Poverty and the Policy Response to the Economic Crisis in Liberia by World Bank Publications

A WORLD BANK STUDY

Poverty and the Policy Response to the Economic Crisis in Liberia Edited by Quentin Wodon

W O R L D

B A N K

S T U D Y

Poverty and the Policy Response to the Economic Crisis in Liberia Edited by Quentin Wodon

© 2012 International Bank for Reconstruction and Development / International Development Association or The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org 1 2 3 4 15 14 13 12 World Bank Studies are published to communicate the results of the Bank’s work to the development community with the least possible delay. The manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally-edited texts. This volume is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to the work is given. For permission to reproduce any part of this work for commercial purposes, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-5222422; e-mail: pubrights@worldbank.org. ISBN (print): 978-0-8213-8979-9 ISBN (electronic): 978-0-8213-8947-8 DOI: 10.1596/978-0-8213-8979-9 Cover photo: © Miss Hibiscus/Vetta Collection/Getty Images. Library of Congress Cataloging-in-Publication Data Poverty and the policy response to the economic crisis in Liberia / edited by Quentin Wodon. p. cm. -- (World Bank studies series ; R67) ISBN 978-0-8213-8979-9 -- ISBN 978-0-8213-8947-8 1. Poverty--Government policy--Liberia. 2. Liberia--Economic conditions--21st century. 3. Global Financial Crisis, 2008-2009. I. Wodon, Quentin. HC1075.L52P68 2011 339.4’6096662--dc23 2011048148

Contents Foreword......................................................................................................................................ix Acknowledgments.....................................................................................................................xi Acronyms and Abbreviations................................................................................................ xii 1. Poverty and the Policy Response to the Economic Crisis in Liberia: Brief Overview..................................................................................................................... 1 Part I: Poverty and Human Development Diagnostic............ 7 2. Poverty in Liberia: Level, Profile, and Determinants................................................... 9 1. Introduction....................................................................................................................... 9 2. Methodology................................................................................................................... 10 3. Poverty Profile and Determinants............................................................................... 15 4. Conclusion....................................................................................................................... 31 Annex: Sensitivity of Poverty Estimates to Caloric Threshold.................................... 32 3. Education in Liberia: Basic Diagnostic Using the 2007 CWIQ Survey................... 35 1. Introduction..................................................................................................................... 35 2. School Enrollment, Reason for Not Enrolling, and Satisfaction with Schools...... 36 3. Benefit Incidence of Public Spending for Education................................................. 51 4. Correlates of School Enrollment................................................................................... 54 5. Conclusion....................................................................................................................... 57 4. Health in Liberia: Basic Diagnostic Using the 2007 CWIQ Survey......................... 60 1. Introduction..................................................................................................................... 60 2. Patterns of Morbidity, Likelihood of Seeking Care, and Reason for Not Seeking Care................................................................................................................ 61 3. Benefit Incidence of Public Spending for Health....................................................... 74 4. Determinants of the Demand for Care........................................................................ 77 5. Conclusion....................................................................................................................... 80 Part II: Impact of Higher Food Prices and Fiscal Measures to Respond to the Crisis.................................................. 83 5. Rice Prices and Poverty in Liberia................................................................................. 85 1. Introduction..................................................................................................................... 85 2. Rice Production and Consumption in Liberia: A Brief Review............................... 87 3. Rice Production and Consumption in the 2007 CWIQ Survey................................ 88 4. Simulating the Impact on Poverty of Changes in the Price of Rice......................... 92 5. Conclusion....................................................................................................................... 97

iii

Contents

6. Benefit Incidence of Fiscal Measures to Deal with the Impact on Households of the Economic Crisis in Liberia: Comparing Import and Income Taxes...................................................................................................................... 99 1. Introduction..................................................................................................................... 99 2. Benefit Incidence of Taxes on Imported Foods........................................................ 100 3. Benefit Incidence of Proposed Income Tax Reform................................................. 103 4. Conclusion..................................................................................................................... 109 PART III: Evaluation of the Cash for Work Temporary Employment Program.................................................................................. 111 7. Ex Ante Assessment of the Potential Impact of Labor-Intensive Public Works in Liberia.............................................................................................................. 113 1. Introduction................................................................................................................... 113 2. Potential Demand for Employment Programs........................................................ 116 3. Potential Poverty Impact of Public Works................................................................ 121 4. Conclusion..................................................................................................................... 125 8. Liberiaâ&#x20AC;&#x2122;s Cash for Work Temporary Employment Project: Responding to Crisis in Low Income, Fragile Countries.................................................................... 128 1. Introduction................................................................................................................... 128 2. The Food Price Crisis in Liberia................................................................................. 129 3. High Need, Low Capacity: Cash for Work in the Liberian Context..................... 131 4. Weaving the Safety Net: Design Elements of the Cash for Work Project............ 132 5. Implementation Elements........................................................................................... 135 6. Measuring Progress: Key Impact Findings and Project Feedback........................ 142 7. Moving Forward: Lessons Learned and Future Planning...................................... 147 9. Impact of Labor-Intensive Public Works in Liberia: Results from a Light Evaluation Survey........................................................................................................... 155 1. Introduction................................................................................................................... 155 2. Evaluation of the Cash for Work Program............................................................... 157 3. Conclusion..................................................................................................................... 166 Figures Figure 2.1: Stochastic dominance by residence area, 2007................................................... 22 Figure 2.2: Poverty and per capita GDP (logarithmic scale)................................................ 23 Figure 2.3: Growth and poverty simulations, 2007............................................................... 26 Figure 3.1: Concentration curve of enrollment in public schools, 2007.............................51 Figure 4.1: Share of population sick or injured in last four weeks, 2007...........................62 Figure 4.2: Share of sick/injured persons who have requested care, 2007.........................62 Figure 4.3: Concentration curves for use of public health facilities, 2007..........................75 Figure 4.4: Concentration curves for use of private health facilities, 2007........................75 Figure 5.1: Cumulative CD curve for selected food itemsâ&#x20AC;&#x201D;Order 2, 2007........................92

Contents

Figure 7.1: Distribution of actual wage and imputed wage, 2007.....................................117 Figure 7.2: Distribution of potential beneficiaries of public works, national, 2007........120 Figure 7.3: Distribution of potential beneficiaries of public works, urban, 2007............120 Figure 7.4: Distribution of potential beneficiaries of public works, rural, 2007.............. 121 Tables Table 2.1: Scale used to compute consumption per equivalent adult................................. 12 Table 2.2: Basic needs food consumption basket, 2007......................................................... 13 Table 2.3: Poverty lines, 2007 (annual in local currency, per equivalent adult)................ 14 Table 2.4: Poverty profile based on consumption per equivalent adult, 2007................... 16 Table 2.5: Extreme poverty profile based on consumption per equivalent adult, 2007... 19 Table 2.6: Comparison of Liberia with FCFA West and Central African countries.......... 23 Table 2.7: Subjective perceptions of poverty and ability to meet basic needs, 2007......... 25 Table 2.8: Subjective indicators on vulnerability to shocks of households, 2007.............. 25 Table 2.9: Liberia—selected economic and financial indicators, 2003–10.......................... 26 Table 2.10: Correlates or determinants of poverty, 2007...................................................... 29 Annex Table 2A.1: Headcount index of poverty and sensitivity to the caloric threshold.............................................................................................................................. 32 Table 3.1: Net and gross enrollment rates in primary and secondary schools, 2007........ 37 Table 3.2: Type of school attended, 2007.................................................................................38 Table 3.3: Reason for never starting going to school, 2007................................................... 40 Table 3.4a: Reason for not enrolling for previously enrolled children by age, 2007........ 41 Table 3.4b: Reason for not enrolling for previously enrolled boys, 2007........................... 42 Table 3.4c: Reason for not enrolling for previously enrolled girls, 2007............................ 43 Table 3.5: Private household expenditure for education, shares, 2007............................... 44 Table 3.6: Private household expenditure for education, amounts, 2007.......................... 45 Table 3.7: Average time (in minutes) to the nearest infrastructure, 2007........................... 45 Table 3.8: Problems encountered at school, primary, 2007.................................................. 47 Table 3.9: Problems encountered at school, secondary, 2007.............................................. 48 Table 3.10: Problems encountered at school, post-secondary, 2007.................................... 49 Table 3.11: Problems encountered at school, all levels, 2007............................................... 50 Table 3.12a: Distribution of enrolled students by grade, all types of schooling, 2007..... 52 Table 3.12b: Distribution of enrolled students by grade, public schools only, 2007........ 53 Table 3.13: Determinants of school enrollment, 2007............................................................ 55 Table 4.1: Patterns of morbidity in last four weeks, 2007..................................................... 63 Table 4.2: Demand for health care and type of provider, 2007............................................ 65 Table 4.3: Measures taken by the household to prevent malaria, 2007.............................. 68 Table 4.4: Payment method for the consultation, 2007......................................................... 69 Table 4.5: Reason for not seeking medical care, 2007............................................................ 70 Table 4.6: Structure of household’s expenditure in health, 2007......................................... 71 Table 4.7: Household’s expenditure on health, 2007............................................................. 72

Contents

Table 4.8: Time (in minutes) to the nearest infrastructure, 2007.........................................72 Table 4.9: Satisfaction/problem with health services, 2007..................................................73 Table 4.10: Benefit incidence analysis for the use of health care facilities, 2007...............76 Table 4.11: Determinants of the demand of health services, 2007.......................................78 Table 5.1: Rice and cassava production, 1990â&#x20AC;&#x201C;2004...............................................................87 Table 5.2: Vulnerability, incomes, and livelihood profile, 2006...........................................89 Table 5.3: Structure of food consumption and role of rice, 2007.........................................90 Table 5.4: Rice consumption for different household groups, 2007.................................... 91 Table 5.5: Impact of a change in consumer or producer prices for rice on poverty, 2007....................................................................................................................... 95 Table 5.6: Impact of a change of both producer and consumer prices of rice on poverty, 2007....................................................................................................................... 96 Table 6.1: Basic statistics and benefit incidence of indirect taxes on imported food...... 101 Table 6.2: Marginal tax rate structure, Liberian dollars, 2007............................................103 Table 6.3: Distribution of households across taxable income group, Liberian dollars, 2007....................................................................................................... 105 Table 6.4: Mean and total sum value of taxable income by residence area and tax bracket, 2007...................................................................................................................... 106 Table 6.5: Contribution of areas/income group in total personal tax income receipts, 2007..................................................................................................................... 107 Table 6.6: Mean value and total taxable income under the baseline and simulated tax reform, 2007................................................................................................................ 108 Table 6.7: Share of taxable income by decile under baseline and simulated tax reform, 2007....................................................................................................................... 108 Table 7.1: Potential beneficiaries of public works among individuals aged 20-40, national, 2007.................................................................................................................... 119 Table 7.2: Estimates of project cost (wages and administrative costs), 2007.................... 121 Table 7.3: Potential leakage of public works for poverty headcount, 2007...................... 122 Table 7.4: Potential leakage of public works for extreme poverty headcount, 2007...... 123 Table 7.5: Potential impact of public works for the reduction of poverty, national, 2007.................................................................................................................... 124 Table 8.1: Beneficiaries and projects per county.................................................................. 133 Table 8.2: Number of projects and types of activities......................................................... 134 Table 8.3: Composition of local wage rates in the sample of counties............................. 137 Table 8.4: Perspectives of stakeholders interviewed on the wage rate............................. 138 Table 8.5: Key elements of the cash for work project payment system............................ 139 Table 8.6: Distribution of project participants by age.........................................................142 Table 8.7: Distribution of project participants by welfare quintile................................... 143 Table 8.8: Use of project income by households.................................................................. 143 Table 8.9: Perception of project value by participants........................................................ 146

Contents

vii

Table 9.1: Percentage of households owning various assets, CfWTEP and national survey.................................................................................................................................158 Table 9.2: Share of program participants by consumption quintile after matching procedure...........................................................................................................................159 Table 9.3: Average wage losses and gains from CfWTEP per matched participant.......159 Table 9.4: Estimated impact on poverty among program participants............................160 Table 9.5: Subjective indicators of program quality among participants.........................161 Table 9.6: Use of program income among participating households (share of funds).................................................................................................................161 Table 9.7: Arguments for and against reducing the wage rate to US$2.50......................164 Table 9.8: Overall assessment of CfWTEP performance.....................................................167

Foreword

his volume provides a collection of papers on poverty and the policy response to the economic crisis in Liberia, one of the poorest post-conflict countries in sub-Saharan Africa. The first part of the study was prepared to inform Liberiaâ&#x20AC;&#x2122;s Poverty Reduction Strategy, while the second and third parts were prepared within the context of the recent economic crisis in order to evaluate the impact on the poor of specific policies implemented by the government. The emphasis throughout is on the analysis of a nationally representative household survey implemented in 2007 by the Liberia Institute of Statistics and Geo-Information Services, and of a second survey of beneficiaries of a new cash for work temporary employment project implemented in 2009. The first part of the study provides estimates of poverty, a poverty profile, and an analysis of the determinants of poverty in Liberia. Almost two thirds of the population is poor, which is high even by sub-Saharan African standards. In addition, many among the non-poor are potentially vulnerable to poverty. The study also provides a basic analysis of the education and health sectors, as seen from the point of view of households in the survey. The chapters on education and health cover the demand for education and health services (using both statistical and regression analysis), as well as the reasons for not going to school or to seek care when ill or sick. Data are also presented on household private spending for education and health, and a benefit incidence analysis of public spending for education and health is conducted. Not surprisingly, there are major differences according to poverty or well-being status in the extent to which education and health services reach various households. The second and third parts of the study are devoted to the impact of the recent economic crisis and an assessment of selected measures taken by the government to respond to the crisis. The recent increase in food prices, and especially rice, is likely to have had a large impact on poverty, in part because a large share of the rice consumed is imported, so that price increases for consumers do not provide substantial additional revenues for local rice producers. For example, an increase of 20 percent in the price of rice alone could lead to an increase of three to four percentage points in the share of the population in poverty. In terms of their distributional impact, while the reduction in taxes on imported rice implemented by the government was not well targeted to the poor, it was still more beneficial for the poor than the reform of the income tax. As for the evaluation of the Cash for Work Temporary Employment Project (CfWTEP), which was financed by the World Bankâ&#x20AC;&#x2122;s Food Price Crisis Response initiative and implemented in 2009â&#x20AC;&#x201C;10 by the Liberian Agency for Community Empowerment, it suggests that the project was fairly well targeted to the poor and had a substantial positive impact on its beneficiaries.

Foreword

Overall, the papers in this volume provide a solid baseline according to which progress in implementing the countryâ&#x20AC;&#x2122;s Poverty Reduction Strategy can be measured in the future. The papers also give useful insights into the distributional impact of various measures implemented by the government to help households cope with the recent economic crisis. These measures include tax cuts and reforms, as well as cash for work programs aimed at increasing job opportunities and incomes for the poor. Ishac Diwan Country Director Africa Region The World Bank

Rakesh Nangia Director for Strategy and Operations Human Development Network The World Bank

Acknowledgments

his study was prepared by a World Bank team led by Quentin Wodon and consisting of Colin Andrews, Prospere Backiny-Yetna, Emily Garin, Errol Graham, Rose Mungai, Clarence Tsimpo, Emily Weedon, and Guizeppe Zampaglione, with various members of the team contributing to the different chapters. The results of the study were presented in Liberia, first at a workshop in December 2007 for the preparation of the countryâ&#x20AC;&#x2122;s first poverty reduction strategy (Part I of the study), and next at a workshop in March 2011 for the preparation of the second poverty reduction strategy (Parts II and III of the study). On the government side, the active collaboration and feedback of Ramses Kumbuyah (Executive Director, Liberia Agency for Community Empowerment) was greatly appreciated for the evaluation of the cash for work temporary employment project. Guidance and support were received from Ishac Diwan (Country Director) and Rakesh Nangia (Director for Strategy and Operations in the Human Development Network). Financial support from the Gender Action Plan, Luxemburg Poverty Reduction Partnership, and Trust Fund for Environmentally and Socially Sustainable Development for this study is gratefully acknowledged.

Acronyms and Abbreviations CBO CEP CESLY CF CFSNS CfWTEP CWIQ DFID FAO FDI GDP GoL IDA ILO LACE LEAP LEEP LGA LISGIS M&E MoE MoF MoL MoPW NGO PRS PRSP TVET UN UNDP UNMIL VAT WFP YES

Community-Based Organization Community Empowerment Project Core Education Skills for Liberian Youth Community Facilitator Comprehensive Food Security and Nutrition Survey Cash for Work Temporary Employment Program Core Welfare Indicators Questionnaire Department for International Development Food and Agriculture Organization Foreign Direct Investment Gross Domestic Product Government of Liberia International Development Association International Labor Organization Liberia Agency for Community Empowerment Liberia Employment Action Programme Liberia Emergency Employment Programme Local Government Authority Liberia Institute of Statistics and Geo-Information Services Monitory and Evaluation Ministry of Education Ministry of Finance Ministry of Labor Ministry of Public Works Non-Government Organization Poverty Reduction Strategy Poverty Reduction Strategy Paper Technical and Vocational Education and Training United Nations United Nations Development Programme United Nations Mission in Liberia Value Added Tax World Food Program Liberia Youth, Employment, Skills Project

xii

CHAPTER 1

Poverty and the Policy Response to the Economic Crisis in Liberia: Brief Overview Quentin Wodon1

After years of violent conflict, Liberia has again benefited from stability since the Accra Comprehensive Peace Agreement of 2003. Free legislative and presidential elections took place in 2005. An additional hurdle to economic recovery was overcome in December 2007 with the clearance of the countryâ&#x20AC;&#x2122;s high level of debt arrears by multilateral organizations. And in 2008, the government finalized the countryâ&#x20AC;&#x2122;s first Poverty Reduction Strategy. Yet despite substantial progress since 2003, Liberia remains one of the poorest countries in the world. The purpose of this study is to provide in one place a set of papers that were written at various points in time over the last four years on poverty and the response to the recent economic crisis in Liberia. More precisely, the objective of the study is twofold. First it is to provide a basic diagnostic of both consumption-based poverty and human development (especially education and health) in the country using the 2007 CWIQ (Core Welfare Indicators Questionnaire) survey. Second, it is to assess the likely impact on the poor of the recent economic crisis, and especially the increase in rice prices, and to document the targeting performance of measures taken by the government in 2008/09 to help the poor cope with the crisis. These measures included a reduction in import taxes for rice, a reform of the personal income tax, and the implementation of a cash for work temporary employment program. This introductory chapter outlines the topics covered in the various chapters of the study and summarizes their main results.

espite substantial progress since 2003, Liberia remains a country with very high levels of poverty and deprivation, and poverty is likely to have been exacerbated by the recent economic crisis, and especially the increase in food prices (on the impact of the crisis, see for example International Labour Organization, 2009). Apart from this brief introduction, this study provides a collection of eight chapters on poverty, human development, and the government response to the recent economic crisis in Liberia. The papers were written at different points in time over the last four years. Some of the papers, especially in part I of the study, were written as early as in 2007 to inform the countryâ&#x20AC;&#x2122;s poverty reduction strategy (Republic of Liberia, 2008). Others were written at the request of government staff form Ministries, as well as for the preparation of economic 1

A World Bank Study

and sector work carried at the World Bank (see for example World Bank, 2009). Still another set of papers were written for the Liberian Agency for Community Empowerment in order to evaluate their cash for work program which was funded by a World Bank grant, as well as for the preparation of a follow-up World Bank operation (World Bank, 2010). The objective of this study is to make these various pieces of analysis publicly and easily available in one place so that it can be used for future reference. The study is structured in three parts. Part I consists of three basic diagnostic chapters for poverty, education, and health. Part II is devoted to assessing the likely impact on the poor of the recent economic crisis, and especially the increase in rice prices, and to document the targeting performance of fiscal measures taken by the government to help the poor cope with the crisis. Part III provides an evaluation of a cash for work temporary employment program also put in place by the authorities to help the poor cope with the crisis. In terms of data sources the emphasis in the first two parts of the study is on the analysis of the nationally representative CWIQ (Core Welfare Indicator Questionnaire) household survey implemented in 2007 by the Liberia Institute of Statistics and Geo-Information Services. The sample size of the survey was 3,600 household at the national level. This was the first survey of its kind implemented in about two decades in the country, and as such it provided crucial information on standards of living and human development outcomes. In part III, much of the analysis is based instead on a special purpose survey of approximately 1,000 beneficiaries of the cash for work temporary employment program. The survey was implemented in order to assess the programâ&#x20AC;&#x2122;s targeting performance (through a comparison with data from the 2007 CWIQ survey) as well as its impact. The first chapter in Part I (chapter 2) provides estimates of poverty as well as a poverty profile and an analysis of the determinants of poverty obtained from an analysis of the CWIQ survey. The analysis suggests that slightly less than two third of the population (63.8 percent) is poor, that is with a level of consumption that does not enable household members to meet their basic food and non-food needs. If patterns of growth that had been observed before the implementation of the survey could be maintained, poverty could be significantly reduced by 2015. In terms of the profile and determinants (or correlates) of poverty, as expected, consumption levels and the probability of being poor vary substantially between households according to characteristics such as geographic location, the education and employment of the household head or spouse, and household size. Chapter 3 is devoted to education. Little has been written on education in Liberia since the start of the conflict in large part because of lack of good data. The chapter provides a basic diagnostic of Liberiaâ&#x20AC;&#x2122;s education system as seen from the point of view of households using the 2007 CWIQ survey. The analysis covers school enrollment rates as well as the reasons for not going to school, and the degree of satisfaction of households with the services received, in each case looking at various age groups and boys and girls separately, as well as at different types of facilities providing education services. Data are also presented on household private spending for education, as well as on distances to facilities. A benefit incidence analysis of public spending for education is conducted, and simple regression analysis techniques are used to assess the determinants or correlates of school enrollment.

Poverty and the Policy Response to the Economic Crisis in Liberia

Chapter 4 is devoted to health. As for education, little has been written on the health system in Liberia since the start of the conflict in large part because of lack of good data. Again, the chapter provides a basic diagnostic of Liberiaâ&#x20AC;&#x2122;s health system as seen from the point of view of households using the 2007 CWIQ survey. The analysis covers rates of illness and injuries in the population, as well as the reasons for not seeking care, and the degree of satisfaction of households with the services received when they do seek care, in each case looking at various age groups and women and men separately, as well as at different types of facilities providing care. Data are also presented on household private spending for health, as well as on distances to facilities. A benefit incidence analysis of public spending for health is conducted, and regression analysis is used to assess the determinants or correlates of the demand for care. In Part II of the study the attention shifts to the impact of the recent economic crisis, and the measures taken to help households cope with the crisis. Chapter 5 first discusses the potential impact that recent increases in food prices, and especially rice, may have had on the poor. There has been a substantial literature on the link between rice and other cereal prices and poverty. The key in this literature is often to assess the double impact that a change in the price of rice may have on producers (who benefit from an increase in prices) and consumers (who lose out when the price increases). In Liberia however, at least under the current conditions, the impact of a change in the price of rice is not ambiguous at all. This is because a large share of the rice that is consumed is imported, while the rice that is locally produced is used mostly for auto-consumption rather than for sale on the market. In such circumstances, an increase in the price of rice will result in higher poverty (even if some local producers will gain from this increase). Furthermore, because rice represents such a large share of the food consumption of households, any change in price is likely to have a rather large effect on poverty measures. Using data from the 2007 CWIQ survey, the chapter suggests that an increase or decrease of 20 percent in the price of rice could lead to an increase or decrease of three to four percentage points in the share of the population in poverty, which is substantial, and the impact on measures such as the poverty gap is even larger proportionately. The chapter also discusses the likely benefit incidence of the import tax cuts implemented by the government to deal with the crisis; for rice, the poor may have benefited from approximately 45 percent of the potential benefits from the reduction in import taxes. Chapter 6 returns to the question of the benefit incidence of import taxes, but in a comparative setting with other West and Central African countries, and it also considers other fiscal measures taken by the government in order to help households cope with the economic crisis. Specifically, apart from the temporary exoneration of import duties on food products, the chapter considers the personal income tax reform announced by Liberiaâ&#x20AC;&#x2122;s President in her January 2009 State of the Union address. The objective of the measure was to reduce top income tax rate from 35 percent to 25 percent together with excluding those with very low income to pay any income tax if they are earning less than L$54,000. The chapter provides an analysis of the consumption and income data from the 2007 CWIQ survey in order to compare the benefit incidence of both measures. While none of the measures appears well targeted to the poor, the first measure is likely to benefit the poor substantially more than the second. Apart from fiscal measures, another initiative taken by the government of Liberia to respond to the economic crisis consisted in the launch of a cash for work temporary em-

A World Bank Study

ployment program. The third part of the study consists of three chapters devoted to the analysis of the program. First, in chapter 7 which was written before the program was actually implemented, an ex ante analysis of the potential impact of such a program is provided, relying on simulation techniques rather than impact evaluation. The approach is simple. We assess who may be potentially interested in participating in the public works program by identifying working individuals without pay, as well as for every level of proposed wage in the public works, those individuals who work but now earn less than the public works wage, since all these individuals may indeed be interested in participating in the program to increase their earnings. We also consider as potential beneficiaries the unemployed whose reservation wage is below the proposed public works wage. Next, we randomly select among the pool of potential beneficiaries of the program a number of participants. Finally, we estimate for the assumed participants to the program two key parameters which affect the potential impact of the program on the poor: the targeting performance of the program, and the substitution effect of the program, whereby only part of the wages paid to beneficiaries generate additional income, because at least some of the beneficiaries would probably have done other work if they had not participated in the program. The results suggest that a cash for work program could be well targeted, but that this is by no means assured ex ante. The analysis from this paper was used at the design phase of the cash for work temporary employment program which is analyzed using ex post data in the next chapters. While chapter 7 looks at ex ante simulations of the potential impact of cash for work programs in Liberia, chapter 8 analyzes ex post the actual performance of the Cash for Work Temporary Employment Program (CfWTEP) that was implemented in 2008/09. The program was funded by the World Bank through an emergency crisis facility in response to the 2007/08 food crisis. Both quantitative and qualitative data are presented, focusing on the operational and policy experiences emerging from program implementation. This chapter analyzes the context that led to the creation and implementation of the CfWTEP in Liberia, the nature and administrative arrangements for the program, and its operational performance. The objective is to share the lessons learned from evaluation findings so that they can be useful for implementing similar programs in the future in Liberia itself or in other countries. Findings from the analysis suggest that Liberiaâ&#x20AC;&#x2122;s program was well targeted to the poor and had a substantial impact towards poverty reduction among beneficiaries. These results highlight the possibilities of implementing public works program in low capacity, post conflict settings and the scope for using the program as a springboard towards a broader and more comprehensive social safety net. Finally, chapter 9 provides a more detailed account of the quantitative component of the evaluation of the CfWTEP. It also provides an assessment of the experience with the program as well as a discussion of options for the expansion and continuation of the program. Over 2009-10, CfWTEP has provided jobs to 17,000 people in the country. In order to assess the program and suggest options for its continuation, a light evaluation survey was implemented in the country in November-December 2009 with four objectives: (i) Assessing the targeting performances of the program; (ii) Measuring the wage substitution effects among the participant; (iii) Analyzing the patterns of use of the wages received by households; and (iv) Documenting other aspects of the program. The results suggest that the performance of the program in all four areas of the evaluation was relatively good. Yet targeting to the poor could be improved in the future.

Poverty and the Policy Response to the Economic Crisis in Liberia

Taken together, the various chapters hopefully provide a good basic diagnostic of poverty and human development in the country, and a first assessment of the targeting performance to the poor of selected measures taken by the government to help the population cope with the economic crisis. While other considerations than targeting should clearly inform policy decisions, targeting to the poor should be one of the considerations, especially in a context where the poor suffered the most from the increase in food prices and the economic crisis. It appears that of the three policy interventions analyzed in this study, the cash for work temporary employment program was the best targeted measure, followed by the tax cuts on imported rice, with the reform of the personal income tax being the least well targeted interventions (but that reform was also considered and implemented for other reasons than poverty reduction).

Note 1. The author is with the World Bank. The views expressed in this chapter are those of the authors and need not reflect those of the World Bank, its Executive Directors or the countries they represent.

References International Labour Organization, 2009, â&#x20AC;&#x153;A Rapid Impact Assessment of the Global Economic Crisis on Liberia,â&#x20AC;? mimeo, Monrovia. Republic of Liberia, 2008, Poverty Reduction Strategy, Monrovia. World Bank, 2009, Liberia: Employment and Pro-Poor Growth, Report No. 51924-LR, Washington, DC. World Bank, 2010, Project Appraisal Document on a Proposed Grant in the Amount of US$16.0 Million from the Africa Catalytic Growth fund (US$10.0 Million) and the Crisis Response Window (US$6.0 Million) to the Government of Liberia for the Liberia Youth, Employment, Skills Project, Report No. 53626-LR, Washington, DC.

Part I Poverty and Human Development Diagnostic

CHAPTER 2

Poverty in Liberia: Level, Profile, and Determinants Prospere Backiny-Yetna, Quentin Wodon, Rose Mungai, and Clarence Tsimpo1 This chapter was originally drafted in 2007 to inform the diagnostic of Liberia’s first Poverty Reduction Strategy. It is based on an analysis of the Core Welfare Questionnaire Indicator survey implemented in 2007 by Liberia’s Institute of Statistics and Geo-Information Services. The chapter estimates the level of poverty and vulnerability in the country, provides a profile of poverty, and analyzes the household level determinants or correlates of consumption and poverty. Slightly less than two thirds of the population (63.8 percent) is estimated to be poor. If patterns of growth that have been observed recently are maintained, poverty could be significantly reduced by 2015. In terms of the profile and determinants of poverty, as expected, consumption levels and the probability of being poor vary substantially between households according to characteristics such as geographic location, the education and employment of the household head or spouse, and household size.

1. Introduction After many years of violent conflict that started with a coup in 1989, Liberia has again benefited from stability since the Accra Comprehensive Peace Agreement of August 2003 (on conflict as well as the transition to democracy in Liberia, see among others Kieh, 2004; Richards et al., 2005; and Sawyer, 2005). Free legislative and presidential elections took place in 2005, and the country was the first African nation to elect a woman President, Ellen Johnson Sirleaf. Demobilization efforts lead more than 100,000 soldiers to be reinserted, and most of the previously displaced population has been able to return (on evidence suggesting an ability of such programs to improve social cohesion in Liberia, see Fearon et al., 2009). An additional hurdle to economic recovery was achieved in December 2007 with the clearance of the country’s very high level of debt arrears by multilateral organizations (World Bank, 2007a). Despite substantial progress since 2003, Liberia remains today one of the poorest countries in the world. The government has recently prepared an Interim Poverty Reduction Strategy (Republic of Liberia, 2006), which organizes the country’s development strategy around four pillars: enhancing national security, revitalizing economic growth (on growth in Liberia, see also Radelet, 2006), strengthening governance and the rule of 9

A World Bank Study

law, and rehabilitating infrastructure and delivering basic services. The final poverty reduction strategy was approved in 2008 (Republic of Liberia, 2008). In order to inform the preparation of a full Poverty Reduction Strategy, a Core Welfare Questionnaire Indicator survey was implemented in 2007 by the Liberia Institute of Statistics and Geo-Information Services (LISGIS). The sample size of the survey was 3,600 household at the national level. The objective of this chapter is to utilize this survey to estimate the level of poverty and vulnerability in the country by providing a profile of poverty, and analyzing the household level determinants of poverty. The key result is that 63.8 percent of the population is estimated to be poor. This estimate of poverty is below the level obtained in a previous study by UNDP Liberia (2001, 2006), according to which 76.2 percent of the population was poor. A number of considerations suggest that the poverty estimate provided in this chapter may not be too far off from the reality of the life of the population. First, the poverty line estimated using the so-called cost of basic needs method in this chapter turns out to be of the order of magnitude of what households themselves say they need to meet their basic needs (self-assessed poverty line). Second, the estimate of poverty is in line with what one might have expected for a country with Liberiaâ&#x20AC;&#x2122;s level of economic development, given the experience of other West and Central African countries. Third, the estimate is also in line with the share of the population declaring having difficulties to live with their current income, as well as the share of the population declaring having unstable incomes. Of course, in a country as poor as Liberia, even those households who may not be poor because they have levels of consumption slightly above the poverty line may still live in precarious conditions. As to the impact of the recent economic crisis on poverty in Liberia, while it is not discussed in this chapter, some information is available in a rapid assessment carried out by the International Labor Office (2009), as well as in a companion chapter in this volume devoted to the impact of the increase in rice prices on poverty (Tsimpo and Wodon, 2012). The rest of the chapter is structured as follows. Section 2 presents our methodology for estimating poverty. Section 3 presents the key results. A brief conclusion follows.

2. Methodology This section provides a description of the methodology adopted for estimating poverty. To compute a poverty measure, three ingredients are needed. First, one has to choose the relevant dimension and indicator of well-being, which is typically the total consumption of the household per capita or per equivalent adult. Second, one has to select a poverty lineâ&#x20AC;&#x201D;that is a threshold below which a given household or individual will be classified as poor. Finally, one has to select a poverty measureâ&#x20AC;&#x201D;which is used for reporting on poverty data for the population as a whole or for a population sub-group only. All three ingredients above are described below in greater detail. 2.1. Indicator of Well-Being

The Liberia CWIQ survey consists of two questionnaires with data among others on socio-demographic variables (household composition, health, education and employment of the members of the household2), housing characteristics, levels of access to the basic services, subjective poverty perceptions, household consumption (including auto-consumption, purchases and gifts) and household income. Our welfare indicator is based on consumption per equivalent adult. Consumption is used rather than income for two

Poverty and the Policy Response to the Economic Crisis in Liberia

main reasons. First, consumption is better measured in household surveys than income. Second, consumption is a better proxy of the well-being of the household as it provides a better picture of a household’s standard of living. Third, in countries where a majority of the population works in the informal sector, net income is very difficult to measure. Various surveys use different methods to collect consumption data. One technique is to record a diary of the exact expenditure of the household over a certain period of time, but this method, while perhaps more precise, requires several visits to the same household over a period of time, and is therefore more time consuming and expensive for data collection. The other approach is to record the expenditure of households by asking them to recall these expenditures over a certain period during visit to the household. This second method may be implemented through a single visit or several visits. In the case of the CWIQ for Liberia, the second technique was adopted with a single visit per household. While it may lead to less precise estimates of poverty, the approach has the advantage to be implemented rather quickly, which was needed to enable the Liberian authorities to complete the work on their PRSP rapidly. Before using expenditure data in poverty analysis, it is important to assess the quality of the data, and whether aggregates obtained for the country as a whole are reliable. This can be done for example by comparing the consumption aggregate with an aggregate obtained from a previous survey with the previous survey being used as the benchmark. However, this type of comparison is not feasible in Liberia, due to the lack of comparable previous surveys, but at least national accounts can be used. That is, one can compare the consumption computed via the survey with GDP or private consumption in the national accounts. In Liberia, the consumption aggregate obtained from the raw data of the CWIQ turned out to be several times higher than total GDP. Therefore, several corrections were carried out in order to correct for outliers in the data. The corrections were made in the three raw data files related to “auto-consumption,” “frequently purchased items,” and “less frequently purchased items.” In the two first files which are related to food items, three important variables are measured: the number of months in the year during which the product is consumed, the quantity consumed per month (according to the unit of measure declared by the household), and the average unit value of the product (according to the specified unit). Corrections have been made on the unit values and on the quantities declared by households. For unit values, there were a number of obvious outliers. Therefore, for the unit values greater than two times the median value, the median value was imputed instead. As for quantities, a classic method of correcting outliers has been used. For each product and each unit, and for all values greater than the mean plus two times the standard deviation, the median value was imputed instead. In the third file related to non-food items, the same types of corrections were implemented. After these corrections, an aggregate file with annual expenditures by household and by item was constructed. The total consumption in the country obtained after the corrections remained high. Therefore, a second type of correction was made in terms of the share of each product in total household consumption. For each product and household, if the share of the product in total consumption was greater than the average share plus two times the standard deviation, the median share was imputed for this household and a new annual expenditure was computed consequently. This gave us the final consumption aggregate on which the poverty measures are based. The measure of total household consumption takes the following components into consideration: monetary consumption (food and non-food); auto-consumption; rent at-

A World Bank Study

tributed to households who are not tenants in their accommodation; and use value of durables. Food spending consists of daily food purchased on markets or received (for example through NGOs or the World Food Program, which is active in Liberia). Food auto-consumption was evaluated using data collected in the questionnaire. Non-food consumption includes among others spending on clothing, housing (including the estimation of imputed rent3), furnishings, education and health, transport, communication, leisure activities, the usage value of durable goods, etc. Certain categories of spending have however been excluded from the household consumption aggregate. First, some categories may be difficult to assign to household consumption due to the significant presence of people from outside the household—this is the case for spending on festivals or ceremonies during the past 12 months. In addition, some categories in the consumption questionnaire do not actually represent household consumption—this is the case for gifts given or received in cash and taxes paid during the past 12 months. Transfers received by the household are excluded from the consumption aggregate as this would lead to double counting since these monies are probably already used for consumption to satisfy household needs. In order to compute consumption per equivalent adult, instead of using the Oxford scale, which is often adopted when the country does not have information concerning the structure and composition of its households, the adult equivalence scale recommended by the FAO was used (table 2.1), which would seem closer to the reality of Africa (scale proposed by the 10th edition of the RDA, National Academy Press, 1989—NAC 89, W.D.C). This scale is not fundamentally different from the scale adopted for example by Cameroon in 2001 to define its poverty threshold. 2.2. Poverty Lines

The poverty lines are based on the cost of basic needs method. First, the food poverty lines were estimated to assess the cost of a food basket providing 2,400 Kcal per day per adult equivalent. The poverty lines were estimated separately for urban and rural areas. As specific data for Liberia were not available in terms of the caloric conversion factors for the various food items, most products in the food questionnaire were allocated the caloric values provided by a study carried out in Guinea in 2004. These caloric equivalents indicate the caloric value for 100 grams or 100 milliliters of products which are in part comestible. Table 2.1: Scale used to compute consumption per equivalent adult Scale of adult equivalence 0–1 year 1–3 years 4–6 years 7–9 years 10–12 years 13–15 years 16–19 years 20–50 years 51 years and over Source: FAO.

Male 0.27 0.45 0.61 0.73 0.86 0.96 1.02 1.00 0.86

Female 0.27 0.45 0.61 0.73 0.73 0.83 0.77 0.77 0.79

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 2.2: Basic needs food consumption basket, 2007

Rice Local rice Maize/corn Cassava flour (fufu, gari, etc.) Gari Bread Chicken Game and insects (porcupine, etc.) Fresh or frozen fish Smoked fish (dried or salted) Fresh milk Eggs Palm oil Banana, plantain Coconuts Palm nut Cassava leaves Bitter balls Okra Green pepper Hot or sweet pepper (fresh or dry) Onions Dried beans Cassava roots Sugar Bouillon cubes Salt Soft/carbonated drinks Total

Initial consumption Quantity Kilo (grams) calories 191 694 136 492 6 21 16 53 5 17 5 13 8 10 1 4 36 23 3 13 1 1 1 1 27 217 23 31 7 25 44 177 21 19 14 5 3 1 7 3 1 0 5 2 4 14 99 148 4 17 3 9 11 36 2 1 2,048

Adjusted consumption Quantity Kilo (grams) calories 224 813 159 577 7 25 18 62 6 19 6 15 9 12 2 4 43 27 4 15 2 1 1 2 32 254 27 36 8 30 52 208 25 23 17 5 3 1 8 3 1 0 6 2 5 16 116 173 5 20 3 10 13 43 3 1 2,400

Conversion coefficient 363 363 359 342 342 249 139 267 64 374 79 140 798 135 388 400 91 32 36 36 53 41 336 149 400 331 337 42

Source: Authors’ calculations using CWIQ 2007, LISGIS.

We defined a basket of food goods consumed on a regular basis (including food auto-consumption) for the entire country (see table 2.2) by the population with consumption between the second and ninth deciles (we do not use the first and last decile to avoid extreme values). The basket includes spending on the 28 food products most often consumed. These products represent more than 87 percent of total household spending on food in the country. Once the basket of food products has been defined, we determine the quantities of each product consumed per day in standard units (primarily kg or litre) per adult equivalent. Each product’s consumption is then converted into calories based on Guinea conversion tables. The amounts actually consumed for all products in the survey are adjusted in order to yield exactly a total of 2,400 Kcal per equivalent adult per day. Using the survey prices observed in the community questionnaire of the survey, we then estimate the total cost of purchasing the resulting food basket. A daily food poverty line is then estimated in urban and rural areas as follows with a normative caloric threshold of 2,400 Kcal (on the sensitivity of poverty measures to the choice of this threshold, see the annex to this chapter):

A World Bank Study

with Qi being the average daily quantity of product i consumed in the country, Ci the caloric value (for 100g or 100 ml) corresponding to product i consumed, and PiU, R being the average price of product i in urban and rural areas. Two sets of nonfood poverty lines were computed by estimating the non-food spending of (1) households whose total expenditure was equal to the food poverty line (more or less 5 percent); and (2) households whose food expenditure was equal to the food poverty line (more or less 5 percent). The total poverty lines are then the sum of the food and nonfood poverty lines. The resulting poverty lines are given in table 2.3. In what follows, the food poverty line will be used to identify the extreme poor, while the total poverty line to measure poverty is based on the second approach to estimate the non-food poverty line. Table 2.3: Poverty lines, 2007 (annual in local currency, per equivalent adult) Food poverty line

Non-food poverty line, (approach 1)

Rural

14,514.49

3,849.18

Urban

14,431.20

5,634.96

Non-food poverty line, (approach 2)

Total poverty line (approach 1)

Total poverty line (approach 2)

6,909.9

18,363.66

21,424.39

15,792.54

20,066.16

30,223.74

Source: Authors’ calculations using CWIQ 2007, LISGIS.

2.3. Poverty Measures

This section provides the mathematical expressions for the poverty measures used in the chapter. Three poverty measures of the FGT class (Foster, Greer, and Thorbecke 1984) are used, namely the headcount, the poverty gap, and the squared poverty gap (for a simple introduction to poverty measurement and profiles, see Coudouel et al., 2002). The poverty headcount is the share of the population which is poor, i.e. the proportion of the population for whom consumption per equivalent adult y is less than the poverty line z. Suppose we have a population of size n in which q people are poor. Then the headcount index is defined as: The poverty gap, which is often considered as representing the depth of poverty, is the mean distance separating the population from the poverty line, with the non-poor being given a distance of zero. Arranging consumption in ascending order y1,...., yq < z < yq+1, ..., yn with the poorest household’s consumption denoted by y1, the next poorest y2, etc. and the richest household’s consumption by yn. The poverty gap is defined as follows:

where yi is the income of individual i, and the sum is taken only on those individuals who are poor (in practice, we often work with household rather than individual consumption). The poverty gap is thus a measure of the poverty deficit of the entire population, where the notion of “poverty deficit” captures the resources that would be needed

Poverty and the Policy Response to the Economic Crisis in Liberia

(as a proportion of the poverty line) to lift all the poor out of poverty through perfectly targeted cash transfers. The squared poverty gap is often described as a measure of the severity of poverty. While the poverty gap takes into account the distance separating the poor from the poverty line, the squared poverty gap takes the square of that distance into account. When using the squared poverty gap, the poverty gap is weighted by itself, so as to give more weight to the very poor. Said differently, the squared poverty gap takes into account the inequality among the poor. It is defined as follows:

The headcount, the poverty gap, and the squared poverty gap are the first three measures of the Foster-Greer-Thorbecke class of poverty measures and a common structure is evident that suggests a generic class of additive measures (additive measures are such that aggregate poverty is equal to the population-weighted sum of poverty in various sub-groups of society). The general formula for this class of poverty measures depends on a parameter Îą that takes a value of zero for the headcount, one for the poverty gap, and two for the squared poverty gap in the following expression:

In what follows, the discussion focuses on the headcount index of poverty, but the results are very similar in terms of key messages with the higher order poverty measures (the use of higher poverty measures such as the poverty gap is often more important when evaluating the impact of policy interventions, as done for example in parts II and III of this study).

3. Poverty Profile and Determinants 3.1. Levels of Poverty and Characteristics of the Poor

Tables 2.4 and 2.5 present overall and extreme poverty estimates as well as a profile of the characteristics of the poor and extreme poor, respectively. The tables first provide the share of the population according to various categories. Next, the headcount of poverty or extreme poverty (share of the population in poverty or extreme poverty within the category) is provided. The number of the poor or extreme poor is also given, as well as the share of the total number of the poor or extreme poor in different categories. At the national level, 63.8 percent of the population is poor. This means that there are 1.7 million individuals in poverty in the country. The share of the population in extreme poverty is 47.9 percent (1.3 million people). The profile of poverty yields expected results. Poverty is higher in rural areas (67.7 percent) than in urban areas (55.1 percent). Given that close to 70 percent of the population lives in rural areas, rural areas account for almost three quarters (73.4 percent) of the poor. The region with the largest share of the poor is the North Central region, followed by Greater Monrovia (although the capital area has a much lower share of the extreme poor, as shown in table 2.5), the South Central region the North Western region, and finally the South Eastern A and B region.

A World Bank Study

Table 2.4: Poverty profile based on consumption per equivalent adult, 2007 Share of the population National

Poverty headcount

Number of poor

Urban

Rural

National

Urban

Rural

National

100.0

55.1

67.7

Urban

Rural

Contribution to poverty

National

Urban

Rural

National

63.8

459,570 1,266,236 1,725,806 100.0

100.0

—

26.6

100.0

73.4

Urban/rural location Urban

30.9

—

30.9

55.1

—

55.1

459,570

Rural

—

69.1

—

67.7

—

Region Greater Monrovia North Central

—

459,570 100.0

1,266,236 1,266,236

—

71.4

—

22.0

48.5

—

48.5

288,695

—

8.1

48.2

35.8

57.5

68.9

68.1

38,936

621,193

288,695

62.8

—

16.7

660,129

8.5

49.1

38.3

North Western

3.7

12.8

10.0

82.4

75.5

76.3

25,794

180,753

206,547

5.6

14.3

12.0

South Central

8.9

19.9

16.5

74.4

55.9

58.9

55,216

207,463

262,678

12.0

16.4

15.2

South Eastern A

5.6

10.2

8.8

76.7

76.6

76.7

35,609

146,104

181,713

7.7

11.5

10.5

South Eastern B

2.3

9.0

6.9

79.2

65.9

67.2

15,320

110,723

126,044

3.3

8.7

7.3

Less than 10

25.0

30.9

29.1

57.5

65.4

63.3

119,873

378,163

498,036

26.1

29.9

28.9

10 thru 19

26.5

22.7

23.8

57.6

72.5

67.4

127,100

307,648

434,748

27.7

24.3

25.2

20 thru 29

18.2

15.6

16.4

51.1

66.1

61.0

77,423

193,226

270,650

16.8

15.3

15.7

30 thru 39

13.7

11.9

12.4

50.8

65.1

60.2

58,155

144,485

202,640

12.7

11.4

11.7

40 thru 49

9.3

9.4

52.7

69.4

64.3

41,092

122,167

163,259

8.9

9.6

9.5

50 thru 59

4.2

4.8

4.6

57.2

67.3

64.5

20,123

60,576

80,699

4.4

4.8

4.7

60 and Over

3.1

4.7

4.2

60.3

68.0

66.2

15,804

59,971

75,775

3.4

4.7

4.4

Male

70.0

76.2

74.3

54.1

68.8

64.6

316,469

981,319 1,297,787

68.9

77.5

75.2

Female

30.0

23.8

25.7

57.2

64.1

61.6

143,102

284,917

428,019

31.1

22.5

24.8

Age of the individual

Gender of the head

Marital status of the head Single or never married

29.4

13.3

18.3

47.6

55.9

51.8

117,074

138,713

255,787

25.5

11.0

14.8

Monogamous

56.3

67.0

63.7

57.0

68.5

65.4

267,839

858,644 1,126,483

58.3

67.8

65.3

Polygamous Widowed, divorced, separated

2.4

8.0

6.3

54.1

75.5

73.0

10,935

112,910

123,844

2.4

8.9

7.2

11.9

11.8

64.4

70.8

68.8

63,723

155,970

219,693

13.9

12.3

12.7

48.0

Education level of head None

24.7

50.1

42.2

73.1

72.4

72.6

150,731

678,415

829,146

32.8

53.6

Some primary

3.9

9.3

7.7

58.7

60.7

60.4

19,291

106,101

125,392

4.2

8.4

7.3

Completed primary

3.1

4.3

3.9

78.0

67.8

70.3

20,075

54,732

74,807

4.4

4.3

Some secondary

19.1

21.8

21.0

53.5

66.0

62.5

85,266

268,988

354,254

18.6

21.2

20.5

Completed secondary

32.2

10.1

16.9

49.4

61.1

54.2

132,846

115,574

248,420

28.9

9.1

14.4

Post secondary

17.0

4.4

8.3

36.3

51.9

42.0

51,362

42,427

93,789

11.2

3.4

5.4

(Table continues on next page)

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 2.4 (continued) Share of the population Urban

Rural

National

Poverty headcount Urban

Rural

Number of poor

Contribution to poverty

National

Urban

Rural

National

Urban

Rural

National

53.2

Education level of spouse 26.2

55.8

46.7

72.5

72.8

72.7

158,268

759,955

918,222

34.4

60.0

Some primary

None

5.1

8.1

7.2

49.0

60.9

58.2

20,959

92,090

113,048

4.6

7.3

6.6

Completed primary

2.0

2.4

2.3

47.5

48.2

48.0

8,030

21,312

29,342

1.7

Some secondary Completed secondary Post secondary No spouse

11.5

5.4

7.3

57.3

69.5

63.6

54,762

70,768

125,530

11.9

5.6

7.3

9.9

2.1

4.5

40.4

47.5

42.7

33,530

18,406

51,936

7.3

1.5

3.0

6.2

0.5

2.3

18.3

13.7

17.6

9,461

1,302

10,763

2.1

0.1

0.6

39.1

25.7

29.8

53.5

62.9

59.1

174,561

302,404

476,965

38.0

23.9

27.6 10.7

Socioeconomic group of head 24.3

9.2

13.9

40.7

59.0

49.1

82,596

101,978

184,574

18.0

8.1

Private formal

Public

5.6

5.2

5.3

37.5

63.0

54.6

17,695

60,958

78,653

3.9

4.8

4.6

Private informal

6.5

3.8

4.6

52.4

52.1

52.2

28,378

36,673

65,051

6.2

2.9

3.8

Self-agriculture Self-other

3.2

46.7

33.3

79.4

71.8

72.0

21,349

627,657

649,006

4.6

49.6

37.6

27.4

16.4

19.8

54.7

62.2

59.0

125,133

190,344

315,477

27.2

15.0

18.3

Unemployed

12.1

2.5

5.4

67.6

62.9

66.1

68,094

29,377

97,471

14.8

2.3

5.6

Inactive, other

20.9

16.2

17.7

66.8

72.2

70.3

116,325

219,250

335,575

25.3

17.3

19.4 42.3

Industry of head Crop farming

3.5

53.0

37.7

80.1

71.3

71.6

23,474

706,604

730,077

5.1

55.8

Forestry/logging

0.5

0.2

0.3

23.0

91.8

56.3

887

3,306

4,193

0.2

0.3

0.2

Fishing

0.7

0.1

0.3

77.4

67.3

74.3

4,525

1,767

6,292

1.0

0.1

0.4

Mining/quarrying

0.4

0.6

0.5

78.9

69.0

71.2

2,576

7,668

10,245

0.6

Manufacturing/ processing

0.5

0.3

70.0

64.7

67.2

3,013

3,055

6,068

0.7

0.2

0.4

Electricity/gas/ water supply

1.6

0.1

0.6

31.8

14.6

30.2

4,352

215

4,566

0.9

0.0

0.3

3.1

0.7

1.5

60.1

52.7

57.5

15,406

7,380

22,786

3.4

0.6

1.3

Wholesale/retail trades

Construction

10.4

3.2

5.4

49.6

38.0

44.8

42,887

23,022

65,909

9.3

1.8

3.8

Transport, storage, communications

2.8

0.3

1.1

36.9

46.4

38.6

8,758

2,420

11,177

1.9

0.2

0.6

Banking/financial services

1.0

0.2

0.4

24.7

34.6

27.6

2,052

1,195

3,247

0.4

0.1

0.2

13.7

7.4

9.3

42.0

57.1

50.3

47,929

78,873

126,802

10.4

6.2

7.3

Community services Other

31.2

18.9

22.7

50.7

65.4

59.2

131,901

231,540

363,441

28.7

18.3

21.1

Unemployed, Inactive

30.6

15.0

19.8

67.2

71.1

69.3

171,811

199,191

371,002

37.4

15.7

21.5

Household owns cultivatable land Yes

20.8

71.9

56.2

65.9

72.0

71.3

114,556

968,365

1,082,920

24.9

76.5

62.7

79.2

28.1

43.8

52.2

56.7

54.2

345,015

297,872

642,886

75.1

23.5

37.3

414,975 1,049,847 1,464,822

84.9

Household uses land it does not own 92.8

80.6

84.4

53.6

69.6

64.2

90.3

82.9

Rented

2.8

1.9

2.2

68.6

67.2

67.8

16,037

23,873

39,910

3.5

1.9

2.3

Sharecropped

0.2

0.7

0.5

52.4

85.3

82.1

703

10,523

11,226

0.2

0.8

0.7

Private land provided free

3.1

7.9

6.4

86.9

67.4

70.3

22,640

99,245

121,885

4.9

7.8

7.1

Open access land

1.1

8.9

6.5

54.4

49.5

49.7

5,215

82,748

87,963

1.1

6.5

5.1

(Table continues on next page)

A World Bank Study

Table 2.4 (continued) Share of the population Urban

Rural

National

Poverty headcount Urban

Rural

National

Number of poor Urban

Contribution to poverty

Rural

National

Urban

Rural

National

Head has a secondary occupation Not working

30.6

15.0

19.8

67.2

71.1

69.3

171,811

199,191

371,002

37.4

15.7

21.5

65.6

74.6

71.8

49.5

69.1

63.6

270,657

964,131 1,234,788

58.9

76.1

71.5

Yes

3.8

10.4

8.4

53.6

52.8

52.9

17,102

3.7

8.1

7.0

102,914

120,016

Spouse has a secondary occupation Not working

24.8

13.3

16.9

63.8

76.7

70.9

131,926

191,094

323,020

28.7

15.1

18.7

34.0

53.6

47.6

51.3

67.5

63.9

145,583

676,696

822,278

31.7

53.4

47.6

Yes No spouse

2.1

7.4

5.7

42.6

69.8

66.7

7,501

96,042

103,543

1.6

7.6

6.0

39.1

25.7

29.8

53.5

62.9

59.1

174,561

302,404

476,965

38.0

23.9

27.6

11.1

11.4

47.2

51.8

50.3

47,414

107,812

155,226

10.3

8.5

9.0

Age of the household head Less than 30

12.0

30 thru 39

29.1

26.2

27.1

50.4

63.4

59.1

122,275

311,281

433,557

26.6

24.6

25.1

40 thru 49

31.0

29.9

30.2

58.3

70.3

66.5

151,059

393,238

544,297

32.9

31.1

31.5

50 thru 59

17.7

17.6

55.8

76.0

69.8

82,269

249,712

331,980

17.9

19.7

19.2

60 and Over

10.2

15.2

13.6

66.6

71.9

70.7

56,554

204,193

260,747

12.3

16.1

15.1

Household size 1 individual

0.8

0.3

0.5

13.9

875

834

1,708

0.2

0.1

2 to 3 individuals

9.7

6.5

7.5

31.4

34.3

33.2

25,567

41,524

67,090

5.6

3.3

3.9

4 to 5 individuals

29.9

33.6

32.4

46.0

57.5

54.2

114,957

360,758

475,715

25.0

28.5

27.6

6 to 7 individuals

28.1

35.3

33.0

65.3

75.8

73.0

153,014

499,724

652,738

33.3

39.5

37.8

8 individuals and more

31.5

24.4

26.6

62.8

79.7

73.5

165,158

363,397

528,555

35.9

28.7

30.6

Number of workers in household None

13.3

8.0

9.6

73.3

72.8

73.0

81,582

108,475

190,057

17.8

8.6

11.0

One

31.2

10.0

16.6

53.4

58.5

55.5

138,883

109,802

248,685

30.2

8.7

14.4

Two

25.4

19.7

21.5

50.3

58.6

55.6

106,545

216,635

323,180

23.2

17.1

18.7

Three and more

30.1

62.3

52.3

52.7

71.4

68.1

132,561

831,323

963,884

28.8

65.7

55.9

Source: Authorsâ&#x20AC;&#x2122; calculations using CWIQ 2007, LISGIS.

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 2.5: Extreme poverty profile based on consumption per equivalent adult, 2007 Poverty headcount

Number of poor

Contribution to poverty

Urban

Rural

National

Urban

Rural

National

Urban

Rural

National

29.0

56.3

47.9

242,055

1,053,240

1,295,295

100.0

National Urban/rural location Urban

29.0

—

29.0

242,055

—

242,055

100.0

—

18.7

Rural

—

56.3

—

1,053,240

—

100.0

81.3

Region Greater Monrovia

22.7

—

22.7

135,338

—

135,338

55.9

—

10.4

North Central

34.4

59.4

57.6

23,243

535,059

558,302

9.6

50.8

43.1

North Western

54.3

63.3

62.2

16,999

151,402

168,401

7.0

14.4

13.0

South Central

46.2

41.4

42.2

34,303

153,843

188,146

14.2

14.6

14.5

South Eastern A

49.6

63.7

60.9

23,006

121,450

144,457

9.5

11.5

11.2

South Eastern B

47.4

54.4

53.7

9,165

91,487

100,652

3.8

8.7

7.8

Age of the individual Less than 10

31.5

53.4

47.6

65,617

308,653

374,270

27.1

29.3

28.9

10 thru 19

29.9

61.7

50.8

66,042

261,681

327,722

27.3

24.8

25.3

20 thru 29

26.7

54.0

44.7

40,390

157,822

198,212

16.7

15.0

15.3

30 thru 39

24.2

53.0

43.2

27,749

117,749

145,499

11.5

11.2

40 thru 49

27.2

57.9

48.5

21,169

101,867

123,037

8.7

9.7

9.5

50 thru 59

33.4

57.6

50.8

11,737

51,834

63,571

4.8

4.9

60 and Over

35.7

60.8

55.0

9,349

53,634

62,984

3.9

5.1

4.9

Gender of the head Male

28.4

57.3

48.9

166,095

817,036

983,131

68.6

77.6

75.9

Female

30.4

53.1

44.9

75,960

236,204

312,164

31.4

22.4

24.1

44.2

33.4

54,958

109,731

164,689

22.7

10.4

12.7 66.8

Marital status of the head Single or never married

22.4

Monogamous

31.0

57.5

50.2

145,415

720,161

865,576

60.1

68.4

Polygamous

38.1

67.3

63.8

7,700

100,556

108,256

3.2

9.5

8.4

Widowed, divorced, separated

34.3

55.7

49.1

33,982

122,792

156,774

14.0

11.7

12.1

51.9

Education level of head None

44.0

62.0

58.8

90,745

580,945

671,691

37.5

55.2

Some primary

31.8

51.7

48.5

10,451

90,347

100,798

4.3

8.6

7.8

Completed primary

50.5

49.7

49.9

12,990

40,088

53,078

5.4

3.8

4.1

Some secondary

27.9

54.1

46.7

44,503

220,304

264,807

18.4

20.9

20.4

Completed secondary

21.3

47.5

32.1

57,456

89,866

147,322

23.7

8.5

11.4

Post secondary

18.3

38.8

25.8

25,910

31,690

57,599

10.7

3.0

4.4

(Table continues on next page)

A World Bank Study

Table 2.5 (continued) Poverty headcount Urban

Rural

Number of poor

Contribution to poverty

National

Urban

Rural

National

Urban

Rural

National 57.5

Education level of spouse None

45.2

61.9

59.0

98,638

646,343

744,982

40.8

61.4

Some primary

16.3

52.2

44.3

6,984

79,035

86,020

2.9

7.5

6.6

Completed primary

20.7

26.6

25.0

3,500

11,757

15,256

1.4

1.1

1.2

Some secondary

29.5

56.9

43.6

28,203

57,880

86,083

11.7

5.5

6.6

Completed secondary

16.3

34.9

22.2

13,558

13,517

27,075

5.6

1.3

2.1

Post secondary No spouse

5.2

13.7

6.5

2,680

1,302

3,982

1.1

0.1

0.3

27.1

50.6

41.1

88,492

243,406

331,898

36.6

23.1

25.6

Socioeconomic group of head Public

20.7

41.4

30.2

41,897

71522

113,420

17.3

6.8

8.8

Private formal

18.8

52.5

41.5

8,845

50839

59,685

3.7

4.8

4.6

Private informal

23.1

38.4

31.7

12,501

27050

39,551

5.2

2.6

3.1

Self-agriculture

51.3

60.9

60.6

13,794

532851

546,646

5.7

50.6

42.2 16.9

Self-other

28.1

50.7

41.0

64,207

155203

219,411

26.5

14.7

Unemployed

32.5

55.0

39.6

32,683

25693

58,376

13.5

2.4

4.5

Inactive, other

39.1

62.6

54.1

68,125

190078

258,203

28.1

18.0

19.9 47.2

Industry of head Crop farming

53.2

60.2

60.0

15,607

596,096

611,703

6.4

56.6

Forestry/logging

23.0

65.5

43.6

887

2,361

3,247

0.4

0.2

0.3

Fishing

66.8

67.3

66.9

3,901

1,767

5,667

1.6

0.2

0.4

Mining/quarrying

13.4

55.5

46.0

436

6,172

6,608

0.2

0.6

0.5

Manufacturing/ processing

70.0

60.1

64.8

3,013

2,837

5,849

1.2

0.3

0.5

Electricity/gas/water supply

12.1

14.6

12.3

1,647

215

1,862

0.7

0.0

0.1

Construction

26.4

39.7

31.1

6,779

5,557

12,336

2.8

0.5

1.0

Wholesale/retail trades

24.9

29.9

27.0

21,553

18,096

39,650

8.9

1.7

3.1

Transport, storage, communications

17.5

38.6

21.3

4,154

2,014

6,168

1.7

0.2

0.5

Banking/financial services

23.2

23.7

23.3

1,929

817

2,746

0.8

0.1

0.2

Community services

18.9

43.8

32.6

21,615

60,574

82,190

8.9

5.8

6.3

Other

26.1

50.6

40.2

67,806

179,270

247,076

28.0

17.0

19.1

Unemployed, Inactive

36.3

63.4

50.4

92,728

177,465

270,193

38.3

16.8

20.9

Household owns cultivatable land Yes

44.3

60.8

58.9

77,075

818,429

895,504

31.8

77.7

69.1

25.0

44.7

33.7

164,980

234,811

399,791

68.2

22.3

30.9 85.9

Household uses land it does not own No

28.3

59.2

48.7

219,043

893,587

1,112,630

90.5

84.8

Rented

35.0

31.9

33.1

8,175

11,329

19,503

3.4

1.1

1.5

Sharecropped

52.4

64.9

63.7

703

8,009

8,712

0.3

0.8

0.7

Private land provided free

47.7

53.7

52.8

12,419

79,126

91,545

5.1

7.5

7.1

Open access land

17.9

36.6

35.6

1,716

61,189

62,905

0.7

5.8

4.9

(Table continues on next page)

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 2.5 (continued) Poverty headcount Urban

Rural

Number of poor

National

Urban

Rural

Contribution to poverty National

Urban

Rural

National

Head has a secondary occupation Not working

36.3

63.4

50.4

92,728

177,465

270,193

38.3

16.8

20.9

26.1

57.1

48.4

142,677

797,389

940,066

58.9

75.7

72.6

Yes

20.9

40.2

37.5

6,651

78,386

85,036

2.7

7.4

6.6

Spouse has a secondary occupation Not working

36.3

64.6

51.8

75,100

160,941

236,041

31.0

15.3

18.2

27.1

56.5

50.0

77,059

566,737

643,797

31.8

53.8

49.7

Yes No spouse

8.0

59.7

53.9

1,404

82,156

83,560

0.6

7.8

6.5

27.1

50.6

41.1

88,492

243,406

331,898

36.6

23.1

25.6

Age of the household head Less than 30

24.4

41.5

35.9

24,534

86,226

110,760

10.1

8.2

8.6

30 thru 39

24.4

50.7

42.0

59,242

249,097

308,339

24.5

23.7

23.8

40 thru 49

30.6

58.5

49.6

79,222

327,005

406,228

32.7

31.0

31.4

50 thru 59

31.5

63.9

53.8

46,440

209,689

256,130

19.2

19.9

19.8

60 and Over

38.4

63.8

57.9

32,616

181,223

213,839

13.5

17.2

16.5

Household size 1 individual

11.6

13.9

12.7

730

834

1,564

0.3

0.1

2 to 3 individuals

12.9

26.0

20.7

10,454

31,434

41,888

4.3

3.0

3.2

4 to 5 individuals

21.2

43.6

37.2

53,028

273,453

326,481

21.9

26.0

25.2

6 to 7 individuals

33.6

63.9

56.0

78,630

421,826

500,456

32.5

40.1

38.6

8 individuals and more

37.7

71.4

59.1

99,212

325,693

424,905

41.0

30.9

32.8

Number of workers in household None

44.3

66.0

56.7

49,297

98,397

147,695

20.4

9.3

11.4

One

27.1

47.7

35.7

70,504

89,444

159,948

29.1

8.5

12.3

Two

23.3

46.1

37.7

49,319

170,109

219,428

20.4

16.2

16.9

Three and more

29.0

59.7

54.3

72,934

695,289

768,224

30.1

66.0

59.3

Source: Authorsâ&#x20AC;&#x2122; calculations using CWIQ 2007, LISGIS.

As shown in figure 2.1, which provides curves representing the share of the population in poverty in urban and rural areas as well as by regions as a function of the poverty line on the horizontal axis, the headcount is higher in rural than in urban areas for all poverty lines, but there are a few reversals in headcount rankings between different regions depending on the choice of the poverty line. There are few differences in poverty measures according to the age of the individuals, while differences according to the gender or the head of household are also small. Poverty seems to be higher among polygamous households than among monogamous households, and individuals who are single or never married tend to have a lower probability of being poor. In terms of demographic variables, household heads who are younger, below 30 or 40 years of age, are less likely to be poor. The larger the household size, the higher the probability of being poor. A higher education for the household head or the spouse is associated with lower levels of poverty, as expected. In terms of the socioeconomic group of the household

A World Bank Study

Figure 2.1: Stochastic dominance by residence area, 2007 1

.2 0

0 0

.5 1 1.5 Normalized expenditure per capita (yi/z)

Urban Rural National

.5 1 1.5 Normalized expenditure per capita (yi/z) Greater Monrovia North Western South Eastern A

North Central South Central South Eastern B

Source: Authorsâ&#x20AC;&#x2122; calculations using CWIQ 2007, LISGIS.

head, households with a head in the public sector or with a wage in the private formal sector have lower rates of poverty. The highest levels of poverty are observed for those household heads who are self-employed in agriculture, followed by inactive heads (who are not working). Poverty rates by industry are lowest in the banking/financial sector, followed by utilities. Poverty is higher for those involved in fishing, crop farming, and mining/quarrying, as well as for those who are unemployed or inactive. Household heads who have a second occupation tend to have lower probabilities of being poor. Poverty also goes down when there is one or two workers in the household, as opposed to none or more than two (in the later case, because this denotes large household who need to have many members working). Cultivation of land is also associated with farming, and thereby poverty. Many of the results obtained with the characteristics of the household head are also similar when using the characteristics of the spouse of the head when there is one. Similarly, the results obtained for the extreme poverty measures display a similar pattern as in the overall poverty in terms of comparisons between various sub-groups groups, although there are more differences in extreme poverty between urban and rural areas than for overall poverty. 3.2. Poverty Comparisons with Other Countries

One way to discuss the level of poverty obtained from the cost of basic needs method is to compare Liberia to other West and Central African countries (Wodon, 2007). This is done in table 2.6 and figure 2.2. There are 17 countries listed in the table, and most belong to the CFA franc zone. For all these countries, the World Bank has recently completed poverty assessments that include poverty measures. These poverty measures are not strictly speaking comparable between countries due to differences in methodologies used for measuring poverty. But at the same time, they can be used to set expectations as to the order of magnitude of poverty estimates that one might expect in any of the 17 countries. Most countries use a poverty line based on the cost of basic needs method, although countries differ in whether they use consumption per capita or per equivalent

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 2.6: Comparison of Liberia with FCFA West and Central African countries Date

GDP, US$

GDP US$

Methodology

Benin

2003

Burundi

2006

Poverty

325

1.18

Relative

39.0

110

0.10

CBN

68.7

Burkina Faso

2003

247

0.90

CBN

46.4

Cameroun

2001

695

1.94

CBN

40.2

Congo

2005

994

2.30

CBN

50.7

Côte d’Ivoire

2002

592

1.78

Relative

38.4

Congo, Dem. Rep. of

2005

120

0.18

CBN

71.3

Gabon

2005

3991

3.69

CBN

33.2

Guinea-Bissau

2002

138

0.33

$1 per day

65.7

Liberia

2007

135

0.30

CBN

63.8

Mali

2001

226

0.82

CBN

55.6

Niger

2005

158

0.45

CBN

62.1

Central African Republic

2003

225

0.81

CBN

67.2

Senegal

2001

442

1.49

CBN

57.1

Sierra Leone

2003

190

0.64

CBN

65.89

Chad

2003

211

0.75

CBN

55.0

Togo

2006

238

0.87

CBN

61.7

Source: Adapted from Wodon (2007). Note: CBN = Cost of Basic Needs.

Figure 2.2: Poverty and per capita GDP (logarithmic scale) 80 75

DRC

Share of population in poverty (%)

70 65 60 55

Niger

Liberia

Tchad

RCA Togo Mali

Senegal "Expected" level of poverty

Rep. Congo

Burkina Faso

45 Benin

Cameroon Côte d'Ivoire

35 30 0.00

Gabon

0.50

1.00

Source: Adapted from Wodon (2007).

1.50 2.00 2.50 Ln (per capita GDP/100), US$

3.00

3.50

4.00

A World Bank Study

adult and the level of the caloric requirement norm used to determine what basic amount of food a person should consume. In two countries, a relative poverty line was chosen to measure poverty—this was done in Benin and Côte d’Ivoire (where the relative poverty line originally adopted to estimate poverty was subsequently regularly adjusted for inflation). In one country (Guinea-Bissau), the poverty line was set by the authorities to match the international benchmark of US$1 per day per person used for monitoring the Millennium Development Goals. Apart from differences in the methodologies used to define the poverty lines, the poverty measures are based on surveys which also differ somewhat between countries, with some surveys tracking the consumption levels of households better than others. Despite differences between countries in methodologies for estimating poverty, an inverse relationship clearly exists between the (natural) logarithm of GDP per capita and the share of the population living in poverty, as shown in figure 2.2. In the figure, GDP per capita has been expressed in constant U.S. dollars for simplicity. The curve was fitted through the scatterplot in order to maximize the explanatory power of a univariate regression using a logarithmic specification. Therefore, the curve gives a very rough idea of the poverty level “expected” for a given level of GDP per capita.4 Quite a few countries appear to have levels of poverty in line with what is expected according to the very simple and rough method used to set expectations, and this is also the case for Liberia. For example, the poorest countries in terms of per capita GDP (Guinea-Bissau and Niger) have very high levels of poverty while at the other extreme, richer countries such as Côte d’Ivoire, Cameroon, and Gabon have lower levels of poverty. But there are also a few countries that seem to have levels of poverty that diverge from what one might have expected. Divergence from the fitted curve may stem not only from issues of data quality or different assumptions used for measuring poverty, but also from different levels of inequality between countries (typically, a more unequal distribution of consumption will be associated with a higher level of poverty). Divergence from the fitted curve will also depend on how the curve is fitted, with alternative ways of fitting the curve leading to different levels of divergence for each country. Still, for most countries that are located “far” from the curve, there are simple data or methodological reasons that help explain why the countries are located far from the curve (see Wodon, 1997 for a discussion). 3.3. Subjective Indicators of Poverty and Vulnerability

It is also interesting to compare objective measures of poverty with subjective perceptions of poverty as well as indicators of vulnerability (tables 2.7 and 2.8). Several observations from the data suggest that the level of poverty measured for Liberia, at 63.8 percent, is realistic. First, the national share of the population living in households where the household head stated that the current income of the household made the household live with difficulty was 57.7 percent and is of the same order of magnitude as the objective poverty estimate. Second, the level of income or consumption deemed by households to be needed in order to be able to satisfy one’s needs, at L$2,049 per month per person according to subjective perceptions, is also of the order of magnitude of the poverty lines per equivalent adult estimated with the cost of basic needs and reported on an annual basis in table 2.3. Third, the share of the population in a vulnerable situation, because their income is very unstable, at 60.6 percent, is also of a similar order of magnitude. By contrast, the shares of the population that needs to borrow money, at 43.6 percent, or that is having always or often difficulties to satisfy basic needs for food,

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 2.7: Subjective perceptions of poverty and ability to meet basic needs, 2007 Inability of households to satisfy needs (always or often)

Total Rural Urban Region Greater Monrovia North Central North Western South Central South Eastern A South Eastern B National quintile 1 2 3 4 5

Perceptions on livelihoods based on current income Living reasonably Living Living with well carefully difficulty 10.0 31.2 57.7 8.1 28.4 62.7 14.1 37.1 47.0

Food 29.9 32.0 25.3

School fees 28.0 28.3 27.5

Health care 27.0 30.8 19.0

Living very well 1.1 0.8 1.8

Subj. poverty line 2,049.3 1,795.3 2,451.5

25.4 38.6 23.3 25.1 25.8 25.3

29.5 38.8 9.1 22.7 19.0 18.8

22.9 33.3 24.3 21.4 29.0 23.1

1.9 0.6 0.2 1.7 1.4 1.0

11.6 8.6 17.6 6.7 7.7 11.0

36.8 26.9 38.1 31.4 29.6 25.9

49.8 63.9 44.1 60.2 61.3 62.2

2,743.8 1,680.3 1,724.0 2,161.2 1,827.5 1,859.1

40.1 32.9 32.8 28.0 20.5

31.5 29.0 28.9 30.6 22.4

30.3 26.2 29.7 27.5 23.1

0.6 0.8 0.6 0.7 2.3

7.1 8.9 8.1 10.5 13.7

18.4 25.7 32.0 34.3 40.4

73.9 64.6 59.3 54.4 43.6

1,500.5 1,625.3 1,679.8 2,122.6 2,995.1

Source: Authorsâ&#x20AC;&#x2122; calculations using CWIQ 2007, LISGIS.

Table 2.8: Subjective indicators on vulnerability to shocks of households, 2007

Total Rural Urban Region Greater Monrovia North Central North Western South Central South Eastern A South Eastern B National quintile 1 2 3 4 5

Save a lot of money 0.1 0.1 0.1 0.2 0.1 0.1 0.4 0.1 0.0 0.1 0.2

Financial situation of households Save Satisfy Need a little basic to use money needs savings 10.5 41.5 4.2 9.1 40.8 3.2 13.5 43.2 6.6

Stability of household income Need to borrow money 43.6 46.9 36.5

Very unstable 60.6 66.7 47.5

Somewhat unstable 36.1 32.0 44.9

Stable 3.3 1.3 7.5

13.1 12.8 6.9 8.1 6.2 6.4

42.5 42.1 39.3 42.1 42.6 36.2

6.1 3.3 4.2 4.1 3.2 5.0

38.1 41.8 49.6 45.7 47.8 52.0

49.0 72.4 50.6 60.0 59.7 54.6

42.7 26.8 48.2 36.5 37.5 42.2

8.3 0.8 1.1 3.6 2.8 3.2

6.5 8.8 10.2 9.8 15.2

39.2 34.9 39.1 43.9 47.5

2.3 3.0 5.0 4.4 5.8

51.9 53.3 45.7 41.8 31.3

72.4 70.5 65.0 54.4 47.7

26.7 27.6 33.5 43.5 44.3

0.9 1.9 1.5 2.2 8.0

Source: Authorsâ&#x20AC;&#x2122; calculations using CWIQ 2007, LISGIS.

A World Bank Study

schooling, or health expenditures, at slightly less than 30 percent, are lower, and closer in magnitude to the estimated measures of extreme poverty. 3.4. Simulations for Future Poverty Reduction

Liberia’s economy has made a strong recovery since 2005 due to higher agriculture production and the return of displaced persons. According to World Bank (2007a), real GDP growth reached 5.3 percent in 2005 and 7.8 percent in 2006, with limited inflation. The macroeconomic framework used in the country suggests that growth could reach 9 percent or even higher in future years (table 2.9). This would translate in a rate of growth of GDP per capita above 6 percent, with limited inflation. Table 2.9: Liberia—selected economic and financial indicators, 2003–10 2004

2005

2006 Est.

2007 Proj.

2008 Proj.

2009 Proj.

2010 Proj.

Real GDP (% growth)

2.6

5.3

7.8

9.4

9.5

11.9

14.0

Consumer prices (annual average % growth)

3.6

6.9

7.2

11.4

9.0

8.0

7.0

Indicator

Source: World Bank (2007a).

Figure 2.3 provides estimates of likely future poverty assuming various growth rates in GDP per capita over the medium term, up to 2015. A number of rather strong assumptions are needed to generate these estimates. First, it is assumed that growth in per capita GDP leads to equivalent growth in average consumption per equivalent adult.

Figure 2.3: Growth and poverty simulations, 2007 80 70 60

Percent

50 40 30 20 10 0 2007

1% annual growth of the GDP per capita 2% annual growth of the GDP per capita 3% annual growth of the GDP per capita 4% annual growth of the GDP per capita 5% annual growth of the GDP per capita 6% annual growth of the GDP per capita

2008

2009

2010

2011 Year

Source: Authors’ calculations using CWIQ 2007, LISGIS.

2012

2013

2014

2015

Poverty and the Policy Response to the Economic Crisis in Liberia

Second, it is assumed that inequality remains unchanged over time. Still, the simulations give an idea of the type of poverty reduction that could be seen in the future with a resumption of higher growth, which is useful for setting targets in the PRSP. For example, with a growth rate in GDP per capita of 6 percent per year, the share of the population in poverty could be slightly above 30 percent in 2015, which would be a remarkable improvement. It may be however that growth will initially favor better off areas, and take some time to fully trickle down to poor areas in the country, in which case the amount of poverty reduction that could be expected by 2015 would be smaller, because inequality could likely increase as the country recovers and some sectors expand more than others. 3.5. Correlates or Determinants of Poverty

Drawing a profile of poverty is a necessary step to identify the characteristics of the population groups that are poor, but it is not sufficient to measure the impact of various household characteristics on poverty. The problem with a poverty profile lies in the fact that it provides information on who are the poor, or on the probability of being poor among various household characteristics, but cannot be used to assess the correlates of poverty. For instance, the variation of poverty rates across regions is sometimes better accounted for by the differences in householdsâ&#x20AC;&#x2122; characteristics than by the specificities of each region. To sort out the correlates or determinants of poverty and the impact of various variables on the probability of being poor, regression analysis is thus required. Table 2.10 provides an analysis of the correlates or determinants of poverty or well-being using standard regression techniques to explain (a) the logarithm of the consumption per equivalent adult of the household which is the variable determining whether a household is poor or not; (b) whether a household is poor or not; and (c) whether a household feels poor or not. The regressions are run separately for Monrovia, other urban areas, and rural areas, with the results mostly as expected in terms of the marginal impacts of various variables on welfare. Apart from a constant, the regressors include: (a) geographic location variables according to key regions; (b) household size variables (number of infants, children, adults and seniors, and their squared value to take into account potential non-linearity in relationships between household size and consumption), whether the household head is a woman, the age of the head, and the marital status of the head; (c) characteristics of the household head, including level of education; socioeconomic group, and whether the head has a second job; (d) the education level of the spouse of the household head when there is one; and (e) other variables including information on land cultivated, migration related to the war, and access to infrastructures. Key findings are as follows:

â&#x2013;

Demographic characteristics: As expected, an additional person in the household tends to reduce consumption per equivalent adult with the impact ranging from no loss to a loss of 25 percent of consumption, depending on the case. Yet the impact on the probability of being poor is less statistically significant in urban areas (except for the number of male adults), and the impact is not present for subjective poverty, as has been observed in other countries. Also as observed in a number of other countries, there are few statistically significant

A World Bank Study

■

differences between male-headed and female-headed households. In terms of marital structure, most of the coefficients are not statistically significant as well, so that no generalizations can be drawn. Finally, the age of the head as well does not seem to make a major difference in consumption levels. Thus, in terms of demographics, the main finding is that households that are larger have a lower consumption per equivalent adult even after controlling for the differences in needs between different persons through the use of the adult equivalence scale. Education level of the head and spouse: As expected, consumption levels increase and the probability of being poor decreases with the education level of the household head, but the effects are statistically significant only as of secondary schooling. The impact of the spouse’s education is in most cases of an order of magnitude similar to that of the head. Still, overall the impacts are not very large, which suggests that opportunities are limited through good employment to benefit from the full returns that an education can provide. Employment of the head: After controlling for other variables, the type of employment does not seem to affect very much the level of consumption of households or their probability of being poor. This is surprising to the extent that in many other countries, when the household head belongs to the public sector or the private formal sector, the household is typically better-off than when the head is self-employed, especially in agriculture. By contrast, if the head is unemployed or inactive, the negative impact on consumption and poverty is rather large in most instances (more so on consumption than on the probability of being poor), and indeed larger than what has been observed in other West and Central African countries. This type of finding may be used for example to advocate policies (as is actually done in Liberia’s Interim Poverty Reduction Strategy) that enable the poor to find employment, for example through public works which are very much needed in Liberia to rebuild the infrastructure destroyed during the civil war. The regression results also suggest that when the head has a second job, consumption is higher, and the probability of being poor lower, at least in rural areas. Other variables: After controlling for other variables, if the household has a larger land size available for cultivation, consumption is higher, and the probability of being poor lower, as expected. Displaced households who have returned to their place of origin actually seem to be better off, after controlling for other variables, than non-displaced persons, perhaps because those that were displaced had higher means to enable them to leave their place of origin. Isolated households, as measured through the time it takes to reach the closest food marker, tend to have lower consumption levels and higher probabilities of being poor. Finally, there is some evidence that households in the South Central A region and to some extent in the North Western region are poorer.

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 2.10: Correlates or determinants of poverty, 2007 Objective poverty (moderate) MCO: ln(yi/z) Monrovia

Other urban

Objective poverty (extreme) Probit (is poor)

Probit (is poor) Rural

Monrovia

Other urban

Rural

Monrovia

Other urban

Subjective poverty Probit (feels poor)

Rural

Monrovia

Other urban

Rural

—

Region Greater Monrovia

—

North Central

—

0.225**

0.067

—

−0.179**

0.014

—

−0.198***

0.046

—

North Western

—

0.033

0.059

—

0.055

0.053

—

0.024

0.052

South Central

—

0.329***

0.272***

—

−0.035

—0.143***

—

South Eastern A

—

Ref.

—

Ref.

—

South Eastern B

—

0.230***

0.208***

—

−0.023

—0.108***

—

−0.133* −0.139*** Ref.

Ref.

−0.171*** −0.082**

−0.197*** 0.138***

—

0.078

0.038

—

0.168**

0.008

—

Ref.

—

−0.130**

0.145***

Household composition Children aged 0 to 5

−0.072

−0.073

−0.059*

0.004

0.011

0.069**

−0.065*

−0.059

0.055*

0.041

−0.038

−0.010

Children aged 0 to 5, squared

−0.011

−0.010

0.003

0.033

0.042

−0.008

0.046***

0.055**

−0.006

0.008

0.017

0.003

Children aged 6 to 14

−0.134***

−0.052

−0.159***

0.071*

0.054

0.139***

0.008

−0.015

0.130***

−0.040

−0.002

0.009

Children aged 6 to 14, squared

0.015**

−0.010

0.015***

−0.005

0.002

−0.014***

0.003

0.024**

−0.011**

0.003

−0.002

−0.005

0.123**

0.165***

0.053

0.073

0.167***

−0.040

−0.019

−0.011

Male adults aged −0.242*** −0.195*** −0.151*** 0.141*** 15 to 60 Male adults aged 0.027*** 15 to 60, squared

0.022

0.003

−0.010

0.007

−0.013***

−0.001

0.005

−0.011**

0.000

−0.012

0.003

Female adults aged 15 to 59

−0.068

−0.107

−0.139***

0.053

0.070

0.098***

−0.020

0.114

0.072**

0.006

−0.023

0.012

Female adults aged 15 to 59, squared

−0.009

0.011

0.015

0.001

−0.013

−0.008

0.011**

−0.015

−0.004

−0.002

0.002

−0.003

Seniors aged over 60

−0.196

−0.263

−0.196***

−0.001

0.049

0.215***

0.103

0.251

0.218***

0.026

0.028

0.011

Seniors aged over 60, squared

0.089

0.118

0.057

0.004

0.094

−0.060**

−0.042

−0.089

−0.057*

−0.027

0.074

0.022

Age of the household head

−0.016

−0.004

0.003

0.014***

0.011

−0.006

0.013**

0.006

0.008

0.001

Age of the household head, squared

0.000

0.000**

0.000

−0.000***

0.000

−0.000**

0.000

Female household head

0.062

0.097

0.058

−0.040

−0.013

0.001

−0.032

−0.133**

0.032

−0.042

−0.096

0.003

Head has No Spouse

0.050

0.073

−0.064

−0.055

−0.294***

0.000

−0.037

−0.019

0.017

−0.016

−0.194*

−0.035

0.021

0.142**

−0.001

0.198**

−0.097*

0.018

−0.061

−0.128**

−0.035

−0.025

0.011

−0.015*** 0.020**

Marital status of the head Single or never married

0.023

Monogamous

Ref.

Polygamous

0.264***

0.114

0.024

−0.231**

0.012

0.005

−0.038

−0.111

0.007

0.096

0.072

−0.052

Widowed or divorced or separated

−0.021

−0.035

0.068

0.043

0.222**

0.045

0.038

0.007

−0.032

0.019

0.099

0.117**

(Table continues on next page)

A World Bank Study

Table 2.10 (continued) Objective poverty (moderate) MCO: ln(yi/z) Monrovia

Other urban

Objective poverty (extreme) Probit (is poor)

Probit (is poor) Rural

Monrovia

Other urban

Rural

Monrovia

Other urban

Rural

Subjective poverty Probit (feels poor) Monrovia

Other urban

Rural

Education level of head None

Ref.

Some primary

0.167

0.170

0.112***

−0.185**

0.066

−0.040

−0.075

−0.159

−0.027

0.007

0.135

0.015

Completed primary

0.012

0.124

0.070

−0.008

0.017

0.006

0.020

−0.022

−0.063

0.021

0.171

−0.109**

Some secondary

0.169**

0.172**

0.074**

−0.151**

−0.051

−0.046

−0.022

−0.011

−0.058

−0.093***

Completed secondary

0.345***

0.202**

0.213*** −0.250***

−0.083

−0.088** −0.137*** −0.141** −0.102** −0.201***

−0.066

−0.223***

Post secondary

0.524***

0.424***

0.321*** −0.341***

−0.038

−0.211*** −0.138*** −0.273*** −0.198*** −0.291*** −0.170* −0.188***

−0.091** −0.131**

Education level of spouse None

Ref.

Some primary

0.137

0.239**

0.091**

−0.134

−0.178

−0.100**

−0.106*

−0.078

−0.093**

−0.020

−0.126

−0.089**

Completed primary

−0.139

0.087

0.146**

−0.152

0.100

−0.176**

−0.018

−0.021

−0.175**

−0.227

0.042

−0.066

Some secondary

0.045

0.067

0.071

−0.031

−0.136

−0.068

−0.050

−0.091

−0.071

−0.061

−0.106

−0.057

Completed secondary

0.238***

0.175**

0.114

−0.171**

−0.087

−0.164* −0.150***

−0.076

−0.136

−0.134* −0.353***

−0.080

Post secondary

0.481***

0.503***

−0.335** −0.295*** −0.386***

−0.153

0.427*** −0.378*** −0.582*** −0.474*** −0.158***

Socioeconomic group of head of household Public

−0.134

−0.042

−0.046

−0.216

−0.060

−0.002

0.087

0.016

−0.006

0.154

−0.102

−0.024

Private formal

−0.078

−0.126

0.069

−0.259*

0.071

−0.042

0.074

0.146

−0.036

0.060

0.017

−0.092*

Private informal

−0.146

0.059

−0.034

−0.175

−0.014

0.006

0.131

−0.023

0.012

0.337**

0.238*

−0.071

Self-agriculture

Ref.

Self-other

−0.115

−0.081

0.033

−0.202

0.012

−0.058*

0.095

−0.016

−0.037

0.322**

−0.046

−0.041

Unemployed

−0.375*

−0.214*

−0.172**

0.011

0.033

0.062

0.261*

0.172

0.131**

0.242

0.124

−0.049

Inactive, other

−0.296

−0.252*** −0.319***

−0.031

0.090

0.122***

0.211

0.199**

0.149***

0.146

0.063

−0.074**

The head has a second job

−0.018

0.166*

0.100**

0.031

−0.002

−0.113**

0.009

−0.091

−0.138***

0.021

−0.125

−0.045

Total Acres of cultivable land owned

0.008**

0.018***

0.003**

0.001

−0.001

−0.018**

0.000

−0.010*

−0.006*

−0.001

−0.110

0.002

0.265***

0.103

0.065

−0.043

−0.083

0.047*

Ref.

−0.011* −0.029***

Migration status due to the war Displaced

−0.024

0.085

0.031

0.048

0.114

0.052

−0.083*

Displaced and has returned to origin

0.018

0.180***

0.110***

0.015

−0.050

−0.072**

−0.018

Never move

Ref.

4.029***

0.156***

1.278**

2.066*

0.191***

1.977**

−0.157*** −0.065**

Accessibility of infrastructures Time to food market (in 1,000 minutes)

−2.315*** −3.865*** −0.131*** 1.700**

Constant

0.966***

−0.044

Observations

816.000

575.000 2204.000 816.000

Adjusted R-squared

0.340

0.260

0.484*** 0.220

575.000 2204.000 816.000

Source: Authors’ calculations using CWIQ 2007, LISGIS. Note: * significant at 10%; ** significant at 5%; *** significant at 1%.

557.000 2204.000 816.000

−2.437** 0.209***

575.000 2204.000

Poverty and the Policy Response to the Economic Crisis in Liberia

4. Conclusion This chapter has relied on data from the 2007 CWIQ survey for Liberia in order to estimate the level of poverty and vulnerability in the country and analyze household level determinants of consumption and poverty. Slightly less than two third of the population (63.8 percent) is estimated to be poor but it is likely that the situation of many other households, who are not considered poor because they have consumption levels above the poverty line, remains precarious. Therefore, poverty and vulnerability can be considered as massive. In recent years, the country has managed to grow at an impressive rate, and according to the macroeconomic framework to be used in the countryâ&#x20AC;&#x2122;s Poverty Reduction Strategy, high growth rates are expected to continue for some time. If this is indeed the case, poverty could be significantly reduced by 2015. As in other developing countries, consumption levels and the probability of being poor vary substantially between households according to their characteristics. Poverty is significantly higher in rural than in urban areas, and there are also important differences in poverty levels between regions. Households who have an educated head or spouse are much less likely to be poor, although it is necessary to go beyond primary education to start to see a significant impact on household consumption. The type of employment of the head does not seem to have a major impact on consumption and poverty, but on the other hand, households with an unemployed or inactive head tend to be poorer. Household size is also a major determinant of poverty, with larger households being poorer, even after adjusting consumption levels for differences in needs between household members through the use of adult equivalence scales.

A World Bank Study

Annex: Sensitivity of Poverty Estimates to Caloric Threshold The threshold of 2,400 kcal per person per day used to define the basket of basic food items that should be consumed by each equivalent adult in the household in order to meet minimum nutritional requirements can be considered as somewhat ad hoc. In some countries, lower caloric thresholds have been used (as low as 2,100 kcal), but in other countries such as Cameroon or Nigeria, higher caloric thresholds have been used (up to 2,900 kcal). On average, countries in West and Central Africa have tended to use slightly lower caloric thresholds than 2,400 kcal per equivalent adult, but there is no universally accepted norm for the choice of the threshold. Annex Table 2A.1 provided below shows how poverty measures would change if one were to adopt a different, slightly lower caloric threshold. If the threshold were to be set at 2,300 kcal, the headcount of poverty at the national level would be reduced to 60.9 percent. If the threshold were reduced further, to 2,100 kcal or 2,200 kcal, the share of the population in poverty would be reduced much more, to 53.7 percent and 52.6 percent, respectively. Annex Table 2A.1: Headcount index of poverty and sensitivity to the caloric threshold Caloric threshold

Poverty line Food

Non-food

Total

Poverty headcount

Urban

2,400

14,514.49

6,909.9

21,424.39

67.7

Rural

2,400

14,431.2

15,792.54

30,223.74

55.1

Urban

2,300

13,909.72

6,297.75

20,207.47

63.6

Rural

2,300

13,829.9

16,272.87

30,102.77

54.8

National

63.8

National

60.9

Urban

2,200

13,304.94

5,169.96

18,474.9

56.9

Rural

2,200

13,228.6

13,585.24

26,813.84

46.3

Urban

2,100

12,700.17

5,430.42

18,130.59

55.4

Rural

2,100

12,627.3

14,186.52

26,813.82

46.3

National

53.7

National

52.6

Source: Authorsâ&#x20AC;&#x2122; calculations using CWIQ 2007, LISGIS.

In Liberia, given that there are no other available and comparable surveys to which the CWIQ can be compared, one could set the threshold at various values, and obtain different levels of poverty. For a number of reasons presented in this chapter, the estimate of poverty of 63.8 percent at the national level is reasonable, even though the caloric threshold is slightly on the high side. It is also believed that consumption in the survey may have been slightly overestimated due to the methods used for gathering data in the survey, and this is another reason not to reduce the caloric threshold. While there is some liberty to change the caloric threshold when one cannot compare results from one survey to another, in future years, it will probably be important for consistent poverty measurement over time to collect similar household survey data and adopt the same norms as those used in this chapter for poverty monitoring and evaluation.

Poverty and the Policy Response to the Economic Crisis in Liberia

Notes 1. The authors are with the World Bank. This chapter was written as an input to Liberia’s Poverty Reduction Strategy. Key results were presented at a workshop organized by Liberia’s core PRSP team in Monrovia on December 10-11, 2007. The views expressed in this chapter are those of the authors and need not reflect those of the World Bank, its Executive Directors or the countries they represent. 2. For an analysis of the employment patterns of the population in Liberia, see Wingfield-Digby (2007). 3. Most households own the dwelling in which they live while a non-negligible share of households is housed free of charge. Both of these categories of households enjoy accommodations that are part of their consumption. It is therefore important to estimate the rent they would have paid if they were tenants. This imputed rent is only estimated for households that are not tenants, based on a regression analysis of the logarithm of the rent paid by households that are tenants. The explanatory variables used for the regression include: the area of residence (region), the type of accommodation, the materials used (walls, floor, roof), the number of rooms in the dwelling, the combustible used for cooking, the lighting source in the dwelling, the water supply source and the waste disposal method. 4. We use the term “very rough” because different techniques could be used to fit a curve between the points in the figure, with a different “expected” level of poverty given the level of GDP per capita resulting from each different way of fitting the curve. In addition, the “expected” level of poverty represented by the fitted curve depends on the normalization used on the horizontal axis of the graphs.

References Coudouel, A., J. Hentschel, and Q. Wodon. 2002. “Poverty Measurement and Analysis.” In J. Klugman, ed., A Sourcebook for Poverty Reduction Strategies: Volume 1: Core Techniques and Cross-Cutting Issues. Washington, DC: World Bank. Fearon, J. D., M. Humphreys, and J. M. Weinstein. 2009. “Can Development Aid Contribute to Social Cohesion after Civil War? Evidence from a Field Experiment in PostConflict Liberia.” American Economic Review: Papers & Proceedings 99(2): 287–291. Foster, J. E., J. Greer, and E. Thorbecke. 1984. “A Class of Decomposable Poverty Indices.” Econometrica 52: 761–766. International Labour Organization (ILO). 2009. “A Rapid Impact Assessment of the Global Economic Crisis on Liberia.” Mimeo, Monrovia. Kieh, G. K., Jr. 2004. “Irregular Warfare and Liberia’s First Civil War.” Journal of International and Area Studies 11(1): 57–77. Radelet, S. 2006. “Reviving Economic Growth in Liberia.” Working paper No. 132. Center for Global Development, Washington, DC. Republic of Liberia. 2006. Interim Poverty Reduction Strategy: Breaking from the Past—From Conflict to Development. Monrovia. Republic of Liberia. 2008. Poverty Reduction Strategy. Monrovia. Richards, P., S. Archibald, B. Bruce, W. Modad, E. Mulbah, T. Varpilah, and J. Vincent. 2005. “Community Cohesion in Liberia: A Post-War Rapid Social Assessment.” Social Development Paper, Conflict Prevention and Reconstruction No. 21. World Bank, Washington, DC. Sawyer, A. 2005. Beyond Plunder: Toward Democratic Governance in Liberia. Boulder, Colorado: Lynne Rienner.

A World Bank Study

Tsimpo, C., and Q. Wodon. 2012. “Rice Prices and Poverty in Liberia.” In Q. Wodon, ed., 85–98. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. UNDP Liberia. 2001. “Poverty Profile of Liberia.” Mimeo, Monrovia. ———. 2006. National Human Development Report 2006—Mobilizing Capacity for Reconstruction and Development. Monrovia. Wingfield-Digby, P. K. 2007. “Employment Statistics for Liberia: Report on Labour-Related Data from the Liberia CWIQ Survey.” Mimeo, Monrovia. Wodon, Q., 2007, “Using Simple Cross-Country Comparisons to Guide Measurement: Poverty in the CFA Franc Zone.” Findings 279. Africa Region. World Bank, Washington, DC. World Bank. 2007a. Program Document for a Proposed Grant in the Amount of SDR 263.1 million (US$416.4 Million Equivalent) to the Republic of Liberia for a Reengagement and Reform Support Program. Report No. 40307-LR. World Bank, Washington, DC. ———. 2007b. Spreading and Sustaining Growth in Africa: Africa Development Indicators 2007. Washington, DC: World Bank. ———. 2009. Liberia: Employment and Pro-Poor Growth. Report No. 51924-LR. Washington, DC: World Bank.

CHAPTER 3

Education in Liberia: Basic Diagnostic Using the 2007 CWIQ Survey Clarence Tsimpo and Quentin Wodon1 This chapter was written in 2007 in order to inform the diagnostic of Liberia’s Poverty Reduction Strategy. Little has been written on the education system in Liberia since the start of the conflict in large part because of lack of good data. The chapter provides a diagnostic of Liberia’s education system as seen from the point of view of households using the new nationally representative Core Welfare Questionnaire Indicator (CWIQ) survey implemented in 2007. The analysis covers school enrollment rates as well as the reasons for not going to school, and the degree of satisfaction of households with the services received, in each case looking at various age groups and boys and girls separately, as well as at different types of facilities providing education services. Data are also presented on household private spending for education, as well as on distances to facilities. A benefit incidence analysis of public spending for education is conducted, and regression analysis is used to assess the determinants of school enrollment.

1. Introduction As Liberia emerges from civil war, a renewed emphasis is being placed by the country’s government as well as by donors on improving the quality of education and health services provided to the population, and on ensuring that more children go to school, and more persons in need of care receive it. Improving the delivery of basic services is one of four key pillars of the interim poverty reduction strategy (Republic of Liberia, 2006) that was adopted in 2006, and this priority was reaffirmed in the full poverty reduction strategy (Republic of Liberia, 2008). It is in order to inform the preparation of this poverty reduction strategy that this chapter was prepared, with a focus on education services as seen from the point of view of users. Very little has been written on the education system in Liberia since the start of the conflict, in large part because of lack of good data. A few recent reports provide a partial diagnostic of the education sector and suggestions for priority actions are available in these reports (Ministry of Education, 2007; Ministry of Education and UNICEF, 2004; and Heninger et al., 2006; see also UNDP, 2006, and Humphreys and Richards, 2005, 35

A World Bank Study

for a broader discussion related to the Millennium Development Goals in Liberia, and International Labour Organization, 2009, for a rapid assessment of the impact of the recent crisis). However, these reports are not based on recent household survey data that provide detailed more descriptive information on who benefits from education services, what households think about the quality of the services that they receive, or why they do not use those services, whether due to their cost or the distance to facilities for example. The objective of this chapter is to provide a basic diagnostic of the education system in Liberia as seen from the point of view of households. The diagnostic is based on the newly available nationally representative CWIQ (Core Welfare Questionnaire Indicator) survey that was implemented in 2007 by the Liberia Institute of Statistics. The survey includes detailed data on school enrollment as well as the reasons for not going to school, and the degree of satisfaction of households with the services received. Data are also available on private spending for education, as well as on distances to facilities. As in other Anglophone countries in West Africa, Liberia’s education system consists of four main levels: primary schools (6 years of study), junior secondary schools (3 years), senior secondary schools (3 years), and tertiary education. Enrollment rates in pre-schools are very low, so that pre-schools are not discussed here. Vocational and technical education and training is available at the secondary and in some cases tertiary levels, but we do not have good data on that in the 2007 CWIQ survey, hence we do not discuss this segment of the education sector either. In this chapter, we focus therefore for the most part on primary and secondary education indicators, given that the share of youths pursuing post-secondary education is very low, but we do discuss some aspects of tertiary education as well, especially in terms of satisfaction rates with the services provided as well as in terms of the benefit incidence of public spending for education. The chapter is structured as follows. Section 2 provides descriptive statistics on school enrollment rates (gross and net), the reasons for not enrolling children (either for children who were never enrolled or for children who were once enrolled but have dropped out of school), and satisfaction with schools services. Section 3 is devoted to a benefit incidence analysis of public spending for education. Section 4 discusses the determinants of school enrollment. A brief conclusion follows.

2. School Enrollment, Reason for Not Enrolling, and Satisfaction with Schools 2.1. School Enrollment Rates and Types of Schools Attended

Table 3.1 provides measures of net and gross enrollment rates as obtained from the 2007 CWIQ survey. At the national level, in primary schools the net enrollment rate is 60.1 percent, while the gross enrollment rate is a much higher 120.7 percent. In secondary schools, the corresponding figures are much lower, at respectively 15.2 percent and 51.3 percent. Enrollment rates are lower in rural areas than in urban areas, and they are also lower among poorer households identified here according to five quintiles of consumption per equivalent adult (for an analysis of poverty in Liberia based on the 2007 CWIQ survey, see Backiny-Yetna et al., 2012). The first quintile “Q1” represents the poorest 20 percent of the population, and the top quintile “Q5” the richest 20 percent. Enrollment rates also remain slightly lower for girls than for boys, but recent efforts to improve girls’ education have helped in reducing the gap so that differences now are relatively small, at least at the primary level (differences remain substantially larger at the secondary level).

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 3.1: Net and gross enrollment rates in primary and secondary schools, 2007 Residence area Urban

Rural

Welfare quintile Q1

Total

Primary enrollment rate Net enrollment rate (6–11) Total

64.8

58.0

51.2

58.3

60.2

65.7

66.1

60.1

Male

64.2

56.9

51.1

58.9

58.2

63.2

66.1

59.0

Female

65.3

59.2

51.3

57.7

62.2

68.5

66.1

61.2

Gross enrollment rate Total

122.9

119.8

109.4

125.6

122.3

121.2

126.1

120.7

Male

117.2

121.7

115.9

131.9

115.6

113.3

126.5

120.4

Female

128.4

117.5

101.3

118.2

129.4

130.2

125.7

121.0

Secondary enrollment rate Net enrollment rate (12–17) Total

25.4

10.1

11.4

12.2

13.0

19.8

21.5

15.2

Male

27.7

11.2

12.6

13.3

15.4

19.1

22.8

16.0

Female

23.4

8.7

9.6

10.7

10.8

20.4

20.3

14.2

39.7

40.1

43.0

42.8

65.3

71.8

51.3

Gross enrollment rate Total

74.4

Male

86.8

44.9

43.4

51.0

54.5

72.5

75.6

57.2

Female

63.7

33.1

35.1

32.6

31.7

59.2

68.0

44.7

Source: Authors’ estimates based on 2007 CWIQ survey.

For a country that only recently emerged from conflict, the above enrollment rates, especially in gross terms, are not as low as one might have feared. As noted by the diagnostic prepared by the Ministry of Education in its program document for the Fast Track Initiative (Ministry of Education, 2007: 7-10), the lack of proper infrastructure and teachers, the lack of security in the country and the high cost for families of education linked to user fees led to a sharp decline in enrollment in the early part of this decade, especially for girls (National Policy of Girls Education, MOE, Government of Liberia, 2005). To stem this decline, an Education Law was adopted in 2001 to make primary education free and compulsory, but resources had been lacking to implement this policy. Renewed efforts by the administration of President Ellen Johnson Sirleaf since 2006 to promote school enrollment as well as a return to peace have led according to administrative data to a dramatic increase in public primary school enrolment of 24 percent for girls and 18 percent for boys between 2006 and 2007. As a result, many children and youth with limited previous exposure to formal education have now returned to school, which also explains the large differences between net and gross enrollment rates observed in table 3.1. A school census implemented in 2006 suggested that only 15 percent of students in the first year of primary school were of the right age (six to seven years of age), and half of the students were between 11 and 20 years of age.

A World Bank Study

Table 3.2: Type of school attended, 2007 Residence area Urban Rural

Quintile Q1

Government Religious organization Private Community Other Total

25.8 8.0 63.0 2.9 0.2 100.0

71.1 1.5 18.0 8.0 1.3 100.0

73.7 2.0 19.4 3.0 2.0 100.0

Government Religious organization Private Community Other Total

28.3 6.0 65.0 0.5 0.1 100.0

65.0 2.7 28.4 3.4 0.5 100.0

66.1 2.5 30.2 0.8 0.4 100.0

Government Religious organization Private Community Other Total

52.3 3.9 42.1 0.0 1.7 100.0

19.7 16.5 63.5 0.0 0.3 100.0

63.6 0.0 19.6 0.0 16.8 100.0

Government Religious organization Private Community Other Total

33.7 7.0 57.6 1.6 0.1 100.0

72.0 1.8 19.0 6.3 0.9 100.0

74.2 1.8 20.4 2.6 1.0 100.0

Government Religious organization Private Community Other Total

23.8 7.1 66.2 2.3 0.5 100.0

65.9 2.1 22.7 7.9 1.4 100.0

68.1 2.6 24.3 2.2 2.9 100.0

Government Religious organization Private Community Other Total

28.8 7.1 61.9 1.9 0.3 100.0

69.3 1.9 20.6 7.0 1.1 100.0

71.8 2.1 21.9 2.4 1.8 100.0

Primary 66.3 56.4 2.2 1.5 22.0 30.7 8.9 10.4 0.5 1.0 100.0 100.0 Secondary 67.7 47.5 0.8 2.4 30.2 45.0 0.9 4.9 0.4 0.1 100.0 100.0 Post-secondary 23.5 53.1 0.0 8.6 76.5 38.2 0.0 0.0 0.0 0.0 100.0 100.0 Boys—all levels 69.0 59.2 1.6 1.5 22.6 31.0 6.4 7.8 0.4 0.6 100.0 100.0 Girls—all levels 62.1 49.0 2.4 2.4 26.8 37.5 8.0 10.0 0.6 1.1 100.0 100.0 All respondents 66.2 54.3 1.9 1.9 24.3 34.1 7.1 8.9 0.5 0.8 100.0 100.0

Source: Authors’ estimates based on 2007 CWIQ survey.

Total

47.3 6.7 40.3 4.9 0.8 100.0

40.4 5.6 49.0 4.4 0.5 100.0

57.0 3.6 32.1 6.4 1.0 100.0

31.6 8.6 57.3 2.0 0.5 100.0

33.7 5.4 59.3 1.5 0.2 100.0

47.3 4.3 46.1 2.0 0.3 100.0

34.5 6.2 59.0 0.0 0.3 100.0

53.4 7.0 38.1 0.0 1.4 100.0

46.1 6.3 46.2 0.0 1.4 100.0

46.2 7.3 42.1 3.9 0.6 100.0

43.4 6.2 47.2 2.7 0.5 100.0

58.7 3.6 32.4 4.7 0.6 100.0

38.6 7.0 49.8 3.9 0.7 100.0

35.9 5.2 54.6 3.9 0.4 100.0

48.9 4.1 40.3 5.6 1.1 100.0

42.4 7.2 45.9 3.9 0.7 100.0

39.6 5.7 51.0 3.3 0.5 100.0

54.2 3.8 36.0 5.1 0.8 100.0

Poverty and the Policy Response to the Economic Crisis in Liberia

Due in part to the inability of the state to provide services during the civil war, nongovernmental schools play a very important role in Liberiaâ&#x20AC;&#x2122;s education system, accounting for about 40 percent of the enrollment of primary school students and more than half the enrollment of secondary and tertiary levels students. This is shown in table 3.2, which provides the share of students by level attending different types of schools. Still, government schools remain the primary service providers for rural students, as well as for the poor. For example, more than three fourths of students belonging to the first quintile of household consumption go to government primary schools, while the proportion is only about 40 percent in the top quintile. Among non-governmental primary schools, private schools are chosen mostly by the better off, while community schools tend to serve more the poor. At the secondary and post-secondary level, the role of private schools is even more pronounced, with these schools serving as many children as government schools. There is also an interesting difference between boys and girls, with boys more likely to go to government schools than girls, while the reverse is true for private schools. Overall, only slightly more than half (54.2 percent) of all students go to government schools, with private schools serving more than a third (36.0 percent) of the students, and the rest using community or religious schools for the most part. 2.2. Reasons for Not Going to School

Despite a rapid increase in enrollment since 2006, many children are still not enrolled in school today. Tables 3.3 and 3.4 provide the reasons invoked by parents for not sending their children to school. In table 3.3, we provide data on the children that have never been enrolled in school. This is done by age group. Despite the gratuity of public schools, costs seem to remain the main factor preventing many children from attending schools. The second main reason invoked is the distance separating the children from the nearest school. As expected, the issue of cost is more serious for children in the bottom quintiles, while the issue of distance is similar in both cases (note however that the data is provided in percentage terms of students who never started schooling; clearly that percentage itself is higher among the poorest segments of the population). There are some differences between boys and girls in reasons for never going to school (for example, among the very poor, cost is more an issue for boys than for girls, while age is more of an issue for girls than boys). Yet overall the reasons for never enrolling are similar between the two sexes. In tables 3.4a to 3.4c, we provide data on children who are not any more going to school, but who had been enrolled in the past. This is done by age group as well as by gender. Again, despite the gratuity of public schools, costs are the main reason for not pursuing oneâ&#x20AC;&#x2122;s education, apart from the fact that many children are waiting for taking an examination in order to be enrolled. The large share of the students who are not enrolled today due to the fact that they still must take an examination seems to be a situation that is peculiar to Liberia, and may have to do with the large influx of new students of many different ages who are returning to school, after the end of the conflict. Apart from these two reasons, the distance to schools is also one factor preventing children to continue their education, even though they did go to school at some point. For the various samples considered in tables 3.4 to 3.6, the issue of cost is actually more serious in urban than in rural areas, perhaps because a larger share of students tend to go to private schools in urban areas. Interestingly, the issue of cost is cited more often for urban boys than for urban girls, while in rural areas, the differences are not

A World Bank Study

large. This may be related to the fact that boys may be seen as more likely to be able to contribute to the income of the household in urban areas than girls. By quintiles, the patterns do not suggest in most cases large and systematic differences in the reasons for stopping to go to school. There is however a clear indication in many (although not all) cases that for the bottom quintile, the issue of the distance from the schools matters more than for the other quintiles. Table 3.3: Reason for never starting going to school, 2007 Residence area Urban

Rural

Quintile Q1

Total

Boys aged 6–11 Too young

8.2

19.7

16.4

15.4

21.1

20.1

19.3

17.9

Too far away

7.5

28.1

31.0

19.9

24.6

26.0

18.2

24.9

Too expensive

72.0

56.5

68.0

70.3

46.1

48.7

45.7

58.8

6.5

1.5

3.2

2.1

4.7

0.0

0.1

2.2

Working (home or job) Useless/uninteresting

0.0

1.2

0.8

0.6

0.0

3.1

0.4

1.0

Illness

2.9

3.6

2.5

4.2

1.3

7.1

3.5

Orphaned Other

0.0

0.8

0.0

2.8

0.0

0.6

19.6

15.1

7.3

8.0

21.2

28.3

26.7

15.8

Girls aged 6–11 10.0

17.9

25.1

7.3

18.1

25.2

16.2

Too far away

Too young

5.7

31.2

17.7

29.1

30.1

13.3

40.7

25.7

Too expensive

78.2

54.7

56.6

64.6

66.4

65.2

43.3

59.8

8.7

2.2

1.7

4.2

7.4

3.7

1.3

3.6

Working (home or job) Useless/uninteresting

1.5

0.6

0.1

0.0

1.0

2.6

2.0

0.8

Illness

4.1

1.8

1.9

0.4

4.8

0.4

5.2

2.3

Orphaned

0.0

10.4

18.3

16.4

15.0

14.0

9.9

29.2

16.6

Too young

9.2

18.9

20.6

11.3

13.9

19.4

22.4

17.1

Too far away

6.5

29.5

24.7

24.6

27.5

21.4

30.0

25.3

Too expensive

75.5

55.6

62.6

67.4

56.7

54.7

44.5

59.3

7.7

1.8

2.5

3.2

6.1

1.3

0.7

2.9

Other

Children aged 6–11

Working (home or job) Useless/uninteresting

0.9

0.4

0.3

0.5

2.9

1.2

0.9

Illness

3.6

2.8

2.2

4.5

1.0

6.1

2.9

Orphaned Other

0.0

0.4

0.0

1.3

0.0

0.3

14.4

16.6

11.6

17.4

21.6

28.0

16.2

Source: Authors’ estimates based on 2007 CWIQ survey.

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 3.4a: Reason for not enrolling for previously enrolled children by age, 2007 Residence area Urban

Rural

Quintile Q1

Total

Children aged 6–11 Completed school

0.0

Too far away

0.3

5.6

8.7

0.0

0.9

5.7

3.2

3.7 18.3

23.6

15.3

17.5

17.3

16.7

14.7

26.0

Working (home or job)

Lack of money/too expensive

0.1

0.2

0.0

0.2

0.0

0.7

0.2

Illness

1.0

1.9

2.1

1.2

1.0

2.2

1.3

1.6

Pregnancy

0.0

Got married

0.0

0.2

0.0

0.8

0.0

0.2

Useless/uninteresting

0.3

1.4

1.6

0.2

0.4

2.7

0.0

1.0

Failed exam

0.1

0.5

0.0

0.3

0.0

1.2

0.3

0.3 74.8

75.0

74.8

69.8

80.9

79.9

76.7

66.2

Dismissed

Awaiting admission

0.1

1.3

1.1

0.0

2.1

1.2

0.9

Orphaned

0.0

Other

3.5

10.6

3.9

10.1

17.1

2.1

5.4

8.0

Children aged 12–17 Completed school

3.3

0.2

1.7

0.0

5.1

0.1

1.3

Too far away

0.3

2.8

0.2

3.4

3.2

1.9

0.0

1.9 19.9

24.2

17.4

20.7

20.9

22.1

16.8

17.9

Working (home or job)

Lack of money/too expensive

3.1

1.9

1.2

0.0

1.9

6.5

2.4

2.3

Illness

2.4

1.2

0.6

3.2

0.5

3.8

0.0

1.6

Pregnancy

6.5

5.1

4.9

3.2

8.3

7.5

5.6

Got married

0.4

0.5

0.6

0.1

0.6

0.0

1.1

0.4

Useless/uninteresting

2.3

4.8

2.7

6.0

4.2

2.9

3.5

3.9

Failed exam

0.2

0.5

0.6

0.0

0.7

0.9

0.4 63.8

59.5

66.2

65.4

59.9

67.7

58.4

67.4

Dismissed

Awaiting admission

0.4

1.5

1.2

2.2

0.0

1.7

0.6

1.1

Orphaned

0.0

Other

7.6

9.6

6.1

12.8

14.5

3.6

5.0

8.9

All respondents 6–17 Completed school

0.9

0.1

0.9

0.0

1.1

0.0

0.4

Too far away

0.2

4.1

5.1

0.8

1.7

4.2

1.9

2.7 18.0

23.9

14.7

17.2

17.4

18.3

14.7

22.5

Working (home or job)

Lack of money/too expensive

1.2

0.3

0.6

0.0

0.3

1.0

1.3

0.6

Illness

1.6

1.7

1.6

2.5

0.8

2.9

0.8

1.7

Pregnancy

2.0

1.5

1.7

1.8

1.2

1.3

2.5

1.7

Got married

0.0

0.2

0.0

0.5

0.3

0.2

Useless/uninteresting

1.1

2.7

1.1

3.5

1.9

2.8

1.4

2.2

Failed exam

0.2

0.5

0.3

0.1

0.0

1.1

0.6

0.4

69.7

73.4

70.1

71.6

76.1

73.9

67.4

72.0

Awaiting admission Dismissed

0.3

1.4

1.3

0.0

1.7

1.0

Orphaned

0.0

Other

4.9

10.4

4.8

12.4

15.6

2.3

5.5

8.5

Source: Authors’ estimates based on 2007 CWIQ survey.

A World Bank Study

Table 3.4b: Reason for not enrolling for previously enrolled boys, 2007 Residence area Urban

Rural

Quintile Q1

Total

Boys aged 6–11 Completed school

0.0

Too far away

0.0

6.5

11.8

0.0

1.5

1.7

6.2

4.1 15.9

Lack of money/too expensive

25.8

10.3

16.3

19.4

16.1

13.5

12.9

Working (home or job)

0.3

0.5

0.0

0.4

0.0

2.3

0.4

Illness

2.0

1.4

0.0

2.3

0.5

4.5

1.4

1.6

Pregnancy

0.0

Got married

0.0

Useless/uninteresting

0.0

2.5

1.7

0.0

6.2

0.0

1.6

Failed exam

0.3

0.0

0.5

0.0

0.1 76.2

Awaiting admission

72.1

78.5

72.4

77.8

80.8

75.2

72.7

Dismissed

0.3

0.2

0.5

0.0

0.7

0.0

0.2

Orphaned

0.0

Other

4.0

13.0

3.2

5.2

21.3

2.6

15.9

9.8 0.5

Boys aged 12–17 Completed school

1.5

0.0

2.1

0.0

Too far away

0.2

2.9

1.1

2.9

4.9

0.7

0.0

2.1

27.1

11.0

18.4

10.1

17.8

16.7

15.9

Lack of money/too expensive Working (home or job)

5.2

0.7

2.2

0.0

0.6

4.9

3.7

2.0

Illness

1.8

1.5

0.7

4.4

0.3

2.6

0.0

1.6 0.2

Pregnancy

0.0

0.3

0.0

0.8

0.0

Got married

0.0

Useless/uninteresting

3.0

2.8

0.0

5.5

6.5

0.5

0.7

2.9

Failed exam Awaiting admission

0.6

0.0

1.7

0.6

62.8

78.9

75.2

69.8

73.3

76.9

76.4

74.1 1.7

Dismissed

0.7

2.1

1.7

4.4

0.0

0.9

1.2

Orphaned

0.0

10.2

6.7

4.4

12.6

11.8

0.8

6.7

7.7

Other

Boys-All respondents 6–17

Completed school

0.7

0.0

1.0

0.0

0.2

Too far away

0.1

4.7

6.7

1.5

3.0

1.3

2.7

3.2 15.9

Lack of money/too expensive

26.4

10.7

17.3

14.6

16.9

14.9

15.1

Working (home or job)

2.4

0.6

1.0

0.0

0.5

2.1

3.1

1.2

Illness

1.9

1.5

0.3

3.4

0.4

3.7

0.6

1.6

Pregnancy

0.0

0.1

0.0

0.4

0.0

0.1

Got married

0.0

Useless/uninteresting

1.3

2.7

0.9

2.8

3.0

3.7

0.4

2.2

Failed exam

0.4

0.3

0.0

0.3

0.0

0.8

0.9

0.3

68.0

78.7

73.7

77.4

75.9

74.8

75.2

Awaiting admission Dismissed

0.5

1.2

1.1

2.3

0.0

0.8

0.7

0.9

Orphaned

0.0

Other

6.7

9.8

3.7

9.0

17.0

1.8

10.7

8.8

Source: Authors’ estimates based on 2007 CWIQ survey.

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 3.4c: Reason for not enrolling for previously enrolled girls, 2007 Residence area Urban

Rural

Quintile Q1

Total

Girls aged 6–11 Completed school

0.0

Too far away

0.8

6.3

4.9

0.0

12.0

2.7

4.1 21.5

22.2

21.0

14.5

15.3

15.4

21.1

37.2

Working (home or job)

Lack of money/too expensive

0.0

Illness

0.0

1.8

6.2

0.0

1.1

Pregnancy

0.0

Got married

0.0

0.7

0.0

1.8

0.0

0.4

Useless/uninteresting

0.4

0.9

0.0

1.2

0.0

0.4

Failed exam

0.0

1.3

0.0

2.9

0.6

0.8 72.4

76.3

69.9

71.5

84.7

81.5

70.9

57.6

Dismissed

Awaiting admission

0.0

2.5

0.0

4.3

2.6

1.5

Orphaned

0.0

Other

2.8

8.5

3.7

19.6

8.8

2.2

0.9

6.3

Girls aged 12–17

Completed school

2.2

0.4

1.3

0.0

4.0

0.0

1.1

Too far away

0.0

0.9

0.0

2.1

0.0

0.6 19.2

20.9

18.2

19.4

24.9

23.7

8.6

19.5

Working (home or job)

Lack of money/too expensive

0.0

Illness

2.3

0.7

2.1

2.3

4.0

1.9

2.3

Pregnancy

7.2

5.7

7.7

6.0

4.9

4.5

8.7

6.3

Got married

0.0

0.4

0.0

0.1

0.0

1.2

0.3

Useless/uninteresting

1.4

4.8

1.6

7.6

0.3

4.0

4.4

3.5

Failed exam

0.0

0.4

1.4

0.0

0.2 65.4

66.8

64.5

58.9

58.0

69.0

73.0

65.6

Dismissed

Awaiting admission

0.3

0.9

3.0

0.0

1.0

0.0

0.7

Orphaned

0.0

Other

3.5

13.4

8.8

15.2

18.2

3.3

2.1

9.7

Girls-All respondents 6–17 Completed school

1.2

0.2

0.7

0.0

2.1

0.0

0.6

Too far away

0.4

3.4

2.9

0.0

6.8

1.3

2.2 20.2

21.5

19.5

17.1

21.0

20.0

14.5

28.3

Working (home or job)

Lack of money/too expensive

0.0

Illness

1.2

2.1

3.3

1.3

2.1

1.0

1.7

Pregnancy

3.8

3.1

4.0

3.6

2.7

2.4

4.4

3.4

Got married

0.0

0.5

0.0

0.1

0.0

0.9

0.6

0.3

Useless/uninteresting

1.0

2.8

1.3

4.5

0.7

2.1

2.2

2.1

Failed exam

0.0

0.8

0.0

1.3

0.3

0.5

71.3

67.0

64.9

68.9

74.6

72.0

61.6

68.6

Awaiting admission Dismissed

0.2

1.7

1.5

0.0

2.6

1.3

1.1

Orphaned

0.0

Other

3.2

11.2

6.4

17.0

14.0

2.8

1.5

8.1

Source: Authors’ estimates based on 2007 CWIQ survey.

A World Bank Study

Table 3.5 provides data on private education spending by households, for those households who have at least one child enrolled in school (the data are not available for each separate child going to school). The largest expenditure in terms of the share of total spending for education is for school fees at the primary school level. This may sound surprising, given the gratuity of public schools, but it may be due to the fact that a large share of children is enrolled in private schools. Still, the share of total education spending allocated to primary school fees does not vary much by quintile of consumption per equivalent adult, which is a bit surprising given the fact that poorer households tend to send their children to public schools more. Schools fees for secondary education also absorb a large part of the private education budget of families, as do school uniforms. Table 3.5: Private household expenditure for education, shares, 2007

School uniforms Raincoats School books Files and file folders Stationery for school Notebooks School bags and knapsacks Other school material Writing and drawing materials School fees—pre-school School fees—primary school School fees—general secondary school School fees—technical secondary school School fees—higher education Professional/vocational training fees Total Share of education in total expenditure

Residence area Urban Rural 30.8 12.3 0.3 0.4 3.7 2.3 0.1 0.5 1.0 0.8 9.3 4.3 4.0 3.1 1.2 0.7 0.8 0.2 6.5 8.3 20.5 25.6 15.1 25.5 1.2 1.1 4.4 13.4 1.1 1.5 100.0 100.0 2.1 5.1

Q1 30.4 0.4 4.1 0.2 0.5 8.2 3.3 0.7 0.5 6.3 23.3 16.4 1.4 3.9 0.6 100.0 3.8

Q2 28.4 0.3 2.7 0.3 1.0 8.7 4.0 1.0 0.6 6.4 24.4 16.8 1.4 3.7 0.4 100.0 3.5

Quintile Q3 Q4 22.3 15.7 0.3 0.4 2.1 2.5 0.4 0.3 0.8 0.7 7.1 5.2 3.1 3.4 1.2 0.7 0.4 0.4 7.2 7.7 25.3 24.5 19.6 23.3 1.6 1.0 7.7 12.0 1.0 2.1 100.0 100.0 3.6 4.1

Q5 13.2 0.4 3.1 0.5 1.1 4.8 3.3 0.8 0.3 8.6 22.6 25.0 0.9 13.9 1.5 100.0 2.9

Total 18.6 0.4 2.8 0.4 0.9 6.0 3.4 0.8 0.4 7.7 23.9 21.9 1.1 10.3 1.4 100.0 3.4

Source: Authors’ estimates based on 2007 CWIQ survey.

As a share of total consumption, table 3.5 shows that education spending has a higher cost for the poor, but in absolute value, better off households tend to spend significantly more on average. The data on total private spending for education is provided in levels in table 3.6. On a per capita basis, households in the top decile of the population (ranked according to consumption per equivalent adult) spend ten times as much as households in the bottom. The total private spending for education is estimated at close to L$1.7 billion (about US$27 million), which is significantly higher than the total budget of the Ministry of Education (at about $10 million; see the discussion in section 3 for more details). In part due to the legacy of the war, the government’s education budget is only a fraction of total spending on the public education system. In many cases, NGOs are topping up salaries for teachers, as well as providing other incentives, books, and school supplies directly to public schools. Unfortunately these aid flows are not being tracked well, so that the government does not have a clear idea of how much is currently spent

Poverty and the Policy Response to the Economic Crisis in Liberia

on public education overall (for health, some have suggested that total public health spending may be of the order of US$100 million for 2007, of which only $15 million is budgeted government expenditure)2. Table 3.6: Private household expenditure for education, amounts, 2007 Deciles of per eq. adult consumption

1 2 3 4 5 6 7 8 9 10 Total

Total population

270,469 270,582 270,477 270,761 269,714 271,127 270,714 269,729 271,538 270,273 2,705,385

Total students

Total expenditure (millions of L$)

Total expenditure in education (millions of L$)

Per capita expenditure (L$)

Per capita expenditure in education (L$)

Per student expenditure in education (L$)

71,053 75,181 77,027 83,768 83,148 76,101 82,414 90,838 74,314 94,097 807,942

1,234 2,133 2,765 3,292 3,802 4,460 5,020 5,938 7,287 13,386 49,316

54 72 91 119 155 140 202 249 222 384 1,687

4,562 7,883 10,222 12,158 14,096 16,451 18,544 22,013 26,834 49,526 18,229

201 267 335 438 573 516 747 923 816 1,420 624

765 962 1,175 1,415 1,860 1,837 2,455 2,741 2,982 4,079 2,088

Source: Authors’ estimates based on 2007 CWIQ survey.

In table 3.7, access is measured by the distance from the nearest school. Remember that in tables 3.4a–3.4b, access is not mentioned as one of the two main reasons for not going to school, once a child has already gone in the past. However, in table 3.3, access was a key reason for never having gone to school. In table 3.7, we provide data on the average time it takes to reach various types of facilities. At the national level, primary schools are on average within half an hour of where children live, but in rural areas, it takes more than three quarters of an hour to reach the primary school. Secondary schools are located much further away, at more than three hours of a rural household’s dwelling on average. These distances to schools are high in comparison of what has been observed in other countries, which justifies an effort on the part of the Ministry of Education as well as donors not only of rehabilitating existing schools, but also of building new schools and classrooms in order to improve access. Table 3.7: Average time (in minutes) to the nearest infrastructure, 2007

Supply of drinking water Food market Public transportation Primary school Secondary school Health clinic/hospital All season road Any road (vehicle)

Residence area Urban Rural 9.7 8.4 23.2 179.1 12.8 161.7 15.5 46.5 24.3 203.0 29.6 151.6 16.7 333.6 6.1 33.0

Quintile Q1 11.4 162.8 145.7 33.4 114.1 124.8 167.9 31.7

Source: Authors’ estimates based on 2007 CWIQ survey.

Q2 8.4 161.0 140.4 46.1 203.0 143.4 322.8 26.6

Q3 8.9 167.6 152.0 46.6 198.9 145.4 323.8 25.0

Q4 7.5 113.5 77.2 27.3 116.2 99.5 227.8 21.4

Q5 8.4 71.0 77.5 32.5 113.3 71.0 153.2 20.5

Total 8.8 129.8 114.6 36.7 146.3 113.0 233.3 24.5

A World Bank Study

2.3. Satisfaction with Education Services and Reasons for Non-satisfaction

The gains that have been achieved recently in enrollment at the primary level are impressive, but this has placed a larger burden on an education system that has only limited resources. As a result, while quality was already an issue in the past, it is probably even more of an issue now, especially as many of the new children that have returned to schools have very limited skills in terms of literacy and numeracy. In addition, many of the schools suffer from dilapidated infrastructure. According a Rapid Assessment of Learning Spaces by MOE/UNICEF (2004) and a subsequent census of public primary schools conducted by the Ministry of Education (2006), one in five schools in Liberia have been destroyed during the war, with the rest of the infrastructure in need of repair. Many schools lack basic functioning amenities such as water and latrines, and desks are available only for one in four children. The report prepared by the Ministry of Education (2007) for the fast track initiative suggests that the pupil textbook ratio is very low, at 27 to one in public primary schools and nine to one at the secondary level. Furthermore, as many qualified teachers have left the country or have been displaced, and as training of new teachers was affected by the conflict as well, more than sixty percent of teachers today lack the formal qualifications in principle required for teaching. Teacher salaries are very low, of the order of only US$200 to US$300 per year, so that teachers are forced to find other means of livelihoods, among others by raising user fees (although this practice has been reduced in recent years). Under such conditions, one might expect satisfaction rates with education services to be low in Liberia. This is indeed the case, especially for children enrolled in public schools as shown in tables 3.8 to 3.11. At the primary level, only about half of all children enrolled in public schools have parents who are satisfied with the services they receive, versus about 60 percent in private schools. In public schools, the main reasons for the lack of satisfaction are the lack of books or supplies, the fact that there are not enough teachers, the fact that facilities are in poor condition, the long distance to schools, and poor teaching. In private schools, low satisfaction in primary schools is due to high fees and the lack of books or supplies. The rates of satisfaction in secondary schools are of the same order of magnitude, that is not very good, and the complaints are similar in both public and private schools. Satisfaction with tertiary education is also low, but there high fees are one of the main reasons for not being satisfied in both public and private schools (in public secondary schools, slightly less than ten percent of students complain about high fees, versus close to 30 percent for private schools). Clearly, at all levels, there is room for improvements in the quality of the education that is provided to children and youth.

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 3.8: Problems encountered at school, primary, 2007 Residence area Urban

Rural

Quintile Q1

Total

Primary—Public None (satisfied)

44.9

50.4

47.6

57.9

42.4

54.4

42.6

49.6

Lack of books or supplies

26.2

20.7

18.1

18.2

25.7

24.4

23.8

21.5

Poor teaching

9.8

9.9

11.2

9.0

7.5

10.6

11.7

9.9

Not enough teachers

8.2

16.3

16.9

13.6

16.6

18.3

9.1

15.2

Teachers often absent

9.8

6.4

9.7

7.6

5.8

4.9

4.6

6.9

Lack of space

3.5

7.0

7.8

5.8

6.8

3.6

9.0

6.5

Facilities in bad condition

6.9

13.6

17.9

11.2

8.8

13.3

10.6

12.6

High fees

4.6

4.4

7.4

1.5

1.8

1.7

12.0

4.4

14.2

11.6

11.8

9.1

18.1

11.2

9.6

12.0

2.6

2.9

3.2

3.6

1.4

1.2

5.3

2.9

Long distance to school Other

Primary—Private None (satisfied)

60.3

60.9

62.2

56.6

59.3

64.6

59.6

60.6

Lack of books or supplies

11.6

17.0

14.4

12.6

15.0

13.6

14.5

14.1

Poor teaching

1.7

4.1

2.4

1.0

3.4

1.5

4.9

2.8

Not enough teachers

1.5

8.2

5.9

4.4

6.1

2.6

4.9

4.6

Teachers often absent

1.4

2.8

3.9

1.5

2.0

2.3

2.1

Lack of space

1.5

8.4

4.1

6.6

2.9

2.4

4.7

Facilities in bad condition

10.9

2.7

9.6

4.1

7.9

9.1

5.2

3.5

5.9

28.8

11.2

17.5

25.1

21.0

16.0

23.3

20.6

Long distance to school

5.4

6.6

3.7

5.0

9.3

5.4

5.9

Other

1.0

1.7

5.9

0.3

0.5

1.2

0.7

1.3

High fees

Primary—All None (satisfied)

56.4

53.4

51.5

57.5

49.7

59.8

52.7

54.3

Lack of books or supplies

15.3

19.6

17.1

16.3

21.1

18.7

18.3

Poor teaching

3.8

8.2

8.8

6.3

5.8

7.7

6.8

Not enough teachers

3.3

14.0

10.5

12.1

10.1

6.6

10.6

Teachers often absent

3.6

5.4

7.1

6.3

3.9

3.4

3.3

4.8

Lack of space

2.0

7.4

8.6

5.2

6.8

3.2

5.1

5.7

Facilities in bad condition

3.8

12.4

14.3

10.1

9.0

6.4

9.7

22.6

6.4

10.1

9.5

10.1

9.3

18.7

11.4

Long distance to school

7.7

10.2

9.6

7.7

14.3

8.1

7.1

9.4

Other

1.4

2.6

3.9

2.5

1.0

1.2

2.6

2.2

High fees

Source: Authors’ estimates based on 2007 CWIQ survey.

A World Bank Study

Table 3.9: Problems encountered at school, secondary, 2007 Residence area Urban

Rural

Quintile Q1

Total

Secondary—Public None (satisfied)

52.3

54.2

54.0

51.5

50.3

68.1

46.4

53.6

Lack of books or supplies

27.6

20.3

17.7

29.7

32.9

14.9

14.1

22.4

Poor teaching

9.5

9.8

9.0

7.7

9.7

11.8

9.5

Not enough teachers

9.7

12.1

13.3

15.8

9.0

8.0

7.8

11.4

Teachers often absent

5.9

5.8

9.3

7.2

3.1

3.5

4.1

5.8

Lack of space

4.3

7.1

4.2

11.4

3.0

4.7

6.6

6.2

Facilities in bad condition

4.5

5.2

4.3

2.2

8.4

6.8

5.0

High fees

2.6

11.7

6.0

8.7

7.7

1.0

23.1

9.1

11.1

13.7

12.6

17.0

15.6

4.9

11.9

12.9

1.9

3.0

0.0

2.3

4.0

2.6

5.6

2.7

Long distance to school Other

Secondary—Private None (satisfied)

53.1

61.1

57.3

50.6

67.4

49.8

56.3

55.8

Lack of books or supplies

16.9

9.7

12.0

19.7

8.7

16.7

14.7

14.5

Poor teaching

1.0

3.1

0.5

2.5

1.9

1.6

2.0

1.7

Not enough teachers

0.8

5.9

5.6

0.0

2.3

4.5

0.5

2.6

Teachers often absent

0.4

2.8

3.6

1.5

0.5

1.3

1.2

Lack of space

1.6

2.5

2.1

0.4

2.8

2.0

1.7

1.9

Facilities in bad condition

2.1

6.3

3.2

0.9

7.5

2.6

3.2

3.5

34.3

20.7

18.7

39.3

24.0

36.5

26.7

29.6

Long distance to school

6.5

10.5

14.5

4.7

7.1

8.1

6.8

7.9

Other

2.6

1.0

3.0

2.5

0.6

2.5

1.9

2.0

High fees

Secondary—All None (satisfied)

52.9

56.6

55.1

51.2

59.3

55.6

53.0

54.8

Lack of books or supplies

20.0

16.6

15.8

26.5

20.2

16.2

14.5

18.2

Poor teaching

3.4

7.3

6.6

6.9

4.7

4.2

5.3

5.4

Not enough teachers

3.3

9.9

10.7

5.5

5.6

3.0

6.7

Teachers often absent

1.9

4.7

7.3

5.3

1.7

1.4

2.2

3.4

Lack of space

2.4

5.5

3.5

7.8

2.9

3.4

4.0

Facilities in bad condition

2.8

5.6

3.9

3.2

5.0

4.4

4.2

25.3

14.8

10.3

18.6

16.2

25.2

25.5

19.9

Long distance to school

7.8

12.6

13.2

13.1

11.2

7.1

8.5

10.3

Other

2.4

2.3

1.0

2.4

2.2

2.5

3.1

2.3

High fees

Source: Authors’ estimates based on 2007 CWIQ survey.

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 3.10: Problems encountered at school, post-secondary, 2007 Residence area Urban

Rural

Quintile Q1

Total

Post-secondary—Public None (satisfied)

49.1

82.9

100.0

82.5

57.6

48.3

43.7

51.9

Lack of books or supplies

18.8

17.1

0.0

17.5

19.8

23.9

18.2

18.7

Poor teaching Not enough teachers Teachers often absent Lack of space Facilities in bad condition

4.6

0.0

5.2

6.1

4.2

12.2

0.0

2.8

25.8

9.3

11.2

5.7

0.0

5.2

8.0

5.2

10.6

0.0

10.3

14.8

9.8

1.8

0.0

3.4

1.7

17.4

0.0

17.5

34.9

5.7

15.9

16.0

Long distance to school

8.1

11.6

0.0

17.5

2.8

2.2

13.9

8.4

Other

0.0

High fees

Post-secondary—Private None (satisfied) Lack of books or supplies

57.0

75.9

53.7

63.4

74.6

54.7

66.3

62.3

5.2

2.9

0.0

11.8

0.0

4.6

Poor teaching

0.0

Not enough teachers

0.0

Teachers often absent

5.1

0.0

3.9

5.7

3.6

Lack of space

3.3

0.0

6.1

0.0

2.4

Facilities in bad condition

0.7

0.0

1.4

0.5

31.7

21.2

46.3

36.6

12.9

30.2

29.9

28.8

Long distance to school

6.5

8.8

0.0

12.5

2.1

12.8

7.2

Other

2.8

0.0

3.4

1.9

2.0

High fees

Post-secondary—All None (satisfied)

52.9

77.3

83.2

67.9

65.6

52.6

54.2

57.5

Lack of books or supplies

12.3

5.7

0.0

4.1

10.5

15.9

9.7

11.0

Poor teaching

2.4

0.0

1.8

3.3

1.9

Not enough teachers

6.3

0.0

1.5

8.7

5.0

5.1

Teachers often absent

5.4

0.0

4.4

7.0

4.4

Lack of space

7.1

0.0

7.5

7.9

5.8

Facilities in bad condition

1.3

0.0

2.5

1.1

24.3

17.1

16.8

32.1

24.6

21.9

22.4

22.9

Long distance to school

7.4

9.4

0.0

4.1

7.4

2.1

13.4

7.7

Other

1.3

0.0

2.3

0.9

1.1

High fees

Source: Authors’ estimates based on 2007 CWIQ survey.

A World Bank Study

Table 3.11: Problems encountered at school, all levels, 2007 Residence area Urban Rural

Quintile Q1

None (satisfied) Lack of books or supplies Poor teaching Not enough teachers Teachers often absent Lack of space Facilities in bad condition High fees Long distance to school Other

47.9 25.6 8.9 9.3 7.9 4.8 5.4 5.8 12.4 2.0

51.2 20.6 9.8 15.5 6.3 7.0 12.0 5.8 12.0 2.9

49.4 17.8 10.8 16.0 9.5 6.9 14.8 7.0 11.9 2.4

None (satisfied) Lack of books or supplies Poor teaching Not enough teachers Teachers often absent Lack of space Facilities in bad condition High fees Long distance to school Other None (satisfied) Lack of books or supplies Poor teaching Not enough teachers Teachers often absent Lack of space Facilities in bad condition High fees Long distance to school Other

57.8 12.9 1.4 1.2 1.3 1.6 2.4 30.7 5.8 1.6

61.3 14.9 3.7 7.4 2.7 6.8 8.6 13.6 7.5 1.5

60.7 13.5 1.8 5.8 1.0 8.3 3.8 18.2 6.6 5.0

54.8 17.4 4.5 4.7 3.5 2.6 2.9 22.0 7.3 1.2

52.5 19.9 8.6 13.6 5.6 7.2 11.6 9.3 11.1 2.3

52.0 16.4 9.2 13.1 7.6 6.1 12.5 12.5 10.0 2.0

None (satisfied) Lack of books or supplies Poor teaching Not enough teachers Teachers often absent Lack of space Facilities in bad condition High fees Long distance to school Other

55.1 15.7 2.7 2.3 2.9 2.5 3.6 25.1 8.1 2.2

56.6 17.6 7.0 12.3 4.6 6.7 10.0 6.7 10.0 2.7

53.6 17.0 6.7 13.1 6.4 9.1 10.6 6.8 11.0 4.9

Source: Authors’ estimates based on 2007 CWIQ survey.

Q2 Q3 Public—All 56.6 44.3 20.7 27.0 8.9 7.4 14.0 14.7 7.5 5.1 7.0 5.9 9.6 7.3 3.1 3.9 10.8 17.2 3.3 1.8 Private—All 55.5 61.8 13.7 12.9 1.3 2.9 3.4 5.0 3.3 1.2 3.3 5.5 6.3 8.4 28.2 21.5 4.9 8.8 0.7 0.5 Boys 56.0 49.9 20.6 21.5 7.4 6.4 12.2 10.7 6.7 3.5 6.9 6.1 8.4 8.2 10.3 11.9 8.9 13.8 2.4 1.1 Girls 56.6 55.0 15.0 19.5 4.9 4.1 7.9 9.8 5.2 3.1 4.0 5.3 8.7 7.4 13.5 11.9 8.6 12.9 2.6 1.4

Total

56.9 22.5 10.2 16.6 4.6 4.1 11.7 1.7 9.6 1.4

43.6 20.9 11.2 8.8 4.9 9.0 9.0 15.0 10.6 4.9

50.5 21.6 9.6 14.3 6.6 6.6 10.7 5.8 12.1 2.7

59.3 14.5 1.5 3.1 1.7 2.8 4.1 23.4 6.0 1.8

59.0 13.7 3.7 3.3 2.2 2.1 3.3 24.7 6.2 1.1

59.3 13.8 2.4 3.9 1.9 3.8 5.0 23.5 6.5 1.5

57.4 18.7 5.4 10.5 2.4 3.9 7.5 13.7 8.0 1.6

50.7 17.8 7.5 5.8 4.0 4.8 6.3 20.8 8.5 2.7

53.3 19.0 7.2 10.5 4.9 5.6 8.6 13.7 9.8 1.9

59.1 17.0 5.0 7.0 3.5 2.8 7.1 14.8 7.1 1.7

55.1 15.4 5.9 5.1 2.5 4.9 4.8 20.9 7.4 2.5

56.0 16.8 5.3 8.3 3.9 5.0 7.5 14.1 9.2 2.5

Poverty and the Policy Response to the Economic Crisis in Liberia

3. Benefit Incidence of Public Spending for Education In this section, we provide an analysis of the benefit incidence of public spending for education. The key data are provided in tables 3.12a (for all students) and 12b (for students in public schools) and visualized in Figure 3.1 in the case of public schools. Tables 3.12a and 3.12b provide estimates of the number of children from households belonging to various deciles of per equivalent adult consumption who are attending various levels of schooling. In the case of table 3.12b devoted to public government schools only, under the simplifying assumption that the unit costs of enrollment are similar for all students attending a given cycle, the estimates of the number of students enrolled provide us with shares of total spending per cycle that are allocated to the various deciles. It can be seen that at the primary, junior secondary, and senior secondary levels, public spending for education seems to be allocated actually more to the poor than to other deciles, while at the post-secondary level most of the public spending goes to students who belong to the wealthier segments of the population. The fact that public spending for education appears to be pro-poor is a somewhat surprising finding, but it is again related to the fact that in Liberia a large share of students are enrolled in private as opposed to public schools, with the poor more likely to use public schools than better off households. Figure 3.1 is simply a representation in terms of concentration curves of the data provided in table 3.12b regarding the shares of public spending estimated to benefit various deciles. Note that the columns “total” in table 3.12a and 3.12b are not weighted by the shares of public spending allocated to the various levels of spending.

Figure 3.1: Concentration curve of enrollment in public schools, 2007 100 90 Primary

80 70

Secondary 2

Percent

60 50

Post-secondary

Secondary 1

40 30 20

Equity

Primary

Secondary 1

Secondary 2

Post-secondary 0

5 Decile

Source: Authors’ estimates based on 2007 CWIQ survey.

Overall public school 8

15,994

22,826

58,397

59,846

65,185

62,712

56,427

55,793

61,652

50,888

58,322

581,428

Total

21,893

4,933

2,649

3,435

1,859

1,162

2,482

1,028

1,156

884

2,305

Secondary 2

Overall students

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

174,487

20,306

20,586

16,981

14,824

16,649

15,209

14,786

16,326

52,205

Secondary 1

Primary

Deciles

30,134

8,016

4,783

5,445

4,176

1,531

3,129

907

816

1,113

216

Post-secondary

807,942

94,097

74,314

90,838

82,414

76,101

83,148

83,768

77,027

75,181

71,053

Total

100.0

10.0

8.8

10.6

9.6

9.7

10.8

11.2

10.3

10.0

9.0

Primary

Table 3.12a: Distribution of enrolled students by grade, all types of schooling, 2007

100.0

13.1

9.2

11.6

11.8

9.7

8.5

9.5

8.7

8.5

9.4

Secondary 1

100.0

22.5

12.1

15.7

8.5

5.3

11.3

4.7

5.3

4.0

10.5

Secondary 2

100.0

26.6

15.9

18.1

13.9

5.1

10.4

3.0

2.7

3.7

0.7

Post-secondary

100.0

11.6

9.2

11.2

10.2

9.4

10.3

10.4

9.5

9.3

8.8

Total

52 A World Bank Study

3,586 3,263 3,218 2,600 1,764 4,181 2,624 2,681 2,085 3,641 29,643

12.1 11.0 10.9 8.8 5.9 14.1 8.9 9.0 7.0 12.3 100.0

16,455 15,052 16,780 18,073 18,774 11,459 14,064 12,911 12,860 8,779 145,207

11.3 10.4 11.6 12.4 12.9 7.9 9.7 8.9 8.9 6.0 100.0

1 2 3 4 5 6 7 8 9 10 Total Â 1 2 3 4 5 6 7 8 9 10 Total

32.9 8.7 0.0 7.7 11.7 2.2 0.3 5.7 13.9 16.9 100.0

981 260 0 228 349 65 8 170 416 505 2,982

Sec. S

1.0 7.4 5.5 0.0 14.1 0.0 3.6 5.5 17.0 46.1 100.0

37 286 211 0 543 0 137 210 654 1,779 3,857

Post-Sec.

Total

11.6 10.4 11.1 11.5 11.8 8.6 9.3 8.8 8.8 8.1 100.0

21,059 18,861 20,209 20,902 21,429 15,705 16,834 15,972 16,014 14,705 181,688

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

Sec. J

Primary

Deciles

Girls in public schools

12.7 14.0 13.3 12.6 12.7 7.3 7.4 7.9 6.4 5.7 100.0

23,423 25,891 24,521 23,376 23,392 13,421 13,736 14,665 11,810 10,585 184,819

Primary

Sec. S

Post-Sec.

Number of students 7,576 929 0 5,825 253 523 8,673 973 124 7,169 183 71 5,153 375 1,309 4,750 210 623 4,960 311 425 2,791 1,063 2,546 4,031 276 1,970 3,871 794 2,437 54,800 5,368 10,028 Shares (%) 13.8 17.3 0.0 10.6 4.7 5.2 15.8 18.1 1.2 13.1 3.4 0.7 9.4 7.0 13.1 8.7 3.9 6.2 9.1 5.8 4.2 5.1 19.8 25.4 7.4 5.1 19.6 7.1 14.8 24.3 100.0 100.0 100.0

Sec. J

Boys in public schools

Table 3.12b: Distribution of enrolled students by grade, public schools only, 2007

12.5 12.7 13.4 12.1 11.9 7.5 7.6 8.3 7.1 6.9 100.0

31,927 32,493 34,291 30,798 30,230 19,003 19,432 21,065 18,088 17,687 255,016

Total

12.1 12.4 12.5 12.6 12.8 7.5 8.4 8.4 7.5 5.9 100.0

39,878 40,943 41,301 41,449 42,166 24,879 27,800 27,576 24,670 19,364 330,026

Primary

13.2 10.8 14.1 11.6 8.2 10.6 9.0 6.5 7.2 8.9 100.0

11,161 9,089 11,891 9,769 6,917 8,931 7,585 5,472 6,116 7,512 84,443 22.9 6.1 11.7 4.9 8.7 3.3 3.8 14.8 8.3 15.6 100.0

1,910 513 973 411 724 275 319 1,233 692 1,299 8,350

Sec. S

0.3 5.8 2.4 0.5 13.3 4.5 4.0 19.8 18.9 30.4 100.0

37 808 334 71 1,852 623 562 2,756 2,624 4,217 13,885

Post-Sec.

Overall in public schools Sec. J

12.1 11.8 12.5 11.8 11.8 7.9 8.3 8.5 7.8 7.4 100.0

52,986 51,353 54,500 51,700 51,659 34,708 36,266 37,037 34,102 32,392 436,704

Total

Poverty and the Policy Response to the Economic Crisis in Liberia 53

A World Bank Study

4. Correlates of School Enrollment In this last section, we look at the correlates or determinants of school enrollment using standard (probit) regression techniques. The analysis is conducted for children aged 6 to 14, for the sample as a whole and for boys and girls separately. By choosing this age bracket, we are implicitly focusing on primary school enrollment, instead of secondary school enrollment. The dependent variable is whether the child is enrolled in school or not. The explanatory variables include the following: (a) characteristics of the child—the age of the child and the age squared, the sex of the child, whether the child is the son or daughter of the household head, whether the child’s father is alive, whether the father lives in the household, and whether the child’s mother is alive and whether she lives in the household; (b) geographic location variables, including urban versus rural areas and a set of dummy variables for various regions (Greater Monrovia, North Central, North Western, South Central, South Eastern A, and South Eastern B); (c) household demographic variables—the number of children aged 0 to 5 (and its square), the number of children aged 6 to 14 (and its square), the number of male adults aged 15 to 60 (and its square), the number of female adults aged 15 to 60 (and its square), the number of seniors aged over 60 (and its square), the age of the household head (and its square), whether the household head is male or female, whether the head has a spouse or not, and the marital status of the head (single or never married, monogamous, polygamous, widowed, divorced or separated); (d) the education level of the head (none, some primary, primary completed, some secondary, secondary completed, post-secondary) and the same variables for the spouse of the head if there is one; (e) the socioeconomic group of the head of household (employment in the public, private formal or private informal sector, self-employment in agriculture or another sector, or inactivity and unemployment, whether the head has a second job), (f) a set of other household variables (the total acres of cultivable land owned, whether the household has migrated due to the war and has been displaced, and whether the household has returned to its place of origin or never moved); and finally (g) a set of variables indicating access to schools (time to nearest primary and secondary schools). The estimations are done for the sample as a whole, as well as separately for boys and girls, and for urban and rural areas. Only the coefficient estimates are provided to save space, with indication as to their level of statistical significance. The results from the estimations are mostly as expected. First, there is an inverted U relationship between the age of the child and the probability of going to school. When running the regression on the sample as a whole, there is no statistically significant difference in enrollment rates between boys and girls (remember though that we are looking here implicitly at primary school enrollment, and that differences between sexes are larger at the secondary school level, as shown in table 3.1). When the father is alive, a child is more likely to go to school (increase in probability of enrollment of 4.1 percent at the national level for the joint sample, but this increase comes from an impact that is statistically significant in rural areas only; said differently, orphans are less likely to be enrolled in school in rural areas). Surprisingly, there is a negative association between living in urban areas and going to school, but this is partly offset by the positive impact of being in the greater Monrovia area. Overall, the geographic location effects, when they are present, are of a limited order of magnitude (three to five percentage point difference in enrollment). Many of the demographic variables for the composition of the household are not significant, although having a higher number of male adults in the household does seem in some cases to improve the likelihood for the child to be enrolled. By contrast, having a female household head leads to an increase in school enrollment.

Characteristic of the child Age Age, squared Girl Son/daughter of head Father is alive Father lives in household Mother is alive Mother lives in household Residence area Urban Rural Region Greater Monrovia North Central North Western South Central South Eastern A South Eastern B Household composition Children aged 0 to 5 Children aged 0 to 5, squared Children aged 6 to 14 Children aged 6 to 14, squared Male adults aged 15 to 60 Male adults aged 15 to 60, squared Female adults aged 15 to 59 Female adults aged 15 to 59, squared Seniors aged over 60 Seniors aged over 60, squared Age of the household head Age of the household head, squared Female household head Head has no spouse Marital status of the head Single or never married Monogamous Polygamous Widowed, divorced or separated 0.151*** −0.009 0.000 0.043 −0.033 0.008 0.023 −0.008 — — — — — — — — 0.020 −0.003 −0.011 0.002 0.059** −0.006 0.050** −0.009** −0.075 0.097** 0.006 0.000 0.105*** −0.036 0.004 Ref. −0.071** −0.039

−0.040*** Ref.

0.038** −0.034*** 0.003 −0.043*** Ref. −0.001

0.010 0.000 −0.003 0.000 0.025** −0.004 0.016 −0.005** 0.019 −0.009 −0.003 0.000 0.042*** −0.017

0.022 Ref. −0.020 −0.017 −0.000*

0.000 Ref.

0.000 0.000 0.000 0.000 0.001** −0.000** 0.000 0.000 −0.001 0.001** 0.000 0.000 0.001 0.002*

— 0.000 0.005*** −0.001*** Ref. 0.002**

— —

0.004*** −0.000*** 0.000 −0.004*** −0.005* 0.000 0.000 0.001**

Boys and Girls (6−14 years old) Monrovia Other urban

0.136*** −0.008*** −0.003 0.017 0.041** −0.007 0.023 0.003

National

Table 3.13: Determinants of school enrollment, 2007

0.031 Ref. −0.013 0.011

0.016 0.000 −0.001 0.000 0.016 −0.004 −0.001 −0.001 0.022 −0.020 −0.003 0.000 0.028 −0.028

— −0.044*** −0.016 −0.044*** Ref. −0.011

— —

0.112*** −0.005* −0.003 0.026 0.064*** −0.019 0.017 −0.002

Rural

0.006 Ref. −0.005 −0.001

0.019 −0.002 0.002 0.000 0.028* −0.004 0.023 −0.005** 0.029 −0.002 −0.001 0.000 0.048** −0.022

0.017 −0.030* 0.024 −0.042*** Ref. −0.012

−0.031** Ref.

0.006 −0.027 0.001

0.000 0.020 −0.004 0.042 0.011

0.002 Ref. −0.005 0.011

0.001 −0.001 −0.003 0.001* 0.008 −0.001 0.015*** −0.003*** 0.003 0.011 0.002 −0.000* 0.005 −0.011*

— — — — — —

— —

−0.001

0.024*** −0.002*

−0.000**

0.000 Ref.

−0.000** 0.000** 0.000 0.000

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

— 0.000 0.000*** −0.000** Ref. 0.000***

— —

0.000 −0.246*** 0.000 0.000 0.000**

0.000*** −0.000***

Boys (6−14 years old) Monrovia Other urban

0.129*** −0.007**

National

0.015 Ref. 0.003 0.037

0.028 −0.002 0.000 0.000 0.025* −0.004 0.002 −0.002 0.025 −0.007 0.000 0.000 0.044 −0.024

— −0.042** 0.011 −0.049*** Ref. −0.026

— —

−0.004 0.066*** −0.025 0.048 0.014

0.106*** −0.003

Rural

0.044* Ref. −0.037** −0.036

−0.011 0.004 −0.004 0.000 0.018 −0.004 0.004 −0.003 0.019 −0.027 −0.006** 0.000*** 0.035* −0.020

0.063** −0.041*** −0.019 −0.040** Ref. 0.011

−0.049*** Ref.

0.038* 0.047** −0.022 0.001 −0.005

0.142*** −0.009***

National

0.000

0.000 Ref.

−0.000* 0.000* 0.000 −0.000* 0.000 0.000 0.000 0.000 0.000 0.000* 0.000 0.000 0.000*** 0.000

— −0.000* −0.000* −0.000* Ref. −0.000**

— —

0.000

−0.000* 0.000 0.000**

0.000* −0.000*

0.042 Ref. −0.026 −0.024

−0.005 0.003 −0.002 0.000 −0.003 −0.002 −0.005 −0.001 0.018 −0.038** −0.008*** 0.000*** 0.020 −0.033

— −0.041*** −0.031** −0.033** Ref. 0.006

— —

0.055*** 0.049*** −0.025 −0.026 −0.020

0.117*** −0.007**

Rural

(Table continues on next page)

0.011 Ref. −0.102** −0.085

0.011 0.002 −0.006 −0.002 0.116** −0.013 −0.006 −0.003 −0.139 0.124* 0.008 0.000 0.237*** −0.054

— — — — — —

— —

0.065 0.022 −0.045 −0.075 −0.025

0.196*** −0.012

Girls (6−14 years old) Monrovia Other urban

Poverty and the Policy Response to the Economic Crisis in Liberia 55

Ref. 0.006 0.096 0.046 0.104*** 0.190*** Ref. 0.010 0.015 0.015 0.004 0.086 0.634*** 0.546** 0.610** Ref. 0.415** 0.460* 0.376* −0.052 0.011*** −0.066*** 0.002 Ref. 0.651 −2.134*** 707

Ref. 0.012 0.036 0.060*** 0.054** 0.099**

0.020 0.008 −0.011 Ref. −0.004 0.002 −0.014

0.005 0.000

0.012 0.019* Ref.

−0.792*** −0.001 3,914

−0.018* 0.001 507

0.008** 0.000 Ref.

0.005** 0.000***

0.000 0.000 0.001 Ref. 0.002* 0.002 0.000

Ref. 0.002 0.008* 0.002 0.001 0.001

Ref. 0.000 −0.001** 0.000 0.001 0.001

−0.843*** 0.003 2,686

0.055* 0.030*** Ref.

0.016 −0.001

0.003 0.024 −0.033 Ref. −0.012 0.009 −0.004

Ref. 0.016 0.045 0.093*** 0.027 0.088

Ref. 0.039** 0.001 0.016 0.057*** 0.067**

Boys and Girls (6–14 years old) Monrovia Other urban Rural

Ref. 0.034** −0.002 0.021* 0.056*** 0.082***

National

Source: Authors’ estimates based on 2007 CWIQ survey. Note: * significant at 10%; ** significant at 5%; *** significant at 1%.

Education level of head None Some primary Completed primary Some secondary Completed secondary Post secondary Education level of Spouse None Some primary Completed primary Some secondary Completed secondary Post secondary Socioeconomic group of head of household Public Private formal Private informal Self-agriculture Self-other Unemployed Inactive, other Other household variables Head has a second job Total acres of cultivable land owned Migration status due to the war Displaced Displaced and has returned to place of origin Never moved Infrastructures accessibility Time to primary school (in 1,000 minutes) Time to secondary school (in 1,000 minutes) Observations

Table 3.13 (continued)

−0.732*** 0.007 1,942

−0.007 0.006 Ref.

0.003 0.000

0.020 0.004 −0.035* Ref. −0.001 0.013 0.001

Ref. 0.003 −0.029 0.048** 0.055 0.056

Ref. 0.058** 0.011 0.033** 0.077*** 0.107***

National

0.127 −0.361*** 346

−0.007*** 0.003 Ref.

−0.005 0.001***

1.000*** 1.000*** 1.000*** Ref. 1.000*** 1.000*** 1.000***

Ref. 0.016 −0.004 0.016* −0.003 0.015

Ref. −0.006* −0.005 0.005 0.014* 0.034**

0.000 0.000 263

0.000*** 0.000 Ref.

0.000** 0.000**

0.000* 0.000 0.001** Ref. 0.000*** 0.003*** 0.000**

Ref. 0.000 0.000 0.000 0.000 0.000

Ref. 0.000 −0.000** 0.000 0.000*** 0.000***

Boys (6–14 years old) Monrovia Other urban

−0.884*** 0.014 1,323

0.037 0.011 Ref.

0.014 −0.001

0.004 0.046 −0.053** Ref. −0.024 0.017 0.010

Ref. 0.005 −0.025 0.059* 0.020 0.079

Ref. 0.071** 0.033 0.040** 0.109*** 0.085*

Rural

−0.746*** −0.004 1,972

0.044 0.033*** Ref.

0.009 0.000

0.008 0.003 0.028 Ref. −0.012 −0.016 −0.027

Ref. 0.021 0.143*** 0.056** 0.047 0.159**

Ref. 0.011 −0.019 0.010 0.041** 0.067**

National

1.427 −2.192** 360

−0.097** −0.035 Ref.

−0.079 0.013**

0.358 0.327 0.330 Ref. 0.110 0.147 0.093

Ref. −0.013 0.355* −0.038 −0.017 0.116

Ref. 0.048 0.372** 0.064 0.162** 0.259**

−0.000** 0.000* 229

0.000* 0.000* Ref.

0.000 0.000

0.000 0.000 0.000 Ref. 0.000 0.000 −0.000**

Ref. 0.000* 0.000* 0.000* 0.000

0.000 0.000** 0.000

Ref. −0.000*

Girls (6–14 years old) Monrovia Other urban

−0.680*** 0.000 1,363

0.074 0.044*** Ref.

0.017 −0.001

−0.007 0.011 0.012 Ref. 0.002 0.004 −0.015

Ref. 0.029 0.138** 0.094*** 0.020 0.164

Ref. 0.016 −0.019 0.003 0.030 0.061*

Rural

56 A World Bank Study

Poverty and the Policy Response to the Economic Crisis in Liberia

The marital status of the head does not affect school enrollment, but the headâ&#x20AC;&#x2122;s (and to a lower extent the spouseâ&#x20AC;&#x2122;s) education do, with large impacts as of the secondary school completed. The impact of the socioeconomic group of the head is mild, except in the Greater Monrovia area, where children in households where the head is involved in wage work (whether in the public, private formal, or private informal sector) have a much higher probability of going to school than otherwise. Land ownership is also associated with higher enrollment, but only in urban areas where such ownership is a clear indication of wealth. There is also some indication that if a household has been displaced and lives in the Greater Monrovia area, the probability of school enrollment for the children is lower, while it is higher if the household has been displaced but has now returned to its place of origin in rural areas. Finally, as expected, the longer it takes to go to the nearest school, the less likely a child is to go to school, at least in rural areas. These regressions provide some useful insights into the determinants of schooling. For policy purposes, the main use of the regressions lies in assessing the potential impact of the construction of new schools on enrollment. For example, the coefficient for the pooled sample of boys and girls of the distance to primary schools is -0.843 in rural areas. Given that the explanatory variable is expressed in 1,000 minutes, this means that a 10 minutes reduction in the time to go to school would increase school enrollment by about 0.8 percentage point. It was mentioned that in rural areas, the average time needed to reach the nearest primary school was approximately 46.5 minutes. If this distance were cut in half, to 26.3 minutes, we would obtain an increase in enrollment of about 2.2 percentage points. While this is not negligible, it is not as large as one is often led to believe, which suggests that policies to increase school enrollment further need to go beyond the simple provision of new schools, even if this is necessary of course.

5. Conclusion This chapter has provided a basic diagnostic of the education system in Liberia on the basis of the analysis of the 2007 CWIQ survey. Several findings show that Liberia stands out in comparison with other countries. First, there are large differences between net and gross enrollment rates due to the fact that many older children have returned to school since the end of the conflict. Second, non-government schools play a major role in the education of children, which is again in part a legacy of the conflict. Third, despite the elimination of school fees for primary education, costs remain an issue for many households, as it is the main reason for never having been enrolled for those students who never went to school. Distance is the second main reason for never having been enrolled. As to the main reason for not continuing oneâ&#x20AC;&#x2122;s education when a child has been enrolled in the past, it is related to the need to wait for admission, but cost shows up again as a significant barrier to further schooling among this group. Public spending for education appears to be pro-poor at the primary and secondary level, at least on the basis of simple statistics on enrollment rates by household consumption deciles. At the tertiary level, public spending does not reach the poor much. While overall public spending seems to be more pro-poor in Liberia than in other countries due in part to the fact that better off households rely heavily on private schools, satisfaction rates with the services received is low, especially in public schools, but also in many private schools. In public schools at the primary level, the main complaints are related to the lack of books or supplies, the fact that there are not enough teachers, the fact that

A World Bank Study

facilities are in poor condition, the long distance to schools, and poor teaching. In private schools, low satisfaction is due to high fees and the lack of books or supplies. The rates of satisfaction in secondary schools are of the same order of magnitude than at the primary level and the complaints are similar in both public and private schools. Satisfaction with tertiary education is also low, but there high fees are one of the main reasons for not being satisfied in both public and private schools. The fact that the quality of the education services is limited is not surprising given the fact that in terms of budget, the Ministry of Education does not have adequate resources to provide basic inputs such as desks, textbooks, and pencils and notebooks to students in public primary schools. Thanks to user fees, secondary schools have more resources, but this works to the detriment of the very poor who often lack resources to pay these fees. Today, a large share of the education costs are borne by NGOs and donors, but budgetary pressures on the government are expected to increase in future years both because the government will probably progressively be expected to take on a larger share of the costs of the education system, and because enrollment in secondary schools will increase rapidly once larger cohorts complete their primary education. Finally, the chapter has provided an analysis of the determinants of school enrollment. Many findings are as expected, with orphans less likely to enroll, and children from better families (as proxied among others by the education level of the father) more likely to enroll. One interesting result to inform policy is that the distances to primary and secondary schools have an impact on the probability to enroll, as expected, but even a substantial reduction in these distances that could be obtained through a program of building new schools would not lead to a dramatic increase in enrollment rates according to our estimations. This type of results underscores the complexity of designing a strategy for further progress in education in Liberia that is both ambitious, and affordable for the government and its partners.

Notes 1. The authors are with the World Bank. Inputs were provided by Rose Mungai and Rebecca Simson. The chapter was prepared as an input to Liberia’s Poverty Reduction Strategy. Key results were presented at a workshop organized by Liberia’s core PRSP team in Monrovia on December 10-11, 2007. The views expressed in this chapter are those of the authors and need not reflect those of the World Bank, its Executive Directors or the countries they represent. 2. We are grateful to Rebecca Simson for pointing this out to us.

References Backiny-Yetna, P., R. Mungai, C. Tsimpo, and Q. Wodon. 2012. “Poverty in Liberia: Level, Profile, and Determinants.” In Q. Wodon, ed., 9–34. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Heninger, L., C. Makinson, F. Richardson, M. Kaiser, and J. A. Duany. 2006. Help Us Help Ourselves: Education in the Conflict to Post-Conflict Transition in Liberia. Women’s Commission for Refugee Women and Children, New York. Humphreys, M., and P. Richards. 2005. “Prospects and Opportunities for Achieving the MDGs in Post-conflict Countries: A Case Study of Sierra Leone and Liberia.” Center on Globalization and Sustainable Development (CGSD) Working Paper No. 27. Earth Institute, Columbia University, New York.

Poverty and the Policy Response to the Economic Crisis in Liberia

International Labour Organization (ILO). 2009. “A Rapid Impact Assessment of the Global Economic Crisis on Liberia.” Mimeo, Monrovia. Ministry of Education. 2007. Liberian Primary Education Recovery Program Prepared for Fast Track Initiative. Mimeo, Monrovia. Ministry of Education and UNICEF. 2004. “Rapid Assessment of Learning Spaces.” Mimeo, Monrovia. Republic of Liberia, 2006, Interim Poverty Reduction Strategy: Breaking from the Past—From Conflict to Development, Monrovia. ———. 2008. Poverty Reduction Strategy. Monrovia. Richards, P., S. Archibald, B. Bruce, W. Modad, E. Mulbah, T. Varpilah, and J. Vincent. 2005. “Community Cohesion in Liberia: A Post-War Rapid Social Assessment.” Social Development Paper, Conflict Prevention and Reconstruction No. 21. World Bank, Washington, DC. United Nations Development Programme (UNDP), Liberia. 2006. National Human Development Report 2006—Mobilizing Capacity for Reconstruction and Development. Monrovia.

CHAPTER 4

Health in Liberia: Basic Diagnostic Using the 2007 CWIQ Survey Clarence Tsimpo and Quentin Wodon1 As for education, little has been written on the health system in Liberia since the start of the conflict in large part because of lack of good data. This chapter was also written in 2007 to inform the diagnostic of Liberia’s Poverty Reduction Strategy. It provides a diagnostic of Liberia’s health system as seen from the point of view of households using the new nationally representative Core Welfare Questionnaire Indicator survey implemented in 2007. The analysis covers rates of illness and injuries in the population, as well as the reasons for not seeking care, and the degree of satisfaction of households with the services received when they do seek care, in each case looking at various age groups and women and men separately, as well as at different types of facilities providing care. Data are also presented on household private spending for health, as well as on distances to facilities. A benefit incidence analysis of public spending for health is conducted, and regression analysis is used to assess the determinants of the demand for care.

1. Introduction Improving the access to, quality of, and affordability of health care is a key priority in a post-conflict and poor country such as Liberia. According to a recent presentation by Liberia’s Ministry of Health and Social Welfare (2007), the conflict has led to the destruction or poor maintenance of a number of health facilities. Out of 521 facilities, only 389 are functional, and among these, 300 are currently being supported by NGOs, some of which may reduce their support in a year or two (see for example Médecins sans Frontières, 2007). Many health facilities, even when they are operational, lack potable water supply, lighting, equipment, refrigeration, and emergency facilities. Public spending for health is very low, at $3.4 per person per year. The country currently has a total of 4,000 health workers, as compared to 13,000 recommended by the World Health Organization. There is a lack of capacity at the central and county levels to implement health policies and programs. Health indicators used for monitoring the Millennium development Goals such as infant and child malnutrition, infant and child mortality, and maternal mortality are low (see UNDP, 2006, and Humphreys and Richards, 2005, for a discussion related to the Millennium Development Goals in Liberia; and International Labour Organization, 2009, for a rapid impact assessment of the recent economic crisis). 60

Poverty and the Policy Response to the Economic Crisis in Liberia

The delivery of basic services is one of four key areas of emphasis in the country’s Interim Poverty Reduction Strategy (Republic of Liberia, 2006) adopted in 2006, and this was reaffirmed in the full poverty reduction strategy (Republic of Liberia, 2008). A national health plan has been approved for the period 2007–11, focusing in part on expanding the ability of providing a basic package of health care for at least 70 percent of the population by 2009. Immunization campaigns are implemented to boost vaccination rates for children. Renovations are being implemented to improve the quality of public hospitals. In order to monitor progress in the delivery of health services, it is important to have good data, among others for establishing a baseline. Although the recent completion of a Demographic and Health Survey has helped to fill many gaps, today there is still a lack of good information on many aspects of the health system and health outcomes in Liberia in part because of the limited data available. In order to help inform the preparation of Liberia’s full Poverty Reduction Strategy, the objective of this chapter is to provide a basic diagnostic of health services as seen from the point of view of users. The diagnostic is based on the newly available nationally representative CWIQ (Core Welfare Questionnaire Indicator) survey that was implemented in 2007 by the Liberia Institute of Statistics. The survey includes detailed data on the incidence of illnesses and sickness, the use of various types of health care facilities by households, as well as the reasons for not seeking care when sick and the degree of satisfaction of households with the services received. Data are also available on private spending for health, as well as on distances to health facilities. The chapter is structured as follows. Section 2 provides descriptive statistics on morbidity (incidence of illnesses and sickness), the likelihood for household to seek care and the types of facilities used, as well as the reasons for not seeking care and satisfaction with health services. Section 3 is devoted to a benefit incidence analysis of public spending for health, with a comparison with private spending. Section 4 discusses the determinants of the demand for care. A brief conclusion follows.

2. Patterns of Morbidity, Likelihood of Seeking Care, and Reason for Not Seeking Care 2.1. Patterns of Morbidity and Likelihood of Seeking Care

Table 4.1 and figure 4.1 provide measures of the share of the population that has been sick or injured during the last four weeks preceding the survey. The rates of morbidity are very high, with slightly more than half of the population reporting an incident. At the national level, 42.9 percent of the population declares having suffered from an illness, with the proportions being higher for women than for men, and higher in rural than in urban areas. Morbidity rates are apparently lower among poorer households identified here according to five quintiles of consumption per equivalent adult (for an analysis of poverty in Liberia based on the 2007 CWIQ survey, see Backiny-Yetna et al., 2012). The first quintile “Q1” represents thus the poorest 20 percent of the population, and the top quintile “Q5” the richest 20 percent. However, it is well known in the health literature that the poor tend to underreport episodes of sickness. The main illness cited is fever/malaria, which accounts for more than 60 percent of the episodes of illness. Next is pain in a person’s back, limbs, or joints, accounting for 15.8 percent of episodes. Diarrhea and abdominal pain accounts for 13.5 percent of epi-

A World Bank Study

Figure 4.1: Share of population sick or injured in last four weeks, 2007 80 70

Liberia

Male

Female

Percent

60 50 40 30 20

0-4

5-9

10-14

15-19

20-29 Age

30-39

40-49

50-59

60 and over

Source: Authors’ estimates based on 2007 CWIQ survey.

sodes, followed by cough and breathing difficulties, for 9.8 percent of episodes. Other symptoms each account for less than five percent of episodes (except for the “other” category, which accounts for 9.2 percent of all episodes). There are relatively few differences by gender, location, or quintile in the types of illnesses that people suffer from. The data by age group reveals as expected that infants and young children (below five years of age) and the elderly (above 60 years of age) are the most likely to be sick, followed by other children (between five and fifteen years of age) and then the adult population aged 15 to 59. This is clearly visible in figure 4.2 which plots the incidence of illnesses by age group and sex. In terms of types of illnesses, children of all ages are the most likely to be affected by malaria, while pain in the back, limbs or joints is much more frequent for the adult and elderly population. Figure 4.2: Share of sick/injured persons who have requested care, 2007 95 94 93

Percent

92 91 90 89 88 87

Liberia

86 85

0-4

5-9

Male 10-14

Female 15-19

20-29 Age

Source: Authors’ estimates based on 2007 CWIQ survey.

30-39

40-49

50-59

60 and over

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 4.1: Patterns of morbidity in last four weeks, 2007 Gender Male Female % of population that has been sick or injured Type of sickness/injury Fever/malaria Diarrhea/abdominal pains Pain in back, limbs or joints Cough/breathing difficulties Skin problems Ear, nose or throat Eye Dental Accident Other % of population that has been sick or injured Type of sickness/injury Fever/malaria Diarrhea/abdominal pains Pain in back, limbs or joints Cough/breathing difficulties Skin problems Ear, nose or throat Eye Dental Accident Other % of population that has been sick or injured Type of sickness/injury Fever/malaria Diarrhea/abdominal pains Pain in back, limbs or joints Cough/breathing difficulties Skin problems Ear, nose or throat Eye Dental Accident Other

40.5

45.3

Residence area Quintile Urban Rural Q1 Q2 Q3 Q4 Total population 36.2 45.9 39.6 42.0 45.0 42.2

60.3 12.9 15.4 9.5 4.2 1.4 2.3 1.0 2.1 9.3

62.4 14.0 16.3 10.1 3.9 1.6 2.5 1.2 0.5 9.0

64.7 11.0 12.7 9.0 3.3 1.2 2.4 1.5 0.9 12.2

60.3 14.3 17.0 10.1 4.3 1.6 2.4 1.0 1.4 8.1

51.0

52.3

50.2

52.1

71.3 12.2 1.5 15.1 4.3 1.0 0.5 0.0 0.2 7.6

76.8 11.9 3.8 15.1 4.9 1.2 1.6 0.2 0.3 6.0

72.3 11.5 2.2 15.4 4.5 0.9 0.7 0.1 0.7 10.0

74.7 12.3 2.8 15.0 4.6 1.2 1.1 0.1 0.2 5.8

37.3

38.6

30.8

41.1

67.3 11.5 5.0 11.5 4.9 1.9 1.5 1.3 1.7 7.1

73.9 11.0 4.7 12.9 6.1 1.5 1.9 1.0 0.4 4.3

74.0 10.4 3.5 10.0 5.0 0.8 1.0 1.0 1.1 5.5

69.3 11.6 5.3 12.9 5.7 2.0 1.9 1.2 1.0 5.8

62.4 55.1 13.1 15.9 16.8 19.1 11.2 8.9 4.2 4.4 1.7 1.3 2.6 2.8 1.1 1.2 0.7 1.3 6.6 9.3 Aged 0â&#x20AC;&#x201C;4 47.8 44.9

70.4 77.6 7.3 11.4 1.8 5.6 14.8 12.5 6.3 5.8 0.6 0.4 1.7 2.7 0.2 0.4 7.0 5.6 Aged 5â&#x20AC;&#x201C;14 33.8 37.1

76.8 9.9 3.5 13.4 4.8 2.0 2.7 1.3 2.8

65.5 10.6 6.9 9.8 6.8 1.0 1.5 2.0 0.9 7.5

Total

45.7

42.9

62.5 13.8 15.0 9.7 3.3 1.7 2.8 1.3 1.5 8.7

65.2 11.2 15.9 10.1 4.5 0.5 1.9 0.7 2.2 10.1

62.0 13.2 12.8 9.3 3.8 2.2 1.7 1.3 0.6 11.0

61.4 13.5 15.8 9.8 4.0 1.5 2.4 1.1 1.3 9.2

55.5

53.8

55.7

51.6

79.2 10.0 1.4 16.7 2.6 3.1 0.2

70.2 18.3 1.3 15.0 4.2 0.8 1.0

4.8

74.0 11.5 3.8 16.0 4.7 0.4 0.1 0.4 0.6 7.2

0.3 8.7

74.2 12.1 2.7 15.1 4.6 1.1 1.0 0.1 0.3 6.8

41.1

37.3

40.8

37.9

69.5 12.6 5.8 10.5 4.4 2.7 1.2 1.3 2.1 5.1

71.2 10.7 3.9 12.9 6.4 0.7 2.0 0.4 0.9 7.0

70.2 12.3 3.8 14.8 5.2 2.0 1.1 0.6 1.1 6.0

70.5 11.3 4.8 12.2 5.5 1.7 1.7 1.1 1.1 5.7

(Table continues on next page)

A World Bank Study

Table 4.1 (continued) Gender Male Female

Residence area Urban Rural

% of population that has been sick or injured Type of sickness/injury Fever/malaria Diarrhea/abdominal pains Pain in back, limbs or joints Cough/breathing difficulties Skin problems Ear, nose or throat Eye Dental Accident Other

37.3

44.7

34.6

44.4

56.3 13.4 22.5 5.9 4.0 1.6 1.7 1.1 2.8 10.2

56.0 16.2 21.4 7.6 2.9 1.7 2.2 1.7 0.6 10.4

60.9 11.3 17.0 6.1 2.5 1.5 2.1 2.1 0.9 14.3

54.3 16.4 23.7 7.1 3.7 1.7 2.0 1.2 1.9 8.7

66.6

75.8

38.6 15.6 36.3 12.9 2.8 0.1 12.2 1.3 3.1 15.4

40.4 10.9 45.5 9.9 1.9 1.9 9.4 0.5 0.4 22.2

Q1 Q2 Aged 15â&#x20AC;&#x201C;59 38.2 41.3

Quintile Q3

Total

42.6

39.7

43.8

41.1

57.6 15.5 19.8 6.6 3.2 1.0 2.6 1.8 1.7 10.7

61.3 12.2 22.2 7.2 4.0 0.4 2.0 0.7 3.0 11.3

57.9 12.3 18.3 5.7 3.0 2.7 1.4 2.2 0.5 12.3

56.1 15.0 21.9 6.8 3.4 1.6 2.0 1.5 1.6 10.3

66.6

57.3 46.3 16.0 19.1 24.1 25.7 8.4 6.6 3.5 3.2 1.9 2.0 1.7 2.4 1.3 1.2 1.1 1.8 6.5 10.2 Aged 60 and over 71.8 68.9 73.9

73.4

70.4

66.1

70.6

43.9 9.2 35.2 16.6 1.1 0.1 14.9 1.1 0.5 25.5

38.2 14.6 42.0 10.1 2.7 1.2 9.8 0.9 2.2 16.7

39.0 19.9 39.0 14.7 1.7

31.1 15.1 47.7 14.1 0.8

10.3 1.6 1.9 15.4

18.3

49.1 5.3 45.5 6.6 1.2 0.0 5.7 2.4 5.2 20.0

42.7 9.3 32.0 4.5 5.1 2.8 10.3

39.4 13.4 40.6 11.5 2.4 0.9 10.9 0.9 1.8 18.6

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

36.4 15.4 38.9 15.4 3.4 2.0 9.9 0.5 0.9 17.2

0.6 17.1

0.7 24.7

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 4.2: Demand for health care and type of provider, 2007 Gender Male Female % of persons who have consulted % of sick/injured persons who have consulted Type of health provider Government hospital Government health center Government health clinic Other public facility Private hospital/clinic Pharmacy Private doctor/dentist Mobile clinic/black bagger/ drug peddler Other private facility Traditional healer Total % of persons who have consulted % of sick/injured persons who have consulted Type of health provider Government hospital Government health center Government health clinic Other public facility Private hospital/clinic Pharmacy Private doctor/dentist Mobile clinic/black bagger/ drug peddler Other private facility Traditional healer Total

Residence area Quintile Urban Rural Q1 Q2 Q3 Total Population 34.2 43.1 35.8 39.5 41.7

Total

40.6

44.2

40.4

38.0

42.7

91.4

91.7

91.5

88.9

91.2

90.6

92.8

94.0

91.6

26.1 7.1 17.7 3.6 19.2 5.9 1.3 10.5

25.3 8.2 18.4 3.8 21.9 5.9 1.1 8.5

35.7 4.4 5.8 2.2 35.7 9.1 1.7 2.0

22.1 8.8 22.4 4.2 15.3 4.7 1.0 12.1

32.4 8.0 23.6 2.2 10.9 4.4 0.9 8.6

23.5 11.8 20.8 5.2 15.4 5.3 1.3 8.4

25.4 7.4 16.2 3.8 22.6 5.8 1.1 10.5

22.6 6.1 18.3 4.9 23.6 5.9 0.8 10.4

25.4 5.4 12.7 2.5 28.5 7.6 1.7 9.1

25.7 7.7 18.1 3.7 20.6 5.9 1.2 9.4

1.6 7.1 100.0

0.9 6.0 100.0

1.3 2.0 100.0

1.2 8.1 100.0

0.8 6.5 100.0

2.5 4.9 100.0

1.2 5.7 100.0

1.2 6.5 100.0

49.1

48.6

47.5

49.3

0.7 0.8 8.2 7.5 100.0 100.0 Aged 0â&#x20AC;&#x201C;4 42.6 43.6

51.2

52.5

53.5

48.9

93.0

91.7

92.6

92.2

87.9

92.8

90.7

94.4

94.7

92.3

27.5 4.9 20.6 4.7 19.3 4.7 1.3 11.1

22.8 8.3 19.6 5.1 22.3 7.9 0.4 10.5

38.1 2.0 6.9 3.3 37.1 6.1 0.8 2.8

21.1 8.1 24.1 5.4 15.8 6.4 0.9 13.3

36.4 5.7 21.3 1.5 9.8 3.0 1.8 11.9

21.6 12.9 22.8 7.8 15.1 6.1 0.0 9.4

20.5 7.1 24.4 4.5 22.5 6.0 0.6 11.6

18.4 5.1 21.0 6.4 23.3 8.3 0.7 11.6

30.0 3.8 12.9 4.2 28.6 7.2 1.1 9.8

25.1 6.7 20.1 4.9 20.8 6.3 0.8 10.8

2.0 3.9 100.0

1.0 2.1 100.0

1.4 1.5 100.0

1.5 3.4 100.0

0.5 8.0 100.0

1.6 2.7 100.0

0.5 2.3 100.0

3.1 2.1 100.0

1.3 1.2 100.0

1.5 3.0 100.0

(Table continues on next page)

A World Bank Study

Table 4.2 (continued)

% of persons who have consulted % of sick/injured persons who have consulted Type of health provider Government hospital Government health center Government health clinic Other public facility Private hospital/clinic Pharmacy Private doctor/dentist Mobile clinic/black bagger/ drug peddler Other private facility Traditional healer Total % of persons who have consulted % of sick/injured persons who have consulted Type of health provider Government hospital Government health center Government health clinic Other public facility Private hospital/clinic Pharmacy Private doctor/dentist Mobile clinic/black bagger/ drug peddler Other private facility Traditional healer Total

Gender Male Female

Residence area Urban Rural

Quintile Q3

Total

35.0

36.2

28.8

38.6

38.3

36.2

38.7

35.6

92.8

91.8

93.0

92.1

91.2

89.6

91.9

94.8

94.2

92.3

24.9 6.1 17.5 3.6 20.2 6.6 0.9 11.9

24.2 9.7 17.7 3.1 21.4 7.3 1.0 8.8

33.1 4.5 5.9 1.4 36.3 11.0 1.1 3.1

21.7 8.9 21.5 4.0 15.6 5.6 0.9 12.8

29.3 10.3 26.7 4.0 8.5 7.1 0.9 6.8

23.6 10.7 17.5 5.2 16.6 3.3 0.9 12.9

22.6 7.7 15.2 2.5 25.8 6.5 1.0 11.6

24.9 4.9 14.5 4.7 22.0 8.3 1.1 12.4

23.0 5.9 15.1 0.5 29.6 9.4 0.7 7.6

24.6 7.8 17.6 3.4 20.8 6.9 0.9 10.4

2.0 6.3 100.0

0.6 6.3 100.0

1.8 1.7 100.0

1.1 7.8 100.0

0.9 1.3 5.5 7.8 100.0 100.0 Aged 15â&#x20AC;&#x201C;59 34.0 39.0

0.1 7.0 100.0

3.0 4.2 100.0

1.1 6.9 100.0

1.3 6.3 100.0

34.6

42.7

33.0

41.7

39.7

38.2

42.8

38.8

89.8

92.1

91.1

87.3

91.8

90.4

91.9

93.5

91.1

26.0 8.1 16.7 3.3 19.1 6.5 1.2 9.8

25.9 7.7 18.0 3.9 22.5 5.1 1.3 7.8

35.6 5.0 5.6 2.4 35.2 9.4 2.3 1.4

22.2 9.0 22.1 4.1 15.5 4.2 0.9 11.5

32.1 7.6 22.9 1.8 12.4 3.7 0.7 8.9

24.4 12.3 21.0 4.6 15.0 6.6 1.1 6.0

27.4 7.1 15.1 4.2 21.1 5.6 1.1 9.5

22.2 6.7 18.8 5.0 25.9 4.5 0.6 9.1

24.5 6.0 11.2 2.6 28.4 7.5 2.5 9.5

26.0 7.9 17.5 3.7 21.0 5.7 1.3 8.7

1.2 8.0 100.0

1.0 6.9 100.0

0.8 2.3 100.0

1.2 9.3 100.0

0.8 8.9 100.0

0.4 8.4 100.0

1.2 7.7 100.0

1.8 5.4 100.0

1.2 6.6 100.0

1.1 7.3 100.0

Q1 Q2 Aged 5â&#x20AC;&#x201C;14 31.2 34.1

(Table continues on next page)

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 4.2 (continued) Gender Male Female % of persons who have consulted % of sick/injured persons who have consulted Type of health provider Government hospital Government health center Government health clinic Other public facility Private hospital/clinic Pharmacy Private doctor/dentist Mobile clinic/black bagger/ drug peddler Other private facility Traditional healer Total

62.8

67.1

Residence area Quintile Urban Rural Q1 Q2 Q3 Q4 Aged 60 and over 59.8 66.1 64.5 67.9 63.6 62.2

93.2

87.7

89.7

90.9

93.7

89.9

86.3

87.4

95.9

90.6

27.7 8.8 17.8 2.4 15.9 2.5 2.5 9.0

31.1 6.5 21.5 2.1 17.0 3.0 2.4 8.3

42.1 4.3 5.0 34.1 6.1 1.7 1.2

25.8 8.7 23.4 2.9 11.7 1.8 2.6 10.7

36.9 7.1 22.2 0.4 11.5 2.6 6.8

20.4 9.2 26.0 4.0 14.3 2.1 6.6 7.6

33.1 9.7 7.8 3.5 21.2 3.2 3.1 11.4

28.8 9.2 21.4 0.4 11.9 1.8 0.9 8.9

27.7 3.5 17.4 3.0 25.0 3.9 1.0 9.5

29.3 7.8 19.5 2.3 16.4 2.7 2.4 8.7

2.2 11.2 100.0

0.5 7.5 100.0

3.3 2.3 100.0

1.0 11.4 100.0

12.5 100.0

0.3 9.4 100.0

1.1 5.9 100.0

5.0 11.5 100.0

1.6 7.4 100.0

1.5 9.5 100.0

Total

64.6

64.7

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

Table 4.2 and figure 4.2 provide data on the demand for care among individuals who have been sick or injured (as well as among the population as a whole). A few important findings emerge. First, a surprisingly high proportion of individuals who were sick did seek care (we will discuss below the reasons for not seeking care). For the sample as a whole, this proportion is 91.6 percent. Second, individuals from richer households are more likely to seek care, as expected. Differences are by contrast negligible between men and women, and between urban and rural areas. The main types of facilities consulted are government hospitals (25.7 percent of the consultations), private hospitals and clinics (20.6 percent), government health clinics (18.1 percent), mobile clinics, black baggers and drug peddlers (9.4 percent), government health centers (7.7 percent) and traditional healers (6.5 percent). As is the case for education, and as expected, private facilities tend to be used comparatively more by richer and urban households, while public facilities are used comparatively more by poorer and rural households. As for education, due in part to the inability of the state to provide services during the civil war, NGOs play today a very important role in Liberiaâ&#x20AC;&#x2122;s health system, but this is not as apparent in the CWIQ survey data as was the case for education, probably because households assimilate NGO-run centers to government facilities (that is, many among the NGOs may well operate public facilities). Data by age group on the demand for care and the type of facilities used are provided for the sake of completeness, but the patterns are very similar across all age groups. One key difference is the fact that older individuals are slightly less likely to seek care than younger individuals, probably because illnesses for small children are potentially more life threatening. The CWIQ survey also has an interesting question on measures taken by households to prevent malaria. The answers are provided in table 4.3. Some 41.7 percent of the popu-

A World Bank Study

Table 4.3: Measures taken by the household to prevent malaria, 2007 Residence area None

Quintile

Urban

Rural

Total

30.4

46.8

52.4

51.1

44.0

35.0

31.6

41.7

Bed net

37.3

32.8

28.6

30.3

29.8

40.1

39.2

34.2

Insecticide

21.4

3.5

2.8

3.5

9.1

10.7

16.1

9.1

Anti-malaria drug

14.6

9.4

10.2

8.1

8.5

12.5

14.3

11.1

0.5

0.1

0.6

0.2

Fumigation Insecticide treated net

4.4

3.8

2.4

7.2

4.7

3.3

4.0

Maintain good drainage

2.0

2.6

4.7

4.0

1.3

1.1

1.6

2.4 10.1

Maintain good sanitation

8.1

11.0

7.8

9.1

9.7

11.4

11.5

Herbs

0.9

4.9

2.1

2.4

4.3

6.4

2.9

3.6

Burn leaf (tobacco, etc.)

1.4

1.8

3.5

1.5

0.6

1.9

1.3

1.7

Window/door net Other

11.6

2.8

5.3

3.7

3.4

6.7

8.0

5.6

1.5

2.4

1.8

2.0

1.7

2.9

2.1

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

lation does not take any measures, and the proportion is above 50 percent among the bottom two quintiles. Bed nets are the most common preventive measure, for a third of the sample, but the likelihood that they will be used by the poor is lower. Anti-malaria drugs are used by 11.1 percent of the population, with some differences across quintiles. Measures to maintain good sanitation are taken by a tenth of the population as well, again with even more limited differences between quintiles. The use of insecticides is of a similar order of magnitude at the national level, but only three percent of the population in the bottom two quintiles uses them, as this is a strategy mostly used in urban areas. Overall, it is clear from the data that additional efforts could be made to help the population protect itself from malaria, which is the first cause of illness in the country. 2.2. Mode of Payment for Care and Reasons for Not Seeking Care

The high share of sick individuals who seek care is probably due in part to the fact that in many instances, health care appears to be free in Liberia. Table 4.4 provides data on the modes of payment for care. In 43.4 percent of consultations, health care is free, while in 54.8 percent of cases, households do pay for care. Very few have insurance, or enjoy benefits so that their employer pays for care. In turn, the high proportion of visits that are free may be related to the important role played by NGOs in the administration of care. The use of free care services is higher among the poor than among better off households, but differences between urban and rural areas and between sexes are negligible. Similarly, the types of modes of payment are similar for the different age groups. Even though health care can be obtained for free in many instances, and even though most individuals do seek care, cost may still be a barrier for care for some households. Table 4.5 provides the reasons invoked by individuals for not seeking care when sick (data are also provided for information for the population as a whole, but the responses

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 4.4: Payment method for the consultation, 2007 Gender Male

Residence area

Female

Urban

Quintile

Rural

Total

Total population Free

43.2

43.5

40.4

44.4

51.6

46.3

43.7

40.9

36.2

43.4

Self/household paid

55.1

54.4

58.2

53.6

45.4

52.5

54.7

57.5

61.9

54.8

Employer

0.4

0.5

0.6

0.4

0.5

0.3

0.1

0.4

0.8

0.4

Insurance

0.6

0.2

0.8

0.3

0.6

0.7

0.8

0.6

Other

0.7

0.9

0.6

0.9

2.2

0.3

0.7

0.4

0.6

0.8

Total

100.0

Free

45.1

44.0

44.7

44.5

56.1

48.6

45.1

40.2

37.3

44.6

Self/household paid

52.8

54.3

53.3

39.6

50.8

52.6

59.2

60.6

53.5

0.5

0.1

0.2

0.5

0.2

0.9

0.5

1.4

0.3

0.7

Aged 0–4

Employer

0.3

0.0

Insurance

0.8

0.5

Other

1.0

1.1

0.4

1.3

3.6

0.2

0.8

0.3

0.9

1.1

Total

100.0

Free

43.6

43.3

40.5

44.5

53.3

49.5

41.5

38.0

36.1

43.5

Self/household paid

55.5

54.7

58.9

53.9

44.3

49.4

56.6

61.0

63.3

55.1

0.0

0.1

0.0 0.7

Aged 5–14

Employer

0.1

Insurance

0.5

0.8

0.9

0.2

0.9

1.0

0.8

0.3

Other

0.4

1.1

0.6

0.7

2.1

0.2

0.9

0.2

0.3

0.7

Total

100.0

Free

41.4

43.5

39.0

44.0

48.9

44.2

43.6

42.3

36.0

42.6

Self/household paid

Aged 15–59 56.6

54.6

59.4

53.9

48.3

54.4

55.4

55.8

61.7

55.5

Employer

0.7

0.8

0.9

0.7

0.8

0.7

0.3

0.8

1.2

0.8

Insurance

0.7

0.5

0.2

0.7

0.4

0.6

0.4

1.0

0.5

0.6

Other

0.7

0.5

0.6

1.7

0.2

0.3

0.2

0.6

100.0

Total

Aged 60 and over Free

48.3

43.3

43.7

46.7

54.2

45.6

50.9

43.2

34.6

46.1

Self/household paid

50.3

52.1

53.4

50.5

42.3

53.5

46.1

52.7

62.5

51.1

0.1

0.4

Employer

0.2

Insurance

0.2

1.4

0.3

0.8

1.0

2.2

0.7

Other

1.0

3.2

2.6

1.8

3.1

0.9

2.2

3.1

0.7

2.0

100.0

Total

Source: Authors’ estimates based on 2007 CWIQ survey.

0.1

A World Bank Study

Table 4.5: Reason for not seeking medical care, 2007 Gender Male

Female

Residence area Urban

Quintile

Rural

Total

Overall population No need

92.3

91.0

92.8

91.1

91.6

90.7

91.3

93.8

91.7

3.3

3.9

4.5

3.1

3.3

3.0

4.0

4.5

3.2

3.6

Too expensive Too far

2.8

3.6

0.3

4.7

3.6

3.7

2.6

4.1

1.8

3.2

Lack of confidence

0.6

0.7

1.0

0.5

0.4

0.5

1.0

0.9

0.6

0.7

Other

2.0

2.5

2.1

2.3

2.6

2.2

2.8

1.7

1.8

2.2

Sick and/or injured population No need

13.3

8.9

9.6

11.5

17.1

6.6

8.0

5.8

17.8

11.0

Too expensive

39.2

41.5

65.6

31.6

32.4

40.5

40.7

55.5

35.9

40.4

Too far

31.8

36.9

2.9

45.4

34.3

38.7

28.2

50.3

21.1

34.5

Lack of confidence Other

3.6

5.8

8.6

3.4

3.7

3.3

6.4

2.0

8.8

4.7

23.0

27.4

27.1

24.7

22.9

24.3

28.0

23.6

28.2

25.3

Aged 0–4 No need

8.5

19.1

18.7

13.3

19.2

10.4

37.5

14.5

Too expensive

39.7

36.5

70.8

28.3

34.7

2.4

60.0

61.6

22.5

37.9

Too far

32.7

33.6

1.1

42.7

48.8

31.1

22.4

45.4

14.1

33.2

49.7

33.5

30.4

26.1

26.3

Lack of confidence

2.4

4.4

3.4

3.6

10.8

Other

25.8

26.6

44.3

20.9

3.4

2.5

3.5

No need

19.6

7.9

12.8

13.7

16.5

3.2

22.9

5.6

20.6

13.5

Too expensive

28.1

37.8

69.7

22.3

24.0

35.8

30.4

49.9

30.6

33.2

36.9

Aged 5–14

Too far

25.2

35.3

0.6

39.4

30.0

Lack of confidence

1.4

3.5

7.4

1.0

2.6

Other

32.8

32.1

20.6

36.0

33.3

30.4

13.3

57.7

21.9

30.5

1.0

3.1

8.8

2.5

32.3

36.2

31.9

32.5

Aged 15–59 No need

12.8

6.1

11.8

8.5

17.3

4.3

5.3

5.9

12.3

9.5

Too expensive

43.2

46.1

63.2

37.4

35.8

52.2

37.2

61.9

42.4

44.7

Too far

33.1

34.6

4.1

45.4

30.7

39.0

33.9

42.8

23.1

33.8

Lack of confidence

4.0

6.2

9.7

3.3

2.0

6.8

8.2

1.7

7.8

5.1

Other

19.9

26.5

26.2

22.1

25.2

16.2

24.7

21.4

27.9

23.2

Aged 60 and over No need

5.1

9.8

10.5

30.4

13.6

8.0

Too expensive

41.1

32.9

65.0

27.1

17.8

43.2

52.7

25.4

25.0

36.1

Too far

42.1

60.0

2.6

68.8

45.4

52.4

42.4

81.8

14.6

53.1

16.3

4.3

60.5

10.7

14.3

27.3

6.7

19.9

17.1

Lack of confidence

9.6

11.4

12.5

10.2

4.4

Other

8.8

22.3

21.9

15.6

19.2

Source: Authors’ estimates based on 2007 CWIQ survey.

Poverty and the Policy Response to the Economic Crisis in Liberia

are more informative when limited to the sample of individuals who experienced an episode of illness). For 11.0 percent of the population, there was no need to seek care, presumably because the illness was mild. Yet for 40.4 percent of those not seeking care, the reason was cost. Distance to facilities was an issue for 34.5 percent of those not seeking care. There is no clear pattern of differences between quintiles for either of the two main reasons not to seek care, but as expected, distance is more an issue in rural areas, while cost was more of a problem in urban areas where a higher share of individuals relies on private care providers and prices may generally be higher. Table 4.6 provides data on private health care spending by households. The largest expenditure in terms of the share of total spending for health is for the purchase of drugs (39.2 percent of total spending). This is followed by spending for medical treatment (injections, bandages, etc.), at 25.8 percent of the total, and spending for consultations, at 22.3 percent. As a share of total consumption, table 4.5 shows that health spending has a higher cost for the poor, but in absolute value, better off households tend to spend significantly more on average. The data on total private spending for health is provided in levels in table 4.7. On a per capita basis, households in the top decile of the population (ranked according to consumption per equivalent adult) spend four times as much as households in the bottom decile. The total private spending for health is estimated at close to L$0.9 billion (about US$ 15 million), which is similar to the total budget of the Ministry of Health and Social Welfare. As is the case for education, in part due to the legacy of the war, the governmentâ&#x20AC;&#x2122;s health budget is only a fraction of total spending on the public health system. In many cases, NGOs are topping up salaries for health professionals, as well as providing other incentives and materials directly to health facilities. Unfortunately these aid flows are not being tracked well, so that the government does not have a clear idea of how much is currently spent on public health overall. It has been suggested that total public health Table 4.6: Structure of householdâ&#x20AC;&#x2122;s expenditure in health, 2007 Residence area Purchase of drugs

Quintile

Urban

Rural

Total

39.8

38.3

44.5

35.6

43.2

36.5

39.2

Consultation by traditional practitioner

4.2

0.8

2.9

1.6

5.3

2.4

2.0

2.7

Vaccination costs

0.0

0.1

0.0

0.1

0.0

Medical consultation

20.9

24.1

18.9

24.5

18.1

24.6

22.7

22.3

Medical treatment (injection, bandages)

23.5

28.8

23.1

26.5

24.6

26.7

26.1

25.8

6.6

1.6

5.4

5.6

4.3

3.3

4.4

Purchase of traditional medications Radiology, EKG, scanner, tests

0.8

1.3

1.6

0.7

0.9

1.1

1.0

Hospitalization

4.1

5.1

3.6

5.5

2.3

4.4

5.5

4.5

100.0

1.9

2.1

2.3

2.0

1.6

1.9

Total Share of health in total consumption

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

A World Bank Study

Table 4.7: Household’s expenditure on health, 2007

Deciles

Total population

Total expenditure (millions of L$)

Total expenditure in health (millions of L$)

Per capita expenditure (L$) 4,561.7

101.8

2.2

Per capita expenditure in health (L$)

Share of health in total expenditures

270,469

1,233.8

27.5

270,582

2,132.9

41.6

7,882.7

153.6

1.9

270,477

2,764.9

61.1

10,222.4

225.8

2.2

270,761

3,292.0

77.5

12,158.4

286.4

2.4

269,714

3,801.9

74.1

14,096.0

274.6

1.9

271,127

4,460.3

93.6

16,450.9

345.3

2.1

270,714

5,020.2

95.8

18,544.1

353.9

1.9

269,729

5,937.7

117.9

22,013.4

437.2

2.0

271,538

7,286.5

124.3

26,834.2

457.9

1.7

270,273

13,385.6

216.5

49,526.4

801.1

1.6

2,705,385

49,315.8

930.0

18,228.7

343.8

1.9

10 Total

Source: Authors’ estimates based on 2007 CWIQ survey.

spending may be of the order of US$100 million for 2007, of which only $15 million is budgeted government expenditure.2 In table 4.8, access is measured by the distance from the nearest health facility. Remember that in table 4.5, access is mentioned as one of the two main reasons for not seeking care, especially in rural areas. In table 4.8, we provide data on the average time it takes to reach various types of facilities. At the national level, health clinics are on average at about two hours of where households live, but in rural areas, it takes almost three hours to reach the nearest clinic or hospital. These distances to health facilities are high in comparison of what has been observed in other countries, which justifies an effort on the part of the Ministry of health and Social Welfare as well as donors not only of rehabilitating existing facilities, but also of building new facilities in order to improve access in rural areas. Table 4.8: Time (in minutes) to the nearest infrastructure, 2007 Residence area

Quintile

Urban

Rural

9.7

8.4

11.4

8.4

8.9

7.5

8.4

8.8

Food market

23.2

179.1

162.8

161.0

167.6

113.5

71.0

129.8

Public transportation

12.8

161.7

145.7

140.4

152.0

77.2

77.5

114.6

Primary school

15.5

46.5

33.4

46.1

46.6

27.3

32.5

36.7

Secondary school

24.3

203.0

114.1

203.0

198.9

116.2

113.3

146.3

Health clinic/hospital

29.6

151.6

124.8

143.4

145.4

99.5

71.0

113.0

All season road

16.7

333.6

167.9

322.8

323.8

227.8

153.2

233.3

6.1

33.0

31.7

26.6

25.0

21.4

20.5

24.5

Supply of drinking water

Any road (vehicle)

Source: Authors’ estimates based on 2007 CWIQ survey.

Total Q4

Poverty and the Policy Response to the Economic Crisis in Liberia

2.3. Satisfaction with Health Services and Reasons for Non-satisfaction

Given the impact of the conflict on many facilities and the lack of resources to run some, one might expect satisfaction rates with health services to be low in Liberia. This is however not necessarily the case, as shown in table 4.9. Approximately 60 percent of the population is satisfied with the services received, which is not very high, but still above what has been obtained using the CWIQ survey for education. There are few differences in satisfaction rates between quintiles, apart from the fact that satisfaction seems to be lower for households in the middle quintile. Satisfaction rates are similar according to sexes, but they are higher in urban than in rural areas. The main reasons for not being satisfied are long waiting times (15.8 percent of those who have obtained care), distances to the facilities (12.3 percent), cost (11.3 percent), and lack of availability of drugs (10.4 percent). The issues of distances and lack of drugs are more prevalent in rural than in urban areas, but there is no obvious pattern of large differences in the reasons for nonsatisfaction between quintiles or by age group. Table 4.9: Satisfaction/problem with health services, 2007 Gender Male

Female

65.7

Residence area

Quintile

Urban

Rural

61.5

76.4

59.6

70.8

1.8

1.1

2.3

1.2

0.0

11.8

13.7

9.6

13.7

14.4

2.2

0.7

1.3

1.5

8.3

13.9

8.6

12.0

11.9

12.2

5.3

14.2

Total

62.3

53.7

64.5

66.7

63.5

1.2

2.3

2.9

0.7

1.5

14.4

14.3

11.7

10.2

12.8

1.9

2.2

2.0

0.9

1.4

12.3

7.9

13.6

13.2

8.9

11.2

10.9

10.0

18.6

6.3

14.2

12.1

Aged 0â&#x20AC;&#x201C;4 No problem (satisfied) Facilities were not clean Long waiting time No trained professionals Too expensive No drugs available Treatment unsuccessful Long distance to health facility Other

2.2

2.0

0.9

2.4

1.1

2.2

4.2

1.6

1.3

2.1

11.2

14.9

2.6

16.4

4.0

11.6

22.5

14.0

11.8

13.1

2.5

1.3

0.5

2.3

2.2

1.9

3.4

1.8

0.5

1.9 61.5

Aged 5â&#x20AC;&#x201C;14 No problem (satisfied) Facilities were not clean Long waiting time No trained professionals

63.0

59.9

62.0

61.3

63.8

61.7

59.3

62.5

60.4

1.6

3.0

1.9

2.4

2.1

0.6

2.6

1.4

4.8

2.3

15.3

16.2

19.5

14.5

19.6

15.9

10.3

13.7

20.1

15.7

1.6

2.0

1.3

2.0

1.7

3.3

2.4

1.3

1.8

11.2

12.0

12.4

11.3

8.4

12.6

11.7

13.7

11.2

11.6

No drugs available

9.1

11.5

6.4

11.5

10.3

6.1

12.2

8.2

14.4

10.3

Treatment unsuccessful

2.9

4.6

4.0

3.6

3.2

4.6

5.9

2.7

1.7

3.7

11.1

9.7

4.7

12.3

15.2

8.2

13.6

6.7

8.5

10.4

1.7

2.3

0.6

2.4

1.9

0.9

2.0

2.5

2.0

Too expensive

Long distance to health facility Other

(Table continues on next page)

A World Bank Study

Table 4.9 (continued) Gender Male

Female

Residence area Urban

Rural

Quintile Q1

Total

Aged 15â&#x20AC;&#x201C;59 No problem (satisfied)

58.8

58.2

63.4

56.5

59.1

53.6

58.1

62.1

59.3

58.5

1.3

1.6

1.5

1.4

1.0

0.6

2.1

2.2

1.3

1.4

17.0

16.0

16.5

16.4

18.7

21.1

13.4

15.5

14.2

16.4

0.8

1.1

0.5

1.2

0.5

1.8

1.3

0.2

1.0

Too expensive

11.5

10.9

12.5

10.7

8.6

11.2

12.2

10.6

12.7

11.2

No drugs available

11.1

9.7

7.5

11.5

10.1

9.3

11.1

8.3

12.6

10.3

Facilities were not clean Long waiting time No trained professionals

Treatment unsuccessful Long distance to health facility Other

4.7

5.8

4.8

5.6

3.1

7.0

5.7

5.6

5.2

5.4

13.2

12.7

4.9

16.1

13.5

17.5

15.4

11.5

7.6

13.0

2.6

2.2

1.1

2.9

1.5

1.9

3.3

2.5

2.6

2.4

58.2

55.4

58.0

56.7

49.2

63.1

56.0

57.0

2.7

1.0

20.4

22.7

18.9

19.8

18.1

19.1

11.1

12.4

Aged 60 and over No problem (satisfied) Facilities were not clean Long waiting time No trained professionals Too expensive No drugs available

1.8 18.1

58.5

18.2

21.5

17.2 0.6

1.5

10.6

6.8

11.9

20.8

1.3

0.8 11.0

57.5

2.2

0.5

0.4 8.9

7.6

11.0

6.7

9.2

7.8

7.9

7.4

6.3

4.9

6.8

14.2

7.8

Treatment unsuccessful

11.3

13.3

9.1

13.0

9.3

12.8

17.4

11.9

10.3

12.2

Long distance to health facility

12.0

11.0

4.1

13.6

10.3

8.1

29.6

5.2

6.3

11.6

1.1

1.9

4.4

0.7

2.5

2.2

0.8

0.1

1.4

1.5

Other

Overall population No problem (satisfied) Facilities were not clean Long waiting time No trained professionals

61.1

58.9

64.7

58.3

62.0

57.3

57.2

62.7

60.7

59.9

1.5

1.7

1.6

1.0

0.8

2.1

2.0

2.1

1.6

15.7

15.9

16.5

15.6

18.5

13.0

14.6

15.0

15.8

1.2

0.8

1.4

0.6

1.0

2.2

1.7

0.6

1.2

Too expensive

10.8

11.7

11.9

11.1

9.0

11.1

12.8

11.7

11.4

11.3

No drugs available

10.3

10.5

6.9

11.7

10.0

8.3

12.2

7.8

13.4

10.4

Treatment unsuccessful Long distance to health facility Other

4.3

5.4

4.3

5.1

3.4

6.1

6.2

4.5

4.0

4.9

12.2

12.3

4.4

15.0

12.2

13.6

16.9

10.4

8.5

12.3

2.2

2.1

1.1

2.5

1.8

1.7

2.8

2.2

2.1

2.2

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

3. Benefit Incidence of Public Spending for Health In this section, we provide an analysis of the benefit incidence of public spending for health. The key data are provided in table 4.10, and visualized in figure 4.3 in the case of public facilities and figure 4.4 in the case of private facilities. Table 4.10 provides estimates of the number of individuals from households belonging to various deciles of per

Poverty and the Policy Response to the Economic Crisis in Liberia

equivalent adult consumption that have obtained care in various types of facilities. For the benefit incidence analysis, we rely on the simplifying assumption that the unit costs of care are similar for all individuals seeking care in a given type of facility. Then the estimates of the number of individuals seeking care gives us the shares of total spending per type of facility that are allocated to the various deciles.

Figure 4.3: Concentration curves for use of public health facilities, 2007 100 90 Government health clinic

80 Government health center

70 Percent

Government

Equity Government health clinic Government hospital Other public facility Government health center Total public health

40 Total public health

30 20 10 0

Other public facility 0

5 Decile

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

Figure 4.4: Concentration curves for use of private health facilities, 2007 100

Equity Private doctor/dentist Traditional healer Private hospital/clinic Mobile clinic/black bagger/drug peddler Total private health Pharmacy Other private facility Private

90 80

Percent

70 60

Mobile clinic/black bagger/drug peddler

doctor/dentist

Total private health

Traditional healer

40 30 10 0

Other private facility

Pharmacy

Private hospital/clinic 0

5 Decile

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

11.5 10.8 9.7 8.2 10.4 10.0 9.5 8.2 10.7 11.0 100.0

1 2 3 4 5 6 7 8 9 10 Total

9.5 8.9 18.3 11.8 10.3 9.6 6.0 10.0 5.3 10.3 100.0

7,925 7,449 15,236 9,857 8,568 8,005 5,040 8,313 4,412 8,583 83,388

10.5 12.6 11.4 11.2 9.8 8.7 12.7 7.6 9.5 6.0 100.0

20,622 24,861 22,375 22,012 19,258 17,100 25,059 14,938 18,687 11,828 196,740

Gvt health clinic

5.5 5.1 12.5 15.0 9.2 11.7 15.5 11.0 7.5 7.2 100.0

2,200 2,037 5,027 6,039 3,715 4,707 6,231 4,414 3,002 2,910 40,282

Other public facility

Source: Authorsâ&#x20AC;&#x2122; estimates based on 2007 CWIQ survey.

32,236 30,248 27,101 22,988 28,974 28,003 26,506 22,886 29,880 30,841 279,663

Gvt hospital

1 2 3 4 5 6 7 8 9 10 Total

Deciles

Gvt health center

4.7 4.7 6.0 8.6 10.7 11.9 12.4 10.6 13.2 17.2 100.0

10,483 10,611 13,579 19,253 23,905 26,742 27,913 23,689 29,559 38,713 224,447

Private hospital or clinic

Private doctor or Pharmacy dentist Number of consultations 4,497 428 3,948 1,252 3,644 1,540 7,725 1,323 4,883 2,108 8,094 369 5,142 511 7,848 1,202 9,570 2,353 8,720 1,809 64,071 12,895 Share 7.0 3.3 6.2 9.7 5.7 11.9 12.1 10.3 7.6 16.3 12.6 2.9 8.0 4.0 12.2 9.3 14.9 18.2 13.6 14.0 100.0 100.0

Table 4.10: Benefit incidence analysis for the use of health care facilities, 2007

4.9 11.3 6.6 10.8 13.9 9.1 9.0 13.1 14.9 6.4 100.0

5,033 11,594 6,814 11,102 14,287 9,353 9,216 13,435 15,344 6,554 102,732 5.3 5.3 7.1 6.3 9.5 3.6 34.6 6.9 10.9 10.6 100.0

712 707 948 844 1,276 479 4,634 921 1,466 1,418 13,405

Mobile clinic, black bagger, Other private drug peddler facility

11.9 10.6 13.7 8.8 8.9 11.8 8.4 6.6 7.4 12.0 100.0

8,399 7,479 9,742 6,230 6,285 8,334 5,976 4,687 5,213 8,508 70,853

Traditional healer

10.5 10.8 11.6 10.1 10.1 9.6 10.5 8.4 9.3 9.0 100.0

62,983 64,595 69,739 60,896 60,515 57,815 62,836 50,551 55,981 54,162 600,073

Total public health

8.5 9.2 9.7 9.9 10.4 10.2 10.7 9.4 11.0 11.0 100.0

92,535 100,186 106,006 107,373 113,259 111,186 116,228 102,333 119,486 119,884 1,088,476

Total

76 A World Bank Study

Poverty and the Policy Response to the Economic Crisis in Liberia

It can be seen that for most public facilities, public spending for health seems to be allocated in roughly similar proportions to the various household groups ranked by consumption decile. For private service providers, traditional healers tend to be used more by the poor, while other types of providers tend to be used more by the better off. The fact that public spending for health does not appears to be regressive is probably related to the fact that in Liberia a large share of health services are provided by private institutions, which tend to be used more by better off households. Note that the columns “total” in table 4.10 are not weighted by the shares of public spending allocated to the various levels of spending.

4. Determinants of the Demand for Care In this last section, we look at the determinants of the demand for care using standard (probit) regression techniques. The analysis is conducted for the population that was sick over the last four weeks, separately in Monrovia, other urban areas and rural areas as well as at the national level. The dependent variable is whether the individual is seeking care or not. The explanatory variables include the following: (a) characteristics of the individual—the age of the individual and his/her sex; (b) geographic location variables, including urban versus rural areas in the national regression and a set of dummy variables for various regions (Greater Monrovia, North Central, North Western, South Central, South Eastern A, and South Eastern B); (c) household demographic variables—the number of children aged 0 to 5, the number of children aged 6 to 14, the number of male adults aged 15 to 60, the number of female adults aged 15 to 60, the number of seniors aged over 60, the age of the household head, and whether the household head is male or female; (d) the education level of the head (none, some primary, primary completed, some secondary, secondary completed, post-secondary); (e) the socioeconomic group of the head of household (employment in the public, private formal or private informal sector, self-employment in agriculture or another sector, or inactivity and unemployment, as well as whether the head has a second job); (f) whether the household has migrated due to the war and has been displaced, and whether the household has returned to its place of origin or never moved); (g) the quintile of consumption per equivalent adult of the household; and finally (h) a variables indicating access to facilities (time to nearest health clinic or hospital). Only the coefficient estimates are provided to save space, with indication as to their level of statistical significance. The results from the estimations shown in table 4.11 are mostly as expected. First, there is an inverse relationship between the age of the individual and the probability of seeking care, but the impact is small, albeit statistically significant in most cases. When running the regression on the sample as a whole, there is a statistically significant difference in the probability of seeking care between urban and rural areas (higher probability in urban areas by 3.6 percentage points). Surprisingly, there is a negative association between living in the Greater Monrovia area and seeking care, but this is partly offset by the positive impact of being in urban areas. Overall, the geographic location effects, when they are present, are of a limited order of magnitude (three to six percentage point difference in the probability of seeking care). Many of the demographic variables for the composition of the household are not significant, although having a higher number of female adults in the household does seem in some cases to improve the likelihood for an individual to seek care, while having a large num-

A World Bank Study

ber of children between 5 and 14 years of age reduces this likelihood. Having a female household head does not lead to an increase in the probability to seek care. The impact of the socioeconomic group of the head is present. When the head is involved in wage work (whether in the public, private formal, or private informal sector), household members have a higher probability of seeking care than otherwise. There is also some indication that if a household has been displaced, the probability of seeking care is lower. Individuals from richer households (who belong to higher quintiles) are more likely to seek care. Individuals who suffer from fever or malaria or who had an accident are also more likely to seek care than if they have been suffering from another illness. There is some indication that in Monrovia, individuals affected by a cough or skin problems are less likely to seek care. Finally, as expected, the longer it takes to go to the nearest facility, the less likely it is that an individual will seek care, at least in rural areas. Table 4.11: Determinants of the demand of health services, 2007 National

Monrovia

Other urban

Age

−0.001***

Female

−0.002

Rural

Individual characteristics 0.000

0.001

−0.003

−0.002

Residence area Urban

0.036***

—

Rural

Ref.

—

Region Greater Monrovia

−0.042***

—

North Central

0.037***

—

−0.062**

North Western

0.011

—

0.003

South Central

0.014

—

−0.058**

— 0.052*** 0.013 0.028**

South Eastern A

Ref.

—

Ref.

South Eastern B

−0.012

—

−0.036**

Ref.

Children aged 0 to 5

−0.001

−0.007

−0.005

0.002

Children aged 6 to 14

−0.009***

−0.005

−0.013***

Male adults aged 15 to 60

−0.001

0.009

−0.009**

0.013**

0.002

−0.005

Household composition

Female adults aged 15 to 59 Seniors aged over 60 Age of head of household Female household head

0.008*** −0.004 0.001** −0.001

0.000

−0.011

0.001

0.004

0.008

0.010*** −0.008 0.001** −0.010

Education level of head None Some primary Completed primary Some secondary Completed secondary Post secondary

Ref.

0.003

0.024

−0.009

0.000

−0.007

−0.024

0.006 −0.004 0.024*

0.018

−0.005

−0.003

−0.018

0.018

0.016

0.021

0.022

0.023

0.035*

(Table continues on next page)

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 4.11 (continued) National

Monrovia

Other urban

Rural

Public

0.040***

0.060*

0.000

0.044***

Private formal

0.033**

0.010

−0.021

0.052***

Private informal

0.047***

0.073***

−0.007

0.036*

Socioeconomic group of head of household

Ref.

Self-other

Self-agriculture

−0.005

0.034

−0.010

−0.002

Unemployed

−0.001

0.033

−0.032

0.011

Inactive, other

0.021**

0.050

0.013

0.016

The head has a second job

0.005

0.037

−0.046*

0.015

Displaced

−0.031*

−0.064**

0.004

−0.045*

Displaced and has returned to origin

−0.008

−0.036**

Migration status due to the war

Never move Time to health clinic/hospital (in 1,000 minutes)

Ref. −0.092***

0.021

−0.007

Ref.

0.132

0.030

−0.106***

Type of sickness/injury Fever/malaria

0.025***

0.010

−0.006

0.039***

Diarrhea/abdominal pains

0.014

−0.001

−0.025

0.027**

Pain in back, limbs or joints

−0.013

−0.018

Cough/breathing difficulties

0.000

−0.086***

−0.007

−0.103**

Skin problems Ear, nose or throat

0.005

−0.014

−0.008

0.024*

0.018

0.008

0.000

−0.024

−0.040

0.013

−0.023

−0.039

−0.046

−0.015

Dental

0.022

−0.016

−0.074

Accident

0.061**

—

Other

0.008

0.003

−0.013

Eye

0.072** 0.061* 0.013

Welfare quintiles Q1

Ref.

0.024***

0.040**

0.011

0.013

0.037**

−0.021

0.034***

0.065***

0.014

0.028**

0.036***

0.075***

0.005

0.027**

8,287

1,617

Observations

Ref.

1,086

Ref. 0.023** 0.016

5,554

Source: Authors’ estimates based on 2007 CWIQ survey. Note: * significant at 10%; ** significant at 5%; *** significant at 1%.

These regressions provide some useful insights into the determinants of the demand for care. For policy purposes, the main use of the regressions lies in assessing the potential impact of the construction of new health facilities on the demand for care. For example, the coefficient for the pooled sample of the distance to health facilities is −0.106 in rural areas. Given that the explanatory variable is expressed in 1,000 minutes, this means that a 100 minutes reduction in the time to go to school would increase the demand for

A World Bank Study

care by about 1.6 percentage point. It was mentioned that in rural areas, the average time needed to reach the nearest health facility was almost three hours. If this distance were cut in half, we would obtain an increase in the demand for care of about one and a half percentage point. This is a small value, and not as large as one is often led to believe, which suggests that policies to increase the demand for care further need to go beyond the simple provision of new health facilities, even if this is necessary of course in some areas.

5. Conclusion This chapter has provided a basic diagnostic of the health system in Liberia on the basis of the analysis of the 2007 CWIQ survey. Several findings show that Liberia stands out in comparison with other countries. First, the incidence of illnesses seems to be higher than in other countries, and at the same time the probability to seek care when ill is also very high. This may be due to the fact that many consultations appear to be free. Second, nongovernment facilities play a major role in the provision of care, which is again in part a legacy of the conflict. Third, while the cost of care is not necessarily high, and many individuals receive free care provided in most likelihood by NGOs, costs remain an issue for some individuals, as it is the main reason for not seeking care when sick. Distance is the second main reason for not seeking care, and is mentioned mostly by rural households. Public spending for health appears to be neither pro-poor, nor pro-rich, at least as measured on the basis of simple statistics on the number of consultations made by various groups of households. The use of private facilities is typically more prevalent among better off households, while the poor rely more than the better off on traditional healers. While overall public spending seems to be less biased against the poor in Liberia than in other countries due in part to the fact that better off households rely in part on private facilities, satisfaction rates, while not very low, are limited, with about 60 percent of care seekers being satisfied. The main complaints are related to long waiting times, distances to the facilities, cost, and lack of availability of drugs. The issues of distances and lack of drugs are more prevalent in rural than in urban areas. The fact that the quality of health services is limited is not surprising given the fact that in terms of budget, the Ministry of Health and Social Welfare does not have adequate resources to provide a basic package of health care to all. NGOs and other groups have stepped in, and are providing valuable services, but as the country completes its transition out of post-conflict stage, several important NGOs have indicated that they would reduce their presence in Liberia. In other words, while today a large share of the health system costs are borne by NGOs and donors, budgetary pressures on the government are expected to increase in future. Finally, the chapter has provided an analysis of the determinants of the demand for care. Many findings are as expected, with older individuals less likely to seek care, and better off households, as well as households whose head is a wage earner more likely to seek care. One interesting result to inform policy is that the distances to health facilities have an impact on the probability to seek care, as expected, but even a substantial reduction in these distances that could be obtained through a program of building new health facilities would apparently not lead to a dramatic increase in the demand for care, perhaps because it is already high in Liberia. This type of results underscores the complexity of designing a strategy for progress in the health sector in Liberia that is both ambitious, and affordable for the government and its partners.

Poverty and the Policy Response to the Economic Crisis in Liberia

Notes 1. The authors are with the World Bank. This chapter was prepared as an input to Liberia’s Poverty Reduction Strategy. Inputs were provided by Rose Mungai. Key results were presented at a workshop organized by Liberia’s core PRSP team in Monrovia on December 10-11, 2007. The views expressed in this chapter are those of the authors and need not reflect those of the World Bank, its Executive Directors or the countries they represent. 2. We are grateful to Rebecca Simson for pointing this to us.

References Backiny-Yetna, P., R. Mungai, C. Tsimpo, and Q. Wodon. 2012. “Poverty in Liberia: Level, Profile, and Determinants.” In Q. Wodon, ed., 9–34. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Heninger, L., C. Makinson, F. Richardson, M. Kaiser and J. A. Duany. 2006. Help Us Help Ourselves: Education in the Conflict to Post-Conflict Transition in Liberia. Women’s Commission for Refugee Women and Children, New York. Humphreys, M., and P. Richards. 2005. “Prospects and Opportunities for Achieving the MDGs in Post-conflict Countries: A Case Study of Sierra Leone and Liberia.” Center on Globalization and Sustainable Development (CGSD) Working Paper No. 27. Earth Institute, Columbia University, New York. International Labour Organization (ILO). 2009. “A Rapid Impact Assessment of the Global Economic Crisis on Liberia.” Mimeo, Monrovia. Liberia Health Sector Partnership Forum Report. 2007. Bridging the Gap and Beyond: Transitioning from Relief to Development. February 16, Washington, DC. Médecins sans Frontières. 2007. From Emergency Relief to Development: No Cheap Solution for Health Care in Liberia. Monrovia. Ministry of Health and Social Welfare. 2007. “Liberia Health System Overview.” PowerPoint presentation, Monrovia. Republic of Liberia. 2006. Interim Poverty Reduction Strategy: Breaking from the Past—From Conflict to Development. Monrovia. ———. 2007. The National Health Plan of Liberia (2007-2011): Standing Together in Support of Good Health. Monrovia. ———. 2008. Poverty Reduction Strategy. Monrovia. Richards, P., S. Archibald, B. Bruce, W. Modad, E. Mulbah, T. Varpilah, and J. Vincent. 2005. “Community Cohesion in Liberia: A Post-War Rapid Social Assessment.” Social Development Paper, Conflict Prevention and Reconstruction No. 21. World Bank, Washington, DC. United Nations Development Programme (UNDP) Liberia. 2006. National Human Development Report 2006—Mobilizing Capacity for reconstruction and Development. Monrovia.

Part II Impact of Higher Food Prices and Fiscal Measures to Respond to the Crisis

CHAPTER 5

Rice Prices and Poverty in Liberia Clarence Tsimpo and Quentin Wodon1 There has been a substantial literature on the link between rice and other cereal prices and poverty. The key in this literature is often to assess the double impact that a change in the price of rice can have through producers (who benefit from an increase in prices) and consumers (who lose out when the price increases). In Liberia however, at least under the current conditions, the impact of a change in the price of rice is not ambiguous at all. This is because a large share of the rice that is consumed is imported, while the rice that is locally produced is used mostly for auto-consumption rather than for sale on the market. In such circumstances, an increase in the price of rice will tend to result in higher poverty in the country as a whole (even if some local producers will gain from this increase), while a reduction in price will lead to a reduction in poverty. Furthermore, because rice represents such a large share of the food consumption of households, any change in its price is likely to have a rather large effect on poverty measures. Using data from the 2007 CWIQ survey, we find that an increase or decrease of 20 percent in the price of rice could lead to an increase or decrease of three to four percentage points in the share of the population in poverty.

1. Introduction Food security remains a major issue in Liberia. As noted in the Comprehensive Assessment of the Agriculture Sector prepared by Liberiaâ&#x20AC;&#x2122;s Ministry of Agriculture (2007), improving rural incomes, food production, food security, safety nets, and nutrition remains a key priority for the country. In part because rice production has fallen substantially during the period of conflict, a large majority of the population today is a net buyer of food, with much of food consumption coming from rice imports. There have been numerous accounts in the press over the years related to the price of rice in the country, including on issues regarding the awarding of import licenses for rice. The issues related to rice are not new in Liberia. Already in 1980, riots about the price of rice led to a coup. Any solutions to the countryâ&#x20AC;&#x2122;s rice and cereal deficit, and more generally lack of food security, will have to be multiple (see the analysis of the Comprehensive food security and nutrition survey in Republic of Liberia, 2006; see also Ejigu, 2006). High on the agenda is the fact that improved technologies must be used by farmers to increase their yields. To this end, the government and its partners are implementing a variety of programs that aim to provide better and more seeds as well as tools to farmers. As part of the 150-day action plan of the new government that took office in January 2006, one 85

A World Bank Study

of the actions for economic revitalization consisted in distributing 20.5 million tons of seed rice to farmers, as well as 41,500 tools. In the medium term, substantial progress is expected from improved rice varieties (e.g., NERICA) and the expansion of small-scale mechanization. But for the immediate years to come, rice imports are likely to continue to remain large, with potential fluctuations in the price for consumers of rice likely also to have a major impact on the poor. In this chapter, our objective is not to advocate a particular policy for increasing local rice production (which should help in the medium term for reducing prices paid by consumers), or for reducing the price of imported rice paid by consumers through import and VAT tax reform or through further regulatory reforms (the country has already liberalized rice imports, so that there is not anymore only one firm only that can import rice as used to be the case in the recent past). Instead, our objective is much more limited. It consists in using the 2007 CWIQ survey to make an assessment of the patterns of consumption and production of rice in the country, and to assess the potential impact on poverty of changes in the price of rice using a very simple methodology. There has been a substantial literature on the link between rice and other cereal prices and poverty. The key in this literature is often to assess the double impact that a change in the price of rice may have through producers (who benefit from an increase in prices) and consumers (who lose out when the price increases). For example, Indonesia is a country that used to import substantial amounts of rice, but where restrictions were progressively placed on imports in order to help local producers, with imports of rice actually banned after 2004. Using a general equilibrium model, Warr (2005) found that the ban on rice imports raised the price of domestically produced rice, and that this led to an increase in poverty by almost one percentage point (on the Indonesia story as well as for a more general discussion on the experience of governments in Asia to stabilize the price of rice, see Timmer and Dawe, 2007). Another paper on Indonesia by (Sumarto et al., 2005) using panel data suggests that the practice of subsidizing rice as part of a social safety net led to a reduction in the risk for household to be poor. Papers on Vietnam by Niimi et al. (2004) and Minot and Goletti (1998) suggest that the liberalization of rice exports probably led to a reduction in poverty despite an increase in the price of rice in the country, thanks essentially to increased production of rice. In Liberia however, at least under the current conditions, the impact of a change in the price of rice is not ambiguous at all. This is because a large share of the rice that is consumed is imported, while the rice that is locally produced is used mostly for autoconsumption rather than for sale on the market. In such circumstances, an increase in the price of rice will tend to result in higher poverty in the country as a whole (even if some local producers will gain from this increase), while a reduction in price will lead to a reduction in poverty. Furthermore, because rice represents such a large share of the food consumption of households, any change in the price of rice is likely to have a rather large effect on poverty measures. The chapter is structured as follows. Section 2 presents basic data on rice production and consumption in Liberia based on two main sources of information: the Comprehensive Assessment of the Agriculture Sector prepared by Liberiaâ&#x20AC;&#x2122;s Ministry of Agriculture (2007), and the results from the Comprehensive Food Security and Nutrition Survey (CFSNS) completed from March to April 2006 with data for 5,409 households. Our own analysis of patterns of rice consumption and production using the 2007 CWIQ survey (which is based on a sample of 3,600 households) is provided in Section 3. In section 4,

Poverty and the Policy Response to the Economic Crisis in Liberia

still using data from the 2007 CWIQ survey, we provide simple techniques and simulations first for illustrating the direction of the potential impact on poverty of a balanced budget tax reform involving import rice, and second for assessing in a bit more detail the impact of changes in the price of rice on poverty among producers as well as consumers. A brief conclusion follows.

2. Rice Production and Consumption in Liberia: A Brief Review The Comprehensive Assessment of the Agriculture Sector prepared by Liberia’s Ministry of Agriculture (2007) suggests that Liberia’s agriculture can be characterized as comprising of three different production systems. First are large plantations which focus on export crops (rubber, palm oil, coffee, and cocoa). Most of the production originates from plantations that are privately owned, but there are also a number of smaller state owned plantations operated by the Liberian Palm Products Corporation and the Liberian Cocoa and Coffee Corporation. A second component of Liberia’s agriculture sector consists of privately owned commercial farms of medium size which also focus on industrial crops for export and to a lesser extent on livestock for the local market. Finally, the bulk of the population engaged in agriculture belongs to small household farms which rely on traditional production techniques that generated low yields due among others to a lack of inputs, and thereby focus on subsistence production. These household farms are small, with most of them being of around one hectare in size or even less (FAO, 2001; CFSNS, 2006).2 In terms of consumption, rice is the main staple food, followed by cassava and other food crops. Production data are scarce, but some estimates are available from the FAO. According to these estimates (table 5.1), cassava production has better resisted to the conflict than rice production, which has fallen from about 180,000 tons at the start of the conflict to 110,000 tons today, while the population has increased substantially over the same period. By contrast, cassava production appears to have increased from 380,000 tons to 490,000 tons. Table 5.1: Rice and cassava production, 1990–2004

Year 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Cassava (fresh and dried) Area harvested Production (1,000 ha) (1,000 mt) Yield (mt/ha) 55.00 380.00 6.91 42.00 270.00 6.43 40.00 280.00 6.67 40.00 245.00 6.13 29.00 250.00 6.25 32.81 175.00 6.03 43.30 213.26 6.50 47.00 282.20 6.52 55.50 307.00 6.53 67.00 361.30 6.51 72.50 440.50 6.57 72.50 480.00 6.62 75.00 480.00 6.62 75.00 490.00 6.53 75.00 490.00 6.53

Area harvested (1,000 ha) 175.00 110.00 120.00 60.00 45.00 50.00 75.60 135.20 161.90 153.70 143.50 130.00 120.00 120.00 120.01

Source: Ministry of Agriculture (2007), based on FAOSTAT data.

Rice Production (1,000 mt) 180.00 100.00 110.00 65.00 50.00 56.20 94.45 168.40 209.40 196.30 183.40 145.00 110.00 100.00 110.00

Yield (mt/ha) 1.03 0.91 0.92 1.08 1.11 1.12 1.25 1.25 1.29 1.28 1.28 1.12 0.92 0.83 0.92

A World Bank Study

Two different systems of rice cultivation co-exist in Liberia. Upland rice cultivation is more prevalent, with 63 percent of producing households using this method of cultivation, as compared to 17 percent of households using swamp rice cultivation methods (the rest, 21 percent of producers, combine both techniques). Upland cultivation is prevalent in River Cess, Grand Kru, and Nimba, while swamp rice is found in Lofa County thanks in part to donor funding for agricultural development projects (CFSNS, 2006). Even in swamp or lowland areas, productivity or yields per hectare are often low, and well below that of neighboring countries, and in the country as a whole, locally produced rice is used mainly for auto-consumption and subsistence. Among the constraints to productivity, households have identified the following: lack of seeds and tools (mentioned by 50 percent of households), lack of financial capital to purchase agricultural inputs (31 percent), lack of household labor (28 percent), and groundhog (pesticide) attacks as well as bird attacks (each cited by 19 percent of households in the CFSNS survey). The inability of the country to produce enough rice and other cereals to feed the population has led to massive imports and has been one of the (many) factors that have led to high levels of food insecurity. The FAO (2006) describes food insecurity as a situation under which some people lack access to enough food of good quality to meet their nutrition needs in order to be able to lead an active and healthy life. According to the results of the CFSNS survey, most rural households are suffering from some forms of food insecurity: As described in Ministry of Agriculture (2007: 15): Nationally 80% of the rural population is either moderately vulnerable (41%), or highly vulnerable to food insecurity (40%), while only 9% of the rural population is food secure, and 11% are food insecureâ&#x20AC;Ś Chronic malnutrition rates reach 39% for children under five, and only 32% of households had access to improved water sources, and other basic services were limitedâ&#x20AC;Ś The most food insecure and highly vulnerable groups [are] involved in palm oil producing and selling (64%) followed by hunters and contract labourers (respectively 61% and 58%). The more food secure and moderately vulnerable groups are among the cash and food crop producers (37%), the petty traders and the employees (44% each) [see table 5.2]. Importantly, even cash and/or food crop producers are considered likely to be food insecure (indeed, this group of households is considered as likely as many other groups to be food insecure in table 5.2), suggesting that food production for auto-consumption often still does not enable many households to meet their food needs.

3. Rice Production and Consumption in the 2007 CWIQ Survey In this section, the 2007 CWIQ survey is used to estimate rice production and consumption, and separate consumption into locally produced rice and imported rice. Table 5.3 provides summary data on rice consumption and production for auto-consumption, as well as a comparison with a number of other food items commonly consumed in Liberia. The total value of food consumption for the items listed in table 5.2 accounts for 87 percent of the total food consumption of households (these items were used for estimating a food poverty line in Liberia using the cost of basic needs method, as discussed in BackinyYetna et al., 2012). The following comments are worth pointing out:

Poverty and the Policy Response to the Economic Crisis in Liberia

Table 5.2: Vulnerability, incomes, and livelihood profile, 2006 % moderately vulnerable and food secure

% highly vulnerable and food insecure

% of income derived from food crop production

% of income derived from cash crop production

Cash and food crop producers

Petty traders

Employees

Food crop farmers

Charcoal producers

Rubber tapers

Fisher folks

Palm oil and food crop producers

Skilled labourers

Contract labourers

Hunters

Palm oil producer/ seller

Livelihood profile

Source: CFSNS (2006).

■

Rice is by far the largest food consumption item, accounting for more than a third of the value of total food consumption. The value of imported rice is estimated at L$6.5 billion (about $100 million at the current exchange rate of L$62 per US$), while that of locally produced rice is estimated at L$4.7 billion (this includes the imputed value of locally produced rice for auto-consumption). In total, rice thus accounts for L$10.2 billion in total consumption, a figure that can be compared to the total food consumption in table 5.3 estimated at L$30.2 billion. The total value of rice imports in 2007 is estimated in the survey at about US$ 100 million. This is probably an overestimation of true imports, which is not surprising given that the 2007 CWIQ survey tends to overestimate consumption (Backiny-Yetna et al., 2011). At the same time, the order of magnitude of the estimation of imports is not completely off, since according to the latest staff report of the International Monetary Fund (2007), imports of rice were estimated at US$57 million in 2006 (the exchange rate between 2006 and 2007 has not changed dramatically), and it is quite possible that not all rice imports are reported in the government’s official statistics. The total production of local rice is estimated at approximately 103,000 tons, which may be on the low side, but is also of an appropriate order of magnitude given that according to table 5.1, production was estimated at 110,000 tons in 2004, and production has probably not increased dramatically since then. Locally produced rice is used mostly for auto-consumption, since only slightly more than one fourth of the locally producer rice is actually purchased. This finding echoes similar results obtained from the 2006 Comprehensive Food Security and Nutrition Survey. The share of rice consumption in terms of the total estimated caloric intake per adult equivalents of households, at 50 percent, is even larger than the share of rice in total consumption, at about a third. This underscores even more the

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ survey.

Imported rice Local rice Maize/corn Cassava flour (fufu, gari, etc) Gari Bread Chicken Game and insects (porcupine, gazelle) Fresh or frozen fish Smoked fish (dried or salted) Fresh milk Eggs Palm oil Banana, plantain Coconuts Palm nut Cassava leaves Bitter balls Okra Green pepper Hot or sweet pepper (fresh or dry) Onions Dried beans Cassava roots Sugar Bouillon cubes (maggi, jumbo, etc) Salt Soft/carbonated drinks (coke,fanta,etc) Total basket Total other food expenditures Total food

Monetary value (millions of L$) Autocons. gifts and Purchase food aids Total 6,492.9 0.0 6492.9 1,256.3 3,478.6 4,734.9 51.8 80.6 132.4 137.6 123.8 261.4 151.1 0.0 151.1 304.2 32.1 336.3 625.6 261.1 886.6 158.5 163.3 321.8 2549.5 640.7 3,190.2 614.3 161.3 775.6 175.9 0.0 175.9 131.4 42.7 174.1 1,234.7 484.4 1,719.1 235.8 252.5 488.4 97.4 50.5 147.9 286.6 217.2 503.8 91.5 112.4 204.0 244.3 134.9 379.1 103.1 69.2 172.3 569.6 334.4 904.0 263.9 0.0 263.9 471.0 30.2 501.2 232.4 0.0 232.4 362.8 580.8 943.7 249.8 0.0 249.8 754.2 0.0 754.2 310.4 0.0 310.4 212.3 26.8 239.0 18,369.0 7,277.5 25,646.5 3,247.7 1,265.9 4,513.6 21,616.7 8,543.4 30,160.1

Share in total consumption (% of L$) Autocons. gifts and Purchase food aids Total 30.0 0.0 21.5 5.8 40.7 15.7 0.2 0.9 0.4 0.6 1.4 0.9 0.7 0.0 0.5 1.4 0.4 1.1 2.9 3.1 2.9 0.7 1.9 1.1 11.8 7.5 10.6 2.8 1.9 2.6 0.8 0.0 0.6 0.6 0.5 0.6 5.7 5.7 5.7 1.1 3.0 1.6 0.5 0.6 0.5 1.3 2.5 1.7 0.4 1.3 0.7 1.1 1.6 1.3 0.5 0.8 0.6 2.6 3.9 3.0 1.2 0.0 0.9 2.2 0.4 1.7 1.1 0.0 0.8 1.7 6.8 3.1 1.2 0.0 0.8 3.5 0.0 2.5 1.4 0.0 1.0 1.0 0.3 0.8 85.0 85.2 85.0 15.0 14.8 15.0 100.0 100.0 100.0

Table 5.3: Structure of food consumption and role of rice, 2007 Quantity (tons) Autocons. gifts and Purchase food aids Total 144898.2 0.0 144898.2 27,267.1 75,500.7 102,767.8 1,737.5 2,707.1 4,444.6 6,211.7 5,589.7 11,801.4 3,663.7 0.0 3,663.7 3,543.0 373.9 3,916.9 4,028.3 1681.0 5,709.4 370.0 536.4 906.4 21,992.1 5,526.4 27,518.5 2,047.1 537.4 2,584.5 1,087.1 0.0 1,087.1 591.9 192.5 784.4 14,809.1 5,810.4 20,619.5 8,323.8 8,914.4 17,238.3 3,257.8 1,687.9 4,945.7 19,097.2 14,472.0 33,569.2 7,265.9 8,924.8 16,190.6 6,898.4 3,808.6 10,707.0 1,233.6 828.0 2,061.6 3,443.7 2,021.5 5,465.2 447.2 0.0 447.2 3,375.7 216.1 3,591.8 3,143.4 0.0 3,143.4 28,923.1 46,299.2 75,222.3 3,305.5 0.0 3,305.5 1,966.6 0.0 1,966.6 8,206.1 0.0 8,206.1 1,552.0 195.8 1,747.8 332,686.8 185,823.8 518,510.6 49,982.7 27,531.8 77,514.5 382,669.6 213,355.6 596,025.1

Daily calories (kcal) per eq adult Autocons. gifts and Purchase food aids Total 694.0 0.0 694.0 130.6 361.6 492.2 8.2 12.8 21.1 28.0 25.2 53.3 16.5 0.0 16.5 11.6 1.2 12.9 7.4 3.1 10.5 1.3 1.9 3.2 18.6 4.7 23.2 10.1 2.7 12.8 1.1 0.0 1.1 1.1 0.4 1.4 155.9 61.2 217.1 14.8 15.9 30.7 16.7 8.6 25.3 100.8 76.4 177.2 8.7 10.7 19.4 2.9 1.6 4.5 0.6 0.4 1.0 1.6 1.0 2.6 0.3 0.0 0.3 1.8 0.1 1.9 13.9 0.0 13.9 56.9 91.0 147.9 17.4 0.0 17.4 8.6 0.0 8.6 36.5 0.0 36.5 0.9 0.1 1.0 1,367.0 680.5 2,047.5 205.4 100.8 306.2 1,572.3 781.3 2,353.7

90 A World Bank Study

Poverty and the Policy Response to the Economic Crisis in Liberia

fundamental role played by rice, including through imports, in the issue of food security in the country. Table 5.4 provides additional data on rice consumption as estimated in monetary terms using the CWIQ 2007 survey. As expected, locally produced rice is consumed mostly by households in rural areas, and the capital of Monrovia is the area that is the most dependent on imported rice. At the same time, an overwhelming majority of households outside of the capital also consume imported rice, so that for most households, local production is apparently not large enough to meet their own needs. There are also interesting differences in terms of the value of consumption according to the level of well-being of households, as measured by their level of total monetary consumption per equivalent adult. For imported rice, the consumption pattern is similar for all households except those located in the bottom quintile of the distribution of consumption, as these households consume only about half of what other households consume. For imported rice, there is a more traditional pattern according to which the richer a household is, the higher the expenditure of that household is as well. Still, while richer households tend to consume much more imported rice than poorer households, the consumption of imported rice among the poor is far from being negligible, so that changes in the price for consumers of imported rice can be expected to have a major impact on the measures of poverty obtained for the population as a whole, the issue to which we turn in the next section. Table 5.4: Rice consumption for different household groups, 2007

% HH consuming rice Locally produced Imported rice rice Total

Average consumption for all HH Locally produced Imported rice Rice Total

Average consumption for households with positive consumption Locally produced Imported rice Rice Total

Residence area Rural

80.0%

79.2%

99.2%

13,201.2

Urban

17.1%

97.3%

98.6%

1,566.6

10,484.8

23,686.0

13,312.5

10,573.1

23,885.6

18,633.0

20,199.6

1,589.6

18,906.9

20,496.6

Region 7.2%

98.2%

98.4%

227.6

19,585.0

19,812.5

231.3

19,905.7

20,137.0

North Central

Greater Monrovia

87.9%

71.6%

99.0%

15,216.9

9,118.9

24,335.8

15,371.0

9,211.3

24,582.3

North Western

69.8%

90.9%

99.7%

9,516.9

9,824.7

19,341.6

9,545.1

9,853.9

19,399.0

South Central

46.3%

90.6%

98.6%

5,796.7

15,287.3

21,084.1

5,877.9

15,501.4

21,379.3

South Eastern A

83.8%

83.0%

99.4%

16,150.2

10,553.5

26,703.6

16,240.5

10,612.5

26,853.0

South Eastern B

75.6%

91.9%

99.9%

10,910.1

15,587.9

26,498.0

10,919.4

15,601.2

26,520.7

Quintile Q1 (poorest)

63.4%

71.8%

96.5%

5,431.4

6,166.7

11,598.2

5,631.1

6,393.5

12,024.7

70.0%

83.0%

99.3%

9,520.0

10,483.4

20,003.3

9,582.7

10,552.4

20,135.1

62.5%

85.7%

99.6%

10,149.5

12,912.5

23,062.0

10,187.9

12,961.4

23,149.2

58.5%

87.6%

99.7%

10,400.1

14,361.1

24,761.2

10,431.3

14,404.2

24,835.5

Q5 (richest)

50.7%

92.3%

99.3%

11,104.2

18,502.0

29,606.2

11,178.2

18,625.3

29,803.5

Total

60.1%

84.9%

99.0%

9,524.0

13,060.1

22,584.1

9,623.1

13,195.9

22,819.0

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ survey.

A World Bank Study

4. Simulating the Impact on Poverty of Changes in the Price of Rice There are two simple so-called partial equilibrium ways to discuss the potential impact of a change in the price of rice on poverty. A first way is rounded in the theory of balanced budget marginal tax reforms, and we illustrate this approach below not to actually advocate for such a reform in Liberia (more information and a more in-depth analysis would be needed before doing so), but to present the idea and technique, because this gives some additional background on the issues, especially in terms of the comparison of the consumption patterns for rice as opposed to other food items. The second approach is to actually simulate the impact of non-marginal absolute or proportionate changes in the price of rice on poverty among both rice consumers and rice producers. The results from this second approach are probably simpler to interpret for the non-specialist reader and thereby for potential reference in Liberiaâ&#x20AC;&#x2122;s Poverty Reduction Strategy, and although we are making some simplifying assumptions in implementing the approach, the orders of magnitude of the impact on poverty that are presented are likely to be correct. 4.1. Marginal Tax and Subsidy Reforms

Figure 5.1 provides consumption dominance curves of the second order for imported rice and selected other food consumption items. As demonstrated by Makdissi and Wodon (2002; see also Duclos, Makdissi and Wodon, 2008), these curves are useful to assess the impact of so-called balanced budget marginal tax reforms on poverty. The idea is to test whether increasing a tax or subsidy for one type of goods, while reducing a tax or subsidy for another type of goods in such a way that the overall tax receipts or subsidy expenditures remain the same, will lead to a reduction or an increase in a wide range of poverty measures. If one consumption dominance curve is above another, then

Figure 5.1: Cumulative CD curve for selected food itemsâ&#x20AC;&#x201D;Order 2, 2007 0.9 0.8

Imported rice

Fresh/frozen fish

Bouillon cubes

Chicken

Palm oil Smoked fish

Total foods

Cumulative CD curve

0.7 Total foods

0.6 0.5

Imported rice

0.4

Smoked fish

Palm oil

0.3

Fresh/frozen fish

0.2

Bouillon cubes

Chicken

0.1 0

0.2

0.4

0.6 0.8 Total per eq. adult consumption/Z

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ survey.

1.2

1.4

Poverty and the Policy Response to the Economic Crisis in Liberia

it is good for poverty reduction to reduce a tax (or increase a subsidy) on the good with the curve that is above the other, while increasing a tax (or reducing a subsidy) for the good that corresponds to the curve that is located below the first curve. When using consumption dominance curves of the second order, we are in practice considering the impact of taxes or subsidies on poverty measures like the poverty gap, which takes into account not only the share of the poor in the population, but also the distance separating the poor from the poverty line or â&#x20AC;&#x153;depthâ&#x20AC;? of poverty (when computing the poverty gap, the non-poor are included in the estimation, but they are given a zero distance separating them from the poverty line since they are not below that line). More simply, the consumption dominance curves of order two actually represent the cumulative share of the total consumption of a good that is made by the poor. The horizontal axis represents the level of consumption per equivalent adult normalized by the poverty line, so that a value of one corresponds to the poverty line actually used in the country. In Figure 5.1 for example, at a value of one on the horizontal axis (which means that we are looking at the consumption share of all the poor taken together), the value of the horizontal axis for total food consumption is about 45 percent. This means that the poor as a whole, who represent 63.8 percent of the population (see Backiny-Yetna et al., 2012) consume about 45 percent of the total food consumed in the country. In terms of comparing different goods and curves in figure 5.1, chicken and smoked fish have the lowest curves. This means that the shares of chicken and smoked fish consumed by the poor are lower for any poverty line we may choose than the shares of the other items presented in the figure. Thus, if this were feasible, it would be better to tax more chicken and smoked fish and tax less other goods. We are of course not suggesting here that this should be doneâ&#x20AC;&#x201D;it would be very difficult in practice to tax chicken or smoked fish, as these goods are locally produced and informally sold and purchased, and thereby typically not subject to taxes. But we want to highlight the properties of the curve representing imported rice, in comparisons with other curves. Imported rice has a consumption dominance curve that is located above the curve for overall food consumption. This means that despite the fact that imported rice is consumed more by better off households than by very poor households (as was already evident from table 5.4), it is still a good that could tentatively be targeted for a reduction in taxes, such as import taxes which are currently levied on this commodity. Indeed, many other food items will be even more consumed by the non-poor than by the poor than imported rice, and in addition, most non-food items, some of which are clearly imported and thereby subject to tax, would probably have consumption dominance curves far below that obtained for food consumption as a whole. Independently of taxation, the consumption dominance curves can also be used to argue that other measures that would help reduce the price of imported rice would be beneficial for poverty reduction, and that imported rice, due not only to its consumption pattern but also to the sheer importance of the commodity in total food consumption, should indeed be a prime candidate for efforts to reduce consumer prices (the issue is less pressing for local rice, not only because consumption of local rice is lower than for imported rice, but also because most of the local rice is actually self-consumed, so that prices pay a less important role there). In Liberia, it has long been argued that liberalizing the imports of rice could lead to more competition (there was until recently a quasi monopoly for such imports), and that

A World Bank Study

this competition could lead to a reduction in the prices ultimately paid by consumers. The data presented in figure 5.1, and more generally in this chapter, certainly suggest that a reduction in prices, if it were to be obtained thanks to more competition, would have a potentially large impact on poverty. In the next section, we estimate exactly how large this poverty impact may be. 4.2. Non-marginal Changes in the Price of Rice

In this section, we provide estimates of the impact of changes in the price of rice (whether imported or locally produced) on the headcount index of poverty, which is simply a measure of the share of the population in poverty (i.e., with a level of consumption per equivalent adult below the poverty line; for an introduction to the concepts and techniques of poverty measurement, see Coudouel et al., 2002). We carry the simulations in a very simple way. First, for rice producers, we measure the additional income or the loss in income obtained from the sale of rice by households due to an increase or reduction in the price of rice. We assume that this difference in income translates into an equivalent difference in the consumption per equivalent adult of households used to measure poverty. We then recomputed the poverty measures keeping the poverty line intact. For consumers, we do essentially the same: we estimate the increase or decrease in the cost of rice following a change in price, taking into account the quantities actually consumed by each household. In the case of a reduction in price, we then add to the consumption aggregate the reduction in the total cost of rice for the household, since this reduction in cost means that the household can actually consume other goods (this is thus as if the household consumption had increased.) In the case of an increase in the price of rice, we subtract from the consumption aggregate the value of this increase, since the household will have to give up other consumption goods in order to be able to purchase the rice it needs. For either an increase or a decrease in the price of rice, we then compute again poverty with the adjusted consumption level. This procedure is admittedly a very rough approach, but it has the merit of being simple. The approach may slightly overestimate the impact on poverty of changes, because we do not take into account the price elasticity of rice consumption, but this price elasticity is likely to be very low in any case, due to the fact that rice is so dominant in the diet of the population. In addition, the approach does not take into account any ripple effects of changes in the price of rice on other parts of the economy. More sophisticated methods could be used to measure the “general equilibrium” effect of a change in the price of rice, but such simulations require a much larger number of assumptions which are the subject of debate. The estimations given here thus provide “first round” likely poverty effects from lower or higher rice prices paid to producing households or paid by consuming households, assuming that households don’t change their consumption patterns for rice after the change in price. Key results from the simulations are provided in tables 5.5 and 5.6. The headcount index of poverty is the share of the population with a level of consumption per equivalent adult below the poverty line. The poverty gap takes in addition into account the distance separating the poor from the poverty line (while giving a zero distance to the non-poor). The squared poverty gap takes in addition into account the square of that distance (and thus inequality among the poor).

1,622.1

Average per eq. adult change in L$

21.3

10.8

Poverty gap

Squared poverty gap

10.5

Squared poverty gap

24.5

12.8

Squared poverty gap

26.7

13.9

Poverty gap

Squared poverty gap

13.8

26.6

70.1

12.7

24.5

63.9

−39.2

23,710.2

10.8

21.4

59.1

11.1

21.8

59.3

1,351.8

25,101.1

−25%

Source: Authors’ estimation using 2007 CWIQ survey.

70.3

Headcount index of poverty

Poverty, rice producers

64.0

Poverty gap

−47.0

Headcount index of poverty

Poverty, population as a whole

Average per eq. adult change in L$

23,702.4

20.9

Poverty gap

Consumption per eq. adult (L$)

58.2

Headcount index of poverty

Poverty, rice consumers

58.4

Headcount index of poverty

Poverty, population as a whole

25,371.5

Consumption per eq. adult (L$)

−30%

13.7

26.4

70.0

12.7

24.5

63.9

−31.3

23,718.0

11.1

21.9

60.2

11.4

22.3

60.4

1,081.4

24,830.8

−20%

−10%

−5%

No change +5%

+10%

+15%

11.7

23.0

61.9

12.0

23.3

62.1

540.7

24,290.0

12.0

23.5

62.7

12.3

23.9

62.8

270.4

24,019.7

12.4

24.1

63.6

12.7

24.4

63.8

—

23,749.3

12.7

24.7

64.4

13.0

25.0

64.6

−270.4

23,479.0

13.1

25.3

65.9

13.4

25.6

66.1

−540.7

23,208.6

13.5

25.9

66.9

13.8

26.2

67.1

−811.1

22,938.3

13.6

26.3

69.9

12.7

24.5

63.9

−23.5

23,725.8

13.5

26.2

69.9

12.7

24.5

63.9

−15.7

23,733.7

13.5

26.1

69.3

12.7

24.4

63.8

−7.8

23,741.5

13.4

26.0

69.3

12.7

24.4

63.8

—

23,749.3

13.3

25.9

68.8

12.7

24.4

63.7

7.8

23,757.2

13.2

25.8

68.6

12.6

24.4

63.7

15.7

23,765.0

13.2

25.7

67.5

12.6

24.4

63.5

23.5

23,772.8

Impact of changes in producer prices only (no impact on consumer prices)

11.4

22.5

60.8

11.7

22.8

61.0

811.1

24,560.4

Impact of changes in consumer prices only (no impact on producer prices)

−15%

Percentage changes in prices

Table 5.5: Impact of a change in consumer or producer prices for rice on poverty, 2007

13.1

25.6

67.1

12.6

24.3

63.4

31.3

23,780.7

13.9

26.5

67.9

14.2

26.8

68.0

−1,081.4

22,667.9

+20%

13.0

25.5

67.1

12.6

24.3

63.4

39.2

23,788.5

14.3

27.2

68.9

14.6

27.5

69.0

−1,351.8

22,397.6

+25%

12.9

25.4

66.8

12.6

24.3

63.4

47.0

23,796.3

14.7

27.8

69.8

15.0

28.1

69.9

−1,622.1

22,127.2

+30%

Poverty and the Policy Response to the Economic Crisis in Liberia 95

1,575.1

10.6

Squared poverty gap

10.8

21.5

Source: Authors’ estimation using 2007 CWIQ survey.

58.6

21.0

Headcount index of poverty

Poverty gap

59.2

11.9

Poverty, rice consumers

Squared poverty gap

11.6

63.1 23.3

62.7

22.9

Headcount index of poverty

Poverty gap

Poverty, rice producers

21.9

21.4

10.9

Poverty gap

Squared poverty gap

11.2

59.4

58.8

Headcount index of poverty

Poverty, population as a whole

25,062.0

25,324.5

Consumption per eq. adult (L$)

Average per eq. adult change in L$

1,312.6

−25%

−30%

11.1

22.0

60.2

12.2

23.8

64.0

11.4

22.4

60.5

1,050.1

24,799.4

−20%

11.4

22.5

60.9

12.5

24.3

64.4

11.7

22.9

61.1

787.6

24,536.9

−15%

11.7

23.0

62.0

12.8

24.9

66.1

12.0

23.4

62.2

525.0

24,274.4

−10%

12.0

23.6

62.7

13.1

25.4

67.1

12.3

23.9

62.9

262.5

24,011.9

−5%

12.4

24.1

63.6

13.4

26.0

69.3

12.7

24.4

63.8

—

23,749.3

No change

12.7

24.7

64.4

13.7

26.6

70.3

13.0

25.0

64.6

−262.5

23,486.8

+5%

Percentage changes in prices

Table 5.6: Impact of a change of both producer and consumer prices of rice on poverty, 2007

13.0

25.2

65.8

14.1

27.2

72.0

13.4

25.5

66.0

−525.0

23,224.3

+10%

13.4

25.8

66.8

14.5

27.9

74.7

13.7

26.1

67.0

−787.6

22,961.8

+15%

13.8

26.4

67.6

14.8

28.6

76.1

14.1

26.7

67.7

−1,050.1

22,699.2

+20%

14.2

27.1

68.8

15.3

29.3

78.5

14.5

27.3

68.9

−1,312.6

22,436.7

+25%

14.6

27.7

69.7

15.7

30.1

79.3

14.9

28.0

69.8

−1,575.1

22,174.2

+30%

96 A World Bank Study

Poverty and the Policy Response to the Economic Crisis in Liberia

Consider first table 5.5, which is based only on data on the consumption of rice. At the time of the survey, the share of the population in poverty was 63.8 percent. If the price of rice could be reduced by 20 percent, and if we look only at the impact on the consumer side, poverty would fall to 60.4 percent. If the price of rice were to increase by 20 percent, poverty would increase to 68.0 percent. If we look at the producer prices, the impacts are much lower, since only locally produced rice that is actually sold must be taken into account for the simulations (for rice auto-consumed, changes in producer prices do not affect household welfare for producers). If the price of rice is reduced by 20 percent, and if we look only at the impact on the producer side, poverty would increase only to 63.9 percent, while if the price of rice for producers were to increase by 20 percent, poverty would decrease to 63.4 percent in the overall population. The total impact of changes in the price of rice on poverty is obtained by taking both consumers and producers into account, and the results are given in table 5.6. If the price of rice is reduced by 20 percent, poverty is reduced in the population as a whole to 60.5 percent, while if the price of rice increases by 20 percent, poverty would increase to 67.7 percent. These are relatively large effects for a single commodity, and they underscore why the population’s feelings about the price of rice run high in Liberia.

5. Conclusion When assessing the potential impact of a change in the price of cereals on poverty, it is important to consider both the impact on producers (who tend to benefit from an increase in prices) and consumers (who tend to lose out when the price increases). If producers tend to be poor and if consumers live in urban areas and are better off, an increase in the price of rice, despite its impact on the cost of food, may very well be poverty reducing. In Liberia however, the impact of a change in the price of rice is not ambiguous at all. A majority of the rice consumed is imported, and a majority of the rice that is locally produced is used by farmers for their auto-consumption. Therefore, any increase (decrease) in the price of rice, whether imported or locally produced, will clearly result in an increase (decrease) in poverty, and this impact is likely to be large given the important share of food consumption allocated to rice in the country. Using data from the 2007 CWIQ survey implemented by Liberia’s Institute of Statistics, we find that a change in the price of rice of 20 percent could lead to an increase or decrease of three to four percentage points in the share of the population in poverty, which is indeed large for a single commodity. The magnitude of the impact on poverty of changes in the price of rice suggests that the issue of what can be done to help reduce the price of rice for consumers would warrant a thorough discussion under the preparation of the country’s Poverty Reduction Strategy.

Notes 1. The authors are with the World Bank. This chapter was written for a workshop that took place in March 2011 in Monrovia for the preparation of the second Liberia Poverty Reduction Strategy. The views expressed in this chapter are those of the authors and need not reflect those of the World Bank, its Executive Directors or the countries they represent. 2. In fact, 53.6 percent of rice farms were between 0.2ha–1.19ha with a further 25 percent of rice farms from 1.2ha–1.69ha. For cassava, 70 percent of farms are less than 0.69ha (CFSNS, 2006).

A World Bank Study

References Backiny-Yetna, P., R. Mungai, C. Tsimpo, and Q. Wodon. 2012. “Poverty in Liberia: Level, Profile, and Determinants.” In Q. Wodon, ed., 9–34. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Coudouel, A., J. Hentschel, and Q. Wodon. 2002. “Poverty Measurement and Analysis.” In J. Klugman, ed., A Sourcebook for Poverty Reduction Strategies: Volume 1: Core Techniques and Cross-Cutting Issues. Washington, DC: World Bank. Duclos, J. Y., P. Makdissi, and Q. Wodon. 2008. “Socially Efficient Tax Reforms.” International Economic Review 49: 1505–1537. Ejigu, M. 2006. “Post Conflict Liberia: Environmental Security as a Strategy for Sustainable Peace and Development.” Working Paper No. 3 2006. United States Agency for International Development (USAID), Washington, DC. Food and Agriculture Organization of the United Nations (FAO). 2008. The State of Food Security in the World 2008. FAO, Rome, Italy. Government of Liberia and United Nations Joint Report. 2010. The State of Food and Nutrition Security in Liberia: Comprehensive Food Security and Nutrition Survey. Monrovia. International Labour Organization (ILO). 2009. “A Rapid Impact Assessment of the Global Economic Crisis on Liberia.” Mimeo, Monrovia. Ivanic, M., and W. Martin. 2008. “Implications of Higher Global food Prices for Poverty in Low-Income Countries.” Agricultural Economics 39: 405–416. Makdissi, P., and Q. Wodon. 2002. “Consumption Dominance Curves: Testing for the Impact of Indirect Tax Reforms on Poverty.” Economics Letters 75: 227–235. Ministry of Agriculture of the Republic of Liberia. 2007. Comprehensive Assessment of the Agriculture Sector in Liberia: Volume 1: Synthesis Report. Monrovia. Minot, N., and F. Goletti. 1998. “Export Liberalization and Household Welfare: The Case of Rice in Vietnam.” American Journal of Agricultural Economics 80(4): 738–49. Niimi, Y., P. Vasudeva-Dutta, and A. L. Winters. 2004. “Storm in a Rice Bowl: Rice Reform and Poverty in Vietnam in the 1990s.” Journal of the Asia Pacific Economy 9(2): 170–190. Republic of Liberia. 2006. Comprehensive Food Security and Nutrition Survey (CFSNS). Monrovia. ———. 2008. Poverty Reduction Strategy. Monrovia. Sumarto, S., A. Suryahadi, and W. Widyanti. 2005. “Assessing the Impact of Indonesian Social Safety Net Programmes on Household Welfare and Poverty Dynamics.” European Journal of Development Research 17(1): 155–77. Timmer, C. P., and D. Dawe. 2007. “Managing Food Price Instability in Asia: A Macro Food Security Perspective.” Asian Economic Journal 21(1): 1–18. Warr, P. 2005. “Food Policy and Poverty in Indonesia: A General Equilibrium Analysis.” Australian Journal of Agricultural and Resource Economics 49(4): 429–51. Wodon, Q., and H. Zaman. 2010. “Higher Food Prices in Sub-Saharan Africa: Poverty Impact and Policy Responses.” World Bank Research Observer 25: 157–176. World Bank. 2009. Liberia: Employment and Pro-Poor Growth. Report No. 51924-LR. World Bank, Washington, DC.

CHAPTER 6

Benefit Incidence of Fiscal Measures to Deal with the Impact on Households of the Economic Crisis in Liberia: Comparing Import and Income Taxes Clarence Tsimpo and Quentin Wodon1 To help households cope with the recent economic crisis, and especially the increase in food prices, the government of Liberia announced a number of fiscal measures. A first measure was to implement a temporary exoneration of import duties on food products, and especially imported rice. A second measure announced by the President in her January 2009 State of the Union address was to reduce the personal income tax top rate from 35 percent to 25 percent together with an exclusion from paying income tax for all individuals earning less than L$54,000. This chapter provides an analysis of the consumption and income data from the nationally representative 2007 CWIQ (Core Welfare Indicator Questionnaire) survey in order to compare the likely benefit incidence of both measures. While none of the measures is well targeted to the poor, the first measure is likely to benefit the poor substantially more than the second.

1. Introduction As discussed in the previous chapter, the increase in rice prices is likely to have affected the poor substantially in Liberia, as well as in a number of other West African countries (Tsimpo and Wodon, 2012; for a rapid impact assessment of the economic crisis in Liberia, see International Labour Organization, 2009). Confronted with rapidly rising food prices, especially for cereals such as rice, many governments in the region implemented reductions in the taxes levied on foods, whether through lower import taxes or lower value added taxes (e.g., Wodon and Zaman, 2010). This was also the case in Liberia, and the issue of who might benefit from a reduction in import tax cuts for rice was already discussed in the previous chapter. The implicit assumption was that a reduction in these taxes would be passed on by intermediaries to consumers, so that the prices paid on markets would be reduced as well. Even if there were such a pass-through or trickle 99

100

A World Bank Study

down, it is not clear that a reduction in indirect taxes is good policy for helping the poor. Reductions in indirect taxes often have large budgetary costs. In addition, if a large share of the targeted food items is consumed by the non-poor, other policy instruments to help the poor cope with a crisis may have a stronger impact on poverty reduction at a lower cost. On the other hand, if much of the imported food for which a tax reduction is provided is consumed by the poor, targeting performance to the poor could potentially be good. In Liberia, in addition to reductions in import taxes on basic foods such as rice, the President announced in her January 2009 State of the Union address a series of other measures to protect households and stimulate the economy. These measures included a reduction in the personal income tax top rate from 35 percent to 25 percent and an exclusion from paying income tax for all individuals earning less than L$54,000. It could seem that this might be a somewhat neutral policy from the point of view of its benefit incidence, given provision to help both those with low incomes and the better off. Yet to the extent that much of the income taxes may be paid by the better off, it could also be that this measure would not benefit the poor very much. The objective of this chapter is to assess and compare the likely distributional implications of both measures. While in the previous chapter some benefit incidence statistics were presented for Liberia only, the analysis of the benefit incidence of the reduction in import taxes presented here was prepared as part of a broader diagnostic for a dozen West and Central African countries, and is discussed in section 2 of the chapter. The analysis of the benefit incidence of the income tax reform was written at the request of the Deputy Minister of Finance in charge of revenue collection in order to help design the details of the tax reform package and assess its potential impact on the population. The results are provided in section 3. The analysis in both sections is based on the consumption and income data from the nationally representative 2007 CWIQ (Core Welfare Indicator Questionnaire) survey. A brief conclusion follows.

2. Benefit Incidence of Taxes on Imported Foods The previous chapter already included a discussion of the likely benefit incidence of import tax cuts for rice in Liberia using the technique of consumption dominance curves of the second order. It was suggested that the poor as a whole, which represent 63.8 percent of Liberiaâ&#x20AC;&#x2122;s population (see Backiny-Yetna et al., 2012) consume approximately 45 percent of the rice consumed in the country. Here, the analysis is done in a simpler way, but comparing the results obtained for Liberia with those obtained for a dozen other West and Central African countries. Table 6.1 provides data on the consumption of imported foods for a number of West and Central African countries. The data have been collected from the most recent available household survey for each country. The survey years range from 2003 in Guinea to 2007 in Liberia, so the data can reasonably be considered as accurately capturing the current consumption patterns of the population in the respective countries. The analysis is focused on rice, flour and bread, maize, vegetable oil, sugar, and milk, because these are food items that tend to be imported to a substantial extent (although not all countries import all those goods significantly). We focus on imported foods because it is more

Poverty and the Policy Response to the Economic Crisis in Liberia

101

Table 6.1: Basic statistics and benefit incidence of indirect taxes on imported food Food item Rice Bread Vegetable oil, butter Sugar Milk Rice Palm oil Wheat Sugar Milk Rice Maize Wheat Palm oil and groundnut oil Rice Bread Flour Rice Local Rice Imported Rice Total Rice Rice Corn Wheat Rice Imported Rice local Maize Rice Vegetable oil Sucre Bread Milk Rice Rice Bread Milk Vegetable oil Sugar Rice Wheat flour and bread

Source: Authorsâ&#x20AC;&#x2122; estimations.

Share in total consumption

Share consumed by Share consumed by Proportion consumers bottom 40% bottom 60% Burkina Faso (2003 survey); base share in poverty at 46.4% 3.6 60.2 13.4 25.6 0.7 35.6 8.3 18.1 1.1 74.9 16.1 31.6 0.9 67.4 19.7 35.3 0.6 18.1 10.3 19.8 Democratic Republic of Congo (2005 survey); base share in poverty at 71.3% 3.2 57.3 15.5 31.7 4.0 96.2 19.7 36.2 1.8 35.1 7.1 17.4 1.4 57.4 10.6 24.6 0.7 23.0 4.1 11.6 Gabon (2005 survey); base share in poverty at 32.7% 3.0 91.4 31.7 51.1 0.3 40.0 14.9 31.7 3.9 93.5 27.9 46.8 1.7 90.6 30.1 48.6 Ghana (2006 survey); base share in poverty at 28.5% 3.1 74.6 16.4 33.0 1.9 84.6 14.2 29.5 0.0 2.8 45.0 60.4 Guinea (2003 survey); base share in poverty at 49.1% 13.0 90.7 23.1 42.8 Liberia (2007 survey); base share in poverty at 63.8% 9.6 60.1 27.5 47.8 13.2 84.9 22.3 41.2 22.8 99.0 24.5 44.0 Mali (2006 survey); base share in poverty at 47.5% 7.2 95.1 11.1 25.1 4.2 91.0 14.4 33.1 1.5 74.0 19.5 36.7 Niger (2005 survey); base share in poverty at 62.1% 4.4 54.7 14.8 31.4 1.7 15.4 20.1 35.9 4.3 30.4 18.2 34.3 Senegal (2006 survey); base share in poverty at 50.8% 6.8 96.3 28.0 47.9 4.5 95.8 22.8 42.1 3.0 99.2 27.1 46.6 4.0 92.7 14.8 32.6 2.1 79.6 10.0 23.4 Sierra Leone (2003 survey); base share in poverty at 66.4% 11.7 96.4 32.0 53.9 Togo (2006 survey); base share in poverty at 61.6% 3.5 92.2 23.0 40.4 0.6 27.0 5.8 15.5 0.7 31.1 7.6 18.4 1.1 81.3 21.3 39.5 0.7 72.3 20.1 36.7 Nigeria (2004 survey); base share in poverty at 54.7% 4.1 73.4 14.0 30.2 1.5 70.4 12.5 27.0

102

A World Bank Study

likely that these food items are taxed, so that it is feasible for governments to indeed reduce these taxes and hope that this will have a downward impact on market prices. By contrast locally produced foods are often not taxed, or at least not taxed to the same extent because they are largely auto-consumed, and when they are sold, this is typically done through transactions taking place through the informal sector which tends to escape taxation. At least two variables are often used by policy makers to assess to what extent a shock in the price of food items are likely to have a large effect on the standard of living of the population, and thereby to determine whether it is necessary for the government to reduce taxes on these food items. The first variable is the share in total consumption represented by the items. This information is provided in the second column of the table. The larger the share is, the more likely it will be that a government will feel pressure to reduce the tax on the good in a time of food price crisis. There are large differences between countries in the extent to which various food items are consumed. For example, rice accounts for less than 5 percent of total consumption in Burkina Faso, the Democratic Republic of Congo, Gabon, Ghana, Niger, Nigeria, and Togo, but it accounts for between 5 percent and 10 percent of total consumption in Mali, and Senegal, and for more than 10 percent of total consumption in Liberia and Sierra Leone. A second important piece of information is the share of the population that is likely to be affected by the price shock. This share maters from a political economy point of view because when a larger share of the population is affected, it is more likely that policy makers will be under pressure to respond to the crisis. The information is provided in the third column of the table. Considering again rice, we see that in many countries more than 90 percent of the population consumed the good (this is the case in Gabon, Guinea, Liberia, Mali, Senegal, Sierra Leone, and Togo), and the proportion remains high in other countries (the minimum share of the population consuming rice is 57 percent in the Democratic Republic of Congo). For other imported foods such as bread, sugar, or milk, the proportion is lower on average, although bread and sugar in some countries are consumed by a very large share of the population. What matters more for poverty reduction though is the share of a goodâ&#x20AC;&#x2122;s consumption that is accounted for by the poor in the population. The share of the population that is poor varies between countries (from 28.5 percent in Ghana to 71.3 percent in the Democratic Republic of Congo according to recent poverty assessments completed at the World Bank), so that for cross-country comparisons, it is easier to consider the share of total consumption accounted for by the bottom 40 percent or 60 percent of the population (these two proportions were chosen because for most countries, the poverty rate falls between these two values). Consider first the share of food consumption in the bottom 40 percent. For rice, this share varies from 11.1 percent in Mali to 32.0 percent in Sierra Leone. This means that if we consider the bottom 40 percent as the poor, out of every dollar spent by a government for reducing indirect taxes on rice, only about 20 cents on average will benefit the poor. This is a rather low proportion, and it also assumes that tax reductions do trickle down

Poverty and the Policy Response to the Economic Crisis in Liberia

103

to lower prices for consumers, which is not necessarily obvious. If we consider that the bottom 60 percent of the population can be considered as poor, the share of subsidies or tax reductions that would reach the poor would be between 25.1 percent and 53.9 percent, which still does not suggest good targeting. For some of the other goods listed in table 6.1, the proportions are even lower. In the case of Liberia, 44.0 percent of the benefits of a reduction in taxes on rice would accrue to the 60 poorest percent of the population, with the proportion being slightly lower if we consider only imported rice. This is one of the higher proportions among all countries considered in table 6.1, but it remains relatively week as targeting performance since the poor, who account for almost two thirds of the population, would benefit from less than half of the potential reduction in rice prices brought about by a reduction in import taxes (assuming trickle down effects).

3. Benefit Incidence of Proposed Income Tax Reform As mentioned in the introduction, another measure proposed by the government was to reduce income taxes. Total revenue from the personal income tax was US$17.3 million in 2007. The tax brackets and tax rates for the personal income tax were set in 2000, and had not been adjusted for inflation since then. This implies that tax rates had increased in real terms over time since 2000 since inflation had likely led a higher number of individuals to move up in the tax brackets, all other things being equal. The marginal tax rate structure in Liberian dollars at the time the President announced the measure is given in table 6.2. In this section, we will simulate one alternative tax structure, as indicated in table 6.2, but it would be easy to simulate alternative tax reforms. The tax reform proposed by Liberia’s President includes an exemption for lower tax brackets, and a reduction in the highest tax rate. The simulation in table 6.2 extends this reduction in tax rates to other brackets, and due to the exemption, the number of tax brackets is also reduced, which helps in simplifying the tax structure. Table 6.2: Marginal tax rate structure, Liberian dollars, 2007 Current structure (baseline)

Simulated

Brackets

Rate

Brackets

Rate

1–12,000

1—12,000

12,001–50,000

240+5% for revenues over 12,000

12,001–54,000

50,001–100,000

2,140+10%

50,001–100,000

5% excess over 54,000

100,001–200,000

7,140+15%

100,001–200,000

2,140+10%

200,001–400,000

22,140+20%

200,001–400,000

7,140+15%

400,001–800,000

62,140+25%

400,001–800,000

22,140+20%

800,001–1,200,000

162,140+30%

>800,000

62,140+25%

>1,200,000

282,140+35%

Source: Government of Liberia for baseline.

104

A World Bank Study

The personal income tax (PIT) is computed on individual income, not household income, although it is not permissible for household members to split their income between several individuals in order to pay lower taxes. Taxable income is computed as gross income minus exclusions and deductions. Gross income is the sum of all income sources of the individual during a given tax year. Income sources include (but are not limited to): (i) earnings from employment, including noncash benefits (such as medical costs paid by an employer or insurance benefits, although there is an after tax credit that provides some relief); (ii) receipts from the operation of a business (noting that the business itself will be taxed at 25 percent of taxable income if it has turnover greater than L$5 million or 2 percent over turnover it has turnover of less than L$5 million); (iii) interests, rents, royalties and dividends; (iv) distributions form a trust or estate; and (v) 100 percent of gains on the disposition of property if used in a business or held for investment, or all amounts over L$1.6 million if held for personal use. Income sources that are not subject to tax (exclusions) include (i) sickness, disability or death benefits; (ii) property received in a donative transfer or transfer by death; (iii) interest accruing from tax-exempt obligations issued by Liberia (not applicable today); (iv) gains on the sale of “personal-use” property up to L$1.6 million; and (v) interest of less than L$200 per year. Deductions from the individual’s tax base include: (i) the cost of producing income but not the costs of personal consumption, Liberian, or foreign income tax, interest relating to any Liberian tax, or any fines or penalties imposed by law; and (ii) charitable contributions to the government or registered charities. In order to use the CWIQ survey for analyzing the benefit incidence of the proposed reform, it is first necessary to verify that the actual income tax collection in the country can be approximated using data from the survey. The CWIQ survey does not have information on individual incomes, but it has a relatively detailed household income module. All estimations and simulations presented here are based therefore on household rather than individual income. This is an approximation, but it is not too damaging because in many households, there is only one individual earning significant income. In countries such as Liberia, most individual are in practice not paying taxes because they have rather small earnings if any (a substantial share of the population makes its livelihood from subsistence agriculture). Many individuals who may have substantial income sources also work in the informal sector, so that it is unlikely that they are taxes in a substantial way. Therefore, we will consider in the survey as taxable income the income of households who have at least one member working in the formal sector, as identified through the survey. The distribution of what we consider as household taxable income in the survey is provided in table 6.3. In table 6.3, the sample data is based on simple tabulations of the number of observations by category without weights. The weighted data takes into account the survey weight, and therefore represents the distribution in the population as a whole. Almost three fourth of households do not have taxable income (either because they don’t have any recorded income, or more often because they do not have household members working in the formal sector). Only about three percent of the population has taxable income above L$100,000.

Poverty and the Policy Response to the Economic Crisis in Liberia

105

Table 6.3: Distribution of households across taxable income group, Liberian dollars, 2007 Sample Income group

Frequency

Weighted Cumulative

Frequency

Cumulative

71.87

Full sample 0

2,495

69.40

357,319

1–12,000

422

11.74

81.14

57,560

11.58

83.45

12,001–50,000

414

11.52

92.66

51,102

10.28

93.73

50,001–100,000

147

4.09

96.75

16,621

3.34

97.07

100,001–200,000

1.61

98.36

6,414

1.29

98.36

200,001–400,000

1.17

99.53

5,546

1.12

99.48

400,001–800,000

0.39

99.92

1,824

0.37

99.84

800,001–1,200,000

0.06

99.97

442

0.09

99.93

>1,200,000

0.03

100.00

331

0.07

100.00

3,595

100.00

497,159

100.00

Total

Sample with positive earnings from formal sector 1–12,000

422

38.36

57,560

41.16

12,001–50,000

414

37.64

76.00

51,102

36.54

77.70

50,001–100,000

147

13.36

89.36

16,621

11.89

89.59

100,001–200,000

5.27

94.64

6,414

4.59

94.18

200,001–400,000

3.82

98.45

5,546

3.97

98.14

400,001–800,000

1.27

99.73

1,824

1.30

99.45

0.18

99.91

442

0.32

99.76

0.09

100.00

331

0.24

100.00

1,100

100.00

139,839

100.00

800,001–1,200,000 >1,200,000 Total

Source: Authors’ estimation using 2007 CWIQ survey.

Table 6.4 provides data on mean and total taxable income. Total taxable income is estimated at L$1,080,474,628. Using an exchange rate of 0.015873 to the dollar, this is equivalent to US$17.15 million, which is very close to the actual tax receipts of the government from the personal income tax. This suggests that at least in first approximation, the CWIQ survey can be used for simulating tax reforms. On the other hand, the number of observations in the upper tax brackets is extremely small (one observation in the top bracket, and two in the bracket just below, as shown in table 6.4), so that the analysis is very sensitive to these observations. In addition, and in part to confirm adequacy of an analysis based on so few observations, it would be useful to know the distribution of the personal tax income receipts by tax bracket and compare this distribution to the CWIQ data. Unfortunately such data on the distribution of income by tax bracket are not available from the Ministry of Finance. The analysis presented in this section should therefore be considered with a lot of caution, as results could have been fairly different from another survey as valid as this one.

106

A World Bank Study

Table 6.4: Mean and total sum value of taxable income by residence area and tax bracket, 2007 Net income

Gross income

Tax

Mean value Location Rural

4,054

4,316

262

Urban

34,859

41,167

6,308

Total

13,790

15,963

2,173

4,262

4,349

Tax brackets 1–12,000 12,001–50,000

27,242

28,297

1,055

50,001–100,000

66,841

71,090

4,249

100,001–200,000

130,652

144,460

13,809

200,001–400,000

232,315

268,069

35,755

400,001–800,000

423,388

514,037

90,649

800,001–1,200,000

676,126

854,665

178,540

2,750,000

4,018,677

1,268,677

15,963

2,173

>1,200,000 Total

13,790

Total sum value

Location Rural

1,378,416,450

1,467,662,108

89,245,658

Urban

5,477,454,208

6,468,683,178

991,228,970

Total

6,855,870,659

7,936,345,287

1,080,474,628

245,324,889

250,331,522

5,006,633

Tax brackets 1–12,000 12,001–50,000

1,392,115,706

1,446,020,084

53,904,378

50,001–100,000

1,110,959,999

1,181,582,520

70,622,520

100,001–200,000

838,001,592

926,573,395

88,571,803

200,001–400,000

1,288,393,528

1,486,679,947

198,286,419

400,001–800,000

772,350,725

937,714,521

165,363,796

800,001–1,200,000

298,620,832

377,475,463

78,854,631

>1,200,000

9101,03,386

1,329,967,834

419,864,449

6,855,870,659

7,936,345,287

1,080,474,628

Total

Source: Authors’ estimation using 2007 CWIQ survey.

Table 6.5 provides data on the share of taxable income by location and income tax brackets. As expected, most of the taxes (91.7 percent) are paid by urban households. Although few households have high level of incomes, those households contribute to a large share of the tax base (but again, key results are based on an extremely small sample of households in upper income tax brackets).

Poverty and the Policy Response to the Economic Crisis in Liberia

107

Table 6.5: Contribution of areas/income group in total personal tax income receipts, 2007 Net income

Gross income

Tax

Rural

20.1

18.5

8.3

Urban

79.9

81.5

91.7

Total

100.0

Location

Tax brackets 1–12,000

3.6

3.2

0.5

12,001–50,000

20.3

18.2

5.0

50,001–100,000

16.2

14.9

6.5

100,001–200,000

12.2

11.7

8.2

200,001–400,000

18.8

18.7

18.4

400,001–800,000

11.3

11.8

15.3

800,001–1,200,000 >1,200,000 Total

4.4

4.8

7.3

13.3

16.8

38.9

100.0

Source: Authors’ estimation using 2007 CWIQ survey.

In tables 6.6 and 6.7, we provide results from the simple simulation indicated in table 6.2. This is done not to suggest or recommend any particular policy, but only to indicate the type of simulations that can be performed. Many other simulations could be implemented, and it would be easy to assess the impact of tax reforms on income inequality, as well as poverty by combining the information in the survey from the income module and the data on consumption. Table 6.6 provides the mean value and total taxable income under the baseline and simulated tax reform. The impact of the simulated reform on total taxes could be large, with a reduction in taxes of almost half. Because upper income bracket households pay a large share of the taxes, they would benefit from a larger share of the reductions in taxes. Table 6.7 provides the share of taxable income by decile under the baseline and simulated tax reform. Even though the simulated tax reform includes an exemption from paying taxes for individuals/households with less than L$ 54,000 in total income, the impact of the overall reform still reduces the share of total after tax income that benefits the lower brackets of the income distribution. In that sense, there is a significant risk that the tax reform would be regressive. All those results should be considered as tentative, given that the sample sizes on which some of the estimates are based remains very small (and extremely so in the upper brackets). But in terms of comparing the benefit incidence of the proposed income tax reform with that of the temporary cut on rice import taxes, the messages tend to be relatively clear.

108

A World Bank Study

Table 6.6: Mean value and total taxable income under the baseline and simulated tax reform, 2007 Net income Baseline Y after tax

Tax

Simulated Y after tax

Gross income

Baseline tax

Simulated tax

Mean values Tax bracket 1–12,000

4,262

4,349

12,001–50,000

27,242

28,297

1,055

50,001–100,000

66,841

70,217

71,090

4,249

873

100,001–200,000

130,652

137,875

144,461

13,809

6,586

200,001–400,000

232,315

250,719

268,069

35,754

17,350

400,001–800,000

423,388

469,090

514,037

90,649

44,947

800,001–1,200,000

676,126

778,859

854,665

178,540

75,806

2,750,000

3,151,868

4,018,677

1,268,677

866,809

13,790

14,846

15,963

2,173

1,117

>1,200,000 Total

Total sum Tax bracket 1–12,000

245,324,889

250,331,522

5,006,633

12,001–50,000

1,392,115,706

1,446,020,084

53,904,378

50,001–100,000

1,110,959,999

1,167,072,102

1,181,582,520

70,622,520

14,510,421

100,001–200,000

838,001,592

884,330,262

926,573,395

88,571,803

42,243,136

200,001–400,000

1,288,393,528

1,390,456,969

1,486,679,947

198,286,419

96,222,978

400,001–800,000

772,350,725

855,720,778

937,714,521

165,363,796

81,993,740

800,001–1,200,000

298,620,832

343,994,495

377,475,463

78,854,631

33,480,972

>1,200,000

910,103,386

1,043,100,186

1,329,967,834

419,864,448

286,867,649

6,855,870,659

7,381,026,399

7,936,345,287

1,080,474,628

555,318,896

Total

Source: Authors’ estimation using 2007 CWIQ survey.

Table 6.7: Share of taxable income by decile under baseline and simulated tax reform, 2007 Net income

Tax

Deciles

Baseline Y after tax

Simulated Y after tax

Gross income

Baseline tax

1.29

1.25

1.16

0.33

0.04

2.67

2.62

2.50

1.45

0.95

2.61

2.52

2.37

0.87

0.32

Simulated tax

4.57

4.45

4.25

2.25

1.55

5.01

4.88

4.67

2.55

1.85

5.37

5.22

4.92

2.08

0.98

6.80

6.66

6.34

3.45

2.15

12.95

12.90

12.54

9.93

7.73

14.42

14.21

13.86

10.26

9.16

44.31

45.28

47.38

66.83

75.26

Total

100.00

Source: Authors’ estimation using 2007 CWIQ survey.

Poverty and the Policy Response to the Economic Crisis in Liberia

109

4. Conclusion This chapter has provided simple estimates of the likely benefit incidence of indirect tax reforms for selected food items in a dozen West and Central African countries. Reducing import taxes or the VAT on food imports is one of the first actions that governments are considering to reduce the impact on the poor of rising food prices. Yet while this is a simple measure to take administratively, it is also costly in budgetary terms, and it is not clear that it necessarily reach the poor. In the case of Liberia, the results suggest that 44 percent of the benefits of reduced taxes on imported rice may have benefitted the bottom 60 percent of the population, with the proportion being reduced slightly further if one considers only the consumption of imported rice. In addition, there is no guarantee that the tax cuts will end up reducing the market prices of the goods targeted. Finally, for many food items, even if there is a one-to-one relationship between taxes and market prices, much of the benefit of the tax cuts could be enjoyed by the non-poor. This does not mean that no tax cuts should not be implemented, but rather than one needs to look closely at the country level data before making decisions. In a country such as Liberia, where a substantial share of the rice is consumed by the poor, a tax cut may make more sense that in some of the Sahelian countries such as Burkina Faso, Mali, and Niger where rice tends to be consume more by better off households residing in urban towns. But even in Liberia, the targeting performance of indirect tax cuts, assuming that they do indeed trickle down to lower the prices for consumers on the markets, remains relatively weak. The chapter has also provided an analysis of household survey data in order to provide an analysis of the benefit incidence of a proposed a personal income tax reform. All results should be considered as tentative, given the small size of the sample used for the estimation in some of the income tax brackets, especially at the upper end. Nevertheless, the analysis suggests that there is an even more significant risk than was the case for import tax cuts on rice that the income tax reform would be regressive. Given that there are plans to adjust other taxes in Liberia, including the sales tax, it could make sense to propose an integrated reform package dealing with the various taxes in a more comprehensive and integrated way to avoid that much of the benefits of the tax reform accrue principally to the better off, especially at a time of crisis.

References Backiny-Yetna, P., R. Mungai, C. Tsimpo, and Q. Wodon. 2012. “Poverty in Liberia: Level, Profile, and Determinants.” In Q. Wodon, ed., 9–34. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. International Labour Organization (ILO). 2009. “A Rapid Impact Assessment of the Global Economic Crisis on Liberia.” Mimeo, Monrovia.

110

A World Bank Study

Tsimpo, C., and Q. Wodon. 2012. “Rice Prices and Poverty in Liberia.” In Q. Wodon, ed., 85–98. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Wodon, Q., and H. Zaman. 2010. “Higher Food Prices in Sub-Saharan Africa: Poverty Impact and Policy Responses.” World Bank Research Observer 25: 157–176.

PART III Evaluation of the Cash for Work Temporary Employment Program

111

CHAPTER 7

Ex Ante Assessment of the Potential Impact of LaborIntensive Public Works in Liberia Clarence Tsimpo, Quentin Wodon, and Errol Graham1 Apart from fiscal measures, another initiative taken by the government of Liberia to respond to the economic crisis consisted in the launch of a cash for work temporary employment program. This part of the study consists of three chapters devoted to the analysis of the program. First, in this chapter which was written before the program was actually implemented, we provide an ex ante analysis of the potential impact of such a program, relying on simulation techniques rather than on impact evaluation. The approach is very simple. We assess who may be potentially interested in participating in the public works program by identifying working individuals without pay, as well as for every level of proposed wage in the public works, those individuals who work but now earn less than the public works wage, since all these individuals may indeed be interested in participating in the program to increase their earnings. We also consider as potential beneficiaries the unemployed whose reservation wage is below the proposed public works wage. Next, we randomly select among the pool of potential beneficiaries of the program a number of participants. Finally, we estimate for the assumed participants to the program two key parameters which affect the potential impact of the program on the poor: the targeting performance of the program, and the substitution effect of the program, whereby only part of the wages paid to beneficiaries generate additional income, because at least some of the beneficiaries would probably have done other work if they had not participated in the program. The results suggest that such a cash for work program could be well targeted, but that this is by no means assured ex ante.

1. Introduction Youth unemployment and underemployment is a major issue in sub-Saharan Africa as in many other areas of the developing world (World Bank, 2007a). In many African countries, children and youth represent up to 40 percent of the population. Thanks to programs such as the Education for All initiative, school enrolment rates are rapidly increasing, but many youth remain out of school, and are often without work or with work that do not build their skills.2 113

114

A World Bank Study

As discussed among others in World Bank (2009) and Backiny-Yetna et al. (2012), Liberia is amongst the poorest countries in the world, with a GDP per capita of less than US$200, 64 percent of the population in poverty and nearly 48 percent in extreme poverty. Poverty is especially high in rural areas but is also widespread in urban areas. Liberia’s labor force is growing rapidly due to population growth, and is for the most part unskilled due to 14 years of civil conflict. The supply of workers in the economy exceeds by a substantial margin the existing demand for workers. Due to the combined effect of unemployment, underemployment, and low productivity work for many workers, the number one priority for the government in the opinion of the population should be to create employment. The Government of Liberia has outlined a Poverty Reduction Strategy that articulates the country’s vision and major strategies for moving towards rapid, inclusive, and sustainable growth and development during the period 2008-2011 (Republic of Liberia, 2008). The Government has indicated that growth will be private sector-led, while the government will focus on reforming public sector institutions and processes to facilitate investment and strengthen market functions. Yet the Liberian economy’s ability to create jobs in the short-to-medium-term has been adversely impacted by the confluence of the three global crises—the food crisis, the financial crisis and the commodity crisis. The crises have further compressed fiscal space, thereby limiting the government’s ability to respond with counter-cyclical fiscal policy. In line with this agenda, an as a response to the global crises, the Government of Liberia is implementing a labor intensive program. The International Labor Organization (ILO) defines employment-intensive or laborintensive projects as those projects where labor is the dominant resource. When considering labor intensive public works to build infrastructure, a first question for policy makers is whether using this type of programs has a negative impact on the infrastructure built, as opposed to using equipment–intensive techniques. Over time, the provision of infrastructure in many countries has shifted from being predominantly labor-based to equipment-based. This shift has been particularly dramatic for developed countries where wage rates have been increasing. However, the shift is also taking place in some developing countries. Nevertheless, work done by the World Bank and the ILO has shown that for countries which are facing strong demand for infrastructure in the face of significant unemployment, labor-based provision of infrastructure remains a viable alternative to equipment-based provision of infrastructure. Labor-intensive employment programs including public works can help Liberia not only to rebuild social and economic assets quickly, but they could also buy crucial time until the private sector expands, and the diversification strategy takes root to allow the economy to absorb a larger proportion of the labor force at reasonable wages. Laborintensive programs should not be seen as a single “silver bullet” but as one of the elements of a comprehensive strategy which has short-term, medium-term and long-term elements to address the issue of the lack of gainful employment in Liberia. At the same time, labor-intensive public works have advantage over other strategies for infrastructure building and employment creation in Liberia. Cross country experience with labor-intensive initiatives shows that the welfare impacts have generally been positive but the results have been mixed in terms of the quality and sustainability of the assets produced. Public employment programs were

Poverty and the Policy Response to the Economic Crisis in Liberia

115

pioneered in South Asia to deal with huge open unemployment. The Maharashtra Employment Guarantee Scheme (MEGS) is known as an effective safety net for the poor in India. Bangladesh has used public employment programs since 1962, largely financed by external donors. In Sri Lanka, labor-intensive public works were used to cushion the adverse effects of structural adjustments. In Latin America, the outcome is equally positive. According to Subbarao (2003), nearly 100 percent of the participants in Chileâ&#x20AC;&#x2122;s public works program belonged to poor households. In Argentinaâ&#x20AC;&#x2122;s Trabajar program, 60 to 70 percent of households participating were poor. In Sub-Saharan Africa, labor-intensive public works programs have been implemented both as free-standing programs as well as components of Social Fund programs. In general, stakeholdersâ&#x20AC;&#x2122; views have been positive on the income and the capacity building impact of these programs. Stakeholders were particularly pleased with the speed with which the jobs were created although they were temporary. The public works program in South Africa, which is considered to be one of the most innovative, has multiple objectives including job creation, poverty reduction, infrastructure development, job training and community capacity building (Adato and Haddad, 2001). What the cross-country experiences show is that projects or programs in support of labor-intensive work focus on a range of assets including roads (mostly rural but also urban), markets, schools, health centers, urban drainage systems, water supply systems, irrigation systems, reforestation, anti-erosion structures, land reclamation, housing, and solid waste management. Roads tend to be the most popular asset of choice for laborintensive public works program across the world. Whichever assets are chosen, a critical success factor in the implementation and sustainability of the project is the ownership by the communities. The cross-country experience clearly shows that those assets which are demand driven and reflect the choice of the communities are more likely to be better implemented and are more sustainable that those assets which are supply driven, even when the communities benefit from the jobs that are created. Social Funds have been successful at encouraging community participation. They have used different participatory tools to get communities involved in deciding priority projects, their location, design supervision of implementation and the maintenance of the projects. There has also been recent successful experimentation with community contracting (for example: the Jamaica Social Investment Fund, the Malawi Social Action Fund and Bolivia Social Fund). Beneficiary participation not only builds ownership of project but it is also an essential component of good governance. The launch of a public works program in Liberia may appear to be a sound idea in order to help youth find employment and improve their skills. Indeed, according to lessons from a Youth Employment Inventory of 289 programs and interventions from 84 countries recently carried out by the World Bank (2007b), public works and training program are more suitable than formal sector wage subsidy programs for youth in developing countries, since wage subsidies do not go far in developing countries due to the small size of the formal wage sector and also do not reach the poor. Public works and training programs are also more likely to succeed than targeted youth entrepreneurship schemes. This is because while these schemes may improve opportunities for young entrepreneurs in low-income countries where job growth in the formal economy tends to be rare, the evidence indicates that not all youth will be well suited for self-employment and that failures rates for young entrepreneurs can be high.

116

A World Bank Study

However, careful targeting and screening for these programs is important to success and cost-effectiveness, and it may well be that training programs are substantially more expensive than public works program, especially if the training programs target relatively better educated workers and pay a high wage for the period of training. Training programs are also more successful when they involve the private sector in providing practical work experience and in identifying the kind of skills required. Engagement of the private sector in training is an effective tool to mitigate the risk of high-cost training disconnected from market demand and to increase on-the-job training. In this chapter, which was originally written before the implementation of Liberiaâ&#x20AC;&#x2122;s cash for work temporary employment project, our objective was to provide some policy guidance to the Government of Liberia in its development of a more strategic approach towards achieving its pro-poor growth objectives and specifically in implementing a public works program. To provide a preliminary assessment of the potential impact of a public works program on poverty (on the impact of public works on poverty in developing countries, see among others Ravallion, 1999), we rely in this chapter on simulation techniques rather than on impact evaluation techniques. The approach is very simple. We assess who may be potentially interested in participating in the public works program by identifying working individuals without pay, as well as for every level of proposed wage in the public works, those individuals who work but now earn less than the public works wage, since all these individuals may indeed be interested in participating in the program to increase their earnings. We also consider as potential beneficiaries the unemployed whose reservation wage is below the proposed public works wage. Next, we randomly select among the pool of potential beneficiaries of the program a number of participants. Finally, we estimate for the assumed participants to the program two key parameters which affect the potential impact of the program on the poor: the targeting performance of the program, and the substitution effect of the program, whereby only part of the wages paid to beneficiaries generate additional income, because at least some of the beneficiaries would probably have done other work if they had not participated in the program. The chapter is organized as follows. In Section 2, using recent household survey data we provide data on the potential demand for public works programs by looking at the number of youths who are either not working but willing to work (the unemployed), or are working but with a level of pay that is below what the program provides. In section 3, we simulate the potential impact on poverty of the program through the payment of wages to the participating youths (we deliberately do not consider the additional impact which may come from the training component of the program since we do not have data to estimate its impact). The simulations take into account the likely targeting performance of the program, as well as the likely substitution effects. A conclusion follows.

2. Potential Demand for Employment Programs To provide an assessment of the potential impact of labor-intensive public works on poverty in Liberia, we rely on simulation techniques using the 2007 CWIQ survey. In a similar way to Coulombe et al. (2008), the approach begins with an assessment of who may be potentially interested in participating in the program by identifying working individuals without pay, as well as for every level of the proposed wage in the program,

Poverty and the Policy Response to the Economic Crisis in Liberia

117

those individuals who work but now earns less than the program wage, since all these individuals may indeed be interested in participating in the program to increase their earnings. The unemployed whose reservation wage is likely to be below the proposed program wage are also considered as potential beneficiaries. Next, we randomly select among the pool of potential beneficiaries of the program a number of participants. Finally, we estimate for the assumed participants in the program a leakage rate which represents the share of program outlays that do not directly contribute to poverty reduction. This leakage rate depends on two key parameters: (i) the targeting performance of the program, and (ii) the substitution effect of the program, whereby only part of the wages paid to beneficiaries generates additional income, because beneficiaries would probably have done other work if they had not participated in the program. Our simulations for the impact of public works on poverty are based on the assumption of a 50 percent substitution effect, so that program participants give up half their current earnings to participate in the public works program which is assumed to take place in the lean season (this may be a high substitution effect given lack of gainful employment in Liberia). Using the 2007 Liberia CWIQ Survey, we provide in this section estimates of the number of youths aged 20 to 40 who could be interested by a national youth employment program. The 2007 Liberia CWIQ Survey did not collect the information on individual wages. Instead, the income section considered wages for the household as a hole. Given the absence of individual data on wages, some assumption has to be made in order to estimate the actual wage of individual and to derive the reservation wage for those unemployed. For each household, the overall household wage is divided by the number wage earner to compute the individual wage. Then, a regression model was estimated to impute the reservation wage for the unemployed. Figure 7.1 gives the distributions of both the actual and the estimated wage. These estimates of wages are critical for the current assessment.

Figure 7.1: Distribution of actual wage and imputed wage, 2007 1.2 Actual wages Imputeed wagesâ&#x20AC;&#x201D;employed

Imputeed wagesâ&#x20AC;&#x201D;unemployed Minimum waqe

Density

0.8

Only one employed

0.6 0.4 0.2 0

5 6 Log of annual wage (US$)

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ data.

118

A World Bank Study

Table 7.1 provides data on the distribution of earnings of individuals who are already working, as well as on the distribution of the imputed reservation wage for individuals who are unemployed and looking for work. The groups of individuals are presented in the first column of the table in terms of their annual wages in U.S. dollar. Table 7.1 enables us assess the potential population that could be interested in a job in a public works program without eligibility condition in terms of education, gender, etc. Consider first the statistics provided in table 7.1. We see for example that there is a very large group of youth who are working but are paid less than half of the minimum wage (52 percent of the youths who are working at the national level are paid less than US$240/year). These individuals are likely to be interested in public works. Clearly, some may not apply for such a program due to various constraints (they may not be paid, but still doing important work that has to be done for their household, and hence they may not be able to participate in the program). Also, depending on the wage paid by public works, additional individuals could be interested in participating in the program if their current wage is below that proposed by the program. We cannot identify those who would actually be interested and those who would not. But for the purpose of the simulations in the next section, all the individuals unpaid for their work, as well as all individuals who earn less than the proposed wage are potential beneficiaries of the program, and we can randomly chose some of these individuals as participants in public works for each proposed wage level in order to simulate the impact of the program on poverty. Finally, among the unemployed, those who have a reservation wage below the proposed wage would also be potential beneficiaries. The estimates in table 7.1 therefore give us an upper bound for the potential number of youths that might be interested in a public works program, depending on the wage provided in the program, and without any eligibility condition as it may be proxied by the gender, the education level or other individual characteristic. Figures 7.2 to 7.4 summarize the data on the potential number of participants by quintiles of per capita consumption of the households to whom the individuals who are potential beneficiaries belong. This is done for three potential wage levels, from US$240 per year to US$720 per year. Two findings stand out from the results presented in figures 7.2 to 7.4. First, the number of individuals who could potentially be interested in the program appears to be very large, especially because many workers are working with low pay (observed or imputed) and might therefore be interested in getting higher cash income through public works. Second, the targeting performance or likely benefit incidence of the program depends on whether the program is implemented mostly in urban or rural areas. In urban areas, the program would probably be regressive, since most of the potential beneficiaries belong to the better off quintiles of the population (this is because urban households tend to have higher levels of consumption than rural households, so that relatively few households in urban areas belong to the bottom quintiles). By contrast, the programs could be well targeted to individuals belonging to households which tend to be poor if the focus is placed on providing employment and reconstructing infrastructure in rural areas. There is also a clear relationship between the wage level of workers and the poverty status of households.

32,377.7 14,513.8 21,253.4 57,138.7 181,690.0 121,557.0 45,818.3 24,594.4 14,947.0 14,284.1 52,966.0 581,140.0

15,180.8 6,549.9 10,934.8 11,978.5 21,755.6 24,687.8 13,060.1 8,350.6 6,130.0 4,182.8 28,967.2 151,778.0

17,196.9 7,963.9 10,318.6 45,160.3 159,934.0 96,868.9 32,758.2 16,243.8 8,817.0 10,101.4 23,998.8 429,362.0

% group

5.6 2.5 3.7 9.8 31.3 20.9 7.9 4.2 2.6 2.5 9.1 100.0

10.0 4.3 7.2 7.9 14.3 16.3 8.6 5.5 4.0 2.8 19.1 100.0

4.0 1.9 2.4 10.5 37.2 22.6 7.6 3.8 2.1 2.4 5.6 100.0

Annual wage 1,299.7 4,065.6 6,133.8 8,718.1 12,542.0 17,298.4 22,085.9 27,299.4 32,710.8 36,989.1 84,797.8 21,170.7 1,073.7 4,177.0 6,210.3 8,590.9 12,967.1 17,471.7 22,407.1 27,383.2 32,673.7 37,542.8 101,128.6 31,203.2 1,499.3 3,974.1 6,052.6 8,751.8 12,484.2 17,254.3 21,957.9 27,256.3 32,736.5 36,759.8 65,086.0 17,624.2

Wage of workers

Source: Authors’ estimation using 2007 CWIQ data.

Liberia < $42 $42–$70 $70–$98 $98–$140 $140–$210 $210–$280 $280–$350 $350–$420 $420–$490 $490–$560 $560+ Total Liberia Urban < $42 $42–$70 $70–$98 $98–$140 $140–$210 $210–$280 $280–$350 $350–$420 $420–$490 $490–$560 $560+ Total Urban Rural < $42 $42–$70 $70–$98 $98–$140 $140–$210 $210–$280 $280–$350 $350–$420 $420–$490 $490–$560 $560+ Total Rural

# of people in group

47.5 47.1 48.4 38.7 43.6 47.7 49.4 44.2 45.5 51.3 46.6 45.2

42.6 54.3 45.9 45.9 45.6 43.9 46.9 47.1 47.5 50.9 51.0 46.9

45.2 50.3 47.1 40.2 43.8 46.9 48.7 45.2 46.4 51.2 49.0 45.6

Weekly hours

87.8 67.6 84.3 67.7 67.8 63.3 66.5 68.2 40.5 65.2 37.2 65.6

56.7 51.3 75.3 65.4 47.8 51.3 41.8 39.1 36.9 38.0 21.6 46.1

73.2 60.2 79.7 67.2 65.4 60.9 59.4 58.3 39.1 57.2 28.7 60.5

% poor

% group — 0.2 3.7 10.3 31.3 31.6 16.1 3.6 2.4 0.2 0.7 100.0 — 0.3 6.3 11.6 16.6 31.8 22.3 5.6 3.9 0.3 1.2 100.0 — — 0.4 8.7 49.9 31.4 8.2 1.0 0.5 — — 100.0 — — 132.4 3,095.5 17,840.7 11,250.4 2,922.5 371.3 168.4 — — 35,781.2

— 129.7 2,844.3 5,250.7 7,521.1 14,361.2 10,100.5 2,546.7 1,761.7 147.8 560.0 45,223.6

— 129.7 2,976.8 8,346.2 25,361.8 25,611.7 13,023.0 2,918.0 1,930.0 147.8 560.0 81,004.8

— — 6,643.9 8,893.0 12,457.3 16,977.9 22,008.5 25,941.7 33,289.5 — — 14,566.9

— 4,687.2 6,408.0 8,554.7 13,092.7 17,431.6 22,361.1 26,523.1 32,229.9 35,300.8 43,864.2 17,524.6

— 4,687.2 6,418.5 8,680.1 12,645.7 17,232.3 22,281.9 26,449.2 32,322.3 35,300.8 43,864.2 16,218.1

— — — — — — — — — — — —

Unemployed reservation wage # of people in Imputed annual group wage Weekly hours

Table 7.1: Potential beneficiaries of public works among individuals aged 20-40, national, 2007

— — 100.0 49.0 61.3 70.3 88.5 92.6 0.0 — — 65.5

— 100.0 68.6 64.9 71.6 60.3 50.1 68.5 40.1 100.0 42.7 60.7

— 100.0 70.0 59.0 64.4 64.7 58.7 71.6 36.6 100.0 42.7 62.8

% poor

Poverty and the Policy Response to the Economic Crisis in Liberia 119

120

A World Bank Study

Figure 7.2: Distribution of potential beneficiaries of public works, national, 2007

Number of beneficiaries, in '000

150

100

Q3 Q4 $240

Q3 Q4 $480

Q3 Q4 $720

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ data.

Figure 7.3: Distribution of potential beneficiaries of public works, urban, 2007

Number of beneficiaries, in '000

$240 Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ data.

Q3 $480

Q3 $720

Poverty and the Policy Response to the Economic Crisis in Liberia

121

Figure 7.4: Distribution of potential beneficiaries of public works, rural, 2007

Number of beneficiaries, in '000

100 80 60 40 20 0

Q3 $240

Q3 $480

Q3 $720

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ data.

3. Potential Poverty Impact of Public Works The cost of a public works program would depend on the number of beneficiaries and the wages paid to program participants. The analysis of the 2007 CWIQ data suggests that only a small share (less than 10 percent) of the population is likely to earn more than the minimum wage, which is US$2/day, or US$480/year. While this suggests that the targeting of public works programs would be better if wages are set lower than the minimum wage, this may not be socially and legally defensible. Therefore we consider three wage levels for the simulations: US$240, US$480 and US$720. These wages are annualized. We assume that public works participants will benefit from the program for six months per year. Table 7.2 provides simple estimates of the potential cost of the program under three scenarios for the number of beneficiaries. Under the first scenario, the program would reach 50,000 beneficiaries, and its cost would then range from US$6.9 Table 7.2: Estimates of project cost (wages and administrative costs), 2007 Parameters Beneficiaries

Scenario I 50,000

50,000

Scenario II 50,000

Min. wage (US$)

100,000

----480----

Paid wage (US$)

240

480

720

240

480

720

% minimum wage

50%

100%

150%

50%

100%

150%

Employment duration Cost (US$M) Adm. cost 15% Total cost (US$M)

---6 Months--6

0.9

1.8

2.7

1.8

3.6

5.4

6.9

13.8

20.7

13.8

27.6

41.4

Cost/GDP (%)

0.8%

1.5%

2.3%

1.5%

3.1%

4.6%

GDP (US$M)

904.35

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ data.

122

A World Bank Study

million (0.8 percent of GDP) assuming program wages of US$240 per year or about half the minimum wage, to US$20.7 million (2.3 percent of GDP) assuming program wages of US$720 or about 150 percent of the minimum wage. Under a second scenario, the program would reach 100,000 beneficiaries at a cost ranging from US$13.8 million (1.5 percent of GDP) to US$41.4 million (4.6 percent of GDP). These costs include wage cost and administrative costs, but do not cover the other costs of public works in terms of materials for construction purposes. In order to assess the potential impact of the public works program, on the basis of the numbers of jobs created, we randomly select among all potential beneficiaries of the program (the number of which depends on the wage provided) a number of participants so as to match the distribution of the actual program participants. This is done for each of the wages assumed to be provided. The results of this procedure and the related statistics on targeting performance are provided in table 7.3 for poverty headcount and table 7.4 for extreme poverty headcount. Consider first table 7.3 which provides data for public works program potential participants given the poverty headcount. The first column provides the estimate of the total number of potential beneficiaries of the program depending on the wage level, as estimated from table 7.1. For example, at a wage level of US$240/year, 165,281 individuals in the North Central region might be potential beneficiaries of public works according to our method for identifying such potential beneficiaries. The second column provides the Table 7.3: Potential leakage of public works for poverty headcount, 2007

$240

$480

$720

Region Greater Monrovia North Central North Western South Central South Eastern A South Eastern B Total Greater Monrovia North Central North Western South Central South Eastern A South Eastern B Total Greater Monrovia North Central North Western South Central South Eastern A South Eastern B Total

# of people 37,331 165,281 27,198 71,549 24,242 17,652 343,252 101,082 229,512 61,478 87,877 54,203 42,298 576,450 114,643 241,544 63,901 93,311 57,130 44,696 615,224

Poverty headcount in% 50.9 68.9 74.2 63.4 78.9 68.6 66.9 49.9 67.9 73.7 58.9 74.4 67.3 64.5 47.6 67.2 73.0 57.7 74.1 66.8 63.3

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ data.

Part time and partial substitution Additional wage in Leakage US$ yearly rate in % 79.6 60.8 73.6 51.5 66.0 53.1 78.4 52.0 72.4 45.5 70.5 52.8 74.4 52.4 173.9 57.9 168.6 45.5 158.4 44.4 176.6 49.4 163.6 41.2 164.8 46.1 168.9 47.8 270.9 58.0 269.3 42.3 260.6 39.4 276.3 48.4 264.3 37.2 266.0 42.9 269.1 45.4

Full time and full substitution Additional wage in Leakage US$ yearly rate in % 108.5 72.5 84.3 72.0 54.0 80.5 103.8 67.4 79.4 69.9 72.1 74.3 87.6 71.7 275.7 65.6 254.4 58.8 213.5 62.4 286.3 57.7 234.2 56.9 239.2 59.5 255.6 60.1 453.8 63.7 447.3 51.9 412.3 51.8 475.3 54.5 427.1 48.6 434.1 52.7 446.3 54.2

Poverty and the Policy Response to the Economic Crisis in Liberia

123

Table 7.4: Potential leakage of public works for extreme poverty headcount, 2007

Region $240

Greater Monrovia

Full time and full substitution Additional wage in Leakage US$ yearly rate in %

37,331

24.5

79.6

81.1

108.5

86.6

North Central

165,281

57.2

73.6

59.9

84.3

77.0

North Western

27,198

59.9

66.0

62.2

54.0

84.4

South Central

71,549

45.8

78.4

65.7

103.8

77.1

South Eastern A

24,242

61.3

72.4

58.0

79.4

77.4

South Eastern B

Total

$480

# of people

Poverty headcount in%

Part time and partial substitution Additional wage in Leakage US$ yearly rate in %

17,652

53.5

70.5

63.8

72.1

81.2

343,252

51.6

74.4

63.7

87.6

78.9

Greater Monrovia

101,082

24.1

173.9

80.1

275.7

84.2

North Central

229,512

55.0

168.6

55.5

254.4

66.1

North Western

61,478

59.1

158.4

55.4

213.5

69.8

South Central

87,877

41.2

176.6

64.5

286.3

70.1

South Eastern A

54,203

58.9

163.6

53.2

234.2

65.4

South Eastern B

42,298

51.4

164.8

58.5

239.2

68.5

Total

576,450

48.0

168.9

61.2

255.6

70.4

$720

Greater Monrovia

114,643

22.6

270.9

80.2

453.8

83.0

North Central

241,544

54.5

269.3

53.1

447.3

60.7

North Western

63,901

58.1

260.6

51.6

412.3

61.4

South Central

93,311

40.2

276.3

63.9

475.3

68.0

South Eastern A

57,130

58.0

264.3

50.5

427.1

59.1

South Eastern B

44,696

51.1

266.0

56.2

434.1

63.5

Total

615,224

46.8

269.1

59.6

446.3

66.1

Source: Authors’ estimation using 2007 CWIQ data.

share of those individuals living in households who are poor. For example, at a wage of US$240/year, 68.9 percent of the potential beneficiaries in the North Central region live in a household in poverty according to the definition of poverty used by the Liberia Institute of Statistics and Geo-Information Services (see Backiny-Yetna and al., 2011). The third column provides the additional wage to be obtained by each individual, on average, depending on the wage proposed for the program. At a wage of US$240/year, out of that amount, on average US$73.6 represents additional income for potential participants to the program in the North Central region. The next column provides the leakage rate, which is computed as the product of the poverty rate times the additional wage divided by the reference wage of the program. The leakage rate represents the share of program outlays that do not directly contribute to poverty reduction. This leakage rate depends on two key parameters: (i) the targeting performance of the program, and (ii) the substitution effect of the program, whereby only part of the wages paid to beneficiaries generates additional income, because beneficiaries would probably have done other work if they had not participated in the program.

124

A World Bank Study

If the public works program is implemented with an annual wage of US$720, the leakage rate for poverty is estimated in Monrovia at 58.0 percent, and for extreme poverty the estimate is at 80.2 percent. These are relatively high leakage rates because the share of those participating in the public works program that are poor or extreme poor is lower in the capital than elsewhere. Overall, however, the variation in leakage rates between the various wage levels is not very high. This is because a higher wage levels implies less targeting to the poor, but on the other hand it reduces the substitution effect through which part of the gains from the public works wage are lost due to the need to give up other work. In terms of results for the country as a whole, at a wage rate of US$240 per year, the overall leakage rate is 52.4 percent, and it remains between 45.4 percent and 52.4 percent when we change the wage rate. However, as already mentioned, the leakage rate is systematically higher in Monrovia than elsewhere, because the share of participants in the program that are poor or extreme poor is lower in the capital. In contrast, the leakage rates are lowest in the South eastern area of the country, where poverty and extreme poverty are higher. The estimated potential impact of the program on poverty is given in table 7.5. The estimates are obtained in a very simple way. For the participants in the program who belong to households living in poverty, we add to the consumption aggregate of the household the gains in earnings obtained by the participants, and we recomputed poverty using the same poverty lines (for a discussion of poverty measurement in Liberia, see Backiny-Yetna and al, 2011). In other words, we assume that the full amount of the earnings gains for program participants translate into additional consumption for their households. Our simulations for the impact of public works on poverty are based on the assumption of a 50 percent substitution effect, so that program participants give up half their current earnings to participate in the public works program which is assumed to take place in the lean season. For higher wages, the impact is higher, since the additional earnings obtained by participants are higher. With the provision of 50,000 jobs and assuming annual public works wage of US$720 (each workers then gets US$360 over a six month period) the headcount index of poverty is reduced by 15.6 percentage points among program beneficiaries. The reduction in the headcount for the population as a whole is 1.61 percentage point. The impact on extreme poverty is similar. While a reduction in the headcount index of poverty of less than two Table 7.5: Potential impact of public works for the reduction of poverty, national, 2007 Beneficiaries

Whole population

Headcount

Poverty gap

Headcount

Poverty gap

$240

3.33

3.29

0.35

0.34

Impact on poverty, 50,000 jobs $480

10.16

6.85

1.05

0.71

$720

15.61

10.06

1.61

1.04

$240

3.40

$480

10.08

6.02

1.04

0.62

$720

16.07

8.62

1.66

0.89

Impact on extreme poverty, 50,000 jobs 3.01

Source: Authorsâ&#x20AC;&#x2122; estimation using 2007 CWIQ data.

0.36

0.31

Poverty and the Policy Response to the Economic Crisis in Liberia

125

percentage points may not appear to be very large as compared to the existing share of the population in poverty (at 64 percent), this is still not negligible and a large share of the population would benefit from improvements in standards of living through the public works program. When 100,000 jobs are created, the impact on poverty is about twice that of the impact with 50,000 jobs.

4. Conclusion Job creation is an imperative for the new Government of Liberia. Fourteen years of civil conflict have not only destroyed the social and economic infrastructure base but they have also ground the economy to a virtual halt and have consequently resulted in large scale unemployment and significant poverty. Growth has picked up since the signing of the Accra Comprehensive Peace Agreement in 2003, spurred in part by increasing foreign direct investment in traditional sectors. However, the current rate of job creation in these sectors, even under the most optimistic scenario, is unlikely to absorb a significant portion of the unemployed and underemployed labor force in the short-term. The lack of gainful employment, including for many unskilled youth (whose education was terminated by the conflict) as well as ex-combatants, poses a significant risk to maintaining peace. One of the major challenges facing the Government is to devise a strategic response to address the immediate employment situation in the context of establishing a strategic framework for more sustainable, long-term jobs created by the private sector. The strategy for job creation that is ultimately adopted by the Government should reflects its social, economic and political needs; its administrative capacity to manage the implementation of the strategy; and the viability of the strategy in terms of it financial and political sustainability given the current and evolving fiscal space and the political situation. In terms of strategy options, labor-intensive public works appear to be one of the most natural fits to the current situation in Liberia. Labor-intensive public works can respond to the countryâ&#x20AC;&#x2122;s dual needs to: (i) create social and economic assets to improve welfare and help create the environment for private sector led growth; and (ii) provide employment for a large number of unskilled workers including women and youth. Wage subsidy schemes may be less appropriate for the current situation in Liberia given the public and private sectorsâ&#x20AC;&#x2122; limited capacity to absorb additional labor in the shortterm. Furthermore, targeted wage subsidy schemes are administratively more difficult to administer and generalized schemes would be more expensive. The opportunities for labor-intensive public urban and rural works cover rehabilitation or reconstruction of schools, health centers and other small civil works as well as the management of solid waste. As for roads, low-volume community and feeder roads also lend themselves to labor-intensive works both for construction and maintenance. Some aspects of the maintenance of high volume, highway roads also lend themselves to labor-intensive operation. However, re-construction of critical high-volume roads with machine surface finish, lend themselves less to labor-intensive operation. In any case, these are also the roads which need to be delivered quickly to crowd in private sector investment. Of the current road network of about 10,000 km, about half of which may be suitable for labor-intensive operation. The cost of a public works program in Liberia would vary greatly depending on its size and the wages paid. For example, a program providing six months of employment

126

A World Bank Study

to 50,000 beneficiaries would cost from US$6.9 million (0.8 percent of GDP) with a very low public works wage of US$240 per year (or about half the minimum wage) to US$20.7 million (2.3 percent of GDP) with a wage of US$720 per year (or about 150 percent of the minimum wage). This cost includes wage costs and administrative costs, but not other costs in terms of materials for construction purposes. A program with 50,000 jobs at an annual wage of US$720 would reduce the share of the population in poverty by about 15.6 percentage points among program beneficiaries and 1.61 point in the population as a whole. The impact on extreme poverty is similar. While this may not seem large, a large share of the population would benefit from improvements in standards of living through the program. Our simulation suggested that the number of individuals who could potentially be interested in the program appears to be very large, especially because many workers are working with low pay (observed or imputed) and might therefore be interested in getting higher cash income through public works. The targeting performance or likely benefit incidence of the program depends on whether the program is implemented mostly in urban or rural areas. In urban areas, the program would probably be regressive, since most of the potential beneficiaries belong to the better off quintiles of the population (this is because urban households tend to have higher levels of consumption than rural households, so that relatively few households in urban areas belong to the bottom quintiles). By contrast, the program could be well targeted to individuals belonging to households which tend to be poor if the focus is placed on providing employment and reconstructing infrastructure in rural areas.

Notes 1. The authors are with the World Bank. This chapter was prepared in part as a background paper for a World Bank report on Employment and Pro-Poor Growth Liberia, and presented in Monrovia among others at a workshop in March 2011. The views expressed here are those of the authors and need not reflect those of the World Bank, its Executive Directors or the countries they represent. 2. On skills training in Africa, see Adams (2007), Haan and Serriere (2002), Johanson and Adams (2004) and Rosholm, Nielsen and Dabalen (2007).

References Adams, A. V. 2007. “The Role of Youth Skills Development in the Transition from School to Work: A Global Review.” HDNCY Discussion Paper No. 5. World Bank, Washington, DC. Adato, M., and L. Haddad. 2001. “Targeting Poverty through Community-Based Public Works Programs: A Cross-Disciplinary Assessment of Recent Experience in South Africa.” IFPRI Discussion Paper 121. International Food Policy Research Institute (IFPRI), Washington, DC. Backiny-Yetna, P., R. Mungai, C. Tsimpo, and Q. Wodon. 2012. “Poverty in Liberia: Level, Profile, and Determinants.” In Q. Wodon, ed., 9–34. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Coady, D. 2004. “Designing and Evaluating Social Safety Nets: Theory, Evidence, and Policy Conclusions. Food Consumption and Nutrition Division Discussion Paper 172. IFPRI, Washington, DC.

Poverty and the Policy Response to the Economic Crisis in Liberia

127

Coulombe, H., M. Temourov, and Q. Wodon. 2008. “Ghana’s National Youth Employment Program and Poverty Reduction.” In Ghana: Job Creation and Skills Development. Report No. 40328–GH, Volume 2. Washington, DC: World Bank. Haan, H. C., and N. Serriere. 2002. “Training for Work in the Informal Sector: Fresh Evidence from West and Central Africa.” Occasional Papers of the International Training Centre of the International Labour Organization (ILO), Turin. Johanson, R., and A. V. Adams. 2004. Skills Development in Sub-Saharan Africa. Washington, DC: World Bank. Ravallion, M. 1999. “Appraising Workfare.” World Bank Research Observer 14: 31–48. Rosholm, M., H. S. Nielsen, and A. Dabalen. 2007. “Evaluation of Training in African Enterprises.” Journal of Development Economics 84: 310–329. Subbarao, K. 2003. “Systemic Shocks and Social Protection: Role and Effectiveness of Public Works Programs.” Mimeo, World Bank, Washington, DC. World Bank. 2007a. World Development Report 2007: Development and the Next Generation. Washington, DC: World Bank. ———. 2007b. Global Inventory of Interventions to Support Young Workers. Washington, DC: World Bank. ———. 2009. Employment and Pro-Poor Growth in Liberia. Washington, DC: World Bank.

CHAPTER 8

Liberiaâ&#x20AC;&#x2122;s Cash for Work Temporary Employment Project: Responding to Crisis in Low Income, Fragile Countries Colin Andrews, Prospere Backiny-Yetna, Emily Garin, Emily Weedon, Quentin Wodon, and Giuseppe Zampaglione1 While the previous chapter looked at ex ante simulations of the potential impact of cash for work programs in Liberia, this chapter analyzes ex post the actual performance of the Cash for Work Temporary Employment Program (CfWTEP) that was implemented in 2008/09. The program was funded by the World Bank through an emergency crisis facility in response to the 2007/08 food price crisis. Both quantitative and qualitative data are presented, focusing on the operational and policy experiences emerging from program implementation. This chapter analyzes the context that led to the creation and implementation of the CfWTEP in Liberia, the nature and administrative arrangements for the program, and its operational performance. The objective is to share the lessons learned from evaluation findings so that they can be useful for implementing similar programs in the future in Liberia itself or in other countries. Findings from the analysis suggest that Liberiaâ&#x20AC;&#x2122;s program was well targeted to the poor and had a substantial impact towards poverty reduction among beneficiaries. These results highlight the possibilities of implementing public works program in low capacity, post conflict setting and the scope for using the program as a springboard towards a broader and more comprehensive social safety net.

1. Introduction Together with reductions in indirect taxes on food imports, cash for work programs were one of the main responses implemented by African governments following the food, fuel, and financial crisis of recent years (Wodon and Zaman, 2010). The main objective of those programs was to help the poor cope with the various shocks by increasing their net earnings through community-level work paid for under the programs. In addition, the programs also helped to build, repair, or maintain local infrastructure, but the main objective was clearly short term poverty reduction. Yet it is unclear whether these 128

Poverty and the Policy Response to the Economic Crisis in Liberia

129

cash for work programs indeed reached their intended beneficiariesâ&#x20AC;&#x201D;that is the poor who were especially affected by rising prices. It is also unclear whether these programs generated other, potentially long-term beneficial impacts. And given the administrative and other costs of implementing the programs, it is also unclear whether a large enough share of the initial budget allocations for the cash for work programs ultimately ended up in additional net incomes for the households whose members benefitted from the programs. In Liberia, in order to assess quickly the performance of the cash for work program it helped fund through an emergency crisis facility, the World Bank implemented in partnership with the Liberian Agency for Community Empowerment a light evaluation of the Cash for Work Temporary Employment Project (CfWTEP). Both quantitative and qualitative data were collected for this evaluation, focusing especially on targeting, wage setting, poverty impact, and other benefits of the program. This chapter analyzes the context that led to the creation and implementation of the CfWTEP in Liberia, the nature and administrative arrangements for the program, and its performance. The objective is to share lessons learned from the evaluation so that they can be useful for implementing similar programs in the future in Liberia itself or in other countries. The chapter is structured as follows. The first section briefly summarizes the impact of the global food price crisisâ&#x20AC;&#x201D;the impetus for the projectâ&#x20AC;&#x201D;on Liberia. The second section places cash for work (CfW) programming in the context of broader social protection needs and programs in Liberia. The third section describes major design and implementation features of the CfW Project. The fourth section lays out key findings from the Liberian experience, building on the qualitative and quantitative data collected. The conclusion provides a series of considerations for responding to similar crises in other low-income, fragile states. It also details the way in which the project is being used to undergird the development of a broader social protection framework within Liberia. In summarizing design elements and feedback from program implementation, the chapter draws on two previous World Bank analyses of the project. The first document is a quantitative assessment, which was based on a light evaluation survey carried out in November and December of 2009 (Backiny-Yetna, Wodon, and Zampaglione, 2012). The quantitative assessment was primarily intended to analyze program results such as targeting performance, wage substitution effects and patterns of wage usage among participating households. The second document is a qualitative analysis, which included summary results from stakeholder interviews and focus group discussions held with both program participants and non-participants.

2. The Food Price Crisis in Liberia Global food prices rose an average of 43 percent between March 2007 and March 2008. These price increases quickly cascaded into a global food crisis with devastating impacts on vulnerable communities throughout the world. The UN Food and Agriculture Organization estimated that 75 million people were added to the global undernourished population in 2007, with most of the rise attributable to price increases (see also Ivanic and Martin, 2009). Liberia was particularly susceptible to the worst effects of the food crisis. Despite its significant potential for agricultural production, Liberia still relies heavily on imported food, with two-thirds of all food coming from outside the country. Even before the cri-

130

A World Bank Study

sis began, 51 percent of the country was classified as having unacceptably low levels of food consumption. On average, food purchases account for half of household cash expenditures. The impact of the crisis on rice—the staple crop of the Liberian diet—provides a clear illustration of why a quick response was necessary. When the crisis began, roughly 60 percent of rice consumed nationwide was imported; in the capital area of greater Monrovia, that number reached 99 percent. When global commodity prices rose, Liberians felt the effects immediately and acutely. A UN Joint Assessment of the crisis found a 40 percent year-over-year increase in the price of rice in Monrovia, despite the suspension of import tariffs. Analysis based on consumption patterns also suggested that the increase in food prices might have a very large impact on the poor, with Liberia being somewhat of an outlier in West Africa in this respect. Specifically, analysis for a dozen African countries by Wodon et al. (2008) suggests that in Liberia, an increase of 50 percent in the price of rice might lead to an increase in the share of the population falling into poverty of six percentage points, which is very large and indeed higher than for most other countries. In addition of driving a large number of households into poverty, the increase in rice price also worsened the living conditions of the two thirds of the population that already lived in poverty before the crisis. Clearly, emergency measures were needed, and were indeed taken, including the launch of CfWTEP. Liberia’s post-conflict status created even greater urgency for a crisis response. In addition to all of the development setbacks traditionally linked to food insecurity, food scarcity in a post-conflict country can threaten the maintenance of a fragile peace. The strong political divides between Liberia’s food producing regions and its largest population centers made the country particularly vulnerable to renewed conflict. Although there was obvious political and economic need for rapid action, Liberia’s long history of civil conflict had left the government with extremely limited capacity for implementing an effective response. Despite the time pressures created by the onset of the food price crisis and the weak national capacity for implementation, Liberia was able to mount a successful response program in the form of the Cash for Work Temporary Employment Project.2 The project was a joint effort of the Government of Liberia (GoL) and the World Bank, designed to provide short term employment and income for households hard-hit by rising food prices. The objective of the CfW Project was to mitigate the short-term effects of the food price crisis by creating 680,000 days of temporary employment for 17,000 vulnerable households. Work performed through the program also provided public services to Liberian communities, including rehabilitating public agricultural land in rural areas and cleaning and clearing roads, drains, and public spaces in urban and rural areas. In response to a request from the Ministry of Agriculture, the CfW Project was introduced as an additional grant to the World Bank’s Community Empowerment Project II (CEPII), alongside a series of crisis response interventions including school feeding and support to agricultural production. By June 30, 2010 the project had reached all 17,000 intended beneficiaries. Based on the success of the early implementation experiences, a project to scale up this intervention was approved to by the Board of the World Bank on June 26, 2010. The second phase will run from July 1, 2010 to June 30, 2013.

Poverty and the Policy Response to the Economic Crisis in Liberia

131

3. High Need, Low Capacity: Cash for Work in the Liberian Context When the food price crisis struck in 2007, Liberia was just four years removed from a fourteen year civil conflict. The war devastated an economy that was already on the decline and, in a nation with only 3.5 million residents, claimed the lives of more than 250,000 people. The entire educational system was disrupted and largely destroyed, leaving a full generation of Liberian students with little to no education and extremely limited job skills. This has impacted both their economic prospects as well as the workforce capacity of the government, private sector businesses and national non-governmental organizations (NGOs). Estimates of the level of unemployment in Liberia are imprecise, but estimates based on nationally representative household surveys suggest that unemployment as well as underemployment affect approximately 20 percent of the population (World Bank, 2010). Of the population that is considered employed, the vast majority are performing low-paying, informal work with little security or opportunity for escaping poverty. Youth are the hardest hit by unemployment, with the highest levels of unemployed youth found in urban influx areas like Monrovia. With an annual per capita gross domestic product estimated at US$222, Liberiaâ&#x20AC;&#x2122;s poverty is endemic. Nationwide, 64 percent of the population falls below the national poverty line and 48 percent of Liberians are below the extreme poverty line (BackinyYetna et al., 2011). While all areas of the country have high levels of poverty, conditions are significantly worse in rural areas, particularly in the southeast of the country. Levels of food insecurity approach or exceed 75 percent in the three counties that make up the southeastern corner of the country. Liberiaâ&#x20AC;&#x2122;s total population consists of approximately 650,000 households, of which 400,000 fall into the category of absolute poverty. According to a 2008 United Nations assessment of social protection needs, 70,000 of those households are labor-constrained, while 330,000 have available labor. Liberiaâ&#x20AC;&#x2122;s high level of need for social protection measures has not yet been met with commensurate levels of protection-related policies and programming. The national Poverty Reduction Strategy (PRS) for the 2008-2011 period makes no reference to social protection, welfare or safety net programming. The topics are also missing from the County Development Agendas, the local counterparts to the national strategy. While there is not yet a coordinated framework for social protection interventions, a variety of actors have undertaken social protection projects. In 2010, UNICEF and the government launched a cash transfer pilot project in Bomi County, aimed at households which were both extremely poor and labor constrained. The World Food Programme (WFP) supports upwards of 600,000 school children through its school feeding program, which is in part supplied by crops purchased from smallholder farmers through the WFP purchase-for-progress scheme. A number of entities have focused on employment-specific interventions. The Government of Liberia (GoL) funded the 2010 Liberia Jobs and Opportunities Initiative, which created 8,000 temporary jobs for youth in Greater Monrovia. Beginning in 2006, various United Nations institutions supported the Liberia Emergency Employment Programme and the Liberia Employment Action Programme (LEEP/LEAP), a public worksbased employment scheme and longer term employment strategy for the country. The

132

A World Bank Study

United Nations Mission in Liberia (UNMIL) has also undertaken temporary employment measures at different points, offering short-term opportunities to 20,000 Liberians. There are encouraging signs about the growing prominence of social protection issues in the consciousness of government planners. The structure for the new PRS includes an entire pillar dedicated to human development as well as a sub-section specifically focused on social protection. A National Social Protection Secretariat has been established within the Ministry of Planning and Economic Affairs and will soon begin development of the National Social Protection Policy. Translating this policy planning into government ownership and financial commitments will remain a serious challenge. Liberiaâ&#x20AC;&#x2122;s annual budget in for the 2010-2011 fiscal year was US$347 million and the country has development needs which require many times that amount each year, meaning that social protection programming faces stiff competition for budget allocations.

4. Weaving the Safety Net: Design Elements of the Cash for Work Project This section of the chapter explores key design and implementation experiences under the CfW Project. The discussion maintains a strong operational focus in order to highlight the main priorities and challenges involved in the introduction of the labor intensive program, particularly in a low capacity context such as Liberia. The CfW Project was financed under a grant agreement between the World Bank and the Government of Liberia, as part of the Global Food Crisis Response Program. Responsibility for overall implementation was assigned to the Liberia Agency for Community Empowerment (LACE). LACE was established by the Community Empowerment Act pursuant to Chapter 50B of Title 12 of the Liberian Code of Laws of July 22, 2004. The objectives of the Agency are to improve the living standards of poor communities through the provisions and strengthening of basic social services and to promote a community-based approach in sub-project identification, preparation, implementation, administration and maintenance. The Agency is a not-for-profit and autonomous organization, but it is accountable to the President of the Republic. During the CfW Project, LACE oversaw the implementation of 34 sub-projects in 15 counties, each consisting of 500 beneficiaries. Coordination of sub-projects was undertaken by local non-governmental or community based organizations, henceforth referred to as Community Facilitators. Project activities were deliberately labor intensive and included work such as roadside brushing and backfilling of potholes. All projects required simple skills and low risk manual labor. This ensured that project workers could be selected from the beneficiary communities. Each worker was paid a daily wage of US$3.00, totaling US$120.00 for forty days of work. Payment was made in US Dollars and disbursed on a monthly basis through a commercial bank. Forty-six percent (46 percent) of the total number of workers were female. Table 8.1 outlines the county-bycounty breakdown of beneficiary and project numbers. At the time of writing, LACE had disbursed US$2,876,923.97 of the US$3,000,000.00 allocated for the CfW Project, representing 95.9 percent of the project budget. Of the total expenditures incurred as of August 25, 2010, 70.7 percent (US$2,035,445.00) was spent on wages for the workers, with an additional 14.1 percent of costs allocated to goods and related subproject costs, including vehicles, motorbikes, tools, and bank com-

Poverty and the Policy Response to the Economic Crisis in Liberia

133

Table 8.1: Beneficiaries and projects per county Number of beneficiaries

Number of projects

Male

Female

Total

Wage disbursements to date (US$)

Montserrado

645

855

1,500

178,979

Margibi

300

700

1,000

120,078

Grand Bassa

510

490

1,000

121,579

Nimba

1,043

457

1,500

180,194

Gbarpolu

220

280

500

60,010

Bomi

840

660

1,500

180,060

Grand Cape Mount

330

170

500

60,080

Bong

524

476

1,000

119,554

Lofa

953

547

1,500

179,277

River Cess

650

350

1,000

119,767

Sinoe

465

535

1,000

120,160

River Gee

759

741

1,500

180,180

Grand Kru

872

628

1,500

178,098

Grand Gedeh

615

385

1,000

119,239

Region

Maryland Total % male & female participation

504

496

1,000

118,200

9,230

7,770

17,000

2,035,455

54%

46%

100%

Source: Authors.

mission fees for worker payment (US$412,146.00). Consultancy and training expenses, including payment of the community facilitators, represented 3.7 percent of total expenditures (US$109,897.00). Project management expenses totaled 11.1 percent of the budget ($319,455.21). Although the CfW Project does not close until June 30, 2012, LACE expected to fully incur all additional expenses by the end of 2010. The Operational Guidelines clearly outlined the types of activities that could be undertaken as sub-projects in both rural and urban communities. Projects had to be simple and could not be imposed on communities by the community facilitators, local authorities, or LACE. Worksite teams each included 500 workers and one supervisor per project. Workers were split into workgroups varying from 20-100 people, with one leader per workgroup. In urban communities, the public work activities were limited to street sweeping and cleaning, drainage clearance, painting of public buildings and street walls, painting of crosswalks, rehabilitation of recreational spaces, rehabilitation of schools, health posts and other community buildings. In rural communities, activities were limited to brushing of bushes along access roads, clearing non-private agricultural land (or for any other purpose that would serve a community or public need), breaking of rocks for road rehabilitation, cleaning and replacement of culverts, and drainage clearance along roads. Beneficiaries as well as non-workers confirmed that these were the types of activities undertaken in their respective communities. Table 8.2 further delineates the types of projects undertaken in a sample of counties.

134

A World Bank Study

Table 8.2: Number of projects and types of activities County

No. of projects

Types of public works

Montserrado

Roadside brushing, drainage clearing, garbage removal, sweeping of streets, backfilling of potholes on roads, grass cutting, painting of public structures and painting of walkways carried out in 26 communities

Grand Bassa

Roadside brushing, sweeping streets, drainage clearing, backfilling of potholes on roads, hoeing of grass and disposing of debris carried out in six communities

Sinoe

Hoeing of grass, roadside brushing, sweeping of streets, backfilling of potholes on roads, drainage clearing, and airstrip clearing

Nimba

Roadside brushing, drainage clearing, backfilling of potholes on roads, and repairing of small log bridges

Source: Authors.

Project activities were undertaken because of the perceived benefit to the communities or implementers. Key considerations in project selection were: Maximizing community participation through labor intensive projects. Heavy machinery was not encouraged given capital to labor intensity ratios; Designing projects which did not divert community members from productive opportunities or appropriate coping mechanisms. To support this goal, projects were organized around task-based work and situated in close proximity to communities; and facilitating simple and quickly executable projects within the timeframe and capacity constraints of the project. Projects such as road rehabilitation and rubbish cleaning helped address existing problems in the affected communities. Rehabilitated roads improved movement of people and goods and clearing potholes and refilling them with dirt led to improvement in community health by reducing the risk of malaria. In terms of institutional design, the Government of Liberia has serious human capacity and resource constraints which limit its ability to effectively implement programming outside the capital. Government structures in the counties are generally weak and overburdened by the competing demands of various sector needs. Given these limitations, GoL and the World Bank decided to delegate overall implementation of the CfW Project to the Liberia Agency for Community Empowerment (LACE). LACE is an independent not-for-profit organization, but it was established under government auspices in 2004 and is accountable to the Executive Branch. LACEâ&#x20AC;&#x2122;s mandate is to improve living standards for poor communities while promoting communityinclusive development processes. The organization was selected because of its excellent track record in project implementation as well as its established presence in each county. Without LACEâ&#x20AC;&#x2122;s existing networks and relationships across the country, a nationwide scale-up of the size and speed of CfWTEP would not have been possible. LACE, in turn, selected Community Facilitators (CFs) to oversee implementation of specific sub-projects in each county. CFs were local NGOs and community based organizations selected on the basis of their capacity and prior project implementation experience. LACE trained a total of 17 organizations for participation as CFs in the CfW Project.3 CFs were responsible for sensitizing local authorities about the project, generating awareness in the selected communities, and recruiting, supervising, and submitting payroll information for workers. LACE worked with the CFs to rollout sub-projects across the country and monitored their performance throughout the CfW Project lifecycle. LACE and the Community Facilitators were responsible for putting into practice

Poverty and the Policy Response to the Economic Crisis in Liberia

135

the specific implementation design that had been agreed to by GoL and the World Bank. These included issues such as geographic project targeting, participant selection, wage setting, payments, and monitoring and evaluation.

5. Implementation Elements 5.1. Targeting of the Poor

A basic building block in safety net programs is the design of effective targeting mechanisms (see Coady et al., 2002, as well as Grosh et al. 2009 for a discussion of targeting mechanisms and other aspects of safety nets, and Adato and Haddad, 2002, as well as Teklu and Asefa, 1997, 1999 for examples of previous evaluation of public works programs in Africa). Given the low capacity and high poverty context, many low income countries will face a hard budget constraint when it comes to determining program eligibility. Difficult triage decisions may often be required when determining how to spend limited finances. To this end, targeting can be a useful instrument to maximize coverage of the poor and vulnerable, as well as promoting transparency in public resource allocation. However, the issue of targeting is not without controversy—especially for public works programs given the selection challenges involved, as well as challenges on feasibility of targeting where data availability is limited. The targeting priority established under the CfW program was to avoid inclusion error i.e. the participation of non-poor groups, and ensure beneficiaries were amongst the neediest of the population. To this end, ‘vulnerability’ criteria were critical in the project design, particularly at the district level as the main indicator of geographic targeting. At the local level, some flexibility was afforded to community facilitators and local communities, to apply vulnerability criteria if required based on the communities own understanding of poverty. A combination of targeting methods was therefore employed. A geographic targeting approach based on food vulnerability criteria determined the number of projects to be assigned to each county. The actual phasing of the implementation was also based on logistical capacity within each targeted area. Additional community criteria were allowed to manage beneficiary selection and ensure a gender balance amongst beneficiaries. Wage setting, a traditional tool for promoting self-selection among the poor, was somewhat constrained by precedents for similar programs and pressure from the government. Given the centrality of wage levels to program design, this issue is discussed in more details later. Survey evidence and community feedback indicates that the project performed well at selecting poor beneficiaries, but less well at reaching the poorest (see section 4). To a large extent this can be attributed to the degree of flexibility afforded to the targeting process, the fact that a high percentage of the population can be considered as poor and the ‘first-come, first-served’ mechanisms used in practice for rationing. While efforts were made to base geographical targeting largely around ‘vulnerability’ information from the Comprehensive Food Security and Nutrition Survey 2006, this was not always possible owing to the transitory nature of the food crisis, combined with sheer logistical and infrastructure deficits. Accordingly, scope was provided to direct targeting decisions around implementation capacity of LACE, as well as other logistical considerations such as banking capacity. The use of different community mechanisms proved valuable in targeting, with potential for even further usage as the future program evolves. Once a county was assigned

136

A World Bank Study

its number of sub-projects through geographic targeting, LACE contacted County Superintendents and mayors of major cities to inform them about the program and seek input on which communities should receive sub-projects. This community input determined the project sites, where CFs then went to recruit individual participants. Once the participants were selected, they were educated about the types of activities that could be chosen for project work. The participants then selected the actual projects to be undertaken as part of the CfW scheme. In this respect, the formal guidelines for individual participation were important: program participation was voluntary and open to community members over the age of 18. The only persons completely prohibited from participation were pregnant women and those on the government payroll. There was also a quota requiring at least 30 percent female participation in the project. To ensure that targeting was appropriate to each context, communities were allowed to adopt additional participation criteria based on their local needs and understanding of vulnerability. They were encouraged to draw-out participation from disabled and women-headed households as well as those households where no family member had employment. While the flexibility was intentionally maintained in the project design, the extent to which communities utilized that leeway transparently and consistently was unclear. The limitations of these somewhat flexible targeting approaches were evident throughout implementation, however. Particular obstacles incurred included excess demand for program participations, weak capacity for identifying the most â&#x20AC;&#x2DC;deservingâ&#x20AC;&#x2122; (in the sense of need or particularly low levels of expected consumption without the program) beneficiaries, limited access to the neediest regions, and seasonal obstacles to completing labor-intensive work. While appropriate wage setting under local market levels (i.e. lower wages facilitating self-targeting) might have helped to remedy the more serious challenges (e.g. excess demand), this was not possibly owing to the political economy constraints under which the program operated. The issue of wage setting is now considered in further detail. 5.2. Wage Setting

Wage setting can be a particularly challenging part of program design, and in this context Liberia is no exception. Under normal circumstances, cash for work interventions set wages just below the prevailing market wage in order to ensure self-selection of program beneficiaries, as well as to prevent wage substitution and other market distorting effects. In response to the food price crisis, a few factors complicated this task. The first challenge was ascertaining a prevailing wage, noting that wage levels may vary across the country. As highlighted in table 8.3, there is indeed variability across counties for payment for unskilled work. Setting program wages below the lowest rural wages would help to reach the poorest, and contribute the most to the reduction in poverty (given that the food price crisis had its strongest impact on those well below the poverty line), but would not necessarily lift participants out of poverty and might prove inadequate in urban or even other rural areas. In addition, there is no clear-cut legal minimum wage rate that could help determine payment levels to workers.4 Given the scale of poverty and absence of formal and informal employment opportunities, it is likely that demand would persist regardless of the rate set, necessitating additional targeting mechanisms with virtually any wage. Table 8.3 compares the daily wage rate offered by the program to rates offered for similar work with local concessionaires or public works programs. The concessionaire

Poverty and the Policy Response to the Economic Crisis in Liberia

137

Table 8.3: Composition of local wage rates in the sample of counties Daily wage in Liberian dollars In-Kind

Daily wage equivalent in US$

Daily wages (in US$) paid by UN/NGOs & other programs

$100.00

$50.00

$2.14

$3.00 $3.00

Geographic location

Description of work/tasks

Cash

Montserrado County

Daily hire for manual labor (unskilled labor) for 8 hours Daily hire for (unskilled labor) for 8 hours

$100.00

$50.00

$2.14

Grand Bassa County

Cutting grass in private yards

$100.00

$50.00

$2.14

Washing clothes for household

$50.00

$40.00

$1.29

Sinoe County

Daily hire for labor (unskilled labor) for 8 hours Clearing of farmland

Nimba County

$3.00 $300.00

$4.29

Daily hire for manual labor (unskilled labor) for 8 hours Farming contract—provision of manual labor

$3.00 $3.00

$125.00

$50.00

$2.50

Source: Making Enterprises (2010).

rates are based on participant-reported local wage rates for unskilled labor in the locations. At the community level, there is no standard daily wage rate. Wage rates are negotiable and dependent on the type of work and economic status of the employer. Local wages are typically a combination of cash, food and extra benefits such as hot meals. On average, the daily wage rate ranges from US$2.00 to US$4.00 depending on the prevailing exchange rate at the time of the transaction. In urban areas, the unskilled wage level is typically around $3.00 per day or higher. In surveyed areas, average wages for unskilled labor varied between US$1.29 and $4.29 per day, with much of the wage consisting of in-kind provisions such as meals. The differences in wages and cost of living between regions could have been used to argue for a variable project wage, but the somewhat unreliable nature of the estimates and the political challenges of wage differentials in different parts of the country precluded this arrangement. Given the estimates that were available, including using data from the 2007 Core Welfare Indicator Questionnaire Survey, the original project design called for wages of US$2.50 per day. But this generated two additional concerns. Although the wage was technically appropriate, given the short term nature of the work and the context of the food crisis, it was unlikely to have as strong a vulnerability-reduction impact as desired (but of course, setting a lower wage would have helped in reaching more beneficiaries within the same budget). More fundamental from a political economy point of view was the fact that in all of Liberia’s recent cash for work style programs (including GoL’s Jobs and Opportunities Initiative, UNMIL’s projects and LEEP), the wage had been set at US$3.00. This made it more difficult for LACE and its CFs to enter communities with a lower wage rate, subjecting them to local suspicion that the total wage was the traditional US$3.00, but that it was being siphoned off by the organizations. When combined with strong pressure from the government for higher wages, all these considerations eventually led to a CfWTEP wage rate of US$3.00 per day.

138

A World Bank Study

Table 8.4 provides perspectives from different stakeholders on the wage level. Given the political context and pay previously offered under similar programs, the wage rate was generally perceived as fair and appropriate. Stakeholders generally believed that this level of payment was justified given the short term nature of the program as well as the absence of fringe benefits such as meals and transportation. There was limited appetite to roll back on this decision, notwithstanding the possible overall benefits that could have been realized through expanded program coverage. While government officials also tended to support this, some concerns were raised regarding the overall affordability of this benefit level. Did the decision to set a relatively high wage (in comparison to prevailing wages) of US$3.0 per day affect strongly the targeting performance of the CfWTEP? We do not have strong counterfactual evidence to assess the impact of the wage level on targeting performance, but the analysis of the light beneficiary survey used to evaluate the program by Backiny-Yetna, Wodon, and Zampaglione (2012) shows that targeting to the poor was fairly good, while targeting to the extreme poor was weaker. This suggests that adopting a lower wage level might have helped in reaching the poorest more, but at the same time the wage level that was adopted did not weaken substantially targeting performance to the poor. Table 8.4: Perspectives of stakeholders interviewed on the wage rate LACE

• Wage setting was intensely debated, with initial rate proposed at $2.50 but a final rate accepted at $3.00 • The wage rate made the selection process difficult because there was an over-supply of labor. Not everybody who wanted to work was accommodated. • The daily wage rate of $3.00 is fair and should be maintained to avoid a backlash. We are concerned for the safety of field staff as some disgruntled persons, knowing that other organizations paid $3.00 for similar works and are now being paid less than $3.00, could incite others to protest payment and during the process, some staff could get harmed.

WFP

• Rural wage is normally below $3.00. • The wage rate of $3.00 had a positive impact on people’s lives.

UNDP

• Up to now, we have not seen any approved wage rate from the Government of Liberia. • UNDP Emergency Employment Program paid $3.00 per day for unskilled labor and $5.00 per day for skilled labor.

UNICEF

• Not aware of the exact criteria used to set wage rate. • Decision to settle on $3.00 was not based on national policy since no clear cut information on minimum wage rate was available at the time.

Ministry of Labor

• The LEEP project used $3.00 as the ideal wage rate. • $3.00 is generally used by all other projects so it is fair given that the intent of the program is to put money in the pockets of the needy. There should be two rates: $5.00 for semi- skilled workers and $3.00 for unskilled workers. This already exists. • Skills training need to be part of the program to help train unskilled workers to make interventions sustainable.

Ministry of Agriculture

• $3/day was adequate, and beneficiaries were very impressed. • It is justifiable looking at our minimum wage rate of $2.50; but MOA cannot afford that wage rate. • The Ministry of Agriculture (MOA) cannot afford the wage rate of $3.00/day for unskilled labor.

Local Authorities

• The wage rate was above our minimum wage of $2.50. • The wage rate was encouraging and attractive due to the poverty level. City Corporation’s workers are paid less than workers of the CfW Project. • The shorter hours of work and output is not commensurate with the daily wage rate of $3.00. Therefore, the number of working hours should be looked at to maximize output. For example, under normal circumstances, a worker would work for 8 hours to earn the $3.00.

Source: Making Enterprises (2010).

Poverty and the Policy Response to the Economic Crisis in Liberia

139

5.3. Payments

The question of payment effectiveness is a major design and operational challenge in the delivery of any safety net program. Payment related mishaps can have serious repercussions on households and may quickly become reputational risks for a program, so timeliness and accuracy of payments should be a key priority in program design. Additionally, poorly devised payment control mechanisms can allow fraud and corruption. As outlined in Section 2, previous implementation experiences have struggled to make cash payments in a timely, secure, and predictable way. Beneficiaries have also complained about fraudulent and corrupt practices such as the inclusion of ‘ghost names’ on the payroll and underpayment for actual participants. In this context, the payment arrangements implemented under the CfW Project are of particular note. Table 8.5 highlights the key elements of these payment arrangements. Careful payment design at the outset of project implementation was critical to its success. The project partnered with a commercial bank with relatively high penetration throughout the country and the ability to make mobile payments where a local branch Table 8.5: Key elements of the cash for work project payment system Element

Description and issues arising

Database of beneficiaries

• All personal information on workers was manually documented. Data included name, age, sex, employment duration and photograph • Data were entered and maintained on a computerized database within the Management Information System (MIS) Unit at LACE.

Identification mechanisms

• Digital Identification Cards were issued upon receipt of signed contract. Cards included a participant’s name, position, community, signature, period of validity and a unique identification number. Agency logos were included to product against fraud. • Workers were required to present identification cards at time of payment. • Workers were issued with a contract for temporary employment

Currency

• US Dollars, through direct cash payments from a commercial bank.

Delivery Instrument and Point of Payment

• Direct cash payments made at local EcoBank branches; or specially commissioned mobile units. • Direct cash payments at EcoBank made in: Greater Monrovia and Paynesville, Montserrado County; Ganta, Nimba County; Buchanan, Grand Bassa County; Kakata, Margibi County; Pleebo, Maryland County; Zwedru, Grand Gedeh County • Mobile payments made in: Rural Montserrado, Sinoe, Grand Cape Mount, Lofa, Gbarpolu, Bomi, River Cess, rural Grand Bassa, Maryland, rural Grand Gedeh, rural Nimba, River Gee, rural Margibi and Grand Kru Counties

Payment Schedule and Frequency

• Initially made on bi-weekly basis. Payments changed to monthly basis given logistical challenges and need to establish mobile bank units. • Dedicated service agents at EcoBank branches were eventually assigned to only issue beneficiary payments, given the large numbers arriving on pay day in Montserrado, Nimba, and Margibi counties.

Reporting and Reconciliation

• Community Facilitators are responsible for processing worker payrolls using the following forms given to them by LACE: 1) Daily Attendance Sheet 2) Workers’ Payroll Form and 3) Table of Indicative Task Rates. Once reviewed by the LNGO, they are forwarded to the Community Finance Unit (CFU) at LACE for final verification. • Payrolls are then reconciled by EcoBank at the end of payment process. A monthly bank statement is submitted to LACE for reconciliation and accounting purposes. Signed copies of the payrolls are returned with a credit advance for the summation of the uncollected/unpaid amounts by beneficiaries.

Communication Aspects

• LACE notifies EcoBank two weeks in advance of payment date. EcoBank is linked with the local facilitators to coordinate payments. Community facilitators advise workers verbally of payment information (date, time and place of payment).

Source: Making Enterprises (2010).

140

A World Bank Study

was unavailable. To support the process, a range of payment control instruments were introduced, including a contract for workers, daily attendance sheets, monthly payroll sheets and identification cards. These elements established a functioning identification and registration system where beneficiary information could be tracked. Daily attendance and payroll sheets were maintained by the work site group leader and collected on a weekly basis by the local community contractor. In the absence of a nationwide identification system, the introduction of a beneficiary ID card was critical, although some isolated identify theft cases were still reported. There were also procedures put in place to handle recurrent problems such as lost identification cards and absences on payment days. Significantly, across all focus group discussions, beneficiaries reported that payments had been made in a timely and accurate manner. The program had initially devised a system of bi-weekly payments to beneficiaries, which was eventually changed to monthly payments because of logistical limitations. Feedback from some focus groups suggested that the use of a commercial bank had inspired beneficiary confidence in the project and reduced payment delays. For many beneficiaries, the interaction with EcoBank was their first with any formal banking institution. The success of this arrangement highlights the value of maintaining proper incentives for commercial bank collaboration, including the possibility of fee payments, the ability to meet corporate responsibility obligations or access to new markets. Despite the general success of the payment scheme, some potential challenges going forward include:

â&#x2013;

Data collection given limited technological and human capacity: Local community facilitators lacked IT capacity or internet connectivity and this caused difficulties when scanning payrolls from LACE to EcoBank. Staff of some local NGOs also lacked computer skills and therefore depended on LACE to computerize the payroll before submission to EcoBank. This underscores the need for simple management information systems and procedures to mitigate bottlenecks in data transfer. Limited ability to provide bank services in rural areas, particularly during the rainy season: For negotiated service charges, EcoBank currently provides mobile banking services for clients where they have no branches.5 According to LACE, other commercial banks could be used in the future to disburse payments to workers in communities where they have branches but EcoBank does not, provided that the World Bank consents to using multiple banks for one project. LACE further suggested that electronic payments could be considered in the future as banking infrastructure develops across Liberia. This would also depend on program workers agreeing to receive their wages through savings accounts maintained with commercial banks. Ensuring adequate security, particularly where payments are being made at mobile bank units: Under the CfW Project, EcoBank was responsible for the safety of its team members and the security of the cash taken into the field. The bank liaised with the Liberian National Police (LNP) to ensure protection of the team. As the program increases visibility and predictability in its payments, this area may require additional vigilance.

Poverty and the Policy Response to the Economic Crisis in Liberia

141

5.4. Monitoring and Evaluation

Monitoring was a key function in the implementation of the CfW Project, for which overall coordination was delegated to a specially hired Monitoring and Evaluation Officer at LACE. Priority areas for monitoring and evaluation were laid out in the project’s Operational Guidelines, with particular emphasis placed on the following issues:

■ ■ ■ ■ ■ ■

Selection of workers: Is the process of beneficiary selection effective and does it succeed in reaching most vulnerable segments of the population? Gender and youth: Are at least 30 percent of the workers women? Are the activities designed in a manner that encourages the employment of women? Are youth benefiting from the project? Production and outputs: Are outputs produced on time? What about quality of outputs? Awareness and communication: Is sufficient information given to communities and other stakeholders? Does this information cover all the key points described in section VII [of the Operational Guidelines] on communication? Payments: Are payments made on time? Do participants receive the correct amount of money? Is cash distribution satisfactory? Markets: Is there any impact of the project on local market prices?

Field level monitoring was the responsibility of the Community Facilitators, who were given a number of tracking tools including worker contracts, daily attendance sheets, a table of indicative task rates and bi-weekly project summary documents for simplified and regular reporting. Some tasks, such as monitoring daily attendance, were further delegated to workgroup leaders. The monitoring indicators set out in the Operational Guidelines were generally comprehensive and sufficient to provide necessary information. The regular Monthly Monitoring Report covered all indicators except the impact of the project on the local market prices. Information gathered on worker selection, female participation and timeliness of payments allowed regular assessment of key program goals. There were some challenges in project monitoring and interviews revealed suggestions for future improvements:

■

Staff Constraints: Monitoring was constrained by the fact that the project had only two program staff—the CfW Project Manager and the CfW Project Monitoring and Evaluation Officer. The broad geographic spread of the projects coupled with bad road conditions made monitoring even more challenging. To address the staffing shortage, LACE periodically seconded two staff members from its regular Community Empowerment Project (CEP) to provide monitoring and evaluation support. The task of monitoring of market prices in beneficiary communities to assess the impact of the program on local economies was delegated to the Ministry of Agriculture but was not completed. When asked why changes in market prices had not been monitored as planned, the focal person at the Ministry of Agriculture cited a lack of communication with field staff and an overall lack of capacity for monitoring projects in the field. Future projects should either consider alternative means of monitoring market impacts or account for constraints within the Ministry of Agriculture.

142

A World Bank Study

■

While it did not appear to be a serious concern in this program, future provisions should be made for a grievance mechanism to handle disputes between any of the actors involved in the program, including workers, non-workers, community facilitators, payment teams and program staff.

6. Measuring Progress: Key Impact Findings and Project Feedback While identifying some areas for improvement, the two previous analyses of the CfW Project found it to be a successful effort, especially in light of the speed with which the intervention was developed and implemented. All 680,000 work days were completed in approximately 20 months. The project far exceeded its target of 30 percent female participation, with a total of 46 percent female workers. This participation was uneven, however, ranging between 30-70 percent by county. Though there were no hard targets for youth participation, the project was successful at engaging young people. Nearly 60 percent of project participants were classified as youth (defined in Liberia as those up to age 35). Table 8.6 provides the complete breakdown of workers by age. Table 8.6: Distribution of project participants by age Ages

% of participants

18–24

12.4

25–34

46.7

35–44

23.7

44+

17.2

Total

100

Source: Administrative data from LACE.

As is explained in more detail below, the wages and projects provided positive dividends for participants and their communities. 6.1. Quantitative Impacts

In reviewing the performance of the project, the quantitative analysis evaluated the project’s targeting efficacy, participants’ use of project wages, the project’s overall level of poverty reduction and the efficiency of the project itself. Targeting

In total, the project review found that roughly 80 percent of project participants could be defined as poor, indicating a generally successful targeting process. The team estimated the targeting performance using a variety of indicators (wealth asset and estimated consumption level) as well as methods (different matching techniques for estimating consumption as well as wealth indicators were considered), with the share of the participants that was estimated to be poor varied from 60 percent to 90 percent, with the estimate of about 80 percent of participants being poor being the most likely. This fairly good performance is partially attributable to the high overall levels of poverty in the country. With 68 percent of the total population considered poor, Liberians in the second and even third consumption quintiles meet the targeting criteria for the project. Table 8.7 lays out the distribution of project participants among the distribution quintiles. The

Poverty and the Policy Response to the Economic Crisis in Liberia

143

program seems to have been successful at preventing significant inclusion errors, but somewhat weaker at reaching the lowest quintile. The fact that the program reached the poor, but less well the extreme poor is likely to be due in part to the ‘first-come, first-served’ nature of the targeting at the community level. This targeting mechanism meant that individuals from better off households were not likely to be favored, but it also meant that some of the poorest individuals may not have been able to participate as much, among others because they tend to live in more isolated areas. Table 8.7: Distribution of project participants by welfare quintile Consumption quintile

% of participants

Poorest quintile

15.2

40.6

28.7

11.2

Wealthiest quintile

4.2

Total

100

Source: Backiny-Yetna, Wodon, and Zampaglione (2012).

Use of Project Wages

The evaluation provided encouraging results about the possible long-term impact of the project on livelihoods and economic opportunity. The survey asked participants about uses of their project wages and revealed that a significant portion of wages were going toward long term investments in children’s education (31 percent of income was used for educational expenses such as school fees) as well as their own future earnings (14.2 percent of income was used for both farm and non-farm investments). Table 8.8 summarizes participants’ reported uses of their wages. These results may be due in part to the timing of payments (in some cases just after school fees were due), but the fact that the incomes received represented a high share of the household’s total. Table 8.8: Use of project income by households Category

Share of funds

Education

Living expenses

Health care

8.4

Farm investment

8.2

House repair

8.2

Non-farm investment

Debt repayment

3.6

Acquiring HH assets

2.9

Transfer

1.3

Funeral

1.2

Celebrations

1.1

Total

100

Source: Backiny-Yetna, Wodon, and Zampaglione (2012).

144

A World Bank Study

Overall, there were no major differences in the use of the wages earned between male and female participants. Men reported somewhat higher rates of education-based spending than women (32.4 vs. 29.3). While men reported equal levels of farm and nonfarm investments (7.3 share for each), women reported much higher levels of farm-based investment than non-farm investments (9.4 vs. 4.5). Women also reported higher utilization of funds for debt repayment than men (4.7 vs. 2.7). But none of these differences were statistically significant. Poverty Reduction

To assess the overall impact of the project on poverty reduction, the quantitative analysis began by estimating the size of wage losses incurred as a result of program participation. Such losses could have been incurred either by leaving a job to take up work with the project or by paying a substitute worker to cover existing obligations such as child care or farm maintenance while participating in the scheme. The evaluation did not find evidence that significant wage losses had occurred, likely due to the high pre-existing levels of unemployment in the project communities. That is, 93 percent of the wages earned through the program were estimated to be net additional wages. Because the benefits of some occupations that program participants may have held is difficult to assess, this estimate is likely to be too optimistic. Still, given the lack of gainful employment for most individuals in Liberia are limited, it is likely that most of the wages obtained through the program contributed directly to additional consumption by households (including investments in human development). The analysis next showed that the project had reduced the number of participants technically living in poverty by 5 percent. This simply means that most participants remained poor, because they had levels of consumption substantially below the poverty line before the program. More importantly though, the project did have a sizable effect on reducing the poverty gap among program participants (the poverty gap takes into account the distance separating the poor from the poverty line apart from the proportion of the population that is poor). The results showed a 21 percent decline in the poverty gap from the baseline on a yearly basis (31 percent for the squared poverty gap), indicating that, while participants were still poor after the program, they were substantially less poor than they had been before it. This result was remarkable given the short duration of benefits. Cost Effectiveness

To evaluate the cost effectiveness of the project, the authors considered three main determinants: the wage share (the total wages paid as a proportion of project costs), the targeting performance (the ratio of wages that reached the poor out of the total wages paid) and the proportionate wage gain (ratio of net wage benefits (accounting for participation opportunity costs) to the total wages paid to the poor) (for a discussion of these parameters, see Ravallion, 1999; Backiny-Yetna, Wodon, and Zampaglione, 2012, for the results). Measured against international markers for similar programs, CfWTEPâ&#x20AC;&#x2122;s performance in each regard was relatively strong. Using project administrative data, the wage share was estimated at 68 percent. Other projects around the world (though all of larger scale) have had rates ranging from 40-85 percent. Liberiaâ&#x20AC;&#x2122;s performance on this indicator may be slightly misleading, however, given the very limited government role in project management, which sets it apart from longer established, government run schemes where costs are absorbed differently in the calculation.

Poverty and the Policy Response to the Economic Crisis in Liberia

145

As discussed above, the targeting performance was estimated at 80 percent, putting it on par with similar programs in Ethiopia and Argentina, but above a range of other programs. The proportionate wage gain was also quite high (93 percent) because, as outlined earlier, there were very few other income-generating alternatives available for participants, meaning that displacement of other paid work was very low. The product of these three ratios produced the final estimate for the projectâ&#x20AC;&#x2122;s cost-effectiveness:

The analysis found a 0.51 overall project cost effectiveness, meaning that the cost of transferring US$1.00 in net wage benefit to a poor participant was US$1.96 (including the US$1.00 net wage). This rate is lower than the 0.55 effectiveness standard in Ethiopiaâ&#x20AC;&#x2122;s Productive Safety Net Program, but higher than rates seen in many other programs with weaker targeting efficiency. Moreover, this calculation does not attribute any value to the actual projects undertaken by CfW Project labor. If future evaluations are able to capture the added value of community improvements produced by the work, the overall cost effectiveness would improve. 6.2. Qualitative Perspectives

In addition to reviewing specific design choices of the CfW Project, the qualitative analysis included summary results from key informant interviews and from focus group discussions held with both participants and other community members. In analyzing the results of the focus groups, it is important to note that the groups were not statistically representative of the program participants or of their communities. 58 percent of focus group participants were female and 43 percent were between the ages of 18 and 35. While each group offered different perspectives on the project, one issue was common to all three groups. Beneficiaries, community members and stakeholders all expressed a desire for future work-based interventions to integrate a training component that would provide workers with skills to help them beyond the end of the project cycle. Beneficiaries

Focus groups with beneficiaries anecdotally confirmed many of the findings of the quantitative survey. Participants reported that the impact of the project had gone beyond immediate consumption smoothing, allowing them to invest for the future. Nearly half of rural female workers reported using wages for small-scale income-generating activities, adding knock-on value to the project after the initial income boost. The focus groups also reiterated the limited wage substitution effect resulting from the program. Respondents noted the lack of employment opportunities before the project and that most of the work that did exist was casual labor, which could be accommodated with the flexible work schedule. Participants were generally satisfied with the project arrangements. The payment system was an area of particular success, with wages regularly delivered on time and accurately. Use of a commercial bank increased participant faith in the payment system. There appeared, however, to be incomplete sensitization about the projectâ&#x20AC;&#x2122;s purpose as a one-off food crisis response. Participants generally understood that the work was

146

A World Bank Study

intended to help the poor, but placed more emphasis on it for the community assets produced than for its intended economic effects. To more accurately understand worker perceptions about the value of the project, the quantitative assessment included a series of short questions on subjective indicators. The results are presented in Table 8.9. Table 8.9: Perception of project value by participants Question

% of respondents answering yes

Has done this type of work before

27.9

Has gained technical skills

75.8

Has learned to be punctual

96.4

Has learned through the program

82.7

Has gained skills to help later

77.2

Assignment was clear

87.6

Has worked too hard

13.1

Has paid a bribe

5.3

The salary was fair

87.9

Experienced payment delays

19.6

Project ID was first ID ever

67.1

First time business with a bank

93.8

Source: Backiny-Yetna, Wodon, and Zampaglione (2012).

Of particular note is the exceptionally high level of participants interacting with the formal banking sector for the first time. This number was very similar between men and women (93.3 vs. 94.3). The only significant gender gap was in response to the question regarding whether the participant has done this type of work before. Some 34.9 percent of men reported they had, while only 19.5 percent of women had undertaken similar work. Despite this initial gap in experience, male and female responses to the other skills and learning questions were very similar. Communities

Non-participants from selected communities were also interviewed in the focus group process. The project was generally well received by community members, who noted the value of the projects undertaken for their communities. Respondents highlighted improved road conditions and access, better maintained public areas and improved drainage. Though there are no reliable statistics to document this perception, multiple participants felt that the increase in work for local residents (particularly youth) had lowered crime rates and improved peace and security. While they generally understood and accepted the basic criteria for project participation, non-participants had a weaker understanding of the participant and project selection processes. When asked to explain how workers were notified and selected, nonworkers frequently had either inaccurate information or did not know what the process had been. This response highlights a possible area of improvement for future iterations of cash for work programming. Community members who had been available for work but not selected for the CfWTEP expressed a strong desire for future projects to help more members of the community.

Poverty and the Policy Response to the Economic Crisis in Liberia

147

Other Stakeholders

In preparing the qualitative assessment, 32 different organizational heads, government officials and local leaders were interviewed about their perceptions of the project. Even when the individuals interviewed had little direct experience with the project, their understanding of its basic framework was good and support for its objectives was strong. As with the other groups, most stakeholders wanted to see work and life skills integrated into CfW programming. They also wanted employment-based projects integrated into a wider set of social protection interventions. Of the groups interviewed, local government authorities (LGAs) raised the most concerns with the project. These issues involved a desire for an increased LGA role in both participant and project selection as well as questions about the productive value of the projects when weighed against their costs. Local authorities also noted that some of the projects (such as roadside brushing in rural areas) displaced work which had previously been undertaken through mandatory unpaid community service. After community members were compensated for undertaking the work, it was more difficult to maintain the projects on an unpaid basis. In general, respondents noted weak asset maintenance after project payments stopped. Subsequent interviews with project implementers revealed more issues to be considered in future project design efforts. The first is an extension of one of the LGA concerns: the productive value of the work undertaken. While some projects (clearing of an agricultural center and an overrun airstrip) ended up creating tremendous community value, others were of lower lasting value. Reconsidering how CFs approach the project selection process with participants could lead to more productive work for laborers and their communities. Rather than simply listing the types of work that could be undertaken (such as roadside brushing), CFs could present examples of more complex projects that have been completed in other areas or scope similar activities and needs in each participating community and present them as possible project options. While the first-come first-served method for participant selection was relatively successful at achieving its targeting goals, implementers and other stakeholders have concerns about the fairness of the procedure as the sole targeting criterion. Because of its reliance on word of mouth and physical availability, first-come first-served selection tends to favor the most informed and connected community members, to the exclusion of those without strong social ties or living in more remote areas. To compensate for this natural bias of the system, stronger outreach efforts should be made to capture those who might otherwise be disadvantaged by a first-come first-served process.

7. Moving Forward: Lessons Learned and Future Planning While the CfW Project was intended as a one-off intervention to address the immediate effects of the food price crisis, its operational successes are informative for future social protection programming both in Liberia as well as in other low-income, fragile countries. This section reviews some of the major lessons drawn from the CfWTEP and considers future CfW interventions in light of these lessons and Liberiaâ&#x20AC;&#x2122;s evolving policy context. 7.1. Operational and Policy Lessons

Implementing social protection programming in every geographic region of a country on tight timelines is difficult under the best of circumstances. In a low-income, post-

148

A World Bank Study

conflict country with poor infrastructure and severely constrained government capacity, the obstacles are immense. The challenges of low-income, low-capacity settings impact both design elements of the projects as well as the operational arrangements necessary for implementation. 7.2. Overcoming Government Capacity Constraints

As has been discussed above, exceptionally limited government capacity for project implementation and oversight necessitated the use of outside implementers for the CfW Project. While not without some constraints of their own, these private actors (LACE for overall coordination, NGOs as Community Facilitators, and EcoBank for payments) offered flexibility and established networks that could be quickly leveraged for project use. Utilizing existing private capacity rather than building entirely new networks was essential for ensuring a timely response. The successful development of a private payment scheme was particularly noteworthy in light of the challenges that previous cash for work projects had faced in assuring safe, timely, and accurate payments in many areas of the country. Future interventions, especially those with broader mandates, could expand upon the relationships being established with commercial banks through CfW programs. Given the exceptionally high number of CfW Project participants using bank services for the first time and the expressed desire for more skills training, future projects could incorporate financial literacy training modules, such as money management, into the project design. The bank contracted for payment delivery could be used to conduct the skills trainings. In areas where banking services are available, participants could access information about savings accounts and other financial tools. While care must be taken to assure that individuals with weak financial literacy skills are not subjected to any inappropriate pressure, this interaction with the banking sector could provide valuable long-term financial benefits for participants. 7.3. Utilizing Local Knowledge and Community Participation

The high level of community participation in key CfW Project decision points was a successful design element that should be replicated in future efforts. Local decision-making was maintained at key steps throughout the process: local government authorities decided the areas where sub-projects would be located, local leaders took a role in defining vulnerability criteria and recruiting vulnerable households for project participation, and the participants themselves selected which projects would be implemented on behalf of their communities. This inclusive approach followed best practices for involving impacted communities in decision-making and also had positive impacts for the efficacy of project management. Despite the efforts of donors and the government, central-level coordination of county-based projects is still weak in Liberia, so allowing local authorities to site the subprojects was an efficient mechanism for assuring that projects were placed in the areas with highest need. This is also an important tool for strengthening the relatively weak hand of local government authorities, a key priority of the Liberian PRS. The CfW Project imposed relatively few criteria on the participant targeting process, allowing community leaders to define additional vulnerability criteria according to local needs and understandings. Together, the basic program guidelines and the community refinements produced a high level of targeting accuracy. This is partially attributable to

Poverty and the Policy Response to the Economic Crisis in Liberia

149

the exceptionally high level of poverty in Liberia, but is also a consequence of community involvement. Communities not only refined their own criteria, but they also were involved in encouraging the participation of vulnerable households. Using this local knowledge to engage target populations was more efficient than using outside sources and increased community acceptance of the selection criteria. As with the site selection process, the transparency and consistency of adopting additional selection criteria could be improved, but the flexibility should be maintained. 7.4. Setting Wages in Complex Local Circumstances

As discussed in the project design section above, setting the wage level for the project involved balancing a complex set of social, political and economic factors. Pressure to maintain consistency with established wage precedents came from the government and other implementing partners. Fears about the negative impact of a non-conforming rate on project implementation also pushed for the higher rate. Despite clear indications that the final project rate was higher than the local market wage, the traditional economic reasons for lowering the rate were not as compelling in the Liberian context. Two factors generally call for setting a wage rate below the local average: the fear of creating upward pressure on market wages by increasing the bargaining power of workers able to command higher wages and the fear of creating large substitution effects if the wages are more attractive than those in the market at large. If realized, the substitution effect would undermine the efficacy of the wage rate as a targeting mechanism, as higher income individuals would find the wage attractive, potentially crowding out poorer individuals for whom the program was intended. In the Liberian context, high rates of poverty and unemployment mitigate those risks. Because of the massive surplus of labor, workers are unlikely to gain any market leverage, regardless of the wage rate. Those same soaring rates of unemployment also cut against the likelihood that workers would have productive work to substitute out of, even if they were willing to take that risk to participate in a temporary, non-repeatable period of work. These factors also point to the weakness of attempting to use the wage rate as a precise targeting tool. Because levels of poverty and demand for the program are so high, a modest reduction in the wage rate would be insufficient to bring demand down to the level of available spots. To accomplish that task, the wage would likely have to be reduced to such a low level that it would undermine the poverty-alleviation intent of the program. Thus, in low-income contexts like Liberia, additional targeting measures must be adopted in tandem with the wage rate. The low-capacity setting of Liberia also blunts the final argument in favor of a lower wage rate. In other circumstances, a strong equity argument could be made that high demand for job opportunities calls for creating as many positions as possible within a given financial envelope. A lower wage rate would allow more vulnerable individuals to participate in the program and spread its benefits further. Greater numbers of work days would also increase the projects that could be completed, benefitting communities with more public works. In the Liberian example, lowering the rate from US$3.00 to US$2.50 would have increased the number of participants and of completed projects by 17 percent. Given the implementation constraints of the country, however, these gains cannot be assumed. Though scaling up of the initial CfW Project is possible, there are real limits to the number of participants and projects that can be included without sacrificing pro-

150

A World Bank Study

gram quality. As such, low-capacity contexts require a careful balancing between meeting local needs and respecting local implementation capabilities. 7.5. Sustaining Participants, As Well As Projects

The Cash for Work Temporary Employment Project was a one-off, crisis response program intended to provide immediate and short term relief from rising food prices. As such, reviewing the sustainability of the effort should focus less on sustaining the emergency response and more on sustaining the participants and communities in need. The responses of program participants and other community members in the focus groups highlight the local desire for this approach. Repeated requests for training in life skills, entrepreneurship and technical fields reflect an understanding that, while temporary work was necessary to make it through the crisis, it was not sufficient to meet the long term needs of vulnerable families. Similarly, the high proportion of wages used for investment is a sign of participant understanding about the need to reduce their own longer-term vulnerability. To this end, future cash for work programming should take these needs into account, building not just short term opportunity but also longer-term systems for self-sustainability. This dual-pronged approach is the essence of productive safety net programming, catching those in greatest need and, where possible, offering them sustainable outlets for escaping poverty traps. Emphasis should be placed on providing skills transfers for participants and added value for communities. While this is more difficult in a crisis context, there are still opportunities to leverage program design elements (such as the use of a commercial bank for financial literacy education, as outlined above) to increase the value of the project with minimal loss of efficiency and potentially substantial rewards. 7.6. Investing in Well-Targeted Programs

How well did CfWTEP perform in comparison to other measures implemented by the government in order to respond to the food, fuel, and financial crisis? Given that the objective of the program was to provide temporary relief to help households cope with shocks, a key statistics obtained in the analysis was that for every dollar spent, 51 cents resulted in increased incomes for the poor. This is a good performance in comparison to the impact of some other programs and policies. For example, on the basis of the rice consumption patterns observed in the 2007 national CWIQ survey, it is likely that less than half of the tax cuts implemented on food imports benefited the poor. As to the income tax cuts also implemented in part to help households cope with the crisis, most of them did not benefit the poor. 7.7. Promoting Gender Equity and Empowerment

Almost half of the program participants were women, and women were paid the same wage as men under CfWTEP. By contrast data from the 2007 CWIQ survey suggests that even though labor force participation rates are similar by sex, 39.5 percent of men work as unpaid family members, versus 56.5 percent of women. The CWIQ survey data also suggests that 25.5 percent of men work as paid employee in Liberia, versus only 8 percent for women. Earnings for men are also much higher when they work that they are for women. This suggests that the program was â&#x20AC;&#x2DC;pro-womenâ&#x20AC;&#x2122; in the sense that the share of the benefits that accrued to women was much larger than what is observed on the

Poverty and the Policy Response to the Economic Crisis in Liberia

151

labor market. This emphasis on benefits for women is likely to be even stronger under the new YES program (discussed below) that sets a target participation rate for women of 50 percent. 7.8. Implementing Light â&#x20AC;&#x153;Q2â&#x20AC;? Evaluations

The results from both the quantitative and qualitative evaluations were obtained at low cost (for example, the quantitative survey cost only $20,000 to implement in terms of data collection) and rapidly, and were very informative for shaping the YES project. This suggests that at least for programs such as CfWTEP that aim to provide short term benefits towards poverty reduction, light evaluation methods can be very useful from an operational point of view, while still maintaining high quality standards in program evaluation. The fact that both quantitative and qualitative data were available apart from the administrative data collected by LACE made it feasible to triangulate the results and sources of information to make sure that the evaluation findings were robust. The fact that the team that carried the evaluation worked closely with the team implementing the project also helped in incorporating the results from the evaluation into the design of the new project. 7.9. The YES Project

In Liberia, the Youth, Employment, Skills (YES) Project is taking these lessons into account. It is the next generation of the CfWTEP, but is not intended exclusively as an emergency response, and has added components to improve the long term value of the project for participants. The YES Project has two primary components. The first, called Community Works, provides temporary employment similar to the CfWTEP for 45,000 individuals. Building on the CfW Project experience, it includes one day per week of non-cognitive skills development for participants, compensated at the same rate as a regular work day. It also includes a few notable program changes from CfWTEP. First, the project has a much stronger focus on youth. Nationwide, 75 percent of the population is under the age of 35, and the youth population is the hardest hit by un- and underemployment. Low practical skill levels and poor educational attainment make it exceedingly difficult for youth to compete for the limited work that is available. To address this need, the YES Project mandates that 75 percent of participants be between the ages of 18 and 35. This still allows the participation of vulnerable older individuals, but focuses the project on addressing youth unemployment and skills development. Second, provisions for gender inclusiveness have been improved. Pregnant women are allowed to participate, and child care is available for women with young children. Alternative work options are provided to accommodate these participants, and require additional sensitization for other workers to maintain cohesion among all participants. Finally, the new project is improving the supervision ratio used in CfWTEP. In that project, there was one supervisor for each sub-project of 500 workers. In the YES Project, that ratio is lowered to 1:100, allowing improved oversight of workers and project implementation. The second main piece of the YES Project is the Employment through Skills Training component, which finances both formal and informal skills training for 4,500 individuals. In line with Liberiaâ&#x20AC;&#x2122;s National Capacity Development Strategy, the training will be

152

A World Bank Study

demand-driven and focused on providing skills that are in immediate need in the formal and informal sectors of the economy. Each component of the YES Project also includes funding to support the institutional development of various GoL actors and policies. These funds will help improve coordination of cash for work programming and the development of standards and policy for technical and vocational education and training. 7.10. Transitioning to a Broader Social Safety Net

The YES Project and other cash for work programming efforts are a vital part of Liberia’s emerging social safety net. While the shape and strength of the net are not yet apparent, the need for one has been explicitly recognized by the government and by its development partners. Strong government leadership will be required to make the transition from wellintentioned but sporadically coordinated interventions to a coherent system for social protection. Steps have already been taken in this direction, with the recent establishment of the National Social Protection Secretariat. The Secretariat will lead the development of the National Social Protection Policy as well as provide the social protection inputs into the next PRS. This represents a significant commitment to protection-focused policy making. The challenge of this arrangement, however, is that the Secretariat is housed in the Ministry of Planning and Economic Affairs, which is not an implementing ministry. Mandates for the implementation of social protection programs are split between at least half a dozen ministries, including Health & Social Welfare, Labour, Education, Agriculture, Public Works and Gender & Development. Coordination among these ministries and other government agencies has been weak in the past, and will require both sustained focus from the government as well as external coordination among development partners. While the obstacles to coordination, financing, and implementation are substantial, the potential impact of a well-functioning set of safety net programs is even greater. Orphans and vulnerable children, the elderly, the disabled, unemployed youth and many others have few options for survival in the absence of social protection programs. For the two out of three Liberians that are living in poverty today, any lifeline will be a welcome one.

Notes 1. This chapter was prepared as part of an evaluation of Liberia’s cash for work program, in preparation of a subsequent World Bank operation. The results of the analysis were presented in Monrovia among others at a workshop in March 2011. The views expressed here are those of the authors and need not reflect those of the World Bank, its Executive Directors or the countries they represent. The chapter is reproduced with minor modifications from Social Protection Paper No. 1114 at the World Bank. 2. Though the Cash for Work project is the focus of this chapter, it was not the Bank’s only crisis response in Liberia. Other components included agricultural supports for farmers and additional resources for the existing school feeding program. 3. One CF was hired in each of Liberia’s 15 counties, with the exception of Lofa County where 3 CFs were retained given the diversity of local dialects. 4. Part II, Chapter 6, Section 513 of the Labor Law of Liberia stipulates that “an unskilled laborer shall be paid for his work at the rate of not less than twenty-five cents ($0.25) an hour if he is an

Poverty and the Policy Response to the Economic Crisis in Liberia

153

industrial laborer, and not less than One Dollar and Fifty Cents ($1.50) per eight-hour day if he is an agriculture laborer, exclusive of fringe benefits”. Also, Chapter 3, Section 21 (c) of the Labor Law of Liberia classifies anyone who works for less than eight hours a day as “casual worker”, for whom there is no minimum wage rate. 5. EcoBank charges US$1,200.00 to deliver payments to workers in remote communities where they have no branches, as compared with US$1,000.00 in communities with nearby branches.

References Adato, M., and L. Haddad. 2002. “Targeting Poverty through Community-Based Public Works Programmes: Experience from South Africa.” Journal of Development Studies 38(3): 1–36. Backiny-Yetna, P., R. Mungai, C. Tsimpo, and Q. Wodon. 2012. “Poverty in Liberia: Level, Profile, and Determinants.” In Q. Wodon, ed., 9–34. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Backiny-Yetna, P., Q. Wodon, and G. Zampaglione. 2012. “Impact of Labor Intensive Public Works in Liberia: Results from a Light Evaluation Survey.” In Q. Wodon, ed., 155–171. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Coady, D., M. Grosh, and J. Hoddinott. 2003. “Targeting of Transfers in Developing Countries: Review of Lessons and Experience.” Regional and Sectoral Studies. World Bank, Washington, DC. Food and Agriculture Organization of the United Nations (FAO). 2008. The State of Food Security in the World 2008. FAO, Rome, Italy. Government of Liberia and United Nations Joint Report. 2010. The State of Food and Nutrition Security in Liberia: Comprehensive Food Security and Nutrition Survey. Monrovia. Grosh, M., C Del Ninno, E. Tesliuc, and A Ouerghi. 2008. For Protection and Promotion: The Design and Implementation of Effective Safety Nets. Washington DC: World Bank. Ivanic, M., and W. Martin. 2008. “Implications of Higher Global Food Prices for Poverty in Low-Income Countries.” Agricultural Economics 39: 405–416. Making Enterprises Inc. 2010. “Liberia Cash for Work Temporary Employment Project: A Qualitative Assessment.” Mimeo, Monrovia. Ministry of Agriculture of the Republic of Liberia. 2007. Comprehensive Assessment of the Agriculture sector in Liberia: Volume 1: Synthesis Report. Monrovia. Ravallion, M. 1999. “Appraising Workfare.” World Bank Research Observer 14: 31–48. Republic of Liberia. 2006. Comprehensive Food Security and nutrition Survey (CFSNS). Monrovia. Subbarao, K. 2003. “Systemic Shocks and Social Protection: Role and Effectiveness of Public Works Programs.” Social Protection Discussion Paper No. 302. World Bank, Washington, DC. Schubert, B. 2008. “Social Protection Issues in Liberia.” Mimeo, Inter-Agency Programming Team of the United Nations. Teklu, T., and S. Asefa. 1997. “Factors Affecting Employment Choice in a Labor-Intensive Public Works Scheme in Rural Botswana.” Economic Development and Cultural Change 46(1): 175–86. Teklu, T., and S. Asefa. 1999. “Who Participates in Labor-Intensive Public Works in SubSaharan Africa? Evidence from Rural Botswana and Kenya.” World Development 27(2): 431–38.

154

A World Bank Study

Tsimpo, C., and Q. Wodon. 2012. “Rice Prices and Poverty in Liberia.” In Q. Wodon, ed., 85–98. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. United Nations Joint Assessment. 2008. “The Impact of High Prices on Food Security in Liberia.” Mimeo, Monrovia. Wodon, Q., C. Tsimpo, P. Backiny-Yetna, G. Joseph, F. Adoho, and H. Coulombe. 2008. “Potential Impact of Higher Food Prices on Poverty: Summary Estimates for a Dozen West and Central African Countries.” Policy Research Working Paper 4745. World Bank, Washington, DC. Wodon, Q., and H. Zaman. 2010. “Higher Food Prices in Sub-Saharan Africa: Poverty Impact and Policy Responses.” World Bank Research Observer 25: 157–176. World Bank. 2009. “Liberia: Employment and Pro-Poor Growth.” Report No. 51924-LR. World Bank, Washington, DC. ———. 2010. “Project Appraisal Document on a Proposed Grant in the Amount of US$16.0 Million from the Africa Catalytic Growth fund (US$10.0 Million) and the Crisis Response Window (US$6.0 Million) to the Government of Liberia for the Liberia Youth, Employment, Skills Project.” Report No. 53626-LR. World Bank, Washington, DC.

CHAPTER 9

Impact of Labor-Intensive Public Works in Liberia: Results from a Light Evaluation Survey Prospere Backiny-Yetna, Quentin Wodon, and Giuseppe Zampaglione1 The President and the Government of Liberia have placed a strategic priority on youth employment and have asked development partners for support in these areas. The Office of the President, the government, and development partners have undertaken over the last two years several steps to generate immediate employment. These include the Cash for Work Temporary Employment Project (CfWTEP) financed by the World Bank Food Price Crisis Response initiative. This chapter provides an assessment of the experience to-date with CfWTEP, as well as a discussion of options for the expansion and continuation of the program. Over 2009-2010, CfWTEP has provided jobs to 17000 people in the country. In order to assess the program and suggest options for its continuation, a light evaluation survey was implemented in the country in NovemberDecember 2009 with four objectives: (i) Assessing the targeting performances of the program; (ii) Measuring the wage substitution effects among the participant; (iii) Analyzing the patterns of use of the wages received by households; and (iv) Documenting other aspects of the program. The results suggest that the performance of the program in all four areas of the evaluation was relatively good, but targeting to the very poor could be improved in the future.

1. Introduction Liberia is at an inflection point, moving from transitional post-conflict recovery to laying the foundations for long-term development. The Liberian economy, institutions, and human capacity suffered the gradual and deep destruction of a protracted civil war, the origins of which correlate with a spiraling pattern of bad governance, reckless political and economic interest, regional instability, and the marginalization of huge sectors of society (on conflict and social cohesion in Liberia, see for example Richards et al., 2005). Young Liberians and the rural population were especially hard hit. However, remarkable progress in the post-conflict economy and political process has moved the country to a turning point. The elections of 2006 ushered in a democratic government intent on redirecting the economy by creating broad ownership of the political structure. Since its election, the government has set in motion a challenging reform agenda centered on its 155

156

A World Bank Study

Poverty Reduction Strategy as complemented by reforms to improve economic governance, overall transparency, economic growth, and social development. In the space of three years, from 2005 to 2008, Liberiaâ&#x20AC;&#x2122;s economy has witnessed a steady uptrend with an average growth rate of 7.4 percent. Furthermore, foreign direct investment has increased, particularly among the mining, rubber, andâ&#x20AC;&#x201D;more recentlyâ&#x20AC;&#x201D;forestry sectors. Despite this progress, the situation in Liberia remains fragile and exposed to the global crisis. Current per capita GDP is estimated at US$249, and an estimated 63.8 percent of Liberia 3.5 million people live below the poverty line, with 47.9 percent living in extreme poverty (Republic of Liberia, 2008; Backiny-Yetna et al., 2011). In addition, recent growth has been adversely affected by the global crisis, with GDP growth slowing to 4.9 percent in 2009. The increase in rice prices is likely to have had a large negative effect (Tsimpo and Wodon, 2012a), and some of the measures taken by the government to help the population cope with the crisis may not have reached the poor very well (Tsimpo and Wodon, 2012a, 2012b; for an overview of the policies implemented in Africa to deal with the crisis, see Wodon and Zaman, 2010). With 75 percent of the population under 35 years, a large segment of the society came of age in the midst of a disrupted, if not destroyed, education system; lack of education and skills likely will relegate them to low-productivity, low-wage jobs. The governance gains, economic growth, and overall peace dividends achieved by the people of Liberia and its government are too great to risk dislocating the vitality of a productive citizenry. Cohesion, peace, and economic prosperity need not succumb to social strife for lack of meaningful income opportunities, crime and violence due to prolonged unemployment, and cronyism and corruption resulting from opaque practices. Poverty reduction and peace stabilization require new employment opportunities, and necessitate accelerated structural changes in the economy. Estimates on the rate of unemployment, underemployment, and unpaid work vary between 20 and 30 percent of the working-age population (World Bank, 2009). The vast majority of the employed are engaged in very low paying jobs thereby underpinning the cycles of poverty or extreme poverty. The current economic structure limits the creation of new, more productive jobs, given the prevalence of low-yield agriculture and the limited size of the formal sector. In addition, this sector has been hit by the global financial crisis, in particular by the fall in international demand for commodities and the slowdown in foreign direct investment. The creation of temporary employment opportunities, for youths in particular, is required to support the emergency response to the repercussions of the financial crisis on the economic and social texture of the country. In this context, communities have an important role to play, given that they have been integral to the peace process and postwar rehabilitation, and key to solving the youth crisis. Despite the nascent (and limited) state capacity, a fragile business environment, and widespread poverty, Liberian communities have provided leadership to youths, identified local development priorities, ensured transparency, mobilized resources, and supported the vulnerable, the poor, and the young. Liberian communities are fundamental to economic and social development, and to new and better jobs targeted to its young citizens. The President and the Government of Liberia have placed strategic priority on youth employment and have asked development partners for support in these areas. The Office of the President, the government, and development partners have undertaken over the last two years several steps to generate immediate employment and develop

Poverty and the Policy Response to the Economic Crisis in Liberia

157

skills for youth. These include preparation of the National Employment Policy and a number of emergency employment programs including the Cash for Work Temporary Employment Project (CfWTEP) financed by the World Bank Food Price Crisis Response initiative, as well as the recent Liberia Jobs and Opportunities Initiative. Yet although promising, these activities have remained limited in scope and in their impact on institutional capacity, as well as fragmented in targeting. The government has requested a more coordinated and cohesive approach to donor partners and Liberian agencies. This chapter provides an assessment of the experience to-date with CfWTEP, as well as a discussion of options for the expansion and continuation of the program (for details, see World Bank, 2010). Over 2009â&#x20AC;&#x201C;10, CfWTEP has provided jobs to 17,000 people in the country. In order to assess the program and suggest options for its continuation, a light evaluation survey was implemented in the country in November-December 2009 with four objectives: (i) Assessing the targeting performances of the program; (ii) Measuring the wage substitution effects among the participant; (iii) Analyzing the patterns of use of the wages received by households; and (iv) Documenting other aspects of the program. The survey implementation was managed by the Liberia Agency for Community Empowerment (LACE) and the Liberia Institute of Statistics and Geo- Information Systems. Apart from this quantitative evaluation, a separate qualitative evaluation of the program was implemented, and an overview of both the quantitative and qualitative components of the evaluation was prepared by Andrews et al. (2012); both chapters are available in this collection. The rest of the chapter is structured as follows. Section 2 presents the results of our evaluation of CfWTEP. A conclusion discusses some of the lessons learned, and the ways in which they are informing the next generation program.

2. Evaluation of the Cash for Work Program 2.1. Key Evaluation Results

The evaluation of CfWTEP presented in this section is based on a light evaluation survey questionnaire fielded in Liberia in November-December 2009. A random sample of beneficiaries from CfWTEP was drawn from most of the regions in which the program is implemented (at the time the survey was fielded, not all regions had completed their implementation of the program, which explains why the sample is not fully nationally representative). Approximately 1,000 participants were interviewed. It consists of a series of modules along the following structure: Section A: Identification of the individual; Section B: Information on data collection and processing; Section C: Characteristics of the head of the household; Section D: Characteristics of the individual; Section E: Employment history before the CfWTEP Program; Section F: Participation in other projects; Section G: Program assessment; Section H: Project impact; and finally Section I: Household assets. The basic principle for the evaluation consisted on assessing the welfare status of participants through a matching procedure whereby the survey data set was appended with the nationally representative 2007 CWIQ survey. Because the questionnaire of the CfWTEP survey was designed in such a way as to be comparable to that of the CWIQ survey, a large number of variables could be used for the matching procedure, including household characteristics, characteristics of the household head, assets and housing

158

A World Bank Study

characteristics, and the like. Since we are trying to estimate the level of consumption per capita of the household to which program participants belong, the matching is implemented at the household level (this is why the survey questionnaire administered to participants includes a module on the characteristics of the household head). Various matching techniques were used to obtain the key results, and the results suggest that program participants tend to be poor. This can be seen by looking at asset ownership in the 2007 CWIQ survey and among participants in table 9.1. In most cases, the level of ownership of participants is smaller than in the national survey. The same type of findings is obtained when looking at the characteristics of the dwelling and other wealth variables. Both traditional propensity score matching and a combination of coarse exact matching (Iacus et al., 2009; King and Stuart, 2007) and propensity score matching were used for the estimations of the level of consumption per capita of the households to which program participants belong. The key results are provided in table 9.2, which is based on implementing coarse exact matching on a few key variables as a first step in order to ensure comparability of the samples, and propensity score matching as a second step on the subsample generated by coarse exact matching. In terms of matching options, knearest neighbor, one-on-one matching, and kernel matching procedures were used. The preferred estimates are those obtained under the k-nearest neighbor procedure (with k equal to 5), as this procedure has been shown to be more robust in the literature. Results Table 9.1: Percentage of households owning various assets, CfWTEP and national survey National 2007 Q1

All

CfWTEP All

Electric iron

0.1

0.3

1.0

2.3

0.9

1.0

Charcoal iron

15.3

20.0

25.0

32.7

40.3

27.7

18.1

Refrigerator

0.2

0.5

0.9

0.6

3.3

1.2

0.1

Television

1.1

1.3

3.0

7.3

15.0

6.2

0.2

VCR

0.5

0.8

1.9

6.2

14.1

5.3

0.3

Radio

32.0

42.7

46.2

57.9

62.8

49.5

39.5

Cell phone

12.7

15.0

22.7

34.6

47.7

28.0

13.6

Computer

0.1

0.2

0.4

1.7

0.6

0.3

Generator

0.8

2.3

3.0

6.2

14.6

6.0

0.4

Fan

0.3

0.7

0.6

2.7

9.6

3.2

0.9

Bed

52.7

58.8

69.1

71.9

76.7

66.7

55.5

Watch

17.5

27.4

36.0

39.4

48.6

34.9

9.0

Sewing machine

0.4

0.2

1.0

2.6

2.5

1.4

0.6

Stove

0.1

0.0

0.3

1.3

0.4

0.0

Boat

0.6

0.7

1.3

0.2

0.6

0.7

0.1

Bicycle

0.8

1.4

2.3

3.7

4.2

2.6

0.9

Motorcycle

0.4

0.1

0.8

2.0

2.6

1.3

0.3

Car

0.1

1.1

3.9

1.2

0.1

Source: Authorsâ&#x20AC;&#x2122; estimation.

Poverty and the Policy Response to the Economic Crisis in Liberia

159

Table 9.2: Share of program participants by consumption quintile after matching procedure National welfare quintiles

Poorest quintile

Target regions welfare quintiles

k-nearest neighbor

One on one

Kernel

k-nearest neighbor

One on one

Kernel

16.5

27.8

11.4

14.5

26.3

4.2

43.3

25.4

67.4

41.5

23.4

59.8

27.8

14.5

21.2

28.5

15.5

36.0

11.4

12.1

0.0

11.4

0.0

1.0

20.2

0.0

4.1

23.5

0.0

100.0

Richest quintile All

Source: Authorsâ&#x20AC;&#x2122; estimation.

are provided with the control group consisting of the national sample or the sample of households located in the regions where the evaluation survey was actually carried out, but differences are small between both sets of results. The results suggest that the program was relatively effective in targeting the poor, given that an estimated 41.5 percent of the program beneficiaries appear to belong to the second poorest quintile of consumption, and another 28.5 percent appear to belong to the third quintile. Since 63.8 percent of the population is considered poor, participants belonging to the second and third quintile of the distribution of consumption per capita can be considered as poor. The program also reaches the poorest quintile, but to a lower extent, given that only 14.5 percent of program participants appear to belong to that quintile. Fewer participants appear to belong to the top two quintiles, and this result was found robust when using other matching techniques, as well as indicators directly based on observed asset indices (using factorial analysis). Thus, overall, probably more than 80 percent of program participants can be considered as being poor. A series of questions in the survey were used to assess the income gains of program participants thanks to the program. The key summary findings from this analysis are provided in table 9.3. The wage losses for program participants due to the fact that they may have had another paying job before the program, or that they may have needed to hire a worker to replace them in their previous occupation or to help for domestic work Table 9.3: Average wage losses and gains from CfWTEP per matched participant

Own salary loss

Salary loss to hire worker

Salary loss to hire domestic worker

Total cost

Gain from CfWTEP

Poorest quintile

558

155

713

8,638

414

510

8,638

460

528

8,640

477

565

8,637

Richest quintile

125

8,640

Total

455

551

8,639

Source: Authorsâ&#x20AC;&#x2122; estimation.

160

A World Bank Study

while they participated in the programs tend to be very small as compared to the incomes gained from program participants. Thus, wage substitution effects do not seem to play an important role in reducing the effectiveness of the program in generating higher additional income for the family, probably in part because there is a large mass of very low paid or unpaid workers in Liberia, so that those who take up a job with CfWTEP are unlikely to have had gainful employment as an alternative to program participation. The estimates of the wage gains can then be used to assess poverty levels before and after program participation among participants, simply by considering that any additional wage gain or loss translates into an equivalent change in consumption level. As shown in table 9.4, the findings of the assessment suggest a substantial reduction in poverty with a drop in the headcount index (share of the population in poverty) of 8.6 percent from 74.1 percent before the program to 67.7 percent after the program among participants. The reductions in the poverty gap and squared poverty gap are also important (in proportionate terms, the reduction from the baseline is 27 percent for the poverty gap). Given that the program provides additional resources to households for a period of only two months, these impacts are substantial. They could be overestimated if we have not been able to assess wage losses from informal unpaid activities properly, but even if we were to account for higher wage losses or substitution effects, the program would still be rather efficient in increasing net wages as a proportion of the wages paid out to participants. Table 9.4: Estimated impact on poverty among program participants Â

Headcount index

Poverty gap

Squared poverty gap

Before the program

74.1

17.2

5.9

After the program

67.7

12.6

3.8

Source: Authorsâ&#x20AC;&#x2122; estimation.

A series of questions on subjective perceptions regarding the program were also asked to program participants, the results of which are provided in table 9.5. Most participants declared that they acquired skills through the public works, and that they learned to be punctual. Participants were paid through local banks contracted to make the payments. Participants and implementers considered this a highly successful part of the program, and for most participants, this was their first contact with a bank. Participants were also issued a program identification card with which they were able to collect their payments. When possible, participants collected their pay at local bank branches; however, for more remote areas, the bank sent representatives to participating communities once a month to make payment, and the laminated ID could be used by participants for other purposes as well. The program budget also provided for the procurement of tools to be purchased through national competitive bidding. The types of tools provided depend on the type of work activity implemented in communities. If another project is implemented afterward in the same area, the tools are to be re-used for the subsequent project. But if no additional projects are planned for a given area, the tools can be handed over to local community structures to continue with similar activities or to the participants if no appropriate local community structure exists, which is an additional benefit appreciated by participants. While some participants experienced delays in being paid, this was limited to about a fifth of the participants.

Poverty and the Policy Response to the Economic Crisis in Liberia

161

Table 9.5: Subjective indicators of program quality among participants Gender

Assignment Clearing of bush

Clearing debris

Other

Total

37.9

19.2

21.6

27.9

74.5

75.6

75.4

79.8

75.8

97.2

95.4

96.7

98.5

96.4

84.1

85.9

81.7

83.7

79.3

82.7

75.1

79.8

75.7

74.5

79.3

82.7

77.2

87.1

88.3

82.9

84.4

89.9

87.6

13.8

12.2

15.9

13.9

9.9

15.1

13.1

Male

Female

Has done type of work before

34.9

19.5

Has gained technical skills

74.4

77.5

Has learned to be punctual

95.8

Has learned through the program

81.6

Has gained skills helping later Assignment was clear Has worked too hard Has paid a bribe

Filling of potholes

All

4.6

6.2

3.5

4.8

8.6

5.3

The salary was fair

86.8

89.2

84.7

85.9

93.2

85.1

87.9

Experience payment delays

13.4

24.4

11.1

35.1

19.6

Laminated ID first ID ever

66.6

67.8

66.1

72.1

63.6

61.3

67.1

Received a hand tool

11.1

6.9

5.3

14.8

5.1

6.8

9.2

9.9

5.6

4.9

12.9

5.9

7.9

93.3

94.3

95.4

95.5

95.2

81.7

93.8

Hand tool for practical use First time business with a bank Source: Authorsâ&#x20AC;&#x2122; estimation.

Finally the results of the survey are encouraging regarding both the short- and longterm impact of the program on households through their expenditure choices for the additional income gained. As shown in table 9.6, as much as 30 percent of the income of participants was spent on education (this may have been influenced by the timing of the program, where most wages were paid when school fees were dues) and 25 percent on various types of investments. This suggests that the program is not only achieving short-term impacts, but is may also have some longer-term impacts on the targeted Table 9.6: Use of program income among participating households (share of funds) Relation to head

Gender

All

Self

Other member

Male

Female

Education

31.2

30.2

32.4

29.3

Health care

8.5

8.3

8.1

8.8

28.8

25.3

27.4

28.5

Funeral

1.2

1.4

1.5

0.9

1.2

Celebration

0.7

2.2

1.2

0.9

1.1

Investment farm

8.3

8.1

7.3

9.4

8.2

Investment, non farm

6.5

4.3

7.3

4.5

Debt repayment

2.3

2.7

4.7

3.6

Transfer

1.1

2.2

1.4

1.3

Repairing the house

8.8

6.2

7.9

8.6

8.2

Acquiring household assets

2.7

3.6

2.7

3.1

2.9

Living expenses

Total Source: Authorsâ&#x20AC;&#x2122; estimation.

100

Total 31 8.4 28

100

162

A World Bank Study

households. The relatively high share of the income used for education and investments is perhaps in part possible because of the wage rate of US$3 a day, which may be considered high for an emergency program aiming to address food insecurity. Yet, given that the program only provides one single episode of employment to participants, a relatively high wage rate that allows not only for increased temporary consumptions but also some level of investment may be warranted, as discussed in more details below in section 2.3. 2.2. Cost-Effectiveness of CfWTEP

Based on the parameters estimated through the light evaluation survey, the program appears to be cost-effective, at least in terms of international comparators. This is due to its high labor intensity of activities, the relatively effective geographical and household level targeting of the poor (even if the extreme poor are less well covered), and the low foregone income from participation. To estimate more precisely the cost-effectiveness ratio or relative efficiency of converting program funding into net wage benefits for the poor, a simple methodology proposed by Ravallion (1999) can be used. Adapting slightly Ravallion’s decomposition, assume that without public works, an individual has a probability F* to find employment at market wage W*. Expected earnings are F*W*. With public works, the individual earns the public works wage W. If the individual can continue to search for private or self-employment while participating in public works, with probability F of finding such employment, the expected wage with public works is FW* + (1 – F)W. The net wage benefit from the program for the worker is NWB = (1 – F) W – (F* – F)W*. If the worker gets unemployment benefits or a subsistence allowance S, the wage benefit is reduced to NWB = (1 – F)W – (F* – F)W* – (1 – F*)S. If the program costs G to the government per worker employed, a measure of cost effectiveness is the share of public expenditures transferred to workers as wage gain NWB/G. This measure is decomposed as:

. The determinants of cost-effectiveness are (i) the leverage ratio C/G, where C is the total cost per worker including community funding; (ii) the wage share (W + L)/C, where W stands for wages paid to the poor and L stands for leakage due to wages paid for the non-poor; (iii) the targeting performance W/(W + L) which is the percentage of wages reaching the poor; and (iv) the proportionate wage gain NWB/W. This model can be extended to take into account the benefits of the infrastructure built by public works, but these benefits tend not to be as immediate. Given that there is very limited budget leverage in Liberia, the calculation provided here is based on three of these four variables: the wage share, targeting performance, and proportional wage gain. The first variable is the wage share, which is estimated using administrative data at 68 percent. This is relatively good by international standards. Other public works programs show wage shares of around 60 percent in India (National Rural Employment Guarantee Scheme), 70 percent in Korea’s public works program, 85 percent in the Pro-

Poverty and the Policy Response to the Economic Crisis in Liberia

163

ductive Safety Net Program in Ethiopia, 40-50 percent in Argentina’s Trabajar Program, and 60-70 percent in Bangladesh’s Food for Work Program. There are two caveats to this nevertheless. The first is that these programs are of a much larger scale than the program in Liberia. The second is that the government contribution in terms of project management in Liberia is very limited. This is important as in most other programs the cost of government officials managing the program is not fully factored into the project costs thus reducing the corresponding wage shares. The second variable is the targeting performance, which is the proportion of the wages paid out that goes to poor workers. By using different matching approaches for analyzing the results of the quantitative survey, it was found that between 74 and 86 percent of program participants were considered as poor. We can thus assume a likely value for the targeting performance parameter of about 80 percent. This again compares fairly well with international experience. The Ethiopian Productive Safety Net Project was found to have 87 percent of its beneficiaries were among the target group, and the corresponding value was 70-80 percent for Argentina’s Trabajar, but some other programs have been less successful in targeting the poor. The third variable is the proportionate wage gain or share of the wages received by the poor that generate additional net income after taking into account foregone income. Because of the high wage paid and lack of other income opportunities in Liberia, the net wage gain for CfWTEP is very high. Approximately 75 percent of program participants had no other income or employment before the program and so the foregone earnings were very low. The net wage gain was found to be 93 percent. In comparison, in the India program mentioned earlier, the estimated proportionate wage gain was at 75 percent, and in Argentina the proportionate wage gain was only 50 percent as there were more work alternatives available. Overall the cost-effectiveness of the program is the product of the wage share (.68) times the targeting performance (.80) times the proportionate wage gain (0.93), or 0.51. Thus the cost of transferring US$1 in net wage benefit through CfWTEP is estimated at US$1.96, which includes the US$1 in net wage. A 0.51 overall cost-effectiveness is somewhat lower than for instance Ethiopia’s PSNP where the cost-effectiveness of the wage transfer was about 0.55, but it is not common to observe cost-effectiveness of only one third, especially when targeting performance is limited. No data is available regarding the benefits of the infrastructure provided by CfWTEP, which could increase overall cost effectiveness by factoring in the benefits of such infrastructure gains. But given the relative effective targeting in the CfWTEP and the fact that projects are generally located within the communities participants are from, it can be assumed that a fair amount of the benefits from improved infrastructure accrue to the poor. 2.3. Discussion of the Wage Rate

The program provides 40 days of employment to each participant and pays a daily wage of US$3, resulting in a total transfer of US$120 to each participant. While the program is relatively well targeted, the comparison of the wages paid to participants with the wages observed in the national 2007 CWIQ survey suggest that the wages paid to participants are relatively high, since the going rate for low-skill workers is closer to US$1 a day. While it does make sense to pay program participants more than the going wage due to the fact that there are costs for workers to shift their employment towards CfWTEP, the premium paid by CfWTEP seems to be too high.

164

A World Bank Study

A number of considerations must be weighted regarding the possibility of changing the wage rate, as summarized in table 9.7. One set of factors in deciding the wage rate is of a political nature. In particular, the Liberia Jobs and Opportunities Initiative launched in December 2009 by the government of Liberia is significant as it represents a commitment to prioritize employment creation from its own resources. This program pays a wage rate of US$3 a day. The scheme plans to create 8,000 temporary jobs for youth. In considering the renewal and extension of CfWTEP for the coming few years, the appraisal team for the operation was strongly advised by the government and development partners to keep the temporary employment project aligned with the governmentâ&#x20AC;&#x2122;s own initiatives in terms of wages paid to program participants. Furthermore, similar programs are also being implemented in Liberia by the World Food Program, United Nations Development Program, and United Nations Mission in Liberia and they are currently all paying a wage rate of US$3 a day. If the extension CfWTEP were to be the only program paying less then US$3 this could create difficulties with communities that may feel short-changed and may result in accusations of LACE holding back some of the funds due to the communities, thereby undermining the relationship between communities and the implementing agency. One potentially important reason for wages of these types of public works programs to be kept low is to enable the wage rate to function as a self-targeting mechanism. However, despite the relatively high wage rate in the CfWTEP the targeting performance was still found to be relatively effective. In fact, the CfWTEP did not rely on the wage rate as the primary targeting mechanism, rather on community involvement in the selection of participants, geographical targeting, and some eligibility criteria such as the condition of Table 9.7: Arguments for and against reducing the wage rate to US$2.50 In favor of reducing wage rate to US$2.50

In favor of maintaining the wage rate at US$3.00 Important factors Could create tensions with other programs because of misaligned wage rates

Political economy Coverage

Under program expansion, 52,500 instead of 45,000 people could benefit Weak local implementation capacity is a major challenge to expand the number of participants above the 45,000 target

Capacity Work outputs

17% more work could be executed by the project Risk of slightly reduced ability of beneficiaries to invest

Impact on beneficiaries

Less important factors

Self-targeting

Likely small improvement in targeting the extreme poor

Labor market distortions

Wage substitution effect could be reduced slightly

Program costeffectiveness Source: Authors.

Risk of small reduction in overall program cost-effectiveness

Poverty and the Policy Response to the Economic Crisis in Liberia

165

being unemployed. Under the CfWETP this approach led close to 80 percent of program beneficiaries classified as poor. Lowering the wage rate to, say, US$2.50 would probably not make the wage rate a much more effective targeting mechanism for reaching the extreme poor better, since it would still be above the market wage rate. Effective targeting thus will still primarily depend on the additional targeting and recruitment activities that are part of the project. From the perspective of improving the targeting therefore, setting the wage at either US$2.50 or US$3.00 is not critical. Another factor often considered in keeping the wage rate low is the substitution effect and the possible impact on local wage rates. In some areas the high wage rate may encourage people to leave other activities to join the program as it is seen as more attractive. Furthermore, if the wage rate is much higher than local wage rate, it may also create an upward pressure on local wage rates, which may impact employers in the agricultural sector in particular. However, given the high degree of labor surplus in Liberia, and the limited size of the programs, these effects are expected to be limited. The results of the evaluation survey suggest that 76 percent of participants where either not active or unemployed prior to the program. Wage substitution effects were also found to be low. In addition, the program provides only a one-off opportunity to participants, as those who have previously participated are not be eligible to participate again if a second project is implemented in their community. This should limit the upward impact on overall wage levels because it does not increase the bargaining power of participants, as the program only provides an alternative to other employment on a once-off basis and for a limited duration. If the wage rate were reduced to US$2.50, a larger number of people would be reached by the program. Given the large number of poor people in Liberia and the relatively small percentage of the poor the program can reach, extending coverage as much as possible would be the more equitable approach to follow, so that this would be an argument in favor of reducing the wage rate to US$2.50. Another argument in favor of reducing the wage rate would be the fact that the total infrastructure outputs of the program would increase. Approximately 17 percent more tasks would be completed by reducing the wage rate to US$ 2.50, resulting in more streets being cleaned, drains cleared, potholes filled, etc. On the other hand, one argument against increasing the number of beneficiaries by reducing the wage rate is the weak in-country capacity to implement larger scale development projects. It is currently expected that there would be a scaling up the current CfWTEP in its next phase from 17,000 beneficiaries in two years to 45,000 people in three years. In addition, a non-cognitive skills training module will be added to the program, which may be quite a challenge to implement well. Increasing the number of beneficiaries even further to 52,500 under a lower wage rate may be difficult given the country context and the fact that scaling-up capacity is something difficult to achieve in Liberia. The assessments done to date on the CfWTEP do indicate that even though the transfer is of a short-term duration, many participants were also able to use their income for investments such as education, farm inputs, household assets, home improvements, or entry into informal economic activities. Their ability to do so is at least partially due to the relatively high wage rate. It could perhaps be expected that reducing the wage rate would reduce the ability of participants to make investments in two ways. Firstly, one could expect that a lower percentage of participants would be able to use the income to make investments, although this effect may be limited as it has been found that in other

166

A World Bank Study

programs where the wage rate was much lower, people were still able to make some investments. The more significant impact would probably be on the actual size of the investments and the percentage of total income actually used for investment, although the impact would be limited if the reduction in the wage rate would itself be limited. Reducing the wage rate could also have an impact on the program cost-effectiveness. As discussed earlier, the cost-effectiveness of the program is estimated to be such that it costs US$1.96, including the wage to transfer US$1 to a poor person. If the wage rate were reduced, the net wage gain of the program could potentially also be reduced which in turn would reduce the cost-effectiveness. The wage share might decrease slightly, but the targeting effectiveness could increase slightly. Overall, the cost-effectiveness of the program could be reduced slightly, perhaps to 0.50 increasing the cost of transferring US$1 to a poor person to US$2. On the other hand, given that the program is essentially a once-off transfer as participants are not allowed to participate more than once, the argument in favor of reducing the wage level is less strong.

3. Conclusion The assessment of CfWTEP is overall positive. A number of characteristics of the program, some of which were not discussed for lack of space, are summarized in table 9.8. While the program scores well, or at least fairly in most dimensions identified in table 9.8, it performs poorly only in terms of one dimension, that of the wage rate. The option of reducing the wage rate to increase the coverage of the program for the same overall budget was carefully considered, including in terms of the local political economy setting. A major issue is that other programs including those funded by the government tend to pay a rate of US$3.00 a day. Reducing the wage rate for CfWTEP would thus create a misalignment between this project and other similar projects operating in Liberia and could create significant tensions, especially at the community level. This misalignment could result in accusation by communities that part of their wages are being unfairly withheld, an accusation that could seriously undermine program implementation. Moreover, while there could be a benefit in reducing the wage rate in order to achieve higher coverage of participants, capacity could become a major implementation constraint. Providing work to 8,500 participants a year under the current program was a challenge, and under the planned expansion of the program, the number of jobs provided per year could reach 15,000 with the current wage of US$3 a day. Reducing the wage would lead, under the planned budget envelope, to an even larger number of jobs to be provided, which could be challenging for LACE and the local NGOs supporting the program. These factors led to the decision to keep the wage rate constant in the future phase of the program, despite some of the costs in doing so. The overall positive assessment of CfWTEP has contributed to the joint decision by the World Bank and the Government of Liberia to expand the program further. In this conclusion, we would like to share some of the decisions that were made for this new project. The project, which will start in 2010 for a period of two years, will be named the Liberia Youth, Employment, Skills (YES) project and it will be financed with a Sector Investment Loan for a total value of US$16.0 million. The proposed project will aim to create additional short-term jobs for the youth, but in addition it will finance demanddriven skills and development programs serving the informal and formal economy, and

Poverty and the Policy Response to the Economic Crisis in Liberia

167

Table 9.8: Overall assessment of CfWTEP performance Best-practice design feature

CfWTEP performance

Wage rate no higher than prevailing market wage for unskilled manual labor

Poor—Wage rates are higher than the prevailing market wage rate. Yet, given the size of the program and the Liberian conditions, the risks of negatively impacting the local labor market are very small.

Restrictions on eligibility should be avoided

Fair—There are few restrictions for participation but given that the wage rate cannot be used as an effective targeting mechanism, screening processes to identify the most vulnerable are applied

Program should be targeted to poor areas, as indicated by a credible “poverty map”

Good—Program is geographically targeted based on the number of extreme poor in all counties

The labor intensity (share of wage bill in total cost) should be as high as possible

Good—Program will achieve 72% labor intensity, which compares well with other programs

Assets created are of maximum value to poor people in those areas. Any assets that largely benefit the non-poor should require co-financing from the beneficiaries

Good—Program focuses on maintenance of assets but they were generally located within poor communities and work activities were generally identified by the communities themselves

Public works should be synchronized to the timing of agricultural slack seasons

Fair—There are practical difficulties with this approach as the slack season (hungry season as referred to in Liberia) coincides with the peak of the rainy season when project implementation is difficult

Encourage female participation. Women can benefit from piece rates or task-based wages; sometimes wages in the form of food have attracted more women to work sites. Provision of childcare can improve female participation.

Fair—Female participation will be around 50%, current restrictions on the participation of pregnant women will be lifted and childcare services will be considered.

Transaction costs to the poor are kept low—one important means to accomplish this is through locating project sites close to villages.

Excellent—Projects are all located within walking distance of communities.

To ensure appropriate mediation of NGOs for protecting the rights of the poor vis -à-vis program managers

Good—Program is implemented by strong local NGOs that have demonstrably paid adequate attention to the needs of the poor

The program should focus on asset maintenance

Good—Program focuses almost exclusively on the maintenance of assets

Sources: Adapted from Ravallion (1999) and Subbarao (1997).

lay the foundations of a stronger and demand-driven institutional framework for technical and vocational education and training (TVET). The new project will aim to be catalytic to the larger poverty reduction strategy of the country in various ways. First, the project will aim to create additional income opportunities, in particular for youths and through active mobilization of communities. It is envisaged that temporary jobs will help mitigate the impact of the current financial and economic global crisis and its repercussions on the poorest segments of society. It also will help mitigate the risk of social unrest resulting from frustration over prolonged unemployment. In this regard, the project will help to extend the window of stability that is required for other elements of the broader poverty reduction strategy to succeed in demonstrating expected peace dividends. Second, the project will provide poor and young Liberians with a second opportunity to get trained and acquire some basic and non-cognitive skills. In the medium term, this should help increase earnings and labor productivity, which remain low compared with other countries in the region. In line

168

A World Bank Study

with the countryâ&#x20AC;&#x2122;s National Capacity Development Strategy, the project will also support the design and launch of a new institutional framework for TVET. The public works component of the program will account for US$8.5 million of the total grant provided to the country with the goal of engaging 45,000 Liberians in temporary employment with a particular emphasis on targeting youths at risk. It will also include a training component, with a focus on a non-cognitive skills module to reinforce basic life skills and workforce readiness behaviors provided by the overall experience of participating in the public works activities. The proposed design is to add one day of training in non-cognitive skills during each week of work. This time will be paid at the same wage as a working day. The training will be linked as much as possible to the experience beneficiaries have during the project, including modules on basic work-related behaviors, attitudes in the workplace, and simple concepts of punctuality, professionalism, respect, and teamwork. The training will also include sessions on identity, highlighting self-awareness, social service, civic engagement, and budgeting and managing money. The latter session would be held in conjunction with participant payments. Public works activities carried out under CfWTEP focused on brushing feeder roads, collecting waste, rehabilitating farmland, and small repairs of roads. Under the new project, the aim will be to increase community participation in project selection to ensure that projects respond to community needs. This will be done by improving monitoring and reporting activities, including increased work supervision, by focusing on these issues in the first training module of the program and by providing additional support to local implementers. Additionally, if the government or other partners, including LACE under its other projects, can supply required material inputs, the project activities could also include other activities building useful and long-lasting social assets such as planting trees, replacing or laying culverts, painting public buildings and street walls (particularly needed after the rainy season), painting cross walks, and small rehabilitation of schools, health posts, community centers, and public markets. Efďťżforts to reach out to youths will be strengthened, given that young Liberians comprise 75 percent of the population and form the backbone of future economic and social growth. Their participation, involvement, improvement of skills, community socialization, and participation in community services and the creation of local useful social assets is a meaningful instrument of empowerment and inclusion. This can lead to increased participation of the youth in peaceful and constructive social and economic dynamics, diverting them from non-constructive and even destructive and violent options. The targeting of youths under CfWTEP was mixed and results differed among counties. On average almost 60 percent of the participants were less than 35 years old. Recognizing the predominately youth focus of the project, the new project will stipulate that at least 75 percent of the participants will be younger than 35 years old. As of February 2010, CfWTEP has exceeded its target of at least 30 percent participation by women with approximately 45 percent of participants being female. The new Liberia YES Project will set a target of 50 percent participation by women. Provisions will be made for women with small children to participate. If there are two or more women on a project with small children who have been selected, one woman will be tasked with looking after these small children, while the other(s) will then be able to work. The woman looking after the children will be paid the same rate as the other unskilled workers on the project. Liberiaâ&#x20AC;&#x2122;s new labor law (currently in draft) is expected to prohibit the

Poverty and the Policy Response to the Economic Crisis in Liberia

169

exclusion of pregnant women from the workforce with a provision that pregnant women be assigned appropriate duties. To anticipate this change, pregnant women will be eligible to participate in the new public works scheme, a change from the participant requirements under CfWTEP. It will be the responsibility of NGOs supporting project implementation to ensure that pregnant women are assigned appropriate tasks for their condition. The activities undertaken in the new project will be similar to CfWTEP, with the main requirements of activities being that they are highly labor-intensive and beneficial to the community. The activities and their locale will be determined in consultation with county officials and local communities. Possible activities include: Clearing of brush along roads or on non-private agricultural land (or for any other purpose that would serve a community or public need); Rock breaking for the roads; Cleaning of culverts; Clearing drains along roads; Filling of potholes and erosion gullies; Improving walking paths and trails, constructing steps and railing on steep sections; and Cleaning and sweeping streets and other public spaces like markets, recreational spaces, and schools yards. In addition, if required, material inputs would be provided by the government, LACE and its other programs, or other agencies. Activities may also include: Planting trees; Replacing or laying culverts; Painting public buildings and street walls (particularly needed after the rainy season); Painting cross walks; and Small rehabilitation of schools, health posts, community centers, and public markets. Many of these activities, which are normally considered part of standard rehabilitation and maintenance, have not been undertaken in Liberia because of lack of funding or institutional structures to manage such activities. It will be important in due time to evaluate the YES project as was done for the CfWTEP.

Notes 1. The authors are with the World Bank. This chapter was prepared in part as a background note for a World Bank operation on cash for work and technical and vocational education and training, with support from the Gender Action Plan. The views expressed here are those of the authors and need not reflect those of the World Bank, its Executive Directors or the countries they represent.

References Adato, M., and L. Haddad. 2002. “Targeting Poverty through Community-Based Public Works Programmes: Experience from South Africa.” Journal of Development Studies 38(3): 1–36. Andrews, C., P. Backiny-Yetna, E. Garin, E. Weedon, Q. Wodon, and G. Zampaglione. 2012. “Liberia’s Cash for Work Temporary Employment Project: Responding to Crisis in Low-Income, Fragile Countries.” In Q. Wodon, ed., 128–154. Poverty and Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Backiny-Yetna, P., and Q. Wodon. 2011. “Measuring the Impact of Labor Intensive Public Works using a Light Evaluation Survey.” Mimeo, World Bank, Washington, DC. Backiny-Yetna, P., R. Mungai, C. Tsimpo, and Q. Wodon. 2012. “Poverty in Liberia: Level, Profile, and Determinants.” In Q. Wodon, ed., 9–34. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Coady, D., M. Grosh, and J. Hoddinott. 2003. Targeting of Transfers in Developing Countries: Review of Lessons and Experience. Regional and Sectoral Studies. Washington, DC: World Bank.

170

A World Bank Study

Food and Agriculture Organization of the United Nations (FAO). 2008. The State of Food Security in the World 2008. FAO, Rome, Italy. Government of Liberia and United Nations Joint Report. 2010. The State of Food and Nutrition Security in Liberia: Comprehensive Food Security and Nutrition Survey. Monrovia. Grosh, M., C. Del Ninno, E. Tesliuc, and A. Ouerghi. 2008. For Protection and Promotion: The Design and Implementation of Effective Safety Nets. Washington DC: World Bank. Iacus, S., G. King, and G. Porro. 2009. “Causal Inference Without Balance Checking: Coarsened Exact Matching.” Mimeo, Harvard University. Ivanic, M., and W. Martin. 2008. “Implications of Higher Global Food Prices for Poverty in Low-Income Countries.” Agricultural Economics 39: 405–416. King, G., and E. Stuart. 2007. “Matching as Nonparametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference.” Political Analysis 15: 199–236. Making Enterprises Inc. 2010. “Liberia Cash for Work Temporary Employment Project: A Qualitative Assessment.” Mimeo, Monrovia. Ministry of Agriculture of the Republic of Liberia. 2007. Comprehensive Assessment of the Agriculture sector in Liberia: Volume 1: Synthesis Report. Monrovia. Ravallion, M. 1999. “Appraising Workfare.” World Bank Research Observer 14: 31–48. Republic of Liberia. 2006. Comprehensive Food Security and Nutrition Survey (CFSNS). Monrovia. ———. 2008. Poverty Reduction Strategy. Monrovia. Richards, P., S. Archibald, B. Bruce, W. Modad, E. Mulbah, T. Varpilah, and J. Vincent. 2005. “Community Cohesion in Liberia: A Post-War Rapid Social Assessment.” Social development Paper. Conflict Prevention and Reconstruction No. 21. World Bank, Washington, DC. Schubert, B. 2008. “Social Protection Issues in Liberia.” Mimeo, Inter-Agency Programming Team of the United Nations. Subbarao, K. 2003. “Systemic Shocks and Social Protection: Role and Effectiveness of Public Works Programs.” Social Protection Discussion Paper No. 302. World Bank, Washington, DC. Teklu, T., and S. Asefa. 1997. “Factors Affecting Employment Choice in a Labor-Intensive Public Works Scheme in Rural Botswana.” Economic Development and Cultural Change 46(1): 175–86. Teklu, T., and S. Asefa. 1999. “Who Participates in Labor-Intensive Public Works in SubSaharan Africa? Evidence from Rural Botswana and Kenya.” World Development 27(2): 431–38. Tsimpo, C., and Q. Wodon. 2012a. “Rice Prices and Poverty in Liberia.” In Q. Wodon, ed., 85–98. Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. Tsimpo, C., and Q. Wodon. 2012b. “Benefit Incidence of Fiscal Measures to Deal with the Impact on Households of the Economic Crisis in Liberia: Comparing Import and Income Taxes.” In Q. Wodon, ed., 99–110, Poverty and the Policy Response to the Economic Crisis in Liberia. Washington, DC: World Bank. United Nations Joint Assessment. 2008. “The Impact of High Prices on Food Security in Liberia.” Mimeo, Monrovia. Wodon, Q., C. Tsimpo, P. Backiny-Yetna, G. Joseph, F. Adoho, and H. Coulombe. 2008. “Potential Impact of Higher Food Prices on Poverty: Summary Estimates for a

Poverty and the Policy Response to the Economic Crisis in Liberia

171

Dozen West and Central African Countries.” Policy Research Working Paper 4745. World Bank, Washington, DC. Wodon, Q., and H. Zaman. 2010. “Higher Food Prices in Sub-Saharan Africa: Poverty Impact and Policy Responses.” World Bank Research Observer 25: 157–176. World Bank. 2009. “Liberia: Employment and Pro-Poor Growth.” Report No. 51924-LR. World Bank, Washington, DC. ———. 2010. “Project Appraisal Document on a Proposed Grant in the Amount of US$16.0 Million from the Africa Catalytic Growth fund (US$10.0 Million) and the Crisis Response Window (US$6.0 Million) to the Government of Liberia for the Liberia Youth, Employment, Skills Project.” Report No. 53626-LR. World Bank, Washington, DC.

ECO-AUDIT

Environmental Benefits Statement The World Bank is committed to preserving endangered forests and natural resources. The Office of the Publisher has chosen to print World Bank Studies and Working Papers on recycled paper with 30 percent postconsumer fiber in accordance with the recommended standards for paper usage set by the Green Press Initiative, a nonprofit program supporting publishers in using fiber that is not sourced from endangered forests. For more information, visit www.greenpressinitiative.org.

In 2010, the printing of this book on recycled paper saved the following: • 11 trees* • 3 million Btu of total energy • 1,045 lb. of net greenhouse gases • 5,035 gal. of waste water • 306 lb. of solid waste * 40 feet in height and 6–8 inches in diameter

overty and the Policy Response to the Economic Crisis in Liberia is part of the World Bank Studies series. These papers are published to communicate the results of the Bankâ&#x20AC;&#x2122;s ongoing research and to stimulate public discussion. Despite substantial progress since 2003 toward peace, economic growth, and better governance, Liberia remains one of the poorest countries in the world. The objective of this study is twofold. First, it provides a basic diagnostic of both consumption-based poverty and human development (especially education and health) in the country using the 2007 CWIQ (Core Welfare Indicators Questionnaire) survey. Second, it assesses the likely impact on the poor of the recent economic crisis, especially the increase in rice prices, and documents the targeting performance of various measures taken by the government in 2008â&#x20AC;&#x201C;09 to help the poor cope with the crisis. These measures included a reduction in import taxes for rice, a reform of the personal income tax, and the implementation of a cash-for-work temporary employment program. World Bank Studies are available individually or on standing order. This World Bank Studies series is also available online through the World Bank e-library (www.worldbank.org/elibrary).

ISBN 978-0-8213-8979-9

SKU 18979