Republic of Congo Poverty Assessment Report Education, Jobs and Social Protection for a Sustainable Reduction of Poverty Republic of Congo Poverty Assessment Report Education, Jobs and Social Protection for a Sustainable Reduction of Poverty May 2017 Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@ worldbank.org. Photography: World Bank, Franck Sidney Chrysantheme Bitemo, Clementine Maoungou and Clarence Tsimpo. Design and layout: The Word Express, Inc. Table of Contents Acknowledgements......................................................................................................................................... ix Abbreviations and Acronyms.......................................................................................................................... xi Executive Summary....................................................................................................................................... xiii Executive Overview........................................................................................................................................xxi Introduction.................................................................................................................................................. xli CHAPTER 1: A  Substantial Reduction of Poverty between 2005 and 2011......................................................... 1 1.1 Introduction.......................................................................................................................................................1 1.2 Country context.................................................................................................................................................1 1.3 Rationale for updating the poverty and inequality figures...................................................................................3 1.4 Substantial poverty reduction between 2005 and 2011.......................................................................................5 1.5 Inequalities remain high.....................................................................................................................................6 1.6 As a consequence of increase in inequalities, prosperity was not shared between 2005 and 2011........................8 1.7 Improvement was also substantial on nonmonetary dimensions of well-being....................................................9 1.8 Outlook toward ending extreme poverty by 2030.............................................................................................12 1.9 Conclusion.......................................................................................................................................................14 CHAPTER 2: Th  e Growing Urban-Rural Dichotomy and Vulnerability are of Concern............................... 17 2.1 Introduction.....................................................................................................................................................17 2.2 A growing urban-rural dichotomy....................................................................................................................17 2.3 Migration as a channel toward better living conditions.....................................................................................20 2.4 Pockets of poverty in Brazzaville and Pointe Noire............................................................................................23 2.5 Growing monetary vulnerability.......................................................................................................................24 2.6 The most vulnerable sociodemographic group(s) in the Republic of Congo......................................................26 2.7 Conclusion.......................................................................................................................................................27 iii CHAPTER 3:  Who are the Poor?................................................................................................................... 31 3.1 Introduction.....................................................................................................................................................31 3.2 The poor live either in rural areas or in Brazzaville suburbs...............................................................................31 3.3 Autochthons, youth, elderly, urban women, and urban disabled are more likely to be poor..............................34 3.4 The poor live in large households with high dependency ratios.........................................................................34 3.5 The poor are more likely to be unskilled...........................................................................................................36 3.6 The poor are either unemployed, employed in agriculture, or employed in informal services............................38 3.7 The poor live in dwellings made of non-improved materials and own little assets.............................................40 3.8 Conclusion.......................................................................................................................................................42 CHAPTER 4:  The Drivers of Poverty Reduction and Inequality.................................................................... 45 4.1 Introduction.....................................................................................................................................................45 4.2 Correlates of poverty........................................................................................................................................45 4.3 Drivers of poverty reduction.............................................................................................................................48 4.4 Main factors explaining inequality....................................................................................................................56 4.5 Conclusion.......................................................................................................................................................57 CHAPTER 5:  Employment and Main Income Sources.................................................................................. 59 5.1 Introduction.....................................................................................................................................................59 5.2 Employment status...........................................................................................................................................60 5.3 Sector of activity and type of occupation..........................................................................................................66 5.4 Earnings profile................................................................................................................................................71 5.5 Wage regressions and returns to education........................................................................................................73 5.6 Main sources of income for households............................................................................................................77 5.7 Conclusion.......................................................................................................................................................85 CHAPTER 6:  Access to Basic Social Services................................................................................................. 87 6.1 Introduction.....................................................................................................................................................87 6.2 Education.........................................................................................................................................................87 6.3 Health and nutrition........................................................................................................................................91 6.4 Electricity.........................................................................................................................................................95 6.5 Water and sanitation........................................................................................................................................99 6.6 Information and communication technology and roads..................................................................................103 6.7 Conclusion.....................................................................................................................................................104 Annexes........................................................................................................................................................ 107 Annex 1: Consumption Aggregate.........................................................................................................................107 Annex 2: Poverty Line...........................................................................................................................................113 Annex 3: Poverty Measures....................................................................................................................................116 Annex 4: A  ssets Index and Assets-Based Poverty....................................................................................................118 Annex 5: Inequality Measure.................................................................................................................................120 Annex 6: Poverty Dominance Analysis..................................................................................................................121 Annex 7: D  etails on the Two Definitions of Unemployment.................................................................................123 References ................................................................................................................................................... 139 iv Republic of Congo – Poverty Assessment Report List of Tables Table 1:  Share of the population in poverty, nonmonetary dimensions, 2005–2011 (%)..................................... xxiv Table 1.1:  Trend in poverty measures, 2005–2011......................................................................................................5 Table 1.2: Share of the population in poverty, nonmonetary dimensions, 2005–2011 (%)........................................11 Table 1.3:  Cost of a hypothetical perfectly targeted cash transfer program that will eliminate poverty.......................13 Table 1.4: Simulated trend in poverty headcount......................................................................................................14 Table 2.1:  Change in annual consumption per equivalent adult by location..............................................................19 Table 2.2: Availability of services, share of households for whom the service is availables within 1 km, 2011.............19 Table 2.3: Correlated on migration, 2011.................................................................................................................24 Table 2.4:  Poverty profile for Brazzaville and Pointe Noire using a World Bank survey, 2008....................................25 Table 3.1:  Household size, composition, and dependency ratios (%) by poverty status, 2011....................................35 Table 3.2:  Confidence interval of monthly consumption per equivalent adult by head’s education level....................38 Table 3.3:  Housing characteristics and assets ownership by poverty status (%)..........................................................40 Table 3.4: Housing characteristics and assets ownership by welfare quintile (%), 2011..............................................41 Table 4.1:  Impact of growth and change in inequality on poverty.............................................................................49 Table 4.2: Sectoral decompositions of change in poverty (%)....................................................................................51 Table 4.3: Household size, composition, and dependency ratios (%) by location, 2005–2011..................................53 Table 5.1: Main causes of poverty according to the population.................................................................................59 Table 5.2: Marginal effects—selected variables—on earnings and employment, 2011 (coefficients)..........................75 Table 6.1: Main source of drinking water by welfare quintile..................................................................................100 Table 6.2: Type of toilet by welfare quintile.............................................................................................................100 Table A1.1: Number of items included in the consumption modules........................................................................110 Table A1.2: FAO equivalent adult scale.....................................................................................................................112 Table A1.3: Population by location............................................................................................................................112 Table A1.4: Budget shares by consumption category and mean consumption aggregate.............................................112 Table A2.1: Food basket composition based on ECOM 2005 data............................................................................114 Table A2.2: Poverty lines for 2005 and 2011.............................................................................................................115 Table A4.1: Construction of the assets index, pooled data for 2005 and 2011...........................................................118 Table SA.1: Correlates of the likelihood of migrating.................................................................................................126 Table SA.2: Population shares by poverty, vulnerability, and middle class groups (%)................................................127 Table SA.3: Household composition and dependency ratios (%)...............................................................................127 Table SA.4: Earnings regressions, specification with fewer controls, 2011..................................................................128 Table SA.5: Earnings regressions, specification with more controls, 2011..................................................................129 Table SA.6: Income sources Regressions-Heckman, 2011..........................................................................................132 Table SA.7: Demand for education regressions, 2011................................................................................................134 Table SA.8: Basics statistics on access to electricity, 2011...........................................................................................135 Table SA.9: Basics statistics on access to residential water network, 2011...................................................................136 Table SA.10: Welfare regressions, 2011.......................................................................................................................136 List of Figures Table of Contents Figure 1:  Sector shares in GDP, 1960–2010 (%).................................................................................................... xxii Figure 2: Trend in per capita GDP........................................................................................................................ xxii Table of Contents v Figure 3: Poverty and per capita GDP trends......................................................................................................... xxii Figure 4: National poverty rate by region.............................................................................................................. xxiii Figure 5: Asset ownership, 2005–2011 (%).......................................................................................................... xxiv Figure 6: Poverty headcount (2011 PPP) versus GNI per capita in a cross-country setting................................... xxiv Figure 7:  Impact of growth and changes in inequality on poverty.......................................................................... xxv Figure 8: Sectoral decompositions of change in poverty (%)................................................................................. xxvi Figure 9:  Educational attainment associated with consumption growth only for upper secondary and tertiary levels....................................................................................................... xxvii Figure 10: Gini inequality for the Republic of Congo and selected countries ....................................................... xxviii Figure 11: Growth incidence curves...................................................................................................................... xxviii Figure 12: Contributions to the poverty headcount by location (%)....................................................................... xxix Figure 13: Poverty maps by department, 2011........................................................................................................ xxix Figure 14: Poverty, insecure nonpoor, and middle class status by year and location.................................................. xxx Figure 15: Distribution of Workers by Type of Employer and Education Level....................................................... xxxi Figure 16:  Employment rate and earnings profile by education level...................................................................... xxxii Figure 17: Performance is below expectation in primary and secondary school completion................................... xxxii Figure 18: Distribution of Student by Type of School........................................................................................... xxxiii Figure 19: Reasons for Lack of Satisfaction With the School................................................................................ xxxiii Figure 20:  Mortality rate, under-five (per 1,000 live births).................................................................................. xxxiii Figure 21:  Maternal mortality ratio (national estimate, per 100,000 live births)....................................................xxxiv Figure 22: Type of facility visited...........................................................................................................................xxxiv Figure 23:  Stunting, height for age, the Republic of Congo vs. peers......................................................................xxxv Figure 24: Access to electricity (% of population)...................................................................................................xxxv Figure 25: Reason for not subscribing to electricity................................................................................................xxxv Figure 26: Satisfaction with the electricity service..................................................................................................xxxvi Figure 27:  Improved water source (% of population with access)...........................................................................xxxvi Figure 28:  Improved sanitation facilities (% of population with access)................................................................xxxvii Figure 29:  Correlation between access to improved water and poverty..................................................................xxxvii Figure 30:  Correlation between access to improved sanitation and poverty...........................................................xxxvii Figure 31:  Population covered by mobile cellular network (%)............................................................................xxxviii Figure 32: Internet users (per 100 people)...........................................................................................................xxxviii Figure 1.1: Trend in per capita GDP............................................................................................................................2 Figure 1.2: Sector Shares in GDP.................................................................................................................................3 Figure 1.3: Sector shares in GDP, 1960–2010 (%)........................................................................................................3 Figure 1.4: Population pyramid for the Republic of Congo in 2011.............................................................................3 Figure 1.5: Headcount poverty rates from 2005 to 2011..............................................................................................5 Figure 1.6: Poverty headcount (2011 PPP) versus GNI per capita in a cross-country setting........................................6 Figure 1.7: International poverty rates US$1.90 and US$3.10 (2011 PPP)..................................................................7 Inequality in the Republic of Congo, 2005–2011.......................................................................................8 Figure 1.8:  Figure 1.9: Growth incidence curves by strata, 2005–2011.........................................................................................10 Mean of growth rates across the distribution, 2005–2011.........................................................................11 Figure 1.10:  Figure 1.11: Asset ownership, 2005–2011 (%).............................................................................................................12 Simulated trend in international (US$1.9 dollar a day) poverty headcount...............................................13 Figure 1.12:  vi Republic of Congo – Poverty Assessment Report Figure 2.1: Distribution of the population by ventile and location (%).......................................................................18 Figure 2.2: Contributions to the poverty headcount by location (%)..........................................................................18 Figure 2.3: Yield of main crops generally below median level among comparators, 2014............................................21 Figure 2.4: Incidence of migration and reasons for migrating, 2011...........................................................................24 Figure 2.5: Poverty, insecure nonpoor, and middle class status by year and location....................................................26 Figure 2.6: Share of households affected by various shocks and coping mechanisms when affected by a shock, 2011 (%)...................................................................................................................26 Figure 2.7: Poverty, insecure nonpoor, and middle class status by vulnerable groups...................................................28 Figure 2.8: Share of population having difficulties satisfying their food needs.............................................................29 Figure 3.1: Poverty Profile by Location and Department, 2011..................................................................................32 Figure B3.1: Map of the Republic of Congo.................................................................................................................33 Figure 3.2: Poverty maps by department, 2011...........................................................................................................33 Figure 3.3: Dependency ratio and fertility rate...........................................................................................................35 Figure 3.4: Population share and poverty headcount by level of education of the head and his/her spouse..................36 Figure 3.5: Population share and welfare aggregate by level of education of the head..................................................37 Figure 3.6: Distribution of the level of education of the head by welfare quintile........................................................37 Figure 3.7: Distribution of the population by head’s sector of occupation and welfare quintile...................................38 Figure 3.8: Population share and poverty headcount by sector of occupation of the head...........................................39 Figure 4.1: Correlates of well-being, 2011 (coefficients)..............................................................................................47 Figure 4.2: Impact of growth and changes in inequality on poverty............................................................................49 Figure 4.3: Sectoral decompositions of change in poverty (%)....................................................................................50 Figure 4.4: Distribution of the population according to household head’s characteristics (%).....................................54 Figure 4.5: Decomposition of the change of poverty into endowment and returns effects...........................................55 Figure 4.6: Share of between- and within-group inequality, 2011 (Theil)....................................................................56 Figure 4.7: Absolute change in between- and within-group inequality, 2005–2011 (Theil).........................................57 Figure 5.1: Employment status by gender and age, 2011............................................................................................61 Figure 5.2: Employment profile with first definition of unemployment, 2005–2011..................................................62 Figure 5.3: Employment profile with second definition of unemployment, 2005–2011.............................................64 Figure 5.4: Employment profile by quintile, 2005–2011............................................................................................65 Figure 5.5: Unemployment rate by selected characteristics, 2005–2011......................................................................65 Figure 5.6: Distribution of employment by sector, 2005–2011...................................................................................67 Figure 5.7: Distribution of workers by type of employer, 2011...................................................................................68 Figure 5.8: Distribution of workers by socio-professional category, 2011....................................................................70 Figure 5.9: Employment rate and earnings profile, 2011............................................................................................72 Figure 5.10: Earnings by occupation and by quintile, 2011..........................................................................................74 Figure 5.11: Share of households with positive income by source (%), 2011................................................................78 Figure 5.12: Type of agricultural activity and type of nonfarm business, 2011..............................................................79 Figure 5.13: Share of households with positive income by source and location (%), 2011............................................80 Share of adult population with account at a formal financial institutionl – cross country comparisons.......... 81 Figure B5.1:  Share of adult population with account at a formal financial institution – distribution by gender, Figure B5.2:  welfare and education ..............................................................................................................................81 Figure B5.3: Ranking of the top business environment obstacle for firms, 2009............................................................82 Figure 5.14: Average household annual income by source (XAF), 2011........................................................................83 Table of Contents vii Figure 5.15: Income source as a share of total income (%), 2011..................................................................................84 Figure 5.16: Gini Income Elasticity, 2011....................................................................................................................84 Figure 6.1: Performance below expectations in primary and secondary school completion.........................................88 Figure 6.2: Reasons for not being enrolled (%)...........................................................................................................89 Figure 6.3: Type of school attended and concentration curves....................................................................................90 Figure 6.4: Satisfaction with schools (%)....................................................................................................................91 Figure 6.5: Dominance curves for education expenditures, 2011................................................................................92 Figure 6.6: Performance below expectations in child and maternal mortality..............................................................92 Figure 6.7: Share of Population Sick or Injured in Last Four Weeks (%), 2011...........................................................93 Figure 6.8: Type of health facility visited and concentration curves, 2011...................................................................94 Figure 6.9: Prevalence of stunting, height for age (% of children under five)..............................................................95 Figure 6.10: Access to electricity (% of population)......................................................................................................96 Figure 6.11: Electricity coverage, access, and take-up rates (%).....................................................................................96 Figure 6.12: Reason for not subscribing and main issue with network electricity, 2011................................................97 Pro-poorness of various energy sources......................................................................................................98 Figure 6.13:  Targeting performance (Ω) of electricity subsidies, selected countries.......................................................98 Figure 6.14:  Figure 6.15: Access to improved water and sanitation (% of population)......................................................................99 Figure 6.16: Correlation between access to improved water and sanitation and poverty..............................................100 Figure 6.17: Residential network water coverage, access, and take-up rates (%)..........................................................101 Figure 6.18: Reasons for not subscribing and main issue with residential network water, 2011...................................102 Pro-poorness of sources of drinking water...............................................................................................102 Figure 6.19:  Targeting performance (Ω) of residential network water subsidies, selected countries.............................103 Figure 6.20:  Figure 6.21: Mobile cellular network and Internet users per 100 people.....................................................................104 Road density (km of road per 100 km2 of land area)..............................................................................104 Figure 6.22:  Figure A1.1: Differences in capturing reference periods for food items between surveys..............................................108 Figure A1.2: Difference in capturing durable goods between surveys...........................................................................109 Assets index and consumption per equivalent adult, 2011......................................................................119 Figure A4.1:  Assets-based and consumption-based poverty headcounts, 2011.............................................................119 Figure A4.2:  Figure A6.1: First order stochastic dominance, national sample, 2005–2011...............................................................121 Figure A6.2: First order stochastic by strata and in rural areas, 2005–2011.................................................................122 Figure SA.1: Republic of Congo vs peers in various dimensions..................................................................................124 Figure SA.1: Republic of Congo vs peers in various dimensions (continued)...............................................................125 Concentration curve of various types of health expenditures ..................................................................125 Figure SA.2:  Dominance curve of various types of health expenditures ......................................................................125 Figure SA.3:  List of Boxes Box 1.1: Poverty estimates in the Republic of Congo are based on relatively old data................................................4 Box 3.1:  Administrative structure of the Republic of Congo...................................................................................33 Box 3.2: Ethnicity and poverty: The case of the autochthons (Pygmies)..................................................................34 Box 4.1: Growth and inequality decomposition......................................................................................................50 Box 4.2: Sectoral decomposition.............................................................................................................................55 Box 5.1: The financial market is almost nonexistent in the Republic of Congo.......................................................81 viii Republic of Congo – Poverty Assessment Report Acknowledgements T he World Bank greatly appreciates the close collaboration with the Government of the Republic of Congo (the Ministry of Planning and Integration1 and the National Statistics Office,2 in particular) in the preparation of this report. The team preparing this report was led by Clarence Tsimpo Nkengne (Senior Economist, GPV01) and consisted of Fulbert Tchana Tchana (Senior Economist, GMF07), Etaki Wa Dzon (Economist, GMF07), and Quentin Wodon (Lead Economist, GEDGE). This report was prepared under the guidance of Ahmadou Moustapha Ndiaye (Country Director, AFCC2), Djibrilla Adamou Issa (Country Manager, AFMCG), Pablo Fajnzylber (Practice Manager, GPV01), Pierella Paci (Acting Practice Manager, GPV01), Yisgedullish Amde (Country Program Coordinator, AFCCD), Emmanuel Pinto Moreira (Program Leader, AFCC2), and Johannes G. Hoogeveen (Lead Economist, GPV01). Kathleen Beegle (Program Leaders, AFCW1), Dean Mitchell Jolliffe (Senior Economist, DECPI), and Vasco Molini (Senior Economist, GPV01) were peer reviewers for the report. In addition, the team is grateful for comments and input from Andrew Dabalen, Prospere Backiny, Paolo Verme, Franck Adoho, Gyorgy Bela Fritsche, Rose Mungai, Mahine Diop, Meskerem Mulatu, Amadou Oumar Ba, Ephraim Kebede, Michel Niama (Director-General of the Economy), Samuel Ambapour Kosso (Executive Director, CNSEE), Théophile Bassissila (Economist, CNSEE), Jean-Christophe Boungou Bazika (CERAPE), Euloge Bikindou-Boueya (UERPOD), and the developments partners’ community. Josiane Maloueki Louzolo (Program Assistant, AFMCG), Aimee Niane (Program Assistant, GPV01), Senait Yifru (Operations Analyst, GPV01), and Arlette Sourou (Senior Program Assistant, GPV01) provided assistance and various supports in the preparation of the report. Thanks are especially due to Martin Buchara (Program Assistant, GPV01) for excellent support during the preparation of this report. In addition, the report benefited from proofreading by Nathan Weatherdon and Rachel Nalubega. 1 Ministère du Plan et de l’Intégration. 2 Centre Nationale de la Statistique et des Etudes Economiques (CNSEE). ix Abbreviations and Acronyms ARV Anti-retroviral CBN Cost of Basic Needs CNSEE Centre Nationale de la Statistique et des Etudes Economiques CERAPE Centre d’Etudes et de Recherche sur les Analyses et Politiques Economiques CONFEMEN Conférence des ministres de l’Éducation des Etats et gouvernements de la Francophonie CPI Consumer Price Index DHS Demographic and Health Survey DPT Diphtheria, Pertussis, and Tetanus DRC Democratic Republic of Congo ECOM Enquête Congolaise auprès des Ménages FAO-STAT The Statistics Division of Food and Agriculture Organization of the United Nations FBO Faith-based Organization FGT Foster-Greer-Thorbecke GAVI Global Alliance for Vaccines and Immunization GDP Gross Domestic Product GIE Gini Income Elasticity GNI Gross National Income GPS Global Positioning System HiA Health in Africa IBRD International Bank for Reconstruction and Development ICT Information and Communication Technology IFC International Finance Corporation MDGs Millennium Development Goals MFM Macroeconomics and Fiscal Management MOHP Ministry of Health and Population (Ministère de la santé et la population) NDP National Development Plan NHA National Health Accounts NIS National Institute of Statistics NSO National Statistical Office OLS Ordinary Least Squares xi PEEDU Projet Eau, Electricité et Développement Urbain PPP Purchasing Power Parity PRSP Poverty Reduction Strategy Paper PSU Primary Sampling Unit RIF Recentered Influence Function ROC Republic of Congo SCD Systematic Country Diagnostic SCI Statistical Capacity Index SDI Service Delivery Indicators SMEs Small and Medium Enterprises SNDE Société Nationale de Distribution d’Eau SNE Société Nationale d’Electricité SSA Sub-Saharan Africa STATCAP Statistical Capacity Building UERPOD Union pour l’Etude et la Recherche sur la Population et le Développement WDI World Development Indicators Vice President: Makhtar Diop Senior Director: Carolina Sanchez-Paramo Country Director: Ahmadou Moustapha Ndiaye Country Manager: Djibrilla Adamou Issa Practice Manager: Pablo Fajnzylber Acting Practice Manager Pierella Paci Task Team Leader: Clarence Tsimpo Nkengne xii Republic of Congo – Poverty Assessment Report Executive Summary T his poverty assessment analyzes recent trends in monetary and nonmonetary aspects of poverty and economic vulnerability in the Republic of Congo (ROC), based on two nation- ally representative and broadly comparable household expenditure surveys conducted by the National Institute of Statistics (NIS) in 2005 and 2011. The study determines the drivers of poverty reduction by systematically looking at demographic, labor, and human capital dimen- sions. The report also discusses cross-cutting issues relevant for poverty reduction, such as service delivery, marginalization of autochthons, and others. This study aims to provide policy makers with the knowledge needed to improve the effectiveness of their programs to reduce and finally eradicate extreme poverty in the Republic of Congo.3 Between 2005 and 2011, the Republic of Congo experienced a strong macroeconomic performance that was driven by oil revenues The country’s history has been characterized by ups and downs related to the performance of the oil sector, changes from communism to market economy, civil war, and social unrest. Since the 1960s, the country has experienced four economic phases: 1960–1972; 1973–1984; 1985–1999; and 2000–2014. These periods correspond, respectively, to the pre-oil economy; the first oil boom; the oil price crisis; and the second oil boom (World Bank 2016a). The country also experienced social unrest with a military coup and civil war. The various governments, also in line with the cold war, have moved back and forth from communism towards a market economy. All these affected the economy and people’s perceptions, especially toward the importance of the state as the main job provider. Dutch disease followed the discovery of oil. In the 1960s, the economy was diversified, but with the discovery of oil followed the reduction of the relative importance of other sectors. Most importantly, the agriculture and the manufacturing sectors shrank due to reduced competi- tiveness. As a consequence, the country imports most of its goods, even food. 3 Economic growth, economic diversification job creation and poverty reduction are at the core of the National De- velopment plan (Republic of Congo 2012a, 2012b). xiii The Republic of Congo experienced strong eco- day had declined in the Republic of Congo from 50.2 nomic growth between 2002 and 2015. This strong eco- percent in 2005 to 37.0 percent in 2011. nomic growth was driven by higher oil prices and political Despite the improvement in living standards, the stability. On average, during this period, The Republic of country is still underperforming given its potential Congo’s growth rate stood at 4.5 percent. Between 2005 and status as a middle-income country. Cross-country and 2011, the country grew strongly at an average annual comparisons suggest that the level of international rate of 5.4 percent mainly driven by high oil revenues from poverty in the Republic of Congo is still much higher oil production and the decision of the Government to step than in other comparable middle-income countries. up its investment in infrastructure in 2006. Countries with a similar level of economic development generally have much lower poverty rates. Moreover, the Between 2005 and 2011, the strong international poverty rate (US$1.90 –2011 PPP) in the Republic of Congo is quite close to the Sub-Saharan macroeconomic performance Africa (SSA) average of 42.6 percent. coincided with substantial poverty reduction Poverty reduction was sustained The strong macroeconomic performance translated beyond 2011, but it will be difficult into a substantial reduction of the proportion of for the country to reach the goals of the population living in poverty between 2005 and eradicating extreme poverty by 2030 2011. The proportion of the population living below the national poverty line declined from 50.7 percent in 2005 Microsimulation results suggest that the poverty to 40.9 percent in 2011, a decrease of 9.8 percentage decline has continued beyond 2011, although at a points which is in line with the gross domestic product slower pace. It is estimated that the slowdown of eco- (GDP) growth rate observed during that period. Overall, nomic growth linked to the oil sector led to a slower around 143,000 people moved out of poverty. Changes reduction in poverty after 2011. Between 2005 and in the poverty gap and squared poverty gap follow similar 2011, poverty declined by 1.63 percentage points patterns to those observed for the poverty headcount. annually. In 2016, due to slower economic growth, it is Nationally, despite population growth between the two estimated that the share of the population living below years, the number of poor decreased to 1,658,000 in the national poverty line was around 34 or 35 percent. 2011, from 1,801,000 in 2005. This corresponds to a poverty decline by 1.52 percentage Similar to the national poverty rate, the inter- points annually between 2011 and 2016. national extreme poverty rates have declined sig- Our projections up to 2030 show that, unless nificantly. National poverty measures are used for performance and inequality improve substantially, within-country poverty analysis and for deriving the it will be difficult but not impossible for the country economic policies for poverty eradication. However, the to reach the goals of eradicating extreme poverty by national poverty figures are generally not comparable 2030. Under a very optimistic scenario where the coun- across countries. To compare the Republic of Congo with try is assumed to achieve a 10 percent annual growth other countries in Africa and worldwide, the so-called rate between 2021 and 2030, the US$1.90 PPP per international extreme poverty measures are generally day poverty rate will be 3.6 percent in 2030. Under a used. The most common international poverty line is more realistic scenario based on projected growth by US$1.90 expressed in 2011 purchasing power parity World Bank staff, we project that the US$1.90 PPP (PPP) U.S. dollars. The share of the extremely poor by per day poverty rate will still be high, around 15 per- international standards living below US$1.90 PPP a cent in 2030. xiv Republic of Congo – Poverty Assessment Report The prosperity that the Republic in 2005 to 69.4 in 2011. In rural areas, the depth and of Congo enjoyed with oil windfall severity of poverty has also increased, that is, the rural poor have become poorer. Between 2005 and 2011, did not trickle down to the entire the number of poor increased in rural areas to 951,000 population up from 795,000. While rural areas accounted for 44.3 percent of the poor in 2005, this had increased to 57.4 Inequality levels remain high. There was a slight increase percent by 2011. of the Gini coefficient, although not statistically sig- Poverty is predominantly a rural phenomenon. nificant (0.460 in 2005 and 0.465 in 2011). This slight In addition, urban poverty remains very important, increase is coherent with the fact that poverty decreased especially Brazzaville. In rural areas, seven out of ten more in the largest cities than in other urban areas and (69.4 percent) people are poor; 57.4 percent of poor peo- rural areas. Per equivalent adult consumption among ple live in rural areas. Brazzaville, despite a relatively low the richest 10 percent of households in the Republic of poverty incidence (21.6 percent), has a big share of poor. Congo was 17.2 times that of the poorest in 2005; it Close to 20 percent of poor people live in Brazzaville. increased to 20.0 times by 2011. Thus, rural areas and Brazzaville account for 70 percent By international standards, inequalities are of the overall population and 77 percent of total poor. higher in the Republic of Congo. Cross-country As a consequence of inequality increase and the comparisons suggest that inequality is high in the urban/rural gap, prosperity was not shared with the Republic of Congo. The Republic of Congo is ranked poorest segments of the population. Growth incidence among the most unequal societies based on the World curves suggest that growth was not pro-poor nation- Development Indicators (WDI) data. Data on inequal- ally. The poorest actually experienced a deterioration in ity in 105 countries are available in the WDI beyond their living standards according to consumption-based 2010. The Republic of Congo ranked 90 out of 105 measures of poverty. Those in the middle of the distri- countries on the Gini. bution and a small share of the wealthiest households The autochthons, who represent about 1 percent experienced the large positive growth. of the population, stand out as the most marginal- Success in reducing poverty has resulted in many ized group in the Republic of Congo. Monetary pov- households that are living just above the poverty line erty headcount for autochthons is more than twice the who remain vulnerable to falling below the poverty poverty rate of the remaining population. Close to nine line in the face of a negative shock. As poverty receded, out of ten autochthons are poor. The marginalization of both the vulnerable (insecure nonpoor) and middle class autochthons is characterized by very limited access to groups expanded, with the middle class group expand- social services, including health and education, as well ing faster than the vulnerable/insecure nonpoor group. as the labor market. Thus, they contribute and benefit While only 20.6 percent of the population had con- very little from economic activities. sumption more than twice the poverty line in 2005, this The story that emerges from poverty estimates is increased to 26.3 percent by 2011. In the meantime, 32.8 that of a dual economy. Poverty is becoming an increas- percent of the population, while technically ‘nonpoor’, ingly rural phenomenon, and this should be a much is consuming at a level below an average XAF 550,000 bigger reason for concern. Most of the poverty reduc- yearly (in nominal 2011 CFA francs) per equivalent tion was observed in the two largest cities of Brazzaville adult. Given that a large share of the vulnerable rely and Pointe Noire. What was most worrisome was an on either agriculture or informal activities, which are increase in poverty headcount as well as in the number susceptible to significant output volatility, many remain of poor people in rural areas. In rural areas, the poverty at serious risk of falling back into poverty, at least on a headcount went up by 4.6 percentage points from 64.8 temporary basis. Executive Summary xv On nonmonetary dimensions of Despite improvement over the last decade, well-being, the Republic of Congo access to electricity is very low compared to expecta- tions. Connection rates in the country remain below is performing less than the main expectations compared to peers. Coverage rates have comparators increased substantially, from 26.7 percent in 2005 to 42.5 percent in 2011. The improvement in coverage Performance is way below expectations on most of the is the result of improvement in both access (network other critical education sector indicators, including availability) and take-up. In 2011, close to seven out primary school completion.4 The Republic of Congo of ten households (68 percent) lived in neighborhoods is performing below expectations with regard to primary with electricity network; this is up from 57 percent in school completion. The primary school completion rate 2005. Reflecting the improvement in monetary poverty, was 74 percent in 2012. The primary school completion take-up rate also increased from 47 percent to 62 percent rate has been fluctuating over the last decade. between 2005 and 2011. There has been substantial improvement in child Although the share of the population with access and maternal mortality. However, the country still to safe water slightly increased during the last decade, performs below expectations with regard to maternal it is still far below what is expected. The Republic of mortality and has not reached any of the health-related Congo is performing below expectations with regard to Millennium Development Goals (MDGs). Child and access to improved water. Between 2005 and 2015, access maternal mortality are often used as a measure of the to improved water increased slightly from 72 percent efficiency of the health sector in a given country. Between to 77 percent. However, the country is still performing 2005 and 2012, under-five mortality, which measures below expectations. the probability of children dying between birth and the Access to improved sanitation remains very fifth birthday, dropped from 95.3 to 52.6 per 1,000 low and as a consequence, the country is performing births. Thanks to this improvement, the country is now below expectations on this dimension as well. The performing as expected given its gross national income Republic of Congo is performing below expectations (GNI) level. The story is a bit different regarding mater- regarding access to a safe toilet. In 2014, only 43 percent nal mortality. Despite improvement, the country is still of the population had access to improved sanitation. The performing lower than its peers on maternal mortality. situation is even worse in rural areas where only 13 per- Maternal mortality rate declined from 781 to 426 deaths cent of the population have access to an improved toilet. per 100,000 live births between 2005 and 2012. In information and communication technol- The country is performing above expectations ogy (ICT), the Republic of Congo is performing with regard to stunting, but malnutrition is still high. very well in terms of the mobile cellular network, Stunting, defined as low height for age and an indica- but performance is below expectations in terms of tor of chronic malnutrition, decreased from 31 percent Internet access. The country seems to have harvested in 2005 to 25 percent in 2011. As a result, the country the low hanging fruit as materialized by the strong per- is now performing as expected in comparison to peers. formance of access to mobile phone network. However, Nevertheless, the level of stunting remains quite high. the country is still struggling to reach the next level. High malnutrition reduces agricultural productivity, Access to Internet is very low. It is estimated that only contributes to poverty, and affects education and intel- 7 percent of the population were using the Internet in lectual potential of school children (for example, stunting causes children to start school late because they look too 4 The findings are not specific to the Republic of Congo; as demons- small for their age, and will also be a cause of absentee- trated by de la Brière et al. (2016), all resource-rich countries in SSA ism and repetition of school years). fare poorly on nonmonetary dimensions of well-being. xvi Republic of Congo – Poverty Assessment Report 2014. Beyond the low quality, prices remain far too high the unemployment rate is 32.7 percent for those ages 15 for the general public and could be the main reason for to 29. The corresponding figures for those ages 30–49 and the low usage of Internet. 50–64 are 15.6 and 8.3 percent, respectively. Therefore, The country is also performing below expecta- a program focusing on youth employment should be tions with regard to connectivity and road density. boosted and scaled up as much as possible, given the In 2014, the country barely had 5 km of roads per 100 importance of the youth in the overall population. km2 of land, which is way below expectations. A pos- Access to formal wage jobs is very limited for the sible explanation for this very low density of road could youth. They are more likely be employed on their own be the fact that the vast majority of the population is account or by a household as family helper. As we saw concentrated in the two main cities. Still, the low den- in the previous section, the youth are more likely to be sity of road will definitively translate into connectivity unemployed. Moreover, even when they manage to find a problems and result in economic inefficiencies. job, they are likely to be employed on their own account as is the case for the whole population. On the other hand, The current generation does not youth, ages 15–29, are more likely to be employed by a household (15 percent) or by small and medium enter- seem to be benefitting from the oil prises (SMEs) (12 percent). Thus, a policy toward support- windfall ing the emergence of strong SMEs could be an indirect way to insure the creation of quality jobs for the youth. During the period of strong economic growth, the The formal sector, either public or private, has economy created jobs, but this was offset by popula- failed to create quality jobs for the population. As tion growth. Between 2005 and 2011, the number of a consequence, the vast majority of the labor force is employed increased by 180,000. In the meantime, the employed on their own account. A bit more than three potentially active population (ages 15–64) increased by out of five workers (63 percent) work on their own 264,000. Population growth appears to be a challenge, account, running a business with no employees or being as not only does the country have to create jobs, but involved in subsistence agriculture, and another quarter there must be enough new jobs to at least match the of workers work for a household without pay. The public fast-growing working-age population. administration is the main provider of formal wage jobs. Unemployment seems to have increased between One out of ten workers (14 percent) is employed in the 2005 and 2011, which could represent a puzzle to public administration or by a parastatal firm. The role understand trends in well-being. It seems that the of the private sector in providing jobs to the popula- impressive economic growth was jobless, or at least did not tion is not negligible. Up to 13 percent of workers are have significant impact on unemployment, especially the employed either in large private firms (5 percent) or in youth. This is not surprising given that growth was driven SMEs (8 percent). by the oil sector which is not labor intensive and has limited linkages with the rest of the economy. Moreover, these findings also suggest that the massive infrastructure Education improves income and program under the ‘municipalization’ slogan also failed, welfare only at the secondary level at least between 2005 and 2011, to create substantial and beyond and enough jobs to have a consequential impact on the labor market. This is not surprising either as most of the The level of education matters a lot on the job market. construction work is just creating temporary jobs. Those with a high level of education are more likely to The youth are severely affected by unemploy- work as managers or skilled workers. On the other hand, ment. If we consider the second definition, for example, those with no education or primary education are more Executive Summary xvii likely to work as unskilled workers or laborers. Up to 80 Investment in human capital percent of those with tertiary education are employed as managers. There is also an important proportion (40 Skills and good health are prerequisites for success on the percent) of those with upper secondary education that job market. Good health and education increase chances work as either managers or skilled workers. Upper second- of finding a job and of being more productive, thus earning ary education appears to be the threshold beyond which higher returns. Our analysis suggests that in the Republic of education actually matters for accessing quality jobs. Close Congo, education really did not make a big difference below to half of those with no education or primary education upper secondary. Thus, the fight against poverty should only work as unskilled workers or laborers. Still, there is focus on dropouts and transition to secondary education. a non-negligible proportion of those with no education Educating and equipping the population with the neces- who are either employed as skilled workers or managers. sary skills will be critical as the country moves ahead with Earnings increase with age (an implicit measure its goal of achieving economic diversification.5 Vocational of experience) and with education. Education affects training should also play a key role, especially in the short earnings only when one reaches the secondary level. The run, for those who already dropped out. gains appear only as of secondary education since there are few differences in earnings between those with less Boost agricultural productivity and than primary education and those who completed pri- commercialization mary school but did not pursue their education further. The better educated an individual is, the higher Increasing agricultural productivity will be critical the level of earnings of the individual. The effect of edu- for rural poverty alleviation. The rural population cation on earnings after controlling for type of employ- relies heavily on agriculture as a main income source. ment is substantially significant for those reaching upper The evidence suggests that the country depends on secondary and tertiary education. This has implications imports to satisfy food needs. Yet, availability of arable for the development of a national strategy for education. land and opportunities in the fishing and livestock areas The objective should be to provide education, including provide huge opportunities for the country to achieve vocational training, up to at least upper secondary level. the food sovereignty. In doing so, the rural population Reducing dropouts and improving transition at each could generate enough income to move out of poverty. sublevel will be critical to achieve such a goal. For increased agricultural productivity to be effective, it should be accompanied with other actions such as For poverty reduction to happen improved connectivity to market, research and develop- ment for better inputs, and so on. in a sustainable way, the following actions should be considered Expand coverage of formal social to overcome some of the many safety nets programs challenges Through a program such as LISUNGI,6 the Government Going forward, one of the main development challenges could aim at providing cash transfers to the poor and for the country is to translate oil wealth into better ser- vices delivery, better human capital, and quality jobs for 5 Commercial farming, agro processing, and services (ICT and its population, and to support the poor and vulnerable tourism, among others). so as to insure pro-poorness and inclusiveness of growth. 6 LISUNGI means helps or assistance in Lingala. The LISUNGI program is a strategic pillar of the social protection system that is The following five areas of actions can be derived from designed to support the country in improving delivery of services the analysis: from multiple sectors. xviii Republic of Congo – Poverty Assessment Report vulnerable, including the autochthons. The financial also be important to take measures to facilitate private cost for a hypothetical cash transfer could be quite high, investments aimed at creating jobs outside of the oil about XAF 171.2 billion in 2011 prices, representing sector, including through efforts to improve electricity about 10 percent of the government budget. In 2011 generation and distribution and increase access to credit prices, the national poverty line was estimated at XAF for the private sector. 274,113 per year and per equivalent adult. On average, the distance of the poor from the national poverty line Provide better services to the is 15.4 percent. This indicates that if it was possible to population perfectly target the poor, it will take average annual payments of XAF 103,260 to eliminate poverty in the As illustrated, the Republic of Congo is performing Republic of Congo.7 Overall, the budget for such a cash below expectations on most of the nonmonetary transfer program is estimated at XAF 171.2 billion in dimensions of well-being. The country is underper- 2011 prices. This represents slightly less than 10 percent forming when compared to peers on access to education, of the government budget and about 14 percent of annual health, electricity, safe water, sanitation, ICT, and roads. oil revenues. Despite the high cost, such a program if Improving the availability and quality of service to the linked to some conditionality could represent investment population will help the country establish itself as a true in human capital that will result in important medium- middle-income country. and long-term benefits for the country. In the case of a conditional cash transfer, for example, the likelihood of a household using the windfall in building human capital will be higher. This will, if all other conditions are also in place,8 result in a better educated, better nourished, and healthier population. Support the expansion of the private 7 Of course this is just a hypothetical assumption, as there are many sector for job creation other factors at play, including education that takes time and the likelihood of some poor spending the allocation on useless items such as consumption of products associated with ill health (alcohol, The private sector is playing an important role in the tobacco, etc.). Republic of Congo and will continue to do so. It will 8 Such as quality of service delivery. Executive Summary xix Executive Overview T his poverty assessment analyzes recent trends in monetary and nonmonetary aspects of poverty and economic vulnerability in Republic of Congo (ROC), based on two nation- ally representative and broadly comparable household expenditure surveys conducted by the NIS in 2005 and 2011. The study determines the drivers of poverty reduction by systemati- cally looking at demographic, labor, and human capital dimensions. The report also discusses cross-cutting issues relevant for poverty reduction, such as service delivery, marginalization of autochthons, and others. This study aims to provide policy makers with the knowledge needed to improve the effectiveness of their programs to reduce and finally eradicate extreme poverty in the Republic of Congo. History has been marked by political instability and dependency on the oil sector The country’s history has been characterized by ups and downs related to the performance of the oil sector, changes from communism to market economy, civil war, and social unrest. Since the 1960s, the country has experienced four economic phases: 1960–1972; 1973–1984; 1985–1999; and 2000–2014. These periods correspond, respectively, to the pre-oil economy; the first oil boom; the oil price crisis; and the second oil boom (World Bank 2016). The country also experienced social unrest with a military coup and civil war. The various governments, also in line with the cold war, have moved back and forth from communism toward a market economy. All these affected the economy and people’s perceptions, especially toward the importance of the state as the main job provider. Dutch disease followed the discovery of oil. In the 1960s, the economy was diversified, but with the discovery of oil followed the reduction of the relative importance of other sectors (Figure 1). Most importantly, the agriculture and the manufacturing sectors shrank due to reduced competitiveness. As a consequence, the country imports most of its goods, even food. The Republic of Congo experienced strong economic growth between 2002 and 2015. This strong economic growth was driven by higher oil prices and political stability. On average, during this period, the Republic of Congo’s growth rate stood at 4.5 percent (Figure 2). Between 2005 and 2011, the country grew strongly at an average annual rate of 5.4 percent mainly driven by high oil revenues from oil production and the decision of the Government to step up its invest- ment in infrastructure in 2006. xxi FIGURE 1: Sector Shares in GDP, 1960–2010 (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Services Extractive Industries Manufacturing Agriculture Forestry Source: World Bank Economic Data, 2014. FIGURE 2: Trend in Per Capita GDP FIGURE 3: Poverty and Per Capita GDP Trends 2500 60 $5,600 GDP per capita (constant 2005 US$) 50.7 50 $5,400 50.2 Poverty Rate (%) 2000 40.9 GDP per capita 40 $5,200 37.0 1500 30 $5,000 20 $4,800 1000 10 $4,600 0 $4,400 500 2005 2006 2007 2008 2009 2010 2011 Year 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 $1.9/day PPP National Poverty line GDP per capita Year Source: Authors’ calculation using WDI, the 2005 and 2011 Enquête Congolaise auprès des Ménages (ECOM) surveys. Source: WDI. decrease of 9.8 percentage points which is in line with Between 2005 and 2011, the strong the GDP growth rate observed during that period. macroeconomic performance Overall, around 143,000 people moved out of poverty. coincided with substantial poverty Changes in the poverty gap and squared poverty gap reduction follow similar patterns to those observed for the pov- erty headcount. Nationally, despite population growth The strong macroeconomic performance translated between the two years, the number of poor decreased into a substantial reduction of the proportion of to 1,658,000 in 2011, down from 1,801,000 in 2005. the population living in poverty between 2005 Most of the reduction in poverty was observed and 2011. The proportion of the population living in the two largest cities of Brazzaville and Pointe below the national poverty line declined from 50.7 Noire. In Brazzaville, the poverty headcount went percent in 2005 to 40.9 percent in 2011 (Figure 3), a down by 20 percentage points from 42.3 in 2005 to xxii Republic of Congo – Poverty Assessment Report FIGURE 4: National Poverty Rate by Region Poverty reduction was sustained 80 beyond 2011, but it will be difficult 70 for the country to reach the goals of 60 eradicating extreme poverty by 2030 50 40 Microsimulation results suggest that the poverty 30 20 decline has continued beyond 2011, although at a 10 slower pace. It is estimated that the slowdown of eco- 0 nomic growth linked to the oil sector led to a slower Brazzaville Pointe Other Semi- Rural ROC reduction of poverty after 2011. Between 2005 and Noire municipalities urban 2011, poverty declined by 1.63 percentage points 2005 2011 annually. In 2016, due to slower economic growth, it is Source: Authors’ calculation using the 2005 and 2011 Enquête Congolaise estimated that the share of the population living below auprès des Ménages (ECOM) surveys. the national poverty line was around 34 or 35 percent. This corresponds to a poverty decline by 1.52 percent- age points annually between 2011 and 2016. Our projections up to 2030 show that, unless 21.6 in 2011. Pointe Noire also experienced a sharp performance and inequality improve substantially, it decrease in poverty (13 percentage points), down to will be difficult but not impossible for the country 20.3 in 2011 from 33.5 in 2011 (Figure 4). By con- to reach the goals of eradicating extreme poverty by trast, in rural areas, there was an increase in poverty. 2030. Under a very optimistic scenario where the coun- In rural areas, the poverty headcount went up by 4.6 try is assumed to achieve a 10 percent annual growth percentage points from 64.8 in 2005 to 69.4 in 2011. rate between 2021 and 2030, the US$1.90 PPP per The story that emerges from these estimates is that of a day poverty rate will be 3.6 percent in 2030. Under a dual economy, with much lower and faster decreasing more realistic scenario based on projected growth by poverty in urban areas, especially the two large cities, World Bank staff, we project that the US$1.90 PPP per than in rural areas. day poverty rate will still be high, around 15 percent Similar to the national poverty rate, the inter- in 2030. national extreme poverty rates have declined sig- nificantly. National poverty measures are used for within-country poverty analysis and for deriving the There were also important economic policies for poverty eradication. However, the improvements in nonmonetary national poverty figures are generally not comparable dimensions of well-being across countries. To compare the Republic of Congo with other countries in Africa and worldwide, the so-called Subjective poverty as well as poverty in terms of assets international extreme poverty measures are generally ownership decreased substantially. There was a 10 used. The most common international poverty line is percentage point decline in the share of those living in US$1.90 expressed in 2011 PPP U.S. dollars. The share households struggling to satisfy their food needs (a proxy of the extremely poor by international standards living definition of subjective poverty). Ownership of modern below US$1.90 PPP a day had declined in the Republic assets increased while ownership of traditional assets of Congo from 50.2 percent in 2005 to 37.0 percent in deteriorated. There was a notable increase in the propor- 2011 (Figure 3). tion of households who own a mobile phone, a TV set, a modern chair, or an iron (Figure 5). As a consequence, the Executive Overview xxiii FIGURE 5: Asset Ownership, 2005–2011 (%) Share of the Population in Poverty, TABLE 1:  Nonmonetary Dimensions, 2005–2011 (%) Mattress or bed Phone Subjective poverty Assets-based poverty Radio 2005 2011 2005 2011 Television Brazzaville 42.6 31.2 50.5 21.3 Modern chair Pointe Noire 29.6 23.0 43.1 20.2 Iron Other municipalities 41.0 35.0 71.6 52.7 Bicycle Semi-urban 54.3 34.7 82.4 59.4 Canoe Rural 50.2 37.6 80.3 69.1 Computer Total 42.9 32.2 62.5 40.7 Motorcycle Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. Car or Truck 0 10 20 30 40 50 60 70 80 90 100 2005 2011 and status as a middle-income country. Cross-country Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. regression estimates suggest that the level of international poverty in the Republic of Congo is still much higher than in other comparable middle-income countries magnitude of the reduction in the share of the population (Figure 6). Countries with a similar level of economic living in households considered assets-poor is even larger.9 development generally have much lower poverty rates. Importantly though, the reduction in assets-based pov- Moreover, the international poverty rate (US$1.90 – erty is again observed for all areas, including rural areas. 2011 PPP) in the Republic of Congo is quite close to the SSA average of 42.6 percent. The country is performing below expectations on monetary poverty 9 A factorial analysis is conducted for possible assets, and the poverty Despite the improvement in living standards, the line is set to be as close as possible to the 2011 monetary estimate. country is still underperforming given its potential Projecting this to 2005 allowed to explore the trend. FIGURE 6: Poverty Headcount (2011 PPP) Versus GNI Per Capita in a Cross-Country Setting 1. Poverty headcount at US$1.90 a day (2011 PPP) 2. Poverty headcount at US$3.10 a day (2011 PPP) 80 100 60 at $1.90 a day (2011 PPP) at $3.10 a day (2011 PPP) 80 Poverty headcount ratio Poverty headcount ratio ROC 2005 ROC 2005 40 ROC 2011 60 ROC 2011 20 40 0 20 –20 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. xxiv Republic of Congo – Poverty Assessment Report Inequality and poverty reduction (SCD) (World Bank 2016a), the focus for the coming were mainly driven by education, years is to consolidate LISUNGI, expand for national coverage, and use it as a vehicle to deliver a bundle of sector of occupation, and a interventions and specific productive, incentive-based, reduction in average household size and income-generating activities. Growth, not redistribution, drove poverty Household size, education level, and reduction, but economic growth only labor market dynamics contributed the benefited the urban population most for poverty reduction Growth in mean expenditures contributed most of the Migration accounted for 5.6 percent of the decline poverty reduction observed between 2005 and 2011. in poverty.10 As illustrated in Figure 8, the population The stories vary depending on the areas of residence. In shift across location accounted for 5.6 percent of the Brazzaville, growth and reduction of inequality con- poverty reduction. Population shift is happening in tributed equally to poverty reduction. In Pointe Noire, favor of Brazzaville. Between 2005 and 2011, the share most of the poverty reduction was driven by growth. In of those living in Brazzaville increased by 8 percentage semi-urban areas, growth contributed substantially to points while the share of those living elsewhere decreased. poverty reduction, but part of this was offset by increased Migration appears to be a channel through which poverty inequality. By contrast, in rural areas, growth in mean is shifting from rural to urban areas. Migration itself is consumption was negative, leading to an increase in not enough to improve living conditions. Other intrin- poverty (Figure 7). Inequalities also increased in rural sic factors, including skills and education, are comple- areas, leading to an increase in poverty as well. mentarily important for migration to be successful. Yet, The overall poor performance on inequality those moving from rural areas to urban areas often lack reflects difficulties for the poor and vulnerable to such endowments. As a consequence, migration did not access quality jobs and a limited use of fiscal policy account for a bigger share of poverty reduction. to directly support them. The only serious social safety There was a positive shift toward more produc- net program is the LISUNGI, which is still in the pilot tive sectors that contributed 11.2 percent of poverty phase. As argued in the Systematic Country Diagnostic reduction. Although the share of people living in a household headed by someone who is unemployed or mpact of Growth and Changes in FIGURE 7: I inactive remained stable (about 18 percent) between Inequality on Poverty 2005 and 2011, there was a positive shift toward more 10 productive sectors, especially in the two main cities 5 (Figure 8). It is important to note that structural trans- 0 formation seems to be happening in the wrong direc- –5 tion. The manufacturing sector is shrinking, probably –10 a consequence of Dutch disease. As a consequence, a –15 growing share of households is relying on agriculture –20 and informal services for livelihood. There was a positive –25 Brazzaville Pointe Other Semi- Rural National 10 The 2011 population structure is questionable. For instance, given Noire municipalities urban the solid performance, it is difficult to understand how the share of Growth component Redistribution component those living in Pointe Noire has reduced. Unfortunately, with no access to the sample frame or detailed census, it was not possible for Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. the team to confirm or invalidate the current figures. Executive Overview xxv FIGURE 8: Sectoral Decompositions of Change in Poverty (%) 1. By region 2. By head’s education 2 0.0 1 0 –0.5 –1 –1.0 –2 –3 –1.5 –4 –5 –2.0 –6 –7 –2.5 Brazzaville Pointe Noire Other municipalities Semi-urban Rural Population- shift effect Interaction effect None Primary Secondary 1 Secondary 2 Tertiary Population- shift effect Interaction effect 3. By head sector of occupation 4. Household size 2 0.5 1 0.0 –0.5 0 –1.0 –1 –1.5 –2 –2.0 –2.5 –3 –3.0 –4 –3.5 –5 –4.0 Agriculture Mining/ Manufacturing Services Not Working Population- shift effect Interaction effect 1 Individual 2 to 3 Individuals 4 to 5 Individuals 6 to 7 Individuals 8 Individuals and more Population- shift effect Interaction effect Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. shift toward more productive sectors in the main cities someone with no or primary education increased in rural and adverse shift in rural areas. In the main cities, labor areas from 46 to 53 percent. Thus, intervention toward shifted out of agriculture and manufacturing toward ser- curbing rural poverty should focus, among other things, vices. In rural areas, labor shifted out of manufacturing/ on skills/education beyond primary education. processing back into agriculture, which is less productive. Further analysis highlights that education Richer data sets and further investigations are needed to makes a significant difference only when one reaches understand this paradigm in rural areas. upper secondary or tertiary levels. In a world where Gains in education contributed 14.0 percent digital competencies are playing a growing role, educa- of the poverty reduction. This is a larger share than tion and skills allow households to improve their living for location and sector of occupation. Most of the standards by accessing more productive jobs and increas- improvement in skills happened in the two main cities. ing their productivity in self-employment activities. The In Brazzaville and Pointe Noire, the share of people liv- slight increase of the share of households headed by ing in households headed by someone with primary or someone with upper or tertiary education emerges as a no education declined, while the corresponding figures significant driver of consumption growth in the decom- for secondary and tertiary education increased. On the position analysis (Figure 9). The strong positive correla- contrary, the share of people in households headed by tion of tertiary education and consumption growth is xxvi Republic of Congo – Poverty Assessment Report particularly important for poor households. However, The country has curbed monetary primary and lower education have been related to dete- poverty substantially, but many rioration of welfare. challenges remain Reduction of the size of households accounted for 38.2 percent of poverty reduction. Although fertil- Inequality remains high because of the ity remains high, the Republic of Congo has started its pattern of growth demographic transition. As a consequence, the house- hold size declined by close to one person from 5.12 in At the national level, inequality levels remain high. 2005 to 4.28 in 2011.11 However, the growing propor- There was a slight increase of the Gini coefficient, tion of children is maintaining the dependency ratio at although not statistically significant (0.460 in 2005 and higher levels, especially in rural areas. Still, results sug- 0.465 in 2011), suggesting a slight increase in inequal- gest that the shift in household size contributed a large ity. This increase is coherent with the fact that poverty percentage to poverty reduction. decreased more in the largest cities than in other urban Gender and disability have an indirect impact on and rural areas. Using an alternative measure, per equiva- poverty as women and disabled persons exhibit lower lent adult consumption among the richest 10 percent levels of education and more limited employment of households in the Republic of Congo was 17.2 times opportunities. Gender and disability affect the ability of that of the poorest 10 percent in 2005; it had increased individuals to acquire the required skills, and thus exclude to 20.0 times by 2011. them from productive jobs. It was not possible to conduct By international standards, inequalities are a trend analysis by ethnicity as the relevant information higher in the Republic of Congo. Cross-country com- was not collected in 2005. Still, women, the disabled, and parisons suggest that inequality is high in the Republic autochthons should be considered as marginalized groups of Congo (Figure 10). The Republic of Congo is ranked as they have limited opportunities to access higher levels among the most unequal societies based on WDI data. of education and better jobs. Data on inequality in 105 countries are available in the WDI beyond 2010. The Republic of Congo ranked 90 out of 105 countries on the Gini. As reflected by higher inequalities, prosperity FIGURE 9: Educational Attainment Associated with was not fairly shared between 2005 and 2011. Another Consumption Growth Only for Upper way to look at growth and changes in the distribution Secondary and Tertiary Levels of welfare consists of using growth incidence curves which plot growth rates in consumption at various lev- 2.5% els of welfare (Ravallion and Chen 2003). As shown in 2.0% Figure 11, growth incidence curves suggest that growth 1.5% was not pro-poor nationally. The poorest actually expe- 1.0% rienced a deterioration in their living standards accord- 0.5% ing to consumption-based measures of poverty. Those 0.0% in the middle of the distribution and a small share of –0.5% the wealthiest households experienced the large posi- –1.0% tive growth. 10th quantile 25th quantile Median 75th quantile 90th quantile Primary Secondary 1 Secondary 2 Tertiary 11 These estimates are very close to the Demographic and Health Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. Survey (DHS) findings. Executive Overview xxvii A growing disparity between the two FIGURE 11: Growth Incidence Curves main cities and the rest of the country 5 Annual growth rate, % Poverty is becoming an increasingly rural phenom- 3 enon, and this should be a much bigger reason for 1 concern. Most of the poverty reduction was observed –1 in the two largest cities of Brazzaville and Pointe Noire. –3 What was most worrisome was an increase in poverty –5 headcount as well as in the number of poor people in 1 11 21 31 41 51 61 71 81 91 rural areas. In rural areas, the poverty headcount went Expenditure percentile up by 4.6 percentage points from 64.8 in 2005 to 69.4 Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. in 2011. In rural areas, the depth and severity of poverty has also increased, that is, the rural poor have become poorer. Between 2005 and 2011, the number of poor increased in rural areas to 951,000 up from 795,000. Beyond the urban-rural dichotomy, significant Consequently, the contribution of the rural areas to spatial differences in welfare exist in the Republic of poverty is on the rise (Figure 12). Congo. Figure 13 visualizes geographical disparities in Poverty pockets still exist in suburban slums. poverty headcount by department with maps. Among The Republic of Congo is one of the most urbanized the 12 ‘departments’ in the Republic of Congo, Pointe countries in the world with more than half of the popu- Noire and Brazzaville have by far the lowest poverty rate lation concentrated in two main cities. In the cities, the with a poverty headcount of 20.3 and 21.6, respectively. poor and vulnerable often are left with no choice but to Cuvette-Ouest is the poorest department, with 79.1 per- settle in slums, with insecurity and difficult living con- cent of population living below the poverty line, followed ditions, including lack of access to basic infrastructure by Lékoumou and Cuvette with 76.1 and 70.2 percent (electricity, piped water, transport, and so on). of poor individuals, respectively. The poverty rate in the FIGURE 10: Gini Inequality for the Republic of Congo and Selected Countries 70 60 50 40 30 20 10 0 Ukraine Norway Slovak Republic Kazakhstan Sweden Romania Kyrgyz Republic Netherlands Montenegro Albania Iraq Serbia Austria Cambodia Luxembourg Cyprus Bangladesh Ireland United Kingdom Nepal France Taiwan, China Canada Sierra Leone Italy Australia Indonesia Tunisia Portugal Iran, Islamic Rep. Tanzania Sri Lanka Vietnam Georgia United States Russian Federation El Salvador Congo, Dem. Rep. Uganda Philippines Benin Djibouti Togo Ecuador Congo, Rep. Costa Rica Chile Rwanda Guatemala Honduras Lesotho Haiti Source: WDI. xxviii Republic of Congo – Poverty Assessment Report FIGURE 12: Contributions to the Poverty Headcount by Location (%) 2005 2011 Brazzaville, Rural, 44% Brazzaville, Rural, 57% 20% 24% Pointe Noire, 10% Pointe Noire, 15% Other municipalities, 7% Other municipalities, Semi-urban, 9% 7% Semi-urban, 6% Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. province of Kouilou is 56.9. The poverty rate is quite Increased vulnerability with many at risk high for the remaining departments as well—the pov- of falling back into poverty erty rate ranges between 62 and 69 percent for Plateaux, Likouala, Bouenza, Sangha, Pool, and Niari. Given the Despite moving out of poverty, many households are high rate in most of the departments, we can rank the living just above the poverty line and remain vulner- departments into two groups: the two main cities and able to falling below the poverty line. A way to look the rest of the country. at vulnerability is to classify the population into three FIGURE 13: Poverty Maps by Department, 2011 1. Consumption-based headcount 2. Map of the Republic of Congo (75.0,80.0] (70.0,75.0] Likouala (65.0,70.0] (60.0,65.0] (55.0,60.0] (50.0,55.0] (45.0,50.0] (40.0,45.0] Sangha (35.0,40.0] (30.0,35.0] Cuvette-Ouest (25.0,30.0] (20.0,25.0] (15.0,20.0] (10.0,15.0] Cuvette (5.0,10.0] [0.0,5.0] Plateaux Niari Lékoumou Pool Kouilou Brazzaville Pointe Noire Bouenza Source: Authors’ calculation using the 2011 ECOM survey. Executive Overview xxix groups: the poor (consumption below the poverty line), group in the Republic of Congo. Monetary poverty the insecure nonpoor or vulnerable (consumption above headcount for autochthons is more than twice the pov- the poverty line but below twice the poverty line), and erty rate of the remaining population. Close to nine the middle class (consumption above twice the poverty out of ten autochthons are poor. This group appears line; this would include some wealthy households as well, to be extremely marginalized. Their marginalization is but most households could be considered middle class characterized by very limited access to social services, for a country like the Republic of Congo). Figure 14 including health and education, as well as the labor shows that as poverty regressed, both the vulnerable market. Thus, they contribute and benefit very little (insecure nonpoor) and middle class groups expanded, from economic activities. The autochthons who are with the middle class group expanding faster than the not poor tend to be close to the poverty line, and thus vulnerable/insecure nonpoor group. While only 20.6 ranked as insecure nonpoor. There is a clear difference percent of the population had consumption more than between autochthons living in rural areas and autoch- twice the poverty line in 2005, this increased to 26.3 thons living in urban areas. All autochthons who man- percent by 2011. In the meantime, 32.8 percent of the age to move from rural areas to urban areas are ranked population, while technically ‘nonpoor’, is consuming as middle class. at a level below twice the poverty line, meaning less that XAF 550,000 yearly (in nominal 2011 CFA francs) per The economy is not creating equivalent adult. Given that a large share of the vul- enough jobs, especially for the youth nerable rely on either agriculture or informal activities which are susceptible to significant output volatility, Issues related to job creation are at the center of the many remain at serious risk of falling back into poverty, development debate in the Republic of Congo. For at least on a temporary basis. most households in poverty, inability to meet their basic needs reflects difficulties faced by its members in Marginalization of autochthons accessing quality jobs. The close relationship between poverty and work is very clear to the population. Lack The autochthons, who represent about 1 percent of of employment is the most cited factor leading to pov- the population, stand out as the main marginalized erty in household responses to a question in the 2011 FIGURE 14: Poverty, Insecure Nonpoor, and Middle Class Status by Year and Location 1. By year 2. By residence area, 2011 only 100% 100% 20.6 26.3 80% 80% 28.7 60% 32.8 60% 40% 40% 50.7 20% 40.9 20% 0% 0% 2005 2011 Brazzaville Pointe Other Semi- Rural National Noire municipalities urban 2011 Poor Insecure nonpoor Middle class Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. xxx Republic of Congo – Poverty Assessment Report ECOM survey about the causes of poverty. Lack of Distribution of Workers by Type of FIGURE 15:  employment was cited as a leading cause of poverty by Employer and Education Level 91.7 percent of respondents and insufficient income, a 100% closely related effect, was cited by 59.0 percent. 90% 80% During the strong economic growth, the 70% economy created jobs, but this was offset by popula- 60% tion growth. Between 2005 and 2011, the number of 50% employed increases by 180,000. In the meantime, the 40% 30% potentially active population (ages 15–64) increased by 20% 264,000. Population growth appears to be a challenge, 10% as not only does the country have to create jobs, but 0% None Primary Secondary 1 Secondary 2 Tertiary there must be enough to at least match the fast-growing Education working-age population. Public administration Public firm/parastatal The youth are severely affected by unemploy- Small and medium enterprise (SME ) Large private company ment. In 2011, the unemployment rate was 32.7 percent Association, Cooperative, Church, NGOs Household International Organization, Embassy Own account for those ages 15 to 29. The corresponding figures for those ages 30–49 and 50–64 were 15.6 and 8.3 percent, Source: Authors’ calculation using the 2011 ECOM survey. respectively. Therefore, a program focusing on youth employment should be boosted and scaled up as much as possible, given the importance of the youth in the same activity, women are paid at the same rate as their overall population. male counterparts. The formal sector, either public or private, has Education is positively correlated with earnings. failed to create quality jobs for the population. As The better educated an individual is, the higher the level a consequence, the vast majority of the labor force is of earnings of the individual. The benefit of education employed on their own account in agriculture or infor- for earnings are very significant for those reaching upper mal services. A bit more than three out of five workers secondary and tertiary education (Figure 16). This has (63 percent) work on their own account, running a implications for the development of a national strat- business with no employees or being involved in subsis- egy for education. The objective should be to provide tence agriculture, and another quarter of workers work education, including vocational training, up to at least for a household without pay (Figure 15). The public upper secondary level. Reducing dropouts and improv- administration is the main provider of formal wage ing transition at each sublevel will be critical to achieve jobs. Fourteen percent of workers are employed in the such a goal. public administration or by a parastatal firm. The role of the private sector in providing jobs to the popula- tion is not negligible. Up to 13 percent of workers are The Republic of Congo is employed either in large private firms (5 percent) or in performing below expectations on SMEs (8 percent). service delivery There is no gender-based pay gap. Analysis sug- gests that possible differences observed between women Education and men are due to other factors, including differences in education/skills, but more importantly, difference in the Performance is way below expectations on most of type of activity. Women are more likely to be employed the critical education sector indicators. For example, on their own account. With equal competencies in the the Republic of Congo is performing below expectation Executive Overview xxxi FIGURE 16: Employment Rate and Earnings Profile Despite improvement in enrollment, the quality of by Education Level education is of concern and seems to be decreasing. Test 80 200,000 results suggest that a high proportion of students leaving 70 180,000 primary school do not have sufficient foundational skills. 160,000 Mothnly wage (FCFA) Share employed (%) 60 End-of-cycle tests in primary education have found that 140,000 50 120,000 two-thirds of students leaving primary school do not have 40 100,000 30 80,000 sufficient foundational skills in literacy and numeracy. Thus 20 60,000 more and more children are going to school, but it does not 40,000 10 20,000 guarantee that they are acquiring the expected knowledge. 0 0 This assessment confirms the importance of the ongoing None Primary Secondary 1 Secondary 2 Tertiary debate on the quality of service delivery, particularly educa- tion services in the country (World Bank 2015).12 Education The private sector is playing a very important Employed Mean and growing role in the provision of education ser- Median vices. Figure 18 provides data on the type of school Source: Authors’ calculation using the 2011 ECOM survey. attended. The share of students in private schools seems to have increased between the last two surveys for 2005 and 2011. In 2011, 35 percent of students were enrolled with regard to primary school completion (Figure 17). The in a private school. This is up by 15 percentage points primary school completion rate was 74 percent in 2012. compared to 2005. Such high shares of private provi- The primary school completion rate has been fluctuating sion in a country where 40 percent of the population is over the last decade. Overall, between 2005 and 2012, the poor is an indication of inadequate success in ensuring primary school completion rate increased for girls, but effective access by the Government. Is the deterioration went down for boys. In 2012, primary school comple- tion for girls was 79 percent, up from 70 percent in 2005. 12 The World Bank Education Global Practice is currently conducting During the same period, the primary school completion a Service Delivery Indicators (SDI) survey that will provide more rate for boys went down from 75 percent to 70 percent. evidence of this. FIGURE 17: Performance is Below Expectation in Primary and Secondary School Completion 1. Primary completion rate 2. Lower secondary completion rate 120 150 Lower secondary completion rate, total (% of relevamt age group) Primary completion rate, total (% of relevamt age group) 100 100 80 ROC 2005 ROC 2012 50 ROC 2012 60 ROC 2004 ROC 1996 ROC 1996 40 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. xxxii Republic of Congo – Poverty Assessment Report FIGURE 18: Distribution of Student by Type of School Reasons for Lack of Satisfaction With FIGURE 19:  the School 100% 16.4 21.6 20.6 20.3 25.7 90% Lack of 31.3 32.9 34.9 36.5 books/supplies 0.1 80% 49.4 3.1 1.4 2.4 70% 0.4 Overcrowding 1.3 3.1 2.9 60% 3.4 50% Lack of 0.2 80.9 teachers 76.2 74.4 73.1 71.4 40% 66.6 63.6 61.1 58.9 30% Teachers often 48.9 absent 20% 10% Facilities in bad condition 0% Poor teaching Primary Secondary 1 Secondary 2 Tertiary All Primary Secondary 1 Secondary 2 Tertiary All Other problem 2005 2011 0 5 10 15 20 25 30 35 40 45 50 Government Religious Organization Private 2005 2011 Community Other Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. and maternal mortality are often used as a measure of the of the quality of education a cause or a consequence efficiency of the health sector in a given country. Between of the growing role of the private sector? Further data 2005 and 2012, under-five mortality, which measures and analysis are needed to answer such an important the probability of children dying between birth and the question. fifth birthday, dropped from 95.3 to 52.6 per 1,000 Satisfaction is low for public schools. Lack of births. Thanks to this improvement, the country is now books/supplies and overcrowding are the main reasons performing as expected given its GNI level (Figure 20). for dissatisfaction. Satisfaction with schools, as measured The story is a bit different regarding maternal mortality. by the share of parents who do not have any complaints Despite improvement, the country is still performing with schools, is low, especially in public schools. Less than 20 percent are satisfied with public schools, against FIGURE 20: Mortality Rate, under-Five close to 60 percent for private schools. The main reasons (per 1,000 Live Births) for dissatisfaction are lack of books/supplies, overcrowd- ing, and lack of teachers (Figure 19). The lack of books/ 150 supplies and the lack of teachers are predominant for Mortality rate, under-5 (per 1,000 live births) ROC 1996 children from lower socioeconomic backgrounds, while 100 ROC 2005 overcrowding is cited more often by parents of children from higher levels of well-being. 50 ROC 2014 Health 0 There has been substantial improvement in child and 0 2000 4000 6000 8000 10000 maternal mortality. However, the country still performs GNI per capita, PPP (constant 2011 international $) below expectations with regard to maternal mortality and Source: Authors’ calculation using the WDI data and based on the Gable, has not reached any of the health-related MDGs. Child Lofgren, and Osorio-Rodarte (2015) approach. Executive Overview xxxiii FIGURE 21: Maternal Mortality Ratio (national FIGURE 22: Type of Facility Visited Estimate, Per 100,000 Live Births) 100% 13.8 11.3 9.1 6.0 4.8 8.8 90% national estimate (per 100,000 live births) 1,500 80% 30.3 35.7 40.1 45.1 44.9 39.5 70% Maternal mortality ratio, 60% 4.2 3.1 1,000 50% 2.0 1.8 1.3 2.4 ROC 2005 40% 30% 500 ROC 2012 20% 51.6 49.9 48.7 47.1 49.0 49.2 10% 0% 0 Q1 Q2 Q3 Q4 Q5 Total 0 2000 4000 6000 8000 10000 Public FBO Private Other GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the 2011 ECOM survey Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. The country is performing above expectations with regard to stunting, but malnutrition is still high. lower than its peers on maternal mortality. Maternal Stunting, defined as low height for age and an indica- mortality rate declined from 781 to 426 deaths per tor of chronic malnutrition, decreased from 31 percent 100,000 live births between 2005 and 2012 (Figure 21). in 2005 to 25 percent in 2011. As a result, the country The distribution of private expenditure on is now performing as expected in comparison to peers. health shows that the poor will be more affected by a Nevertheless, the level of stunting remains quite high. fiscal reform concerning drugs, hospitalization, and High malnutrition reduces agricultural productivity, consultation fees. In August 2011, the Government contributes to poverty, and affects education and intel- decided that malaria drugs, anti-retroviral (ARV), and lectual potential of school children (for example, stunting C-section will be free of charge in public facilities. The causes children to start school late because they look too analysis here shows that these measures were actually pro- small for their age, and will also be a cause of absenteeism poor, but new data and additional analytical work are and repetition of school years). As shown in Figure 23, needed to access the effectiveness and the impact of such in the case of the Republic of Congo, malnutrition a measure. Since 2014, financing of inputs for these free increases with poverty. Affordability of food might be health care programs has stagnated and many services the main cause of malnutrition. In principle, this makes are either not provided (free C-section or free health it easier to design a program to fight malnutrition, as it care for pregnant women), not free, or provided against will mainly imply making sure the poor access food, for out-of-pocket payments (in the case of malaria drugs). example, through school canteens. The private sector is playing a very important and growing role in the provision of health services. Electricity Slightly more than half of the population seeks care in private facilities. Only six out of ten individuals who were Despite improvements over the last decade, access sick did seek care. A bit more than half of sick people to electricity is very low compared to expectations. seek care in nongovernment facilities (Figure 22). Such Given that the Republic of Congo is a country with high shares of private provision in a country with high substantial hydro potential, electricity could in principle poverty rate is an indication of inadequate success in be generated and distributed at a relatively low cost to ensuring effective access by the Government. a large share of the population. Coverage rates have xxxiv Republic of Congo – Poverty Assessment Report FIGURE 23: Stunting, Height for Age, the Republic FIGURE 25: Reason for Not Subscribing to Electricity of Congo vs. Peers 100% 60 90% Prevalence of stunting, height for age 80% 70% (% of children under 5) 40 60% 50% ROC 2005 40% ROC 2011 30% 20 20% 10% 0% 0 Q1 Q2 Q3 Q4 Q5 0 2000 4000 6000 8000 10000 Welfare quintile Total GNI per capita, PPP (constant 2011 international $) Too expensive to subscribe Consumption too expensive Source: Authors’ calculation using the 2011 ECOM and the 2012 DHS; and Complicated process Remote network (not accessile) using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte Useless Network not avaliable in the area (2015) approach.. Source: Authors’ estimates based on the 2011 ECOM survey. increased substantially, from 26.7 percent in 2005 to 42.5 percent in 2011. Unfortunately, connection rates Frequent shortages are by far the main reason in the country remain below expectations compared to for dissatisfaction. Poor quality of electricity and peers (Figure 24). high cost are also important reasons for dissatisfac- Those who are not connected to the network tion. Satisfaction with electricity service is very low. quote issues related to affordability and supply Slightly less than one out of three households is satis- as main reasons for not subscribing. Poor people fied with the electricity network service. Seven out of often don’t have electricity in their neighborhood, ten households point to shortage as reason for dissatis- and when the network is present, they cannot afford faction (Figure 26). The second reason for dissatisfac- it (Figure 25). tion is issues related to quality (quote by 13 percent of FIGURE 24: Access to Electricity (% of Population) 1. Urban areas 2. Rural areas 120 100 Access to electricity, urban Access to electricity, rural 100 80 (% of urban population) (% of rural population) 80 60 60 ROC 2012 40 40 ROC 2000 20 ROC 2012 20 ROC 2000 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. Executive Overview xxxv FIGURE 26: Satisfaction with the Electricity Service FIGURE 27: Improved Water Source (% of Population with Access) 100% 90% 100 80% (% of population with access) 70% Improved water source 60% 50% 80 ROC 2014 40% ROC 2005 ROC 1997 30% 20% 60 10% 0% Q1 Q2 Q3 Q4 Q5 40 Welfare quintile Total 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) Too expensive Frequent shortages Poor quality Service interruption/landslide Not concerned Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. Source: Authors’ estimates based on the 2011 ECOM survey. households). The cost of electricity is the third reason differences between those using electricity and those pay- (11.5 percent). Cost issues are more predominant in ing for it are big. In 2011, 42.5 percent of households poor households. For the bottom quintile, cost of elec- were connected to the grid, but only 30 percent were pay- tricity is quoted by three out of ten households. ing for electricity, generating a 12.5 percentage point gap Electricity tariffs are poorly designed in the between coverage and payment. The Société Nationale de Republic of Congo. The implicit consumption subsi- Distribution d’Eau (SNDE) is facing a similar challenge. dies are poorly targeted. Connection subsidies clearly have the potential to be better targeted. A robust mea- Water and sanitation sure of the targeting performance of subsidies is the share of the subsidy benefits received by the poor divided by Although the share of the population with access the proportion of the population in poverty (Ω).13 The to improved water slightly increased during the last tariffs design is pro-poor if Ω>1. For the Republic of decade, it is still far below what is expected. The Congo, the value was 0.62 in 2011, that is, implicit Republic of Congo is performing below expectations electricity subsidies are not pro-poor in the Republic with regard to access to improved water. Between 2005 of Congo. To our knowledge, electricity tariffs have not and 2015, access to improved water increased slightly been revised for a very long time. Pro-poorness of tariffs from 72 percent to 77 percent. However, the coun- design should be considered in the future. try is still performing below expectations (Figure 27). Evidence suggests that Société Nationale Poverty is correlated with the lack of access to safe water d’Electricité (SNE) is struggling to effectively col- (Figure 29). lect payments from residential customers, including Access to improved sanitation remains very through the installation of new meters in some areas. low and as a consequence, the country is perform- The share of households paying for their electricity is ing below expectations on this dimension as well. systematically lower than the share of households who As illustrated in Figure 28, the Republic of Congo is declare using electricity. This may be an indication of illicit connections, but it may also reflect late payment, lack of recovery of late payment, or other issues. The 13 See Angel-Urdinola and Wodon (2007) for more details. xxxvi Republic of Congo – Poverty Assessment Report FIGURE 28: Improved Sanitation Facilities (% of Correlation between Access to FIGURE 29:  Population with Access) Improved Water and Poverty 50 100 45 Cuvette-Ouest (% of population with access) Improved sanitation facilities 40 Monetary poverty (FGT1) 80 35 Lékoumou Cuvette 30 Likouala Plateaux Sangha 60 25 Pool Niari Bouenza 40 ROC 2012 20 Kouilou 15 20 10 y = 123.84e–0.03x 5 Brazzaville R² = 0.7462 Pointe-Noire 0 0 30 50 70 90 110 0 2000 4000 6000 8000 10000 Access to Improved Water Source (%) GNI per capita, PPP (constant 2011 international $) Source: Authors’ estimates based on the 2011 ECOM survey. Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. still struggling to reach the next level. Access to Internet is very low and way below expectations (Figure 32). It performing below expectations with regard to access to is estimated that only 7 percent of the population were a safe toilet. In 2014, only 43 percent of the population using the Internet in 2014. Beyond the low quality, prices had access to improved sanitation. The situation is even remain far too high for the general public and could be the worse in rural areas where only 13 percent of the popula- main reason for the low usage of Internet. At US$1,200, tion have access to an improved toilet. More of concern the Republic of Congo was ranked as one of the most is the share of the population with no toilet at all. The expensive countries in the world with regard to the price situation is particularly of concern in four departments: of international bandwidth (Mbps/month) (Jacquelot, Plateaux, Lékoumou, Cuvette, Cuvette-Ouest. The share quote by World Bank 2016a). of households with no toilet ranges from 42.8 to 30 per- cent in these departments. Here again, the gap between the two main cities and the rest of the country is very pronounced (Figure 30). Correlation between Access to FIGURE 30:  Information and communication Improved Sanitation and Poverty technology 50 45 y = 10.522e0.0335x R² = 0.51722 Cuvette-Ouest The Republic of Congo is performing very well with 40 Monetary poverty (FGT1) regard to the mobile cellular network, but performance 35 Cuvette Likouala 30 Niari Lékoumou is below expectations with regard to Internet access. Sangha Plateaux 25 Bouenza Pool ICT provides new opportunities to boost productivity and 20 Kouilou save costs. Moreover, applications in various areas such as 15 10 education, agriculture, health, and so on have proven to Brazzaville 5 Pointe-Noire be critical for economic development and well-being. For 0 now, the country seems to have harvested the low hanging 0 10 20 30 40 50 fruit as materialized by the strong performance of access Share of households with no toilet (%) to mobile phone network (Figure 31). The country is Source: Authors’ estimates based on the 2011 ECOM survey. Executive Overview xxxvii FIGURE 31: Population Covered by Mobile Cellular Investment in human capital Network (%) Skills and good health are prerequisites for success 100 on the job market. Good health and education increase ROC 2011 mobile cellular network (%) chances of finding a job and of being more productive, Population covered by 80 thus earning higher returns. Our analysis suggests that in the Republic of Congo, education really did not make 60 a big difference below upper secondary. Thus, the fight 40 ROC 2005 against poverty should focus on dropouts and transi- tion to secondary education. Educating and equipping 20 the population with the necessary skills will be critical 0 2000 4000 6000 8000 10000 as the country moves ahead with its goal of achieving GNI per capita, PPP (constant 2011 international $) economic diversification.14 Vocational training should Source: Authors’ calculation using the WDI data and based on the Gable, also play a key role, especially in the short run, for those Lofgren, and Osorio-Rodarte (2015) approach. who already drop out. FIGURE 32: Internet Users (per 100 People) Boost agricultural productivity and commercialization 100 Internet users (per 100 people) Increasing agricultural productivity will be critical 80 for rural poverty alleviation. The rural population relies heavily on agriculture as a main income source. 60 The evidence suggests that the country depends on imports to satisfy food needs. Yet, availability of arable 40 land and opportunities in the fishing and livestock areas ROC 2014 ROC 2005 provide huge opportunities for the country to achieve 20 the food sovereignty. In doing so, the rural population 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) could generate enough income to move out of poverty. For increased agricultural productivity to be effective, Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. it should be accompanied with other actions such as improved connectivity to market, research and develop- ment for better inputs, and so on. Looking forward: how to overcome Expand coverage of formal social some of the many challenges safety nets programs Going forward, one of the main development challenges Through a program such as LISUNGI, the Government for the country is to translate oil wealth into better ser- could aim at providing cash transfers to the poor and vices delivery, better human capital, and quality jobs for vulnerable, including the autochthons. The financial its population, and to support the poor and vulnerable cost for a hypothetical cash transfer could be quite high, so as to insure pro-poorness and inclusiveness of growth. The following five areas of actions can be derived from 14 Commercial farming, agro processing, and services (ICT and the analysis: tourism, among others). xxxviii Republic of Congo – Poverty Assessment Report about XAF 171.2 billion in 2011 prices, representing also be important to take measures to facilitate private about 10 percent of the government budget. In 2011 investments aimed at creating jobs outside of the oil prices, the national poverty line was estimated at XAF sector, including through efforts to improve electricity 274,113 per year and per equivalent adult. On average, generation and distribution and increase access to credit the distance of the poor from the national poverty line for the private sector. is 15.4 percent. This indicates that if it was possible to perfectly target the poor, it will take average annual Provide better services to the payments of XAF 103,260 to eliminate poverty in the population Republic of Congo.15 Overall, the budget for such a cash transfer program is estimated at XAF 171.2 bil- As illustrated, the Republic of Congo is performing lion in 2011 prices. This represents slightly less than below expectations on most of the nonmonetary 10 percent of the government budget and about 14 dimensions of well-being. The country is underper- percent of annual oil revenues. Despite the high cost, forming when compared to peers on access to education, such a program if linked to some conditionality could health, electricity, safe water, sanitation, ICT, and roads. represent investment in human capital that will result Improving the availability and quality of service to the in important medium- and long-term benefits for the population will help the country establish itself as a true country. In the case of a conditional cash transfer, for middle-income country. example, the likelihood of a household using the windfall in building human capital will be higher. This will, if all other conditions are also in place,16 result in a better educated, better nourished, and healthier population. Support the expansion of the private sector for job creation 15 Of course this is just a hypothetical assumption, as there are many other factors at play, including education that takes time and the likelihood of some poor spending the allocation on useless items such The private sector is playing an important role in the as consumption of products associated with ill health. Republic of Congo and will continue to do so. It will 16 Such as quality of service delivery. Executive Overview xxxix Introduction T his poverty assessment analyzes trends in monetary and nonmonetary aspects of poverty and economic vulnerability in Republic of Congo (ROC), based on two nationally representative and comparable household expenditure surveys conducted by the NIS in 2005 and 2011. The study determines the drivers of poverty reduction by systematically looking at demographic, labor, and human capital dimensions. The report also discusses cross-cutting issues relevant for poverty reduction, such as service delivery, marginalization of autochthons, and others. This study aims to provide policy makers with the knowledge needed to improve the effectiveness of their programs to reduce and finally eradicate extreme poverty in the Republic of Congo. This report examines the Republic of Congo’s progress in reducing poverty over the last decade, with a specific focus on the period 2005 to 2011. The focus on this period is due to data avail- ability. The Republic of Congo’s progress in reducing poverty from 2005 and 2011 is substantial. Still, empirical evidence to sustain this claim is limited. Annual reductions in the national poverty rate of 1.5 percentage points resulted from strong growth driven by oil prices and investments in services and infrastructures, especially in the two metropolises (Brazzaville and Pointe Noire). The Republic of Congo is one of the most urbanized countries in the world. This has many implications on issues related to shared prosperity and poverty. Latest estimates suggest that 61.8 percent of the population live in urban areas, with a bigger concentration in the two metropolises (Brazzaville and Pointe Noire). A concentration in urban areas makes it easier to provide public service to the population. At the same time, there is a risk that those living rural areas will be left out. This is exactly what is happening in the Republic of Congo. Between 2005 and 2011, all the poverty indices increased in rural areas. Unfortunately, data collection effort has been low in the Republic of Congo and as a con- sequence, the country has only two household consumption surveys: the 2005 and 2011 ECOM surveys. The core analysis undertaken in this report uses two series of surveys: (a) The ECOM undertaken in 2005 and 2011 and (b) the DHS undertaken in 2005 and 2011/2 (henceforth referred to as 2005 and 2012). All these surveys are a nationally representative cross section and it is from this series that the official monetary poverty and other MDGs estimates are derived. Data from the WDI are also used to assess the country’s performance compared to its peers. Despite lack of availability of recent data, the exercise of conducting this poverty assess- ment has proven to be critical in understanding poverty in the Republic of Congo. It has served xli as a learning experience for World Bank staff as well as Chapter 3 relies on descriptive statistics to provide government officials, and builds the ground for future the poverty profile for 2011. Chapter 4 abandons the collaboration in planning and analysis of the upcoming descriptive angle and uses regressions and decomposi- ECOM 3 survey. tions to examine the drivers of consumption growth This poverty assessment consists of six chapters. and poverty reduction, focusing in particular on four The first chapter presents the main trends in mon- important factors that are derived from the previous etary and nonmonetary poverty, as well as inequali- chapter: location, ethnicity, education, and labor. The ties. In addition, the chapter examines the incidence fifth chapter provides an assessment of the labor mar- of consumption growth and simulates poverty trends ket and households’ income sources. Finally, Chapter beyond 2011. The second chapter highlights some 6 covers the country performance on service delivery, downsides of the positive performance, including the specifically education, health, water and sanitation, expanding urban-rural gap and increased vulnerability. ICT, and roads. The chapter looks at the relationship In particular, the chapter assesses the degree to which between service delivery and poverty reduction. It also households who are classified as nonpoor are at risk highlights that the Republic of Congo is underperform- of falling back into poverty if hit by a negative shock. ing compared to its peers. xlii Republic of Congo – Poverty Assessment Report A Substantial Reduction of Poverty between 2005 and 2011 1 1.1 Introduction The Republic of Congo (ROC) experienced strong macroeconomic performance between 2002 and 2015. Between 2002 and 2015, the Republic of Congo’s economy grew at an average rate of 4.5 percent per year. During the sub-period 2005–2011, the country was among the top performers in Africa with an average annual growth rate of 5.4 percent. The strong growth performance was accompanied by substantial improvements in living standards. For instance, under-five mortality rate dropped by two-thirds between 2005 and 2011. A notable improvement was also registered in many other dimensions of well-being, including primary school enrollment and completion, access to improved water, access to electricity, and so on. Monetary poverty headcount followed the general trend, with the share of population living below the national poverty line dropping from 50.7 percent in 2005 to 40.9 percent in 2011. This chapter will use several sources of data to paint a detailed picture of the evolution of poverty and household living standards in the Republic of Congo, with a focus on the period for which household surveys are available, that is between 2005 and 2011. The chapter has seven parts. In the first part, we will revisit the country context. In the second part, we will offer a rationale for updating the poverty and inequality figures. In the third part, we will offer a detailed description of the evolution of poverty using data from the only two available surveys. The fourth part will present a detailed description of the evolution of inequalities. Shared pros- perity will be discussed in the fifth part. Trend in nonmonetary dimensions of well-being will be presented in the sixth part. The last part will rely on microsimulations to illustrate the challenging path the country will have to follow to eradicate extreme poverty by 2030. 1.2 Country context The country’s history has been characterized by ups and downs related to the performance of the oil sector, changes from communism to market economy, civil war, and social unrest. Since the 1960s, the country has experienced four economic phases: 1960–1972; 1973–1984; 1985–1999; 2000–2014. These periods correspond, respectively, to the pre-oil economy; first oil 1 boom; oil price crisis; and the second oil boom (World FIGURE 1.1: Trend in Per Capita GDP Bank 2016a). The country also experienced social unrest 2500 with a military coup and civil war (Bhattacharya, and GDP per capita (constant 2005 US$) Ghura 2006). The various governments, also in line 2000 with the cold war, have moved back and forth from 1500 communism to a market economy. All these affected the economy and people’s perceptions, especially toward 1000 the importance of states as the main provider of jobs. 500 From 2002 to 2015, driven by higher oil prices and political stability, the Congolese economy grew 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 fairly strong. On average, during this period, the Republic of Congo’s growth rate stood at 4.5 percent Year (Figure 1.1). High oil prices translated into higher Source: World Development Indicators (WDI). revenue to both the Government as well as to the private sector related to the oil sector. This ultimately led to higher demand for goods and services, thereby This has implications for the distribution of income boosting the non-oil sector. Importantly, between 2005 and wealth, with urban areas faring much better than and 2011, the country grew strongly at an average rural areas. Gaps are widening. annual rate of 5.4 percent mainly driven by high oil As other resource-rich countries in Sub-Saharan revenue from oil production and the decision of the Africa (SSA), the economic growth is not generating Government to step up its investment in infrastruc- enough quality jobs. Because of the limited linkages ture in 2006. between the oil sector and the rest of the economy, recent However, while growth has been strong, it is economic growth failed to generate enough quality jobs. not sustainable as it is highly dependent on oil, The Republic of Congo is one of the most leading to volatility in gross domestic product urbanized countries in the world. This has many (GDP) growth over time and the risk of Dutch dis- implications on issues related to shared prosperity and ease.17 In 2013, oil accounted for almost two-thirds of poverty. Latest estimates suggest that 61.8 percent of GDP (63 percent) and contributed about 77 percent the population live in urban areas, with a bigger con- of government revenues and almost 90 percent of centration in the two metropolises (Brazzaville and merchandise exports. Manufacturing and agriculture Pointe Noire). A concentration in urban areas makes contribute a much smaller share of GDP despite large it easier to provide public service to the population. reserves of arable land (Figure 1.2 and Figure 1.3). This At the same time, there is a risk that the small share in oil dependency implies that the combined volatility of rural areas will be left out. This is exactly what is hap- oil production and prices can lead to significant fiscal pening in the Republic of Congo. planning issues and to a reduction in the quality of public spending. In the 1960s, the Republic of Congo was a diversified and relatively industrialized economy. 17 The Dutch disease is the apparent relationship between an increase Manufacturing accounted for about 25 percent of in the economic development of natural resources (or inflows of foreign aid) and a decline in the manufacturing or agriculture sector. GDP, dominated by the textile industry, concrete The mechanism at work is that an increase in revenues from natu- production, and agro-business. The emergence of the ral resources (or inflows of foreign aid) tends to lead to a stronger oil sector, coupled with socioeconomic problems and currency (as manifested through the exchange rate), resulting in a decline in the competitiveness of a nation’s other exports as well as civil wars, have contributed to the current undiversi- cheaper imports, making the national manufacturing and agriculture fied status of the economy and high oil dependency. sectors less competitive. 2 Republic of Congo – Poverty Assessment Report FIGURE 1.2: Sector Shares in GDP FIGURE 1.4: Population Pyramid for the Republic of Agriculture, Congo in 2011 4% 80–84 Services, 70–74 24% 60–64 50–54 40–44 Crude oil, 30–34 Industry, 63% 9% 20–24 10–14 0–4 300 200 100 0 100 200 300 Population (in thousand) Source: World Bank Economic Data, 2014. Male Female Source: Authors’ calculation using the 2011 Enquête Congolaise auprès des Ménages (ECOM) surveys. The Republic of Congo’s population is very young, about two persons out of five are below 15 years. The 2011 age pyramid reflects the typical pat- tern of a developing country—wide at the bottom and  ationale for updating the 1.3 R narrowing fast as one moves up the age distribution poverty and inequality figures (Figure 1.4). Such a situation presents a lot of challenges to the Government in terms of service provision, manag- The frequency of a household survey in the Republic ing congestion, economic growth, and planning for the of Congo is very low and does not allow for up-to-date next generation. The high proportion of youth will also assessment of poverty and living conditions of the pop- put pressure on the labor market in the medium run, as ulation. Unfortunately, the only two available household they will be need more jobs. A very young population also surveys for the Republic of Congo are the 2005 and 2011 results in the country having a high dependency ratio. ECOM surveys. The country has not been doing well in implementing its statistical strategy. All the analyses in this report are limited by data availability. Still, the avail- FIGURE 1.3: Sector Shares in GDP, 1960–2010 (%) able data can be put into contributing to understand past 100% drivers of poverty and their policy implications. 90% Initial analysis by World Bank staff provided 80% 70% a number of results that were surprising. First, the 60% overall decline in the headcount index of poverty at the 50% national level was relatively small, at less than 4 per- 40% 30% centage points over the six-year period. This decrease 20% is relatively small given the level of economic growth 10% observed in the country at the time. Second, the poverty 0% lines used for the estimates are such that the poverty line 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 in 2011 is 82 percent higher than the poverty line in Agriculture Forestry Extractive Industries 2005 (an increase from XAF 198,704 in 2005 to XAF Manufacturing Service 362,705 in 2011). Yet, over the same period, the cumu- Source: World Bank Economic Data, 2011. lative increase in the consumer price index (CPI) was A Substantial Reduction of Poverty between 2005 and 2011 3 substantially lower at 31.5 percent. Third, the estimates Poverty estimates in the Republic Box 1.1:  suggest a large decrease in inequality between the two of Congo are based on relatively survey years. For example, the estimates suggest that the old data Gini index of inequality may have been reduced from The ECOM survey is used to estimate official poverty figures 0.46 in 2005 to 0.38 in 2011, a large drop which would in the Republic of Congo. This survey was conducted in be surprising if poverty reduction benefitted more large 2005 and 2011.1 The ECOM is a nationally representative cities as opposed to other urban areas and rural areas. survey collected by the National Institute of Statistics (NIS). Finally, the analysis suggests that the share of household The ECOM employs a two-stage sampling framework and it covered approximately 5,000 households in 2005 and 10,400 consumption allocated to food increased substantially in 2011. In 2005, the sample design allowed for estimation from 38.6 percent in 2005 to 46.1 percent in 2011, of up to five strata: Brazzaville, Pointe Noire, other town, which would again be surprising if observed during a semi-urban, and rural. In 2011, the sample was substantially period of sustained economic growth. increased, allowing for estimations disaggregation up to the 12 departments level. The ECOM survey’s aim was to provide up- It turned out that the inconsistencies were to-date information on household incomes and expenditures, related to differences in methodology. In 2005, the determine households’ sources of income, and provide World Bank suggested a robust approach in estimating information on households’ demographic characteristics. poverty figures. This approach was not followed in con- The survey had a modular structure, collecting information on individual demographic characteristics, economic activity, ducting the analysis for 2011. Yet, when it comes to the education, health, and incomes. At the households’ level, the poverty and inequality dynamic, similar approaches must survey collected information on sources of income, all types be used to construct the welfare aggregate. In addition, of expenditures, possession of durable goods, and access to it is recommended not to update the poverty line in a basic services. The 2005 and 2011 ECOM surveys served as the base for our analysis. time span that is less than 10 years. Under the current poverty assessment, the 2011 1 The ECOM survey was supposed to be conducted every five years, but this objective has not materialized due to limited funding of the statistical poverty estimates where revisited to have robust com- apparatus. Statistical capacity in the Republic of Congo is weak for a parability between the two ECOM. A new welfare aggre- middle-income country. In 2015, the Republic of Congo’s overall Statistical Capacity Index (SCI) score was 50, which is well below the global, SSA, gate was constructed taking into account the exact same and International Bank for Reconstruction and Development (IBRD) assumptions as in 2005 (see Annexes 1 and 2). In addition, countries’ average scores of 66, 59, and 74, respectively (WDI 2016). The situation is expected to improve in the medium term with the support of the poverty basket of 2005 was kept and the poverty line the World Bank under the Statistical Capacity Building (PSTAT) project. of 2011 was derived by applying the CPI to the cost of the basket. During the exercise, it also transpired that the story on inequality is very sensitive to the treatment of outliers. the indicator of well-being.18 Annex 2 explains how the We did little adjustment of outliers as this often reflects the poverty line was estimated, and it mentions differences actual state of wealth of a part of the population. between the surveys that make it difficult to obtain fully To compute poverty measures, three ingredients comparable poverty estimates over time, even though are needed (see, for example, Coudouel, Hentschel, those differences are not too large. Annex 3 explains and Wodon 2002). First, one has to choose the relevant how the poverty measures were used in this study—the dimension and indicator of well-being. Second, one has headcount index of poverty, the poverty gap, and the to select a poverty line—that is a threshold below which squared poverty gap, are defined. a given household or individual will be classified as poor. Finally, one has to select a poverty measure—which is used for reporting for the population as a whole or for 18 It is important to note that in 2011, because of archiving issues, a population sub-group only. Annex 1 explains how regional prices were not available. Therefore, the fisher price index as estimated by Afristat was used as regional deflator, but experience the ECOM household surveys were used to estimate from other countries suggests that this will have negligible impact household consumption per equivalent adult used as on the storyline. 4 Republic of Congo – Poverty Assessment Report 1.4  Substantial poverty reduction TABLE 1.1: Trend in Poverty Measures, 2005–2011 between 2005 and 2011 Poverty Poverty Squared poverty headcount (%) gap (%) gap (%) As a consequence of the strong macroeconomic per- 2005 formance, the proportion of the population living Brazzaville 42.3 15.5 7.6 in poverty reduced substantially between 2005 and Pointe Noire 33.5 10.6 4.9 2011. The proportion of individuals living below the Other municipalities 58.4 24.2 12.9 national poverty line declined from 50.7 percent in Semi-urban 67.4 29.4 16.2 2005 to 40.9 percent in 2011 (Figure 1.5), a decrease Rural 64.8 25.7 13.0 of 9.8 percentage points which is in line with the GDP National 50.7 19.4 9.7 growth rate observed during that period. Overall, around 2011 143,000 people moved out of poverty. Changes in the Brazzaville 21.6 6.2 2.7 poverty gap and squared poverty gap follow similar Pointe Noire 20.3 4.8 1.8 patterns to those observed for the poverty headcount Other municipalities 52.8 18.1 8.4 (Table 1.1). Nationally, despite population growth Semi-urban 59.7 22.5 11.0 between the two years, the number of poor decreased Rural 69.4 30.2 16.4 to 1,658,000 in 2011, down from 1,801,000 in 2005. National 40.9 15.4 7.8 Similar to the national poverty rate, the inter- Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. national extreme poverty rate has also fallen sig- nificantly. National poverty measures are used for within-country poverty analysis and for deriving the international extreme poverty measures are generally economic policies for poverty eradication. However, the used. The most common international poverty line is national poverty figures are generally not comparable US$1.90 expressed in 2011 purchasing power parity across countries. To compare the Republic of Congo with (PPP) U.S. dollars. The share of the extremely poor by other countries in Africa and worldwide, the so-called international standards living below US$1.90 PPP a FIGURE 1.5: Headcount Poverty Rates from 2005 to 2011 1. Poverty and per capita GDP trends 2. National poverty rate by region 60 $5,600 80 50.7 50 50.2 $5,400 40.9 60 40 $5,200 Poverty Rate (%) GDP per capita 37.0 30 $5,000 40 20 $4,800 20 10 $4,600 0 $4,400 0 2005 2006 2007 2008 2009 2010 2011 Brazzaville Pointe Other Semi-urban Rural ROC Noire municipalities Year $1.9/day PPP National poverty line GDP per capita 2005 2011 Source: Authors’ calculation using WDI, the 2005 and 2011 ECOM surveys. A Substantial Reduction of Poverty between 2005 and 2011 5 FIGURE 1.6: Poverty Headcount (2011 PPP) Versus GNI Per Capita in a Cross-Country Setting 1. Poverty headcount at US$1.90 a day (2011 PPP) 2. Poverty headcount at US$3.10 a day (2011 PPP) 80 100 at $3.10 a day (2011 PPP) at $1.90 a day (2011 PPP) Poverty headcount ratio 60 80 Poverty headcount ratio ROC 2005 ROC 2005 40 ROC 2011 60 ROC 2011 20 40 0 20 –20 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. Note: GNI = Gross national income. day had declined in the Republic of Congo from 50.2 these estimates is that of a dual economy, with much percent in 2005 to 37.0 percent in 2011 (Figure 1.5). lower and faster decreasing poverty measures in urban Despite the improvement in living standards, areas, especially the two large cities, than in rural areas. the country is still underperforming given its poten- Chapter 2 will analyze in more detail the growing urban/ tial and status as a middle-income country. Cross- rural dichotomy. country regression estimates suggest that the level of international poverty in the Republic of Congo is still 1.5 Inequalities remain high much higher than in other comparable middle-income countries (Figure 1.6 and Figure 1.7). Countries with At the national level, inequality levels remain high.19 a similar level of economic development generally have There was a slight increase of the Gini coefficient, much lower poverty rates. The share of the popula- although not statistically significant (0.460 in 2005 and tion living on less than US$1.90 a day is higher in the 0.465 in 2011), suggesting a slight increase in inequal- Republic of Congo compared to other middle-income ity (Figure 1.8). This increase is also confirmed by other countries. Moreover, the international poverty rate measures as provided in Figure 1.8, and it does seem to (US$1.90–2011 PPP) in the Republic of Congo is quite be coherent with the fact that poverty decreased more close to the SSA average of 42.6 percent. in the largest cities than in other urban and rural areas. Most of the reduction in poverty was observed Using an alternative measure, per equivalent adult con- in the two largest cities of Brazzaville and Pointe sumption among the richest 10 percent of households Noire. In Brazzaville, the poverty headcount went down in the Republic of Congo was 17.2 times that of the by 20 percentage points from 42.3 in 2005 to 21.6 in poorest in 2005; it had increased to 20.0 times by 2011. 2011. Pointe Noire also experienced a sharp decrease in The trend across regions was mixed, inequality poverty (13 percentage points), down to 20.3 in 2011 reduces in the big cities and increases in rural areas. from 33.5 in 2011. By contrast, in rural areas, there Within geographic areas, the trends are more complex, was an increase in poverty. In rural areas, the poverty as shown in the inequality measures in Figure 1.8. In the headcount went up by 4.6 percentage points from 64.8 in 2005 to 69.4 in 2011. The story that emerges from 19 In this report, we are focusing on consumption inequality. 6 Republic of Congo – Poverty Assessment Report Madagascar Congo, Dem. Rep. Malawi Guinea-Bissau Zambia Rwanda Lesotho Togo Haiti Benin Sierra Leone Niger Low income Tanzania Bangladesh Sub-Saharan Africa (developing only) Fragile and conflict affected situations IDA only Chad Senegal Congo, Rep. Guinea Sao Tome and Principe Ethiopia Uganda IDA total Lao PDR IDA blend India South Asia Lower middle income Djibouti South Africa Indonesia $3.10 a day (2011 PPP) Honduras Vanuatu Nepal Low & middle income IDA & IBRD total Philippines World Middle income Guatemala Georgia China IBRD only FIGURE 1.7: International Poverty Rates US$1.90 and US$3.10 (2011 PPP) Pakistan $1.90 a day (2011 PPP) Bolivia East Asia & Pacific (developing only) Cambodia Colombia Latin America & Caribbean (developing only) Upper middle income Brazil Ecuador Peru El Salvador Vietnam Panama Mexico Dominican Republic Bhutan Paraguay Europe & Central Asia (developing only) Tunisia Bulgaria Kyrgyz Republic Armenia Sri Lanka Costa Rica Argentina Latvia Albania Estonia Chile Lithuania Mauritius Mongolia Turkey Slovak Republic Uruguay Jordan Serbia Moldova Iran, Islamic Rep. Hungary Thailand Russian Federation Source: WDI.. Czech Republic Kazakhstan 100 90 80 70 60 50 40 30 20 10 0 A Substantial Reduction of Poverty between 2005 and 2011 7 largest cities of Brazzaville and Pointe Noire, inequality  s a consequence of increase in 1.6 A appears to have decreased. The same is observed for other inequalities, prosperity was not municipalities. By contrast, in semi-urban areas and in shared between 2005 and 2011 rural areas, inequality may have increased. By international standards, inequalities are higher in the Republic of Congo. Cross-country com- At the national level, prosperity was not shared with parisons suggest that inequality is high in the Republic of the poorest segments of the population. Another Congo (Figure 1.8.2). The Republic of Congo is ranked way to look at growth and changes in the distribution among the most unequal societies based on WDI data. of welfare consists of using growth incidence curves Data on inequality in 105 countries are available in the which plot growth rates in consumption at various WDI beyond 2010. The Republic of Congo ranked 90 levels of welfare (Ravallion and Chen 2003). As shown out of 105 countries on the Gini. in Figure 1.9, growth incidence curves suggest that nequality in the Republic of Congo, 2005–2011 FIGURE 1.8: I 1. Trend in Inequality 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Gini coefficient Theil index (GE(a), a = 1) Mean Log Deviation Entropy index Half (Coeff.Var. squared) (GE(a), a = 0) (GE(a), a = –1) (GE(a), a = 2) 2005 2011 2. Gini inequality index for the Republic of Congo and other countries in the world 70 60 50 40 30 20 10 0 Ukraine Norway Slovak Republic Kazakhstan Sweden Romania Kyrgyz Republic Netherlands Montenegro Albania Iraq Serbia Austria Cambodia Luxembourg Cyprus Bangladesh Ireland United Kingdom Nepal France Taiwan, China Canada Sierra Leone Italy Australia Indonesia Tunisia Portugal Iran, Islamic Rep. Tanzania Sri Lanka Vietnam Georgia United States Russian Federation El Salvador Congo, Dem. Rep. Uganda Philippines Benin Djibouti Togo Ecuador Congo, Rep. Costa Rica Chile Rwanda Guatemala Honduras Lesotho Haiti Source: Authors’ calculation using WDI data, the 2005 and 2011 ECOM surveys. (continued on next page) 8 Republic of Congo – Poverty Assessment Report Inequality in the Republic of Congo, 2005–2011 (continued) FIGURE 1.8:  3. Absolute difference D10-D1 (XAF) 2,500,000 2,000,000 1,500,000 FCFA 1,000,000 500,000 0 Brazzaville Pointe Noire Other municipalities Semi-urban Rural Total 2005 2011 4. Relative difference D10/D1 (number) 25 20 15 Number 10 5 0 Brazzaville Pointe Noire Other municipalities Semi-urban Rural Total 2005 2011 Source: Authors’ calculation using WDI data, the 2005 and 2011 ECOM surveys. growth was not pro-poor nationally. The poorest actu- growth rates observed across the distribution by area, ally experienced a deterioration in their living standards focusing on the 10th, 40th, and 60th percentiles. according to consumption-based measures of poverty. Those in the middle of the distribution and a small mprovement was also 1.7 I share of the wealthiest households experienced the large substantial on nonmonetary positive growth. Prosperity was shared with the poor in Brazzaville, dimensions of well-being Pointe Noire, and other municipalities. By contrast, growth was not pro-poor in semi-urban and rural areas. Trend in nonmonetary dimensions can be used to The results provided by growth incidence curves by area assess the reliability of monetary trend. One way to reveal differences. Within each area, growth was pro-poor assess whether these revised estimates of the poverty in Brazzaville, Pointe Noire, and the other municipalities. trends seem reasonable is to rely on other indicators In other urban areas, growth was higher for the nonpoor. of well-being apart from household consumption, and In rural areas, the growth incidence lies below zero, sug- determine whether the trends observed for the con- gesting losses in welfare. Figure 1.10 provides the mean sumption-based poverty measures are similar to trends A Substantial Reduction of Poverty between 2005 and 2011 9 FIGURE 1.9: Growth Incidence Curves by Strata, 2005–2011 National Brazzaville 15 5 10 Annual growth rate, % Annual growth rate, % 3 5 1 –1 0 –3 –5 –5 –10 1 11 21 31 41 51 61 71 81 91 1 11 21 31 41 51 61 71 81 91 Expenditure percentile Expenditure percentile Pointe Noire Other Municipalities 15 8 13 6 11 Annual growth rate, % Annual growth rate, % 4 9 7 2 5 3 0 1 –2 –1 –4 –3 –5 –6 1 11 21 31 41 51 61 71 81 91 1 11 21 31 41 51 61 71 81 91 Expenditure percentile Expenditure percentile Semi–urban Rural 10 4 8 2 Annual growth rate, % Annual growth rate, % 6 0 4 –2 2 –4 0 –6 –2 –8 1 11 21 31 41 51 61 71 81 91 1 11 21 31 41 51 61 71 81 91 Expenditure percentile Expenditure percentile Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. 10 Republic of Congo – Poverty Assessment Report  ean of Growth Rates Across the Distribution, 2005–2011 FIGURE 1.10: M 10 8 6 4 Percentage 2 0 –2 –4 –6 National Brazzaville Pointe Noire Other municipalities Semi-urban Rural Bottom 10% Bottom 40% Top 60% Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. observed for other indicators of household well-being. TABLE 1.2: Share of the Population in Poverty, This is done in Table 1.2 by considering first subjective Nonmonetary Dimensions, 2005–2011 (%) perceptions of poverty and next assets-based poverty Subjective poverty Assets-based poverty measures (the methodology used for computing assets- 2005 2011 2005 2011 based poverty measures is explained in Annex 4).20 Brazzaville 42.6 31.2 50.5 21.3 Between 2005 and 2011, subjective poverty, Pointe Noire 29.6 23.0 43.1 20.2 measured by the inability of households to meet Other municipalities 41.0 35.0 71.6 52.7 their basic food needs, decreased by 10.7 percentage Semi-urban 54.3 34.7 82.4 59.4 points. Using comparable subjective perceptions data Rural 50.2 37.6 80.3 69.1 for the 2005 and 2011 surveys, Table 1.2 provides the Total 42.9 32.2 62.5 40.7 share of the population in households who tend to have Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. difficulties in meeting basic food needs. Nationally, sub- jective poverty measured in that way declined by 10.7 percentage points from 42.9 percent of the population households possess additional assets, this can quickly in 2005 to 32.2 percent in 2011. One difference though lead to a rapid reduction in the measures of assets-based with the consumption-based measures of poverty is that poverty. Importantly though, the reduction in assets- the decline in subjective perceptions of poverty among based poverty is again observed for all areas, including households was observed for all areas, including for rural areas (on the methodology for the construction of rural areas. assets-based poverty measures, see Annex 4; the meth- Poverty in terms of assets ownership also odology is such that by construction assets-based and decreased substantially. Ownership of modern assets consumption-based headcount poverty measures in 2011 increased while ownership of traditional assets deterio- are very close to each other). rated. There was a notable increase in the proportion of households who own a mobile phone, a TV set, a modern 20 Such an approach is commonly used and its robustness has been chair, or an iron (Figure 1.11). As a consequence, the well documented in the literature; see, for instance, Sahn and Stifel magnitude of the reduction in the share of the popula- (2000, 2003). 21 A factorial analysis is conducted for possible assets and the poverty tion living in households considered assets-poor is even line is set to be as close as possible to the 2011 monetary estimate. larger,21 but this is explained in part by the fact that when Projecting this to 2005 allowed to explore the trend. A Substantial Reduction of Poverty between 2005 and 2011 11 FIGURE 1.11: Asset Ownership, 2005–2011 (%) (0.40), a high pass-through (0.8) and double digit growth Mattress or bed rate (10%) between 2021 and 2030. The pass-through Phone measures how macroeconomic performances translate into Radio actual improvement of living standards of the population. Television An increase of the pass-through reflects better linkages. Modern chair Microsimulation results suggest that the pov- Iron Bicycle erty decline had continued beyond 2011, although Canoe at a slower pace. It is estimated that the slowdown Computer of economic growth linked to the oil sector led to a Motorcycle slower reduction in poverty after 2011. Between 2005 Car or Truck and 2011, poverty declined by 1.63 percentage points 0 20 40 60 80 100 annually. In 2016, due to slower economic growth, it is 2005 2011 estimated that the share of the population living below Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. the national poverty line is around 34 or 35 percent. This corresponds to a poverty decline by 1.52 percentage points annually between 2011 and 2016. 1.8  Outlook toward ending extreme Projection up to 2030 shows that, unless eco- poverty by 2030 nomic performance and inequality improve substan- tially, it will be difficult but not impossible for the Combining information on actual and projected per country to reach the goal of eradicating extreme pov- capita GDP growth, it is possible to simulate the pov- erty by 2030. Under the more than optimistic scenario 7, erty level beyond 2011. In the literature, there are many where the country is assumed to post double-digit growth ways of conducting such a projection. In this chapter, we between 2021 and 2030, the international US$1.90 pov- follow Ravallion (2004)22 who expressed the growth of the erty rate will be 3.6 percent in 2030 (Figure 1.12). The poverty headcount capita GDP as a function of the GDP more realistic scenario 2 or 3 shows that the international per capita growth and the Gini inequality index. US$1.90 poverty rate will be around 15 percent by 2030. A total of seven simulations were conducted, with Market mechanism will not be enough for the assumption on per capita GDP growth, pass-through, country to achieve the goal of eradicating extreme and Gini inequality index. The first scenario (the base- poverty by 2030. Conditional or unconditional cash line) assumes a pass-through of 0.7, a constant Gini of transfer to the poor is something to consider. Direct 0.465, and the baseline scenario of actual and projected cash transfer to poor households is often considered as economic growth as estimated by World Bank staff. The a mechanism to eradicate poverty.23 Such programs can second scenario is similar to scenario 1, with the only dif- ference that GDP per capita growth is assumed to reach and remain stable at 5 percent between 2021 and 2030. 22 Ravallion (2004) demonstrates that the following equation ex- presses the growth of the poverty headcount capita GDP as a function The third scenario is a reproduction of scenario 1, but this of the GDP per capita growth and the Gini inequality index: time we assume a reduction in inequality, with the Gini going down to 0.40 from 0.465. The fourth scenario is D FGT 0 3 D GDPcap , = −9.3 ∗ (1 − Gini ) ∗ FGT 0 GDPcap a replication of scenario 2, but with a reduction of Gini to 0.40. Scenarios 5 and 6 replicated scenarios 1 and 2, where FGT0 stands for poverty headcount ratio and GDPcap for respectively, with the only difference that there is improve- per capita GDP. 23 There is a large volume of literature on conditional and unconditio- ment in the pass-through from 0.7 to 0.8. The seventh nal cash transfer. Relevant examples include Baird et al. (2011, 2012); scenario is the most optimistic, assuming a lower Gini Haushofer and Shapiro (2014); Stampini and Tornarolli (2012). 12 Republic of Congo – Poverty Assessment Report  ost of a Hypothetical Perfectly Targeted Cash Transfer Program That Will Eliminate Poverty TABLE 1.3: C Population share Transfer per poor Total transfer Share in total Population (%) Poverty gap (XAF) (XAF, billions) transfer (%) Region Brazzaville 1,502,793 37.1 6.2 78858 25.6 15.0 Pointe Noire 777,740 19.2 4.8 65006 10.2 6.0 Other municipalities 232,633 5.7 18.1 94066 11.6 6.7 Semi-urban 169,814 4.2 22.5 103485 10.5 6.1 Rural 1,370,266 33.8 30.2 119114 113.3 66.2 Total 4,052,841 100.0 15.4 103260 171.2 100.0 Source: Authors’ calculation using the 2011 ECOM surveys. Note: Amount in XAF 2011 prices. be categorized into two types: conditional cash transfer in 2011 prices, representing about 10 percent of and unconditional cash transfer. Under conditional cash the government budget. In 2011 prices, the national transfer programs, beneficiaries receive the transfer only poverty line was estimated at XAF 274,113 per year if they meet certain criteria. These criteria may include and per equivalent adult. On average, the distance of enrolling children in school, making sure the children the poor from the national poverty line is 15.4 per- receive vaccinations, and others. Leakage, operating cost, cent. This indicates that if it was possible to perfectly and sustainability are some of the challenges faced by target the poor, it will take an average annual payment such programs. Some of the success stories in SSA are of XAF 103,260 to eliminate poverty in the Republic documented by Garcia and Moore (2012). of Congo (Table 1.4).24 Overall, the budget for such a The financial cost for a hypothetical cash trans- cash transfer program is estimated at XAF 171.2 bil- fer could be quite high, about XAF 171.2 billion lion in 2011 prices. This represents slightly less than 10 percent of the government budget and about 14 percent of annual oil revenues. Most of the poor live FIGURE 1.12: Simulated Trend in International in rural areas, and the distance to the poverty line is (US$1.9 Dollar a Day) Poverty higher in rural areas. As a consequence, two-thirds of Headcount a hypothetical cash transfer to the poor should go to 40 those living in rural areas. 35 Despite the high cost, such a program if linked 30 to some conditionality could represent investment in 25 19.7 human capital that will result in important medium- 20 15.7 and long-term benefits for the country. In the case of 15 15.1 13.2 10.9 conditional cash transfers, for example, the likelihood 10 9.1 of a household using the windfall in building human 5 3.6 capital will be higher. This will, if all other conditions 0 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6 Scenario 7 24 Of course this just a hypothetical assumption, as there are many other factors at play, including education that takes time and the Source: Authors’ calculation using macroeconomics and fiscal management likelihood of some poor spending the allocation on useless items (MFM) projections and the 2011 ECOM survey. such as alcohol and tobacco. A Substantial Reduction of Poverty between 2005 and 2011 13 TABLE 1.4: Simulated Trend in Poverty Headcount Importantly, between 2005 and 2011, the country grew 2011 2015 2020 2025 2030 strongly at an average annual rate of 5.4 percent mainly National poverty line driven by high oil revenue from oil production and the decision of the Government to step up its investment Scenario 1 40.9 34.5 28.9 25.7 21.7 in infrastructure in 2006. Scenario 2 40.9 34.5 28.9 22.4 17.3 The strong growth performance was accompa- Scenario 3 40.9 32.2 25.0 21.2 16.7 nied by substantial improvements in living standards. Scenario 4 40.9 32.2 25.0 17.4 12.1 The proportion of the population living in poverty Scenario 5 40.9 31.0 23.2 19.2 14.6 reduced substantially between 2005 and 2011. The Scenario 6 40.9 31.0 23.2 15.3 10.1 proportion of individuals living below the national pov- Scenario 7 40.9 31.0 23.2 9.7 4.0 erty line declined from 50.7 percent in 2005 to 40.9 per- US$1.90 PPP cent in 2011, a decrease of 9.8 percentage points which Scenario 1 37.0 31.2 26.1 23.2 19.7 is in line with the GDP growth rate observed during that Scenario 2 37.0 31.2 26.1 20.2 15.7 period. Overall, around 143,000 people moved out of Scenario 3 37.0 29.1 22.6 19.1 15.1 poverty. Changes in the poverty gap and squared poverty Scenario 4 37.0 29.1 22.6 15.7 10.9 gap follow similar patterns to those observed for the pov- Scenario 5 37.0 28.1 21.0 17.4 13.2 erty headcount. Nationally, despite population growth Scenario 6 37.0 28.1 21.0 13.8 9.1 between the two years, the number of poor decreased Scenario 7 37.0 28.1 21.0 8.7 3.6 to 1,658,000 in 2011, down from 1,801,000 in 2005. Source: Authors’ calculation using the 2011 ECOM surveys. Similar to the national poverty rate, the interna- tional extreme poverty rate has also fallen significantly. National poverty measures are used for within-country are also in place,25 result in a better educated, nourished, poverty analysis and for deriving the economic policies and healthy population. for poverty eradication. However, the national poverty figures are generally not comparable across countries. To 1.9 Conclusion compare the Republic of Congo with other countries in Africa and worldwide, the so-called international This chapter has documented the poverty trends extreme poverty measures are generally used. The most between 2005 and 2011. The poverty estimates are common international poverty line is US$1.90 expressed robust despite some caveats related to surveys com- in 2011 PPP U.S. dollars. The share of the extremely parability. Poverty estimates for 2011 were revisited to poor the international standards living below US$1.90 produce figures based on a similar methodology that was PPP a day had declined in the Republic of Congo from used by the World Bank in 2005. Perfect comparability 50.2 percent in 2005 to 37.0 percent in 2011. was not possible, especially regarding the special price Despite the improvement in living standards, the deflator which could not be recomputed as the raw data country is still underperforming given its potential sets were not properly archived. Nevertheless, based on and status as a middle-income country. Cross-country countrywide experiences, current estimates are reliable regression estimates suggest that the level of international as possible bias from spatial deflators will not change poverty in the Republic of Congo is still much higher the main storyline. than in other comparable middle-income countries. The From 2002 to 2015, driven by higher oil price share of the population living on less than US$1.90 a day and political stability, the Congolese economy grew fairly strong. On average, during this period, the Republic of Congo’s growth rate stood at 4.5 percent. 25 Such as quality of service delivery. 14 Republic of Congo – Poverty Assessment Report is higher in the Republic of Congo compared to other nationally. The poorest actually experienced a deteriora- middle-income countries. Moreover, the international tion in their living standards according to consumption- poverty rate (US$1.90 – 2011 PPP) in the Republic of based measures of poverty. Those in the middle of the Congo is quite close to the SSA average of 42.6 percent. distribution and a small share of the wealthiest house- By international standards, inequalities are holds experienced the large positive growth. higher in the Republic of Congo. Cross-country com- Projection up to 2030 shows that, unless eco- parisons suggest that inequality is high in the Republic nomic performance and inequality improve substan- of Congo. The Republic of Congo is ranked among the tially, it will be difficult but not impossible for the most unequal societies based on WDI data. Data on country to reach the goals of eradicating extreme inequality in 105 countries are available in the WDI poverty by 2030. Under the more than optimistic beyond 2010. The Republic of Congo ranked 90 out of scenario, where the country is assumed to post double- 105 countries on the Gini. digit growth between 2021 and 2030, the international At the national level, prosperity was not shared US$1.90 poverty rate will be 3.6 percent in 2030. More with the poorest segments of the population. Growth realistic scenarios show that the international US$1.90 incidence curves suggest that growth was not pro-poor poverty rate will be around 15 percent by 2030. A Substantial Reduction of Poverty between 2005 and 2011 15 The Growing Urban-Rural Dichotomy and Vulnerability are of Concern 2 2.1 Introduction Despite the substantial poverty reduction described in Chapter 1, many challenges remain, including poverty pockets in the main cities, a growing urban-rural dichotomy, and vul- nerability. At the national level, poverty indices went down substantially, but this performance was concentrated in Brazzaville and Pointe Noire. Monetary poverty increases in rural areas and poverty pockets in big cities remain. Moreover, some groups of the population are lagging behind. These include autochthons, youth, elderly, women, and disabled. This chapter will illustrate some of the apparent drawbacks of the current trend in pov- erty reduction. The chapter has five parts. In the first part, we will analyze the growing urban/ rural dichotomy. The second part describes migration flow and possible implications for welfare. In the third part, we will look at the pockets of poverty in the two main cities. In the fourth part, we will describe the vulnerability from a monetary poverty point of view. The last part will look at vulnerable socioeconomic groups. 2.2 A growing urban-rural dichotomy As pointed out in the previous chapter, poverty is becoming an increasingly rural phenom- enon, and this should be a much larger reason for concern. Most of the poverty reduction was observed in the two largest cities of Brazzaville and Pointe Noire. What was most worrisome was an increase in poverty headcount as well as in the number of poor people in rural areas. In rural areas, the poverty headcount went up by 4.6 percentage points from 64.8 in 2005 to 69.4 in 2011. In rural areas, the depth and severity of poverty has also increased, that is, the rural poor have become poorer. Between 2005 and 2011, the number of poor in rural areas increased to 951,000 up from 795,000. The story that emerges from these estimates is that of a dual economy. As it will be illustrated in Chapter 4, difference across the main cities, the remaining urban areas, and the rural areas account for a big part of the overall inequality. Over time, rural areas account for a lower share of the population but a larger share of the poor. Figure 2.1 shows the distribution of the population by ventile of welfare and location (each ventile accounts for 5 percent of the 17 FIGURE 2.1: Distribution of the Population by Ventile and Location (%) 2005 2011 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Welfare ventiles Welfare ventiles Brazzaville Pointe Noire Other municipalities Semi-urban Rural Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. national population, from the poorest to the richest). increases in rural areas is the fact that welfare decreased Clearly, rural areas account for a larger share of lower for those living in rural areas. Between 2005 and 2011, socioeconomic groups in 2011 than in 2005. This is welfare of those living in rural areas decreased at a pace shown in Figure 2.2 in terms of changes in the contri- of −0.3 annually (Table 2.1). In the meantime, those bution of various areas to the headcount of poverty at in urban areas, especially the two big cities, benefitted the national level. While rural areas accounted for 44.3 from an important increase in welfare. As a conse- percent of the poor in 2005, this had increased to 57.4 quence of this, the monetary gap between those living percent by 2011. in urban areas and those in rural areas is widening. The monetary gap between those living in In 2005, welfare was 71 percent higher in Brazzaville urban areas and those in rural areas is widening. compared to rural areas. In 2011, the gap widened to Those in urban areas experienced a substantial growth reach 111 percent. A similar path is observed when in their welfare, while for those in rural areas, there comparing rural areas with Pointe Noire and other was a welfare decrease. The major reason for poverty municipalities. FIGURE 2.2: Contributions to the Poverty Headcount by Location (%) 2005 2011 Brazzaville, Rural, 44% Brazzaville, Rural, 57% 20% 24% Pointe Noire, 10% Pointe Noire, 15% Other municipalities, 7% Other municipalities, Semi-urban, 9% 7% Semi-urban, 6% Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. 18 Republic of Congo – Poverty Assessment Report The urban-rural gap is very important in non- Change in Annual Consumption Per TABLE 2.1:  monetary dimensions of well-being as well. For Equivalent Adult by Location example, the availability of social services and modern Welfare Welfare Growth rate Annual growth infrastructures is very limited in rural areas. As shown 2005 2011 period (%) rate (%) in Table 2.2, the availability of electricity, piped water, Brazzaville 518,608 628,834 21.3 3.5 and other services is minimal in rural areas. For example, Pointe Noire 529,728 627,618 18.5 3.1 only one in ten households in rural areas states that the Other 330,720 348,799 5.5 0.9 electricity network is available in their community. The municipalities corresponding figure for Pointe Noire and Brazzaville are Semi-urban 273,216 376,175 37.7 6.3 way higher (nine out of ten for Pointe Noire and close to Rural 302,952 297,642 −1.8 −0.3 seven out of ten for Brazzaville). Similar urban-rural gaps Total 420,198 492,548 17.2 2.9 are observed for many other social services, including Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. availability of electricity, piped water, a market, public Note: Welfare in 2011 XAF. TABLE 2.2: Availability of Services, Share of Households for Whom the Service is Availables Within 1 Km, 2011 Residence area Other Brazzaville Pointe Noire municipalities Semi-urban Rural Total Sanitation (collection and treatment of waste) 36.9 31.5 19.7 0.6 3.0 21.7 Primary school 81.6 88.8 88.0 69.0 71.7 79.3 Middle school 61.3 74.4 65.7 47.6 28.1 51.8 High school 32.6 30.0 32.1 23.7 4.6 22.0 Covered market 52.1 51.9 53.5 38.7 25.2 42.2 Open-air market 73.3 61.5 77.1 51.4 26.7 54.0 Fixed phone 24.7 18.9 7.1 14.5 3.0 14.7 Mobile phone 79.1 95.8 87.9 83.7 55.7 74.6 Electricity (SNE) 65.5 91.6 67.3 69.5 9.8 51.1 Other electricity supply 24.7 42.5 14.4 22.4 18.0 25.1 Drinking water supply 35.4 52.3 47.4 47.7 25.7 36.3 Piped water (SNDE) 66.1 88.9 75.6 46.7 6.0 49.0 Other sources of water supply 32.4 68.0 38.2 45.0 65.5 51.5 Registry office 28.4 17.4 49.3 41.0 22.5 25.9 Health center/hospital 63.5 74.3 71.6 56.1 39.1 57.1 Police station 58.4 30.4 63.6 48.5 24.7 41.2 Justice 13.2 13.6 27.0 34.0 11.6 14.3 Public transport 72.9 94.4 71.6 50.2 23.7 58.7 Long-distance transport 25.0 16.7 36.5 51.1 45.6 32.3 Source: Authors’ calculation using the 2011 ECOM survey. Note: SNDE = Société Nationale de Distribution d’Eau; SNE = Société Nationale d’Electricité. The Growing Urban-Rural Dichotomy and Vulnerability are of Concern 19 transport, and a health center. As a consequence, living a year (IRIN 2012). Government projects aiming at conditions are not attractive in rural areas. Chapter 6 achieving import substitution have failed due to poor will provide a deeper analysis on access to social services. planning and poor governance (IRIN 2012; World Bank Low agricultural production and lack of link- 2016a). New Agricultural Village and the partnership ages between rural activities (agriculture) and urban with South African farmers are examples of such failures. activities (manufacture and services) seem to be To limit imports and ensure food security, the Republic important reasons driving monetary poverty in rural of Congo launched in 2010 a US$26 million project to areas. As pointed out in Chapter 1, the Republic of build ‘New Agricultural Villages’. With this project, the Congo economy is not diversified. The expansion of country only managed to halve the import bill for eggs. the oil sector was accompanied by the negative effects They produced 6.6 million eggs in 2011, while imports of the Dutch disease. As a consequence, the agriculture are estimated at 13 million eggs per year (IRIN 2012). and manufacturing sectors contracted. Agriculture has In 2011, the Republic of Congo leased 180,000 ha of become less competitive and now the Republic of Congo arable land to a group of South African farmers who relies heavily on importation to satisfy the country’s food have managed to plant 1,200 ha of maize. needs. Safoulanitou and Ndinga (2010) show that the Republic of Congo’s massive food imports represented  igration as a channel toward 2.3 M a strategy not only for ensuring the country’s food secu- better living conditions rity, but also for supplying food products to neighboring countries, notably the Democratic Republic of Congo The welfare gap between urban and rural areas is a and Angola. They further show that repeated armed con- factor driving migration in the Republic of Congo. flict and Dutch diseases have lasting important negative Those in rural areas are more likely to migrate. Because consequences on agricultural production. of the difficult conditions in rural areas, individuals With regard to productivity of most of the main tend to leave their households for greener pastures in crops, the country is among the least performers in urban areas, especially toward the two main cities. An the world. Figure 2.3 provides data on yield for the analysis of the correlates of the likelihood of migration main crops (cassava, banana, maize, bean, groundnut, shows that the likelihood of a household sending a and yam). For maize, bean, and groundnuts, the country member is higher for rural households only (Table 2.3 exhibits one of the lowest yields in the world. For example, and Figure 2.4.1). In addition, for households who Congolese farmers produce 10.5 tons of maize per ha, move entirely, Pointe Noire seems to be the main while neighboring Gabon and Cameroon outperform destination. Recall that Pointe Noire is doing better with a productivity of 16.6 tons and 19.5 tons per ha, than Brazzaville with regard to living conditions. The respectively. A similar situation is observed for beans, likelihood of migration increases with poverty rate, the yield is 7.6 tons per ha, while farmers in Cameroon population density, and the distance to the two main produce close to double with 14.5 tons of beans per ha. cities (Brazzaville/Pointe Noire). One can deduct from A higher agricultural production and better such results that the national wealth is concentrated in commercialization has the potential of boosting the these cities, and the farther from them, the less is the income of the rural population. However, government likelihood of benefitting from the economic growth. projects aiming to achieve this goal have not succeeded People are less likely to migrate if their household owns due to corruption and poor management. The Republic a piece of land. Migration also appears to be a coping of Congo is blessed with good weather and immense mechanism in case of a shock. Those who experienced arable land. Agriculture is still not benefitting from such a shock are more likely to migrate. potential. It is estimated that every year, the Republic Issues related to work, education, housing, and of Congo imports over US$240 million worth of food health are the main factors driving migration. When 20 Republic of Congo – Poverty Assessment Report 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 400,000 0 20,000 40,000 60,000 80,000 100,000 120,000 140,000 160,000 180,000 200,000 2. Banana 1. Cassava Cayman Islands Burkina Faso Timor-Leste South Sudan Solomon Islands Saint Lucia Central African Republic Trinidad and Tobago Equatorial Guinea Montserrat Haiti Réunion Gambia Zambia Togo Equatorial Guinea Zambia Guinea Guinea-Bissau French Guiana Fiji Antigua and Barbuda United Republic of Tanzania Mozambique Swaziland Comoros Congo, Rep. Madagascar French Polynesia Honduras Burundi Mozambique Dominica Wallis and Futuna Islands Nicaragua Ethiopia Puerto Rico Bolivia (Plurinational State of) Congo, Rep. Saint Vincent and the Grenadines Bolivia (Plurinational State of) ditions also registered important scores. These development Senegal who migrate. Issues related to education and housing con- work are by far the main reasons (Figure 2.4.2). Access to asked why the member left the household, issues related to quality jobs appears then to be the challenge faced by those Bahrain Guinea Cuba Cabo Verde Guinea-Bissau Grenada Burundi Colombia Nepal Chad Bermuda Guyana Seychelles United Arab Emirates Papua New Guinea Philippines Philippines Argentina Peru Puerto Rico Sri Lanka Senegal Kenya Réunion Kenya Mali Malawi Cameroon Maldives Mauritius Cyprus Malaysia Panama Belize El Salvador FIGURE 2.3: Yield of Main Crops Generally Below Median Level among Comparators, 2014 Occupied Palestinian Territory Brunei Darussalam alleviation in the Republic of Congo. Peru Paraguay Morocco French Polynesia Martinique Niger Suriname Ghana Sao Tome and Principe Thailand Indonesia Egypt The Growing Urban-Rural Dichotomy and Vulnerability are of Concern Cook Islands Costa Rica Lao People's Democratic Republic Indonesia Suriname 21 (continued on next page) education and quality jobs, are very important for poverty and the reduction of the growing urban-rural gap. As we concerns should be at the center of the fight against poverty will see in the coming chapters, all these factors, especially 22 –10,000 10,000 30,000 50,000 70,000 90,000 110,000 130,000 150,000 0 10,000 15,000 20,000 25,000 30,000 3. Maize 4. Beans Eritrea Cabo Verde Swaziland Sao Tome and Principe Morocco India Mozambique Ecuador Haiti Niger Eritrea Kenya French Guiana Congo, Rep. Uruguay Jamaica Thailand Senegal Mexico Chad Guinea Congo, Rep. Dominican Republic Costa Rica United Republic of Tanzania Bangladesh Gabon Paraguay Central African Republic Burkina Faso Haiti Côte d'Ivoire Australia Djibouti Guatemala Republic of Congo – Poverty Assessment Report Romania Georgia Rwanda Suriname El Salvador Panama Jamaica Guam Such migration flow creates more problems in rity. Many who move to cities are often not equipped cities, including expansion of slums, congestion on social services, increased unemployment, and insecu- Viet Nam El Salvador Syrian Arab Republic Serbia Zambia China, mainland Lebanon Indonesia Belize Ethiopia Cambodia Sri Lanka Ghana Maldives Mauritania Thailand Timor-Leste Iraq The former Yugoslav Republic of Macedonia Sweden Bosnia and Herzegovina Lao People's Democratic Republic Montenegro Ukraine Kazakhstan Azerbaijan China, Taiwan Province of China, mainland France Kyrgyzstan United States of America Ukraine Cyprus Argentina Armenia Italy Australia Hungary Egypt Turkey Croatia Portugal Yemen Mauritius Armenia Réunion Luxembourg Chile Germany FIGURE 2.3: Yield of Main Crops Generally Below Median Level among Comparators, 2014 (continued) Netherlands New Zealand Tajikistan Tajikistan Barbados Jordan Kuwait that are of poor quality and difficulties with regard conditions. This includes living in slums with shelters with the right skills and assets for a smooth transition- (continued on next page) insertion, thus ending up experiencing vulnerable living FIGURE 2.3: Yield of Main Crops Generally Below Median Level among Comparators, 2014 (continued) 5. Groundnut 50,000 45,000 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5,000 0 Swaziland Namibia Uruguay South Sudan Madagascar Guyana Congo, Rep. Uganda Senegal Eritrea Sudan Georgia Burkina Faso Gambia Gabon Cuba Colombia Timor-Leste Nigeria Dominican Republic Comoros Guinea Cambodia Guatemala Belize Mali Thailand Sri Lanka Kyrgyzstan Indonesia Morocco Iran (Islamic Republic of) Syrian Arab Republic China, Taiwan Province of Tajikistan China Iraq Malaysia Nicaragua Cyprus 6. Yam 250,000 200,000 150,000 100,000 50,000 0 Fiji Niue Martinique Cayman Islands Grenada Sudan South Sudan Antigua and Barbuda Panama American Samoa Saint Lucia New Caledonia Congo, Rep. Congo Cuba Samoa Trinidad and Tobago Philippines Guyana Sao Tome and Principe Comoros Mauritania Gabon United Republic of Tanzania Dominican Republic Burkina Faso Côte d'Ivoire Togo Central African Republic Nigeria Wallis and Futuna Islands Liberia Venezuela (Bolivarian Republic of) Chad Brazil Burundi Saint Kitts and Nevis Solomon Islands Barbados Cameroon Puerto Rico Guinea Colombia Belize Dominica Costa Rica Haiti Tonga Saint Vincent and the Grenadines Benin Rwanda Kenya Ghana Jamaica Portugal Papua New Guinea Guadeloupe Japan Mali Ethiopia Source: The Statistics Division of Food and Agriculture Organization of the United Nations (FAO-STAT). Note: Banana data are for 2013. to access to sanitation. For example, one out of three  ockets of poverty in Brazzaville 2.4 P households in Brazzaville or Pointe Noire doesn’t have and Pointe Noire a safe toilet. As a consequence, for the new migrant, the living conditions in the city can be worse because There are pockets of poverty in the two main cities. of the lack of or limited formal safety net. In rural areas, In Brazzaville, the poor are concentrated in Makélékélé a traditional safety net mechanism can help them cope and in Pointe Noire, they are concentrated in Tietie and with difficulties. Loandjili. Currently, there is not enough information to The Growing Urban-Rural Dichotomy and Vulnerability are of Concern 23 ncidence of Migration and Reasons for Migrating, 2011 FIGURE 2.4: I 1. A member of the household left 2. Reason for leaving 20 Work (assignment) 18 Studies 16 Job search 14 Housing problem 12 10 Health condition 8 Acquisition of own housing 6 Return of peace 4 Other 2 Insecurity problem 0 Brazzaville Pointe Other Semi-urban Rural 0 5 10 15 20 25 Noire municipalities Source: Authors’ calculation using the 2011 ECOM survey. TABLE 2.3: Correlated on Migration, 2011 (PEEDU). This survey was representative up to the A member of the ‘arrondissement’ level. The estimates (Table 2.4) suggest household left definitively that for Brazzaville, Makélékélé was the arrondissement Coef. t which contributed the most to poverty (36 percent of Household experienced a shock 0.268*** 0.036 the poor). In Brazzaville, Talaingai, Mfilou, and Ouenze Region also presented double-digit contribution to poverty. For Brazzaville Ref. Ref. Brazzaville, 80 percent of the poor live in those four arrondissements. Pointe Noire also exhibited a concen- Pointe Noire −0.083 0.101 tration of poor in specific arrondissements. In Pointe Other municipalities 0.252 0.186 Noire, around 78 percent of the poor live in Loandjili Semi-urban 0.140 0.215 and Tietie—46.7 and 32.1 percent, respectively. Rural 0.422** 0.211 Household own land −0.089** 0.036 2.5 Growing monetary vulnerability Population density (province) 0.104*** 0.032 Distance to Brazzaville/Pointe Noire 0.055** 0.028 Success in reducing poverty has resulted in many Province adjacent to Brazzaville/Pointe 0.134 0.082 households that are living just above the poverty line Noire who remain vulnerable to falling below the poverty Poverty headcount (province) 0.009** 0.004 line in the face of a negative shock. A way to look Constant −3.236*** 0.404 at vulnerability is to classify the population into three Number of observations 10,307 groups: the poor (consumption below the poverty line), Source: Authors’ calculation using the 2011 ECOM survey. Note: See Statistical appendix (Table SA.1) for the detailed regressions. *** p<0.01, ** p<0.05, * p<0.1 26 There are two possible ways of building a poverty map: (a) the parametric approach, which combines household consumption survey and census data and (b) the Bayesian approach that builds build a poverty map for the Republic of Congo.26 Luckily, on global positioning system (GPS) coordinates of households and the World Bank conducted a survey in Brazzaville and social infrastructures. Unfortunately, the country lost track of the census data set due to limited capacity in archiving. On the other Pointe Noire in 2008. This was the baseline survey hand, the survey did not collect data on GPS, thus it is impossible for the Projet Eau, Electricité et Développement Urbain to use the Bayesian approach to build a poverty map. 24 Republic of Congo – Poverty Assessment Report TABLE 2.4: Poverty Profile for Brazzaville and Pointe Noire Using a World Bank Survey, 2008 Poverty headcount Poverty gap Squared poverty gap Contribution to poverty (percent) (percent) (percent) (percent) Brazzaville Makélékélé 48.0 16.4 26.2 36.7 Bacongo 33.1 7.6 8.5 8.2 Poto-Poto 25.4 6.8 8.1 6.0 Moungali 19.9 5.7 9.8 5.7 Ouenze 28.1 8.9 13.4 11.0 Talaingai 27.1 7.4 23.1 18.2 Mfilou 44.3 16.4 10.9 14.1 Total 34.2 10.7 100.0 100.0   Pointe Noire Lumumba 29.6 10.4 10.8 8.9 Mvou-Mvou 42.4 11.2 10.4 12.3 Tietie 36.2 12.6 31.9 32.1 Loandjili 35.9 10.8 46.9 46.7 Total 36.0 11.4 100.0 100.0 Source: World Bank 2008; PEEDU baseline survey. the insecure nonpoor or vulnerable (consumption above The poor are more likely to experience nega- the poverty line but below twice the poverty line), and tive shocks. Another way to look at the issue of vul- the middle class (consumption above twice the poverty nerability is to rely on the information available in the line; this would include some wealthy households as well, 2011 ECOM survey on household exposure to nega- but most households could be considered middle class tive shocks. Questions are asked about the following for a country like the Republic of Congo). Figure 2.5 shocks: death in the household, serious illness, loss of shows that as poverty receded, both the vulnerable employment, bankruptcy of a business, floods or other (insecure nonpoor) and middle class groups expanded, disasters, loss of crops, and loss of livestock. As shown with the middle class group expanding faster than the in Figure 2.6.1, the most common shock is a serious vulnerable/insecure nonpoor group. While only 20.6 illness (one in five households is affected), followed by percent of the population had consumption more than bankruptcy of a business, a death in the household, and twice the poverty line in 2005, this increased to 26.3 a loss of crops. Some households are also affected by the percent by 2011. In the meantime, 32.8 percent of the other types of losses. The poor are especially vulnerable population, while technically ‘nonpoor’, is consuming at (with the exception of bankruptcies, since the poor often a level below an average XAF 550,000 yearly (in nomi- do not have businesses). Half of those in the bottom nal 2011 CFA francs) per equivalent adult. Given that quintile are affected by at least one of the shocks. a large share of the vulnerable rely on either agriculture The ability of households, and especially the poor, or informal activities which are susceptible to significant to cope with these various shocks seems to be limited. output volatility, many remain at serious risk of falling Households who were affected by shocks were asked in back into poverty, at least on a temporary basis. the 2011 survey whether they received help provided by The Growing Urban-Rural Dichotomy and Vulnerability are of Concern 25 FIGURE 2.5: Poverty, Insecure Nonpoor, and Middle Class Status by Year and Location 1. By year 2. By residence area, 2011 only 100% 100% 20.6 26.3 90% 80% 80% 70% 28.7 60% 32.8 60% 50% 40% 40% 50.7 30% 20% 40.9 20% 10% 0% 0% 2005 2011 Brazzaville Pointe Other Semi-urban Rural National Noire municipalities Poor Insecure nonpoor Middle class Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. the Government; whether they sold livestock, capital, or  he most vulnerable 2.6 T property to cope with shock; whether they were able to rely sociodemographic group(s) in on savings or obtain a loan; whether help was provided by the Republic of Congo relatives or friends; whether they had other means to cope, for example, through insurance; or whether they simply had to cope without support from any of these mecha- When analyzing vulnerability, the common attitude is nisms. As shown in Figure 2.6.2, for all shocks considered to assess the situation of particular sociodemographic together, almost half of the households declared that they groups who are often perceived as vulnerable: youth, had no support mechanism in place. That proportion was, elderly, disabled, women, autochthons. A series of vul- as expected, even higher for the poor. nerable groups are often considered given the expected  hare of Households Affected by Various Shocks and Coping Mechanisms when Affected by a FIGURE 2.6: S Shock, 2011 (%) 1. Share of households affected by various shocks 2. Coping mechanisms when affected by a shock Any of the 7 shocks No help Help provided by Serious illness relatives/friends Bankruptcy of business Relied on savings Obtained a loan Death Sold its property Crop loss Other (insurance, etc..) Loss of livestock Sold livestock Flood/diaster Help provided by government Loss of Employment Sold its capital 0 5 10 15 20 25 30 35 40 45 50 0 10 20 30 40 50 60 Bottom 40 Top 60 Total Source: Authors’ calculation using the 2011 ECOM survey. 26 Republic of Congo – Poverty Assessment Report challenges they may face in satisfying their basics needs. nonpoor. In urban areas, the disabled are more likely For example, some ethnic groups, due to cultural bar- to be insecure nonpoor and are more likely to suffer riers and low endowments, are often excluded from the in case of a negative shock on their income source. In economic sphere. Similarly, the lack of a social safety rural areas, households headed by a disabled person are net can negatively affect the elderly. Cultural division of slightly better off even when taking into account the labor and preference for boys regarding expenditure on insecure nonpoor. education, often play against women. Disabled people There is a U-shaped relationship between face similar challenges in accessing social services and the the household head’s age and vulnerability. labor market. In this section, we look at the situation Households headed by young people are more vulner- of these various groups of the monetary welfare ladder. able (Figure 2.7.1). Poverty rate reaches 60 percent The autochthons, who represent about 1 percent in households headed by those ages 15–19. On the of the population, stand out as the main marginalized contrary, poverty incidence is lower for households group in the Republic of Congo. Monetary poverty headed by someone ages 30–50. Beyond the age of 50, headcount for autochthons is more than twice the pov- poverty starts increasing. The elderly appear thus to be erty rate of the remaining population. Close to nine out a vulnerable group as well. Households headed by the of ten autochthons are poor. Clearly, this group appears elderly (ages 60 and above) are more likely to be ranked to be extremely marginalized (Figure 2.7.3). As we will as poor or most insecure poor. The youth constitute a document in the next chapters, their marginalization is disproportionate share of the poor, with especially large characterized by very limited access to social services, increases for young people ages 0 to 10 (Figure 2.7.6). including health and education, as well as to the labor The vulnerability of autochthons, the disabled, market. Thus, they contribute and benefit very little from women, youth, and the elderly is confirmed with a economic activities. The autochthons who are not poor measure of subjective well-being. Considering another tend to be close to the poverty line, and thus ranked as dimension of welfare, households headed by autochthons insecure nonpoor. There is a clear difference between and youth are more likely to have difficulties satisfying those living in rural areas and those living in urban areas. their food needs (Figure 2.8). Those in rural areas which All those who manage to move from rural areas to urban are headed by the disabled or women also experience areas are ranked as middle class. challenges accordingly in satisfying their food needs. There seems to be very little difference between The already high and generalized poverty in households headed by men and those headed by rural areas can explain that other factors seem not to women. In urban areas, households headed by women matter in rural areas. Except ethnicity, all other possible are slightly more vulnerable. In the Republic of Congo, discriminatory characteristics seem to matter in urban 21.5 percent of households are headed by women. At the areas only. This is in part due to the already very high national level, there is little difference with regard to wel- incidence of poverty in rural areas. This means that there fare between households according to the gender of the is no need to try and use these factors (gender, age, dis- household head (Figure 2.7.2), but this situation is true ability, and ethnicity) for targeting of potential programs only in rural areas. In urban areas, households headed by in rural areas. Given the already high poverty rate in rural women are more likely to be poor or insecure nonpoor. areas, universal coverage of a social program is preferable. Disabled people living in urban areas are more vulnerable. About 3 percent of households are headed 2.7 Conclusion by a person with disability. The incidence of poverty is lower for households headed by a disabled person This chapter has provided an assessment of the urban- (Figure 2.7.4), but this picture changes if we consider rural dichotomy and growing vulnerability in the the insecurity, meaning proximity to poverty line for the country. The Growing Urban-Rural Dichotomy and Vulnerability are of Concern 27 FIGURE 2.7: Poverty, Insecure Nonpoor, and Middle Class Status by Vulnerable Groups 1. Poverty rate by head’s age and gender, 2011 2. By head’s gender, 2011 70 100% 60 80% 50 40 60% 30 40% 20 20% 10 0 0% 15–19 20–29 30–39 40–49 50–59 60+ Men Women Total Men Women Total Men Women Total Rural Urban Poor Insecure nonpoor Middle class 3. By head’s ethnicity, 2011 4. By head’s disability, 2011 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% Bantou Pygmy Total Bantou Pygmy Total Other Disabled Total Other Disabled Total Rural Urban Rural Urban Poor Insecure nonpoor Middle class 5. Poverty by age group of the overall population 6. Contribution to poverty by age group of the overall population 60 18 16 50 Poverty headcount (%) Poverty headcount (%) 14 40 12 10 30 8 20 6 4 10 2 0 0 0–4 9–10 11–14 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84 85 et + 0–4 9–10 11–14 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84 85 et + Age group Age group 2005 2011 2005 2011 Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. The story that emerges from these estimates poverty reduction was observed in the two largest cities is that of a dual economy. Poverty is becoming an of Brazzaville and Pointe Noire. What was most worri- increasingly rural phenomenon, and this should some was an increase in poverty headcount as well as in be a much larger reason for concern. Most of the the number of poor people in rural areas. In rural areas, 28 Republic of Congo – Poverty Assessment Report Share of Population Having Difficulties Satisfying their Food Needs FIGURE 2.8:  70 60 50 40 30 20 10 0 Bantou Pygmy Other Disabled Male Female 15–19 20–29 30–39 40–49 50–59 60+ Head ethnicity Head disablility Head gender Head age Urban Rural Source: Authors’ calculation using the 2011 ECOM survey. the poverty headcount went up by 4.6 percentage points the agriculture and manufacturing sectors contracted. from 64.8 in 2005 to 69.4 in 2011. In rural areas, the Agriculture has become less competitive and now, the depth and severity of poverty has also increased, that is, Republic of Congo relies heavily on importation to satisfy the rural poor have become poorer. Between 2005 and their food needs. Safoulanitou and Ndinga (2010) show 2011, the number of poor increased in rural areas to that the Republic of Congo’s massive food imports rep- 951,000 up from 795,000. While rural areas accounted resented a strategy not only for ensuring the country’s for 44.3 percent of the poor in 2005, this had increased food security, but also for supplying food products to to 57.4 percent by 2011. neighboring countries, notably the Democratic Republic The urban-rural gap is very important in non- of Congo and Angola. They further show that repeated monetary dimensions of well-being as well. For armed conflict and Dutch diseases have lasting impor- example, the availability of social services and modern tant negative consequences on agricultural production. infrastructures is very limited in rural areas. For example, The gap between urban and rural areas is a factor only one in ten households in rural areas states that the driving migration in the Republic of Congo. Those electricity network is available in their community. The in rural areas are more likely to migrate. Because of the corresponding figure for Pointe Noire and Brazzaville are difficult conditions in rural areas, individuals tend to way higher (nine out of ten for Pointe Noire and close to leave their households for greener pastures in urban areas, seven out of ten for Brazzaville). Similar urban-rural gaps especially toward the two main cities. The likelihood of are observed for many other social services, including the migration increases with poverty rate, population den- availability of electricity, piped water, a market, public sity, and the distance to the two main cities (Brazzaville/ transport, and a health center. As a consequence, living Pointe Noire). One can deduct from such results that the conditions are not attractive in rural areas. national wealth is concentrated in these cities, and the Low agricultural production and lack of link- farther from them, the less is the likelihood of benefit- ages between rural activities (agriculture) and urban ting from the economic growth. Migration also appears activities (manufacture and services) seem to be to be a coping mechanism in case of shock. Those who important reasons driving monetary poverty in rural experienced a shock are more likely to migrate. areas. The Republic of Congo economy is not diversi- Such migration flow creates more problems fied. The expansion of the oil sector was accompanied by in cities, including expansion of slums, congestion negative effects of the Dutch disease. As a consequence, on social services, increased unemployment, and The Growing Urban-Rural Dichotomy and Vulnerability are of Concern 29 insecurity. Many who move to cities are often not For Brazzaville, 80 percent of the poor live in those four equipped with the right skills and assets for a smooth arrondissements. Pointe Noire also exhibited a concen- transition-insertion, thus ending up experiencing vul- tration of poor in specific arrondissements. In Pointe nerable living conditions. This includes living in slums Noire, around 78 percent of the poor live in Loandjili with shelters that are of poor quality and difficulties with and Tietie—46.7 and 32.1 percent, respectively. regard to access to sanitation. For example, one out of Success in reducing poverty has resulted in many three households in Brazzaville or Pointe Noire doesn’t households that are living just above the poverty line have a safe toilet. As a consequence for the new migrant, who remain vulnerable to falling below the poverty the living conditions in the city can be worse because of line in the face of a negative shock. As poverty receded, the lack of or limited formal safety net. In rural areas, both the vulnerable (insecure nonpoor) and middle class a traditional safety net mechanism can help them cope groups expanded, with the middle class group expand- with difficulties. ing faster than the vulnerable/insecure nonpoor group. There are pockets of poverty in the two main While only 20.6 percent of the population had con- cities. Currently, there is not enough information to sumption more than twice the poverty line in 2005, this build a poverty map for the Republic of Congo. Luckily, increased to 26.3 percent by 2011. In the meantime, 32.8 the World Bank conducted a survey in Brazzaville and percent of the population, while technically ‘nonpoor’, Pointe Noire in 2008. This was the baseline survey for is consuming at a level below an average XAF 550,000 the PEEDU. This survey was representative up to the yearly (in nominal 2011 CFA francs) per equivalent ‘arrondissement’ level. The estimates suggest that for adult. Given that a large share of the vulnerable rely on Brazzaville, Makélékélé was the arrondissement which either agriculture or informal activities which are sus- contributes the most to poverty (36 percent of the ceptible to significant output volatility, many remain poor). In Brazzaville, Talaingai, Mfilou, and Ouenze at serious risk of falling back into poverty, at least on a also presented double-digit contribution to poverty. temporary basis. 30 Republic of Congo – Poverty Assessment Report Who are the Poor? 3 3.1 Introduction The two previous chapters summarized the poverty and inequality dynamics between 2005 and 2011. The country experienced a substantial reduction in poverty, but growing inequality and dichotomy between urban and rural areas are of concern. More alarming is the fact that poverty indices are on the rise in rural areas. This chapter will focus on providing a detailed profile of the poor in terms of a range of sociodemographic characteristics. We are interested in the most recently available charac- teristics of the poor. Therefore, this chapter will focus on data from the 2011 ECOM survey to derive the poverty profile. At times, the 2005 ECOM survey will be used to provide trends of key changes over time. This chapter focuses on the univariate profile of poverty in the Republic of Congo. A (univariate) poverty profile is a set of tables giving the probability of being poor or other measures of poverty according to various household characteristics, such as the geographic area in which a household lives or the level of education of the household head. In addition to measuring poverty by household characteristics, it is also useful to assess the correlates of poverty using regression techniques. This multivariate analysis will be done in the next chapter on determinants and driv- ers of poverty reduction. This chapter has six parts. In the first part, we will revisit the relationship between poverty and location in the Republic of Congo. The second part analyzes the link between demographic characteristics (disability, gender, ethnicity, and age). The third part looks at the relationship between household size—including household composition—and poverty. The fourth part dis- cusses education characteristics along the welfare distribution. The fifth part analyzes the relation- ship between poverty and the sector of activity. The last part will look at housing characteristics and assets ownership along welfare distribution. 3.2 The poor live either in rural areas or in Brazzaville suburbs Poverty has a strong regional dimension. As mentioned in Chapters 1 and 2, there are large differences in poverty by location. There was a much lower and more rapidly decreasing 31 FIGURE 3.1: Poverty Profile by Location and Department, 2011 1. Location 2. Departments 40 80 40 90 35 70 35 80 70 Poverty Headcount (%) Poverty Headcount (%) 30 60 30 Population Share (%) Population Share (%) 60 25 50 25 50 20 40 20 40 15 30 15 30 10 20 10 20 5 10 5 10 0 0 0 0 Brazzaville Pointe Noire Other municipalities Semi-urban Rural Kouilou Niari Lékoumou Bouenza Pool Plateaux Cuvette Cuvette-Ouest Sangha Likouala Brazzaville Pointe Noire Population share Poverty headcount 3. Access to covered market 4. Access to open-air market 60 60 70 60 50 50 60 50 Poverty Headcount (%) Poverty Headcount (%) Population Share (%) Population Share (%) 50 40 40 40 40 30 30 30 30 20 20 20 20 10 10 10 10 0 0 0 0 No Yes No Yes Population share Poverty headcount Source: Authors’ calculation using the 2011 ECOM survey. poverty measure in Brazzaville and Pointe Noire. In the Poverty is predominantly a rural phenomenon. In meantime, there was higher poverty level in other urban addition, urban poverty remains very important, espe- municipalities and semi-urban areas and the highest lev- cially Brazzaville. In rural areas, seven out of ten (69.4 els of poverty in rural areas and no decrease over time percent) people are poor; 57.4 percent of poor people live in in those areas. The first panel in Figure 3.1 provides a rural areas (Figure 3.1). Brazzaville, despite a relatively low visualization of the levels of poverty (according to the poverty incidence (21.6 percent), has a big share of poor. headcount index, represented by the line/curve) as well Close to 20 percent of poor people live in Brazzaville. Thus, as population shares (represented by the vertical bars) by rural areas and Brazzaville account for 70 percent of the location in 2011. Since these results were already dis- overall population and 77 percent of total poor. The second cussed in Chapters 1 and 2, they do not need to be panel in Figure 3.1 displays the share of the population in repeated here. poverty by department as well as each department’s share in the overall population of the country. There are some 32 Republic of Congo – Poverty Assessment Report Box 3.1: A  dministrative structure of the differences in poverty measures between the 10 depart- Republic of Congo ments excluding Brazzaville and Pointe Noire, but those differences are much smaller than the differences observed The Republic of Congo is divided administratively into 12 between the departments where the two main cities are departments, themselves divided into communes and districts. The ECOM 2011 survey is representative at the level of the 12 located and the ten other departments. departments, but this is not the case for the 2005 ECOM survey. Beyond the urban-rural dichotomy, significant Therefore, in this study, results at the level of departments spatial differences in welfare exist in the Republic of are provided only for 2011. The departments are Bouenza, Congo. Figure 3.2 visualizes geographical disparities in Brazzaville, Cuvette, Cuvette-Ouest, Kouilou, Lékoumou, Likouala, Niari, Plateaux, Pointe Noire, Pool, and Sangha. The poverty by department with maps. This is done for the location of the departments is indicated in Figure B3.1. first three consumption-based poverty measures of the Foster–Greer–Thorbecke (FGT) class. Among the 12 FIGURE B3.1: Map of the Republic of Congo ‘departments’ in the Republic of Congo, Pointe Noire and Brazzaville have the lowest poverty rate with a pov- Likouala erty headcount of 20.3 and 21.6, respectively. Cuvette- Ouest is the poorest department, with 79.1 percent of Sangha the population living below the poverty line, followed Cuvette-Ouest by Lékoumou and Cuvette with 76.1 and 70.2 percent of poor individuals, respectively. The poverty rate in the Cuvette province of Kouilou is 56.9. The poverty rate is quite high for the remaining departments—the poverty rate Plateaux ranges between 62 and 69 percent for Plateaux, Likouala, Niari Lékoumou Bouenza, Sangha, Pool, and Niari. Pool Kouilou Poverty is typically higher in areas with limited Brazzaville access to services. Still, another way to look at the geog- Pointe Noire Bouenza raphy of poverty is to consider poverty in areas according to the extent to which these areas have access to a range FIGURE 3.2: Poverty Maps by Department, 2011 1. Consumption-based headcount 2. Consumption-based poverty gap 3. Consumption-based squared poverty gap (75.0,80.0] (45.0,50.0] (45.0,50.0] (70.0,75.0] (40.0,45.0] (40.0,45.0] (65.0,70.0] (35.0,40.0] (35.0,40.0] (60.0,65.0] (30.0,35.0] (30.0,35.0] (55.0,60.0] (25.0,30.0] (25.0,30.0] (50.0,55.0] (20.0,25.0] (20.0,25.0] (45.0,50.0] (15.0,20.0] (15.0,20.0] (40.0,45.0] (10.0,15.0] (10.0,15.0] (35.0,40.0] (5.0,10.0] (5.0,10.0] (30.0,35.0] [0.0,5.0] [0.0,5.0] (25.0,30.0] (20.0,25.0] (15.0,20.0] (10.0,15.0] (5.0,10.0] [0.0,5.0] Source: Authors’ calculation using the 2011 ECOM survey. Who are the Poor? 33 of public services. As an example, Figure 3.1 displays the or a woman are more likely to be poor or vulnerable. relationship between access to markets, whether open-air Poverty incidence is higher for the youth and elderly. or covered, and poverty. Access to markets is associated with a lower likelihood of being poor, suggesting local  he poor live in large households 3.4 T geographic effects on standards of living. with high dependency ratios 3.3  Autochthons, youth, elderly, urban In this section, the analysis will focus on household size and composition and their relationship with women, and urban disabled are poverty. By definition, consumption-based measures more likely to be poor of poverty are supposed to have some relationship with household size and its composition. Under the meth- The previous chapter on vulnerability already odological approach to derive a consumption-based explored some of the relationships between demo- measure of poverty profile, the welfare indicators are graphic characteristics and poverty, including dis- conceived as the overall household consumption divided ability, gender, ethnicity, and age. In Chapter 2, it was by the number of equivalent adults (or the number of documented that ethnicity, gender, age, and disability household members). Explicitly, this means the more the matter for poverty and vulnerability. As illustrated in members, the more resources will be needed to satisfy Chapter 2, autochthons are the most vulnerable group the household’s basic needs. of the population (see Box 3.2 for more on this), with The poor live in large households with high poverty headcount that is double the national average. dependency rates. In 2011, poor households had on In urban areas, households headed by a disabled person Box 3.2: Ethnicity and poverty: The case of the autochthons (Pygmies) The Pygmies are considered to be among the oldest inhabitants in Central Africa. Their seminomadic lifestyle has remained largely unchanged for thousands of years, living by hunting, fishing, and gathering wild fruits and nuts. In the last two or three decades, however, under the influence of multiple factors, these populations have gone through a process of semi-sedentarization. Traditionally, the Pygmies in Central Africa were closely attached to the rain forest. They were the “Forest People” (Turnbull 1961) and the forest was the source of their religion, their livelihood, and their protection. They used to lead a nomadic life in camps of 30 to 40 families, which maintained regular links and exchanges with each other. Their mostly egalitarian and horizontal society acknowledged the wisdom of elders who preserved the community’s knowledge of the sites, plants, animals, ghosts, and spirits as well as their entire cultural heritage (rituals, music, dances, holy sites) and practices (pharmacopeia, hunting, and fishing). Elders occupied prominent positions within the community and settled disputes. They lived in simple huts made out of leaves and branches. This traditional lifestyle should not necessarily be equated with a life of poverty. It had its own dignity, noblesse, and coherence and it is part of the universal heritage of humanity. Yet today, the traditional Pygmy lifestyle is in danger: as a population, they are losing what constitutes their identity and the richness of their culture and knowledge due to gradual sedentarization. Their access to the forest as well as to land for cultivation is increasingly at risk. Estimates by Wodon et al. (2012) for the Democratic Republic of Congo, the Central African Republic, and Gabon suggest that Pygmy populations tend to be very poor. Many children are not enrolled in schools and adult literacy is low. Health outcomes are weak and vulnerability is high. In addition, qualitative data from the Democratic Republic of Congo suggest that many among the Pygmies perceive themselves negatively, in part because aspects of their culture (type of housing, religious beliefs, rites, and practices) are considered “bad” by their Bantu neighbors. In the Democratic Republic of Congo, the Bantu and state institutions may not treat the Pygmies in a fair manner that would allow them to make informed changes and adaptations to improve their general living conditions and live in harmony with their neighbors while preserving their own culture (World Bank 2009). In the Republic of Congo, the ECOM 2005 survey did not identify Bantus and Pygmies separately, but the 2011 ECOM survey did. The results suggested patterns of disadvantage among Pygmy populations similar to those observed in other Central African countries. Source: Wodon et al. (2012). 34 Republic of Congo – Poverty Assessment Report average 1.34 members more than nonpoor households TABLE 3.1: Household Size, Composition, and (Table 3.1). This difference is entirely driven by the Dependency Ratios (%) by Poverty higher number of children in poor households: while Status, 2011 the average extremely poor household had 2.35 children Poverty status (below the age of 15), the average nonpoor household Nonpoor Poor Total only had 1.36. The number of adults does not differ Children aged 0 to 14 1.36 2.35 1.70 significantly across poverty status. Adult aged 15 to 64 2.33 2.61 2.43 The Republic of Congo has started it demo- Elderly aged 65 plus 0.13 0.20 0.16 graphic transition but fertility still remains high, Household size 3.83 5.16 4.28 especially among poor households. The total fertility Dependency 64.10 97.70 76.30 was very high in the 1970s and 1980s (a bit more than Child dependency 58.30 90.10 69.90 six children per women [Figure 3.3.2]). Toward the early Aged dependency 5.80 7.60 6.40 1980s, the country started its demographic transition. Source: Authors’ calculation using the 2011 ECOM survey. The number of children per woman has since been going down at a steady but slow pace to reach 5.1 children per woman in 2005. Further reduction was observed As a consequence of the high fertility among between 2005 and 2011. In 2014, the total fertility the poor, dependency ratios are strongly correlated rate was estimated at 4.9. As expected, fertility rate is with poverty. The distribution of dependency across higher in rural areas and for the poor (CNSEE and ICF welfare quintile shows a strong and negative correlation International 2012). The total fertility for women in (Figure 3.3.1). In 2011, for example, the dependency rural areas is 6.5, which is 2 children more than those in ratio for the poorest quintile was 109.5 against only 52.3 urban areas. The 2012 Demographic and Health Survey for the richest quintile. This means that an adult in the (DHS) results suggest that total fertility for women in the poorest quintile must cater to the needs of many more poorest quintile is 7 children, while those in the richest people. Between 2005 and 2011, dependency ratios quintile have only 3.8 children. have increased substantially for the poorest, while there was a slight reduction in dependency for the rich. High FIGURE 3.3: Dependency Ratio and Fertility Rate 1. Dependency by welfare quintiles 2. Fertility rate, total (births per woman) 120 7 100 6 5 80 4 60 3 40 2 20 1 0 0 Q1 Q2 Q3 Q4 Q5 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2005 2011 Source: Authors’ calculation using the 2011 ECOM survey. Source: WDI. Who are the Poor? 35 FIGURE 3.4: Population Share and Poverty Headcount by Level of Education of the Head and His/her Spouse 1. Education of the head 2. Education of the spouse 35 70 35 70 30 60 30 60 Poverty Headcount (%) Poverty Headcount (%) Population Share (%) Population Share (%) 25 50 25 50 20 40 20 40 15 30 15 30 10 20 10 20 5 10 5 10 0 0 0 0 None Primary Secondary Secondary Tertiary None Primary Secondary Secondary Tertiary No 1 2 1 2 spouse Population share Poverty headcount Source: Authors’ calculation using the 2011 ECOM survey. fertility has numerous implications not only for depen- education (Figure 3.4). The likelihood of poverty also dency ratio, but it also represents serious challenges for decreases with the education level of the spouse. improving human development outcomes as cohorts of In the Republic of Congo, the vast majority of children are getting larger and for empowering women the population lives in households with a head who (Canning, Raja, and Yazbeck 2015). has at least completed primary education. However, Lower fertility can have positive effects on transition to a higher level of education is low, especially household living standards in both the short and lon- for those in the lowest quintiles of welfare. Eight out of ger run. In the short run, lower fertility rates translate ten people live in households with a head who had com- into smaller households and lower dependency ratios. pleted at least primary education. As expected, the poor Fewer dependent children in a household means less study less than the nonpoor. For instance, only one out strain on household resources and a mechanical increase of ten people in the poorest quintile has a head with at in per equivalent consumption. In the longer run, a drop least upper secondary level education. The corresponding in fertility tends to lead to increased female labor market figure for the richest quintile is six out of ten. participation and better human capital outcomes for the The poor households are characterized by a low younger generation. When families have fewer children, level of education of the head, with more than half of they also tend to invest more in the education of their the heads with either no education or primary edu- children, laying the foundations for improved household cation at most. Higher levels of education are for the living standards in the next generation.27 better off exclusively (Figure 3.6). Affordability issues and trade-offs that the poor face are some of the factors 3.5  The poor are more likely to be behind this situation. Chapter 6 will look at constraints to schooling in more detail. unskilled Education beyond the primary level to at least upper secondary level seems to have the biggest returns As is the case worldwide, there is a strong link between in terms of poverty reduction. Households whose head education and poverty in the Republic of Congo. The probability of being poor decreases with the education level of the household head, from no education to pri- 27 Becker, Murphy, and Tamura 1990; Bloom, Canning, and Sevilla mary, lower secondary, upper secondary, and tertiary 2003; Bloom et al. 2009; Galor 2005; Galor, O. & Weil 1999. 36 Republic of Congo – Poverty Assessment Report FIGURE 3.5: Population Share and Welfare Aggregate by Level of Education of the Head 1. Education of the head by quintile 2. Welfare indicator by head’s education 100% 1000000 90% 900000 Population distribution (%) 80% 800000 Annual consumption per equivalent adult (XAF) 70% 700000 60% 600000 50% 500000 40% 400000 30% 300000 20% 200000 10% 100000 0% 0 Q1 Q2 Q3 Q4 Q5 None Primary Secondary Secondary Tertiary Superior 1 2 Welfare quintile Education of the household head None Primary Secondary 1 Secondary 2 Superior 2005 2011 Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. Distribution of the Level of Education of the Head by Welfare Quintile FIGURE 3.6:  1. Head with at least primary education by quintile 2. Head with at least lower secondary education by quintile 100 100 80 80 60 60 (%) (%) 40 40 20 20 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Welfare quintile (%) Welfare quintile (%) 2005 2011 3. Head with at least upper secondary education by quintile 4. Head with at least tertiary education by quintile 80 40 60 30 40 20 (%) (%) 20 10 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Welfare quintile (%) Welfare quintile (%) 2005 2011 Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. Who are the Poor? 37 TABLE 3.2: Confidence Interval of Monthly Between 2005 and 2011, there was a notable Consumption Per Equivalent Adult by improvement in the education level of the household Head’s Education Level head. This improvement happened mainly for those at [95% conf. interval] the top of the welfare distribution. For instance, the Mean Standard (in XAF) error Minimum Maximum proportion of those living in a household with a head None 345,854 17,786 310,954 380,755 who has completed at least upper secondary education increases by 5 percentage points from 30.6 to 35.9 Primary 334,555 11,438 312,111 357,000 percent. Chapter 4, on drivers of poverty, will further Secondary1 415,942 14,579 387,334 444,549 investigate how this difference in skills endowments for Secondary2 575,026 28,862 518,391 631,660 the poor and nonpoor has affected poverty dynamics. Superior 862,162 58,076 748,201 976,123 Total 492,548 17,466 458,276 526,821 Source: Authors’ calculation using the 2011 ECOM survey.  he poor are either unemployed, 3.6 T employed in agriculture, or employed in informal services has no education or only primary education have similar standard of living (Figure 3.5 and Table 3.2). The welfare There is a clear urban-rural dichotomy with regard to aggregate for those with secondary and tertiary education sector activity of the household head. In rural areas, most are statistically different from those with no education of the household heads are employed in agriculture. There or primary education only. Households headed by some- is a strong difference in the type of job that people have one with no education or with only primary education access to or are attracted to according to the residence areas. consumed on average XAF 345,000 and XAF 334,000 As is the case for most SSA countries, the rural population per equivalent adult, respectively. The difference between relies heavily on subsistence agriculture as a main income these two is not statistically different (Table 3.2). The source. Agriculture is important for all welfare quintiles welfare increased significantly with the education of the (Figure 3.7). The services sector is also growing in rural head, from XAF 415,000 for those with lower secondary areas, and those who manage to move out of agriculture to XAF 575,000 for those with upper secondary and XAF into services seem to be doing well. The share of house- 862,000 for those with tertiary education. hold heads in services increased with welfare, and is quite FIGURE 3.7: Distribution of the Population by Head’s Sector of Occupation and Welfare Quintile 1. Urban areas only 2. Rural areas only 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Agriculture Mining/Manufacturing Services Unemployed/Inactive Source: Authors’ calculation using the 2011 ECOM survey. 38 Republic of Congo – Poverty Assessment Report important for the richest quintile. Up to 35 percent of rural Households headed by an unemployed person or some- households in the top quintile are in services. one engaged in informal services are also experiencing In rural areas, the poor are either unemployed or high poverty rates. In urban areas, if we characterize poor rely on subsistence agriculture as main income source. households based on the activity of the head, three main In rural areas, poverty is very high for households with groups will emerge: (a) those in agriculture; (b) those a head who is unemployed or engaged in agriculture. in informal services; and (c) those who are unemployed The lowest poverty rate in rural areas is registered for or inactive. those in civil service and those in the transport sector Given the situation on the labor market, one (Figure 3.8.2). Surprisingly, those in mining are not can conclude that poverty alleviation reforms in the doing that well in rural areas. This might be due to Republic of Congo should focus on at least three artisanal mining or that they don’t have the skill or the things: (a) boosting agricultural productivity and bargaining power to negotiate better wages. income; (b) boosting informal services productivity In urban areas, services delivery is by far the and income; and (c) providing social safety nets for main sector of activity. Unemployment is a big con- the unemployed/inactive. It appears from the previous cern in urban areas. As illustrated in Figure 3.8.1, ser- analysis that the poor are concentrated in three groups vices is by far the main sector of occupation of household that reflect survival strategy: agriculture, informality, heads in urban areas. This is true across the welfare lad- unemployment. Helping growth and productivity in these der, but often, these are informal activities concentrated sectors will definitively contribute to poverty alleviation. in sales/trades and other services. Surprisingly, most of In summary, the occupation of the household the people in urban areas live in households with a head head seems to have a strong influence on the house- who is unemployed or inactive. In a situation of almost hold living standard. Wages earned vary across the inexistent formal safety, this can be a big concern for sector and therefore induce the available resource. If we shared prosperity. ranked households according to the main sector of activ- In urban areas, the highest probability of ity of the head, it systematically appears that poor house- being poor is among heads working in agriculture. holds are those whose heads are engaged in agriculture, FIGURE 3.8: Population Share and Poverty Headcount by Sector of Occupation of the Head 1. Urban areas only 2. Rural areas only 30 60 70 80 Poverty Headcount (%) Poverty Headcount (%) 70 Population Share (%) Population Share (%) 25 50 60 50 60 20 40 50 40 15 30 40 30 10 20 30 20 20 5 10 10 10 0 0 0 0 Agriculture Unemployed/Inactive Services Trade/Sales Other services Construction Transport Education/Health Production/processing Administration Mines/quarries Unemployed/Inactive Agriculture Other services Services Construction Mines/quarries Trade/Sales Education/Health Production/processing Administration Transport Population share Poverty headcount Source: Authors’ calculation using the 2011 ECOM survey. Who are the Poor? 39 engaged in informal services, or unemployed. A more lead poisoning, injuries, and mental health (Krieger and detailed analysis of employment patterns, including Higgins 2002). The type of materials used for construc- estimation of the probability of employment and wage tion is therefore not only an indicator of the economic regressions to estimate the returns to education, is pro- situation of households, but also an indicator of poten- vided in Chapter 5. tial exposure of household members to these health conditions. Housing quality has also been associated 3.7  The poor live in dwellings made with child development. According to Jaramillo (2014), basic infrastructure related to housing quality is crucial for of non-improved materials and achieving child development with regard to health, nutrition, own little assets and education outcomes in the long run. Moreover, some of Beyond the direct discomfort, living in a poor stan- the low quality construction material (mud and poles, thatch, dard dwelling can have a negative consequence on and so on) are not resistant and will collapse easily in case the population, especially on their health. The low of heavy rain or floods, which are likely given the country’s quality of construction materials can also make it dan- location in straddling the equator. gerous as the dwelling might collapse easily in case of As expected, poor households are more likely to bad weather. It has been documented that low housing live in dwellings made of unimproved materials. In conditions are associated with a wide range of health 2011, a bit more than nine out of ten (92.9 percent) non- conditions, including respiratory infections, asthma, poor households where living in dwellings with improved TABLE 3.3: Housing Characteristics and Assets Ownership by Poverty Status (%) 2005 2011 Wald test Wald test Nonpoor Poor Total (p-value) Nonpoor Poor Total (p-value) Dwelling amenities Electricity 38.9 10.8 26.7 0.000 56.3 15.5 42.5 0.000 Piped water 37.4 12.1 26.5 0.000 59.6 24.8 47.8 0.000 No toilet 5.2 12.6 8.4 0.000 6.7 20.5 11.3 0.000 Improved roof 89.5 77.1 84.1 0.000 92.9 72.7 86.0 0.000 Improved wall 61.3 42.6 53.2 0.000 67.2 44.4 59.4 0.000 Improved floor 70.7 40.9 57.8 0.000 78.8 36.7 64.5 0.000 Asset ownership Television 33.6 9.6 23.2 0.000 60.5 17.3 45.9 0.000 Computer 1.4 0.1 0.9 0.000 5.1 1.3 3.8 0.000 Radio 63.9 46.8 56.5 0.000 59.5 43.3 54.0 0.000 Fixed/mobile phone 45.5 14.3 32.0 0.000 88.3 65.9 80.7 0.000 Bicycle 6.6 5.6 6.2 0.224 4.3 7.9 5.5 0.000 Motorcycle 2.3 0.8 1.7 0.001 3.9 3.4 3.7 0.334 Motor vehicle 3.1 0.2 1.9 0.000 3.1 0.5 2.2 0.000 Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. Note: The definition of improved roof material includes iron sheets, tiles, and cement. Improved wall material includes cement blocks, stabilized bricks, burnt bricks, and clay cemented. Improved floor material includes tiles, cements, and marble. Wald test are for the difference non-poor vs non poor. 40 Republic of Congo – Poverty Assessment Report roof. The corresponding figure for poor households was for piped water. Chapter 6 will look at challenges faced seven out of ten (Table 3.3). Differences are even more by the poor in connecting to electricity or piped water pronounced when it comes to wall and floor materi- network. als. Close to eight out of ten nonpoor households have Ownership of assets is very low for poor improved floor while the corresponding figure for poor households. Yet, some assets are critical endowment households is four out of ten. that can help the households be more productive. For Usage of time-saving technology such as elec- example, information and communication technology tricity and piped water is predominantly reserved (ICT) equipment such as televisions, radios, and mobile for nonpoor households. Electricity, piped water, and phones can provide very useful information on market critical services not only provide a better living stan- opportunities, prices, available technology, weather, and dard for a household, but also have gender and youth so on that households can build on to improve their implications. Often, children and youth are in charge of livelihood. Unfortunately, many poor households still domestic chores including fetching water. Electricity can lack these basic equipment in the Republic of Congo. also help children to do their homework at home and In 2011, only 17 percent of poor households had a TV thus increase their performance in school, but often, the set, against 60 percent of nonpoor. Despite substantial poor households are excluded from the usage of these improvement since 2005, many still don’t have a mobile critical services. For instance, only 15 percent of poor phone. Up to 35 percent of poor households did not households had access to electricity in 2011, against up have a mobile phone in 2011, against only 12 percent to 56 percent for the nonpoor. A similar gap is observed for the nonpoor. TABLE 3.4: Housing Characteristics and Assets Ownership by Welfare Quintile (%), 2011 Welfare quintile Q1 vs Q5   Q1 Q2 Q3 Q4 Q5 Total Wald test (p-value) Dwelling amenities Electricity 8.6 21.3 36.8 52.7 72.3 42.5 0.000 Piped water 12.9 34.3 48.3 59.9 67.2 47.8 0.000 No toilet 28.1 14.5 9.1 7.8 3.9 11.3 0.000 Improved roof 63.0 80.4 88.2 92.1 96.8 86.0 0.000 Improved wall 36.0 51.0 58.2 64.6 75.6 59.4 0.000 Improved floor 21.0 49.4 66.8 78.0 87.5 64.5 0.000 Asset ownership             Television 9.3 23.4 43.1 59.0 73.7 45.9 0.000 Computer 0.6 1.9 1.2 2.4 10.0 3.8 0.000 Radio 39.0 46.4 52.3 59.4 64.7 54.0 0.000 Fixed/mobile phone 56.9 73.2 81.9 88.2 92.7 80.7 0.000 Bicycle 9.3 6.8 4.3 4.2 4.4 5.5 0.000 Motorcycle 2.6 4.1 3.7 3.5 4.3 3.7 0.008 Motor vehicle 0.4 0.6 0.6 1.5 6.2 2.2 0.000 Source: Authors’ calculation using the 2011 ECOM survey. Note: The definition of improved roof material includes iron sheets, tiles, and cement. Improved wall material includes cement blocks, stabilized bricks, burnt bricks, and clay cemented. Improved floor material includes tiles, cements, and marble. Who are the Poor? 41 The differential in housing quality between women. However, in urban areas, households headed the poor and nonpoor is even bigger when looking by women are slightly more vulnerable. At the national at the whole welfare distribution through welfare level, there is little difference in terms of welfare between quintiles. Gap between the poor and the nonpoor is households according to the gender of the household even bigger (Table 3.4). By far, the biggest gap between head, but this situation is only true in rural areas. In the poorest and the richest quintile is observed in the urban areas, households headed by women are more ownership of TV sets. In the top quintile, up to 73.7 likely to be poor or insecure nonpoor. percent of households own a TV set, while the cor- Poverty has a strong regional dimension. There responding figure for the poorest quintile is only 9.3 are large differences in poverty by location. There was a percent. Very important differences are also observed much lower and more rapidly decreasing poverty mea- on ownership of radio and mobile phone. Thus, the sure in Brazzaville and Pointe Noire. In the meantime, digital revolution is happening, but has yet to deeply there was higher poverty level in other urban municipali- reach the poorest. ties and semi-urban areas and the highest levels of poverty Households tend to replace traditional devices in rural areas and no decrease over time in those areas. such as radio and bicycle with more upgraded goods, Poverty is predominantly a rural phenomenon. such as a TV set or motorcycle. Between 2005 and In addition, urban poverty remains very important, 2011, the share of households owning a radio decreased especially Brazzaville. In rural areas, seven out of ten (from 56.5 to 54 percent) while the share of those (69.4 percent) people are poor; 57.4 percent of poor owning a TV set increased substantially (from 23.2 people live in rural areas. Brazzaville, despite a relatively to 45.9 percent). During the same period, the share low poverty incidence (21.6), has a big share of poor. of those owning a bicycle decreased (from 6.2 to 5.5 Close to 20 percent of poor people live in Brazzaville. percent) while the share of those owning a motorcycle Thus, rural areas and Brazzaville account for 70 percent increased (from 1.7 to 3.7 percent). These are also signs of the overall population and 77 percent of total poor. of improved welfare and people aiming for better style Beyond the urban-rural dichotomy, significant of living. Still, ownership of modern assets, especially a spatial differences in welfare exist in the Republic of TV set or vehicle/motorcycle, is still low and reflects the Congo. Among the 12 ‘departments’ in the Republic middle income status of the country. of Congo, Pointe Noire and Brazzaville have the low- est poverty rate with a poverty headcount of 20.3 and 3.8 Conclusion 21.6, respectively. Cuvette-Ouest is the poorest depart- ment, with 79.1 percent of population living below This chapter has provided a basic profile of the poor in the poverty line, followed by Lékoumou and Cuvette the Republic of Congo. both with 76.1 and 70.2 percent of poor individuals, The autochthons stand out as the most margin- respectively. The poverty rate in the province of Kouilou alized group in the Republic of Congo. Monetary pov- is 56.9. The poverty rate is quite high for the remaining erty headcount for autochthons is more than twice the departments—the poverty rate ranges between 62 and poverty rate of the remaining population. Close to nine 69 percent for Plateaux, Likouala, Bouenza, Sangha, out of ten autochthons are poor. The marginalization of Pool, and Niari. autochthons is characterized by very limited access to Fertility remains high, especially among poor social services, including health and education, as well households. The poor live in large households with as the labor market. Thus, they contribute and benefit high dependency rates. In 2014, the total fertility rate very little from economic activities. was estimated at 4.9. In 2011, poor households had There seems to be very little difference between on average 1.34 members more than nonpoor house- households headed by men and those headed by holds. This difference is entirely driven by the higher 42 Republic of Congo – Poverty Assessment Report number of children in poor households: while the aver- There is a clear urban-rural dichotomy with age extremely poor household had 2.35 children (below regard to sector activity of the household head. As is the age of 15), the average nonpoor household only had the case for most SSA countries, rural population relies 1.36. The number of adults does not differ significantly heavily on subsistence agriculture as a main income across poverty status. source. The services sector is also growing in rural areas, As is the case worldwide, there is a strong link and those who manage to move out of agriculture into between education and poverty in the Republic of services seem to be doing well. In urban areas, services Congo. The probability of being poor decreases with the is by far the main sector of activity. Unemployment is a education level of the household head, from no educa- big concern in urban areas. Services is by far the main tion to primary, lower secondary, upper secondary, and sector of occupation of household heads in urban areas. tertiary education. This is true across the welfare ladder, but often, these are Education beyond the primary level to at least informal activities concentrated in sales/trades and other upper secondary seems to have the biggest returns services. Surprisingly, most of the people in urban areas in terms of poverty reduction. Households whose live in households with a head who is unemployed or head has no education or only primary education inactive. In a situation of almost inexistent formal safety, have similar standard of living. The welfare aggregate this can be a big concern for shared prosperity. for those with secondary and tertiary education are Given the situation on the labor market, one statistically different from those with no education can conclude that poverty alleviation reforms in the or only primary education. Households headed by Republic of Congo should focus on at least three someone with no education or only primary educa- things: (a) boosting agricultural productivity and tion consumed on average XAF 345,000 and XAF income; (ii) boosting informal services productivity 334,000 per equivalent adult, respectively. The differ- and income; and (c) providing social safety nets for ence between these two is not statistically different. the unemployed/inactive. It appears from the previous The welfare increased significantly with the education analysis that the poor are concentrated in three groups of the head, from XAF 415,000 for those with lower that reflect survival strategy: agriculture, informality, secondary education to XAF 575,000 for those with unemployment. Helping growth and productivity in upper secondary education and XAF 862,000 for those these sectors will definitively contribute to poverty with tertiary education. alleviation. Who are the Poor? 43 The Drivers of Poverty Reduction and Inequality 4 4.1 Introduction The objective of this chapter is to use several methods, including econometric analysis and standard decomposition methods, to quantify the contribution of different factors toward poverty reduction and inequality. The chapter has three parts. The first part will provide evi- dence of the correlates of poverty using multivariate econometrics models. In the second part we will take a closer look at some of the factors that have been associated with the reduction in poverty or associated with poverty increase in the case for rural areas. Based on the findings from previous chapters, we will focus on three key factors: households’ composition, education, and the type of occupation. The last part will provide evidence on factors that play a significant role in explaining inequality. 4.2 Correlates of poverty The univariate profile of poverty provided in the previous chapter is a useful step but has some limitations. The previous chapters suggest that a number of household characteristics are associated with a higher or lower probability for the household to be poor. Providing univariate profiles of poverty is a useful step to identify the characteristics of the population groups that are poor, but it is not sufficient to measure the impact at the margin of various household characteristics on poverty. The limitation with a poverty profile lies in the fact that it provides information on the probability of being poor according to various household categories, but this cannot be used to assess the magnitude of the correlates of poverty at the margin, controlling for other variables. Multivariate analysis based on regression techniques are a more robust way of assessing correlates of poverty. For instance, the variation in poverty measures across regions may be due to regional effects, but it may also be due to differences in household characteristics in various regions, and these differences in characteristics may have a larger effect on poverty than broader and specific characteristics of the regions themselves. To sort out the correlates of poverty and the impact (loosely speaking) at the margin of various variables on the probability of being poor, regression analysis is needed. When estimating such regressions, it is better to rely on linear regres- sions for the correlates of consumption per equivalent adult than on categorical regressions for the poverty status of households. This is because using a probit or logit model for poverty status 45 implies throwing away valuable information contained living in Pointe Noire have higher levels of consump- in the distribution of household consumption. tion than those living in Brazzaville, while households The regression for the correlates of poverty is in other urban areas and in rural areas have lower levels estimated nationally. The dependent variable is the of consumption. Many of the coefficients are large. For logarithm of the welfare aggregate. Apart from a constant, example, in 2011, the coefficient for rural areas is esti- the independent variables include (a) geographic location mated at −0.399, implying a reduction in consumption (Pointe Noire, other municipalities, semi-urban, and rural of close to forty percent for rural households in compari- with Brazzaville as reference category); (b) household size son to otherwise similar households living in Brazzaville. variables (number of infants, children, adults, and elderly Larger households tend to have substantially household members and their squared value to take into lower levels of consumption per equivalent adult, account potential non-linearity in relationships between which is not surprising. The effects of additional house- household size and consumption); (c) characteristics of hold members are, however, non-linear and decreasing the household head, including his/her ethnicity, disability at the margin (the linear coefficient for each of the age status, gender, age, level of education (primary, lower sec- groups is negative while the quadratic term is positive). ondary, upper secondary, or tertiary, with no education as As argued in the previous chapter, larger families tend to reference category) and his/her sector of activity (mines- have more children. As a result, the dependency ratio is quarries, production-processing, construction, transport, higher, thus placing a burden on household disposable trade-sales, services, education-health, administration, income. Issues related to responsible parenthood should other services, or unemployed/inactive, with agriculture- be part of the poverty alleviation interventions. livestock-fisheries-forestry as reference category); (d) the After controlling for other factors, there is no education level of the spouse of the household head with difference in terms of welfare between male- and the same categories; and (e) other variables, including female-headed households. One of the important whether the household owns land and whether it has findings here is that the gender of the head does not access to markets, both covered and open air, whether a matter for the overall household welfare. Female-headed member left the households, and whether the household households are doing as well (or as poorly, depending) as suffers from a shock. male-headed households. Robustness check across resi- It is important to note that the econometric dential areas suggests that this finding holds for both results presented here are not implying causality but rural and urban areas. correlates of poverty. The discussion of the results pro- A disability for the head is not associated with vided here is meant only to suggest how various house- a statistically significant change on the level of con- hold characteristics tend to affect levels of welfare. Actual sumption of households. Once we control for all other causality is not necessarily implied. Given the log-linear possible households characteristics, it appears that dis- specification of the regression, the coefficients of the first ability has no correlation with welfare. Still, disability regression on the logarithm of welfare can essentially be could be influencing other important determinants of interpreted as percentage gains in household consump- welfare such as education and access to quality jobs, as tion associated with the various explanatory variables. illustrated in the next two chapters. Figure 4.1 provides econometric results for selected vari- There is an inverted U-shape relationship ables. The detailed results are in the statistical appendix between the head’s age and the age of the household. (Table SA.10). This is in line with what one would expect given an Location is strongly correlated to welfare. individual’s life cycle. At the beginning of their adult Controlling for other independent variables, geographic life, individuals have limited assets and their income is location still has a large and statistically significant cor- still low. As they grow in age and experience, earnings relation with household consumption. Households increase. In addition, they also accumulate assets that 46 Republic of Congo – Poverty Assessment Report FIGURE 4.1: Correlates of Well-Being, 2011 (coefficients) 1. Region 2. Household composition 0.2 0.05 0.1 0.00 0.0 –0.05 –0.1 –0.10 –0.2 –0.15 –0.3 –0.20 –0.4 –0.25 –0.5 –0.30 Brazzaville Pointe Other Semi-urban Rural kid aged 0 to 5 kid aged 0 to 5, squared Boys aged 6 to 18 Boys aged 6 to 18, squared Girls aged 6 to 17 Girls aged 6 to 17, squared Adult aged 18 to 60 Adult aged 18 to 60,squared Elderly aged 60 plus Elderly aged 60 plus, squared Noire municipalities 3. Demographic characteristics and ethnicity 4. Education of the head 0.00 0.50 –0.05 0.45 –0.10 0.40 –0.15 0.35 –0.20 0.30 –0.25 0.25 –0.30 0.20 –0.35 0.15 –0.40 0.10 –0.45 0.05 –0.50 0.00 Head is pypmy Head is disable Head is female None Primary Secondary 1 Secondary 2 Tertiary 5. Head sector of occupation 6. Age of the head, prediction of log welfare 0.35 0.40 0.30 0.35 Predicted log of welfare Poverty headcount (%) 0.25 0.30 after controls 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0 0 0 10 20 30 40 50 60 70 80 90 Agriculture, livestock, fisheries, forestry Mines/quarries Production/processing Construction Transport Trade/Sales Services Education/Health Administration Other services Unemployed/Inactive Age of the household head Source: Authors’ estimations using the 2011 ECOM survey. will boost their income and overall welfare. Toward has a large effect on consumption, with Pygmy house- retirement age, the likelihood of working decreases and holds having a reduction in welfare of more than 40 an individual starts relying on savings or low pension. percent versus Bantu households with otherwise simi- The results of the econometric analysis con- lar household characteristics (the coefficient estimate firm the marginalization of autochthons. Ethnicity is −0.451). The Drivers of Poverty Reduction and Inequality 47 As expected, consumption levels increase with Households from which there was a migration the education level of the household head. The better out are doing better. There are at least three factors at educated a household head is, the higher the level of con- play here. First, the simple fact of someone living in sumption of the household. For example, a household the household automatically increases the consump- having a head with tertiary education had a level of con- tion per equivalent adult, assuming of course that it is sumption about 50 percent higher than a household with not the main breadwinner, who is often the household a head who had no education or less than primary edu- head. The second factor is related to the fact that most cation. The gains associated with education are actually people migrate for economic purposes and often will probably larger than those estimates since the controls send remittances back, thus increasing the disposable include the occupation of the head, and education affects income of the original household. The third factor is occupation. If occupation were not controlled for in the related to the fact that, according to the literature (see, regression, the returns to education would be larger. The for example, Mensah and O’Sullivan 2016), there is a gains associated to spouse’s education are important as minimum requirement for a household to be able to send well. According to the literature (Agénor and Otaviano a migrant (skill of the person, minimum cash at least for 2013; Malhotra, Pande, and Grown 2003), beyond transport). Often, the poorest households are not able increasing the likelihood of adding to the household’s to meet these criteria and therefore are excluded from overall disposable income through labor market partici- the migration process. pation, the spouse’s education also equips females with relevant skills to care for children. As a consequence, 4.3 Drivers of poverty reduction female education has an overall important impact on human capital development. Most of the poverty reduction was The sector of employment of the head is also an attributable to economic growth, but important correlate of the household consumption this growth only benefitted the urban per equivalent adult. Given that agriculture is the refer- population ence category, working in most other sectors brings in a How much of poverty reduction is due to growth in gain in consumption per equivalent adult, with the gain mean expenditure and how much to redistribution? being largest for those working in administration (this As noted by Datt and Ravallion (1992) and others would include the public sector and suggests a premium (including Kolenikov and Shorrocks 2005, who also for public sector jobs). Those working in mining and consider price shifts), changes in poverty over time manufacturing also enjoy important gain in consumption can be attributed to the impact of growth (a change per equivalent adult. It is important to note that house- in the mean of household consumption per equivalent holds headed by a person who is unemployed, inactive, adult) and the impact of changes in inequality (change in agriculture, or in other services all have the same and in the distribution of household consumption). Results lowest welfare. This confirms that agriculture is purely for of such decompositions are provided in Table 4.1 and subsistence, so are the informal services in cities. Figure 4.2. Access to market has a positive relation with the Growth in mean expenditures contributed most consumption per equivalent adult. Access to markets, of the poverty reduction observed between 2005 whether open air or covered, has a strong correlation and 2011. The stories vary depending on the areas with consumption per equivalent adult, with the coef- of residence. In Brazzaville, growth and reduction of ficient estimates a little under 10 percent for both types inequality contributed equally to poverty reduction. In of markets. This is as expected, as markets are the main Pointe Noire, most of the poverty reduction was driven place where goods and services are exchanged and, con- by growth. In semi-urban areas, growth contributed sequently, wealth created. substantially to poverty reduction, but part of this was 48 Republic of Congo – Poverty Assessment Report FIGURE 4.2: Impact of Growth and Changes in Inequality on Poverty 10 5 0 –5 –10 –15 –20 –25 Brazzaville Pointe Noire Other municipalities Semi-urban Rural National Growth component Redistribution component Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. offset by increased inequality. By contrast, in rural areas, Reduction in household size and improved growth in mean consumption was negative, leading to education level contributed the most for an increase in poverty. Inequalities also increased in rural poverty reduction areas, leading to an increase in poverty as well. Various decompositions can be used to explain what The overall poor performance on inequality happened to poverty over time in terms of the con- reflects a pattern of growth that was not favorable to tribution of various sectors of the population. One poor households. Clearly, economic growth was con- such simple decomposition was suggested by Ravallion centrated in sectors that are favorable to the nonpoor. and Huppi (1991) to look at the contribution of various If the country had implemented and scaled up a large locations to poverty reduction over time. The decompo- cash transfer program, this could have tackled inequality sition is applied to the contribution of various groups challenges. The only serious social safety net program is to poverty reduction between 2005 and 2001, as shown LISUNGI, which is still in the pilot phase. As argued in in Figure 4.3 and Table 4.2, with the groups defined by the SCD (World Bank 2016a), the focus for the coming location, the education level of the household head, the years is to consolidate LISUNGI, expand for national sector of activity of the household head, and the house- coverage, and use it as a vehicle to deliver a bundle of hold size. Intra-sectoral effects capture changes in poverty interventions and specific productive, incentive-based, measures within a sector, while population shift effects and income-generating activities. capture changes in poverty that result from households mpact of Growth and Change in Inequality on Poverty TABLE 4.1: I   Brazzaville Pointe Noire Other municipalities Semi-urban Rural National Headcount in 2005 42.3 33.5 58.4 67.4 64.8 50.7 Headcount in 2011 21.6 20.3 52.8 59.7 69.4 40.9 Change in headcount −20.7 −13.2 −5.6 −7.7 4.7 −9.8 Growth component −10.4 −9.7 −3.2 −15.9 1.4 −8.6 Redistribution component −10.2 −3.5 −2.3 8.2 3.3 −1.3 Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. The Drivers of Poverty Reduction and Inequality 49 FIGURE 4.3: Sectoral Decompositions of Change in Poverty (%) 1. By region 2. By head’s education 2 0.0 1 0 –0.5 –1 –1.0 –2 –3 –1.5 –4 –5 –2.0 –6 –7 –2.5 Brazzaville Pointe Noire Other municipalities Semi-urban Rural Population- shift effect Interaction effect None Primary Secondary 1 Secondary 2 Tertiary Population- shift effect Interaction effect 3. By head’s sector of occupation 4. By household size 2 0.5 1 0.0 –0.5 0 –1.0 –1 –1.5 –2 –2.0 –2.5 –3 –3.0 –4 –3.5 –5 –4.0 Agriculture Services Not working Population- shift effect Interaction effect 1 individual 2 to 3 individuals 4 to 5 individuals 6 to 7 individuals 8 individuals and more Population- shift effect Interaction effect Mining/ Manufacturing Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. Box 4.1: Growth and inequality decomposition Changes in poverty rates can also be decomposed into changes due to economic growth (or mean income) in the absence of changes in inequality (or income distribution) and changes in inequality in the absence of growth. Denoting by P(μt , Lt), the poverty measure corresponding to a mean income in period t of μt and a Lorenz curve Lt, the decomposition is ∆P = [P(µ2, Lr) – P(µi, Lr)] + [P(µr, L2) – P(µr, L1)] + Rr | | | Growth effect Inequality effect Residual The first component is the change in poverty that would have been observed if the Lorenz curve had remained unchanged, while the second component is the change that would have been observed if mean income had not changed. The last component is a residual (Datt and Ravallion 1992), but in subsequent decompositions it has been shown that this residual can be avoided. 50 Republic of Congo – Poverty Assessment Report TABLE 4.2: Sectoral Decompositions of Change in Poverty (%)   2005 to 2011 Poverty rate in 2005   50.7   Poverty rate in 2011   40.9   By region Sector Population share in 2005 Change in poverty Percentage change Brazzaville 29.0 −6.0 61.1 Pointe Noire 23.5 −3.1 31.6 Other municipalities 5.9 −0.3 3.4 Semi-urban 7.0 −0.5 5.5 Rural 34.6 1.6 −16.4 Total intra-sectoral effect −8.4 85.2 Population-shift effect −0.5 5.6 Interaction effect −0.9 9.3 Change in poverty headcount   −9.8 100.0 By education of the head Sector Population share in 2005 Change in poverty Percentage change None 17.2 −1.6 16.6 Primary 19.7 −1.6 16.5 Secondary 1 32.5 −2.3 23.6 Secondary 2 18.5 −1.0 10.0 Tertiary 12.2 −1.8 18.2 Total intra-sectoral effect −8.3 84.8 Population-shift effect −1.4 14.0 Interaction effect −0.1 1.2 Change in poverty headcount   −9.8 100.0 By sector of occupation of the head Sector Population share in 2005 Change in poverty Percentage change Agriculture 27.7 0.4 −4.3 Mining/Manufacturing 15.1 −4.4 44.6 Services 39.4 −4.2 43.2 Not working 17.7 −1.8 17.9 Total intra-sectoral effect −9.9 101.4 Population-shift effect −1.1 11.2 Interaction effect 1.2 −12.6 Change in poverty headcount   −9.8 100.0 (continued on next page) The Drivers of Poverty Reduction and Inequality 51 TABLE 4.2: Sectoral Decompositions of Change in Poverty (%) (continued)   2005 to 2011 By household size Sector Population share in 2005 Change in poverty Percentage change 1 individual 1.5 0.0 −0.3 2 to 3 individuals 11.8 −0.4 3.8 4 to 5 individuals 25.5 −1.9 18.9 6 to 7 individuals 27.4 −2.7 27.4 8 individuals and more 33.7 −0.9 9.2 Total intra-sectoral effect −5.8 59.1 Population-shift effect −3.7 38.2 Interaction effect −0.3 2.8 Change in poverty headcount −9.8 100.0 Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. shifting from one group to another. Interaction effects are productive sectors, especially in the two main cities essentially correlation terms or remainders—they typi- (see Figure 4.4). As we will see in Chapter 5, structural cally account for a smaller share of the overall changes transformation seems to be happening in the wrong in poverty. direction. The manufacturing sector is shrinking, prob- Migration accounted for 5.6 percent of the ably a consequence of Dutch disease. As a consequence, decline in poverty. As illustrated in Figure 4.3 and a growing share of households is relying on agriculture Table 4.2, the population shift across location accounted and informal services for livelihood. In the two main for 5.6 percent of the poverty reduction. Population shift cities, there seems to be a shift toward better quality is happening in favor of Brazzaville. Between 2005 and service type of work. The poverty increase in rural areas 2011, the share of those living in Brazzaville increased seems to be related to a surprising shift of population by 8 percentage points while the share of those living out of manufacture back to agriculture. Richer data sets elsewhere decreased. Migration appears to be a channel and further investigation are needed to understand this through which poverty is shifting from rural to urban paradigm in rural areas. areas. Migration itself is not enough to improve living Gain in education contributed for 14.0 percent conditions. Other intrinsic factors, including skills and of the poverty reduction. This is a larger share than for education, are complementarily important for migration location and sector of occupation. The two main cities to be successful. Yet, those moving from rural areas to are where most of the improvement in skills took place. urban areas often lack such endowments. As a conse- In Brazzaville and Pointe Noire, the share of people liv- quence, migration did not account for a bigger share of ing in households headed by someone with primary or poverty reduction. no education declined, while the corresponding figures Changes in occupations between sectors of for secondary and tertiary education increased. On the activity accounted for 11.2 percent of poverty contrary, the share of people in households headed by reduction. Although the share of people living in a someone with no or primary education increased in household headed by someone who is unemployed rural areas from 46 to 53 percent. Thus intervention or inactive remains stable (about 18 percent) between toward curbing rural poverty should focus on, among 2005 and 2011, there was a positive shift toward more other things, skills/education beyond primary education. 52 Republic of Congo – Poverty Assessment Report TABLE 4.3: Household Size, Composition, and Dependency Ratios (%) by Location, 2005–2011 Location   Brazzaville Pointe Noire Other municipalities Semi-urban Rural Total 2005 Children, ages 0 to 14 1.78 2.01 2.20 2.21 2.12 2.01 Adult, ages 15 to 64 3.13 3.39 3.02 2.71 2.53 2.93 Elderly, ages 65 and above 0.15 0.13 0.17 0.20 0.24 0.18 Household size 5.06 5.52 5.39 5.13 4.89 5.12 Dependency 61.6 63.1 78.6 89.0 93.2 74.6 Child dependency 56.7 59.3 73.0 81.6 83.8 68.4 Aged dependency 4.8 3.8 5.6 7.5 9.4 6.2 2011 Children, ages 0 to 14 1.55 1.55 2.02 1.98 1.85 1.70 Adult, ages 15 to 64 2.64 2.64 2.56 2.25 2.08 2.43 Elderly, ages 65 and above 0.12 0.11 0.16 0.16 0.21 0.16 Household size 4.31 4.31 4.74 4.40 4.15 4.28 Dependency 63.4 62.9 85.0 95.1 98.9 76.3 Child dependency 58.8 58.6 78.6 88.0 88.7 69.9 Aged dependency 4.6 4.3 6.4 7.2 10.2 6.4 Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. Among the variables that are considered here, collected in 2005. Still, women, disabled, and autoch- the household size is the one that contributed the thons should be considered as marginalized groups, as most to poverty reduction. A reduction in average they have limited opportunity to improve their endow- household size accounted for 38.2 percent of poverty ment through education. reduction. As illustrated in Chapter 3, although fertil- ity remains high, the Republic of Congo has started its Most of the poverty reduction is explained demographic transition. As a consequence, the house- by change in returns hold size reduced by close to one person from 5.12 in It is possible to decompose the change of poverty 2005 to 4.28 in 2011. However, the growing proportion into changes in the characteristics of households and of children is maintaining the dependency ratio at higher individuals (‘endowments’) compared to the chang- levels, especially in rural areas (Table 4.3). Still, results ing nature of how these characteristics are valued on for the decomposition suggest that the shift in household the market (‘returns’). Decomposition methods can size contributed a large percentage to poverty reduction. be used to assess whether the underlying distribution of Gender and disability do not seem to matter. It is characteristics can be associated with changes in poverty because they affect welfare indirectly. Gender and disabil- or whether we observe the relationship between key char- ity each contributed to 1.5 percent of poverty reduction. acteristics and poverty changing over time. In addition Gender and disability affect the ability of individuals to to the classic Oaxaca-Blinder decomposition methods acquire the required skills and thus excluded them from that focus on differences between means, we will use the productive jobs. It was not possible to conduct a trend improved version by Firpo, Fortin, and Lemieux (2009): analysis by ethnicity as the relevant information was not the Recentered Influence Function (RIF) regressions The Drivers of Poverty Reduction and Inequality 53 FIGURE 4.4: Distribution of the Population According to Household Head’s Characteristics (%) 1. Sector of occupation of the household head 100% 7.6 5.9 7.8 10.8 17.7 13.7 11.2 18.6 90% 29.7 25.2 14.5 23.5 26.0 80% 32.5 13.3 70% 20.5 40.5 6.1 45.0 60% 39.5 47.7 44.8 50% 54.2 16.1 6.1 65.9 59.7 40% 57.8 15.5 15.1 30% 14.7 8.6 20% 10% 7.9 15.6 8.1 12.6 0% 4.6 5.0 28.7 43.8 59.1 27.7 2.6 1.7 23.9 45.7 69.4 28.0 Brazzaville Pointe Other Semi-urban Rural Total Brazzaville Pointe Other Semi-urban Rural Total Noire municipalities Noire municipalities 2005 2011 Agriculture, livestock, fisheries, forestry Production/processing Services Unemployed/inactive 2. Education of the household head 100% 4.8 3.1 14.2 8.3 12.2 16.1 8.6 14.7 90% 18.8 19.3 13.4 24.8 17.8 10.7 14.8 18.0 80% 23.0 18.5 70% 22.2 16.1 28.8 19.9 33.1 21.2 60% 27.0 36.0 27.3 32.5 32.6 50% 32.0 29.9 30.4 31.3 40% 30.6 19.6 23.3 22.2 28.5 33.2 30% 17.7 17.4 19.7 15.9 14.3 20% 16.5 19.4 9.4 13.4 10% 0% 11.9 11.4 17.4 26.6 23.6 17.2 10.0 8.6 16.0 23.2 33.5 18.6 Brazzaville Pointe Other Semi-urban Rural Total Brazzaville Pointe Other Semi-urban Rural Total Noire municipalities Noire municipalities 2005 2011 None Primary Secondary 1 Secondary 2 Tertiary 3. Household size 100% 90% 20.8 18.1 16.7 20.3 20.3 32.7 27.3 27.3 80% 37.8 36.5 33.7 42.6 70% 25.5 32.0 27.3 26.5 26.8 60% 29.3 27.4 50% 28.1 25.8 27.4 24.1 28.5 40% 33.3 35.9 30% 32.6 31.5 33.0 24.6 28.6 29.8 20% 25.4 22.3 21.8 25.5 10% 12.0 9.6 10.6 11.3 13.5 11.8 16.7 18.5 13.7 16.2 17.3 17.1 0% 1.9 1.4 1.4 1.7 1.4 1.5 2.8 2.0 1.9 2.4 3.7 2.9 Brazzaville Pointe Other Semi-urban Rural Total Brazzaville Pointe Other Semi-urban Rural Total Noire municipalities Noire municipalities 2005 2011 1 individual 2 to 3 individuals 4 to 5 individuals 6 to 7 individuals 8 individuals and more Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. approach. This approach allows decompositions to be decompositions analysis, the focus is on a counterfactual constructed at various points along the distribution, of a constant relationship between endowments and pov- which allows for a more informative analysis. In the erty in the Republic of Congo between 2005 and 2011. 54 Republic of Congo – Poverty Assessment Report Box 4.2: Sectoral decomposition Two main decompositions have been used in the literature to analyze changes in poverty over time. The first decomposition deals with shifts in poverty between sectors or groups (Ravallion and Huppi 1991). The second decomposition deals with the contribution of growth and changes in inequality to changes in poverty (Datt and Ravallion 1992; Kakwani 1997). Consider first the sectoral decomposition. Poverty measures of the FGT class are additive. This means that the poverty measure for the population as a whole is equal to the weighted sum of poverty measures for population subgroups, with the weights defined by the population shares of subgroups. This property makes it feasible to analyze the contribution of population subgroups to changes in overall poverty over time. Assume that households or individuals can be classified according to various sectors in the economy. These may be industrial sectors, geographic sectors (urban versus rural), or any other sectors that the analyst may suggest. The overall change in poverty over time can be decomposed into (a) changes in poverty within specific sectors or intra-sectoral changes, (b) changes in poverty due to changes in the population shares of sectors or inter-sectoral changes, and (c) changes due to the possible correlation between intra-sectoral and inter-sectoral changes, or interaction effect. Denote by Pit the poverty measure in sector i at time t; there are m sectors (i = 1,…, m), with population share ni in sector i and two periods (1 and 2). Then, the overall change in poverty is equal to m m m D P = ∑ ni 1 (Pi 2 − Pi 1) +∑ Pi 1 ( ni 2 − ni 1) +∑ (Pi 2 − Pi 1)( ni 2 − ni 1) . i =1 i =1 i =1 | | | Intra-sectoral Inter-sectoral Interaction effect This counterfactual is used to determine which changes differences across regions. In Brazzaville and semi- in endowments could have contributed to poverty urban areas, poverty reduction was essentially due to an reduction and how much poverty reduction could have improvement of returns. The observed deterioration of changed as a result of a changing relationship between living standards in rural areas was due to deterioration poverty and endowments. The latter is sometimes in returns. This is certainly a symptom of the Dutch dis- referred to as changes in the returns to endowments. ease. In Pointe Noire, poverty reduction was the result The results of this exercise are presented in Figure 4.5. of improvement in endowments. At the national level, 62 percent of the poverty Clearly the contrast between Brazzaville and reduction is attributable to an improvement of the Pointe Noire illustrates a situation that is mainly returns, while 38 percent of the reduction is attribut- explained by the change in household size. It is pos- able to improvement in endowments (characteristics sible that the demand for skills is larger in Pointe Noire of households and individuals). There are important and semi-urban areas, which has incentivized people to FIGURE 4.5: Decomposition of the Change of Poverty Into Endowment and Returns Effects 100% 12.9 9.8 90% 80% 70% 62.1 60.9 52.3 60% 77.0 50% 40% 30% 20% 10% 37.9 39.1 87.1 47.7 90.2 23.0 0% National Brazzaville Pointe Noire Other municipalities Semi-urban Rural Endowments Returns Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. The Drivers of Poverty Reduction and Inequality 55 FIGURE 4.6: Share of between- and Within-Group Inequality, 2011 (Theil) 100% 95% 90% 85% 80% 75% Region Head age Head is disable Head is female Head marital status Head education Head industry No spouse Spouse education Spouse industry Household size Household onw land Covered market Open-air market Within group contribu;on to inequality (%) Between group contribu;on to inequality (%) Source: Authors’ estimations using the 2011 ECOM survey. invest more in those skills. This seems to also have led Disparities among region, education level, and to a higher reduction of household size, particularly in industry of activity all contributed to an increase of Pointe Noire. In Pointe Noire, the average household size inequalities. Among the factors that do have some mea- reduced by 1.22 against 0.74 for Brazzaville. The aware- surable impact are region, head and spouse education ness on the importance to invest in children’s education, level, head and spouse industry, household size, land given the market demand, must be at play here. ownership, and access to markets. Other factors like age, disability, gender, and marital status have surpris- 4.4 Main factors explaining inequality ingly little explanatory power. Moreover, these between- group determinants—in particular, the head’s age and The decomposition of inequality enables one to marital status—contributed significantly to convergence explore how the differences in households’ charac- between 2005 and 2011 (Figure 4.7). The household size teristics affect the level of inequality and provide also had a positive effect on convergence. important clues for understanding the underlying What matters for inequality is not the group to structure of per equivalent consumption distribution. which an individual belongs but rather its endow- The decomposition follows the approach of Cowell and ment and the opportunities that are available to that Jenkins (1995) and consists of separating total inequality person. The fact that the between-group inequalities is in the distribution of per equivalent consumption into small for most factors is not surprising. For instance, inequality between the different household groups in education will increase the likelihood of someone to each partition and the remaining within-group inequality. be a wage earner. Living in the main cities increases Inequality is explained mainly by within-group opportunities to access quality jobs. The importance of differences28 than by any aggregate determinant. For most factors, the within-group inequality is by far the main explanatory factor behind welfare. Still, some key 28 Inequality is decomposed here to differences between categories factors stand out with very high between-group inequali- (for example, being in one region versus another) and differences within categories (differences among individuals in a specific region); ties (Figure 4.6). this decomposition follows the methods described in Cowell and Jenkins (1995). 56 Republic of Congo – Poverty Assessment Report FIGURE 4.7: Absolute Change in between- and Within-Group Inequality, 2005–2011 (Theil) 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000 –0.005 –0.010 Region Head age Head is disable Head is female Head marital status Head education Head industry No spouse Spouse education Spouse industry Household size Household onw land Covered market Open-air market Change within Change between Source: Authors’ estimations using the 2005 and 2011 ECOM surveys. Note: Data on ethnicity not available in the 2005 ECOM survey. household size for inequality is just a mathematical con- thus placing a burden on household disposable income. sequence of the fact that the welfare indicator decreases Issues related to responsible parenthood should be part with an additional household member. Other factors of the poverty alleviation interventions. might still be indirectly important for inequalities as they There is no apparent difference in terms of wel- can be barriers for certain groups of people in acquiring fare between male- and female-headed households. the necessary endowments. In the end, human capital is One of the important findings here is that the gender what actually matters. of the head does not matter for the overall household welfare. Female-headed households are doing as well 4.5 Conclusion (or as poorly, depending) as male-headed households. Robustness check across residential areas suggests that This chapter uses decomposition and econometric this finding holds for both rural and urban areas. techniques to provide a robust analysis of the drivers of Econometric analysis confirms the marginaliza- poverty reduction and inequality. tion of autochthons. Ethnicity has a large effect on con- Location is strongly correlated with welfare. sumption, with Pygmy households having a reduction in Households living in Pointe Noire have higher levels welfare of more than 40 percent versus Bantu households of consumption than those living in Brazzaville, while with otherwise similar household characteristics. households in other urban areas and in rural areas have As expected, consumption levels increase with lower levels of consumption. For example, in 2011, esti- the education level of the household head. The bet- mations suggest a reduction in consumption of close to ter educated a household head is, the higher the level of a quarter (−0.223) for rural households in comparison consumption of the household. For example, a house- to otherwise similar households living in Brazzaville. hold having a head with tertiary education had a level Larger households tend to have substantially of consumption about 50 percent higher than a house- lower levels of consumption per equivalent adult, hold with a head who had no education or less than which is not surprising. Larger families tend to have primary education. The gains associated with education more children. As a result, the dependency ratio is higher, are actually probably larger than those estimates since The Drivers of Poverty Reduction and Inequality 57 the controls include the occupation of the head, and seems to be happening in the wrong direction. The education affects occupation. If occupation were not manufacturing sector is shrinking, probably a conse- controlled for in the regression, the returns to education quence of Dutch disease. As a consequence, a grow- would be larger. ing share of households is relying on agriculture and The sector of employment of the head is also informal services for livelihood. In the two main cities, an important correlate of the household consump- there seems to be a shift toward better quality service tion per equivalent adult. Given that agriculture is the type of work. The poverty increase in rural areas seems reference category, working in most other sectors brings to be related to a surprising shift of population out of in a gain in consumption per equivalent adult, with manufacture back to agriculture. Richer data sets and the gain being largest for those working in administra- further investigation are needed to understand this tion (this would include the public sector and suggests paradigm in rural areas. a premium for public sector jobs). Those working in Gain in education contributed for 14.0 per- mining and manufacturing also enjoy important gain cent of the poverty reduction. The two main cities are in consumption per equivalent adult. It is important to where most of the improvement in skills took place. In note that households headed by a person who is unem- Brazzaville and Pointe Noire, the share of people living ployed, inactive, in agriculture, or in other services all in households headed by someone with primary or no have the same and lowest welfare. This confirms that education declined, while the corresponding figures agriculture is purely for subsistence, so are the informal for secondary and tertiary education increased. On the services in cities. contrary, the share of people in households headed by Migration accounted for 5.6 percent of the someone with no or primary education increased in decline in poverty. The population shift across loca- rural areas, from 46 to 53 percent. Thus, intervention tion accounted for 5.6 percent of the poverty reduction. toward curbing rural poverty should focus on, among Population shift is happening in favor of Brazzaville. other things, skills/education beyond primary education. Between 2005 and 2011, the share of those living in Among the variables that are considered here, Brazzaville increased by 8 percentage points while the the household size is the one that contributed the share of those living elsewhere decreased. Migration most to poverty reduction. A reduction in average appears to be a channel through which poverty is shift- household size accounted for 38.2 percent of poverty ing from rural to urban areas. Migration itself is not reduction. Although fertility remains high, the Republic enough to improve living conditions. Other intrinsic of Congo has started its demographic transition. As a factors, including skills and education, are comple- consequence, the household size reduced by close to one mentarily important for migration to be successful. Yet, person, from 5.12 in 2005 to 4.28 in 2011. However, those moving from rural areas to urban areas often lack the growing proportion of children is maintaining the such endowments. As a consequence, migration did not dependency ratio at higher levels, especially in rural areas. account for a bigger share of poverty reduction. Gender and disability do not seem to matter Changes in occupations between sectors of for poverty reduction. It is because they affect welfare activity accounted for 11.2 percent of poverty indirectly. Gender and disability each contributed to reduction. Although the share of people living in a 1.5 percent of poverty reduction. Gender and disability household headed by someone who is unemployed affect the ability of individuals to acquire the required or inactive remains stable (about 18 percent) between skills and thus excluded them from productive jobs. Still, 2005 and 2011, there was a positive shift toward more women, disabled, and autochthons should be considered productive sectors, especially in the two main cities. as marginalized groups, as they have limited opportunity As we will see in Chapter 5, structural transformation to improve their endowment through education. 58 Republic of Congo – Poverty Assessment Report Employment and Main Income Sources 5 5.1 Introduction For most households in poverty, inability to meet their basic needs reflects difficulties faced by the members in accessing quality jobs. The close relationship between welfare and access to quality jobs is very clear to the population. As shown in Table 5.1, lack of employment is the most cited factor in household responses to a question in the 2011 ECOM survey about the causes of TABLE 5.1: Main Causes of Poverty According to the Population Location Brazzaville Pointe Noire Other municipalities Semi -urban Rural Total Lack of employment 89.8 96.3 95.4 93.7 90.5 91.7 Lack of education 30.4 51.9 51.8 64.8 52.4 44.7 Lack of connectivity 27.5 56.8 60.3 61.0 69.5 50.8 Witchcraft or laziness 23.5 33.2 38.2 40.3 45.6 34.5 Poor governance 57.1 68.8 73.0 74.3 63.1 62.9 Corruption 48.1 59.9 62.6 67.4 47.7 51.8 Insufficient income 52.4 64.3 63.2 67.4 61.4 59.0 Other 5.3 15.9 9.9 15.2 15.6 11.5 Welfare quintile Q1 Q2 Q3 Q4 Q5 Total Lack of employment 90.6 93.1 92.2 93.4 89.6 91.7 Lack of education 51.8 47.0 43.1 43.4 41.2 44.7 Lack of connectivity 66.6 53.8 52.4 43.8 44.3 50.8 Witchcraft or laziness 39.9 33.9 33.1 31.1 35.7 34.5 Poor governance 61.2 61.0 61.8 63.4 65.7 62.9 Corruption 50.5 50.0 54.5 49.0 54.0 51.8 Insufficient income 61.1 60.1 62.0 56.2 57.0 59.0 Other 14.6 12.1 10.3 11.1 10.6 11.5 Source: Authors’ calculation using the 2011 ECOM survey. 59 poverty. Lack of employment was cited as a leading cause and Patrinos 2004). Yet, while the literature is rich, esti- of poverty by 91.7 percent of respondents and insufficient mates for countries such as the Republic of Congo tend income, a closely related effect, was cited by 59.0 percent. to be scarce. This chapter aims to fill this gap. Other factors play a role as well but less so than employ- This chapter will use several sources of data to ment. Just under two-thirds (62.9 percent) of households present a detailed picture of the evolution of poverty also identified poor governance as a major factor leading to and household living standards in the Republic of poverty, with corruption cited by half of the respondents Congo, with a focus on the period for which house- (51.8 percent). Lack of connectivity (50.8 percent), lack of hold surveys are available, that is between 2005 and education (44.7 percent), and even witchcraft or laziness 2011. The chapter has six parts. In the first part we will (34.5 percent) were also cited. Perceptions about the main provide basic statistics on employment patterns for the causes of poverty tend to be similar across groups, even population. The second part will look at sectors of activity if for the very poor (bottom quintile), lack of education and type of occupation of the employed population. The and connectivity play a larger role. third part will present the earnings profile. The fourth In a country such as the Republic of Congo part will use multivariate regressions to provide correlates which is dominated by oil production, GDP growth of the probability of wage employment and the level of does not necessarily lead to more employment or bet- wages earned when employed, with a focus on returns to ter employment. Despite substantial growth during the education. The last part will present a basic profile as well last ten years, the employment structure of the country as correlates of income sources at the household level. has not changed fundamentally, but this could change in the future. Similar to many other African countries, the 5.2 Employment status Republic of Congo has a relatively young population and education attainment is rising rapidly (even if learning A key objective of this chapter is to provide basic performance remains low).29 If the Republic of Congo statistics on employment patterns for the population continues to grow and manages to diversify its economy, and how those patterns may have changed between the industrial and service sectors could grow at higher 2005 and 2011. This is done using two different mea- rates. This in turn would require a skilled workforce. sures of unemployment based on the survey data. For This means that education and skills—including digital the first definition, unemployment is restricted in the competencies—are becoming increasingly important for traditional sense to the subset of individuals who are not individuals to obtain good jobs and earn decent wages, employed and are looking for a job. There are, however, as well as for countries to compete internationally. many discouraged workers in the survey who state that In this context, the objective of this chapter is they are not looking for a job because there are no jobs threefold: to describe trends in employment patterns, available, they are not qualified, they do not know where provide an analysis of households’ main income to look for a job, or they are waiting for the outcome of sources, and estimate the returns to education for their application. These are typically only temporarily the population. There is a large volume of literature on inactive. Therefore, for the second definition of unem- the returns to education. As noted by Montenegro and ployment, discouraged workers are also included in the Patrinos (2014), estimates of these returns have been a unemployed category.30 mainstay of the education literature for at least 40 years, Overall, differences in employment status with hundreds of studies and many reviews conducted between men and women tend to be less marked in the over the years (these reviews include Banerjee and Duflo Republic of Congo than in some other SSA countries. 2005; Colclough et al. 2010; Harmon et al. 2003; Psacharopoulos 1972, 1973, 1985, 1989, 1994, 1995; 29 See Chapter 6 for more details on the education sector. Psacharopoulos and Layard 1979, 2012; Psacharopoulos 30 See annex 7 for more details. 60 Republic of Congo – Poverty Assessment Report Whether one uses the first or second definition of unem- still in school at that age. For those between 50 and 64 ployment, there are important life cycle effects at work years of age, there is a small drop in employment share in terms of who is employed, unemployed, or inactive. versus those in the 30 to 49 age bracket. Unemployment, These life cycle effects are illustrated in Figure 5.1, where as traditionally defined, has increased between the two the age of the individual is on the horizontal axis. Both years according to the first definition despite the fact panels in the figure rely on the first, traditional defini- that poverty has been reduced. tion of the unemployed, but qualitatively, the results Based on the first definition of unemployment, would not be fundamentally different from the second it seems that the impressive economic growth was job- definition in terms of age profiles by gender. Younger less or at least did not have a significant impact on individuals are obviously more likely to be studying, but unemployment, especially for the youth. This is not one should note the substantial share of inactive and surprising given that growth was driven by the oil sector unemployed youth. Among older individuals, inactivity which is not labor intensive and has limited linkages with starts to rise after the age of 54 and even more so after the rest of the economy. The trend in the employment the age of 59. When comparing men and women, it profile suggests that the economic growth was jobless. can be seen that the proportion of women employed is This is not surprising given that this economic growth smaller than that of men. In addition, among relatively was essentially driven by the oil economy. Moreover, young adults such as those between the ages of 25 and these findings also suggest that the massive infrastructure 29, unemployment is higher among women than men. program under the ‘municipalization’ slogan also failed, Unemployment seems to have increased between at least between 2005 and 2011, to create substantial 2005 and 2011, which could represent a puzzle to and enough jobs to have a consequential impact on the understand trends in well-being. Figure 5.2 relies on labor market. This is not surprising either as most of the first definition to display a basic employment profile the construction work is just creating temporary jobs. not only according to age but also by education level, Looking at level, it is clear that the economy ethnicity, disability status, gender, and location of indi- actually creates jobs, but these were offset by the pop- viduals. Individuals are categorized as employed, unem- ulation growth. Between 2005 and 2011, the number ployed, or inactive. As expected, employment is lower of employed increased by 180,000. In the meantime, the in the bottom age group (ages 15–29) because many are potentially active population (ages 15–64) increased by FIGURE 5.1: Employment Status by Gender and Age, 2011 1. Men 2. Women 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 10–14 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70/max 10–14 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70/max Inactive Student Employed Unemployed Source: Authors’ calculation using the 2011 ECOM survey. Employment and Main Income Sources 61 FIGURE 5.2: Employment Profile with First Definition of Unemployment, 2005–2011 1. Age 2. Education 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 15–29 30–49 50–64 None Primary Secondary 1 Secondary 2 Tertiary Age group Education 3. Ethnicity 4. Disability 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Bantou Pygmy 2005 2011 2005 2011 Ethnicity Other Disabled Disability 5. Gender 6. Location 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 Men Women Brazzaville Pointe Noire Other municipalities Semi- urban Rural Gender Region Employed Unemployed Inactive Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. 62 Republic of Congo – Poverty Assessment Report 264,000. Population growth appears to be a challenge, Unemployment rate is very low among autoch- as the country has to not only create jobs, but they must thons. When looking at ethnicity, it is clear that the be enough to at least match the fast-growing working- autochthon populations, while poorer, are much more age population. likely to work, simply to survive. Moreover, when looking at the same statistics Disability is associated with a lower likelihood according to the second definition, the changes in to be employed and a higher rate of inactivity. The unemployment are much smaller between the two disabled seem to be marginalized on the labor market. years. In other words, the increase in unemployment Unemployment rate for disabled individuals is 29.6 per- as traditionally defined between 2005 and 2011, in cent, which is 10 percentage points higher compared to Figure 5.2, seems to reflect at least in part the fact that the remaining population. many discouraged workers may have returned to the There is little difference across gender. There labor market and were looking for employment in 2011, seems to be little or no difference between men and while this was not the case in 2005. Under the second women in terms of unemployment. However, a clear definition (Figure 5.3), these workers are included in the look at the second definition suggests that if one takes unemployed category, while under the first definition, into account discouraged workers, then unemployment they are not. This would then suggest a better labor mar- rate will be significantly higher for women. This suggests ket in 2011 in comparison to 2005, which would be in that most women are more likely to get discouraged and line with the gains in poverty reduction between the two drop out of the labor force. years. Since the estimates under the second definition Differences in employment patterns are larger by tend to better reflect trends in the labor market, the rest location, with the lowest employment rates observed of the discussion will focus mostly on findings obtained in the largest cities. Probably for a number of reasons, under the second definition. young adults may still be studying, and more individu- The youth are severely affected by unemploy- als can afford not to work if they have high reservation ment. If we take the second definition for example, the wages; in rural areas by contrast, the pressure imposed unemployment rate is 32.7 percent for those who are 15 by poverty makes employment a necessity. to 29 years of age. The corresponding figures for those ages The pressure of poverty tends to result in higher 30–49 and 50–64 are 15.6 and 8.3, respectively. Therefore, employment in the bottom quintiles, as the poor typi- programs focusing on youth employment should be pro- cally cannot afford any prolonged period of unem- moted and scaled up as much as possible, given the impor- ployment. When looking at the quintiles of well-being tance of the youth in the overall population. in Figure 5.4, gaps in employment, unemployment, and Rates of inactivity are higher among those with inactivity rates are less pronounced than when looking mid-level education. This could reflect the fact that at other dimensions as various factors, some leading to a those with no education at all simply have no choice higher likelihood of employment and others to a lower but to work, while those with a tertiary education tend likelihood, may offset each other. to have access to good jobs and thereby also work in Clearly, unemployment is higher for younger higher proportions. At the same time, other effects could populations, the better educated, Bantus, individuals also be at work, in that different age cohorts have dif- with a disability, women, and those living in the larg- ferent education levels on average. Unemployment was est cities. While unemployment rates may be computed higher, especially in 2011, among the better educated, from the data in Figures 5.2 to 5.4 by comparing the suggesting that reservation wages among those groups employed population to the combined population of the may lead to some not working unless they find a fairly employed and unemployed, it is easier to visualize unem- good job (Figure 5.3). For the less educated and poorer ployment rates directly. This is done in Figure 5.5 under individuals, this may not be a luxury. the same two definitions of unemployment provided Employment and Main Income Sources 63 FIGURE 5.3: Employment Profile with Second Definition of Unemployment, 2005–2011 1. Age 2. Education 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 15–29 30–49 50–64 None Primary Secondary 1 Secondary 2 Tertiary Age group Education 3. Ethnicity 4. Disability 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Bantou Pygmy 2005 2011 2005 2011 Ethnicity Other Disabled Disability 5. Gender 6. Location 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 Men Women Brazzaville Pointe Noire Other municipalities Semi- urban Rural Gender Region Employed Unemployed Inactive Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. 64 Republic of Congo – Poverty Assessment Report FIGURE 5.4: Employment Profile by Quintile, 2005–2011 1. First definition of unemployment 2. Second definition of unemployment 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Welfare Quintile Welfare Quintile Employed Unemployed Inactive Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. FIGURE 5.5: Unemployment Rate by Selected Characteristics, 2005–2011 1. Age 2. Education 45 35 40 30 35 25 30 25 20 20 15 15 10 10 5 5 0 0 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 15–29 30–49 50–64 None Primary Secondary 1 Secondary 2 Tertiary Age group Education 3. Ethnicity 4. Disability 25 35 30 20 25 15 20 10 15 10 5 5 0 0 Bantou Pygmy 2005 2011 2005 2011 Ethnicity Other Disabled Disability 5. Gender 6. Location (continued on next page) 30 45 40 25 35 20 30 Employment and Main Income Sources 65 25 15 20 10 15 10 5 10 5 5 0 0 Bantou Pygmy 2005 2011 2005 2011 Ethnicity Other Disabled FIGURE 5.5: Unemployment Rate by Selected Characteristics, 2005–2011 (continued ) Disability 5. Gender 6. Location 30 45 40 25 35 20 30 25 15 20 10 15 10 5 5 0 0 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 Men Women Brazzaville Pointe Noire Other municipalities Semi- urban Rural Gender Region 7. Welfare quintiles 30 25 20 15 10 5 0 2005 2011 2005 2011 2005 2011 2005 2011 2005 2011 Q1 Q2 Q3 Q4 Q5 Welfare Quintile Definition 1 Definition 2 Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. earlier. Clearly, unemployment is highly correlated to shrinking—probably consequences or symptoms of age, education, gender, disability status and location. the Dutch disease. There have been shifts in the sectors In addition, the results also suggest that unemployment of activity of the population between the two years. rates rise slightly with household welfare, although in Figure 5.6 shows the distribution of the employed popu- 2011, according to the second definition, the unemploy- lation by sector and by ventile of welfare (each ventile ment rate in the bottom quintile was high. accounts for 5 percent of the national population, from the poorest to the richest). As shown in the figure, the 5.3  Sector of activity and type of manufacturing sector appears to be shrinking in favor occupation of services and agriculture. The shifts between the two years seem large for a relatively short period, and there Structural transformation is happening in the may be comparability issues between the two years. wrong way. The manufacturing sector seems to be At the same time the direction of the shifts away from 66 Republic of Congo – Poverty Assessment Report FIGURE 5.6: Distribution of Employment by Sector, 2005–2011 1. 2005 distribution 2. 2011 distribution 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Welfare ventiles Welfare ventiles Agriculture Mining/Manufacturing Services Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. manufacturing, if confirmed with other data, could be of employer for various subgroups of the population. It a cause of concern. Overall, 55 percent of the workforce seems that the vulnerable are often excluded from formal worked in services in 2011, with 38 percent engaged jobs and must rely on the informal sector by setting up in agriculture and only 7 percent in manufacturing. their own businesses or subsistence agriculture. Mining accounts for less than 1 percent of the overall Access to formal wage jobs is limited for the labor force. While an analysis of sectoral employment youth. They are more likely to be self-employed or shares is beyond the scope of the present study, it would employed by a household as family helper. As we saw be worth investigating the factors leading to shifts. in the previous section, the youth are more likely to be The formal sector, either public or private, has unemployed. Moreover, even when they manage to find failed to create quality jobs for the population. As a job, it is likely to be self-employment as is the case for a consequence, the vast majority of the labor force is the whole population. On the other hand, youth, ages employed on their own account. Slightly more than three 15–29, are more likely to be employed by a household out of five workers (63 percent) are self-employed, run- (15 percent) or by SMEs (12 percent). Thus, a policy ning a business with no employees or being involved in toward supporting the emergence of strong SMEs could subsistence agriculture, and another quarter of workers be an indirect way to insure the creation of quality jobs work for a household without pay (Figure 5.7). The pub- for the youth. lic administration is the main provider of formal wage On the other hand, the elderly, ages 50 to 64, jobs. One out of ten workers (14 percent) is employed are more likely to be in civil service. As is the case for in the public administration or by a parastatal firm. The the population as a whole, most of the elderly work in role of the private sector in providing jobs to the popula- their own businesses or own a farm. In addition, those tion is not negligible. Up to 13 percent of workers are between the ages of 50 and 64 are more likely to be employed either in large private firms (5 percent) or in employed as civil servants (Figure 5.7.1). These results small and medium enterprises (SMEs) (8 percent). are not surprising given the freezing of hiring and the There are important differences in the type of introduction of a quota system, which makes it harder employer across age, education level, ethnicity, dis- for the youth to enter public service. ability status, gender, location, and welfare quintile. Autochthons seem to have been excluded The analysis suggests a lot of differences in terms of type from civil service. This might be related to their low Employment and Main Income Sources 67 FIGURE 5.7: Distribution of Workers by Type of Employer, 2011 1. Age 2. Education 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 15–29 30–49 50–64 None Primary Secondary 1 Secondary 2 Tertiary Age group Education 3. Ethnicity and disability 4. Gender 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Bantou Pygmy Other Disabled Men Women Ethnicity Disability Gender 5. Location 6. Welfare quintile 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Brazzaville Pointe Other Semi- Rural Q1 Q2 Q3 Q4 Q5 Noire municipalities urban Welfare Quintile Region Public administration Public firm/parastatal Large private company Small and medium enterprise (SME) Association, Cooperative, Church, NGOs International Organization, Embassy Household Own account Source: Authors’ calculation using the 2011 ECOM survey. endowment in terms of skills. Autochthons are either by households or on their own account. Although employed on their own account (77 percent) or by a differences in terms of type of employer by disability household (21 percent) (Figure 5.7.3). status are present, they are less pronounced. In addi- The disabled are less likely to be employed by tion, SMEs tend to discriminate against the disabled SMEs and are slightly more likely to be employed (Figure 5.7.3). 68 Republic of Congo – Poverty Assessment Report C o m p a re d t o m e n , w o m e n s e e m t o h a ve The employment profile by type of provider more difficulties accessing formal wage jobs. A bit along the welfare distribution mirrors the educa- more than seven out of ten women are self-employed tion. Those in the top quintiles are more likely to (75 percent) against only five out of ten men (52 per- be employed by the formal sector (public or private cent) (Figure 5.7.4). Employers in the formal sector firm, SMEs), while those in the bottom quintile are (public or private) seem to have a preference for or bias more likely to be employed by a household or are self- in favor of men. employed. There is a very clear correlation between As expected, location does matter. Jobs from the welfare quintile and the type of employer. Individuals public sector, large private firms, and SMEs are more from the top quintiles are more likely to be employed likely to be available for those living in the two main by formal type of providers. One can assume that these cities. The biggest share of workers in the public sector jobs are not only more secure but provide adequate is observed for the capital city of Brazzaville. This is salary. The next section will look into earnings by type not surprising given that this is the center of the public of providers in detail. On the other hand, the poor are institutions. On the other hand, the share of workers more likely to rely on employment from a household or employed by large private firms or by SMEs is higher self-employment (Figure 5.7.6). Overall, the distribu- for Pointe Noire. These results are also quite intuitive tion of workers across socio-professional categories is given that Pointe Noire is the economic capital. With as expected. About 12 percent are senior managers, 31 10 percent of workers, SMEs are also important in percent are mid-level managers, 30 percent are skilled Brazzaville (Figure 5.7.5). In the remaining locations, workers, 11 percent are unskilled workers, 11 percent self-employment and households are by far the main are laborers, and 5 percent are employers. However, it job ‘providers’. Still, even in the two main cities, half of should be noted that 73 percent of workers are skilled; the workers are self-employed. this is more proof of the importance of education. The education level of an individual is strongly Thus, programs to boost youth employment must correlated with the likelihood of being employed have a strong component on skills, vocational training by the public administration, large private firms, among others. or SMEs. Education makes a difference, starting from Age, education, ethnicity, disability, gender, upper secondary. Education seems to matter a lot in location, and welfare are also correlated to the socio- terms of providing the required skills to access quality professional category. As it was the case for the type of jobs in the formal sector. In the Republic of Congo, it employer, these socio-demographic characteristics seem really makes an important difference when one reaches to have a clear relationship with the socio-professional upper secondary. Up to 21 percent of those with upper occupation of those who are employed. Figure 5.8 pres- secondary education are employed in the public admin- ents the distribution of workers by socio-professional istration, and 13 percent of them are employed by category. To have a clear picture, we have excluded SMEs (Figure 5.7.2). These numbers are even higher those working on their own account or for a household, for those with tertiary education. In fact, 46 percent apprentices, or family helpers from this part of the analy- of those with tertiary education are employed in the sis. By doing so, we are increasing the likelihood that this public administration. Moreover, 20 percent of them part of the analysis focuses mainly on the formal sector. are employed either by large private firms or by SMEs. Youth are more likely to be employed as skilled/ Only 19 percent of those with tertiary education are unskilled workers or laborers, while the elderly are self-employed. On the contrary, up to 81 percent of more likely to be employed as managers. As illustrated those with no education are self-employed. Education in Figure 5.8.1, age does matter for the type of position then appears as a clear barrier or key asset to entering that workers will have in a job. The elderly are more the quality job market. likely to serve as managers. This is not surprising given Employment and Main Income Sources 69 FIGURE 5.8: Distribution of Workers by Socio-Professional Category, 2011 1. Age 2. Education 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 15–29 30–49 50–64 None Primary Secondary 1 Secondary 2 Tertiary Age group Education 3. Ethnicity and disability 4. Gender 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Bantou Pygmy Other Disabled Men Women Ethnicity Disability Gender 5. Location 6. Welfare quintile 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Brazzaville Pointe Other Semi- Rural Q1 Q2 Q3 Q4 Q5 Noire municipalities urban Welfare Quintile Region Senior manager/Engineer Middle manager/Mastering agent Employee skilled Employee unskilled worker (s) Laborer Employer Source: Authors’ calculation using the 2011 ECOM survey. Note: We have excluded those working on their own account, as apprentices, or as family helpers. By doing so, we are trying to focus on formal employment only, although it is not a perfect method. the experience required to get such a position. These no education or primary education are more likely to results are also as expected. work as unskilled workers or laborers. Up to 80 percent The level of education also matters; those with of those with tertiary education are employed as manag- high-level education are more likely to work as man- ers (Figure 5.8.2). There is also an important proportion agers or skilled workers. On the other hand, those with (40 percent) of those with upper secondary education 70 Republic of Congo – Poverty Assessment Report that work as either managers or skilled workers. Once (Pointe Noire), and intensity of agriculture and other again, upper secondary education appears to be the rural activities (for the rural sector). threshold beyond which education actually matters for As expected, there is a strong correlation accessing quality jobs. Close to half of those with no edu- between household welfare and the socio-professional cation or primary education only work as unskilled work- category of its members. As illustrated in Figure 5.8.6, ers or laborers. Still, there is a non-negligible proportion the nonpoor are more likely to be managers. Meanwhile, of those with no education who are either employed as the poor have higher probability to be laborers or skilled workers or managers. Favoritism, experience, and unskilled workers. There is no clear distribution of skilled other factors could be at play there. workers across welfare quintiles. There is no autochthon who has the rank of manager or even skilled worker in the country. 5.4 Earnings profile Autochthons appear to be excluded from quality jobs. As illustrated in Figure 5.8.3, none of the autochthons has For poverty reduction, more than employment per a position beyond unskilled worker. They are limited to se, it is the level of earnings of individuals that mat- unskilled worker and laborer levels. Education could be ters. Unfortunately, data on earnings at workers levels the main factor behind this. Primary school enrollment were collected only in 2011. It is therefore not possible and completion are very low among autochthons.31 As a to conduct trends analysis on wages levels. Figure 5.9 consequence, they are not equipped with the minimum provides data on employment shares as well as earnings skills to access quality jobs. according to the age, education level, ethnicity, dis- Workers with disability present a mixed situa- ability, gender, and location status of individuals. The tion with regard to the socio-professional occupation. statistics on employment shares were already discussed, Compared to others, they are more likely to be senior so the focus here is on earnings. Both mean and median managers or laborers. Such findings depict the situation earnings for each group are provided (means may be where the disabled face difficulties in attending schools, affected by extreme values, hence medians tend to be but once they are enrolled, they are likely to perform well. more appropriate for general comparisons). This is confirmed with econometric analysis of school As expected, earnings increase with age (and enrollment and dropout by level. implicit experience) and with education. Education Female workers have a higher probability of affects earnings only when one reaches secondary level. being employed as mid-level managers. Gender seems The gains appear only as of secondary education since to matter as well. Up to 40 percent of female workers there few differences in earnings between those with less are mid-level managers, against only 27 percent for their than primary education and those who completed pri- male counterparts. However, female workers are less mary school but did not pursue their education further. likely to be senior managers or laborers. Autochthons have lower earnings than Bantus. There is also a relationship between location and If we look at the median, on average, autochthons earn the socio-professional category. In Brazzaville, there is XAF 10,000 per month while the others earn five times higher probability for workers to be managers. In Pointe more, which is XAF 50,000 monthly. Noire, workers have more chance to be employed as skilled workers. In other municipalities, the predomi- nant category is the one of mid-level managers. Finally, in rural areas, workers are more likely to be employed as laborers. This distribution is as expected and has to 31 Our own estimation using the 2011 ECOM survey shows that primary net enrollment is only 46 percent for autochthons against do with availability of public jobs (for Brazzaville and up to 89 percent for Bantus. Moreover, net secondary school enroll- other municipalities), the intensity of the private sector ment is 2.6 percent for autochthons against 59 percent for Bantus. Employment and Main Income Sources 71 FIGURE 5.9: Employment Rate and Earnings Profile, 2011 1. Age 2. Education 90 140,000 90 200,000 80 80 180,000 120,000 70 70 160,000 Mothnly wage (FCFA) Mothnly wage (FCFA) Share employed (%) Share employed (%) 100,000 140,000 60 60 80,000 120,000 50 50 100,000 40 60,000 40 80,000 30 30 60,000 40,000 20 20 40,000 10 20,000 10 20,000 0 0 0 0 15–29 30–49 50–64 None Primary Secondary 1 Secondary 2 Tertiary Age group Education 3. Ethnicity 4. Disability 100 100,000 61 120,000 90 90,000 60 80 80,000 59 100,000 Mothnly wage (FCFA) Mothnly wage (FCFA) Share employed (%) Share employed (%) 70 70,000 58 57 80,000 60 50,000 56 50 60,000 60,000 55 40 40,000 54 30 30,000 40,000 53 20 20,000 52 20,000 10 10,000 51 0 0 50 0 Bantou Pygmy Other Disabled Ethnicity Disability 5. Gender 6. Location 64 140,000 90 120,000 120,000 80 62 100,000 70 Mothnly wage (FCFA) Mothnly wage (FCFA) Share employed (%) Share employed (%) 100,000 60 60 80,000 80,000 50 58 60,000 60,000 40 56 30 40,000 40,000 20 54 20,000 20,000 10 52 0 0 0 Men Women Brazzaville Pointe Other Semi- Rural Noire municipalities urban Gender Region Employed Mean Median Source: Authors’ calculation using the 2011 ECOM survey. 72 Republic of Congo – Poverty Assessment Report Those with a disability have similar level of senior managers and engineers are at the top (median earnings as those without. The median earnings of the range between XAF 116,000 and XAF 150,000). The disabled are exactly the same as the earnings for those big difference between the mean and the median for without disability. Moreover, on average, the disabled those in services and mining reflects the existence of tend to earn a bit more. The challenge for the disabled large inequalities in these sectors. is to find a job. Once they are get one, their salary is similar to the others.  age regressions and returns to 5.5 W Women earn less than men. There is an impor- education tant gender gap on the labor market, with men earning about twice more than women. On average, women earn Multivariate regressions are useful to look at the cor- XAF 60,000 per month, while men earn double that, relates of the probability of wage employment and XAF 119,000. A similar gap remains if one considers the level of wages earned when employed. Two types the median. This may be related to the type of activity of regressions can be used. One can estimate the corre- than a clear discrimination. The next section will explore lates of (the logarithm of ) wages only on the sample of this gender gap issue a bit more using econometric those with a positive wage using Ordinary Least Squares approaches. (OLS) models. However, if one wants to estimate wage Those in Brazzaville and Pointe Noire earn more regressions that are representative of the sample as a than those in other parts of the country. This is not whole, one can also rely on the Heckman model with very surprising given that these are the two main cities. a first equation for the probability of wage work and a Moreover, earnings are slightly higher in Pointe Noire, second for the (logarithm of ) wages earned. In addition, the economic capital, compared to Brazzaville. Median one can estimate those regressions for hourly wages or earnings in Pointe Noire stand at XAF 75,000, against monthly wages. Finally, different types of specifications XAF 70,000 in Brazzaville. The corresponding figures can be used, with a smaller or larger set of controls. Each for other locations are much lower: XAF 45,000, XAF of the methods will generate different results, but if the 30,000, and XAF 20,000 for other municipalities, semi- different models generate relatively stable coefficients, urban areas, and rural areas, respectively. some confidence can be placed in the results as being Earnings are strongly and positively correlated robust to the specifications. with welfare quintiles. In Figure 5.10, data is provided Two different specifications are considered. The by welfare quintiles, with individual-level earnings baseline specification includes, as occupation controls, increasing monotonically with the quintiles of house- only the sector of employment of the individual in the hold consumption per equivalent adult. For instance, wage regression, while the more complete specification the median earnings for those in the fifth quintile is 4.5 also adds the type of employer and the socio-professional times higher compared to those in the poorest quintile: category. Table 5.2 provides estimates of the marginal XAF 90,000 against XAF 20,000. effects of education on both the probability of wage Those employed in public administration and work and the level of earnings using a baseline and a large companies are the best paid, while those work- more complete specification (details are provided in ing in agriculture or self-employed have the lowest the statistical appendixes [Tables SA.4 and SA.5]) with earnings. The administration and mining sectors pay both OLS and the Heckman model. The more variables the highest salaries (median range between XAF 100,000 are included as controls, the more the marginal effect of and XAF 120,000), while those working in agriculture education is reduced, but it should be kept in mind that have the lowest pay (median of XAF 20,000). With a education affects the type of employment of individuals, median of XAF 30,000, self-employed, family helpers, so in essence, specifications with many controls under- and apprentices are at the bottom of the pay scale, while play the importance of education. The identification of Employment and Main Income Sources 73 FIGURE 5.10: Earnings by Occupation and by Quintile, 2011 1. Type of Employer 2. Sector of Activity 180,000 250,000 160,000 140,000 200,000 120,000 100,000 150,000 80,000 100,000 60,000 40,000 50,000 20,000 0 0 Own account Household Association, Cooperative, Church, NGOs Small and medium enterprise (SME) Large private company Public administration Agriculture Trade/Sales Other services Construction Education/Health Transport Services Administration Mines/quarries Public firm/ parastatal Production/ processing 3. Socio-Professional Category 4. Welfare Quintiles 250,000 65 140,000 64 63 120,000 Mothnly wage (FCFA) 200,000 Share employed (%) 62 100,000 150,000 61 60 80,000 59 60,000 100,000 58 57 40,000 50,000 56 20,000 55 0 54 0 Q1 Q2 Q3 Q4 Q5 Apprentice Working on own account Family helper Employee unskilled worker Laborer Employee skilled Employer Middle manager/ Mastering agent Senior Manager/ Engineer Welfare Quintile Employed Mean Median Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. Note: For the type of employer, earnings for those working for international organizations and embassies are not displayed in the figure because the earnings are much higher and would affect the readability of the data. the Heckman model is achieved, as is common practice, The better educated an individual is, the higher through the use of demographic variables (number of the level of earnings of the individual. At the same infants and children, essentially). time, as already noted earlier, the marginal effect of edu- Positive gains from education are observed only cation on earnings after controlling for type of employ- as of secondary school, and they increase thereafter ment is substantially significant for those reaching upper with the level of schooling of the individual. Overall, secondary and tertiary education (see Table 5.2).This has the results seem robust across specifications. As hinted implications for the development of a national strategy by the basic descriptive statistics provided earlier in for education. The objective should be to provide educa- Figure 5.9.2, medium levels of education are associated tion, including vocational training up to at least upper with a reduction in employment, an issue that would secondary. Reducing dropouts and improving transition warrant additional exploration. at each sublevel will be critical to achieve this goal. 74 Republic of Congo – Poverty Assessment Report TABLE 5.2: Marginal Effects—selected Variables—on Earnings and Employment, 2011 (coefficients) Hourly Wage Monthly Wage OLS Heckman OLS Heckman Earnings, smaller number of controls Education None Ref. Ref. Ref. Ref. Primary completed –0.110 –0.195** –0.066 –0.199* Junior secondary completed 0.042 0.116* 0.121 0.301*** Senior secondary completed 0.196*** 0.313*** 0.296*** 0.580*** Tertiary education 0.504*** 0.422*** 0.600*** 0.555*** Location Brazzaville Ref. Ref. Ref. Ref. Pointe Noire 0.113* 0.179** 0.226*** 0.345*** Other municipalities –0.034 –0.217*** –0.082 –0.451*** Semi-urban 0.055 –0.371*** –0.060 –0.895*** Rural –0.098 –0.400*** –0.330*** –0.979*** Pygmy –0.583*** –0.823*** –0.766** –1.215*** Disable –0.226 0.240 –0.489** 0.449* Female –0.100 –0.023 –0.162 0.003 Employment, smaller number of controls Education None — Ref. — Ref. Primary completed — 0.094* — 0.057 Junior secondary completed — –0.131*** — –0.157*** Senior secondary completed — –0.228*** — –0.247*** Tertiary education — 0.067 — –0.018 Location Brazzaville — Ref. — Ref. Pointe Noire — –0.109** — –0.101** Other municipalities — 0.220*** — 0.163*** Semi-urban — 0.605*** — 0.485*** Rural — 0.494*** — 0.466*** Pygmy — 0.409** — 0.368*** Disable — –0.464*** — –0.324*** Female — –0.065* — –0.058 Earnings, larger number of controls Education None Ref. Ref. Ref. Ref. Primary completed –0.097 –0.184** –0.051 –0.191 (continued on next page) Employment and Main Income Sources 75 TABLE 5.2: Marginal Effects—selected Variables—on Earnings and Employment, 2011 (coefficients) (continued) Hourly Wage Monthly Wage OLS Heckman OLS Heckman Earnings, smaller number of controls Junior secondary completed 0.042 0.113* 0.124 0.296*** Senior secondary completed 0.122* 0.239*** 0.206** 0.492*** Tertiary education 0.251*** 0.168** 0.334*** 0.295** Location Brazzaville Ref. Ref. Ref. Ref. Pointe Noire 0.139** 0.205*** 0.236*** 0.353*** Other municipalities –0.031 –0.210*** –0.097 –0.454*** Semi-urban 0.074 –0.347*** –0.054 –0.874*** Rural –0.080 –0.380*** –0.321*** –0.959*** Pygmy –0.578*** –0.819*** –0.788** –1.229*** Disable –0.208 0.251 –0.458* 0.461* Female –0.106 –0.029 –0.160 0.012 Employment, larger number of controls Education None — Ref. — Ref. Primary completed — 0.091* — 0.052 Junior secondary completed — –0.134*** — –0.159*** Senior secondary completed — –0.230*** — –0.253*** Tertiary education — 0.057 — –0.036 Location Brazzaville — Ref. — Ref. Pointe Noire — –0.112** — –0.107** Other municipalities — 0.220*** — 0.161*** Semi-urban — 0.608*** — 0.485*** Rural — 0.491*** — 0.462*** Pygmy — 0.400** — 0.361** Disable — –0.464*** — –0.325*** Female — –0.066* — –0.059 Source: Authors’ calculation using the 2011 ECOM survey. Note: *** p<0.01, ** p<0.05, * p<0.1; see statistical appendixes (Tables SA.4 and SA.5) for detailed econometric results. Location actually matters in determining earn- living in Brazzaville, while individuals in other urban ings levels. Controlling for other independent variables, areas and in rural areas have lower levels of earnings. This geographic location is strongly correlated with earnings, ranking mirrors the ranking of region by poverty level. at least with the Heckman model. Individuals living in Many of the effects are large. For example, the coeffi- Pointe Noire have higher levels of earnings than those cient for rural areas in the Heckman model is estimated 76 Republic of Congo – Poverty Assessment Report at −0.321, implying a reduction in monthly earnings of autochthon individuals having a massive reduction in a third for rural individuals in comparison to otherwise earnings versus Bantu individuals. similar individuals living in Brazzaville. The sector of employment of the individual also After controlling for various individual charac- has strong correlation to earnings. Those in agriculture teristics, it appears that there is no gender gap. For all earn less. Civil servants and managers earn the most. In the models that were estimated, the coefficient on females the earnings regression, agriculture is the reference cat- is not significant, implying that the eventual differences egory, and working in most other sectors brings in a gain observed between female and male should be due to other in earnings. In terms of socio-professional categories, factors, including difference in education/skills—but the reference category is senior managers and engineers, more important, difference in the type of activity. As with other categories faring less well. In terms of type of noted earlier in the chapter, females are more likely to employer, the reference category is public administration, be self-employed. It is also important to note that one of with other categories again faring less well. the Heckman model hinted at a possibility that gender discrimination against women happens at the selection  ain sources of income for 5.6 M level. With equal competencies, women are less likely to households be employed, but once they manage to be employed, they are paid at the same rate than their male counterparts. This section focuses on household income sources. This results are similar to what is being found The ECOM survey collected data on major household in related literature. Gender norms influence the type income sources. Overall, households’ incomes can be of activities women engage in, causing them to go into classified in seven main categories: transfers, wages, non- lower productive sectors. In the case of Uganda for farm business, agriculture, pension, property income, example, Campos et al. (2015), controlling for the sec- and other. The objective of this section is to provide a tor in which a woman works, found that women earn description of household income by the socio-demo- just as much as men. However, women tend to choose graphic characteristics of the head. Econometric analyses less profitable sectors. Women are more likely to work are also conducted for robustness. The Congolese society in sectors that are considered ‘feminine’, such as hair- appears to be a generous one, with up to 41 percent of dressing and retail trade. Women who cross over into households declaring that they have received a transfer male-dominated sectors earn as much as men and three during the last 12 months. With 34 percent, nonfarm times more than women who stay in female-dominated business registered the second highest income source, sectors. This study suggests that women are self-selecting followed by wages (32 percent), agriculture (24 percent), themselves into less productive sectors. This is likely to other (12 percent), pension (5 percent), and property be the case in the Republic of Congo as well. income (2 percent). The effect of disability on earnings is not conclu- Households headed by autochthons are more sive. After controlling for a range of variables denoting likely to get income from agriculture activities. The the type of occupation of the individuals, disability does probability of earning income from agriculture is much show opposite effects in both specifications. Note that higher when the head is autochthonous (Figure 5.11.1). this is for individuals with a disability who are working, This is not surprising given the living mode of autoch- suggesting that the disability may not be too severe. thons, who still depend primarily on nature, including Individual’s endowments must be the main factors. The picking and hunting. disabled who manage to go to school will equally do Households headed by females are more likely to better on the job market. receive transfers. Transfers are more frequent for house- There is a clear discrimination against autoch- holds headed by females. Such results are quite general- thons. Ethnicity has a large effect on earnings with ized and well known in the literature on migration and Employment and Main Income Sources 77 FIGURE 5.11: Share of Households with Positive Income by Source (%), 2011 1. Ethnicity and disability 2. Gender 80 60 50 60 40 40 30 20 20 10 0 0 Bantou Pygmy Other Disabled Male Female Head ethnicity Head disability Head gender 3. Location 4. Welfare quintile 60 50 45 50 40 40 35 30 30 25 20 20 15 10 10 5 0 0 Brazzaville Pointe Other Semi- Rural Q1 Q2 Q3 Q4 Q5 Noire municipalities urban Welfare Quintile Region Transfer Wage Nonfarm business Agriculture Pension Property income Other Source: Authors’ calculation using the 2011 ECOM survey. remittances. Households headed by females often benefit percent of households declared that they do earn income more from remittances because the male partners migrate from agriculture (Figure 5.13). On the other hand, agri- for work and send transfers back home frequently. A culture seems to be very important in the department similar situation must be at play here. Unfortunately, of Plateaux where up to 72 percent of the households ECOM data do not allow to test this assumption. declared earning income from agriculture. Agriculture is In urban areas, households are more likely to earn also as important in Lekoumou (67 percent), Bouenza income from wages, transfers, and nonfarm business, (54 percent), and Pool (51 percent). Wages, transfers, while in rural areas, households are more likely to earn pension, and property income are more likely to be income from agriculture. This is not surprising based on earned by those living in the two main cities. An impor- earlier results on labor market. Wage jobs are located in tant share of households receive transfers in Lekoumou, urban areas, especially the two main cities. In rural areas, Bouenza, and Plateaux (more than 40 percent each). people rely on subsistence agriculture for survival. In addition to Brazzaville and Pointe Noire, an impor- Beyond the urban-rural differences, there are tant fraction of households declared receiving a wage important regional specificities in terms of house- in Sangha and Niari (29 and 23 percent, respectively). holds’ income sources. Obviously, households in the Plateaux, Cuvette-Ouest, and Pool are the departments two main cities are less likely to earn income from agri- with the lowest share of households receiving a wage culture. For Brazzaville and Pointe Noire, less than 6 (less than 10 percent). 78 Republic of Congo – Poverty Assessment Report FIGURE 5.12: Type of Agricultural Activity and Type of Nonfarm Business, 2011 1. Agricultural activity Cassava (tuber) Groundnut Maize Leafy vegetable Banana plantain Bush pear (Plum) Banana Tomato Yam Avocado Pineapple Bean Sweet potato Taro Palm nuts Orange Mango Onion Ginger Scallion Cocoa Irish potato Coffee Rice 0 5 10 15 20 25 30 35 2. Nonfarm business Sales/trade Agro-processing Other services Other manufacturing Personal care Construction Transport Extractives 0 10 20 30 40 50 60 70 Source: Authors’ calculation using the 2011 ECOM survey. The poor earn their income from agriculture, bill somewhere. This was one of the findings from the while the nonpoor earn income from wages, trans- SCD consultation (World Bank 2016). The analysis of fers, and, to some extent, nonfarm business. At the survey data seems to give more understanding to this household level, there is a clear and strong correlation perception, as those with wage income are less likely to between income source and welfare. Wage income is be poor or vulnerable. particularly related to poverty. Up to at least 40 percent Cassava, groundnuts, maize, leafy vegetables, of households in the two richest quintiles declared receiv- banana, bush pear (plum), and tomatoes are the main ing wage earnings, against only 9 percent for the bottom crops. These crops are cultivated by at least 5 percent of quintile. Surprisingly, the nonpoor are also more likely households. By far, cassava, is the main crop, with up to receive transfers. As noted earlier, agriculture earnings to 35 percent of households in the Republic of Congo are more likely to be earned by the poor. declaring that they did cultivate it (Figure 5.12.1). This In the Republic of Congo, a common belief is share is even higher if we look at the poorest. Up to 60 that someone is employed only if he appears on a wage percent of the bottom quintile did cultivate cassava. Employment and Main Income Sources 79 FIGURE 5.13: Share of Households with Positive Income by Source and Location (%), 2011 1. Transfers 2. Wages 3. Nonfarm business (45.0,50.0] (45.0,50.0] (45.0,50.0] (40.0,45.0] (40.0,45.0] (40.0,45.0] (35.0,40.0] (35.0,40.0] (35.0,40.0] (30.0,35.0] (30.0,35.0] (30.0,35.0] (25.0,30.0] (25.0,30.0] (25.0,30.0] (20.0,25.0] (20.0,25.0] (20.0,25.0] (15.0,20.0] (15.0,20.0] (15.0,20.0] (10.0,15.0] (10.0,15.0] (10.0,15.0] (5.0,10.0] (5.0,10.0] (5.0,10.0] [0.0,5.0] [0.0,5.0] [0.0,5.0] 4. Agriculture 5. Pension 6. Property income (4.5,5.0] (75.0,80.0] (9.0,10.0] (4.0,4.5] (70.0,75.0] (8.0,9.0] (3.5,4.0] (65.0,70.0] (7.0,8.0] (3.0,3.5] (60.0,65.0] (6.0,7.0] (2.5,3.0] (55.0,60.0] (5.0,6.0] (2.0,2.5] (50.0,55.0] (4.0,5.0] (1.5,2.0] (45.0,50.0] (3.0,4.0] (1.0,1.5] (40.0,45.0] (2.0,3.0] (0.5,1.0] (35.0,40.0] (1.0,2.0] [0.0,0.5] (30.0,35.0] [0.0,1.0] (25.0,30.0] (20.0,25.0] (15.0,20.0] (10.0,15.0] (5.0,10.0] [0.0,5.0] Source: Authors’ calculation using the 2011 ECOM survey. Similarly, 33 to 37 percent of the bottom quintile did related to the transformation of cassava along the value cultivate groundnut or maize. One can conclude that chain (World Bank 2015b). it is a very good thing for the poor that the National The financial market seems to be almost absent Development Plan (NDP) focuses on, among other in supporting private initiative. In agriculture, up to things, these crops, plus others. 85 percent of farmers declared that they relied solely on Most of the nonfarm businesses are in sales/ their own savings or parents’ support to finance their trade. When analyzing the type of nonfarm business, businesses. A similar situation is observed for those one notices that these businesses are dominated by households owning a nonfarm business. Most of them services. Most of the households deal in sales/trade rely on their own savings as well. As a consequence, (Figure 5.12.2). Agro-processing also has an important access to finance is cited as the second major constraint share of nonfarm business (19 percent). There is insuf- to conduct business, electricity being the first obstacle ficient evidence to develop on this, but it is likely to be (see Box 5.1 for more details). 80 Republic of Congo – Poverty Assessment Report Box 5.1: The financial market is almost nonexistent in the Republic of Congo The only available global data set on financial inclusion—Global Findex—shows that only 16.7 percent of the adult population in the Republic of Congo own an account at a formal financial institution. This places the Republic of Congo way below the SSA average (29 percent), as well as far below the lower-middle-income countries average (42 percent). As expected, there is a big gap in financial inclusion between the bottom 40 percent and the top 60 percent of the population. Only 6 percent of the adult population in the bottom 40 percent has access to a formal financial institution (compared to an average of 24 percent in the top 60). The gap by education level is also wide, with about 23 percent of those with secondary or higher education banked versus just 10 percent for those with primary education or lower. Gender differences exist and are quite important as well, about 5 percentage point difference. Lack of access to the formal financial market makes it difficult for households to cope with short-term shocks such as illness, loss of employment, drought/irregular rains, landslides, and so on. It may also be a significant barrier to their ability to invest to build human and other physical assets or to start and operate a small business. Estimations from the 2011 ECOM survey suggest that the majority of the population rely on their own savings or on their parents to start and run their economic activity. Share of Adult Population with Account at a Formal Financial Institutionl – Cross Country FIGURE B5.1:  Comparisons 80 14 % of adult population (15+) 70 12 60 10 50 8 Ratio 40 6 30 20 4 10 2 0 0 Madagascar Burundi Chad Congo, Dem. Rep. Cameroon Senegal Cambodia Mali Burkina Faso Sierra Leone Cote d'Ivoire Sudan Benin Congo, Rep. Togo Tanzania Mauritania Ethiopia Low income Uganda SSA (developing only) Bangladesh Angola Gabon Vietnam Zambia Nepal Ghana Rwanda Lower middle income Nigeria Botswana LAC (developing only) Kenya World South Africa Overall Ratio: top 60% to bottom 40% (right axis) Source: World Bank, Global Findex Database of Financial Inclusion. Share of Adult Population with Account at a Formal Financial Institution – Distribution by FIGURE B5.2:  Gender, Welfare and Education 25 23.8 23.1 20 19.2 15 14.2 10.2 10 6.3 5 0 Male Female Bottom 40% Top 60% Primary or lower Secondary or higher Gender Income Education Source: World Bank, Global Findex Database of Financial Inclusion. (continued on next page) Employment and Main Income Sources 81 Box 5.1: The financial market is almost nonexistent in the Republic of Congo (continued) FIGURE B5.3: Ranking of the Top Business Environment Obstacle for Firms, 2009 Electricity 31.9 Access to finance 15.6 Political instability 15.5 Corruption 8.7 Customs and trade regulations 5.4 Transportation 3.7 Practices of the informal sector 3.6 Tax administration 3.1 Access to land 2.9 Inadequately educated workforce 2.9 Business licensing and permits 2.1 Tax rates 1.6 Courts 1.0 Labor regulations 1.0 Crime, theft and disorder 0.9 Source: Enterprise Surveys, The World Bank. On average, wage earnings represent the highest 330,000 against XAF 260,000 for women). Pointe Noire part of income earnings by households, followed by being the economic capital, it appears that the average income from nonfarm business. In 2011, households earnings from nonfarm business is also very high there. earned, on average, XAF 650,000 annually as wage in the Households headed by autochthons, those liv- Republic of Congo. This represents 48 percent of total ing in rural areas, and the poor are more likely to earn earnings. Earnings from nonfarm business are a distant income from agriculture. If we look at the contribu- second at XAF 311,000, which represent 23 percent tions in terms of share of the overall household income, of total income. In total, income from wages and non- we find that agriculture contributes for a bigger share of farm businesses represents 71 percent of total incomes. income for those groups (Figure 5.15). Earnings from Agriculture accounts for a small share of overall incomes. agriculture represents 60 percent of income for autoch- Agriculture only represents 8 percent of income. thons, 36 percent for those in rural areas, and 28 percent Annually, Congolese households earn, on average, XAF for the bottom quintile. Note that agriculture accounts 105,000 from agricultural activities. for a very small proportion (8 percent) of income for As one could expect based on previous analyses, the disabled. there are important differences across the various Nonfarm business is also very important for dimensions. First, if we focus on income levels only, autochthons, the disabled, women, and both urban it appears that Bantus, those with no disability, men, and rural poor. In addition to agriculture, it appears those living in urban areas, and those in the top quin- clearly that the poor and vulnerable rely on nonfarm tile have higher wage and nonfarm business earnings businesses for livelihood. Earnings from nonfarm busi- (Figure 5.14). Nonfarm earning is very important for the ness represents 36 percent of income for autochthons, disabled. On average, they earn XAF 405,000, against 35 percent for the disabled, 32 percent for women, 30 XAF 310,000 for Bantus. Nonfarm business earnings percent for those in rural areas, and 39 percent for the are also very high for households headed by men (XAF bottom quintile. 82 Republic of Congo – Poverty Assessment Report FIGURE 5.14: Average Household Annual Income by Source (XAF), 2011 1. Ethnicity and disability 2. Gender 700,000 900,000 600,000 800,000 700,000 500,000 600,000 400,000 500,000 300,000 400,000 200,000 300,000 200,000 100,000 100,000 0 0 Bantou Pygmy Other Disabled Male Female Head ethnicity Head disability Head gender 3. Location 4. Welfare quintile 1,400,000 1,400,000 1,200,000 1,200,000 1,000,000 1,000,000 800,000 800,000 600,000 600,000 400,000 400,000 200,000 200,000 0 0 Brazzaville Pointe Other Semi- Rural Q1 Q2 Q3 Q4 Q5 Noire municipalities urban Welfare Quintile Region Transfer Wage Nonfarm business Agriculture Pension Property income Other Source: Authors’ calculation using the 2011 ECOM survey. Clearly, wage earnings are for the better-off, those We can use the Gini Income Elasticity (GIE) to who are not vulnerable. Wages contribute for a bigger assess which income source has a positive (or nega- share of income for those living in urban areas, male- tive) effect on inequality, thus their correlation with headed households, and those in the top quintiles. There poverty. As discussed in Wodon and Yitzhaki (2002), is an exception for the disabled. Wages account for a big source decomposition of the Gini index has been used proportion of their earnings. For those in the top two quin- extensively to analyze how various sources of income tiles, wages represent up to 55 percent of their incomes. affect the inequality.32 Results from a decomposition by Most of these univariate results in terms of prob- source of the Gini index are presented in Figure 5.16. ability of a given household to earn a type of income When an income source has a GIE of 1, it means that it and the level of income are confirmed by econometric moves perfectly in synchronization with total income so analysis. Econometric analysis confirms all the descriptive that a change in the source does not affect the inequality analysis results, including the fact that households headed by autochthons are more likely to get income from agri- culture activities. Econometric analysis also confirms that 32 The change in the Gini as a proportion of the initial Gini resulting households headed by females are more likely to receive from a 1 percent increase in income or consumption from source k, denoted by ∆G/G , is equal to the share of the source k in total income transfers. In urban areas, households are more likely to or consumption, denoted by Sk times GIE minus 1. In formal terms, earn income from wages, transfers, and nonfarm business. ∆G/G = Sk* (GIEk – 1)/100. The division by 100 is a normalization. Employment and Main Income Sources 83 FIGURE 5.15: Income Source as a Share of Total Income (%), 2011 1. Ethnicity and disability 2. Gender 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Bantou Pygmy Other Disabled Male Female Head ethnicity Head disability Head gender 3. Location 4. Welfare quintile 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Brazzaville Pointe Other Semi- Rural Q1 Q2 Q3 Q4 Q5 Noire municipalities urban Welfare Quintile Region Transfer Wage Nonfarm business Agriculture Pension Property income Other Source: Authors’ calculation using the 2011 ECOM survey. FIGURE 5.16: Gini Income Elasticity, 2011 more. Thus, if an income source has a GIE larger than 1, a marginal increase in the income from that source 50% Wage results in higher inequality. The larger the GIE is, the Share in total income per 40% larger the increase in overall inequality will be. A source equivalent adult (%) 30% with a GIE equal to 0 is not correlated with total income. Nonfarm business For example, a universal allocation identical for all indi- 20% viduals would have a GIE of 0. 10% Agriculture Other Results from the GIE suggest that four income Transfer Pension Property income sources are pro-rich: wages, other, pension, and prop- 0% 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 erty income, as they are contributing to inequality Gini Income Elasticity (GIE) increase. The most inequality increasing sources at the Source: Authors’ calculation using the 2011 ECOM survey. margin are wages, other, pension, and property income. This means that a large share of income per equivalent adult from wages, other, pension, and property income in total income. A source with a GIE larger than 1 is are benefitting the better-off. affecting the richer part of the population more, while The remaining three income sources are pro- a source with a GIE smaller than 1 is affecting the poor poor: nonfarm business, agriculture, and transfers, 84 Republic of Congo – Poverty Assessment Report as they are contributing to reduction of inequalities. agriculture, and another quarter of workers work for a Nonfarm business, agriculture, and transfers are the most household without pay. The public administration is the inequality decreasing sources at the margin. This means main provider of formal wage jobs. One out of ten work- that a large share of income per equivalent adult from ers (14 percent) is employed in the public administration nonfarm business, agriculture, and transfers are benefit- or by a parastatal firm. The role of the private sector in ting poor households. In the short run, to have a higher providing jobs to the population is not negligible. Up impact on poverty, any intervention aiming at increasing to 13 percent of workers are employed either in large income for the poor should focus on these items. private firms (5 percent) or in SMEs (8 percent). The education level of an individual is strongly 5.7 Conclusion correlated with the likelihood of being employed by the public administration, large private firms, or This chapter has provided a basic diagnostic of employ- SMEs. Education makes a difference starting from upper ment and main income sources. secondary. Education seems to matter a lot in term of Unemployment seems to have increased between providing the required skills to access quality jobs in the 2005 and 2011, which could represent a puzzle to formal sector. In the Republic of Congo, it really makes understand trends in well-being. It seems that the an important difference when one reaches upper second- impressive economic growth was jobless or at least did not ary. Up to 21 percent of those with upper secondary have significant impact on unemployment, especially for education are employed in the public administration, the youth. This is not surprising given that growth was and 13 percent of them are employed by SMEs. These driven by the oil sector which is not labor intensive and has number are even higher for those with tertiary educa- limited linkages with the rest of the economy. The trend tion. In fact, 46 percent of those with tertiary education in the employment profile suggests that the economic are employed in the public administration. Moreover, growth was jobless. This is not surprising given that this 20 percent of them are employed either by large private economic growth was essentially driven by the oil econ- firms or by SMEs. Only 19 percent of those with tertiary omy. Moreover, these findings also suggest that the massive education are self-employed. On the contrary, up to 81 infrastructure program under the ‘municipalization’ slogan percent of those with no education are self-employed. also failed, at least between 2005 and 2011, to create sub- Education then appears as a clear barrier or key asset to stantial and enough jobs to have a consequential impact entering the quality job market. on the labor market. This is not surprising either as most Results from the GIE suggest that four income of the construction work is just creating temporary jobs. sources are pro-rich: wages, other, pension, and prop- Clearly, unemployment is higher for younger erty income, as they are contributing to inequality populations, the better educated, Bantus, individu- increase. The most inequality increasing sources at the als with a disability, women, and those living in the margin are wages, other, pension, and property income. largest cities. For robustness, two definitions of unem- This means that a large share of income per equivalent ployment are used. The same conclusions emerge: unem- adult from wages, other, pension, and property income ployment rates are highly correlated with age, education, are benefitting the better-off. Nonfarm business, agri- disability status, ethnicity, and location. culture, and transfers are the most inequality decreasing The formal sector, either public or private, has sources at the margin. This means that a large share of failed to create quality jobs for the population. As a income per equivalent adult from nonfarm business, consequence, the vast majority of the labor force are self- agriculture, and transfers are benefitting poor house- employed. Slightly more than three out of five workers holds. In the short run, to have a higher impact on (63 percent) work on their own account, running a busi- poverty, any intervention aiming at increasing income ness with no employees or being involved in subsistence for the poor should focus on these items. Employment and Main Income Sources 85 Access to Basic Social Services 6 6.1 Introduction Understanding dynamics in access to and quality of basic social services can provide a valu- able complement to monetary measures of poverty. Access to basic infrastructure services is paramount for meeting basic needs and improving human capital indicators related to health and education. Given that it has direct linkages with other dimensions of well-being, access to improved sources of drinking water and sanitation services is very crucial, particularly for the poor. A connection to the electricity or piped water network eases access to time-saving technol- ogy and therefore reduces domestic work, especially for women. Education and health are key priorities sectors for the Government of the Republic of Congo. The development of a wealthier, literate, and healthy society is a fundamental goal of the Republic of Congo’s NDP 2012–16 and Poverty Reduction Strategy Paper (PRSP) 2012–16. Reaching the education and health Millennium Development Goals (MDGs), institutional strengthening of both education and health systems, and improving the quality of service deliv- ery in education and health provision are thus key aims of the NDP and PRSP 2012–16. The Congolese authorities have been developing and implementing policies and programs aimed at achieving such objectives (World Bank 2014). This chapter will use data from the two ECOM surveys, the DHS, and the WDI to provide a diagnostic of key social sectors in the Republic of Congo. Following Gable, Lofgren, and Osorio-Rodarte (2015), econometric techniques are used to show that in many aspects, the country performs below expectations compared to its peers. The chapter has five parts. In the first part, we will offer a diagnostic of the education sector. The second part analyzes the health sector, including nutrition. The third part presents the country status in terms of access to electricity. The fourth part covers access to water and sanitation, with an emphasis on residential network water. The last part assesses the country performance on roads and ICT. 6.2 Education The link between education, well-being, and social development is not a new phenomenon. As shown in previous chapters, education is a strong correlate of poverty in the Republic of Congo. Those with higher level of education are more likely to access quality jobs. Human capital theory 87 suggests that education raises incomes by increasing the Republic of Congo also performs better than expected productivity of workers. The accumulation of human with regard to net primary school enrollment. Nine out capital can improve the efficiency of labor input in of ten primary school age children are enrolled. Girls’ net terms of quality and can also enhance overall technical primary enrollment is 87 percent, which is much lower efficiency in production and allocative efficiency of the than the net enrollment of boys (95 percent). household (Jolliffe 2002; Kurosaki and Khan 2006). However, performance is way below expectations In the 1970s, the Republic of Congo had a on most of the other critical education sector indicators, relatively well-developed education system compared including primary school completion. As illustrated in with most other SSA countries. However, civil war, Figure 6.1.1, the Republic of Congo is performing below social unrest, and economic crisis contributed to destroy expectation in terms of primary school completion. The the achievement. The Republic of Congo expanded the primary school completion rate was 74 percent in 2012 coverage of its education system early on, in the 1960s and has been fluctuating up and down over the last decade. and 1970s. In the 1980s and 1990s, conflict and eco- Overall, between 2005 and 2012, the primary school nomic crises led to a substantial drop in school partici- completion rate increased for girls, but went down for pation, but since around 2000, school attainment has boys. In 2012, the primary school completion rate for girls been steadily improving with growing coverage at all was 79 percent, up from 70 percent in 2005. During the levels of education. same period, the primary school completion rate for boys The country is performing above expecta- went down from 75 percent to 70 percent. tions on literacy and primary school enrollment. One of the direct consequences of the low pri- Nevertheless, there are important gender gaps. The mary school completion is that secondary school Republic of Congo seems to have relatively high adult enrollment and are below expectations. Slightly less literacy rates compared to many of its income peers, per- than six out of ten (58 percent) of secondary school age forming better than expected. Literacy rates for both men children were enrolled in 2011. Moreover, the Republic and women are above expectations compared to peers. of Congo performs below expectations on lower second- On average, close to eight out of ten adults can read and ary school completion (Figure 6.1.2.). Lower second- write, but literacy is slightly higher for men compared ary school completion is low, at 52 percent. Secondary to women (86 percent against 73 percent in 2011). The school completion is similar for both girls and boys. FIGURE 6.1: Performance Below Expectations in Primary and Secondary School Completion 1. Primary completion rate 2. Lower secondary completion rate Lower secondary completion rate, 120 150 total (% of relevant age grroup) total (% of relevant age group) Primary completion rate, 100 100 80 ROC 2012 ROC 2005 50 ROC 2012 ROC 2004 60 ROC 1996 ROC 1996 40 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. 88 Republic of Congo – Poverty Assessment Report FIGURE 6.2: Reasons for Not Being Enrolled (%) 1. Ages 6–11 years 2. Ages 12–18 years, with primary Other Too expensive Too expensive Useless/uninteresting Useless/uninteresting Failed exam Not of school age Illness/Pregnancy Failed exam Other Too far away Got married Too old/Completed school Too far away Illness/Pregnancy Too old/Completed school Got married Is working (home or job) Is working (home or job) Not of school age 0 5 10 15 20 25 0 5 10 15 20 25 30 35 40 Total Girls Boys Source: Authors’ calculation using the 2011 ECOM survey. Despite improvement in enrollment, the quality children are much less likely to go to school than Bantu of education is of concern and seems to be decreasing. children, with virtually none enrolled in secondary Test results suggest that a high proportion of students schools. Children with a disability are at a clear disad- leaving primary school do not have sufficient founda- vantage. There are finally large differences in enrollment tional skills. Survey estimations suggest that the propor- rates, especially again at the secondary level, by location tion of students enrolled in primary grade 4 to grade 6 or geographic area. who can read and write dropped from to 89 percent to Affordability and early marriage/pregnancies are 85 percent between 2005 and 2011. End-of-cycle tests the main reasons for children not being enrolled. As in primary education have found that two-thirds of stu- shown in Figure 6.2, cost is the main reason not to go dents leaving primary school do not have sufficient foun- to school for both children of primary school age and dational skills in literacy and numeracy. Results from children of secondary school age who have completed the international standardized tests also show downward primary cycle and are thus eligible for secondary school trend in student competencies. Moreover, the student (at the primary level, ‘other’ reasons not specified in the proficiency is low performing compared to peer countries data play a role for boys). Lack of interest in schooling (CONFEMEN 2012). Thus, more and more children comes up next, which may denote a lack of opportunity are going to school, but it does not guarantee that they associated with schooling, at least in some areas. This is are acquiring the expected knowledge. This assessment especially the case for boys, but for girls at the secondary confirms the importance of the on-going debate on the level the issue of an illness or more likely a pregnancy quality of service delivery, particularly education services is a more important factor for dropping out, especially in the country (World Bank 2014).33 when one also factors in marriage as another reason to Children from poor families, autochthons, and those with disability are less likely to be enrolled. Enrollment rates are higher for the nonpoor than for the 33 The World Bank Education Global Practice is currently conducting poor, especially at the secondary level, and the same is a Service Delivery Indicators (SDI) survey that will provide more observed when considering welfare quintile. Autochthon evidence on this. Access to Basic Social Services 89 drop out.34 Other factors, including work (opportunity on who attends which schools in terms of welfare level cost of schooling) and distance to school, also play a role. through concentration curves by type of school (for pri- The private sector is playing a very important mary schools). As expected, children in public schools and increasing role in the provision of education ser- tend to be poorer than those in private schools, with vices. Figure 6.3.1 provides data on the type of school students in faith-based schools in-between in terms of attended. The share of students in private schools seems welfare levels. In other words targeting public schools to have increased between the last two surveys for 2005 helps in reaching the poor. and 2011. In 2011, 35 percent of students were enrolled Satisfaction is low for public schools. Lack of in a private school. This is up by 15 percentage points books/supplies and overcrowding are the main rea- compared to 2005. Such high shares of private provi- sons for dissatisfaction. Satisfaction with schools, as sion in a country where 40 percent of the population is measured by the share of parents who do not have any poor is an indication of inadequate success in ensuring complaints with schools is low, especially in public effective access by the Government. schools (Figure 6.4.1). Less than 20 percent are satis- The growing importance of private provision of fied with public schools, against close to 60 percent education service calls for more effort in regulating for private schools. Importantly, parents of children the sector. Often, private investors are driven by profit from lower socioeconomic background, as measured maximization, which may not be good for the overall from the quintile of well-being of the household, are country benefit. Thus, effort to make sure that the pro- much less likely to be satisfied than parents from higher vision by private sector is beneficial for the student and quintiles. The main reasons for dissatisfaction are lack the country as a whole in terms of curriculum, alignment with the labor market, and effectiveness. As expected, public schools are doing much bet- 34 In Brazzaville, 27 percent of girls ages 15–19 are either pregnant ter in reaching the poor. Figure 6.3.2 gives information or already mothers (DHS 2012). FIGURE 6.3: Type of School Attended and Concentration Curves 1. Market share by type of school 2. Pro-poorness of various type of schools 100% 100 16.4 90% 21.6 20.6 20.3 90 25.7 31.3 32.9 34.9 36.5 80% 1.4 0.1 80 49.4 3.1 2.4 70% 0.4 Share in total users (%) 1.3 3.1 60% 3.4 2.9 70 50% 0.2 60 80.9 40% 76.2 50 74.4 73.1 71.4 66.6 63.6 61.1 58.9 30% 40 48.9 20% 30 10% 0% 20 10 Primary Secondary 1 Secondary 2 Tertiary All Primary Secondary 1 Secondary 2 Tertiary All 0 0 10 20 30 40 50 60 70 80 90 100 2005 2011 Welfare percentile Government Religious Organization Equity Government FBO Private Community Other Welfare 2011 Private Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. 90 Republic of Congo – Poverty Assessment Report FIGURE 6.4: Satisfaction with Schools (%) 1. Share of households satisfied with schools 2. Reasons for lack of satisfaction 70 Lack of books/supplies 60 Overcrowding 50 Lack of teachers 40 Teachers often absent 30 Facilities in bad condition 20 Poor teaching 10 Other problem 0 Q1 Q2 Q3 Q4 Q5 Total 0 5 10 15 20 25 30 35 40 45 50 Public schools Non public shcools All 2005 2011 Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. of books/supplies, overcrowding, and lack of teachers 6.3 Health and nutrition (Figure 6.4.2). The lack of books/supplies and the lack of teachers are predominant for children from lower There has been substantial improvement in child and socioeconomic backgrounds, while overcrowding is maternal mortality. However, the country still performs cited more often by parents of children from higher below expectations with regard to maternal mortality and levels of well-being. has not reached any of the health-related MDGs. Child Subsidizing school uniforms and school sup- and maternal mortality are often used as a measure of the plies would be the most pro-poor. Average private efficiency of the health sector in a given country. Between spending per child in school ranges from a low of 2005 and 2012, under-five mortality, which measures XAF 7,775 per year in the bottom quintile to a high the probability of children dying between birth and the of XAF 117,473 per child in the top quintile. The fifth birthday, dropped from 95.3 to 52.6 per 1,000 national average is XAF 47,111. As a share of total births. Owing to this improvement, the country is now consumption, households allocate 2.5 percent of their performing as expected, given its GNI level. The scenario resources to education in the bottom quintile versus 3.7 is a bit different regarding maternal mortality. Despite percent in the top quintile. The national average is 2.8 percent. For the very poor, who are much more likely not to enroll in school, cost is the main constraint, as 35 Consumption dominance curves are useful to assess the impact of balanced budget marginal tax reforms on poverty (Duclos, Makdissi, mentioned earlier. Strategies for lowering costs could and Wodon 2008; Makdissi and Wodon 2002). The idea is to test include transfers to poor households conditional on whether increasing a tax or subsidy for one type of goods—while enrollment of the children, but another alternative 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 adopted in some countries is to ensure that private lead to a reduction or an increase in a wide range of poverty measures. costs of schooling are reduced. As shown in Figure 6.5, If one consumption dominance curve is above another, then it is good among different types of private expenditures, subsi- 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 dizing school uniforms and school supplies would be (or reducing a subsidy) for the good that corresponds to the curve the most pro-poor.35 that is located below the first curve. Access to Basic Social Services 91 FIGURE 6.5: Dominance Curves for Education Expenditures, 2011 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000 900,000 1,000,000 Annual expenditure per equivalent adult Uniforms School and registration fees School supplies Books Poverty line Source: Authors’ calculation using the 2011 ECOM survey. improvement, the country is still performing lower than who are immunized against DPT increased from 62 to its peers on maternal mortality. Maternal mortality rate 86 percent during that period. Owing to this strong declined from 781 to 426 deaths per 100,000 live births performance, the country is now performing as expected between 2005 and 2012 (Figure 6.6). with regard to DPT vaccine. During the same period, Performance is mixed regarding immunization. immunization against measles also improved among The country is performing as expected on diphtheria, children ages 12–23 months, from 54 to 78 percent. pertussis, and tetanus (DPT) and below expectation for Despite this strong performance, the country is still measles vaccines. Between 2005 and 2014, there was a performing below expectation regarding measles vac- substantial improvement of the share of children who are cine. Also, the country is unable to finance its part in the immunized. The share of children ages 12–23 months graduation plan from the Global Alliance for Vaccines FIGURE 6.6: Performance Below Expectations in Child and Maternal Mortality 1. Mortality rate, under-five (per 1,000 live births) 2. Maternal mortality ratio (national estimate, per 100,000 live births) 150 1500 estimate (per 100,000 live births) Maternal mortality ratio, national under-5 (per 1000 live births) ROC 1996 100 1000 Mortality rate, ROC 2005 ROC 2005 50 ROC 2014 500 ROC 2012 0 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. 92 Republic of Congo – Poverty Assessment Report FIGURE 6.7: Share of Population Sick or Injured in Last Four Weeks (%), 2011 1. Morbidity by age group 2. Main diseases 80 Malaria Fever 70 Headache 60 Accident Other health problem 50 Stomach aches Diarrhea 40 Ear/nose/throat problem Eye problem 30 Dental problem Skin problem 20 0–4 5–9 10–14 15–19 20–29 30–39 40–49 50–59 60 and 0 5 10 15 20 25 over All Male Female Source: Authors’ estimates based on the 2011 ECOM survey. and Immunization (GAVI)/Global Fund and this has led high-level facilities. For instance, 38 percent of people to nationwide vaccine shortages during 2015. seek care in public hospitals and 29 percent in private As expected, infants and young children and the hospitals. Integrated health center is a distant third, elderly are the most likely to be sick. This is clearly with 12 percent. visible in Figure 6.7, which plots the incidence of ill- Faith-based organizations (FBOs), traditional nesses by age group and gender. At the national level, healers, and integrated health centers are pro-poor. 37 percent of the population declares having suffered Figure 6.8.2 gives information on health providers that from an illness in 2011. Except for the infants and young were visited by the population. Faith-based providers are children, morbidity is higher for women. performing better in reaching the poor, so are traditional Similarly to other SSA countries, malaria and healers and integrated health centers. Other types of fever are the main diseases. The main illness cited is health facilities, although not pro-poor, are progressive, fever/malaria, which accounts for more than 38 per- as they lie above the Lorenz curve. The performance of cent of the episodes of illness. Next are headaches and informal service providers, especially traditional healers accidents, each accounting for 13.5 percent of episodes. should be of concern, as it might be a reflection of the The top three causes of disability adjusted life years lost poor being excluded from formal health services. In were HIV/AIDS, lower respiratory tract infections, and the poorest departments (Lékoumou, Bouenza, Pool, malaria. There are relatively few differences by gender, Plateaux, Cuvette, Cuvette-Ouest), about one in ten location, or quintile in the types of illnesses that people individuals seeks the care of a traditional healer. suffer from. The private sector plays an important role in Slightly more than half of the population seeks providing essential health services, especially in large care in private facilities. Only six out of ten individu- urban centers. Generally, the quality of public providers als who were sick did seek care. A bit more than half is perceived to be better than private providers, particu- of sick people seek care in nongovernmental facilities larly for maternal care. However, people often choose (Figure 6.8.1).The high concentration of the popula- to seek care at private providers because of proximity in tion in urban areas makes it easier for them to access urban areas, extended opening hours and shorter waiting Access to Basic Social Services 93 FIGURE 6.8: Type of Health Facility Visited and Concentration Curves, 2011 1. Type of facility visited 2. Pro-poorness of various health providers 100% 9.1 6.0 4.8 8.8 100 13.8 11.3 90% 90 Cumulative distribution of users (%) 80% 80 30.3 35.7 40.1 45.1 44.9 39.5 70% 70 60% 60 4.2 3.1 50% 2.0 1.8 1.3 2.4 50 40% 40 30% 51.6 30 49.9 48.7 47.1 49.0 49.2 20% 20 10% 10 0% 0 Q1 Q2 Q3 Q4 Q5 0 10 20 30 40 50 60 70 80 90 100 Welfare quintile Total Cumulative distribution of the population (%) Public FBO Private Other Equity Medical office/Private Hospital Public Hospital Integrated Health Centre Doctor/private dentist Traditional healer FBO Pharmacy / Pharmacist Welfare 2011 Source: Authors’ estimates based on the 2011 ECOM survey. times, and better perceived quality of laboratory services sector, to facilitate the development of public-private as compared to the public sector (Makinen, Deville, and partnerships. Additionally, a new law on the private sector Folsom 2012). Price is also an important determinant of was signed in May 2015. These efforts are already bear- health care demand: in some cases, the private for-profit ing results, as the country’s Performance-Based Financing sector charges lower prices than the public sector (par- program includes contracting out with 70 percent of ticularly for some preventive services which are offered private sector facilities in Pointe Noire and Brazzaville free of charge at point of use but for a fee in the public to deliver the basic package of services. A public-private sector) and is more flexible in offering payment terms to roundtable on pharmaceuticals, held in October 2015, is the poorest patients who have difficulty paying. further evidence of increased public-private cooperation. The Government is committed to working with Out-of-pocket spending is high due to the impo- the private sector to strengthen the health system, and sition of user fees and a charge for medicines. These as part of the World Bank’s private sector Health in high costs prohibit utilization of government health facili- Africa (HiA) initiative, the World Bank/International ties, especially among the poor. Currently government Finance Corporation (IFC) is working with the spending in the health sector is very low, (2.5 percent of Government to revise the legislative and regulatory total GDP, and US$67 per capita per year—National frameworks for the private sector to facilitate more Health Accounts [NHA] 2009–2010). This means that engagement with the private sector in service delivery. health facility budgets are insufficient to cover the actual To maximize the contribution of the private sector, the costs of services (World Bank 2014). Cost is by far the Ministry of Health and Population (Ministère de la santé main reason for dissatisfaction with health services. et la population, MOHP) has conducted a review of its The distribution of private expenditure on health legislative and regulatory framework for the private health shows that the poor will be more affected by a fiscal 94 Republic of Congo – Poverty Assessment Report FIGURE 6.9: Prevalence of Stunting, Height for Age (% of Children under Five) 1. Performance above expectations on prevalence of stunting 2. Stunting, height for age, is highly correlated to poverty 45 60 Cuvette-Ouest Prevelance of stunting, height for y = 3.594e0.0622x age (% of children under age 5) 40 R² = 0.3257 Monetary poverty (FGT1) 35 Cuvette Likouala Lékoumou 40 30 Bouenza Pool Sangha ROC 2005 25 Niari Plateaux ROC 2011 20 Kouilou 20 15 ROC 10 Brazzaville 0 5 Pointe-Noire 0 0 2000 4000 6000 8000 10000 15 20 25 30 35 40 GNI per capita, PPP (constant 2011 international $) Height-for-age Source: Authors’ calculation using the 2011 ECOM survey, the 2012 DHS, and the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. reform concerning drugs, hospitalization, and con- In the Republic of Congo, malnutrition is sultations fees. Concentration and dominance curves all strongly correlated to poverty. Affordability of food show that these are pro-poor type expenditures (see sta- might be the main cause of malnutrition. In some coun- tistical appendixes SA.2 and SA.3). In August 2011, the tries, malnutrition is often related to diet and therefore Government decided that malaria drugs, anti-retroviral not correlated to poverty. As shown in Figure 6.9.2, in (ARV), C-section will be free of charge in public facilities. the case of the Republic of Congo, malnutrition increases The analysis here shows that these measures were actually with poverty. In principle, this makes it easier to design pro-poor. However, new data and additional analytical programs to fight malnutrition, as it will mainly imply work are needed to access the effectiveness and impact of making sure the poor have access to food through social such measures. Since 2014, financing of inputs for these programs such as cash transfer, public works, school can- free health care programs has stagnated and many services teens for example. Such programs are in theory easier to are either not provided (free C-section or free health care design and implement as oppose to situation where diet for pregnant women), are not free, or provided against and behavior have to be changed. out-of-pocket payments (in the case of malaria drugs). As a result of recent improvement, the country 6.4 Electricity is performing above expectations on stunting, but malnutrition is still high. Stunting, defined as low Despite improvement over the last decade, access to height for age and an indicator of chronic malnutri- electricity is very low compared to expectations. The tion, decreased from 31 percent in 2005 to 25 percent Republic of Congo is a country where owing to sub- in 2011. As a result, the country is now performing stantial hydro potential, electricity could, in principle, as expected in comparison to its peers (Figure 6.9.1). be generated and distributed at a relatively low cost to a Nevertheless, the level of stunting remains quite high. large share of the population. Unfortunately, connection High malnutrition reduces agricultural productivity, rates in the country remain below expectations compared contributes to poverty, and affects education and intel- to the peer countries (Figure 6.10). Coverage rates have lectual potential of schoolchildren (for example, stunting increased substantially, from 26.7 percent in 2005 to causes children to start school late because they look too 42.5 percent in 2011 (Figure 6.11.1). In Figure 6.11, small for their age and will also be a cause of absenteeism coverage or connection rates, denoted as C, are decom- and repetition of school years). posed as the product of access rates at the neighborhood Access to Basic Social Services 95 FIGURE 6.10: Access to Electricity (% of Population) 1. Urban areas 2. Rural areas 120 100 Access to electricity, urban Access to electricity, rural (% of urban population) (% of rural population) 100 80 80 60 ROC 2012 60 40 40 20 ROC 2012 ROC 2000 ROC 2000 20 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. FIGURE 6.11: Electricity Coverage, Access, and Take-Up Rates (%) 1. Trend in grid coverage, access, and take-up rates 2. Grid access, take-up, and coverage rates by decile, 2011 100 100 80 80 60 60 40 40 20 20 0 0 2005 2011 1 2 3 4 5 6 7 8 9 10 ROC Welfare deciles Access to piped water at the PSU level Access to piped water at the PSU level Connected to piped water Take up rate Take up rate Connected to piped water Source: Authors’ estimates based on the 2005 and 2011 ECOM surveys. level, denoted by A, and take-up rates among households Noire and other municipalities, connection rates are who have access, denoted by U, so that C = A × U.36 The lower (50.3 and 45.7 percent respectively). On the other improvement in coverage is the result of improvement in hand, only 5.3 percent of rural households are connected both access (network availability) and take-up. In 2011, to the network. In rural areas, access and take-up rate are close to seven out of ten households (68 percent) lived in neighborhoods with electricity network; this is up from 57 percent in 2005. Reflecting improvement in mon- 36 In the survey, we consider that a household has access to electricity in its neighborhood or village if at least one household living in the etary poverty, take-up rate also increased from 47 percent same primary sampling unit (PSU) of the survey has access to the to 62 percent between 2005 and 2011. grid. In other words, neighborhoods are identified in the household Coverage is much higher in urban areas. In surveys through the PSU to which households belong. These PSUs are typically based on an administrative unit according to census Brazzaville, three out of four households (75 per- data, from which households are randomly selected to be included cent) have a connection to electricity network. In Pointe in the survey. 96 Republic of Congo – Poverty Assessment Report FIGURE 6.12: Reason for Not Subscribing and Main Issue with Network Electricity, 2011 1. Reasons for not subscribing 2. Satisfaction with the service 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Welfare quintile Total Welfare quintile Total Too expensive to subscribe Consumption too expensive Too expensive Frequent shortages Complicated process Remote network (not accessile) Poor quality Service interruption / landslide Useless Network not avaliable in the area Not concerned Source: Authors’ estimates based on the 2011 ECOM survey. low. Only 18 percent of rural households live a neighbor- demand-side factors (14.6 percent).There are differences hood with electricity network, and when the network is across location. In rural areas, the combined factor is available, only 28.9 percent can afford to connect. dominant, while the supply factor is dominant in urban As expected, poor households are less likely to areas. This means that for those in rural areas, making the be connected to the electricity network. Coverage rates network available will not be enough because the afford- are much higher among households in the top deciles of ability (demand) issue will remain. Rural electrification the distribution of consumption per capita than among should take into account pro-poor policies such as con- poorer households (Figure 6.11.2). In fact, connection nection subsidies and cross-urban-rural consumption rates are virtually nonexistent in the bottom 10 percent of subsidization in tariff design. the population in terms of welfare levels. Poor people often Satisfaction with electricity service is very low. do not have electricity in their neighborhood, and when Frequent shortages are by far the main reason for dis- the network is present, they cannot afford to connect. satisfaction. Poor quality of the electricity service and In urban areas, affordability is the main barrier cost are also important reasons for dissatisfaction. Poor to access, while in rural areas both availability of net- households also quote high cost as a reason for dissatis- work and affordability are present. Poor households faction. Slightly less than one out of three households is are more likely to state availability of network as the satisfied with the electricity network service. The poor main reason for not being connected (Figure 6.12.1). seem more satisfied. Satisfaction decreases with welfare. However, further analysis based on Foster and Araujo (2004) and Wodon et al. (2009),37 suggests that even if the network was to be made available to them, afford- 37 Foster and Araujo (2004) use a statistical framework to explain ability issues will prevent them from connecting. With the deficit in coverage into three components: (a) pure demand-side the Wodon et al. (2009) approach for instance, supply- problems; (b) pure supply-side problem; and (c) combined demand and supply-side problems. Wodon et al. (2009) proposed an alter- side factors account for a majority of the gap (63.7 per- native econometric method to try to better identify demand and cent), followed by combined factors (21.6 percent) and supply-side problems. Both methods were used here. Access to Basic Social Services 97 FIGURE 6.13: Pro-Poorness of Various Energy FIGURE 6.14: Targeting Performance (Ω) of Sources Electricity Subsidies, Selected 100 Countries Cumulative distribution of lighting source (%) 90 Burkina 0.06 80 Burundi 0.10 70 Cameroon 0.36 60 Cape Verde 0.48 50 CAR 0.27 40 Chad 0.06 30 ROC 0.62 20 Côte d'ivoire 0.51 10 Gabon 0.78 0 Ghana 0.31 0 10 20 30 40 50 60 70 80 90 100 Guinea 0.22 Cumulative distribution of the population (%) Malawi 0.02 Mozambique 0.31 SNE with own meter Gas lamp Nigeria 0.79 Solar energy Candle/other Rwanda 0.01 Other SNE (direct connection) Equity Sao Tome 0.41 SNE, the neighbor Kerosene lamp Senegal 0.41 Togo 0.47 Welfare 2011 Generator Uganda 0.02 Source: Authors’ estimates based on the 2011 ECOM survey. 0.0 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 Omega Source: Tsimpo and Wodon 2014a. Nearly 43 percent of those in the bottom quintile are satisfied as opposed to only 28 percent for those in the richest quintile. By far, frequent shortage is the main Electricity tariffs are poorly designed in the reason of dissatisfaction (Figure 6.12.2). Seven out of ten Republic of Congo. The implicit consumption subsi- households point to shortage as reasons of dissatisfac- dies are poorly targeted. Connection subsidies clearly tion. Second in the reasons for dissatisfaction are issues have the potential to be better targeted. A robust related to quality (quoted by 13 percent of households). measure of the targeting performance of subsidies is The cost of electricity in third place (11.5 percent). Cost the share of the subsidy benefits received by the poor issues are more predominant in poor households. For the divided by the proportion of the population in poverty bottom quintile, cost of electricity is quoted by three out (Ω).38 The tariff design is pro-poor if Ω > 1. For the of ten households. Republic of Congo, the value was 0.62 in 2011, that Kerosene lamp and candle are the only two is, implicit electricity subsidies are not pro-poor. Still, pro-poor sources of lighting. As shown in Figure 6.13, the Republic of Congo is among the best performers in which provides concentration curves for various sources Africa (Figure 6.14). To our knowledge, electricity tariffs of lighting, kerosene lamps and candles are the only two have not been revised for a long time. Tariff designs that sources that lie above the equity line. In relation to the are pro-poor should be considered in the future. Further Lorenz curve, it appears that solar energy, gas lamp, simulations suggest that connection subsidies will have other SNE, and generator are progressive. SNE with own meter and, to some extent, SNE from neighbor are regressive. 38 See Angel-Urdinola and Wodon (2007) for more details. 98 Republic of Congo – Poverty Assessment Report much better targeting performance than consumption 6.5 Water and sanitation subsidies. For example, if connection subsidies were given only to households with access but no connec- Improved water and sanitation tion, then the targeting indicator will be close to 1.2, Although the share of the population with access and thus pro-poor. While connection subsidies clearly to improved water slightly increased during the last have the potential to be better targeted than consump- decade, it is still far below what is expected. The tion subsidies, they should be implemented at scale Republic of Congo is performing below expectations in only when generation capacity is sufficient, and when terms of access to improved water. Between 2005 and considered, they need to be implemented well to ensure 2015, access to improved water increased slightly from good targeting and limit costs. 72 percent to 77 percent. However, the country is still Evidence suggests that the SNE is struggling performing below expectations (Figure 6.15). to effectively collect payments from residential cus- As expected, the poor are less likely to have tomers, including through the installation of new access to improved water. Usage of an improved meters in some areas. The share of households paying source of water increases with welfare (Table 6.1 and for their electricity is systematically lower than the share Figure 6.16.2). Only half of those in the poorest quintile of households who declare using electricity. This may be have access to safe water while for the richest quintile, an indication of illicit connections, but it may also reflect close to nine out of ten households have access to an late payment, lack of recovering late payment, or other improved water source. The share of households with no issues. The differences between those using electricity toilet ranges from 42.8 percent to 30 percent in these and those paying for it is big. In 2011, 42.5 percent of departments. The gap between the two main cities and households are connected to the grid, but only 30 per- the rest of the country is very pronounced. cent are paying for electricity, generating a 12.5 percent- Access to improved sanitation remains very age point gap between coverage and payment. In 2005, low, and as a consequence, the country is perform- the gap was at 7.8 percentage points. Thus, as a share of ing below expectations on this dimension as well. As the coverage rate, the 29 point gap has remained stable illustrated in Figure 6.17.2, the Republic of Congo is (29.3 in 2005 against 29.5 in 2011). This may suggest performing below expectations regarding access to a safe no improvement in the ability of the SNE to collect toilet. In 2014, only 43 percent of the population had payments from residential customers in recent years. access to improved sanitation (Table 6.2). The situation FIGURE 6.15: Access to Improved Water and Sanitation (% of Population) 1. Improved water source (% of population with access) 2. Improved sanitation facilities (% of population with access) 100 100 (% of population with access) (% of population with access) Improved sanitation facilities Improved water source 80 80 ROC 2014 60 ROC 2005 ROC 1997 ROC 2012 40 60 20 40 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. Access to Basic Social Services 99 TABLE 6.1: Main Source of Drinking Water by Welfare Quintile Welfare quintile Q1 Q2 Q3 Q4 Q5 Total Improved sources 49.2 66.9 78.7 83.4 87.5 75.4 Piped schemes 15.4 36.9 53.4 62.3 72.1 51.5 Of which on premises 12.9 34.3 48.3 59.9 67.2 47.9 Borehole 7.1 8.7 8.1 9.3 8.1 8.3 Other improved sources 26.6 21.3 17.2 11.7 7.2 15.6 Unimproved sources 50.1 32.0 19.3 14.1 6.8 21.9 Other 0.7 1.1 2.0 2.6 5.7 2.7 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source: Authors’ estimates based on the 2011 ECOM survey. TABLE 6.2: Type of Toilet by Welfare Quintile is even worse in rural areas where only 13 percent of the population has access to an improved toilet. Households Population Poverty is correlated with lack of toilets. More Urban Rural Total Urban Rural Total concerning is the share of the population with no toi- Improved, not 15.2 4.7 11.3 18 5.5 13.5 let at all. In the poorest quintile, close to a quarter of shared facility households simply do not have a toilet at all. The situ- Shared facility 43.5 7.6 30.3 41.3 7.7 29.2 ation is particularly of concern in four departments: Non-improved 41.4 87.8 58.5 40.6 86.7 57.4 facility Plateaux, Lékoumou, Cuvette, Cuvette-Ouest. The share Total 100.0 100.0 100.0 100.0 100.0 100.0 of households with no toilet ranges from 42.8 percent to Source: DHS 2012. 30 percent in these departments (Figure 6.16.2). Here FIGURE 6.16: Correlation between Access to Improved Water and Sanitation and Poverty 1. Improved water and poverty 2. Improved sanitation and poverty 50 50 45 45 y = 10.522e0.0335x Cuvette-Ouest Cuvette-Ouest 40 40 R² = 0.5172 Monetary poverty (FGT1) Monetary poverty (FGT1) 35 Cuvette 35 Cuvette Likouala 30 Lékoumou Likouala Sangha 30 Lékoumou Niari Sangha 25 Pool Plateaux Niari Bouenza 25 Bouenza Pool Plateaux 20 Kouilou 20 Kouilou 15 15 10 y = 123.84e–0.03x 10 R² = 0.7462 Brazzaville 5 Brazzaville 5 Pointe-Noire Pointe-Noire 0 0 30 40 50 60 70 80 90 100 0 5 10 15 20 25 30 35 40 45 50 Access to Improved Water Source (%) Share of households with no toilet (%) Source: Authors’ estimates based on the 2011 ECOM and DHS surveys. 100 Republic of Congo – Poverty Assessment Report FIGURE 6.17: Residential Network Water Coverage, Access, and Take-Up Rates (%) 1. Trend in grid coverage, access, and take-up rates 2. Grid access, take-up, and coverage rates by decile, 2011 60 100 50 80 40 60 30 40 20 10 20 0 0 2005 2011 1 2 3 4 5 6 7 8 9 10 ROC Welfare deciles Access to piped water at the PSU level Access to piped water at the PSU level Connected to piped water Take up rate Connected to piped water Take up rate Source: Authors’ estimates based on the 2005 and 2011 ECOM surveys. again, the gap between the two main cities and the rest Supply factors, as well as the combination of of the country is very pronounced. supply and demand factors, are the main barrier to access. Poor households are more likely to state availabil- Residential network water ity of network as the main reason for not being connected Over the last decade, there was no improvement (Figure 6.18.1), while the better-off are more likely to in access to piped water. Access remains very low. state cost as the reason for not been connected. Further Connection rates to residential network water remain analysis based on the Wodon et al. (2009) approach sug- low, with only 27 percent of households who were con- gests that supply factors account for a majority of the nected in 2011 (Figure 6.17.1). gap (52.1 percent), followed closely by combined factors Coverage is much higher in urban areas. In (43.7 percent) and demand-side factors (4.3 percent). Brazzaville, close to half of the households (47.3 per- There are differences across location. In rural areas, the cent) have a piped water connection. In Pointe Noire combined factor is dominant, while the supply factor and other municipalities, connection rates are a bit lower is dominant in urban areas. This means that for those (37.8 and 27.9 percent, respectively). On the other in rural areas, making the network available will not be hand, only 1 percent of rural households are connected enough because the affordability (demand) issue will to the network. In rural areas, access and take-up rate remain. In addition, pro-poor policies such as connection are low. Only 4.4 percent of rural households live in subsidies, and cross-urban-rural consumption subsidiza- a neighborhood with the network, and when the net- tion in tariff design should be considered. work is available, only 23.1 percent can afford to have In contrast to electricity, satisfaction with a connection. residential network water service is high. Frequent As expected, poor households are less likely shortages and poor quality of water are the main rea- to be connected to the residential network water. sons for dissatisfaction. Close to seven out of ten (66 Coverage rates are much higher among households in percent) households are satisfied with the residential the top deciles (Figure 6.17.2). In fact, connection rates network water service. The poor seem more satisfied. are virtually inexistent in the bottom 30 percent of the Satisfaction decreases with welfare; 73 percent of those in population. Poor households often do not have residen- the bottom quintile are satisfied as opposed to 64 percent tial network water in their neighborhood, and when the of those in the richest quintile. By far, frequent shortage network is present, they cannot afford a connection. is the main reason for dissatisfaction (Figure 6.18.2). Six Access to Basic Social Services 101 FIGURE 6.18: Reasons for Not Subscribing and Main Issue with Residential Network Water, 2011 1. Reasons for not subscribing 2. Satisfaction with the service 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Welfare quintile Total Welfare quintile Total Too expensive to subscribe Consumption too expensive Too expensive Frequent shortages Complicated process Remote network (not accessile) Poor quality Service interruption / landslide Useless Network not avaliable in the area Not concerned Source: Authors’ estimates based on the 2011 ECOM survey. FIGURE 6.19: Pro-Poorness of Sources of Drinking Rivers, unprotected wells, protected wells, and Water public taps are the only sources of water that are pro-poor. As shown in Figure 6.19, which provides 100 concentration curves for various sources of water, rivers, Cumulative distribution of source (%) 90 80 unprotected wells, protected wells, and public taps are 70 the only sources that lie above the equity line. In rela- 60 tion to the Lorenz curve, it appears that water tanks, 50 40 rainwater, village pumps, and piped from neighbor are 30 progressive. SNDE piped into dwelling and SNDE else- 20 where are to some extent regressive. 10 0 Residential network water tariffs are poorly 0 10 20 30 40 50 60 70 80 90 100 designed in the Republic of Congo. The implicit Cumulative distribution of the population (%) consumption subsidies are poorly targeted. Connection SNDE piped water into dwelling Equity subsidies clearly have the potential to be better targeted. SNDE piped water elsewhere Protected well For the Republic of Congo, the value Ω was 0.54 in SNDE piped water from neighbors Water tank 2011, that is, implicit subsidies are not pro-poor in River / creek / source Drilling / Pump Village the Republic of Congo. Still, the Republic of Congo is Unprotected well Public tap among the best performers in Africa (Figure 6.20). To our knowledge, residential network water tariffs have Welfare 2011 Rainwater not been revised for a very long time. Tariff designs Source: Authors’ estimates based on the 2011 ECOM survey. that are pro-poor should be considered in the future. Further simulations suggest that connection subsidies out of ten households point to frequent shortage as the will have much better targeting performance than con- reason for dissatisfaction. Poor quality is mentioned by sumption subsidies. As noted for electricity, connec- close to three out of seven households. tion subsidies have the potential to be better targeted 102 Republic of Congo – Poverty Assessment Report than consumption subsidies. However, they should be FIGURE 6.20: Targeting Performance (Ω) of implemented at scale only when generation capacity is Residential Network Water Subsidies, sufficient, and when considered, they need to be imple- Selected Countries mented well to ensure good targeting and limit costs. Burkina 0.02 Evidence suggests that the SNDE is struggling Burundi 0.15 to effectively collect payments from residential cus- Cameroon 0.30 tomers, including through the installation of new Cape Verde 0.24 CAR 0.66 meters in some areas. The share of households paying Chad 0.26 for piped water is systematically lower than the share of ROC 0.54 households who declare having a piped water connec- Côte d'ivoire 0.28 tion. This may be an indication of illicit connections, Gabon 0.64 but it may also reflect late payment, lack of recovering Ghana 0.12 late payment, or other issues. The differences between Guinea 0.12 those having a connection and those paying for it is Malawi Blantyre 0.15 Malawi Lilongwe 0.04 significant. In 2011, 26.7 percent of households were Niger 0.27 connected to the piped water network, but only 18.3 Nigeria FCT 0.36 percent were paying for it, generating a 8.4 percentage Nigeria Kaduna 0.53 point gap between coverage and payment. In 2005, the RDC 0.43 gap was at 9 percentage points. Thus, as a share of the Rwanda 0.01 coverage rate, the gap has remained quite stable (33.8 Senegal 0.77 in 2005 against 31.5 in 2011). This may suggest no Togo 0.49 Uganda 0.0002 improvement in the ability of the SNDE to collect pay- ments from residential customers in recent years. 0.0 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 Omega 6.6  Information and communication Source: Tsimpo and Wodon 2014b. technology and roads With regard to ICT, the Republic of Congo is per- of the Internet. At US$1,200, the Republic of Congo forming very well in terms of the mobile cellular was ranked as one of the most expensive countries in the network, but performance is below expectations in world in terms of price of international bandwidth Mbps terms of Internet access. ICT provides new opportu- per month (Jacquelot quote by World Bank 2016). nities to boost productivity and save costs. Moreover, The country is also performing below expecta- applications in various areas such as education, agri- tions in terms of connectivity and road density. In culture, health, and so on have proven to be critical for 2014, the country barely had 5 km of roads per 100 km2 economic development and well-being. For now, the of land, which is way below expectations (Figure 6.22). country seems to have harvested the low-hanging fruit A possible explanation for this very low density of roads as materialized by the strong performance of access to a could be the fact that the vast majority of the popula- mobile phone network (Figure 6.21). The country is still tion is concentrated in the two main cities. Still, the low struggling to reach the next level. Access to the Internet density of roads will definitively translate into a connec- is very low. It is estimated that only 7 percent of the tivity problem and will result in economic inefficiencies, population were using the Internet in 2014. Beyond the for example, high transport cost, difficulties for farmers low quality, prices remain far too high for the general to access markets, and many more challenges related to public and could be the main reason for the low usage transport and accessibility. Access to Basic Social Services 103 FIGURE 6.21: Mobile Cellular Network and Internet Users Per 100 People 1. Population covered by mobile cellular network (%) 2. Internet users (per 100 people) 100 60 Internet users (per 100 people) ROC 2011 mobile cellular network (%) Population covered by 80 40 60 20 40 ROC 2005 ROC 2014 ROC 2005 0 20 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. FIGURE 6.22: Road Density (km of Road Per 100 rate was 74 percent in 2012 and has been fluctuating up Km2 of Land Area) and down over the last decade. Children from poor families, autochthons, and (km of road per 1000 sq. km of land area) 200 those with disability are less likely to be enrolled. Enrollment rates are higher for the nonpoor than for the 150 poor, especially at the secondary level, and the same is Road density observed when considering welfare quintile. Autochthon 100 children are much less likely to go to school than Bantu children, with virtually none enrolled in secondary 50 schools. Children with a disability are at a clear disad- 0 ROC 2009 vantage. There are finally large differences in enrollment ROC 2000 rates, especially again at the secondary level, by location 0 2000 4000 6000 8000 10000 or geographic area. GNI per capita, PPP (constant 2011 international $) Affordability and early marriage/pregnancies Source: Authors’ calculation using the WDI data and based on the Gable, are the main reasons for children not being enrolled. Lofgren, and Osorio-Rodarte (2015) approach. Cost is the main reason not to go to school for both children of primary school age and children of sec- ondary school age who have completed the primary 6.7 Conclusion cycle and are thus eligible for secondary school (at the primary level, ‘other’ reasons not specified in the This chapter has provided a basic diagnostic of access to data play a role for boys). Lack of interest in schooling basic social services with a focus on cross-country com- comes up next, which may denote a lack of opportu- parison and constraints to access, especially for the poor. nity associated with schooling, at least in some areas. Performance is way below expectations on most This is especially the case for boys, but for girls at the of the other critical education sector indicators, secondary level, the issue of an illness or more likely including primary school completion. The Republic of a pregnancy is a more important factor for dropping Congo is performing below expectations in terms of pri- out, especially when one also factors in marriage as mary school completion. The primary school completion another reason to drop out. Other factors, including 104 Republic of Congo – Poverty Assessment Report work (opportunity cost of schooling) and distances to The country is performing above expectations schools, also play a role. of stunting, but malnutrition is still high. Stunting, The private sector is playing a very important defined as low height for age and an indicator of chronic and increasing role in the provision of education malnutrition, decreased from 31 percent in 2005 to 25 services. The share of students in private schools seems percent in 2011. As a result, the country is now perform- to have increased between the last two surveys for 2005 ing as expected in comparison to peers. Nevertheless, the and 2011. In 2011, 35 percent of students were enrolled level of stunting remains quite high. High malnutrition in a private school. This is up by 15 percentage points reduces agricultural productivity, contributes to poverty, compared to 2005. Such high shares of private provi- and affects education and intellectual potential of school sion in a country where 40 percent of the population is children (for example, stunting causes children to start poor is an indication of inadequate success in ensuring school late because they look too small for their age effective access by the Government. and will also be a cause of absenteeism and repetition There has been substantial improvement in child of school years). and maternal mortality. However, the country still Despite improvement over the last decade, performs below expectations with regard to maternal access to electricity is very low compared to expecta- mortality and has not reached any of the health-related tions. Connection rates in the country remain below MDGs. Child and maternal mortality are often used as expectations compared to peers. Coverage rates have a measure of the efficiency of the health sector in a given increased substantially, from 26.7 percent in 2005 to country. Between 2005 and 2012, under-five mortal- 42.5 percent in 2011. The improvement in coverage is ity, which measures the probability of children dying the result of improvement in both access (network avail- between birth and the fifth birthday, dropped from 95.3 ability) and take-up. In 2011, close to seven out of ten to 52.6 per 1,000 births. Owing to this improvement, households (68 percent) lived in neighborhoods with the country is now performing as expected given its electricity network; this is up from 57 percent in 2005. GNI level. The story is a bit different regarding mater- Reflecting improvement in monetary poverty, take-up nal mortality. Despite improvement, the country is still rate also increased from 47 percent to 62 percent between performing lower than its peers on maternal mortality. 2005 and 2011. Maternal mortality rate declined from 781 to 426 deaths In urban areas, affordability is the main barrier per 100,000 live births between 2005 and 2012. to access, while in rural areas both availability of net- The private sector plays an important role in work and affordability are present. Poor households providing essential health services. The private sector, are more likely to state availability of network as the especially in large urban centers, plays an important role main reason for not being connected. Further analysis in basic health service provision. Generally the quality suggests that even if network was to be made available to of public providers is perceived to be better than private them, affordability issues will prevent them from being providers, particularly for maternal care. However, people connected. This means that in some areas, making the often choose to seek care at private providers because network available will not be enough because the afford- of proximity in urban areas, extended opening hours ability (demand) issue will remain. Electrification pro- and shorter waiting times and better perceived quality grams should take into account pro-poor policies such as of laboratory services as compared to the public sector connection subsidies and cross-urban-rural consumption (Makinen, Deville, and Folsom 2012). Price is also an subsidization in tariff design. important determinant of health care demand: in some Although the share of the population with access cases, the private for-profit sector charges lower prices than to improved water slightly increased during the last the public sector and is more flexible in offering payment decade, it is still far below what is expected. The terms to the poorest patients who have difficulty paying. Republic of Congo is performing below expectations Access to Basic Social Services 105 in terms of access to improved water. Between 2005 harvested the low-hanging fruit as materialized by the and 2015, access to improved water increased slightly strong performance of access to a mobile phone network. from 72 percent to 77 percent, but the country is still However, the country is still struggling to reach the next performing below expectations. level. Access to Internet is very low. It is estimated that Access to improved sanitation remains very only 7 percent of the population were using the Internet low, and as a consequence, the country is perform- in 2014. Beyond the low quality, prices remain far too ing below expectations on this dimension as well. high for the general public and could be the main reason The Republic of Congo is performing below expecta- for the low usage of the Internet. tions regarding access to a safe toilet. In 2014, only The country is also performing below expecta- 43 percent of the population had access to improved tions in terms of connectivity and road density. In sanitation. The situation is even worse in rural areas 2014, the country barely had 5 km of roads per 100 where only 13 percent of the population has access to km2 of land, which is way below expectations. A pos- an improved toilet. sible explanation for this very low density of roads could With regard to ICT, the Republic of Congo is be the fact that the vast majority of the population is performing very well in terms of the mobile cellular concentrated in the two main cities. Still, the low den- network, but performance is below expectations in sity of road will definitively translate into a connectivity terms of Internet access. The country seems to have problem and will result in economic inefficiencies. 106 Republic of Congo – Poverty Assessment Report Annexes Annex 1: Consumption Aggregate To compute poverty measures, three ingredients are needed (see, for example, Coudouel, Hentschel, and Wodon 2002). First, one has to choose the relevant indicator of well-being. Second, one has to select a poverty line—the threshold below which a given household or individual will be clas- sified as poor. Finally, one has to select a poverty measure for reporting for the population as a whole or for a population subgroup only. This annex documents choices made for the estimation of the household-level welfare aggregate as well as some issues of comparability between the two surveys and whether the issues are major or not. Changes in Questionnaires A number of changes in survey questionnaires were made between 2005 and 2011. These changes make it impossible to estimate household expenditures in exactly the same way in both years. At the same time, however, these changes are not so severe as to invalidate poverty comparisons over time based on the two surveys. A first change in the questionnaires relates to the frequency of acquisitions of food items. In 2005, the questionnaire had two questions to capture the frequency of purchases. One question captured the number of times that a household bought a good, while the second question captured the period considered (daily, weekly, monthly, and so on). In 2011, instead of two questions, the questionnaire collapsed the two questions into a single one (Figure A1.1). Another difference in the questionnaires for the two years relates to the way in which infor- mation on durable goods was collected. In 2005, the questionnaire included a question on the number of durable goods owned by the household. In 2011, that question was removed, so it is not possible to know the number of items owned by the households (see Figure A1.2). This omis- sion is likely to have a negligible effect on poverty measures for two reasons. Most poor do not own (many) durable goods. In addition, in 2005, most households who did own durable goods only owned one item per category (with exceptions for chairs, beds, in some cases air condition- ers, eyeglass frames, and crutches, but these exceptions are not too significant). 107 FIGURE A1.1: Differences in Capturing Reference Periods for Food Items between Surveys 108 ECOM 2005 DEPENSES ET ACQUISITIONS QUOTIDIENNES DE PRODUITS OU DE SERVICES POUR LA CONSOMMATION DU MENAGE Mode d’acquisition Unité Avec Fraction ? Quelle quantité de (PRODUIT, SERVICE) Quel est le montant total de cette Fréque nce Unité de temps 1. achat 1= Oui avez-vous consommée ce jour ? dépense ? de renouve 1. Jour 2. auto-consommation 2= Non QUANTITE PAYEE PRIX UNITAIRE (en francs Fcfa) llement 2. Semaine 3. cadeau reçu (en francs CFA) 3. Mois 4. Année (4) (5) (6) (7) (8) (9) (10) (11) |__| |__|__| |__| |__|__|__|__|__| |__|__|__|__|__|__|__| |__|__|__|__|__|__|__|__| |__|__| |__| Republic of Congo – Poverty Assessment Report ECOM 2011 Section N°GRAPPE N°MENAGE JOUR DATE Nombre de lignes DEPENSES MONETAIRES ET ACQUISITIONS QUOTIDIENNES DE PRODUITS OU DE SERVICES |__|__| |__|__| |__|__| 03 |__|__|__|__| |__|__| |__|__| |__|__| POUR LA CONSOMMATION DES MENAGES PENDANT UNE PERIODE DE 15 JOURS Jour Mois Année N°Ligne Qu’avez-vous effectivement consommé ? Quelle quantité de (produit, service) avez-vous Quel est le Fréquence de Lieu Origine du Etat à consommé ce jour ? montant total de renouvellement d’achat produit l’achat Code du DESCRIPTION CODE PRODUIT Quantité Unité Prix unitaire cette dépense ? 0. Moins de 15 1. Local 1.Neuf titulaire PRECISE DU PRODUIT A inscrire au plus payée (en francs CFA) jours 2. Importé 2.Usagé du carnet OU DU SERVICE tard chaque soir 1. Quinzaine (CEMAC) 3. Service CONSOMME après vérification du 2. Mois 3. Importé RDC/ 4. Non (y compris les cadeaux questionnaire 3. Trimestre Angola applicable donnés et reçus en 4. Semestre 4. Importé (Afrique) nature) 5. Année 5. Importe (autres 9. Ne sait pays) pas ou achat 6. Service exceptionnel 9. NSP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) |__|__| |__|__|   |__|__|__|__|__|__|__| |__|__|__|__| |__|__| |__|__|__|__|__|__|__|__| |__|__|__|__|__|__|__|__| |__| |__|__| |__| |__| |__|__| |__|__|   |__|__|__|__|__|__|__| |__|__|__|__| |__|__| |__|__|__|__|__|__|__|__| |__|__|__|__|__|__|__|__| |__| |__|__| |__| |__| Source: National Statistical Office (NSO) – Republic of Congo. FIGURE A1.2: Difference in Capturing Durable Goods between Surveys ECOM 2005 SECTION 08 : BIENS DURABLES     Q1 Q2 Q3 Q4 Q5 Biens Codes Votre ménage possède-t-il un/ Combien votre ménage Depuis combien d’années le Quel est le prix d’achat de votre A combien revendriez-vous une [Nom Bien] ? possède-t-il de [Nom Bien]? dernier [Nom Bien] a-t-il été [Nom Bien] le plus récent ? aujourd’hui votre [Nom Bien] le 1= oui acquis ? (âge) (en milliers F CFA) plus récent ? 2=Non=> suivant (en milliers F CFA) Meubles, articles d’ameublement, tapis et autres revêtements de sol     Table 001 |__| |__|__| |__|__| |__|__|__|__|__|__| |__|__|__|__|__|__| ECOM 2011 Section N°GRAPPE N°MENAGE Nombre de Lignes POSSESSION DES BIENS DURABLES 35 |__|__|__|__| |__|__| |__|__| en état d’utilisation par le ménage N° ligne Votre ménage possède-t-? Année Valeur d’acquisition Etat à A combien revendriez-vous DESCRIPTION PRECISE DU BIEN CODE PRODUIT d’acquisition (Inscrire la valeur 0 si c’est l’acquisition aujourd’hui votre DURABLE DU MENAGE A inscrire au plus tard chaque soir après cadeau reçu en FCFA) 1. Neuf [Nom Bien] ? vérification du questionnaire 2. Usagé (en francs Fcfa) (1) (3) (4) (5) (6) (7) (9) |__|__|   |__|__|__|__|__|__|__| |__|__| |__|__|__|__|__|__|__|__| |__| |__|__|__|__|__|__|__|__| Source: NSO – Republic of Congo. Annexes 109 Number of Items Number of Items Included in the TABLE A1.1:  Consumption Modules Between 2005 and 2011, the number of consumption Section 2005 2011 items on which information was collected increased Food, beverages, tobacco, and narcotics 202 699 considerably. A total of 2,038 categories were included Clothing and footwear 94 248 in the 2011 questionnaire versus only 767 categories in Housing, water, electricity, gas, and other fuels 34 103 2005 (see Table A1.1). For example, in 2005 rice was Furnishings, household equipment, and 87 239 considered as a single item, but in 2011 rice was split routine maintenance into nine varieties (local long grain white rice, long Health 49 185 grain rice, regular rice with differences, imported rice Transport 56 130 uncle bens, imported rice lustucru, medium grain rice, Communication 17 34 short grain rice, basmati rice, and broken rice). This Recreation and culture 120 120 could result in an increase in the measures of household Education 20 46 consumption from 2005 to 2011, but at the same time, Restaurants and hotels 15 75 the increase may not be large. To consider the example Miscellaneous goods and services 73 159 of rice, households typically know how much they con- Total 767 2,038 sume, and collecting information on different varieties Source: Authors’ compilation based on survey questionnaires. of rice need not necessarily imply that recorded con- sumption will increase. In addition, most households, especially the poor, tend to spend the most on a few Auto-consumption items, and these items are included in both surveys. The surveys include a module on agricultural crops, Purchase of Food Items with data on quantities produced, sales, and the value of sales for 24 crops. The list includes two cash crops (cof- The ECOM surveys capture data on food purchases fee and cocoa) and crops such as cassava, maize, banana, in two modules. One module is for frequent food beans, irish potato, sweet potato, and yam, among others. purchases and the other for less frequent purchases. Unfortunately, the modules do not capture well enough the For the consumption aggregate, only purchases possible uses of agriculture production, so there are risks of renewed every two weeks or less for the first mod- overestimation or underestimation of auto-consumption. ule and purchases renewed less often than every two The assumption used to estimate auto-consumption was weeks for the second module were considered. In each that the difference between production and sales was con- case, appropriate expansion coefficients were used to sumed by the household. This is an overestimation given extrapolate the data into annual expenditure. When that other alternative uses such as losses, investments in errors were detected for the recorder frequencies, cor- stocks, and gifts are also potential uses of production. The rections were made. Extreme outliers were also cor- quantities assumed to be auto-consumed were valued at rected. Such corrections can have implications for the the median sales price computed for each strata. estimation of the Gini coefficient, the share of food in overall expenditures, and the daily calorie intake Durable Goods per equivalent adult of households. Relatively few corrections had to be made partly because the data When computing the welfare aggregate, it is recom- set already had gone through a cleaning process with mended to rely on the usage value of durable goods the National Statistical Office. because durable goods last for several years. For example, if a household buys a car, it will be inappropriate to add 110 Republic of Congo – Poverty Assessment Report the total value of the purchase to the welfare aggregate. so on) should be accounted for in the durable good sec- Instead, a usage value that takes into account depreciations tion. On the other hand, it may be misleading to add should be used. The computation of usage values takes all health expenditures, especially hospitalization, to the into account the purchase value, the resale value, the age welfare aggregate, as this reflects a regrettable necessity of the good, interest rates, and inflation. Assuming an that does nothing to increase welfare per se. Thus hos- annual real interest rate of 6 percent, the usage value is pitalization expenditures are often excluded. This is also given by the following formula: Usage value = Quantity × the approach followed here. Current value × (Depreciation rate + 0.06). Note that in the 2011 survey, there is a more exhaustive list of durable Education Expenditures goods (116 goods in 2011 versus 51 in 2005). On the other hand, the 2011 ECOM did not ask the question A few education expenditures (uniforms and books) were on the number of goods owned; hence, the assumption listed in different sections of the questionnaire than is was made that households owned only one item per good. usually the case. Information on school uniforms were collected under clothing. Similarly, expenditures on Imputed Rent and Other Housing school books were collected under the culture and leisure Expenditures section. All education expenditures were reflected in the consumption aggregate. The importance of using econometric modeling for rent imputation for welfare measurement is well recognized Spatial Deflator (for example, Balcazar et al. 2014). Surveys often collect information on monthly rent for those renting their To account for spatial price differences, a deflator was dwelling. The literature39 recommends imputing a rent used to adjust regional expenditure into same monetary value for those who are not subject to rent (owners and metrics. In 2005, the food basket was valued for each those for whom housing is provided free of charge). OLS of the five strata. Nonfood expenditure was computed and Heckman two-stage estimation methods can be used. to derive a cost of basic needs (CBN) poverty line for The same econometric model was used in both surveys. each strata. Pointe Noire was then used as reference. The The parameter estimates were then used to calculate rents ratio of the CBN between each strata and Pointe Noire for nonrenter. The dependent variable is the logarithm of was used as the regional deflator. For 2011, because of monthly rent. The explanatory variables include location, archiving issues, regional prices were not available. It neighborhood, and dwelling characteristics. was, therefore, not possible to estimate the CBN for the Household expenditures on housing are considered basket in 2011. Instead, the Fisher price index as esti- as investments and excluded. The underlying principle mated by Afristat was used as the regional deflator. It is is the same as for other durable goods. The estimation important to note that in 2005, 5 strata were considered, of the rental value of the dwelling implicitly reflects the while in 2011, 12 strata/regions were considered; hence, value to households of expenditures on housing. the spatial deflator was disaggregated. Health Expenditures Adult Equivalence Scale While computing the welfare aggregate for a given house- Due to economies of scales and the fact that calorie hold, the literature recommends that particular attention intakes differ across life span and gender, consumption should be paid to exceptional health expenditures such as hospitalization and health equipment. Health equip- 39 See for example, Haughton and Khandker (2009) and Balcazar ment and accessories (dental prostheses, eye glasses and et al. (2014). Annexes 111 TABLE A1.2: FAO Equivalent Adult Scale TABLE A1.3: Population by Location Equivalent adult scale 2005 2011 Male Female   Population Share (%) Population Share (%) 0–1 year old 0.27 0.27 Brazzaville 1,029,980 29.0 1,502,487 37.1 1–3 years old 0.45 0.45 Pointe Noire 833,109 23.5 777,676 19.2 4–6 years old 0.61 0.61 Other 210,626 5.9 232,560 5.7 7–9 years old 0.73 0.73 municipalities 10–12 years old 0.86 0.73 Semi-urban 250,069 7.0 169,599 4.2 13–15 years old 0.96 0.83 Rural 1,227,715 34.6 1,370,004 33.8 16–19 years old 1.02 0.77 Total 3,551,500 100.0 4,052,326 100.0 Source: Authors’ estimation from the expansion factors in the surveys. 20–50 years old 1.00 0.77 51 and above 0.86 0.79 Source: FAO. TABLE A1.4: Budget Shares by Consumption Category and Mean Consumption Aggregate aggregates are often constructed in terms of consumption 2005 2011 per equivalent adult rather than in per capita terms. The Food and nonalcoholic beverages 38.6% 38.7% FAO scale used to account for different caloric needs by Alcoholic beverages, tobacco, and 1.7% 1.3% age and gender is provided in Table A1.2. narcotics Clothing and footwear 5.6% 7.1% Outliers Housing, water, electricity, gas, and 18.5% 17.0% other fuels Expenditure levels above the mean plus three times the Furnishings, household equipment, and 3.5% 4.2% routine household maintenance standard deviation were adjusted to three times the stan- Health 4.2% 1.7% dard deviation. Although this affected very few observa- Transport 6.4% 10.0% tions, the way outliers are corrected can have important Communication 3.6% 7.3% effect on the Gini coefficient. Recreation and culture 1.7% 1.4% Education 2.2% 3.8% Sampling Restaurants and hotels 3.0% 2.8% The sample frame for the 2011 survey was updated based Miscellaneous goods and services 11.2% 4.7% on the 2007 census results. The new population structure Total expenditure 100.0% 100.0% seems to suggest a shift toward Brazzaville. The census Annual expenditure per capita, nominal 256,223.2 417,390.1 (CFA francs) results are questionable. Unfortunately, with no access to the sample frame or detailed census results, it is not Annual expenditure per equivalent adult, 328,397.5 535,283.7 nominal (CFA francs) feasible to confirm whether a population shift toward Gini index 0.460 0.465 Brazzaville indeed took place. Source: Authors’ estimation. Budget Shares categories, the budget shares are similar between the two Budget shares resulting from the estimation of the con- years, but there are also some changes that may result sumption aggregate are provided in Table A1.4. For many from behavioral changes or data comparability issues. 112 Republic of Congo – Poverty Assessment Report Annex 2: Poverty Line provide 2,400 kcal. Using prices observed in 2005, the food poverty line is thus estimated as: Following standard practice, the CBN approach was n used to estimate poverty lines. The approach includes k ∑ Q i × Pik two main steps: (a) construction of the food poverty ZF = 2400 × i− n 1 line and (b) construction of the nonfood poverty line. ∑i−1 Q i ×C i The food poverty line is obtained on the basis of the main food items consumed in the country. Expenditure with Qi being the average daily quantity of product i on these items and price data are used to compute the consumed, n Ci the caloric value corresponding to product quantities consumed. Conversion tables are then used k ∑ i consumed, Q k i × Pi being the average price of product and to convert daily quantities consumed into kilocalories Z F =i 2400 × k. in area i−1 n (kcal). For most African countries, the main food items ∑ Once i−1 Q thei× C i poverty line has been estimated, food typically consumed by the population generate between various approaches can be used to estimate the nonfood 1,500 and 2,000 kcal per equivalent adult. The nor- poverty line. One approach is to consider spending on mative food basket used for the food poverty line is nonfood items by households whose welfare aggregate obtained by adjusting upward the actual food basket is at the food poverty line. Then nonfood expenditure consumed by the population to meet basic needs as rec- of these households is then used as the nonfood poverty ommended by nutritionists. Often the kcal threshold is line. The sum of the food and nonfood poverty lines set at 2,100–2,500 kcal per equivalent adult. The mode generates the overall poverty line. Other techniques can of the thresholds used by African countries as well as the also be used. Republic of Congo is 2,400 kcal. As mentioned earlier, the normative food basket The food basket used to estimate the food poverty used for the food poverty line was based on data from lines is based on data from the 2005 ECOM survey. The the 2005 ECOM survey and this basket was retained basket consists of 43 food items that represent 85 percent in 2011 as it is often recommended not to change the of food consumption (Table A2.1; the names of the food food basket for comparisons of poverty within a decade. items used have been kept in French for clarity of iden- Because detailed price information for all goods in the tification in the survey). In the ECOM 2005 survey, the basket was not available in 2011, the food CPI was used basket generates only 1,031 kcal, which is very low, and to update the food poverty line between 2005 and 2011. may suggest that food consumption is underestimated. The resulting poverty line in 2011 was XAF 274,113 per Still, the food basket represents the items actually con- equivalent adult and per year. Table A2.2 provides the sumed by households. It can therefore be scaled up to food and nonfood poverty lines for both years. Annexes 113 TABLE A2.1: Food Basket Composition Based on ECOM 2005 Data Observed consumption Adjusted consumption Quantity (kg) Energy (kcal) Quantity (kg) Energy (kcal) Farine de manioc (foufou, attiéké, gari) 0.078 284.566 0.182 662.187 Manioc cuit (Moungouélé, gros manioc) 0.023 58.583 0.053 136.323 Riz 0.017 60.311 0.039 140.345 Huile d’arachide 0.010 90.431 0.024 210.433 Autres poissons salés et séchés 0.003 9.149 0.008 21.289 Pain de blé industriel en baguette (gro) 0.016 42.823 0.038 99.649 Chinchard frais 0.010 11.281 0.022 26.251 Hareng fume 0.003 10.601 0.007 24.669 Autres poissons d’eau douce fumés 0.002 7.391 0.005 17.199 Sel 0.003 0.000 0.008 0.000 Autres poissons d’eau douce frais 0.001 1.331 0.003 3.097 Viande de bœuf fraîche 0.003 13.124 0.007 30.540 Poulet congelé 0.004 5.750 0.010 13.380 Huile de palme 0.004 31.331 0.008 72.906 Mfumbu/coco (gnetum) 0.002 1.395 0.005 3.245 Morceaux de poulet (ailes, cuisses, gés) 0.006 7.917 0.013 18.423 Hareng frais 0.001 0.873 0.002 2.032 Autres poissons de mer frais 0.002 2.460 0.005 5.725 Oignon/ciboule frais 0.004 1.548 0.009 3.602 Haricots secs 0.003 9.769 0.007 22.733 Lait en poudre 0.050 249.715 0.116 581.088 Sucre en poudre 0.008 32.052 0.019 74.586 Autres légumes frais n.d.a. (aubergine) 0.007 6.043 0.016 14.062 Pâte d’ arachide locale 0.003 15.554 0.006 36.195 Feuille de manioc 0.006 5.899 0.015 13.727 Beignet à base de farine de blé 0.002 5.260 0.005 12.240 Gibier frais 0.000 0.083 0.000 0.192 Sucre en morceaux 0.001 1.985 0.001 4.620 Noix de palme 0.004 13.026 0.009 30.312 Pain de blé industriel en baguette (pet) 0.004 10.711 0.010 24.924 Abats et tripes de bœuf (estomac, foie) 0.005 12.188 0.012 28.362 Concentré de tomate 0.020 6.082 0.047 14.152 Pain de blé local artisanal 0.000 0.075 0.000 0.174 Cube (Maggi, Jumbo, and so on) 0.000 0.443 0.001 1.031 Tomate fraîche 0.003 0.571 0.006 1.330 Autres poissons de mer fumés 0.001 2.373 0.001 5.522 (continued on next page) 114 Republic of Congo – Poverty Assessment Report TABLE A2.1: Food Basket Composition Based on ECOM 2005 Data (continued) Observed consumption Adjusted consumption Quantity (kg) Energy (kcal) Quantity (kg) Energy (kcal) Endive 0.003 1.892 0.006 4.403 Banane plantain 0.000 0.247 0.001 0.576 Thon salé et séché 0.000 0.515 0.000 1.198 Ntsinga fume 0.000 1.358 0.001 3.161 Conserves de poissons (sardines, thons) 0.003 9.740 0.006 22.665 Lait concentré 0.001 2.139 0.002 4.976 Abats et tripes de porc (estomac, foie) 0.001 2.784 0.003 6.477 Total 1,031 2,400 Source: Authors’ estimation. TABLE A2.2: Poverty Lines for 2005 and 2011 2005 2011 Food poverty line 136,613 188,458 Nonfood poverty line 62,091 85,655 Overall poverty line 198,704 274,113 Source: Authors’ estimation. Annexes 115 Annex 3: Poverty Measures with z − yq I= This annex is reproduced with minor changes from z Coudouel, Hentschel, and Wodon (2002). It provides where expressions for commonly used poverty measures and 1 q for their decomposition by sector or by group. The note yq = ∑ yi is the averageconsumption of the poor. q i=1 focuses on the first three poverty measures of the FGT class, namely the headcount, the poverty gap, and the It must be emphasized that the consumption gap squared poverty gap. ratio I in itself is not a good measure of poverty. Assume that some households that are poor but close to the Poverty measures poverty line are improving their standards of living over time, and thereby become nonpoor. The consumption Poverty Headcount. This is the share of the population gap ratio will increase because the mean distance sepa- which is poor, that is, the proportion of the population rating the poor from the poverty line will increase (this for whom consumption per equivalent adult y is less than happens because some of those who were less poor have the poverty line z. Suppose we have a population of size emerged from poverty—so that those still in poverty are n in which q people are poor. The headcount index is: on average further away from the poverty line), suggest- q. ing a deterioration in welfare, while nobody is worst-off H= and some people are actually better-off. Although the n consumption gap ratio I will increase, the poverty gap Poverty Gap. The poverty gap, which is often considered PG itself will decrease because the headcount index of as representing the depth of poverty, is the mean distance poverty will decrease, suggesting an improvement toward separating the population from the poverty line, with the poverty reduction. The problem with the consumption nonpoor being given a distance of zero. The poverty gap gap ratio is that it is defined only on the population is a measure of the poverty deficit of the entire popula- that is poor, while the poverty gap is defined over the tion, where the notion of ‘poverty deficit’ captures the population as a whole. resources that would be needed to lift all the poor out of poverty through perfectly targeted cash transfers. It Squared Poverty Gap. This is often described as a measure is defined as follows: of the severity of poverty. While the poverty gap takes into account the distance separating the poor from the 1 q ⎡ z − yi ⎤ , PG = ∑⎢ n i=1 ⎢⎣ z ⎥⎦ ⎥ poverty line, the squared poverty gap takes the square of that distance into account. When using the squared where yi is the consumption of household i, and the sum poverty gap, the poverty gap is weighted by itself, so as is taken only for those households that are poor (with to give more weight to the very poor. Stated differently, appropriate weights). The poverty gap can be written the squared poverty gap takes into account the inequality as being equal to the product of the consumption (or among the poor. It is obtained as follows: income when that metric is used) gap ratio and the 2 headcount index of poverty, where the consumption (or 1 q ⎡ z − yi ⎤ . P2 = ∑⎢ n i=1 ⎢⎣ z ⎥⎦ ⎥ income) gap ratio is defined as: The headcount, the poverty gap, and the squared PG = I × H , poverty gap are the first three measures of the FGT class 116 Republic of Congo – Poverty Assessment Report of poverty measures. The general formula for this class It is important to use the poverty gap and perhaps of poverty measures depends on a parameter α that even the squared poverty gap in addition to the head- takes a value of 0 for the headcount, 1 for the poverty count for evaluation purposes because these measure gap, and 2 for the squared poverty gap in the follow- capture different aspects of poverty. Basing an evaluation ing expression: on the headcount only could consider as more effective policies that lift the least poor (those close to the line) out α 1 q ⎡ z − yi ⎤ . of poverty to the detriment of poorer households. The Pα = ∑ ⎢ ⎥ n i=1 ⎢⎣ z ⎥⎦ poverty gap PG and the squared poverty gap P2, on the other hand, put more emphasis on helping those who are further away from the line, the poorest of the poor. Annexes 117 Annex 4:  Assets Index and Assets- in estimates of the headcounts under the two approaches Based Poverty in 2011 because there may be ties in assets indexes. Third, FGT poverty measures were computed by comparing To assess the robustness of trends in consumption-based the assets index to the assets-based poverty line. Trends poverty, an assets index, also including education vari- in assets indexes and assets-based poverty measures were ables, was constructed, and assets-based poverty measures obtained by using 2011 as based year, and projecting the were estimated. The assets index was obtained with stan- 2011 factorial analysis parameters to 2005 data. 2011 dard factorial analysis. The key variables used are shown been the base year, the assets-based poverty line for 2011 in Table A4.1. Assets-based measures of poverty were was used for the two surveys. obtained in the following way. First, the assets index The approach used for estimating assets-based obtained through factorial analysis was normalized to poverty is arguably approximate, but it provides an take a value between zero and one. Second, for each of additional useful check to assess whether poverty trends the five strata in 2011, an assets-based poverty line was are reasonable. Importantly, as shown in Figures A4.1– derived in such a way that the assets-based headcount A4.2, there is a high degree of correspondence between of poverty would be equal to the consumption-based consumption-based poverty measures and assets-based equivalent poverty measure. There are small differences measures, suggesting that both move in concert. TABLE A4.1: Construction of the Assets Index, Pooled Data for 2005 and 2011 Factor loadings Scoring coefficients (pattern matrix) and unique variances (method = regression) Factor1 Uniqueness Factor1 Primary education −0.158 0.662 –0.020 Lower secondary –0.045 0.513 –0.006 Upper secondary 0.257 0.537 0.042 Tertiary education 0.432 0.539 0.071 Flat iron 0.716 0.394 0.133 Television 0.786 0.315 0.175 Computer 0.275 0.742 0.038 Radio 0.292 0.740 0.035 Mattress/bed 0.203 0.848 0.024 Modern armchair 0.584 0.601 0.079 Bicycle –0.059 0.900 –0.004 Motorcycle 0.100 0.901 0.010 Car or truck 0.206 0.793 0.029 Canoe –0.185 0.859 –0.023 Phone 0.459 0.606 0.072 Refrigerator/freezer 0.641 0.479 0.103 Land –0.458 0.675 –0.056 Piped water 0.618 0.513 0.095 118 Republic of Congo – Poverty Assessment Report TABLE A4.1: Construction of the Assets Index, Pooled Data for 2005 and 2011 Factor loadings Scoring coefficients (pattern matrix) and unique variances (method = regression) Factor1 Uniqueness Factor1 Electricity 0.777 0.314 0.168 Flush toilet 0.398 0.715 0.052 No toilet –0.326 0.820 –0.041 Improved roof 0.473 0.588 0.076 Improved wall 0.562 0.587 0.077 Improved floor 0.736 0.379 0.163 Source: Authors’ estimation. FIGURE A4.1: Assets Index and Consumption Per FIGURE A4.2: Assets-Based and Consumption- Equivalent Adult, 2011 Based Poverty Headcounts, 2011 700,000 Brazzaville 90 Cuvette-Ouest Pointe-Noire 80 600,000 Cuvette Lékoumou Monetary Poverty (FGT0) 70 Likouala Annual expenditure per Bouenza Plateaux 500,000 Kouilou ROC Niari equivalent adult 60 Kouilou Sangha Pool 400,000 Cuvette 50 Likouala Bouenza 300,000 Pool Niari 40 ROC Plateaux Sangha Lékoumou 30 Brazzaville y = 0.8908x + 4.4067 200,000 Cuvette-Ouest y = 820579x + 143664 20 Pointe-Noire R² = 0.9438 100,000 R² = 0.73126 10 0 0 0.1 0.2 0.3 0.4 0.5 0.6 20 30 40 50 60 70 80 90 Assets ownership index Assets ownership (FGT0) Source: Authors’ estimation. Source: Authors’ estimation. Annexes 119 Annex 5: Inequality Measure v −1 −v cov( y ,[1− F ] ) Gini(v ) = This annex is reproduced with minor changes from y Coudouel, Hentschel, and Wodon (2002). It provides Another family of inequality measures is the General expressions for commonly used inequality measures. As Entropy (GE) measure, defined as: is the case for poverty measures, some inequality mea- sures can be decomposed, but these decompositions are 1 ⎡⎢ 1 n ⎛ ⎜ yi ⎞ ⎟ ⎤ not used in this study. GE (α) = ∑ ⎜ α 2 − α ⎢⎢⎣ n i=1 ⎜ ⎝y⎟ ⎟ ⎟ − 1 ⎠ ⎥⎥⎦ ⎥, The standard Gini index or coefficient measures twice the surface between the Lorenz curve, which maps with the cumulative income share on the vertical axis against the distribution of the population on the vertical axis, 1 n y 1 n y y GE (0) = ∑ n i=1 log , GE (1) = ∑ i log i , yi n i=1 y y and the line of equal distribution. A large number of mathematical expressions have been proposed for the Gini index, but the easiest to manipulate is based on the and covariance between the consumption Y of a household n and the rank F of the household in the distribution of 1 ∑( y − y ) 2 GE (2) = i . consumption (this rank takes a value between zero for 2ny 2 i =1 the poorest and one for the = Gini , F ) / y denoting 2cov (YWith richest). mean consumption, the standard Gini index is: Measures from the GE class are sensitive to changes at the lower end of the distribution for α close to 0, equally sensitive to changes across the distribution for Gini = 2cov (Y , F ) / y . α equal to 1 (the Theil index), and sensitive to changes at the higher end of the distribution for higher values. The Gini index has attractive theoretical and sta- A third class of inequality measures was proposed tistical properties that other inequality measures do not by Atkinson. This class also has a weighting parameter ε have, which explains why it is used by many research- (measuring aversion to inequality) and some of its theo- ers. The extended Gini uses a parameter n to emphasize retical properties are similar to those of the extended Gini various parts of the distribution. The higher the weight, index. The Atkinson class is defined as follows: the more emphasis is placed on the bottom part of the 1 distribution (n = 2 for the standard Gini index): ⎡ 1 n ⎛ y ⎞1−ε ⎤ (1−ε) i⎟ Aε = 1− ⎢⎢ ∑ ⎜ ⎜ ⎟ ⎟ ⎥ ⎥ . n ⎜ ⎝ y ⎟ ⎠ ⎢⎣ i=1 ⎥⎦ 120 Republic of Congo – Poverty Assessment Report Annex 6: Poverty Dominance Analysis Figure A6.1 provides the cumulative density func- tions of consumption at the national level for the two Simple poverty profiles such as the statistics presented surveys (in real terms). The horizontal axis represents in Figure 1.5 provide a ranking for various household the welfare level of households (annual consumption per groups (or various time periods) in terms of their level equivalent adult), while the vertical axis represents the of poverty. However, it is important to test whether the cumulative percentage of the population with a welfare ranking is robust to the choice of the poverty line. This level below the value on the horizontal axis. Essentially, leads to a special type of robustness test, referred to as sto- the two curves for 2005 and 2011 do not intersect (there chastic dominance, which deals with the sensitivity of the is actually a point of intersection at about XAF 90,000, ranking of poverty measures between groups or between but this is a very low level of consumption and the periods of time to the use of different poverty lines (see intersection is largely because at that level, the surveys for example, Coudouel, Hentschel, and Wodon 2002). have few observations). Thus, except possibly for those The simplest way to do this (for the robustness who have extremely low levels of consumption, it can of poverty comparisons based on the headcount index be stated that poverty at the national level was lower in of poverty) is to plot the cumulative distribution of 2011 than in 2005. consumption per equivalent adult for two household Figure A6.2 provides the cumulative density func- groups or for the same group over two periods of time, tions of consumption by strata in 2011 and for rural as shown in different ways in Figures A6.1 and A6.2. areas for both survey years. The data by strata suggest One can then see whether the curves intersect. If they similar levels of poverty in Brazzaville and Pointe Noire do not intersect, then poverty is higher for the group or and higher levels of poverty in other areas. The data for period with the highest curve for all the poverty lines rural areas suggest an increase in poverty over time but that could be considered (as a proportion of the poverty recall that the increase is not statistically significant and line used). If the curves intersect, comparisons of poverty with data on perceptions of poverty and assets, a decrease levels then depend on the choice of the poverty line. is instead suggested. FIGURE A6.1: First Order Stochastic Dominance, National Sample, 2005–2011 A. Cumulative density functions B. Lower part of the distribution 1.0 0.35 0.9 0.30 Cumulative percentage of 0.8 Cumulative percentage of 0.25 population in poverty (P0) population in poverty (P0) 0.7 0.20 0.6 0.15 0.5 0.10 0.4 0.05 0.3 0.3 0.2 0.2 0.1 0.1 0 0 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000 900,000 1,000,000 0 20,000 40,000 60,000 80,000 100,000 120,000 140,000 160,000 180,000 200,000 Expenditure per equivalent adult Expenditure per equivalent adult 2011 2005 Poverty line Source: Authors’ estimation. Annexes 121 FIGURE A6.2: First Order Stochastic by Strata and in Rural Areas, 2005–2011 A. By strata, 2011 only B. In rural areas, 2005–2011 1.0 1.0 0.9 0.9 Cumulative percentage of 0.8 Cumulative percentage of 0.8 population in poverty (P0) population in poverty (P0) 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000 900,000 1,000,000 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000 900,000 1,000,000 Expenditure per equivalent adult Expenditure per equivalent adult Brazzaville Pointe Noire Other municipalities Rural 2011 Rural 2005 Poverty line Semi-urban Rural Poverty line Source: Authors’ estimation. 122 Republic of Congo – Poverty Assessment Report Annex 7:  Details on the Two definition also considers ‘discourage’, those stating that Definitions of Unemployment they are not looking for a job because there is no job available, they are not qualified, they do not know where The difference in the two definitions rely on the way to look for a job, or they are waiting for the outcome of unemployment is measured. For the first definition, their application. unemployment is restricted to the subset of those who The information needed to build this variable from are not employed, and are looking for a job. The second the two surveys includes the following questions: 2005 2011 Employment Over the last 7 days, did [NAME] work at least an hour? Over the last 7 days, did [NAME] work at least an hour? In the last week, did [NAME] work for a wage, salary, commission, or Although [NAME] did not work last week, does he has a job? any payment in kind? Why did [NAME] not work in the last 7 days? Was [NAME] absent from work in the last 7 days? Although [NAME] did not work last week, has he worked in the last 12 months? Unemployment Definition 1 Was [Name] looking for job in the last 4 weeks? Was [Name] looking for job in the last 7 days? Those saying YES Was [Name] looking for job in the last 30 days? Those saying YES Definition 2 Definition 1 PLUS Definition 1 PLUS What is the main reason [NAME] was not working in the last seven What is the main reason [NAME] was not working in the last days? seven days? Those saying NO JOB AVAILABLE Those saying NO JOB AVAILABLE/ NOT QUALIFY/DON’T KNOW HOW TO LOOK FOR/AWAITING APPLICATION OUTCOME Inactive Remaining population Remaining population Annexes 123 Statistical Appendix FIGURE SA.1: Republic of Congo vs Peers in Various Dimensions School enrollment, primary (% net) School enrollment, secondary (% net) 100 100 ROC 2012 School enrollment, primary, 80 School enrollment, secondary (% net) 80 total (% net) 60 ROC 2011 60 40 ROC 2005 20 40 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) School enrollment, primary, female (% net) School enrollment, primary, male (% net) 100 100 ROC 2012 School enrollment, primary, School enrollment, primary, ROC 2012 80 80 female (% net) male (% net) 60 ROC 2005 60 ROC 2005 40 40 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Primary completion rate, female (% of relevant age group) Primary completion rate, male (% of relevant age group) 120 120 Primary completion rate, female Primary completion rate, male 100 (% of relevant age group) (% of relevant age group) 100 80 ROC 2012 ROC 2005 80 ROC 2005 60 ROC 2012 ROC 1996 60 40 ROC 1996 20 40 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. (continued on next page) 124 Republic of Congo – Poverty Assessment Report FIGURE SA.1: Republic of Congo vs Peers in Various Dimensions (continued) Lower secondary completion rate, female (% of relevant age group) Lower secondary completion rate, male (% of relevant age group) 150 150 Lower secondary completion rate, Lower secondary completion rate, female (% of relevant age group) male (% of relevant age group) 100 100 50 ROC 2012 50 ROC 2012 ROC 2004 ROC 2004 ROC 1996 ROC 1996 0 0 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Immunization, DPT (% of children ages 12–23 months) Immunization, measles (% of children ages 12–23 months) 100 100 (% of children ages 12–23 months) (% of children ages 12–23 months) ROC 2014 Immunization, measles 80 ROC 2014 Immunization, DPT 80 60 ROC 2005 ROC 2005 60 40 ROC 1996 40 ROC 1996 20 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 GNI per capita, PPP (constant 2011 international $) GNI per capita, PPP (constant 2011 international $) Source: Authors’ calculation using the WDI data and based on the Gable, Lofgren, and Osorio-Rodarte (2015) approach. FIGURE SA.3: Dominance Curve of Various Types of FIGURE SA.2: Concentration Curve of Various Types Health Expenditures of Health Expenditures 0.7 100 0.6 90 0.5 Cumulative distribution 80 of expenditures (%) 70 0.4 60 % 50 0.3 40 0.2 30 0.1 20 10 0 0 0 100000 200000 300000 400000 500000 600000 0 10 20 30 40 50 60 70 80 90 100 Consumption per equivalent adult (FCFA) Cumulative distribution of the population (%) Modern drugs (include vaccine) Usual drugs Equity Modern drugs (include vaccine) Medical products (include malaria drugs) Usual drugs Medical products (include alcohool) (include alcohool) Therapeutic appliances (include malaria drugs) Consultation fees Consultation fees and equipment Therapeutic appliances Laboratory and radiology services Laboratory and radiology services Dental services and equipment Hospitalization (include delivery fees) Hospitalization Medical auxiliary services Dental services Medical auxiliary services (include delivery fees) Poverty line Source: Authors’ estimates based on the 2011 ECOM survey. Source: Authors’ estimates based on the 2011 ECOM survey. Annexes 125  orrelates of the Likelihood of TABLE SA.1: C Correlates of the Likelihood of TABLE SA.1:  Migrating Migrating (continued) A member of the household A member of the household left definitively left definitively coefficient t coefficient t Household experienced a shock 0.268*** 0.036 Head industry Region Agriculture, livestock, fisheries, Ref. Ref. Brazzaville Ref. Ref. forestry Pointe Noire −0.083 0.101 Mines/quarries −0.084 0.144 Other municipalities 0.252 0.186 Production/processing −0.057 0.082 Semi-urban 0.140 0.215 Construction −0.273*** 0.098 Rural 0.422** 0.211 Transport −0.078 0.108 Household composition Trade/sales −0.108* 0.060 Child ages 0 to 5 −0.046 0.038 Services −0.067 0.077 Child ages 0 to 5, squared 0.004 0.012 Education/health −0.198** 0.085 Boys ages 6 to 18 0.090** 0.038 Administration −0.085 0.075 Boys ages 6 to 18, squared −0.020* 0.012 Other services −0.209** 0.098 Girls ages 6 to 17 −0.019 0.040 Unemployed/inactive −0.003 0.050 Girls ages 6 to 17, squared 0.015 0.013 Head marital status Adult ages 18 to 60 0.049 0.039 Never married Ref. Ref. Adult ages 18 to 60, squared 0.002 0.006 Married monogamous 0.018 0.066 Elderly ages 60 plus 0.189*** 0.072 Married polygamist 0.069 0.092 Elderly ages 60 plus, squared −0.024 0.028 Free union −0.064 0.066 Head age 0.028*** 0.007 Divorced/separated −0.073 0.071 Squared head age −0.000*** 0.000 Widower/widow −0.004 0.079 Head is pygmy −0.260** 0.122 Household own land −0.089** 0.036 Head is disable 0.006 0.085 Population density (province) 0.104*** 0.032 Head is female 0.062 0.053 Distance to Brazzaville/Pointe Noire 0.055** 0.028 Head education Province adjacent to Brazzaville/ 0.134 0.082 Pointe Noire None Ref. Ref. Poverty headcount (province) 0.009** 0.004 Primary 0.060 0.048 Constant −3.236*** 0.404 Secondary 1 0.086* 0.044 Number of observations 10,307 Secondary 2 0.154*** 0.055 Source: Authors’ calculation using the 2011 ECOM survey. Tertiary 0.266*** 0.071 Note: *** p<0.01, ** p<0.05, * p<0.1. 126 Republic of Congo – Poverty Assessment Report TABLE SA.2: Population Shares by Poverty, Vulnerability, and Middle Class Groups (%)   Brazzaville Pointe Noire Other municipalities Semi-urban Rural National 2011 Poor 21.6 20.3 52.8 59.7 69.4 40.9 Insecure nonpoor 39.3 41.9 33.3 23.2 21.7 32.8 Middle class 39.1 37.8 13.9 17.0 8.8 26.3 Total 100.0 100.0 100.0 100.0 100.0 100.0 2005 Poor 42.3 33.5 58.4 67.4 64.8 50.7 Insecure nonpoor 31.2 34.7 29.3 24.9 23.2 28.7 Middle class 26.5 31.9 12.3 7.7 12.1 20.6 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source: Authors’ calculation using the 2005 and 2011 ECOM survey. TABLE SA.3: Household Composition and Dependency Ratios (%) Poverty status Location Welfare quintiles Total Pointe Other Semi- Nonpoor Poor Brazzaville Noire municipalities urban Rural Q1 Q2 Q3 Q4 Q5 2005 Children ages 0 to 14 1.60 2.54 1.78 2.01 2.20 2.21 2.12 2.89 2.34 2.14 1.75 1.35 2.01 Adult ages 15 to 64 2.70 3.25 3.13 3.39 3.02 2.71 2.53 3.37 3.10 3.16 2.81 2.48 2.93 Elderly ages 65 plus 0.15 0.22 0.15 0.13 0.17 0.20 0.24 0.22 0.24 0.20 0.16 0.12 0.18 Household size 4.45 6.01 5.06 5.52 5.39 5.13 4.89 6.48 5.69 5.51 4.72 3.96 5.12 Dependency 65.0 85.1 61.6 63.1 78.6 89.0 93.2 92.1 83.3 74.1 67.8 59.3 74.6 Child dependency 59.3 78.3 56.7 59.3 73.0 81.6 83.8 85.6 75.5 67.7 62.1 54.5 68.4 Aged dependency 5.7 6.8 4.8 3.8 5.6 7.5 9.4 6.6 7.8 6.4 5.7 4.8 6.2 2011 Children ages 0 to 14 1.36 2.35 1.55 1.55 2.02 1.98 1.85 2.72 2.06 1.74 1.45 1.03 1.70 Adult ages 15 to 64 2.33 2.61 2.64 2.64 2.56 2.25 2.08 2.67 2.57 2.54 2.37 2.15 2.43 Elderly ages 65 plus 0.13 0.20 0.12 0.11 0.16 0.16 0.21 0.21 0.19 0.19 0.13 0.10 0.16 Household size 3.83 5.16 4.31 4.31 4.74 4.40 4.15 5.59 4.82 4.46 3.95 3.28 4.28 Dependency 64.1 97.7 63.4 62.9 85.0 95.1 98.9 109.5 87.6 76.0 66.6 52.3 76.3 Child dependency 58.3 90.1 58.8 58.6 78.6 88.0 88.7 101.8 80.2 68.4 61.0 47.8 69.9 Aged dependency 5.8 7.6 4.6 4.3 6.4 7.2 10.2 7.8 7.4 7.6 5.6 4.5 6.4 Source: Authors’ calculation using the 2005 and 2011 ECOM surveys. Annexes 127 TABLE SA.4: Earnings Regressions, Specification with Fewer Controls, 2011   Earnings, OLS Work and Earnings, Heckman Hourly Monthly Hourly Monthly Earnings Work Earnings Work Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Region Brazzaville Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Pointe Noire 0.113* 0.061 0.226*** 0.083 0.179** 0.075 −0.109** 0.052 0.345*** 0.105 −0.101** 0.044 Other municipalities −0.034 0.070 −0.082 0.101 −0.217*** 0.082 0.220*** 0.053 −0.451*** 0.111 0.163*** 0.044 Semi-urban 0.055 0.094 −0.060 0.132 −0.371*** 0.100 0.605*** 0.054 −0.895*** 0.135 0.485*** 0.051 Rural −0.098 0.064 −0.330*** 0.096 −0.400*** 0.076 0.494*** 0.050 −0.979*** 0.110 0.466*** 0.042 Pygmy −0.583*** 0.173 −0.766** 0.317 −0.823*** 0.195 0.409** 0.171 −1.215*** 0.329 0.368*** 0.142 Disability −0.226 0.146 −0.489** 0.240 0.240 0.155 −0.464*** 0.084 0.449* 0.242 −0.324*** 0.082 Demographics Female −0.100 0.075 −0.162 0.118 −0.023 0.079 −0.065* 0.038 0.003 0.111 −0.058 0.036 Married 0.237*** 0.058 0.288*** 0.089 −0.492*** 0.073 0.906*** 0.046 −1.051*** 0.114 0.726*** 0.041 Married and female −0.395*** 0.078 −0.559*** 0.123 −0.204** 0.083 −0.366*** 0.046 −0.262** 0.121 −0.199*** 0.042 Divorced/separated 0.129 0.101 0.163 0.159 −0.560*** 0.113 0.847*** 0.099 −1.084*** 0.183 0.790*** 0.118 Divorced and female −0.201* 0.122 −0.203 0.193 −0.195 0.132 −0.125 0.107 −0.245 0.206 −0.180 0.123 Widow/widower 0.385* 0.204 0.629** 0.284 −0.156 0.225 0.507*** 0.141 −0.398 0.317 0.431*** 0.126 Widow and female −0.356* 0.211 −0.530* 0.294 −0.506** 0.232 0.191 0.149 −0.821** 0.327 0.152 0.131 Age 0.055*** 0.011 0.087*** 0.018 −0.133*** 0.013 0.215*** 0.006 −0.276*** 0.020 0.182*** 0.006 Age squared −0.001*** 0.000 −0.001*** 0.000 0.002*** 0.000 −0.002*** 0.000 0.003*** 0.000 −0.002*** 0.000 Education None Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Primary −0.110 0.072 −0.066 0.115 −0.195** 0.076 0.094* 0.052 −0.199* 0.118 0.057 0.048 Lower secondary 0.042 0.059 0.121 0.091 0.116* 0.065 −0.131*** 0.047 0.301*** 0.099 −0.157*** 0.041 Upper secondary 0.196*** 0.064 0.296*** 0.097 0.313*** 0.072 −0.228*** 0.051 0.580*** 0.108 −0.247*** 0.045 Tertiary 0.504*** 0.073 0.600*** 0.105 0.422*** 0.081 0.067 0.059 0.555*** 0.117 −0.018 0.052 Sector Agriculture Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Mines/quarries 0.382 0.403 0.257 0.712 0.454 0.322 0.682* 0.381 Production/ 0.879*** 0.096 1.278*** 0.135 0.764*** 0.098 0.857*** 0.107 processing Construction 0.831*** 0.086 1.125*** 0.132 0.747*** 0.082 0.826*** 0.093 Transport 0.892*** 0.085 1.462*** 0.112 0.728*** 0.082 0.932*** 0.093 Trade/sales 0.707*** 0.062 1.108*** 0.091 0.598*** 0.058 0.648*** 0.061 Services 0.781*** 0.074 1.158*** 0.113 0.675*** 0.071 0.799*** 0.085 Education/health 1.336*** 0.070 1.626*** 0.095 1.191*** 0.070 1.104*** 0.084 (continued on next page) 128 Republic of Congo – Poverty Assessment Report TABLE SA.4: Earnings Regressions, Specification with Fewer Controls, 2011 (continued)   Earnings, OLS Work and Earnings, Heckman Hourly Monthly Hourly Monthly Earnings Work Earnings Work Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Administration 1.363*** 0.069 1.706*** 0.099 1.223*** 0.066 1.192*** 0.080 Other services 0.714*** 0.096 0.978*** 0.144 0.674*** 0.093 0.756*** 0.118 Missing −0.154 0.322 −0.032 0.581 −0.147 0.279 0.229 0.409 Household size Children < 6 −0.026 0.019 −0.018 0.014 Children < 6 and 0.034* 0.019 0.018 0.015 female Share of dependents −0.108* 0.060 −0.056 0.043 Constant 3.706*** 0.228 7.729*** 0.368 8.844*** 0.281 −4.297*** 0.115 17.573*** 0.409 −3.763*** 0.109 /athrho −1.209*** 0.042 −1.966*** 0.042 /lnsigma 0.614*** 0.024 1.084*** 0.030 Number of 13,109 13,325 21,321 21,537 observations R2 0.172 0.144 Source: Authors’ estimation. Note: *** p<0.01, ** p<0.05, * p<0.1. TABLE SA.5: Earnings Regressions, Specification with More Controls, 2011 Wage regressions, OLS, unweighted Wage regressions, Heckman Hourly wage Monthly wage Hourly wage Monthly wage Model Selection Model Selection Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Region Brazzaville Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Pointe Noire 0.139** 0.058 0.236*** 0.081 0.205*** 0.071 −0.112** 0.052 0.353*** 0.101 −0.107** 0.044 Other municipalities −0.031 0.068 −0.097 0.100 −0.210*** 0.080 0.220*** 0.053 −0.454*** 0.109 0.161*** 0.044 Semi-urban 0.074 0.093 −0.054 0.130 −0.347*** 0.099 0.608*** 0.054 −0.874*** 0.133 0.485*** 0.051 Rural −0.080 0.062 −0.321*** 0.095 −0.380*** 0.074 0.491*** 0.050 −0.959*** 0.108 0.462*** 0.042 Pygmy −0.578*** 0.178 −0.788** 0.323 −0.819*** 0.198 0.400** 0.171 −1.229*** 0.334 0.361** 0.143 Disability −0.208 0.146 −0.458* 0.241 0.251 0.154 −0.464*** 0.084 0.461* 0.242 −0.325*** 0.083 Demographics Female −0.106 0.073 −0.160 0.116 −0.029 0.077 −0.066* 0.038 0.012 0.110 −0.059 0.036 Married 0.175*** 0.058 0.217** 0.089 −0.544*** 0.073 0.905*** 0.046 −1.093*** 0.114 0.727*** 0.041 Married and female −0.346*** 0.077 −0.493*** 0.122 −0.161* 0.082 −0.365*** 0.046 −0.220* 0.120 −0.201*** 0.042 (continued on next page) Annexes 129 TABLE SA.5: Earnings Regressions, Specification with More Controls, 2011 (continued) Wage regressions, OLS, unweighted Wage regressions, Heckman Hourly wage Monthly wage Hourly wage Monthly wage Model Selection Model Selection Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Divorced/separated 0.110 0.101 0.135 0.159 −0.567*** 0.112 0.853*** 0.100 −1.082*** 0.183 0.802*** 0.124 Divorced and female −0.185 0.121 −0.175 0.193 −0.185 0.131 −0.130 0.107 −0.240 0.207 −0.193 0.128 Widow/widower 0.379* 0.206 0.630** 0.286 −0.154 0.226 0.515*** 0.142 −0.378 0.318 0.440*** 0.127 Widow and female −0.350* 0.212 −0.528* 0.295 −0.501** 0.233 0.187 0.150 −0.828** 0.328 0.142 0.133 Age 0.052*** 0.011 0.082*** 0.018 −0.134*** 0.013 0.215*** 0.006 −0.276*** 0.020 0.183*** 0.006 Age squared −0.001*** 0.000 −0.001*** 0.000 0.002*** 0.000 −0.003*** 0.000 0.003*** 0.000 −0.002*** 0.000 Education None Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Primary −0.097 0.071 −0.051 0.114 −0.184** 0.075 0.091* 0.052 −0.191 0.117 0.052 0.048 Lower secondary 0.042 0.059 0.124 0.092 0.113* 0.064 −0.134*** 0.047 0.296*** 0.099 −0.159*** 0.041 Upper secondary 0.122* 0.063 0.206** 0.097 0.239*** 0.070 −0.230*** 0.051 0.492*** 0.108 −0.253*** 0.045 Tertiary 0.251*** 0.072 0.334*** 0.105 0.168** 0.080 0.057 0.059 0.295** 0.117 −0.036 0.052 Main employer Public administration Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Public firm/parastatal 0.091 0.074 0.142* 0.082 0.041 0.079 0.063 0.108 Large private −0.025 0.072 0.053 0.083 −0.022 0.080 0.053 0.089 company SME −0.500*** 0.072 −0.518*** 0.099 −0.511*** 0.074 −0.446*** 0.095 Associations and −0.303*** 0.109 −0.502*** 0.163 −0.289*** 0.109 −0.397** 0.161 similar Int’l Org., Embassy −0.070 0.437 −0.126 0.680 −0.005 0.320 0.019 0.425 Household −0.391*** 0.113 −0.540*** 0.158 −0.382*** 0.113 −0.500*** 0.128 Own account −0.516*** 0.094 −0.623*** 0.128 −0.511*** 0.094 −0.523*** 0.108 Missing −0.759 0.490 −0.555 0.770 −0.777* 0.446 −0.490 0.446 Sector Agriculture and Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. similar Mines/quarries 0.148 0.394 −0.119 0.695 0.236 0.318 0.386 0.386 Production/ 0.576*** 0.099 0.809*** 0.140 0.492*** 0.106 0.492*** 0.114 processing Construction 0.594*** 0.095 0.716*** 0.139 0.550*** 0.091 0.542*** 0.100 Transport 0.552*** 0.097 0.935*** 0.132 0.430*** 0.094 0.561*** 0.106 Trade/sales 0.704*** 0.062 1.090*** 0.091 0.598*** 0.058 0.648*** 0.062 Services 0.527*** 0.082 0.770*** 0.125 0.450*** 0.077 0.518*** 0.087 Education/health 0.583*** 0.086 0.707*** 0.121 0.485*** 0.085 0.367*** 0.112 (continued on next page) 130 Republic of Congo – Poverty Assessment Report TABLE SA.5: Earnings Regressions, Specification with More Controls, 2011 (continued) Wage regressions, OLS, unweighted Wage regressions, Heckman Hourly wage Monthly wage Hourly wage Monthly wage Model Selection Model Selection Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Coef. t-st. Administration 0.488*** 0.095 0.648*** 0.135 0.395*** 0.094 0.332*** 0.117 Other services 0.532*** 0.096 0.699*** 0.145 0.515*** 0.092 0.553*** 0.117 Missing −0.309 0.310 −0.298 0.561 −0.301 0.279 −0.013 0.437 Type of occupation Sr. Mgt./Engineer Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Mid Mgt./Master −0.190*** 0.064 −0.170** 0.076 −0.194*** 0.067 −0.161* 0.087 Empl., qualified −0.454*** 0.080 −0.416*** 0.102 −0.458*** 0.080 −0.408*** 0.095 Empl., unskilled −0.692*** 0.094 −0.658*** 0.119 −0.689*** 0.093 −0.635*** 0.112 Blue collar −0.769*** 0.097 −0.469*** 0.118 −0.782*** 0.106 −0.536*** 0.130 Employer-Patron −0.324** 0.131 −0.258 0.166 −0.285** 0.143 −0.205 0.167 Own account −0.766*** 0.114 −0.832*** 0.157 −0.732*** 0.112 −0.734*** 0.128 Apprentice −1.218*** 0.198 −1.395*** 0.318 −1.117*** 0.190 −1.050*** 0.204 Family helper −1.035*** 0.208 −1.500*** 0.352 −0.842*** 0.200 −0.774*** 0.256 Missing −0.890*** 0.219 −0.603** 0.280 −0.908*** 0.216 −0.682*** 0.196 Household size Children < 6 −0.026 0.019 −0.020 0.013 Children < 6 and 0.034* 0.019 0.020 0.015 female Share of dependents −0.110* 0.060 −0.057 0.042 Constant 5.091*** 0.248 9.326*** 0.388 10.132*** 0.302 −4.297*** 0.115 18.831*** 0.430 −3.764*** 0.109 /athrho −1.210*** 0.042 −1.965*** 0.042 /lnsigma 0.606*** 0.024 1.079*** 0.030 Number of 13,109 13,325 21,321 21,537 observations R2 0.185 0.153 Source: Authors’ estimation. Note: *** p<0.01, ** p<0.05, * p<0.1. Annexes 131 TABLE SA.6: Income Sources Regressions-Heckman, 2011 Nonfarm Total HH Transfer Wage business Agriculture Pension Other income coef coef coef coef coef coef coef Region       Brazzaville Ref. Ref. Ref. Ref. Ref. Ref. Ref. Pointe Noire 0.007 0.206*** 0.060 −0.076 0.105 0.231 0.199*** Other municipalities −0.271*** 0.040 −0.283*** −0.897*** −0.241 −0.394** −0.128** Semi-urban −0.432*** −0.003 −0.542*** −1.071*** −0.209 −0.819*** −0.601*** Rural −0.592*** −0.150*** −0.555*** −1.153*** −0.448** −0.719*** −0.540*** Household composition       Child ages 0 to 5 −0.012 0.025 −0.076 −0.013 0.138 −0.092 0.034 Child ages 0 to 5, squared 0.006 −0.019 0.027* −0.003 −0.062** 0.044 −0.009 Girls ages 6 to 17 0.089* 0.024 −0.042 −0.030 −0.140 0.049 −0.021 Girls ages 6 to 17, squared −0.025 −0.003 0.021 0.003 0.012 0.013 0.014 Boys ages 6 to 18 0.040 0.068 −0.017 −0.071 0.033 −0.235** 0.019 Boys ages 6 to 18, squared −0.008 −0.014 0.010 0.018 −0.006 0.100*** 0.007 Adult ages 18 to 60 0.076 0.144** 0.167*** 0.279*** 0.110 0.226** 0.255*** Adult ages 18 to 60, squared −0.001 −0.007 −0.015* −0.023*** −0.010 −0.030 −0.022*** Elderly ages 60 plus 0.322*** −0.484*** −0.092 0.028 0.080 −0.053 0.065 Elderly ages 60 plus, squared −0.035 0.207** 0.018 0.060** −0.049 0.073 0.041 Head age −0.008 0.011 0.029*** 0.035*** 0.076* 0.021 0.030*** Square head age 0.000 −0.000 −0.000*** −0.000*** −0.001* −0.000 −0.000*** Head is pygmy −1.156*** −0.131 −0.338* −0.414*** 0.411 −0.268 −0.529*** Head is disabled −0.036 −0.079 −0.001 −0.160 −0.135 −0.158 −0.176* Head is female 0.494*** −0.203*** −0.281*** −0.233*** −0.865*** −0.085 −0.203*** Head education       None Ref. Ref. Ref. Ref. Ref. Ref. Ref. Primary 0.014 −0.086 −0.071 −0.063 0.194 −0.239* −0.091* Secondary 1 0.156** 0.021 −0.025 0.041 0.339* 0.191 0.015 Secondary 2 0.390*** 0.160* −0.023 −0.021 0.578*** 0.200 0.120** Tertiary 0.701*** 0.419*** −0.063 0.029 0.606*** 0.490*** 0.396*** Head industry       Agriculture, livestock, fisheries, Ref. Ref. Ref. Ref. Ref. Ref. Ref. forestry Mines/quarries −0.014 0.944*** 0.606*** −0.151 0.790* 0.097 0.796*** Production/processing 0.522*** 0.682*** 0.481*** −0.177 −1.272* 0.383 0.778*** Construction −0.011 0.629*** 0.639*** −0.502*** −0.206 0.470*** 0.718*** Transport 0.141 0.646*** 0.316** −0.392* −0.482 0.544** 0.794*** (continued on next page) 132 Republic of Congo – Poverty Assessment Report TABLE SA.6: Income Sources Regressions-Heckman, 2011 (continued) Nonfarm Total HH Transfer Wage business Agriculture Pension Other income coef coef coef coef coef coef coef Trade/sales 0.248*** 0.452*** 0.799*** 0.046 0.145 0.397* 0.576*** Services 0.070 0.552*** 0.414*** −0.236* −0.016 0.377** 0.701*** Education/health 0.207 0.545*** 0.311** −0.081 0.468 0.156 0.956*** Administration 0.397*** 0.737*** −0.024 0.035 0.284 0.525*** 1.083*** Other services 0.127 0.521*** 0.260* −0.168 0.617 0.462** 0.717*** Unemployed/inactive 0.446*** 0.335*** 0.295*** 0.052 0.132 0.357*** 0.282*** No spouse −0.060 0.034 −0.001 −0.035 −0.001 −0.300* −0.097 Spouse education       None Ref. Ref. Ref. Ref. Ref. Ref. Ref. Primary −0.224*** −0.075 0.190** −0.045 −0.284* −0.422** 0.123** Secondary 1 −0.172** −0.003 0.109 −0.034 0.146 −0.283* 0.090* Secondary 2 0.011 0.173** 0.286** 0.014 −0.035 −0.254 0.265*** Tertiary −0.010 0.372*** 0.044 0.329 0.589* −0.028 0.455*** Household own land −0.037 −0.069 −0.039 0.060 −0.173 −0.165* −0.036 Access to market       Covered market 0.163*** 0.098** 0.149*** −0.177*** 0.069 0.164* −0.019 Open-air market 0.173*** −0.063 0.059 0.126*** 0.089 0.079 −0.004 Constant 9.062*** 12.590*** 11.653*** 11.998*** 10.708*** 11.700*** 12.032*** athrho 1.000*** −0.096*** 0.087** 0.019 −0.047 0.035 −0.990*** lnsigma 0.434*** −0.171*** 0.180*** 0.131*** 0.004 0.279*** 0.327*** Number of observations 10,393 10,393 10,393 10,393 10,393 10,393 10,393 Source: Authors’ estimation. Note: *** p<0.01, ** p<0.05, * p<0.1; the model for Property income could not be run because of low sample size. Annexes 133 TABLE SA.7: Demand for Education Regressions, 2011 Complete Drop out Complete Drop out Complete Drop out junior of junior Start senior of senior Start primary of primary Start junior secondary secondary senior secondary secondary primary cond. on cond. on secondary cond. on cond. on secondary cond. on cond. on   (SP) SP SP (SJS) SJS SJS (SSS) SSS SJS Time to the primary school Less than 15 minutes Ref. Ref. Ref. 15 to 30 minutes 0.0002 0.0160 −0.0160 30 to 45 minutes −0.0261*** 0.0164 −0.0164 45 minutes and above −0.0686*** 0.0437 −0.0437 Time to the secondary school Less than 15 minutes Ref. Ref. Ref. Ref. Ref. Ref. 15 to 30 minutes 0.0130 0.0604 −0.0604 0.0323 −0.0697 0.0697 30 to 45 minutes 0.0048 0.1091** −0.1091** −0.0201 −0.0102 0.0102 45 minutes and above −0.0611** 0.1133** −0.1133** −0.0236 0.0510 −0.0510 Household composition Child ages 0 to 5 −0.0017 −0.0462*** 0.0462*** −0.0292*** −0.0237 0.0237 −0.0205 −0.0355 0.0355 Girls ages 6 to 17 0.0024 −0.0268*** 0.0268*** −0.0164* 0.0014 −0.0014 −0.0002 −0.0494 0.0494 Boys ages 6 to 18 −0.0013 0.0169* −0.0169* −0.0062 0.0072 −0.0072 0.0172 −0.0151 0.0151 Adult ages 18 to 60 −0.0011 0.0020 −0.0020 0.0034 0.0006 −0.0006 −0.0117 0.0430** −0.0430** Elderly ages 60 plus 0.0016 0.0050 −0.0050 −0.0042 0.0998*** −0.0998*** 0.0139 0.0899 −0.0899 Welfare quintile Q1 Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Q2 0.0102* 0.0403* −0.0403* 0.0669*** 0.1882*** −0.1882*** −0.0517 0.0693 −0.0693 Q3 0.0250*** 0.0833*** −0.0833*** 0.0911*** 0.2206*** −0.2206*** 0.0188 0.0774 −0.0774 Q4 0.0234*** 0.1241*** −0.1241*** 0.1019*** 0.2498*** −0.2498*** 0.0163 0.1368 −0.1368 Q5 0.0412*** 0.1564*** −0.1564*** 0.1519*** 0.2825*** −0.2825*** 0.1064** 0.1047 −0.1047 Region Brazzaville Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Pointe Noire −0.0086 −0.0623 0.0623 −0.0626 −0.0675 0.0675 −0.1480*** −0.0715 0.0715 Other municipalities −0.0062 −0.1167*** 0.1167*** −0.1298*** −0.1050** 0.1050** −0.1938*** −0.1188 0.1188 Semi-urban −0.0223 −0.1589*** 0.1589*** −0.1604*** −0.1271** 0.1271** −0.1912*** −0.2053** 0.2053** Rural −0.0222* −0.2120*** 0.2120*** −0.1556*** −0.2968*** 0.2968*** −0.2683*** −0.1623* 0.1623* Head age 0.0000 0.0007 −0.0007 −0.0004 −0.0040** 0.0040** 0.0016 −0.0026 0.0026 Head female 0.0011 0.0828*** −0.0828*** −0.0027 0.0097 −0.0097 0.0993*** 0.1103** −0.1103** (continued on next page) 134 Republic of Congo – Poverty Assessment Report TABLE SA.7: Demand for Education Regressions, 2011 (continued) Complete Drop out Complete Drop out Complete Drop out junior of junior Start senior of senior Start primary of primary Start junior secondary secondary senior secondary secondary primary cond. on cond. on secondary cond. on cond. on secondary cond. on cond. on   (SP) SP SP (SJS) SJS SJS (SSS) SSS SJS Head education None Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Primary 0.0121* −0.0283 0.0283 −0.0738** −0.0876 0.0876 0.0203 0.0525 −0.0525 Secondary 1 0.0290*** 0.0591** −0.0591** −0.0123 0.0313 −0.0313 −0.0154 −0.0712 0.0712 Secondary 2 0.0412*** 0.2036*** −0.2036*** 0.0677** 0.0441 −0.0441 0.1841*** −0.0953 0.0953 Tertiary 0.0378*** 0.2851*** −0.2851*** 0.0830** 0.1953*** −0.1953*** 0.1796*** 0.0851 −0.0851 HH has a nonfarm business −0.0013 0.0129 −0.0129 0.0301 −0.0251 0.0251 0.0012 −0.0567 0.0567 HH has livestock 0.0017 −0.0143 0.0143 −0.0101 −0.0873** 0.0873** −0.0221 0.0503 −0.0503 HH in crop farming −0.0002 −0.0935*** 0.0935*** −0.0618*** −0.0273 0.0273 0.0223 −0.2464*** 0.2464*** Child is Pygmy −0.3210*** −0.4246*** 0.4246*** −0.0375 — — — — — Child is disabled −0.3226*** 0.0016 −0.0016 −0.0042 −0.2127* 0.2127* −0.3659** −0.1486 0.1486 Child age 0.0216*** 0.1823*** −0.1823*** 0.1152*** 0.1484*** −0.1484*** 0.0008 0.0062 −0.0062 Child is a girl 0.0052 −0.0457** 0.0457** 0.0187 0.0204 −0.0204 −0.0225 −0.0217 0.0217 Child is head of household 0.0057 0.0355 −0.0355 0.0656*** 0.1030*** −0.1030*** 0.0549 −0.0053 0.0053 Child is orphan −0.0113 −0.0246 0.0246 0.0356 0.0387 −0.0387 — — — Observations 7,728 3,994 3,994 2,751 1,154 1,154 1,179 477 477 Source: Authors’ estimation. Note: *** p<0.01, ** p<0.05, * p<0.1. HH = Household. TABLE SA.8: Basics Statistics on Access to Electricity, 2011 Decile expenditure Cumulative share of Share paying for Access to electricity per equivalent adult Kwh (Qn > 0) Access to electricity electricity at the PSU level Take up rate 1 0.2% 6.1 1.1 27.4 22.1 2 0.9% 10.9 4.2 37.5 29.1 3 2.9% 16.4 10.4 50.5 32.5 4 6.8% 26.0 18.5 58.3 44.7 5 13.3% 32.0 21.2 64.3 49.8 6 22.1% 41.3 30.7 73.0 56.6 7 32.6% 44.5 34.2 74.3 60.0 8 47.1% 60.0 44.3 83.0 72.4 9 66.2% 68.9 50.2 87.3 78.9 10 100.0% 75.3 52.2 90.3 83.4 Total ROC   42.5 29.9 68.1 62.3 Source: Authors’ estimation. Annexes 135 TABLE SA.9: Basics Statistics on Access to Residential Water Network, 2011 Decile expenditure Cumulative share of Share paying for Access to water at per equivalent adult m3 (Qn > 0) Access to water water the PSU level Take-up rate 1 0.6% 1.9 0.1 13.6 14.0 2 2.2% 5.0 1.8 18.9 26.4 3 4.8% 9.3 6.4 32.6 28.6 4 10.9% 18.5 13.4 47.5 38.9 5 19.1% 25.0 17.0 53.7 46.5 6 28.5% 24.3 18.3 56.3 43.2 7 41.0% 30.5 18.6 61.2 49.9 8 56.6% 38.1 27.0 70.5 54.0 9 73.7% 40.6 29.3 75.5 53.8 10 100.0% 46.8 31.7 79.7 58.7 Total ROC   26.7 18.3 54.8 48.7 Source: Authors’ estimation. TABLE SA.10: Welfare Regressions, 2011 OLS: log welfare Coef. t Migration status A member of the household left definitively 0.092*** 0.018 Region Brazzaville Ref. Ref. Pointe Noire 0.159*** 0.027 Other municipalities −0.337*** 0.026 Semi-urban −0.320*** 0.032 Rural −0.399*** 0.027 Household composition Child ages 0 to 5 −0.199*** 0.016 Child ages 0 to 5, squared 0.026*** 0.005 Boys ages 6 to 18 −0.261*** 0.016 Boys ages 6 to 18, squared 0.034*** 0.005 Girls ages 6 to 17 −0.199*** 0.017 Girls ages 6 to 17, squared 0.026*** 0.006 Adult ages 18 to 60 −0.176*** 0.018 Adult ages 18 to 60, squared 0.013*** 0.003 Elderly ages 60 plus −0.143*** 0.031 Elderly ages 60 plus, squared 0.001 0.011 Head age 0.013*** 0.003 (continued on next page) 136 Republic of Congo – Poverty Assessment Report TABLE SA.10: Welfare Regressions, 2011 (continued) OLS: log welfare Coef. t Squared head age −0.000*** 0.000 Head is pygmy −0.451*** 0.047 Head is disabled 0.011 0.042 Head is female 0.007 0.024 Head education None Ref. Ref. Primary 0.060*** 0.021 Secondary 1 0.140*** 0.021 Secondary 2 0.249*** 0.027 Tertiary 0.475*** 0.034 Head industry Agriculture, livestock, fisheries, forestry Ref. Ref. Mines/quarries 0.219*** 0.062 Production/processing 0.241*** 0.037 Construction 0.063* 0.037 Transport 0.125*** 0.041 Trade/sales 0.166*** 0.027 Services 0.094*** 0.036 Education/health 0.145*** 0.039 Administration 0.310*** 0.035 Other services 0.014 0.037 Unemployed/inactive −0.029 0.023 No spouse −0.020 0.028 Spouse education None Ref. Ref. Primary −0.020 0.024 Secondary 1 0.049** 0.023 Secondary 2 0.191*** 0.038 Tertiary 0.307*** 0.058 Household own land −0.012 0.017 Access to market Covered market 0.066*** 0.015 Open-air market 0.079*** 0.016 Constant 12.961*** 0.078 Number of observations 10,328 R2 0.357 Source: Authors’ estimation. Note: *** p<0.01, ** p<0.05, * p<0.1. References 137 References Agénor, Pierre-Richard, and Otaviano Canuto. 2013. “Gender Equality and Economic Growth in Brazil.” Economic Premise 109, Washington, DC: The World Bank. Angel-Urdinola, D., and Q. Wodon. 2007. “Do Utility Subsidies Reach the Poor? Framework and Evidence for Cape Verde, Sao Tome, and Rwanda.” Economics Bulletin 9 (4): 1–7. Baird, Sarah J., Richard S. Garfein, Craig T. McIntosh, and Berk Özler. 2012. “Effect of a Cash Transfer Programme for Schooling on Prevalence of HIV and Herpes Simplex Type 2 in Malawi: A Cluster Randomised Trial.” Lancet 379: 1320–1329. Baird, Sarah J., Craig Mcintosh, and Berk Ozler. 2011. “Cash or Condition? Evidence from a Cash Transfer Experiment.” Quarterly Journal of Economics 126: 1709–1753. Balcazar, Carlos Felipe, Ceriani Lidia, Olivieri Sergio, and Ranzani Marco. 2014. “Rent Imputation for Welfare Measurement: A Review of Methodologies and Empirical Findings.” Policy Research Working Paper 7103. Banerjee, Abhijit V., and Esther Duflo. 2005. “Growth Theory through the Lens of Development Economics.” In Philippe Aghion and Steven Durlauf, (eds.), Handbook of Economic Growth. Amsterdam: Elsevier. Becker, G., K. Murphy and R. Tamura. 1990. “Human capital, fertility, and economic growth”, Journal of Political Economy, vol. 98, No. 5. Bhattacharya, and Ghura. 2006. “Oil and Growth in the Republic of Congo.” IMF Working Paper WP/06/185. Bloom David E, Canning David, Fink Günther, Finlay Jocelyn E. 2009. “Fertility, Female Labor Force Participation, and the Demographic Dividend”, Journal of Economic Growth, 14(2), 79–101. Bloom, D., D. Canning and J. Sevilla. 2003. “The demographic dividend: a new perspective on the economic consequences of population change”, RAND Population Matters Program, No. MR-1274, Santa Monica, California. de la Brière, Bénédicte, Filmer Deon, Ringold Dena, Rohner Dominic, Samuda Karelle, and Denisova Anastasiya. 2016. “From Mines and Wells to Well Built Minds: Turning Sub-Saharan Africa’s Natural Resource Wealth into Human Capital.” Washington, DC: The World Bank. Campos, Francisco, Markus P. Goldstein, Laura Mcgorman, Ana Maria Munoz Boudet, and Obert Pimhidzai. 2015. “Breaking the Metal Ceiling: Female Entrepreneurs Who Succeed in Male- Dominated Sectors.” Policy Research Working Paper 7503, Washington, DC: The World Bank. Canning, David, Sangeeta Raja, and Abdo S. Yazbeck, eds. 2015. Africa’s Demographic Transition Dividend or Disaster? Africa Development Forum Report. 139 Centre National de la Statistique et des Études Galor, O. 2005. The demographic transition and the Économiques (CNSEE) and ICF International. emergence of sustained economic growth. Journal of 2012. “Enquête Démographique et de Santé du Congo European Economic Association 3(2-3): 494–504. (EDSC-II) 2011-2012”. Garcia, Marito, and Charity M. T. Moore. 2012. “The Colclough, Christopher, Gaeta Gandhi Kingdon, and Cash Dividend: The Rise of Cash Transfer Programs Harry A. Patrinos. 2010. “The Changing Patterns in Sub-Saharan Africa.” Directions in Human of Wage Returns to Education and its Implications.” Development No. 67208. Development Policy Review 28(6):733–47. Harmon, Colm, Hessel Oosterbeek, and Ian Walker. CONFEMEN. 2012. L’enseignement primaire au Congo: 2003. “The Returns to Education: Microeconomics.” à la recherche de la qualité et de l’équité. Rapport Journal of Economic Surveys 17(2):115–56. PASEC Congo-Brazzaville. Haughton, Jonathan, and Shahidur R. Khandker. 2009. Coudouel, A., J. Hentschel, and Q. Wodon. 2002. Handbook on Poverty and Inequality. Washington, “Poverty Measurement and Analysis.” In A DC: World Bank. Sourcebook for Poverty Reduction Strategies, Volume Haushofer, Johannes, and Shapiro, Jeremy. 2014. 1: Core Techniques and Cross-Cutting Issues, edited “Household Response to Income Changes: Evidence by J. Klugman. Washington, DC: The World Bank. from an Unconditional Cash Transfer Program in Cowell, F.A., Jenkins, S.P., 1995. “How much inequality Kenya.” can we explain? A methodology and an application IRIN (Integrated Regional Information Networks). 2012. to the USA”. Economic Journal, 105, 421–430. “Congo-Brazzaville: Imports, Corruption Drive Up Datt, G., and M. Ravallion. 1992. “Growth and Food Prices.” http://www.irinnews.org/report/96889/ Redistribution Components of Changes in Poverty congo-imports-corruption-drive-food-prices. Measures: A Decomposition with Applications Jaramillo, Sandra G. 2014. “Putting Children and to Brazil and India in the 1980s.” Journal of Adolescents at the Center of Housing Policy: A Latin Development Economics 38 (2): 275–295. American Perspective, in Housing, Contexts, and Duclos, J. Y., P. Makdissi, and Q. Wodon. 2008. the Well-Being of Children and Youth.” Cityscape: “Socially Efficient Tax Reforms.” International A Journal of Policy Development and Research 16 (1): Economic Review 49 (4): 1505–1537. 197–201. Firpo, S., N. Fortin, and T. Lemieux. 2009. “Unconditional Jolliffe, D. 2002. “Whose Education Matters in the Quantile Regressions.” Econometrica 77 (3): 953–973. Determination of Household Income? Evidence Foster, J. E., J. Greer, and E. Thorbecke. 1984. “A Class from a Developing Country.” Economic Development of Decomposable Poverty Indices.” Econometrica and Cultural Change 50 (2): 287–312. 52: 761–766. Kurosaki T. and Khan H. 2006. “Human capital, Foster, V., and M. C. Araujo. 2004. “Does Infrastructure productivity, and stratification in rural Pakistan,” Reform Work for the Poor? A Case Study from Review of Development Economics, vol. 10, no. 1, Guatemala.” Policy Research Working Paper 3185, pp. 116–134. Washington, DC: The World Bank. Kolenikov, Stanislav, and Anthony Shorrocks. 2005. Gable, Susanna, Hans Lofgren, and Israel Osorio-Rodarte. “A Decomposition Analysis of Regional Poverty in 2015. Trajectories for Sustainable Development Goals: Russia.” Review of Development Economics 9 (1): 25–46. Framework and Country Applications. Washington, Krieger, J., and D. L. Higgins. 2002. “Housing and DC: The World Bank. Health: Time Again for Public Health Action.” Galor, O. & Weil, D. N. 1999. From Malthusian stagna- American Journal of Public Health 92 (5): 758–768. tion to modern growth. American Economic Review Makdissi, P., and Q. Wodon. 2002. “Consumption 89(2): 150–154. Dominance Curves: Testing for the Impact of 140 Republic of Congo – Poverty Assessment Report Indirect Tax Reforms on Poverty.” Economics Letters Psacharopoulos, George, and Harry A. Patrinos. 2004. 75: 227–235. “Returns to Investment in Education: A Further Makinen, Marty, Leo Deville, and Amanda Folsom. Update.” Education Economics 12(2):111–34. 2012. Assessment of the Private Health Sector in the Ravallion, Martin. 2004. “Pro-Poor Growth: A Primer.” Republic of Congo. A World Bank Study No. 72464. Policy Research Working Paper 3242, World Bank. Washington, DC: The World Bank. Ravallion, M., and S. Chen. 2003. “Measuring Pro-Poor Malhotra, Anju, Rohini Pande, and Caren Grown. Growth.” Economic Letters 78 (1): 93–99. 2003. Impact of Investments in Female Education on Ravallion, M., and M. Huppi. 1991. “Measuring Gender Equality. International Center for Research Changes in Poverty: A Methodological Case Study on Women. Washington, DC. of Indonesia during an Adjustment Period.” World Mensah, Edouard, and Michael O’Sullivan. 2016. Bank Economic Review 5: 57–82. “Moving Out and Up: Panel Data Evidence on République du Congo. 2012a. “Plan National de Migration and Poverty in Uganda.” Background paper Développement PND 2012-2016, livre 1 (Document for the Uganda Poverty Assessment, World Bank. de Stratégie pour la croissance, l’Emploie et la réduc- Montenegro, Claudio E. and Patrinos Harry A. 2014. tion de la pauvreté).” “Comparable Estimates of Returns to Schooling ———. 2012b. “Plan National de Développement PND Around the World.” Policy Research Working Paper 2012-2016, livre 2 (Condensé des PAS/CDMT No 7020. Washington, DC: The World Bank. ministériels.” Psacharopoulos, George. 1972. “Rates of Return on Safoulanitou, Léonard Nkouka, and Mathias Marie Investment in Education around the World.” Adrien Ndinga. 2010. “An Empirical Analysis of the Comparative Education 16(1):54–67. Determinants of Food Imports in Congo.” AERC ———. 1973. “Returns to Education: An International Research Paper 195, African Economic Research Comparison.” Amsterdam, San Francisco: Elsevier Consortium, Nairobi. and Jossey-Bass. 38 Sahn, David E., and David C. Stifel. 2000. “Poverty ——— . 1985. “Returns to Education: A Further Comparisons Over Time and Across Countries in International Update and Implications.” Journal of Africa.” World Development 2000; 28(12):2123–55. Human Resources 20(4):583–604. ——— . 2003. “Exploring Alternative Measures of ———. 1989. “Time Trends of the Returns to Education: Welfare in the Absence of Expenditure Data.” Cross-National Evidence.” Economics of Education Review of Income and Wealth. Series 49, Number 4. Review 8(3):225–31. Stampini, Marco, and Leopoldo Tornarolli. 2012. “The ———. 1994. “Returns to Education: A Global Update.” Growth of Conditional Cash Transfers in Latin World Development 22(9):1325–43. America and the Caribbean: Did They Go Too Far?” ——— . 1995. “The Profitability of Investment in IZA Policy Paper No. 49. Education: Concepts and Methods.” Washington, Tsimpo, C., and Q. Wodon, eds. 2014a. Residential DC: The World Bank. Electricity in Uganda. Washington, DC: The World Psacharopoulos, George, and Richard Layard. 1979. Bank. “Human Capital and Earnings: British Evidence ——— . 2014b. Residential Network Water in Uganda. and a Critique.” Review of Economic Studies Washington, DC: The World Bank. 46(3):485–503. Turnbull, Colin M. 1961. The Forest People: A Study of the ——— . 2012. “Rates of Return to Investment in Pygmies of the Congo. New York: Simon & Schuster. Education: An International Comparison.” The Wodon, Q. and S. Yitzhaki, 2002, “Inequality and World Bank, Human Development Network. Social Welfare” in J. Klugman, editor, A Sourcebook Washington, DC. Processed. for Poverty Reduction Strategies, Volume 1: Core References 141 Techniques and Cross-Cutting Issues, Washington, ——— . 2014. “Enhancing Efficiency in Education DC: The World Bank. and Health Public Spending for Improved Quality Wodon, Q., S. Banerjee, A. Diallo, and V. Foster. 2009. Service Delivery for All.” A Public Expenditure “Is Low Coverage of Modern Infrastructure Services Review of the Education and Health Sectors. in African Cities Due to Lack of Demand or Lack Washington, DC: The World Bank of Supply?” Policy Research Working Paper 4881, ——— . 2015a. “Implementing Public Financial Washington, DC: The World Bank. Management Reforms to Stimulate Growth and Wodon, Q., P. Backiny-Yetna, A. Ben-Achour. 2012. Achieve Shared Prosperity.” Public Expenditure “Indigenous Peoples in Central Africa: the Case Management and Financial Accountability Review of the Pygmies” In G. Hall & H. Patrinos (Eds.), (PEMFAR). Washington, DC: The World Bank Indigenous peoples, poverty, and development (pp. ——— . 2015b. “Policy Note Cassava Value Chain”, 125–176). Washington, DC: The World Bank. Washington, DC: The World Bank. World Bank. 2008. “Infrastructures et pauvreté au ———. 2016a. “Systematic Country Diagnostic: Ending Congo” Projet Eau et Développement Urbain Extreme Poverty and Boosting Share Prosperity on (PEDU). Washington, DC: The World Bank Road to the Emerging Status”, Washington, DC: ——— . 2009. Democratic Republic of Congo: Strategic The World Bank. Framework for the Preparation of a Pygmy Development ——— . 2016b. “Mitigating the Impact of Oil Market Program, Report No. 51108–ZR, Washington, DC: Fluctuations on Fiscal Revenue”, Third Economic The World Bank. Update, Washington, DC: The World Bank. ——— . 2011. “Republic of Congo: Employment and Growth Study - From Jobless to Inclusive Growth.” 142 Republic of Congo – Poverty Assessment Report 1818 H Street, NW Washington, DC 20433 USA Telephone: +1 202 473 1000 Internet: www.worldbank.org