Africa Povert y Report Poverty in a RISING Africa OV E RV I E W K at h l e e n B e e g l e Lu c C h r i s t i a e n s e n A n d r e w Da b a l e n Isis Gaddis This booklet contains the overview, as well as a list of contents, from Poverty in a Rising Africa, doi: 10.1596/978-1-4648-0723-7. A PDF of the full-length book is available at https://openknowledge .worldbank.org/ and print copies can be ordered at http://Amazon.com. Please use the full version of the book for citation, reproduction, and adaptation purposes. © 2016 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank with external contributions. The findings, interpre- tations, and conclusions expressed in this work 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. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution —Please cite the work as follows: Beegle, Kathleen, Luc Christiaensen, Andrew Dabalen, and Isis Gaddis. 2016. “Poverty in a Rising Africa” Overview. World Bank, Washington, DC. License: Creative Commons Attribution CC BY 3.0 IGO Translations —If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations —If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned indi- vidual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to the Publishing and Knowledge Division, The World Bank, 1818 H Street N W, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Cover design: Bill Pragluski, Critical Stages LLC. Cover image: Africa Footprints, lithograph © Richard Long/World Bank Art Program. Further permission required for reuse. Contents Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii About the Authors and Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Abbreviations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Key Messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Assessing the Data Landscape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Improving Data on Poverty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Revisiting Poverty Trends. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Profiling the Poor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Taking a Nonmonetary Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Measuring Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Contents of Poverty in a Rising Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Figures O.1 Good governance and statistical capacity go together. . . . . . . . . . . . . . . . . . . . . . . . . . . 7 O.2 Adjusting for comparability and quality changes the level of and trends in poverty . . . . 8 O.3 Other estimates also suggest that poverty in Africa declined slightly faster and is slightly lower. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 O.4 Fragility is associated with significantly slower poverty reduction. . . . . . . . . . . . . . . . . 10 O.5 Acceptance of domestic violence is twice as high in Africa as in other developing regions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 O.6 Residents in resource-rich countries suffer a penalty in their human development . . . . 14 O.7 Declining inequality is often associated with declining poverty. . . . . . . . . . . . . . . . . . . 16 iii iv   C o n t e n t s Maps O.1 Lack of comparable surveys in Africa makes it difficult to measure poverty trends. . . . . 5 O.2 The number of violent events against civilians is increasing, especially in Central Africa and the Horn. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 O.3 Inequality in Africa shows a geographical pattern. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Foreword A fter two decades of unprecedented evidence on core measures of poverty and e c o no m i c g row t h , how mu c h inequality, along both monetary and non- have the lives of African families monetary dimensions. The findings are both improved? The latest estimates from the encouraging and sobering. World Bank suggest that the share of the Considerable progress has been made in African population in extreme poverty did terms of data for measuring the well-being decline—from 57 percent in 1990 to 43 per- of the population. The availability and qual- cent in 2012. At the same time, however, ity of household survey data in Africa has Africa’s population continued to expand improved. At the same time, not all coun- rapidly. As a result, the number of people ple and comparable surveys to tries have multi­ living in extreme poverty still increased by track poverty trends. Reevaluating the trends more than 100 million. These are stagger- in poverty, taking into account these data ing numbers. Further, it is projected that the concerns, suggests that poverty in Africa may world’s extreme poor will be increasingly be lower than what current estimates suggest. concentrated in Africa. In addition, Africa’s population saw progress With the adoption of the Sustainable in nonmonetary dimensions of well-being, Development Goals, including the eradica- particularly in terms of health indicators tion of extreme poverty by 2030, successful and freedom from violence. While the avail- implementation of the post-2015 develop- able data do not suggest a systematic increase ment agenda will require a solid understand- in inequality within countries in Africa, the ing of poverty and inequality in the region, number of extremely wealthy Africans is across countries and population groups, and increasing. Overall, notwithstanding these in different dimensions. broad trends, caution remains as data chal- Poverty in a Rising Africa is the first of lenges multiply when attempting to measure two sequential reports aimed at better under- inequality. standing progress in poverty reduction in While these findings on progress are Africa and articulating a policy agenda to encouraging, major poverty challenges accelerate it. This first report has a modest, remain, especially in light of the region’s but important, objective: to document the rapid population growth. Consider this: even data challenges and systematically review the under the most optimistic scenario, there v vi   F o r e w o r d are still many more Africans living in pov- To maintain and accelerate the momen- erty (more than 330 million in 2012) than in tum of progress of the past two decades, con- 1990 (about 280 million). Despite improve- certed and collective efforts are also needed ments in primary school enrollment rates, to further improve the quality and timeliness the poor quality of learning outcomes, as of poverty statistics in the region. Domestic evidenced by the fact that two in five adults political support for statistics can be the most are illiterate, highlights the urgency of poli- important factor in the quest for better data. cies to improve educational outcomes, par- Development partners and the international ticularly for girls. Perpetuation of inequality, community also have an important role to in the absence of intergenerational mobility play in terms of promoting regional coop- in education, further highlights the long-run eration, new financing models, open access consequences of failure to do so. Not surpris- policies, and clearer international standards. ingly, poverty reduction has been slowest in This volume is intended to contribute toward fragile states. This trend is compounded by improving the scope, quality, and relevance the fact that violence against civilians is once of poverty statistics. Because, in the fight again on the rise, after a decade of relative against poverty in Africa, (good) data will peace. There is also the paradoxical fact that make a difference. Better data will make for citizens in resource-rich countries are expe- better decisions and better lives. riencing systematically lower outcomes in all human welfare indicators controlling for Makhtar Diop their income level. Clearly, policies matter Vice President, Africa Region beyond resource availability. World Bank Acknowledgments T his volume is part of the African Isabel Almeida, Prospere Backiny-Yetna, Regional Studies Program, an ini- Yele Batana, Abdoullahi Beidou, Paolo Brun- tiative of the Africa Region Vice ori, Hai-Anh Dang, Johannes Hoogeveen, Presidency at the World Bank. This series La-Bhus Jirasavetakul, Christoph Lakner, of studies aims to combine high levels of Jean-François Maystadt, Annamaria Mi­­ analytical rigor and policy relevance, and lazzo, Flaviana Palmisano, Vito Peragine, to apply them to various topics important Dominique van de Walle, Philip Verwimp, for the social and economic development and Eleni Yitbarek. of Sub-Saharan Africa. Quality control and The team benefited from the valuable oversight are provided by the Office of the advice and feedback of Carlos Batarda, Chief Economist for the Africa Region. Haroon Bhorat, Laurence Chandy, Pablo This report was prepared by a core team Fajnzylber, Jed Friedman, John Gibson, Jéré- led by Kathleen Beegle, Luc Christiaensen, mie Gignoux, Ruth Hill, José Antonio Mejía- Andrew Dabalen, and Isis Gaddis. It would Guerra, Berk Ozler, Martin Ravallion, not have been possible without the relentless Raul Santaeulalia-Llopis, and Frederick Solt. efforts and inputs of Nga Thi Viet Nguyen Valentina Stoevska and colleagues from the and Shinya Takamatsu (chapters 1 and 2), ILO provided valuable data. Umberto Cattaneo and Agnes Said (chapter Stephan Klasen, Peter Lanjouw, Jacques 3), and Camila Galindo-Pardo (chapters 3 Morisset, and one anonymous reviewer pro- and 4). Rose Mungai coordinated the vided detailed and careful peer review massive effort to harmonize data files; Wei comments. Guo, Yunsun Li, and Ayago Esmubancha The World Bank’s Publishing and Knowl- Wambile provided valuable research assis- edge team coordinated the design, type­ tance. Administrative support by Keneth setting, printing, and dissemination of the Omondi and Joyce Rompas is most grate- report. Special thanks to Janice Tuten, fully acknowledged. Stephen McGroarty, Nancy Lammers, Abdia Francisco H. G. Ferreira provided gen- Mohamed, and Deborah Appel-Barker. eral direction and guidance to the team. ­ Robert Zimmermann and Barbara Karni Additional contributions were made by edited the report. vii About the Authors and Contributors Kathleen Beegle is a lead economist in the Luc Christiaensen is a lead agriculture econ- World Bank’s Africa Region. Based in Accra, omist in the World Bank’s Jobs Group and she coordinates country programs in Ghana, an honorary research fellow at the Maas- Liberia, and Sierra Leone in the areas of edu- tricht School of Management. He has writ- cation, health, poverty, social protection, ten extensively on poverty, secondary towns, gender, and jobs. Her broader area of work and structural transformation in Africa and includes poverty, labor, economic shocks, East Asia. He is also leading the “Agriculture and methodological studies on household in Africa: Telling Facts from Myths” project. survey data collection. She was deputy direc- He was a core member of the team that pro- tor of the World Development Report 2013: duced the World Development Report 2008: Jobs. She holds a PhD in economics from Agriculture for Development. He holds a Michigan State University. PhD in agricultural economics from Cornell University. Umberto Cattaneo is a research assistant at the World Bank and a doctoral fellow at the Andrew Dabalen is a lead economist in the European Center for Advanced Research in World Bank’s Poverty and Equity Global Economics and Statistics at the Université Practice. His work focuses on policy analysis Libre de Bruxelles. His research interests and research in development issues, such as include development economics, civil war, poverty and social impact analysis, inequal- poverty analysis, applied microeconomet- ity of opportunity, program evaluation, risk rics, and agricultural and environmental and vulnerability, labor markets, conflict, economics. He recently completed a study and welfare outcomes. He has worked in the on the impact of civil war on subjective and World Bank’s Africa and Europe and Cen- objective poverty in rural Burundi. He holds tral Asia Regions on poverty analysis, social a master’s degree in development econom- safety nets, labor markets, and education ics from the School of Oriental and African reforms. He has coauthored regional reports Studies of the University of London and a on equality of opportunity for children in master’s degree in economics and finance Africa and vulnerability and resilience in the from the University of Genova. Sahel and led poverty assessments for several ix x   About the Authors and Contributors countries, including Albania, Burkina Faso, Nga Thi Viet Nguyen is an economist in the Côte d’Ivoire, Kosovo, Niger, Nigeria, and World Bank’s Poverty and Equity Global Serbia. He has published scholarly articles Practice, where her work involves poverty and working papers on poverty measure- measurement and analysis, policy evaluation, ment, conflict and welfare outcomes, and and the study of labor markets and human wage inequality. He holds a PhD in agricul- development. She was part of the team that tural and resource economics from the Uni- produced the 2013 report Opening Doors: versity of California-Berkeley. Gender Equality and Development in the Middle East and North Africa. In Africa she Isis Gaddis is an economist in the World investigated the impact of Nigeria’s import Bank’s Gender Group. She previously served bans on poverty, the role of social safety net as a poverty economist for Tanzania based programs in rural poverty in Malawi, and in Dar es Salaam. Her main research inter- the contribution of labor income to poverty est is empirical microeconomics, with a focus reduction in five African countries and con- on the measurement and analysis of poverty tributed to various poverty assessments. She and inequality, gender, labor economics, and holds a master’s degree in public policy from public service delivery. She holds a PhD in Harvard University. economics from the University of Göttingen, where she was a member of the development Agnes Said is a lawyer who has been work- economics research group from 2006 to ing with the World Bank since 2009. Her 2012. work focuses on public sector governance and social protection. She is part of the manage- Camila Galindo-Pardo worked as a research ment team of a multidonor trust fund for the analyst in the Chief Economist’s Office of the Middle East and North Africa Region that is Africa Region of the World Bank, where she striving to strengthen governance and increase studied the link between sectoral economic social and economic inclusion in the region. growth and poverty, income inequality and Her work on justice and fundamental rights extreme wealth, gender based-violence, and has been published by the European Commis- the prevalence of net buyers of staple foods sion and the European Parliament. She holds among African households. She is a PhD a master of laws degree from the University student in economics at the University of of Gothenburg and a master’s degree in inter- Maryland-College Park. national relations and international econom- ics from the School of Advanced International Rose Mu nga i is a sen ior econom ist / Studies of the Johns Hopkins University. statistician with the Africa Region of the World Bank and the region’s focal point on Shinya Takamatsu is a consultant in the poverty data. She has more than 15 years of Poverty and Equity Global Practice of the experience designing household surveys and Africa Region of the World Bank, where he measuring and analyzing poverty. For sev- is a core member of the region’s statistical eral years she led production of the Bank’s development team. He has published several annual Africa Development Indicators working papers on poverty imputations and report. Before joining the World Bank, she survey methodology and conducted research worked as a senior economist/statistician on the educational spillover effect of a condi- at the Kenya National Bureau of Statistics, tional cash transfer program and the poverty where her core role was measuring pov- impacts of food price crises. He holds a PhD erty. She holds a master’s degree in devel- in agricultural and resource economics, with opment economics from the University of a minor in statistics, from the University of Manchester. Minnesota. Abbreviations AIDS acquired immune deficiency syndrome CPI consumer price index GDP gross domestic product HIV human immunodeficiency virus SDG Sustainable Development Goal xi Key Messages Measuring poverty in Africa remains a challenge. The coverage, comparability, and quality of household surveys to monitor living standards •  have improved. Still, by 2012, only 27 of the region’s 48 countries had conducted at least two comparable surveys since 1990 to track poverty. Regular and good-quality GDP, price, and census data are also lacking. •  Technical approaches can fill in some gaps, but there is no good alternative to regular and •  good-quality data. A regionwide effort to strengthen Africa’s statistics is called for. Poverty in Africa may be lower than current estimates suggest, but more people are poor today than in 1990. The latest estimates from the World Bank show that the share of Africans who are poor fell •  from 57 percent in 1990 to 43 percent in 2012. Limiting estimates to comparable surveys, drawing on nonconsumption surveys, and applying alternative price deflators suggest that poverty may have declined by even more. Nonetheless, even given the most optimistic estimates, still many more people are poor •  because of population growth: more than 330 million in 2012, up from about 280 million in 1990. Poverty reduction has been slowest in fragile countries, and rural areas remain much poorer, •  although the urban-rural gap has narrowed. Chronic poverty is substantial. Nonmonetary dimensions of poverty have been improving. Health, nutrition, education, and empowerment have improved; and violence has diminished. •  But the challenges remain enormous: more than two in five adults are still illiterate, and the •  quality of schooling is often low; after a decade of relative peace, conflict is on the rise. Nonmonetary welfare indicators are weaker in resource-rich countries, conditional on •  income, pointing to the unmet potential of natural resource wealth. Inequality in Africa has many dimensions. The data do not reveal a systematic increase in inequality across countries in Africa. But these •  data do not capture extremely wealthy Africans, whose numbers and wealth are increasing. Spatial inequalities (differences between urban and rural areas and across regions) are large. •  I ntergenerational mobility in areas such as education and occupation has improved, but •  mobility is still low and perpetuates inequality. 1 Overview P erceptions of Africa changed dramati- and projections that the world’s poor will cally over the past 20 years. Viewed be increasingly concentrated in Africa even as a continent of wars, famines, and if the average 1995–2014 growth rates are entrenched poverty in the late 1990s, there maintained suggest the need to focus the is now a focus on “Africa rising” and an global poverty agenda on Africa. “African 21st century.”1 At 4.5 percent a This report is the first of a two-part vol- year, average economic growth was remark- ume on poverty in Africa. This study docu- ably robust, especially when contrasted with ments the data challenges and revisits the the continuous decline during the 1970s and core broad facts about poverty in Africa; the 1980s. second report will explore ways to accelerate Substantial improvements in well-being its reduction. should have accompanied this expansion. The report takes a broad, multidimen- Whether or not they did remains unclear sional view of poverty, assessing progress given the poor quality of the data (Devara- over the past two decades along both mon- jan 2013; Jerven 2013), the nature of the etary and nonmonetary dimensions. The growth process (especially the role of natural dearth of comparable, good-quality house- resources) (de la Briere and others 2015), the hold consumption surveys makes assessing emergence of extreme wealth (Oxfam 2015), monetary poverty especially challenging. the heterogeneity of the region, and persis- The report scrutinizes the data used to tent population growth of 2.7 percent a year assess monetary poverty in the region and (Canning, Raja, and Yazbeck 2015). explores how adjustments for data issues Expectations are also rising. All develop- affect poverty trends. 2 ing regions except Africa have reached the At the same time, the remarkable expan- Millennium Development Goal (MDG) of sion of standardized household surveys on halving poverty between 1990 and 2015 nonmonetary dimensions of well-being, (UN 2015). Attention will now shift to the including opinions and perceptions, opens set of new global development goals (the Sus- up new opportunities. The report examines tainable Development Goals [SDGs]), which progress in education and health, the extent include the ambitious target of eradicating to which people are free from violence and poverty worldwide by 2030. The poten- able to shape their lives, and the joint occur- tial for a slowdown in economic growth rence of various types of deprivation. It also 3 4   POVERTY IN A RISING AFRICA reviews the distributional aspects of poverty, ranks second to South Asia in terms of the by studying various dimensions of inequality. number of national household surveys per To shed light on Africa’s diversity, the country, according to the International report examines differences in performance Household Survey Network catalog. The across countries, by location, and by gen- region has an average of 24 surveys per coun- der. Countries are characterized along four try conducted between 1990 and 2012— dimensions that have been shown to affect more than the developing world average growth and poverty: resource richness, fra- of about 22. This expansion was confined gility, landlockedness (to capture geographic almost entirely to surveys that do not collect openness and potential for trade), and income consumption data, however. status (low, lower-middle, upper-middle, and The increase in household consump- high income). tion surveys, which are the building blocks for measuring poverty and inequality, was sluggish, though coverage increased. Since Assessing the Data Landscape 2009 only 2 countries did not conduct a According to World Bank estimates from single consumption survey over the past household surveys, the share of people liv- decade (down from 10 in 1990–99). The ing on less than $1.90 a day (in 2011 inter- number of countries that either did not national purchasing power parity [PPP]) conduct a consumption survey or do not fell from 57 percent in 1990 to 43 percent allow access to the microdata declined in 2012, while the number of poor still from 18 in 1990–99 to 4 in 2003–12; and increased by more than 100 million (from the number of countries with at least two 288 to 389 million). consumption surveys increased, from 13 These estimates are based on consumption in 1990–99 to 25 in 2003–12. Many frag- surveys in a subsample of countries cover- ile states—namely, Chad, the Democratic ing between one-half and two-thirds of the Republic of Congo, Sierra Leone, and region’s population. Poverty rates for the rest Togo—were part of this new wave of sur- of the countries are imputed from surveys veys. Nonetheless, fragile states still tend to that are often several years old using gross be the most data deprived. domestic product (GDP) trends, raising ques- The lack of consumption surveys and tions about the accuracy of the estimates. On accessibility to the underlying data are obvi- average only 3.8 consumption surveys per ous impediments to monitoring poverty. But country were conducted in Africa between the problems do not end there. Even when 1990 and 2012, or one every 6.1 years. In available, surveys are often not comparable the rest of the world, one consumption sur- with other surveys within the country or are vey was conducted every 2.8 years. The aver- of poor quality (including as a result of misre- age also masks quite uneven coverage across porting and deficiencies in data processing). countries. For five countries that together rep- Consequently, countries that appear to be resent 5 percent of the African population, no data rich (or have multiple surveys) can still data to measure poverty are available (either be unable to track poverty over time (exam- because no household surveys were con- ples include Guinea and Mali, with four sur- ducted or because the data that were collected veys each that are not comparable). are not accessible, or, as in the case of one At a country level, lack of comparability survey for Zimbabwe, were collected during between survey rounds and questions about a period of hyperinflation and unsuitable for quality issues often prompt intense technical poverty measurement). As of 2012, only 27 of debates about methodological choices and, 48 countries had conducted at least two com- national poverty estimates within countries parable surveys since 1990 to track poverty. (see World Bank 2012 for Niger; World Bank To be sure, the number of household sur- 2013 for Burkina Faso; World Bank 2015b veys in Africa has been rising. Africa now for Tanzania). But much regional work in O V E R V I E W   5 MAP O.1  Lack of comparable surveys in Africa makes it difficult to measure poverty trends Cabo Mauritania Verde Mali Niger Sudan Eritrea Senegal Chad The Gambia Guinea-Bissau Burkina Faso Guinea Benin Nigeria Côte Ethiopia Sierra Leone d’Ivoire Ghana Central African South Sudan Republic Liberia Cameroon Somalia Togo Equatorial Guinea Uganda São Tomé and Príncipe Rep. of Kenya Gabon Congo Rwanda Dem. Rep. of Burundi Congo Tanzania Seychelles Comoros Number of comparable surveys conducted, Angola 1990–2012 Malawi 0 or 1 survey (9 countries) Zambia No comparable surveys (12 countries) 2 comparable surveys (17 countries) Zimbabwe Mauritius Mozambique More than 2 comparable surveys (10 countries) Namibia Botswana Madagascar Swaziland South Lesotho Africa IBRD 41865 SEPTEMBER 2015 Source: World Bank data. IBRD 41865 SEPTEMBER 2015 Africa and elsewhere disregards these impor- conducted only 1.6 comparable surveys in the tant differences, relying on databases such 23 years between 1990 and 2012. as the World Bank’s PovcalNet that has not The challenge of maintaining compara- consistently vetted surveys on the basis of bility across surveys is not unique to Africa comparability or quality. or to tracking poverty (see, for example, If surveys that are not nationally repre- UNESCO 2015 for data challenges in track- sentative (covering only urban or rural areas, ing adult literacy). However, in Africa lack for example), that were not conducted at of comparability exacerbates the constraints similar times of the year (in order to control imposed by the already limited availability of for seasonality in consumption patterns), consumption surveys. It becomes especially and that collected consumption data using problematic when the challenges concern different instruments or reporting periods populous countries, such as Nigeria. Only 27 are dropped, the typical African country countries (out of 48) conducted two or more 6   POVERTY IN A RISING AFRICA comparable surveys during 1990–2012 (map is driven by capital-intensive sectors such as O.1). On the upside, they represent more than mining and oil production (Loayza and Rad- three-quarters of Africa’s population. datz 2010) and may lead to poverty reduc- The estimation of poverty also requires tion being overestimated. Caution is therefore data on price changes. For cross-country counseled, especially when extrapolating to a comparisons of poverty in a base year, distant future (or past). 2011 in this case, nominal consump- tion must be converted to 2011 price lev- els. The main method used to make this Improving Data on Poverty adjustment is the consumer price index Lack of funding and low capacity are often (CPI), which relies on both the collection of cited as main drivers for the data gaps in country-specific price data and basket weights Africa. But middle-income status is not associ- of consumer items to measure inflation. The ated with the number of consumption surveys CPI suffers from three specific problems in a country conducts, and countries receiv- Africa, in addition to the more general techni- ing more development aid do not have more cal difficulties. First, in many countries prices or higher-quality poverty data. In terms of are collected only from urban markets. Sec- capacity, the production of high-quality con- ond, the basket weights rely on dated house- sumption surveys and statistics is technically hold surveys and sometimes only on market complex, involving the mobilization of finan- purchases (excluding home-produced foods). cial and human resources on a large scale Third, computational errors sometimes bias and requiring the establishment of robust the data, as in Tanzania (World Bank 2007) quality-control mechanisms. But many coun- and Ghana (IMF 2003, 2007).3 tries that do not conduct household surveys to Across the globe, when surveys are not available in a given year, researchers use GDP measure poverty at the same time undertake to compute annual poverty estimates. Mis­ other activities that are more or equally com- sing data are interpolated (between surveys) plex (delivering antiretroviral drugs to people and extrapolated (to years before and after with AIDS and conducting national elections, the previous and latest surveys) using GDP for example) (Hoogeveen and Nguyen 2015). growth rates (see World Bank 2015a). Not Good governance is strongly correlated with all of these GDP data are reliable, however. higher-quality data (figure O.1). Countries Ghana, for example, leapt from low-income that have better scores on safety and rule of to low-middle-income country classifica- law also have superior statistical capacity. tion after rebasing its GDP in 2010; follow- Many researchers have recently suggested ing rebasing, Nigeria surpassed South Africa that problems with the availability, compara- overnight as the biggest economy in Africa. bility, and quality of data reflect the political These examples suggest that GDP growth preferences of elites (Carletto, Jolliffe, and rates—and by extension the extrapolated Banerjee 2015; CGD 2014; Devarajan 2013; poverty reductions—may be underestimated. Florian and Byiers 2014; Hoogeveen and Another issue is that imputation based on Nguyen 2015). Political elites may not favor GDP growth rates assumes that GDP growth good-quality statistics for several reasons. translates one-to-one into household con- First, where clientelism and access to poli- sumption and that all people see their con- tics are limited, a record of achievement that sumption expand at the same pace. But GDP can be supported by good-quality statistics includes much more than household con- is unnecessary because support from a small sumption: on average across a large sample group of power brokers suffices. Second, of African countries, household consumption maintaining a patronage network is costly, surveys captured just 61 percent of GDP per and high-quality statistics come at a high capita. The assumption that growth is evenly opportunity cost. Third, poor-quality sta- distributed can also be tenuous when growth tistics reduce accountability. The prevailing O V E R V I E W   7 FIGURE O.1  Good governance and statistical capacity go together 90 Mauritius 80 Rwanda Malawi Mozambique Nigeria Tanzania Senegal South Africa São Tomé and Príncipe Burkina Faso Lesotho 70 The Gambia Niger Cabo Verde Statistical capacity indicator Uganda Benin Mali Chad Madagascar Ghana Togo 60 Central African Zimbabwe Ethiopia Swaziland Seychelles Mauritania Zambia Republic Congo, Sierra Leone Dem. Rep. Cameroon Burundi Guinea Kenya 50 Namibia Botswana Angola Côte d’Ivoire Congo, Rep. Liberia Guinea-Bissau Gabon 40 Comoros Equatorial Guinea 30 Eritrea 20 Somalia 0 10 20 30 40 50 60 70 80 Safety and rule of law score Source: Hoogeveen and Nguyen 2015. political arrangements thus favor less (or less by failure to adhere to methodological and autonomous) funding for statistics because it operational standards. While this problem represents one way to exercise influence over partly reflects the lack of broader political statistical agencies. In some countries donor support domestically, regional cooperation financing has replaced domestic financing, and peer learning, as well as clear interna- but the interests of donors are not always tional standards, could help improve techni- aligned with the interests of governments. cal quality and consistency. The Program for This problem highlights the need for alterna- the Improvement of Surveys and the Measure- tive financing models, including cofinancing ment of Living Conditions in Latin America arrangements, preferably under a coordi- and the Caribbean (known by its acronym in nated regional umbrella and with adequate Spanish, MECOVI) provides a compelling incentives for quality improvements. model for achieving better poverty data. Politics and funding are not the only rea- sons statistics are inadequate. The evidence presented here suggests that better outcomes Revisiting Poverty Trends were possible even with the set of surveys that were conducted. African countries collected Various technical approaches can be applied on average 3.8 consumption surveys in the to address some of the data shortcomings past two decades, but many of them could in tracking regional poverty trends. They not be used to track poverty reliably because include limiting the sample to comparable of comparability and quality concerns caused surveys of good quality, using trends in other 8   POVERTY IN A RISING AFRICA nonconsumption data rather than GDP to (37 percent instead of 43 percent). The series impute missing poverty estimates, and gaug- of comparable and good-quality surveys only ing inflation using alternative econometric excludes some of the surveys from Burkina techniques. Faso, Mozambique, Tanzania, and Zambia Taking these steps affects the view of how and replaces the poverty estimates of the poverty has evolved in Africa. The estimate two comparable but poorer-quality surveys from PovcalNet in figure O.2 shows the of Nigeria (Nigeria Living Standards Sur- now-familiar trend in poverty from surveys veys 2003/04 and 2009/10) with the estimate in the World Bank PovcalNet database. It from the General Household Survey Panel provides the benchmark. These estimates are 2010/11, which has been deemed of good ­ population-weighted poverty rates for the 48 quality. Poverty gap and severity measures countries, of which 43 countries have one or follow similar trajectories, after correction more surveys.4 For years for which there were for comparability and quality. no surveys, poverty was estimated by impu- In the series depicted based on the subset tation using GDP growth rates. of comparable and good-quality surveys, the The estimate based on only comparable information base for Nigeria, which encom- surveys shows the trends when only com- passes almost 20 percent of the population parable surveys are used and the same GDP of Africa, shifts. The 2003/04 and 2009/10 imputation method is applied. It largely mir- surveys showed no change in poverty in Nige- rors the PovcalNet estimate. In contrast, ria. The poverty rate indicated by the alter- when in addition to controlling for compara- native survey for 2010/11 (26 percent) is half bility, quality is taken into account, the 2012 the estimate obtained from the lower-quality estimate of poverty in Africa is 6 percentage survey (53 percent) in 2009/10. Given that points lower than the PovcalNet estimate only one survey is retained, the estimated poverty trend for Nigeria also relies more FIGURE O.2  Adjusting for comparability and quality changes the on the GDP growth pattern (which was high level of and trends in poverty during the 2000s) as well as a lower poverty rate for 2010/11. Reesti­ mating the poverty 65 rate with only comparable surveys of good quality but without Nigeria indicates that Nigeria accounted for a large fraction of the 60 additional decline observed using the cor- rected series (the red line). Without Nigeria, Poverty rate (percent) 55 the corrected series declines from 55 percent to 40 percent (a 15 percentage point drop), 50 compared with 57 percent to 43 percent (a 14 percentage point drop) in PovcalNet. Confi- 45 dence in the revised regional series depends significantly on how reliable the trends in 40 Nigeria’s poverty obtained using the good- quality survey and greater dependence on GDP imputation are considered. 35 Consumption data gaps can also be filled 1990 1993 1996 1999 2002 2005 2008 2010 2012 by applying survey-to-survey (S2S) imputa- PovcalNet tion techniques to nonconsumption survey Comparable surveys only data. In this method, at least one survey with Comparable and good-quality surveys only Comparable and good-quality surveys only without Nigeria consumption and basic household character- istics is combined with nonconsumption sur- Sources: World Bank Africa Poverty database and PovcalNet. veys with the same basic characteristics for Note: Poverty is defined as living on less than $1.90 a day (2011 international purchasing power parity). different years. Consumption for the years O V E R V I E W   9 with no survey is then estimated based on FIGURE O.3  Other estimates also suggest that poverty in Africa the evolution of the nonconsumption house- declined slightly faster and is slightly lower hold characteristics as well as the relation between those characteristics and consump- 65 tion, as estimated from the consumption survey. Where they have been tested, these prediction techniques perform mostly well Poverty rate (percent) 55 in tracking poverty, although, as with GDP extrapolation, caution is counseled when pre- dicting farther out in the past or the future (Christiaensen and others 2012; Newhouse 45 and others 2014; World Bank 2015a). Apply- ing this method to the 23 largest countries in Africa (which account for 88 percent of both the population and the poor) and keeping 35 only good-quality and comparable consump- 1990–94 1995–99 2000–04 2005–09 2010–12 tion surveys suggests that poverty declined Survey to survey PovcalNet from 55 percent in 1990–94 to 40 percent in Comparable and good-quality surveys 2010–12 (figure O.3, blue line). This decline is slightly larger than the one obtained from Source: World Bank Africa Poverty database; calculations using additional household surveys for the 23 largest countries in Africa. the World Bank’s PovcalNet for the same 23 countries (which showed the poverty rate falling from 57 percent to 43 percent) (green Engel curve (which shows households’ food line) but smaller than the 19 percentage point budget share declining as real consumption reduction obtained using the comparable and rises) remains constant over time, so that good-quality surveys and GDP imputation deviations indicate over- or underestima- for these countries (red line). tion of the price deflator used. Application Another approach to addressing consump- to urban households in 16 African countries tion data gaps is to forgo using consumption with comparable surveys during the 2000s data entirely and examine changes in house- suggests that CPIs in Africa tend to overstate hold assets. However, although changes in increases in the (urban) cost of living. Poverty asset holdings may be indicative of some in many African countries may have declined aspects of household material well-being, this faster than the data indicate if the CPI is approach does not yet serve well as a proxy or overestimated. Research on many more coun- replacement for what consumption measures. tries as well as rural areas and time periods is A final issue concerns how consumption needed to confirm these results. data from a given survey year are adjusted Taken together, this set of results sug- to the year of the international poverty line, gests that poverty declined at least as much which is 2011. National CPIs are typically as reported using the World Bank database used to inflate/deflate nominal consumption PovcalNet and that the poverty rate in Africa to this benchmark year. To address concerns may be less than 43 percent. This news is about applying CPI to adjust consumption of encouraging. Nonetheless, the challenges households, researchers can look for evidence posed by poverty remain enormous. As a of the potential level of CPI bias and the result of rapid population growth, there are implications of any bias for poverty trends. still substantially more poor people today An overestimated (underestimated) CPI will (more than 330 million in 2012) than there result in flatter (steeper) poverty trends. were in 1990 (about 280 million), even under One way to assess CPI bias is by using the most optimistic poverty reduction sce- the Engel approach (Costa 2001; Hamilton nario (that is, using comparable and good- 2001). It is based on the assumption that the quality surveys only). 10   POVERTY IN A RISING AFRICA This exercise also underscores the need gap in performance is 12 percentage points for more reliable and comparable consump- in favor of nonfragile countries. Conditional tion data to help benchmark and track prog- on the three other country traits, the differ- ress toward eradicating poverty by 2030, ence in poverty reduction between fragile as envisioned under the SDGs. More gener- and nonfragile countries rises to 15 percent- ally, it counsels against overinterpreting the age points (figure O.4). Middle-income coun- accuracy conveyed by point estimates of tries as a group did not achieve faster poverty poverty—or other region- or countrywide reduction than low-income countries, and statistics of well-being. These estimates pro- being resource rich was associated with pov- vide only an order of magnitude of levels and erty reduction that was 13 percentage points changes, albeit one that becomes more pre- greater than in non-resource-rich countries cise the more comparable and reliable is the after controlling for other traits. The main underlying database. driver for the difference in poverty reduction in resource-rich and resource-poor countries, however, is corrections to the Nigeria data. Profiling the Poor More surprisingly, once resource richness, What distinguishes countries that have suc- fragility, and income status are controlled for, ceeded in reducing poverty from those that landlocked countries did not reduce poverty have failed? What are the effects of income less than coastal economies (the effect is not status, resource richness, landlockedness, statistically significant and the point estimate and fragility? is even negative). This finding contradicts the Not surprisingly, fragility is most detri- common notion that landlocked countries mental to poverty reduction. Between 1996 perform worse than coastal countries because and 2012, poverty decreased in fragile states transport costs impede trade and lower com- (from 65 percent to 53 percent), but the decline petitiveness (Bloom and Sachs 1998). was much smaller than in nonfragile econo- Although Africa is urbanizing rapidly, in mies (from 56 percent to 32 percent). The the majority of countries, 65–70 percent of the population resides in rural areas (Can- ning, Raja, and Yazbeck 2015). Across coun- FIGURE O.4  Fragility is associated with significantly slower tries rural residents have higher poverty rates poverty reduction (46 percent in rural areas in 2012 versus 18 percent in urban areas, using corrected data for all countries). But the gap between –1.1 Middle income the poverty rate in rural and urban areas declined (from 35 percentage points in 1996 to 28 percentage points in 2012). Among the –7.1 Landlocked four geographic regions, only urban areas in West Africa halved poverty. Poverty among rural populations in West and Southern –12.6*** Resource rich Africa declined about 40 percent. Africa is distinguished by a large and rising share of female-headed households. Fragile 15.1*** Such households represent 26 percent of all households and 20 percent of all people in –15 –10 –5 0 5 10 15 20 Africa. Southern Africa has the highest rate Change in poverty rate (percentage points) of female-headed households (43 percent). compared to alternative category West Africa exhibits the lowest incidence (20 percent), partly reflecting the continu- Source: World Bank Africa Poverty database. ing practice of polygamy, together with high Note: Figure shows results of a regression on the change in the poverty rate for 43 countries from 1996 to 2012 based on estimated poverty rates using comparable and good-quality surveys. remarriage rates among widows. The poverty *** Statistically significant at the 1% level. rates among people living in male-headed O V E R V I E W   11 households (48 percent) are higher than Taking a Nonmonetary in female-headed households (40 percent), Perspective except in Southern Africa, where poverty among female-headed households is higher Many aspects of well-being cannot be prop- (Milazzo and van de Walle 2015). erly priced or monetarily valued (Sandel Two caveats are warranted. First, the 2012; Sen 1985), such as the ability to read smaller household size of female-headed and write, longevity and good health, secu- households (3.9 people versus 5.1) means that rity, political freedoms, social acceptance and using per capita household consumption as status, and the ability to move about and con- the welfare indicator tends to overestimate nect. Recognizing the irreducibility of these the poverty of male-headed households rela- aspects of well-being, the Human Develop- tive to female-headed households if there are ment Index (HDI) and the Multidimensional economies of scale among larger households Poverty Index (MPI) (Alkire and Santos (Lanjouw and Ravallion 1995; van de Walle 2014) focus on achievements in education, and Milazzo 2015). But household composi- longevity and health, and living standards tion also differs: the dependency ratio is 1.2 (through income, assets, or both), which they among households headed by women and 1.0 subsequently combine into a single index. among households headed by men. Counting This study expands the scope to include children as equivalent to adults can lead to freedom from violence and freedom to decide an underestimation of poverty in male versus (a proxy for the notion of self-determination female-headed households. Understanding that is critical to Sen’s capability approach).5 the differences in poverty associated with the It also examines jointness in deprivation, gender of the household head is intertwined by counting the share of people deprived in with how one defines the consumption indi- one, two, or more dimensions of poverty. cator used in measuring poverty. Second, This approach achieves a middle ground woman household heads are a diverse group. between a single index of nonmonetary pov- Widows, divorced or separated women, and erty (which requires weighting achievements single women frequently head households in the various dimensions) and a dashboard that are relatively disadvantaged , as opposed approach (which simply lists achievements to households with a temporarily absent male dimension by dimension, ignoring jointness head (van de Walle and Milazzo 2015). in deprivation) (Ferreira and Lugo 2013). The evidence examined above captures The focus in selecting indicators was on snapshots of poverty. Looking at the body outcomes (not inputs) that are measured at of evidence on the evolution of households’ the individual (not the household) level. Infor- poverty over time (that is, taking movies of mation on these indicators is now much more people’s poverty status) reveals large varia- widely available than it once was, although tion across countries. Panel data estimates of some of the comparability and quality issues chronic poverty (the share of households stay- highlighted above also apply (see, for exam- ing poor throughout) range from 6 percent ple, UNESCO 2015 for a review of data chal- to almost 70 percent. Countries with similar lenges in tracking adult literacy). poverty rates can also be quite dissimilar in Overall, Africa’s population saw substan- terms of their poverty dynamics. A system- tial progress in most nonmonetary dimen- atic assessment using synthetic two-period sions of well-being, particularly health and panels (which are less prone to measurement freedom from violence. Between 1995 and errors) constructed for 21 countries reveals 2012, adult literacy rates rose by 4 percent- that about 58 percent of the poor population age points. Gross primary enrollment rates was chronically poor (poor in every period), increased dramatically, and the gender gap with the remaining poor being poor only in education shrank. Life expectancy at transiently (in only one period) (Dabalen and birth rose 6.2 years, and the prevalence of Dang 2015). Chronic poverty remains perva- chronic malnutrition among children under sive in the region. 5 fell by 6 percentage points. The number 12   POVERTY IN A RISING AFRICA of deaths from politically motivated vio- At the other end of the spectrum, obesity is lence declined by 75 percent, and both the emerging as a new health concern. incidence and tolerance of gender-based Africans enjoyed considerably more peace domestic violence dropped. Scores on voice in the 2000s than they did in earlier decades, and accountability indicators rose slightly, but the number of violent events has been on and there was a trend toward greater par- the rise since 2010, reaching four times the ticipation of women in household decision- level of the mid-1990s (map O.2). Violence is making processes. increasingly experienced in terms of political These improvements notwithstanding, unrest and terrorism rather than large-scale the levels of achievement remain low in all civil conflicts. domains, and the rate of progress is leveling Africa also remains among the bottom off.6 Despite the increase in school enroll- performers in terms of voice and account- ment, today still more than two out of five ability, albeit with slightly higher scores than adults are unable to read or write. About the Middle East and North Africa and East three-quarters of sixth graders in Malawi Asia and the Pacific. Tolerance of domestic and Zambia cannot read for meaning—just violence (at 30 percent of the population) is one example of the challenge of providing still twice as high as in the rest of the devel- good-quality schooling. The need to rein- oping world (figure O.5), and the incidence vigorate efforts to tackle Africa’s basic educa- of domestic violence is more than 50 percent tional challenge is urgent. higher. Higher tolerance of domestic violence Health outcomes mirror the results for lit- and less empowered decision making among eracy: progress is happening, but outcomes younger (compared with older) women sug- remain the worst in the world. Increases in gest that a generational shift in mindset is immunization and bednet coverage are slow- still to come. ing. Nearly two in five children are malnour- Around these region-wide trends there is ished, and one in eight women is underweight. also remarkable variation across countries MAP O.2  The number of violent events against civilians is increasing, especially in Central Africa and the Horn a. 1997–99 b. 2009–11 c. 2014 50–300 (6) 50–400 (6) 50– 650 (9) 10–50 (12) 10–50 (9) 10–50 (14) 0–10 (25) 0–10 (28) 0–10 (20) IBRD 41867 SEPTEMBER 2015 Sources: Armed Conflict Location and Events Dataset (ACLED); Raleigh and others 2010. Note: Maps indicate annual number of violent events against civilians; number in parentheses indicates the number of countries. For the following countries there are no data: Cabo Verde, Comoros, Mauritius, São Tomé and Príncipe, and the Seychelles. O V E R V I E W   13 FIGURE O.5  Acceptance of domestic violence displaced persons—have traits that may is twice as high in Africa as in other developing make them particularly vulnerable. In 2012, regions 3.5 million children in Africa were two- parent orphans (had lost both parents), and 50 another 28.6 million children were single- parent orphans, bringing the total number 41 of orphans to 32.1 million. The prevalence Tolerance of domestic violence 40 of orphanhood is particularly high in coun- (percent of population) 30 30 tries in or emerging from major conflict and in countries severely affected by HIV/AIDS. 22 Because it can be correlated with wealth and 20 14 urban status, orphanhood does not always confer a disadvantage on children in terms of 10 schooling. Data on school enrollment among 10- to 14-year-olds in the most recent Demo- 0 graphic and Health Surveys show that in half 2000–06 2007–13 of the countries surveyed, orphans were less Developing countries in other regions likely to be enrolled than nonorphans. Sub-Saharan Africa In a sample of seven African countries for which comparable data are available, almost Source: Data from Demographic and Health Surveys 2000–13. Note: Figures are population-weighted averages of 32 African and 28 1 working-age adult in 10 faces severe dif- non-African developing countries. ficulties in moving about, concentrating, remembering, seeing or recognizing people across the road (while wearing glasses), or and population groups. Literacy is especially taking care of him- or herself. People with low in West Africa, where gender dispari- disabilities are more likely to be in the poor- ties are large. High HIV prevalence rates are est 40 percent of the population, largely holding life expectancy back in Southern because of their lower educational attainment Africa. Conflict events are more concentrated (Filmer 2008). They score 7.2 percent higher in the Greater Horn of Africa and the Demo- on the multidimensional poverty index than cratic Republic of Congo. people without disabilities (Mitra, Posärac, Rural populations and the income poor are and Vick 2013). Not unexpectedly, disability worse off in all domains, although other fac- rates show a statistically significant correla- tors, such as gender as well as the education tion with HIV/AIDS and conflict. of women and girls, often matter as much or Africa had an estimated 3.7 million refu- more (at times in unexpected ways). Women, gees in 2013, down from 6.7 million in 1994 for example, can expect to live in good health but up from 2.8 million in 2008. In addition, 1.6 years longer than men; and, among chil- there were 12.5 million internally displaced dren under 5, boys, not girls, are more likely people, bringing the number of people dis- to be malnourished (by 5 percentage points).7 placed by conflict to 16.2 million in 2013, or At the same time, illiteracy remains substan- about 2 percent of Africa’s population (May- tially higher among women, women suffer stadt and Verwimp 2015). The main source more from violence (especially domestic vio- of refugees is the Greater Horn of Africa, lence), and they are more curtailed in their although the number of refugees from Cen- access to information and decision making. tral Africa is still about 1 million, about half Multiple deprivation characterizes life for a of them from the Democratic Republic of sizable share of African women (data on men Congo. are not available). Although the suffering associated with dis- Several groups—including orphans, placement is tremendous, the displaced are the disabled, and refugees and internally not necessarily the poorest; and fleeing often 14   POVERTY IN A RISING AFRICA helps them mitigate the detrimental effects points), suffer more from domestic violence of conflict (Etang-Ndip, Hoogeveen, and (by 9 percentage points), and live in countries Lendorfer 2015). Refugee status is also not that rank low in voice and accountability always associated with weaker socioeconomic measures (figure O.6). outcomes. Finally, local economies often also Third, better-educated women (secondary benefit from the influx of refugees (Maystadt schooling and above) and children in house- and Verwimp 2015) through increased holds with better-educated women score demand for local goods (including food) and decisively better across dimensions (health, services, improved connectivity (as new roads violence, and freedom in decision). More are built and other transport services pro- rapid improvement in female education and vided to refugee camps), and entrepreneur- women’s socioeconomic opportunities will be ship by refugees themselves. game changing in increasing Africa’s capabil- Three overarching aspects stand out from ity achievement. a review of the nonmonetary dimensions of poverty in Africa. First, fragile countries tend to perform worse and middle-income coun- Measuring Inequality tries better. This unsurprising finding con- Although not all aspects of inequality are firms the pernicious effects of conflict and is necessarily bad (rewarding effort and risk consistent with the widely observed associa- taking can promote growth), high levels of tions with overall economic development. inequality can impose heavy socioeconomic Second, controlling for these factors, costs on society. Mechanically, higher initial there is a worrisome penalty to residing in inequality results in less poverty reduction a resource-rich country: people in resource- for a given level of growth. Tentative evidence rich countries tend to be less literate (by 3.1 also suggests that inequality leads to lower percentage points), have shorter life expec- and less sustainable growth and thus less tancy (by 4.5 years) and higher rates of poverty reduction (Berg, Ostry, and Zettel- malnutrition among women (by 3.7 percent- meyer 2012) (if, for example, wealth is used age points) and children (by 2.1 percentage to engage in rent-seeking or other distortion- ary economic behaviors [Stiglitz 2012]). The pathway by which inequality evolves thus FIGURE O.6  Residents in resource-rich countries suffer a matters for poverty reduction and growth. penalty in their human development The report measures inequality using the Gini index, which ranges from 0 (perfect equality) to 1 (perfect inequality). It shows Incidence of domestic violence (% points) 9 that inequality is especially high in Southern Africa (Botswana, Lesotho, Namibia, South Children’s malnutrition (% points) 2.1 Africa, Swaziland, and Zambia), where Gini indices are well above 0.5 (map O.3). Women’s malnutrition (% points) 3.7 Of the 10 most unequal countries in the world today, 7 are in Africa. Excluding these Life expectancy (years) –4.5 countries (five of which have populations of less than 5 million and most of which are in Southern Africa) and controlling for Literacy (% points) –3.1 country-level income, Africa has inequality levels comparable to developing countries –6 –4 –2 0 2 4 6 8 10 in other parts of the world. Inequality levels do not differ significantly between coastal Source: Staff calculations based on World Health Organization and multiple Demographic and Health Surveys. and landlocked, fragile and nonfragile, or Note: Figure shows the gap between resource-rich and other countries in Africa. Results control resource-rich and resource-poor countries, for demographic factors, education, poverty, and other country characteristics (income, fragility, landlockedness). controlling for subregion. O V E R V I E W   15 MAP O.3  Inequality in Africa shows a geographical pattern Cabo Mauritania Verde Mali Niger Senegal Sudan Eritrea The Gambia Chad Guinea-Bissau Burkina Faso Guinea Benin Nigeria Côte Ethiopia Sierra Leone d’Ivoire Ghana Central African South Sudan Republic Liberia Cameroon Somalia Togo Equatorial Guinea Uganda São Tomé and Princípe Rep. of Kenya Gabon Congo Rwanda Dem. Rep. of Burundi Congo Tanzania Seychelles Comoros Gini index Angola Malawi 0.60–0.63 Zambia 0.50–0.59 0.46–0.49 Mozambique Zimbabwe Madagascar Mauritius 0.41–0.45 Namibia 0.36–0.40 Botswana 0.31–0.35 No data Swaziland South Lesotho Africa IBRD 41869 SEPTEMBER 2015 Source: World Bank Africa Poverty database. For the subset of 23 countries for which Although declines in inequality are associ- comparable surveys are available with which ated with declines in poverty, poverty fell, to assess trends in inequality, half the coun- despite increasing inequality, in many coun- tries experienced a decline in inequality and tries (figure O.7, quadrant 1). the other half saw an increase. No clear pat- For Africa as a whole, ignoring national terns are observed by countries’ resource boundaries, inequality has widened. The status, income status, or initial level of Africa-wide Gini index increased from inequality. While one might have expected a 0.52 in 1993 to 0.56 in 2008. A greater more systematic increase in inequality given share of African inequality is explained by Africa’s double decade of growth and the role gaps across countries, even though within- the exploitation of natural resources played country inequality continues to dominate. in that growth, the results presented here do These results stand in contrast to changes not provide strong evidence for such a trend. in global inequality (Lakner and Milanovic 16   POVERTY IN A RISING AFRICA FIGURE O.7  Declining inequality is often associated with declining between households. In Rwanda, South poverty Africa, and Zambia, educational attain- ment of the household head explains about Annualized percentage change in Gini index Quadrant 1 40 percent of overall inequality. Countries 2 Malawi Ethiopia 04-10 Rwanda 00-05 Togo Nigeria Zambia 98-04 with higher inequality tend to have a high Chad Ghana 98-05 Mozambique 96-02 Madagascar 05-10 Côte d’Ivoire share of their inequality driven by unequal Uganda 05-09 0 Ghana 91-98 Zambia 04-06 Cameroon education, which is an association that is not South Africa Rwanda 05-10 Senegal Ethiopia 99-04 Mozambique 02-09 Swaziland Mauritania Namibia Dem. Rep. Congo Mauritius observed for most of the other socioeconomic Uganda 09-12 Botswana Tanzania groupings. –2 Uganda 02-05 The demographic composition of the Sierra Leone Burkina Faso household also explains a large share of –4 inequality (30 percent in Senegal and 32 per- Quadrant 4 Madagascar 01-05 cent in Botswana). In countries for which data –10 0 are available to study trends in horizontal –5 5 Annualized percentage change in poverty rate inequality from the mid-1990s to the present, the main drivers—geography, education, and Survey mean increased Survey mean decreased demographics—have not changed, though Source: Countries in World Bank Africa Poverty database with comparable surveys. some variations exist at the country level. Note: Ethiopia 1995–99, an outlier, is excluded. Survey years are indicated for countries with more Inequality in Africa is the product of many than one pair of comparable surveys. forces. The circumstances in which one is born (for example, in a rural area, to unedu- cated parents) can be critical. Inequality of 2015). Not surprisingly, the wealthiest Afri- opportunity (what sociologists call ascrip- can households are much more likely to live tive inequality)—the extent to which such in countries with higher per capita GDP. circumstances dictate a large part of the out- Inequality can be decomposed into two comes among individuals in adulthood—vio- parts: inequality between groups (horizon- lates principles of fairness. tal inequality) and inequality within groups The evidence on inequality of economic (vertical inequality). Among the range of opportunity in Africa has been limited. groups one can examine, geography, educa- This report draws on surveys of 10 African tion, and demography stand out as groups for countries to explore the level of inequality which a large share of overall inequality is of economic opportunity by looking at such explained by the group to which one belongs. circumstances as ethnicity, parental educa- From the decomposition method, spatial tion and occupation, and region of birth. inequalities (by region, urban or rural, and The share of consumption inequality that so forth) explain as much as 30 percent of is attributed to inequality of opportunity is total inequality in some countries. Perhaps a as high as 20 percent (in Malawi) (because more straightforward approach to assessing of data limitations, this estimate is a lower spatial inequality is simply to look at mean bound). But inequality of opportunity is not consumption per capita across geographic necessarily associated with higher overall domains. The ratio of mean consumption inequality. between the richest and the poorest regions Another approach to measuring inequality is 2.1 in Ethiopia (regions), 3.4 in the Demo- of opportunity is to examine persistence in cratic Republic of Congo (provinces), and intergenerational education and occupation. more than 4.0 in Nigeria (states). Price differ- Does the educational attainment of a child’s ences across geographic areas drive some of parents affect a child’s schooling less than it this gap; adjusted for price differences, spa- did 50 years ago? Is a farmer’s son less likely tial inequalities are lower but are still large. to be a farmer than he was a generation ago? Education of the household head is asso- Among recent cohorts, an additional year ciated with even larger consumption gaps of schooling of one’s parents has a lower O V E R V I E W   17 association with one’s own schooling than Notes it did for older generations, suggesting more T hroughout this report, Africa refers to Sub- 1.  equal educational opportunities for younger Saharan Africa. cohorts. Intergenerational mobility trends 2. The focus is on a range of measurement issues, are comparable to trends estimated for other including the limited availability, comparabil- developing countries. For occupation the ity, and quality of consumption data and the findings are more mixed for the five coun- remedies used to overcome these constraints. tries for which data are available. Intergener- For a range of other measurement issues— ational occupational mobility has been rising including the measurement of service flows rapidly in the Comoros and Rwanda. In con- from housing and durable goods, the conver- trast, it remains rigid in Guinea. The shift in sion of household into individual consumption the structure of occupations in the economy measures (to account for differential needs (sometimes called structural change) is not and economies of scale), and methodological the sole reason for changes in intergenera- differences in constructing poverty lines—the tional occupational mobility. Other factors, report adopts standard approaches. such as discrimination, social norms, and A n additional aspect to measuring cross- 3.  country poverty is converting local currency impediments to mobility (poor infrastruc- measures into a common currency. This report ture, conflict, and so forth), are also chang- adopts the new international poverty line of ing in ways that can affect mobility. $1.90/day in 2011 based on the latest round These results tell only part of the story of the purchasing power parity (PPP) exercise because household surveys are not suited to and discusses the complicated set of issues that measuring extreme wealth. Data on holders PPPs entail. of extreme wealth are difficult to collect, but The five countries for which no survey data are 4.  such people are increasingly on the radar in available to estimate poverty (Eritrea, Equato- discussions of inequality around the globe. rial Guinea, Somalia, South Sudan, and Zim- Africa had 19 billionaires in 2014 accord- babwe) were assigned the regional poverty rate ing to the Forbe’s list of “The World’s Bil- based on the other 43 countries. lionaires.” Aggregate billionaire wealth S en’s capability approach provides the philo- 5.  increased steadily between 2010 and 2014 in sophical foundations for the nonmonetary Nigeria (from 0.3 percent to 3.2 percent of perspective. B elow-average performance in Africa’s three 6.  GDP) and South Africa (from 1.6 percent to most populous countries (Nigeria, the Demo- 3.9 percent). The number of ultra-high-net- cratic Republic of Congo, and Ethiopia) partly worth individuals (people witha net worth of drives the high levels of nonmonetary poverty at least $30 million) also rose. Few detailed in the region. studies explore the level of extreme wealth of H igher life expectancy for women is possible 7.  nationals. One exception comes from Kenya, even in an environment that is disadvantageous where 8,300 people are estimated to own 62 to them, given that women are genetically pre- percent of the country’s wealth (New World disposed to live longer (Sen 2002; World Bank Wealth 2014). The share of extreme wealth 2011). derived from areas prone to political cap- ture, including extractives, has been declin- ing, while the share derived from services References and investment has been increasing. Between ACLED (Armed Conflict Location and Event 2011 and 2014, 4 out of 20 billionaires in Data Project) http://www.acleddata.com Africa derived their wealth mainly or par- /about-acled/. tially from telecommunications. Data limi- Alkire, Sabina, and Maria Emma Santos. 2014. tations make it difficult to draw conclusions “Measuring Acute Poverty in the Developing about whether the emergence of extreme World: Robustness and Scope of the Multidi- wealth in Africa is driven less by political mensional Poverty Index.” World Develop- connections than it used to be. ment 59: 251–74. 18   POVERTY IN A RISING AFRICA Berg, Andrew, Jonathan D. Ostry, and Jeromin People.” Policy Research Working Paper 7253, Zettelmeyer. 2012. “What Makes Growth World Bank, Washington, DC. Sustained?” Journal of Development Econom- Ferreira, Francisco H. G., and María Ana Lugo. ics 98 (2): 149–66. 2013. “Multidimensional Poverty Analysis: Bloom, David, and Jeffrey Sachs. 1998. “Geog- Looking for a Middle Ground.” World Bank raphy, Demography, and Economic Growth Research Observer 28 (2): 220–35. in Africa.” Brookings Papers on Economic Filmer, Deon. 2008. “Disability, Poverty, and Activity 2: 207–95. Schooling in Developing Countries: Results Canning, David, Sangeeta Raja, and Abdo from 14 Household Surveys.” World Bank Yazbeck, eds. 2015. Africa’s Demographic Economic Review 22 (1): 141–63. Transition: Dividend or Disaster? Africa Florian, Krätke, and Bruce Byiers. 2014. “The Development Forum Series. Washington, DC: Political Economy of Official Statistics: World Bank. Implications for the Data Revolution in Sub- Carletto, Calogero, Dean Jolliffe, and Raka Saharan Africa.” PARIS21 Discussion Paper 5. Banerjee. 2015. “From Tragedy to Renais- Hamilton, Bruce W. 2001. “Using Engel’s Law sance: Improving Agriculture Data for Better to Estimate CPI Bias.” American Economic Policies.” Journal of Development Studies 51 Review 91 (3): 619–30. (2):133–48. Hoogeveen, Johannes, and Nga Thi Viet Nguyen. CGD (Center for Global Development). 2014. 2015. “Statistics Reform in Africa: Align- ing Incentives with Results.” Working Paper, Delivering on the Data Revolution in Sub- World Bank, Poverty and Equity Global Prac- Saharan Africa. Final Report of the Data for tice, Washington, DC. African Development Working Group. Center IMF (International Monetary Fund). 2003. for Global Development and African Popula- Ghana: First Review under the Three-Year tion and Health Research Center, Washington, Arrangement under the Poverty Reduction DC. and Growth Facility. IMF Country Report Christiaensen, Luc, Peter Lanjouw, Jill Luoto, and 03/395, Washington, DC. David Stifel. 2012. “Small Area Estimation- ———. 2007. Ghana: Article IV Consultation: Based Prediction Methods to Track Poverty: Staff Report. IMF Country Report 07/210, Validation and Applications.” Journal of Eco- Washington, DC. nomic Inequality 10 (2): 267–97. Jerven, Morten. 2013. “Comparability of GDP Costa, Dora L. 2001. “Estimating Real Income in Estimates in Sub-Saharan Africa: The Effect of the United States from 1888 to 1994: Correct- Revisions in Sources and Methods since Struc- ing CPI Bias Using Engel Curves.” Journal of tural Adjustment.” Review of Income and Political Economy 109 (6): 1288–310. Wealth 59 (S1): S16–S36. Dabalen, Andrew, and Hai-Anh Dang. 2015. Lakner, Christoph, and Branko Milanovic. 2015. “The Transition of Welfare over Time for “Global Income Distribution: From the Fall of Africa: Evidence from Synthetic Panel Analy- the Berlin Wall to the Great Recession.” World sis.” Background paper prepared for this Bank Economic Review. Advance access pub- report, World Bank, Washington, DC. lished September 26, 2015. de la Briere, Benedicte, Deon Filmer, Dena Lanjouw, Peter, and Martin Ravallion. 1995. Ringold, Dominic Rohner, Karelle Samuda, “Poverty and Household Size.” Economic and Anastasiya Denisova. 2015. From Mines Journal 105 (433): 1415–34. to Minds: Turning Sub-Saharan’s Mineral Loayza, Norman V., and Claudio Raddatz. 2010. Wealth into Human Capital. Washington, “The Composition of Growth Matters for DC: World Bank. Poverty Alleviation.” Journal of Development Demographic and Health Surveys. 2015. Calver- Economics 93 (1): 137–51. ton, MD: ICF International. Maystadt, Jean-François, and Philip Verwimp. Devarajan, Shantayanan. 2013. “Africa’s Statisti- 2015. “Forced Displacement and Refugees in cal Tragedy.” Review of Income and Wealth Sub-Saharan Africa: An Economic Inquiry.” 59 (S1): S9–S15. Policy Research Working Paper 7517, World Etang-Ndip, Alvin, Johannes Hoogeveen, and Bank, Washington, DC. Julia Lendorfer. 2015. “Socioeconomic Impact Milazzo, Annamaria, and Dominique van de of the Crisis in North Mali on Displaced Walle. 2015. “Women Left Behind? Poverty O V E R V I E W   19 and Headship in Africa.” Policy Research UN (United Nations). 2015. The Millennium Working Paper 7331, World Bank, Washing- Development Goals Report 2015. New York: ton, DC. United Nations. Mitra, Sophie, Aleksandra Posärac, and Brandon UNESCO (United Nations Educational, Scientific Vick. 2013. “Disability and Poverty in Devel- and Cultural Organization). 2015. Education oping Countries: A Multidimensional Study.” for All 2000–2015: Achievements and Chal- World Development 41: 1–18. lenges. EFA Global Monitoring Report. Paris: New World Wealth. 2014. Wealth in Kenya: The UNESCO. Future of Kenyan HNWIs. Johannesburg. van de Walle, Dominique, and Annamaria Mila- Newhouse, David, S. Shivakumaran, Shinya zzo. 2015. “Are Female-Headed Households Takamatsu, and Nobuo Yoshida. 2014. “How Poorer? New Evidence for Africa.” mimeo, Survey-to-Survey Imputation Can Fail.” Policy DECRG. World Bank, Washington, DC. Research Working Paper 6961, World Bank, World Bank. 2007. “Underreporting of Consumer Washington, DC. Price Inflation in Tanzania 2002–2006.” Raleigh, Clionadh, Andrew Linke, Håvard Hegre, World Bank Policy Note, Washington, DC. and Joakim Karlsen. 2010. “Introducing ———. 2011. World Development Report 2012: ACLED-Armed Conflict Location and Event Gender and Development. Washington, DC: Data.” Journal of Peace Research 47 (5). World Bank. Oxfam. 2015. “Wealth: Having It All and Want- ———. 2012. Niger: Investing for Prosperity: A ing More.” Oxfam Issue Briefing, January. Poverty Assessment. Washington, DC: World Oxford. Bank. Sandel, Michael J. 2012. What Money Can’t Buy: ———. 2013. “Burkina Faso: A Policy Note: The Moral Limits of Markets. New York: Poverty Trends and Profile for 2003–2009.” Farrar, Straus and Giroux. World Bank, Washington, DC. Sen, Amartya. 1985. Commodities and Capabili- ———. 2015a. A Measured Approach to End- ties. Amsterdam: North-Holland. ing Poverty and Boosting Shared Prosperity: ———. 2002. “Why Health Equity?” Health Concept, Data, and the Twin Goals. Policy Economics 11 (8): 659–66. Research Report. Washington, DC: World Stiglitz, Joseph E. 2012. The Price of Inequality: Bank. How Today’s Divided Society Endangers Our ———. 2015b. “Tanzania Mainland Poverty Future. New York: W. W. Norton. Assessment.” World Bank, Washington, DC. ECO-AUDIT Environmental Benefits Statement The World Bank Group is committed to reducing its environmental footprint. In support of this commitment, the Publishing and Knowledge Division leverages electronic publishing options and print-on-demand technology, which is located in regional hubs worldwide. Together, these initiatives enable print runs to be lowered and shipping distances decreased, resulting in reduced paper consumption, chemical use, greenhouse gas emissions, and waste. The Publishing and Knowledge Division follows the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on Forest Stewardship Council (FSC)–certified paper, with nearly all containing 50–100 percent recycled content. The recycled fiber in our book paper is either unbleached or bleached using totally chlorine-free (TCF), processed chlorine-free (PCF), or enhanced elemental chlorine-free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www .worldbank.org/corporateresponsibility. Contents of Poverty in a Rising Africa Foreword Acknowledgments About the Authors and Contributors Abbreviations Key Messages Overview Assessing the Data Landscape Improving Data on Poverty Revisiting Poverty Trends. Profiling the Poor Taking a Nonmonetary Perspective Measuring Inequality Introduction 1.  The State of Data for Measuring Poverty Types of Data for Measuring Monetary Poverty The Political Economy of Data Production. Reappraising the Information Base on Poverty Concluding Remarks and Recommendations 2.  Revisiting Poverty Trends Trends Using Comparable and Better-Quality Data Robustness to Reliance on GDP Imputation. Profiling the Poor The Movement of People into and out of Poverty Concluding Remarks 3.  Poverty from a Nonmonetary Perspective The Capability Approach Levels of and Trends in Well-Being Multiple Deprivation. Concluding Remarks 4.  Inequality in Africa Perceptions of Inequality Measurement of Inequality Inequality Patterns and Trends Unequal Opportunities Extreme Wealth and Billionaires Concluding Remarks Africa Povert y Report “Our knowledge of global poverty has long been compromised by inadequacies in the underlying data, in which people report how much they spend. This is nowhere more true than in Africa, where national accounts data are also very weak. So I am delighted that this report and the World Bank are directing serious attention to these issues and hope that, before too long, our basic database will be much stronger than it is today. In the meantime, this report provides an insightful synthesis of what is known about poverty in Africa today, in both its monetary and nonmonetary dimensions.” — ANGUS DEATON, Nobel Laureate in Economic Sciences, 2015; Dwight D. Eisenhower Professor of Economics and International Affairs, Princeton University “Drawing on data to profile well-being in Africa is no easy task. I strongly commend the efforts of synchronizing and reviewing many data sources in this report. It sheds a bright light—based on data—on the poverty profile of Sub-Saharan African countries over time. Looking forward, it is critical that we continue the work of strengthening Africa’s statistics and filling in data gaps that this report carefully outlines. While investing in consumption data is paramount, strengthening the methodological and analytical capacity for measuring nonmonetary dimensions of poverty must also be given a due priority. We need to focus on the multidimensions of well-being, but without high-quality and frequent data, knowledge gaps will remain.” — SAMIA ZEKARIA GUTU, Ethiopian Ambassador to Nigeria; Former Director General, Ethiopia Central Statistical Agency “Following Africa’s rise, are people better off? Scrutinizing the available information, this report does an exemplary job of addressing this pertinent question. It also draws our attention to the seminal crisis at the heart of global decision making around development issues—the crisis of poor data. Bad data feed bad management, weak accountability, loss of resources, and then, bad results. Embracing the more and better data agenda is key for better governance and faster poverty reduction.” — MO IBRAHIM, Founder and Chair, Mo Ibrahim Foundation SKU 33009