove growth-to make social spending munity-based public work programs offer- Jordanians must have the tools that ed- more affordable and to directly expand op- ing low-wage jobs and by unemployment ucation provides. So continued emphasis portunities for poor people. insurance and assistance. The National Aid on ensuring access for poor people to Government assistance is impressive- Fund (in charge of the government safety publicly provided basic education services targeted cash transfers favoring female net) could identify other means of assis- is critical both for the short and the long and elderly heads of household and the tance by soliciting ideas from beneficia- run. 36 WORLD DEVELOPMENT REPORT 2000/2001 Box 2.1 On interacting with state institutions: the voices of the poor Although there are pockets of excellence, poor people in the Voices shouted at, told they smelled bad and were lazy and good-for- of the Poor study generally rated state institutions low on fair- nothing. ness, honesty, relevance, effectiveness, responsiveness, and ac- Poor people in many countries spoke about being kept end- countability. Nevertheless, they view government agencies as lessly waiting while the rich went to the head of the queue. having an important role in their lives, and they have a clear picture In Europe and Central Asia pensioners trying to collect their of the qualities they would like to see in the institutions with which meager pensions experience endless red tape, rude and unre- they interact. sponsive officials, and withholding of information. Poor people in In India the characteristics of credit institutions can deter poor the region criticize mayors and local authorities for their arbi- people from seeking loans. Poor people in many regions also trariness, inefficiency, and often corruption (though there are no- report widespread corruption in health care systems. But when table exceptions). facing serious health conditions, they feel they have no choice Poor people hunger for institutions that are fair, polite, honest, but to comply with demands for bribes. In Macedonia people listening, trustworthy, and neither corrupt nor corrupting. A poor conclude, "nobody wants you to come with empty hands." woman in Vila Junqueira, Brazil, summed it all up: The behavior of health care providers becomes another de- terrent to those needing health services. In Tanzania in many An institution should not discriminate against people because areas, men, women, and youth stated over and over that they they are not well dressed or because they are black. If you wear are treated like animals, worse than dogs. They report that a suit you are treated as sir; if you are wearing sandals they even before they could explain their symptoms, they would be send you away. Source: Narayan, Chambers, Shah, and Petesch 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 2000. place a large burden on poor people. They are prevented Vulner ability from taking advantage of new economic opportunities or engaging in activities outside their immediate zone ofse- Threeyears ago it was a ve?y badyear. Theflood washed curity. Threats ofphysical force or arbitrary bureaucratic away all of our crops, and there was a lot of hunger power make it difficult for them to engage in public af- a,-ound here, to the point that nmany people actually died fairs, to make their interests known, and to have them of hunger. They must have been at least a dozen, mostly taken into account. And unaccountable and unrespon- children and old people. NVobody could help them. Their sive state institutions are among the causes of relatively relatives in the village had no food either; nobody had slow progress in expanding the human assets of poor enough food for his own children, let alone for the people. children of his brother or- cousin. Andfew had a richer In agrarian societies poor people's lack of assets and relative somewhere el-e IVho could help. income-earning opportunities ties them to rich land- ----lw 'iThV , Lic, owners in patron-client relationships. And for women, a lack of savings and assets precludes a more indepen- Vulnerability is a constant companion of material and dent role in decisionmaking in the household and the human deprivation, given the circumstances of the poor community. and the near-poor. They live and farm on marginal Social norms and barriers can also contribute to voice- lands with uncertain rainfall. They live in crowded lessness and powerlessness. While local cultures have in- urban settlements where heavy rains can wipe out their trinsic value, they can sometimes be inimical to reducing homes. They have precarious employment, in the for- human deprivation. Pervasive in almost all societies is in- mal or informal sector. They are at higher risk of dis- equality between men and women. Poor women are dis- eases such as malaria and tuberculosis. They are at risk criminated against in the household and in land, labor, of arbitrary arrest and ill treatment at the hands of local and credit markets. This both causes poverty and un- authorities. And they-women in particular-are at dercuts development-for women's agency is a powerful risk of being socially excluded and victims of violence source of human gains, especially for children. Discrim- and crime. ination based on ethnicity, religious beliefs, social status, The risks that poor people face as a result of their cir- and race has similar effects. cumstances are the cause of their vulnerability. But the CAUSES OF POVERTY AND A FRAMEWORK FOR ACTION 3,7 deeper cause is the inability to reduce or mitigate risk or such as volatile capital flows, global climate change, and cope with shocks-a cause that both draws from and feeds arms sales-are the causes of shocks and disruptions iin into the causes of other dimensions of povertv. Low lev- poor countries, the inability or unwillingness of the els of physical, natural, and financial assets make poor peo- global community to address them increases the vulner- ple especially vulnerable to negative shocks-those with ability of poor people (chapter 10). more assets can weather these shocks as long as they are temporary. Lack of adequate assets can set up a vicious A framework for action downward spiral in which actions to cope in the short term worsen deprivation in the long term. Pulling children out What framework for action is needed to effectively reduce of school to earn extra income during an economic cri- poverty in all its dimensions? National economic devel- sis. Depleting natural resources beyond the sustainable opment is central to success in poverty reduction. But level. Making quick sales of land or livestock at desper- poverty is an outcome of more than economic processes. ately low prices. Lowering nutritional intake below the It is an outcome of economic, social, and political processes levels necessary to sustain health (chapters 8 and 9). that interact with and reinforce each other in ways that Another underlying cause of vulnerability is the in- can worsen or ease the deprivation poor people face every ability of the state or community to develop mecha- day. To attack poverty requires promoting opportunity, nisms to reduce or mitigate the risks that poor people face. facilitating empowerment, and enhancing security-- Irrigation, infrastructure, public health interventions, with actions at local, national, and global levels. Making honest police and a fair legal system, public work schemes progress on all three fronts can generate the dynamics for in times of stress, microcredit to tide people through the sustainable poverty reduction. aftermath of an adverse shock, social networks of support The areas for action illustrate the complexity of de- and insurance, famine relief in extreme circumstances- velopment. How can priorities be decided in practice? all reduce vulnerability for poor people. The diverse Do all actions have to be carried out in all three areas? cross-country experience with each of these mechanisms Both the strategic approach and the areas of suggested can help in developing actions to address vulnerability in action are only a guide. Actual priorities and actions need specific circumstances. to be worked out in each country's economic, sociopo- Poor people also are exposed to risks beyond their litical, structural, and cultural context-indeed, each community-those affecting the economy, the environ- community's. But even though choices depend on local ment, and the society in which they live. Civil conflict conditions, it generally is necessary to consider scope for and wars, economic crises, and natural disasters affect not action in all three areas-opportunity, empowerment, and only their current living standards but also their ability security-because of their crucial complementarities. to escape poverty. And to the extent that global forces- The country examples in this chapter illustrate how Ito RUSSIAN FEDERATION Like other countries of the former Soviet of elite capture of the state. As the new oli- Today's structural inequality, closely linked Union, Russia has had a dramatic rise in both garchs have also captured privatized assets to the political structure, runs the risk of be- poverty and inequality and a worsening of and resource rents, the rise in inequality is coming deeply embedded, if it has not al- adult mortality. The Russian people have ex- the product not of the market-oriented re- ready become so. Dealing with associated perienced large increases in insecurity- forms themselves but of the interactions be- issues of governance is likely to be a pre- through macroeconomic volatility, the loss tween the reforms and the political and requisite to reduced macroeconomic volatil- of old job-related forms of security, and the institutional structures during the transition ity and a business environment that fosters sharp rise in violence-and often acute psy- process. the investment needed to counter the ex- chological stress from the rise in poverty. What are the priorities for action to re- traordinary collapse in formal sector jobs. While the electoral process has been im- duce poverty? Fundamental to improving It is also a prerequisite to pro-poor budget portant in empowering the citizenry, this the overall environment is reducing the allocations, backed by decentralization and has been offset by the profound feelings of elite's capture of the state at the national participatory engagement to foster greater disempowerment stemming from the new level, including through further market re- accountability and responsiveness in service sources of insecurity and by the problems forms to deconcentrate economic power. provision. 38 WORLD DEVELOPMENT REPORT 2000/2001 identify priorities and areas of action in country-specific effects on poor people. This does not necessarily mean poverty reduction strategies. going slow: rapid reforms can be important to bring gains to poor people and to break down monopoly priv- Opportunity ileges for the rich. Adverse effects of reforms on poor peo- Growth is essential for expanding economic opportunity ple can be compensated for by action in other areas, for poor people-though this is only the beginning of the such as safety nets to ease the transition costs. story of public action (chapter 3). The question is how Relatively neglected is market reform targeted to poor to achieve rapid, sustainable, pro-poor growth. A busi- people. Such reforms may have a different focus than other ness environment conducive to private investment and reforms-eliminating or simplifying the regulations af- technological innovation is necessary, as is political and fecting microenterprises and small and medium-size social stability to underpin public and private invest- firms, strengthening registries to allow small producers ment. And asset and social inequalities directly affect to use land as collateral, or developing the policy frame- both the pace of growth and the distribution of its ben- work for small-scale insurance. efits. The distribution of growth benefits matters, not least Key in expanding economic opportunities for poor because distributional conflict can undermine the stability people is to help build up their assets (chapter 5). Human needed for overall growth. capabilities such as health and education are of intrinsic Markets are central to the lives of poor people (chap- value, but also have powerful instrumental effects on ter 4). The evidence shows that on average countries material well-being. Also important to the material that are open to international trade and have sound prospects of poor people is ownership of-or access to- monetary and fiscal policy and well-developed financial land, infrastructure, and financial services. And social as- markets enjoy higher growth. Where market-friendly re- sets, including social networks, often also play an forms have been successfully implemented, on average instrumental role. stagnation has ended and growth resumed. But at times A range of actions can support poor people in ex- reforms to build markets fail entirely. The impact of panding their assets. The state-because of its power to market reforms on economic performance and inequal- raise revenues and use them as an instrument of ity depends on institutional and structural conditions, in- redistribution-has a central role, especially in provid- cluding the comparative advantage of countries and ing basic social services and infrastructure. Where access patterns of asset ownership. And the impact of market to land is highly unequal, there is a social and economic reforms differs for different groups in an economy- case for negotiated land reforms. For many services the there are winners and losers, and the losers can include state's role in provision can be complemented by market poor people. The design and sequencing of such reforms mechanisms, civil society, and the private sector, in- thus need to take account of local conditions and the likely creasing the benefits to poor people. And for local ser- SIERRA LEONE Sierra Leone, by the latest price-adjusted teract with ethnic and other social divides with citizens' disempowerment by those measures, is the poorest country in the to cause internal strife. The effects of with guns, and re-create the institutions world. But this statement fails to convey conflict-destruction of fragile institutions for mourning and for managing psycho- the true depth of human deprivation in of governance, flight of skills, personal logical losses. that nation. The people of Sierra Leone losses, and social wounds that could take International action will be important. remain caught in a tragic conflict-one generations to heal-create a vicious cycle Once there is a basis for some development, that has taken a terrible toll through lost of continued poverty and strife. concerted external support will be crucial. lives, rape, mutilation, and the psycholog- Sierra Leone has a desperate need for The Enhanced Heavily Indebted Poor Coun- ical harm to boys abducted into the army assets-human, physical, and social-and tries Debt Relief Initiative provides transi- and militias. for greater market opportunities. And the tional support to postconflict societies for Work on the sources of conflict in de- personal insecurity is unimaginable. But economic reconstruction. Much more chal- veloping countries suggests that material there can be no progress without mecha- lenging will be the delicate task of social and poverty and weak democratic structures in- nisms to resolve the social conflict, deal institutional reconstruction. CAUSES OF POVERTY AND A FRAMEWORK FOR ACTION 39 vice delivery, engaging poor people and communities The burden of petty corruption falls disproportionately can have a powerful impact on effectiveness. on poor people, who generally have common cause with an anticorruption agenda. Empowerment Social interactions between individuals and comr- Empowerment means enhancing the capacity of poor peo- munities also have an important influence on poverty ple to influence the state institutions that affect their lives, outcomes. Culture's part in the development process is by strengthening their participation in political processes complex. The beliefs and practices that are part of and local decisionmaking. And it means removing the local culture can be a source of sustainable development. barriers-political, legal, and social-that work against But customary practices and discrimination on the particular groups and building the assets of poor people basis of gender, ethnicity, race, religion, or social sta- to enable them to engage effectively in markets. tus can also be a source of inequality in many countries. Expanding economic opportunities for poor people in- Removing discrimination and managing these divi- deed contributes to their empowerment. But efforts are sions can help reduce poverty. Confronting gender in- needed to make state and social institutions work in the equities is a fundamental part of this, with direct interests of poor people-to make them pro-poor (chap- benefits for women (and men) and instrumental effects ter 6). Formal democratic processes are part of empow- on growth and development. Recent evidence shows erment. As important are the mechanisms through which that greater gender equity is associated with faster everyday state interventions help or hurt poor people. growth (chapter 7). Here, more detailed processes of accountability come into play-mobilizing poor people in their own organi- Security zations to hold state institutions accountable and ensur- Enhancing security for poor people means reducing their ing the rule of law in their daily lives. vulnerability to such risks as ill health, economic shocks, Empowering poor people is part of the broader agenda and natural disasters and helping them cope with adverse of sound governance and accountability of state institu- shocks when they do occur (chapters 8 and 9). tions to their citizens. National empowerment of citizens Poverty reduction strategies can lessen the vulnera- can have important indirect effects on poor people, by in- bility of poor households through a range of approaches fluencing the quality and pace of economic and social de- that can reduce volatility, provide the means for poor velopment. But the outcome for poor people depends on people to manage risk themselves, and strengthen mar- the political and social structures within a society. Gov- ket or public institutions for risk management. The tasks ernments are often more responsive to the concerns of elites include preventing or managing shocks at the national than to the needs of poor groups. So the extent to which and regional level-such as economic downturns and the concerns of nonpoor and poor groups coincide will natural disasters-and minimizing the impact on poor frequently determine whether governance is pro-poor. people when they do occur. Improving governance also requires building admin- Supporting the range of assets of poor people-human, istrative and regulatory capacity and reducing corruption. natural, physical, financial, and social-can help them UGANDA Having emerged from a period of destruc- poverty, impressive efforts toward univer- countability and participation in resource al- tive conflict just over a decade ago, Uganda sal primary education, and a major effort to location and strengthening central and local suffers deep poverty in many dimensions. ensure transparent, poverty-focused bud- state institutions, to provide a basis for But it also shows what an immensely poor gets, both centrally and locally. One of its sound local investment programs in so- Sub-Saharan African country can achieve. main vulnerabilities is health. HIV/AIDS hit cial and physical capital. Tackling the per- The first country to receive enhanced Uganda early: a tenth of adults are now in- ceived risks in the business environment debt relief on the basis of its poverty re- fected, and AIDS orphans are straining tra- so that job-creating growth can take off. duction strategy, Uganda stands out for its ditional systems of fostering children. And furthering current efforts to stop the steady growth in the 1990s. It also stands Three areas are priorities for future spread of H IV/AIDS and such diseases as out for significant reductions in income action. Consolidating and deepening the ac- tuberculosis. 40 WORLD DEVELOPMENT REPORT 2000/200I manage the risks they face. And supporting the institu- Box 2.2 tions that help poor people manage risk can enable them Preventing famines: the local press matters to pursue the higher-risk, higher-return activities that can lift them out of poverty. Improving risk management in- Famines are often the result of crises that affect agricultural stitutions should thus be a permanent feature of poverty production-floods or droughts. How quickly governments respond to such crises depends on many factors. A funda- reduction strategies. A modular approach IS needed, with rsodt uhcie eed nmn atr.Afna reduction stratgies. A modula mental one is the level of democracy and the extent to different schemes to cover different types of risk and dif- which politicians are held accountable for the efficiency of ferent groups of the population. The tools include health relief programs. A recent study in India shows that the dis- insurance, old age assistance and pensions, unemployment tribution of newspapers can play a big part. insurance, workfare programs, social funds, microfi- India has a relatively free press, with only 2 percent of newspapers controlled directly by central or state govern- nance programs, and cash transfers. Safety nets should be ments. The study looked at the interaction between gov- designed to support immediate consumption needs-and ernment responsiveness to floods and droughts (measured to protect the accumulation of human, physical, and so- in public relief funds) and the circulation of newspapers cial assets by poor people. across Indian states. The hypothesis: an informed population can link inefficiency to a particular politician and therefore elicit In addressing risk and vulnerability, the issue once again a greater response to crises. is whether public interventions and institutions work The results confirmed the hypothesis: for a given shock well-and in the interests of poor people. Famines are a (drought or flood), higher newspaper circulation leads to constant threat in many parts of the world, yet some coun- greater public food distribution or relief spending. A 10 per- cent drop in food production due to a crisis is associated with tries have been able to avoid mass deaths. In the 20th cen- a 1 percent increase in public food distribution in the states tury no democratic country with a free press and a free with the median newspaper circulation per capita but a more political opposition ever experienced famine (box 2.2). than 2 percent increase in states in the 75th percentile of Access to information and participation can reduce newspaper circulation. vulnerabiliy. .Separating newspapers by language yields an interest- vulnerability. ing result. Among three types of papers-those in Hindi, Eng- lish, and local languages-only those in local languages Interconnections at local and national levels seem to enhance government responsiveness to crisis. The Just as the dimensions and causes of poverty are interlinked, state governments' responses to local crises are thus very so the areas for action are interconnected. Action to ex- sensitive to the distribution of local newspapers, typically read 50teaesfrato r mecnetd clnt x by the local electorate. pand opportunity is itself a potent source of empowerment, in a deep, intrinsic sense with respect to basic human ca- Source: Besley and Burgess 2000. pabilities, but also instrumentally-for as the asset base, incomes, and market opportunities of poor people increase, so will their potential political and social influence. Im- enough. Global economic advance, access to interna- proving material conditions is also instrumental in en- tional markets, global financial stability, and technolog- hancing security: adverse shocks have lower costs when a ical advances in health, agriculture, and communications person is above the margin of bare survival, and assets are are all crucial determinants of poverty reduction (chap- at the heart of people's risk management strategies. Em- ter 10). International cooperation is thus needed to re- powerment is fundamental in determining action in mar- duce industrial countries' protectionism and avert global ket reforms and the expansion of assets that affect the financial volatility. And the growing importance of such pattern of material opportunities and in shaping the de- international public goods as agricultural and medical re- sign of policies and institutions that help poor and non- search calls for a shift in the focus of development co- poor people manage the risks they face. Finally, reducing operation. Furthermore, because of the importance of vulnerability, with all its debilitating consequences, is international actions in poverty reduction, the voices of central to improving material well-being (or preventing poor countries and poor people should be strengthened reversals) and empowering poor people and communities. in international forums. Country-focused aid programs remain essential-to Inter-national actions help countries implement poverty reduction strategies With global forces having central-and probably rising- that empower poor people, enhance their security, and importance, actions at the local and national level are not expand their opportunities (chapter 11). Aid should be CAUSES OF POVERTY AND A FRAMEWORK FOR ACTION 41 Box 2.3 Attacking poverty in Vietnam Vietnam has made striking progress against poverty, reducing drought) remain ever-present threats. Women, ethnic mi- the share of its population in income poverty from 58 percent to norities, and unregistered urban migrants remain especially 37 percent in 1993-98. A recent analysis by the Poverty Working disadvantaged. Group, with members from government, donors, and NGOs, The analysis-which involved a wide range of stakeholders- found that: identified three areas of policy action: * The main engine of the rapid poverty reduction was reform. * Launch a new round of reforms that unleash the dynamism of Especially important were the land reforms that Vietnam the private sector and create opportunities for employment and launched in the mid-1 980s, which created enormous oppor- productivity growth, so that incomes rise and poor people are tunities for people to improve their lives and livelihoods. able to escape poverty. * Despite the gains, poor people expressed a sense of voice- * Implement the Grassroots Democracy Decree, which aims to lessness and powerlessness. Participatory poverty assess- empower people by authorizing their direct participation in ments (done jointly with Oxfam, Actionaid, and Save the local decisionmaking and improving local governance. Children) found that people were hungry for a two-way flow * Strengthen safety nets and targeted programs, such as the of information-from the government to them about the na- Hunger Eradication and Poverty Reduction Program, to reduce ture and timing of public policies and programs affecting their the vulnerability of poor people to risks (illness, poor harvests). lives and from them to the government to influence those poli- The report on the study has been disseminated widely in Viet- cies and programs. nam, including to all 450 members of the National Assembly and * Poverty remains deep and widespread-and the gains frag- to all 61 provinces. The prime minister has asked the Poverty Work- ile. Millions of people are still vulnerable to poverty. Illness, ing Group to translate the findings into a comprehensive poverty the death of a family member, and natural disasters (flooding, reduction strategy for Vietnam before the end of 2000. Source: World Bank 1999bb. directed to countries with high levels of poverty. But This chapter has presented an overall framework for that should be only part of the criteria for allocating actions in three areas-opportunity, empowerment, and aid. Also essential is having the right policy and insti- security-to reduce poverty in its different dimensions. tutional framework in place to make poverty reduction Actions need to be taken by the full range of agents in a success. society-poor people, government, the private sector, Debt relief for the world's poorest nations, the heav- and civil society organizations-and at local, national, and ily indebted poor countries, has been the most prominent global levels. The country examples in this chapter il- issue in development cooperation in recent years. This re- lustrate three fundamental points: port recognizes that debt reduction must play a central * Actions affecting opportunity, empowerment, and se- part in an overall strategy for attacking poverty. curity are interconnected-there is no hierarchy, The chapters that follow outline different sets of ac- and advances in one area generally depend on gains tions to consider in devising a poverty reduction strategy. in the others. The priorities cannot be set in the abstract. They must * In all cases the social, political, and institutional in- fit the context-and reflect a broad national consensus. derpinnings for action are offundamental importance. Recent experience in Vietnam shows how this process can * Context matters. While it generally is always desirable to be set in motion (box 2.3). take or sustain action in all three areas, the design of ac- tion, and the agents that matter, depend on the economic, social, and political conditions prevailing in the country. PART Il Opportunity CHAPTER 3i Growth, Inequality, and Poverty ,A _ s countries become richer, on average the in- more than tenfold in real terms, in China more than cidence of income poverty falls. Other indicators of fourfold, and in South Asia threefold. The conse- well-being, such as average levels of education and quences for poverty have been dramatic. In the rich health, tend to improve as well. For these reasons, countries of Europe the fraction of the population liv- economic growth is a powerful force for poverty re- ing on less than $1 a day has fallen to nil. In China, duction. This observation is not the end of the story, where growth was slower, less than 20 percent of the for it raises the questions of what causes economic population now lives on less than $1 a day. In South growth and why countries with similar rates of eco- Asia, where growth was slower still, around 40 per- nomic growth can have very different rates of poverty cent of the population does. Today roughly a fifth of reduction. the world's people fall below this austere income Until the mid- 1 8th century improvements in liv- threshold. ing standards worldwide were barely perceptible. Most But differences in rates of economic growth, and societies were resigned to poverty as an inescapable fact in the rates at which that growth translates into of life.' As late as 1 820 per capita incomes were quite poverty reduction, are not the consequence of sim- similar around the world-and very low, ranging pie choices. Countries do not choose to have slow from around $500 in China and South Asia to growth or to undergo painful crises. Nor do they sim- $1,000-i ,500 in the richest countries of Europe. 2 ply choose how equitable growth will be. Instead, the Roughly three-quarters of the world's people lived patterns of growth, the changes in the distribution on less than $1 a day.3 of income and opportunities, and the rates of poverty The onset of modern economic development reduction reflect a complex set of interactions among opened the possibility that growth could significantly the policies, institutions, history, and geography of improve the living standards of poor people-and countries. Understanding the forces underlying coun- everyone else. Over the next two centuries per capita tries' disparate growth experiences, and the mecha- incomes in the richest countries of Europe increased nisms through which this growth has reached poor 45 46 WORLD DEVELOPMENT REPORT 2000/2001 people, is essential for formulating poverty reduction Figure 3.1 strategies. In general, the wealthier a country, the lower This chapter takes up these issues in turn. It first doc- the incidence of poverty uments the strong links between economic growth and Average annual per capita consumption of poorest fifth the income and nonincome dimensions of poverty. It of population next turns to the policies and institutions that under- 1993 u.s. dollars (PPP); log scale pin growth and provide the basis for poverty reduction. 10,000 It recognizes that there are substantial deviations from these general relationships reflecting the wide diversity of country experience-and that these deviations reflect a further set of interrelationships between distribu- 1 tional outcomes, policies, and institutions. It therefore 1 00. discusses how cross-country differences in the poverty- growth nexus are a consequence of initial inequalities in the distribution of income and opportunities -and of changes in the distribution of income that . occur with growth. These inequalities themselves reflect an array of factors, which in turn have consequences for 100 o economic growth. Last, the chapter explores the inter- 100 1,000 10,000 Average annual per capita consumption actions between growth and two nonincome dimensions 1993 U.S. dollars (PPP); log scale of poverty-health and education. Share of population living on less than $1 a day Percent Economic growth and poverty 80 reduction 70 0 Today close to a fifth of the people in the world sur- 60 vive on less than $1 a day. The incidence of this de- 50 privation varies greatly across countries. Not surprising, . the richer the country, the higher the average con- 40 sumption of the poorest fifth of its population-and 30 the smaller on average the fraction living on less than 30 $1 a day (figure 3.1). There are also significant varia- 20 * tions around this relationship. Countries with the same 10* average consumption have quite different proportions . of the population living on less than $1 a day, reflect- 0 1,000 10,000 ing substantial differences in inequality across countries. Average annual per capita consumption Education and health indicators are also better on av- 1993 U.S. dollars (PPP), log scale erage for richer countries. In rich countries fewer than Note: The data cover 65 developing countries and refer to various 1 child in 100 does not reach its fifth birthday, while years in the 1990s. Source: World Bank staff estimates based on data from Chen and in the poorest countries as many as a fifth of children Ravallion (2000). do not (figure 3.2). Similarly, in the poorest countries as many as half of children under five are malnourished -in rich countries fewer than 5 percent. Again, how- [Poverty is] . . low salaries and lack ofjobs. And it's also ever, there can be striking deviations from the average. not having medicine, food, and clothes. For example, the United States is vastly richer than --f-ron: a discusalon groiqp, Brazil China and India, but the life expectancy of African Americans is about the same as that in China and in Still, the stark differences in poverty outcomes between some states in India.4 rich and poor countries point to the central role of eco- GROWTH, INEQUALITY, AND POVERTY 417 Figure 3.2 Figure 3.3 Health indicators improve as incomes rise Poverty trends tracked growth trends in the 1980s and 1990s Under-five mortality rate, 1995 - Per 1,000 live births Average annual growth in per capita consumption of 300 poorest fifth of population * Percent 250 40 * ecce 30 C 200 20 -3 1* . *: 20ercen . * c *t *omlc evelopent l poverty reauctlo. 1 nese Percent10 50 r ct ce c ~~~~~~~-20 Pero capta GDP, 1996 -40 U.S. dollars (PPP); log scale -40 -30 -20 -10 0 10 20 30 40 Source: World Bank data. Average annual growth in per capita consumption Percent Average annual growth in share of population living on nomh c pdevelopment in poverty reduction. These airrer- less t h ens p erally reflect cross-country differences in eco- 80 nomc lgrowth over the very long run. But the benefits 60 ° of growth in reducing income poverty can also be seen over shorter periods. Chapter 1 discusses the highly 40 variable evolution of income poverty across countries in 20 the past two decades. Differences in economic growthco- _ across countries account for much of this variation: as 0ug in the very long run, growth in the 1980s and 199aOs was -20 a powerful force for reducing income poverty. On aver- age, growth in the consumption of the poorest fifth of the population tracked economic growth one-for-one over A60 this period (figure 3.3). In the vast majority of cases growth led to rising consumption in the poorest fifth of -840 -30 -20 -10 0 10 20 30 40 the population, while economic decline led to falling Average annual growth in per capita consumption consumption. Percent The pattern is similar for the share of people living on Note: The data cover 65 developing countries. lessUthan of pove y. Con aversely , lowory adtiv ronl erS- Source: World Bank staff estimates based on data from Chen and less than 1 a day. O average, very additonal per- Ravallion (2000). centage point of growth in average household con- sumption reduces that share by about 2 percent. Although the deviations from this average relationship show that sulting from the collapse of the state, natural disaster, war, in some countries growth is associated with much more or economic crisis can have a devastating impact on poor poverty reduction than in others, the relationship high- people. lights the importance of economic growth for improv- As chapter 1 shows, national poverty figures hide much ing the incomes of poor people and for moving people variation in outcomes within countries. But just as cross- out of poverty. Conversely, low or negative growth re- country differences in economic growth do much to cx- 48 WORLD DEVELOPMENT REPORT 2000/2001 plain cross-country differences in poverty outcomes, re- an important source of poverty reduction, and where it gional and subregional growth does much to explain sub- has not, poverty has often stagnated. Understanding why national poverty outcomes. World regions, countries, and countries and regions have had such disparate growth ex- provinces within countries have grown at very different periences, and how this growth reaches poor people, is es- rates (figure 3.4). Where growth has occurred, it has been sential for formulating poverty reduction strategies. Figure 3.4 Economic growth was a force for poverty reduction in the 1980s and 1990s ... ... across regions of the world ... ... across countries in East Asia ... Average annual change in incidence of poverty Average annual change in incidence of poverty Percent Percent 12 5 * Europe and i Central Asia 8 0 _ Mongolia Philppines 4 \ | 5 tLao PDR * Vietnam * China \ Latin America andi\ \the Caribbean MalaysIa\ o --------- -10 * South Asia Indon -4 , \East Asia -15 and Pacific -8 -20 * Middle East and North Africa Thailand. -12 -25 -5 0 5 10 -5 0 5 10 Average annual growth in per capita GDP Average annual growth in per capita GOP Percent Percent ... across regions in the Philippines ... ... and across provinces in southern Luzon Average annual change in incidence of poverty Average annual change in incidence of poverty Percent Percent " '.Autonomous 20 Mindoro Region *Caraga Region Occidental in Muslim 15 Mindanao-, O -Eastern Visayas------------ ICentral Mindanao -2 Central Luzon* * Ilocos 10 F Cagayan.* * Bicol 4 Csestern Mindanao C i Laguna \ Southern Luzon * Cavite. * Ara -6 Western C vte* a -~~~~~~~~~ ~~~Visay as * 0 Cordillera Southern 0 -8 t Administrative Mindanao lan Romblon Region Northern Marinduque * -10 Mindanao Mindoro Oriental- Q Metro Manila . Batangas o -12 -10 -3 -2 -1 0 1 2 3 4 5 -5 -4 -3 -2 -1 0 1 2 3 4 5 Average annual growth in per capita consumption Average annual growth in per capita consumption Percent Percent Note: The incidence of poverty is the share of the population living on less than $1 a day. Source: Top left panel, World Bank data; top right panel, World Bank 1 998f; bottom two panels, World Bank 2000o. GROWTH, INEQUALITY, AND POVERTY 49 What drives economic growth? Some economic policies-such as openness to inter- national trade, sound monetary and fiscal policies (re- Understanding the policies and institutions that lead flected in moderate budget deficits and the absence of high to sustained and sustainable economic growth is a first inflation), a well-developed financial system, and a modL- step in developing strategies for improving the lot of erately sized government-are also strongly conducive to poor people. Wide divergences in growth reflect the out- economic growth.8 Aid can boost growth if such policies come of interactions among countries' initial conditions, are in place, but not if they are absent.9 Both domestic their institutions, their policy choices, the external and external shocks matter as well. Not surprising, wars, shocks they receive, and no small measure of good civil unrest, and natural disasters all lower growth rates luck. (box 3.2). Less dramatically, so do macroeconomic volatil- There is evidence that growth depends on education ity, adverse terms of trade shocks, and slower growth and life expectancy, particularly at lower incomes.5 For among trading partners.'0 Poorly sequenced and badly example, it has been shown that female literacy and girls' implemented reforms can lead to sudden reversals in education are good for overall economic growth.6 There capital flows or other macroeconomic disruptions, also is also some evidence that rapid population growth is slowing growth (chapter 4). These collapses in growth can negatively associated with per capita GDP growth and be particularly devastating for poor people, who have that the changing age structure of the population can weaker support mechanisms and generally lead a more also affect growth (box 3.1).7 precarious life than the better-off (chapter 9). Box 3.1 Population, growth, and poverty Many studies have documented that as countries become richer, the potential of a growing workforce was due to a variety of fac- both fertility and mortality decline on average, with reductions in tors, including strong educational attainment and a supportive pol- mortality typically preceding reductions in fertility.' The interac- icy and institutional environment. In other regions of the world, tions between this demographic transition and economic devel- notably Latin America, a similar change in the composition of the opment are complex. They have fueled heated debate at least since population occurred without a comparable growth benefit. This 1798, when Thomas Malthus argued that since "food is neces- failure is disappointing since the demographic "bonus" of a larger sary to the existence of man" and "the passion between the sexes workforce is temporary and is followed by a period of higher old is necessary and will remain nearly in its present state" (1985, age dependency rates that place greater demands on the social p. 70), population growth would inevitably lead to an imbalance security institutions that provide support for the elderly. between people and available resources. Second, there is evidence that better education is associ- Malthus's grim prediction on the effects of population growth ated with higher contraceptive use and lower fertility.3 This evi- on economic development failed to materialize-since the turn dence may reflect a variety of mechanisms. More education of the 19th century the world's population has increased more expands economic opportunities for women and so can raise the than fivefold, and thanks to improvements in technology of all opportunity cost of having more children (Becker 1960). Infant mor- kinds, per capita incomes have increased by even greater multi- tality is often lower in families in which women are better edu- ples. The links between demographic change and development cated, and so fewer births are required to achieve a desired are more subtle than this. Two issues are noteworthy: the effects number of children. And better education can improve the ef- of changes in the age structure of the population induced by this fectiveness of contraceptive use, Investments in improving poor demographic transition, and the links between investments in people's access to education and health can therefore have a dou- health and education, growth, and demographic outcomes. ble impact. These investments have been shown to improve First, in many countries sharp declines in fertility have been growth and reduce poverty directly. To the extent that they are followed by sharp increases in the working-age share of the pop- associated with lower fertility and population growth, they can also ulation. In some countries, notably in East Asia, the increase in contribute to a virtuous circle of improved maternal health and bet- the number of workers per capita was accompanied by faster ter investment in children's health and education, which reinforce growth in GDP per capita.2 These countries' success in tapping these gains. 1. See Livi-Bacci (1997) for a historical survey and Birdsall (forthcoming) for a modern review of the literature on demography and economics. 2. For example, Young 11995) provides a careful assessment of the contribution of a growing labor force and greater participation rates to the rapid growth in per capita GDP observed in four Asian economies. 3. Schultz (1994) provides cross-country evidence on the links between female education and fertility. See Feyisetan and Ainsworth (1996) for micro- economic evidence on education and contraceptive use and Ainsworth, Beegle, and Nyamete (19961 on education and fertility. Pritchett and Summers (1994) provide a more cautious assessment of the magnitude of the effect of contraceptive availability on fertility. 50 WORLD DEVELOPMENT REPORT 2000/200I Box 3.2 How war devastates poor people Wars are devastating wherever they occur. Since they occur Nor are children exempt, for they are often recruited to fight. disproportionately in poor countries, the devastation falls dis- Children lucky enough to survive a conflict bear deep psycholog- proportionately on the world's poor people (see figure). More wars ical scars. They also pay a heavy price for their abandoned school- are now civil. During 1987-97 more than 85 percent of conflicts ing in permanently diminished economic opportunities. were fought within national borders (14 were in Africa, 14 in Wars cripple economies by destroying physical, human, Asia, 1 in Europe). Tragically, 90 percent of war deaths are not and social capital-reducing investment, diverting public spend- military (Pottebaum 1999). In Cambodia 1.7 million people died ing from productive activities, and driving highly skilled work- in 20 years of fighting and political mass murder-among them, ers to emigrate. In civil war a country's per capita output falls most of the country's doctors, lawyers, and teachers. Civilian vic- an average of more than 2 percent a year relative to what it tims are also singled out because of their ethnic identity: as would have been without conflict. In more severe and protracted many as 800,000 Tutsis and moderate Hutus were killed by ex- wars, the economic and human costs are even greater (Collier tremist Hutus in Rwanda in 1994. 1999b). Conflict is overwhelmingly concentrated in poor countries Share of incidents of civil war and strife, 1990-95 Percent 100 80 60 40 20 0 East Asia Europe Latin America Middle East South Asia Sub-Saharan Developing High-income and and and the and Africa countries countries Pacific Central Asia Caribbean North Africa Source: USAID, OFDA 1999. Institutional factors are also important for growth. For provide fewer-and poorer quality-public goods, es- example, there is evidence that strong rule of law and the pecially education. Such areas are also more prone to vi- absence of corruption contribute to growth-by pro- olent ethnic conflict. Institutions that guarantee minority viding a fair, rule-based environment in which firms and rights and provide opportunities to resolve conflicts have households can invest and grow."t Strong institutions can been shown to offset the side effects of polarized societies also have powerful indirect benefits. For example, adjusting (chapter 7).13 to adverse shocks often requires painful but necessary Other exogenous factors, such as geography and ini- changes in domestic economic policies. In countries tial incomes, matter as well. There is some evidence that where conflicts between competing interests are pro- geographic characteristics affect growth-for example, nounced, and the institutions to resolve these conflicts a remote or landlocked location acts as a drag on are weak, recovery from shocks is often slower than it is growth.t4 On average, initially poor countries have where these institutions are strong."2 grown more slowly than rich countries, so that the gap Similarly, there is growing evidence that ethnic frag- between rich and poor countries has widened (box 3.3). mentation has adverse effects on growth. Ethnically frag- However, there is strong evidence that, controlling for mented countries and regions within countries tend to some of the factors mentioned above, growth is faster in GROWTH, INEQUALITY, AND POVERTY 5I Box 3.3 Divergence and worldwide income inequality Given the importance of growth for poverty reduction, the failure disappointing, contributed less to worldwide inequality be- of growth to take root in some of the poorest countries with the tween individuals. highest incidence of poverty is particularly disappointing. One Income inequality within countries shows less pronounced symptom of this failure is the widening gap in average incomes trends: in some countries inequality has increased, while in oth- between the richest and poorest countries. In 1960 per capita GDP ers it has fallen. Recent studies have found that across countries in the richest 20 countries was 18 times that in the poorest 20 coun- increases and decreases in inequality are roughly equally likely tries. By 1995 this gap had widened to 37 times, a phenomenon (Deininger and Squire 1996b). Again, however, country size mat- often referred to as divergence (see left-hand panel of figure). ters: changes in inequality in populous countries such as China, Such figures indicate that income inequality between coun- India, or Indonesia will contribute more to changes in worldwide tries has increased sharply over the past 40 years. What has inequality between individuals than will changes occurring in happened to worldwide inequality between individuals? Trends small countries. in worldwide inequality between individuals reflect trends in Trends in worldwide income inequality between individuals both inequality between countries and inequality between in- reflect both these factors, with the between-country component dividuals within countries. The contribution of inequality be- typically more important than the within-country component. In tween countries depends on differences in country growth light of the difficulties with measuring income described in chap- performance and country size: rapid growth in a few large ter 1, it is not surprising that estimates of worldwide inequality and initially poor countries can offset the disequalizing effect between individuals are subject to substantial margins of error. of slow growth in other poor countries. In China, for example, But available estimates indicate that there have been some in- rapid growth from a very low base has helped a fifth of the creases in worldwide inequality between individuals in past world's population halve the gap in average per capita in- decades (see right-hand panel of figure). While the size of these comes with the world as a whole, significantly reducing world- increases depends on the methodology used and the period wide inequality between individuals. In contrast, the 20 poorest considered, the evidence suggests that the increases in world- countries in the world in 1960 accounted for only about 5 per- wide inequality in recent years are small relative to the much larger cent of the world's population, and so their failure to grow, while increases that occurred during the 19th century. Widening gaps between rich and poor countries account for much of the increase in worldwide income inequality across individuals over the past 40 years Per capita GOP Income inequality among individuals 1985 U.S. dollars (PPP) Theit index 20,000 1.00 Bourguignon and Morrisson 1999 15,000 0.75 Richest Milanovic 1999 10,000 20 countries 0.50 5,000 0.25 Poorest 20 countries 1960 1995 1800 1850 1900 1950 2000 Note: The left panel refers to population-weighted averages of per capita GDP in the indicated groups, based on a sample of 123 countries with complete data on per capita GDP over the period 1960-95. China is excluded from the poorest 20 in 1960. The Theil index is a measure of income inequality; higher values indicate higher inequality. Source: Summers and Heston 1991; World Bank data; Bourguignon and Morrisson 1999; Milanovic 1999. 52 WORLD DEVELOPMENT REPORT 2000/200I countries that are initially poor. This relationship may Changes in the distribution of income not be linear, with higher growth kicking in only after How growth affects poverty depends on how the addi- countries reach a threshold level of income. This raises tional income generated by growth is distributed within the possibility of poverty traps at very low levels of de- a country. If economic growth is accompanied by an in- velopment.15 Finally, initial inequality can influence crease in the share of income earned by the poorest, later growth, with implications for how growth translates incomes of poor people will rise faster than average in- into poverty reduction. This important issue is discussed comes. Similarly, if economic growth is accompanied by in the following section. a decline in this share, growth in the incomes of poor What determines the sustainability of growth? In ad- people will lag behind growth in average incomes. dition to the policy, institutional, and geographic factors The same is true for poverty rates. For a given rate of mentioned above, a further important consideration is economic growth, poverty will fall faster in countries whether or not growth is accompanied by environmen- where the distribution of income becomes more equal than tal degradation, which can in turn undermine growth.16 in countries where it becomes less equal. For example, in Environmental degradation can exact a heavy toll on the Uganda growth with rising equality delivered strong economy through poor health and reduced agricultural poverty reduction, while in Bangladesh rising inequality productivity. For example, heavy reliance on coal with- tempered the poverty reduction from growth (box 3.4). out effective controls on particulate, sulfur, and other emis- Another example is Morocco, where the number of poor sions can cause high rates of lung disease, and sulfur people increased by more than 50 percent between 1990 emissions lead to acid rain, which reduces agricultural pro- and 1998, mainly because of declining real per capita pri- ductivity.17 In the long run especially, attending to the vate consumption (-1.4 percent a year). In urban areas quality of the environment and the efficiency of resource the increase in poverty was dampened by a decline in in- use is likely to boost investment, accumulation, and equality, while in rural areas rising inequality reinforced growth. Rapid growth and environmental protection the increase in poverty.19 can go together-because new additions to industrial ca- Does growth itself lead to systematic increases or de- pacity can take advantage of cleaner technologies and ac- creases in income inequality? Do the policies and insti- celerate the replacement ofhigh-pollution technologies.18 tutions that contribute to higher growth increase or decrease inequality? Does the regional or sectoral com- Water is lift, and because we have no water, lift is position of growth affect changes in income inequality? miserable. To answer these questions the chapter first looks at the -4---j-l i Ji:ifsiwlY,a,A.a2il available cross-country evidence-and then turns to more detailed country-specific evidence, which high- Why are similar rates of growth lights the fact that changes in income inequality are often associated with different rates driven by a complex array of opposing forces. of poverty reduction? Many studies show that on average there is no sys- tematic relationship across countries between growth The general relationship between economic growth and and summary statistics of income inequality such as the poverty reduction is clear. But there are also significant Gini coefficient (figure 3.5).2° While this average rela- differences across countries and over time in how much tionship is of interest, so are the substantial deviations poverty reduction occurs at a given rate of economic around it. growth. The bottom panel of figure 3.3 shows that there The differences in inequality at a given rate of growth can be large variation in poverty reduction for the same could reflect the fact that the combination of policies and growth rate in per capita consumption (though extreme institutions that led to this growth differed across values should be considered outliers). What explains countries-and that these differences in policies matter these large differences? For a given rate ofgrowth, the ex- for income distribution. But at the aggregate cross- tent of poverty reduction depends on how the distribu- country level, there is not much evidence that this is the tion of income changes with growth and on initial case. A recent study of growth and poverty reduction in inequalities in income, assets, and access to opportuni- a sample of 80 industrial and developing countries found ties that allow poor people to share in growth. that macroeconomic policies such as a stable monetary GROWTH, INEQUALITY, AND POVERTY 53 Box 3.4 Figure 3.5 Inequality trends and poverty reduction Inequality varied widely in the 1980s and 1990s but showed no systematic association with growth In Uganda growth with rising equality delivered strong poverty reduction ... Average annual growth in Gini coefficient After decades of war and economic collapse, growth re- 20en covered in Uganda in the 1 990s, averaging more than 5 per- . . cent a year. In just six years (1992-98) the share of Ugandans t *X ;* ' in poverty fell from 56 percent to 44 percent. The benefits * . of growth were shared by all income groups, by rural and 0 . urban households, and by nearly all economic sectors. Real .,. ** * per capita consumption rose for all deciles of the population, * *: implying a reduction in poverty regardless of the poverty line. - 0* * **e Modest reductions in income inequality made growth es- .20 pecially effective in reducing poverty, with the Gini coefficient 5 falling from 0.36 to 0.34 during the five years. Living standards improved more among poorer households. Consumption (per -40 adult equivalent) rose 27 percent for the poorest decile, com- pared with 15 percent for households in the richest decile. Among cash crop producers-especially coffee farmers, ini- tially as poor as the average Ugandan-poverty fell more 60 -20 -10 0 10 20 than twice as fast as for the country as a whole. Average annual growth in per capita consumption ... while in Bangladesh rising inequality tempered the Percent poverty reduction from growth Note: The data cover 65 developing countries. In Bangladesh per capita GDP grew at about 2 percent a year Source: World Bank staff estimates based on data from Chen and during the 1990s, and poverty declined quite slowly. Between Ravallion 12000). 1983 and 1996 the share of people in extreme poverty fell _ from 40.9 percent to 35.6 percent-and the share in mod- erate poverty from 58.5 percent to 53.1 percent. Rural poverty in particular remains very high. vors growth and may lower icome iequality by im- Why the slow decline? Part of the answer lies in rising proving access to credit.22 inequality, in both urban and rural sectors, especially be- tween 1992 and 1996, when the Gini coefficient rose from When I retired, I had 20,000 rubles in my savings 0.26 to 0.31. Depending on the poverty measure used, a fifth account. . . . But what the government did with it-the to a third of the potential poverty reduction from growth may have been lost because of higher inequality. If inequality government we truisted with our money! They re-indexed had not increased, the poverty rate would have been about savings so that iniflation ate it! That money is now not 7-10 percentage points lower in 1995-96 than it actually was. enough for hread and water. The higher inequality in Bangladesh does not imply that -From a discussion .noup, IJkrainie growth should not be pursued. To the contrary, faster growth is needed to reduce poverty faster, because growth's net ef- fect on poverty reduction is positive. Also required are efforts Another possible explanation for the lack of associa- to limit rising inequality and to ensure that growth reaches tion between growth and inequality is that countries with rural areas, where many of the country's poor people live. similar overall growth rates could experience very differ- Source: Appleton and others 1999; Wodon 1997, 1999, 2000c. ent changes in income distribution because of differences in the regional and sectoral composition of growth. If growth bypasses poor regions and poor people cannot eas- ily migrate to regions where opportunities are expanding, policy, openness to international trade, and a moderate- growth can lead to rising inequality. If growth is concen- size government raise the incomes of poor people as trated in sectors from which poor people are more likely much as average incomes.21 In other words, these poli- to derive their income, such as agriculture, growth can be cies did not systematically affect income distribution. associated with declining income inequality. Other policies, such as stabilization from high In China much of the sharp increase in income in- inflation, may even disproportionately favor poor peo- equality between the mid-1980s and mid-1990s reflects ple (chapter 9). And greater financial development fa- the much swifter growth in urban areas relative to rural 54 WORLD DEVELOPMENT REPORT 2000/200I areas.23 India's states tell a similar story ofthe importance 9 Market forces, such as changes in the demand for of rural growth in poverty reduction (box 3.5). So does labor. Indonesia.24 A study of 38 developing countries found that * Policies, such as public investment in education. the variation in inequality reflects the abundance of arable * Social forces, such as higher participation of women land, the prevalence of smallholder farming, and the pro- in the labor force or changes in practices discriminating ductivity of agriculture.25 These findings underscore the against women and ethnic minorities. importance of removing policy biases against agriculture A Institutional forces, such as changes in legal restrictions for generating more equitable growth (chapter 4). on the ownership of property by women or ethnic Cross-country evidence can take us only so far in un- groups. derstanding the factors underlying changes in the distri- Not every increase in income inequality should be seen bution of income that make growth more or less pro-poor. as a negative outcome. As economies develop, income Careful country-specific analyses paint a more nuanced inequality can rise because the labor force shifts from agri- picture, highlighting a complex set of reinforcing and culture to more productive activities. For example, if countervailing forces. These include changes in the dis- wages are lower in agriculture than in industry and ser- tribution of education, changes in the returns to educa- vices and the labor force shifts toward those two sectors, tion, labor market choices, and demographic changes (box many summary statistics, especially those sensitive to 3.6). Those changes are the result of: changes at the bottom end of the income distribution, will show increases in inequality despite an overall de- Box 3.5 cline in poverty. These trends should not be seen as neg- What makes growth pro-poor in India? ative if: i The incomes at the bottom rise or at least do not fall. Consistent with cross-country evidence for developing * Thedevelopmentprocessexpandsopportunitiesforall. countries, consumption poverty in India has fallen with the growth in mean household consumption. Moreover, the re- Figure 3.6 gional and sectoral composition of growth affects the na- Initial inequalities influence the pace of poverty tional rate of poverty reduction, with far stronger responses reduction to rural economic growth than to urban. And within rural areas growth in agriculture and services has been particu- larly effective in poverty reduction, while industrial growth Average annual reduction in incidence of poverty has not. associated with 1 percent increase in average per capita has not. ~~~~~~~~~~~~consumption In rural India higher agricultural productivity is crucial Percent for pro-poor economic growth. Data spanning 1958-94 3.0 show that higher real wages and higher farm yields raised average living standards and did not affect income distrib- 2.5 ution. The result: less absolute poverty. The effectiveness of nonfarm growth in reducing poverty has varied widely across states, reflecting systematic dif- 2.0 ferences in initial conditions. In states with low farm pro- ductivity, low rural living standards relative to urban areas, 1.5 and poor basic education, poor people were less able to participate in the growth of the nonfarm sector. The role of 1.0 initial literacy is notable: more than half the difference between 1.0 the elasticity of poverty to nonfarm output for Bihar (the I state with the lowest elasticity in India) and that for Kerala 0.5 (the highest) is attributable to Kerala's substantially higher ini- tial literacy rate. Women's literacy is a slightly more signifi- 0 cant predictor of growth's contribution to poverty reduction 0.2 0.4 0.6 than men's literacy. Initial Gini coefficient For poor people to participate fully in India's economic growth, agriculture, infrastructure, and social spending (es- Note: The data cover 65 developing countries in the 1980s and pecially in lagging rural areas) need to be higher priorities. 1990s. The incidence of poverty is the share of the population living on less than $1 a day. Source: World Bank staff estimates based on the methodology of Source: Ravallion and Datt 1996, 1999. Ravallion 11997a) and data from Chen and Ravailion (2000). GROWTH, INEQUALITY, AND POVERTY 55 Box 3.6 Complex patterns of distributional change in three economies Observed changes in the distribution of income reflect a complex changes in the distribution of education did not offset this. While array of factors-among them, changes in the distribution of as- educational attainment rose faster for the less educated, the re- sets, changes in the returns to these assets, labor market choices, turns to higher education were sufficiently high that the addi- and demographics. Brazil, Mexico, and Taiwan, China, show how tional earnings due to greater education disproportionately favored these forces can reinforce and offset one another to result in in- the more educated. Superimposed on this were important re- equality that is respectively lowered, increased, and unchanged. gional effects, with widening rural-urban real wage differences con- tributing substantially to inequality, despite some convergence of Brazil-inequality lowered urban and rural returns to education and experience. Income inequality declined in Brazil between 1976 and 1996, with the Gini coefficient falling from 0.62 to 0.59. During the same pe- Taiwan, China-inequality unchanged riod the returns to education became more unequal: both wage earn- Noted for its low and stable level of inequality, Taiwan, China, has ers and self-employed workers with more education saw larger had a Gini coefficient of about 0.30 for the past 30 years. As in Brazil, increases in earnings than their less-educated counterparts, even this outcome reflects a variety of opposing forces. Despite a rapid after controlling for age and gender. There were no changes in the increase in their supply, more-educated workers saw larger increases returns to experience and only small declines in the pay gap be- in earnings than less-educated workers. This was more than offset tween men and women, so overall earnings inequality increased. by greater equality in the distribution of education and greater labor This disequalizing effect was more than offset by three factors: market participation by women. The pattern of taxes and transfers X The distribution of education became more equal. was also equalizing, with the effect that the distribution of individ- * Average educational attainment rose from 3.8 to 5.9 years of ual income became more equal. Interestingly, however, income in- schooling, and higher levels of schooling (particularly for women) equality at the household level increased, as many of the new female contributed to a noticeable reduction in family size, with the av- entrants to the labor force came from initially better-off households. erage household falling from 4.3 to 3.5 members. Since fam- ily size fell more for poorer households, inequality fell. * Inequality in the returns to characteristics other than education seems to have fallen, suggesting a reduction in labor market These examples show that simple trends in summary measures segmentation during 1976-96 and a possible decline in regional of income inequality can disguise major structural forces. Some inequalities. of them, such as changes in the distribution of education, can be influenced by policy-though this takes time. Others, such as Mexico-inequality increased changes in the returns to education, reflect primarily market forces Mexico's Gini coefficient rose sharply between 1984 and 1994, and are less amenable to direct policy interventions. And as Tai- from 0.49 to 0.55. As in the previous two examples, changes in wan, China, shows, tax and transfer policies can counter increases the returns to education were a strongly disequalizing force. But in primary income inequality. Source: For Brazil, Ferreira and Paes de Barros (1999b); for Mexico, Legovini, Bouillon, and Lustig (1999); andforTaiwan, China, Bourguignon, Fournier, and Gurgand (1998). * The observed trends are not the result of dysfunctional this: when initial inequality is low, growth reduces poveirty forces such as discrimination. nearly twice as much as when inequality is high (figure 3.6). u The number of poor people falls. Initial inequality in income is not the whole story-- for inequality in other dimensions matters too. The sen- Initial inequality and poverty reduction sitivity of poverty to growth depends a great deal on Even when the distribution of income itself does not initial inequality in poor people's access to opportunities change with growth, countries with similar rates of growth to share in this growth. If disparities in educational at- can have very different poverty outcomes, depending on tainment mirror disparities in income, poor people may their initial inequality. Other things being the same, growth not have the skills to find employment in dynamic and leads to less poverty reduction in unequal societies than in growing sectors of the economy. This effect is coin- egalitarian ones. If poor people get a small share of exist- pounded by gender inequality in access to education ing income and if inequality is unchanged, they will also (chapter 7). In addition, if fixed costs or overt policy bar- get a small share of the new income generated by growth, riers hinder movement from remote, rural, and eco- muting the effects of growth on poverty. Evidence confirms nomically depressed regions to more vibrant urban centers, 56 WORLD DEVELOPMENT REPORT 2000/ZOOI poor people will be less likely to take advantage of op- Box 3.7 portunities to migrate (box 3.7). Diversification and migration in rural China They have alwnys excluded uis Mlavas, theY have For rural agricultural households in China, opportunities for discriiminated against us. They cut down the tree, bitt off-farm employment have been an important source of growth in incomes. These opportunities can be equalizing or *forgot to pll down the roots. That tree is nowsprouting. disequalizing. To the extent that diversification into nonfarm g Oig l il I8-.Uli. 'n1/'~ ( , -'I ;ilemployment reflects a pull factor-higher returns off the farm-diversification can be disequalizing as richer and better- If social inequities-such as caste systems or discrim- educated workers take advantage of these opportunities. To ination against indigenous peoples-confine members the extent that diversification reflects a survival mechanism ination against indigenouspeoplescon emembefor the poorest, it can be equalizing. of disadvantaged groups to employment in stagnant sec- Evidence from four provinces in China suggests that the tors, poor people will benefit less from growth (chapter pull factor has been more important than the survival mech- 7). Or if ethnic discrimination in the marketplace leads anism, with access to nonfarm employment accounting for to different returns to the same level of education, growth a nsing share of income inequality in rural areas between 1985 and 1990. Evidence also suggests that even the modest gap will be less effective in reducing poverty for the group dis- (by international standards) between female and male edu- criminated against. A study in Latin America found that cational attainment exacerbates these trends, with less- in several countries differences in earnings between in- educated women less likely to find off-farm employment. In digenous and nonindigenous people cannot be explained contrast, migration has had equalizing effects on income. Sur- vey data from the four provinces show that private transfers by differences in skills or experience, suggesting that dis- (largely reflecting migrants' remittances) have been an equal- crimination in the labor market may be to blame.26 These izing force. results bring to the fore the importance of eliminating so- cial barriers for women, ethnic minorities, and socially dis- Source: World Bank 1 997b. advantaged groups in making growth broad based. Initial inequality and growth the impact of inequality in assets-and gender High initial inequality reduces the poverty impact of a inequality-is generally clearest. A recent study of sugar given rate of economic growth. It can also undermine cooperatives in India found that those that are most un- poverty reduction by lowering overall economic growth. equal (in land ownership among cooperative members) Early thinking on the effects of inequality on growth sug- are the least productive.77 Various studies have also gested that greater inequality might be good for growth- found an adverse effect of land inequality on growth.28 for example, by redistributing income to the rich, who A study in China found that living in a high-inequality save, from the poor, who do not. This view implied a area reduced growth rates at the farm household level, tradeoff-more growth could be bought for the price of controlling for a household's human and physical cap- more inequality, with ambiguous effects on poor people. ital.29 Other studies have found evidence of a link be- More recent thinking-and empirical evidence- tween education and gender inequality and growth.30 weaken the case for such a tradeoff: lower inequality can In contrast, evidence on the effect of initial income in- increase efficiency and economic growth through a va- equality on subsequent growth is more mixed. Some riety of channels. Unequal societies are more prone to dif- studies have found negative effects.31 Others have found ficulties in collective action, possibly reflected in positive effects.32 Still others have found different effects dysfunctional institutions, political instability, a propen- over different ranges.33 sity for populist redistributive policies, or greater volatil- These results open the possibility that policies to improve ity in policies-all of which can lower growth. And to the the distribution of income and assets can have a double extent that inequality in income or assets coexists with benefit-by increasing growth and by increasing the share imperfect credit markets, poor people may be unable to of growth that accrues to poor people. This is not to say that invest in their human and physical capital, with adverse every pro-equity policy will have such desired effects. If the consequences for long-run growth. reduction in inequality comes at the expense of the other The effects of inequality on growth have been sub- factors conducive to growth (discussed in the early part of jected to considerable empirical scrutiny. Evidence on this chapter), the gains from redistribution can vanish. Ex- GROWTH, INEQUALITY, AND POVERTY 'j7 Box 3.8 Redistribution can be good for efficiency Redistribution need not compromise efficiency and growth. In sev- Universal policies (such as pricing of government services) can eral instances redistributive policies can increase asset accumu- have redistributive and efficiency effects as well. Abolishing sec- lation by poor people-while improving efficiency and growth. A ondary school fees in Taiwan, China, in 1968 and introducing few recent studies illustrate the possibilities for win-win out- compulsory education benefited poorer children more than richer comes, further strengthening the case for redistribution. children (Spohr 2000). It also substantially increased school Land reform is a classic example of a redistributive policy. Op- attainment (0.4 year for males) and labor force participation, trans- eration Barga, a tenancy reform in the Indian state of West Ben- lating into higher earnings (Clark and Hsieh 1999). gal in the late 1970s and early 1980s, is one of the few examples Direct income redistribution (through cash transfers) is rare in of large-scale transfers of property rights not accompanied by developing countries. A concern is that cash may not be spent in major social upheaval. The operation was associated with an 18 the most efficiency-enhancing ways. In South Africa at the end of percent increase in agricultural output in the state (Banerjee, apartheid, the small pension program was dramatically expanded Gertler, and Ghatak 1998). for the black population. In 1993 the pension amounted to twice Redistribution can also be a source of efficiency gains if trans- the median income for blacks in rural areas (Case and Deaton 1998). fers to poor people improve their human capital. Public provision When the pension was received by the maternal grandmothers of infrastructure targeted to poor people is an important example. of girls, it had large effects on nutrition-halving the gap in height Massive primary school construction (61,000 new schools built and between these girls and those of the same age in the United States staffed in five years) under Indonesia's INPRES (presidential in- (Duflo 2000a). Other studies have shown, however, that the pen- structions) program, the main mechanism for redistributing the gain sion, when received by an elderly woman, also led to a reduction from the oil boom in Indonesia, substantially increased education in prime-age male labor supply (Bertrand, Miller, and Mullainathan and income. The primary school graduation rate rose 12 percent, 1999). The results suggest that cash transfers can-but may and male wages 5 percent (Duflo 2000b). not-lead to efficiency gains. propriation of assets on a grand scale can lead to political Across countries, and across individuals within coun- upheaval and violent conflict, undermining growth. And tries, there are strong correlations between health and ed- sometimes attempts to redistribute income can reduce in- ucation outcomes and incomes. Richer countries arLd centives to save, invest, and work. But there are a number richer individuals within countries have lower rates of mor- of win-win possibilities (box 3.8). Policies should focus on tality and malnutrition.34 Within and between counI- building up the human capital and physical assets of poor tries both the quantity and the quality of education people by judiciously using the redistributive power of improve with income-although quality is difficult t-o government spending and, for example, market-based and measure.35 Disparities in educational attainment also other forms of land reforms (chapter 5; box 5.12). decline with income. These strong correlations reflect reinforcing causal Economic growth and nonincome effects from higher income to better health and educa- poverty tion outcomes-and from better health and education to higher income. For individuals, this is not surprising. Just as income poverty declines as average incomes in- Ill health and malnutrition reduce productivity and time crease, so does nonincome poverty, such as in health and spent working, effects that vary with the level of educa- education. Just as with income poverty, there are signif- tion. For example, a study of Brazilian men showed that icant deviations around these general relationships: coun- adult height was strongly associated with wages-and that tries and regions with similar per capita incomes can wages increased faster with height among individuals have quite different outcomes in nonincome poverty as with some (as opposed to no) education.36 Conversely, well. And just as with income poverty, these deviations individuals with higher incomes can better afford to in- reflect a wide array of forces-including initial inequal- vest in health and education.37 Many studies document ity, the effectiveness of public interventions, and the level the positive effects of parental education on children's of development. Conversely, there is strong evidence health and education. that better health and education outcomes contribute to Similar patterns hold for countries, with positive ef- faster economic growth. fects of higher per capita income on infant mortality.38 58 WORLD DEVELOPMENT REPORT 2000/200I Other studies have documented the benefits of lower Figure 3.7 mortality for faster growth, with most of the growth Across countries, the ratio of female to male payoff at low levels of income.39 And we have already seen literacy rises with per capita income the evidence that better education outcomes lead to faster Ratio of female to male literacy rate growth. Average, 1990-98 Moreover, there is some evidence that these relation- 120 ships are not linear, with stronger increases in health as- . 0 sociated with growth in poorer countries and regions. 100 Fairly small differences in economic growth rates can thus have large impacts on human development out- 0 . comes in such countries. One study estimated that had 60 *. t ** S growth rates in the developing world (excluding China e . * and India) been as high in the 1 980s as they were in the 40 * 1960s and 1970s, 656,000 deaths could have been averted 5 . during the 1980s among children under five.40 20 These reinforcing effects from human development to economic development and back suggest the possibility 0 of vicious and virtuous circles. Poor countries and poor 100 1,000 10,000 people can be locked in a vicious circle, as low human Per capita GDP development diminishes economic opportunities, mak- 1995 U.S. dollars (PPP); log scale ing it more difficult to invest in health and education. In Source: World Bank data. contrast, well-targeted public interventions in health and education can contribute to a virtuous circle of greater economic opportunities generating resources for further Nonincome inequalities matter as well. Discrimina- investments (chapter 5). tion by gender and ethnicity-in the allocation of pub- The considerable variations in country experience lic spending for education and health or in the operation around these general relationships again reflect a com- of education and health facilities-can lead to differ- bination of factors. One is inequality in income." We have ences in education and health achievements. Gender dis- seen that the effects of income on health are most pro- parities in educational attainment are especially nounced at low levels of income. This implies that the pronounced in poor countries (figure 3.7). In the Indian same rate of economic growth can have very different state of Kerala-which has a long history of equitable gen- health and education outcomes, depending on the ini- der relations-education and mortality differ little between tial distribution of income and on how that distribution men and women. But in such states as Uttar Pradesh- changes with growth. In particular, growth accompa- where gender discrimination is high-the female literacy nied by a reduction in inequality is more likely to lead rate is less than half the male rate and the female to male to better health outcomes. ratio in the population is a disturbing 87.9 to 100.43 Cross- Research has found evidence that the correlation country studies have also identified geographic factors, across countries between average health indicators and ethnic fragmentation, and especially female educational average income vanishes after controlling for differ- attainment as important in explaining differences in ences in the incidence of income poverty and in pub- health outcomes at a given income.44 Finally, the qual- lic spending.42 The same research has found that ity and quantity of public spending matter as well, though cross-country differences in public health spending the size of the impact on poor people depends greatly on matter more to the health of the income-poor than to supportive policies and institutions (chapter 5). others: the nonpoor are better able to protect their health from lower public spending. These results sug- * * gest that growth improves average health attainments through its ability to reduce income poverty and per- This chapter has shown the importance of growth mit more pro-poor social spending. for poverty reduction, particularly for income and GROWTH, INEQUALITY, AND POVERTY 'j9 human development. It has also shown how low and ple and building up their assets. It also requires mak- declining inequality enhances the impact of growth on ing state institutions work better for poor people, re- poverty. Growth can be made more equitable by re- moving social barriers, and supporting poor people's ducing inequality in access to assets and opportunities. organizations. These issues are taken up in subsequent This requires opening market opportunities to poor peo- chapters. CHAPTER 4 Making Markets Wlork Better for Poor People ..~l..arkets matter for the poor because poor reduction required active participation of the state and people rely on formal and informal markets to sell protection of local industry. This inward-looking, state- their labor and products, to finance investment, and to led development path was adopted by a wide array of insure against risks. Well-functioning markets are im- countries throughout the world, with varying degrees portant in generating growth and expanding oppor- of success. Many countries adopted protectionism, tunities for poor people. That is why market-friendly government control of investment, and state monop- reforms have been promoted by international donors olies in key sectors. In countries such as India this and by developing country governments, especially strategy resulted in persistently slow growth. In other those democratically elected. countries, particularly in Latin America, this strategy But to develop markets and the institutions that sup- initially delivered strong growth through the 1 960s, but port them is difficult and takes time. At times, reforms growth eventually faltered as countries were buffeted to build markets fail entirely. When they succeed, they by oil shocks in the 1970s and the debt crisis of the frequently impose costs on specific groups in society. 1980s. And in China in the late 1 970s there was a grad- When the losers from reforms indude poor people, who ual realization that the economy, especially the agri- are particularly vulnerable to shocks, countries have a cultural sector, had not realized its full potential under special obligation to ease the burden of reform. And even heavy state control. when markets work, societies have to help poor peo- The increasing disenchantment with inward- ple overcome the obstacles that prevent them from looking, state-led development led national govern- freely and fairly participating in markets. ments to implement reforms that replaced state In the 1950s and 1960s many of those shaping pol- intervention in markets with private incentives, pub- icy believed that economic development and poverty lic ownership with private ownership, and protection 61 62 WORLD DEVELOPMENT REPORT 2000/2001 of domestic industries with competition from foreign pro- This chapter addresses these issues in turn. It first ducers and investors.2 Where such market-friendly reforms considers the widely varying experience of countries that have been successfully implemented, on average eco- implemented market-oriented reforms over the past 20 nomic stagnation has ended and growth has resumed. years, highlighting both success stories and the severe con- But in some cases reforms were not successfully im- sequences of failed reforms. It then illustrates the com- plemented, often with particularly severe consequences plex effects that market reforms have on poor people, with for poor people. The broad diversity of failed reforms does examples from three areas: agriculture, fiscal policy, and not lend itself to easy generalization.3 Some reforms pro- trade. Last, it discusses how lightening the regulatory bur- ceeded too quickly and failed for want of supporting in- den, promoting core labor standards, and expanding stitutions. Others proceeded too slowly and were captured microfinance can be beneficial in improving the terms on and undermined by special interests. Yet others were im- which poor people participate in markets. posed by government elites and foreign donors and foundered for lack of strong domestic leadership and a Have market reforms delivered broad-based commitment to reform.4 growth? The debate about reforms is therefore not over a choice between reforms or no reforms: the absence of re- In the 1980s and 1990s much of the developing world forms to develop vibrant, competitive markets and cre- moved toward implementing market-friendly reforms. The ate strong institutions condemns countries to continued motivation for reforms and their scope and pace varied stagnation and decline. Nor is the debate over a simplistic widely. In China, for example, the "household responsibility dichotomy between gradualism and shock therapy: re- system" replaced communal farming and created new forms can proceed either too quickly or too slowly to suc- incentives for rural households to produce, invest, and in- ceed. Rather, the debate is on how reforms to build novate. These reforms were provoked neither by macro- markets can be designed and implemented in a way that economic crisis nor by ideological epiphany; rather, they is measured and tailored to the economic, social, and po- reflected a growing realization that China's agricultural po- litical circumstances of a country.5 tential was not being fulfilled. These initial agricultural re- Inevitably, market-oriented reforms have different ef- forms were followed by the introduction of market fects on different segments of society. Every reform pro- mechanisms throughout the economy. In other countries gram has its winners and losers, and poor people may be macroeconomic crises provided the catalyst for reform: in found in either group. The particular vulnerability of poor Mexico, for example, the debt crisis of the 1 980s was fol- people demands a careful assessment of the likely poverty lowed by the introduction of wide-ranging economic re- impact and the implementation of appropriate compen- forms. And in the countries of Eastern Europe and the sating policies.6 It also calls for careful consideration of the former Soviet Union the political transition precipitated a pace of reforms in the light of the likely effects on poor peo- dramatic progress toward markets that succeeded as spec- ple. Experience shows that direct dialogue with poor peo- tacularly in some countries as it failed in others.7 ple can be particularly effective in informing this process. As a result of this move toward reforms the economic Even when markets function, they do not always serve landscape in many developing countries-but not all- poor people as well as they could. Physical access to mar- has been significantly altered. Government involvement kets can be difficult for poor people living in remote in economic activity has been scaled back. Domestic areas. Regulatory barriers often stifle economic activity in markets are more open to international trade and capi- sectors and regions where poor people are likely to seek tal flows. Revised tax codes are in place. And generally jobs. And access to some markets, especially for financial markets, not governments, determine prices, output, services, can be difficult for poor people since they often and the allocation of resources. Many-but not all-of engage in small transactions, which traditional market par- these reforms reflected the principles of the so-called ticipants find unprofitable or insignificant. Investments Washington consensus, which laid out 10 policy priori- in infrastructure, lighter regulatory burdens, and innov- ties that were adopted in different combinations by many ative approaches to improving access to financial markets countries (box 4.1). can therefore do much to ensure that the benefits of mar- Given the wide diversity of reforms implemented by kets are shared by poor people. different countries at different times and under different MAKING MARKETS WORK BETTER FOR POOR PEOPLE 63 Box 4.1 Figure 4.1 The Washington consensus Indications of successful policy reforms in the developing world The Washington consensus of market-friendly reforms refers to the following 10 objectives of policy: Median annual inflation rate * Fiscal discipline. 25 Percent * Redirection of public expenditure toward education, health, and infrastructure investment. * Tax reform-broadening the tax base and cutting marginal 20 tax rates. * Interest rates that are market determined and positive (but moderate) in real terms. 15 * Competitive exchange rates. * Trade liberalization-replacement of quantitative restric- tions with low and uniform tariffs. 10 * Openness to foreign direct investment. * Privatization of state enterprises. * Deregulation-abolishment of regulations that impede 5 entry or restrict competition, except for those justified on safety, environmental, and consumer protection grounds, and prudential oversight of financial institutions. 0 I Legal security for property rights. Source: Williamson 1993. Median black market premium on foreign exchange Percent 50 circumstances, summarizing overall progress is difficult. Nevertheless, encouraging indicators are clear (figure 4.1). For example, typical inflation rates in developing countries fell from around 15 percent in the early 1980s 30 to 7 percent in 1997, indicating a broad trend toward more disciplined monetary policy. More important, many 20 countries have escaped the scourge of chronic bouts of high inflation and hyperinflation. The black market pre- mium on foreign exchange-a sure indicator of unreal- 10 istic and nonmarket exchange rates-fell from 25 percent for a typical developing country in the mid- 1980s to only 0 1960 1970 1980 1990 1997 5 percent in the late 1990s. Reducing barriers to international trade and capital Note: Data are for all developing countries. The band around the movements has been a central part of many reform pro- central relationship is the 95 percent confidence interval. Data on black market premiums for 1994 and 1995 are unavailable. grams. In Latin America average tariffs were reduced Source: Easterly 2000b. from 50 percent in 1985 to 10 percent in 1996, and max- imum tariffs fell from an average of 84 percent to just 41 percent.8 By 1996 nontariffbarriers affected only 6 per- on average, countries with market-friendly policies such cent of imports, down from 38 percent before reform.9 as openness to international trade, disciplined monetary Reforms have also been widespread in other areas, such and fiscal policy, and well-developed financial markets as liberalizing investment regulations, reducing or elim- enjoy better long-run growth performance than countries inating a large assortment of subsidies to bring down fis- where such policies are absent (chapter 3). cal deficits, and privatizing many state enterprises. Only There is also evidence that reforms that move countries in labor markets have reforms generally been slow. iO closer to such market-friendly policies also contribute to Have these reforms delivered the expected growth better growth performance in the medium term. Cross- payoff? A large empirical literature has documented that, country studies of the impact of reforms typically either 64 WORLD DEVELOPMENT REPORT 2000/200I Figure 4.2 predict cross-country differences in economic perfor- Reforms delivered growth in Latin America, mance. But many developing countries were buffeted by although the gains varied large external shocks. World interest rates rose sharply, in- Additional per capita growth due to reforms in the 1990s creasing the burden of debt service obligations. Growth Percentage points in the industrial countries slowed, lowering growth in their 8 Easterly, Loayza, developing country trading partners. In some cases these 7 and Montiel 1997 shocks eroded the benefits of reforms that were being im- 6 Lora and plemented concurrently. Barrera 1997 At times, however, reform programs have failed to de- 5 Fernandez Ar as liver as much as expected-and at times reforms have 4 and Montiel 1997 r failed entirely. Consider what went wrong in East Asia, 3 countries of the former Soviet Union, and Africa (box 4.2). 2 The grim lessons of these failures, and the heavy burdens 2 l rl l I n f they placed on poor people, underline the importance of 1 I l I a measured and realistic approach to reforms to ensure o_ U _i z _^ L ^ - --- *that their objectives are attained.14 A note of caution on the future of reforms. In many -1. Argentina Brazil Costa Mexico Peru Latin cases the reforms discussed above are straightforward Rica America "first-generation" reforms, such as stabilizing from high Source: As noted in the figure. inflation, moderating chronic budget deficits, and dis- mantling the most egregious trade barriers. Consolidat- ing the gains from these reforms often requires institution compare the performance of countries before and after re- building in much more difficult areas, such as develop- forms or else examine whether changes in measures of re- ing an independent judiciary, creating independent and forms explain changes in growth rates. Reforms are effective regulatory agencies, and instilling professional- measured indirectly as changes in such variables as trade ism in the public sector. Such "second-generation" reforms volumes, tariff rates, inflation rates, or budget deficits. are not only much more complex and take much more Such studies often find a strong growth payoff from reforms. time-they are also often likely to be opposed by pow- Figure 4.2 summarizes the results of three such studies for erful and entrenched interests. 15 This is not to say that Latin America, which found a significant growth impact such second-generation reforms should be postponed- of reforms. Similar studies of the transition economies of precisely because they take time to bear fruit, it is im- Eastern Europe and the former Soviet Union, where suc- portant to embark on them as promptly as possible. cess in implementing market reforms has varied widely, In sum, market-oriented reforms have been wide- found that countries that implemented reforms forcefully spread though uneven throughout the developing world. and early (and enjoyed favorable initial conditions) achieved On average they have delivered lower inflation and higher stronger growth than reform laggards."1 A 1999 study of growth, both powerful forces for reducing income poverty. India finds that the states that implemented reforms saw But reforms can also go awry, with painful consequences faster growth and stronger improvements in education for poor people. Lack of supporting institutions, mistakes and primary health care than those that did not."2 in sequencing reforms, and the capture of the reform This does not mean that the developing world as a process by powerful individuals or groups lie at the bot- whole enjoyed rapid growth as a result of reforms in the tom of most failed reforms. 1980s and 1990s. Indeed, growth in the developing world has been disappointing, with the typical country Have market reforms delivered registering negligible growth. benefits to poor people? A recent study argues that this disappointing growth should not be attributed to the failure of reforms. 13 De- Even when market-friendly reforms have succeeded in de- spite slow overall growth, the study found that differences livering growth, the effects on the incomes of poor peo- in indicators of market-friendly policies continued to ple have varied. This reflects both initial inequalities in MAKING MARKETS WORK BETTER FOR POOR PEOPLE 6(5 Box 4.2 Why do reforms sometimes fail? Reforms can go awry when supportive institutions are absent or nopoly rights and to bias the awarding and pricing of public con- powerful individuals or groups manipulate the results. tracts. State capture runs counter to the premises of a free and fair competitive market economy-and contributes to increasing Incomplete financial sector reforms contributed to the inequality. State capture is also widespread. In several countries East Asian crisis of the former Soviet Union more than 30 percent of firms surveyed During the 1990s several emerging economies in East Asia liber- in a business environment survey reported that they had suffered alized their domestic financial markets and lifted capital account as a result of successful state capture by their competitors (Hell- restrictions. In the Republic of Korea and Thailand especially, a surge man and others 2000). of capital inflows, often through newly formed nonbank financial Market economies cannot function well where the institu- institutions, placed heavy financial stresses on banks. Prudential tional and incentive environment permits such corruption to flour- regulation of banks and nonbank financial institutions did not keep ish. Worse, countries may fall into vicious circles, with incomplete pace with these developments, and there was rapid growth in often- reforms creating new incentives for corruption. Fighting the cor- unhedged short-term foreign currency liabilities. Sudden exchange rosive effects of state capture requires much deeper institutional rate fluctuations in the summer of 1997 wreaked havoc on these development-in the organization of the political system, the foreign currency exposures, contributing to the depth of the en- checks and balances among core state institutions, and the rela- suing crisis (World Bank 1998f). tionships between state and firms and between state and civil This experience matches a broader pattern emerging from society. cross-country analysis: financial reforms unaccompanied by ade- quate supervisory institutions are a significant determinant of Inadequate public investment and excessive bureaucracy banking crises worldwide (Demirgug-Kunt and Detragiache 1998). have undermined market reforms in Sub-Saharan Africa At the root of the 1995 Mexican peso crisis were inadequacies in Several African countries have failed to grow since the mid-1 980s, the bank privatization process and in financial liberalization (Lustig when, with the support of international financial institutions, they 1998). These experiences do not invalidate the importance of re- began implementing market reforms, especially in agriculture. The forms in developing financial markets. In fact, the effective inter- results have been less than spectacular, in part due to inadequate mediation of savings to productive investment was a contributing public investment and persistent red tape (World Bank 2000b). factor in East Asia's remarkable development success, a success African farmers, like those in other parts of the world, re- that dwarfs the setbacks of the recent crisis. But incautious and spond vigorously to price and nonprice incentives. But if public excessively rapid reforms can culminate in crises. infrastructure-such as roads to remote agricultural areas-is un- developed or underdeveloped, the impact of pricing and marketing Grand corruption subverted reforms in countries of the reforms on output is muted. Inadequate infrastructure affects former Soviet Union other sectors as well. Business surveys carried out in a number The state steals from us all the time, so deceiving the state is of African countries in 1996-97 consistently point to the poor not a sin, quality of infrastructure services as a critical barrier to expansion -From a discussion group, Ukraine into labor-intensive exports in response to trade reform. In Uganda transport and other costs increased the cost of capital goods by What kind of government do we have? One hand g~ives and the almost half. And in Zimbabwe poor transport services mean that other takes away! delivery of inputs is unreliable, forcing firms to hold large inven- -From a discussion group, Ukraine tories despite high interest rates. These difficulties have been compounded by a lack of im- In the countries of the former Soviet Union market reforms and provement in transparency and accountability. Although legal and perceptions of corruption are inextricably intertwined (see, for regulatory changes are often integral parts of reform packages, example, Narayan, Patel, Schafft, Rademacher, and Koch-Schulte their implementation is often flawed or half-hearted. As a result, 2000). This is understandable: most of these countries score very regulatory barriers to competition remain serious obstacles, poorly in cross-country comparisons of corruption, and encounters and corruption, red tape, and lack of transparency continue to with corruption are dispiritingly frequent for many firms and indi- impede trade and investment by raising costs. Business surveys viduals. Corruption has coincided with worse macroeconomic often also identify corruption and bureaucratic red tape as bar- performance and deeper output declines as these countries have riers to business expansion and diversification in several African wrestled with the transition to a market economy. countries. For example, it can take more than a week for inter- A particularly pernicious form of corruption is "state capture," mediate inputs to clear customs on the Ugandan border, and de- referring to the ability of firms and powerful individuals to influence lays of more than a day are routine at customs checkpoints in the formation of new laws and regulations to their own advantage. southern Africa. These obstacles are symptomatic of larger in- This may involve manipulating the judicial, executive, and legisla- stitutional failures that policymakers must address if reforms are tive branches of government to obtain special privileges and mo- to be effective. U 66 WORLD DEVELOPMENT REPORT 2000/2001 income and opportunity, and the effects of reforms on Not surprising, case studies of reform episodes show growth and inequality. What has actually happened? that market-friendly reforms have uneven costs and And what can be learned from this experience with mar- benefits-especially in the near term-with the costs ket-friendly reform? concentrated on particular groups and the benefits spread Cross-country evidence suggests that macroeconomic broadly over the economy as a whole. Costs and bene- reforms on average have had little effect on income dis- fits can also be distributed unevenly over time. For ex- tribution. For example, recent studies have examined ample, trade liberalization can lead quickly to reductions the impact of market-friendly policies-such as openness in employment in previously protected sectors, but it may to international trade, low inflation, a moderate-size gov- take time for affected workers to develop the skills required ernment, and strong rule of law-on the incomes of to take advantage of growing opportunities in other sec- poor people in a large cross-country sample. The find- tors. In Hungary the average duration of unemployment ings: these policies on average benefit poor people as for those laid off from state enterprises between 1990 and much as anyone else.16 Some policies, notably stabiliza- 1992 was more than four years.'9 tion from high inflation, may even benefit poor people more than others. This outcome is consistent with sur- Our leaders announced a transition to new market vey evidence showing that poor people are more likely to relations and then left us to the merry offate. single out high inflation as a pressing concern. --Fr ln a discluiOn g0roup, (;coigya Where reforms have adverse distributional effects, these are generally small compared with the growth ben- On the whole these costs do not negate the benefits of efits that reforms deliver, especially over periods of sev- the reforms discussed above. But they do point to the im- eral years or more.17 So the macroeconomic evidence does portance of social policies to ease the burdens that reforms not suggest that the benefits of reform have bypassed poor impose (see chapter 8). This is particularly so for poor peo- people-nor even that the benefits only gradually "trickle ple, whose assets, particularly the human capital of their chil- down" to them. Instead, it suggests a pattern in which all dren, can be irreversibly affected by even short-term costs. income groups on average benefit equally from reforms. The costs also remind us that success or failure is not mea- Even among the countries of the former socialist bloc, sured only by changes in average incomes. Survey evidence where reforms have often gone awry, inequality increased from Latin America indicates that reforms can be unpop- least in countries that successfully implemented reforms. ular if they are associated with the perception-and often It increased most in countries that introduced reforms only the reality-of greater risk and uncertainty.20 partially or not at all.'8 Who wins? And who loses? The winners are often those This kind of cross-country evidence provides only a in rural areas, those in countries where the enabling envi- partial picture of the effects of reforms on poor people. ronment for the private sector is strong and private sector The same reforms may have very different effects in dif- capacity to seize new opportunities is good, those with the ferent countries, and so such average results provide skills to be absorbed into new activities, and those who are only a rough guide to the likely future impact of reforms geographically mobile and willing to look for work in new in a particular country. Furthermore, even when re- occupations and sectors. The losers have often been in urban forms on average have no effect on aggregate income in- areas (where services have been hit), in government jobs, and equality, there will still be winners and losers from in jobs where protected insiders once earned more than reform. And when the main effects of reforms are on the market wages would support. The losers might also indude provision of public goods such as health, education, or the unskilled, the immobile, and those without access to the infrastructure, it may take time before the effects on in- new market opportunities-because they lack human cap- come distribution and human development outcomes are ital, access to land or credit, or infrastructure connecting far- felt. Detailed case studies of reforms in specific countries flung areas. The losers may also indude otherwise viable firms shed light on some of the complexities of reform. While hit by economic crises not of their own making. it is as difficult to generalize from an individual coun- try's experience as it is to generalize from an average cross- As the state sector contracts, employment opportunities country relationship, both types of evidence provide are evaporating. useful insights into the effects of reforms. -From a a'iscussion group, Ukraine MAKING MARKETS WORK BETTER FOR POOR PEOPLE 67 Table 4.1 Impact of reforms on agricultural prices, output, and productivity in seven countries Percentage change, five-year postreform period compared with five-year prereform period Agricultural productivity Real GDP growth Real growth rate (percentage agricultural Real (percentage Agricultural point Country prices exchange ratea point change) output change) Chile 120 105 2.8 40 8.2 Ghana 5 230 3.9 50 12.2 Hungary -10 -23 .. -15 25.4 Indonesia 20 75 -0.6 42 2.3 Madagascar 11 94 2.0 15 2.9 Mexico -24 22 -3.7 14 1.3 New Zealand -31 -2 0.4 5 0.8 .. Not available. a. An increase indicates depreciation. Source: Meerman 1997. Since poor people are represented among both the win- As chapter 5 discusses, access to land plays an important ners and the losers described here, there can be no gen- part in poverty reduction. Better access to land, accom- eral lesson that reforms are good (or bad) for all poor panied by access to such assets as credit and infrastruc- people all the time. But examples of reforms in three ture, can improve the productivity of land and labor f'or areas-agriculture, fiscal policy, and trade-yield im- poor people. Thus liberalizing land markets has large portant insights into what determines success and fail- potential benefits. Evidence from Mexico, for example, ure, how reforms affect poor people, and whether it is indicates that land market reforms expanded small farm- possible to mitigate the adverse effects on losers. ers' access to land through the rental market (box 4.4). Beyond these direct benefits, growth in agricultural in- Agriculture comes appears to have been particularly effective at re- Under inward-oriented models of development the ducing rural poverty because of demand spillovers to structure of tariffs and nontariff barriers and often the local markets in which the nonfarm rural poor have a large exchange rate were biased against agriculture. Market- stake. Rural construction, personal services, simple man- oriented reforms that reduced this antiagriculture bias- ufacturing, and repair have been major channels through and dismantled various forms of state intervention which poor people have shared in agricultural booms, even (price supports, input and credit subsidies, support for when they have not been direct beneficiaries of higher crop marketing products)-have generally increased agri- prices. In Ghana the big beneficiaries of reform-cocoa cultural growth. Policy reforms such as privatization, re- producers-constitute less than 8 percent of the poor, yet duced regulation, and trade and price liberalization rural poverty fell sharply. have had a positive impact for many countries.21 Agri- cultural output and productivity growth have generally All our problems derive from lack of land. If we have risen in the postreform period, sometimes substantially enough land we will be able to produce enough tofeed our (table 4.1). Because many poor people are small agri- househokls, build houses, and train our children. cultural producers, they have benefited directly from -Poor man, Nigeria these reforms. Case studies of Chile, China, Ghana,22 Uganda, and Vietnam show that reforms have helped Another example of the indirect benefits of market re- raise producer prices for small farmers by eliminating forms comes from cotton smallholders in Zimbabwe.23 marketing boards, changing real exchange rates through Before the reforms the Cotton Marketing Board used its broader economic reforms, lowering tariffs, and elim- power as the sole buyer to impose low producer prices on inating quotas (box 4.3). farmers to subsidize the textile industry. Large farmers di- 68 WORLD DEVELOPMENT REPORT 2000/2001 Box 4.3 Agricultural reforms in Chile and China help small farmers Chile dramatically illustrates how incomplete reforms can harm more than doubled and fertilizer production had increased signif- agriculture-and how completed reforms can have large benefits. icantly. But arable land per capita declined from about 0.2 to 0.1 The military government that took power in 1973 implemented a hectare over those 30 years. State-imposed cropping patterns sustained program of policy reform. Agricultural production in- forced most cultivation into rice and other cereals. Collective creased by a quarter in 1974, but then stagnated through 1983, farms had to fulfill grain quotas for delivery to the cities, and the thanks to the uncertainty over future policies and the incom- national grain market was fragmented into 30 self-sufficient regions. pleteness of reforms. In 1978-82 elimination of credit and input Starting in 1979, family farming through the "household re- subsidies and appreciation in the real exchange rate hit agriculture sponsibility system" swept the country, replacing communal farm- hard, while delays in implementing reforms in land, labor, and water ing. Although initially farmers were still obliged to deliver grain at rights markets prevented an effective response (Valdes 1994). low prices, they were otherwise permitted to produce what they In 1984 an aggressive devaluation and completion of reforms wished at mostly market prices. Commerce in rural areas and be- led to a vigorous recovery. The sector responded strongly. Agri- tween farm and city, previously repressed by the state, was allowed cultural labor force participation quickly rose-from a low of 14 to flourish. Rather than have self-sufficient provinces, the state en- percent of the total labor force to more than 19 percent, sub- couraged regional and national markets. Effective demand in- stantially higher than at any time in the previous decade. Agri- creased rapidly for high-value products (vegetables, fruits, meat, cultural growth increased from 0.2 percent a year in 1960-74 to fish, eggs) that had been repressed by the earlier state-directed poli- 4.9 percent in 1974-90. Greater land productivity was a major cies. China's peasant farmers-skilled, hard working, and strongly factor. motivated-responded to the new opportunities with great vigor China's agricultural liberalization led to a swift response. Be- and launched five years of the fastest sustained agricultural growth fore the reforms in 1979, China had good roads and irrigation in- ever recorded anywhere. Between 1978 and 1984 net agricultural frastructure, excellent technical packages for grains and other output increased 7.7 percent annually and grain output 4.8 percent crops, and effective application of fertilizer and other inputs. Be- (Lin 1995). With the vast majority of China's poor people in rural tween the 1940s revolution and the 1970s, irrigation capacity had areas, the incidence of poverty fell dramatically. Source: Meerman 1997. Box 4.4 Land markets and poor peasants in Mexico Liberalization of land rental and reorganization of the property rights the land rental market could be offset by reduced access to credit system in the ejidos (communal lands) in 1992 formed the backbone for the land-poor, who were less able to use land as collateral. This of structural reforms to transform the Mexican economy. Liberal- could have shifted the benefits from smaller to larger holders. izing land markets and better defining and enforcing land property The increased supply of land from large farmers in the rental mar- rights were expected to drastically reduce the costs of transactions ket allowed the rural poor a small but statistically significant in- in both land and credit markets, improving access to land and credit crease in access to land. After controlling for the greater access to for poor, small-scale (and perhaps more efficient) producers. credit, it appears that large farmers increased their demand for Policies aimed at activating land rental markets would bene- land rentals. But small farmers appear to have increased their de- fit the landless and the land-poor by increasing their access to land mand for land even more, suggesting that had their access to credit through rental and sharecropping transactions. But with both land not worsened, land-poor farmers might have benefited even more and credit markets being liberalized, the easing of restrictions on from land market liberalization. Source: Olinto, Davis, and Deininger 1999. versified into unregulated crops, such as horticulture and ' io u k thbe ab, is A 'nerous, bhut what is the incentive tobacco. After the reforms cotton prices rose. In absolute ! X, mOo/lfe wo rhb the fiiutni17 n1eeds if there are nio terms the gains would be greater for larger farmers, simply ,i,A- ' ;'Ouad to gf t protdi'c to the mnarket? because they produce more cotton. But there have been par- .. ,; . u , 'U) '"'l ticular gains for smaliholders, as newly privatized cotton buy- ers have chosen to compete with one another in part by Market-friendly reforms have also sometimes hurt providing new extension and input services to smaliholders. the rural poor. In some countries financial reforms tight- MAKING MARKETS WORK BETTER FOR POOR PEOPLE 69 Box 4.5 Listening to farmers in Zambia Since 1991 Zambia has radically changed the policy and institutional fering from staff shortages and lack of operating funds, transport, environment for agriculture. With liberalization and privatization, and equipment, were also failing to respond well to farmers' needs. private suppliers have replaced state agricultural services for Farmers want better infrastructure (especially roads and credit, inputs, and marketing. bridges) and more effective regulation of the private sector. They Using participatory rural appraisals and beneficiary assess- also want more information on markets for agricultural products ments, the World Bank-assisted Agricultural Sector Investment and easier access to more flexible and responsive credit facilities. Program has established systematic and regular feedback be- And they want advice on subsistence crops and storage methods, tween policymakers, service providers, and those affected by which they prefer to get through group extension. programs. Talking to farmers has helped policymakers under- Talking to farmers also identified ways to help those who are stand the farmers' resource constraints, service delivery problems, economically vulnerable take part in agricultural markets- and strategies for dealing with the vicissitudes of transition. extending microcredit, promoting local seed production systems, Participatory assessments also examine local perceptions of the and offering research and extension services for subsistence effectiveness of agricultural infrastructure and services. These con- crops and low-input agriculture. To create the local organizational sultations revealed that agricultural credit and marketing, now han- basis for participatory extension and economically viable joint dled by the private sector, were uneven and unpredictable-because activities-such as marketing, local financial services, and cattle of poor infrastructure, lack of capacity, and inadequate enforce- dipping-support needs to go to producer associations, service- ment mechanisms. Public extension and animal health services, suf- providing NGOs, and other organizations active in communities. Source: World Bank 1998a. ened credit and closed rural bank branches, reducing These examples suggest at least two lessons. The first the availability of credit.24 And in some cases research, is simple: reforms can benefit poor people but also hurt data collection, reporting, and quality monitoring dis- them. Listening to stakeholders through participatory pol- appeared after the abolition of state enterprises and mar- icymaking can do much to identify and avoid unin- keting boards. In Cameroon the marketing board had been tended consequences for poor people (box 4.5). Second, maintaining rural roads, but the responsibility was not when reforms leave an institutional vacuum, perfor- reassigned after the reforms. In Zambia remote farmers mance suffers. As with other reforms, agricultural mar- had been implicitly subsidized by a uniform pricing pol- ket liberalization without the proper institutional icy that did not take into account transport costs, while framework will not deliver the expected results-and small farmers without storage faciities were implicitly sub- could have serious consequences for poor people. sidized by prices held constant across seasons. After the reforms market forces eliminated the implicit subsidies, Fiscal policy and transport infrastructure deteriorated significantly, In many countries fiscal reforms to strengthen revenue leaving many farmers worse off. collection capacity and control unsustainable spending The converse of the gain to small producers from rel- have been a central element of broader reform programs. ative price shifts is the cost to poor urban dwellers. Take Since raising revenues takes time, fiscal reforms often show Ghana. The rural sector gained from higher export up first in spending cuts. When those cuts are felt in so- prices and greater rural demand as cocoa farmers spent cial sectors and in subsidies, they can hurt poor people. their windfall, but urban residents grew poorer. Living As chapter 5 discusses, there is evidence that the intro- standards in Accra deteriorated in 1988-92, even while duction of user fees in health services hurts the poor conditions were improving elsewhere in the country. more than the rich. In Madagascar the real incomes of Poor and middle-income urban residents suffered from poor households in the capital city declined substan- higher food prices. Moreover, the dismantling of the tially when food prices were decontrolled.25 But elimi- old export marketing system removed an important nating subsidies does not always hurt the poor. A study source of public revenues that was not quickly re- in Guinea and Mozambique found that eliminating food placed. This led to higher inflation and public sector subsidies did not hurt poor people because the subsidies retrenchment, with the costs of both felt most by urban had not reached them in the first place.26 The lesson is residents. clear: lower overall subsidies need not be inconsistent with 70 WORLD DEVELOPMENT REPORT 2000/2001 helping poor people, if the subsidies are better targeted shifted the burden of taxation toward the poor: the tax bur- or replaced by other forms of assistance. den on the richest income group declined 4.3 percentage In the 1 990s governments in Eastern Europe and the points, while that on the poorest group rose 10.3 percentage former Soviet Union introduced a rapid phaseout of util- points.29 In contrast, when tax reforms reduce the re- ity subsidies across the board. The urgency was dictated liance on inflationary finance, they can be progressive by the need to reduce unsustainable fiscal deficits. This because of the heavy burden high inflation places on poor had a huge impact on the welfare of all families, especially people. Moreover, most of the redistributive power of poor families. In Ukraine household energy tariffs in- public finance lies on the expenditure side rather than the creased four- to twelvefold (in real terms) between 1992 revenue side. Therefore, even a slightly regressive tax re- and 1995, while average household income dropped to form can have progressive results if the additional revenue less than half its prereform level. To help cushion the im- is devoted to expenditures targeted toward poor people. pact, a cap of 20 percent of family income was put on what households pay in utility bills and rent. The state Trade budget is supposed to pay any bills in excess of this limit Trade reforms-reducing tariffs and nontariff barriers- (although arrears in payments continue to be a problem). have had profound effects in many developing coun- In Moldova the average winter heating bill would have tries. As chapter 3 discusses, there is now substantial exceeded 60 percent of the income (cash and in kind) of evidence that open trade regimes support growth and de- a typical family of four in the lowest fifth of the income velopment and that moving toward an open regime and distribution living in a small apartment. Aware that this its attendant benefits is the reason for trade reform. But was unsustainable, the government eventually introduced the consequences for poor people depend crucially on how mechanisms to subsidize families, ranging from tolerat- trade liberalization affects the demand for their greatest ing nonpayment to establishing different tariff rates for asset: their (often unskilled) labor. Furthermore, trade re- poor families.27 forms in the developing world have not always been Experience in the countries of the former Soviet Union matched by complementary reforms by rich countries, also shows that fiscal adjustment could have been done where the remaining protection imposes a heavy burden in different, and much more pro-poor, ways. For exam- on the developing world (chapter 10). ple, before the political transition the ratio of health and The initial push for trade liberalization as an instru- education personnel and facilities to the total population ment for poverty reduction was influenced by a narrow was above OECD standards. During the 1990s public rev- reading of predictions from trade theory: removing trade enues and expenditures fell as a share of GDP. And since barriers in developing countries would increase demand GDP collapsed as well, government spending in real for their abundant low-skilled labor and expand unskilled terms fell dramatically. Rather than downsize personnel, employment and earnings. Not only would trade liberal- rationalize facilities, and institute some cost-recovery mea- ization raise average incomes-it was also expected to be sures, governments allowed real public sector wages to particularly pro-poor through this effect on unskilled erode, and spending on maintenance and material inputs labor. The evidence shows that the actual results in the past collapsed. Public sector wages were often in arrears, and 15 years have been mixed. Trade reforms have delivered public employees responded to their personal financial pres- growth, and thus poverty reduction-but their distribu- sures by demanding under-the-table payments for pub- tional effects have been more complex. Careful analysis lic services, something poor people could ill afford.28 suggests three main factors at work. Revenue-raising measures, such as a growing reliance First, in some countries trade restrictions had benefited on value added taxes, can also hurt poor people if not im- poor people by artificially raising the prices of the goods plemented carefully. Strong efficiency arguments for value they produced. In such cases it is not surprising that trade added taxation are being heeded throughout the developing liberalization would hurt poor people. For example, a world. But introducing such taxes can have either pro- study of Mexico found that the wages of unskilled labor gressive or regressive effects. If value added taxes replace relative to those of skilled labor declined over 1986-90, progressive income taxes or if poor people either avoided and that about a quarter of the decline was from the re- or did not qualify for other taxes, such reforms are re- duction in tariffs and the elimination of import license gressive. Pakistan's introduction of a value added tax requirements (figure 4.3).3° The authors explain this ap- MAKING MARKETS WORK BETTER FOR POOR PEOPLE 71 parent anomaly by noting that Mexico, despite its com- Figure 4.3 parative advantage in low-skilled industry, had protected The gap between skilled and unskilled wages labor-intensive sectors-such as textiles and clothing- widened in Mexico before adopting trade reforms. Supporting the incomes of the unskilled through trade barriers is very inefficient. 130 Support can often be given in other ways at lower social cost, although designing and implementing such better- 120 Skilled targeted programs take time. But it is not surprising in these circumstances that trade liberalization-unaccompanied 110 by compensatory programs-would hurt poor people. In some other countries, however, the pattern was different: 100 urban manufacturing workers protected by trade barriers were more skilled and less likely to be poor. 90 Second, some countries that liberalized trade were not particularly abundant in unskilled labor. In Africa and Latin 80 Unskilled America land is relatively abundant, and in Eastern Eu- rope skilled labor is plentiful. Although this does not de- 70 1984 1990 1995 tract from the efficiency and growth arguments for trade reform, it does call into question the earlier presumption Source: Lustig 1998. that trade reform might also deliver equalizing effects by raising the demand for unskilled labor. But in countries where unskilled labor is abundant, such as Bangladesh, skill-intensive employment in the industrial world in China, and Vietnam, the gains from integrating into the the 1 970s and 1 980s is being matched by a similar, later world economy can be significant for unskilled labor, shift in the developing world.33 Third, trade reforms were often accompanied by other This is of course not to say that technological change developments that were disequalizing rather than equal- should be avoided because it hurts poor people. On the izing. In many developing countries that opened to trade, contrary, technological change is a fundamental deter- as in many industrial countries, the wages of skilled work- minant of growth and rising living standards, powerful ers have grown faster than those of unskilled workers. In forces for poverty reduction. Instead, the importance of the United States the wages of unskilled workers have fallen a rising relative demand for skills points to the need to in- in real terms by 20 percent since the 1 970s, despite rapid vest in the skills of poor people, to enable them to take growth in the overall economy.31 Studies for countries as advantage of the new opportunities that technological diverse as Chile, Colombia, Mexico, Turkey, and Venezuela change brings. show a similar phenomenon-premiums paid to skilled workers have increased in all these countries.32 Private sector response Is trade the culprit behind this widening inequality? These examples of agricultural, fiscal, and trade reforms The balance of evidence suggests that it is not. More im- show that reforms can have complex distributional out- portant has been technological change favoring workers comes. But remember that the objective of market-friendly with better education and skills, sometimes in the form reforms-a vibrant and dynamic private sector-can be of imported foreign technologies. This can be seen from one of the most effective antidotes to the costs of reforrn. several pieces of evidence. Even though the relative wages New job creation, technological change that raises labor of skilled workers have risen in many countries, there has productivity and wages, and institutions that ensure equal also been a shift toward greater employment of skilled opportunities for gaining access to the new jobs do much workers-contrary to what simple trade theory would pre- to ensure that the benefits from reform are widely shared. dict. This shift has been pervasive across industries-again Fortunately, a strong private response appears to be the contrary to simple trade models, which would have pre- general experience in developing countries after reform, dicted increases in some sectors and declines in others. especially when labor market regulations are not onerous And there is evidence that the pattern of shifts toward more and do not inhibit adjustment.34 A retrospective study of 72 WORLD DEVELOPMENT REPORT 2000/2001 trade liberalizations found that in 12 of 13 cases where data viewing regulations and exploring possibilities for more were available formal manufacturing employment in- flexible requirements can ease the burden. In Chile the creased within one year after liberalization was com- government recently simplified the duty-drawback sys- pleted.35 The exception was Chile, where increased tem to reduce the administrative costs for small firms. In employment in agriculture offset the decline in manu- Bolivia parts of the tax system were drastically simplified facturing employment. In Estonia a flexible labor market for small firms.40 In the Philippines there are much lower created many new jobs, leading to minimal unemployment minimum capital requirements for small thrift and rural despite the intense job destruction and labor turnover as- banks than for commercial banks. sociated with reform.36 In Panama unemployment fell In contrast, in Indonesia official and unofficial levies steeply after liberalization. In South Asia growth in the for- are estimated to raise the costs faced by small and medium- mal manufacturing sector accelerated from 3.8 percent a size enterprises by as much as 30 percent.41 In some sec- year to 9.4 percent after liberalization, as many workers tors small enterprises have to secure as many as eight pulled out of informal sector employment.37 And in licenses-some of which have identical functions but Africa the micro and small enterprise sector is the most are issued by different agencies. Obtaining licenses takes dynamic in five economies considered in a recent study. so long and procedures are so complicated that some busi- Annual employment growth was strong in these enterprises ness owners choose to operate illegally.42 In the Indian after reform, and new enterprises started up at a high rate.38 state of Gujarat licensing requirements for gum collec- tors are a barrier that hinder women's collector groups.43 Making markets do more Reforms to reduce levies and to simplify and shorten li- for poor people censing and entry procedures for small and medium-size enterprises could ease this burden. Even where market-friendly reforms have taken hold, there Given the opportunity, small and medium-size en- is much that countries can do to improve the benefits that terprises might serve some segments of the markets nor- markets offer to the poor. To reach poor people, many mally thought of as natural monopolies. In many urban reforms need to be accompanied by institutional support, areas in Africa and Latin America small independent investment in infrastructure, and complementary re- water providers bring basic water services at low cost to forms at the micro level. The incentives for policymak- poor marginal communities. Small enterprises have also ers to undertake such reforms are small because the been effective in solid waste management.44 But they markets involving poor people are typically small. So often face barriers-such as requirements for experi- the reforms get little attention, even though they can be ence, complex or expensive procedures for registration powerful forces for poverty reduction. But increasing ac- and tendering, and noncompetitive behavior in markets. cess to productive assets and lightening and improving Removing these constraints could allow small and regulation can do much to involve poor people more di- medium-size enterprises to expand their activities in rectly in markets.39 New technologies can help as well, this area, increasing employment opportunities for low- especially information technologies that break down income groups while expanding access to services for poor some of the barriers of physical remoteness that many poor communities. people face (box 4.6). Better regulation does not always mean less regulation. The potential of reforms to improve access to markets Take the privatizarion of gas, water, electricity, and for poor people can be seen from examples in three areas: telecommunications utilities in Argentina in the early lifting the heavy hand of regulation, especially on the small 1990s. Privatization improved performance, and poor peo- businesses that often provide the poor with employment; ple, as direct consumers, benefited along with the rest of promoting core labor standards; and improving access to the economy-and more than proportionately for gas and financial markets for the poor, especially through electricity, major components of their consumption bas- microfinance. ket. But because privatized utilities are often monopolies, appropriate regulatory institutions were essential to fair Lightening the regulatory burden pricing. The new regulations to ensure that utility prices Compliance with regulations imposes fixed costs that yielded only normal rates of return had important indi- are particularly onerous for small firms. Carefully re- rect benefits for poor people, by encouraging investment MAKING MARKETS WORK BETTER FOR POOR PEOPLE 73 Box 4.6 Attacking poverty with information Virtual Souk expands market access for artisans in the gain access to opportunities for empowerment, capacity building, Middle East and North Africa income generation-and for the use of their skills with dignity. Fadma Aoubaida, a Moroccan weaver from Taliouine and a mother Cellular phone technology gives bargaining power to of seven-with the money she eaMed from selling her products on the Virtual Souk-repaired her roof and started building an in- w door latrine, one of the few in her village. lj]a Aittalblhsen, another I always sell eggs to middlemen. In the past, whatever prices woman artisan in Morocco, spent her profits to buy cement and they offered, I accepted because I had no idea about the windows for her house. With future profits, she wants to buy a going prices of eggs.... Last week, the middleman came ... truck to transport rugs from her village to the market or buy bi- and desired to pay me 12 taka per hali [four units]... Keep- cycles that women can ride. ing him waiting, I rushed to check the prices through the Vil- -BBC Online News, 14 October 1999 lage Phone. The price was 14 taka per ha/i of eggs in nearby markets. I came back and refused to sell to him at the lower Artisans in the Middle East and North Africa have always crafted high-quality products using traditional techniques and ancestral prices. ... After a brief haggling, we agreed to buy and sell know-how. But shrinking local markets and difficulties in gainang ath13ntaka plteaeloa _Halirna Khatuon, a poor, iliiterate woman access to more lucrative national and international markets are lead- who sels eggs, Bangladesh ing to a gradual disappearance of culturally rich crafts-and with them an important source of income for poor people. A subsidiary of Grameen Bank, Grameen Telecom operates a vil- The Virtual Souk is bucking this trend. Since 1997 this Internet- lage pay phone program that leases cellular telephones to selected based marketplace has been providing direct access to international bank members, mostly women in rural areas, who use the tele- markets for several hundred artisans from Egypt, Lebanon, Mo- phone to provide services and earn money. Today around 2,000 rocco, and Tunisia, many of them women. The network is expanding village pay phones are in place. The target is to install 40,000 tele- to other countries in the region, and there is demand to adapt the phones by 2002, introducing telefax and email services as well. concept to East Asia and Latin America. These phones have helped lower the cost of information gath- Online sales soared tenfold between the first and last quarters ering. This can be seen in lower prices for poultry feed, more sta- of 1999, reaching markets around the world, including countries in ble diesel prices, and less spoilage of perishable goods due to more Europe and North America and as far as Australia, Japan, and South precise shipment dates. Women providing the phone services have Africa. Participating artisans receive 65-80 percent of the proceeds, gaii'red confidence and new status as "phone ladies." Telephone a much larger margin than through traditional channels. And the gains users include both rich and poor, but poor people make more calls are more than simply financial. Through the Virtual Souk, artisans for economic reasons. Source: For the Virtual Souk, see www.peopbnk/vsouk/; for the Grameen Telecom cellular phone program, see Burr (2000). and job creation throughout the economy. One study Promoting core labor standards found that these indirect gains for poor people-reflecting Core labor standards have been set out in the Declaration the power of appropriate regulation-were five times as on the Fundamental Principles and Rights at Work large as the direct gains from lower utility prices and bet- adopted by the members of the International Labour Or- ter service.45 ganization in 1998. They include freedom of association An appropriate and generally lighter regulatory frame- and the right to collective bargaining, elimination of work in labor markets could also benefit poor people. In forced labor, effective abolition of child labor, and the elim- general, excessively burdensome labor market regula- ination ofdiscrimination in employment and occupation.46 tions can limit job creation and thus opportunities for poor The goals underlying these core labor standards are im- people to productively employ one of their most important portant, and it is widely agreed that the standards them- assets-their labor. These constraints are especially im- selves represent worthy targets for economic development. portant when reforms in other areas create temporary em- This consensus is especially strong for the most ex- ployment dislocations. But the benefits of deregulating ploitative forms of child labor and forced labor. However, labor markets should not be overstated. Often labor mar- there is no consensus regarding the best way to achieve ket regulations are not well enforced, especially in the in- the labor conditions envisaged by these core standards. formal sector, so relaxing them would have little effect on How best to implement the objectives set out in these stan- employment opportunities for poor people. dards is difficult to determine and depends a great deal 74 WORLD DEVELOPMENT REPORT 2000/200I on the circumstances of individual countries. Some in- tions is unlikely to produce the desired outcomes for dustrial countries take the position that the standards workers.51 Rather, promoting them as part of a broad-based should be enforced through trade agreements or devel- development strategy through information, technical as- opment cooperation. Many developing countries argue- sistance, capacity building, and complementary initia- and rightly so-that applying trade sanctions in this way tives is likely to yield the greatest benefits. Using can serve protectionist purposes for industrial countries incentives-such as programs to keep children in school- and that conditioning development cooperation will un- to address the causes of suboptimal labor practices must fairly hamper development. be a key part of this strategy. Along these lines, and also It is clear that simply adopting core labor standards deserving close attention, are interesting new ideas about will not guarantee their realization. In developing coun- complementing public standards with private (market-dri- tries problems meeting these standards may be a conse- ven) standards that encourage employers to adopt desir- quence of poverty, able labor practices.52 Consider child labor. Too often children's time spent at work comes at the expense of their formal schooling-with Improving access tofinancial markets likely adverse long-term consequences. But a child's earn- for poor people ings may make the difference between survival and star- Access to financial markets is important for poor people. vation for the family, or they may help provide the resources Like all economic agents, low-income households and mi- for a sibling to stay in school.47 In these circumstances sim- croenterprises can benefit from credit, savings, and in- ple bans on child labor can have adverse consequences for surance services. Such services help to manage risk and poor families' incomes and can even have the unintended to smooth consumption in the face of sharp fluctuations effect of pushing children from work in the formal sector in agricultural yields and prices, economic shocks, and to more exploitative work in firms outside the reach of for- even natural disasters. Savings and credit facilities can help mal regulations. As a complement to standards against the to make larger investments more affordable, and so allow most exploitative forms of child labor, programs that pro- people to take advantage of profitable business oppor- vide financial incentives that make it affordable to keep chil- tunities and increase their earnings potential. For dren in school can be a very effective strategy.48 economies as a whole, a large literature has documented Implementation of the standards on freedom of asso- the importance of well-functioning financial markets for ciation and collective bargaining also raises complex issues growth.53 for economic development. Enshrining such rights can help But financial markets, because of their special fea- eliminate abusive workplace practices and ensure fair tures, often serve poor people badly. Asymmetric infor- compensation, especially for poor people, whose desper- mation between lenders and borrowers creates problems ate need for employment places them most at risk of un- of adverse selection and moral hazard. The traditional so- fair and exploitative employers. Unions also are an lution to these problems is for lenders to demand collateral important dimension of civil society, and consultation with from borrowers. Since poor people have insufficient tra- unions can provide a valuable input into policy formula- ditional forms of collateral (such as physical assets) to offer, tion. However, empirical evidence on the economic ben- they are often excluded from traditional financial mar- efits of unionization and collective bargaining is generally kets. In addition, transactions costs are often high rela- quite mixed and suggests that both costs and benefits are tive to the small loans typically demanded by poor people. complex and context specific.49 Particularly important are And in areas where population density is low, physical ac- the rules that govern collective bargaining and resolution cess to banking services can be very difficult: in the of labor disputes. Some forms of collective bargaining rules mountains of Nepal people must walk six hours to and may be better at producing efficient and equitable out- from the nearest bank branch at an opportunity cost of comes than others.50 In any case, the exercise of these rights a day's wages.54 Facing such hurdles, poor people are will best serve development objectives when unions and often discouraged and simply do not seek loans since they employers are knowledgeable and independent and bar- believe that they will be denied credit or will not be able gain in good faith. to fulfill bank requirements. At the same time, conven- The core labor standards, then, set an important tar- tional banks often find it unprofitable to provide services get, but a simple strategy of enforcing them through sanc- to poor people using traditional lending practices. MAKING MARKETS WORK BETTER FOR POOR PEOPLE 75 These failures have been used to justify a high level of ing these risks among microfinance organizations and pos- government intervention in the form of targeted credit, sibly encouraging a greater role for larger and more ge- with government-owned financial institutions channel- ographically diversified and established financial ing sizable resources at subsidized interest rates. Often, institutions can help in this respect. this approach assumed that poor people required only Careful measurement of the economic impact of mi- cheap credit, ignoring their demand for savings instru- crofinance programs or institutions is fraught with ments.55 Outcomes were disappointing. The lending in- methodological difficulties, and the results of studies are stitutions were not financially viable, and in countries from often contradictory.60 Nevertheless, evidence is gradually Indonesia to Peru government-sponsored rural credit emerging. For example, a recent review of 13 microfinance programs collapsed under the weight of their losses. Sub- institutions found that borrower households above or on sidized interest rates distorted the financial markets. Tar- the poverty line experience a higher impact than house- get groups were not reached.56 holds below the poverty line, suggesting that while ef- fective, such institutions are not necessarily well targeted So many lending institutions have emerged, but their toward the poorest households.61 Another study found operations are hardly transparent. People do not know that the majority of microfinance programs reviewed how to access them. Those who have tried have been let still required financial subsidies to be viable.62 Increas- down by high levels of collateral demanded. ingly, the performance of these institutions is evaluated -From a discussion group. Malawi by two primary criteria: their outreach to target clientele and their dependence on subsidies.63 Although these cri- Over the past two decades new approaches known teria do not provide a full assessment of the economic im-- collectively as microfinance have emerged, applying sound pact of microfinance institutions, they highlight the economic principles in the provision of financial services social cost at which microfinance institutions have reached to low-income clients and using group as well as individual their objectives. lending. Pioneers such as Grameen Bank in Bangladesh These results on targeting and the prevalence of sub.- and the village banks (unit desas) of Bank Rakyat In- sidy dependence point to the challenges faced by mi- donesia captured attention worldwide by providing fi- crofinance programs: continuing to move toward financial nancial products matching the needs of low-income viability while extending their outreach to their target clients, using innovative collective monitoring through clientele. Best-practice design features of such institutions group lending to strengthen repayment performance, and as the village banks of Bank Rakyat Indonesia-interest charging interest rates that fully cover operational costs.5' rates that fully cover costs, availability of well-rewarded In many cases these innovations led to much higher re- voluntary savings, performance-based compensation for payment rates than under previous schemes-and were staff, intensive staff training, innovative low-cost distri- particularly effective in reaching women.58 bution networks, frequent loan collection, products While such programs have become popular and rep- matching the demand of low-income groups, and effec- resent a major step forward from previous public inter- tive management information systems-are all associated ventions, they are still no panacea for poverty. Not with good financial performance. Stronger capacity buikl- surprising, simply providing access to credit does not ing and better dissemination of these best practices can create investment opportunities: a study of rural house- help microfinance institutions wean themselves from holds in Nicaragua and Romania found that removing subsidies without compromising their ability to provicde credit constraints would have only moderate impacts on services to poor people. the number of households making investments and on Governments can improve financial intermediation for the amounts invested.59 In addition, small, locally based the poor by providing complementary public goods and microfinance organizations can be particularly vulnera- improved regulation that recognize the special needs of ble to shocks such as natural disasters or fluctuations in microfinance schemes. For example, better investment in agricultural yields, which affect a large proportion of rural infrastructure and literacy promotion can help ex- their clientele at once. This can raise the riskiness of pand the reach of microfinance organizations, and creclit their loan portfolios and make it more difficult for them information registries can lower informational costs atid to provide more sophisticated financial products. Shar- allow borrowers to build reputational collateral. 76 WORLD DEVELOPMENT REPORT 2000/2001 On the regulatory and supervisory fronts outdated forms create winners and losers. And when the losers in- usury laws that prevent microfinance institutions from clude poor people, societies have an obligation to help establishing sufficiently high spreads between savings and them manage the transition. lending rates to allow them to cover the high transactions However, there is no presumption that making reforms costs on small loans should be eliminated. Improving the pro-poor means making reforms slowly. In some cases poor legal framework for secured transactions, as is being people will benefit more from rapid market-oriented re- done in Argentina, Mexico, and Romania, can widen forms, especially in areas that directly affect their economic credit opportunities for low-income people. opportunities or that help break down entrenched mo- nopoly privileges. In view of the urgent need to get coun- tries onto dynamic, job-creating development paths, it is critical that the difficulty of reform and the impossi- Well-functioning markets create opportunities for bility of compensating every loser not lead to policy poor people to escape poverty. But establishing such paralysis. markets where they are absent, making them work bet- Furthermore, to make markets work better for poor ter, and ensuring that poor people have free and fair ac- people, macro reforms must be complemented by micro cess are difficult and take time. At times, market reforms reforms and improvements in poor people's access to fail entirely-or have unintended consequences for poor markets and information-through investment in in- people. The lessons of these failures point to the impor- frastructure and modern technologies-as well as sources tance of designing and implementing reforms in a way of credit. Reducing labor market restrictions that limit that is measured and tailored to the economic, social, and job creation and stifle competition while promoting core political circumstances of a country. Market-friendly re- labor standards remains a key challenge. CHIAPTER 5 Expanding Poor People's Assets and Tackling Inequalities L acking assets is both a cause and an outcome also lack assets because of stark inequalities in the dis- of poverty. Poor health, deficient skils, scant access to tribution of wealth and the benefits of public action. In basic services, and the humiliations of social exclusion Bolivia the under-five mortality rate of the poorest 20 reflect deprivations in personal, public, and social assets. percent of the population is more than four times that Human, physical, and natural assets also lie at the core of the richest 20 percent.' In West and Central Africa of whether an individual, household, or group lives in the rich-poor gap in school enrollment ranges from 19 poverty-or escapes it. These assets interact with mar- percentage points in Ghana to almost 52 percentage ket and social opportunities to generate income, a bet- points in Senegal.2 And in Ecuador 75 percent of house- ter quality of life, and a sense of psychological well-being. holds among the poorest fifth lack piped water, com- Assets are also central to coping with shodcks and reducing pared with 12 percent amonig the richest fifth.3 Poor the vulnerability that is a constant feature of poverty. women and members of disadvantaged ethnic or racial groups may lack assets because of discrimination in the Assets and their synergies law or customary practices. Low assets and low income are mutually reinforcing: low education translates into If we get a road we would get everything else: community low income, which translates into poor health and re- center, employment, post office, water, telephone. duced educational opportunities for the next generation. -Young woman in a discussion group, There are powerful complementarities across assets-- Little' Bay, Jamaica the benefits of one asset can depend crucially on access to another. The synergies between human capital as- Poor people have few assets in part because they live in sets-such as a mother's education and her offspring's poor countries or in poor areas within countries. They nutrition levels-are well documented. In Vietnam 77 78 WORLD DEVELOPMENT REPORT 2000/2001 Box 5.1 Box 5.2 Interactions between human and physical Links between the environment and health capital There are many critical interactions between the environment A study of irrigation infrastructure in Vietnam uncovered im- and poverty. Among the best documented is the link between portant complementarities between education and the gains the environment and the health of poor people. Pollution- from irrigation. The study tried to explain differences in farm dirty water and air -s a major contributor to diarrhea and res- profits as a function of irrigated and nonirrigated land allo- piratory infections, the two most frequent causes of death cations with controls for the observed factors that determined for poor children. the administrative land allocations to households on decol- Research has consistently shown that improving drinking lectivization. Assuming that placement of irrigation is not water has less pronounced health benefits than improving san- based on expected rates of return, the results suggest that itation. But the benefits from cleaner water are enhanced when households with high levels of primary schooling benefit sanitation is improved and water quantity is optimal. When most f rom irrigation. The figure shows how the marginal ben- hygiene is also improved, increasing the quantity of water does efits from irrigation would vary across per capita consump- more to improve health than improving its quality. Education tion expenditures if there were no differences in the education on hygiene is often necessary, though, before communities levels of adults across households. The baseline shows the realize its potential health benefits. gains at actual levels of education and compares those with A study of 144 water and sanitation projects found that the simulated amounts that would result if each head of improved water and sanitation services were associated household had the maximum five years of primary withamedianreductionof 22percentintheincidenceof di- education-or if all adults had the full five years. More edu- arrhea and 65 percent in deaths from diarrhea. But improved cation raises the returns to irrigation, and the effect is par- excreta disposal and hand washing can reduce under-five mor- ticularly strong for the poor, who tend to have the least tality rates by 60 percent and cases of schistosomiasis by education. 77 percent, of intestinal worms by 29 percent, and of tra- Lack of irrigation infrastructure is only one of the con- choma by 27-50 percent. Other work has found significant straints to reducing rural poverty in Vietnam. But the full re- relationships between air quality and health. turns from irrigation investments will not be realized without These critical interactions between the environment and concomitant investments in education. health highlight the importance of working across sectors to improve poverty outcomes. Irrigation delivers greater benefits to Source: World Bank forthcoming b; Klees, Godinho, and Lawson-Doe better-educated households in Vietnam 1999. Marginal benefits from irrigation, 1993 Dong per square meter per year 400 All adults have full primary by 15 percent.5 In Morocco places with better rural roads education also have much higher girls' primary school enrollment rates and twice the use of health care facilities.6 Another important example of the interactions between 300 assets lies in the influence of the environment on health / /Baseline / \ \ (box 5.2). Such interactions suggest that poor health in- dicators in an urban slum, for example, may not be sig- Head of household nificantly improved by a local health center without the has full primary education benefits of an effective sewage system. Increasing human 200 ) 0.5 1 3 6 well-being is thus likely to require action to simultane- Annual consumption expenditure per capita ously expand complementary assets. Millions of dong; log scale Source:vandeWalle2000a. Public action to facilitate the accumulation of assets research found that households with higher education Poor people are central agents in building their assets. Par- levels had higher returns to irrigation, with the largest ben- ents nurture, care for, socialize, teach skills to, and help efits going to the poor (box 5.1).. In rural Philippines elec- finance the education of their children. Small farmers in- trification was estimated to increase the returns to education vest in their land and livestock, while the self-employed EXPANDING POOR PEOPLE S ASSETS AND TACKLING INEQUALITIES 79 invest in materials, equipment, and market contacts im- Box 5.3 portant to their business. Workers migrate to cities and Win-win policies in the health sector to other countries, and their remittances are often used to invest in household assets. Poor women and men save The combined objectives of greater equity and efficiency are in housing, in rotating savings and credit societies, and easier to achieve in some programs than in others, as illus- (where available) in local banks and credit associations. Still, trated in the following two examples from the health sector. (where available) in local banks and credit associations. Still, Some health services, such as mosquito and other pest the scope for asset accumulation by poor households is se- control and health education on basic hygiene and nutrition, verely constrained by inadequacies in the markets they face are pure public goods. Others, such as combating infectious and by weaknesses in public and private institutions for diseases, have clear positive externalities. Still others, such service delivery. It is also constrained by lack of income, as curative care for noninfectious conditions, are private goods. Governments are responsible for infectious disease with poor people suffering severe handicaps in financing control on efficiency grounds. But such policies have important health, education, and other asset-related investments. equity benefits as well. While the poor suffer more from al- Why does the state have a role in expanding poor peo- most all diseases than the nonpoor do, the difference is ple's assets? For two basic reasons. First, markets do not greatest for infectious diseases. In India the poorest tenth of the population is seven times as likely to suffer from tu- work well for poor people, because of their physical iso- berculosis as the richest tenth. lation and because of market failures in the financial, The general inadequacy, if not total absence, of health in- health, and insurance sectors, for example. Second, pub- surance markets in most developing countries exposes both lic policy can reduce initial inequalities and increase the the poor and the nonpoor to substantial financial risk and in- security (chapter 8). (The systemic reasons for this failure were opportunities for poor people to benefit from growth. originally discussed in detail by Arrow 1963.) While public pro- Equity and efficiency considerations can be largely in- vision of insurance is one policy option, managing such pro- dependent, but they generally overlap. Poor people, faced grams is not easy. A common way of addressing the insurance with a failure of a private market, can be left with only the problem is to subsidize expensive care, usually through pub- lic hospitals. But on equity grounds, the value of subsidizing hospitals is not as clear as that of attacking infectious disease. poor will seek alternative solutions, the nonpoor have Hospitals are usually in urban areas, and the nonpoor end up more resources at their disposal and so will clearly have more benefiting more from the subsidies. Subsidized provision of options. Among the most effective antipoverty policies are essential hospital care can enhance equity as well as efficiency. those that achieve more equity through redistribution ad .But it will be a win-win policy-enhancing both equity and the those that achieve more equity through redistribution and efficiency of markets used by poor people-only if poor peo- simultaneously enhance the efficiency of markets used by ple can effectively access hospital care, and that poses a sig- poor people (box 5.3, see also box 3.8 in chapter 3). nificant challenge. While there is a case for state involvement in ex- panding poor people's assets, there is no guarantee that Source: Hammer 1997; Devarajan and Hammer 1998; World Bank panding poor people's assets, there is no guarantee that 1ss8t. the state will be effective. Ineffective state action and unsatisfactory human outcomes partly reflect the fact that government can influence only a few of the multi- ple sources of well-being. But they also reflect the diffi- households and poor communities in decisions on the culty many governments have in delivering goods and choice, operation, monitoring, and evaluation of programs services. Governments are constrained by the fiscal re- and services that build their assets. sources at their disposal and their administrative capac- The rest of this chapter focuses on these three prin- ity to deliver services effectively. Also, even if they have ciples, with examples from different asset categories. It the resources and the capacity, state institutions may not then discusses the important complementarities in set- be responsive to the needs of poor people. ting priorities across areas of action, using land reforn How can public action enable poor people to expand and the provision of urban water and sanitation services their assets, and how can it tackle asset inequalities? In as illustrations. (Expanding access to financial assets is dealt three complementary ways. By using the power of the state with in chapter 4. Mechanisms to improve the account- to redistribute resources, especially in services that cre- ability and responsiveness of state institutions are the ate assets, such as education, health, and infrastructure. focus of chapter 6. How to build social assets is covered By implementing policy and institutional reforms to en- in chapter 7. Using assets and protecting the assets of poor sure effective delivery of services. And by engaging poor people during adverse shocks are discussed in chapters 8 8o WORLD DEVELOPMENT REPORT 2000/2001 and 9. And the role of the international community in good for growth (see box 3.8 in chapter 3). The outcomes asset building is covered in chapters 10 and 11.) of redistributive policies for equity and for efficiency and growth thus need to be carefully evaluated. Redistributing public spending The most important domain for state action in build- ing the assets of poor people is the budget. Evidence The coercive power of the state can be a potent force sup- suggests, for example, that public spending on education porting asset formation by poor people. But there are clear and health is not progressive but is frequently regressive bounds to state action. In today's globally integrated (tables 5.1 and 5.2). This is also an area where redistri- world intrusive state action can undercut the function- bution is technically feasible and where tradeoffs be- ing of markets and the incentives for private investment- tween redistribution and aggregate growth may be low killing job opportunities, not creating them. or negative (indeed, growth can be spurred). But trade- In some instances there are significant tradeoffs bet veen offs also arise in choosing between competing redistrib- efficiency and equity. States generally mirror the un- utive actions on which public funds could be spent. The equal political structure they are founded on, and gov- choice has to be guided by an assessment of the relative ernment action often reflects this. Coercive land reform effectiveness of different instruments in realizing the ob- under highly unequal land ownership, while potentially jectives of redistribution and poverty reduction. good for poor people and good for efficiency, rarely has Budgetary action in support of asset redistribution re- enough political support to be effectively implemented. quires two things. The first is a willingness and capacity to Nationalizing industrial assets is rarely good for poor raise adequate revenues and devote a significant share of those people (who do not enjoy the benefits) and can damage revenues to development (not to military spending, subsi- efficiency. But there are many other instances in which dies to the nonpoor and to loss-making public enterprises, addressing asset inequality can enhance efficiency and be or illicit transfers to foreign bank accounts). The second is Table 5.1 Public spending on education by income quintile in selected developing countries, various years Percent Quintile ~~~~~~1 ~~~~~~~~5 Country Year (poorest) 2 3 4 (richestl Armenia 1996 7 17 22 25 29 C6te d'lvoire 1995 14 17 17 17 35 Ecuador 1998 11 16 21 27 26 Ghana 1992 16 21 21 21 21 Guineaa 1994 9 13 21 30 27 Jamaica 1992 18 19 20 21 22 Kazakhstan 1996 8 16 23 27 26 Kenya 1992/93 17 20 21 22 21 Kyrgyz Republic 1993 14 17 18 24 27 Madagascar 1993/94 8 15 14 21 41 Malawi 1994/95 16 19 20 20 25 Morocco 1998/99 12 17 23 24 24 Nepal 1996 11 12 14 18 46 Nicaragua 1993 9 12 16 24 40 Pakistan 1991 14 17 19 21 29 Panama 1997 20 19 20 24 18 Peru 1994 15 19 22 23 22 Romania 1994 24 22 21 19 15 South Africa 1993 21 19 17 20 23 Tanzania 1993 13 16 16 16 38 Vietnam 1993 12 16 17 19 35 a. Includes only primary and secondary education. Source: Li, Steel, and Glewwe 1999; World Bank 1997i (for Romania), 2000f (for Ecuador). EXPANDING POOR PEOPLE'S ASSETS AND TACKLING INEQUALITIES 8I Table 5.2 Public spending on health by income quintile in selected developing countries, various years Percent Quintile 1 5 Country Year (poorest) 2 3 4 (richest) Argentina 1991 33 6 Brazil 1990 8 18 30 25 20 Bulgaria 1995 13 16 21 26 25 Chile 1982 22 11 Ghana 1994 12 15 19 21 33 Indonesia 1987 12 14 19 27 29 Kenya 1992 14 17 22 22 24 Malaysia 1989 29 11 Mongolia 1995 18 20 19 19 24 South Africa 1993 16 17 Uruguayb 1989 37 21 17 14 11 Vietnam 1993 12 16 21 22 29 a. Distribution across these quintiles not distinguished in original source. b. Quintiles defined by households rather than individuals. Source: Filmer and Pritchett 1999b. the allocation and management of development spending nonpoor. In middle-income developing countries withr to support asset formation for poor people-especially high inequality, the nonpoor are often reluctant to con- human and infrastructure assets. Increasing transparency in tribute their fair share. Unblocking this resistance re- budgets at both the national and the local level can reveal quires actions to build pro-poor coalitions (chapter 6). the extent to which public resources are used in a pro-poor Low-income countries have the added problem of low manner and can improve local accountability (chapter 6). public revenues: in 1997 government revenues in these Public financing of services is a core element of poverty countries averaged about 17.5 percent of GDP (exclud-- reduction policy and practice. Experience suggests two ing China and India).7 Compare that with around 29.6 lessons. First, higher public spending on social services percent for high-income countries. and infrastructure may not translate into more or better One reason for the disparity is the high cost of rais-- services for poor people because programs for poor peo- ing revenues in poor countries, costs that sometimes ple are too often of low quality and unresponsive to their outweigh the benefits of public spending.8 This in turn needs, and because the incidence of public expenditures is explained by a combination of factors: a narrow tax base, is often regressive. However, subsidies to the nonpoor can- high (and distorting) tax rates, weak tax administration, not be fully avoided because gaining political support for and poor public sector governance. So the payoff from quality programs may sometimes require providing ser- tax reforms that seek to broaden the base, lower rates, and vices to a broader segment of the population than just the strengthen revenue collection (often by contracting iv poor alone. Second, it is important to use public re- out to the private sector) can be substantial, making ad- sources to relax demand-side constraints. Even when ditional resources available for effective redistribution. In health, education, and infrastructure services are publicly unequal societies, making tax collection as progressive as financed, poor people face constraints that limit their abil- possible without seriously compromising efficiency is ity to benefit from them (for example, complementary also desirable. For example, inheritance and real estate taxes costs, such as transportation to medical care). could be very progressive. Often, however, the real problem is that the limitecd Raising resources and making public resources are not spent on activities-such as ed- public spending pro-poor ucation, health, slum upgrading, and rural development-- As just noted, effective public redistribution requires a will- that help poor people accumulate assets. Part of the ingness and capacity to raise revenues, especially from the reason is that many low-income countries are simply 82 WORLD DEVELOPMENT REPORT 2000/2001 spending too much on other areas, such as debt service, used more by the poor and with the greatest market subsidies to the nonpoor, loss-making or inefficient pub- failures-typically not the case. For example, education lic enterprises, and the military. In the heavily indebted and health resources go disproportionately to tertiary ed- poor countries more than a fifth of public funds goes to ucation and to hospital and curative care, used more by debt repayments.9 Six heavily indebted poor countries in better-off groups. 15 Africa spend more than a third of their national budgets Several studies confirm that public resources favor on debt service and less than a tenth on basic social ser- the better-off. In Nepal the richest quintile receives four vices.10 Niger spends more than twice as much servicing times as much public education spending as the poorest debt as it does providing primary health care. For several quintile (see table 5.1). In Ghana the richest quintile re- other low-income countries, debt service is not the con- ceives nearly three times the public health spending re- straint because they are not meeting their debt obligations. ceived by the poorest quintile (see table 5.2). 6 Still, debt levels and debt service obligations are unsus- Infrastructure spending also tends to disproportionately tainable for several countries and incompatible with help- benefit wealthier groups.17 Subsidizing electricity in ing poor people accumulate assets. Croatia and water in Russia helps the rich much more Military spending in developing countries fell from 4.9 than the poor. 18 In Bangladesh infrastructure subsidies percent of GDP in 1990 to 2.4 percent in 1995." In sev- for the better-off are about six times those for the poor.'9 eral countries this lower military spending permitted Governments face important political issues in redis- greater spending on health and higher education.12 But tributing public spending to support asset accumulation in other countries-especially those experiencing armed by poor people. With finer targeting, public funds may conflicts or facing unresolved tensions with neighbors- in principle reach more poor people. But such targeting military spending continues to cut into pro-poor spend- may lack political support from powerful groups that may ing. Many such countries have some of the worst health lose out. Hence the importance of building pro-poor and education indicators in the world but spend more than coalitions (chapter 6). This may require allocating some twice as much on the military as on education and health of the resources to actions and programs that also bene- combined. High military spending also has significant fit the nonpoor. costs in lost opportunities for asset building. Beyond Making public spending more pro-poor will involve this crowding-out effect, the destruction of physical and reducing military spending and subsidies to the non- social infrastructure and the slowdown in growth often poor. Privatizing loss-making or inefficient public en- associated with military conflicts further limit asset ac- terprises releases resources that can potentially be used to cumulation and poverty reduction (see box 3.2 in chap- address poor people's needs. Simplifying bureaucratic ter 3). procedures reduces not only wasteful spending but the Spending on the military and on broader security opportunities for corruption and diversion of resources nevertheless has a role in development. The challenge is to illicit activities as well. Prudent macroeconomic man- to inform budget allocation by making an intelligent as- agement can lower debt payments and make space for pro- sessment of the threats that a country faces. Better gov- poor spending. Periodic reviews of overall public ernance and transparency in managing military spending expenditure outcomes can shed light on how efficiently can help keep it in check.'3 So can the peaceful resolu- public resources are used and how well they benefit poor tion of ongoing or potential conflicts. people (see box 9.2 in chapter 9). Between the mid- 1 980s and mid- I 990s public spend- For the poorest countries, domestic actions will not ing on education and health increased in a large num- suffice. These actions will have to be complemented by ber of low-income countries, though slowly. For 118 efforts from the international community to bring about developing and transition economies, real per capita debt relief and expand government resource bases through spending increased on average by 0.7 percent a year for development cooperation (chapter 11). education and 1.3 percent a year for health. Such spend- ing also rose as a share of total spending and national in- Providing services and targeting subsidies come. 14 But allocating more funds to these sectors is not Public spending can provide services directly to poor enough. To support asset accumulation by poor people, people-through the construction of roads, schools, the distribution within sectors must favor basic services health clinics, or water supply schemes. But redistribu- EXPANDING POOR PEOPLE'S ASSETS AND TACKLING INEQUALITIES 83 tion can also be achieved by relaxing demand-side con- income households spend 14-20 percent of their budgets straints for poor people by subsidizing the consumption on water-costs reflecting the minimal water infrastruc- of privately provided services and covering complemen- ture and the higher cost of the small quantities poor peo- tary and opportunity costs. The Voices of the Poor study ple buy.24 A cubic meter of water from private water shows how the cost of services can prevent poor house- vendors in Port-au-Prince, Haiti, costs 6-1 0 times as much holds from obtaining them (box 5.4). as a cubic meter from the public water service.25 Similarly, Case studies confirm the cost constraints that poor peo- high connection costs prevent poor households from en- ple face in accumulating a wide range of assets. In rural joying energy services.26 In rural areas, connecting to an areas in the Kyrgyz Republic 45 percent of patients sold electricity grid can cost $20-1 ,000.27 In too many cases assets (produce or livestock) to pay for hospital care.20 In- poor people simply do not have the choice of consuming direct costs to households in forgone income from child cheaper water and energy from a commercial network.,8 labor or in household chores no longer performed by Redistribution, by providing services for free or subsi- children who are in school can also be significant.21 In rural dizing their demand, can help poor people expand their as- Madagascar, where access to water is poor, the high op- sets. Free primary education for poor people is critical for portunity cost of girls' school attendance in time spent expanding their human assets, especially for girls. Similarly, fetching water significantly reduces girls' education.22 In subsidizing prevention of infectious diseases and helping poor Uganda primary enrollment nearly doubled in the households finance the costs ofcatastrophic health episodes 1997/98 school year when the requirement that parents need to be key elements in strengthening poor people's health pay half the cost of school fees was lifted and parent- assets and reducing their vulnerability to health shocks (see teacher association levies were banned.23 box 5.3; chapter 8). Poor people often pay enormous amounts for infra- In both education and health services-even when they structure services. In Nouakchott, Mauritania, most low- are provided for free-demand-side subsidies can help Box 5.4 Locked out by health and education fees Whethertoseekmedicaltreatmentoreducationfortheirchildrenpre- Difficulties with paying school fees and other costs associ- sents agonizing choices for poor people. Among participants in the ated with sending children to school are also widely reported. A Voices of the Poor study, illness was the most frequent trigger of a mother from Millbank, Jamaica, explained that she could not send slide into deeperpoverty. Nha, a 26-year-dd father in Vietnam, reported her six-year-old daughter to school because she could not afford that he had had to sell four buffalo, a horse, and two pigs to pay for the uniform and other costs. Another daughter had to drop out his daughter's operation. The operation failed to cure her, and the need of school because the family could not afford the $500 for school forfurther treatmenttransformed his family from one of the most pros- fees. The woman said, "My son will be ready for school in Sep- perous in the community to one of the poorest. tember but I can't see how I'll be able to send all three of them In Pakistan many households reported that they had borrowed to school." large sums of money, sold assets, or removed a child from school In some countries children are pulled out of school because at least once to cover medical costs. Said an old woman from Ghana, fees are due when families can least afford them. In Ethiopia pay- "If you don't have money today, your disease will take you to ments are due at the start of the school year in September, a time your grave." of two important festivals and the harvest. Amadi, a 14-year-old Although the greatest fear for poor people is the risk of large boy in Nigeria, said that he had been in and out of school because hospital fees, illegal payments for primary care can also be painful. his parents could not pay his school fees regularly and promptly. Corruption in health care is widely reported. Poor women from He missed his promotion exams several times and remains in pri- Madaripur, Bangladesh, said that the doctor in the government mary school while others his age have gone on to secondary health care center ignored them, giving preferential treatment to school. patients wearing good clothes and to those who could afford side In formerly centrally planned economies the cost of school- payments referred to as "visit fees." A study participant from ing is a serious concern for poor families because education was Vares, Bosnia and Herzegovina, exclaimed, "Before, everyone free in the past. People also reported problems with teachers could get health care. But now everyone just prays to God that soliciting bribes and special "tutoring fees" in exchange for pass- they don't get sick because everywhere they just ask for money. " ing grades and diplomas. Source: Narayan, Chambers, Shah, and Petesch 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 2000. 84 WORLD DEVELOPMENT REPORT 20001200t poor families invest further in the human capital of their cause of social stigma attached to declaring oneself indi- children (to cover transport costs, for example). To increase gent.33 Sometimes private and nongovernmental providers access to education for girls or minority, indigenous, or poor are in a better position to implement sliding scales, since children, public funds-in vouchers, stipends, scholar- they frequently know their patients' background and have ships, grants, and so on-are paid directly to individuals, an incentive to charge what the market will bear.34 institutions, and communities. In Bangladesh the govern- For water and energy many developing countries use ment pays stipends covering 30-54 percent of direct school increasing block tariffs, charging a low tariff (often below expenses for girls in grades 6-10. In Colombia in the past cost) for the first block of consumption and rising tar- poor children received public vouchers to attend the sec- iffs for additional blocks. In Asia 20 of 32 urban water ondary school of their choice. In Balochistan, Pakistan, com- utilities use this tariff structure.35 Such tariffs appear to munity grants are provided for girls to attend community be more equitable (since they force firms and wealthier schools. These programs raise the demand for education consumers to subsidize consumption by poor house- among poor households. Mexico's Progresa,29 for example, holds). They also discourage waste. But there are prob- has boosted enrollments among beneficiaries compared lems. In many developing countries few poor households with similar families not in the program (box 5.5). are served by network utilities, and governments choose An alternative to transfers is subsidies on the price of large initial consumption blocks, putting most of the fi- services. Few developing countries, however, have suc- nancial benefit in the hands of middle- and upper-income cessfully implemented price discrimination in health ser- consumers. Ironically, increasing block tariffs discrimi- vices through sliding scale fees.30 In most African countries nate against poor households that share a water connec- such exemptions tend to benefit wealthier groups (such tion with several other households, because even if the as civil servants).31 In Ghana's Volta Region in 1995 less consumption of each household is low, total consump- than 1 percent of patients were exempt from health user tion is high. In addition, tariffs charged to industries fees, and 71 percent of exemptions went to health service have often been so high that they choose to self-provide, staff.32 In Indonesia and Vietnam poor people can have undermining the financial viability of service providers.36 user fees waived through an affidavit of indigence, but few Still, in countries where network access is high, a well- people seem to take advantage of this-partly perhaps be- designed increasing block tariff can outperform cash trans- fers administered by poorly funded social protection offices, Box 5.5 as in some countries in Eastern Europe and the former So- Mexico's Progresa: paying parents to send viet Union.37 But in countries where poor households children~~~~~~~~~~~~ to Buschooliswer oo oseod children to school have limited access, the subsidy of choice should be sup- port for a connection, not for consumption. One way to Mexico's Progresa, an integrated poverty reduction program mitigate high connection costs is to extend credit to poor initiated in 1997, subsidizes education, health care, and nutri- 38 tion for poor rural households. It aims to reduce current poverty users. Another IS to subsidize all or part of the connec- and increase investment in human capital, breaking intergen- tion fee. Infrastructure subsidies can also be made more pro- erational poverty. Progresa covers 2.6 million iamilies-about poor if financed through the general budget or through 80 percent of the population in extreme poverty in rural areas. industry levies in ways that are not discriminatory.39 This Progresa provides grants to poor families for each child under 18 enrolled between the third grade of primary and the approach, compatible with free entry, provides strong i- third grade of secondary school. The grants increase for higher centives to serve the poorest if the subsidy is paid to the grades and are slightly higher for girls than boys. For a child provider only after service has been delivered. in the third year of secondary school, grants are equal to 46 How redistribution is best achieved through transfers percent of the average earnings of an agricultural worker.r Families of children who miss more than 15 percent of the a p s i vi wn school days in a month do not receive the grant that month. identifv the poor and administer subsidies. If it is possi- Progresa has pushed up enrollments at all levels, with the ble to identify poor people individually, any number of largest effect (17 percent) on the transition from sixth grade policies can help in redistribution. Direct cash payments, to the first year of secondary school (traditionally when many or the subsidy of any good at all, are fine if the benefits children drop out). can be restricted to poor people alone. But it usually is Source: IDB 2000. not possible to tell precisely who is eligible, necessitating more indirect means (box 5.6). EXPANDING POOR PEOPLE'S ASSETS AND TACKLING INEQUALITIES 8 5 Box 5.6 Once countries have settled the political problem of howv Some general principles on how to design much should be redistributed and the more technical ques- subsidies tion of what is to be redistributed, the next step is to make Even when poor people cannot be identified individually by ad- sure that services do in fact reach the poor. How can poor ministrative means, subsidies can be designed to reach the poor. people get effective delivery of the services they need to * Self-targeting. Programs can be designed to ensure "self- form assets? The old model of universal state provisioll selection"-say, by paying wages below prevailing mar- too often fails because of lack of financial and adminis- ket rates. The Maharashtra Employment Guarantee trative resources or the failure to respond to poor peo- Scheme in India relies on providing work that only poor people would find attractive (Ravallion 1 999a; see box 8.9 ple's needs. in chapter 8). Part of the problem may be technical and logistica:l. * Geographic targeting. Subsidies can go to specific loca- Poor people often live in remote, low-density rural areas tiQns, so that rural and remote areas receive most of the that are expensive and difficult to serve. Resources for benefits. This works best if the correlation between poverty and location is high-less well if poor and non- poverty reduction may simply not stretch far enough in poor live close together. It also works best if the subsidy these environments. is attached to goods that are hard to transport, such as But the problem is quite frequently management and direct services in education and health. motivation, with inadequate incentives for conscientiouLs * Commodity targeting. Subsidies should go to commodi- ties that poor people consume proportionately more of service delivery (chapter 6). For services that require the than other people, ensuring that they will receive most presence of an educated professional-education, health of the subsidy. Food and primary education usually rank care, judicial services-it is often difficult to induce skilled high on this criterion. civil servants to live in remote or rural areas.42 With chil- dren of their own, they often resist living where the edu- cational and cultural opportunities are limited. Besides Pooling risk through insurance is another way to address geography there is another kind of "distance" between cost constraints on demand for health care. Several middle- providers of services and the poor. Since doctors, teachers, income countries are pursuing universal health insurance and judges are highly educated, they are often from very (chapter 8). Chile managed to reach the 15 percent of its different social classes than the communities they serve, popuLation not covered by social insurance by creating a na- making interaction difficult and strained. If pay is deteir- tional health fund (Fonasa) that collects both payroll de- mined by civil service rules and differentials for difficult ductions for social insurance and a general revenue subsidy postings do not fully compensate for living conditions, it for health care. Still, public resources may be better spent is very difficult for the public sector to serve poor people.'L3 and poor people may benefit more if governments focus on Public provision is generally only part of the picture insuring against catastrophic health incidents-which most for services supplied to poor people, however, and other poor households are less able to finance.40 Social insurance actors can often overcome the limitations on state pro- schemes, even when intended to be universal, frequendy serve vision. Religious groups often do much in providing ecd- the better-off first, with poor people receiving coverage ucation. NGOs are also a major force in many countries: late. Indeed, before coverage becomes universal poor peo- in Bangladesh such agencies as the Bangladesh Rural ple may suffer-since the demand and prices for private care Advancement Committee play a substantial role in de- can increase as a result of the insurance program, as was the livery, with better results than the government's. And the case in the Philippines.41 private sector has always delivered services. More than half the health services in developing countries are private. In Institutional reforms for effective Bolivia almost three-quarters of visits to health clinics for delivery: governance, markets, treatment of diarrhea or acute respiratory infections are and competition to private facilities.44 The private sector also began play- ing a bigger role in infrastructure provision in the 1990s. I heard rumors about assistance for the poor, but no one Sound governance, competition, and markets-and seems to know where it is. free entry for multiple agents, whether government, -Fromn a discuission gro1up. nongovernment, or private-are essential for effective Tanjugrejo, Indonesia service delivery, especially to poor people. (Indeed, 86 WORLD DEVELOPMENT REPORT 2000/200I nonpoor communities and people are more likely to paying extra allowances for hardship posts. None of them make effective use of state systems.) This is not an is problem free. issue of the state versus the market, but of the use of In recent years there have been more attempts to de- different agents and mechanisms depending on the centralize health services to subnational levels of gov- type of activity. In education the national curriculum ernment. This also changes the incentives for providers and exams are a public function, but multiple agents because they have to satisfy a different set of employers. can provide schooling and communities can hold teach- Local governments may be more responsive to feedback ers accountable. from clients. But the jury is still out on the benefits of The mix of state and market-and the mix of agents- decentralization in health. Sometimes decentralization has depends on the nature of the service and the institu- simply shed responsibilities from the central govern- tional context.45 Where governments are weak, there ment-not an example of good decentralization. Suc- might be a stronger case for open entry and reliance on cessful decentralization relies on increased participation private and nongovernment agents. But this, too, may re- of people as monitors of quality (see next section on quire more monitoring and regulatory capacity than the participation). government can muster. The importance of institutional Rather than provide services directly, governments reforms, good governance, and markets in providing can make better use of the private sector and NGOs. quality services to poor people is illustrated here with ex- In many countries even very poor people prefer to amples from health and telecommunications. spend money on services from the private sector (or from NGOs) if they perceive the quality to be higher Improving the delivery of health services than that of public services.46 This preference can be exploited by changing the role of government from Sometimes I stayfor long hours until I can see one of the provider to financier. doctors, then afterwards the nurse comes and tells me Effective partnerships draw on the strengths of each that he is not coming or he came but he will not be able sector-public, private, nonprofit-in improving pro- to see me. vision of health services to poor people. The World -l'ooir womall, FlAl ailail, ! t/ Health Organization's immunization program has been a remarkable success (box 5.7). In Brazil, by having Despite impressive advances in health in recent decades, NGOs compete for funding, the government has har- and despite the potential effectiveness of policies and nessed the private sector's energy and expertise in the bat- programs, health services often fail to reach poor people. tle against AIDS and other sexually transmitted diseases. A fundamental problem: it is difficult to maintain staff NGOs can often reach segments of society that shy away in rural areas and to ensure conscientious care by those from official contact yet run the highest risk of HIV in- who do show up. In countries as diverse as Brazil, India, fection. By the end of 1994 NGOs financed under this Indonesia, and Zambia, staff vacancies in health posts are system had distributed an estimated 2.6 million con- much more numerous and last longer in poor and rural doms and taken 11,000 calls to hotlines. Stringent gov- areas than in richer and urban ones. ernment supervision has also been important in ensuring Improving health services in poor communities might that all but 4 of the 191 NGO-run projects financed have involve changing the incentive structure for public gone forward without a hitch.47 providers, switching from public provision to public fi- Governments might rethink the types of services they nancing of private or NGO providers, or changing the choose to offer, based simply on what they can credibly type of services the government is committed to offer, fa- promise. Maintaining permanent staff in rural primary voring those whose delivery is easier to manage. health care clinics may be too difficult for some gov- Changing incentives in the public sector is often dif- ernments to manage. Changing the mode of delivery or ficult, with civil service rules often tightly constraining the types of services may be called for. India recently in- hiring, firing, promotion, and pay (chapter 6). Different troduced a campaign to combat polio, reducing reliance methods, all with their own risks, have been used, such on permanent clinics with short trips by medical per- as allowing private practice to supplement incomes, mak- sonnel to rural areas-with good results.45 Similarly, ing education subsidies contingent on public service, or rather than relying on permanent health staff who are EXPANDING POOR PEOPLE'S ASSETS AND TACKLING INEQUALITIES 87 Box 5.7 reform, the number of fixed lines had increased more than Effective public-private partnership 165 percent, the number of mobile lines had risen fromn in immunization 20,000 to nearly 500,000, and the number of locales with access to telephones had more than doubled. Between In 1974 the World Health Organization (WHO) launched the 1995 and 1996 the share of households in the poorest Expanded Immunization Program, aiming for 80 percent cov- quintile with telephones increased from 1 percent to 7 erage of children under five by 1990. Although the program started as a WHO initiative, it soon included many multina- percent.54 tional organizations, and volunteers, private entrepreneurs, Large and small providers can offer services side-by- and government workers did the actual work within coun- side, facilitated by a wide range of innovative technolo- tries. The Rotarians, for example, raised more than $240 mil- lion to provide polio vaccine to some 500 million children in 103 countries. Private manufacturers also took part, provid- telecommunications services in Senegal, costs dropped and ing largevolumes of vaccine at a low price to the United Na- access more than doubled (box 5.8).55 Provision of pay tions Children's Fund's global distribution network and still phones can greatly enhance poor people's access to making a profit. telecommunications services, in countries The results have been spectacular. By 1990 the program t ele phon call rates particularnynecontrges had achieved its goal of immunizing 80 percent of children where telephone call rates are low but connection charges against the most common childhood diseases. The initiative are high.56 Advances in cellular technology have also succeeded because it combined public, private, and multi- dramatically increased access to telecommunications in national efforts, with each organization using its comparative countries where laws and regulations encourage geo- strength to fit each country's circumstances. graphically widespread coverage. Grameen Telecom, a Source: van der Gaag 1995. nonprofit in Bangladesh, uses cellular technology, coni- bined with the entrepreneurial talents of rural women, to provide services to villages (see box 4.6 in chapter 4). difficult to monitor and motivate, governments might India, Peru, South Africa, and Thailand have seen dra- find infrastructure projects providing clean water and san- matic growth in privately owned and operated telecen- itation both easier to manage and more likely to improve ters, providing rural inhabitants with new sources of health conditions.49 And rather than fighting doctors' information and new opportunities.57 preference to work in hospitals, governments might Because private providers focus on the most prof- choose to focus more on gaining access to hospital ser- itable market segments, some pockets of the population-- vices for poor people.50 particularly poorer groups-may not receive access because of the high cost and low revenue potential of extending Providing telecommunications services to poor people Box 5.8 Given the right policies and regulations, the private sec- Local entrepreneurs increase access ror is well positioned to provide telecommunications to telecommunications services services to poor people. Better communications, bring- ing new influences and broader views of the world, can Allowing local entrepreneurs to offer telecommunications raise the earnings of poor people. 51 In Sri Lanka telephone services is an important first step in lowering the costs of pub- service in rural areas increased farmers' share of the price lic access Pay phones in particular benefit those who can- received for crops sold in the capital city from 50-60 per- more than 2,000 private telecenters, each with a telephone cent to 80-90 percent.52 and many with a fax machine-four times the number just Since the mid-i 980s developing countries have been two years before. By 1998 it had 6,000. Sonatel (the Sene- opening telecommunications to private participation galese public telecommunications company) franchises phone service to the telecenter owner, who may charge a tariff up and competition.53 The arrangements range from private to 140 percent above the Sonatel price per call unit. On av- investment in publicly owned companies (China) to erage, telecenters have paid $3,960 to Sonatel and kept complete privatization and widespread competition, leav- $1,584 each year. The result: public access to telephones has ing the state to focus on regulation (Chile). Private par- more than doubled. ticipation has generally resulted in rapid growth in access, Source: Ernberg 1998; CSIR 1998. lower prices, and better service. In Peru five years after 88 WORLD DEVELOPMENT REPORT 2000/2001 service to them. To avoid leaving out poor people, in- ture of benefits by local elites (chapter 6). In Bangladesh novative public-private partnerships and well-targeted the extent to which food-for-education transfers go to poor government subsidies may be needed.58 Chile used gov- or nonpoor households depends on the relative strength ernment resources to improve access for low-income of organizations for the poor and the nonpoor.62 Simi- households, people with disabilities, and public schools, larly, some local values may be inimical to some groups health centers, and libraries and auctioned subsidies to of poor people-as with biases against women in many private providers to pay for rural telecommunications roll- parts of the world, against lower castes in India, or against out. In about half the chosen locations, bids to provide other disadvantaged ethnic or social groups. Shame, de- service did not require subsidies as initially expected; the nial, and stigma over HIV/AIDS are in some countries demand analysis done by the privatization group per- reasons for local inaction. suaded private investors of the profitability of providing The complexities of participation imply that it needs services in these areas. Within two years 90 percent of roll- to be fostered by actions that strengthen the voice of poor out objectives had been achieved for about half the ini- groups in confronting social stratification or stigma. tial budget. This implies that participation needs to be shaped in a Despite successes in extending telecommunications broader institutional context. Local government is in the services to poor people, privatization is unlikely to sig- middle of the picture, with core interactions between mu- nificantly increase access in the absence of greater com- nicipalities and communities-as in the design of the De- petition and more effective regulation to prevent abuse centralization and Popular Participation Laws in Bolivia of market power. To make private participation pro- in the past decade. But local governments often need to poor in telecommunications-and in infrastructure be strengthened and made accountable: too often dis- more broadly-policymakers may need to refocus reg- empowered in the past, they face difficulties of weak ca- ulations and transaction processes. A study of telecom- pacity and local capture. Civil society organizations can munications provision in 30 African and Latin American also increase the influence of poor people and poor countries found that strong competition is correlated with communities. These broader issues are taken up in chap- per capita increases in mainlines, pay phones, and con- ters 6 and 7; here the importance of participation in en- nection capacity-and with decreases in the price of abling poor people to expand their assets is illustrated local calls. It also found that well-designed regulation was with examples from education, local infrastructure, and important in improving connection capacity.60 A study forest management. on infrastructure reform in Argentina suggests that public-private partnerships can, with the right policies Increasing local participation and regulations, also improve access to infrastructure for and accountability in education poor people.61 Parents and local communities are demanding more of a say in children's education. As education systems have Participation: choice, monitoring, expanded in many developing countries, concerns have and accountability mounted about the quality of instruction. Central struc- tures are weak in dealing with daily administrative tasks The third principle for public action to promote asset ac- and too distant to take effective action against teachers cumulation involves engaging poor communities and who do not perform. poor people. Participation has three main objectives: Community participation in primary education fre- * To ensure that the preferences and values of commu- quently focuses on monitoring teacher performance and nities are reflected in the choice and design of inter- ensuring the availability of school supplies. In the func- ventions. tions most suitable for local management-in-service * To use community and participant monitoring to im- training and pedagogical supervision-teachers unions prove implementation, transparency, and accountability,. can complement local parent-teacher associations. But * To give poor people more influence over their lives. teachers unions often fiercely oppose devolving control Participation, while potent, is no panacea. Depend- of hiring and firing to local levels, because that has often ing on local organizations and power structures, shifting resulted in delayed salary payments and at times abuse influence to local communities can lead to greater cap- by local officials, inciting teacher strikes in Nigeria and EXPANDING POOR PEOPLE'S ASSETS AND TACKLING INEQUALITIES 89 Zimbabwe.63 Ongoing reform efforts in Nicaragua are improvement.65 There is also evidence that supervision seeking to overcome such problems (box 5.9). incentives are the most cost-effective input for rural and The overall trend in education is to decentralize. In indigenous schools. Nicaragua's experience with school 1993 Sri Lanka established school development boards autonomy reform indicates that a higher level of deci- to promote community participation in school manage- sionmaking by schools is associated with higher stude.nt ment. In Bangladesh school management committees test scores, particularly in schools exerting greater aa- have been reactivated by the Social Mobilization Cam- tonomy in teacher staffing, monitoring, and evalua- paign, to involve communities in education. El Salvador tion.66 In Nigeria teachers have shown up on time since started involving rural communities in school manage- local supervision was introduced.67 And social assessments ment in 1995.64 In several communities in developing of Brazil's Minas Gerais program and El Salvador's Corn- countries parent groups have responsibility for hiring munity Participation in Education program (Educaci6n and firing teachers and for supplying and maintaining con participaci6n de la comunidad, or Educo) indicate equipment, under contract from education ministries. So that as a result of the programs, teachers meet more far, however, there has been little experience with full-scale often with parents and are regarded more highly by decentralization of teacher management to schools. them. Their attendance is also better, which lowers stu- One of the few available quantitative impact evalu- dent absences.68 ations, for the Primary Education Project in Mexico, Other evidence suggests that community manage- shows that educational achievement improved signifi- ment of education can increase efficiency. Preliminary r e- cantly during decentralization and that the lower the ini- sults for the Philippines show that primary schools thLat tial achievement level, the greater the scope for rely more on local support have lower costs, holding quality and enrollments constant.69 In Mauritius parent- Box 5.9 teacher associations have been so successful that govern- Local participation in Nicaragua's decentralized ment funds are being used to further stimulate this education system partnership.70 Effective community management in education may, Since 1993 Nicaragua's Ministry of Education has been de- however, be hard to achieve. Finding qualified people to centralizing public primary and secondary education to local manage schools can be difficult, and the results are un- management boards, based on the following model: even. Botswana has had trouble attracting talented peo- * Legal responsibility for public education rests with the 1 71 ministry, but some teacher management is delegated to pIe to school boards. In the Zambezia provice of other levels of the system. Mozambique parent management committees have led * The center controls teacher preparation, establishes staffing to beneficial partnerships between communities and levels, funds teacher salaries, and sets standards for schools in some villages, but in others they barely func- teacher qualifications and pedagogical performance. It . . . also drafts regulations and financial controls. tion. Many villagers are afraid to openly criticize school * The departmental level is responsible for supervisory func- staff, and committees have been co-opted by corrupt of- tions, providing pedagogical supporttoteachers, and mon- ficials.72 There is also evidence that community in- itoring compliance with standards. volvement may have little impact where adults are barely * Municipal education councils composed of local repre- literate.73 Despite Educo's success in expanding access in sentatives discharge the administrative functions dele- El Salao D ite has os deliv her acheveent gated to the local level. These councils pay teacher salaries El Salvador, it has not delivered higher achievement (with central funds) and approve teacher appointments, scores than traditional schools in the poor rural coim- transfers, leaves of absence, and dismissals in accord munities that were the top priorities.74 with central laws and regulations. They also oversee Overall, experience suggests that a strong regulatory teacher incentives and issue incentive payments to eligi- ble teachers. framework is needed and that training parents is vital to * Teachers are accountable for what happens in their class- make local monitoring of schools effective. Many other room. Parents monitor their attendance and report to concerns about greater household involvement in eclu- the municipal education councils, which inform teachers cation can be addressed through public funding.75 Con- weekly of their status in relation to incentives. tinued monitoring and evaluation of local participation Source: Gaynor 1998. in education can tell much about what works and whiat does not. 90 WORLD DEVELOPMENT REPORT 2000/2001 Fostering ownership through participation minority-must be able to voice their demands. Local and choice in local infrastructure communities reflect existing social, ethnic, gender, and economic divisions, and unless the question of who con- The policy of the party is that the people know, the people stitutes the community is understood and addressed up discuss, the people do, but here people only implement the front, men and local elites may dominate decisionmak- last part, which is the people do. ing and capture project benefits. In rural Sub-Saharan --Fromza diciissiow g1ro0p, -Fa 7T01b. V'iet;u,an Africa men often identify roads as high-priority inter- ventions, while women, when consulted, prefer to improve Community involvement in planning and managing the footbridges and paths that make up the local trans- local infrastructure services can greatly increase owner- port system on which they rely.79 In a village water sup- ship and sustainability-if communities make informed ply project in rural India, water supply points were placed choices. While local infrastructure is scarce in most de- near influential households.S° And in Honduras benefi- veloping countries, the infrastructure that exists is often ciary assessments showed that in places where the social poorly conceived and maintained. Why? fund had financed piped sewerage, the choice of better- Past efforts to provide local infrastructure have often off households with water connections, most community failed to involve communities in key decisions, with members had wanted roads and bridges.8' Provider central ministries deciding on what local communities agencies-local governments, NGOs, project facilitators- needed and communities learning of a project only can reduce the risk of capture by elites, but these agen- when the bulldozer showed up. Another problem has cies sometimes also try to capture benefits. been a lack of choice. Faced with take-it-or-leave-it, few Requiring beneficiaries to share in the cost of invest- communities turn down a free or heavily subsidized in- ment can also improve ownership."2 Contributions usu- vestment.76 This can kill local ownership, and upkeep of ally come more readily when the communities and local the investment becomes somebody else's responsibility. governments responsible for operation and maintenance The incentive structures for agency staff can also com- are given a voice in design and implementation. Local con- promise community participation-when the prefer- tributions vary. In Ghana communities contribute 5 per- ence is speed in implementation. True community cent of the cost for improved water systems and 50 participation processes take time-it took years for the percent for sanitation systems. Cost sharing in menu- Orangi community in Pakistan to agree on the sewers driven or social fund-type projects is typically between it wanted to install77-but they generate ownership and 5 and 20 percent of project costs."3 Significant financial sustainability. contributions-between 20 and 55 percent of project Involving beneficiaries in decisionmaking is the start- costs-have been suggested as important for sustain- ing point in creating local ownership of infrastructure ability in a study of rural water projects."4 There is also assets, ownership that is important in three key dimen- compelling evidence that ownership is a function of the sions. First, it helps in choosing priorities, particularly in institutional relationship between communities and ser- areas difficult to tackle through economic analysis, such vice providers (chapter 6).85 as the relative value of social and productive investment, There is a trend toward providing local infrastructure the complementarity between investments, and targeting through community-based multisectoral approaches.Y6 By within communities.78 Second, it is essential for good op- giving greater choice, such approaches have the poten- eration and maintenance, because governments can rarely tial to respond better to the priorities of each commu- be relied on to perform timely upkeep of local infra- nity, contributing to ownership and sustainability.87 But structure. Third, given the precarious budget situation of it may not be desirable to offer open menus in every pro- most developing country governments and the vast in- ject. If there is a critical need for institutional or policy frastructure needs (and total costs), local ownership is re- reform in water or transport, for example, multisectoral quired for community cost sharing in investments and approaches are unlikely to address it (box 5.10). Similarly, operation. some types of infrastructure involving more than one com- For the participatory process to generate ownership, munity, such as roads linking many communities, will all groups in the community-men and women, those rarely be demanded by individual communities even if well represented in the community and those in the they are needed. Such infrastructure is thus best supplied EXPANDING POOR PEOPLE S ASSETS AND TACKLING INEQUALITIES 9I Box 5.10 Single-sector and multisector arrangements for improving rural roads in Zambia In Zambia in 1997 there was a critical need to clarify the institu- ers of each road. Improving the roads owned by different levels of tional arrangements for managing and financing rural roads. Many government took a vertical single-sector approach working through communities had constructed roads without the involvement of local governments and the Ministries of Transport and Communi- the local council, motivated by food aid from NGOs or by free- cations, Public Works, and Local Government and Housing. The ap- standing projects. But these communities, which had been fully proach for the community roads was a horizontal multisectoral compensated for their work, were unwilling to carry out mainte- approach working with the Zambian social fund. nance on a voluntary basis. Local councils, strapped for resources, Only by working through the social fund, which has a well- were unable to assume responsibility for the roads. As a result, established system for facilitating community participation, scarce infrastructure assets were going back to bush, leaving could Roadsip ensure that communities would choose their communities in isolation. priority investment. When communities request improvements To improve rural accessibility, the government of Zambia in- to roads, footbridges, or paths, they pay 25 percent of project cluded district and community roads as part of a road sector in- costs and the social fund and Roadsip pay 75 percent. On com- vestment project (Roadsip) in 1998. Recognizing the importance pletion of a road project, the community forms a road owners of local ownership in infrastructure, Roadsip addresses the insti- association and applies to the national road board for grant sup- tutional arrangements for the entire road sector-from the main port for maintenance (75 percent community, 25 percent road highways to the community roads-and the government is ex- fund). Providing cost-sharing grants to communities for main- ploring ways to put community ownership of roads into law. taining their roads does not have to cost much. At $300 a kilo- Experience shows that efforts for sustainable improvements in meter, the cost of supporting the maintenance of 5,000 rural accessibility at the community level also have to address sec- kilometers of community roads would be less than 2 percent tor policy and institutional reform and must involve the future own- of annual road fund revenues. Source: World Bank 1998u, 1999ee. and managed by local governments, although in con- prospects for equitable and sustainable infrastructure sultation with communities,88 using a unified invest- services for poor people.92 ment planning process. Such a process can be followed by multisector or single-sector projects. Promoting local management offorests Among the attempts to introduce participatory Common property resources, because they possess chair- processes that allow choice, social funds have been the acteristics of both public and private goods, are subject to most widely studied. Social htnds aim to empower com- free-rider problems that may lead to degradation or de- munities by promoting their participation in the selec- pletion in a free market. Poor people suffer the most fromn tion, implementation, and operation and maintenance these problems when they depend heavily on natural re- of their development projects, usually for local infra- source assets.93 The common response has been state structure.89 But merely making financing available for in- management, widh regulations to induce user behavior con- vestments in a variety of sectors is not enough to ensure sistent with resource conservation. But the deplorable that beneficiaries exercise their choice.90 In many social environmental outcomes under state-led programs, dwin- fund projects community members are unaware of the dling public resources, and the general shift from top-down full range of options eligible for financing. In Peru only to bottom-up partnership approaches have recendy in- 16 percent of beneficiaries could cite more than 5 of 19 creased the emphasis on community-based natural re- eligible project types.91 Furthermore, the mere fact that source management. This approach recognizes and communities have a choice does not necessarily mean that reinforces the role of communities living in and around it will be an informed one. vulnerable natural resources, tapping their ideas, experi- For local infrastructure investments to be effective ence, values, and capabilities for preserving their natural and sustainable, the demand-based approach gener- resources.94 ally has to be complemented by supply-side inputs Communities often manage natural resources in co- (capacity building, information, outreach). Balancing operation with-and with support from-other coni- a bottom-up identification of investments with care- munities and higher (or external) entities, such as local fully selected supply-side inputs will enhance the or district governments, government agencies, or NGOs.95 92 WORLD DEVELOPMENT REPORT 2000/2001 Often, such arrangements apply to forests. At the center Box 5.11 of joint forest management is an agreement between Rejuvenating India's forests through joint action governments and communities on the distribution of use rights and the sharing of benefits,96 usually with The state government of Andhra Pradesh has introduced joint communities getting a larger share of forest assets if they forest management on a massive scale. People on the communities getting a larger share of.forest.assets.if.they.fringes of forests are forming village organizations to pro- achieve agreed conservation and sustainability objec- tect forests-vana samrakashna samithi (VSSs). The orga- tives. 97The state, through the forest department, is often nizations work with the state forest department, sharing the the owner of the forest and also regulates the system.9t responsibilities and benefits of forest restoration, protection, The approach has been widely applied in South Asia- and management. for example, in the Indian state of Andhra Pradesh (box The forest department is responsible for organizing and providing technical and administrative support to the VSSs. 5.1 1). In Africa community participation has helped re- Villages and VSSs are selected carefully, but people from store forest resources in The Gambia and led to broader scheduled castes and tribes are automatically eligible for participation in rural development in Zimbabwe. membership. The VSSs protect the forest from encroach- But the mixed record in other cases signals the chal- ment, grazing, theft, and fire, improving it according to a joint forest management plan. As compensation, the VSSs are en- lenges in making joint management an effective tool for titled to all the forest's products (nontimber products as promoting poor people's access to key natural assets. well as all the income from the harvest of timber and bam- Overly centralized administrative structures have been one boo) as long as they set aside half the income for the future reason for failure. The experience of many countries con- development of the forest. The program got off to a slow start in 1992 because vil- firms that powerful resistance at the national level to de- lagers were hesitant to assume responsibility for forest man- volving rights to forest users can blunt effective community agement. In addition, forest department staff had reservations participation. And forest users or communities, often aboutjoint forest management. But in 1999 more than 5,000 unorganized and with diverging interests, may lack the VSSs were managing more than 1.2 million hectares of de- capacity, interest, and incentives to manage large forestate. Results are impressive, and the program is expanding areas.99 The applicability and success of joint management rapidly. The degraded forests have sprung back to life, tim- will in each case be determined by the institutional ber smuggling has almost stopped, and cattle grazing is context-including private interests, local norms, and under control. There has been no further encroachment by traditions-and by the quality of state and local organi- agriculture on lands managed by the VSSs. Many villagers now work in the forests, and outmigration has declined. Soil zations and institutions. Part of the challenge of joint man- conservation has resulted in higher water tables in many areas, agement is identifying the sociogeographic units that increasing agricultural production. And local plants and ani- can work together to manage and conserve natural mals are flourishing. resources. l° Source: Venkataraman and Falconer 1999. Several approaches have been used to overcome these obstacles. Providing incentives for stakeholder partici- pation is essential: granting secure tenure and rights to forest users, lot more fully transferring management au- multiple land uses.104 Where local capabilities to control thority over forests (rather than user rights alone) to and manage resource use have been eroded or have bro- communities, sharing benefits, and using socially ac- ken down, external assistance can help strengthen and ceptable technologies that provide adequate revenue.t02 monitor resource sharing and management. °5 Effective mechanisms for resolving conflicts are also crit- As in other examples of local participation, social in- ical, especially where resource users' livelihoods compete equalities can reinforce the influence of politically pow- with other objectives, such as biodiversity protection or erful and better-off groups, further reducing access for sustainable forestry.103 Contracts between the government, other groups.106 In particular, women and poor people villages, and fuelwood collectors in Burkina Faso and who depend on communal assets for their livelihoods can Madagascar specify which subgroups of users manage op- be pushed into deeper poverty if they are excluded. 107 tions in watershed and protected areas. Effective en- While the forest management groups in Andhra Pradesh forcement of these contracts is essential. In the Czech successfully involved women, a few programs in other In- Republic, Ecuador, the Slovak Republic, and Ukraine sites dian states still allow only one household member to outside protected areas are rezoned to accommodate participate-effectively excluding women. In several In- EXPANDING POOR PEOPLE'S ASSETS AND TACKLING INEQUALITIES 513 dian villages women were barred from collecting any ership remains concentrated, and efforts to increase lan.d forest products on protected lands. 108 equality have often generated conflict. 113 New approaches Even when women are not excluded, their numbers to land reform stress the importance of bringing together and their influence in management committees are various stakeholders-the landless and their associations, low.'09 A policy and legal framework that promotes the private sector (landowners), and government institu- participation by poor users in the management of nat- tions at the local and national levels. ural resources can help change this."'1 Greater inclusion There are many ways to gain secure access to land-some of women will also require awareness-raising activities informal and others formal, some spontaneous and oth- to break through societal norms that keep women from ers requiring extensive government intervention.114 ]n playing an equal role with men. general, secure access to land can be gained through own- ership, tenure, or customary use rights. 115 Ownership Complementarities in public action rights are the most secure but also the least likely to be en- joyed by the poor and other socially excluded groups. Recall the three principles for building assets for poor Poor people often gain access to land through the rental people-redistribution by the state, effective governance and market and customary use rights. 116 In India, of the esri- use of markets, and participation. Effective action gener- mated 19 percent of rural households that lease land, ally involves applying all three principles in a particular area more than 90 percent are landless or own very little land." 17 or sector. The extent to which each principle is applied de- When land is relatively abundant or poor people are well pends on the structural conditions, the type of action, the organized and influential, they can have secure land access state of governance, and the extent of participatory in- without formal property rights and registration."8 This is volvement and social inequality. Take education. There is still the case in many parts of Sub-Saharan Africa, though a case for public redistribution that seeks to ensure free or in recent decades population growth and market integra- subsidized basic education for all. Effective delivery often tion have accelerated the shift to individual land rights."19 involves multiple agents providing schooling-public, pri- Public action is critical to ensuring secure access to vate, and civil society. And in poor areas participation of par- land for poor people. Land reform that enhances equity ents is important for increasing coverage, quality, and and productivity through government-supported pro- accountability. grams is usually what first comes to mind (see box 3.8 in In this section the mix of actions is illustrated for a nat- chapter 3). But many other actions can improve access to ural asset (land) and for two physical assets (urban water land. Policies and laws that clearly define land rights and and sanitation). The complementarities imply that action protect poor people against land grabbing can greatly en- is needed on several fronts. But the priorities should de- hance their ability to use land as collateral and invest in land pend on what poor people lack most relative to their po- they already "own." Well-functioning rental markets can tential opportunities. In poor rural areas this may be basic raise the efficiency of land use and help the landless clinib economic infrastructure, land-enhancing investment, water the "agricultural ladder" to ownership.'20 Providing legal and sanitation services, and basic education and health care. assistance to poor people enables them to press their legal In urban slums it may be infrastructure. And secure prop- claim to a plot of land. Similarly, public support to ins-ti- erty rights on land are important for both. tutions that protect women's rights can be instrumental where deep-seated social norms and customs inhibit women Enabling good governance, active markets, from exercising effective control over land even when there and broadparticipation in land reform are legal provisions for them to do so (chapter 7). 121 Land reform has returned to the policy agenda in the past Attempts at land reform often fail because they rely on decade, as many developing countries move beyond im- government alone. New approaches emphasize continu- plementing macroeconomic reforms to addressing the ous mechanisms of adjustment in land access, greater re- often weak micro-level supply responses in agriculture. "' liance on traditional forms of access, and greater use of land It is easy to see why farmland is a key asset for the rural markets.'22 Brazil, Colombia, the Philippines, and South poor. But secure access to land and for whom and under Africa are experimenting with decentralized, community- what conditions remains a thorny issue.1 12 Poor people, based, demand-driven negotiations between stakeholders especially poor women, often lack land rights. Land own- to find less antagonistic ways to improve access to lanl. 94 WORLD DEVELOPMENT REPORT 2000/2001 Such negotiated land reform relies to a great extent on Box 5.12 experience gained in past attempts at land reform and on A new approach to land reform in Brazil the successful aspects of demand-driven social funds. It addresses both supply and demand. Landlords are paid Brazil has been expropriating and redistributing land since the in cash rather than with highly discounted government mid-1 960s, reaching an impressive number of beneficiaries, bonds, as in the past. Beneficiaries receive grants for but with high costs and uneven quality. prod .ictivirv. enhancing investments, because experience To increase quality, lower costs, and speed reform, five productivity-enhancing investments, because experience state governments in the Northeast started a pilot program shows that unless those who receive land can make pro- of negotiated, decentralized, community-driven land reform ductive use of it, reforms will be undermined.123 in 1997. The program provides loans for land purchases to The negotiated approach has several innovative com- landless rural dwellers (sharecroppers, renters, landless workers, labor tenants) or to smallholders who organize ponents, including strong involvement of local governments themselves in beneficiary associations. The associations and communities in organizing land transfers and assist- have to identify landlords interested in selling them land, an ing the beneficiaries-and a high degree of transparency approach that fosters direct negotiations between owners 5.12). In Colombia municipalities must develop a and the associations and reduces government intervention. (box 5.12). In Colombia municipalities must develop a The pilot program also provides grants for productivity- comprehensive plan to identify potential sellers and ben- enhancing community projects identified by the beneficiary eficiaries, and the plan is widely circulated to avoid cor- associations, drawing on well-established poverty allevia- ruption. Potential beneficiaries are offered training in farm tion projects in the five states. The philosophy of the projects management and assistance in developing land use plans. is that beneficiary associations are best placed to identify, rank, and implement investments, drawing on technical as- These plans must then be approved in public meetings of sistance as necessary. The same participatory philosophy has municipal councils.124 Beneficiaries are supported by na- been adopted in the land reform pilots. tional, regional, or state councils that provide technical guid- The program has three grant elements constituting about ance and resolve administrative obstacles. 50 percent of the cost of the land: an inherent subsidy in the interest rate, a grant for the complementary infrastructure in- The results on the ground from the community-based vestment, and an installation grant. A credit with a maximum approach are encouraging. Yet it Is still too early to fully term of 20 years is provided for the land. The land guaran- evaluate this new generation of reforms. While the ne- tees the loan and has to meet a number of requirements, in- gotiated land reform has been criticized for, among other cluding a price comparable to what prevails in the local things, burdening beneficiaries with loans they cannot market. Since there is a maximum grant per beneficiary fam- ily and the infrastructure investment grant is a residual, there repay and trying to replace expropriative land reform, re- is an incentive for the association to buy the land at the low- visions to the approach have sought to address these and est possible cost. other concerns. Another issue is costs. Although in Brazil Although the number of beneficiary families is still modest and elsewhere this approach has achieved savings of as (about 10,000 in 330 projects), the pilot program has picked up momentum, and its success has prompted the central gov- much as 40 percent relative to expropriative reforms, it ernment to expand the program nationwide. Complementing still requires significant public outlays. 125 Its effectiveness the larger, expropriative program, the new program focuses on needs to be assessed relative to other instruments for re- purchasing properties that cannot be legally expropriated. ducing povertv. 126 ducing poverty.126 Source: Deininger forthcoming. Providing water and sanitation services to the urban poor exploring alternative approaches involving the private The sewage runs right in your front door, and when it sector and local neighborhood and civil society groups. rains, the water floods into the house and you need to lift Working together, these actors seek to provide quality things . .. the waste brings some bfugs. Here we bave rats, services to poor consumers while ensuring financial cockroaches, spiders, and even snakes and scorpions. sustainability. lYouiwa o o/i ousWfl ,r-oup, Water and sanitation services have traditionally been Xova(U ;nm, Br.zzil provided by public agencies. While there are exceptions, few developing countries have elicited strong, sustained To improve water and sanitation services to urban resi- performance from public water and wastewater utilities. dents, governments and municipalities the world over are Weak performance incentives and difficulties in shelter- EXPANDING POOR PEOPLE S ASSETS AND TACKLING INEQUALITIES 95 ing management decisions from political interference cient, and the early results are encouraging. Coverage has have locked many utilities into a cycle of poor service, low increased rapidly, and in some cases tariffs have fallen (as user willingness to pay, and insufficient maintenance. The in Manila, Philippines). The water supply system in Cbte inability of supply to keep pace with increasing demand d'Ivoire, which introduced the first private concession in from growing urban populations has forced poor house- Sub-Saharan Africa, performs better than other urban holds to find their own solutions. In many countries water systems in West Africa.129 small informal water vendors and sanitation providers But large-scale private participation in water and sani- reach poor urban areas unserved by government utilities. tation does not automatically mean better services for poor In West African cities independent entrepreneurs supply people. Unless carefully crafted, contracts may preclude the most poor households (box 5.13). Similarly, in Guatemala extension of services to low-income areas and create local City and Lima, Peru, which both have major utility com- monopolies. Contracts often mandate tariff structures and panies, most families depend on private informal set connection fees that do not vary with the true cost of providers.127 Although local suppliers can be more ex- connection. These features discourage concessionaires from pensive than public providers, households would be delivering service to low-income areas.130 In Guayaquil, worse off without them. Ecuador, residential water tariffs did not cover collection Since the early 1 990s there has been a marked increase costs. Every new connection, even if fully grant financed, in large-scale private participation in water and sanitation was a net revenue drain on the utility.'31 To serve low-incomre in developing countries, reflecting a desire to deliver bet- households better, concessions in Buenos Aires, Argentina, ter services at lower cost-including services to poor and La Paz-El Alto, Bolivia, have been restructured based urban neighborhoods. 128 Private participation can boost on negotiations between governments and private providers service coverage and make utility operations more effi- and input from local stakeholders. Good pricing policy is a key element of pro-poor pol- Box 5.13 icy.'32 Whether water and sanitation utilities are publicly West African businesses pioneer water and or privately managed, those most successful in expanding sanitation services for the urban poor these services charge tariffs that cover costs. Such tariffs can increase access for low-income households by attracting pri- Africa's independent water and sanitation providers suggest vate investment to expand supply and enhance quality. They that the market has found solutions that benefit everyone: can also end general government subsidies that go mostly providers, utilities, and, above all, low-income customers. Recent studies in seven West African cities show that half to the nonpoor, releasing public resources for more targeted the residents rely on private independent providers for their assistance to poor people. Notwithstanding the overall water and at least three-quarters rely on independent oper- trend in water and sanitation toward greater cost recovery, ators for sanitation. Depending on the city, independent providers cover up to 85 percent of marginal and low-income governments can ensure greater access for poor people by neighborhoods, and they serve many better-off families as well. subsidizing connections or, where network access is higlh, Independent sanitation providers, working without an using well-designed block tariffs. 133 official mandate or arrangement with local governments, in- Additional measures to benefit poor households and clude small informal operators as well as a few that have attract private investors to water and sanitation include grown and become "legitimate.' Providers rely on good client relations, since their operations are completely simplifying contracts, contracting out some regulatory demand driven. Consumer ability to pay and competition functions, and increasing the predictability of regula- among providers determine prices. tory discretion. 134 The design of regulation-particular]y Independent sanitation providers charge higher prices than to reduce monopoly power-is also critical for pro-poor subsidized public companies, but public companies rarely re- outcme135 Regulation can enhance competition by coup their operating costs-let alone the costs of installing outcomes. sewer networks. And the independent providers are gener- permitting greater entry, including by nonconventional ally reliable and responsive to their customers. They extend suppliers, and by changing service standards to fit local credit (for a few days at least) and spread collections over needs-for example, focusing on the potability of water days and weeks, far easier for poor clients to fathom than rather than on technical construction standards set at in- the three-month bills from public companies. dustrial country levels. '36 Source: Solo 1999. Also important is involving users and local institutions in designing private sector options with user preferences in 96 WORLD DEVELOPMENT REPORT 2000/2001 mind-and fully leveraging the presence of alternative ser- Box 5.14 vice providers. This is the approach of the Water and San- Sustainable water and sanitation for Brazil's itation Program-a partnership of donors, governments, urban poor and NGOs that focuses on poor rural and periurban areas. In each context this demand-responsive approach must be The first phase of Prosanear (1992-97) was a period of learn- tailored to local conditions. Brazil's Prosanear (Water and ing that led to six guiding principles for sustainable provision of water and sanitation in poor urban neighborhoods in Brazil: Sanitation Program for Low-Income Urban Populations) * Start community participation at the very beginning of pro- follows six principles to provide sustainable water and san- ject preparation. itation services to poor households (box 5.14). * Ensure that cost-recovery and subsidy rules are clear A similar approach aims to cover 35 towns in an and transparent. * Make formal, long-term arrangements for operating and urban water and sanitation project in the Philippines. maintaining systems an integral part of the design. Communities decide to participate (borrowing money * Discuss all feasible technical options and their costs with from the Development Bank of the Philippines) after ex- communities. tensive consultations involving consumers, the mayor, and * Coordinate projects with the local government's urban de- velopment plan from the outset of preparation. the town council. Service charges are used to repay the * Confirm that the local government has a strong com- loan and cover operation and maintenance. The in- mitment to the project and to poverty reduction. volvement of users, together with local government de- cisionmaking on participation, appears to ensure the Source:KatakuraandBakalian1998. ownership needed to enhance prospects for long-term sustainability. 137 But because there is a two-way causal relationship between 3 * * political and social structures and the assets of poor peo- ple, it may be necessary to tackle exclusionary or weak The assets that poor people possess-or have access social structures in order to form assets (chapter 7). to-directly contribute to their well-being and have a po- Public action is essential to expand poor people's as- tent effect on their prospects for escaping poverty. Human, sets and to tackle asset inequalities-particularly in the physical, natural, financial, and social assets can enable distribution of human assets. Effectively using the re- poor people to take advantage of opportunities for eco- distributive power of the state and involving multiple nomic and social development (just as their lack can agents (civil society, markets, and the state) and stake- prevent this). Expanding the assets of poor people can holders in the provision of services are crucial to this end. strengthen their economic, political, and social position Expanding poor people's assets is at the core of getting and their control over their lives. Assets empower the poor. the benefits of growth to reach poor people faster. And And assets help people manage risks (chapters 8 and 9). it can be pro-growth (chapter 3). PART III Emp owe rm eent f . , 1^4 CHAPTER 6 Making State Institutions M'ore Responsive to Poor People iP overty is an outcome not only of economic able to pro-poor actions and coalitions, facilitating processes-it is an outcome of interacting economic, the growth of poor people's associations, and in- social, and political forces. In particular, it is an out- creasing the political capacity of poor people. come of the accountability and responsiveness of state * Political regimes honor the rule of law, allow the ex- institutions.1 pression of political voice, and encourage the par- As this chapter discusses, the state will deliver more ticipation of poor people in political processes. effectively to all its citizens, but to poor people in par- ticular, if: Public administration and poverty * Public administrations implement policies effi- reduction ciently and are accountable and responsive to users, corruption and harassment are curbed, and the It is hard to get to the rghtperson in the munwapaity, and power ofthe state is used to redistribute resources when you do he says, "I'm sorry, Lam not abl to he4p you" for actions benefiting poor people (chapter 5). -From a discussion group, Zenica, * Legal systems promote legal equity and are acces- Bosnia and Herzegovina sible to poor people. * Central and local governments create decentralized In most developing countries poor people have trou- mechanisms for broad participation in the delivery of ble getting prompt, efficient service from the public ad- public services and minimize the scope for capture by ministration (box 6. 1). To change this, the first step is local elites. building the capacity of public administration. Officials * Governments generate political support for public also need tractable regulatory frameworks, with proper action against poverty by creating a climate favor- performance incentives and mechanisms to ensure ac- 99 IOO WORLD DEVELOPMENT REPORT 2000/2001 Box 6.1 Poor people are often harassed by public officials Poor women and men in the Voices of the Poor study stressed The officials of the social assistance department are impolite that officials are often unresponsive to them. They shared count- andeven crude with ordinarypeople from the village. Igo there less examples of criminality, abuse, and corruption in their en- formy social benefit formychildren. I have to wait for two hours; counters with public institutions and said they have little recourse they treat me very badly. If I cry and shout that my child is ill, to justice. In describing their encounters with institutions, poor peo- they'll give me something. But it happens seldom. ple also drew attention to the shame and indignity of being treated -Woman, Novy Gorodok, Russian Federation with arrogance, rudeness, and disdain. We in the country get up at 6 a.m. to take the collective bus. When they assist you they treat you like a beggar .. . but we We arrive. We go to the doctor at the hospital. You arrive at aren't ... we pay taxes.... There must be transparency in 8 a.m. or sometimes not until I p.m. You are stuck there until government actions, tax money has to be well employed. the afternoon, without eating, without being able to drink ... ... They invent these useless constructions and grab our you spend hours and hours hungry. You have to go back be- money ... fore the doctor has seen you. You miss the bus. You have to -Poor man, Vila Junqueira, Brazil go however you can ... Some receive us, others don't. It's awful.... They are abusive -Twenty-five-year-old mother, Los Juries, Argentina ... They treat one almost like a dog.... The municipality only We would rather treat ourselves than go to the hospital, where serves the high-society ones.... the angry nurse might inject us with the wrong drug. -From a discussion group, Esmeraldas, Ecuador -Poor youth from Kitui, Tanzania Source: Narayan, Chambers, Shah, and Petesch 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 2000. countability and responsiveness to clients, including poor crucial to ensure steady progress toward a fully respon- people.2 Poor organizational design engenders inefficiency sive and accountable state. and corruption, typically hurting poor people the most. Enabling and motivating public Focusing public action on social priorities administrations In nearly every country the public sector often pursues ac- Having the right performance incentives smooths the de- tivities that are not socially justified and, in some cases, that livery of public services. Key incentives include merit- generate rents for the elite. During the past two decades, based recruitment, clear specification of tasks, rewards for as societies and their governments have become aware of good performance, and insulation from excessive political this problem, they have launched public sector reforms to pressure.3 Together with skilled technocrats and close col- focus public action and programs on social priorities and laboration with the business community, these make up increase the capacity of the state to reduce poverty. what has been termed "the developmental state."4 Public sector reform and modernization have great po- Merit-based recruitment goes a long way toward im- tential to reduce poverty, if they are at the core of a de- proving administrative performance. When nepotism or velopment strategy that establishes clear priorities for cronyism exists, it is difficult to motivate staff to perform public action. The functional and organizational struc- well.5 Cross-country analyses indicate that merit-based ture of the public sector needs to be rationalized to im- recruitment is associated with less corruption and fewer prove resource allocation for programs that are social delays (figure 6.1). Merit-based promotion is also essen- priorities and have greater capacity to reduce poverty. Most tial for motivating staff. If there are few opportunities for important is to streamline and "rightsize" public ad- promotion, or if promotion is unrelated to performance, ministrative entities and privatize public enterprises and staff have much less incentive to perform. What is im- other operational public programs. portant is to promote an evaluation culture, for staff and Beyond rationalizing the structure of the public sec- for agencies. Also important to good performance are tor there is a need to improve public management sys- clearly specified and tractable tasks and competitive tems to make public programs more efficient and salaries. Compensation of public servants that is severely accountable. Involving civil society in planning, moni- out of line with that in the private sector affects perfor- toring, and evaluating public programs and policies is also mance incentives and encourages corruption.6 MAKING STATE INSTITUTIONS MORE RESPONSIVE TO POOR PEOPLE IOI Figure 6.1 Legislative oversight of the executive, carried out ac- Merit-based recruitment in government is cording to transparent procedures, is an important part associated with less corruption and of monitoring and improving performance. Public ad- bureaucratic delay ministrations also need to be supported and actively monitored by political leaders. Surveys in several devel- oping countries show that public officials' performance Corruption improves when they know that elected representatives are High overseeing their work. But sometimes this process becomes subject to the personal goals or whims of elected repre- sentatives, resulting in excessive political interference. The quality of public service is reduced when public of- ficials are held accountable more to their hierarchical superiors than to the people they serve.7 Making the public sector more responsive to client needs Many different kinds of measures help improve public sector service delivery. One important measure is sim- plifying procedures and making them transparent to LLow clients. In the Philippines several public agencies have Low Merit-based recruitment High streamlined procedures to curb corruption. At the out- set of a transaction clients receive a list of required doc- High Bureaucratic delay uments along with a timetable showing how long the process will take and a schedule of fees.8 More generally, simplifying and improving regulatory and tax systems and privatizing state-owned enterprises can reduce the op- . . . . portunities and scope for corruption. * . . Another important measure is disseminating informa- . . . * tion to allow people to monitor public services. Using newspapers and other popular information sources to dis- seminate information on budget allocations and spending enables people to hold civil servants accountable, reducing inefficiency and corruption. In Uganda, when primary en- rollments did not improve despite substantial increases in budget alLocations, a survey of schools examined public Low. spending on primary education. The study found that Low High budget allocations may not matter when institutions or their Merit-based recruitment popular control is weak: in 1991-95 on average less than Note: The figure is based on the responses to a survey sent to 30 percent of the intended nonsalary public spending onl experts in developing countries, with 126 complete responses from primary education reached schools. The government has 35 countries. Merit-based recrurtmentcaptures the share of higher officials in core economic agencies who enter the civil service since improved performance by increasing the flow of in- through a formal examination system or have university or formation within the system. A major breakthrough was postgraduate degrees. A high score on corruption indicates a strong likelihood that high government officials will demand special achieved by makmg regular announcements m local news- payments and strong expectations of illegal payments throughout papers and on the radio of the public funds transferred to low levels of govemment. A low score on bureaucratic delay districts and posting information on transfers at each schoo:l. indicates greater speed and efficiency of the civil service. Source: Rauch and Evans 1999. A follow-up survey in 1999 showed dramatic improve- ments since 1995, with schools receiving close to 100 per- cent of the nonwage public funding.9 102 WORLD DEVELOPMENT REPORT 2000/2001 Fostering communication between civil servants and Figure 6.2 their clients is also important. Many developing coun- Corruption is a regressive tax try administrations have poor mechanisms for learning about and responding to users' demands. In India the "re- 10Percent port card" on Bangalore's public services shows how a pub- Bribe cost as a share Bribe cost as a share lic feedback mechanism can make public agencies more of firms' revenue of households' income accountable to their clients. Launched in 1993 by a 8 group of committed citizens, the report card provided cit- 7 izens' views of public service delivery in the city. Re- 6 spondents focused on agencies they dealt with to redress 5 a problem or to get a service-ranking their satisfaction 4 and indicating the time spent. The findings were dis- 3 seminated to public agencies, the media, and NGOs, trig- gering some service providers to become more efficient 2 and accountable, The Bangalore City Corporation helped I set up an informal network of city officials and non- 0 , , ~~~~~Micro Small Large Low Middle High governmental groups to meet periodically and work out (<11) and (>99) income income income answers to prioritv problems.'l medium size (<$110) ($110- (>$329) (11-99) 329) t_urbing corruption Note: Results are for Ecuador. For firms, the values in parentheses refer to the number of employees; for households, the values refer Corruption takes a toll on economic performance, un- to monthly household income. The figure is based on preliminary dermines employment opportunities, and clouds prospects data from a 1999 survey of 1,164 enterprises and another of 1,800 households. for poverty reduction. Even petty corruption dramatically Source: Kaufmann, Zoido-Lobaton, and Lee 2000. raises the cost of engaging in productive activities. In West Africa bribes in the transport industry are crippling. The estimated cost of transporting goods from C6te d'Ivoire public servants can help reduce the opportunities for cor- to Niger includes bribes to customs, police, and transport ruption. And community participation and monitoring can officials that represent three-quarters of payments to the keep it in check. administration.1" Similarly, a transport trip in Benin en- countered 25 roadblocks over 753 kilometers-road- Poor people and the rule of law blocks staffed by state agents who demanded bribes that added up to 87 percent of the cost of the trip. '2 There are four dragons: law court, prosecutor's office, The burden of petty corruption falls disproportionately khokimiat, and head of police. Vobody can get anything on poor people (figure 6.2). For those without money and until they are satiated. connections, petty corruption in public health or police --i F ir c ra li,nlo,)i) gorp. ()itano,lali, I /.Zblkistan services can have debilitating consequences. Corruption affects the lives of poor people through many other chan- The rule of law means that a country's formal rules are nels as well.'3 It biases government spending away from made publicly known and enforced in a predictable way socially valuable goods, such as education. It diverts pub- through transparent mechanisms. Two conditions are es- lic resources from infrastructure investments that could sential: the rules apply equally to all citizens, and the state benefit poor people, such as health clinics, and tends to is subject to the rules. How state institutions comply with increase public spending on capital-intensive investments the rule of law greatly affects the daily lives of poor peo- that offer more opportunities for kickbacks, such as de- ple, who are very vulnerable to abuses of their rights. fense contracts. "4 It lowers the quality of infrastructure, The rule of law is upheld through many channels, the since kickbacks are more lucrative on equipment purchases. most formal being the legal and judicial system. The Corruption also undermines public service delivery. legal and judicial system constrains and channels gov- Streamlining bureaucratic procedures, simplif,ring tax sys- ernment action-and maintains clear rules and procedures tems, eliminating excessive regulations, and motivating for upholding an individual's constitutional rights. This MAKING STATE INSTITUTIONS MORE RESPONSIVE TO POOR PEOPLE 103 system is essential for guarding against abuse of power by tecting their interests. Legal systems, the product of the state or other actors, and it requires that the judiciary power relations between different groups in society, typ- be independent of the executive and legislative branches. ically focus on protecting the interests of those with pc- The rule of law protects life and personal security and litical strength and representation. Making laws and their guards against human rights abuses. Thus defined, the rule interpretation more sensitive to the needs of the disad- of law is tremendously important for all citizens-but es- vantaged requires building coalitions to this end. This is pecially for poor people, who have few private means of the goal, for example, of efforts to make laws more eq- protecting their rights (box 6.2). uitable in their treatment of women and minorities The rule of law is associated with better overall eco- (chapter 7). nomic performance (figure 6.3), and in this sense it also Legal obstacles leave poor people vulnerable to ex- promotes poverty reduction. It does this by creating a pre- ploitation by local bosses and the police, and arbitrary ha- dictable and secure environment for economic agents to rassment, lawlessness, and violence are constants in thei r engage in production, trade, and investment, thereby lives. For poor people, a crucial aspect of the rule of law expanding poor people's employment opportunities and is the ability to live without fear of lawlessness and ha- incomes.15 Market mechanisms depend on credible rassment. An effective modern police force is needed to threats of punishment for breaking contractual obligations, maintain order by enforcing the law, dealing with po- backed by prompt methods for resolving disputes and en- tentially disorderly situations, and attending to citizens forcing contracts. Without these deterrents, the transac- in distress. tions costs of doing business can be very high. Although the rule of law benefits poor people in many Making the legal system more responsive ways, laws and statutes are not necessarily geared to pro- to poor people Even when the legal system is well run, poor people face Box 6.2 constraints in tising it.16 Poor people typically have lil;- Lawlessness contributes to poverty Figure 6.3 Ata hospital in Babati district in Tanzania a newdeliveryofes- Better rule of law is associated with higher per sential medical supplies purchased with foreign currency dis- capita income appears from the public dispensary within hours but is available for purchase at the doctor's home that evening. The poor do Per capita GDP not receive the free medical care promised by the government, u.s. dollars (PPP) but those with the right connections and the ability to pay can 25,000 secure pharmaceuticals in abundance. In Johannesburg, South Africa, rates of theft and violent crime are among the highest in the world. Wealthy resi- 20,000 * dents can afford sophisticated alarms, security guards, and other forms of private policing to protect their property and 0. persons. Poor people are stuck in poorly built homes, some- 15,000 * times without even simple locks, and are vulnerable to theft, , assault, murder, and other violent crimes. I In Pakistan a man too old to work is left without assets 10,000 °°e. or income after his son is murdered. To gain access to his . son's estate, he needs a succession certificate from the J e civil court in Lahore-more than 160 kilometers away. The 5 * t * train ticket and the bribe demanded by the clerk of the court . * _ * ,iA * send the man deeper into debt, yet after five trips to the court 0 _ in as many months, he still has not received the stamped -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5 1.0 1.5 2.0 piece of paper to which he is legally entitled. The clerk re- Rule of law index fuses to produce the certificate, while the authorities in the man's home village refuse to give him access to his son's Note: The figure shows the relationship between an indicator of assets until the certificate is produced. rule of law and per capita GDP for 166 countries in 1997-98. Source: Kaufmann, Kraay, and Zoido-Lobaton 1999. Source: Michael Anderson 1999. I04 WORLD DEVELOPMENT REPORT 2000/2001 tle knowledge of their rights and may be deliberately process), shortening proceedings, allowing parties to rep- misinformed. Contemporary legal systems are written and resent themselves. Broader reforms, such as changing the are conducted on the basis of written documents- structure of the courts, also help increase poor people's ac- making access inherently difficult for poor people, who cess to justice. Small claims courts and other informal pro- usually have little formal education. Language, ethnic, ceedings can reduce the backlog and widen access.22 And caste, and gender barriers and other exctusionary prac- the teaching and practice of law can be amended to sen- tices add to these problems. sitize the legal profession to the needs of poor people and The intrinsic complexity of legal systems is exacerbated to the use of the law to further the public interest. in many developing countries by the superimposition of Alternative dispute resolution mechanisms hold con- new laws and constitutional rights over colonial legisla- siderable potential for reducing the delays and corruption tion and customary law. The resulting confusion makes that characterize much dispute settlement. In El Salvador it difficult to know one's rights, introduces arbitrariness mediation provides parties a means to settle disputes with- in law enforcement, and enables the powerful to choose out a lawyer and within two months.23 In Sri Lanka the which legal system to apply. 18 This reduces poor people's Asia Foundation has assisted the Ministry of Justice since confidence in the legal system. It also gives enormous dis- 1990 in establishing a national network of communitv- cretion to authorities, often making connections and based mediation boards. In 1998, 100,000 cases were re- bribes central to negotiating the legal system. Making the ferred to the mediation boards, with two-thirds of cases rules simpler and clearer is especially important in the areas resolved to both parties' satisfaction. An independent eval- of greatest concern to poor people, such as labor disputes, uation found that the boards enjoyed an outstanding rep- land titling, human rights abuses, and police violence. utation and successfully provided low-cost, accessible Although poor people need to access the legal system justice to poor people in rural areas.24 In Bangladesh some for registration and other administrative purposes, they NGOs have adopted the shalish (an indigenous practice that use the judicial system much less frequently than the uses outside parties to help resolve disputes) to aid women nonpoor. Court systems in many developing countries are and other disadvantaged groups, such as low-income farm- poorly funded and equipped, and mechanisms for en- erswith land-related disputes. A 1999 study in Dhaka shows forcing judgments often weak. These add to the other prob- that women who have been through NGO-initiated me- lems poor people face in using the judicial system, such diation expressed satisfaction with the results by a four-to- as financial costs. Waiving court fees for people with low one margin.25 That the NGOs can back up the mediation incomes could provide some relief. Ecuador and Peru process with litigation is a factor in that success. provide exemptions for court fees in certain cases. To as- These alternative mechanisms may provide more pre- sist poor people, legal aid is provided in many develop- dictable outcomes than the formal system, because com- ing countries, but often more in principle than in practice. munity mediators are typically more familiar with the To be effective, such aid must be delivered promptly: in details of cases than are judges.26 The risk of such mech- Trinidad and Tobago it takes the legal office about three anisms is that they can give undue power to conservative months to process applications for legal assistance, in ef- forces in a community (which might, for example, be bi- fect denying access to those unable to wait that long.19 ased against gender equity) and be subverted to serve the In addition to government-provided services, legal interests of local elites. To minimize these risks, alternative aid can be provided through alternative sources. Many dispute resolution mechanisms need to be carefully regu- countries require law school graduates to provide legal aid lated and supervised by more formal legal structures. They before becoming attorneys-others require practical can also be introduced gradually-for example, through training of law students. In Chile and Peru lawyers must pilot programs sponsored and supervised by regular courts. complete a specified amount of practical training after law school, often in legal aid offices, thereby providing im- Promoting legal service organizations portant resources for poor people.20 Civil society organizations such as legal service organi- Streamlining the operation of the judicial system to re- zations seek to help poor people gain access to the ben- duce costs and delays will address some of the problems efits and protection of the legal system inside and outside poor people face in the courts.21 Reforming court proce- the court system (box 6.3). Protecting individuals against dures helps-simplifying rules (while respecting due unlawful discrimination at work and eviction from their MAKING STATE INSTITUTIONS MORE RESPONSIVE TO POOR PEOPLE 105 homes, such organizations help people collect their en- This goes far beyond the conventional idea of offering free titlements, obtain basic services, and get court orders to legal representation to poor individuals and helping peo- protect women from domestic violence. They can also pro- ple or communities assert their rights through the courts. tect communities from being dispossessed. More generally, the work of legal service organizations Legal service organizations can help poor people by tak- helps create a culture of rights that changes the way peo- ing legal action on behalf of a group of plaintiffs. Often, ple think about themselves relative to those who have large numbers of poor people suffer from similar injuries, power over their lives-spouses, landlords, employers, gov- so seeking redress as a group provides poor people with oth- ernment agencies. This encourages poor people to avavil erwise inaccessible judicial protection. Legal advocacy or- themselves of the protection that the formal legal system ganizations in Bangladesh helped avert the eviction of offers. These organizations also generate pressure for urban slum dwellers. Evicted residents became petitioners changing the way the rules are applied by judges, bu- in litigation, where the basic argument rested on funda- reaucrats, and the police. Legal literacy and legal aid mental constitutional guarantees: demolishing their home have the maximum benefit if they help create a process deprives the poor of a livelihood, in violation of the con- of self-empowerment and social empowerment that stitution.27 Public interest litigation can also benefit poor moves citizens to activate their rights and to redefine people. In India it has improved the delivery of some pub- and reshape inequitable laws and practices. lic services and reduced environmental contamination.28 Legal service organizations help change the rules that The most effective legal service organizations work out- affect poor people, whether in constitutions, statutes, side the judicial system, protecting rights without re- regulations, municipal ordinances, or myriad other codes. sorting to lawsuits-important, because the costs of In Thailand the Women and the Constitution Netwo:rk lawsuits can sometimes outweigh any resulting gains. was very active in constitutional reform that led ito Box 6.3 Legal service organizations help poor people gain access to the protections of the legal system Almost any form of legal assistance is of value. But comprehen- Cambodia sive services from independent legal service organizations are es- Legal aid organizations in Cambodia are struggling to create a pecially valuable to poor people, and the demand for such services justice system-from almost nothing. The Cambodia Defenders is high. Standards should be developed to make them even more Project, established in 1994, focuses on criminal defense and effective. Legal service organizations also need financial support community legal education. It collaborates with NGOs to provide from donors and civil society, but they have to be allowed to func- services and represent women in court, especially in domestic vi- tion autonomously, taking direction from poor people themselves. olence cases. The organization's lawyers run training programs, comment on draft laws, and work with civil society groups to ex- Bangladesh plore legal tools for influencing government. The Legal Aid Soci- The Ain-O-Salish Kendra (ASK), established in 1986, seeks to re- ety of Cambodia works to increase public understanding of and form the law through its representation of poor women and chil- respect for the law, while providing free legal services in criminal dren, organized groups of workers, the rural poor, and slum and civil cases. It is especially active in defending farmers being dwellers. It provides legal aid primarily on family matters, includ- dislodged from their land by powerful business interests. ing violence against women. Litigation on behalf of victims is un- dertaken in criminal cases and when basic legal rights are violated. South Africa ASK investigates and monitors violations of law and human rights, The Legal Resources Center, a national organization founded in including police torture, murder, rape, and deaths in garment fac- 1979 to serve poor people, initially used legal advocacy to exploit tories. It also monitors police stations to collect information on vi- contradictions in the apartheid legal system. Since the end of olence against women and children and to track cases reported apartheid the center has used legal advocacy to address land and at the stations. The-work by ASK is significant because of the sub- housing issues. It successfully represented the Makuleke com- stance of what it does-work on basic issues fordisenfranchised munity in its restitution claim to land in Kruger National Park. people-and because of the way it does it-through mediation, Other activities have included cases to restore water service ter- discussion groups, legal awareness training, individual court cases, minated because residents were too poor to pay and to protect administrative and legal lobbying, group representation, and pub- the land rights of an aboriginal community in the privatization of lic interest litigation. a diamond mine. Source: Manning 1999. io6 WORLD DEVELOPMENT REPORT 2000/2001 amendments recognizing equal rights for women. The net- viding public services, and organizing their operation and work followed up this success by launching a mass cam- maintenance. A study in South Africa found that com- paign to educate Thai citizens-women and men-about munitv involvement reduced the cost of creating jobs and the new constitution and its implications.29 improved the cost-effectiveness of transferring resources to poor people (figure 6.4). Moreover, knowing what local How can decentralization be made needs are most pressing can help the disadvantaged. In In- pro-poor? donesia greater local control over funds led to more spend- ing on health and education in prioritv areas for poor State institutions are often accused of being too remote people-and to more spending on small infrastructure, from the daily realities of poor people's lives, and de- boosting nonfarm emplovment and incomes.34 centralization is often recommended as a solution. De- Local monitoring and supervision for many types of centralization can be powerful for achieving development projects and programs are more effective and less ex- goals in ways that respond to the needs of local com- pensive because of proximity to the point of provision munities, by assigning control rights to people who and better interactions at the local level (box 6.4). In have the information and incentives to make decisions Nicaragua students attending schools that were best suited to those needs, and who have the responsi- "autonomous"-as measured by the share of decisions bility for the political and economic consequences of on teacher staffing made by the school-achieved bet- their decisions.30 It is not in itself a goal of development, ter test scores than students in schools with limited or but a means of improving public sector efficiency. And no local autonomy.35 there are important caveats. The most important is that decentralization can bolster the power of elites in set- What is required to reach poor people? tings with highly unequal power structures.31 To ben- Decentralization can greatly enhance the state's capacity ctit poor people, it must have adequate support and to accelerate local development and reduce poverty, but safeguards from the center and effective mechanisms of only if it is effectively designed. Local authorities and agen- participation. Decentralization can mean different things. Here it Figure 6.4 refers to the formal devolution of power to local deci- Decentralization lowers the cost of raising the sionmakers. Less extensive forms of decentralization in- income of poor people in South Africa elude deconcentration (the central government posts Program cost of transferring 1 rand to a poor person, employees at the local level) and delegation (powers are 1995-97 delegated to the local level).32 The size of decentralized Rand units of government can vary enormously: decentraliza- 15 tion to the state or province in Brazil, China, and India merely breaks government into units the size of many 12 countries. Decentralization to smaller units increases the scope for interaction with the citizenry served. 9 Decentralization can make state institutions more re- sponsive to poor people, but only if it allows poor peo- ple to hold public servants accountable and ensures their 6 participation in the development process. The pace and design of decentralization affect its impact on efficiency, 3 accountability, participation, and ultimately poverty re- duction. But only general principles from successful mod- * els can be transferred from one setting to another.33 None Limited Partner Sole implementer Type of community involvement in program Moving,programs closer to users Local information has many advantages. It can help iden- Source: Adato and others 1999. tifv more cost-efficient ways of building infrastructure, pro- MAKING STATE INSTITUTIONS MORE RESPONSIVE TO POOR PEOPLE 107 Box 6.4 While a certain degree of fiscal devolution is needed for Community monitoring can reduce effective decentralization, it carries the risk of exacerbating environmental pollution inequalities between regions.37 In China, where provinces and local bodies are expected to be self-financing, social ser- Poor communities benefit directly from regulations that re- vices are greatly underfunded in poorer provinces."8 Mech- duce pollution, but where enforcement is weak-as in many developing countries-companies run little risk of being anisms for redistribution from the central budget can caught and punished. Polluting firms thus have little incen- mitigate these inequalities, but this is politically contentious. tive to clean up their activities, and firms that respect legal The problem needs to be addressed through consensus limits have less incentive to cut their pollution. building and tax sharing so that the central government has A new approach combines public information disclosure buidga to sarn tralsgovernmentshay with market-based incentives to encourage factory man- resources to make transfers where necessary. agers to improve their environmental performance. In some Sitpportand safeguards from the center. Central support countries local community representatives negotiate with gov- is required to ensure that national policies are adhered to ernment regulators and factory managers to agree on ac- and to coordinate the interregional interests of different- ceptable pollution levels and set pollution charges accordingly. administrative units-as with highway charges and ac- Elsewhere, public information enables consumers, bankers, and stockholders to evaluate a company's environmental cess to common water resources. Common macroeco- record before deciding whether to buy a product, lend money, nomic and redistributive goals also need to be supported. or trade the company's shares. The danger of decentralization without safeguards is il-- Because poor people are less able to protect themselves lustrated by the situation of Brazil in January 1999, from industrial pollution, their communities particularly value public information about which companies pollute and how when one state's action threatened the macroeconomic their discharges affect health. Where governments have stability of the entire country.39 provided local communities with reliable pollution data, poor Support for training is also required. Studies of successful people living in the vicinity of industrial polluters have ne- decentralization indicate the importance of creating ad-- gotiated better arrangements for compensation and cleanup. 40 Many local governments The results so far are promising. In Indonesia the gov- ministrative capacity.lack the ernment has cut industrial water pollution sharply by moni- administrative capacity for large-scale decentralizatiori toring factory discharges and bringing public pressure on and need training in accounting, public administration, factories by publicizing their emissions data. financial management, public communications, and com-- Source: World Bank 1999k. munity relations. If subnational governments have strong administrative capacity and accountability mechanisms, decentralization can reduce the scope for corruption. IF they do not, it can increase corruption and reduce access cies need considerable autonomy, including on fiscal to basic social services,41 as in Central Asia, the South Cau-- matters, as well as considerable support and safeguards casus, and the Baltics.42 from the center. Moreover, decentralized government Safeguards are also needed to monitor financial pro-- needs mechanisms to ensure high levels of participation bity and discourage the capture of local bodies by pow-- in the design and monitoring of programs and policies erful elites. One of the most serious pitfalls o:F by all sections of the population to be served. decentralization occurs when power imbalances are Autonomy andfiscal decentralization. Local authorities large at the local level. In such a situation higher lev-- need to have enough fiscal control to plan their activities. els of government, less subject to local political pres- But locally raised revenues are often only a small part of the sures, may be more motivated than local bosses to help budget of decentralized units, weakening ownership of lo- the disadvantaged. For example, the U.S. federal gov-- cally designed policies and threatening their sustainability. ernment has a long history of doing more to proteci: While decentralized units need an adequate budget base, minority civil rights than state governments, which enforcing hard budget constraints is also essential, to make have greater representation of those interested in sub-- them accountable. If ad hoc funds from outside units are verting those rights.43 Studies of Argentina indicate that- available to meet budget shortfalls, local bodies can lose their subnational governments can sometimes be less effec-- incentive to function efficiently. Moreover, suchi funding tive than central governments at targeting poor areas. 44 erodes the real power of the local body and its ability to ef- Similar problems are noted elsewhere in Latin Amer- fect change, as attention goes to extracting these benefits.36 ica and South Asia.45 108 WORLD DEVELOPMENT REPORT 2000/2001 Participation. Widespread popular participation is The politics of poverty reduction: vital to successful decentralization-without it, the po- pro-poor coalitions tential benefits of local information cannot be realized. Moreover, participation creates a virtuous circle. Partic- If we aren't organized and we don't unite, we can't ask ipating in local government helps build civil society and for anything. ensure that majority needs are heard and goals are Poor womnani, Florencio Varela, Argentina achieved. It also helps increase the voice of poor people in local affairs. Pro-poor coalitions that link the interests of the poor and One direct way of ensuring participation is to hold the nonpoor are important for poverty reduction. Im- regular elections for local government. Electoral rules proving the capacity of poor people to participate pro- can also foster broad participation by reserving seats for ductively in economic activity also helps lay the foundation marginal groups. In India a third of panchayat (local for faster growth. The state can support the growth of pro- council) presidents, vice presidents, and elected mem- poor coalitions by: bers must be women. In addition, certain other disad- * Fostering a political climate favorable to pro-poor ac- vantaged groups must be allotted memberships and tions and coalitions. executive positions proportional to their number in the * Removing legal barriers to pro-poor associations and area. Such measures can transform power relations over offering them technical and other support to scale up time.'6 their activities. Participation on a more frequent basis than just at elec- * Fostering state-community synergies and increasing the tion time also needs to be fostered. In Bolivia, Brazil, and capacity of poor people to participate in development the Philippines decentralization laws require local gov- and local governance. ernments to incorporate or formally associate grassroots Such transformations are essentially political and have to organizations with their deliberative procedures and to be effected through political processes involving changes goive such organizations a role in administering services in political configurations and power balances.5 and projects.47 Successful participatory budgetmaking in Porto Alegre, Brazil, shows that having local communi- Creating political supportfor pro-poor ties decide on the use of municipal resources can be very actions and coalitions effective for local development.48 Good information The interests of the poor and the nonpoor are intertwined channels between governments and communities are in many ways, making it beneficial for the nonpoor to also necessary for good results. In Chile, where calls for take an interest in redistributive measures and pro-poor community fund proposals are publicly broadcast and for- actions. This interest can be motivated by a recognition mats for project presentation are distributed through that efforts to reduce poverry can promote social and eco- municipalities, a survey of beneficiaries found that fund nomic development for the whole nation, thereby also disbursements were biased toward neighborhoods and so- raising the living standards of the nonpoor. The indus- cial organizations well connected with municipal and trializing economies of East Asia, where the creation of regional governments. Those with weaker connections re- a skilled, healthy workforce was crucial to success, show ceived fewer funds.49 that investing in mass education and human capital for- Decentralizing powers and resources to the mation provides a significant boost to national eco- submunicipal level-such as neighborhoods or villages- nomic growth. requires special effort, bur the benefits can be consider- Control of communicable diseases is another case in able.50 In South Africa partnerships between communities which all citizens benefit from programs focused on the poor, and local governments sharply increase the probability of as it is almost impossible for any group to avoid these dis- long-term returns to the community.51 In Guinea a pilot eases unless the sources of contagion are eradicated. Poor project showed that communities are adept at designing people, undernourished and living in environments with and managing such projects as building and maintain- greater exposure to disease, are especially vulnerable to in- ing new infrastructure. Communities mobilized local fection. They are also less likelyto receive adequate preventive resources, used grant funds equitably and efficiently, and and curative health services. Thus poor people tend to targeted funds to help vulnerable women and children.52 form pockets of contagion from which diseases can spread MAKING STATE INSTITUTIONS MORE RESPONSIVE TO POOR PEOPLE 109 to other groups. This was one of the main driving forces Box 6.5 behind the public health movements in the West at the turn National coalitions against communicable of the 20th century (box 6.5). The spread of disease is in- diseases in the West tensified today by the vastly increased volume of travel: if Sanitary neglect is mistaken parsimony: the physical strength health services are of poor quality or unaffordable for poor of a nation is among the chief factors of national prosperity. people in one country, drug-resistant strains of malaria -John Simon (1858) as cited in Rosen (1993) and tuberculosis can spread around the globe. So both na- The public health movement in Europe and the United States tional and global efforts are needed to address some of the brought rapid improvements in the health conditions of poor health problems of poor people (chapter 10). and rich alike in the late 19th and early 20th centuries, long Another motivation for the nonpoor to support pro-poor before the discovery of antibiotics. The politics of public re- sponsibility for reducing communicable diseases were mo- action is the specter of mass migration to urban areas, with tivated by a blend of economic, political, and humanitarian attendant problems of growing slums and rising demands interests. Industrialists were concerned with reducing the on already overburdened urban services. China and India drain on labor force productivity. States were concerned have reduced incentives for urban migration by providing with having enough fit young men to serve in the army and expand spheres of influence. Elites felt that their environment was detrimentally affected by the ill health of poor people- plying schools, health services, electricity, and other basic and that the dangers for the population as a whole needed amenities, but also ensuring that employment creation is to be reduced. Intellectuals pointed out the connections be- geographically dispersed and that transport networks allow tween ill health and poverty, demanding radical change as a people to commute to work from their villages. solution to the problems of endemic and epidemic diseases. people to commute to work from their villages. To reduce everyone's exposure to communicable dis- To build political support for public action against poverty, eases, strenuous efforts had to be made to improve the health governments have to enhance the perception of common in- of poor people. Measures included control of food and drugs, terests between the poor and the nonpoor."4 Key to this is smallpox vaccinations, and quarantine. Central to the en- systematiayntroucing the public debate the notion deavor was securing a pure water supply, effective waste disposal, clean streets, and reduced pollution. Housing reg- that poverty reduction is a public good and can further the ulations were enforced to ensure adequate ventilation, toi- well-being of the nonpoor. How these issues are framed in let facilities, drainage, and sewerage in homes. Restrictions the public debate can greatly influence the outcome. Povertys on private behavior included forbidding spitting and urinat- character, causes, and solutions are malleable concepts, which ing in public spaces and banning livestock from domestic premises. Massive health education campaigns used ex- can be reinterpreted and represented m a vanrety of ways, many tensive outreach to change personal health behavior and in- of them conducive to public action against poverty. In the crease people's understanding of how to avoid ill health and early 20th century state governments in the United States were care for the sick. These state interventions, combined with persuaded, mainly by national middle-class women's orga- rising living standards, dramatically improved health and life nizations, to spend public money to support poor famlies- expectancy between 1880 and 1920. Paradoxically, improvements in curative technologies in on the grounds that this was the only way to protect the moral recent decades may have led to less vigilance against com- and physical integrity of the nation.55 municable diseases in some developing countries. These Understanding the benefits of helping the less for- powerful curative technologies need to be combined with tunatecantusbeapowrful stimulus for pu c astrong public health policies aimed at improving environ- .uate can thus beapowrfustmulmental sanitation and encouraging healthy lifestyles. This tion. Without such understanding, the living conditions will help increase economic growth and reduce poverty, of the disadvantaged are sometimes used to justify their averting negative consequences for national and global health further exclusion. Latin American elites have some- as drug-resistant strains of diseases multiply. times viewed poor people as a danger to public well- Source: Rosen 1993; Preston and Haines 1991;Schofield, Reher, and being. This mindset makes it more difficult to eradicate Bideau 1991; Caldwell and others 1990. poverty and mitigate its negative impact on the econ- omy and society. Facilitating the growth ofpoor people's forming associations and to provide an administrative associations and judicial framework supportive of such associations.56 The state's most important task in fostering poor people's Without this, it is very difficult for poor people's associ- organizations is to remove legal and other barriers to ations to flourish and to influence public policy. Poor 110 WORI. ) DEVELOPMENT REPORT 2000/2001 people face enormous constraints in forming associations agencies.59 To counter these problems, some peasant or- to increase their voice and improve their circumstances. ganizations in Bolivia and Ecuador have worked through Thev usuallv take little part in politics because participa- NGOs to link up with national agricultural agencies, enor- tion seems irrelevant to their primary concerns, futile, or mously expanding their range and effectiveness.60 both. They often have low expectations of their In most developing countries NGOs are central actors government-and may even fear reprisals from state or in antipoverty policies and programs.61 The social and ed- local authorities if they organize. Even for matters where ucational background of many NGO staff enables them the government is viewed as relevant, poor people see in- to interact easily with the staff of national institutions, and dividual and collective efforts to exert influence as hav- they can help create bridges between these institutions, out- ing little effect. When poor people do participate, their side agencies, and grassroots organizations. NGOs can also class identity is not the only influence on their decision be very effective in delivering technical assistance to poor to do so. As with other citizens, the forces that move people,62 as Mopawi has been in Honduras (box 6.6). them to action are often tangible, short term, and local. In Bolivia a Dutch NGO helped a campesino federa- Reducing asymmetries in information can do much tion link with research institutions involved in the national to change poor people's hesitance to participate-and to empower them. Formal education enables people to gain Box 6.6 access to better economic opportunities (chapter 5) and NGOs can help mobilize and empower gives them the means to articulate their needs and de- communities mands in public forums and in political processes. All this is enhaniced by widespread dissemination of informa- Since 1985 the NGO Mopawi (Moskitia Pawisa, or Devel- *on. TodayIs information technology and lowr II Inor- opment of La Mosquitia) has been working alongside the in- digenous communities of La Mosquitia, a remote area of mation costs, combined with rising demand for greater western Honduras and one of the last remaining areas of trop- access to public documents, can have powerful benefits ical forest in Central America. Over the years Mopawi has for the poor. developed a large and complex development program. It Major impediments to organization by poor people are has worked to change government policy for the region through continuous lobbying and advocacy, helping to form lack of rimc, resources, information, and access to outside links between government, international NGOs, research sources of help. Added to that are phvsical constraints on organizations, and indigenous organizations to raise aware- collaboration, such as geographic dispersion and poor ness and inspire action. It has worked with local communi- traiisport aiid con*tlLinications infrastrL CtLIre. EhIc ad Ities to find ways of improving livelihoods without harming transport and communicationsinfrastructu. Ethnic the environment. Mopawi has also addressed deforestation other social divisions are another impediment (chapter 7).9 in La Mosquitia, combining advocacy with practical preven- Despite these difficulties, many countries have seen an ex- tion. Most of its staff are from La Mosquitia, which has plosion of community-based, participatory grassroots or- proved to be a major strength. ganizations in the past few decades. Throughout Latin The organization has helped improve people's livelihoods by identifying alternative models of resource use and in- America popular and indigenous organizations, some- volving local communities in decisionmaking and manage- times based on traditional forms of association, now give ment. For example, small businesses have been set up, and voice to the underprivileged and deal with immediate experiments conducted with agriculture and agroforestry. An needs in health, schooling, and public infrastructure. agroforestry and pasture project being conducted with set- tlers and indigenous communities includes experiments with Such grassroots organzatons require many forms of sustainable management of the forest and restoration of support from the state and from civil society. They often degraded areas. And efforts targeting women are aimed at need technical assistance and skill building to become sus- cultivating vegetable plots to improve health and nutrition. tainable and effective. Thev also need help in scaling up To strengthen local organizations, Mopawi has worked ing up with Masta (Mosquitia Alsa Tanka), the federation of repre- their membership, range of functions, and political en- sentative indigenous organizations in La Mosquitia. With gagement. Many grassroots initiatives are limited in Mopawi's help, local organizations have taken on legalization scope and depth and never reach the national political of land ownership and use rights and developed their capacity arena. Studies in Latin America have found that some or- for advocacy. ganizations are effective in addressing some of the im- t~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Source: Brehm 2000. mediate concerns of poor people, but their sustainability is constrained by difficulties in linking up with external MAKING STATE INSTIT'UTIONS MORE RESPONSIVE TO POOR PEOPLE III potato program by contracting with an international ex- sources. In the 1987 drought the state used agricultural pert to work with the federation at the onset of the pro- extension workers to break the grip of patronage in the ject. The consultant had no difficulty establishing distribution of drought relief. But sustaining such suc- high-level contacts with research institutions, and on cesses requires much continuing effort. the basis of these meetings the federation established The Brazilian experience also shows that many of these strong links with the national potato program. The re- obstacles can be overcome by bringing grassroots electoral sult was higher crop incomes for federation members.63 pressure to bear on local governments. Political interference Sometimes NGOs reflect the political system in has been kept at bay by the state governments' insistence which they thrive, or local interest groups, and thus may that municipal councils for disbursing development funds not serve the interests of poor people as well as they have at least 80 percent representation from end-user com- might.64 NGOs are no panacea-it is important that munities. Moreover, if communities feel they are treatedun- they be accountable for their actions, especially to the fairly by the municipal councils, they can apply for funds poor groups that they seek to represent. directly from the state government.67 The state can undertake several key actions to foster Fostering state-community synergies developmental synergies between communities and local in growth andpoverty reduction governments (figure 6.5): The state can facilitate interactions between local admin- * Generating community demand for better public ad- istrations and communities to engender development and ministration and service delivery, through intensive dis- reduce poverty.65 There are two main aspects to this role: semination of information. reducing obstacles to collective action in communities * Forming dense networks between the state and com- and encouraging greater collaboration between commu- munities and making available to communities the in-- nities and local governments. To forge ties within com- formation and technical, marketing, credit, and othe.r munities and facilitate local collective action, the state can support they need to implement programs. initiate programs that build up the assets of poor people * Changing local agencies' mode of operation by putting and make public services more accessible. Such programs pressure on them from above and below. In Brazil the reduce the perception among poor people that their sur- state used official job recognition to motivate staff. vival depends on avoiding risks and keeping their patrons a Motivating grassroots workers and leaders through happy-releasing their energies to pursue actions for up- positive and negative sanctions, including respect: ward mobility and to collaborate with others on a more from peers. Where the workers are also community equal footing. members, as in the Republic of Korea, the potential The combination of a more egalitarian social organi- sanctions are especially strong. zation at the community level and better local adminis- * Adjusting the roles of higher levels of government, tration enables the creation of powerful coalitions for rapid training and motivating their staff to focus on man- development. Strong links between local administrations aging overall strategy, and providing technical support.. and communities improve service delivery and reduce the regulations, and facilitation. potential for local capture of development programs. These initiatives yield substantial political payoffs for This arrangement has been used successfully in very dif- the governments in legitimacy and in popular support. ferent political and administrative settings: Brazil in the In a municipal election in Brazil candidates said that if 1 980s, the Republic of Korea in the 1 960s and 1970s, they wanted to get elected, they had to support the new and Taiwan, China, in the 1 950s.66 arrangements for increasing public accountability of local The example of Brazil shows that institutional change government and improving public service delivery.68 is considerably more difficult in highly unequal settings. This helped to strengthen potentially weak governments Lacking the prior extensive land reform of East Asian and motivate them to engage in these difficult tasks. At countries, the state in its efforts to reform local govern- the same time the conditions for a plural polity were ment had to tackle problems of landlord interests and po- strengthened. litical connections with local government. This created Collaboration between communities and local gov- problems, because large landowners, private contractors, ernments can also promote many different forms of de- and relief suppliers were accustomed to cornering re- velopment. In addition to local improvements in 112 WORLD DEVELOPMENT REPORT 2000/2001 Figure 6.5 State-community coalitions can foster rapid development and better service delivery Central or state government a Ordevelrpliting (incentives for action: a Political and economic a Raise aspirations, generate and grassroots workers demand (incentives for action: aOffer information On services public image, material benefits) communities (incentives for action: Pressure for quality services development and and peer monitoring political empowerment) Source: Das Gupta, Grandvoinnet, and Romani forthcoming infrastructure and living conditions, such collaborative Figure 6.6 efforts have delivered health and drought relief in Brazil Good political and administrative institutions and supported industrial production for export markets go hand in hand with economic growth in Taiwan, China. Changes are incremental and can often take time, Economic growth index but as success stories accumulate in a given setting, they 0.03 create a demonstration effect for others. The examples sug- 0.02 * gest that it is possible, in the space of decades, to reengi- .* neer state institutions to quicken the pace of development, 0.01 * 0 growth, and poverty reduction. They also show that with * * creative political thinking these changes are possible even 0 .-_--- --- - in relatively weak institutional settings. S jS* Political regimes and poverty -0.02 Voicelessness and powerlessness are key dimensions of -0.03 * poverty, and an important aspect of voice relates to po-04 litical rights and civil liberties.69 Democracy is intrinsi- -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5 1.0 1.5 2.0 cally valuable for human well-being as a manifestation of Quality of institutions index human freedom. Political freedoms have enormous im- 70 Note: The quality of institutions index is estimated on the basis of pact on the livcs and capabilities of citizens . measures of civil liberties and political rights, the quality of Participatory political processes can also help build a government institutions, the rule of law, and public spending on good institutional base for the polity, society, and econ- social insurance. Source: Rodrik 1999a. omy, enabling all voices to be heard and to interact in de- termining outcomes (figure 6.6).71 Civil and political MAKING STATE INSTITUTIONS MORE RESPONSIVE TO POOR PEOPLE 113 liberties, along with competitive elections, are powerful The checks and balances of participatory democratic instruments for holding governments accountable for regimes-and the procedures for consensus building- their actions. To translate this potential into reality, many limit the scope for rent seeking and drastic policy rever- institutions need to be in place to ensure that democra- sals, offering a much more reliable and sustainable path tic processes function as they should-among them, to development.76 Participatory political regimes are as- independent media to monitor electoral and administra- sociated with more stable growth77-very important for tive processes, an independent judiciary to uphold the con- poverty reduction, given the highly adverse effects that stitution and rule of law, and strong parliamentary shocks have on poor people (chapters 8 and 9). There are institutions with the capacity to monitor the executive several reasons for this association. through such mechanisms as public accounts committees. First, participatory political processes encourage the Building these institutions takes time, and constant vig- use of voice rather than violence to negotiate conflict. ilance is required to ensure that democratic processes Combined with guaranteed political rights, these function as they should. But it is worth the effort, for these processes reduce the potential for ethnic and other in- processes offer the most effective means of guaranteeing tergroup conflict, averting major sources of social and voice and participation. economic vulnerability for poor people.78 For example, India's strong democratic political institutions help me- Promoting democratic politics to foster stable diate the potentially conflicting demands of its highly environments for growth heterogeneous population. Evidence on the relationship between the type of politi- Second, political and civil rights and a free press cal regime and the rate of economic growth is mixed.72 allow people to draw attention to their needs and de- In part this reflects measurement problems,73 and in part, mand appropriate public action.79 This is especially the experiences of growth with significant poverty re- important for averting or responding quickly to major duction in a few notably development-oriented coun- disasters. And third, democratic elections confer legit- tries, such as the Republic of Korea, before they became imacy on governments, fostering sociopolitical and eco- pluralist democracies. These countries developed the pre- nomic stability. conditions for a developmental state-notably, political elites committed to development and supported by an ef- How can democracy deliver more effectively ficient public administration that was insulated from po- for poor people? litical pressures and had close links to the business Democracy-both representative and participatory- community.74 is a good in itself. But democratic political processes Another major factor in the success of these economies alone are not enough to ensure that poverty reduction was their early emphasis on equity-especially extensive is taken as a key priority in society's efforts. Political and land reform and universal education, which laid the foun- social ideologies shape the extent to which democra- dation for rapid and equitable growth.75 These efforts were tic systems actually reduce poverty. Different philoso- facilitated in some cases by the devastation caused by war phies underlying welfare policies in OECD countries and the attendant disempowerment of entrenched elites. produce very different outcomes in poverty reduction- These circumstances lowered the resistance of politically despite the fact that all these countries have a long powerful vested interests to drastic land reform. history of democratic political institutions and high per In most nondemocratic settings, however, lack of in- capita income (box 6.7). stitutionalized accountability has resulted in poor per- Representative politics allow all citizens' interests to formance in growth and poverty reduction. Even be expressed, but the outcomes depend on how differ- successful developmental states point to an important ent group interests play out.80 Groups that are politically lesson: undemocratic regimes face serious abuses of connected or better educated have a natural advantage state power, and they are prone to rapid policy rever- over others in influencing public policy. This is reflected sals that can make their development gains fragile. in the United States in the large discrepancies between These states are moving to resolve some of these prob- affluent and poor communities in funding appropriations lems by changing their political institutions to increase for law enforcement and public schools. In developing official accountability. countries, where the distribution of education and po- 114 WORLD DEVELOPMENT REPORT 2000/2001 Box 6.7 litical know-how is far more skewed than in the United Politics and poverty in OECD countries States, large segments of the population remain under- served.81 In C6te d'Ivoire 35 percent of public educa- Poverty is not restricted to developing countries. There are tnon spending goes to the richest 20 percent of the significant pockets of poverty in some OECD countries. Al- population (see table 5.1 in chapter 5), and 55 percent though these countries are all affluent market economies with m i 82 democratic systems, ideological differences and corre- of tertiary students come from this group. Democra sponding differences in popular support for poverty reduc- tic politics are also subject to manipulation by political tion programs result in very different levels of povertyfor their leaders. Leaders may favor spending resources on im- citizens. mediate consumption rather than investment, on pop- A comparative study of poverty trends in Germany, the ulist rather than productive measures..3 In addition, Netherlands, and the United States examined the incidence of poverty and how it was affected over a 1 0-year period by interest groups can be bought off or co-opted with fa- government programs. The study found wide differences in vors from politicians.8' poverty incidence among the three countries, differences that There are three main wavs to strengthen the institu- were widened by government programs. Levels of "pre- government" poverty (based on earned and unearned per- sonal income excluding taxes and transfers from government) them more effective at reducing poverty. First, democ- vary, largely a result of marked differences in labor laws and ratic processes must permeate all major levels of deci- other market factors. Pension payments reduce poverty in sionmaking. Some regimes are more democratic in all three countries, but the design and impact of other pub- lic transfers and taxes aimed at reducing poverty differ. principle than in practice. Others, like India, are gen- Especially striking are the low levels of poverty in the umnely democratic at most levels but have historically Netherlands, a result of universal benefits. Although the found it difficult to ensure that political accountability transfers have large targeting errors-they go to the non- reaches all levels of decisionmaking, particularly for the poor as well as the poor-they do not appear to have re- poor. Indias ongoing panchavati raj drive toward de- sulted in slower economic growth compared with the other countries. ~~~~~~~~centralization and community empowerment is an ef- countries.I fort to correct this by increasing the powers of local Government programs widen differences in elected councils. poverty among OECD countries Second, citizens must be given svstematic access to Share of population under age 60 living in poverty, information so that thev can hold their civil servants and 1985-94 politicians accountable. If information on budgets and Percent 15 on the use of funds-from the federal to the local level-is made available in newspapers and other in- formation sources, people can hold their leaders ac- countable for results. Such public accountability can help 10 ~ to reduce inefficiency and corruption. This dissemina- tion of information needs to be legally mandated to en- sure that it does not stop with a change of government. Germany Progress in information technology and increasing ex- posure to global currents help to create a new environ- ment of public awareness that reinforces democratic politics. \ Netherlands Third, strong civil societv organizations can promote 0 1 2 3 the political empowerment of poor people, pressuring the Government policy steps state to better serve their interests and increasing the ef- Note: Policy steps: 0 = "pregovernment" poverty rate, fectiveness of antipovertv programs.85 Case studies in 1 = social insurance pensions, 2 = 1 + other public transfers, the Indian state of Kerala and elsewhere show that a 3 = 2 + taxes. Poverty is defined as having less than half the highlv engaged civil societv contributes to better outcomes median disposable household income of the country. Incomes . '6 (net of inflation) were cumulated for the 10-year period. in health and education. What is needed is an enabling Source: Goodin and others 1999. institutional environment for civil society to develop and thicken (box 6.8).8" MAKING STATE INSTITUTIONS MORE RESPONSIVE TO POOR PEOPLE II5 Box 6.8 The evolution of civil society and state reform in Mexico Traditionally, Mexico has had well-institutionalized systems for chan- manded that the basic social contract between state and society neling and controlling political activities-and using state resources be renegotiated, political leaders and parties attempted to re- to cement political support for the regime. These systems were spond in ways that would give them the leverage to determine more concerned with controlling society's demands than re- the scope and nature of that contract. This conflict remained un- sponding to them. The state had developed a highly effective and resolved in the early 1990s, with the potential to develop a more sophisticated machinery for co-opting and managing demands open political system depending on the capacity of civil society to and dissent. Although the capacity of civil society to demand re- force change. sponsiveness was limited, the state was sensitive to the ongo- In recent years prospects of real change have emerged in Mex- ing need to cement loyalties, garner support, and resolve conflict. ico. A much more open and democratic political process has de- In the 1980s these relatively strong political capacities came veloped, and an independent electoral commission and civil society under siege: responsiveness, representation, and participation organizations widely encouraged citizens to vote according to their became issues of great contention. Day-to-day management of po- conscience, free of coercion and inducements, in the July 2000 elec- litical and economic conflict became an increasingly difficult task tion. While much remains to be done to open the door further to for public officials. Financial resources reached historically low lev- civil society participation and the expression of citizen demands, this els, and government legitimacy plummeted. While civil society de- is a cautious, lurching, but in the end irreversible, first step. Source: Grindle 1996. * * * poor people are often voiceless-and their interests figure little in public policy. Poor people need direct Respect for the rule of law, an efficient public ad- voice in the interventions that affect their daily lives, ministration, and high-quality political systems facil- as well as the ability to organize and vote. Actions are itate the emergence of state institutions inclusive of poor needed to bring down barriers-legal, political, ad- people. But the impact of these factors on poverty de- ministrative, social-that work against particular groups pends on how effectively they are translated into em- and to build up the assets of poor people to prevent their powerment at the community level. Even in states with exclusion from the market. Some major social barriers extensive political and civil liberties and with govern- to poverty reduction are discussed in the following ments that are neither captured by elites nor corrupt, chapter. N~~~~~~~~~~ CHAPTER 7 Removing Social Barriers and Building Social . 0 Institutions SJ ocial institutions-kinship systems, community come these inequalities must be accompanied by ef- organizations, and informal networks-greatly affect forts to raise awareness about culturally based atti-- poverty outcomes. They do so by affecting the pro- tudes such as those toward women and people of ductivity of economic assets, the strategies for coping different races, religions, or ethnic origin. Otherwise with risk, the capacity to pursue new opportunities, and these measures will be unable to produce real change. the extent to which particular voices are heard when Social barriers can take many forms. Here the focus is important decisions are made. Social institutions can on key barriers arising from gender inequality, social help poor people get by and get ahead.' But they can stratification, and social fragmentation.3 also place barriers between poor people or the socially disadvantaged and the opportunity and resources they Gender discrimination and poverty need to advance their interests. Discrimination on the basis of gender, ethnicity, race, religion, or social sta- Until we became organized as a SEWA cooperative, the tus can lead to social exclusion and lock people in long- middlemen could cheat us. But now I can negotiate with term poverty traps. them as the representative of our cooperative and as an Values, norms, and social institutions may reinforce elected member of our local council. One day near the bus persistent inequalities between groups in society-as stop, I heard a couple of men saying, "There's the woman with gender-based prejudice throughout much of the who is giving us all this trouble. Shall we beat her up?" I world, the caste system in India, and race relations in told them, "Go ahead and just try it. I have 40,000 South Africa and the United States.2 In the extreme, women behind me. " these social divisions can become the basis of severe de- -Woman laborer, speaking at World Summitfor privation and conflict. Legal and other measures to over- Social Development and Beyond, Geneva, June 2000 117 II8 WORLD DEVELOPMENT REPORT 20002/001 The extent and manifestations of gender inequality vary vorce laws were changed to allow women equal rights to among societies, shaped to a considerable degree by kin- property acquired during marriage, and child custody is ship rules.4 Rules of inheritance determine ownership of no longer granted automatically to the father. But the law productive resources. Rules of marriage determine women's continues to insist on male household headship, which domestic autonomy: if these rules require that women join women's organizations see as the main source of gender their husband's family, women have far less autonomy than inequality in the family and in other social institutions. if they are able to form a new household or live with their So while women in Korea have become educated and par- own family (which is uncommon). The most pervasive ticipate actively in the labor force, their unequal status forms of gender inequality appear where both inheritance serves to maximize their economic contribution while and marriage rules are heavily weighted in favor of men. minimizing advances in gender equity.9 By contrast, where such rules are more gender balanced, In many countries women continue to be denied even women have greater voice in the household and in pub- basic legal rights. In Botswana, Lesotho, Namibia, and lic spaces and face fewer constraints on becoming inde- Swaziland married women, according to both customary pendent economic and social actors.5 and common law, are under the permanent guardianship Norms for gender roles and rights form part of the of their husband and have no independent right to man- moral order of a community and permeate other insti- age property (except under prenuptial contract)."0 In tutions, including those of the state. This further rein- Guatemala men can restrict the kind of employment forces gender inequities, unless conscious efforts are made their wife can accept outside the home. In some coun- to avoid it. Legal systems play a key part, either reinforcing tries women need their husband's permission to obtain customary gender rights and roles-or deliberately seek- a passport and move about freely.1l ing to alter them. Also important is the provision of Poor women face a double disadvantage in access to public goods and services, which often bypass women un- resources and voice-they are poor, and they are women. less specific efforts are made to reach them. Poor people have much less access to education and health care than the nonpoor, and the gender gap in . X i :: ,li ,e. ; z rc d ZC access tv /resturces these services is larger among poor people.12 The same Customary gender norms and values can lead to politi- is true for credit and agricultural extension services: un- cal, legal, economic, and educational inequalities that per- less strong countervailing measures are taken, the poor petuate women's lack of access to resources, control over receive less than the nonpoor, and women receive the least. decisionmaking, and participation in public life. Greater Studies from many countries show that agricultural ex- political representation could help change this-in no tension agents focus on male farmers, even though women country do women hold more than a very small share of are often the primary cultivator because husbands work the seats in parliament.6 off the farm. 13 So women face disadvantages not only in Legal systems can constrain women from becoming land ownership, but in gaining access to the resources and independent economic actors. In many countries family information that would improve yields. laws are heavily stacked against women, restricting their rights in divorce and in inheritance of land and other pro- 7'he toll of gender inequality on society ductive resources. In most developing countries titles to If the rights of men and women are flagrantly unequal, land are normally vested in men.7 Since the great majority it is very difficult to establish a democratic and partici- of the world's poor people live in agrarian settings, this patory sociopolitical order and an environment of equal is a fundamental source of vulnerability for poor women. opportunity. Moreover, the more extreme manifesta- Some countries use the legal system to formalize cus- tions of power inequality between men and women tomary rules that explicitly limit women's rights. In the constitute gross violations of human rights. Domestic vi- Republic of Korea, for example, customary laws re- olence has been shown to be startlingly prevalent around stricting women's rights were formalized in the Civil the world-among people at all income levels (see box Code of 1962, and women's legal rights have been very 8.1 in chapter 8). slow to improve. After decades of struggle by women's or- In some societies the lower value assigned to women ganizations, key amendments in 1990 gave women the and girls translates into excess mortality. Estimates based right to inherit their parents' and husband's property.8 Di- on official national censuses find that as a result of ex- REMOVING SOCIAL BARRIERS AND BUILDING SOCIAL INSTITUTIONS 119 cess female mortality, about 7 percent of girls under age Figure 7.1 five are "missing" in China and Korea and more than 4 Closing the gender gap in schooling more percent in India and other parts of South Asia. 14 With- rapidly would boost economic growth out such discrimination there would be an estimated 60-100 million more women in the world. 15 Average annual growth in per capita GNP, 1960-92 60-100 million more women in the world.15 Percent Gender inequality also has strong repercussions for 3.5 human capital in the next generation, because the bur- 3.0 Closing the gap den of bearing and rearing children falls largely on at East Asian pace women. Women deprived of education and decision- 2.5 making power in the home face serious constraints in rear- l ing healthy, productive children. They also tend to have 2.0 Actual more children than they wish, compounding the pressures 1.5 on themselves and their family. Better-educated women are able to communicate better with their spouse about 1.0 family size decisions, use contraception more effectively, and have higher aspirations for their children.16 0.5 Low autonomy for women takes an independent toll. 0 Studies in China and India find that even controlling for Middle East South Asia Sub-Saharan education, household income, and other socioeconomic and North Africa Africa characteristics, low domestic autonomy is associated with Source: World Bank (forthcoming a), estimated from Klasen (1 999). higher infant and child mortality rates.17 Studies con- sistently show that women's education improves child sur- vival.18 And longitudinal studies in the United Kingdom country analysis indicates that countries that invest in girls' and the United States find that, controlling for other education have higher rates of economic growth (figure household-level factors, mother's education is associated 7.l).24 Country studies show the benefits of increasing with better child cognitive development.19 women farmers' access to agricultural extension, credit ser- Among children of women who have greater financial vices, and other productive inputs.25 Raising their edu- autonomy, either because they earn cash incomes of their cation increases their efficiency as producers, by increasing own or have a greater role in domestic decisionmaking, their adoption of new technologies and their efficiency nutrition and education are higher. Studies in Brazil in using resources. Analysis from Kenya suggests that giv- show that more income in the hands of mothers is asso- ing women farmers the same education and inputs as men ciated with better nutritional outcomes and physical de- increases yields by as much as 22 percent.26 For Burkina velopment.20 Microcredit programs in Bangladesh find Faso analysis of household panel data suggests that farm that giving income-generating loans to women improves output could be increased 6-20 percent through a more the nutritional status of their children, a result that does equitable allocation of productive resources between not hold for men.21 male and female farmers.27 Further analysis is needed to Education and autonomy reinforce each other. Women determine the impact of such a reallocation on overall with more education and greater domestic autonomy household income and nutritional well-being. are better able to nurture and protect their children.22 Low education and low autonomy make it more difficult for Scope for change women to obtain medical care, comply with instruc- While political and legal equality between men and tions, and follow up with the health care provider if the women have increased in most regions, it takes effort instructions seem ineffective. They also make it more dif- and perseverance to change people's gender values and ficult for women to obtain healthcare information, pre- beliefs.28 But much can be done-and much has been vent illness, and care for the sick. done-to improve women's voice and access to re- More equitable distribution of opportunities and re- sources by increasing their political representation, sources between men and women also leads more directly their legal rights, and their command over physical, fir- to higher economic growth and productivity.23 Cross- nancial, and human capital (figure 7.2). Efforts are 120 WORLD DEVELOPMENT REPORT 2000/2001 Figure , 2 Trends in female education and life expectancy reflect increasing equality between women and men Female gross primary enrollment rate Female to male gross primary enrollment ratio Percent 120 1.1 10 1.0__ 0.9 0.8 0.7 40 0.6 1970 1980 1990 1995 1970 1980 1990 1995 - East Asia - Europe and - Latin America - Middle East - South Sub- -- OECD and Pacific Central Asia and the and North Africa Asia Saharan Caribbean Africa Female life expectancy Female to male life expectancy ratio Years 85 1.15 80 _____-- 75 ' 1.10 70 65 1.05 55 / 1.00 50 45 0.95 1970 1980 1990 1997 1970 1980 1990 1997 Source: World Bank forthcoming a. under way in at least 32 countries to increase women's Women's legal rights have been broadened consider- political representation by reserving seats for them in ably in many countries. In a growing number of countries local and national assemblies.29 In India two amend- daughters and sons now have equal legal rights to inherit ments to the constitution reserve a third of local coun- from their parents. The existence of such legal rights does cil seats for women, giving rise to a new class of women not mean that deeplv rooted cultural norms immediately (some 600,000 strong) with political influence; simi- change, however. Moreover, the legal system often gives lar reservation is under consideration for higher polit- people scope for implementing their own norms. For ex- ical levels.30 In Argentina at least a third of the ample, the option ofwritingawill allows people to main- candidates on national election lists must be women.31 tain cultural norms on inheritance favoring sons.32 When REMOVING SOCIAL BARRIERS AND BUILDING SOCIAL INSTITUTIONS 121 Box 7.1 legislation conflicts too sharply with customary law, prob- Making land titling less gender biased lems can surface.33 Still, even if laws are not self-enforcing, in Latin America they are a necessary first step toward gender equity. More direct efforts to ensure women s access to pro- The land titling process, rife with inequities, has often reduced ductive resources include recent land titling programs to women's access to land. Statutory law in several Latin Amer- ican countries required that the beneficiaries of earlier land reform programs be heads of household. Since custom dic- ian Law gave top priority to redistributing land to house- tated that men were the head of the household, it was diffi- holds headed by women and to women who lacked cult for women to benefit from such programs. During the protection or had been displaced by war (including sin- 1980s and 1990s, however, reform measures changed, and gle and childless women).34 The scheme-"a parcel of the more progressive agrarian codes of the 1990s gave spe- cial attention to this problem. one's own"-was the only guarantee of secure livelihood A study based on gender-disaggregated data for six coun- for women and their children upon separation or divorce. tries (Chile, Colombia, Ecuador, Honduras, Mexico, and Peru) Several other Latin American countries have been work- shows that women make up a larger share of beneficiaries ing on this issue, with varying success (box 7. I). under current land titling programs than under past agrarian Wn men ths ise, wh ry su ccess (o 7.1). reforms. Still, several obstacles to improving women's formal Women also need more equitable access to credit and rghts to land remain: associated productivity-enhancing services. Studies of' * Women often are unaware of their rights or of the land the effect of networking schemes, such as group-based mi- titling program. crocredit, suggest that these schemes have enormous po- * Land titling projects are often arbitrary. The problem usu- . . r ally starts with lack of clarity about the bundle of property tential for reducig poverty Some of these credit programs, rights to land within one household: those of the wife, such as Grameen Bank in Bangladesh, are targeted more those of the husband, and those to jointly acquired prop- to women than to men.35 Using peer pressure and group erty. At the enforcement stage, this confusion often works obligations rather than legal contracts, group-based to women's disadvantage. . * - * Some legal dispositions are gender biased. Procedures schemes rely on social collateral rather than traditonal ceding rights to land often aim at individualizing land financial assets as security.36 The schemes have helped rights-one person per household. To work in favor of women acquire nonland assets and have also been asso- women, land titling programs need to give female heads ciated with positive effects on girls' schooling.37 of household priority, as in Chile. Critical to these programs are services that comple- Two sets of measures are particularly important for pre- venting gender bias in land titling and promoting the rights ment credit and savings facilities, such as training in en- of women: trepreneurial skills-especially for women, who are * Making joint titling of land to couples mandatory. Joint typically cut off from the normal paths for acquiring such. titling guarantees married women property rights to land that has been jointly acquired. In Colombia land titled jointly s G t to couples accounted for 60 percent of land adjudications cessful entrepreneurs. In southern Africa women own. in 1996, up from 18 percent in 1995. Land titled exclu- an impressive share of small, informal sector businesses: sively to men declined from 63 percent to 24 percent over 67 percent in Zimbabwe, 73 percent in Lesotho, and the same period. 84 percent in Swaziland.38 The next step is to ensure * Fostering partnerships between government depart- ments and NGOs that defend the rights of women-to greater access for women to business opportunities inL increase women's awareness of their rights and sup- the formal sector. port them in claiming title to land in the face of a pos- Recognizing the constraints women face in gaining ac- sibly hostile bureaucracy or family. In Bolivia and Ecuador, cess to public services and other opportunities makes where women's land rights featured little in the nego- tiations leading to new agrarian codes and where there antipoverty interventions more effective. In education,, was no movement toward joint titling or special rights female teachers and separate sanitary facilities-or even for women, the reforms did not improve women's land single-sex schools-can boost girls' enrollments in some rights. regions.39 Demand-side interventions can also be effec- Source: World Bank forthcoming a {based on Deere and Leon 1997 tive (box 7.2). In agricultural extension, efforts to hire and 1999); Deere and Leon forthcoming. train female extension agents-and to focus extension ef- forts on women farmers-help make new agricultural methods and technologies more accessible to them and increase productivity. 122 WORLD DEVELOPMENT REPORT 2000/2001 .iox 7.3 Using subsidies to close gender gaps Toward stronger female voice in policymaking: in education women's budget initiatives in southern Africa Evaluations of recent initiatives that subsidize the costs of The South African Women's Budget Initiative began as an schooling indicate that demand-side interventions can in- innovative "joint venture" between several NGOs and new crease girls' enrollments and close gender gaps in education. parliamentarians in the first post-apartheid government. The A school stipend program established in Bangladesh in 1982 parliamentarians were members of the Gender and Eco- subsidizes various school expenses for girls who enroll in sec- nomic Policy Group of the Joint Standing Committee on Fi- ondary school, In the first program evaluation girls' enrollment nance, while many of the NGO representatives were involved rate in the pilot areas rose from 27 percent, similar to the na- in budget-related and more general policy research. The pur- tional average, to 44 percent over five years, more than twice pose of the initiative has been to highlight the gender di- the national average (Bellew and King 1993). After girls' tu- mensions of the government's budget-including in taxation, ition was eliminated nationwide in 1992 and the stipend pro- expenditure, and the budget process itself-and to ensure gram was expanded to all rural areas, girls' enrollment rate that gender equity is better served by the budget process climbed to 48 percent at the national level. There have also and allocations. been gains in the number of girls appearing for exams and in The initiative has undertaken four rounds of budget analy- women's enrollments at intermediate colleges ILiang 1996). sis on a range of sectors. While the early rounds focused While boys' enrollment rates also rose during this period, they largely on the national budget process, the fourth has begun did not rise as quickly as girls'. to focus on local government and on dissemination of findings Two recent programs in Balochistan, Pakistan, illustrate and messages to a broader constituency of South Africans- the potential benefits of reducing costs and improving phys- to better equip ordinary citizens to engage in policy discussions. ical access. Before the projects there were questions about The South African initiative has inspired several others. A whether girls' low enrollments were due to cultural barriers three-year gender budget initiative was started in Uganda in that cause parents to hold their daughters out of school or 1997, led by the Parliamentary Women's Caucus in cooper- to inadequate supply of appropriate schools. Program eval- ation with the Forum for Women in Democracy, an NGO. Like uations suggest that improved physical access, subsidized the South African program, the Ugandan initiative involves the costs, and culturally appropriate design can sharply increase coordinated efforts of parliamentarians and NGO researchers. girls' enrollments. Already a powerful force in Uganda, the Women's Caucus has The first program, in Quetta, the capital of Balochistan, pushed through several legislative changes, including the uses a subsidy tied to girls' enrollment to support the cre- clause in the local government law requiring that women ation of schools in poor urban neighborhoods by local NGOs. constitute at least a third of executive committee members The schools admit boys as long as they make up less than at the parish and village levels. The gender budget initiative half of total enrollments. In rural Balochistan the second has focused on macroeconomic policy and gender, including program has been expanding the supply of local, single-sex the effects of structural adjustment on poor women. primary schools for girls by encouraging parental involvement In Tanzania another three-year initiative, also started in in establishing the schools and by subsidizing the recruitment 1997, is spearheaded by a coalition of NGOs led by the Tan- of female teachers from the local community. The results: zania Gender Networking Program. It focuses on under- girls' enrollments rose 33 percent in Quetta and 22 percent standing the budget processes of the National Planning in rural areas. Interestingly, both programs appear to have Commission and the Ministry of Finance, how those also expanded boys' enrollments, suggesting that increas- processes affect government spending on basic services, and ing girls' educational opportunities may have spillover ben- how government spending decisions affect women's and efits for boys. men's access to health and education services. The initiative has begun disseminating key findings in simple language to Source: World Bank forthcoming a; Kim, Alderman, and Orazem 1998. make them broadly accessible. Source: World Bank forthcoming a; Budlender 1999; TGNP 1999. Increasing gender equity has enormous benefits in establishing a culture of human rights as well as more im- mediate material benefits through its effects on produc- public support for such efforts (box 7.3). All these in- tivity and the human capital of the next generation. terventions need to be backed by efforts to increase the Paths to gender equity include giving men and women political participation of women, so that they can con- equal rights under the law, equal access to education and tribute more fully to society. health care, and equal access to services related to income Antidiscriminatory legal, institutional, and policy generation. Gender budgeting and publication of gender- reforms for increasing gender equality have both in- disaggregated development indicators can help generate strumental value for development and poverty reduc- REMOVING SOCIAL BARRIERS AND BUILDING SOCIAL INSTITUTIONS I23 tion and intrinsic value for furthering human rights and Box 7.4 well-being. More equitable access to material resources Using development programs to break the and to needed services increases economic productiv- power of agrarian elites: a case study from iry and growth. More generally, increasing gender eq- eastern Uttar Pradesh, India uity is an important component of efforts to encourage greater citizen participation in public life and in mon- The socioeconomic hierarchy in the village is apparent: there greater citizen participation in public life and in mon- is one large white-washed brick mansion standing out among itoring state institutions. a sea of mud huts. The mansion is the home of the talukdar, the large landowner whose duty it was to collect land revenues Social stratification and poverty for the colonial power. The talukdar's family lost some of its holdings when land ceilings were imposed in the 1950s, al- though it held onto much of its land through fictitious division. Because we had no schooling we are almost illiterate. For the next couple of decades the talukdars family con- Sometimes we cannot even speak Spanish; we can't add. solidated its relationships with the new power structures of Store-owners cbeat uts, because we don't know how to the state. In a typical pattern of diversifying family networks, count or aniything else. They buy at the prices they want the father arranged for one son to be in the police service, while another managed the land. They continued to be the and pay less. They cheat us because we are not educated, main source of credit and employment for the villagers, who -1iirka'enous woman in Asocilicion de 10 Ago:to, fcuador acknowledged their social superiority by prefacing every in- teraction with the greeting "Touching your feet, Lord." Economic inequalities reinforced by social barriers make For this family, well educated and well connected, it was easy to divert development funds for its own benefit. The other villagers generally never knew about the entitlements poverty. When social distinctions between groups are they were being deprived of. Even if they did know, they could used to perpetuate inequalities in access to material hardly protest, because the talukdar's family had guns and resources, they generate rigid sociopolitical hierarchies, was known to rape and maim at will. which constitute powerfuil social barriers explicitly aimed Around 1970 agricultural extension agencies brought in- formation on tubewells to the village. Some middle-level at preserving the status of the better-off. They place crip- peasants pooled their resources to sink a tubewell to irrigate pling constraints on individuals. For poor people, naturally their contiguous plots. Eager to maximize their profits, they risk averse because they live close to the margin of sur- began to sow cash crops and raised the wages they offered vival, the prospect of incurring the wrath of powerful elites agricultural laborers. The talukdars son responded by strid- ing around at the weekly market with a gun slung over his by challenging these barriers is intimidating. Rigid strat- shoulder, threatening to shoot anyone who offered laborers ification also creates obstacles to collective action: if the more than the going rate. That temporarily thwarted the distribution of power in a community is too skewed, peasants' efforts. prospects for trust and cooperation are low. But new opportunities offered by tubewells and the opening of a government milk collection center in the village Social inequality in villages undermines efforts to made it more difficult for the talukdars family to retain its po- manage collective goods such as water.40 In the hands of sition. Over time the middle-level peasants increased their village elites, control of these resources can be used to fur- incomes and offered new sources of employment and credit ther discriminate against poor people. One of the most for poor people. The village shifted away from a bipolar polity glaring manifestations of inequality is in access to land. toward a broader distribution of power. A study of another Uttar Pradesh village noted similar ten- In most developing countries large inequalities in land sions. There the talukdars family had tried such methods as ownership make it virtually impossible for poor people arson and election rigging to maintain its power, but the to rise from the bottom of the agrarian hierarchy. But land middle-level peasants' determined use of new agrarian op- reform and broader efforts to diversify economic op- portunities eventually weakened that power. By the 1990s the middle-level peasants had become prosperous and ed- portunities can break some of these barriers and reduce ucated and formed a serious political challenge to the taluk- rural poverty (chapter 5; box 7.4). dar's family, defeating them in local elections. The hegemony In many settings discrimination and social inequality of the old colonial landowning elite has been effectively are the outcome of entire social groups having little po- challenged through continuing development programs and participatory political institutions. litical voice. These groups are discriminated against or ne- glected in the distribution of public goods, which translates Source: Das Gupta, Grandvoinnet, and Romani forthcoming; Dreze, into lower access to education and health-and lower in- Laniouw, and Sharma 1998. come. Most damaging are the poverty traps that arise from 124 WORLD DEVELOPMENT REPORT 2OOO/2001 Figure 7.3 human capital.42 Living in a better-off neighborhood Minority groups in Vietnam have less access exposes individuals to social and cultural factors that in- to services than nonminorities h43 Neghborhood effects can crease their productivity. - egbrod fet a also reduce economic mobility and widen income dis- Distance to facility, 1992-93 parities across communities, as in Ethiopia.44 Similar re- 20 Kilometers suits have been reported in industrial countries, where the rich often live apart from the rest of the population. Other poverty traps result directly from prolonged dis- Minority crimination against minority groups, as in the United 15 ethnic groups *' Majority e States, or even against majority groups, as during the ethnic groups .. apartheid regime in South Africa.45 In these countries, 10 0 as in Latin America, blacks have lower education and income than whites. But their disadvantages run even deeper: their life expectancy at birth is also lower, a gap 5 not explained by socioeconomic disparities alone.46 The cumulative effects of discrimination in education, employment opportunities, and information weaken the Up Ps Hi Dr M aropportunities for members of these groups to find good Upper Post Hospital District Market p by psy- secondary office center jobs.4 This dynamic is powerfullv boosted by the psv- school chological damage from discrimination-and the psy- Source: Adapted from van de Walle and Gunewardena 2000. chological obstacles to upward mobility add to the physical and financial obstacles to obtaining qualifications. People cease to believe in their abilities and stop aspir- ing to join the economic and social mainstream. This so- active discrimination, which can inflict psychological cial dynamic emerges forcefully in the context of race damage on those discriminated against. relations in the United States (box 7.5). Some poverty traps are created in part by geographic Mitigating the impact of social stratification requires isolation. Differential outcomes based on geographic iso- multifaceted approaches. Ensuring that public agencies lation are a form of stratification, even if not consciously and other state institutions serve all sectors of the pop- designed. For example, the disproportionately high ulation equally can make a big difference. This practice poverty among indigenous groups in Latin America can be furthered bg mobilizing excluded groups to be more partly reflects their greater distance than others from assertive of their needs and rights. In situations of active markets, schools, hospitals, and post offices. Similar con- discrimination, carefully designed affirmative action poli- straints are documented for minority ethnic groups in cies can help equalize access to opportunities. Vietnam (figure 7.3). Indigenous groups in Latin Amer- ica also receive less education on average than non- Refor7ming institutions indigenous groups. Ethnic discrimination exacerbates In societies not deeply stratified, reform of state institu- the effects: returns to schooling are lower among in- tions can increase social equity. A fairly simple reform is digenous groups. Indigenous people are more likely than to ensure that delivery of public services does not neglect others to be sick and less likely to seek medical treatment, disadvantaged groups. Broader reforms involve making which may also help account for the difference in earn- legal systems equitable and ensuring that administrative ings.41 This is a vicious circle, as low income reduces the and political institutions are accessible and responsive to probability of improving one's health. all. Rather than create barriers, these systems should fa- Isolation and lack of education can create poverty cilitare the full participation of the entire population. Cit- traps that persist over generations, as children living in izenship laws may also need reform-to reduce social different locations experience different types of human tensions and enable disadvantaged groups to participate capital accumulation. Even the neighborhood in which in political life, which is important to their ability to or- one lives can have a powerful influence on income and ganize on their own behalf48 In some countries, having REMOVING SOCIAL BARRIERS AND BUILDING SOCIAL INSTITUTIONS I2, 5 Box 7.5 crease their access to health, education, and other pub- Discrimination is psychologically devastating lic services, improving their living conditions and rais- ing their incomes. Early results from innovative In an analysis of social exclusion and the need for affirma- "ethno-development" programs in Ecuador show the tive action in the United States, Glenn Loury draws attention importance of cultivating genuine demand, enhancing self- to the psychological havoc that long-standing discrimination can wreak on black ghetto dwellers: management, and building local capacity-instructive lessons for development practitioners and policymakers.49 Here is a youngster to whom one says, "Why don't you marry the girl you got pregnant? Instead of stand- Taking afl7rmative action ing on the street corner hustling, why don't you go It to the community college and learn how to run one Ii s of these machines in the hospital? You could learn that plemented by affirmative action programs-to counter thie with a couple of years at the community college in- disabilities from long-standing discrimination. To com- stead of being a misfit," and the answer is not, "I have pete in economic and political arenas, those discrimi- done my sums and the course you suggest simply nated against need special assistance in acquiring education, does not pay." Instead, his answer s, "Who, me?" natin and seLf-confidnce mative actionb He cannot see himself thus. information, and self-confidence. Affirmative action be- Black ghetto dwellers in the United States are a gins with legislation against discrimination in access to pub- people apart, susceptible to stereotyping, ridiculed lic and private goods and services, such as housing, credit, for their cultural styles, isolated socially, experiencing transport public places and public office. an internalized sense of helplessness and despair, , p p , with limited access to communal networks of mutual Prominent in affirmative action are efforts to reduce assistance. In the face of their despair, violence, and the cumulative disadvantages of low access to education self-destructive behavior, it is morally obtuse and sci- and employment. This typically involves helping mem- entifically naive to argue that if "those people" would bers of discriminated-against groups acquire skills and just get their acts together we would not have such . . t . r a horrific problem. Social processes encourage the de- access to opportunites through fnancial support for velopment of self-destructive behavior. This is not to education, preferential admission to higher education, and say that individuals have no responsibility for the job quotas.50 These policies, of two main types, make a wrong choices they may make. Instead, it is to rec- big difference in outcomes:51 ognize a deep dilemma, one that does not leave us with .Developmental policies seek to enhance the perfor- any good choices. Because the creation of a skilled workforce is a mance of members of disadvantaged groups. Exam- social process, the meritocratic ideal-that in a free ples are financial and other inputs to improve society individuals should be allowed to rise to the educational qualifications, and management assistance level of their competence-should be tempered for those establishing their own business. with an understanding that no one travels that road alone. "Merit" is produced through social processes. * Preferential policies seek to reduce cumulative For this reason, there should be a collective public disadvantages more rapidly by giving members of effort to mitigate the economic marginality of those disadvantaged groups opportunities even when they blacks who languish in the ghettos of America. Pub- lic goals ought not to be formulated in race-neutral may be less qualified than others. ALthough the quick- terms, even if the instruments adopted for the pur- est way to social and economic mobility, these policies suit of those goals are, in themselves, color-blind. can backfire by reinforcing negative stereotypes about the lower abilities of the disadvantaged.52 Even qua.L- Source: Passages excerpted from various sections of Loury 12000). ified members of disadvantaged groups cannot es- cape this shadow. A crucial role for affirmative action policies is to cre- accountable judicial institutions would also help pro- ate role models who can alter the deep-rooted beliels tect disadvantaged groups from discrimination. about different worth and abilities that permeate segre- Poor, marginalized communities can be mobilized to gated societies (box 7.6). Such beliefs, psychologically help reduce their poverty by drawing on and strength- devastating for the disadvantaged, are also shared by those ening their social institutions. Groups with a strong col- who offer jobs and promotions, reducing the likelihood lective identity-and a willingness to collaborate with that they will give equal consideration to minority can- outside agents to forge new solutions-can work to in- didates, even when they have the necessary qualifications. 1z6 WORLD DEVELOPMENT REPORT 2000/2001 1ox a negative effects are largelv associated with preferential Using affirmative action against caste-based policies and can be averted through greater use of discrimination in India developmental policies. In the United States affirmative action has redistributed income to women and minori- The caste system in India separated people into economic and ties, with minimal loss of efficiency.53 Preferential poli- social strata by birth, reinforcing these divisions through dif- ferences in ritual status. This rigid hierarchy remained largely ces may be costly i the long term, Job quotas for in place for many centuries, despite periodic challenges from minorities may distort the allocation of labor, impede ef- social and religious reform movements. But in 1950 the ficiency, and create tensions between the "favored" and newly independent government of India set out to transform the others.54 Preferential policies can also have negative the system. The constitution abolished untouchability in pri- . . vate or public behavior and empowered the government to poltical repercussions. Poltical elites, seekig to bene- take corrective action by reducing the social and educational fit from political clientelism, can manipulate policies disadvantages faced by lower-caste people and introducing aimed at reducing segregation or reserving employment affirmative action in employment. Seats in the national par- for particular groups. Developmental policies, less likely liament and state assemblies were reserved for members of to elicit resentment from other groups, are politically less scheduled (lower) castes and tribes, and an act was adopted making the practice of untouchability a criminal offense. challenging than preferential policies, and have enor- The process of change has been fraught with difficulties. mous potential for reducing the cumulative disadvantages Legal challenges have been mounted against the policies on of longstanding discrimination. grounds also reflected in the public debate-that lower- caste people have no monopoly on poverty and that the Social fragmentation and conflict gains of affirmative action have been cornered by a subgroup of the lower castes. And political resistance arose when the scope of preferential policies was expanded in recent decades Group differentiation by such characteristics as eth- to reserve larger shares of government sector jobs for lower- nicity, race, religion, and language can sometimes result caste people. By contrast, the developmental policies aimed at helping lower-caste people gain access to education for in social fragmentation, with groups perceiving them- upward mobility have been effective and less contentious. selves as having distinct interests even though they may Despite these difficulties, the affirmative action programs have similar socioeconomic status. Ethnicity-a multi- have done much to lower the barriers faced by lower castes. dimensional phenomenon and a controversial notion- Lower-caste people now occupy positions in the highest is based on perceived cultural differences between walks of life, serving as role models for others. Still, a great groups deal remains to be done, as economic and educational in- in a society, differences that form a powerful source of equalities persist. A survey in 1992-93found that 57 percent identity and a base for political mobilization.55 Some of heads of household were illiterate in scheduled castes, scholars have treated ethnicity as a form of capital-a compared with 35 percent in other castes. And special ef- forts are needed in the few remaining regions where the po- resource or asset on whch members of a partcular eth- lice are still dominated by upper-caste interests. Nevertheless, nic community call in their business and political deal- the experience of affirmative action in India illustrates how, ings.56 Common ethnic affiliations can be a basis for wlth political will, the effects of long-standing patterns of dis- bonding social capital (see next section), providing crimination can be overcome, community members with a range of benefits (credit, Source: Deshpande 2000; Dushkin 1972; Galanter 1972; Srinivas employment, marital partners) while imposing signif- 1987; Tumma a 1999. icant obligations and commitments (financial support, conformity). Membership in an ethnic community can also generate negative externalities, as with conflict be- Affirmative action seeks to alter these perceptions of tween ethnic groups (box 7.7).7 Such divisions can be different worth by bringing some members of obstacles to collective action: in the United States greater discriminated-against groups into the mainstream econ- ethnic fragmentation is associated with lower partici- omy and society. This has an important demonstration ef- pation in civic activities. 58 fect: having black or low-caste doctors, for example, shows Ethnicity can become a basis for competition for po- everyone, including their own group, that members of this litical power and for access to material resources.59 Un- group can be good doctors. less institutions of the state and civil societv offer forums Do affirmative action programs reduce efficiency or for mediating intergroup rivalries and forging cross- engender political strife? Evidence shows that these cutting ties among diverse ethnic groups, these ethnic REMOVING SOCIAL BARRIERS AND BUILDING SOCIAL INSTITUTIONS 127 cleavages can lead to conflicts, tearing a society and econ- unrest and poor performance in the sector.61 Such dis- omy apart, leaving everyone vulnerable to poverty. tortions in the distribution of resources and the effi- The extent to which social fragmentation leads to ciency of their use show up in development outcomes. conflict depends largely on administrative and political In several African countries, for example, child survival institutions. To create a functioning society, a whole is higher in dominant ethnic groups.62 range of social and political institutions must work to- gether. By contrast, breakdowns in governance and in the Buildi7ng political allianices delivery of public goods and related social services cre- Countries with high ethnic diversity need to build the po- ate conditions for social unrest and conflict-as do break- litical conditions for integrating diverse groups so that they downs in the institutions of conflict mediation, such as can function collectively.63 With well-functioning ad- representative politics and the rule of law. ministrative and political institutions, multiethnic soci- Ethnic cleavages can affect development outcomes in eties can be effectively shaped into an "imagined many ways. They can influence the internal organization community" of nation and state.64 Knitting diverse com- of government and the allocation of public spending, lead- munities together through a multiplicity of civil and staite ing to unequal distribution of public goods and services. channels-to avert conflict-was a major goal of the They can encourage rent seeking, reducing the efficiency early designers of European unity.65 The communist of public spending.60 Further economic distortions enter regimes of the Soviet Union and Yugoslavia, despite their when powerful ethnic groups use their political power to economic and political failures, not only reduced economic increase their incomes relative to those of others. Recent inequalities but also managed ethnic conflict. With their studies in Ghana show that locally dominant groups re- collapse, violent ethnic conflicts broke out because no al- ceive a 25 percent premium over the wages of other ternative ideological and institutional framework had groups in the public sector-a discrepancy that leads to evolved to mediate them. In Sub-Saharan Africa nation-states were fashioned Box 7.7 out of arbitrary divisions of territory by colonial powers-- Ethnic divisions and civil conflict divisions often based on convenient geographic markers such as lines of latitude and longitude, with no consid- Ethnic fragmentation, in its most extreme form and under con- eration of the social units of local populations. With dis- ditions of economic deprivation and nondemocratic govern- parate groups and few supraethnic institutions to mediate ment, can descend into civil conflict. Ethnic conflict intensified I v v c in the second half of the 20th century, as the pattern of con- among them, the creaton of nation and state has been flict shifted from wars between nations to conflicts within fraught with problems. Colonial rulers and local politri- states. Civil conflict is both a cause and a consequence of poor cians have often manipulated ethnic tensions for private economic performance. Research has shown that during civil wars gain, sometimes leading to gruesome civil wars.66 In- per capita output falls by more than 2 percent a year on average, flaming ethnic tensions and civil unrest is a frequent strat- The most important cost of civil conflict is loss of life-a humanitarian tragedy and an obstacle to reconstruction. Other egy for gaining and keeping power in these circumstances, costs include destruction of physical, human, and social cap- since it justifies expanding brutal military forces while un- ital; lower investment in physical and human capital; disrup- dermining the capacity of opposition groups demanding tion of markets and other forms of economic and social order; reform. Over time ethnic minorities, especially those fac- diversion of human resources and public expenditure away from productive or productivity-enhancing activities; migration of ing discrimination, inequality, or conflict, can become highly skilled workers; and transfers of financial assets abroad. ethno-classes,67 groups whose ethnicity-based sensibilities These costs can trap countries in poverty-and in conflict. and demands become independent causes of conflict.68 Civil conflict can also accelerate the collapse of the state, disproportionately hurting poor people. And the problems of civil conflict spill across borders, increasing the burdens of neighboring countries. In 1998 there were an estimated 12.4 Constructing high-quality public institutions is essential million international refugees and 18 million internationally dis- for ensuring that diverse identities become a develop- placed people, almost half of them in Africa. mental asset, not a source of political division and violence Source: Collier and Hoeffler 1998; Austin 1999; Stewart, Humphreys, (figure 7.4) 69 This is especially important in countries witi and Lea 1997; Collier 1999c, Luckham 1999. abundant natural resources, such as oil, diamonds, and minerals.70 In environments with little institutional 128 WORLD DEVELOPMENT REPORT 2000/2001 Figure 7.4 Ethnic diversity is associated with violence In addition to removing social barriers, effective efforts where institutional quality is low to reduce poverty require complementary initiatives to build up and extend the social institutions of the poor. Probability of state-sanctioned ethnic killings, 1960-90 Social institutions refer to the kinship systems, local or- ,r/ ---- -- -H-igh-~ - -~ ~ ~ ~~ ~ ~ -~~ ~ 5 ganizations, and netwvorks of the poor and can be usefully 0.5 / ', fragmentation discussed as different forms or dimensions of social I - capital. 0.4 Bondin?g, bridging, and linikinig social capital Low Distinguishing among different dimensions of social cap- 0.3 ethnolinguistic .dfeetosc fragmentation ital within and between communities is useful for un- 0.2 derstanding the problems faced by poor people (box 7.8). 0 The strong ties connecting family members, neighbors, close friends, and business associates can be called 0.1 ?-l l ____ _ _ _ ________ _ __ bonding social capital. These ties connect people who share similar demographic characteristics. 0 Low High * The weak ties connecting individuals from different Institutional quality ethnic and occupational backgrounds can be referred Source: Easterly 2000a. to as bridging social capita /i2 Bridging social capital implies horizontal connections to people with broadly comparable economic status and political power. A accountability and transparency, the exorbitant rents from theory of social capital that focuses only on rela- these resources become a primary source of competition tions within and between communities, however, among ruling factions. opens itself to the criticism that it ignores power.73 Civil society organizations and the state can do much * A third dimension, linking social capital, consists of the to lay the institutional foundation for groups to cooper- vertical ties between poor people and people in posi- ate for the common good. Institutions need to be par- tions of influence in formal organizations (banks, ticipatory, credible, and accountable, so that people can agricultural extension offices, the police).74 This di- see the benefits of cooperation. Underpinning these in- mension captures a vitally important additional fea- stitutions need to be constitutional and legal systems ture of life in poor communities: that their members and representative political systems, which allow groups are usually excluded-by overt discrimination or lack to work out their interests through mechanisms other than of resources-from the places where major decisions violence. Some social integration can be achieved by en- relating to their welfare are made. couraging people to learn each other's languages. Another Research on the roles of different types of social net- important requirement for effectively helping excluded works in poor communities confirms their importance. groups is to collect accurate data on them.7t An analysis of poor villages in rural North India, for example, shows that social groups play an important role Bujidding social institutions and in protecting the basic needs of poor people and mediating social capital against risk (chapter 8). In contrast, the more extensive, leveraged networks of the nonpoor are used for strategic d b?: Ul, e? r e is ti ,ineral. we work togetber . womlen advantage, such as procuring better jobs and higher wages (it a wari- iw . ('cllecf tn ewiood, and collect maizeflour firo77z and seizing new economic opportunities (in some cases while the men dig thpe grae rUld bllr y the by directly mobilizing to secure a disproportionate share dei/> 1, . t uork together on colmmunity projects like of public resources and services).75 ?nottu jcs' a 1ks for a school . . .'onomen also work Strikingly similar results emerge from work on the ?jhtr/rn cleaning around the boreboles. relationship between enterprise performance and the structure of business networks in Africa. Poor entre- REMOVING SOCIAL BARRIERS AND BUILDING SOCIAL INSTITUTIONS IZ9 preneurs operating small local firms in traditional in- household social capital have found a positive associa- dustries form "solidarity networks," sharing personal in- tion with household consumption, asset accumulation, formation about members' conduct and intentions in and access to credit.78 order to reduce risk and uncertainty. In contrast, larger Researchers and practitioners have long recognized regional firms form "innovation networks," which share that the bonding and bridging social capital in local or- knowledge about technology and global markets in ganizations is necessary but insufficient for long-term order to increase productivity, profits, and market development. In Kenya a participatory poverty assess- share.76 Studies of agricultural traders in Madagascar ment found more than 200,000 community groups in show that social relationships are more important to rural areas, but most were unconnected to outside re- traders than input prices. Close relationships with other sources and unable to help poor people rise out of traders are used to lower the transactions costs of ex- poverty.79 The creation of linking social capital is es- change, while ties to creditors and others who can help sential, and external support has often been important out during times of financial hardship are vital sources in its emergence.80 External support-from NGOs anid of security and insurance.77 In Bolivia, Burkina Faso, religious organizations, for example-can help create so- and Indonesia field surveys attempting to measure cial capital that increases the voice and economic op- Box 7.8 How does social capital affect development? There are at least four views on the relationship between social complement one another. Macro institutions can provide an capital and development (Serageldin and Grootaert 2000; Wool- enabling environment in which micro institutions develop and cock and Narayan 2000). The narrowest holds social capital to be flourish. In turn, local associations help sustain regional and na- the social skills of individuals-one's propensity for cooperative tional institutions by giving them a measure of stability and behavior, conflict resolution, tolerance, and the like (Glaeser, Laib- legitimacy-and by holding them accountable for their actions son, and Sacerdote 2000). (Evans 1996; Woolcock 1998; Narayan 1999; Serageldin and A more expansive meso view associates social capital with fam- Grootaert 2000; Putnam 1993). ilies and local community associations and the underlying norms While the mechanisms by which social capital operate are (trust, reciprocity) that facilitate coordination and cooperation for generally well understood, there is less consensus on whether they mutual benefit. This view highlights the positive aspects of social qualify social capital as "capital." In many cases norms and insti- capital for members of these associations but remains largely tutions have the durability and lasting effects associated with silent on the possibility that social capital may not impart benefits capital (Collier 1998; Narayan and Pritchett 1999). Some argue, how- to society at large and that group membership itself may entail sig- ever, that the sacrifice of a present for a future benefit, typical of nificant costs. traditional forms of capital, is not present in social networks-to A more nuanced meso view of social capital recognizes that the extent that these networks are built for reasons other than their group membership can have both positive and negative effects (Cole- economic value to participants (Arrow 2000). Even so, social net- man 1990; Burt 1992; Portes 1995; Massey and Espinoza 1997). works and organizations are clearly key assets in the portfolio of This approach broadens the concept of social capital to include as- resources drawn on by poor people to manage risk and opportu- sociations in which relationships among members may be hierar- nity. They are also key assets for the rich, who advance their in- chical and power sharing unequal. These forms of associations and terests through such organizations as country clubs and professional networks address a wider range of objectives: some of them serve associations, but their relative importance is greater for poor only the private interests of members, while others are motivated people. by a commitment to serve broader public objectives. This view em- Social capital has its dark side, however. Where groups or net- phasizes that groups, in addition to providing benefits to members, works are isolated, parochial, or working at cross-purposes to so- can make significant noneconomic claims on them. ciety's collective interests (gangs, drug cartels), the social capital A macro view of social capital focuses on the social and polit- within them serves perverse rather than productive purposes, ical environment that shapes social structures and enables norms undermining development (Rubio 1997; Levy 1996; Portes and Lan- to develop. This environment includes formalized institutional re- dolt 1996). Organized crime syndicates, such as those in Latin Amer- lationships and structures, such as government, political regime, ica and Russia, generate large negative externalities for the rest rule of law, the court system, and civil and political liberties. Insti- of society-lost lives, wasted resources, pervasive uncertainty tutions have an important effect on the rate and pattern of economic (Rose 1999). And in India, for example, obligations to family mem- development (North 1990; Fukuyama 1995; Olson 1982). bers and pressures to fulfill community expectations lead many An integrating view of social capital recognizes that micro, young girls to drop out of school (Dreze and Sen 1995; PROBE meso, and macro institutions coexist and have the potential to Team 19991. 130 WORLD DEVELOPMENT REPORT 2000/2OOI Box 7.9 U,sing social capital to improve program The federation of comedores in Peru: effectiveness the creation of linking social capital The state plays a vital role in shaping the context and cli- mate in which civil society organizations operate (chap- The comedores (community kitchens) movement, one of the ter 6).82 In some cases the state can also create social capital. most dynamic women's groups in Peru, emerged in the mid-1980s. Participants sought to move beyond their tradi- In 1987 the Department of Health of the state of Ceara, tional survival strategy and make demands on the political Brazil, launched a rural health program-since then ex- system. Federations were formed at the neighborhood level, panded to most of the country-that increased vaccina- then at the district level, and finally at the metropolitan and tion rates significantly and reduced infant mortality. The national levels. Centralization of the movement lowered the cost of inputs, such as food and kitchen equipment, and in- success of the program has been attributed largely to the creased the availability of educational workshops. building of trust between government workers and poor The highest-level organization, the CNC (National Com- people. The program made building trust an explicit mission of Comedores), became the officially recognized part of the health workers' mandate by adopting a client- representative of the comedores, One of its central de- centered, problem-solving approach to service delivery. mands was to include all poor women in welfare programs, not just those with connections to the ruling party. Besides Workers were helped by government media campaigns influencing policymaking, the comedores have had a signif- that publicized the program regularly and gave them a icant impact on local power relations in the shantytowns and, sense of calling. The result was a total reversal of attitude: by extension, on the structure of the political system. mothers who once hid their children from government Although the movement's actions have been limited by 8 the structure of the Peruvian state (with few formal channels workers saw the agents as true friends of the community.83 for political action), the network of comedores represents a Many case studies show that social capital can improve form of social capital that has enhanced poor women's value project design and sustainability (box 7. 10). Recent eval- as an electoral constituency. The comedores have also in- uations ofWorld Bank rural development projects show that c reased women's negotiating power In their families. outcomes turn heavily on the nature of the power relations Source: Houtzager and Pattenden 1999. between key stakeholder groups and on the fit between ex- ternal interventions and local capacities. How relations be- tween stakeholders evolve over time has an important portunities of poor people (box 7.9). This support is most bearing on the generation of trust. Project and community effective when it is sustained over time, emphasizes ca- leaders who create confidence and goodwill are crucial, pacity building, and is based on a sensitive under- suggesting that high turnover among field staff can un- standing of the local conditions and a relationship of trust dermine project effectiveness. The Gal Oya irrigation pro- and partnership. ject in Sri Lanka has succeeded in a destitute region with This approach characterizes the work of Myrada, an high levels of ethnic violence because of the patience and Indian NGO delivering microfinance services. Myrada long-standing commitment of field staff (aptly called in- acts as a medium-term intermediary between poor people stitutional organizers). The project's key contribution has and commercial banks.81 Its initial task is to mobilize the been integrating local knowledge with external expertise and bonding social capital within village communities to forging cooperation between NGOs and government of- form credit management groups and then over time to ficials.84 In Africa recent innovations in community-driven form regional federations made up of representatives development programs have shifted responsibility for main- from each credit group (thereby enhancing each group's taining hand pumps and latrines directly to communi- bridging social capital). From the outset credit man- ties.85 Where previously such items broke down quickly and agement groups hold accounts with commercial banks, took months to repair, they are now in good condition. progressively gaining the confidence and skills they need A key lesson for practitioners and policymakers is the to participate independently in formal institutions (link- importance of using existing forms of bridging social ing social capital). After five years of training and hard- capital in poor communities as a basis for scaling up the won experience, group members are able to manage efforts of local community-based organizations.86 Cre- these accounts-and even arrange for annual external ating more accessible formal institutions helps poor audits-without the involvement of Myrada staff, who people articulate their interests to those in power more move on to start the process afresh. clearly, confidently, and persuasively. REMOVING SOCIAL BARRIERS AND BUILDING SOCIAL INSTITUTIONS 131 Box 7.10 Mobilizing and creating social capital in development projects Development programs have relied on local groups of project ben- projects. In Cote d'lvoire rural water supply improved significantly when eficiaries or local associations to improve the success of develop- responsibility for maintenance was shifted from the national water ment projects for more than two decades.' What is new is the distribution companyto communitywater groups. Breakdown rates umbrella label social capital to refer to the underlying social force were reduced from 50 percent to 11 percent, while costs fell nearly or energy. 70 percent. These results were sustained, however, only in villages In Bangladesh Grameen Bank relies on groups of poor women in which well-functioning community organizations existed and de- to implement programs, and the Bangladesh Rural Advancement mand for water was high (Hino 1993). Committee on groups of village workers with little or no land. In Pak- In many cases, challenging existing norms and practices in- istan the Aga Khan Rural Support Program gives assistance to vil- creases the social capital of previously excluded groups while de- lage organizations to supplement their self-help efforts. The Kenya creasing the power of local elites, helping reduce obstacles to Tea Development Authority worked with grower committees to poverty reduction. Development programs such as women's mi- promote production, obtaining a one-third share of the country's tea crofinance in Bangladesh change the social relationships in a exportswithin 15years. The6-S movementin nineWestAfricancoun- village-indeed, their success depends on it. Breaking the grip of tries organized peasant federations in more than 2,000 communities moneylenders, overcoming the resistance of certain religious to help farmers overcome the hardships of the dry season. The leaders, and giving women more decisionmaking power within their Center for Social and Economic Development in Bolivia has supported household all require a fundamental realignment of traditional so- more than 250 peasant organizations that promote programs in agri- cial relationships. Many development programs are inherently po- culture, livestock, forestry, artisan production, and community in- litical (Fox and Gershman 1999), and powerful vested interests can frastructure lUphoff 1993; Krishna, Uphoff, and Esman 1997). be expected to mobilize against reforms that seek to erode their Local groups have also been used frequently in irrigation, water position in the name of poor people. Development researchers, supply, and sanitation programs. The Orangi Pilot Project in Pakistan policymakers, and practitioners must recognize these tensions and provided low-cost self-help sewerage facilities and other services to respond appropriately. poor settlements and helped autonomous local institutions implement 1. Among the first systematic evaluations of community participation was Esman and Uphoff (1984). practices and create visible role models for others to fol- low. Where there is considerable ethnic heterogeneity Many aspects of social norms and practices help gen- and social fragmentation, conflict can be averted through erate and perpetuate poverty. Discriminatory practices as- efforts to increase the civic interaction of different groups sociated with gender, ethnicity, race, religion, or social and engage them in resolving potential conflicts through status result in the social, political, and economic exclu- political processes. Gender-based discrimination is qual- sion of people. This creates barriers to upward mobility, itatively different from these other forms of discrimina- constraining people's ability to participate in economic tion because it involves intrahousehold distinctions in opportunities and to benefit from and contribute to eco- assigning value to people and allocating resources ac- nomic growth. It also constrains their effective partici- cordingly. Reducing gender-based social barriers requires pation in political processes and civil action to ensure that changing deep-rooted beliefs about appropriate gender state institutions are accountable to citizens and respon- roles, as well as taking action to ensure greater gender eq- sive to their needs. uity in the functioning of formal public institutions. Policies and programs for mitigating social exclusion Increasing the participation of the poor in develop- depend on the nature of the exclusion. In some cases ex- ment and reducing social barriers are important com- clusion can be addressed simply by improving the outreach plements to creating an environment in which they have of public services to neglected areas. Where more active greater opportunity and security. This empowerment is discrimination is involved, it is important to ensure eq- enhanced by scaling up social institutions, increasing the uity in the law and in the functioning of state institutions. capacity of poor people and the socially disadvantaged In addition, affirmative action policies may be needed to to engage society's power structure and articulate their reduce the cumulative disadvantages of discriminatory interests and aspirations. PART IV 0 Securit f~~~~ CHAPTER 8 Helping Poor People Manage Risk To be well is to know what will happen with me ity lessened if mechanisms to reduce, mitigate, and cope tomorrow. with risks are available to them. -,'1id/lc-agH 7ma, Razzad, Bulgaria Poor people have developed elaborate mechanisms for dealing with risk. But the mechanisms are far from capable of eliminating vulnerability. Many of the mechanisms offer short-term protection at long-term J overty means more than inadequate consump- cost, preventing any escape from poverty. tion, education, and health. As the voices of the poor The policy response to vulnerability must be aimed cry out, it also means dreading the future-knowing at helping poor people manage risk better by reduc- that a crisis may descend at any time, not knowing ing and mitigating risk and lessening the impact of whether one will cope. Living with such risk is part shocks. Such policies address the immediate problems of life for poor people, and today's changes in trade, of shocks and the inability to cope with them. But they technology, and climate may well be increasing the risk- also lay the foundations for investment by poor peo- iness of everyday life. Poor people are often among the ple that can take them out of poverty. This report ad- most vulnerable in society because they are the most vocates a modular approach to risk management that exposed to a wide array of risks. Their low income adapts safety nets to the specific pattern of risk in means they are less able to save and accumulate assets. each country and complements existing risk manage- That in turn restricts their ability to deal with a cri- ment arrangements. This chapter briefly reviews ex- sis when it strikes. perience with seven tools especially relevant for poor Economic growth is one way of reducing the vul- people: health insurance, old age assistance and pen- nerability of poor people. As their incomes rise, they sions, unemployment insurance and assistance, work- are better able to manage risks. However, at any point fare programs, social funds, microfinance programs, and in time those who are poor will see their vulnerabil- cash transfers. 135 136 WORLD DEVELOPMENT REPORT 2000/200I Table 8.1 Main sources of risk Idiosyncratic Covariant Risks affecting an Risks affecting Risks affecting individual or groups of households regions or Type of risk household (micro) or communities Imeso) nations (macro) Natural Rainfall Earthquake Landslide Flood Volcanic eruption Drought High winds Health Illness Epidemic Injury Disability Old age Death Social Crime Terrorism Civil strife Domestic violence Gang activity War Social upheaval Economic Unemployment Changes in food prices Resettlement Growth collapse Harvest failure Hyperinflation Balance of payments, financial, or currency crisis Technology shock Terms of trade shock Transition costs of economic reforms Political Riots Political default on social programs Coup d'etat Environmental Pollution Deforestation Nuclear disaster Source: Adapted from Sinha and Lipton 11999) and World Bank (2000q). A typology of risks This distinction by level of risk is critical. A risk that affects an entire village, for example, cannot be insured One way to understand risks better and design appro- solely within the village. It requires pooling with areas not priate policy responses is through a typology of risks and subject to the risk. In practice, many shocks have both shocks to which people are vulnerable (table 8. 1). Risks idiosyncratic and covariant parts, though most empiri- can be classified by the level at which they occur (micro, cal studies find that the idiosyncratic part of income risk meso, and macro) and by the nature of the event (nat- is large.' This chapter focuses on risks that usually have ural, economic, political, and so on). Micro shocks, large idiosyncratic components: illness and injury, old age, often referred to as idiosyncratic, affect specific indi- violence, harvest failure, unemployment, and food price viduals or households. Meso shocks strike groups of risk (box 8. 1). Covariant risks are discussed in chapter 3 households or an entire community or village. These (box 3.2) and chapter 7 (war and civil strife) and chap- shocks are common (or covariant) to all households in ter 9 (macroeconomic shocks and natural disasters). the group. Shocks can also occur at the national or in- The extent to which a risk is covariant or idiosyncratic ternational level. depends considerably on the underlying causes. For HELPING POOR PEOPLE MANAGE RISK 137 Box 8.1 Poor people's exposure to risk Poor people are exposed to a wide range of risks. der rate for adolescent males in poor neighborhoods was 11 times that in wealthier ones (Sinha and Lipton 1999). Poor peo- Illness and injury ple frequently voice their fear of violence and the resulting pow- Poor people often live and work in environments that expose erlessness, "I do not know whom to trust, the police or the them to greater risk of illness or injury, and they have less access criminals." to health care (Prasad, Belli, and Das Gupta 19991. Their health risks Crime also hurts poor people indirectly. Children exposed to are strongly connected to the availability of food, which is af- violence may perform worse in school (Morrison and Orlando fected by almost all the risks the poor face (natural disasters, 1999). A study of urban communities in Ecuador, Hungary, the wars, harvest failures, and food price fluctuations; de Waal 1991). Philippines, and Zambia showed that difficult economic conditions Communicable diseases are concentrated among the poor, with lead to destruction of social capital as involvement in community respiratory infections the leading cause of death (Gwatkin, Guil- organizations declines, informal ties among residents weaken, and lot, and Heuveline 2000). A recent study of poverty in India found gang violence, vandalism, and crime increase (Moser 1998). Vi- that the poor are 4.5 times as likely to contract tuberculosis as the olence and crime may thus deprive poor people of two of their rich and twice as likely to lose a child before the age of two best means of reducing vulnerability: human and social capital. (World Bank 1998t). Rich and poor women alike are victims of domestic violence, Illness and injury in the household have both direct costs (for but the incidence is often higher in poor households. In Santiago, prevention, care, and cure) and opportunity costs (lost income or Chile, 46 percent of poor women and 29 percent of wealthy schooling while ill; Sinha and Lipton 1999). The timing, duration, women suffer from domestic violence; in Managua, Nicaragua, 54 and frequency of illness also affect its impact. A study of South percent and 45 percent (Morrison and Orlando 1999). India found that households can compensate for an illness during the slack agricultural season, but illness during the peak season Unemployment and other labor market risks leads to a heavy loss of income, especially on small farms, usu- Labor market risks include unemployment, falling wages, and ally necessitating costly informal borrowing (Kochar 1995). having to take up precarious and low-quality jobs in the informal sector as a result of macroeconomic crises or policy reform. The Old age first workers to be laid off during cutbacks in public sector jobs Many risks are associated with aging: illness, social isolation, in- are usually those with low skills, who then join the ranks of the ability to continue working, and uncertainty about whether trans- urban poor, a pattern observed in Africa and Latin America during fers will provide an adequate living. The incidence of poverty the structural adjustment reforms of the 1980s and early 1990s among the elderly varies significantly. In most Latin American (ECLAC 1991; Sinha and Lipton 1999). The East Asian crisis also countries the proportion of people in poverty is lower for the el- had pronounced effects on labor markets, with real wages and non- derly than for the population at large (IDB 2000). In contrast, in many agricultural employment falling in all affected countries (World countries of the former Soviet Union the incidence of poverty is Bank 1999j). As state enterprises in Eastern Europe and the coun- above average among the elderly, particularly among people 75 tries of the former Soviet Union were privatized, poverty increased and older (Grootaert and Braithwaite 1998; World Bank 20001). among displaced workers with low education and obsolete skills, Women, because of their longer life expectancy, constitute the not qualified to work in emerging industries. Wage arrears in Rus- majority of the elderly, and they tend to be more prone to poverty sia intensified the problem (Grootaert and Braithwaite 1998). in old age than men (World Bank forthcoming a). The number of Fluctuations in demand for labor often disproportionately af- elderly people in the developing world will increase significantly fect women and young workers. Most public sector retrench- in coming decades with the rapid demographic transition. ment programs have affected women's employment more than Consultations with poor people show that income security is men's (World Bank forthcoming a), and women are more likely than a prime concern of the elderly, followed closely by access to men to work for small firms, which tend to be more sensitive to health services, suitable housing, and the quality of family and com- demand fluctuations (Horton and Mazumdar 1999). As incomes munity life. Isolation, loneliness, and fear all too often mark old peo- fall, poor households try to increase their labor market participa- ple's lives (Narayan and others 1999). As an elderly woman in tion, especially for women and children. This response has been Ukraine put it, " If I lay down and died, it wouldn't matter, because documented in many countries (Horton and Mazumdar 1999; nobody needs me. The feeling of being unnecessary, of being Grootaert and Patrinos 1999). unprotected, is, for me, the worst of all." Harvest failure and food price fluctuations Crime and domestic violence Weather-related uncertainties (mainly rainfall), plant disease, and Crime and domestic violence reduce earnings and make it harder pests create harvest risk for all farmers, but technologies for re- to escape poverty. While the rich can hire private security guards ducing such risks (irrigation, pesticides, disease-resistant and fortify their homes, the poor have few means to protect themselves against crime, In Sao Paulo, Brazil, in 1992 the mur- (box continues on next page) 138 WORLD DEVELOPMENT REPORT 2000/2001 Box 8.1 Poor people's exposure to risk (continued) varieties) are less available in poor areas. In 1994-96 less than 20 over the year through arbitrage, but such infrastructure is lack- percent of all cropland was irrigated in low- and middle-income ing in many areas. In Madagascar the mean price of rice, the main countries (only 4 percent in Sub-Saharan Africa). staple, rose 42 percent and the variance increased 52 percent Fluctuations in food prices are a related risk. Since poor house- after the price liberalization of the 1980s. Two-thirds of rice holds spend a large part of their income on food, even small price farmers were hurt because they consumed more rice than they increases can severely affect food intake. Households that meet produced, and poverty deepened (Barrett 1996, 1 998a). their food needs through subsistence agriculture are less vulner- For the rural poor, crop diversification and income diversifica- able than households that have to buy all their food. tion into nonfarm activities hold the greatest promise for reduc- Liberalization of markets often boosts the price of staples- ing food price and harvest risks. Reducing consumption as food a benefit to small farmers if they are net sellers of food. Hurt are prices rise can have major and lasting adverse health effects, es- the urban poor and the landless rural poor, as net food buyers, pecially for children. Successive harvest failures because of in- and farmers who engage in seasonal switching, selling food sufficient monsoons in Sri Lanka in 1995 and 1996 led to increased after the harvest when food is plentiful and cheap and buying it indebtedness in 80 percent of households in eight villages, and when it is scarce and expensive (Sinha and Lipton 1999). Where 30 percent of households reported increased incidence of illness transport facilities are good, traders can step in and equalize prices (Sinha and Lipton 1999). example, job loss can be an individual risk, or it can be The nature and magnitude common to most workers in a country if it is the result of vulnerability of a macroeconomic crisis. The risk of becoming ill can be idiosyncratic, or it can have a large common compo- Vulnerability affects everyone (box 8.3). Even well-paid civil nent if there is an epidemic. The HIV/AIDS pandemic servants are vulnerable to losing their jobs and sliding into is a health risk at the global level, with devastating effects poverty. For the poor, and for people just above the poverty on poor people and poor countries (box 8.2). line, vulnerability is a graver concern because any drop in Knowing the source of shocks is important for pre- income can push them into destitution. As a result, poor venting them, but identify-ing the source is not always people are highly risk averse and reluctant to engage in the straightforwvard. Many exogenous events can have simi- high-risk, high-return activities that could lift them out of lar effects on household income. A macroeconomic shock, poverty. One slip could send them deeper into poverty. a hurricane, or a civil war can all lead to severe decline Large fluctuations in income are common for poor peo- in income and deplete a household's assets. But how a ple.3 For South Indian villages estimates of the coefficient shock is transmitted to households is greatly affected by of variation of annual income from the main crops range a countrys institutions. Not every drought causes famine, between 0.37 and 1.1 and are as high as 1.27 for total illness, and death. The effect of a disaster depends on how farm profits.5 In rural Ethiopia three of four households well the government functions, whether there is peace or suffered a harvest failure over a 20-year period, resulting civil strife, how well the safety net and other institutions in significant fluctuations in farm income.6 include the poor, and so on. Furthermore, because poor people have fewer assets The rvpology can be refined by distinguishing the and less diversified sources of income, these fluctua- severity and frequency of shocks. Consumption smooth- tions affect them more than other groups. In South In- ing is more difficult with repeated shocks, because house- dian villages an increase in risk (from the monsoon holds may have depleted their assets in coping with the arriving too soon or too late) reduced farm profits for initial shock, leaving them unable to absorb subsequent the poorest quarter of households by 35 percent but left shocks.2 And one shock might lead to another. A natural the wealthiest farmers nearly unaffectedi. In Vietnam par- disaster could wipe out poor people's food supply, leav- ticipants in the Voices of the Poor study said of harvest ing them weak and susceptible to illness. Severity can range losses due to floods: from catastrophic (a natural disaster, death of the bread- winner) to minor (a slight illness, a few days without work The wealthy can recover losses in one year, but poor for casual laborers). people, who have no money, will never recover. HELPING POOR PEOPLE MANAGE RISK 139 Box 8.2 AIDS and poverty More than 34 million people worldwide are infected with HIV, and With the more educated responding to the information available more than 18 million people have died of AIDS. More than 90 per- on AIDS and adopting protective sexual practices (condoms), the cent of people infected with HIV/AIDS are in the developing world. share of new infections is rising among low-income and less ed- Cross-country evidence indicates that both low income and unequal ucated people. distribution of income are strongly associated with HIV infection With 5 million people becoming infected annually, urgent ac- rates. Countries with high gender inequality also have higher in- tion is needed to stop the spread of HIV/AIDS. Successful inter- fection rates. Sub-Saharan Africa has more cases of existing and vention programs require strong government commitment and new infections than the rest of the world combined, though the partnerships with the private sector, NGOs, and community lead- rate of increase is now steepest in Asia and in the countries of the ers. Interventions shown to be effective include conducting pub- former Soviet Union. lic information campaigns to change individual behavior and social All 20 countries with the highest HIV prevalence are in Sub- norms for sexual contact; making condoms more available and af- Saharan Africa. In Botswana and Zimbabwe 1 in 4 adults is in- fordable; providing voluntary counseling, testing, and treatment of fected. In 10 other African countries more than 1 in 10 adults sexually transmitted diseases; ensuring a safe blood supply; and are infected. The effect on life expectancy will be devastating. taking measures to reduce mother-to-child transmission. In addi- Had AIDS not affected these countries, life expectancy would tion, care activities need to be scaled up to support the vast num- have reached 64 years by 2010-15. Instead, it will have re- bers of people infected and affected. gressed to 47 years, reversing the gains of the past 30 years. AIDS has a devastating impact on poor people. During the ill- The impact on child mortality is also enormous. In Zambia and ness it leads to loss of labor and causes poor households to dis- Zimbabwe 25 percent more infants are dying than would have pose of productive assets to pay for treatment. The impact of an without HIV. adult death from AIDS is more severe in poor households. The Despite the strong correlation at the country level between recommended policy approach is to concentrate on poor house- poverty and AIDS, the evidence for individuals does not suggest holds most in need of survivor assistance, focusing on the period that poor people are most likely to be infected. Indeed, early on, immediately after a death, when food consumption has fallen but the disease struck mainly the better-off groups. Evidence for the there has not yet been a permanently damaging impact. 1 980s and the first half of the I 990s indicates a positive corre- The view that HIV/AIDS is a central development issue is em- lation between HIV infection and education, income, and socio- bodied in the International Partnership against HIV/AIDS in Africa, economic status, probably because wealthier and better-educated launched in 1999 by the cosponsors of the Joint United Nations people were more likely to have multiple sexual partners. Non- Programme on HIV/AIDS (UNAIDS), including the World Bank. sexual modes of transmission-intravenous drug use and mother- In collaboration with African governments, the program aims to to-child transmission-are associated more with poverty, In increase resources and technical support, establish targeted recent years the profile of HIV-infected people has been chang- prevention and treatment efforts, and expand the knowledge base ing rapidly, and AIDS is becoming a disease of poor people. to assist countries. Source: Ainsworth and Semali 1998; Basu 1995; Over 1998; Rugalema 1999; UNAIDS 2000; World Bank 1997d, 1999m . Box 8.3 Some key terms: risk, risk exposure, and vulnerability As traditionally defined and measured, poverty is a static concept- being-the risk of becoming ill, or the risk that a drought will a snapshot in time. But insecurity and vulnerability are dynamic- occur, The uncertainty can pertain to the timing or the magnitude they describe the response to changes over time. Insecurity is of the event. For example, the seasonal fluctuation of farm income exposure to risk; vulnerability, the resulting possibility of a decline is an event known in advance, but the severity is not always pre- in well-being. The event triggering the decline is often referred to dictable. Risk exposure measures the probability that a certain risk as a shock, which can affect an individual (illness, death), a com- will occur. Vulnerability measures the resilience against a shock- munity, a region, or even a nation (natural disaster, macroeconomic the likelihood that a shock will result in a decline in well-being. As crisis). this chapter explores, vulnerability is primarily a function of a Risk, risk exposure, and vulnerability are related but not syn- household's asset endowment and insurance mechanisms-and onymous. Risk refers to uncertain events that can damage well- of the characteristics (severity, frequency) of the shock. I40 WORLD DEVELOPMENT REPORT 2000/2001 In China 40 percent of an income decline is passed on Table 8.2 as lower consumption for the poorest tenth of households, Shocks faced by rural households in Ethiopia but onlv 10 percent for the richest third of households, Percentage of households because they have better access to insurance.8 reporting a hardship One measure of the vulnerability of the poor and Event episode in past 20 years near-poor is how often a household falls below the poverty Harvest failure line. A studv of seven countries for which panel surveys (drought, flooding) 78 labl' .Policy shock (taxation, are available found that in six of them the "sometimes forced labor) 42 poor group was significantly larger than the always Labor problems poor" group.9 A nine-year panel survey of households in (illness, death) 40 South Indian villages found that 20 percent of households Oxen problems were poor in each of the nine years and that only 12 per- Other livestock problems cent were never poor, with movement in and out of (illness, death) 35 poverty the norm for the vast majority of households.10 Land problems (land a) ~~~~~~~expropriation, reform) 17 These findings show both the high vulnerability and the Asset losses 16 strong resilience of poor households-the ability to es- War 7 cape poverty again after suffering an income shock. Rel- Crime (theft, violence) 3 ative income mobility can be quite large. In South Africa Source: Dercon 1999. 29 percent of households in the poorest quintile moved up two or more quintiles from 1993 to 1998, while in Peru 37 percent of households did so between 1985 and taxation, land expropriation, and war (table 8.2). Rainfall- 1990.11 induced income shocks have idiosyncratic components Another approach is to define long-term poverty as av- of 23 percent, but crop damage from other sources (pests, erage long-term consumption below the poverty line animals, weeds) have idiosyncratic components of 65-87 and then to ask how much of measured povertv is tran- percent. Income shocks from illnesses have an even larger sitory. This approach implicitly considers the duration and idiosyncratic component.16 The cumulation of different depth of transitions into and out of poverty. By this shocks is a source of significant stress for households: method about half the estimated poverty in South Indian villagest2 and about half the severe poverty in China are As if land shortage is not bad enough, we live a life of transitory. 13 tension wor-rying about the rain: wvill it rain or not? We Both methods suggest that transitory poverty is a live hour to hour. large part of total poverty in many settings. Generally, -lt "oITUi kaliima, Fthiopia households with the fewest assets are most likely to be chronically poor. Education almost always reduces chronic Responses to risk by households poverty, but its effects on transitory poverty differ. Bet- and communities ter educated households in Cote d'Ivoire and Hungary were found to recover better from downward income fluc- For poor people, dealing successfully with the range of tuations, but in China education is not correlated with risks they are exposed to is often a matter of life or death. transitory poverty. 4 The duration of transitory poverty To manage risks, households and communities rely on also depends on the frequency of shocks: households are both formal and informal strategies (table 8.3). Informal more likely to bounce back from a single shock than from strategies include arrangements that involve individuals repeated income shocks.15 or households or such groups as communities or vil- Vulnerability is multidimensional, and poor households lages. Formal arrangements include market-based activ- face manifold risks, so variations in income and con- ities and publicly provided mechanisms. Informal and sumption can occur for a variety of reasons. Rural house- formal strategies are not independent: public policies holds in Ethiopia, for example, face natural shocks such and the availability of formal mechanisms heavily influ- as harvest failure, health-related shocks such as illness or ence how extensively informal arrangements are used disability, and macro-level shocks such as the effects of and which kinds are used. HELPING POOR PEOPLE MANAGE RISK 141 Table 8.3 Mechanisms for managing risk Informal mechanisms Formal mechanisms Individual Group Market Publicly Objective and household based based provided Reducing risk * Preventive health * Collective action * Sound macroeconomic practices for infrastructure, policy * Migration dikes, terraces * Environmental policy * More secure income * Common property * Education and training sources resource policy management * Public health policy * infrastructure (dams, roads) * Active labor market policies Mitigating risk Diversification * Crop and plot * Occupational * Savings accounts in * Agricultural extension diversification associations financial institutions * Liberalized trade * Income source * Rotating savings and * Microfinance * Protection of property diversification credit associations rights * Investment in physical and human capital Insurance * Marriage and * Investment in social Old age annuities * Pension systems extended family capital Inetworks, * Accident, disability, * Mandated insurance for * Sharecropper associations, rituals, and other insurance unemployment, illness, tenancy reciprocal gift giving) disability, and other risks * Buffer stocks Coping with * Sale of assets * Transfers from * Sale of financial * Social assistance shocksa * Loans from money- networks of assets * Workfare lenders mutual support * Loans from financial * Subsidies * Child iabor institutions * Social funds * Reduced food * Cash transfers consumption * Seasonal or temporary migration Note: The white shaded area shows household and community responses through informal mechanisms to improve risk mitigation and coping. The dark shaded area shows the publicly provided mechanisms for insuring against risk and coping with shocks-the social safety net. a. Publicly provided coping mechanisms can also serve risk mitigating purposes if they are in place on a permanent basis. Source: Adapted from Holzmann and Jorgensen (2000). Risk management strategies can be further classified as Risk mitigation aims at reducing the impact of shocks. risk reduction and mitigation measures (actions in antic- Households mitigate risk through diversification (acquiring ipation of a shock) and coping measures (actions in re- assets whose returns are not perfectly correlated) and in- sponse to a shock).t7 Risk reduction aims at reducing the surance. Common diversification strategies are planting dif- probability of a shock or negative fluctuation. Individu- ferent crops and plots, combining farm and nonfarm income als or households can sometimes take such action them- in rural areas, and combining wage income and income frorn selves (digging wells, getting vaccinated). But to reduce household enterprises in urban areas. Households can take most risks effectively, action is also needed at the meso or most of these actions on their own-though group or gov- macro level. The risk of flooding can be reduced if the com- ernment action (agricultural extension, infrastructure) can munity builds a dike or the government builds a dam. sometimes facilitate diversification. Households also miti.- Sound economic and environmental policies, education gate risk through insurance, including self-insurance, informal and training, and other measures can also reduce a wide insurance, and formal insurance-though market-based variety of risks (and are discussed elsewhere in the report). formal insurance plays a minor role for poor people. I42 WORLD DEVELOPMENT REPORT 2000/200I Coping strategies aim to relieve the impact of a shock Evidence suggests, however, that the net effect of these after it occurs. Actions by individuals include drawing efforts is limited and that the variability of farmers' in- down savings or selling assets, borrowing, and calling on come remains high. The income options typically open support networks. Actions by government include acti- to farmers tend to move together during crises. Drought, vating the transfers or workfare mechanisms that con- for example, reduces nonfarm income as well as harvest stitute the social safety net. If these measures prove income because crop failure leads to a generalized drop insufficient, households may need to reduce consump- in income that reduces demand for nonfarm services.20 tion or increase labor supply. Many of these coping re- The range of income options available to farming sponses force a high long-term cost on households for households is often quite restricted. Evidence from a short-term benefit. Burkina Faso, Ethiopia, India, Kenya, and Tanzania This chapter focuses primarily on how to improve risk shows entry constraints-including lack of working cap- mitigation and coping by poor people. It examines ital, skills, and inputs-for many activities that could allow households' and communities' own responses through farmers to diversify their incomes. Startup costs for set- informal mechanisms. The chapter then explores the con- ting up a shop or providing services are often 10-20 ditions for public action to supplement poor people's times the cost of other activities that poor people typi- own risk management efforts-and the forms this in- cally undertake, such as charcoal making, dung cake col- tervention can take. In particular, it discusses the range lection, or simple food processing, activities that provide of safety nets that can be used for risk mitigation and only weak income diversification.2' As a result, poor coping (see table 8.3). farmers in Africa tend to be less effectively diversified than rich farmers (table 8.4).22 Poor farmers in other parts of Mitigating risk through diversification the world have had more success in diversifying income Many studies document how households throughout the sources. In Pakistan 55 percent of farmers' income in developing world diversify their income sources to 1986-89 came from nonfarm sources, and this share smooth the flow of income over time." A review of 25 was three times as high for poor as for rich farmers.23 In studies in Africa shows that rural households receive an Egypt as well, poor farmers were found to be more di- average of 45 percent of income from nonfarm activi- versified than rich farmers.24 ties, with the share ranging from 15 to 93 percent.'9 Where the possibilities for effective diversification are Farmers also diversify across crops and plots and by limited, poor farmers will specialize in low-risk, low- working for other farmers. return activities, making it hard to escape poverty. Poor Table 8.4 Income diversification among African farmers Average share of nonfarm Ratio of rich farmers' income in total income nonfarm share to poor Country Period (percent) farmers' Botswana 1985-86 77 2.5 Burkina Faso 1981-84 37 2.5 Ethiopia 1989-90 36 1.2 Gambia 1985-86 23 1.3 Malawi 1990-91 34 1.0 Mozambique 1991 15 2.5 Niger 1989-90 52 2.0 Rwanda 1990 30 5.0 Senegal North 1988-89 60 2.0 Central 1988-90 24 1.0 South 1988-90 41 2.6 Sudan 1988 38 1.0 Zimbabwe 1988-89 42 1.0 Source: Reardon 1997. HELPING POOR PEOPLE MANAGE RISK I43 Indian farmers devote a larger share of land to tradi- Simulations with household risk models suggest that tional varieties of rice and castor than to high-return va- self-insurance quickly loses effectiveness when the co.r- rieties.25 Tanzanian farmers without livestock grow more relation between income and the terms of trade of assets sweet potatoes, a low-risk, low-return crop, than do farm- exceeds 0.5. Households then have to curtail the sale of ers who own livestock. As a result, returns to farming per assets during crises because they gain so little extra con- adult household member are 25 percent higher for the sumption in return. During the 1984-85 famine in wealthiest group than for the poorest.26 Poor farmers Ethiopia asset terms of trade collapsed, and householcls are at a further disadvantage because harvest shocks are cut their consumption drastically rather than sell as- typically covariant over a fairly large area. This limits the sets.30 During the 1981-85 drought in Burkina Faso usefulness of group-based strategies and networks of mu- livestock sales compensated for only 15-30 percent of the tual support, because all or most group members are shortfall in crop income.31 likely to be affected simultaneously.27 Buying and selling cattle, though a common strategy for coping with income fluctuations, is not a feasible one Mitigating risk through insurance for many poor households. Buying a cow requires a large, In principle, any shock with a probability that can be cal- one-time outlay (and significant prior saving). In western culated from historical records is insurable. In practice, Tanzania a cow costs about a fifth of mean annual crop there are almost no insurance markets in developing income, explaining why only half of households own cat- countries because of problems of contract enforcement tle.32 Where possible, poor households use smaller animals and asymmetric information. People, especially poor (goats, sheep) or more divisible items as buffer stocks. In people, have to rely largely on self-insurance and infor- three South Indian villages farmers held buffer stocks of mal insurance instead. These problems have been over- grains and currency as their main risk management strat- come in developed countries through strong legal and egy.33 In rural China, by contrast, households increased other institutions. their holdings of unproductive liquid assets only slightly Self-insurance. Households insure themselves by ac- in response to income risk.34 cumulating assets in good times and drawing on them Because the indivisibility and riskiness of many assets in bad. The strategy is effective if assets are safe and have (price risk, survival risk for cattle) limit asset-based risk a positive rate of return, especially if the rate of return managenment strategies, poor people need a wider range exceeds the rate of time preference (of present con- of assets and greater stability of asset values. This would sumption over future consumption). In practice, re- allow them to take better advantage of opportunities for turns to assets may be negative, and many poor income growth (described in part II of this report). Sav- households have very high rates of time preference (they ings accounts hold great promise as a divisible asset with are "impatient," often out of necessity), which impedes a fixed value and positive return. Given some assurances asset accumulation.28 about the safety of the financial institution holding the Another problem is that asset values and income are accounts, the main risk would be inflation. Several recent often covariant following a macro shock, so that the experiences have underscored the great demand by poor value of assets is lowest just when they are needed most. households for safe savings accounts. Bank Rakyat In- A drought that destroys a harvest may also weaken and donesia has more than 16 million low-income depositors. kill cattle, which farmers in many poor countries use as SafeSave, an NGO in Dhaka, Bangladesh, has adapted a buffer stock. The terms of trade of assets relative to con- the principles of a traditional rotating savings and credit sumption goods may also deteriorate as a result of the association; its agents collect small sums of money daily shock, as everyone tries to sell assets and buy staples at for deposit in members' accounts.35 the same time. Both supply and demand factors push Informal insurance. Households also use group-based down asset prices: the income shock induces everyone to mechanisms of informal risk sharing that rely on the so - sell assets, and the decline in purchasing power reduces cial capital of groups of households. Typically, informal demand (unless buyers from outside the shock zone show insurance involves a mutual support network of mem-- up). In good times the process works in reverse: every- bers of a community or extended household, often within one wants to buy the buffer asset, pushing up its price ethnic groups; among members of the same occupation; and making the strategy very costly.29 or between migrants and their households of origin. 144 WORLD DEVELOPMENT REPORT 2000/2001 Like consumption smoothing, which aims to equal- uations of high economic stress, norms and social pres- ize marginal utilities over time, group-based insurance aims sure may not be enough to ensure that members of the to equalize marginal utilities across members of the group do in fact transfer resources to other members. In- group.36 When one member's consumption falls, the formal insurance works best where people value future pro- others transfer resources to rebalance marginal utilities. tection highly (rates of time preference are low) and fear These nerworks are effective only against shocks common of future exclusion from the insurance scheme keeps com- to some members but not all. So the wider the group, the pliance high. But this works against poor people, who tend less likely a shock is to affect all members, and the more to value current consumption highly relative to future con- effective the}' all are at risk pooling.37 sumption (usually out of necessity). For this reason, poor A network operates through transfers, gifts, or loans people, even though they need insurance most, are more between members, typically with expectations of reci- likely to drop out of informal arrangements. Informal in- procitv. Transfers respond to an emergency befalling a surance also works better when the rate of transfers is high member of the network, thus serving risk management (because frequent interactions create trust in future com- purposes, but they also fulfill a social function in forg- pliance) and shocks are idiosyncratic (because covariant ing community cohesion.38 The importance of gifts and shocks can wipe out the entire network's resources).39 transfers varies greatly. In Bulgaria fewer than a fifth of To determine the need for a formal safetv net, re- households receive transfers; in Jamaica more than half searchers have tried to measure how well informal insur- do (table 8.5). In most countries the bulk of transfers goes ance works, but measurement has proved difficult. It is to the poorest households, often representing a large hard to distinguish between the effects of informal in- share of income. Private transfers increase the poorest quin- surance and those of self-insurance. And because mea- tile's share of aggregate income by about 50 percent in surement requires information about consumption and Jamaica and Nepal and by almost 70 percent in Russia trends for all members (or a statistically valid sample of (figure 8. 1). them), it is especially difficult when a network extends past The occurrence of transfers is not always a sign of ad- equate protection against crises. The kev feature of informal Figure 8.1 insurance is reciprocity, self-enforced by the group. In sit- Private transfers represent a large share of the income of the poor Table 8.5 Private cash and in-kind transfers for poor Income share of poorest quintile, mid-1990s households 8 Percent After private transfers Share Share of receiving transfers 6 households All Poor Before private Country giving house- house- 5 transfers -X Iyear) transfers holds holds8 4 Jamaica (1997) 13.1 53.0 65.0 Nepal (1996) 17.4 44.7 55.3 3 Peru (1994) 14.3 37.3 46.7 Panama (1997) 15.5 37.8 40.9 2 Kazakhstan (1996) 20.2 27.5 33.8 1 * * Kyrgyz Republic 3 0 (19961 15.7 35.5 31.7 0 Russian 'A . ; ' Q Federation P" 5' / ' (1997) 23.7 25.2 31.5 11,P Bulgaria (1995) 15.0 17.0 21.4 Note: Quintiles are based on per capita income distribution. a. Households in the lowest quintile of the per capita income Transfers are those accruing to the poor. Data are for most recent distribution. year available. Source: Cox, Galasso, and Jimenez 2000. Source: Cox, Galasso, and Jimenez 2000. HELPING POOR PEOPLE MANAGE RISK I45 the boundaries of a village or other geographical entity. are especially vulnerable to fluctuations in household in- Evidence from C6te d'Ivoire, India, Thailand, and Uganda come and consumption. They are more likely than other suggests that informal insurance exists, but is far from per- children to be underweight, so that further declines in food fect.40 Evidence from China and India indicates that the consumption can cause irreversible harm. In Bangladesh poor and the landless are much less protected from income children's growth suffered during major floods.45 In rurao fluctuations than the rich and the large landholders.41 India child mortality rates increased in times of drought, especially in landless households.46 Coping with shocks The evidence on gender bias in the responses to such1 When a shock hits, people cope by cashing in their in- shocks is mixed. No such bias was found following surance: selling livestock or other assets, or calling on sup- floods in Bangladesh.47 Studies in India, however, found port networks for transfers or loans. If these mechanisms that girls' nutrition suffered more than boys' during pe- fail or fall short, households may increase their labor riods of low consumption in the slack agricultural sea- supply, working more hours, involving more household son.48 Price changes also were found to affect girls' members (women or children), or migrating to unaffected consumption more than boys'.49 For children under the areas. If all else fails, households reduce consumption and age of two, rainfall shortages were associated with more go hungry.42 deaths for girls than for boys.50 The poor have fewer options than the wealthy for cop- Some studies have found that women suffer more ing with shocks. Because they own fewer physical assets, than men from adverse shocks. Rising food prices led to poor people are more likely to increase their labor sup- larger reductions in nutrient intake for women than for ply. If the shock is covariant and the local labor market men in Ethiopia and India.51 Cultural and traditional fac- has collapsed, migration is the only answer. And if the tors can increase women's exposure to risk. Divorced men in the household migrate, women and children and widowed women in South Asia often face higher may need to pick up the slack locally.43 health risks and are more likely than married women to Coping with shocks often requires more than these eco- be poor because they lose access to their husband's prop- nomic responses. During a prolonged crisis people may erty.52 In some African countries women may lose access delay marriage and childbearing, families mav move in to household land when their husband dies.53 There is together (especially in urban areas), and people may re- also evidence of a pro-male bias in household health and sort to illegal activities (theft, robbery, prostitution). Ul- nutrition expenditures, but it is not clear whether the bias timately, the social mechanisms meant to help households affects poor households more than others. A recent study cope with shocks may come undone under the contin- in Pakistan found limited evidence that gender bias in uing pressure of a prolonged shock. health expenditures decreases with rising income.54 On balance, the evidence points to important differ- Effects within households ences in intrahousehold effects from shocks. But the ev- So far, the discussion of risk management has viewed the idence comes mainly from South Asia. Whether similar household as the unit of impact and decision. Yet risk shar- effects occur elsewhere is still unknown. ing within the household may not be equal, and the burden of the household's response may fall dispropor- The poverty trap and the long-term tionately on the weakest members, especially women consequences of inadequate risk management and children. Two situations are possible. A shock affecting As households move closer to extreme poverty and des- the household as a whole may have different effects on titution, they become very risk averse: any drop in income different household members. Or an individual shock (ill- could push them below the survival point. The poorest ness, loss of job) may have different effects on con- households try to avoid this even if it means forgoing a sumption depending on whether the affected person is large future gain in income. Despite facing the highest a man or a woman. There is more evidence on the first risk, they have the fewest resources for dealing with that situation than the second.44 risk. And forced onto the most marginal lands (floodplains, Because poor households tend to have many chil- hillsides) and into areas with poor infrastructure, they are dren, children are more exposed to poverty and vulner- most at risk from natural disasters and usually far frorn ability than other groups. Children in poor households health facilities. I46 WORLD DEVELOPMENT REPORT 2000/200I Extreme poverty deprives people of almost all means forced to engage in defensive actions to protect the as- of managing risk by themselves. With few or no assets, sets they do have. One study estimated that poor house- self-insurance is impossible. With poor health and bad holds engaging in this strategy could have boosted their nutrition, working more or sending more household incomes by 18 percent with a more entrepreneurial man- members to work is difficult. And with high default agement strategy (but one that requires access to credit).60 risks, group insurance mechanisms are often closed off. Dysfunctional factor markets can also create or ag- The poorest households thus face extremely unfavor- gravate poverty traps. Take child labor. When a crisis strikes able tradeoffs. When a shock occurs, they must obtain and households cannot borrow or when adult unem- immediate increases in income or cut spending, but in ployment is high or wages low, children are pulled out so doing they incur a high long-term cost by jeopardiz- of school and sent to work. The lost schooling leads to ing their economic and human development prospects. a lifelong loss in earning ability for these children. Fail- These are the situations that lead to child labor and mal- ures in the credit or labor markets thus transmit poverty nourishment, with lasting damage to children, and the and vulnerability across generations. breakdown of families. In C6te d'Ivoire severe economic recession caused Policy responses for improving households, especially the poorest, to sharply increase the risk management labor supply of children.55 In rural India child labor was found to play a significant role in households' response Since poor people cannot fully manage risk on their to seasonal variations in household income.56 In every part own, any poverty reduction strategy needs to improve risk of the world participants in the Voices ofthe Poor study men- management for the poor-reducing and mitigating risk tioned child labor as an undesirable coping mechanism. and coping with shocks. The strategy should include In Egypt children were sent to work in a storehouse pack- formal and informal mechanisms, provided by both the ing vegetables. During periods of drought in Ethiopia chil- public and the private sector. dren were taken out of school and sent to towns to be In principle and excluding cost considerations, the best employed as servants, with their earnings sent back to their approach is to reduce the risk of harmful shocks.62 Next families. In the lean season in Bangladesh children work would be risk mitigation to reduce the possible impact on farms, tend cattle, or carry out household tasks in ex- of a shock. Coping would be a residual approach to ad- change for food. Parents are often aggrieved by the undue dress the failures of the first two.63 In practice, different physical labor of their children and worry especially about direct and opportunity costs may well change the rank- the vulnerability of girls to beatings and sexual assaults.57 ing of options. Some risk reduction and mitigation strate- Inadequate risk management can also compromise gies are prohibitively expensive, especially those for nutrition in poor households. After the devastating floods dealing with infrequent but catastrophic shocks. of 1988 in Bangladesh, many households took out loans Comparative cost data and cost-benefit analyses are to meet consumption needs, but landless households generally not available to help policymakers choose from were less able to do so and their children suffered more different types of risk management interventions. Fur- severe malnutrition.58 A study of rural Zimbabwe found thermore, the distributional implications of different that the 1994-95 drought caused a 1.5-2-centimeter strategies need to be considered. A comparative study in decline in annual growth among children one to tvo years India found that, at the margin, public work programs ben- old. Although this study found the reduction to be per- efit the poorest quintile the most, while credit programs manent, other studies have found evidence of catch-up benefit the second and third poorest quintiles the most.64 during subsequent good periods.59 Most developing countries pay too little attention to What do households suffering these unfavorable long- risk reduction and mitigation and rely too much on in- term effects on the education and nutrition of their chil- terventions after disaster strikes. Efforts to cope with the dren have in common? Low asset endowments (physical, Mexican peso crisis of 1995 and the East Asian financial human, social) and little or no access to credit and in- crisis of 1997 have shown how difficult it is to put effective surance markets-a chronic trap for poor people, unable safety nets in place after the fact (chapter 9). to accumulate enough assets to escape poverty. When The balance needs to shift from policies for coping households do not have some threshold of assets, they are to those for reducing and mitigating risk. That means HELPING POOR PEOPLE MANAGE RISK 147 ensuring that social safety nets such as workfare programs, Box 8.4 targeted human development programs, and social funds Managing risk: the modular approach to social are in place on a permanent basis and can be scaled up safety nets when a shock occurs (see table 8.3). Interventions fol- lowing the 1998 floods in Bangladesh were effective be- Constructing a social safety net is far from an exact science, caue o th exstig etwrk f N Osnd thr and the process will vary from country to country depend- cause of the existing network of NGOs and other ing on the context, data availability, and political urgency. But mechanisms ready to be activated to help poor peo- the process should have certain analytic elements, including ple.65 It also means providing better access to credit and establishing the country context, constraints, and challenges; financial assets, facilitating income diversification, man- identifying sources of risk, vulnerable groups, and potential aging labor market risk better (especially child labor), and interventions; and identifying the optimal mix of programs. Malawi illustrates the mix of preferred programs that can re- providing health insurance. Such actions would allow sult, depending on prevailing conditions. poor people to pursue higher-risk, higher-return activ- Malawi is a low-income country, with more than half its ities that could pull them out of poverty.66 Social safety population in severe poverty. The vast majority of the pop- nets can also serve as an automatic compensatory mech- ulation depends on subsistence agriculture. There is little gov- anism for the unwanted distributionalefernment revenue surplus to redistribute and limited anism for the unwanted distributional effects of policy administrative capacity to manage complex programs. There reforms (chapter 4). By doing so, they will help make is no formal social safety net. reform socially and politically feasible. While a new bal- ance is needed, coping mechanisms will remain vital Identifying sources of risk and vulnerable groups for disngewit unfoe see rand sinfeq enshc where Vulnerable groups in Malawi were identified on the basis of for dealing with unforeseen and infrequent shocks where a poverty analysis conducted in the early 1990s. Four groups it is prohibitively expensive to put mechanisms in place were found to be most at risk: rural households with small ahead of time. landholdings, female-headed households, AIDS orphans and Not every country needs to set up a comprehensive so- their relatives, and those who could not care for themselves. In addition, four major risks were identified: seasonal price cial safety net. But each does need to construct a mod- icessadfo hrae,proi ruh,lrepn r . r . . ~~~~~~~~~~~Increases and food shortages, periodic drought, large peri- ular system of programs based on its own patterns of risk odic macroeconomic shocks, and the threat of HIV/AIDS. Po- and to cultivate a suitable mix of providers (public and tential interventions to address these risks were developed. private) and administrative arrangements (box 8.4). The first step in selecting and designing programs is to un- intervention$ derstand the general principles of how safety nets com- A cost-effectiveness analysis of existing programs was con- plement existing risk management arrangements. The next ducted before potential new interventions were ranked by is to identifv specific types of risk (illness, old age, un- priority. The results, together with consideration of the vul- employment) and the mechanisms for dealing with them. nerable groups, the risks, and the need to focus on productivity-enhancing interventions, led to the following modular system of programs: Generalprinciples of safety nets * Public work (risk mitigation and coping). and risk management * Transfers for orphans in poor communities (risk mitiga- Reducing risk is possible for some categories of risk but tion and coping). * Nationwide nutrition program (risk reduction and coping). not all. For example, building a dam can reduce the risk * Targeted cash transfers to the needy (coping). of flooding. Immunizations and other public health cam- paigns can reduce the risk of illness. Policies undertaken Source: World Bank forthcoming b. primarily for other purposes can also contribute to risk reduction. Good education policies, including scholar- ships for poor families, can reduce child labor. Environ- available to farmers can help rural residents to diversify. mental policies can limit deforestation, reducing damage Opening trading opportunities through investments itn from hurricanes and deaths from mudslides. Sound infrastructure and other means can also stimulate diver- macroeconomic policies can reduce the risks of high in- sification. But liberalizing markets (say, by privatizing state flation and unemployment. commodity boards) can have mixed effects and will nolt But the focus in this chapter is primarily on mitigat- always benefit poor people. Sometimes dealers step in be- ing risk (diversification and insurance) and on coping. tween farmers and export traders and capture most of the Making a wider variety of crops and extension services gains from open trade. I48 WORLD DEVELOPMENT REPORT 2000/2001 Policies should also make it easier for poor people to public sector can pool risks over a larger area, the possi- build up assets while reducing the covariance betveen asset biliry exists for providing insurance at a lower cost than values and income. Covariance is a big problem in rural informal agents can (assuming that information problems areas, where asset values (livestock) often move in tandem can be dealt with; see below). Publicly provided insurance with farm income. This could be addressed through bet- could thus yield a net gain to society-if the state is per- ter integration of asset markets with the wider economv- ceived as credible and the insurance scheme is fiscally sus- by investing in transport infrastructure, disseminating price tainable. information, and removing structural and institutional But if trust in the state is low, few people will put their market barriers. Macroeconomic stability promotes more faith in the government system and give up their personal stable asset prices, reducing inflation-driven deterioration or group insurance. And even if credibility is not an in the terms of trade of assets relative to consumption issue, fiscal constraints may prevent the state from mak- goods. And easier access to credit would facilitate the ac- ing payments during a crisis. People who had given up quisition of costly indivisible assets, such as cattle.67 their informal insurance mechanisms would then be left Another critical intervention is the provision of in- worse off than before the state offered insurance. Rela- surance, especially for covariant risk. Self-insurance has tive cost-effectiveness, trust, and sustainability thus all need limits, mainly because poor people cannot accumulate to be considered in deciding on government intervention. enough assets, especially after successive shocks. And in- Government spending on social safety nets varies con- formal insurance, which relies on risk sharing across a siderably. Figure 8.2 illustrates this with one component: community or network, is ineffective for covariant spending on social security by the central government. But shocks. costs are only part of the picture. These expenditures are The first question with insurance is whether market also investments in human capital formation. By provid- or government provision is more cost-effective than ing poor people access to basic services and allowing them informal mechanisms. Can the state provide less costly to undertake higher-risk, higher-return activities, the in- insurance for risks that are self-insured by poor people vestments can have positive effects on poverty and eco- or insured through group-based risk sharing? Because the nomic development. Costs are still likely to be an issue, Figure 8.2 Central government spending on social security varied greatly in 1995 Percent 35 30 25 Share of government 20 . i expenditure 15 10 t t | I t V t I 0 n Share 10~~~ of GNP] l | | * l '_0_ 5 7 0 Source: IMF 1995. HELPING POOR PEOPLE MANAGE RISK 149 but they can often be reduced by more explicitly consid- dividuals or households. Most safety nets target spe- ering the role of private providers.68 cific types of people or households: the ill, the el- Concerns that formal safety nets will displace self-in- derly, the women heading households with many surance or group-based mechanisms also need to be children, and the like. The danger is that improving considered. Empirical estimates of this effect vary, so the the risk position of one person belonging to a group- country context is important. A study of the urban based insurance scheme creates an incentive to drop Philippines estimated that government-provided un- out of the group. If this leads to the collapse of the employment insurance would displace 91 percent of group scheme, members not covered by the safety private transfers to the unemployed.69 Another study es- net could end up worse off. The solution is to target timated that providing a basic pension benefit to black broad groups (say, a credit program for the entire South Africans displaced only 20-40 percent of private community or specific groups within it), although transfers to the elderly.70 Studies of other countries also doing so can be difficult because insurance groups do found displacement rates on the order of 20-40 percent.71 not always coincide with communities or other eas- Displacements of private transfers need not imply a so- ily identifiable target groups. Of course, if the safety cial loss. If poverty reduction objectives are considered along net protects almost everyone, the disappearance of with insurance objectives, there may well be a net social informal insurance arrangements may not matter, aLt gain, despite the displacements.72 In South Africa many least if the formal safety net is more cost-effective and of the displaced transfers were from young to old house- sustainable.77 holds, both of them poor. The new pension program left In the end, decisions on safety nets need to weigh the more money in the pockets of poor young households and negative effects of displacement against the positive ef- also covered many elderly residents who had not been re- fects of long-term improvements in the welfare of poor ceiving private transfers. Overall, then, the pension scheme households. Safety nets are not the only way to improve significantly strengthened South Africa's social safety net.73 poor households' ability to manage risk and to engage in When should the state step in and provide a social higher-risk, higher-return activities. Stable macroeco- safety net for poor people-and how? The general answer nomic policies may do more to reduce employment risk is that it depends on the types of shocks likely to occur than public work programs. But sound economic poli- and the kinds of private insurance arrangements in place. cies may increase the risk for some categories of house- * If informal arrangements insure adequately against idio- holds. Trade liberalization may lower the cost of imported syncratic risk, the state should step in to insure against clothes and utensils, reducing demand for weaving and covariant risk. In most circumstances providing this handicrafts-two activities with low entry costs fre- coverage will improve overall risk management and in- quently used by poor people to diversify income.78 So the crease welfare, without crowding out informal insur- decision about providing safety nets needs to be viewed ance.74 But since households' overall risk exposure in the full context of economic and social policies and of will have declined, self-insurance (precautionary sav- the impacts on household risk. ings or other asset buildups) may decline.75 Where there is a strong concern for the poor, especially * Where informal insurance is ineffective-because of en- the very poor, the formal-informal, public-private balance forcement problems or because shocks are too frequent generally shifts in favor of public provision of insur- or too large-household welfare could be increased if ance.79 Concerns for sustainability and other incentives the social safety net insured against both idiosyncratic in group-based insurance and credit schemes generally and covariant risks. Whether coverage should come work against inclusion of the poorest, who have a higher from the state or private insurers depends largely on the perceived risk of default.80 Similar concerns tend to ex- type of risk. The state is often best able to cover covariant clude poor people from market-based insurance. Thus risks, but most idiosyncratic risks may be better han- public insurance provision is not likely to undercut any dled by private providers (communities, insurance informal arrangements involving the poorest households. firms). The government's role should then be to facil- Further strengthening the case for public intervention itate and, if necessary, regulate private provision.76 is the ineffectiveness of the insurance mechanisms used * Where group-based informal insurance works well, the by poor people against repeated shocks-mechanisms that state should avoid safety net programs targeted to in- also tend to be costly.81 A study of six South Indian vil- 150 WORLD DEVELOPMENT REPORT 2000/2001 lages found that farmers sacrifice as much as 25 percent sured, coproduction is frequently recommended: the of average income to reduce exposure to harvest risk.82 government provides the financial and technical means, Several practical issues have to be considered in setting and local institutions or peer groups take care of imple- up state insurance programs. These include obtaining in- mentation and monitoring. Or the government provides formation about the people to be insured and dealing with funds to communities, which are responsible for identi- the political economy issues in providing insurance. fying poor beneficiaries (box 8.5). Obtaining information about people to be insured is The political economy may strengthen or weaken the costly. That is why so many traditional credit and insur- case for publicly provided risk mitigation. The state may ance institutions are local. Moneylenders or members of well be the best agent to provide insurance, but lack the a rotating savings and credit association have a better necessary institutional strength, financial resources, or chance of knowing who is a bad risk than would an out- management capacity. Capacitv building may then be re- side insurance program.8" Asymmetric information creates quired inside the government. The political support to problems of moral hazard and adverse selection, leading allocate resources may also be lacking, since it requires to the underprovision of insurance (relative to the social getting the rich to support a program that does not ben- optimum) by private providers.84 Because information efit them. If the insurance program is not self-support- problems are especially acute for poor people, the social gains ing, it may have to be funded out of general tax revenue, from government provision of insurance may be large. at the expense of other programs that benefit the rich. Because the government has no comparative advan- (Chapter 6 discusses the political economy of poverty re- tage in obtaining local information on who should be in- duction further.) Box 8.5 Is targeting by the community a good idea? In most social safety net programs the central government provides school textbooks are allocated locally by parent-teacher associa- funds and sets the eligibility criteria, ostensibly guaranteeing equal tions or the school principal. The program has not been formally eval- treatment across the country. But local needs may vary across the uated, but informal appraisals suggest that the system has been well country, and benef its may leak to ineligible households in varying de- accepted by parents, and it may be expanded to other types of aid. grees. In an effort to improve targeting, an increasing number of pro- The Kecamatan Development Project in Indonesia provides grams rely on communities to determine eligibility rules and identify block grants to 10,000 villages. Each community decides on the beneficiaries. The success of this approach depends in part on the use of the funds through an extensive process of information degree of social cohesion in the community and whether the com- dissemination, community facilitation, and proposal preparation and munity can be effectively mobilized in a consultative process to al- selection. Field assessments indicate that the process works best locate benefits. when both traditional and official community leaders are on board Targeting efficiency also depends on the entity charged with al- from the start (KDP Secretariat 1999). locating benefits. In Uzbekistan quasi-religious community groups In Albania the Economic Support Program helps poor rural known as maha/las target child benefits and other types of social as- households and people who lose their jobs in the transition. Local sistance to low-income families. They have considerable discretion governments receive block grants to allocate within their com- over amounts and criteria for assistance. An extemal review concluded munes. Local targeting compares favorably with that of safety net that benefits were targeted fairly well. In Armenia subsidies for programs in other countries. Advantages and disadvantages of allowing communities to allocate benefits Advantages Disadvantages Better information is available on needy households Program may be used to serve interests of the elite Allocation criteria are adapted to local needs Participation of community leaders may have opportunity Decentralized administration is more efficient cost Community mobilization may build social capital Allocation rules may cause increased divisiveness in the community Externalities across communities may not be taken into account Source: Conning and Kevane 1999. HELPING POOR PEOPLE MANAGE RISK I5I Box 8.6 Insurance options for the informal sector The simple expansion of statutory coverage of formal sector social delivery of welfare services from individual employee-employer re- insurance programs (pensions, unemployment insurance, disability lationships, removing a major bottleneck to including informal insurance) to small enterprises will not meet the risk management sector workers in contributory schemes. needs of the informal sector. Schemes need to accommodate the Several Indian states recently tried a more experimental group lower contributory capacity and greater earnings volatility of self- insurance scheme. In Gujarat about 20,000 landless agricultural employed and informal workers. laborers received life and accident insurance coverage. Most Lines may need to be blurred between strict contributory self- group insurance schemes are not fully self-financed and require financed insurance schemes and assistance paid out of general state government contributions. tax resources. Also called for are flexible partnerships between dif- In surveys informal sector workers regularly single out health ferent providers: the state, private insurance companies, com- insurance as their greatest insurance need. In China rural health in- munities, NGOs, and organizations representing the informal surance covers hospital and primary health care costs through pri- sector workers. vate and public contributions. Premiums paid by beneficiaries are Many contributory, and often self-managed, schemes for in- supplemented through a village public welfare fund and government formal sector workers have emerged in recent years. They are ei- subsidies. ther adapted from formal systems or based on cooperatives and In Tanzania a pilot project in Dar es Salaam provides health in- mutual benefit societies. In some cases they complement surance through five mutual associations of informal sector work- indigenous risk management arrangements, such as burial and ro- ers. In Igunga, a town in the northwest, a community health fund tating savings and credit societies. covering primary health care services has achieved 50 percent par- In the Indian state of Andhra Pradesh pension coverage has ticipation. Since the scheme relies on matching funds, sustainability been given to about 425,000 home-based workers in the beedi (leaf- remains an issue. rolled cigarettes) industry, under the Employees' Provident Fund Key to the success of contributory insurance schemes for the Act. A simple procedural mechanism-issuing identity cards-was informal sector are their organization around an association based crucial to success. In addition, a welfare fund for beedi workers on trust and mutual support (professional group, village) and the ad- was set up by the central government, funded through an ear- ministrative capacity to collect contributions and provide benefits. marked tax collected from employers and manufacturers in the Administrative capacity can be fortified by an umbrella organization beedi industry. This delinks the collection of contributions and the that joins local groups and provides them with technical support. Source: van Ginneken 1999. Public risk mitigation may also reduce profit oppor- programs can. Community-based and integrated provi- tunities for the rich (from money lending) or undercut sion of insurance are two innovative approaches showirLg patron-client relationships between rich and poor by promise. Such programs recognize the strong links between making the poor more independent.85 Allowing the rich labor market risks and other risks in the informal econ- (or at least the middle class) to participate in some insurance omy.87 Packages that combine different types of insurance programs and showing that insurance is less costly than or assistance for the self-employed may be particularly at- other poverty reduction efforts can boost political support tractive. In Chile many self-employed people participate for publicly funded insurance. Above all, as chapters 5 and in the pension system to get health insurance.88 6 argue, public risk mitigation will succeed only if poor people have a channel for dialogue with government on Specific instruments and the lessons issues of risk and vulnerability. of experience Special considerations stem from the large (and grow- While the general principles discussed here are useful in ing) informal sectors in many developing countries (box framing choices for policymakers, it is their application 8.6). Employment in the informal sector in 12 Latin to specific cases and the lessons of experience that really American countries rose from 50 percent of the eco- matter. Many tools are available for public interventions nomically active population to 54 percent between 1990 to improve the ability of households to manage risk. Tlhe and 1997.86 A large "gray" economy has sprung up dur- rest of this chapter covers seven tools especially relevant ing the economic transition in Eastern Europe and the for- for poor people: health insurance, old age assistance and mer Soviet Union. Unemployment insurance will not pensions, unemployment insurance and assistance, work- reach workers in the informal sector, but social assistance fare programs, social funds, microfinance programs, and 152 WORLD DEVELOPMENT REPORT 2000/2001 cash transfers. Some of these instruments address primarily Injuries and chronic illnesses that result in long-term one type of risk-others are useful for a wide range of risks. disability affect an estimated 5-10 percent of people in Health insurance. Several studies have shown that developing countries.96 Disability is associated with low many households in developing countries cannot insure education, poor nutrition, high unemployment and un- against major illness or disability. Significant economic deremployment, and low occupational mobility-all fac- costs are associated with these conditions, both in income tors that increase the likelihood of being poor. And being losses and in medical expenses. The policy response poor adds to the risk of becoming disabled. Much dis- should be to provide health insurance and to direct pub- ability in developing countries is caused by injuries or by lic health spending to facilities serving primarily poor peo- communicable, maternal, and prenatal diseases, some ple (or poor areas).8'9 of them preventable. Medical prevention of disease be- Some middle-income countries have set up universal comes easier with rising incomes, of course. health insurance, as Costa Rica and the Republic of In the long run policy efforts need to focus on pre- Korea did in the 1980s. But most low-income countries vention, especially on maternal and child health care. Pro- can offer only limited health insurance, usually provid- grams to eradicate measles, to fight onchocerciasis (river ing minimum benefits for all illnesses ("first dollar cov- blindness), and to reduce micronutrient deficiencies have erage") rather than full insurance for infrequent but very already greatly reduced disabilities.97 Preventive pro- costly illnesses.'0 This choice may look pro-poor (bene- grams that keep simple diseases from becoming chronic fits are provided regardless of income and there is no de- disabilities are especially important for children.98 War ductible or copayment), but the evidence suggests that and civil conflict have also caused many disabilities. Land catastrophic illnesses and disabling injuries create much mine accidents have increased sharply over the past 15 greater problems for poor people than frequent, minor years: a study of four war-affected countries found that illnesses. Households in Indonesia were able to smooth 6 percent of households had a member who had been more than 70 percent of consumption fluctuations caused killed or permanently disabled by land mines.99 by moderate health shocks, but only 40 percent of those People with disabilities incur extra medical costs and caused by large health shocks.9' An average hospital stay are often excluded from services and community activi- in Indonesia costs 131 percent of the annual income of ties.100 Most people with disabilities depend on their the poorest quintile of households, but only 24 percent families for support and cannot increase their labor sup- of the income of the richest quintile."2 In China house- ply in response to income crises. One study found that holds could smooth only 6 percent of consumption fluc- 61-87 percent of land mine victims went into debt to pay tuations caused by overall medical care costs, but 100 their medical bills, and 12-60 percent had to sell assets.101 percent of fluctuations involving health care expenses of Prevention and better health care hold the key to reduc- less than 50 yuan.93 ing disabilities in the future. Those who are already dis- Public provision of insurance against catastrophic abled need community-based rehabilitation programs health risks could thus significantly improve the welfare and public transfers to the families that provide care.102 of poor people where households are unable to insure Old age assistanceandpensions. The risks associated with against these risks themselves. The evidence further sug- old age have social as well as economic dimensions, and gests that premiums can be quite low (because major ill- policies need to address both. To reduce the social isola- ness is rare) and well below households' willingness to tion of many of the elderly, social policies should facili- pay.94 Countries as diverse as Costa Rica and Singapore tate access to community groups or associations that have implemented health insurance schemes with near cater to the elderly. Proximity to health facilities is also universal coverage (box 8.7). Where administrative ca- a major concern, since elderly people have difficulty pacity or other constraints make catastrophic health in- reaching faraway clinics. surance infeasible for poor people, subsidies for hospital On the economic side, many elderly are poor because care can be used instead. For this to be pro-poor, how- they have been poor all their lives.103 Poverty reduction ever, there must be equity in referrals and access to hos- policies that increase people's income during their work- pitals.95 In both approaches the objective is to avoid a need ing lives will also make them better off during retirement. for poor people to pay for medical emergencies through Well-functioning financial markets that facilitate saving debt, distress sales of assets, or cuts in consumption. and investment will help workers accumulate financial as- HELPING POOR PEOPLE MANAGE RISK 153 Box 8.7 Two universal health insurance systems: Costa Rica and Singapore . Costa Rica and Singapore have vastly different income levels and ties. Each health care team is supported by a health committee administrative capacity, but each has succeeded in establishing set up by the community. universal health care coverage. They also have some common char- Still, all is not well in Costa Rican health care. Waiting times acteristics that are helpful in targeting fee waivers to poor people, are long, and there have been complaints of improper treatment such as almost universal literacy and a system of formal docu- of users. As a result, many people entitled to public services go mentation of vital events (births, marriages) and transactions (em- to private providers for low-cost procedures. And because eligi- ployment contracts, utility bills). bility is not linked to a specific number of premiums, some peo- ple pay premiums only when they need costly treatment. This Costa Rica violates the solidarity principle of an insurance system. Costa In Costa Rica the public sector designs and carries out health care Rica has introduced reforms to deal with these problems. policies. The role of the private sector in health care is very limited: barely 2 percent of hospital beds in the country are in private faci- Singapore ities. The Costa Rican Social Security Fund was created in 1943, and Between 1984 and 1993 Singapore set up a three-tiered system coverage for health services was extended to the entire population of health insurance: Medisave, Medishield, and Medifund. The pro- in 1971. About 85 percent of the population actually participates. Fund- gram insures against intermediate-level health risks through indi- ing comes through payroll deductions and voluntary, income-based vidual or household Medisave accounts. These mandatory savings contributions of the self-employed. Public spending on health care accounts, part of Singapore's compulsory social security system, has remained high, ranging from 4.7 to 6.8 percent of GDP during are funded by a 40 percent payroll tax (shared equally by employers 1975-93. Universal health insurance went hand in hand with health and employees). Of this contribution, 6-8 percent is allocated to care strategies aimed at preventing disease, addressing specific risk Medisave accounts, which can be used to pay hospitalization ex- factors, and extending service coverage to rural and urban areas. penses of up to about $170 a day. Individuals are expected to cover Health indicators responded. Between 1975 and 1990 infant mor- minor health costs out of pocket or through private insurance. tality declined from 37.9 to 15.3 per 1,000 live births, and medically Catastrophic health risks are covered through Medishield, optional assisted births rose from 82.5 percent to 95.2 percent. backup insurance for expenses exceeding the maximum coverage The 15 percent of the population not covered by the national provided by Medisave. Eighty-eight percent of Medisave account hold- health insurance program is concentrated at the lowest end of the ers have opted for Medishield coverage. The coinsurance rate is 20 income distribution. A free health insurance program covers more percent, and the deductible amount varies with the comfort class of than three-fourths of this group. Eligibility is verified through sys- the medical facility. tematic evaluations by social workers, based on documentation Equity backup is provided through subsidies from Medifund, to provided by applicants on household composition, earnings, and remedy the nonprogressive nature of Medisave accounts and housing conditions. The administrative reviews of applicants are Medishield. A catastrophic health shock would cost 55 percent of methodical and effective: 55 percent of program benefits go to the annual per capita expenditures for the poorest quintile of households, poorest quintile. and just 21 percent for the richest quintile. The Medifund subsidies Equity concerns are further addressed in the primary health care are differentiated by class of facility and thus are self-targeted to poorer reform started in 1995. The country has 800 health zones, each users. As a last resort, patients who are unable to pay all their med- served by a comprehensive health care team that ensures universal ical bills can apply for a means-tested grant from their hospital Med- access to primary care and suitable referral to higher-level facili- Ifund committee, financed from the government's budget surplus. Source: Grosh 1994; Prescott and Pradhan 1999; Sauma 1997. sets over their lifetime. This is especially important for percent of the labor force are usually seen only in countries informal sector workers and the self-employed, who with annual per capita income exceeding $5,000.106 rarely participate in pension plans. Higher incomes and The general recommendation for pension reform is to better risk management for today's prime-age workers will establish a multipillar system: combining a publicly man- also help them support their parents financially. aged defined-benefit plan with a privately managed Formal pension systems are limited in most developing defined-contribution plan, supplemented by voluntary countries, covering only 16 percent of the labor force in the retirement savings. The publicly managed plan, funded developing world.'04 In the poorest countries in South from general tax revenues, can address poverty and eq- Asia and Sub-Saharan Africa pensions cover less than 10 per- uity concerns. The privately managed plan, fully funded cent of the labor force. 15 Coverage can be increased through by participant contributions, serves as wage replacement suitable reform, but this takes time: coverage rates above 50 after retirement.107 Several countries, mainly in Latin 154 WORLD DEVELOPMENT REPORT 2000/2001 America and Eastern Europe, have multipillar pension sys- Box 8.8 tems. But successful management of such systems requires Social pensions in Chile and Namibia considerable administrative capacity. Even a well-structured pension system will not initially Countries as different as Chile and Namibia have established reach the poor. Coverage in formal pension systems tends social pension schemes to cover the most vulnerable elderly. to be much greater for high-income workers: in Chile Chile has a multipillar pension system, with pensions that to be much greater for high-income workers: in Chile depend on years of employment and contributions. Gender more than 40 percent of workers in the poorest income differences in earnings and in years in the labor market lead to decile do not participate in the pension system, compared wide differences in pension payments. Take a woman with an with fewer than 20 percent of workers in the richest incomplete primary education and with average tenure in the decile.108 In general, coverage is lowest among the poor, labor market who retires at her statutory retirement age of 60: she would receive only 29 percent of the pension of an equally the uneducated, the self-employed, and women who qualified man who retires at his statutory retirement age of 65. have worked in the household rather than in the labor In addition to this formal pension system, the government market for most of their lives. finances a social assistance pension intended for poor women Contributing to this lower coverage are market and in- and men over 65 not covered by the formal system. Because the program is means tested rather than employment based, stitutional failures and incentives that discourage mdi- benefit amounts are not differentiated by gender. Since el- viduals from seeking coverage. The profile of risks that derly women are generally poorer than elderly men, the pro- poor households face may mean that illness or harvest fail- gram benefits women proportionately more than men, ure are of much more concern to them than old age in- especially in rural areas. Namibia administers a social pension program for indi- come security. In a credit-constrained environment viduals overage 60. Unlike Chile's program, Namibia's is uni- mandatory contributions to a pension system may be dif- versal, not means tested. In practice, 88 percent of eligible ficult for poor or self-employed households to meet. If pensioners receive the pension. The social pension con- in addition the public pension svstem lacks credibility, tributes significantly to poverty reduction. It is the main source of income for 14 percent of rural households and 7 many households will continue to rely for old age income percent of urban households. security on traditional informal arrangements, often The social pension program also indirectly helps chil- based in the household, extended family, or tribe.109 dren, because many elderly people in Namibia look after Addressing the needs of the elderly poor thus requires grandchildren and pay their school fees when the parents are more than pensions. Preventive measures include facili- away working or looking for work or because the children are more tan pesions Prevntive easurs incude fcili-AIDS orphans or disabled. tating saving and investment and providing poverty re- duction programs during people's working lives. Different Source: Cox Edwards 2000; Subbarao 1998; World Bank forthcoming a. forms of direct and indirect support are needed for today's elderly. Programs can provide assistance to families that care for live-in elderly."'' Retraining and workfare pro- should be a primary objective.'13 In the informal sector, grams adapted to older workers can make it easier for them where laws and regulations are seldom applied, public ac- to continue working. i And social assistance or social pen- tion can complement customary informal arrangements sions should cover the poorest and the very old (categories to improve the environment in which workers operate. that frequently overlap) and those without family sup- Reform and enforcement need to be combined with port (box 8.8). Widows will often make up a large part programs of skill enhancement, job search assistance, and of this group.112 microenterprise development. Since experience with Unemployment insurance and assistance. Labor market government-run training programs is mixed, partner- risk can be reduced significantly by improving the func- ships with the private sector need to be explored. 14 Labor tioning of labor markets and by adopting sound macro- markets can also be made more effective by improving re- economic policies. Many labor markets in developing lationships among labor market partners (employers or- countries are segmented (effectively barring entry to some ganizations, trade unions, and government) and by groups) and excessively regulated. Reform of labor laws strengthening collective bargaining and contracting.' 5 and regulations needs to balance greater efficiency in the Even a well-functioning labor market will not fully labor market with promotion and enforcement of core eliminate the risk of unemployment or underemployment, labor standards to protect vulnerable workers (chapter 4). however. Displaced workers will need unemployment Eliminating the most exploitative forms of child labor benefits to protect them from large income losses and HELPING POOR PEOPLE MANAGE RISK 155 poverty. In some countries the link between unemiploy- Workfare programs are not necessarily an inexpensive ment and poverty is very strong. A study of poverty in wa) of delivering benefits to poor people. Their cost- countries of Eastern Europe and the former Soviet Union effectiveness needs to be compared with that of alterna- found a 40-80 percent higher incidence of poverty tive transfer programs. The cost per person-day of among households that had one unemployed member employment created varies greatly across countries, rang- than among households that had no unemployed mem- ing from as low as $1-2 in several South Asian programs ber. Households with several unemployed members had to $8 in Bolivia. The cost depends on the wage rate, type poverty rates twice the national average or more in some of projects undertaken, costs of local private contrac- countries.116 Typical unemployment programs in the re- tors, and administrative effectiveness. Wages typically gion include retraining, wage subsidies, job counseling represent 30-60 percent of total costs.124 and referral services, public work and community em- Socialfinds. Social funds help finance small projects ployment, and small business creation programs. 117 identified and implemented by poor communities, which Unemployment insurance, the traditional means of usually provide cofinancing. Almost 50 countries, most mitigating the risk of job loss, is not appropriate for in Latin America and Sub-Saharan Africa, operate social most developing countries because of their low admin- funds or similar entities. The world's largest is in Egypt. istrative capacity and large informal sectors. The irregu- Recently, Eastern European and Central Asian countries lar and unpredictable earnings typical of the informal have begun to set up social funds, with 10 already in op- sector make it hard for workers to participate in a con- eration or under preparation. tributory insurance program.1'8 Many of the market and The first social fund was set up by the Bolivian gov- institutional failures discussed under pension systems ernment in 1987 as an emergency response to a general apply also to unemployment insurance. Better options for economic dowvnturn. Generally, however, social funds are assisting the unemployed are means-tested social assistance not coping instruments. Instead, they address a wide and public work programs (workfare)."9 Means testing range of objectives, including infrastructure, commu- has proved difficult in most settings, but promising ap- nity development, social services, and support for de- proaches that use easily observable indicators for target- centralization. 25 But some have been used to respond to ing are being pilot tested.'20 emergencies-Hurricane Mitch in Central America Workfare programs. Public work programs are a useful (chapter 9), civil war in Cambodia, an earthquake in Ar- countercvclical instrument for reaching poor unemployed menia, drought in Zambia. Social funds have also grad- workers. They can easily be self-targeting by paying wages ually assumed a greater role in risk mitigation-supporting below market rates. A well-designed and well-funded work- income generation projects, stimulating school enrollment fare program is a mix of risk mitigation and coping. To mit- and health center use, and strengthening the social cap- igate risk, the program must inspire confidence that it will ital of communities. They have proved to be flexible, quick continue to be available after a crisis. Only if the govern- to respond, and cost-effective. But the record is mixed ment is perceived as credible will such programs induce when it comes to sustainability and poverty reduction.16 households to give up costly self-insurance or group in- Social funds use three targeting devices to reach poor surance, freeing resources for other productive purposes. 12' people: investment selection (mainlv basic services), pro- The program functions as a coping mechanism by providing ject screening (to ensure that most beneficiaries are poor), jobs when a crisis strikes. Providing households with income and geographical targeting (of poor areas). The poverty following a crisis helps them avoid costly and damaging targeting strategy and the demand-driven approach of strategies (selling assets, reducing food intake). Some work- social funds are sometimes in conflict, To enhance their fare programs-such asTrabajar in Argentina, theTfemporary effectiveness, many funds initially financed projects in Employment Program in Mexico,'22 and the Maharashtra better-off communities with good organizational skills. The Employment Guarantee Scheme in India-have succeeded poorest communities, which often have difficulties putting in creating employment for poor people (box 8.9). Other investment proposals together, received fewer benefits. programs not originally designed as workfare programs To address this problem, some social funds (Ar- may actually perform very similar functions. This is the case gentina, Chile, Mali, Romania) have supported capac- for Mexicols self-targeted Probecat, which provides train- ity building in poor communities. Others have ing to the urban unemployed.'23 temporarily assumed some implementation responsi- I56 WORLD DEVELOPMENT REPORT 2000/2001 Box 8.9 Principles of successful workfare programs In many programs for the poor a large share of benefits go to the Benefits to the poor nonpoor. This problem has stimulated interest in self-targeting Since poor people can rarely afford to be totally idle, they often schemes, such as public work programs (workfare), which have give up some form of income to join a workfare scheme. Estimates been especially effective. Two successful workfare programs are suggest that forgone income could represent as much as 50 per- the Maharashtra Employment Guarantee Scheme in India and cent of the wages paid by workfare schemes. But because the Trabajar in Argentina. employment is guaranteed, it provides major insurance benefits Launched during the severe drought of 1970-73, the to poor people. Incomes in villages where the Maharashtra scheme Maharashtra scheme expanded rapidly to reach some 500,000 operates have just half the variability of incomes in villages with- workers monthly. In a typical year the scheme provides 100 mil- out the scheme. Poor people also derive indirect gains from a work- lion person-days of employment. Argentina set up Trabajar II in the fare program if the infrastructure created by the program benefits mid-1 990s (as an expanded and reformed version of an earlier pro- them. Experience is mixed. In some cases better-off households gram) to cope with sharply rising unemployment, which reached have appropriated the assets created (not an unqualified liability, 18 percent in 1996-97 and was concentrated among poor people. since it may increase the political acceptance of the scheme by the rich, apparently the case in Maharashtra). Project selection Both programs concentrate on infrastructure projects (roads, irri- Principles of success gation schemes, embankments). Local authorities, in collaboration Workfare programs can improve their effectiveness by adhering with communities and NGOs, propose projects, which must use to several principles. labor-intensive technologies, benefit the local community, and * The wage rate should be determined by the local market wage target poor areas. for unskilled labor, not by the program's budget. If resources are insufficient to meet demand, the program should target Wage rate and self-targeting areas with a high concentration of poor people. Using additional To ensure that most participants are poor and to maintain incentives eligibility criteria should be avoided. for workers to move on to regular work when it becomes available, * Wage schedules should be gender neutral. Women can be en- programs should pay no more than the average wage for unskilled couraged to participate through suitable project selection, decen- labor. Trabajar set the wage rate at about 75 percent of average tralized work sites, and the provision of child care. monthly earnings from the main job of the poorest 10 percent of * Labor intensity should be higher than the local norm for sim- households in Greater Buenos Aires. The Maharashtra scheme ilar projects. uses the average wage rate of rural unskilled labor. Both programs * Communities should be involved in project selection to maximize have been highly successful in reaching the poorest of the poor. About the capture by the poor of indirect benefits of the infrastructure 9 of 10 Maharashtra scheme participants were living below the local created. poverty line; 4 of 5 Trabajar participants were poor by Argentine stan- * To get the most risk mitigation, the program should be avail- dards. For the poorest 5 percent of participants, program benefits able at all times, expanding automatically during crises as de- were 74 percent of their pre-program income. mand increases. Source: Jalan and Ravallion 1999c; Lipton 1998; RavaliBon 1991, 1999a. bilities while communities increased their capacity (box shocks-they can also provide capital to create or expand 8.10). Several funds are improving their poverty target- microenterprises. Microfinance thus helps households ing. In Malawi and Zambia social funds are introduc- diversifv their sources of income and reduces their vul- ing poverty mapping to identify pockets of poverty 127 nerability to income shocks. Microfinance programs have Beneficiary assessments have identified community ori- been especially important for women and households entation (responsiveness to community priorities, help- headed by women, who often have difficulty getting fulness in promoting social cohesion) as one of the credit. However, microfinance institutions, depending on strengths of social funds.'28 their size and diversification, are unlikely to be effective Microfinance (credit, savings, and insurance). Micro- against large covariant shocks (chapter 4). finance programs can help poor households smooth con- Microfinance programs have been more successful in sumption during an adverse shock. Access to credit may reaching moderately poor and vulnerable (not necessar- help them avoid distress sales of assets and replace pro- ily poor) households than extremely poor households. ductive assets destroyed in a natural disaster. But micro- Most programs reach clients just above or just below the finance programs do more than help households cope with poverty line. Efforts to direct microcredit programs ex- HELPING POOR PEOPLE MANAGE RISK 157 plicitly to poor households often fail, although there is that they do not lose access to future loans. Clients con- evidence that some programs successfully use geographic tinued to repay loans even during and after the floods in targeting to reach poor people. 129 Having appropriate local Bangladesh. 132 Evidence suggests that microcredit has es- groups identify beneficiaries or targeting beneficiaries pecially improved the lives of poor women, by strength- by size of landholdings (as in the 0.5-acre limit used by ening their bargaining position with their husbands, Grameen Bank) has proved more successful.130 boosting their self-confidence, and increasing their par- Empirical studies find that clients often use loans to re- ticipation in public life. 133 duce risk rather than to cope with shocks, meaning that The success of microfinance in reducing vulnerabil- loans are not usually "diverted" to consumption. Poor and ity through income diversification and asset accumula- nonpoor clients alike use loans to smooth consumption by tion suggests that these programs should be a priority for smoothing income flows, mainly by increasing diversifi- government and donor support.134 But expanding the cation. Loans help households accumulate a variety of as- client base to poorer households remains a challenge. To sets: physical and productive (vehicles, equipment, housing, some degree, microfinance products could be redesigned livestock), financial (savings accounts), human (educa- to reach poorer households. Loan size and repayments tion, health care), and social (contributions to funerals and could be made more flexible to better match the income weddings or to networks of mutual support).131 flows and repayment capacity of borrowers.135 There As a risk management tool, the key strength of micro- probably is a practical limit to this accommodation, since finance programs is the knowledge that loans will be at some point the increasing costs of making such loans available in time of need, making it possible for house- will undermine the sustainability of microfinance insti- holds to dispense with less effective and less desirable strate- tutions. The very poorest may well be more effectively gies (child labor, money under the mattress). There is a helped with targeted cash transfers. parallel here with employment guarantee schemes: the Program effectiveness would be increased by combining confidence in future availability is the key to the success microcredit with savings and insurance products so that cients of microfinance programs as a risk management tool. would not have to take out loans to cope with illness or deati The availability of microfinance services enables poor (box 8.1 1). Bank Rakyat Indonesia and SafeSave in households to move from reactive to proactive approaches: Bangladesh demonstrate the potential of combining they can plan to mitigate risk. Most clients, well aware of this benefit, go to great lengths to repay their loans so Box 8.11 The Self-Employed Women's Association Box 8.10 of India The Eritrean Community Development Fund Established in 1972, the Self-Employed Women's Association After the war of independence, the government of Eritrea (SEWA) is a registered trade union for women in India's infor- mal sector. SEWA's 220,000 members are hawkers, vendors, promised to provide each province with basic economic and me-bsedtorkErs, andlbers adt towits ven- social infrastructure. But many poor communities lacked the homebased workers, and iaborers In addition to its conven- capacity to implement the projects themselves. Eritrea's in- tionar sabor union functions (ensuring minimum wages and novative solution was to combine social fund and public work work security)c uEtA provides Tegal aid and operates a bank mechanisms in the Eritrean Community Development Fund. and a social security scheme The bank offers savings ac- counts and loans to members, The social security scheme, The fund combines the bottom-up selection of projects with which insures ao members, coers the top-down selection of intervention areas. Contracting which insures about 14 percent of SEWA memberst covers procedures are kept flexible to reach even communities with- heaoth life and asset insurance. Slightly more than half the cost out implementation capacity. If a community cannot form a of the insurance program is covered by premiums. The rest is project committee to supervise a project, the fund takes over financed by SEWA and a public subsidy SEWA views this procurement, contracting, and technical supervision. If nec- arrangement as a first step toward increased contributions by members and self-sustainability. The combination of banking, essary, the fund even manages the community's contribution, nuac,aduinsrieshshle.nraeSW' This flexible approach is combined with an ambitious capacity- Insurance, and union services has helped increase SEWA's building program, which trains community and local govern- planshto and heath bnefits a d a pension c on ment staff in project design, maintenance, and operation. plans to expand health benefits and add a pension component. Source: Frigenti, Harth, and Huque 1998. Source: Lund and Srinivas 1999b; Mirai Chatterjee, general secretary, SEWA, emal communication, 3 May 2000. 158 WORLD DEVELOPMENT REPORT 2000/2001 microcredit with savings. Other microfinance programs uation results suggest that the program is able to target ben- have successfully introduced life insurance, at low rates and efits to the poorest households and that it has raised the en- with limited benefits (burial costs and repayment of debts).136 rollments of children in beneficiary households (see box 5.5 Cash transfers. Cash transfers (excluding transfers in chapter 5).139 through such contributory systems as regular pensions and Fee waivers can be effective in counteracting falling unemployment insurance) include social assistance pay- school enrollment in the aftermath of a crisis or shock. ments for the elderly, child allowances, targeted human Following the crisis in Indonesia, primarv school enroll- development programs, and fee waivers for basic ser- ment of boys in the poorer areas of Jakarta fell 8.3 per- vices. In countries with large informal sectors, where cent and junior secondary enrollment fell countrywide, formal unemployment insurance is not feasible, means- with the greatest drops in poorer areas. In 1998 the In- tested social assistance is an important way of assisting donesian government abolished entrance fees for public the unemploved and underemployed. schools and lowered monthly fees and exam fees at the The role of cash transfers in a social risk management primary level, providing relief for many parents who had strategy depends on a country's income. In high-income fallen behind on fee payments as a result of the crisis. An countries cash transfers are part of social insurance, of- individual scholarship program and block grants to fering a broad guarantee of minimum income. In tran- schools, both targeted to poorer areas, supplemented the sition economies family assistance payments represented fee waivers to restore school enrollment rates.140 0.4-5.1 percent of GDP in 1992-93. Cash social assis- tance programs operate in only a few Asian countries, where they account for less than 1 percent of GDP, and are negligible in Africa and Latin America. Poor people are exposed to a wide array of risks that Cross-country experience suggests that family assistance make them vulnerable to income shocks and losses of well- and targeted social assistance are effective for reducing being. This chapter argues that helping poor people man- poverty in the short term, especially in countries with rel- age risk is thus an essential part of poverty reduction atively little poverty. The difficulty is finding an appro- programs-and should complement efforts to increase av- priate targeting mechanism compatible with the country's erage income and improve the distribution of income, administrative capacity. Decentralized solutions may be which are discussed elsewhere in this report. The focus preferable if communities have better information on who has been on risks occurring primarily at the individual, is needy (see box 8.5).13 household, and community (micro and meso) levels, Targetedhumandevelopmentprogramsforpoorhouse- such as illness and injury, crime and domestic violence, holds with children transfer income in cash or in kind on the old age, harvest failure, and fluctuations in food prices basis of such observable criteria as children's age, attendance and demand for labor. (Chapter 9 discusses macro-level in school, or participation in a health care program. They thus risks such as macroeconomic crises and natural disasters.) serve the dual objectives of poverty reduction and human de- Poor people respond to their risk exposure through di- velopment. When effective, they prevent the long-term dam- versification of assets and sources of income and various age to children that occurs when households, unable to types of self-insurance (buffer stocks, savings) and informal adequately manage risk, respond to shocks by underfeeding insurance (networks of mutual support)-all means to their children or pulling them out of school to work. reduce the risk or soften its impact. Where these pre- In the Bangladesh Food-for-Education program the emptive mechanisms prove inadequate, households cope transfer to a household of 100 kilograms of rice increased with shock by increasing or diversifying labor supply the probability of boys' schooling by 17 percent and girls' (child labor, migration), selling assets, or reducing schooling by 160 percent.138 The Brazilian Bolsa Escola pro- consumption. gram targets scholarships to regions and communities where These mechanisms work, but not well enough. Volatil- child labor is greatest, seeking to keep children in school by ity in household income remains high in many areas, compensating parents for the income children would have and many households suffer episodic declines in well- earned. The Mexican scheme Progresa provides health and being. Some recover, but not all do. Shocks common to education benefits for 2.6 million households in 2000. Eval- a large area, which can wipe out an entire network's re- HELPING POOR PEOPLE MANAGE RISK 159 sources, are most likely to overwhelm the risk management nets put in place before adverse shocks hit can serve both tools of poor households. And because shocks do not af- risk mitigation and coping purposes. fect all members of poor households equally, with women To counter the incentive and information problems thair and children frequently the most at risk, inadequate risk exclude poor people from many market-based insurance management can cause long-term harm to children mechanisms, the state has a special role in providing or reg-- through malnourishment, child labor, and loss of schooling. ulating insurance and setting up safety nets. This reporir In most developing countries today, risk management advocates a modular approach that adapts the safety net emphasizes interventions after a disaster strikes. The bal- to the specific pattern of risk in each country or area and ance needs to shift to favor policies to reduce and miti- complements existing risk management arrangements. gate risk. Health, environmental, labor market, and Many solutions will involve partnerships among poor macroeconomic policies can all reduce risk. And safety communities, the private sector, and the state. CHAPTER 9 Mlanagcing Econom'ic Cises and N~*atural Disasters There is nowhere to work. We get sick and we don't dren may rise during economic crises and natural have the money to get cured, we don't have medicines disasters. Poor households are often forced to sell because they are expensive. The government makes their meager assets at depressed prices. These re- everything expensiv... We don't have money to buy sponses perpetuate chronic poverty, possibly reducing fertiizers, seeds, everything is in dollrs. We don't future economic growth because of the irreversible have anything to eat. Everything is so expensive, losses in human and physical capital. That is why -From a discussion group of aduilt women at preventing economic crises and natural disasters is so the rime of a banking crisis, Juncal, Ecuador crucial. And that is why, when they occur, among the top priorities should be to protect poor people. Re- [Security is]. . the ability of persons to cope with quired for that protection are not only resources but disasters. also the instruments (safety net programs) to channel -From a discussion group, Little Bay, Jamaica those resources to poor households. While develop- ing countries and transition economies in general are vulnerable to crises and natural disasters, SMAl states are especially vulnerable to adverse external events cnomic crises and natural disasters can bring because of their remoteness and isolation, high degree deep and sudden collapses in national output--and sharp of openness, susceptibility to natural disasters, and lim- increases in income poverty; Together with violent con- ited diversification.1I flicts (see box 3.2 in chapter 3), they are great sources of vulnerability and inseurity. Worse, because ofdthe collateral Preventing and coping with damage they cause, such as irreversible loss of human cap- economic crises ital, they affect not only the current living standards of poor people but their ability to escape from poverty as well. Even our limited access to schools and health is nou' Malnutrition and dropout rates among poor chil- beginning to disappear. We fear for our childrens' 161 i6z WORLD DEVELOPMENT REPORT 2000/200I fiiture.. What is tbe justice ini sending our children tu the garbage site every daey to suipport thefiunilvy? external shocks (terms of trade shocks, higher international i>/s/ l',l .! si.l ljirts'cf -;; li.' li. '' . interest rates, sudden movements in capital flows as a re- sult of contagion). In such circumstances restrictive fis- cal and monetary policies are inevitable and less costly than Economywide crises entail sharply falling output, de- the alternative of delaving such measures, which could clining incomes, and rising unemployment. Pervasive in lead to a larger crash. the 1 990s, they came in different forms: fiscal crises, Once adjustment policies are accepted as inevitable, balance of payments crises, terms of trade shocks, currency the way governments introduce fiscal austerity can crises, banking crises, hyperinflation. The economic crises worsen the adverse effects on the living standards of the in Mexico in 1995, in East Asia in 1997, and in Brazil poor and near-poor. For example, removing food or and Russia in 1998 received wide media coverage. But fuel subsidies would exacerbate the effects on poor peo- they were not the only episodes of economic distress.2 ple-unless compensatory measures are taken (chapter Most crises have been brought on by varying combina- 4). So would increasing the rates and sometimes the cov- tions of policy mismanagement and such external factors erage of indirect taxes on food and other products that as terms of trade shocks, volatile capital flows, and con- figure large in the consumption basket of poor people. tagion in international capital markets. Net government transfers may decline as governments Economic crises hurt both the poor and the nonpoor, cut social assistance as part of a fiscal austerity program. but they are far more devastating for those already in Reducing the quantity and quality of public services poverty or nearly poor. even if they are not hurt dis- used by the poor and near-poor would also worsen their proportionately. The welfare losses are larger for poor situation. households and those who fall into poverty than for the But government actions can also mitigate the impact rest of the population. Poor people are unlikely to have of crises on poor people. The task of the policymaker enough savings or self-insurance to see them through is to implement the combination of macroeconomic bad times, and they have little or no access to insur- measures that results in the lowest cost in forgone out- ance schemes, whether social or market based (chap- put and affords the greatest protection to the living ter 8). standards of poor people. A key element of a poverty- An economic crisis affects the living standards of poor sensitive response is the right composition of revenue- people and those living close to poverty through differ- raising measures and fiscal cuts. A poverty-sensitive ent channels: response should also allow for the expansion of safety * Typically, real wages fall and unemployment rises, nets targeted to poor people (the "social insurance" driving down labor earnings. component of social spending) during periods of macro- * Nonlabor incomes fall as economic activity slows, economic adjustment. and the prices of the goods and services produced by poor people may fall relative to other prices. Social i7npact of cises * Private transfers, particularly from family members, are There is a strong link between macroeconomic down- likely to shrink as living standards fall across the nation. turns and rising income poverty (table 9. 1; see also fig- * The meager assets of poor people are exposed to in- ure 2. 1).3 During crises many people become temporarily flation or a collapse in prices. poor, and social indicators tend to worsen or to im- * Macroeconomic crises slow the accumulation of prove more slowly. Data suggest that the human capi- human, financial, and physical capital, weakening the tal of poor people, particularly poor children, can ability of poor people to escape poverty. deteriorate. The damage can be irreversible, affecting the Is the observed fall in incomes during crises made ability of these children to escape poverty when they reach worse by the policies to respond to the crises? The debate adulthood. on this is long-standing. That rising poverty coincides with In most countries in East Asia poverty rose as a result the policy responses does not mean that the policies of the financial crises of the late 1990s: it is estimated that caused the rise. Crises can occur because of past unsus- it rose almost 50 percent in Indonesia and that urban tainable macroeconomic policies or inability to adjust to poverty doubled in the Republic of Korea.4 In both MANAGING ECONOMIC CRISES AND NATURAL DISASTERS I63 Table 9.1 Effect of economic crises on incidence of poverty in selected countries Percent . Before Year After Country and type of crisis crisis of crisis crisis Argentina, hyperinflation and currency 25.2 47.3 33.7 (1987) (1989) (1990) Argentina, contagion 16.8 24.8 26.0 (1993) (1995) (1997) Indonesia, contagion and financial 11.3 18.9 11.7 (1996) (1998) (1999) Jordan, currency and terms of trade 3.0 .. 14.9 (1986-87) (1989) (1992) Mexico, currency and financial 36.0 .. 43.0 (1994) (1995) (1996) Russian Federation, financial 21.9 32.7 (1996) (1998) Thailand, currency and financial 11.4 12.9 a (1996) (1998) Not available. Note: Based on national poverty lines and per capita household income except for Indonesia (per capita expenditure), Mexico (household income), and Russia (household expenditure per equivalent adult). Data for Argentina refer to Greater Buenos Aires. For Indonesia poverty estimates before and during the crisis are based on the full SUSENAS (the national socioeconomic survey) conducted in February 1996 and 1999; estimates after the crisis are based on a smaller sample. Figures are not comparable across countries because poverty lines differ. a. Based on the socioeconomic survey conducted between February 1998 and January 1999, which does not fully reflect the impact of the crisis. Estimates from a smaller survey conducted during June-September 1999 put the poverty incidence at 15.9 percent. Source: Ministerio de Economia de Argentina 1998; World Bank 1994c, 1999dd; ECLAC 1999b; Lokshin and Ravallion 2000b. countries, however, poverty fell as the economies recov- rose, sometimes substantially.6 ered. In Russia the incidence of poverty rose from 21.9 The impact of economic crises on living standards is percent to 32.7 percent between 1996 and 1998. In not fully captured by measures of inequality and income every crisis in Latin America and the Caribbean the in- poverty. Economic crises are characterized by extensive cidence of poverty increased and several years later re- mobility: previously nonpoor people may fall into poverty, mained higher than it had been before the crisis. and previously poor people may escape it. Evidence of Inequality may rise, fall, or remain unchanged during sharp downward and upward mobility was found after a crisis. In Latin America inequality (as measured by the the 1998 crisis in Russia, for example.7 Mean expendi- Gini coefficient) rose in 15 of 20 crisis episodes for which tures of people classified as poor in 1996 actually rose, there are data. In East Asia during the recent crisis, how- and 42 percent of them escaped poverty after the crisis. ever, inequality remained practically unchanged, and in By contrast, 61 percent of those who were poor after the Mexico following the peso crisis in 1995 it fell. When crisis had not been poor in 1996. Put another way, 20 crises are accompanied by increases in inequality, economic percent of the population fell into poverty as a result of contractions can more than reverse previous gains in the economic downturn. Even though overall inequalitv poverty reduction. In Latin America the poverty reduc- fell and a large share of the poor escaped poverty after the tion from a 3.7 percent increase in per capita income for crisis, there was substantial downward mobility for many urban areas and a 2 percent increase for rural areas in the who were not previously poor and for some who were al- 1970s was reversed by just a 1 percent decline in per capita ready poor. Those who become poor during economic income in the 1980s.5 Even if inequality increases, the crises often have different characteristics than the chron- poorest fifth of the population is not always hurt dis- ically poor. For example, they may be better educated. A proportionately. In Latin America the income share of the study in the Philippines found that households with middle fifths of the population often fell most during the more education are more vulnerable to wage and emr- 1980s debt crisis, but the share of the top tenth always ployment shocks.8 I64 WORLD DEVELOPMENT REPORT 2000/2001 Table 9.2 Social impacts of economic crises in selected countries Main crisis indicators Health indicators Education indicators Argentina * Per capita GDP fell 4.1%. a Per capita daily protein intake * Growth in gross primary 1995 * Per capita private consumption fell 3.8% in 1995, but enrollment declined from fell 5.6%. increased 1.9% in 1996. 2.2% in 1993 to 0.8% in 1996. Mexico * Per capita GDP fell 7.8%. * Among children under age * Gross primary enrollment 1995 * Per capita private consumption 1, mortality from anemia increased 0.44% in 1994, fell 11. 1 %. increased from 6.3 deaths but fell 0.09% in 1995. per 100,000 live births in 1993 to 7.9 in 1995. * Among children ages 1-4, the mortality rate from anemia rose from 1.7 to 2.2 per 100,000. Indonesia * Per capita GDP fell 14.6%. a The share of women whose a The dropout rate for children 1998 * Per capita private consumption body mass index is below the in the poorest fourth of the fell 5.1 %. level at which risks of illness population rose from 1.3% in 1997 and death increase rose 25%. to 7.5% in 1998 for those ages * Most indicators of child 7-12 and from 14.2% to 25.5% nutritional status remained for those ages 13-19. In both stable. The exception may cohorts the poorest fifth be the weight (conditional on experienced the largest height) of children under age increase. 3, suggesting that families may m The share of children be investing in some members in the poorest fourth of the at the expense of others. population not enrolled in school rose from 4.9% in 1997 to 10.7% in 1998 for those ages 7-12 and from 42.5% to 58.4% for those ages 13-19. In both cohorts the poorest fourth had the largest increase. Note: Gross enrollment ratios are used because net ratios were not available. These data should be used with caution. Source: World Bank 1999cc; IDB Statistical and Social Database; PAHO 1998; Thomas 1999; Frankenberg, Thomas, and Beegle 1999. Most social indicators either deteriorate or improve at increase rose by a quarter in 1998, and the average weight a slower pace during a macroeconomic crisis (table 9.2). of children under age three declined. Social indicators such as infant mortality rates continued School attendance and literacy also take hits during to improve in Latin America in the 1980s, though more crises. In the Philippines secondary school enrollments slowlv than in the previous decade. But health indicators increased only 0.9 percent between the 1997/98 and more sensitive to consumption or income downturns 1998/99 academic years, after growing at an average an- worsened. In Chile the share of low-birthweight infants nual rate of 2.6 percent in the previous five years. In Mex- and undernourished children rose as the economy de- ico the proportion of each graduating class that enrolled clined. In Mexico infant and preschool mortality caused in the next education level declined during the 1980s debt by nutritional deficiency rose in the 1980s, reversing the crisis, particularly among high school and university stu- trend of the previous decade, and rose again with the eco- dents. The percentage of age-appropriate children entering nomic crisis of 1995. In Argentina and Venezuela the daily primary school also declined. In rural areas the dropout per capita intake of protein declined as per capita GDP rate rose by 40 percent. In Argentina and Mexico growth fell. In Indonesia the share of women whose body mass in gross primary enrollment slowed in 1995. A study for index is below the level at which risks of illness and death South India found that children are often taken out of MANAGING ECONOMIC CRISES AND NATURAL DISASTERS I65 school in response to adverse shocks.9 example, ending hyperinflation brought about a signif- There is no question that economic crises increase tran- icant one-time drop in the incidence of poverty: ini sitory poverty " They can also increase persistent or chronic Greater Buenos Aires the incidence of poverty dropped poverty because of hard-to-reverse effects on the human from 34.6 percent in 1989 to 22.6 percent in 1991.15 capital of poor people. While the trends cited for malnu- The 1990s saw various types of crises, triggered by trition, infant mortality, and enrollment are national av- weak banking systems and weak financial regulation in erages, they most likely reflect a deterioration in these a world of large and volatile international capital flows. indicators among poor people. For Indonesia, information Liberalizing the financial sector was expected to put by income group shows that the dropout rate in the low- economies on a more stable footing. But the transition est fourth of the income distribution rose from 1.3 per- from more repressed to more open financial systems in cent in 1997 to 7.5 percent in 1998 among children ages the developing world has been difficult to manage. 7-12 and from 14.2 percent to 25.5 percent among those Banking crises have been more numerous in the past two ages 13-19. The proportion of poor children not enrolled decades, when stroke-of-the-pen financial liberalizatio n in school increased from 4.9 percent to 10.7 percent. became popular. 16 Recent research shows a link between macroeco- Some of the reforms introduced in the financial sec- nomic downturns and education indicators. The aver- tor backfired because the institutional rules allowed ex- age annual increase in years of schooling in 18 Latin cessively risky behavior while the costs of that behavior American countries fell from 1.9 years in the 1950s and had to be paid by society as a whole. A vivid example is 1960s to 1.2 in the 1970s and 1980s. Worsening macro- the Mexican financial crisis of 1995.17 At the root of the economic conditions (short-term GDP shocks, volatil- crisis was a weak banking system, its fragility traced to ity, and adverse trade shocks) explain 80 percent of the the privatization process used for the banks, some aspects decline, according to one study." 1 As evidence from of the financial liberalization program, and weak regu- Mexico shows, the negative "income effect" of falling in- latory institutions. Rescuing the banking sector will cost come tends to outweigh the positive "price effect" of the Mexican taxpayers an amount equal to about 20 percent lower opportunity cost of attending school. 12 Simulation of GDP (in present value terms). results suggest that the gross secondary enrollment rate To prevent financial crises, governments need to imt- in Mexico would have been 11 percentage points higher prove the prudential regulation and supervision of financial in 1991 if the economy had grown during the 1980s at intermediaries, introduce new standards for data dis- half the rate of the 1970s. semination, and implement corporate bankruptcy re- form."8 These measures are already under way in many Avoiding crises developing countries, but there is still a long way to go. Clearly, avoiding crises should be a top priority in any anti- At the same time, a cautious approach should be taken poverty strategy. There is wide agreement on the kind of to capital account liberalization. Controls on capital macroeconomic and financial policies governments need inflows-such as those Chile used until recently-can he to reduce vulnerability to policy-induced crises or adverse an appropriate instrument for tempering the volatility of external shocks.13 They should avoid profligate fiscal capital flows. There is evidence that capital controls can and monetary policies, overvalued exchange rates, and un- shift the composition of capital flows toward longer- sustainable current account deficits-all problems in the maturing investments.19 1970s and 1980s. Other initiatives and measures are also important for Many parts of the world have made great progress in avoiding crises, such as mechanisms to diversify and in- steering away from irresponsible fiscal policy. Leading ex- sure against risk. Some governments, such as Chile, self- amples are the large economies in Latin America and some insure using fiscal stabilization funds. Others, such aLs of the transition economies, where the ensuing fall in in- Argentina, negotiate contingent credit lines between the flation rates has helped build investor confidence and re- central bank and private international financial institu.- duced, if not eliminated, the potential long-term effects tions to ensure access to foreign currency in the event of of inflation on efficiency and growth. 14 Lower inflation a sudden slowdown in capital inflows.20 has also helped reduce poverty, since high inflation often However, actions at the national level may not be hurts the poor more than the nonpoor. In Argentina, for enough to prevent economywide crises. Domestic actiorns i66 WORLD DEVELOPMENT REPORT 2000/2001 will have to be complemented by actions at the interna- omy crashed, with per capita GNP falling by 13.4 per- tional level to foster global financial stability (chapter 10) cent and real wages by 40.6 percent. Altogether, real and help countries, particularly the poorest and the small- wages fell by 67 percent between 1988 and 1990.21 est, manage commodity price shocks. The 1997 crisis in Thailand shows what happens when there are no corrective measures to address the buildup of Formulating a crisis response that protects vulnerability.22 True, the financial panic of domestic and poor people international investors suddenly concerned about the fate No matter how skillful the economic management, crises of their portfolios lit the fuse for the explosion. But the are likely to affect the developing world and transition buildup of structural vulnerabilities provided the economies for some time to come. That is why articu- dynamite-sharply rising short-term debt that far ex- lating a response to crises must take into consideration ceeded international reserves, a financial sector that had done its impact on poor people. A poverty-sensitive response a poor job of intermediating capital inflows and found it- to crisis should steer toward: self saddled with hugely mismatched assets and liabilities, * Helping poor households maintain their consumption. and corporations that were massively overleveraged and ex- * Ensuring that poor people do not lose whatever ac- posed to interest and exchange rate fluctuations. cess they have to basic social services. Not all problems arise from a failure to adjust to an * Preventing permanent reversals in the accumulation adverse shock or from unsound macroeconomic policies. of human and physical capital. In some cases the policy response errs in the direction of * Averting self-defeating behavior, such as criminal ac- too much adjustment, with fiscal and monetary policy tivity, prostitution, and exploitative forms of child more restrictive than necessary to restore equilibrium in labor. the currency market, the current account, or the capital A poverty-sensitive response should also provide mech- account. Overreaction can cause more pain than neces- anisms for those at risk of becoming poor as a result of sary and in some circumstances can be self-defeating. An the crisis. initial overreaction on the fiscal front can lead to a higher What does it take to protect those who are already poor fiscal deficit down the road because the larger-than- and those at risk of becoming so from sharp declines in expected recession lowers government revenues, defeat- short-term income? Appropriate macroeconomic re- ing the purpose of the initial austerity measures. The sponses and well-functioning safety nets can enhance reason for overshooting often is that cautious policy- equity and result in better growth outcomes. Some of the makers prefer to err on the side of excessive adjustment, recommendations here are already being incorporated in since timid adjustment can be far more devastating. the standard approach for dealing with crises. The Re- Although it may be hard to tell whether a policy pack- public of Korea, for example, introduced or expanded age is excessively restrictive, there are some indications safety nets relatively quickly in the wake of the 1997 fi- that those in place in East Asia during the recent crisis were nancial crisis (box 9.1). But in general the response con- just that. In Thailand the tax increase in September 1997 tinues to be ad hoc-with measures thrown together in made the ensuing recession worse. In Korea the restrictive the heat of a crisis. fiscal policy initially made room for the expected costs of Adopting the right macroeconomic policy mix. Re- bank restructuring. But the fiscal target was subsequently sponding with the right macroeconomic policy mix after relaxed as both the authorities and the international financial an adverse shock is one of the biggest challenges policy- institutions recognized that it was unrealistic in light of the makers face. Driven by political considerations, policy- larger-than-expected slowdown in growth. Aiming toward makers may postpone needed adjustment and stabilization the original target in the face of worsening economic con- measures because they are painful-making the situation ditions would have been self-defeating. And for Malaysia far worse. Peru was an extreme case in the 1980s. The gov- and the Philippines the trend of cyclically adjusted deficits ernment refused to implement an adjustment program (for both revenues and expenditures) suggests that they did and in July 1985 announced a cap on external debt pay- not relax their fiscal policy, even though the actual deficit ments (a de facto unilateral moratorium) equal to 10 per- made it look as though they had. cent of exports. Peru did well for a while, but the Even if excessively restrictive policies are later cor- disequilibria continued to mount and in 1988 the econ- rected, the short-term costs can be significant, particu- MANAGING ECONOMIC CRISES AND NATURAL DISASTERS I67 Box 9.1 Providing social protection in response to crisis in the Republic of Korea The Republic of Korea was making sustained progress in reduc- in January 1998 to 174,000 in March 1999, still only 10 per- ing poverty in the 1990s: the urban poverty rate fell an average cent of the unemployed workforce. 20 percent a year during 1990-97, and there were no increases * Public work. Since most of Korea's jobless did not benefit from in inequality. But then economic crisis struck, sharply increasing the expansion of unemployment insurance, the government in- unemployment and poverty. The incidence of poverty in urban areas troduced a temporary public work program in May 1998, en- doubled from 9 percent in 1997 to 19.2 percent in 1998.1 Unem- rolling 76,000 workers. By January 1999 the program was ployment rose from 2.6 percent in the second quarter of 1997 to providing 437,000 jobs, though the number of applicants was a peak of 8.7 percent in early 1999. Real wages declined 20.7 per- higher still, at 650,000. By the first quarter of 1999 the public cent. Most of the newly unemployed were low-paid workers: in work program was benefiting around 2.5 times as many peo- December 1998 three-quarters were temporary, daily, self- ple as the unemployment insurance program. employed, or unpaid family workers, and about 20 percent were * Livelihoodprotection. In May 1998 the government introduced the head of a household with no other income earners. a temporary livelihood protection program, with funding to Expansionary fiscal policies in 1998 and 1999 were critical in cover 750,000 beneficiaries. It also introduced a means-tested stemming the economic downturn. Social protection spending noncontributory social pension for 600,000 elderly people. was increased threefold-from 0.6 percent in 1997 to 2.0 per- Although the government's social protection response was quite cent in 1999. The government used three main instruments of exemplary, public spending on health and education did not increase social protection to help the unemployed, the poor, and the in line with the overall budget, and real spending either fell or re- elderly: mained constant. But even within the smaller envelope for health, * Unemployment insurance. Korea expanded its nascent un- spending on primary care was protected. employment insurance program-the only such program The government is now focusing on consolidating social safety among the East Asian crisis countries-from firms with more nets, reducing income disparities, and creating the basis for a com- than 30 employees to all firms. It also included temporary and petitive and knowledge-based economy. Policies to achieve these daily workers, shortened the contribution period required for objectives include a law guaranteeing a minimum standard of liv- eligibility, and extended the duration of unemployment bene- ing, to take effect in October 2000. The law will entitle all Koreans fits. This expanded the eligible workforce from 5.7 million living underthepovertylinetoreceiveincomesupportforliving,ed- workers at the beginning of 1998 to 8.7 million at the end of ucation, and housing expenses. Nearly 2 million poor people are ex- the year. Beneficiaries increased tenfold-from around 18,000 pected to benefit, four times the current number. 1. The poverty rates were calculated using seasonally adjusted expenditure data and a national poverty line equivalent to about $8 a day (in 1993 PPP dollars). Source: World Bank 1999w, 2000d. larly for poor people. If there are vicious cycles of poverty, higher for poor people. Moreover, because poor people live low education, and poor health, a recession can cause per- close to the subsistence level, their preference may reflect manent damage for the poor. application of the safety principle (minimizing the prob- Do the macroeconomic responses to crises that are best ability that their income will fall below a certain level). Or for the overall output levels of the economy differ from the poor and nonpoor may discount future consumption those that would be best for the incomes of poor people? differently, with the poor putting a larger premium on pre- Perhaps. Different policy combinations imply different sent consumption than the nonpoor. costs for the poor than for the nonpoor because of the The distributive and intertemporal implications of al- way the reduction in per capita output is distributed.23 ternative adjustment policies are important, but policy- But even if distributive outcomes were the same, the makers rarely have the luxury ofchoosing among different poor and the nonpoor could well prefer different policy adjustment paths. In general, the optimal combination of packages.24 policies-to achieve the necessary balance of payments acl- Poor people are more likely to prefer an adjustment that justment with the smallest decline in output-depends leads to the smallest drop in GDP at any point in time on initial conditions.25 When a currency is under speCL- even if it implies a slower recovery. Nonpoor people are lative attack, a spike in interest rates will in most cases be more likely to prefer a program that reduces income more needed to stop the attack. But when a country intro- severely in the short run but yields higher growth in the duces adjustment measures early on, the government may medium run. This difference results simply from the fact have more freedom to choose among different policy that the welfare losses from an economic downturn are combinations and thus be more likely to manage a so.ft 168 WORLD DEVELOPMENT REPORT 2000/2001 landing. Unfortunately, macroeconomic analysis in its to poor people should be protected if not expanded. current state can offer little guidance in assessing the dis- It may seem obvious that governments should protect tributive and intertemporal implications of alternative spending that benefits poor people and expand the safety policy packages, clearly an area in need of far more ana- net programs targeted to them. But this does not neces- lyrical and applied research. sarily happen in practice. Recent research in some coun- Protecting spending that benefits poor people. How gov- tries in Latin America has found that a 1 percent decline ernments raise revenues and cut public (nondebt) spend- in per capita GDP leads to an estimated 2-3 percent de- ing has important policy implications for who bears the cline in targeted public spending per poor person.27 And burden of adjustment and whether poor people are pro- a study on the Argentine employment program Trabajar tected.26 To design a poverty-sensitive fiscal adjustment found that its performance in reaching poor people de- to avoid or respond to a crisis, policymakers need to as- teriorated sharply with cuts to its budget.28 sess the distributional effects of spending programs. A use- There may be several reasons for such "antipoor" pat- ful tool for this is the public expenditure review (box 9.2). terns in fiscal adjustment. Without budgetary guidelines As a general rule, areas important for poor people- to direct fiscal austerity, governments may go for pro- basic education, preventive health care, water and sani- portional cuts to minimize bureaucratic infighting and ease tation, rural infrastructure-should be protected from acceptance by the legislature. Another reason may be that budget cuts to ensure that services are adequate. That governments lack the instruments to target resources to means ensuring that schools and health posts in poor areas the poor-instruments that are difficult to put in place have at least the basic minimum of supplies. General sub- in the heat of a crisis. Even if the instruments exist, po- sidies on food staples might need to be maintained in the titical forces may be such that the resources going to poor short run-even if the benefits leak to the nonpoor- people are cut more than proportionately. In some coun- unless they can be effectively replaced by targeted pro- tries information can be the major constraint: governments grams. Safety nets and social assistance programs targeted may lack reliable records of their budget or programs. What can be done to counter these factors? One way Box 9.2 to protect spending that benefits poor households is for Public expenditure reviews to assess the impact the government and legislature to rank current programs of fiscal retrenchment on poor people by their importance as part of the budget approval process. When spending cuts are needed, the order of the cuts Public expenditure reviews-assessments of public sector would be determined bv the priority assigned to each pro- issues that focus on the efficiency and rationale of the public budget-could be useful tools for evaluating the impact of gram. Government agencies could be required to evalu- fiscal adjustment programs and public sector reforms on ate social programs to help policymakers identify those social programs and safety nets. In economywide crises that are most cost-effective in reducing poverty and there- that lead to spending cuts, these reviews could help estab- fore should be protected during a crisis. lish a transparent budget mechanism for rationalizing, allo- Peru has introduced guidelines for protecting pro- cating, executing, and managing public spending to protect poor people and ensure private sector efficiency. grams that benefit poor people as part of its public finance Public expenditure reviews typically analyze and project reform law (box 9.3). The guidelines combine fiscal rules public revenues and determine the level and composition of with measures to increase fiscal transparency and ac- public spending, assessing the allocation of resources among countabilitv. T and within sectors. When planning fiscal retrenchment, a he program creates a stablizaton fund with short review should be done, focusing on the sectors that ac- the proviso that programs benefiting poor people should count forthe bulk of the public budget (agriculture, education, be protected. Although such budget protocols may not health, infrastructure). The review should rank expenditures be classified as antipoverty programs, they can have an on social programs, considering the tradeoff between these impor rant effect on overt b rotecting o-poor programs and other nonessential spending (such as military p e y u g prp spending) that could be minimized during a crisis. This type spending during fiscal retrenchment. of adjustment is clearly more efficient in protecting vulnera- If benefits targeted to poor people are cut for politi- ble groups and maintaining private sector efficiency than the cal economy reasons, a third party-such as the multi- typical uniform spending cut. lateral lending organizations-could advocate for the Source: World Bank 1999v. poor and help governments implementing austerity mea- sures design a viable way to protect programs and spend- MANAGING ECONOMIC CRISES AND NATURAL DISASTERS i6-9 ing that benefit the poor. This happened to some degree One way to avoid this political economy constraint is to in several countries in the 1 990s. design programs with low marginal benefits or high mar- Changes in the incentive system embedded in targeted ginal costs for the nonpoor.29 programs could also facilitate cuts for nonpoor benefi- Evaluating different types of spending can be difficult ciaries during periods of austerity. The argument is this: when data are poor, the case in most developing coun- it is often said that for political economy reasons some tries. Efficiency indicators are almost nonexistent, and data of the benefits of targeted programs have to go to the non- on actual spending, as opposed to budgeted amounts, are poor-through "leakage"-to ensure continuing sup- available only after long lags. Usually an evaluation should port for programs. The same forces will presumably act take the available intermediate information and com- to limit the welfare losses to the nonpoor from cuts. plement it to determine whether public resources reach the intended beneficiaries effectively. A social monitor- Box 9.3 ing and early response unit, such as the one set up in In- Protecting poor people during fiscal adjustment: donesia during its recent crisis, can help ensure quick and Peru's Fiscal Prudence and Transparency Law reliable information for evaluating spending in specific social programs.30 Where field surveys are infeasible (be- Peru's Fiscal Prudence and Transparency Law, overwhelm- cause of budget or time constraints), recent household sur- ingly approved by the national congress in 1999, does much b to ensure that social protection is maintained during a fiscal veys can e used to try to determie an efficent and adjustment. rational allocation of government resources among social First, the law established fiscal rules on the maximum an- programs and safety nets. nual deficit of the consolidated public sector, capping it at 2 Puttting safety nets in place before a crisis. If the problern percent of GDP in 2000, 1.5 percent in 2001, and 1 percent is a lack of instruments for protecting poor people, the so- thereafter. (The consolidated public sector includes the cen- tral and regional governments, decentralized agencies, and lution is to introduce, during normal times, safety net pro- national public enterprises; it excludes local governments and grams that can operate as insurance in times of economic their agencies and enterprises.) In the event of a national distress. Safety nets are important for several reasons. emergency, international crisis, or fall in GDP, the fiscal deficit can increase to 2 percent of GDP. The law also set Theycanplayacrucal rolei mitigatigtheeffectsofcrses limits on increases in public spending and debt. The maxi- on the poor and protecting the near-poor from falling into mum annual growth of nonfinancial public spending is equiv- poverty. A study estimated that if the targeted programn alent to the inflation rate plus 2 percentage points, implying Progresa (see box 5.5 in chapter 5) had existed when the a future reduction in the relative size of the public sector. 1995 crisis hit Mexico, the poverty gap index in rural areas Second, the law created a fiscal stabilization fund, to be and he s ed povery gap index m gives greas funded from three sources: the revenues above the average and the squared povery gap index (which gives greater collected during the previous three years, three-fourths of weight to the poverty of the poorest)31 would have declined future privatization proceeds, and half of all revenues from fu- by 17 percent and 23 percent in the year after the crisis.-52 ture concessions. (Savings accumulated in the fund in excess Safety nets can also help prevent irreversible damage to tlhe of 3 percent of GDP will be transferred to the public pension . . fund or used to reduce public debt.) Up to 40 percent of the human capital of poor people. And they can aid poltical fund's resources can be used in a given year if current rev- acceptance of stabilization and reforms, preventing con- enues fall below the average collected over the previous flicts over resource distribution that can create stalemates, three years. Fund resources can also be used in emergencies, deepen economic crises, even cause governments to fall. such as an economic crisis or a natural disaster. Third, the law mandates that the fund's spending on tar- Recent work has shown that istitutional weaknesses, i- geted poverty reduction programs be given priority over cluding lack of safety nets, have been responsible for spending on other programs. many crises over the past 25 years.33 To enhance fiscal transparency, the law introduced a Most developing countries lack effective safety nets that three-year fiscal framework to be developed, approved, and published by the government. And to improve fiscal ac- protect poor people from the output, employment, and countability, it requires that the finance minister submit to price risks associated with systemic adverse shocks. When congress and publish annual reports assessing the execution these mechanisms are not in place before a crisis occurs, of the fiscal goals in the multiyear framework. policymakers are often forced to improvise or to use pro- Source: Ruprah 1999. grams designed for other purposes and other beneficia- ries. Emergency responses to emergency situations are often prepared without technical analysis to identify the 170 WORLD DEVELOPMENT REPORT 2000/2001 groups most vulnerable to the shocks and to evaluate the ures are, the fiill human and economic costs are even cost-effectiveness of different social protection options. greater. Human costs include injuries and temporary and Programs put in place and operating-even on a small permanent disabilities, temporary and permanent dis- scale-before crises hit do better at protecting poor peo- placement of people, the breakup of families and social ple than ad hoc emergency measures. networks, increased poverty and disease, and psycholog- To be effective, safety nets should include a wide range ical scars. Economic costs, based largely on direct physi- of programs-public work programs, scholarships for poor cal impacts or losses of fixed capital and inventory, are also children, cash transfers, food-related transfers, food subsi- underestimated. Many indirect and secondary effects on dies, social funds, and fee waivers for essential services economic activity-such as changes in fiscal policies, the (chapter 8). Social programs that focus on long-term de- long-term consequences of the reallocation of investment velopment (for example, such targeted human development resources, or the losses in human capital-go unrecorded. programs as Mexico's Progresa) can also perform a safety net Over the past 10 years the incidence of natural disas- function during economic downturns. The appropriate ters has increased.36 This could be due in part to social fac- mix of safety net programs will depend on the characteris- tors, as settlements have sprung up in hazardous areas. The tics of the poor and vulnerable, the type of crisis, and the urban poor in megacities-for example, in Rio de Janeiro government's institutional and administrative capacity. and its favelas-are often forced to build on steep, mar- The international community can play an important ginal land prone to landslides that kill or leave homeless part by providing policy advice, contributing financial sup- thousands of people every year. But there are also natural port, and helping policymakers design and fund safety factors. The El Nifio events, associated with anomalous nets. International financial institutions can help coun- floods, droughts, and storms, are getting larger and more tries design pro-poor fiscal adjustment programs and frequent.37And warming of the surface of the Atlantic is safety nets and, for countries too poor to fund a safety increasing the frequency and severity of hurricanes.38 net during a crisis, can provide financing.34 Still, it is often asked whether it would be more correct to label many of these disasters as "human-made" rather Reducing vulnerability to natural than "natural." They are probably both. disasters Inpact of natural disasters on poor countries The biggest shoek we ever had was Hurricane Gilbert; . .. and poor people all what we found after Gilbert was one wooden chair. - ll Knnau 'ThYbw,R jbJunZ..; Z AUnfortunatelyfir mne, the land on which I made myyfarm was a swampy area and when it rained the whole farm Economic development is repeatedly interrupted by nat- submerged with water and was destroyed. ural disasters-by earthquakes, droughts, floods, land- -F---Haol'ywa una, At1/nsu Bokr, (7/7an/a slides, volcanic eruptions, windstorms, forest fires. Like economic crises, natural disasters can cause sharp in- Developing countries, especially their most densely pop- creases in poverty and slow the pace of human develop- ulated regions, suffer the brunt of natural disasters. Be- ment. And like economic crises, they hurt poor people tween 1990 and 1998, 94 percent of the world's 568 major in the short run and diminish their chances of escaping natural disasters and more than 97 percent of all natural poverty in the longer run. disaster-related deaths were in developing countries (fig- The damage to agriculture and infrastructure varies by ure 9.1). In Bangladesh alone three storms, four floods, type and intensity of natural disasters, as do the impli- one tsunami, and two cyclones killed more than 400,000 cations for their indirect and secondary impacts. Droughts, people and affected another 42 million. In southern for example, can result in heavy crop and livestock losses Africa in 1991-92, Malawi, South Africa, Zambia, and while leaving infrastructure and productive capacity Zimbabwe experienced severe droughts.39 In Latin Amer- largely unaffected. ica and the Caribbean major natural disasters associated Between 1988 and 1997 natural disasters claimed an with El Nifio, Hurricane Mitch, Hurricane Georges, and estimated 50,000 lives a year and caused damage valued the Quindio earthquake in Colombia claimed thousands at more than $60 billion a year.35 Dramatic as these fig- of lives and caused billions of dollars of damage between MANAGING ECONOMIC CRISES AND NATURAL DISASTERS 171 Figure 9.1 Developing countries bore the brunt of natural disasters in 1990-98 U Percent 100 80 Disaster-related deaths 80 40 20 Major disasters East Asia Europe and Latin America Middle East South Asia Sub-Saharan Developing High-income and Pacific Central Asia and the and North Africa Africa countries countries Caribbean Note: A disaster is classified as major if it caused more than 50 deaths or affected more than 100,000 people. Source: USAID, OFDA 1999. 1995 and 1998.4° In 1998 severe flooding of the Yangtze ity of infrastructure in poor communities increases their River caused devastation in China, and a large earthquake vulnerability. occurred in Armenia. Another long series of disasters While natural disasters hurt everyone affected by struck in 1999-a major earthquake in Turkey, a cy- them, poor families are hit particularly hard because in- clone in Orissa, India, floods in central Vietnam, torrential jury, disability, and loss of life directly affect their main rains and catastrophic mudslides in parts of Venezuela, asset, their labor. Disasters also destroy poor households' floods in Mozambique. The list goes on. natural, physical, and social assets, and disrupt social as- Poverty and lagging development amplify the ad- sistance programs.45 Long-term disabilities and the de- verse effects of natural disasters. Developing countries struction of assets can trap families in chronic poverty. are particularly vulnerable, because they have limited ca- Malnutrition impairs children's ability to learn. pacity to prevent and absorb these effects. People in The few studies that have analyzed the impact of nait- low-income countries are four times as likely as people ural disasters on poverty show that the harm to current in high-income countries to die in a natural disaster.41 and future living standards can be significant. In Ecuador Despite similar patterns of natural disasters in Peru and El Nifio may have increased the incidence of poverty in Japan, fatalities average 2,900 a year in Peru but just 63 affected areas by more than 10 percentage points.46 In in Japan.42 Average costs as a proportion of GDP are 20 Honduras Hurricane Mitch caused an estimated 7 per- percent higher in developing countries than in industrial cent decline in agricultural output in 1998.47 Loss of crops economies.43 was extensive, affecting a quarter to a half of households. Poor people and poor communities are frequently Rural households, most dependent on agriculture, lost the the primary victims of natural disasters, in part because most.48 they are priced out of the more disaster-proof areas and In the 1984 drought in Burkina Faso the income of live in crowded, makeshift houses.44 The incidence of dis- the poorest third of the rural population fell 50 percent asters tends to be higher in poor communities, which are in the Sahelian zone, the poorest agroclimate, and 7 per- more likely to be in areas vulnerable to bad weather or cent in the Sudanian zone.49 There was also evidence that seismic activity. And there is evidence that the low qual- poor people sold livestock out of desperation. Because they I72 WORLD DEVELOPMENT REPORT 200012001 had very small stocks of animals to begin with, these dis- The need to replace damaged infrastructure in disaster- tress sales may have dangerously depleted their buffer stricken countries diverts government resources from stocks, leaving them extremely vulnerable to future longer-term development objectives and consumes a sig- drought and other shocks and possibly trapping them per- nificant share of multinational lending resources. In manently in dire poverty.50 Mexico as much as 30 percent of the funds approved by Studies of the impact of the 1994-95 drought in Zim- the World Bank for improving rural water supply over babwe found that women and young children were the the past decade have been diverted to postdisaster most affected. For women, the drought's effect on health rehabilitation.58 (as measured by body mass) was temporary. With good rains the following year, they regained much of the lost Risk reduction aizd mitigation: body mass. But for children ages 12-24 months the lessening vulnerability to disasters drought will probably have a permanent effect. These Cumulative experience with natural disasters points to an young children lost an average 1.5-2.0 centimeters of urgent need to move from fatalism to prevention, from linear growth in the aftermath of the drought. The im- response to preparation, from mobilizing resources after pact was most severe among children in households with the fact to reducing and transferring risk before the fact. little livestock, the principal asset of these households for There is a distinct difference in approach to emergency smoothing consumption.51 The drought had no impact management between many developing and developed on men's health. countries. Developing countries emphasize preparedness On balance, female-headed households fare worse and response-making sure that the resources to respond than male-headed households following a natural disas- to emergencies are available and ready for dispatch and ter, in part because of their smaller average resource then that they are dispatched quickly and used efficiently base.52 Customary or formal laws can make this worse. after an emergency has occurred. Developed countries in- Among the Tonga of Zambia, for example, a widow has creasingly emphasize reducing or mitigating the impacts no entitlement to any of the household's possessions.53 of disasters (box 9.4). The effect of a natural disaster on poverty can go well Disaster reduction and mitigation can lessen the beyond the households directly affected. Research on disruption caused by natural disasters, save lives, and Sub-Saharan Africa suggests that both agricultural and protect property. From a purely economic point of overall GDP are sensitive to downward fluctuations in view, investing in risk reduction pays off. For example, rainfall. The 1991-92 drought in southern Africa slowed a cost-benefit analysis for eight cities in the Argentina growth in agricultural and total output in Malawi, South Flood Rehabilitation Project found an internal economic Africa, Zambia, and Zimbabwe.54The impact of drought rate of return of 35 percent. The estimated $187 mil- shocks on GDP and the recovery time depend in part on lion (1993 dollars) in avoided damages from the 1997 the economic importance of the agricultural sector and flood more than covered the $153 million in investment. its integration and links with industry. The second-round By installing flood control dams and improving and subsequent effects are more pronounced in more in- drainage, the Rio Flood Reconstruction and Prevention tegrated economies. In Senegal and Zimbabwe the effect Project reduced total floodable areas by 40 percent, of droughts spilled over from agriculture to manufac- achieving an estimated 6.5 benefit-cost ratio for seven turing.55 The value of Zimbabwe's manufacturing out- subbasins of the Iguacu and Sarapui Rivers.59 Com- put declined 9.5 percent in 1992, largely as a result of the prehensive disaster risk management can be integrated 1991-92 drought, and export receipts from manufactures into development investment decisions. In Turkey in- declined 6 percent.56 ternational lenders and donors worked with the gov- The destruction of infrastructure by catastrophic nat- ernment to develop a new disaster management ural disasters also has both immediate implications and framework in the aftermath of the 1999 earthquakes longer-term, second-round poverty effects. In Asia, for (box 9.5). example, where 70 percent of the world's floods occur, Resettlement-tailored to the needs of poor people- the average annual cost of floods over the past decade was is often the appropriate risk reduction strategy in flood- estimated at $15 billion, with infrastructure losses ac- prone or volcanic areas. Where resettlement is not feasible counting for 65 percent.57 or desirable, neighborhood improvement programs are an MANAGING ECONOMIC CRISES AND NATURAL DISASTERS 173 Box 9.4 Box 9.5 Mitigation is the cornerstone of emergency Mitigating the risks of natural catastrophes: management in the United States lessons from the 1999 earthquakes in Turkey Mitigation-the ongoing effort to lessen the impact disasters A powerful earthquake shook northwestern Turkey in the have on people and property-is the cornerstone of emergency summer of 1999, killing more than 17,000 people, injuring management in the United States. It involves keeping homes tens of thousands, and razing several population centers. away from floodplains, engineering bridges to withstand Three months later a second quake hit, raising the number earthquakes, creating and enforcing effective building codes of victims and the social and economic losses. Industry and to protect property from hurricanes, and more. businesses in the areas hit by the quakes had contributed Over the past 10 years the U.S. Federal Emergency Man- more than 35 percent of the country's GDP. Their destruc- agement Agency (FEMA) has spent $25 billion to help peo- tion is likely to affect growth in Turkey for many years to come. ple repair and rebuild communities after natural disasters. The international community assisted Turkey in relief Other government agencies and insurance companies have and immediate recovery efforts. In partnership with the Turk- responded with billions of dollars more. Beyond this, the costs ish government, the European Investment Bank, the Coun- of emergencies also include lost lives, jobs, and business op- cil of Europe's Social Development Fund, and other donors, portunities. A big emergency can reduce local GDP by as the World Bank coordinated the preparation of a framework much as 10 percent. for a $1.7 billion reconstruction program. A crucial part of the In 1995 the high and escalating costs of emergencies led framework is a disaster management and response system FEMA to adopt a national mitigation strategy, with two goals: to prevent similar losses in the future. to protect people and structures from disasters and to min- Disaster and land development laws will be reviewed and imize the costs of disaster response and recovery. FEMA es- modified, and the capacity of municipalities to regulate, timates that every dollar spent on mitigation saves two in plan, and implement disaster-resistant development will be response and recovery. strengthened. Pilot projects in selected municipalities will The strategy promotes a community-based approach to help planning and building departments develop risk-based reducing vulnerability to natural hazards: municipal master plans, means for effective implementation I Altering the hazard (seeding clouds during a drought). of building codes, municipal regulations to ensure that I Averting the hazard (building dams to control floodwaters). builders follow appropriate licensing procedures, and pro- I Avoiding the hazard (moving parts of communities out of grams for evaluating existing buildings. floodplains). The government's earthquake insurance program will 3 Adapting to the hazard (constructing earthquake-proof expand its catastrophic risk management and risk transfer buildings). capabilities. The program will create an insurance mecha- In February 2000 FEMA announced Project Impact: Buil- nism to make funds readily available to owners (those pay- ding Disaster-Resistant Communities, a project to provide ex- ing real estate taxes) who need to repair or replace a pertise and technical assistance to about 200 communities dwelling destroyed or damaged by an earthquake. It will also striving to become disaster resistant. Three principles drive ensure the financial solvency of the insurance pool after all the project: Preventive actions must be decided at the local but the most catastrophic events and reduce the govern- level. Private sector participation is vital. Long-term efforts ment's financial dependence on donors following major and investments in prevention are essential. earthquakes. Source: Olsson 2000. Source: Kreimer 1999. alternative. In these programs residents of low-income Low-cost local initiatives can also reduce the vulner- urban areas improve their houses themselves or with com- ability of communities' income to natural disasters. In munity help. The programs reduce building code viola- rural areas such initiatives might focus on environmen- tions by training informal sector construction workers in tal conservation and reforestation. For places prone to mitigation techniques and by providing finance for low- droughts and floods, community food banks can help. cost improvements that bring housing to stipulated stan- In Burkina Faso local cereal banks were introduced to dards. International assistance, channeled through local improve storage, lower food prices, and stabilize themn NGOs, has often helped turn housing reconstruction ef- over the year, including during the drought season.60 forts into low-cost opportunities for mitigating risks in fu- Community agricultural cooperatives can help sma'll ture disasters (box 9.6). Other important neighborhood farmers obtain credit or crop insurance. And various upgrading activities include constructing drainage works strategies can help diversify the economic activities and reducing the risk of flooding and mudslides. within a community. I74 WORLD DEVELOPMENT REPORT 2000/2001 Box 9.6 Box 9.7 Turning reconstruction into risk mitigation with Mitigating risk with catastrophe bonds the help of a local NGO Catastrophe bonds-or cat bonds-offer an alternative to in- In a poor area of Peru partly destroyed by an earthquake in surance in countries lacking active private insurance markets. 1990, Caritas, a local NGO, initiated a reconstruction program A before-the-fact risk transfer mechanism, cat bonds provide that was also designed to mitigate earthquake-related risks. financial protection against disaster losses. After consulting with the community, Caritas decided to Consider a government that wants protection against construct housing from quincha, a local material capable of the risk of flood damage to one of its water treatment plants withstanding earthquakes. To directly assist the neediest fam- in the next year. Experts estimate the chance of a flood at ilies, such as households headed by women, Caritas provided 1 in 100, a risk low enough to induce an institutional in- materials in exchange for participation in communal work. An vestor to purchase a cat bond whose payoff is tied to flood earthquake in 1991 showed the advantage of using quincha: damage to the treatment plant. The investor buys the bond most houses resisted the earthquake, which registered 6.2 at the beginning of the risk period at par. At the end of the on the Richter scale. risk period the investor loses the entire principal if the water treatment plant is damaged. But if no damage occurs, the Source: Schilderman 1993. investor recovers the principal plus interest, normally above the market rate to reflect the risk of losing the principal. The government invests the funds, which will be used Reducing economic vulnerability also involves only if a catastrophe occurs, in risk-free securities. The cost encouraging-or mandating-the purchase of private to the government is equal to the difference between the interest rate it receives from the risk-free securities and the insurance for those who can afford it and identifying interest rate it pays the bondholder-a cost analogous to pay- mechanisms for transferring risk, such as catastrophic rein- ing an insurance premium. The value of the bond-and the surance and catastrophe bonds (box 9.7). While risk government's interest payments-would be lower if the transfer mechanisms can efficiently cover much of the cost government flood-proofs the treatment plant. So, besides per- forming an insurance function, the cat bond gives the gov- of repairing and rebuilding infrastructure, freeing up ernment an incentive to invest in mitigation efforts. scarce government resources, they may not be easy to apply A potential problem with catastrophe bonds is the diffi- in poor countries. For one thing, they require systems for culty of verifying damage. The public agency operating the verifving damage that cannot be easily manipulated by water treatment plant might exaggerate damage to ensure that the bondholder pays. One way to deal with this moral those (governments, for example) who would collect the hazard is to tie payouts to an objective index (such as flood insurance benefits. To deal with this problem in floods, height) rather than to actual damage. for example, a country could establish a high-quality measuring and ieporting system. This would facilitate in- Source: Kunreuther 1999. surance contracts that link payment schedules to a rain- fall index.61 of human capital and perhaps to provide shelter for dis- Coping with natural disasters placed people. Cash transfers to poor families reduce the In the emergency phase following a disaster, efforts should likelihood that they will need to pull their children out focus on providing food, water, shelter, and medicine. That of school. Where children need to participate in recov- makes temporary repair of such infrastructure as roads and ery efforts, schools can adopt flexible schedules. water supply critical. Priorities need to be based on the Following a widespread natural disaster, national magnitude ofdamages and level ofvulnerability. The most and local governments need to establish a macroeco- vulnerable groups-women, children, and the elderly- nomic management scheme to tackle fiscal and current need special attention. Involving women in the man- account effects-lower tax revenues and higher public agement of shelters, establishing workfare programs spending, lower exports and higher imports. A calamity adapted to women's needs, and ensuring gender neu- fund like that in Mexico can improve governments' trality in housing acquisition can improve recovery for ability to cover the costs of coping with natural disas- women and households headed by women. Expanding ters (box 9.8). Calamity funds should focus on ab- early childhood development programs, particularly sorbing the catastrophic risks that cannot be absorbed mother and child feeding programs, is also important. Re- by third parties, such as disaster-related damage affect- building schools should be a top priority-to avoid loss ing farmers and urban dwellers unable to afford private MANAGING ECONOMIC CRISES AND NATURAL DISASTERS I75 Box 9.8 insurance, and providing social assistance to poor dis- Sharing the costs of catastrophes: the Mexican aster victims. fund for natural disasters In the aftermath of a natural disaster targeted inter- national assistance can help maintain macroeconomic sta- With tremendous diversity in geography and ciimate, Mex- bility, accelerate recovery, and protect poor people. But ico is susceptible to a wide range of natural disasters- . e floods, droughts, earthquakes, wildfires, tropical cyclones, to enable countries to accept financal assistance, inter- volcanic eruptions. Since 1980 direct damage from natural national financial institutions may have to relax some ad- disasters has totaled some $6.5 billion, and about 7,000 justment targets during crises. In Zambia in 1992-93 the people have lost their lives. tight public spending policy being implemented as part In 1996, to help reduce the country's vulnerability to tf pdjubi spendin poic beinghimplemented raisp natural disasters, the government established Fonden o adjustment restricted the governments ablity to raise (Fondo para desastres naturales, or Fund for Natural Dis- external financing because of the lack of counterpart re- asters). This federal fund was to be financier of last resort sources in the local currency. As a result, international for emergency response equipment, disaster relief activ- funds for drought relief were not fully used.62 ities, and reconstruction of public infrastructure and pro- The reconstruction period provides an opportunity to tected areas. In 1998, following a period of particularly heavy losses from reduce vulnerability to natural events (see box 9.6). Targeted natural disasters, the government decided to use Fonden assistance locally for the most affected populations (with the more strategically, to provide incentives for insurance use and poor the top priority) and consultation with affected com- disaster mitigation. After broad consultation with stake- munities and households should be a key strategy Includ- holders, in March 1999 the government changed Fonden's ing local people in reconstruction activities can foster operating guidelines to: * Increase clarity and transparency in the decision rules for leadership and promote solidarity, helping reduce the psy- granting access to the fund and in loss assessment chological trauma caused by natural disasters (box 9.9).63 processes. Countries that have social or rural infrastructure in- * Limit moral hazard by encouraging greater use of private insurance by Fonden's beneficiaries and establishing vestment funds can use them to channel resources effi- clear cost-sharing formulas for financing disaster losses ciently.64 Given their experience in building infrastructure falling under the responsibility of state and municipal and providing social services to communities, these funds governments. can quickly identify local spending priorities and help min- * Encourage mitigation in the reconstruction programs fi- prioriIesthelp mu- nanced by Fonden and in beneficiaries' regular investment Imlze corruption. The Honduras Social Investment Fund programs. performed this role in the aftermath of Hurricane Mitch. * Refinance disaster response activities initially financed It financed the opening of several important secondary through emergency liquidity facilities to speed disaster roads, began the rehabilitation of some water systems in recovery. secondary towns and undertook cleanup activities. The These changes are being formalized through voluntary s agreements between the federal government and the state fund rapidly put in place a decentralized operating struc- governments that set out the parties' rights and responsi- ture and responded to the pressing needs of local munic- bilities, Fonden's rules, and agreed cost-sharing formulas for ipalities and communities. Procedures established by the disaster relief and reconstruction activities. The agreements fund immediately after the disaster simplified project will also lead to the establishment of trusts between the fed- eral government and each state. Under the terms of each preparation and authorization and expedited contracting trust, spending decisions and contracting of eligible emer- and disbursement of funds. Close monitoring by local res- gency activities will be carried out by a technical committee idents of private contractor equipment and employmenr consisting of state and municipal representatives, acting on of local residents in cleanup activities helped ensure that advice from federal entities. If successful, these measures will increase trans- funds were well accounted for.65 parency, accountability, and efficiency in the use of The experience produced valuable lessons on how to Fonden's resources and redistribute the costs of natural improve the capacity of such funds to cope with disaster: disasters between government and the private sector. providing adequate financial resources to meet post- Over time they will also reduce the share of costs borne g q by the federal government for mitigating and coping with disaster needs, ensuring that ivestment projects go be- disasters. yond rehabilitation and cleanup to include expansion of existing facilities and new construction, and expanding Source: Barham 2000. the fund's mandate to allow direct social assistance to vul-- nerable people.66 I76 WORLD DEVELOPMENT REPORT 2000/2001 Box 9.9 Involving communities in postdisaster reconstruction: lessons from the Maharashtra Emergency Earthquake Rehabilitation Program On 30 September 1993 an earthquake struck the Indian state of effective tool for dealing with problems that arise during Maharashtra, killing some 8,000 people and damaging 230,000 implementation. houses in Latur, Osmanabad, and 11 other districts. With the help Participation also had a positive psychological effect on com- of the World Bank, the government of Maharashtra created the munities. Involving local people in the reconstruction helped them Maharashtra Emergency Earthquake Rehabilitation Program. The overcome the trauma caused by the earthquake. Recognizing this, program institutionalized community participation and formal con- the government began reconstruction in small villages even before sultation with beneficiaries at all stages. the rehabilitation program began, appealing to donors, corporations, The program divided communities into two categories: those NGOs, and religious organizations to "adopt" villages for recon- that needed to be relocated-the 52 villages that sustained the struction. Some organizations also worked on social issues, such worst damage-and those that needed to be reconstructed, re- as schooling for children. paired, or strengthened. The Tata Institute of Social Sciences Information on the program, its processes, and the mecha- worked in the 52 relocation villages, which had some 28,000 fam- nisms for redress was accessible-and awareness was high. The ilies. The Society for Promotion of Area Resource Centers orga- participatory process opened many informal channels of commu- nized community participation in the 1,500 villages-with some nication between the people and the government, helping to nar- 190,000 families-in which rebuilding or repair was to take place. row the gap between them. Beneficiaries learned of their Over time the program became a people's project. As re- entitlements and worked hard to secure them. People who felt that sults materialized, community participation received greater ac- their grievances were not addressed appropriately in the village or ceptance. Initially skeptical, officials in the project management taluka (an administrative unit that includes several villages) could take unit later came to acknowledge community participation as an them to the district authorities and the government in Mumbai. Source: Vatsa 1999. Workfare programs can usefully be introduced or ex- * * panded in disaster areas in conjunction with reconstruc- tion operations, providing a livelihood to people who Large adverse shocks-economic crises and natural can no longer support themselves (chapter 8). They can disasters-cause poor people to suffer not only in the short also help people affected by the less visible impacts of a run. They undercut the ability of the poor to move out of disaster, such as the poor fishers in Ecuador and Peru who poverty in the long run as well, by depleting their human fell deeper into poverty as fish fled the waters warmed by and physical assets. Particularly harmful are the effects on El Nifio. In Northeast Brazil the program Frente de Tra- poor children, who may suffer irreversible damage if a cri- balho (Work Front) provided similar employment op- sis or natural disaster increases malnutrition or forces them portunities in periods of drought. During the 1979-84 to drop out ofschool. Integral to any poverty reduction strat- drought it employed up to 3 million workers in con- egy should thus be measures to prevent and manage economic struction and drought-related jobs.67 Public work programs crises and natural disasters-and to establish safety nets, with that build social or community infrastructure or help in ensured financing, to help poor people cope when these ad- cleanup and reconstruction can also be a good option. verse shocks do occur. PART V International 0 Actions Ft~~~~~~~~~ t ~ ~~ .- 4 CHAPTER 10 Harnessing Global Forces for Poor People T hroughout this report we have seen that poli- in chapter 11. Other global forces that affect the poor cies and institutions at the country and local level are include international labor migration, commodity the keys to enhancing the opportunity, empowerment, price volatility, global warming and environmental and security of poor people. But the lives of poor peo- degradation, promotion of political and human rights, ple are also affected by forces originating outside their and the international arms sales and trade in illicit gems countries' borders-global trade, capital flows, offi- that spur or prolong conflict in countries. Several of cial development assistance, technological advance, dis- these were discussed in last year's World Development eases, and conflicts, to name just a few. Actions at the Report. global level are therefore crucial complements to country-level actions. They can accelerate poverty re- Expanding market access duction and help narrow the gaps-in income, health, in high-income countries and other dimensions-between rich countries and poor. At first glance, it seems that rich countries benefit This chapter discusses four key areas of international more from the opportunities of the global economy. action for poverty reduction: After all, they have averaged faster growth than poor * Expanding market access in rich countries for de- countries over the past 40 years. But it is also true that veloping countries' goods and services. poor countries that are more integrated with interna- * Reducing the risk of economic crises. tional markets have grown as fast as or faster than * Encouraging the production of international pub- rich countries.' As chapter 3 detailed, trade can pro- lic goods that benefit poor people. vide a powerful engine for growth and poverty re- * Ensuring a voice for poor countries and poor peo- duction. It has also been argued that trade with richer ple in global forums. countries can speed the process of "catch-up."2 Also important for poverty reduction is development Expanding access to rich country markets can thus cooperation-foreign aid and debt relief-discussed do much to help poor countries grow faster and to 179 I80 WORLD DEVELOPMENT REPORT 2000/2001 Figure 10.1 cent of the official development assistance given to de- High-income countries protect manufacturing veloping countries in 1998.6 This is a serious setback to and agriculture development efforts in poor countries. Taffs on mohancr ifnported by high-income countries,19In general, trade reforms in poor countries have Percent failed to deliver their full benefits because they have not 20 been matched by reforms in rich countries. For man- ufactured goods (including food products), which now account for almost three-quarters of developing coun- 15 try exports, tariffs facing developing country exports Developing to high-income countries are, on average, four times country those facing industrial countrv exports to the same 10 exports maret _ _ ~~~~~market. income High-income countries' tariffs are not only higher for country World manufactures from developing countries, they also es- 5 exports exports calate with the level of processing. For example, in Japan and the European Union fully processed food o _ ' * * 1 F * products face tariffs twice as high as those on products Manufactures Agricultural products in the first stage of processing. In Canada the ratio is Source: Hertel and Martin 1999a. even higher, with tariffs on fully processed food prod- ucts 12 times those on products in the first stage. This escalation can discourage industrialization efforts in reduce poverty in the developing world. This is particu- developing countries. larlt so for agricultural products, since more than two- Developed countries' trade barriers can place signif- thirds of the developing world's poor people live in rural icant constraints on poor countries' efforts to grow. areas. Not only do foreign markets represent important Finding ways to unblock the political obstacles to re- sources of demand for developing countries' agricultural moving such barriers would do much to aid poverty re- goods-because the demand for basic food products is duction in the developing world. By some estimates the inelastic-but exporting can expand nonfarm employ- welfare losses for high-income countries from their own ment and stimulate the entire rural economy. Agricultural distortionary trade policies are large-$63 billion a year exports have been shown to be a strong determinant of for agricultural distortions alone. It should be feasible overall agricultural growth.3 So it is disturbing that while to put in place compensatory mechanisms for the rela- world trade in manufactured products expanded at 5.8 tively small-but politically powerful-groups of pro- percent a year from 1985 to 1994, agricultural trade ducers as part of an agreement to lower trade barriers. grew at only 1.8 percent. But more than anything, reducing trade barriers will re- One reason for this slow growth is the continuing pro- quire real political will on the part of the leaders of de- tection of agricultural products by developed countries- veloped countries. Special priority should go to reducing protection not only through tariffs and quotas but also the scope and scale of protection on agricultural goods, through export subsidies.4 The tariffs that high-income labor-intensive manufactures, and services. countries impose on agricultural goods from developing countries, especially such staples as meat, sugar, and Reducing the risk of economic crises dairy products, are almost five times those on manu- factures (figure 10. 1). The European Union's tariffs on As chapter 9 details, economic crises in developing coun- meat products peak at 826 percent.5 These barriers are tries can be devastating for poor people. So creating the huge obstacles for developing countries striving to break conditions for macroeconomic stability is essential for en- into export markets. High-income countries' agricultural hancing the security of the poor and avoiding reversals tariffs and other distortions, such as subsidies, have been in poverty reduction. estimated to cause annual welfare losses of $19.8 billion Countries can take measures on their own to reduce for developing countries-equivalent to about 40 per- the risk of macroeconomic crises (chapter 9). Among the HARNESSING GLOBAL FORCES FOR POOR PEOPLE I81 most important are sound macroeconomic policies and government and perhaps creating a significant fiscal bur- adequate prudential regulation and supervision of finan- den, even large reserves are likely to be inadequate in some cial institutions. But even if a country follows such poli- situations. An alternative is to impose higher liquidity re- cies, it can still be hit by contagion and by waves of panic quirements on the banking sector, effectively shifting or herd behavior in world capital markets. A premium must the burden of holding reserves to the private sector (and therefore be placed on ensuring stability in the interna- possibly making banks safer, with beneficial long-term ef- tional economy, particularly in the financial sector. fects). Another is to contract with an institution for a con- International efforts to achieve stability, intense dur- tingent credit line. Both private banks and the IMF offer ing the Asian crisis, have tapered off as the crisis eased. such arrangements, which provide varying degrees of One focus has been to create and enforce international automatic access to credit at predetermined interest rates. standards for financial data dissemination and financial Even if these short-term safeguards are put in place practices. The goal is to ensure that financial markets and countries will often be unable to withstand serious in- the public have timely and reliable data for making ternational volatility. That is why priority must be placecd decisions-and to ensure that financial institutions run on increasing the momentum for international systemic effectively. Toward this end, the International Monetary financial reforms that promote stability and ensure the Fund (IMF) has developed standards on financial data availability of liquidity for countries facing severe adverse dissemination, financial sector soundness, and fiscal, shocks or hit by economywide crises. monetary, and financial transparency. Other standard- setting bodies are working on bankruptcy, corporate gov- Producing pro-poor international ernance, securities market regulation, and accounting public goods and auditing. But efforts have stalled in other areas. For example, Many of the challenges facing poor countries have solu- there has been little progress in setting up early warning tions that involve the production of international pub- devices that could alert the international community to lic goods. One important characteristic of public goods danger.8 Efforts have been similarly unproductive in de- is the difficulty of restricting people from consuming the m signing clear guidelines for private sector involvement in without paying-free riding-once they are produced. crisis prevention and resolution, which can limit moral This characteristic means that if production of public hazard, strengthen market discipline by fostering better goods were left to the market, there would be an under-- risk assessment, and improve the prospects for both supply unless the government stepped in to produce the debtors and creditors in debt workouts. There is a risk goods or to provide incentives (such as subsidies) for that an apparent lack of urgency in the aftermath of the their production. Governments have long intervened in Asian recovery could lead to inaction-but history teaches this way, providing such national public goods as defense, that more crises are a real possibility. infrastructure, law and order, and rules and standards. Recognizing this, developing countries may wish to The problem is more complex for international pub- implement short-term safeguards to limit their expo- lic goods, such as control of communicable diseases or sure.9 These safeguards are of two types: controls on research to raise yields in agriculture. Just as for national capital flows and measures to enhance liquidity. Controls public goods, the incentives-for countries or for the pri- on capital-including Chilean-type taxes on inflows, vate sector-to produce international public goods are quantitative controls on the banking sector's interna- weak or absent. But there is no world government to help tional short-term liabilities, and restrictions on capital spur the production of these goods-countries must de- outflows-have their problems, ranging from evasion to cide to cooperate to produce them. Today, as international. implementation difficulties and opportunistic imposition. problems grow more pressing, attention is focusing on. They can also restrict a country's access to much-needed how this cooperation can be achieved. 10 capital. But each type of control can be effective in some Indeed, international cooperation has had some re- situations in dampening the volatility of capital flows, thus markable successes in producing and spreading public helping to prevent crises. goods. The green revolution-one of the 20th century's One way of enhancing a country's liquidity is to main- most important development advances-was an out- tain higher reserves. But besides being expensive for the come of international research on high-yielding plant I82 WORLD DEVELOPMENT REPORT 2000/200I Box 10.1 Figure 10.2 A success story: the fight against river blindness The burden of HIV/AIDS is heavily concentrated in Africa in Sub-Saharan Africa Share of adults infected with HIV, 1997 The international effort to control river blindness (oncho- Percent cerciasis) is one of the most successful programs in the his- 8 tory of development cooperation. A painful and debilitating disease caused by a parasitic worm, river blindness has been 7 virtually eliminated in the 11 West African countries included in the Onchocerciasis Control Program. Before the program 6 began in 1974, more than a million people were infected with 5 the disease, suffering from itching, disfigurement, eye le- sions, and, for 100,000 of them, blindness. When the program 4 winds down in 2002, after a 28-year effort to eliminate the black flies that carry the parasite, 34 million people will be pro- 3 tected, 600,000 cases of blindness will have been prevented, and 5 million years of productive labor will have been saved. 2 Partners in the program have included African govern- ments, local communities, international organizations, bilat- 1 eral donors, corporations, foundations, and NGOs. A key *_*_ _ _ _ contributor has been the Merck Corporation, which has dis- 0 Low-income Middle-incoe High-income Sub-Saharan tributed the drug ivermectin free of charge. countries countries countries Africa While the program has been highly successful, on- chocerciasis remains a problem in countries outside the pro- Source: World Bank 2000s. gram area. So in 1996 the African Program for Onchocerciasis Control was created, extending the effort to control river blind- ness to the 19 remaining African countries where it is en- demic. Seventy development partners participate in this Controlling infectious diseases project. The potential benefits of international cooperation to con- trol infectious diseases are exemplified in the AIDS pan- Source: World Bank lwww.worldbank.org/gper). demic. More than 34 million people worldwide are infected with HIV, and more than 18 million have died of AIDS.'3 The epidemic continues largely unabated: 5.4 varieties at institutes established around the world expressly million people were infected with HIV in 1999, and to develop technologies to relieve the world's food prob- some 15,000 are infected every day. AIDS has no cure- lem. More recently, international cooperation in the nor is there yet a preventive vaccine. More than 90 per- campaign against river blindness in Africa brought tremen- cent of the infections are in the developing world, nearly dous benefits to 11 poor countries (box 10.1). Another 70 percent in Sub-Saharan Africa (figure 10.2). Despite success story is the Montreal Protocol on ozone deple- its concentration in developing countries, AIDS poses a tion: 165 parties to the protocol agreed to full phaseout risk to all countries-not only through its health effects of 94 ozone-depleting substances. but also through its destabilizing economic and social ef- Still, international public goods have received relatively fects. 4 The United States this year classified AIDS as a little attention in international cooperation."l And there national security risk. have been failures-the Kyoto Protocol on greenhouse While preventive behavior is the key to controlling the gases that contribute to global warming, for example, has epidemic, an effective vaccine could help dramatically. 15 languished. 2 Given the potential that some public goods But progress in developing a vaccine has been slow. More hold for poverty reduction, more attention to ensuring than 25 candidate vaccines have been tested, but only one their provision is warranted. The benefits of such goods, is in large-scale efficacy trials in humans. There are two main and the difficulty of creating the right incentives for reasons for the slow progress. The first is scientific: the cor- their production, are well illustrated by the attempts to relates of HIV immunity are unknown, and many differ- control infectious diseases and boost agricultural yields- ent approaches will probably have to be tested in parallel, two international public goods that would do much to with little certainty about their effectiveness. This raises the help poor people. There are many others, as well. up-front costs and risks for investors in AIDS vaccine de- HARNESSING GLOBAL FORCES FOR POOR PEOPLE I8,3 velopment. The second reason is economic: investors Vaccine Initiative, an international nonprofit that stim- would likely take the risks associated with research if de- ulates investment in and demand for AIDS vaccines for mand were sufficient, but there are too few market incentives global use. The initiative works with the public and pri- to invest in an AIDS vaccine that would be effective and vate sectors on targeted support to research and devel- affordable in developing countries. Africa, for example, ac- opment for novel vaccine approaches and on measures counts for only 1 percent of world drug sales. to reduce obstacles to private investment. Donor gov- The result is that international investment in research ernments, for their part, could provide tax breaks or sub- and development for an AIDS vaccine is quite low- sidies for product development relevant to poor countries. $300-350 million a year."6 Of this, $50-120 million is Second, the international community could demon- estimated to come from the private sector, which has a strate or ensure a substantial future market in developing crucial role in converting research to product development countries for vaccines. It could pledge to fully implement and distribution. And most of the research focuses on a programs for the childhood vaccines already on the mar- vaccine that could be marketed in North America and ket (immunization rates in many countries have slipped Western Europe. Only about $ 10-25 million is spent an- in the past decade). To ensure a large market for vaccines nually on development of a vaccine for the virus subtypes in poor countries, it could create a fund or other credible and health systems of developing countries. 17 In contrast, precommitment mechanism for purchasing, for the poor- more than $2 billion is spent each year on research and est countries, many doses of vaccines shown to be both development for AIDS treatment, much of it in the pri- effective and affordable.21 Prices should cover not just pro- vate sector, driven primarily by the market represented duction costs but some of the research costs as well. Mul- by the 3 million people with HIV/AIDS in industrial tilateral development banks might also issue contingent countries. loans for vaccine purchase to developing countries, to be What is true for AIDS is true for other diseases as well. released once a vaccine is developed. Similar arrange- The World Health Organization estimates that only 10 per- ments could be put in place for other medical advances. cent of the $50-60 billion in health research worldwide each year goes for the diseases that afflict 90 percent of the Boosting agricultural yields world's people.18 Developing countries account for only Like advances in medical research, advances in agricul- about 8 percent of world spending on research and de- tural technology can have profound effects on the lives velopment, mainly because they lack resources.19 Of the of poor people (box 10.2). The green revolution is among 1,233 new medicines patented between 1975 and 1997, the most famous examples of an international public only 13 (1 percent) were for tropical diseases. The effect good used for development. The revolution began when of the research and spending gaps is devastating: malaria, foundations, governments, and NGOs took the lead in tuberculosis, and AIDS cause 5 million deaths a year- trying to transfer to farmers in developing countries what about 9 percent of all deaths in the world-most of them scientists already knew about plant genetics and new in developing countries. Even when medical remedies high-yielding varieties of grains. Private companies had exist, countries may not be able to afford them. Despite shown little interest because of the difficulty of making an effective vaccine, hepatitis B still kills some 92,000 an adequate return on investments in new varieties-- people a year, and chronic hepatitis B contributes to an- farmers could simply collect seeds from the original other 700,000 deaths through cirrhosis and liver cancer.20 plants. Complementary public efforts at the national About 350 million people are chronically infected hepatitis- level were essential. Many developing countries (such as B carriers, able to transmit the disease for many years. Brazil and India) established national agricultural re- The international community could accelerate progress search organizations to develop second-generation mod- on vaccines in two ways. First, international organizations ern varieties better suited to local conditions. They also and national governments could "push" research and de- set up agricultural extension services to disseminate the velopment by subsidizing or reducing the costs of vac- knowledge to farmers and get feedback on the new va- cine development and strengthening the capacity of rieties and cultivation techniques. developing countries with a strong scientific base to be These efforts had a dramatic effect on the lives of the partners in vaccine research. For example, in 1996 the rural poor. In Africa the adoption of improved maize raised Rockefeller Foundation launched the International AIDS yields an estimated 12-14 percent, with gains as high as I84 WORLD DEVELOPMENT REPORT 2000/2OOI Box 10.2 (an example of such traits is increased nutritional qual- Research, maize, and pigs in rural Guizhou ity, as in vitamin A rice). More research is needed on the potential benefits and risks of specific uses of biotech- Anyone who doubts the impact of agricultural research on nology in developing countries. But it is likely that farm income and household food security (and thus poverty) biotechnology, if steered by the right policies, including should visit rural areas in Guizhou, the poorest province of biosafety measures could be a key part of the solution China. In remote villages, on small farms set in the moun- be tainous countryside, there has been an almost miraculous turn- to the problems of food security and poverty. around in the lives of poor people thanks to the introduction So far, however, biotechnology has had little impact of quality protein maize. in most developing countries. Unlike the advances of the Until recently annual incomes were less than $50 per capita, and for up to three months a year families had virtu- green revoluton, much of the progress i biotechnology ally no food. Then hybrids were introduced in Guizhou in 1994. has been concentrated in the private sector. Govern- Quality protein maize is higher yielding than conventional va- ment funding of agricultural research, so crucial in the rieties, but more important, it has higher levels of two es- green revolution, has stagnated or even declined, a ca- sential amino acids vital for the growth of children. Today the sualty of general fiscal restraint and a more skeptical view local people are better fed, and surplus maize has been used to produce pork, increasing food security and dispos- of the social benefits of investing in science (despite the able incomes. The extra income has been used for yield- high returns on agricultural research).27 Private institu- enhancing investments such as irrigation. tions now hold a majority of the patents in biotechnol- Having transformed the lives of 25,000 families in Guizhou, cultivation of the hybrid variety of maize is being adapted to neighboring provinces. 10.3). Because the knowledge is private, the cost of ac- quiring it is much greater. Figuring out how to allow de- Source: Bale 1999. veloping countries to capitalize on advances in biotechnology research remains a key challenge for pol- icymakers concerned with food security and poverty. Part 40 percent reported in areas with favorable conditions.22 of the answer may lie in how intellectual property rights A survey in southern India concluded that the average real are used. income of small farmers rose 90 percent in 1973-94 and that of the landless-among the poorest in farming Safeguarding the interests ofpoor people communities-125 percent.23 Higher productivity also in the intellectualpropeerty rights regime brought lower prices. It has been estimated that wheat Intellectual property rights are important for encouraging prices would have risen 34 percent more in 1970-95 with- innovation, particularly in such areas as medicine and out the international agricultural research efforts-rice agriculture. When creators of knowledge do not retain ex- prices, 41 percent more. And because of the lower prices, clusive rights of ownership for a period of time, there is far 1.5-2 percent fewer children in developing countries less incentive to produce new knowledge. This was one of are malnourished.24 the arguments for laying down standards under the Agree- Despite these advances, the growth rate of cereal ment on Trade-Related Aspects of Intellectual Property yields in developing countries has been declining steadily, Rights (TRIPS), negotiated in the Uruguay Round of from 2.9 percent a year in 1967-82 to 1.8 percent in trade negotiations in 1986-94. But intellectual property 1982-94. With demand for foodgrains in developing rights can sometimes prevent the distribution of potential countries predicted to increase 59 percent in the next international public goods helpful to poor countries, which 25 years, the challenge for agriculture remains signifi- can seldom afford the prices charged by patent owners.28 cant, particularly if yield growth is to be environmen- Three trends in intellectual property rights are partic- tally sustainable.25 ularly worrying to developing countries. The first is that One tvpe of technology that might make a significant basic research and knowledge are increasingly being gen- difference is biotechnology-using living organisms to erated by private companies alone. The second is that in- make or modify products to improve plants and ani- dustrial countries continue to account for the vast majority mals. With far greater speed and accuracy than conven- of patents worldwide-97 percent.29 Only 31 of the tional technology, biotechnology can identify desirable 26,088 applications for patents filed in 1997 under the traits and introduce them into plant and animal strains auspices of the African Intellectual Property Organization HARNESSING GLOBAL FORCES FOR POOR PEOPLE I85 Box 10.3 Developing countries have responded to these trends Most biotechnology patents are private by proposing safeguards for the intellectual property rights regime. Among them: The public sector is often instrumental in pioneering biotech- * Recognizing the rights of farmers cultivating traditional nology research, later transferring it to private firms. That pat- varieties. tern is evident in the utility and plant patents directly involving insect toxicity of the Bacillus thuringiensis (Bt) microorgan- * Prohibiting the patentng of lfe forms or biological ism. Until 1987 the public sector held the majority of the processes. patents. Since then the ownership of patents in force (whose * Reconciling World Trade Organization (WTO) provi - overall number has increased) has shifted dramatically toward sions on intellectual property rights with the Interna- the private sector (see figure). Patents are now particularly concentrated in the "big 6, " the six large corporations actively tional Conventon on Biodiversiy and the International consolidating their global positions in agricultural biotechnology Undertaking on Plant Genetic Resources. research, intellectual property, and markets (Dow, Novartis, * Ensuring access to essential medicines at reasonable cost. Aventis, Monsanto, AstraZeneca, and DuPont). Negotiating a new intellectual property rights regime that encourages private innovation while safeguarding the Holdings of biotechnology patents have shifted interests of poor countries and poor people in the benefits sharply toward the private sector of that innovation will take time and much debate. Like Share of Bacillus thuringiensis patents by the production of all international public goods, it will type of holder cetn atcpto Percent require creating incentives for participation by all those 100 * Universities with an interest in the outcome, including the private and public sector. institutionssetr 80 El Independent biotechnology Ensuring a voice for poor people 60 firms and i lblfrm individuals in global forums n Other corporate 40 entities Actions with a global reach are generally discussed in global _ Big 6 and international forums, such as nation groups, inter- 20 * d d national organizations, and United Nations conferences r0 * [ [ and other gatherings. Ensuring that poor countries, and 1987 1994 1999 especially poor people in these countries, have a strong voice in these forums will help ensure that these institu- tions respond to the needs of poor people. Productive partnerships-whether to agree on standards, produce public goods, or work toward other common goals-- were from residents of Africa. And only 7 of 25,731 ap- require that all partners have an effective voice. plications registered that year by the African Regional Industrial Property Organization were filed by residents.30 Strengthening the capacity ofpoor countries The third trend is that genetic science-enabling to represent their interests companies to patent such innovations as recombinant Not all partnerships should be global-because not all DNA techniques, monoclonal antibodies, and new cell international problems are global. Solutions to an in- and tissue technologies-is gaining primacy. This raises ternational problem-such as river blindness or pollu- a concern that a system of property rights designed to pro- tion in a lake bordering two countries-should be guided tect industrial machinery may not be able to cope fairly primarily by the countries affected.32 If those countries and effectively with the complexities of genetically ma- need assistance, financial or otherwise, the assistance nipulated organisms.31 In some cases breeders of plant va- should go to the smallest relevant group-for example, rieties protected by patents can prevent farmers from the Economic Community of West African States for reusing harvested seed. And if broadly written, patents cross-border problems involving only its member states. on biotechnology processes such as research tools can deter This principle of subsidiarity can be applied all the way invention in other fields using the same processes. up the geographic scale of international public goods, but i86 WORLD DEVELOPMENT REPORT 2000/2001 it must also be reconciled with economies of scale and ronment. Half the representatives on its council are from scope.33 developing countries (box 10.4). Subsidiarity implies that regional institutions should In addition to participating in discussions and solu- be significantly strengthened to handle cross-border prob- tions, developing countries must be able to represent lems. Given the importance of ownership, such institu- their own interests well-and this requires capacity tions would in many cases be a better choice for solving building. For example, poor countries are at a signifi- local problems than such global institutions as the World cant disadvantage in WTO negotiations on such issues Bank and the United Nations. And because most re- as labor, the environment, and intellectual property gional institutions lack wide-ranging expertise, sector- rights. Why? Negotiating in the WTO is a continuous specific organizations should also be strengthened to process, involving as many as 45 meetings or more a week assist when needed. by one estimate. Yet only two-thirds of developing But many problems are global, and participation by countries even have offices in Geneva, including only developing countries in finding solutions is just as im- 12 of the 29 least developed WTO members, and these portant as for regional problems. Since international in- offices frequently must represent the country at other stitutions will generally facilitate the discussions of international organizations as well. Moreover, devel- global problems, these institutions need to take the lead oping country officials often lack the expertise to par- in making information available, ensuring all parties a ticipate in the increasingly technical trade debates. It has seat at the table, and strengthening countries' capacity been estimated that almost 60 percent of the develop- to analyze issues and effectively communicate their ing country members of the WTO are handicapped in interests. their participation.35 Because knowledge is essential to decisionmaking, One attempt to address such problems is the Integrated international organizations must place a premium on Framework for Trade-Related Assistance to Least Devel- transparency in information and in their operations. In oped Countries, which seeks to enhance the trade-related addition to publishing as much information as possible, assistance provided by the six participating international they need to ensure independent evaluation of their ac- agencies and other development partners.36 Despite tions-to make themselves more accountable and more "needs assessments" submitted by 40 poor countries, effective. This is the direction in which international or- progress has been slow, with new donor projects in just ganizations have been moving in the past few years. one country (Uganda). Developing countries have ex- Even with all the right information, developing coun- pressed disappointment with the limited financial tries cannot represent their interests without a seat at the pledges.37 At the request of donors, an independent re- table. Many global decisions continue to be made mainly view is being conducted with the hope that the pro- by the group of seven largest industrial democracies (the gram's weaknesses can be corrected. If the problems can G-7). Mechanisms are needed to ensure that developing be resolved, the program could be a model for capacity countries contribute effectively to those decisions.34 Bet- building in' other areas to help developing countries rep- ter progress has been made in discussions about the in- resent their interests. ternational financial architecture. In 1999 the Group of 20 was established to conduct ongoing discussions on pre- Building global networks ofpoor people's venting and managing systemic financial crises. Seven de- organizations veloping countries (Argentina, Brazil, China, India, the Republic of Korea, Mexico, and South Africa) are part At last those above will hear us. Before now, no one ever of this group. Still, the arrangement lacks formal provi- asked us what we think. sions for including any of the poorest or smallest coun- -Poor man, Guatemala tries, which, though not yet integrated enough into the global economy to present a risk of starting systemic Like the voices of poor countries, the voices of organi- crises, can certainly be affected by them. A better model zations of poor people are essential in ensuring that for integrating developing countries into global problem global actions are targeted toward poverty reduction. solving is the Global Environment Facility, which works Such organizations, particularly when linked up in global to foster international cooperation to protect the envi- coalitions amassing strength and capacity, can have a HARNESSING GLOBAL FORCES FOR POOR PEOPLE 187 Box 10.4 The Global Environment Facility: a model for developing country participation -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The Global Environment Facility (GEF) is a financial mechanism for and implement them through executing agencies. They partner with fostering international cooperation and action to protect the global a wide variety of organizations to execute the projects, including environment. Through grants and concessional financing, it funds government agencies, other international organizations, private in- the additional costs incurred when a national, regional, or global stitutions, and international, national, and local nongovernmental development project also addresses environmental concerns re- and civil society organizations. lated to biological diversity, climate change, international waters, Each participating country has a political focal point-the and depletion of the earth's ozone layer. Efforts to stem land contact point with the GEF Secretariat and other participating degradation are also eligible for funding. countries-and an operational focal point, which identifies project The GEF was started in 1991, and after a trial period was cap- ideas that meet country priorities and ensures that GEF propos- italized by 34 nations (including 13 developing countries) at $2 bil- als are consistent with them. These organizations help to ensure lion for four years. In 1998, 36 countries donated a total of $2.75 country ownership, as do the 16 regional NGOs that disseminate billion to keep the facility running until 2002. Its governing struc- information and provide coordination between national and local ture ensures representation by all stakeholders. The GEF as- NGOs and the GEF. sembly, with representatives from all 165 participating countries, A recent independent evaluation of the GEF found that in a short meets every three years to review general policfes. The GEF time and with few resources, it had performed effectively in cre- council, with representatives from 32 countries (16 developing, 14 ating new institutional arrangements and approaches and in lever- developed, and 2 transition economies), meets every six months aging cofinancing for GEF projects. It has also had a positive on operational policies and programs. The GEF Secretariat trans- impact on policies and programs in recipient countries, Although lates the decisions of the assembly and council into action. there is room for improvement, particularly in efforts to main- The GEF's three implementing agencies-the United Nations stream attention to the environment, the evaluators concluded that Development Programme, the United Nations Environment Pro- the GEF had potential for much greater success and that donors gramme, and the World Bank-develop projects for GEF funding should strengthen it. Source: Porter and others 1998. major influence on international debates. For example, countries and concerned about the adverse impact of a coalition of the Jubilee 2000 movement and other globalization on the livelihoods of poor women in the in- groups concerned with debt reduction worked closely with formal economy. HomeNet's objective was international international financial institutions and industrial coun- recognition of the rights of home-based workers, em- try governments to forge a consensus for deeper, faster, bodied in an International Labour Organization (ILO) and broader debt relief for heavily indebted poor coun- convention. That convention was ratified by the ILO in tries (chapter 1 1). 1996, thanks in part to an alliance of researchers at Har- Innovative solutions are needed to increase poor peo- vard University and the United Nations Development ple's connections to each other and to global decision- Fund for Women (UNIFEM), who compiled statistics makers. The most important shift needed is in the for HomeNet to make the informal economy visible. In mind-set of global actors-to be directly informed by the 1997 the alliance of grassroots organizations, researchers, experiences of poor men and women who will be affected and international organizations gave birth to WIEGO by or are expected to benefit from global actions. Also crit- (Women in Informal Employment: Globalizing and Or- ical is information technology, which can help build net- ganizing), a global network to promote better statistics, works to channel the voices of the poor to global research, and policy in support of poor women in the in- decisionmakers. With the right tools and organization, formal economy. HomeNet, with active member orga- these networks can be powerful in spurring the integra- nizations in more than 25 countries, publishes a newsletter tion of poor people's priorities and analyses into global that reaches organizations in more than 130 countries. discussions. Strengthening such networks will fortify a much- One such global network of poor people is HomeNet. needed voice in international cooperation: the voice of It was created in the mid- I 990s by unions, grassroots or- the poor themselves. Just as for national policies, their ganizations, and NGOs working with home-based work- voice is essential in ensuring that global policies meet their ers and street vendors in developing and developed needs. I88 WORLD DEVELOPMENT REPORT 2000/200I * * * for poverty reduction depend on policy changes in high- income countries and cooperative actions at the global The four areas of action highlighted in this chapter il- level. These include lowering trade barriers, increasing fi- lustrate the importance of international cooperation in nancial stability, producing international public goods that the fight against poverty. Many of the most pressing particularly benefit poor people, and ensuring a voice for problems in developing countries-from trade barriers poor countries and poor people in global forums. to financial crises to infectious diseases-can be solved The need for these international actions should re- only with cooperation from high-income countries. Yet define the role of international cooperation in poverty in the past, international cooperation has consisted pri- reduction. Even with more effective aid, the subject of marily of financial transfers from rich countries to poor the next chapter, progress against poverty will be slower countries, notably aid. But aid is not enough-prospects without the international actions recommended here. CHAPTER 11 Reforming Development Cooperation to Attack Poverty D evelopment cooperation is being reformed. duction. And donors are working to resolve differences From the relationship between donor and recipient to in approaches to poverty reduction through the the way in which aid is delivered and the framework OECD's Development Assistance Committee (DAC), for debt relief for the poorest countries, many of the which expects by mid-2001 to agree on guidelines for old ways of assisting development are beginning to be poverty reduction to help donor agencies make their replaced by new forms. programs more effective.3 Much of this is due to a reaffirmed commitment But while the international community's com- by the international community to fight poverty. The mitment to attack poverty was strengthened in the World Summit for Social Development in Copen- 1990s, official development assistance shrank. This, hagen in 1995 set forth the goal of eradicating poverty despite the optimism at the start of the 1990s that in the world through decisive national actions and development cooperation would reap a post-cold international cooperation. Donors have included halv- war "peace dividend" from cutbacks in military spend- ing poverty between 1990 and 2015 and other targets ing.4 Indeed, after peaking in 1992 (in real terms), among their international development goals (see box official development assistance fell consistently over 2 in the overview).' In the 12th replenishment of the the decade-despite the robust economic growth of International Development Association (IDA) in 1998, DAC countries-rebounding only slightly in 1998 donors reaffirmed their mission to support programs during the global financial crisis (figure 11.1). Six- to reduce poverty and improve the quality of life in teen of the 21 DAC countries spent a smaller share IDAs poorest member countries.2 The Jubilee 2000 of their GNP on development assistance in 1997-98 movement helped put deeper debt relief at the heart than in 1988-92.5 The regional distribution of this of development cooperation strategies for poverty re- aid remained roughly constant between 1987 and 189 I90 WORLD DEVELOPMENT REPORT 2000/200I Figure 11.1 Figure 11.2 While donor countries' economies grew after With the exception of Europe and Central Asia, 1992, their development assistance shrank the regional distribution of official development assistance remained roughly constant ... GNP and official development assistance per capita in DAC countries Average annual share of net receipts of official U.S. dollars development assistance 30,000 Percent 35 29,000 28,000 30 27,000 25 26,000 20 25,000 20 1987-88 15 1997-98 10 Official development assistance .992-93 75 ~~~~~~~~~~~~~~~~~10.- 50 25 5 2519932 1993 1994 1995 1996 1997 1998 0 Eas Asia Europe Lati n Middle South Sub- Note: GNP is in 1995 U.S. dollars, official development assistance and and America East and Asia Saharan in 1998 U.S. dollars. Pacific Central andthe North Africa Source: World Bank staff estimates based on OECD and World Asia Caribbean Africa Bank data. Note: Shares do not sum to 100 percent, as some aid is not allocated by country or region. Source: OECD, DAC data. 1998, apart from an increase in the share going to Eu- rope and Central Asia (figure 11.2). But total develop- ment assistance fell in every region except Latin America private capital flows to developing countries, leaving and the Caribbean after 1992-93 (figure 11.3). Pre- some 140 developing countries and territories (with liminarv estimates show that official development as- about 1.7 billion people) to share the remainder. Almost sistance rose again in 1999, by about 5 percent, though entirely left out were the 61 low-income countries be- it is too soon to know whether this reflects more than sides China and India.6 For example, all of Sub-Saharan the response to the Asian crisis and indicates a much- Africa received only 1.2 percent of flows to developing needed real and sustained reversal of the downward countries in 1998. These are the countries that need aid trend in the 1990s. most, and they are hit hard by its decline. The decline has been costly for many countries. Al- There is no single reason for the decline. Donors ini- though it has coincided with massive inflows of private tially cited their fiscal deficits as a large part of the prob- capital to developing countries, very little of that capi- lem. Yet even as these deficits declined (from 4.3 percent tal goes to the poorest countries. Net private capital of GDP in 1993 to 1.3 percent in 1997), official devel- flows to low- and middle-income countries reached opment assistance continued to shrink, dropping 14 per- $268 billion in 1998 and now dwarf aid flows in some cent from 1996 to 1997.' A more likely explanation is countries. Overall, private flows to developing coun- that donors continue to view development cooperation tries surged during the 1990s, from 43 percent of total through a strategic lens rather than a poverty lens, see- resource flows in 1990 to 88 percent in 1997, just be- ing other uses for their money as strategically more im- fore the East Asian financial crisis. However, inflows of portant. Historically, aid flows have been determined private capital have been concentrated in relatively few more by political and strategic interests than by poverty countries; a large number of countries receive little or reduction goals.' nothing. In 1997, before the financial crisis, the top 15 Perhaps more noteworthy is the decline in support developing country recipients received 83 percent of from the traditional proponents of official development REFORMING DEVELOPMENT COOPERATION TO ATTACK POVERTY 191 Figure 11.3 cutting debt for poor countries to support poverty re- ... while receipts fell after 1992-93 in all duction and human development. So there is clearly regions but Latin America and the Caribbean continuing support for the principle of providing re- sources for improving the lives of poor people in the de- Billions of U.S. dollars veloping world, but widespread questioning of the 20 traditional mechanisms for providing such resources. Is aid working? Can it work better? What is the role of debt reduction in concessional support? Developing 15 countries themselves will largely determine through their own policies whether they achieve the international de- 10 1992-93 velopment goals. But aid and debt relief can provicle | | 1997-98 * i crucial support. So finding out how to make these moire *997-98 effective-and then doing what it takes-remains vital. In answering these questions, this chapter outlines a vision for a better system of development cooperation, one based on new thinking and new practices. This vi- sion includes a reformed framework for country-focused East Asia Europe Latin Middle South Sub- aid and debt relief for the poorest countries-under- and and America East and Asia Saharan pinned by a renewed emphasis on the policy and insti- Pacific Central and the North Africa uional envirent and the fundamental prioit of Asia Caribbean Africa tutional environment and the fundamental priority of Source: OECD, DAC data. poverty reduction. Donors would work in partnership with countries, directing aid and debt relief along the lines of a broad-based poverty reduction framework (as ad- vanced by many donors and laid out in this report), sup- assistance. The preeminence of geopolitical interests is porting countries that can put these resources to good use not new.9 But what is new is the falloff in countervail- for poor people. ing support from advocates for development assistance Supporting good policies and institutions is important, on humanitarian grounds. Many fell victim to "aid fa- but it is not enough. We learned in the 1 990s that process tigue" and were far less vocal supporters in the 1 990s than is as important as policy in foreign aid and the manage- before. ment of unsustainable debt burdens. The way donors and Not every country was affected by aid fatigue-indeed, recipients interact strongly influences the effectiveness of aid flows increased from some countries-but its symp- development cooperation. Relationships have tended to toms were clearly evident. For example, in the United follow the preferences of donor countries, leaving recip- States a comprehensive poll found that an overwhelm- ient countries with little sense of ownership of the aidl- ing majority of the population favored foreign aid in financed activities. Along with advancing a broad-based principle-and that only 35 percent thought it should poverty reduction framework, this report emphasizes be cut from current levels.10 Yet more than 80 percent how much local realities matter in development. That aid of respondents believed that waste and corruption kept relationships have too often failed to take local realities foreign aid from reaching the people who need it. This into account, undermining ownership, is an important kind of public disillusionment may have made it harder flaw. for donor governments to maintain foreign aid, let alone If development cooperation is to attack poverty ef- increase it. If aid is not working, the sentiment goes, the fectively and efficiently, donors will need to: money could be better spent elsewhere. * Pay more attention to local conditions and country In contrast to the rise in aid fatigue in some places was ownership. a major upsurge in support and activism around debt re- * Deliver aid in ways that intrude less on government duction, most notably under the auspices of the Jubilee functions, including greater use of sectorwide ap- 2000 movement of religious organizations and other proaches and a movement away from old forms of aid civil society groups. They rallied around the cause of conditionality. 192 WORLD DEVELOPMENT REPORT 2000/200I * Provide sustained support for policy and institutional pacts on developing countries. It was widely believed that environments that are strongly conducive to poverty government interference in the economy had prevented reduction, in preference to ones that are not. developing countries from adjusting to these shocks. Sub- The chapter begins by exploring how these new ap- sequently, a new consensus began to form, eventually to proaches can make aid more effective. It then examines be known as the "Washington consensus" (see box 4.1 in the issues associated with relieving the debt problem of chapter 4) .4 To many, including staff at the World Bank poor countries. and other multilateral financial institutions, fiscal prudence, free markets, and outward orientation had clearly demon- Making aid more effective in reducing strated their superiority as the most efficient way for poverty countries to grow and develop.15 But it has become clear that simple strategies for de- Recent studies confirm what anecdotal evidence has long velopment and poverty reduction are elusive. While mar- hinted: the experience of aid has been mixed. 1 Early pre- kets are a powerful force for poverty reduction, institutions dictions that aid would close the financing gap that pre- that ensure that they operate smoothly and that their ben- vented developing countries from moving ahead have not efits reach poor people are important as well. As the 21st come to pass. If all the aid that went to Zambia between century begins, donors are coalescing around a develop- 1961 and 1994 had gone into productive investment, and ment strategy that includes investing in people through if investment had been as important to growth as initially health and education services, promoting inclusive and predicted, the country's per capita income would have equitable growth, supporting good governance, and pro- been more than $20,000 in 1994, not $600.12 tecting the environment. IG This strategy also recognizes And yet there have been many aid successes. The On- the centrality of local conditions: that the most effective chocerciasis Control Program is but one example (see development policies will vary by situation. box 10. 1). Aid was important, in different periods, in East Despite this growing consensus on the broad devel- Asias extraordinary success in poverty reduction over the opment framework, agreement on the right policies in par- past few decades. The rapid progress in Vietnam in the ticular conditions has tended to elude donors and 1990s is another example. So aid can work. The challenge recipients. Donors come to development problems with for the international community is to understand how to their own mandates, histories, ideologies, and political re- make it work consistently-and then to do what it takes. alities and often do not see situations in the same way as other donors or the recipient countries. Even in health The key problems with aid and education, which all donors agree are essential, the Aid's difficulties in reducing poverty go deeper than the right reforms are open to debate. As an analyst com- sway of geopolitical interests over development inter- mented, there is "a bewildering multitude of national sys- ests, which has often directed aid to countries whose tems and experiences, with varied (and hotly debated) policies were not focused on reducing poverty. Aid has advantages and disadvantages associated with each."17 So been hindered by the frequent differences in donors' while the days of adhering strictly to either state-led or perspectives on development policies, even though the past market-led solutions are over, between these extremes lie 50 years have been punctuated by times of relatively a host of options, and the debate on them is far from over. wide consensus on the best way to pursue development.i3 The lack of consensus on the broad outlines and the Donor differences have played a key role in preventing details of national and local policies and projects has re- aid from achieving full effectiveness. Donors have often duced the effectiveness of development assistance. 1 This failed to coordinate their efforts, countries have not taken effect is especially evident in problems of ownership, ownership, and there has been heavy use of condition- donor coordination, fungibility, and conditionality- ality both at the project level and economrywide. the four main issues affecting aid in the 1990s. In the first two decades after World War II state-led in- Ownership. Because donors and recipients often dis- dustrialization was generally seen as the best way to pur- agree, donors have looked for ways to ensure that their sue development, a consensus undone in the 1970s by money is spent as they intend. They have run their own world events, including the demise of the fixed exchange projects, required detailed reports from countries on pro- rate system and two oil shocks, which had devastating im- jects, and attached conditions-usually policy oriented- REFORMING DEVELOPMENT COOPERATION TO ATTACK POVERTY 193 to the use of funds. A major study on relations between ment because of the importance of recurrent spending in donors and African recipients found that "in spite of maintaining basic social and economic services.24 some improvements, donors still tend to dominate the Conditionality. Donors know that even properly im- project cycle and pay inadequate attention to the pref- plemented projects will have limited impact in poor pol- erences of the government or project beneficiaries."19 icy environments.25 Awell-built school will be useful only These efforts to ensure that aid is spent effectively, evi- if money is budgeted annually for teachers, books, and dence now shows, have often had the opposite effect by supplies-and if the economic environment enables chil- diminishing ownership by the recipient country. dren to go to school. The role of good policies and in- Analyses show that ownership is a key ingredient of aid stitutions in ensuring sustainable results suggests that effectiveness.20 How strongly a country believes that a pro- aid should flow more to countries with a good overall pol- ject or reform will bring benefits affects the effort put into icy environment and good policies for poverty reduction. the activity, the domestic resources contributed, and the com- But the relationship between good policies and aid flows mitment to the activity after the donor has left-all sub- has not been strong.26 stantial determinants of success. To succeed, reforms and This finding would be understandable if aid were projects must foster ownership by the people for whom the spurring policy reform by influencing countries to change policy or project is ostensibly being implemented. their policies or by helping them do so. This has been the Donor coordination. When different donor priorities intention of many donors, and it is one reason (fungibility and project-related conditions (including donor-specific is another) that many of them have reduced the share of reporting and procurement requirements) are multi- their portfolio allocated to projects and increased the plied many times over, they can create an unworkable en- share allocated to program and policy-based aid.27 Most vironment for a recipient government. Just the sheer number program and policy-based aid has been tied to the en- of donors and donor projects can be challenging. At one actment of certain policy reforms. But studies in the point there were 405 donor-funded projects in the Mozam- 1 990s showed little systematic relationship between con- bican Ministry of Health alone. In the early 1 990s in Tan- ditionality and policy changes, though case studies do find zania there were 40 donors and more than 2,000 projects. positive effects under some conditions, especiaLly where In Ghana during the same period 64 different government conditionality supports the hand of reforming groups.283 or quasi-government institutions were receiving aid.21 Co- The dynamics between aid donors and recipients ex- ordinating these efforts to support a coherent development plain why conditionaLity fails. Recipients do not see the con - strategy-even at the sector level-is nearly impossible. ditions as binding, and most donors are reluctant to stop Fungibility. Studies show that aid funds allocated to giving aid when conditions are not met.29 As a result, a particular sector tend to free up for other purposes compliance with conditions tends to be low, while the re-- money that the government would otherwise have spent lease rate of loan tranches remains high.30 Thus aid has often in that sector.22 This means that in funding specific pro- continued to flow despite the continuation of bad policies. jects or sectors, donors may actually be helping to increase In addition to performing poorly in influencing spending on sectors they do not want to finance, such as policy reform, policy-related conditions, often combined the military. This has profound implications for devel- with project-related conditions, severely burden de- opment cooperation. Project-level evaluations will not re- veloping country administrators-a problem that has flect the true impact of aid, since aid is likely to be freeing become more pronounced as conditionality has ex- up resources for other activities.23 panded. Conditions on World Bank adjustment loans,, Even where resources are fungible, donor support can having mushroomed in the 1 980s, continued to grow still have some impact, from the design of certain poli- in the 1990s along with the expanding development: cies to institutional development. Moreover, in countries agenda.31 As one recent assessment put it: "Although highly dependent on aid, donors as a group could lead much has been added to the conditionality menu since to shifts in government resource allocations, because of 1981, nothing has been taken off."32 The time gov- the sheer size of flows. A potentially important part of this ernment officials spent negotiating and monitoring is the preference of donors to support development bud- these conditions is time they could better have spent gets, which can lead to a net shift in resources out of the analyzing development problems and designing de- recurrent budget-not always a good thing for develop- velopment strategies. Ownership has been shown to be I94 WORLD DEVELOPMENT REPORT 2000/2001 central to the sustainability of both projects and pol- nership, or contract, in a way that provides the incentives icy reform, and the fact that the delivery of aid weak- for country-driven, long-term poverty reduction strategies ens it is a fundamental flaw of current development while also strengthening the internal partnerships neces- cooperation mechanisms. sary for social stability and economic development. Consultations between governments and civil society Solutions that accommodate different and between governments and donors have been carried perspectives out in a number of countries piloting the World Bank's While the dominant forms of donor-recipient relations Comprehensive Development Framework, the European have allowed donors to pursue their own priorities, the Union's partnership approach, and other such approaches. result has generally been a fragmented system that un- The consultations under the Comprehensive Develop- dermines their efforts. The challenge in reforming in- ment Framework have proved fruitful in several countries ternational development cooperation is to accommodate -such as Bolivia, the Dominican Republic, and Ghana- different perspectives on development without overbur- but have also highlighted the need for government com- dening the recipient or undermining ownership. mitment and for capacity as key ingredients of successful Achieving global uniformity in development strategies consultations (box 1 1. 1). might be one solution, but history shows that uniformity This emerging approach to development cooperation is undesirable. Development is determined to a great ex- has been incorporated into the new initiative by the tent by local conditions, including social institutions, so- World Bank and International Monetary Fund (IMF) to cial capability, ethnic fragmentation, inequality, and link their support of low-income countries to nationally geography.33 In studies these variables significantly explain designed poverty reduction strategies, working within the the variation in growth rates over the past 30 years.34 Stud- principles of the Comprehensive Development Frame- ies also show that external shocks-and the ability to re- work (box 11.2). Concessional funds and debt relief spond to them-can have as much effect on growth as from the World Bank and IMF will be linked to the goals policies do.35 The approach to designing development of poverty reduction strategies prepared by governments strategies should therefore be flexible enough to adjust in consultation with civil society organizations, the pri- to both internal and external conditions. vate sector, and donors. Based on a good understanding This perspective began to take hold in the development of the poverty situation in the country, the strategies will community in the late 1 990s. Combined with new think- identify actions with the greatest expected impact and set ing on aid effectiveness, it has prompted proposals to ad- up monitoring and evaluation processes. The goal is for dress the problems of aid. Three prominent themes are these strategies, described in poverty reduction strategy ownership and partnership, less intrusive aid delivery papers, to form the basis for assistance not only from the mechanisms that focus on the overall policy and expen- World Bank and IMF, but from other assistance agencies diture framework, and selectivity Together, they form the as well.37 Similar initiatives are under way in the regional agenda for the international community to improve de- development banks. velopment cooperation in the coming decade. Less intrusive aid delivery mechanisms focusing on the Ownership andpartnership. Recognizing the importance overallpolicy and expenditure environment Donors have of ownership and the problem of donor coordination, most used many means to influence recipient country policies. donors have embraced partnership as a guiding principle Old forms of policy conditionality have often had dis- in interactions among donors, governments, and citizens appointing results, depending on country circumstances in developing countries.36 Most partnership frameworks and how the conditionality was used. Policy review have two parts. The first is a partnership between the re- processes also have had limited success. Public expendi- cipient government and its citizens, who share responsi- ture reviews, for example, have evaluated the level and bility for developing their national development strategy. composition of countries' expenditures and identified This strategy can take shape through a consultation process ways to improve expenditure policy and use donor funds involving government, civil society, and the private sec- more efficiently (see box 9.2 in chapter 9). But several stud- tor. The second is a partnership between the government ies have found this type of intervention to be ineffective and donors, with donors designing their assistance strate- in many cases, largely because recipient countries have not gies to support the government's strategy. In the new been closely involved in the reviews-and so have felt lit- thinking the focus is on how to shape this external part- tle inclination to comply with the findings.38 REFORMING DEVELOPMENT COOPERATION TO ATTACK POVERTY 195 Box 11.1 Learning about the consultative process through the Comprehensive Development Framework In 1999 the World Bank announced its Comprehensive Develop- All discussions with donors now take place in the context of ment Framework, a tool for improving country ownership and donor the national action plan. At a consultative group meeting in April coordination in development cooperation. The framework is based 1998 donors pledged 45 percent more than they had in 1997. on four principles: country ownership of the policy agenda, part- Donors have also been encouraged to formulate their strategies nership with all stakeholders, attention to social and structural con- in support of the national action plan. The World Bank recently cerns as well as macroeconomic and financial issues, and a redesigned its country assistance strategy to align it with the plan, long-term, holistic approach built on national consultations. choosing to support three of the plan's four pillars. The govern- The country develops its national strategy in consultation with ment continues to lead donor coordination, chairing the consul- civil society and the private sector-and then, with donors, designs tative group meeting in Paris in 1999, where it presented its a matrix linking development goals and development actors. The version of the Comprehensive Development Framework. It has activities of actors in support of each goal are listed in the matrix, also agreed with donors on intermediate indicators for monitoring revealing any gaps or overlaps. outcomes. The framework is being implemented in 13 countries, encour- Other countries have not progressed as far. The difficulties of aging wide consultation between governments and their citizens and some highlight potential problem areas. For example, it is clear that enhancing partnerships with donors in the design of comprehen- country ownership depends largely on national capacity. The coun- sive national development strategies. But progress has been var- try must be able to hold broad consultations with all elements of ied, reflecting different starting dates and country circumstances. society and to conduct the complex analysis necessary to design Bolivia is an early case. In late 1997 the new government em- national strategies that balance macroeconomic and financial is- barked on an analysis of the country's development challenges and sues with social, structural, and institutional concerns. And, of the preparation of a national action plan to address them. A key part course, the country must be able to implement the strategy. was a national consultation with a wide range of representatives Without this ownership-and the country leadership from of civil society-NGOs, unions, religious organizations, opposition it-donor coordination will remain difficult. While there is some parties, and academics-and the private sector to discuss devel- evidence that some donor countries are gradually aligning their opment constraints and propose solutions. The results of this na- strategies with those of recipient countries, stronger leadership tional dialogue were presented to the government as input to the by the recipient government will be required to accelerate national action plan. progress. Source: Wolfensohn 1999; World Bank 1 999d, 1 999u. Box 11.2 The new poverty reduction strategy initiative The poverty reduction strategy initiative of the World Bank and In- years, all embedded in a long-term framework for poverty re- ternational Monetary Fund seeks to link external support to do- duction. While the actual form of the strategy will be decided mestically developed, results-based poverty strategies. It is also by the country-there is no single blueprint-most strategies intended to improve the effectiveness of World Bank and IMF re- would likely include: lations (and those of other donors as well) with recipient countries. * Long-term goals for key poverty reduction targets, and the As important as the recipient country strategy is the process lead- macroeconomic, structural, and institutional framework for ing up to it. A broad, participatory dialogue with representatives achieving them (see, for example, Uganda's goals in box 1.7). of civil society and the private sector is expected to: * Mechanisms for monitoring and evaluating progress toward * Help national authorities develop a better understanding of the the poverty reduction targets, linked to public actions. obstacles to poverty reduction and growth-and devise good * A consistent policy and institutional f ramework that includes the indicators of progress in poverty reduction. underpinnings for rapid, sustained growth and poverty reduc- * Deepen a shared vision of desired poverty reduction goals tion (including macroeconomic policies, institutional reforms, across society. sector strategies, and associated domestic and external fi- * Lead to formulation of priorities for public actions to achieve nancing needs). the desired poverty reduction outcomes. Donors can help by providing technical assistance in some * Encourage the development of participatory processes for areas. Initial experience in Africa and Latin America indicates that setting poverty reduction goals and monitoring implementa- countries are strong in laying out a poverty profile and a general tion and progress. poverty reduction strategy but weaker in preparing quantified tar- The results will be periodically reported in poverty reduction gets, costing the strategy, and evaluating tradeoffs under lim- strategy papers expected to reflect a broadly owned develop- ited resources. As in other aspects of development cooperation, ment strategy. The strategies will generally focus on three- the country should determine its own need for assistance-to year cycles, with annual progress reports in the intervening maintain ownership of this important process. Source: IMF and IDA 1999; World Bank and IMF 2000a. 196 WORLD DEVELOPMENT REPORT 2000/2001 Perhaps more surprising, donor compliance has been Figure 11.4 weak as well. A recent evaluation found public expendi- Aid does not go only to poor countries ture reviews to have had little effect on either recipient country policies or donor lending practices." So donors Average annual net recipts of offcial development , t 0 1 ~~~~~~~~~~~~~~~assistance as a share of total are searching for new mechanisms for strengthening pol- Percent icy environments that encourage country ownership rather than undermine it. They have begun, for exam- 35 1997-98 ple, to encourage countries to participate fully in the 30 public expenditure review process, and they are experl- 1987-88 I menting with new instruments. 25 One new instrument that has received much atten- 2 l tion is the sectorwide approach: the government designs 20 an overall sector strategy, and donors sign on to fund 15 the sector, not individual projects. This resolves the 10 problem of donor coordination by eliminating the need for it: all activity in the sector is conducted by the re- 5 cipient country, using its own funds in addition to n those of donors. This instrument responds to a broader Least Other Middle- High-income developed low-income income countries policy environment while also ensuring ownership. Al- countries countries countries though the approach is too new to have a track record, Note: Least developed countries are the 48 countries so defined by some early experiences are promising (box 1 1.3). the United Nations on the basis of certain thresholds for income, Some proponents have suggested applying the prin- economic diversification, and social development. p ~~~~~~~~~~~~~Source: OECD, DAC 2000. ciples of the sectorwide approach to all development cooperation (box 1 1.4). Others consider project-based lending to be desirable and consistent with the new dle-income countries (most donors target aid to the thinking on development cooperation for poverty re- poorer countries), that still means that global aid is not duction. Project support can be effective for results- heavily targeted to areas where the incidence of poverty based sector development-if it falls within a sector is greatest. Aid, and especially nonconcessional develop- framework that systematically links investments and pol- ment flows, still has a role in reducing poverty in middle- icy and institutional development to poverty outcomes income countries, when the policy environment is sound (and to intermediate indicators for tracking and in- and the resources are well targeted. terpreting progress). The choice of instrument will de- In addition to targeting poverty, donors should allo- pend on the policy and institutional conditions of cate aid on the basis of the policy environment. Aid has particular countries (or sectors within countries) and been shown to be effective in promoting growth and the preferences of individual donors. But a premium poverty reduction in poor countries with sound eco- should be placed on putting the country in charge and nomic policies and sound institutions-ineffective where ensuring that the mechanisms of aid delivery do nor these are lacking.4' Aid driven by political and strategic compromise its ownership. interests rather than by the recipient country's develop- Selectivity. For aid to be most effective at reducing ment policy environment is largely wasted from a poverty poverty, it must be well targeted. If all aid money were reduction perspective. Several instruments have been de- allocated on the basis of high poverty rates and reasen- veloped to assess the policy and institutional environment ably effective policies and institutions, a recent study es- in recipient countries, generally covering macroeconomic timates, even today's small aid flows could lift 19 million management, structural policies, policies for social in- people out of poverty each year-almost twice the esti- clusion (poverty, gender), and public sector management mated 10 million now being helped.40 (box 1 1.5). Currently, about a third of aid goes to middle-income Factoring in the level of poverty and the quality of countries, whose average GNP per capita is roughly six policies should make aid much more efficient in reducing times that of low-income countries (figure 1 1.4). While poverty, and there is evidence that donors began to do only a few major donors target more of their aid to mid- this in the 1990s.42 In replenishing IDA in 1998, for REFORMING DEVELOPMENT COOPERATION TO ATTACK POVERTY 19 7 Box 11.3 Sectorwide development cooperation To address problems of ownership, donor coordination, and fun- The approach ensures fult ownership by the country and elim- gibility, donors are experimenting with pooling their resources to inates problems of donor coordination. With the country having more support sectorwide strategies designed and implemented by the ownership and control over what happens, the use of resources can recipient government. The country, in consultation with key stake- be much more efficient. But it also means great changes in donor- holders, designs a sector strategy and a budget framework extending recipient relations and perhaps greater difficulties in implementa- several years forward, and donors put their money into the cen- tion. Several sectorwide programs have stumbled because of the tral expenditure pool for the sector. The approach encourages recipient country's inadequate institutional capacity. Lack of con- country ownership of sector strategies and programs. It also links sistency with the macroeconomic program has been another prob- sector expenditure with the overall macroeconomic framework. And lem. And donors often have too many requirements and thus too it ensures coordination of donor and recipient activities. much of a problem (or too little interest) in harmonizing them (Har- Some benefits of a sectorwide program are evident in the Zam- rold and associates 1995). Furthermore, these arrangements greatly bian health sector. In 1994 the government presented its national diminish donor control and monitoring of exactly how money is spent. health policy and strategy to donors and-to ensure equitable dis- The changes required imply that gaining support for the ap- tribution of services and coherent implementation of the strategy- proach will be difficult. The recipient government has to be very con- asked them not to fund specific provinces or projects but to fund fident, because strict adherence to a sectorwide approach means the Ministry of Health centrally. Hesitant at first, donors began to donors that do not participate in common implementation arrange- comply. An independent evaluation in 1997 found that "health ments are not allowed to act in the sector (that is, they do not have workers are better motivated; clinics are functioning; funds are their own projects). The result may be less donor funding for a sec- flowing to the districts; some modicum of decentralization is in place; tor. Governments might therefore opt for less strict sectorwide pro- land) an important part of the private sector has become formally grams, choosing instead to allow donors to implement projects as involved." long as they fit into the overall sector strategy. Box 11.4 The common pool for development cooperation Seeing the potential of the sectorwide approach, some propose and the confidence to follow through even if donors do not sup- extending the idea to the country level (Kanbur, Sandier, and Mor- port it. In addition, donors might resist common pools at the na- rison 1999). Donors would cede complete control to the recipient tional level because they would likely mean a reduction in donor country government-advancing their own perspective on de- staff, since donor agencies would no longer be developing and mon- velopment strategy through dialogue with the country and with itoring projects or negotiating and monitoring conditions. one another rather than through specific programs and projects. However, like the sectorwide approach, the common pool Rather than fund their own projects, donors would give central bud- approach would ensure full ownership by the country and elimi- get support to countries with good development strategies (and nate donor coordination problems. It would also preserve two im- the capacity to implement them). portant benefits of the current development cooperation approach: A country would first develop its own strategy, programs, and * The knowledge transferred in donor-implemented projects, an projects in consultation with its people and with donors. It would important side effect of aid. A road building project, for example, then present its plans to donors, which would put unrestricted might transfer knowledge of engineering or even project ac- financing into a common pool of development assistance, to be counting to local workers. This transfer would not be lost in a com- used along with the government's own resources to finance the mon pool arrangement. Recipient countries could still ensure development strategy. Earmarking would disappear. Donor mon- knowledge transfer through their choice of companies and the itoring and control of specific projects and programs would not terms of contracts. be permitted. And no conditions would be placed on donor aid. * The support that conditionality gives to reform factions in gov- How much donors give would depend on their assessments ernments. Support for reform elements in a country is perhaps of the country's policy environment, including how the country the only effective part of the present system of conditionality. came to agreement on the strategy and how capable it is of im- Donor-imposed conditions can strengthen the position of re- plementing the strategy and monitoring progress. In this way the formers in national debates or serve as a "self-imposed" con- common pool approach would be a more rigorous form of condi- straint on government officials. The approach to conditionality tionality, because donors would need to evaluate the overall pol- in a common pool arrangement would be far different, but it icy environment, direction, and capacity of countries. These would not sacrfice this benefit. Donors could strengthen the hand assessments would be made known to the country and to other of reformers by publicizing the criteria used to assess country donors during the dialogue leading up to the financing decision, strategies and adjusting the volume of their assistance. This would This approach would entail many of the same challenges fac- form the basis for a more open and honest relationship between ing the sectorwide approach, including the need for recipient donors and recipients and preserve the benefits of the current countries to have both the capacity to implement their strategy conditionality while eliminating its problems. 198 WORLD DEVELOPMENT REPORT 2000/2001 Box 11.5 Assessing country policies and institutions The World Bank has designed a measure of policy and institutional Competitive environment for the private sector soundness-the Country Policy and Institutional Assessment, which Factor and product markets gives equal weight to 20 components that have evolved as the mea- Policies and institutions for environmental sustainability sure has been refined. Each component is rated by country specialists on a scale of 1-6 using standard criteria. Although care is taken to Policies for social inclusion and equity ensure that the ratings are comparable within and between re- Equality of economic opportunity gions, the scores include an irreducible element of judgment. But Equity of public resource use when the measure has been included in regression analyses of Building of human resources growth along with other commonly used policy variables, it has had Safety nets statistical significance, while other policy measures have not. It Poverty monitoring and analysis thus appears to be a good summary indicator of the overall policy environment for economic development. The 20 components: Public sector management and institutions Property rights and rule-based governance Economic management Quality of budgetary and financial management Management of inflation and the current account Efficiency of revenue mobilization Fiscal policy Efficiency of public expenditures Management of external debt Transparency, accountability, and corruption in public services Management and sustainability of the development program Developing a consistent basis for rating economic and struc- Structural policies tural policies has been relatively straightforward, but doing so Trade policy and foreign exchange regime for social inclusion and public sector management has proved Financial stability and depth more challenging. Work to refine the indicators and reference Banking sector efficiency and resource mobilization points continues. Source: Coltler and Dollar 2000; World Bank 1999h. example, donors called for allocating funds on the basis provide the framework for substantially improved inter- of each country's policy performance.43 national development cooperation. But progress toward How selectivity is applied will likely evolve as the in- that vision will not be easy. Each component of improved ternational community continues to learn about the en- development cooperation brings great challenges in vironments in which aid is most effective.44 Some analysts implementation. stress that the level of poverty in a country is more im- For example, while almost everyone agrees that part- portant to aid effectiveness than the policy environment, nership is a good idea, there is no consensus on how to though both are crucial.45 Others show that external implement it.48 Some analysts note that ownership is rel- shocks-such as declining terms of trade, volatility in ative and that reaching consensus on strategies is essen- export prices, and even climate change-can impede tially a political process, involving the same power countries' efforts in growth and poverty reduction (chap- relations that exclude poor people from discussions or ter 9).46 It has been argued that aid can make a larger dif- discriminate against them (as seen in chapter 6).9 Oth- ference in these countries (and therefore be more effective) ers voice doubts that donors will really come to terms than in countries not experiencing shocks.47 Refining the with the implications of ownership and partnership for criteria for selectivity should continue. Adhering to the basic their actions: that donors should interfere less in recip- principle that aid should go where it is most effective in ient country policymaking.50 Many donor practices- reducing poverty will be key if the international community such as maintaining control over resource monitoring and is to achieve the international development goals. tying aid to specific procurement requirements-run con- trary to the idea of partnership.5i The recipient coun- Implementation di]ficulties try's capacity to design and implement development and practical steps strategies and its ability (and willingness) to hold broad These three components-ownership and partnership, aid consultations with all elements of society also pose sig- delivery mechanisms that are less intrusive, and selectivitv- nificant challenges. REFORMING DEVELOPMENT COOPERATION TO ATTACK POVERTY I99) The combination of greater selectivity and a broader, Box 11.6 less intrusive approach to delivering development assistance How aid can help in countries with a weak presents its own challenges. Determining how much sup- policy environment port to give to a sector or national budget is difficult-and likely to prove contentious. Some country expenditures may When a country has poor policies and no coherent political not sm fio" torr cmovement to change the situation, aid can have a limited but not seem tofinta"es"poe t, bt effective role, as Ghana, Uganda, and Vietnam all illustrate. donors will have to evaluate the poverty reduction impact In their prereform periods (before 1983 for Ghana, 1986 for of the overall program, not the individual expenditures. Uganda, and 1991 for Vietnam), these countries received very A more fundamental problem arises when a country little aid, probably reflecting their governments' political es- does not have an overall policy environment worth sup- trangement from the West. But the aid was instrumental in porting, so that aid is largely ineffective. How should laying the foundation for policy reform. For example, when Ghana was dealing with a macro- donors proceed? economic crisis in the early 1980s, its well-trained econo- Most important, they must understand that policies are mists found the policy dialogue with international financial driven primarily by the domestic political economy-and institutions to be helpful in working out plans. A few years later, when Uganda's leaders were trying to design new poli- that donors are simply not very effective in influencing cies, donors financed helpful study tours to Ghana. In 1991 them.52 But donors can have some influence by tailoring their the United Nations Development Programme and World involvement to a countrys commitment to reform. Until a Bank organized a meeting for Vietnamese leaders with eco- country commits seriously to reform, the best that donors nomic ministers from Indonesia, the Republic of Korea, and Malaysia, who laid out some key policies that had worked can do is to provide technical assistance and policy dia- for them and also some of the detailed issues in stabiliza- logue, without large budget or balance of payments support tion, trade liberalization, foreign investment, and other eco- (box 11.6). If donors pour large amounts of aid into poor nomic policies. policy environments, they are likely to sustain poor policies In successful cases political leaders learn from other longer. When the country finally commits to reform, evi- countries and from their own mistakes. Low-key assistance can help with this policy learning, which generally has to take dence shows that finance should be increased as policies im- place at a country's own pace. Even in countries that do not prove.53 reform for a long time, technical assistance can lay the foun- In addition to this more nuanced approach to influ- dation for policy learning. In Kenya, for example, donors are encing policy reform, donors can address the challenges supporting the Institute of Policy Analysis and Research to enc,ng pir r d ohelp develop local capacity in research and policy analysis. of the new development cooperation framework by tak- This kind of capacity building is not going to have large ben- ing several other steps: efits as long as vested interests resist serious reform. But * Molve the donor-recipient dialogue to the country and turn it is an essential foundation if a political movement for change its leadership over as welL Donor-recipient consultations- develops. consultative groups or roundtables-have traditionally Source: Devarajan, Dollar, and Holmgren 2000; World Bank 1998b. taken place in donor countries, chaired by the World Bank, the United Nations Development Programme, or another donor institution. Meetings are now beginning to be held in recipient countries and chaired by their gov- own practices-for example, by harmonizing proce- ernments, to foster ownership. dures and reporting requirements among themselves-- • Continue to experiment with sectorwide approaches. so that they can contribute effectively to these new aid National capacity-and donor-recipient partner- relationships. ships-can be built up sector by sector. While many * Strengthen monitoringand evaluationpractices. Donors' countries will for some time not have the overall tech- systems of monitoring and assessing the impacts of nical capacity, accountability, and transparency to their own projects have failed to focus on how poor peo- monitor funds to the satisfaction of donors, these ple benefit.55 But doing this will be even more impor- may be more advanced in some sectors than in oth- tant (and challenging) when looking at a sectorwide or ers. The advanced sectors could be funded through the nationwide program. Donors should encourage local sectorwide approach as soon as possible, taking into monitoring by participants, to ensure ownership of the account the lessons from experience with this ap- results. Furthermore, donors tend to be weak in dis- proach.54 And donors should continue to improve their seminating information and incorporating knowledge 200 WORLD DEVELOPMENT REPORT 2000/200I from evaluations.56 Feedback and learning are essential Figure 11.5 to successful aid practices, and donors must ensure NGOs channeled some $10 billion to that they happen effectively As part of this, donors and developing countries in 1998, about half Ihtte apnercley spr ftl,unr n of it from the official sector recipients should continue to strengthen their efforts against corruption, a major obstacle to economic per- Billions of U.S. dollars formance that occasionally also affects donor agencies. 6 * End tied aid. In 1998 almost a quarter of official devel- opment assistance was tied, meaning that the procure- 5 ment contracts were limited to the donor country or a 4 group of countries. Driven by domestic political inter- ests, this practice goes against the very free-market prin- * ciples that most donors are trying to encourage in developing countries and results in inefficient use of aid. 2 It has been estimated that tying aid reduces its value by 15-30 percent.57 The practice should be ended as quickly 1 as possible, and contracts should go to the best bids.58 * * Make technical assistance demand driven. Turning more 0 responsibility over to recipient countries for designing NGOs' own DAC Official national development strategies and leading consulta- grants to contributions development national development strategies and leading consulta- developing to NGOs assistance tion meetings will require rapid capacity development. countries administered Recipient countries will also need strong auditing and by NGOs accounting skills if donors are to relinquish monitor- Note: These data cover money used for development or relief and control * ~~~~~~~~~~~purposes, The amount of official development assistance ing and control of projects. But technical assistance, the administered by NGOs is a rough estimate only, because many obvious choice for building capacity, has a spottv donors do not report this information. Souirce: OECD, DAC data. record at best, particularly in countries where capac- iry is already weak. The main reason is that it has often not been demand driven-it has often been tied aid gone into funding their evaluation and monitoring ef- and designed to develop capacity only in donor-sup- forts.62 Even with better monitoring, though, NGO ported activities. 59 Instead, technical assistance should projects face the same problems of fungibility as donor be incorporated into a national strategy and expendi- projects, and policy environments strongly influence ture plan, with the recipient government deciding their effectiveness. Donors and NGOs should continue what assistance it needs and who should provide it. This to improve their working relationships, sharing best is likely to require initial support to countries on how practices for making aid more effective in the long term. to use the market for technical assistance. * Relieve more debt. Debt relief for the poorest countries * Continue to learn about how to work effectively with is essential for effective aid. Heavy debt burdens reduce NGOs. Relationships between donors and NGOs are incentives for policy reform, while debt negotiations and complex, with much room for improvement.60 Good the constant circulation of new aid money to service old data on the extent and effectiveness of donor-NGO debt distract government officials from the needs of their relationships are scarce, but an estimated $5 billion in citizens. The next section turns to this issue. aid is now channeled through NGOs, either in sub- sidies to their activities or in contracts to implement Relieving the debt burden donor activities (figure 11.5). NGOs appear to be an of poor countries effective channel for aid when they are involved early in projects (at the design phase), when they are cho- The most prominent issue in development cooperation at sen for their proven capacity and experience, and the end of the 20th century and the beginning of the 21st when they are treated as partners rather than con- has been debt relief for the poorest countries. There has been tractors.61 The long-term impact of NGO projects re- a steady increase over two decades in the indebtedness of mains unknown, perhaps because so little money has a group of poor countries now referred to as the heavily REFORMING DEVELOPMENT COOPERATION TO ATTACK POVERTY ZOI Figure 11.6 Box 11.7 As per capita income in the heavily indebted The Enhanced Heavily Indebted Poor Countries poor countries has gone down, debt has gone Debt Relief Initiative up-and vice versa The Heavily Indebted Poor Countries (HIPC) Debt Relief Ini- Median per capita Median present value of external tiative was announced in late 1996. Realizing that the initia- 997 U.S. dolars debt as a share of GDP tive did not go far enough, leaders of the Group of Seven (G-7) 350 80 countries endorsed an Enhanced HIPC Initiative at a summit in Cologne, Germany, in July 1999. The enhanced initiative External debt was approved by the full membership of the World Bank and International Monetary Fund in September 1999 as an inte- 70 gral part of the new poverty reduction strategy initiative (see box 11.2). The Enhanced HIPC Initiative changed the eligi- Per capita / ~ \ | \ bility requirements for debt relief and the timing of relief. 30income rN \ 6 300 6 Eligibility To be eligible, a country must be very poor, have an unsus- tainable debt burden, and pursue good policies. 50 * Poor is defined as both eligible for support under the IMF's Poverty Reduction and Growth Facility (the reformed and renamed Enhanced Structural Adjustment Facility, or 250 ESAF) and eligible only for concessional financing from 1979 1985 1990 1995 1997 the World Bank, through the International Development Association. Note: The observed association between declining income and * An unsustainable debt burden is defined as a stock of debt rising debt should not be viewed as implying that debt reduction that is more than 150 percent of exports in present value will automatically result in higher incomes. Government policies are terms after the full use of traditional debt relief mecha- the key to growth and poverty reduction, and bad policies can lead nisms or (for countries with certain structural character- to both higher debt and lower incomes. istics) a ratio of debt to government revenue of more than Source: Easterly 1999c 250 percent. I Good policies are interpreted to mean macroeconomic, structural, and social policies consistent with poverty re- duction and sustained growth. indebted poor countries (figure 11.6). Public attention has These new eligibility criteria increase from 26 to 33 the num- been drawn to their plight in large part through the tire- ber of countries likely to qualify for relief. less efforts of NGOs in developed and developing coun- tries, whose campaign for debt cancellation by 2000 has Tmn frle tries, whose campaign for debt cancellation by 2000hasThe Enhanced HIPC Initiative provides for the possibility of captured the world's interest. 63 At the 1999 annual meet- interim relief for countries after they pass the decision point, ings of the World Bank and IMF, member countries agreed when the World Bank and IMF determine a country's eligi- on an enhanced plan for debt relief, an acknowledgment bility. A reduction in debt service payments is therefore pos- of the detrimental effects of debt on country policy envi- sible even before a country reaches the completion point, when the stock of debt is reduced. Under the earlier HIPC ronments and overall expenditure frameworks (box 11.7). agreement the debt stock was reduced only after comple- tion of two full ESAF programs-a minimum of six years. Now The effects of heavy debt burdens the completion point can be moved up if the country's per- Many heavily indebted poor countries spent as much as formance is particularly good. Relief is intended to be front- loaded as much as possible. a fifth of their annual budgets on debt service in the 1 990s, Combined with traditional debt relief arrangements, the and some spent much more.64 Because this is often more Enhanced HIPC Initiative is likely to cut by half the net pre- than the amount spent on social programs, debt servic- sent value of public debt for the 33 countries likely to qual- ing is viewed by many as a severe impediment to im- ify. As many as 20 countries may reach a decision point on proving the lives of the world's poor. debt relief by the end of 2000, depending on progress in de- veloping their poverty reduction strategies and on how much It has been argued, however, that debt servicing is financing is available from donors. not really a problem because heavily indebted poor countries receive more money from donor countries Source: World Bank lwww.worldbank.org/hipc). than they pay back. Actual debt service payments are 202 WORLD DEVELOPMENT REPORT 2000/200I Figure 11.7 Concessional transfers largely compensate for negative net transfers of nonconcessional resources Net transfers to heavily indebted poor countries by creditor, 1988-97 Billions of U.S. dollars 25 20 15 10 5 - --- -5 -10 International Bilateral Other Other International Private Bilateral International Bank Development concessional multilateral multilateral Monetary Fund nonconcessional for Reconstruction Association concessional nonconcessional and Development Source: Easterly 1999c. almost always far less than scheduled payments, because itoring take up much of the already stretched time and the countries cannot make the full payments. The capacity of government officials. debts are serviced by rescheduling some loans and fi- These resource inflows can also be unstable, making nancing the rest through a combination of new loans it difficult for governments to manage their spending and and grants.65 Overall, while net transfers of noncon- maintain sound fiscal policies.67 Furthermore, if resource cessional resources tend to be negative because new non- flows are positive because countries have to rely on con- concessional borrowing is strongly discouraged, transfers tinuous recheduling and on grants and concessional lend- of concessional resources tend to more than compen- ing, their access to private capital flows will remain very sate (figure 11.7). low. And where debts are not serviced in full, countries' However, heavy debt burdens bring additional prob- debt stocks continue to grow, creating a potential disin- lems that can affect a country's growth performance and centive to investment, since investors may fear that fu- ability to focus government action on social priorities. ture profits will be affected by debt-related macroeconomic Debt service is financed largely by scarce domestic bud- problems or higher taxes to service debt.68 getary resources and thus competes with domestic re- Debt is therefore as much a problem of how gross flows current spending, while concessional assistance goes to and debt management affect ownership, policy, and ca- new investment projects. This mix can mean resources pacity as it is a problem of net flows. In this, it shares for new health centers and roads but not for nurses or many of the problems that have diminished the effec- maintenance. In addition, many grants go to donor- tiveness of aid. Debt relief can play an important role here managed activities that are not included in the budget. by reducing the burden on recurrent budgets and allowing These are subject to all the problems of ownership and government officials to focus on sound spending strate- donor coordination discussed above and can contribute gies rather than continual renegotiation of debt. And it to the further institutional weakening of an already weak- can be particularly crucial for countries emerging from ened, insolvent state.66 And debt negotiations and mon- civil conflict and war. REFORMING DEVELOPMENT COOPERATION TO ATTACK POVERTY 2C3 There is also some evidence that high debt service ily indebted poor countries.71 That is why the principile obligations (including those to international financial is to grant debt relief on the basis of reputation-an es- institutions) tend to weaken the link between conces- tablished track record in using resources effectively for sional flows and the quality of the policy and institu- poverty reduction. tional framework-and so the effectiveness in reducing The design of the Enhanced Heavily Indebted Poor poverty. This could be because donors try to avoid de- Countries (HIPC) Debt Relief Initiative incorporates faults on loans, and as countries become more in- these lessons. Debt relief will be granted to eligible coun- debted, donors give new loans to cover the old ones. tries with a viable and comprehensive poverty reduction (Between 1989 and 1997 debt relief for the 41 heav- strategy and a framework for linking public actions to ily indebted poor countries totaled $33 billion and monitorable results in poverty reduction. The strategy is new borrowing $41 billion.)69 Not only does this com- to be defined through a participatory process involving promise the ability of donors to target aid to where it government, the private sector, and civil society. The will be most effective, but it may also deter reform in participatory process is important for the design of the countries with poor policies, because they have less strategy-and to help ensure good use of external (and incentive to reform if they can expect relief and resources internal) resources. Debt relief will be integrated with other anyway.70 sources of external finance in the country's overall bucd- Debt relief can ease all these problems by reducing getary framework for poverty reduction, rather than the gross flows and, if structured correctly, encouraging being earmarked for certain expenditures. The goal of the a structure of new inflows that is more effective for Enhanced HIPC Initiative is to contribute directly to poverty reduction. poverty reduction and to ensure that countries that re- ceive debt relief do not have policies that will lead themn An improved initiative for debt reliejf deeply into debt again. To be effective, debt relief needs to be delivered in ways In May 2000 Uganda became the first country to re- that encourage country ownership, using instruments ceive debt relief under the Enhanced HIPC Initiative that provide incentives to use the resources for poverty (box 11.8). The relief was based on several years of reduction. This is the same issue as for traditional aid flows, progress in the participatory formulation of its poverty but in the context of a one-time decision to reduce debt. reduction strategy, results in key areas (getting children How much impact debt relief has on net transfers to a into school, reducing income poverty through agricul- country depends, of course, on what happens to gross aid tural and aggregate growth), and mechanisms to help flows-on whether the resources for debt relief are ad- increase accountability for public funds and reduce ditional or not. But even if the resources are not entirely leakages. additional, debt relief can ease policy and budgetary con- The cost of the Enhanced HIPC Initiative is esti- straints for the recipient country, since it frees up resources mated at $28 billion. If the debt relief is to be add.i- from the recurrent budget. What will guarantee that tional, financing must come from outside the normal these resources are used for poverty reduction? There are aid and concessional lending budgets of donor insti- two related challenges: tutions. Under current plans the cost will be financed * Linking resources from debt relief to results in roughly equally by bilateral and multilateral credi- poverty reduction. tors. Although many donors have endorsed the En- * Strengthening accountability in the use of public re- hanced HIPC Initiative and made political sources, to minimize diversion to other uses (especially commitments for funding, the mobilization of re- through corruption). sources has been slow, and some donors have not yet The lessons from the past-including those from the committed to the initiative. Because a key principle un- experience with aid outlined above-indicate that both derlying the initiative is that debt relief should be co- are best tackled through their links to the overall policy ordinated among all creditors, with broad and equitable and institutional environment, especially for public re- participation, this lagging of resources and commit- source use. Experience also shows that debt relief alone ments seriously endangers the initiative. Donors need will not improve policies. Twenty years of gradually in- to give high priority to securing sufficient funding creasing debt relief have not improved policies in heav- for the Enhanced HIPC Initiative. 204 WORLD DEVELOPMENT REPORT 2000/200I Box 11.8 How debt relief fits into a poverty reduction strategy: Uganda's Poverty Action Fund Fundamental in the fight against poverty is improving the overall and district-level water and sanitation. Specific outcome targets allocation of resources, including those from debt relief, through have been identified, such as the construction of 1,000 additional more poverty-oriented and transparent budgets. There are many classrooms to support the primary education program. ways of achieving this end, and in Uganda a special fund to use Two crucial features of the Poverty Action Fund are its integration the savings from debt relief is proving useful. into the overall budget and the Ugandan government's effort to cre- The government chose to create the Poverty Action Fund as ate a transparent and accountable structure of management. Re- a conduit for the savings from debt relief under the HIPC Initia- ports on financial allocations are released at quarterly meetings tive (about $37 million a year; the Enhanced HIPC Initiative is ex- attended by donors and NGOs. The Inspector General's office pected to double this amount). The fund has been earmarked for monitors the use of funds at the district and national levels. This priorities of the poverty eradication action plan adopted in 1997 self-imposed conditionality reflects the government's strong com- to address poverty and social conditions. The plan emphasizes main- mitment to tackling corruption. But it is also an attempt to address taining macroeconomic stability while increasing the incomes and creditor concerns about the capacity of a debtor country to link debt the quality of life of poor people by developing rural infrastructure, relief to poverty reduction. Several measures have been proposed promoting small businesses and microenterprises, creating jobs, for improving monitoring, ranging from including district-level offi- and improving health services and education. The Poverty Action cials in the quarterly meetings to having local NGOs do community- Fund focuses on schools, rural feeder roads, agricultural extension, based monitoring of the poverty fund's spending. Source: UNICEF and Oxfam International 1999. their procedures and interventions along the lines laid out in their poverty reduction strategies.73 All these issues re- Many questions remain about the implementation flect the state of international development cooperation of debt relief and of the new development cooperation at the turn of the 21st century. There is profound, on- framework advanced in this chapter. Despite the fi- going change in the way developing and developed coun- nancing difficulties of the Enhanced HIPC Initiative, some tries work together to fight poverty. observers call for even deeper and faster debt relief, ar- While many issues remain, the right direction for the guing that the debt deemed "sustainable" under the En- international community is clear. Country-focused as- hanced HIPC Initiative is still too burdensome.72 How sistance should incorporate a greater emphasis on part- to move quickly to relieve debt while still allowing enough nership between donors and developing countries. It time to build country ownership of the poverty reduc- should apply less intrusive mechanisms of aid delivery that tion strategy is another concern. Some countries wonder focus on the overall policy and expenditure environ- about their capacity to prepare their own poverty assess- ment. And it should exercise greater selectivity in allocating ments and poverty reduction strategies. Others question aid where it will be most effective. More aid and debt re- whether donors can support the formulation and im- lief need to be available to countries with effective poverty plementation of poverty reduction strategies without reduction programs. Donor evaluations of these pro- undermining country ownership. Questions also remain grams must be informed by an awareness of the condi- about the participatory process-how best to consult tions each country faces and by the new approach to with poor people, how to fit consultative processes into poverty reduction presented in this report. And to relieve the context of national political processes, and how to de- the burden of the heavily indebted poor countries, donor velop effective feedback and monitoring systems. And countries should finance the Enhanced HIPC Initiative countries wonder how well donors will be able to realign with money additional to their aid budgets. Bibliographic Note T his report draws on a wide range of World Bank White, Laurence Whitehead, L. Alan Winters, Quentin documents and on numerous outside sources. Back- Wodon, Shahin Yaqub, and David Zilberman. ground papers and notes were prepared by Daron Background papers for the report are available ei- Acemoglu, Michelle Adato, Mary B. Anderson, Michael ther on the World Wide Web (wwwworldbank.org/ R. Anderson, Simon Appleton, Gareth Austin, Michael poverty/wdrpoverty) or through the World Develop- Banton, Pranab Bardhan, Paolo Belli, Timothy Besley, ment Report office. The views expressed in these pa- Pilwha Chang, Monique Cohen, Michelle Connolly, pers are not necessarily those of the World Bank or of Richard C. Crook, Robert A. Dahl, Partha Dasgupta, this report. Shelton Davis, Alain de janvry, Stefan Dercon, Ann Many people, both inside and outside the World Elwan, Gary S. Fields, Gary Gereffi, Gregory Graff, Bank, gave advice and guidance to the team. Valuable George Gray-Molina, Lawrence Haddad, John Har- comments and contributions were provided by Taoufik riss, Ronald J. Herring, John Hoddinott, Naomi Hos- Ben Abdallah, Richard Adams, Nisha Agrawal, Sadiq sain, Peter P. Houtzager, Rajshri Jayaraman, Noushin Ahmed, Martha Ainsworth, George Akerlof, Harold Kalati, Marcus Kurtz, Edward E. Leamer, Jennifer Alderman, Titus Alexander, Jock Anderson, Hutton Leavy, David Lindauer, Michael Lipton, Frances Lund, Archer, Anthony Atkinson, Gareth Austin, Robert Daniel S. Manning, James Manor, Martha Argelia Ayres, Malcolm Bale, Namrata Bali, Andrew Balls, Martinez, Jacob Meerman, Mick Moore, Samuel A. Abhijit Banerjee, Pranab Bardhan, Christopher Barham, Morley, Kimberly J. Niles, Anthony Oliver-Smith, Douglas Barnes, Tamsyn Barton, Ananya Basu, Kaushik Jonathan Pattenden, Anan Pawasuthipaisit, Louis Basu, Amie Batson, Anthony Bebbington, Alan Berg, Pouliquen, Kameshwar Prasad, James Putzel, Danny Timothy Besley, Gordon Betcherman, Andre Beteille, Quah, Elisa Reis, James A. Robinson, Francisco Ro- Surjit Bhalla, Vinay Bhargava, Ela Bhatt, Mihir R. driguez, Elisabeth Sadoulet, Sombat Sakuntasathien, Bhatt, Hans Binswanger, Nancy Birdsall, Yonas Biryu, Peter K. Schott, Jennefer Sebstad, Saurabh Sinha, Lina Mark Blackden, Rebecca Blank, David Bloom, Zelco Song, Smita Srinivas, Alan Sturla Sverrisson, Robert Bogeic:, Jan Bojo, Rene Bonnel, Ed Bos, Cesar Bouil- M. Townsend, Ben Turok, Ashutosh Varshney, Howard l6n, Francois Bourguignon, Samuel Bowles, Carlos A. 205 206 WORLD DEVELOPMENT REPORT 2000/2001 Primo Braga, John Briscoe, Penelope Brooks, Stephen Leete, Arianna Legovini, Danny Leipziger, Brian Levy, Brushett, Robin Burgess, Sara Calvo, Sarah Cambridge, Maureen Lewis, Michael Lipton, Jennie Litvack, Laszlo Roy Canagarajah, Gerard Caprio, Teresa Carbo, Guy Lovei, James Christopher Lovelace, Landis Mackellar, Carrin, Soniya Carvalho, Robert Chambers, Jacques Fran,ois Regis Mahieu, Nick Manning, Tamar Manuelyan Charmes, Celine Charveriat, Mirai Chatterjee, Mrinal Atinc, Timothy Marchant, Rachel Marcus, Tiffany Mar- Datta Chaudhuri, Rodrigo Chaves, Sandeep Chawla, lowe, Ricardo Martin, Will Martin, Antonio Martin del Shaohua Chen, Susan Chen, Kenneth Chomitz, Alberto Campo, Keith Maskus, Andrew Mason, Simon Maxwell, Chong, Ralph Christy, Mariam Claeson, John Clark, Bill Mayville, Elizabeth McAllister, Milla McLachlan, Monique Cohen, Paul Collier, Tim Conway, Giovanni An- John Mellor, Jean-Roger Mercier, Tom Merrick, Rick drea Cornia, Uri Dadush, Dana Dalrymple, Amit Dar, Messick, Dilip Mookherjee, William Moomaw, Michael Koen M. Davidse, Adrian Davis, Gloria Davis, Alain de Moore, Mick Moore, Jonathan Morduch, Daniel Mor- Janvry, Samantha De Silva, Naa dei Nikoi, Angus Deaton, row, Robert Moulie, Peter Mousley, Ranjana Mukherjee, Klaus Deininger, Lionel Demery, Stephen Denning, Ste- Joseph Mullen, Rinku Murgai, Edmundo Murrugara, fan Dercon, Mahendra Dev, Shantayanan Devarajan, Philip Musgrove, David Nabarro, Mustapha Nabli, Reena Ishac Diwan, David Dollar, Philippe Dongier, Donna Nanavaty, Deepa Narayan, Richard Newfarmer, Juan Dowsett-Coirolo, Jean Dreze, Jean-Luc Dubois, Steven Pablo Nicolini, Michel Noel, Barbara Nunberg, Veronic Durlauf, Chris Dye, Tim Dyson, William Easterly, Judith Nyhan, Abena D. Oduro, Marcelo Olarreaga, Jonathan Edstrom, Dag Ehrenpreis, Lars Ekengren, Ibrahim El- Olsson, Azedine Ouerghi, Mead Over, Margaret Owen, badawi, David P. Ellerman, Diane Elson, Gunnar Eske- Howard Pack, Truman Packard, Sheila Page, Robert Pala- land, Wolfgang Fengler, Marco Ferroni, Deon Filmer, Ben cios, Ok Pannenborg, Sulekha Patel, Harry Anthony Pa- Fine, Ariel Fiszbein, Ann Florini, Emmanuel Forestier, trinos, Guillermo Perry, Jean Pesme, Patti Petesch, Guy Justin Forsyth, Paul Freeman, Jose Furtado, Andreas Pfeffermann, Claire Pierangelo, Jean-Philippe Platteau, Galanakis, Emanuela Galasso, Joaquin Garcia, Michel Boris Pleskovic, Louis Pouliquen, Alexander Preker, Gio- Garenne, Roberta Gatti, Guido Geissler, Alan Gelb, Paul vanna Prennushi, William C. Prince, Lant Pritchett, Fe- J. Gertler, Coralie Gevers, Ashraf Ghani, Maitreesh licity Proctor, James Putzel, Dagmar Raczynski, Atiqur Ghatak, Alan Gilbert, Michael Goldberg, Jeff Goldstein, Rahman, Mamphela Ramphele, James Rauch, Martin Fr. Xabier Gorostiaga (and his colleagues from Asociaci6n Ravallion, Susan Razzaz, Thomas Reardon, Ritva Reinikka, de Universidades Confiadas a la Compaifa de Jesus en Ana L. Revenga, Carolyn Reynolds, Helena Ribe, Michelle America Latina), Vincent Gouarne, Heather Grady, Peter Riboud, Peter Roberts, Richard D. Robinson, Alberto Ro- Grant, Stefanie Grant, Cheryl Gray, Duncan Green, Mar- driguez, John Roemer, Halsey Rogers, Andrew Rogerson, garet Grosh, Sumit Guha, Patrick Guillaumont, Sanjeew Jaime Ros, Jaime Saavedra, Elisabeth Sadoulet, David E. Gupta, Davidson R. Gwatkin, Lawrence Haddad, Peter Sahn, Joanne Salop, Susana Sanchez, Todd Sandler, Sven Hakim, Gillette Hall, Kristin Hallberg, Jeffrey Hammer, Sandstrom, Filomeno Santa Ana, Justine Sass, David Sat- Lucia Hanmer, Nancy Happe, Caroline Harper, Ricardo terthwaite, Dieter Schelling, Anita Schwarz, Christopher Hausmann, Yujiro Hayami, John Healey, Gerry Helleiner, Scott, Jennefer Sebstad, Marcelo Selowsky, Amartya Sen, Jesko Hentschel, Alicia Herbert, Norman L. Hicks, John Elena Serrano, Nemat Shafik, Shekhar Shah, Jim Shea, Hoddinott, Robert Holzmann, Peter P. Houtzager, Albert Geoffrey Shepherd, Lynne D. Sherburne-Benz, John D. D. Howlett, Chia-Hsin Hu, Gregory Ingram, Keiko Itoh, Shilling, Paul Bennett Siegel, Hilary Silver, William Sil- VijayJagannathan, SelimJahan, K. Jankovsky, Mahieujar- verman, Marcia Simoes, John Sinclair, Saurabh Sinha, ret, Renana Jhabvala, Emmanuel Jimenez, Ian Johnson, Richard Skolnick, Tova Maria Solo, Paul Spray, Lyn Gerd Johnsson, Ben Jones, Christine Jones, Steen Jor- Squire, T. N. Srinivasan, Nicholas Stern, David Stiedl, gensen, Sonia Kapoor, Dani Kaufmann, Masahiro Kawai, David Stifel, Joseph E. Stiglitz, Kalanidhi Subbarao, Allen Kelley, Charles Kenny, Michel Kerf, Christine Kes- Parita Videt Suebsaeng, Eric Valdeman Swanson, Vinava sides, Roger V. Key, Anupam Khanna, Stuti Khemani, Swaroop, Simon Szreter, Cecilia Tacoli, Kazuo Takahashi, Tony Killick, Ronald Kim, Elizabeth King, Stephan Vito Tanzi (and the team at the Fiscal Affairs Department Klasen, Jeni G. Klugman, Steve Knack, Grzegorz Kolodko, of the International Monetary Fund), David Tarr, Ju- Valerie Kozel, Annette Krauss, Alcira Kreimer, Jean-Louis dith Tendler, Sumeet Thakur, Duncan Thomas, Kirsten Lamboray, Jack Langenbrunner, Patricia Laverley, Richard Thompson, Robert Thompson, Erik Thorbecke, Mari- BIBLIOGRAPHIC NOTE 207 ano Tommasi, Lee Travers, Kerstin Trone, Carrie Tudor, Chapter 1 Wendy Tyndale, Zafiris Tzannatos, Christopher Udry, Al- Unless otherwise noted, all quotations in the chapter are drawn from berto Valdes, Dominique van de Walle, Julie van Dome- Voices of the Poor (Narayan, Chambers, Shah, and Petesch 2000; len, M. Willem van Eeghien, Wouter van Ginneken, Narayan, Patel, Schaft, Rademacher, and Koch-Schulte 2000). 1. Sen 1999, p. 87. Warren Van Wicklin, Jan Vandemoortele, Krishna Vatsa, 2. Szekely and others forthcoming. Anthony Venables, Mathew A. Verghis, Louis-Charles 3. Haddad and Kanbur 1990. Viossat, Tara Vishwanath, Milan Vodopivec, Joachim 4. Ravallion and van de Walle 1991. von Amsberg, Jayshree Vyas, Robert Wade, Mike 5. The poverty gap index is the sum of the income shortfalls Waghorne, Adam Wagstaff, Michael Walton, Kevin of all poor people-the amount bywhich their incomes fall short Watkins (and an Oxfam team), Catherine Wart, Richard of the poverty line-divided by the total population. The squared poverty gap index is the sum of the squared shortfalls. The poverty Webb, L. Alan Winters, Quentin Wodon, Adrian Wood, gap index is John Worley, Gustavo Yamada, Jacob Yaron, Shahid Q Yusuf, Roberto Zagha, and Elaine Zuckerman. N (y Y ), The team was assisted by students from-n rhe In- where N = the total population, y7 = the poverty line, ternship Program of the Washington Center: Anju Ag- y, = income of individual i, Q= the total population below the garwal, Waldo Aleriano, Juan Carlos Arandia, Hector poverty line, and a = 1. For the squared poverty gap index, Cabrera, Mario de la Cruz, Celeste de la Huerta, x = 2. When a = 0, the measure is the well-known headcount ratio. Joaquin de la Torre, Alison Drury, Nilima Gulrajani, See Foster, Greer, and Thorbecke (1984) and Foster and Shorrocks Tomoko Hagimoto, Daniel Hernandez Ruiz, Vir- (1988). ginia Iglesias, Mika Iwasaki, Alejandra Lua, Felix 6. Rowntree 1901. Marklein, Nadia Montiel, Mark Schlueter and Neil 7. World Bank 1999j. MaklnNdi one arSh uete, a8. For more details see the information on poverty reduction Thompson. strategy papers at www.worldbank.org/poverty/strategies/ Despite efforts to be as comprehensive as possible in index.htm. compiling the list above, some who contributed may 9. Mecovi is Programa para el Mejoramiento de las Encucs- have been inadvertently omitted. The team apologizes to tas y la Medici6n de las Condiciones de Vida en America Latina ayw weenreea s ra y el Caribe (Program for the Improvement of Surveys and the Mea- any who were and reiterates its gratefulness to all who surement of Living Conditions in Latin America and the contributed. Caribbean). See www.iadb.org/sds/pov. 10. Townsend 1985. Endnotes 11. Atkinson and Bourguignon forthcoming; Chen and Raval- lion 2000. Overview 12. World Bank 1999t. Unless otherwise noted, all quotations in the overview are drawn 13. World Bank 20001. from Voices of the Poor (Narayan, Chambers, Shah, and Petesch 14. Demery 1999. 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 15. Wodon, Ayres, Barenstein, Lee, Peeters, Siaens, arLd 2000). Yitzhaki 2000. 1. Sen 1999. 16. The incidence of urban poverty in Colombia has increased 2. This account is of a visit by Ravi Kanbur, director of the steadily since 1996, however. report until May 2000. 17. World Bank 2000e. 3. Note that 1998 is the most recent year for which data are 18. Suryahadi and others 1999. Poverty appears to have fallen available. However, figures for 1998 are preliminary, from February 1996 to around the third quarter of 1997. There 4. While these numbers provide a sense of broad trends, they are several alternative estimates of Indonesian poverty, discussed should be treated with caution in light of the data shortcomings in detail in Suryahadi and others (1999), but they tell a reason- mentioned in chapter 1 and the fact that figures for 1998 are ten- ably consistent story. The increase in the poverty rate from the tative because of the limited number of surveys available (see box lowest point (August-October 1997) to the highest (Septem- 1.2). ber-December 1998) is around 164 percent, which can be regarded 5. Hanmer and Naschold 1999. as the maximum impact of the crisis on poverty. The poverty rate 6. Hanmer and Naschold 1999; McGee 1999. peaked around the middle of the second half of 1998, following 7. For a discussion of the relationship between environment the large surge in the price of rice and before the beginning of the and growth, see chapter 4 of World Bank (2000p). stabilization of general inflation. 8. For further information see the UNICEF statistical data- 19. Data on the distribution of consumption are not available base at www.unicef.org/statis. for both 1996 and 1998. Thus an estimate of consumption was 208 WORLD DEVELOPMENT REPORT 2000/2001 obtained by multiplyinlg all incomes by the share of aggregate pri- Naravan, Patel, Schafft, Rademacher, and Koch-Schulte 2000). vate consumption in national income based on national accounts Reference material for this chapter is cited in the relevant data. Actual consumption data, available for 1998, suggest that chapters of the report rather than repeating it here. the estimation procedures yield broadly accurate headcount ratios. 20. World Bank 1999bb. 21. World Bank 1999t. Chapter 3 22. Demery 1999. Unless othervise noted, all quotations in the chapter are drawn 23. iLUstig 1998. from Voices of tbe Poor (Narayan, Chambers, Shah, and Petesch 24. Yao 1999. 2000; Naravan, I'atel, Schafft, Rademacher, and Koch-Schulte 25. Minot 1998; Jalan and Ravallion 1 999b. 2000). 26. Lopez and della Maggiora forthcoming. 1. Lipton and Ravallion 1995. 27. Bonilla-Chacin and Hammer 1999. 2. Figures are in constant 1990 U.S. dollars adjusted for dif- 28. Wagstaff forthcoming b. ferences in purchasing power parity, as reported in Maddison 29. Wagstaff forthcoming b. (1995). 30. World Bank 1998t. 3. Bourguignon and Morrisson 1999. 31. World Bank 1999t. 4. Sen 1999. 32. Filmer and Pritchett 1998. 5. On the effects of human capital on growth, see Barro 33. Filmer 1999b; Ghana Statistical Service 1'999. (1997) and Bhargava and others (2000). 34. Liu, Hsiao, and Eggteston 1999. 6. Summers 1993. T'here is debate on this. See Benhabib and 35. Vclla 1997. Spiegel (1994). who find no effect of human capital growth on out- 36. World Bank forthcoming a. put growth, and Krueger and Lindahl (1999). who argue that the 37 . Filmer 1999b. lack of association is due to measurement error. 38. World Bank 1998t. 7. See, for example, Mankiw, Romer, and Weil (1992), who 39. In China ethnic minority groups make up less than 9 per- find a consistently negative and sometimes significant effect of pop- cent of the population, but are believed to account for about 40 ulation growth on output growth consistent with the predictions percent of the remaining absolute poor (World Bank 2000e). of the Solow model. Levine and Renelt (1992) find more mixed 40. Psacharopoulos and Patrinos 1994. evidence of the robustness of this result, while Kelley and Schmidt 41. Psacharopoulos and Patrinos 1994. (1994) argue more strongly for a negative effect. Young (1995) 42. World Bank 1999s. and Bloom and Williamson (1997) provide evidence on the im- 43. Gragnolati 1999. portance of demographic change for growth in East Asia. 44. U.S. Census Bureau 1999. 8. On openness and growth, see Sachs and Warner (1995) and 45. World Bank 1997g. A 1994 survey by the National Coun- Frankel and Romer (1999). A critical review of the evidence is pro- cil of Applied Econonmic Research shows that in rural India roughly vided by Rodriguez and Rodrik (1999), and a rebuttal of this re- halfthe members ofscheduled castes and scheduled tribes lived below view can be found in Bhagwati and Srinivasani (1999). On fiscal the poverty line. and that while these groups represent about a third policy and growth, see Easter-ly and Rebelo (1993). On inflation of the population, they make up 43 percent of poor people. and growth, see Bruno and Easterly (1998) and Barro (1997). And 46. Kozel and Parker 2000; PROBE Team 1999; World Bank on finanicial development and growth, see Levine (1997). I 998r, table 2: World Bank 1 999z, subnational database of India. 9. See Burnside and Dollar (forthcoming). 47. Dreze and Sen 1995. 10. On civil uLnrest, terms of trade shocks, and growth, see East- 48. Deaton 2000. erly and others (1993). On volatility and growth, see Ramey and 49. Baulch and Hoddinott forthcoming. Ramey (1995). And on slow trading partner growth, see Easterly 5)l. In addition, one caveat in interpreting these results is that (2000b). the pcople who move frequently in and out of poverty may be pri- II. On corruption and growth, see Mauro (1995). On rule marily those whose income is permanently close to the poverty of law and growth, see Kaufmann, Kraav, and Zoido-Lobaton line. lfthat is the case, mansv who are considered transitorily poor (1999). could actuallv be chronically poor. 12. Rodrik 1998. 51. Jalani and Ravallion 1998b, 1999b. 13. On ethnic fragmentation and growth, see Easterly and 52. Fields 1999. Levine (1997). On the role of institutions, see Easterly (2000a). 53. Falkingham 1999. 14. See Gallup, Sachs, and Mellinger (1999). 54. Okrasa 1999. 15. Ros 2000. 55. Braithwaite 1997. 16. World Bank 2000p. 56. Klugmani and Braithwaitc 1998. 17. World Bank 1997c. 18. World Bank 1997a. 19. Rovaume du Maroc, Ministere de la Prevision Economique Chapter 2 et du Plan, Direction de la Statistique 1999; World Bank 2000j. Unless othervise noted, all quotations in the chapter are drawn from 20. Bruno, Ravallion, and Squire 1998; Deininger and Squire Voices of tbe Poor (Narayan, Chambers, Shah, and Petesch 2000: 1996b; Ravallion and Chen 1997; Dollar and Kraay 2000. BIBLIOGRAPHIC NOTE 209 21. Dollar and Kraay 2000. to impatience, aggressiveness, and arrogance.... The transition 22. Li, Squire, and Zou 1998. from socialism to capitalism .. . is a trial and error process. 23. World Bank 1997b. Each element in the process might be very rapid, fairly rapid or 24. Thorbecke and Jung 1996. slow. Each has its own appropriate speed." 25. Bourguignon and Morrisson 1998. 6. This point is emphasized in World Bank and IMF (2000b). 26. Psacharopoulos and Patrinos 1994. 7. ForMexico, see Lustig (1998); for transition economies, see 27. Banerjee and others 2000. EDRB (1999). 28. Birdsall and Londofio 1997; Deininger 1999a. 8. Birdsall, Graham, and Sabot 1998. 29. Ravallion 1998. 9. IDB 1997. 30. Thomas and Wang 1998; Klasen 1999; Dollar and Gatti 10. Lindauer 1999. 1999. 11. World Bank 1996f. 31. Alesina and Rodrik 1994; Persson and Tabellini 1994; Per- 12. Bajpai and Sachs 1999. otti 1996a; Clarke 1995; Deininger and Squire 1998 (with land 13. Easterly 2000b. inequaliy) . Rodriguez (1999) provides a survey of the cross-coun- 14. Berg and Taylor 2000. try evidence. 15. Tommasi and Velasco (1996) suggest this classification of 32. Li and Zou 1998; Forbes forthcoming. reforms and provide a thorough review of the political economy 33. Banerjee and Duflo 2000. of reform. 34. World Bank 1993a; Pritchett and Summers 1996; Filmer 16. Dollar and Kraay 2000; Gallup, Radelet, and Warner and Pritchett 1999a; WHO 1999b; van Doorslaer and Wagstaff 1998. There is some debate, however, over the cross-country ev- 1997; Gwatkin and others 2000; Schalick and others 2000; idence on the distributional impact of trade liberalization. See, for Wagstaff 2000; Wagstaff and Watanabe 2000; Kakwani 1993; example, Lundberg and Squire (2000). Jamisonandothers 1996; Kim andMoody 1992;Anandand Raval- 17. See, for example, Morley (1999) and Dollar and Kraay lion 1993. (2000). 35. Behrman and Knowles 1997; Oxfam International 1999; 18. World Bank 20001. Filmer and Pritchett 1999a; Behrman 1987; Hanushek and Kimko 19. Haltiwanger and Singh 1999. forthcoming; PROBE Team 1999. 20. de Ferranti and others 2000. 36. Strauss and Thomas 1998. 21. This discussion draws heavily from Meerman (1997). 37. Lavv and others 1996; Thomas, Lavy, and Strauss 1996. 22. Sarris 1994; Alderman 1994. 38. Pritchett and Summers 1996. 23. Winters 1999. 39. Barro 1997; Bhargava and others 2000; Hamoudi and Sachs 24. This discussion draws heavily on Akiyama and others 1999. (forthcoming). 40. Pritchett and Summers 1996. 25. Sahn, Dorosh, and Younger 1997. 41. On income inequaliry and average health, see Rodgers 26. Sahn, Dorosh, and Younger 1997. (1979), Flegg (1982), Waldmann (1992), and Filmer and Pritch- 27. World Bank 20001. ett (1999a). 28. World Bank 20001. 42. Anand and Ravallion 1993; Bidani and Ravallion 1996. 29. Kemal 1994. 43. Dreze and Sen 1995. 30. Hanson and Harrison 1999. 44. Filmer and Prichett 1999a. 31. Freeman 1995. 32. The country studies are V&ez, Kugler, and Bouill6n (1999) for Colombia; Contreras and others (2000) and Legovini, Bouil- Chapter 4 16n, and Lustig (1999) for Mexico; and Ruprah and Marcano Unless otherwise noted, all quotations in the chapter are drawn (1999) for Venezuela. Also see Revenga (1997). from Voices of the Poor (Narayan, Chambers, Shah, and Petesch 33. Berman and Machin 2000. 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 34. Pessino (1997) provides evidence on how labor market reg- 2000). ulations hindered labor market adjustment in Argentina in dte 1. Rodrik (2000) presents evidence that democratically elected 1990s. governments are more likely to undertake reforms. 35. Papageorgiou, Choksi, and Michaely 1995. 2. See Collier, Dollar, and Stern (2000) for a discussion of these 36. Haltiwanger and Vodopivec 1999. shifting perspectives. 37. CUTS 1999. 3. Eichengreen (1999), IDB (1997), and Tommasi and Ve- 38. Parker, Riopelle, and Steel 1995. lasco (1996) all provide careful discussions of the experience with 39. Birdsall and de la Torre (2000) provide a valuable discu s- reforms. sion of regulatory burdens in Latin America. 4. Tommasi and Velasco 1996. 40. Hallberg 1999. 5. Kornai (2000, p. 24-25) emphasizes this point and cautions 41. Hallberg 1999. against evaluations based on short-term results. "The transformation 42. Asia Foundation and AKATIGA 1999. of society is not a horse race. The main indicator of success is not 43. SEWA 1997. who passes the winning post first. Excessive emphasis on speed leads 44. Haan, Coad, and Lardinois 1998. 210 WORLD DEVELOPMENT REPORT 2000/200I 45. Chisari, Estache, and Romero 1999. Suthiwart-Narueput, and Thierfelder (2000) calculated the cost 46. ILO 2000. of raising Cameroon's export tax as 1.7 times the revenue raised. 47. Basu 1999a. 9. The debt levels of the 41 heavily indebted poor countries 48. Examples of such programs include Progresa (Programa on average far exceed their GDP and are about twice the levels de Educaci6n, Salud y Alimentaci6n, or Program of Education, considered to be sustainable (van Trotsenburg and MacArthur Health, and Nutrition) in Mexico and Bolsa Escola in Brazil. ITzan- 1999). natos (1998) shows that different incentives may be required to 10. UNICEF and Oxfam International 1999. keep children of different ages in school. 11. Gupta, Schiff, and Clements 1996. 49. Martin and Maskus (2000) review the economic arguments 12. On average in 18 heavilv indebted poor countries the for core labor standards with an emphasis on implications for trade. share of military spending in GDP fell by the same amount (0.6 Aidt, Schlemmer-Schulte, and Tzannatos (2000) provide a lengthy percent) as the increase in the share of education and health review of the empirical evidence on the benefits of the right to spending (IMF 1999). unionize and bargain collectively. 13. Costa Rica shows that low military spending need not com- 50. Pencavel (1997) reviews arguments that the gains won by promise external or internal security. unionized workers come at the expense of nonunionized work- 14. Experience varied across regions, however. There were ers rather than from profits. large increases in real per capita spending on education and health 51. It is increasingly clear that it is counterproductive to use in Asia and Latin America and the Caribbean. In Sub-Saharan Africa trade sanctions as an international enforcement mechanism for core per capita spending fell in education and rose only moderately in labor standards, as the costs of such sanctions are too often borne health. In transition economies both education and health spend- by the poor. See Marcus (2000) for a review of arguments. ing declined sharply (Gupta, Clements, and Tiongson 1998). 52. Sable, O'Rourke, and Fung 2000. 15. Patrinos and Ariasingam 1997. 53. Levine 1997. 16. Benefit incidence analvsis is a well-established tool for un- 54. Bennett, Goldberg, and Hunte 1996. derstanding who benefits from public spending, but it has limi- 55. Adams 1984; Paxton and Cuevas 1996. tations. Average benefits, even when correctly measured, may 56. See von Pischke, Adams, and Donald 1984; Yaron, Ben- not be a reliable guide to the change in aggregate spending on a jamin, and Piprek (I1997); Braverman and Guasch (1993). given program or to the distributional impact of a reallocation be- 57. See Chaves and Gonzalez-Vega (1996) and Yaron (1992), tween programs. Program participation may be nonhomoge- for example. neous, causing the marginal impacts of program expansion or 58. Morduch 1999c. contraction to differ greatly from the average impacts (Lanjouw 59. Chaves and Sanchez 2000; Sanchez 2000. and Ravallion 1999). 60. See, for example, Khandker (1998) and Morduch (I 999b). 17. van de Walle 1996. 61. Hulme and Mosley 1996. 18. In Croatia in 1998 more than 90 percent of the energy sub- 62. Morduch 1999c. sidy went to nonpoor households. In Russia in 1997 about 74 per- 63. For example, the subsidy dependence index, developed in cent of the water subsidy went to middle- and high-income Yaron (1992), measures the extent to which the lending interest consumers. In both cases subsidies increased rather than reduced rate would have to be raised to cover all operating costs in the ab- inequality (Lovei and others 2000). sence of subsidies. 19. World Bank 1994e. 20. Lewis 2000. 21. In some cases child labor may be a consequence rather than C,hapter 5 a cause of children being out of school (Oxfam International 1999). Unless othenvise noted, all quotations in the chapter are drawn 22. Bredie and Beehary 1998. from Voices of'the Poor (Naravan, Chambers, Shah, and Petesch 23. Oxfam International 1999. 2000; Naravan, Patel, Schafft, Rademacher, and Koch-Schulte 24. Azandossessi 2000; Pouliquen 1999b. 2000). 25. Constance 1999. 1. Wagstaff 2000. 26. Barnes and Halpern 2000. 2. Ghana Statistical Service 1999; Filmer 1999b. 27. Barnes, van der Plas, and Floor 1997. 3. Instituto Nacional de Estadistica v Censos 1998. 28. In energy, for example, the prospects of better services for 4. van de Walle 2000a. many poor people in the foreseeable future will depend on find- 5. Barnes, van der Plas, and Floor 1997. ing cheaper off-grid sources (Villagran 2000). 6. Levy 1996. Strong associations are often found between roads 29. Programa de Educaci6n, Salud y Alimentaci6n, or Program and social indicators, though the direction of causality is unclear of Education, Health, and Nutrition. (van de Walle 2000b). 30. Nolan and Turbut 1995. 7. China raised about 5.8 percent of GDP in central govern- 31. Gilson 1998. ment revenues, while India raised 12.2 percent (World Bank 32. Nyonator and Kutzin 1999. 2000s). 33. Gertler and Hammer 1997. 8. Ahmad and Stern (1987) calculated the additional cost of 34. World Bank 1996a. raising a rupee through the sales tax in India as 60 paise. Devarajan, 35. Boland and Whittington 2000. BIBLIOGRAPHIC NOTE 2I.[ 36. This is the case for water supply in Jakarta. Indonesia nity organizer Akhter Hameed Khan (World Bank 1992b). Know- (Boland and Whittington 2000). ing the tradeoff berween ownership and speed of implementation, 37. Lovei and others 2000. he allowed the participatory decisionmaking processes to take all 38. Barnes, van der Plas, and Floor 1997. the time needed. 39. Irwin 1997. 78. Pouliquen 1999a. 40. World Bank 1998t. 79. Barwell 1996. 41. Gertler and Solon 1998. 80. White 1997. 42. Filmer, Hammer, and Pritchett 1998. 81. Walker and others 1999. 43. Chomitz and others (1998) show that the pay differential 82. Pouliquen 1999a. needed to get medical personnel to outer areas in Indonesia is sev- 83. Rawlings. Sherhurne-Benz, and van Domelen forthcoming. eral times any realistic amount. 84. Sara and Katz 1997. 44. See www.worldbank.org/html/extdr/hnp/health/ppi/ 85. Tendler and Freedheim 1994. contents.htm. 86. Since 1987 the World Bank has approved funding for about 45. Girishankar 1999a. 1 00 multisectoral social fund-type projects worth a total of $3.4 46. World Bank 1998t; Filmer, Hammer, and Pritchett 1998. billion in more than 60 countries (Parker and Serrano forthcom- 47. van der Gaag 1995. ing). Many other donors are also supporting social funds. 48. World Bank 1998t. 87. While choice among investments can help ensure sustain- 49. World Bank 1998t; Bonilla-Chacin and Hammer 1999. ability, unrestricted choice is rare in multisectoral projects. Yet a truly 50. Filmer, Hammer, and Pritchett 1998. demand-driven approach requires an open menu of investments. 51. Better communications also have a positive impact on so- 88. Community ownership is difficult to instill for infra- cial capital (Pouliquen 1999a). structure that serves many communities. Such infrastructure is usu- 52. World Bank 1998aa. ally better managed by local or higher-level government (Malmberg 53. Izaguirre 1999. More than 90 developing countries opened Calvo 1998). their telecommunications sector to private participation in 1990-98. 89. Social funds are cross-sectoral financing mechanisms that 54. Osiptel 1996. focus on poor communities but also channel grants to local gov- 55. For more on telecenters in Senegal, see www.idrc.ca/ ernments and NGOs. Originally emergencyfinancing mechanisms, acacia/engine/eng_6.htm and www.sonatel.sn/c-telece.htm. they have evolved into long-term mechanisms for development fi- 56. World Bank 1994e. nancing (chapter 8). A common pitfall is to assume that social fund 57. Ernberg 1998. projects provide open choice. Some studies of social funds indi- 58. Subsidies may be required in telecommunications and cate that the menus of eligible projects may not be inclusive such sectors as water and energy (Wellenius 1997; Barnes and enough and that projects may be too narrowly defined (Owen and Halpern 2000). van Domelen 1998; Walker and others 1999). 59. Wellenius 1997. 90. Pouliquen 1999a. 60. Wallsten 1999. 91. Owen and van Domelen 1998. 61. Chisari, Estache, and Romero 1999. 92. Carvalho 1999a. 62. Ravallion and Wodon forthcoming. 93. In Zimbabwe a study showed that roughly 35 percent of 63. Gaynor 1996. income in rural households comes from freely provided environ- 64. World Bank 1995b. mental goods and that the share increases as income declines 65. Gaynor 1998. (Cavendish 1999). 66. King and Ozler 1998. 94. Uphoff 1998. 67. Gaynor 1996. 95. In 1990 the government of India issued a circular to state 68. Gaynor 1998. governments recommending the adoption of joint forest man- 69. Jimenez and Paqueo 1996. agement on areas of state forestland. By 1995, 15 states had 70. World Bank 1995b. adopted such programs, involving local communities in manag - 71. World Bank 1995b. ing and protecting forestlands in return for rights to use specified 72. Oxfam International 1999. forest products (Arnold forthcoming). 73. Jimenez and Sawada 1998. 96. For example, allowing village communities to own neigh- 74. Educo's failure to raise standardized test scores in mathe- borhood forests and grasslands while the government owns far- matics and languages may reflect the absence of direct incentives off forestlands (Murty 1994). for teachers, parents, and parent-teacher associations to do so 97. Arnold forthcoming. Jimenez and Sawada 1998). 98. Arnold forthcoming. 75. These concerns include increased difficulties in imple- 99. In addition, very high population growth rates may so re- menting systemwide education policies, enforcing broader national duce the benefits to members that their incentive to participate objectives, avoiding social segregation, and promoting equity if is eroded. schools accept students based on ability to pay (World Bank 1995b). 100. Uphoff 1998. 76. Pouliquen 1999b. 101. A forestry project in Nepal allowed user communities to 77. The Orangi project was founded in 1980 by the commu- take over forest management, following approval of forest man- ZI2 WORLD DEVELOPMENT REPORT 2000/200I agement plans, and issued certificates ensuring long-term rights 124. de Janvry, Gordillo, Platteau, and Sadoulet forthcoming. to forest benefits (World Bank 1989). 125. Deininger forthcoming. 102. Experience also shows that community participation is 126. de janvry, Gordillo, Platteau, and Sadoulet forthcoming. higher when appropriate technologies allow for an adequate re- 127. Manv local suppliers have network services that under- turn from activities. For example, fuelwood plantations that use sell even subsidized public companies (Solo 1998a). closely spaced planting, which results in poor survival rates for trees, 128. The number of large-scale water and sanitation man- elicit low levels of community involvement. By contrast, com- agement contracts, leases, concessions, and divestitures in devel- munity participation is high when trees are planted farther apart oping countries increased from 4 in 1993 to 29 in 2000. Private and the plantations allow for an annual income flow from non- investment in the sector peaked in 1997 at $8.4 billion (Roger timber products (agricultural intercrops, fodder or thatch grass, 1999). commercially valuable seeds or leaves; Banarjee and others 1997). 129. The main reason for this performance is the good fit be- 1 03. Devising effective, autonomous, and honest court systems tween the institutional endowment, with the long-standing ac- is as important as designing appropriate property rights systems ceptance of private participation, and the guarantees it provides (Ostrom forthcoming). to investors (Menard and Clarke 2000). 104. Banarjee and others 1997. 130. Brook Cowen and Tynan 1999. 105. Carnev 1998. 131. The system sought to cover costs through industrial cross- 106. Banarjee and others 1997. subsidies, but the resulting tariffs were so high that most industries 107. Women and children are often the primaty collectors of fuel opted out. While service declined to all users, poor households- and fodder for home consumption and sale to urban markets and both served and unserved-suffered the most (Yepes 1999). so act as de facto managers of forests. Forest products are especially 132. Price reforms should be implemented before switching important where women and children are unable to obtain sufficient to private providers or introducing programs to increase public income from agriculture or wage employment and few other options utility efficiency (World Bank forthcoming b). exist (Arnold forthcoming). In such cases forest-related activities are 133. If needed in countries where connections are few, sub- likely to be labor intensive and household based (Arnold 1998). sidies should be used for new connections, not for consumption. 108. Agarwal 1997. Where connections are widespread, such as in many countries in 109. Agarwal 1997. Eastern Europe and the former Soviet Union, well-designed block 110. Carney 1998. tariffs can be more pro-poor (Lovei and others 2000). 111. de Janvry, Gordillo, Platteau, and Sadoulet forthcoming. 134. Brook Cowen 1997. Regulatory structures need to be both 112. de Janvry, Gordillo, Platteau, and Sadoulet forthcoming. sufficiently rigid to provide investors and managers with the cer- A secure land title facilitates access to credit and induces more tainty they need to judge long-term profitability and sufficiently environment-friendly farming practices, particularly in forest flexible to adjust to changing conditions (Savedoff and Spiller areas. It also provides an incentive to invest in productivity- 1999). enhancing techniques and improve local infrastructure (Hoff, 135. While much efforr has gone into regulation to stop util- Braverman, and Stiglitz 1993; Schneider 1995). ities from abusing their monopoly power, relatively little has been 113. Deininger and Binswanger 1999. Secure property rights done to reduce that monopoly power (Webb and Ehrhardt 1998). are also a major issue for the urban poor. 136. Brook Cowen and Tynan 1999. 114. de Janvry, Platteau, Gordillo, and Sadoulet forthcoming. 137. World Bank 1998o. 115. Mearns (1999) distinguishes individuals' rights, claims, or interests in land according to whether they may be legally up- held under prevailing law (strict legality), are socially perceived to ChJapter 6 be legitimate regardless of their strict legality (social legitimacy), Unless otherwise noted, all quotations in the chapter are drawn or are exercised in practice and therefore translate into effective from Voices of the Poor (Narayan, Chambers, Shah, and Petesch control over land (effective control). 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 116. Mearns 1999. Sharma and Dreze (1996), however, found 2000). that in Palanpur, India, tenancy has lost much of its equalizing 1. Skocpol 1992; Bates 1989; North 1990. influence. As agriculture has become more capital intensive, there 2. World Bank 19971. has been growing equality between landlords and tenants, in part 3. Evans and Rauch 1999. reflecting the exclusion ofthe landless from tenancy contracts and 4. Woo-Cumings 1999; Evans 1999; Cheng, Haggard, and the increased leasing-in by large landowners. Kang 1999. 117. Parthasarthy as cited in Mearns (1999). 5. Collier forthcoming. 118. de Janvry, Gordillo, Platteau, and Sadoulet forthcoming. 6. van Rijckeghem and Weder 1997; Adams 1998. 119. Platteau and Baland forthcoming. 7. Wade 1985. 120. Sadoulet, Murgai, and de Janvry forthcoming. See box 8. Grandvoinner and Tamesis 2000. 4.4 in chapter 4. 9. Ablo and Reinikka 1998. 121. Mearns 1999. 10. Paul 1998. 122. de Janvrv, Gordillo, Platteau, and Sadoulet forthcoming. 11. Marches Tropicaux et Mediterraneens 1995. 123. Deininger 1999b. 12. IRIS-USAID 1996. BIBLIOGRAPHIC NOTE 2I3 13. Wei 1999a, 1999b. 66. For Brazil, see Tendler (1997); for Korea, Whang (1981); 14. Tanzi and Davoodi 1997; Gupta. Davoodi, and Alonso- and for Taiwan, China, Fei, Ranis, and Kuo (1979). Terme 1998; Gray and Kaufmann 1998. 67. Coirolo 2000. 15. Norton 1998. 68. Coirolo 2000. 16. Grandvoinnet 2000. 69. Sen 1997b. 17. Berkovitz, Pistor, and Rischard 1999; Posner 1997. 70. Sen 1999. 18. Bouderbala and Pascon 1970. 71. The empirical result shown in figure 6.6, however, has to 19. Dakolias 1996. be taken with caution. It could also be the case that higher growth 20. Dakolias 1996. leads to institutions of better quality. Nevertheless, historical 21. Garro 1999. analysis reveals that the creation of some foundational institutions 22. World Bank 1999a. (property rights, for example) was key in the process of development. 23. Dakolias 1996. 72. For example, Sah (1991), Bardhan (1997b), Harriss (1999), 24. Asia Foundation [www.asiafoundation.org]. Varshney (1999a), Niles (1999), and Moore and Putzel (1999). 25. Mc Clymont and Golub 2000. 73. Alesina and Rodrik 1994; Barro 1996b; Brunetti 1997; 26. Dakolias 1996. Campos 1994; Bardhan 1999a. 27. Manning 1999. 74.Johnson 1982; Wade 1991; H. Stein 1999; Woo-Cumings 28. Michael Anderson 1999; Manning 1999. 1999. 29. Yost 1999. 75. Acemoglu and Robinson 1999. 30. Bardhan 1997b. 76. Dethier 1999. 31. Bardhan 1997b; R. Adams 1986. 77. Rodrik 1999a. 32. Ranis and Stewart 1994. 78. Collier 1999c; Collier and Gunning 1999. 33. Tommasi and Weischelbaum 1999. 79. Sen 1997b; Besley and Burgess 2000. 34. Ranis and Stewart 1994. 80. Diamond, Linz, and Lipset 1988; Joseph 1999. 35. King and Ozler 1998. 81. Bardhan and Mookherjee 1999. 36. Bardhan 1997b. 82. Li, Steele, and Glewwe 1999. See also Castro-Leal and others 37. Brown and Oates 1987. (1999). 38. Hsiao 1995. 83. Przeworski and Limongi 1993; Bratton and van de Walle 39. EIU 1999a. 1997; Sah 1991; Varshney 1999a. 40. Crook and Sverisson 1999. 84. Burki and Perry 1998. 41. Fisman and Gatti 1999; Treisman 1998. 85. de Waal 1999. 42. World Bank 20001. 86. Dreze and Sen 1995. 43. Foner 1989. 87. Joshi and Moore forthcoming. 44. Ravallion 1999b, forthcoming; Litvack and Rondinelli 1999. 45. Bardhan 1997b; Bardhan and Mookherjee 1999; Harriss 1999; Dreze and Sen 1995; Burki, Perry, and Dillinger 1999. Chapter 7 46. Blair 2000. Unless otherwise noted, all quotations in the chapter are drawn 47. Manor 1999. from Voices ofthe Poor (Narayan, Chambers, Shah, and Petesch 48. World Bank 1996b. 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 49. Owen and van Domelen 1998. 2000). 50. Fox 1995. 1. Briggs 1998; Hirschman 1984. 51. Adato and others 1999. 2. Banton 1999. 52. World Bank 1999b. 3. World Faiths Development Dialogue 2000. 53. Bhatt 2000. 4. This section draws heavily on World Bank (forthcoming a) 54. Moore and Putzel 1999. for examples and other materials. 55. Skocpol 1992. 5. Das Gupta 1995, 1999; Goody 1976, 1990. 56. Moore and Putzel 1999. 6. World Bank forthcoming a. 57. Huntington and Nelson 1976. 7. Gray and Kevane 1996; Deere and Leon 1997; Agarwal 1994; 58. Uvin 1995. Sairo, Mekonen, and Spurling 1994. 59. Paerregaard 1998. 8. KIHASA and UNDP 1998; Kim 1991. 60. Bebbington 1996. 9. Das Gupta and others 2000. 61. Riddell and Robinson 1995. 10. World Bank forthcoming a. 62. Brown and Ashman 1996. 11. UNDP 1995. 63. Bebbington 1996. 12. Filmer 1999b. 64. van de Walle forthcoming; Narayan, Chambers, Shah, and 13. Quisumbing 1994; Chi and others 1998; Saito and Spurl- Petesch 1999; Narayan, Patel, Schafft, Rademacher, and Koch- ing 1992. Schulte 2000. 14. Sen and Dreze 1989; Coale 1991; Klasen 1994. 65. Das Gupta, Grandvoinnet, and Romani forthcoming. 15. Das Gupta and others 1997; Zeng and others 1993. 214 WORLD DEVELOPMENT REPORT 2000/2001 16. Jejeebhoy 1995. 64. Anderson 1983. 17. Ren 1996; Das Gupta 1995. 65. Monnet 1988. 18. Hobcraft 1993; Hill and King 1995; Klasen 1999. 66. Austin 1999; Reno 1998. 19. Joshi and others 1999. 67. Gurr and others 1993. 20. Thomas 1990, 1997. 68. See Chua (1998) on "ethno-nationalism" in Kazakhstan, 21. Khandker 1998. the Philippines, South Africa, Thailand, and Vietnam. 22. Jejeebhoy 1995. 69. Easterly 2000a. 23. Dollar and Gatti 1999. 70. Collier and Hoeffler 2000; Woolcock, Pritchett, and 24. Klasen 1999. Isham 1999. 25. Saito, Mekonen, and Spurling 1994; Quisumbing 1994; 71. IDB 2000, p.27. Of 26 countries in the region, only four Due and Gladwin 1991. have adequate data on Afro-Latin populations: Bolivia, Brazil, 26. Quisumbing 1996. Guatemala, and Peru. 27. Udry and others 1995; Udrv 1996. 72. Girtell and Vidal 1998; Narayan 1999. 28. Gopal 1999. 73. Fine 1999. 29. World Bank forrhcoming a. 74. Woolcock 2000. 30. Sen 1999; Singh 1999. 75. Kozel and Parker 2000. 31. Jones 1998. 76. Barr 1998, 1999. See also van Dijk and Rabellotti (1997). 32. Das Gupta and others 2000; Agarwal 1994. 77. Fafchamps and Minten 1999. 33. Agarwal 1994; Andors 1983; Das Gupta and others 2000; 78. Grootaert 1999b; Grootaert and Narayan 2000; Grootaert, Davin 1976, 1995; Uberoi 1999; Wolf 1985. Oh, and Swamy 1999. For an overview of these studies, see 34. Deere and Leon 1999. Grootaert (1999a). 35. Pitt and Khandker 1998. 79. Narayan and Nvamwaya 1996. 36. Besley and Coate 1995. 80. Bebbington and Carroll 2000. 37. Khandker, Khalily, and Khan 1996. 81. Fernandez 1994. 38. Rhyne and Holt 1994. 82. Singerman 1995. 39. See World Bank (forthcoming a). 83. Tendler 1997. 40. Wade 1987. 84. Uphoff 1992. 41. Psacharopoulos and Patrinos 1994. 85. World Bank forthcoming c. 42. However, spatial poverty traps can (and do) emerge purely 86. Fox 1992; Bebbington 1999. out of the interaction of factor market imperfections with exter- nalities. See Jalan and Ravallion (1999a). 43. Borjas 1992. C1hapter 8 44. Jones 1998. Unless otherwise noted, all quotations in the chapter are drawn 45. Borjas 1997; Borjas and Sucyoshi 1997; Turok 1999. from Voices of the Poor (Narayan, Chambers, Shah, and Petesch 46. Davey Smith and others 1998; Lowell and Wood 1998; 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte Meerman 1999. 2000). 47. The strongest empirical evidence for this comes from 1. Deaton 1997; Townsend 1995; Udry 1991. poor urban communities in the United States. See, among 2. The recurring nature of shocks is also referred to as the de- others, Wilson (1996) and Sampson, Morenhoff, and Earls greeofautocorrelation.Deaton(1991)developsatheoreticalmodel (1999). showing the effect of repeated shocks on buffer stock behavior. 48. Herbst 1999. 3. Some downward income fluctuations are predictable, such 49. van Nieuwkoop and Uquillas 2000. as preharvest food shortages in rural areas. While this food short- 50. Bardhan 1997a; Coate and Loury 1993. fall is strictly speaking not a risk (although its magnitude may be 51. The distinction between the two types of affirmative ac- uncertain), households will still try to smooth consumption, and tion policies is made powerfully by Loury (2000). this carries costs. They may need to borrow, to build and fill a grain 52. Coate and Loury 1993; Steele 1999. store, or to call on a social netrvork for support, activities that may 53. Holzer and Neumark 1999. be harder for the poor (Sinha and Lipton 1999). Poor farmers may 54. Bardhan 1997a. be forced to "presell" their crops to wealthier farmers or middle- 55. Horowitz 1999. men at below-market prices (Search 1999). Similarly, income 56. Borjas 1997; Casella and Rauch 1997. declines associated with old age are often predictable, but their mag- 57. Bates 1999. nitude may not be because of uncertainties of health and life ex- 58. Alesina and La Ferrara 1999. pectancy. The chapter's discussion of risk management includes 59. Turton 1997. management of predictable fluctuations because many of the 60. See Alesina, Baqir, and Easterly (1998) on U.S. cities. tools available for dealing with risk also help households smooth 61. Collier and Garg 1999. predictable fluctuations. 62. Brockerhoff and Hewett 1998. 4. Townsend 1994. 63. Varshney 1999b. 5. Rosenzweig and Binswanger 1993. BIBLIOGRAPHIC NOTE 215 6. Dercon 1999; Kinsey, Burger, and Gunning 1998. 40. Ravallion and Chaudhuri 1997; Deaton 1997; Town- 7. Rosenzweig and Binswanger 1993. send 1995; McDonald, Schiller, and Ueda 1999. 8. Jalan and Ravallion 1997a. 41. Jalan and Ravallion 1997a; Townsend 1994. 9. Baulch and Hoddinott forthcoming. 42. See, for example, Kochar (1999) and Jacoby and Skoufias 10. Gaiha and Deolalikar 1993. (1997) for India, Moser (1998) for Ecuador and Zambia, and Der- 11. Maluccio, Haddad, and May 1999; Glewwe and Hall con (1999) for Ethiopia. 1998. 43. Migration is a coping strategy but also serves insurance and 12. Ravallion 1988. income diversification purposes. Of course, much migration takes 13. Severe poverty was measured by the squared poverty gap place for reasons other than risk management. A detailed discus- index Ua-lan and Ravallion 1997a). sion of the role of migration (domestic and international) is be- 14. Grootaert, Kanbur, and Oh 1997; Jalan and Ravallion yond the scope of this chapter. 1999c; Lokshin and Ravallion 2000a. 44. Udry 1999. 15. Lokshin and Ravallion 2000a. 45. Foster 1995. 16. Dercon and Krishnan 2000a. 46. E. Rose 1999. 17. The three-tiered framework of risk reduction, risk miti- 47. Foster 1995. gation, and coping is developed in Holzmann and Jorgensen 48. Behrman 1988. (2000) and World Bank (2000q). Some of the literature on risk 49. Behrman and Deolalikar 1990. uses a different classification that labels only ex ante strategies as 50. E. Rose 1999. "risk management" and ex post strategies as "coping." Risk man- 51. Dercon and Krishnan 2000b; Behrman and Deolalikar agement is seen as all efforts at reducing income variability (di- 1990. versification, migration). Coping consists of consumption 52. Sinha and Lipton 1999; Lanjouw and Stern 1991. smoothing, either across time (borrowing, self-insurance, saving) 53. FAO 1995. The effects of AIDS may change the customary or across households (risk sharing, through formal or informal in- treatment of widows. In a Tanzanian village AIDS widows are no surance). See Alderman and Paxson (1992). longer dispossessed, so that they can take care of the many AIDS 18. Households diversify their income sources not only to deal orphans and compensate for the reduction in the male labor sup- with risk but also to increase income. Through such activities as ply (Rugalema 1999). intercropping and seasonally varying their labor supply, households 54. Alderman and Gertler 1997. increase their mean income and in some cases may reduce their 55. Grootaert 1998. risk as well (Dercon 1999). 56. Jacoby and Skoufias 1997. 19. Reardon 1997. 57. Narayan, Chambers, Shah, and Petesch 2000. 20. Fafchamps, Udry, and Czukas 1998. 58. Foster 1995. 21. Dercon 1999. 59. Hoddinott and Kinsey 1998. 22. Reardon 1997; Dercon and Krishnan 1996. 60. Zimmerman and Carter 1999. 23. Adams 1995. 61. Barrett and Carter 1999; Grootaert and Kanbur 1995a. 24. Adams 1999. 62. This priority ranking reflects a concern for preventing the 25. Morduch 1990. hardship and deprivation that frequently result from shocks, es- 26. Dercon 1996. pecially for the poor. It does not imply that all risk should be elim- 27. Sinha and Lipton 1999. inated, because this would have strong, undesirable incentive 28. The theoretical model of precautionary saving and asset effects. In the extreme, the absence of all income risk would lead building is discussed in Deaton (1991) and Dercon (1 999). to an incentive to undertake no activity at all-the essence of tht 29. The covariance between asset values and income is not just moral hazard problem (Devarajan and Hammer 1997). a problem for the poor. As the recent East Asian experience has 63. This three-pronged policy response to risk is elaborated fur- shown, an economic crisis usually coincides with a collapse of the ther in Holzmann and Jorgensen (2000). securities market, undercutting the use of financial assets for pre- 64. World Bank 1998t. cautionary purposes. 65. Zaman 1999. 30. Dercon 1999. 66. World Bank 2000q. 31. Fafchamps, Udry, and Czukas 1998. 67. Dercon 1999. 32. Dercon 1998. 68. Holzmann and Jorgensen 2000. 33. Lim and Townsend 1998. 69. Cox and Jimenez 1998. 34. Jalan and Ravallion 1998a. 70. Jensen 1998. 35. Morduch 1999a. 71. CoxandJimenez 1998. 36. Dercon 1999. The extent to which marginal utilities are 72. Morduch 1999a. equalized depends on the internal homogeneity of the network. 73. Morduch 1999a; Jensen 1998. 37. Wider networks may find it more difficult to enforce the 74. Dercon 1999. reciprocal obligations among members, however. 75. Deaton 1991. 38. Morduch 1999a. 76. Morduch 1999a; Baulch and Hoddinott forthcoming. 39. Dercon 1999; Coate and Ravallion 1993. 77. Dercon 1999. 216 WORLD DEVELOPMENT REPORT 2000/2001 78. Dercon 1999. 118. van Ginneken 1999. 79. Sinha and Lipton 1999. 119. World Bank 2000q. 80. Besley 1995a; Sebstad and Cohen 1999. 120. Grootaert and Braithwaite 1998. 81. Sinha and Lipton 1999. 121. Ravallion 1991; Baulch and Hoddinott forthcoming. 82. Walker and Ryan 1990. 122. Wodon 2000b. 83. Sinha and Lipton 1999. 123. Wodon and Minowa 2000. 84. Moral hazard, in this context, means the possibility that 124. Subbarao 1997. people will engage in riskier behavior once they know the risk is 125. This diversity of objectives means that social funds can insured. Adverse selection refers to the tendency of people most be a delivery vehicle for a wide range of programs. In that sense exposed to a certain risk to apply for insurance. they differ from specific interventions, such as pensions or health 85. Sinha and Lipton 1999. insurance, discussed elsewhere in the chapter. 86. Holzmann, Packard, and Cuesta 1999. 126. Jorgensen and van Domelen 1999; Frigenti, Harth, and 87. Lund and Srinivas 1999b. Huque 1998; Bigio 1998. 88. Holzmann, Packard, and Cuesta 1999. 127. Jorgensen and van Domelen 1999; Frigenti, Harth, and 89. Birdsall and Hecht 1995. Huque 1998. 90. Gertler 2000. 128. van Domelen and Owen 1998. 91. Gertler and Gruber 1997. 129. Ravallion and Wodon forthcoming. 92. Prescott and Pradhan 1999. 130. Sebstad and Cohen 1999; Lipton 1998. 93. Gertler 2000. 131. Sebstad and Cohen (1999) provide many sources docu- 94. Gertler 2000. menting these different uses of loans. 95. World Bank 1998t. 132. Zaman 1999. 96. Part of the difficultv in measuring the incidence of dis- 133. Sebstad and Cohen 1999. ability stems from definitional confusion. The World Health Or- 134. The specific roles government, donors, and NGOs should ganization (WHO 1980) defines disability as any restriction or play in the delivery of microfinance remains an open question. Ex- lack (resulting from an impairment) of rhe ability to perform an periences across countries vary widely. But programs operated di- activity considered normal for a human being. What constitutes rectly by governments tend to have higher default rates (Morduch "normal" activity is clearly open to debate. In addition, many 1999a). surveys measure impairment rather than disability. And there is 135. Sebstad and Cohen 1999. a long-standing debate on whether a corrected impairment con- 136. Morduch 1999a, 1999c; Zeller 1999. stitutes a disability. See the discussion in Elwan (1999). 137. Subbarao and others 1997. 97. llwan 1999. 138. World Bank 1999n. 98. Khan and Durkin 1995. 139. IDB 2000. Progresa is Programa de Educaci6n, Salud y 99. Andersson, Palha da Sousa, and Paredes 1995. Alimentaci6n, or Program of Education, Health, and Nutrition. 100. Abu-Habib 1997; Harris-White 1996. 140. Filmer and others 1998. 101. Andersson, Palha da Sousa, and Paredes 1995. 102. Elwan 1999. 103. Flanagan 1999. Chapter 9 104. By comparison, in OECD countries pension coverage is Unless otherwise noted, all quotations in the chapter are drawn from 85 percent of the labor force (Palacios and Pallares-Miralles 1999). Voices of the Poor (Narayan, Chambers, Shah, and Petesch 2000; 105. Palacios and Pallares-Miralles 1999. Narayan, Patel, Schafft, Rademacher, and Koch-Schulte 2000). 106. Holzmann, Packard, and Cuesta 1999. 1. Joint Commonwealth Secretariat-World Bank Task Force 107. World Bank 1994b. In some circumstances the defined- on the Small States 2000. contribution pillar could also be managed publicly. 2. World Bank 20001. 108. Holzmann, Packard, and Cuesta 1999. 3. See, for example, Lustig (forthcoming), World Bank (1999j), 109. Holzmann, Packard, and Cuesta 1999; James 2000. Poppele, Sumarto, and Pritchett (1999), Reyes and Mandap 110. James 2000. (1999), and Lokshin and Ravallion (2000b). 111. Flanagan 1999. 4. The estimates for Indonesia should be treated cautiously, 112. Flanagan 1999; Grootaert and Braithwaite 1998; World as the large devaluations of the rupiah made it difficult to estimate Bank forthcoming a. price levels, particularly in rural areas. 113. Extensive discussions of child labor issues and policies can 5. de Janvry and Sadoulet 2000a. be found in GrooLaert and Kanbur (1995a), Grootaert and Pa- 6. See Lustig (1995). trinos (1999), and Fallon and Tzannatos (1998). Also see chap- 7. Lokshin and Ravallion 2000b. ter 4. 8. Datt and Hoogeveen 2000. 114. World Bank 1995c, 2000q; Dar and Tzannatos 1999. 9. Jacoby and Skoufias 1997. 115. World Bank 1995c. 10. See, for example, Jalan and Ravallion (1997a) for China 116. Grootaert and Braithwaite 1998. and Gaiha and Deolalikar (1993) for India. 117. World Bank 2000q. 11. Behrman, Duryea, and Szekely 1999. BIBLIOGRAPHIC NOTE 217 12. Binder 1999. 46. Vos, Velasco, and de Labastida 1999. 13. See Eichengreen (1999) and IDB (1995). 47. IDB 2000. 14. Tommasi 1999. 48. Morris and others 2000. 15. Morley 1994. These numbers differ from those in table 49. Reardon and Taylor 1996. 9.1 because they are from a different source. 50. Asset reduction was also the coping strategy of many 16. Caprio and Klingebiel 1996. households in Zimbabwe in the afternath of the 1991-92 drought. 17. Lustig 1998. Households were forced to sell their goats-a form of savings used 18. These are not matters for emerging markets only. Banks to pay for secondary education-to sustain short-term con- and other financial actors in developed countries have acted irre- sumption (Hicks 1993). sponsibly. In 1998 the U.S. Federal Reserve was forced to or- 51. Hoddinott and Kinsey 1998, 2000. Owens 2000. chestrate a bailout of Long-Term Capital Management, a hedge 52. Bonitatibus and Cook 1996. fund based in Connecticut. 53. Ainsworth and Over 1994. 19. De Gregorio, Edwards, and Valdes forthcoming. 54. Benson and Clay 1998. 20. Kiguel 1999. 55. Benson and Clay 1998. 21. Dornbusch and Edwards 1991. 56. Robinson 1993. 22. World Bank 1998f, 2000e. 57. Freeman and MacKellar 1999b. Major destruction of in- 23. See, for example, Bourguignon and Morrisson (1992), frastructure can have an asymmetric effect on output: the loss of World Bank (1991 b), Thorbecke (1991), deJanvry, Fargeix, and output caused by the destruction of infrastructure can exceed the Sadoulet (1991), and Bruno, Ravallion, and Squire (1998). increase in output caused by the eventual expansion of infra- 24. See the example in Lustig (2000). structure (World Bank 1994e). 25. See, for example, the discussion by Perry and Lederman 58. Kreimer 1999. (1999). 59. World Bank 1999c, 1999e. 26. Gupta and others 1998. 60. Anderson and Woodrow 1989. 27. Wodon, Hicks, Ryan, and Gonzalez 2000; Hicks and 61. The government of Nicaragua, for example, is consider- Wodon 2000. ing establishing such a system (World Bank 2000m). 28. Ravallion 1999c. 62. Benson and Clay 1998. 29. Ravallion 1999b, pp. 13-14. 63. Levav 1999. 30. Social Monitoring and Early Response Unit (www.smeru.or. 64. For a discussion on social funds, see chapter 8. id/about.htm). 65. Davis and Oliver-Smith 1999. 31. See note 5 for chapter 1. 66. Davis and Oliver-Smith 1999. 32. Davis, Handa, and Soto 1999. 67. Magalhaes and Glantz 1992. 33. Rodrik 1998. 34. See, for example, World Bank (forthcoming b). 35. International Federation of Red Cross and Red Crescent Chapter 10 Societies 1999, tables 5 and 12. Unless otherwise noted, all quotations in the chapter are drawn 36. Freeman 1999. from Voices of the Poor (Narayan, Chambers, Shah, and Petesch 37. Patterns vary greatly, however, even within regions. The 2000; Narayan, Patel, Schafft, Rademacher, and Koch-Schulte Sahel, for example, has experienced a significant decline in aver- 2000). age rainfall between 1960 and 1990 compared with the previous 1. Ades and Glaeser 1999. three decades. In contrast, southeastern Africa has experienced an 2. Quah 1999. 18- to 20-year cycle of relatively drier and wetter periods, and East 3. Scandizzo 1998. Africa has not shown a change in trends or any cyclical patterns 4. Another reason is that developing countries may not respond (Benson and Clay 1998). to trade opportunities. This underlines the importance of policy 38. Gray and others 1999. and institutional reforms in developing countries to create a more 39. Benson and Clay 1998. favorable incentive framework. 40. USAID, OFDA 1999. The database maintained by 5. World Bank and IMF 2000b. USAID's Office of U.S. Foreign Disaster Assistance includes all 6. Anderson, Hoekman, and Strutt 1999. natural hazards declared a disaster by the U.S. government as well 7. Anderson, Hoekman, and Strutt 1999. as major undeclared disasters causing a substantial number of 8. Goldstein, Kaminsky, and Reinhart 2000. deaths and injuries and substantial damage to infrastructure, agri- 9. World Bank 2000g. cultural production, and housing. 10. Kaul, Grunberg, and Stern 1999a; Kanbur, Sandler, and 41. International Federation of Red Cross and Red Crescent Morrison 1999. Societies 1993, p. 34. 11. Cook and Sachs (1999) estimate that less than 10 percent 42. Anderson and Woodrow 1989. of aid has gone to international public goods. Although it is dif- 43. Funaro-Curtis 1982. ficult to disaggregate aid flows in this way, the 10 percent figure 44. Bhatt 1999b. is calculated on the basis of how much aid goes to regional rather 45. Benson 1997a, 1997b, 1997c. than national projects. 2I8 WORLD DEVELOPMENT REPORT 2000/2001 12. See World Bank (1 999dd) for a comparison and analysis 9. See, for example, Schraeder, Hook, and Taylor (1998) and of the Kyoto and Montreal Protocols. Maizels and Nissanke (1984). 13. UNAIDS 2000. 10. Kull, Destler, and Ramsay 1997. 14. National Intelligence Council 2000. 11. See Boone (1994) and World Bank (1998b). 15. Pilot prevention projects and national programs in Brazil, 12. Easterly 1999a. India, Senegal, and Thailand show that greater condom use and 13. For a discussion of the changing consensus on development safer injecting and sexual behavior can have substantial impact. strategies over time, see Thorbecke (2000). t6. World Bank 2000a. 14. Williamson 1990. 17. Estimate by the International AIDS Vaccine Initiative. 15. While much structural adjustment lending, especially in 18. WHO 1999b. the 1 980s, emphasized these principles, most aid, including most 19. These expenditures include but are not limited to health World Bank lending, continued to be in traditional projects, es- research. pecially in infrastructure, rural development, and, increasingly, the 20. WHO 1999a. social secto