Report No. 36546-MW Malawi Poverty and Vulnerability Assessment Investing in Our Future Full Report December 2007 Poverty Reduction and Economic Management 1 Africa Region Document of the World Bank Republic of Malawi The World Bank Malawi Poverty and Vulnerability Assessment Investing in Our Future Full Report December 2007 CURRENCY EQUIVALENTS (Exchange Rate Effective December 1, 2007) Currency Unit = Malawi Kwacha (MK) MK1 = US$0.0072 US$1 = MK139 MEASURES Metric System FISCAL YEAR July 1 to June 30 (as of July 1998) ABBREVIATIONS AND ACRONYMS ADD Agricultural Development Division ADMARC Agricultural Development and Marketing Corporation AGOA African Growth and Opportunity Act AHL Auction Holding Limited ARET Agriculture Research and Extension Trust ART Anti-Retroviral Therapy BOP Balance of Payments CEM Country Economic Memorandum COMESA Common market for Eastern and Southern Africa CPI Consumer Price Index DFID Department for International Development DHS Demographic Health Survey EBA Everything But Arms EPZ Export Processing Zone ESCOM Electricity Supply Commission of Malawi EU European Union FAO Food and Agriculture Organization of the United Nations FBO Faith-Based Organization FEWS Famine Early Warning System FTA Free Trade Area GDP Gross Domestic Product GOM Government of Malawi HAART Highly Active Anti-Retroviral Therapy HIPC Highly Indebted Poor Countries IDA International Development Association IFC International Finance Corporation IHS Integrated Household Survey IMF International Monetary Fund LCE Local Commodity Exchange LDC Less Developed Countries ii M&E Monitoring and Evaluation MASAF Malawi Social Action Fund MDGs Millennium Development Goals MDHS Malawi Demographic Health Survey MEPD Ministry of Economic Planning and Development MGDS Malawi Growth and Development Strategy MOAI Ministry of Agriculture and Irrigation MOHP Ministry of Health and Population MPRS Malawi Poverty Reduction Strategy MRA Malawi Revenue Authority MRFC Malawi Rural Finance Company NAC National AIDS Commission NACP National AIDS Control Program NASFAM National Small Farmers Association NFRA National Food Reserve Agency NGO Non-Governmental Organization NPV Net Present Value NRA National Road Authority NSNS National Safety Nets Strategy NSO National Statistical Office ORT Other Recurrent Costs PA Prime Age PRGF Poverty Reduction and Growth Facility PRSP Poverty Reduction Strategy Paper PSIA Poverty and Social Impact Analysis PWP Public Works Program RBM Reserve Bank of Malawi SADC Southern Africa Development Community SFFRFM Smallholder Farmers' Fertilizer Revolving Fund of Malawi SGR Strategic Grain Reserve SME Small and medium Scale Enterprise TAMA Tobacco Association of Malawi TCC Tobacco Control Commission TIP Targeted Inputs Program TLU Tropical Livestock Unit TNP Targeted Nutrition Program USAID United States Agency for International Development VCT Voluntary Counseling and Testing The Malawi Poverty and Vulnerability Assessment was prepared by a core team of Government and World Bank officials led by Antonio Nucifora (AFTP1, Senior Economist, the World Bank), and Time Fatch (Ministry of Economic Planning and Development), and comprising Kathleen Beegle (DECRG, World Bank), Gero Carletto (DECRG, World Bank), Rhoda Eliasi (Ministry of Economic Planning and Development), Shelton Kanyanda (National Statistical Office), Khwima Nthara (AFTP1, World Bank), and Diane Steele (DECRG, World Bank). iii iv Acknowledgements This report was prepared jointly between the Government of Malawi and the World Bank, and in collaboration with Statistics Norway and the FAO. The report was prepared under the guidance of Patrick Kamwendo (Principal Secretary, Ministry of Economic Planning and Development), Charles Machinjili (Commissioner, National Statistical Office), and Emmanuel Akpa (Sector Manager AFTP1, World Bank). The report also benefited from the input and guidance of Ben Botolo (Director M&E, Ministry of Economic Planning and Development) and Mercy Kanyuka (Deputy Commissioner, National Statistical Office). Michael Baxter (Country Director, World Bank), Timothy Gilbo (Country Manager, World Bank), Hartwig Schafer (Director of Operations, Africa Region, World Bank), Sudhir Shetty (PREM Director, Africa Region, World Bank) and John Page (Chief Economist, Africa Region, World Bank) also provided overall direction and guidance for the analysis. This report has been prepared by a joint core team led by Antonio Nucifora (AFTP1, World Bank) and Time Fatch (Ministry of Economic Planning and Development), and comprising Kathleen Beegle (DECRG, World Bank), Gero Carletto (DECRG, World Bank), Rhoda Eliasi (Ministry of Economic Planning and Development), Shelton Kanyanda (National Statistical Office), Khwima Nthara (AFTP1, World Bank), and Diane Steele (DECRG, World Bank). Though many people contributed to each chapter, and are too numerous to name and thank individually, the principal authors of each chapter are as follows: Chapter One: Kathleen Beegle (DECRG, World Bank); Rhoda Eliasi (Ministry of Economic Planning and Development), Time Fatch (Ministry of Economic Planning and Development), Shelton Kanyanda (National Statistical Office), Antonio Nucifora (AFTP1, World Bank), Diane Steele (DECRG, World Bank), Thomas Pave Sohansen (AFTP1, World Bank); Chapter Two: Diane Steele (DECRG, World Bank), Kathleen Beegle (DECRG, World Bank), Rhoda Eliasi (Ministry of Economic Planning and Development), Time Fatch (Ministry of Economic Planning and Development), Shelton Kanyanda (National Statistical Office), Antonio Nucifora (AFTP1, World Bank); Chapter Three: Kathleen Beegle (DECRG, World Bank); Chapter Four: Gero Carletto (DECRG, World Bank), Benjamin Davis (Food and Agriculture Organization), Time Fatch (Ministry of Economic Planning and Development), Marcella Vigneri (Food and Agriculture Organization), Carlo Azzarri (Food and Agriculture Organization); Chapter Five: Kathleen Beegle (DECRG, World Bank), Flora Nankhuni (University of Harvard); Chapter Six: Antonio Nucifora (AFTP1, World Bank); Chapter Seven: Guido Porto (DECRG, World Bank), Jorge Balat (DECRG, World Bank); Chapter Eight: Time Fatch (Ministry of Economic Planning and Development), Antonio Nucifora (AFTP1, World Bank), Hardwick Tchale (AFTAR, World Bank); Chapter Nine: Kathleen Beegle (DECRG, World Bank), Antonio Nucifora (AFTP1, World Bank); v Chapter Ten: Khimwa Nthara (AFTP1, World Bank), Rhoda Eliasi (Ministry of Economic Planning and Development); Chapter Eleven: Ronnie Hamad (AFTRL, World Bank), Diana Masone (AFTRL, World Bank), Rhoda Eliasi (Ministry of Economic Planning and Development), Time Fatch (Ministry of Economic Planning and Development). Other contributors to the various chapters of the report deserve special recognition. From the World Bank: Harold Aldermann, Sushenjit Bandyopadhyay, Uwe Deichmann, Michael Lokshin, Alexander Lotsch, Pavel Lukyantsau, Denis Nikitin, Priya Shyamsundar, Martin Ravallion, Helene Carlsson Rex, Quentin Wodon. From beyond the World Bank: Ephraim Chirwa (Chancellor College, University of Malawi), Maxton Tsoka, John Kadzandira (Center for Social research, University of Malawi), Manohar Sharma (IFPRI), Astrid Mathiassen, Bjorn World, Moyo Gunvor Iversen (Statistics Norway), Lindsay Mangham (DFID/ODI). Samira Leakey provided extensive and invaluable comments and advice in compiling and shaping the report. Excellent and extensive advice was also received from Peter Lanjouw, Jesko Hentchel, (World Bank), Todd Benson (IFPRI) and Ephraim Chirwa (Chancellor College, University of Malawi), who were the peer reviewers for this report. The report also benefited from numerous valuable comments. From the World Bank: Wim Alberts, Francisco Carneiro, Alfred Chirwa, Sudhir Chitale, Louise Fox, Rae Galloway, Valerie Kozel, Mungai Lenneiye, Renos Vakis, Trond Vedeld, Jos Verbeek. From beyond the World Bank: Nicholas Staines (IMF), Stephen Devereux, Bob Baulch, Rachel Sabates-Wheeler (Institute of Development Studies, Sussex), Isabel Cardinal, Leigh Stubblefield, Bernabe Sanchez, Alan Whitworth (DFID, Malawi), Domenico Scalpelli, Blessings Mwale (WFP, Malawi), Benjamin Banda (Chancellor College, University of Malawi), Eyob Zere (WHO Malawi), as well as from numerous participants at the discussion workshop of the draft report in May 2006 in Lilongwe, including members of the Government, development partners, civil society and parliamentarians in Malawi. Valuable input was also received from participants to the various Thematic Working Groups and the various discussion workshops (of the concept note in May 2005, preliminary findings on poverty lines and poverty profile in October 2005, and preliminary findings on risk and social protection in December 2005). Much of the background research was supported by funding from the World Bank and from a multi-donor trust fund (the Trust Fund for Environmentally and Socially Sustainable Development, TFESSD), which is funded by Norway and Finland, and provides grant resources for World Bank activities aimed at mainstreaming the environmental, social and poverty reducing dimensions of sustainable development into overall Bank work. Funding was also received from the FAO/WB Cooperation Program to undertake the analysis of food security and malnutrition. Statistics Norway contributed to the preparation of the report. The cover page has been designed by Samira Leakey, and the photograph in the cover was taken by Kathleen Beegle. Dotilda Sidibe, Annie Jere, Grace Soko, Rose Ndalama, Ethel Kuniya provided excellent assistance in various aspects of the management and administration of the project, and the preparation of the final document. vi MALAWI POVERTY AND VULNERABILITY ASSESSMENT INVESTING IN OUR FUTURE TABLE OF CONTENTS INTRODUCTION......................................................................................................................... 1 CHAPTER 1: DIMENSIONS OF POVERTY IN MALAWI................................................... 3 INTRODUCTION..........................................................................................................................................................3 INCOME MEASURES OF POVERTY AND INEQUALITY..................................................................................................3 Income poverty and ultra-poverty in Malawi .....................................................................................................3 Geographical variations in poverty levels...........................................................................................................4 Income Inequality in Malawi..............................................................................................................................8 Depth of poverty measures................................................................................................................................10 EVOLUTION OF POVERTY AND INEQUALITY BETWEEN 1998 AND 2005 ...................................................................12 EVOLUTION OF NON-INCOME DIMENSIONS OF POVERTY ........................................................................................14 Knowledge: changes in education indicators for Malawi................................................................................15 A long and healthy life: changes in health indicators for Malawi..................................................................16 MALAWI'S PROGRESS TOWARDS THE ACHIEVEMENT OF MILLENNIUM DEVELOPMENT GOALS.............................18 CHAPTER 2: POVERTY PROFILE AND THE DETERMINANTS OF POVERTY ....... 21 INTRODUCTION........................................................................................................................................................21 THE CHARACTERISTICS OF POOR HOUSEHOLDS IN 2005: THE POVERTY PROFILE...................................................21 Demographic characteristics of poor households: household size and dependency ratio..............................21 Key characteristics of the household head: gender, age and level of education.............................................23 Occupation of the Household Head .................................................................................................................25 Education and the poor: school attendance, and enrollment in primary, secondary, and tertiary education, and literacy rate.................................................................................................................................................26 Health and the poor: morbidity, births attended by skilled health personnel, chronic health problems, and child malnutrition..............................................................................................................................................31 Housing: overall quality of the dwellings, sanitation, water, cooking fuel and lighting fuel.........................34 Household assets: ownership of durable goods, livestock, land and labor .....................................................38 Gender dimensions in labor and income earnings ..........................................................................................43 Access to roads, transport, and distance from markets....................................................................................45 Access to communications ................................................................................................................................47 THE DETERMINANTS OF POVERTY IN MALAWI IN 2005...........................................................................................49 Modeling the determinants of poverty..............................................................................................................49 Results of the analysis.......................................................................................................................................51 CHAPTER 3: RISK AND VULNERABILITY TO SHOCKS IN MALAWI....................... 56 INTRODUCTION........................................................................................................................................................56 RISKS AND MOVEMENTS IN AND OUT OF POVERTY..................................................................................................56 Risks and shocks as a cause of poverty ............................................................................................................56 The dynamic dimension of poverty...................................................................................................................58 VULNERABILITY TO RISK AND POVERTY IN MALAWI..............................................................................................60 Extent of Risk in Malawi ..................................................................................................................................60 Major Types of Risks in Malawi.......................................................................................................................62 Characteristics of Shocks in Malawi................................................................................................................67 CHRONIC POVERTY AND TRANSIENT POVERTY IN MALAWI....................................................................................68 Chronic poverty in Malawi ...............................................................................................................................69 Transient poverty and household vulnerability to shocks................................................................................70 COPING WITH SHOCKS: HOUSEHOLD COPING STRATEGIES......................................................................................73 vii CHAPTER 4: FOOD AND NUTRITION SECURITY IN MALAWI .................................. 80 INTRODUCTION........................................................................................................................................................80 MEASURING FOOD SECURITY AND MALNUTRITION AT THE HOUSEHOLD LEVEL.....................................................80 Undernourishment............................................................................................................................................80 Caloric availability ............................................................................................................................................80 Depth and severity of hunger............................................................................................................................84 Self assessment of food adequacy.....................................................................................................................84 Dietary diversity.................................................................................................................................................86 Seasonality issues in Food Security..................................................................................................................88 Coping via changes in consumption patterns...................................................................................................91 Anthropometric measures.................................................................................................................................92 A PROFILE OF MALNUTRITION IN MALAWI...............................................................................................................94 THE DETERMINANTS OF CHILD MALNUTRITION AND HOUSEHOLD CALORIC AVAILABILITY......................................98 Model specifications for child malnutrition.....................................................................................................98 Model for household calorie availability........................................................................................................103 AGRICULTURE AND FOOD SECURITY.....................................................................................................................105 Access to Land.................................................................................................................................................105 Land and staple crops .....................................................................................................................................106 Maize yields .....................................................................................................................................................108 What role for other staple crops?....................................................................................................................108 Tobacco and food security ..............................................................................................................................109 CHAPTER 5: THE IMPACT OF CHRONIC ILLNESS AND HIV/AIDS ON HOUSEHOLDS IN MALAWI ................................................................................................ 113 INTRODUCTION......................................................................................................................................................113 THE HIV/AIDS EPIDEMIC IN MALAWI..................................................................................................................113 HIV/AIDS prevalence in Malawi....................................................................................................................114 Knowledge of HIV/AIDS, safe sex practices and HIV testing.......................................................................114 DEMOGRAPHIC IMPACTS OF HIV/AIDS IN MALAWI .............................................................................................118 THE IMPACT OF HIV/AIDS AT THE HOUSEHOLD LEVEL .......................................................................................119 HIV/AIDS and Poverty ...................................................................................................................................119 Other differences in household characteristics..............................................................................................120 HIV/AIDS and Agriculture.............................................................................................................................123 HIV/AIDS and Gender ...................................................................................................................................124 HIV/AIDS and Household Assets...................................................................................................................124 Economic Implications of Illness in Households...........................................................................................126 Household Coping Mechanisms.....................................................................................................................129 Orphans and HIV/AIDS .................................................................................................................................129 IMPACTS OF HIV/AIDS AT THE MACRO LEVEL.....................................................................................................134 HIV/AIDS and Economic Growth..................................................................................................................134 HIV/AIDS and Access to Public Goods and Services....................................................................................135 Institutional Response to HIV/AIDS..............................................................................................................136 CHAPTER 6: MACROECONOMIC DEVELOPMENTS AND IMPLICATIONS FOR POVERTY................................................................................................................................. 138 INTRODUCTION......................................................................................................................................................138 MACROECONOMIC DEVELOPMENTS SINCE THE EARLY 1990S...............................................................................138 Economic growth and its composition............................................................................................................138 Changes in inflation since the early 1990s.....................................................................................................140 Interest Rates movements over the past decade..............................................................................................141 Movements in the Exchange Rate ..................................................................................................................142 PUBLIC EXPENDITURES AND THE PROVISION OF SOCIAL SERVICES.........................................................................143 Impact of Government Borrowing..................................................................................................................145 Impact of Increased Debt on Expenditure .....................................................................................................146 CONCLUSIONS AND POLICY RECOMMENDATIONS ..................................................................................................148 viii CHAPTER 7: AGRICULTURE GROWTH AND PRODUCTIVITY IN MALAWI........ 151 INTRODUCTION......................................................................................................................................................151 BRIEF OVERVIEW OF MALAWI'S AGRICULTURAL SECTOR....................................................................................151 MAIN CHARACTERISTICS OF SMALLHOLDER AGRICULTURAL PRODUCTION IN MALAWI ......................................153 Land availability and smallholders' access to land........................................................................................153 Crops grown by smallholders..........................................................................................................................157 Labor use in smallholders' agricultural production......................................................................................162 Smallholder Fertilizer use...............................................................................................................................165 Access to Credit ...............................................................................................................................................168 Access to agricultural extension services and benefit incidence analysis of public spending in extensions services.............................................................................................................................................................169 PRODUCTIVITY OF SMALLHOLDER AGRICULTURE IN MALAWI..............................................................................171 An analysis of smallholder farmers' efficiency in hybrid maize and burley tobacco production ................172 MARKETING OF SMALLHOLDERS AGRICULTURAL PRODUCTS.................................................................................183 Maize marketing in Malawi: The findings of the Poverty and Social Impact Analysis (PSIA) of the restructuring of ADMARC..............................................................................................................................184 Tobacco marketing in Malawi: The findings of the Poverty and Social Impact Analysis (PSIA) of the reform of tobacco marketing arrangements...................................................................................................188 GOVERNMENT POLICIES FOR SMALLHOLDER AGRICULTURAL GROWTH...............................................................191 CHAPTER 8: POVERTY AND TRADE IN AGRICULTURAL COMMODITIES......... 201 INTRODUCTION......................................................................................................................................................201 EXPORT PARTICIPATION, COMMODITY EXPORTS GAINS AND POVERTY................................................................201 Income Gains in Commodity Exports.............................................................................................................203 Constraints to Household Participation in Agricultural Export Commodities.............................................206 ASSESSING THE IMPACT ON POVERTY OF REMOVING TRADE CONSTRAINTS.........................................................211 CHAPTER 9: SOCIAL PROTECTION AND DISASTER MANAGEMENT FOR POVERTY ALLEVIATION AND GROWTH IN MALAWI.............................................. 213 INTRODUCTION......................................................................................................................................................213 A REVIEW OF THE EXISTING SOCIAL PROTECTION SYSTEM IN MALAWI ...............................................................213 Direct Food Transfers and subsidized food sales...........................................................................................215 Public Works Programs ..................................................................................................................................222 TIP and Fertilizer Subsidies ...........................................................................................................................225 Main findings from the review of the existing safety nets system..................................................................230 POSSIBLE PRINCIPLES FOR REFORM OF SOCIAL PROTECTION SYSTEM IN MALAWI ...............................................233 Measures to alleviate chronic poverty and build the assets of the chronically poor.....................................234 Measures to enhance productivity and strengthen risk management for the transient poor .......................244 CHAPTER 10: ACCESS TO PUBLIC SERVICES AND THE DISTRIBUTION OF BENEFITS FROM PUBLIC EXPENDITURES IN HEALTH AND EDUCATION ........ 251 INTRODUCTION......................................................................................................................................................251 ACCESS TO EDUCATION SERVICES AND BENEFIT INCIDENCE ANALYSIS OF PUBLIC SPENDING IN THE EDUCATION SECTOR..................................................................................................................................................................251 Primary Education..........................................................................................................................................251 Access to Secondary Education ......................................................................................................................254 Access to Tertiary Education..........................................................................................................................258 Summary of Access to Education, and Benefit Incidence of Public Spending in Education ......................259 ACCESS TO HEALTH SERVICES AND BENEFIT INCIDENCE ANALYSIS OF PUBLIC SPENDING IN THE HEALTH SECTOR ..............................................................................................................................................................................260 Access to Health Services................................................................................................................................260 Benefit Incidence Analysis of Public Spending in the Health Sector ...........................................................263 POLICY IMPLICATIONS...........................................................................................................................................265 ix CHAPTER 11: MONITORING AND EVALUATION SYSTEMS AND INSTITUTIONS ..................................................................................................................................................... 268 INTRODUCTION......................................................................................................................................................268 THE POLITICAL AND INSTITUTIONAL FRAMEWORK FOR M&E ..............................................................................268 The Role of Civil Society in Holding Government Accountable ...................................................................271 MALAWI'S NATIONAL SURVEY PROGRAM ............................................................................................................272 MALAWI'S MPRS/MGDS IMPLEMENTATION MONITORING SYSTEM....................................................................275 MALAWI'S BUDGET AND EXPENDITURE MONITORING SYSTEM ............................................................................278 Public Financial Management Work in Progress..........................................................................................280 POLICY RECOMMENDATIONS.................................................................................................................................281 Short- to medium-term reforms......................................................................................................................281 Long-term reforms ..........................................................................................................................................282 REFERENCES.......................................................................................................................... 283 TABLES Table 1.1: Poverty line in Malawi Kwacha per person per year....................................................................................4 Table 1.2: Poverty headcount and distribution of Malawi's poor by place of residence in 2005..................................6 Table 1.3: Gini coefficient in 1998 and 2005.....................................................................................9 Table 1.4: Poverty Headcount, Income Gap, and Severity of Poverty estimates in 1998 and 2005...........................11 Table 1.5: Malawi, key human development indicators, 2003 ....................................................................................14 Table 1.6: Summary of Malawi's progress towards the MDGs as of end-2005..........................................................20 Table 3.1: Shocks in the past 5 years as reported by households (percentage of households reporting).....................61 Table 3.2: Sources of income earnings (percent of households reporting)..................................................................74 Table 3.3: Sources of income earnings (percent of households reporting)..................................................................76 Table 3.4: First response to major shocks reported by households (% of households) ...............................................78 Table 4.1: Caloric availability per capita per day: by region, location and poverty status .........................................82 Table 4.2: Nutritional status of children aged 6-59 months (by region and location) ................................................93 Table 4.3: Moderate stunting, by survey ....................................................................................................................93 Table 4.4: Comparing malnutrition in selected countries in Sub-Sahara Africa.........................................................94 Table 4.5: Share of household caloric availability from home production, by region, location and poverty status. 106 Table 5.1: Incidence of deaths in past two years, by age of deceased ......................................................................119 Table 5.2: Characteristics of households with and without deaths...........................................................................122 Table 5.3: Probit regressions for household no longer producing tobacco, among household growing tobacco 5 years earlier.........................................................................................................................................................................124 Table 5.4: Asset and land losses related to deaths in last 2 years.............................................................................125 Table 5.5: Funeral expenses in last 12 months ..........................................................................................................126 Table 5.6: Incidents of illnesses in households.........................................................................................................127 Table 5.7: Responses to deaths of household members that occurred in past 5 years ...............................................129 Table 5.8: Characteristics of households with and without children and orphans aged 0-17 ....................................131 Table 5.9: Accessibility of community programs that may mitigate the impacts of HIV/AIDS, in 2005 .................136 Table 7.1: Malawi land use in 1998 (million hectares)........................................................................151 Table 7.2: Share of households with access to land, by type of land, region and poverty status...............................153 Table 7.3: Structure of landholdings by region, poverty status, and wealth quintiles (percent)................................155 Table 7.4: Percentage of households cultivating different crops, by poverty status and land and expenditure quintiles..............................................................................................................................158 Table 7.5: Percentage of households selling each type of food crop group*.............................................................160 Table 7.6: Farmer's Access to Credit by Land and Expenditure Quintiles................................................................168 Table 7.7: Proportion of farmers that received advice from field assistant by quintile .............................................170 Table 7.8: Estimates of efficiency from empirical studies in developing countries ..................................................174 Table 7.9: Mean technical, allocative and economic efficiency of Malawian smallholder farmers (monocrop) .....175 Table 7.10: Efficiency levels by expenditure and land quintile (percent) .................................................................181 x Table 8.1: Export Cropping and Income Gains................................................................................204 Table 9.1: Receipt of program benefits (percent of IHS2 households reporting).....................................................215 Table 9.2: Receipt of program benefits 2001-2003 (percent of IHS2 households reporting)...................................218 Table 9.3: Number of households who benefited from social protection intervention (calculated from IHS2 households reporting) ................................................................................................................................................219 Table 9.4: TIP fertilizer receipt and use in 2003/2004 cropping season (percent of households reporting)..............229 Table 9.5: Estimates of the poverty gap in MK.........................................................................................................235 Table 9.6: Targeting using proxy means testing (percent).........................................................................................239 Table 9.7: Poverty rates for categories of households ...............................................................................................241 Table 9.8: Categorical targeting under alternative scenarios.....................................................................................242 Table 10.1: Distance to Nearest Government Primary School (percent)...................................................254 Table 10.2: Number of Teachers and Students in Government Primary Schools (average)............................254 Table 10.3: Proportion of students enrolled in government primary school, by quintile...........................................253 Table 10.4: Percentage share of public expenditure on primary education by quintile .............................................254 Table 10.5: Distance to the Nearest Secondary School (percent)...........................................................257 Table 10.6: Number of Teachers and Students in Secondary Schools (average)......................................................255 Table 10.7: Proportion of students enrolled in government secondary schools by quintile.......................................256 Table 10.8: Proportion of students enrolled in government secondary schools by type of secondary school and by quintile (percent)........................................................................................................................................................257 Table 10.9: Proportion of students enrolled in boarding government secondary schools, by quintile ......................257 Table 10.10: Percentage share of public expenditure on secondary education by quintile........................................258 Table 10.11: Proportion of students enrolled in tertiary education institutions by quintile.......................................258 Table 10.12: Distance to Nearest Health Clinic and presence of Nurse/Midwife (percent).............................263 Table 10.13. Distance to Nearest Health Clinic with Medical Doctor (percent)..........................................263 Table 10.14: Proportion of people who went to a government health facility when sick..........................................262 Table 10.15: Proportion of people hospitalized at a government health facility when sick ......................................262 Table 10.16: Percentage share of public expenditure in curative health services, by quintile...................................264 FIGURES Figure 1.1: Proportion of poor and ultra-poor persons by region in 2005 .....................................................................6 Figure 1.2: Map of poverty headcount in urban areas...........................................................................7 Figure 1.3: Map of poverty headcount at Traditional Authority level.........................................................8 Figure 1.4: Distribution of per capita consumption expenditure in 2005 .....................................................................9 Figure 1.5: Per capita consumption expenditure by decile in 2005............................................................9 Figure 1.6: Lorenz Curve in 2005..................................................................................................................................9 Figure 1.7: Proportion of the population deemed poor and ultra-poor in 1998 and 2005...........................................12 Figure 1.8: Malawi Human Development Index, 1975-2005 .....................................................................................15 Figure 1.9: Life expectancy at birth in Malawi, 1967-2002 ........................................................................................16 Figure 1.10: Physicians per 100,000 people in Malawi, 1967-2002............................................................................16 Figure 1.11. Trends in Infant and Under-five Mortality, 1992 to 2004.......................................................................17 Figure 2.1: Household size, number of children, and dependency ratio, by decile, 2005............................................23 Figure 2.2: Demographic composition of poverty in Malawi, 2005............................................................................23 Figure 2.3: Population Poverty Rates by Sex of Household Head and Residence .....................................................24 Figure 2.4: Population poverty rates by age group of household head........................................................................24 Figure 2.5: Education of Household Heads by wealth decile (percent)......................................................................24 Figure 2.6: Occupation of the Household Head...........................................................................................................25 Figure 2.7: School attendance by poverty status .........................................................................................................26 Figure 2.8: Primary Gross Enrollment Rate in 2004-2005 (percent)..........................................................................28 Figure 2.9: Secondary Gross Enrollment Rate by Wealth Decile (richest and poorest), 2004-2005 (percent)...........30 Figure 2.10: Adult Literacy Rate (percent)..................................................................................................................31 Figure 2.11: Proportion of Persons Reporting Having a Chronic Illness.....................................................................32 Figure 2.12: Caloric Intake (daily kcal per capita) and Child Nutritional Status (percent underweight)....................33 Figure 2.13: Housing Quality by Wealth Decile and Location (percent) ....................................................................34 xi Figure 2.14: Proportion of Population with Improved Water Source.........................................................................35 Figure 2.15: Households Cooking with Firewood by Wealth Decile (percent)..........................................................36 Figure 2.16: Household Lighting Fuel by Residence and Wealth Decile (percent)....................................................36 Figure 2.17: Household Ownership of Selected Assets (percent of households) .......................................................38 Figure 2.18: Average Livestock Ownership in Malawi (TLU/household by household type and residence), and in Southern Africa (TLU/100 people)..............................................................................................................................39 Figure 2.19: Land Holdings: Average Hectares of Land Per Capita ..........................................................................40 Figure 2.20: Average weekly time spent working (adults, age 15 plus), by wealth quintile .......................................42 Figure 2.21: Distance to nearest tar/asphalt road and number of months road is passable..........................................46 Figure 2.22: Estimated household travel time to nearest trading center by wealth deciles..........................................47 Figure 2.23: Proportion of households with telephones (percent of households)........................................................47 Figure 2.24: The correlates of poverty in Malawi in 2005 (percentage change effect) ...............................................52 Figure 3.1: Cumulative Density Function of per capita consumption associated with consumption increases/decreases of 20 percent ................................................................................................................................60 Figure 3.2: Ratio of Oct-April cumulative rainfall to 40-year average at weather stations, 1996-2006......................62 Figure 3.3: Average rainfall level and variability (1962-2005) by income status .......................................................63 Figure 3.4: Monthly average maize price (MK/Kg) in nominal and real terms, 2001-2006 .......................................64 Figure 3.5: Most severe shocks in the past 5 years (percent of households reporting)................................................67 Figure 3.6: Ratio of consumption per capita to the poverty line, by wealth decile......................................................69 Figure 3.7: Poverty rates by orphan status...................................................................................................................71 Figure 3.8: Household reports of the number of different shocks occurring in the last five years, by urban and rural .....................................................................................................................................................................................72 Figure 3.9 Ganyu earnings as a share of household expenditure, by wealth decile.....................................................75 Figure 3.10: Sources of income earnings (percent of households reporting)...............................................................76 Figure 3.11: Temporary withdraw from school among students 10-15 years .............................................................79 Figure 4.1: Calories per person per day (by expenditure quintile) .............................................................................82 Figure 4.2: Self-reported versus Estimated Inadequacy in Food Consumption...........................................................85 Figure 4.3: Self assessment of food inadequacy versus estimated caloric shortfall by expenditure quintile...............85 Figure 4.4: Measures of dietary diversity in different households..............................................................................86 Figure 4.5: Diet diversity by expenditure quintile......................................................................................................86 Figure 4.6: Food group shares by region....................................................................................................................87 Figure 4.7 Share of cereals in total household caloric intake (by expenditure quintile)..............................................87 Figure 4.8: Calorie consumption and Seasonality ......................................................................................................88 Figure 4.9: Calorie consumption and Seasonality, by poverty status .........................................................................89 Figure 4.10: Maize expenditures and prices, by month..............................................................................................89 Figure 4.11: Calories from Maize versus prices, by month.........................................................................................89 Figure 4.12: Shares of households buying maize by month ........................................................................................90 Figure 4.13: Shares of households buying maize by month ........................................................................................90 Figure 4.14: Seasonal consumption patterns of cassava and green maize, by month and region...............................92 Figure 4.15: Malnutrition indices (by age group and gender) .....................................................................................95 Figure 4.16: Child malnutrition and selected characteristics of the mother.................................................................95 Figure 4.17: Child malnutrition and selected household characteristics......................................................................96 Figure 4.18: Child malnutrition and household wealth................................................................................................96 Figure 4.19: Child malnutrition and geographical distribution...................................................................................97 Figure 4.20: Malnutrition indices (by quintile of total pc exp. and urban rural divide)..............................................97 Figure 4.21: The determinants of child malnutrition (unit change effect on height-for-age Z-scores).....................102 Figure 4.22: The Determinants of calorie intake (percentage change effect) ...........................................................104 Figure 4.23: Incidence of calorie inadequacy and malnutrition by size of landholdings (by quintiles) ...................106 Figure 4.24: Share of home production in total calorie expenditure.........................................................................107 Figure 4.25: Maize production by land quintile.........................................................................................................107 Figure 4.26: Maize production by calorie quintile.....................................................................................................108 Figure 4.27: Calorie availability by maize yield quintiles.........................................................................................108 Figure 4.28: Calorie availability by land quintiles, cassava production and by regions............................................109 Figure 4.29: Malnutrition by burley tobacco adoption (total and by region).............................................................110 Figure 4.30: Stunting by extent of tobacco adoption.................................................................................................111 Figure 4.31: Stunting by tobacco adoption and profitability (total and by region)....................................................111 xii Figure 4.32: Stunting by length of tobacco adoption (total and by region) ...............................................................112 Figure 5.1: General knowledge of HIV/AIDS versus Condom Use by location ......................................................115 Figure 5.2: Condom Use by Education Level and by Age Group .............................................................................115 Figure 5.3: Life Expectancy at Birth and Crude Death Rates over time for Malawi and selected sub-Saharan African Countries....................................................................................................................................................................118 Figure 5.4: Poverty and HIV prevalence rates in Malawi..........................................................................................120 Figure 5.5: Probit model for children's school attendance (Age 6-17)......................................................................132 Figure 5.6: Probit model for children's school attendance (Ages 15-17)..................................................................133 Figure 6.1: GDP growth and changes in GDP per capita in Malawi, 1980-2005......................................................138 Figure 6.2 Maize production and GDP growth in Malawi, 1984-2005 .....................................................................140 Figure 6.3: Movements in Inflation Rate, 1992-2005................................................................................................141 Figure 6.4: Movements in Interest Rate, 1992-2003 .................................................................................................142 Figure 6.5: Movements in the Exchange Rate (Malawi Kwacha to US dollar), 1994-2005......................................143 Figure 6.6 Fiscal Balance during 1998/99 to 2004/05..............................................................................................144 Figure 6.7 Domestic Debt Stock and Interest Bill during 1998/99 to 2004/05.........................................................145 Figures 6.8 and 6.9: Major trends and composition of domestic expenditures in Malawi, 1998/99 to 2004/05 .......147 Figure 6.10: Shares of Health and Education Sectors in Recurrent Expenditure 2000/01 to 2004/05 ......................148 Figure 7.1: Contribution of the Agriculture Sector to GDP (1994-2005) and Employment (2003)..........................152 Figure 7.2: Malawi's main agricultural exports prices and earnings, 1994 to 2005 ..................................................153 Figure 7.3: Average size of households' landholdings by wealth and land quintiles ................................................155 Figure 7.4: Percentage of households selling some portion of agricultural production, by status, location and wealth quintile.......................................................................................................................................................................159 Figure 7.5: Agricultural work by gender and month .................................................................................................162 Figure 7.6: Seasonality of labor hours among rural households by land holdings ....................................................164 Figure 7.7: Household use of hired labor (percentage) and amount (man days per season), by status and location, and land quintile...............................................................................................................................................................164 Figure 7.8: Malawi National Fertilizer Consumption (1991 to 2002) and International Average Fertilizer Use (2004/05) ...................................................................................................................................................................165 Figure 7.9: Price of fertilizer in Malawi, 1990 to 2005, in nominal and real terms...................................................166 Figure 7.10: Smallholder Fertilizer Consumption (kg/ha) by Region and by Land and Expenditure Quintiles.......166 Figure 7.11: Smallholders hybrid maize yields and fertilizer use..............................................................................167 Figure 7.12: Farming Households Receiving Visits from Agricultural Field Assistants...........................................169 Figure 7.12: Concentration curve for agriculture extension services ........................................................................171 Figure 7.13: Yield trends in major smallholder crops, 1990 to 2005 ........................................................................172 Figure 7.14: Average technical efficiency of Malawian smallholder farmers (monocrop) ......................................175 Figure 7.15: Determinants of technical efficiency among Malawian smallholder farmers (OLS estimates based on the monocrop sample for hybrid maize and burley tobacco).....................................................................................179 Figure 7.16: Maize yields and land quintiles: inverse relationship?..........................................................................180 Figure 7.17: Fitted values from non parametric regression: ......................................................................................181 ln (maize yields) = F (ha under maize cultivation)....................................................................................................181 Figure 7.18: Maize Price Fluctuations: Chicago Board of Trade (CBOT), South African Futures Exchange SAFEX (RSA Spot), Malawi, and Zambia, 1996-2004 ..........................................................................................................185 Figure 7.19: Distribution of revenues from1984-2004..............................................................................................189 Figure 7.20: Tobacco prices in Malawi tobacco production in 2004.........................................................................189 Figure 8.1: Poverty and cash crops............................................................................................................................202 Figure 8.2: Tobacco growing and income gains........................................................................................................205 Figure 8.3: Cash crop growing and income gains......................................................................................................206 Figure 8.4: Determinants of participation in tobacco: land share (Tobit Marginal effects).......................................209 Figure 9.1: Number of people who benefited from WFP food aid distribution during Emergency Operations, July 2002-June 2003 .........................................................................................................................................................218 Figure 9.2: Receipt of free food by deciles and type in 2003 (percent of IHS2 households reporting)....................220 Figure 9.3: Participation in Targeted Nutrition Programs (TNP) in 2003 (percent of IHS2 households reporting)221 Figure 9.4: Proportion of Households Receiving Public Works Program (PWP) in 2003 (percent of IHS2 households reporting..................................................................................................................................................224 Figure 9.5: Proportion of Households Receiving Targeted Input Program (TIP) in 2003/04 by wealth deciles and type of household (percent of IHS2 households reporting) .......................................................................................228 xiii Figure 9.6: Farming households in 2003/04 season who received the TIP and used it for agricultural production (percent of IHS2 households reporting).....................................................................................................................229 Figure 9.7: Poverty Gap in Malawi in 2005 .............................................................................................................235 Figure 10.1: Concentration curves for enrolment in primary, secondary and tertiary education, compared to Income distribution.................................................................................................................................................................259 Figure 10.2: Concentration curves for primary and secondary education expenditure..............................................260 Figure 10.3: Concentration curves for access to health services ...............................................................................263 Figure 10.4: Concentration curves for benefit incidence of treatment.......................................................................264 BOXES Box 1.1: Key Concepts in Measuring Poverty.................................. .................................. ............................... 5 Box 1.2 Can we Compare Poverty and Inequality in Malawi to Neighboring Countries?.................................. 10 Box 1.3: Measuring Progress by Comparing the IHS1 and IHS2.................................. .................................. .. 13 Box 2.1: Malawi's population demographic projections to 2025.................................. ..................................... 22 Box 2.2: Child Labor in Malawi.................................. .................................. .................................. ................. 27 Box 2.3: Biomass Availability and poverty in Malawi.................................. .................................. .................. 37 Box 2.4: Labor Shortages Despite Underemployment? Seasonality in Time Use.................................. .......... 41 Box 2.5: Access to credit (*)................................................................................................................................ 43 Box 2.6. Subjective Versus Objective Poverty.................................. .................................. ............................. 48 Box 3.1: Key Concepts of Vulnerability, Risk and Shock .................................. .................................. ........... 57 Box 3.2: Moving Out of Poverty - Understanding Growth from the Bottom-Up.................................. ............ 59 Box 3.3: Unintended consequences of government actions that exacerbate food insecurity *........................... 66 Box 3.4 Risk Exposure and Poverty Traps.................................. .................................. .................................... 73 Box 4.1: The UNICEF Conceptual Framework of the Determinants of Nutritional Status................................ 81 Box 4.2: Key Concepts in Food Security.................................. .................................. ....................................... 83 Box 5.1: Econometric problems associated with measuring HIV/AIDS impacts.................................. ............ 121 Box 7.1: The Origins of Malawi's Dual Agricultural Sector.................................. .................................. ......... 154 Box 7.2: The Tobacco Marketing Chain in Malawi.................................. .................................. ..................... 190 Box 7.3: Can Malawi Leapfrog the Need for Agricultural Development and Proceed Directly to 193 Industrialization? *............................................................................................................................................... Box 7.4 Designing and Implementing Market-Smart Fertilizer Interventions.................................. ................ 195 Box 7.5: An Innovative Weather Insurance Program in Malawi.................................. ...................................... 197 Box 7.6: Warehouse Receipt System: Key to Improved Crop Marketing ................................... ...................... 199 Box 9.1. Selected Findings of 2005 report on implementation of Safety Nets.................................. ................. 214 Box 9.2: Ineffectiveness of Targeting in Previous Fertilizer Transfers.................................. ........................... 230 Box 9.3: Honduras Community-Based Integrated Child Care Program.................................. ......................... 233 Box 9.4: Best-Practices in Safety Net Design.................................. .................................. .............................. 234 Box 9.5: In-Kind or Cash Transfers for the Ultra-poor in Malawi? .................................. ............................... 236 Box 9.6: Is Free Fertilizer better than Food Aid? .................................. .................................. ........................ 238 Box 9.7 Targeting of Social Assistance Programs .................................. .................................. ....................... 240 Box 9.8: Innovations in Ex-Ante Risk-Management against Drought.................................. ............................ 249 Box 11.1: A Brief History of Poverty Monitoring in Malawi.................................. .................................. ........ 270 Box 11.2: Malawi Social Action Fund Monitoring & Evaluation System: Linking Global, National and 273 Local Development Objectives .................................. .................................. .................................. .................. Box 11.3: Key Benchmarks for Assessing the National Survey Program.................................. ........................ 274 Box 11.4: Adopting a Results-based Approach to Managing and Monitoring Implementation of the MGDS 277 Box 11.5: Public Expenditure Management: Getting the Basics Right.................................. ............................ 280 Box 11.6: Lessons from Experience with MTEF.................................. .................................. .......................... 281 xiv ANNEXES (ON CD-ROM) Annex 1A: Source of Data available to Measure Poverty and Living Standards in Malawi..................................CD-3 Annex 1B: Note on Construction of Expenditure Aggregate and Poverty Lines for IHS2 ...................................CD-5 Annex 1C: Evaluating Alternative Equivalence Scales........................................................................................CD-20 Annex 1D: Methodology use in preparation of the poverty maps, and Maps of Gini coefficient and Poverty Gap at Traditional Authority level ...................................................................................................................................CD-28 Annex 1E: Comparable Poverty Estimates for 1998 and 2005.............................................................................CD-47 Annex 1F: Malawi's progress towards the Millennium Development Goals (as of 2005)...................................CD-53 Annex 2A: Gender inequities in Malawi..............................................................................................................CD-87 Annex 2B: Forests, Biomass Use and Poverty in Malawi..................................................................................CD-137 Annex 2C: Distribution and breakdown of time use by location and gender ....................................................CD-178 Annex 2D: Subjective wellbeing........................................................................................................................CD-185 Annex 2E: OLS Regression on Determinants of Household Log Per Capita Expenditure................................CD-196 Annex 3A: Shocks reported by households between 1999-2004 in IHS1 complementary panel surveys..........CD-199 Annex 3B: Probit Regression on Determinants of Ultra-Poor Status of Households.........................................CD-200 Annex 3C: Characteristics of shocks and Association between specific household characteristics and likelihood of experiencing a shock ..........................................................................................................................................CD-206 Annex 3D: Temporary withdraw from school, students 10-15...........................................................................CD-216 Annex 4A: Details of the Estimation of Malnutrition Indicators and Comparison with 1998 IHS1 ..................CD-218 Annex 4B: The Foster, Greer, and Thorbecke 1984 poverty measures..............................................................CD-223 Annex 4C: Different indices of dietary diversity: definitions and a full break down of results .........................CD-225 Annex 4D: The determinants of child malnutrition and household caloric availability .....................................CD-227 Annex 5A: Prevalence of households in which a child resides without either parent (either due to being orphaned or fostered)..............................................................................................................................................................CD-233 Annex 5B: Results of the model probit on likelihood of orphans attending school ...........................................CD-237 Annex 6A: Tables on National Accounts Sectoral Composition of GDP, and Employment .............................CD-243 Annex 7A: Land Holdings by household and per capita ....................................................................................CD-246 Annex 7B: Proportion of farmers that received advice from field assistant by type of advice and by quintile..CD-247 Annex 7C: Smallholder farmers' efficiency in hybrid maize and burley tobacco production............................CD-248 Annex 7D: Weather-based Insurance .................................................................................................................CD-258 Annex 8A: Determinants of Commercialization (and adoption of tobacco production) ....................................CD-267 Annex 9A: Weather-based Insurance: National-level drought insurance...........................................................CD-281 Annex 10A: Details of the methodology for Benefit Incidence Analysis...........................................................CD-287 Annex 11A: National Data Sources for Monitoring the Millennium Development Goals.................................CD-290 xv xvi MALAWI POVERTY AND VULNERABILITY ASSESSMENT INVESTING IN OUR FUTURE INTRODUCTION 1. This study builds a profile of the status of poverty and vulnerability in Malawi. Malawi is a small land-locked country, with one of the highest population densities in Sub-Saharan Africa, and one of the lowest per capita income levels in the world. Almost 90 percent of the population lives in rural areas, where most people are engaged in smallholder, rain-fed agriculture, and therefore highly vulnerable to annual rainfall volatility. A majority of households cultivates very small landholdings, largely for subsistence. As a result, poverty is pervasive and not merely the situation of the lowest economic groups. Therefore, while this report focuses on the least-well-off sections of the population, the analysis provides valuable information to accelerate wealth creation and economic growth for the whole of Malawi. 2. This Full Report presents the detailed analysis and results of the study. Due to the length and detail of this Full Report, a Synthesis Report presenting the main findings and policy recommendations stemming from the analysis is available as a separate publication (and also includes an executive summary).1 The Full Report has three main sections. The first five chapters comprise the first part of the report, where we take an in depth look at poverty on the micro-level. Chapter One provides an overview of the income and non-income dimensions of well being in Malawi, including progress towards the Millennium Development Goals (MDGs). Chapter Two builds a profile of poverty and models the determinants of poverty at the household level. Chapter Three looks at the role of risk and vulnerability to shocks in both causing poverty, and hindering the ability of households to break free of the poverty trap. Chapter Four and Five take a more detailed look at two of the most severe and prevalent types of shocks faced by households, namely those relating to food security and the impact of chronic illness, respectively. 3. Part II of the report, comprising five chapters, focuses on the macro-level and provides policy recommendations to address some of the key findings of Part I of the report. Chapter Six briefly overviews macro-economic policy in Malawi and its bearing on economic growth, and identifies the large role of weather-related shocks in determining economic performance. Chapter Seven looks at the smallholder agriculture in detail, given the predominant role of agriculture both in household income and the economy at large. Given the sector's high degree of susceptibility to weather shocks, policies to mitigate climate shocks are explored. Chapter Eight looks at ways to boost trade as a poverty reduction strategy, focusing on the main export crop, tobacco, as an example. Chapter Nine examines the current social protection system in Malawi and recommends ways to improve the social safety nets, both to mitigate chronic poverty and as a means for breaking the poverty trap. Chapter Ten focuses on access to public services, (namely, health and education services), and looks at the distribution of public expenditure on these services across the population's income distribution. 1The complete set of reports (Synthesis Report, Full Report, and Annexes) is available on CD-ROM. 1 4. In Part III of the report, Chapter Eleven looks at the role of monitoring and evaluation systems in measuring poverty and in targeting and tracking poverty reduction efforts to maximize their effectiveness, and suggests ways to improve monitoring and evaluation in Malawi. 5. The main source of information used in this report is the new second Integrated Household Survey 2005 (IHS2), carried out by the National Statistical Office (NSO) in 2004/05, with technical support from the World Bank. This survey provides a wealth of information on household living conditions. The information has been analyzed to identify the major characteristics of poor households and the main constraints to wealth creation in Malawi. In addition, other data from the 1998 first Integrated Household Survey (IHS1 1998) is used to see how poverty and its characteristics have changed over time. The analysis has also been complemented with information from other sources, including the 1998 Population Census, the 2004 Malawi Demographic and Health Survey (MDHS 2004), as well as previous MDHS in 1992, 1996 and 2000. 2 CHAPTER 1: DIMENSIONS OF POVERTY IN MALAWI INTRODUCTION 6. Malawi is a land-locked country in southern Africa, with little arable land, high population density, and a young and rapidly growing population. The Malawian economy has been very fragile and sustained growth has been elusive. The economy remains highly dependent on agriculture which has been subject to natural calamities, increasing the vulnerability of the largely rural population. As will be discussed in this chapter, the poverty situation has not improved since the 1990s, and income inequality remains fairly high. 7. This chapter discusses income and non-income dimensions of well-being in Malawi. While the focus of the chapter is to understand the situation of the least-well-off sections of the population, the analysis provides valuable information on the overall level of wealth and well- being for the whole of Malawi. The chapter presents poverty and inequality figures, at the national level and by region, and also briefly describes progress in improving living conditions using a suite of basic human development indicators. It gives updated information on the MDGs indicators that can be derived from the household survey, and briefly discusses Malawi's progress towards the MDGs. 8. The main source of information used is the recently completed second Integrated Household Survey (IHS2) that was carried out during March 2004 to March 2005. The IHS2 data has been analyzed to produce up-to-date and time-consistent poverty and inequality figures. In addition to the 2005 IHS2, the analysis draws on data from the 1998 first Integrated Household Survey (IHS1), as well as Demographic and Health Surveys in Malawi (MDHS in 1992, 1996, 2000, and 2004) and other surveys to review the main changes since the mid-1990s (for details of data sources available, see Annex 1A). INCOME MEASURES OF POVERTY AND INEQUALITY 9. The most commonly used income measures of poverty and inequality rely on income or consumption-expenditure data collected from a sample survey of households. Consistent, high- quality data such as these have hitherto been lacking for Malawi, and the recently completed 2005 IHS2 provides the first high-quality set of data. In this section, the IHS2 dataset has been analyzed to produce up-to-date poverty and inequality figures, based on household consumption expenditures as a proxy for income (Box 1.1).2 Where possible, time-consistent comparisons are made with estimates derived from the 1998 IHS1. Income poverty and ultra-poverty in Malawi 2It should be noted that the consumption expenditure aggregate for the Dowa district has been replaced by imputed values. Preliminary calculations using the IHS2 data suggest that Dowa district is the least poor district in Malawi. According to the NSO, this observation does not match reality, and it is more likely that the IHS2 data collection for Dowa district may have been affected by enumeration problems. The values for Dowa district have therefore been substituted by an imputed value. The procedure to impute the values for the Dowa district is explained in Annex 1B. 3 10. Using the methodology outlined in Box 1.1, the poverty lines for identifying the poor and ultra- poor in Malawi are presented in Table 1.1.3 The poverty line in Malawi has been calculated at 16,165 Malawi Kwacha (MK) per person per year, or 44.3 MK per person per day.4 The line was calculated with a food component that was derived by estimating the cost of buying a sufficient amount of calories to meet a recommended daily calorie requirement. The food poverty line is 10,029 MK per person per year, or 27.5 MK per person per day. The non-food component of that total poverty line is 6,136 MK per person per year, or 16.8 MK per person per day. Following the standard methodology, the non-food component is calculated based on the non-food expenditure for those close to the food poverty line. Table 1.1: Poverty line in Malawi Kwacha per person per year Poverty line (MK per person per year) Poor MK16,165 Ultra-Poor MK10,029 Source: Our calculations based on National Statistical Office data from IHS2 11. Using this poverty line, the headcount poverty rate for the population of Malawi in 2005 is 52.4 percent (see Table 1.2). Given that Malawi's total population in 2005 is estimated at 12.3 million (see Chapter Two for details of Malawi's demographics), this implies that in 2005 about 6.4 million Malawians were living in poverty. It is also possible to calculate the portion of the population living below an ultra-poverty line. The ultra-poor are those households whose total per capita expenditure levels are below the food poverty line. In Malawi, 22.4 percent of the population lives below the ultra-poverty rate. That is, as many as 2.7 million Malawians, about one in every five people, lives in such dire poverty that they cannot even afford to meet the minimum standard for daily-recommended food requirement. Geographical variations in poverty levels 12. Malawi is divided into three main regions, North, Central and South.5 The regional rates of poverty mask a striking difference in poverty rates between urban and rural areas: poverty is predominantly a rural phenomenon. Therefore, to calculate the geographic distribution of poverty in this study, urban areas are extracted as a separate category.6 While the national poverty rate is 52 percent, there is variation across regions (Table 1.2 and Figure 1.1). The South region has the largest poverty rate (64 percent) implying that two out of three people live in poverty in the rural areas of the South. The North region has the second highest proportion of poor people (56 percent). The Central region has the lowest proportion (47 percent) of poor. A similar pattern is observed for ultra-poor people. 3The poverty line level is based on average national prices for February-March 2004. A more detailed explanation of how the consumption aggregate and poverty lines were calculated is presented in Annex 1B. 4At the time of the IHS2, MK44.3 was roughly equivalent to US$0.50. 5The North region is made up of Chitipa, Karonga, Nkhata Bay, Rumphi, and Mzimba/Mzuzu City districts. The Central Region is made up of Kasungu, Nkhotakota, Ntchisi, Dowa, Salima, Lilongwe/Lilongwe City, Mchinji, Dedza, and Ntcheu districts. The South Region is made up of Mangochi, Machinga, Zomba/Zomba City, Chiradzulu, Blantyre/Blantyre City, Mwanza, Thyolo, Mulanje, Phalombe, Chikwawa, Nsanje, and Balaka districts. 6Hence, unless otherwise specified, references in the text to `Central Region', `North Region' and `South Region' exclude the relevant urban areas. 4 BOX 1.1: KEY CONCEPTS IN MEASURING POVERTY WHAT IS POVERTY? Poverty is a multidimensional concept encompassing numerous aspects of well-being. In practice, no one indicator can capture all its dimensions. Nevertheless, measures of poverty are routinely constructed to help policy-makers and researchers understand the poor are. The poverty measure in this report is based on standards adopted in many World Bank reports. An income measure of an individual's consumption-related expenditures is compared to a cost-of-basic-needs threshold, below which a person is deemed to be poor. WHY NOT USE INCOME? While welfare is measured by income in other settings (for example, income-based welfare is standard in the United States and other developed economies), measuring income is problematic in developing economies. First, many people do not have regular income, making current income difficult to assess at any point in time. Second, income from farm activities may be hard to enumerate if households do not keep formal accounts of revenues and expenditures. Third, households are likely to intentionally under-report earnings from informal activities. In Malawi, agriculture's share of GDP has remained steady at about 40 percent, and there is a large informal sector. Income from self-employment agricultural activity accounts for a sizeable share of an average household's income. Thus, given the considerable measurement issues, income is deemed not to be a suitable standard to assess poverty in Malawi. Instead, household welfare for the Malawi poverty assessment is based on total household consumption and expenditures (including implicit expenditures on home-produced food items). HOW DO WE MEASURE POVERTY? In order to compute a poverty indicator for each individual in the household survey, it is necessary to: (a) chose a welfare indicator, and (b) compute a threshold for this welfare indicator. This threshold level of welfare that distinguishes poor households from non-poor households is the poverty line. The welfare indicator used in analysis is the total annual per capita consumption expenditure reported by a household, expressed in Malawi Kwacha deflated to February/March 2004 prices. The poverty line is a subsistence minimum based on the cost-of-basic-needs methodology. It is comprised of two parts: minimum food expenditure based on the food requirements of an individual and critical non-food consumption. Food needs are tied to the recommended daily calorie requirement. Non-food needs are estimated based on the expenditure patterns of households whose total expenditure is close to the minimum food expenditure. In this way, a poverty line and an `ultra-poverty' line can be constructed. Individuals who reside in households with consumption lower than the poverty line are labeled "poor". The "ultra-poor" live in households whose total consumption per capita on both food and non-food items is lower than the subsistence minimum food expenditure. In other words, the ultra-poverty line is the same as the food poverty line. The baseline poverty threshold in this report estimates the cost of a minimum basket, which is calculated using the WHO recommended calorie requirements for moderate activity levels as described in Annex 1B. These calorie requirements were applied to the IHS2 sample to yield a median calorie requirement of 2,400 calories per day per person. Choosing a more generous minimum food basket (in terms of more expensive calories) will result in a higher poverty line and higher poverty rate; likewise a higher calorie standard will increase the poverty threshold since more calories cost more. HOW DOES OUR MEASURE OF POVERTY ACCOUNT FOR HOUSEHOLD SIZE AND COMPOSITION? To accurately identify households whose total per capita consumption on food and non-food items is lower than the minimum threshold, we need to consider that there may be economies of scale in expenditure. For example, a 2-person household does not double expenditures on housing, utilities or other non-food items for which expenditure can be shared (these are `public' goods whose costs do not vary whether one or more persons use them). Larger households might also buy items in bulk, which can mean lower prices or discounts. In addition, the age structure of household members is also considered. A small child is assumed not to be equivalent to an adult in terms of consumption needs. Therefore, household needs are computed based on the use of equivalence scales which account for the different size and composition of households. The choice of equivalence scale reflects judgments about differences in needs. Adjusting for household size and composition can be done in numerous ways, and there is not a clear dominant choice. Rather, it is important to ascertain that the general profile is robust to choice of scale. Details of the equivalence scale used in this study are provided in Annex 1C. HOW DO WE CHOOSE THE POVERTY LINE? ABSOLUTE VS. RELATIVE POVERTY MEASURES By assessing the poverty line using a cost-of-basic needs approach, we will have a poverty line that indicates an absolute measure of poverty. An alternative approach measures relative poverty within a country, using a poverty line based on the distribution of a welfare measure, such as 60 percent of median income (the standard in Western Europe). In low-income economies, however, a relative measure is not a useful indicator of the fraction of the population unable to meet minimum living standards. Moreover, relative poverty measures do not provide a clear indication of trends in poverty over time, as they also reflect distributional changes. 5 13. About 25 percent of the population in urban areas is living in poverty, compared to 56 percent of the rural population. That is, a person in a rural area is more than twice as likely to be poor. The difference is more dramatic among the ultra-poor. Overall, 22.4 percent of the population is ultra-poor. Of this group, 24.2 percent of the rural population is ultra-poor while only 7.5 percent of the urban population is ultra-poor (Table 1.2 and Figure 1.1). 14. It is important to note that distribution of the population is slightly different from that of the poor. In terms of population distribution, the Southern rural area has 40 percent, the Central rural has 38 percent, and the Northern rural has 10 percent while the urban areas contribute 11 percent. Hence, the Southern rural areas have a disproportionate share of the poor, reflecting both the higher population and the higher poverty rate in this region. The urban areas are contributing only 6 percent of all the poor people in the country. Table 1.2: Poverty headcount and distribution of Malawi's poor by place of residence in 2005 Poverty Percent of Percent of Malawi's headcount Ultra-Poor Malawi's poor population (%) (%) Malawi 52.4 22.4 100.0 100.0 North rural region 56.3 25.9 10.9 10.2 Central rural region 46.7 16.2 33.9 38.1 South rural region 64.4 31.5 49.7 40.4 Urban 25.4 7.5 5.5 11.3 Notes: `Central region', `North region' and `South region' exclude the relevant urban areas. Malawi's population in 2005 is estimated at 12.4 million (Source: National Statistical Office) Source: Our calculations based on NSO data from IHS2 Figure 1.1: Proportion of poor and ultra-poor persons by region in 2005 70 Southern rural 60 region 50 Central rural 40 region 30 20 Northern rural region 10 0 Urban Poor Ultra Poor Note: The solid line indicates the national poverty level. Source: Our calculations based on NSO data from IHS2 15. The poverty maps in Figures 1.2 and 1.3 provide a breakdown of the headcount poverty rates at the Traditional Authority administrative level, to give a detailed overview of where the pockets of deepest poverty are in the country. The idea of a poverty map is to combine census information and household survey information to estimate poverty and inequality in greater 6 geographical detail than the survey is able to alone.7 In this study, we have created a poverty map using the 2005 IHS2 survey and the 1998 census.8 16. The poverty map shows that the highest levels of poverty are concentrated in the southernmost and northernmost areas of the country. The central region displays consistently lower rates of poverty (except for two very isolated and small pockets of poverty). The Northern region exhibits the highest variation in poverty rates, including both some of the areas with the highest concentration of poverty as well as some of the relatively better off areas. Urban areas also exhibit substantial of variation. 17. Headcount poverty rates indicate the share of the population below a minimum income level (the poverty line), but they don't reveal any information about the distribution of income above and below that threshold. Inequality measures, instead, consider the entire distribution of income levels. Figure 1.4 shows the distribution of income across Malawi's population. It shows the extreme disparity between incomes of the richest and poorest in the country, with the distribution skewed towards the lower end of the scale. The graph also shows that the population is concentrated around the poverty line, such that only few Malawians have incomes which are more than double the poverty line. Figure 1.2: Map of poverty headcount in urban areas Source: National Statistical Office, IHS2 2005 and 1998 Census 7The first poverty maps of Malawi were prepared by IFPRI based on the 1998 IHS1 data and the 1998 Population and Housing Census. The methodology for this map generally follows previous work in Malawi, Madagascar and other countries. The basic steps are described in Annex 1D. For additional information and description of the methodology see Mistiaen et al. (2002), Demombynes et al. (2002) and Elbers et al. (2002). For information on the first Malawi poverty map see Malawi Social Atlas and background papers by Benson et al. (2002). 8The predictions of poverty in this poverty map build on the same census data as used for the 1998 map (a new census will not be conducted until 2008), but use the more recent IHS2 household survey data. As economic growth has been quite moderate since 1998, it is reasonable to assume that geographical picture of poverty has not changed much. However, since the IHS2 2005 survey offers much improved data quality, the new poverty map takes advantage of the better data. 7 Figure 1.3: Map of poverty headcount at Traditional Authority level Source: National Statistical Office, IHS2 2005 and 1998 Census Income Inequality in Malawi 18. Figures 1.5 and 1.6 also display the extent of inequality. Figure 1.5 plots the median level of (expenditure as a proxy for) income per capita for each of the deciles of the population. As shown, the richest 10 percent of the population has a median per capita income that is eight times higher (MK50,373 per person per annum) than the median per capita income of the poorest 10 8 percent (MK6,370 per person per annum). Moreover, the richest 10 percent of the population has a median income that is three times higher than the overall median income in the country. 19. Figure 1.6 displays the Lorenz Curve in 2005 which shows the share of income (again proxied by expenditures per capita) associated with a given share of the population. The diagonal line in the graph represents perfect equality, since it indicates that any percentage of the population would receive the same percentage of total income. The curved line below the diagonal represents how far the population is from perfect equality. The closer the curved line is to the diagonal, the more equal the distribution is. The Lorenz curve shows that while the bottom 50 percent of the population accounts for only 25 percent of total income, the richest 5 percent accounts for 20 percent of the total income. Figure 1.4: Distribution of per capita Figure 1.5: Per capita consumption consumption expenditure in 2005 expenditure by decile in 2005 Per Capita Consumption Expenditure 60,000 .1 Richest Ultra-Poverty Line 50,000 8 .0 Poverty Line noital raeYreP 40,000 puoP 6 .0 30,000 of nosr noit 4 or .0 Per 20,000 oprP Pe K 2 M .0 10,000 Poorest 0 0 0 20000 40000 60000 80000 100000 1 2 3 4 5 6 7 8 9 10 MK Per Person Per Year Population fromPoor to Rich (by decile) Source: National Statistical Office, IHS2 Source: National Statistical Office, IHS2 Figure 1.6: Lorenz Curve in 2005 Table 1.3: Gini coefficient in 1998 and 2005 atpi Lorenz Curve for Malawi 2004-05 1998 2005 1 car 0.39 0.39 pe Malawi (0.005) (0.004) resutidnep .8 Urban 0.44 0.48 .6 (0.012) (0.009) exfo 0.33 0.34 onirtop .4 Overall Rural (0.005) (0.004) pro .2 0.36 0.34 veitalu North (0.021) (0.001) muC 0 0.31 0.32 0 .2 .4 .6 .8 1 Central Cumulative proportion of population (0.006) (0.004) Lorenz curve Line of Perfect Equality South 0.33 0.35 (0.008) (0.014) Source: National Statistical Office, IHS2 Source: National Statistical Office, IHS2 9 20. The Gini coefficient is a standard measure of the amount of inequality, and is based on the mathematical measure of the Lorenz curve. In general, the Gini coefficient can take a value from 0 (perfect equality) to 1 (perfect inequality). Table 1.3 shows the levels of Gini coefficient across the country. The extent of inequality does not differ much across rural areas but is substantially higher in urban areas. A detailed map of the Gini coefficient across Malawi is provided in Annex 1D. BOX 1.2 CAN WE COMPARE POVERTY AND INEQUALITY IN MALAWI TO NEIGHBORING COUNTRIES? International comparisons of poverty and inequality rates across countries cannot be made easily. In fact the calculation of household `consumption-related expenditures' will not follow the same identical definitions across countries, and will depend on a variety of factors (including the length of the recall period over which consumption is recorded, the degree of disaggregation of consumption items, the methods for imputation of housing and durables consumption), and also because countries set different subsistence minimum standards, and use different methodologies, to calculate national poverty estimates. Additional factors confounding comparability include differences in survey nonresponse rates across countries (see Korinek, Mistiaen, and Ravallion 2005). Differences across countries in the availability of spatial price indexes can also affect conclusions. Across countries there tends to be little uniformity in whether, and how, spatial price variation is accommodated. Without a concerted effort to harmonize data collection across countries, it is unlikely that such global databases can be relied on to provide more than a tentative picture of differences in poverty and inequality across countries.* Cross-country datasets on monetary poverty and inequality generally incorporate some attempts to improve comparability, but they typically fall far short of achieving strict comparability. For instance, poverty comparison based on a common `subsistence minimum standard' have been calculated by using a fixed poverty line, notably the well-known "$1 per day" estimates. In this approach, one dollar is converted into local currency units using the purchasing power parity (PPP) conversion factor. PPP is a form of exchange rate that takes into account the cost and affordability of common items in different countries. This conversion is defined as the number of units of a country's currency required to purchase a standard basket of goods and services collected in all countries. In this study, we calculate the "$1 per day" poverty line. Specifically, the 1993 PPP conversion factor (1.5221) was updated using Malawi CPI inflation rates from 1993 to 2004. The results indicate that in 2004, one US dollar was equivalent in terms of purchasing power to 28.13 Malawi Kwacha. This translates to a "$1 per day" poverty line of MK 11,051 per person per year. In 2005, the portion of the population living below this poverty line was 28 percent, which puts Malawi in the middle of the range of values for neighboring sub-Saharan countries. As discussed in the previous paragraph, however, it should be emphasized that this calculation is tentative, and falls short of achieving strict comparability. * Source: Adapted from Box 2.5 in World Bank (2005). World Development Report 2006: Equity and Development. World Bank, Washington D.C Depth of poverty measures 21. The poverty headcount measures the number of people below the poverty line, but does not measure the distance from the poverty line. The poverty gap shows how far below the poverty line households are found, on average, expressed as a percentage of the poverty line. Those households that are close to the poverty line could be improved out of poverty with less effort than those that are far below the line. In 2005, the poverty gap was 17.8 percent overall and 5.3 percent for the ultra-poor (Table 1.4). In other words, the poor, on average, subsist on 17.8 percent less than the MK16,165 poverty line, and the ultra-poor, on average, survive on 5.3 10 percent less than the MK10,029 ultra-poverty line. A detailed map of the poverty gap is found in Annex 1D. It shows that the poor are much poorer in the northernmost and southernmost areas of the country, while they tend to be relatively closer to the poverty line in the central region. Table 1.4: Poverty Headcount, Income Gap, and Severity of Poverty estimates in 1998 and 2005 1 1998 2005 Headcount Gap Severity Headcount Gap Severity Malawi Malawi Poor 54.1 18.6 8.5 52.4 17.8 8.0 (standard_error)2 (0.8) (0.5) (0.3) (1.0) (0.5) (0.3) Ultra-poor 23.6 5.7 2.0 22.4 5.3 1.8 (standard_error) (0.9) (0.3) (0.1) (0.9) (0.3) (0.1) By Region By Region Poor Poor Urban 18.5 4.8 1.8 25.4 7.1 2.8 (standard_error) (1.7) (0.6) (0.3) (2.8) (1.0) (0.5) Rural overall 58.1 20.2 9.2 55.9 19.2 8.6 (standard_error) (0.9) (0.5) (0.3) (1.0) (0.5) (0.3) North 56.3 19.5 8.9 56.3 19.6 8.8 (standard_error) (3.7) (2.0) (1.2) (2.7) (1.4) (0.8) Central 47.6 14.4 6.0 46.7 14.1 5.9 (standard_error) (1.7) (0.8) (0.4) (1.6) (0.6) (0.3) South 68.4 25.7 12.3 64.4 23.8 11.2 (standard_error) (1.4) (0.9) (0.6) (1.5) (0.8) (0.5) Ultra-Poor Ultra-Poor Urban 4.9 1.1 0.5 7.5 1.6 0.5 (standard_error) (1.0) (0.3) (0.2) (1.4) (0.3) (0.1) Rural overall 25.7 6.2 2.2 24.2 5.8 2.0 (standard_error) (0.9) (0.3) (0.1) (0.9) (0.3) (0.1) North 24.9 6.0 2.1 25.9 5.9 1.9 (standard_error) (3.7) (1.2) (0.5) (2.4) (0.7) (0.3) Central 16.3 3.5 3.2 16.1 3.5 1.1 (standard_error) (1.4) (0.4) (0.2) (1.0) (0.3) (0.1) South 34.6 8.9 1.2 31.5 7.9 2.8 (standard_error) (1.6) (0.6) (0.3) (1.5) (0.5) (0.2) Notes: 1. These estimates have been calculated using IHS1 and IHS2 data. Poverty measures from the 1998 IHS1 were calculated in 2000 and the estimate of the poverty headcount was 65.3 percent (National Economic Council, 2000). In this study the 1998 estimates (from IHS1) have been recalculated because the survey instruments & methods of calculating the poverty rates have been revised and improved to meet local and international standards (see Boxes 1.1, Box 1.2, and 1.3). In order to ensure comparability with the 2005 IHS2 estimates there we have recalculated the 1998 IHS1 estimates (using the same revised methodology as for the 2005 IHS2). 2. The statistical precision of the estimates is measured by the `standard errors', which are shown in parenthesis. A `confidence interval' which gives a 95 percent probability of including the real value can be calculated around each estimated value as follows: The lower bound is equal to the estimated value minus (1.9644 * standard error) and the upper bound is equal to estimated value plus (1.9644 * standard error). The resulting confidence interval around the estimate provides a range which is likely to include the real values at 95 percent probability. It is also worth noting that in comparing the 1998 and 2005 estimates, if the confidence intervals around the two estimates (for 1998 and 2005, respectively) are overlapping, then there is a possibility that there may have been no change over the period. Source: National Statistical Office, IHS1 and IHS2 11 22. The severity figure is a more sophisticated, weighted measure of poverty. It takes into account the income gap and the inequality amongst the poor, whereby a dollar of income gap for the extreme poor is given more weight than a dollar of income gap for those who are just under the poverty line. As a result, the index increases both with respect to the income gap and with respect to the existence of extreme poverty. Unfortunately there is no simple interpretation of the severity measure, beyond the fact that the lower the measure the better. The poverty severity in Malawi in 2005 is 8 on average, with large regional differences ranging from 11.2 in the South region to 5.9 in the Central region, again confirming that the South holds the highest number of poor and ultra-poor (Table 1.4). Severity of poverty is much lower in the urban areas, confirming that poverty in urban areas is not as extreme as in rural areas. EVOLUTION OF POVERTY AND INEQUALITY BETWEEN 1998 AND 2005 23. As highlighted above, the IHS2 estimate of the poverty rate in 2005 is 52.4 percent. It should be emphasized that this rate should not be directly compared to the 65.3 percent estimate from the 1998 IHS1 (National Economic Council, 2000). This is because the survey instruments & methods of calculating the poverty rates have been revised and improved to meet local and international standards (see Boxes 1.1, 1.2, and 1.3). Despite this change in survey techniques, an effort was put in place to compute the poverty rates for the previous IHS using the current methodology. In this exercise, poverty estimates from IHS1 were recalculated using regression models to impute expenditure per capita based on comparably measured household characteristics (see the Annex 1D for details of the methodology applied). The poverty rates calculated from IHS1 using this methodology can be compared directly to the poverty rate calculated from the IHS2. The results are shown in Figure 1.7, Table 1.3 and Table 1.4. Figure 1.7: Proportion of the population deemed poor and ultra-poor in 1998 and 2005 Comparison of poverty and ultra-poverty in 1998 and 2005 ( with 95% confidence intervals ) 70 60 54.1 52.4 oni at 50 popul 40 of 30 23.6 22.4 entc 20 erP 10 0 1998 2005 1998 2005 Poor Ultra-Poor Notes: 1. Estimates for 1998 have been recalculated using the same methodology as for 2005 to allow comparisons across the two datasets (and these estimates cannot be compared to the estimates for 1998 which were presented by National Economic Council in 2000). 2. The `confidence intervals' around each estimate has been calculated to give a 95 percent probability of including the real value. Source: National Statistical Office, IHS1 (1998) and IHS2 (2005). 12 24. The overall poverty rate remained about the same between 1998 and 2005. About 54.1 percent of the population was deemed poor in 1998 while the rate is at 52.4 percent in 2005. The slight decrease is not statistically significant.9 The distribution of poverty has also not changed much. Rural poverty was much higher than urban poverty already in 1998. The South region had the highest rate of poverty followed by the North and Central regions.10 25. There have been some movements in relative levels of poverty, however. Urban poverty has increased from 18 percent in 1998 to 25 percent in 2005.11 This increase is more than offset by a decrease in rural poverty in the South from 68 to 64 percent. Hence while rural areas remain disproportionally poorer than urban areas, urban poverty has been rapidly increasing. Also, the gap between the South and the other rural regions is diminishing. These changes are BOX 1.3: MEASURING PROGRESS BY COMPARING THE IHS1 AND IHS2 CAN WE COMPARE IHS1 AND IHS2? The IHS1 and IHS2 surveys, while similar in many respects, are based on two different methodologies, and some of the questions included in the surveys are not identical. While the IHS2 survey was designed, in part, to have sections that would be directly comparable to the IHS1, the major difference between the two surveys was the method used to collect food consumption data. In IHS1, a diary was used to collect information over 14 days, whereas in IHS2, recall questions covering food consumed during the past week were used. This and other improvements in the design of the IHS2 make it difficult to calculate consistent estimates of household welfare and poverty over time.(*) Nevertheless, as described below, considerable effort has been made to develop comparable indicators, as well as to re-estimate Malawi's poverty lines such that they can be compared. WHAT ARE THE KEY STEPS IN THE METHODOLOGY? Because of the differences in the food consumption data collected, we could not simply take the poverty measure developed from the IHS1 data and compare this with the poverty measure calculated in the IHS2. Instead, we took the data from the IHS1 survey and recalculated an estimate of 1998 poverty, which we then compared to our IHS2 results. The steps involved in computing a comparable IHS1 poverty estimate are: (1) estimate per capita expenditure for IHS1 households based on a regression model of per capita expenditure developed from IHS2 using a set of household characteristics measured in both surveys, and (2) estimate poverty rates for households using the imputed per capita expenditure, and applying the IHS2 poverty lines. The main assumption imbedded in this approach is that the correlation between poverty and the set of household characteristics has not changed significantly over time. (A second important assumption is that the heteroskedastic process for the error term also remains the same across the two years.) The approach used follows recently-developed statistical techniques that originated in Elbers et. al. (2002, 2003), and has since been widely applied in different countries, in particular for poverty maps, but also for survey-to- survey imputations as used here. See, for example, Kijima and Lanjouw (2003), Luoto (2005). Additional details of the methodology adopted are provided in Annex 1F. This method allows the calculation of a complete set of comparable poverty indicators ­ headcount, gap, severity, and Gini Index. It should be emphasized again, however, that the results reported here for the IHS1 will not be the same as results reported in earlier reports, because of the change in methodology. (*) For more information on the differences in the questionnaires used in the IHS1 and the IHS2, see the Basic Information Document for the IHS2. This is available from the Malawi NSO (http://www.nso.malawi.net/). 9A `confidence interval' which gives a 95 percent probability of including the real value can be calculated around each estimate by using the `standard errors' shown in Table 1.4. Since the confidence intervals around the 1998 and 2005 estimates overlap, we conclude that there is a significant statistical probability that there may have been no change over the period between the two surveys. 10As discussed above, it is important to note that distribution of the population is slightly different from that of the poor. In both 1998 and 2005 the South region has a disproportionately high share of the poor reflecting the higher poverty rate in this region while the Central region has a slightly lower poverty rate than its population share. 11It should be noted that the two confidence intervals (for 1998 and 2005 urban poverty headcount estimates, respectively) marginally overlap, indicating that there is less than 95 percent confidence that the increase is statistically true. 13 consistent with the anecdotal information about a substantial increase in migration from rural areas into urban areas. 26. Similar patterns can be observed when comparing ultra-poverty, as well as changes in the poverty gap, the severity measure and the Gini coefficient. At the national level changes between 1998 and 2005 are not statistically significant. Ultra-poverty, poverty gaps and severity have increased in urban areas, and this increase is more than offset by decreases in the South (Table 1.4). Little has changed in the Central and North regions of the country. Inequality has also not changed much (Table 1.3). Urban inequality has increased, reflecting the increased proportion of poor people in urban areas. EVOLUTION OF NON-INCOME DIMENSIONS OF POVERTY 27. The previous sections have highlighted that income measures of poverty have not changed significantly over the past decade. This section briefly highlights recent trends in living conditions of Malawi's population using a suite of non-income human development indicators, such as literacy, school enrollment, malnutrition, infant mortality, and maternal mortality. Table 1.5: Malawi, key human development indicators, 2003 Least Sub- Developed Saharan Malawi1 Countries Africa Human poverty index (HDI):2 rank out of 177 countries 165 .. .. Adult literacy rate (% ages 15 and above)3 64.1 53.65 60.5 Combined gross enrolment ratio for primary, secondary, tertiary4 72 456 50 Births attended by skilled health personnel (%) 57 345 415 Physicians (per 100,000 people) 7 1.1 .. .. Life expectancy at birth (years)7 37.5 52 46 Under-five mortality rate (per 1000 live births) 133 156 179 Infant mortality rate (per 1000 live births)7 76 97 104 One-year-olds fully immunized against tuberculosis (%) 91 79 75 One-year-olds fully immunized against measles (%) 79 67 62 Maternal mortality ratio adjusted (per 100,000 live births) 984 .. .. Children underweight for age (% under age 5) 22 .. .. Children under height for age (% under age 5) 48 .. .. Notes: 1. Data for Malawi refer to the most recent year available between 2000 and 2005. Due to differences in methodology and timeliness of underlying data, comparisons with other countries should be made with caution. 2. A composite index measuring average achievement in three basic dimensions of human development: a long and healthy life, knowledge and a decent standard of living. 3. Data refer to national estimates from censuses or surveys between 2000 and 2004, unless otherwise noted. 4. Data refer to the school year 2002/03, unless otherwise noted. 5. Data refer to a year between 1995 and 1999. 6. Preliminary UNESCO Institute for Statistics estimate, subject to further revision. 7. Data refer to the most recent year available during the period 1995 to 2005 Source: Human Development Reports, and Malawi DHS various reports. 28. Malawi's current status in terms of key human development indicators is summarized in Table 1.5. In general, Malawi scores above the average for Africa in terms of education (adult literacy and gross enrolment ratios). In most health indicators, however, Malawi scores well below the average for Africa, notably with low life expectancy, low child immunization rates, 14 and exceedingly high maternal mortality. The current status of these human indicators is given greater scrutiny in the poverty profile built in Chapter Two. 29. The Human Development Index (HDI) is a widely used multi-dimensional summary indicator of development.12 The HDI is a comparative measure of poverty, literacy, education, life expectancy, childbirth, and other factors for countries worldwide. It is a standard means of measuring well-being, especially child welfare. The HDI measures the average achievements in a country in three basic dimensions of human development: A decent standard of living, as measured by gross domestic product (GDP) per capita at purchasing power parity (PPP) in USD. Knowledge, as measured by the adult literacy rate (with two-thirds weight) and the combined primary, secondary, and tertiary gross enrolment ratio (with one-third weight). A long and healthy life, as measured by life expectancy at birth. 30. Malawi's Human Development Index has stagnated during the past decade. Each year, UN member states are listed and ranked according to these measures. In 2005, Malawi ranked 165 out of 177 countries, reflecting the extremely low achievement in the three basic dimensions of human development measured by the HDI. The HDI value in Malawi increased steadily between 1975 and 1995, but has stagnated since 1995 (Figure 1.8). Unfortunately, the pattern in education and health indicators mirrors the absence of any changes in income measures of poverty during the last decade, as discussed in previous sections. Figure 1.8: Malawi Human Development Index, 1975-2005 0.50 180 170 xednItnempolev )$SU 0.45 160 0002 150 0.40 antt 140 130 Denam 0.35 onsc(at 120 apic Human Development Index 110 Hu 0.30 per GDPper capita (constant 2000 US$) 100 PD 0.25 90 G 1975 1980 1985 1990 1995 2000 2005 Source: UNDP statistics, available online at http://hdr.undp.org/statistics/data/countries.cfm?c=MWI Knowledge: changes in education indicators for Malawi 31. Over the last decade there has been a significant improvement in the proportion of Malawians who have received some formal education (World Bank 2004). According to the 1992 Malawi Demographic and Health Survey (MDHS),13 the proportion of men and women 12Notably, the index has been used since 1993 by the UNDP in its annual report. 13A total of 3 DHS have been carried out in Malawi in 1992, 2000 and 2004. The Malawi DHS are part of the world-wide MEASURE/Demographic and Health Surveys (DHS) Program, funded by the United States Agency for 15 who had never attended school was 21 and 47 percent, respectively. In the 2004 MDHS, these percentages had dropped to 12 and 23 percent, respectively. Differences by sex remain noteworthy, however, and are exacerbated when looking at secondary school. In the 2004 MDHS, while 26 percent of men had attended secondary school, the corresponding proportion for women was only 16 percent. 32. In line with the trends in education, literacy among adults has been increasing (DHS EdData, 2002). In 1992, 44 percent of women were literate (ability to read), compared to 56 percent in 2000. For men, literacy increased from 75 percent in 1990 to 79 percent in 2000. A long and healthy life: changes in health indicators for Malawi 33. Life expectancy has been decreasing sharply during the last 20 years, from 46 years in 1987 to around 37 years in 2002 (Figure 1.9). As discussed in Chapter Five this trend is common to many Sub-Saharan African countries as a result of the spread of the HIV/AIDS epidemic. There is little difference between the life expectancy for males and females in Malawi. Figure 1.9: Life expectancy at birth Figure 1.10: Physicians per 100,000 in Malawi, 1967-2002 people in Malawi, 1970-2003 50 e) 3.0 48 opl pe )s 2.5 46 eary(yc 44 0,000 2.0 42 10 1.5 ant 40 pecxeefLi 38 per( 1.0 36 34 ansicis 0.5 32 0.0 30 hyP 1967 1972 1977 1982 1987 1992 1997 2002 1970 1981 1990 1993 2003 Source: UNDP statistics Source: UNDP statistics 34. The number of doctors per 100,000 people remains extremely low, and has been decreasing rapidly during the past decade (Figure 1.10). This reduction is the result of both the loss of skilled personnel due to the HIV/AIDS and the brain drain to developed countries as a result of the wage differentials. In order to stop this trend, the remuneration package for physicians (and nurses) has recently been improved by the Government, with support by the donor community. 35. The availability of medical assistance at delivery helps to lower the risk of adverse pregnancy outcomes, including lowered rates of maternal morbidity, maternal mortality, and perinatal mortality. However, the percentage of births assisted by a doctor or nurse/midwife has not changed much since the early 1990s. The proportion of births that were assisted by a doctor International Development (USAID). The program is designed to collect data among others on fertility, family planning and maternal and child health. 16 or nurse/midwife at delivery in the 2004 MDHS is 57 percent. This is about the same level observed in the 1992 MDHS (55 percent) and the 2000 MDHS (56 percent). 36. Over the course of the 1990s Malawi's maternal mortality ratio (MMR) doubled to one of the highest in the world. According to the 2000 MDHS, the MMR reached 1,120 per 100,000 live births, nearly double the MMR of 620 per 100,000 live births estimated from the 1992 MDHS.14 This decline has been attributed to several factors, including poor health care, health systems deficiencies, poor access to care and harmful patient-related behavior (see inter alia, McCoy et al., 2004). Hence, by the year 2000 Malawi faced one of the highest maternal mortality rates in the world. Since the year 2000 the situation appears to have improved, however. The 2004 MDHS indicates that the maternal mortality has declined to 984 per 100,000 live births in 2004. 37. During the 1980s there was very little change in childhood mortality. During the 1990s, however, a gradual decrease in infant and under-five mortality is observed, which appears to have accelerated in the most recent five-year period.15 This is true for all measures, but most importantly during the first month of life. As shown in Figure 1.11, the infant mortality rate (IMR) has decreased from 135 deaths per 1,000 live births in 1988-1992 to 76 deaths in 2000- 2004. Similarly, the under-five mortality rate (U-5) has decreased from 234 deaths per 1,000 live births in 1988-1992 to 133 deaths in 2000-2004. Figure 1.11. Trends in Infant and Under-five Mortality, 1992 to 2004 300 hstrbievil 250 200 1,000 150 per 100 50 eathsD 0 1980 1985 1990 1995 2000 U-5 1992 MDHS U-5 2000 MDHS U-5 2004 MDHS IMR 1992 MDHS IMR 2000 MDHS IMR 2004 MDHS Sources: 1992, 2000, and 2004 Demographic and Health Surveys. 38. The 2004 MDHS data show that 64 percent of children age 12-23 months has received the full series of recommended vaccinations, a decrease from 1992 and 2000 levels (82 and 70 14The United Nations Children's Fund (UNICEF), World Health Organization (WHO) and United Nations Population Fund (UNFPA) provide an even higher estimate of maternal mortality in Malawi (adjusted for well- documented problems of underreporting and misclassifications) at 1,800 deaths for 100,000 live births in the year 2000. 15The recent decline in childhood mortality has also been observed in neighboring countries (NSO and ORC Macro 2005). 17 percent, respectively).16 This is true for all types of vaccines. For example, BCG (vaccination against tuberculosis) coverage has decreased from 97 percent in 1992, to 92 percent in 2000, to 91 percent in 2004. 39. Nutritional status is an important human development indicator as it allows evaluation of the susceptibility of the population to disease, impaired mental development, and early death. In the MDHS surveys, the height and weight of children under age five were measured in order to estimate their nutritional status.17 Children's nutritional status in the 2004 MDHS is virtually identical to the status in 1992 MDHS and 2000 MDHS, indicating that there has been no improvement in the nutritional status of children under age five since 1992. As discussed in detail in Chapter Four, as many as 48 percent of children under five years of age in Malawi are stunted (too short for their age), and 22 percent are severely stunted. Five percent of children are wasted (or too thin), and 22 percent are underweight. These numbers are extremely high even for Sub-Saharan Africa, and underscore that child malnutrition remains one of the biggest development challenges facing Malawi. MALAWI'S PROGRESS TOWARDS THE ACHIEVEMENT OF MILLENNIUM DEVELOPMENT GOALS 40. In this section we briefly discuss Malawi's progress towards the MDGs as of end-2005.18 Achievement of the targets for these goals will be monitored through a set of 48 indicators.19 41. In Malawi, the MDGs are to be achieved through implementation of the Malawi Poverty Reduction Strategy (MPRS, 2002-2005) and the Malawi Growth and Development Strategy (MGDS, 2006-2010), which express the country's overarching economic and development targets for poverty reduction. The overall monitoring of the MDGs is expected to be aligned under the Monitoring & Evaluation Master Plan launched in November 2004, which lays the basis for the monitoring of the MPRS and MDGS (see Chapter Eleven). 16 The World Health Organization guidelines for childhood immunizations call for all children to receive: a BCG vaccination against tuberculosis; three doses of the DPT vaccine to prevent diphtheria, pertussis and tetanus; three doses of polio vaccine (not considering polio given at birth); and a measles vaccination. 17 Three standard indicators of growth are used in this report. A child is considered stunted if he is too short for his age. Stunting indicates chronic under-nutrition, typically due to poor nutrition over an extended period. A child is considered wasted if s/he is too thin, i.e., weighs too little for his height. Wasting is an indicator of acute or recent nutritional deficits and is closely tied to mortality risk. Finally, a child is considered underweight if s/he weighs too little for his age. A child can be underweight for his age because s/he is stunted, wasted, or both. To allow standardized measurements over time and in different settings, height and weight data are routinely compared to a reference population. The World Health Organization (WHO) recommends using the child population data maintained by the NCHS (U.S. National Center for Health Statistics) as the reference. The status of a child with regard to stunting, wasting, and underweight is determined by how many statistical units, standard deviations, the child's measurements are below the mean of the NCHS reference population. If a child is between 2 and 3 standard deviations below the mean, the child is considered moderately malnourished (stunted, wasted, or underweight); if the child is 3 or more standard deviations below the mean, the child is considered severely malnourished. 18 Malawi is one of the countries that signed the Millennium Declaration that was adopted in September 2000. The Millennium Development Goals (MDGs) commit countries to an expanded vision of development that promotes human development as key to sustaining social and economic progress, and recognizes the importance of creating a global partnership. The Declaration outlines eight (8) goals and eighteen (18) targets to be achieved by the year 2015. Achievement of these targets will be monitored through a set of 48 indicators. 19 For a full list of goals, targets and indicators the reader is referred to: http://www.un.org/millenniumgoals/ 18 42. Malawi has completed one report on the progress towards the MDGs in 2003.20 The report highlighted that Malawi was falling short in a number of ways towards reducing poverty and advancing other human developments. The work done in 2003 has been updated using results from MDHS 2004 and the IHS2 2005, and detailed results are presented in Annex 1F. The update is not comprehensive, in that it does not attempt to analyze the progress on every indicator. Only those indicators for which new data is available have been covered. A summary of progress towards the MDGs is provided in Table 1.6. Below we briefly discuss progress towards the MDGs 1 to 7. 43. Malawi is unlikely to meet Goal 1 (to eradicate extreme poverty and hunger). In fact, as discussed in this chapter, little progress has been made in reducing poverty levels and ultra- poverty over the past decade. Similarly, progress towards Goal 2 (to achieve universal primary education) has been limited. While Malawi has improved its GER ratios in primary education, the NER remains at around 80 percent. More important only 60 of children who start school actually complete a full course of primary education (see Chapter Two). Little progress has also been made in achieving Goal 5 (to improve maternal health). As discussed above, maternal mortality actually increased between 1992 and 2000, but has recently begun to diminish. The current rate of reduction is not rapid enough to meet this MDG by 2015, however. 44. Malawi could reach Goal 3 (to promote gender equality and empower women). Notably, good progress has been made in reaching equality of enrollment in primary education and in reducing gender disparity in youth literacy. More progress is needed in reducing the gender gap in higher education, however, and also in increasing women participation in the workforce and in position of authority. Similarly, good progress has also been made towards achieving Goal 4 (to reduce child mortality), with under-five mortality projected to decrease by more than two-thirds, between 1990 and 2015. 45. Some progress has been made in reaching Goal 6 (to combat HIV/AIDS, malaria and other diseases). Notably, Malawi appears to have halted and begun to reverse the spread of HIV/AIDS, and has begun to reverse the incidence of malaria, by increasing the proportion of population using effective malaria prevention. Finally, mixed progress has been made in achieving Goal 7 (Ensure environmental sustainability). Deforestation is continuing at an alarming rate, and the proportion of population using solid fuels remains very high. On a more positive note, the proportion of people who have access to safe drinking water and improved sanitation has increased significantly. 46. In sum, Malawi is well placed to achieve three of the MDGs by 2015, provided additional progress is made. Achieving the other MDGs by 2015 looks unlikely. 20Government of Malawi and UNDP (2003) "Malawi Millennium Development Goals Report 2003". 19 Table 1.6: Summary of Malawi's progress towards the MDGs as of end-2005 MDG Most Target Target Baseline Intermediate Recent for 2015 feasibility 1990-1992 1998-2000 2004-2005 Goal 1: Eradicate extreme poverty and hunger Target 1: Halve, between 1990 and 2015, the proportion of people Unlikely under the poverty line (Indicator 1) 54.0 53.9 52.4 27.0 Target 2: Halve, between 1990 and 2015, the proportion of people who suffer from hunger (under the ultra-poverty line, Indicator 5) 28.0 25.4 22.2 14.0 Goal 2: Achieve universal primary education Unlikely Target 3: Ensure that, by 2015, children everywhere, boys and girls alike, will be able to complete a full course of primary schooling (completion rate, Indicator 7) - 69.0 60.0 100.0 Goal 3: Promote gender equality and empower women Potentially Feasible Target 4: Eliminate gender disparity in primary and secondary education, preferably by 2005, and to all levels of education (Gender ratio in primary, Indicator 9a) 0.9 0.9 1.0 1.0 Goal 4: Reduce child mortality Potentially Target 5: Reduce by two-thirds, between 1990 and 2015, the under- Feasible five mortality rate (Indicator 12) 234.0 189.0 133.0 78.0 Goal 5: Improve maternal health Unlikely Target 6: Reduce by three-quarters, between 1990 and 2015, the maternal mortality ratio (Indicator 16) 620.0 1120.0 960.0 155.0 Goal 6: Combat HIV/AIDS, malaria and other diseases Target 7: Have halted by 2015 and begun to reverse the spread of HIV/AIDS (HIV prevalence among 15-24-year-old pregnant Potentially women, Indicator 18) 17.4 24.1 15.3 <17.4 Feasible Target 8: Have halted by 2015 and begun to reverse the incidence of malaria and other major diseases (Proportion of population in malaria risk areas using effective malaria prevention (percent of under five children using bednets, Indicator 22a) - - - - Goal 7: Ensure environmental sustainability Target 9: Integrate principles of sustainable development into country policies and programmes and reverse loss of environmental resources (Proportion of forested land area, Indicator 25) 34.7 27.2 - - Unlikely Target 10: Halve, by 2015, the proportion of people without sustainable access to safe drinking water (Indicator 30) 47.0 62.0 66.1 73.5 Target 11: By 2020, to have achieved a significant improvement in the lives of at least 100 million slum dwellers (Proportion of households with access to secure tenure, Indicator 32) 95 91.0 - 87.7 Source: Authors based on various data (see Annex 1F for details). 20 CHAPTER 2: POVERTY PROFILE AND THE DETERMINANTS OF POVERTY INTRODUCTION 1. While a large share of Malawi's overall population is poor, this population is still a diverse group with diverse problems and conditions. This chapter adopts a multidimensional approach in the analysis of living conditions of the Malawian population. First, it identifies salient characteristics such as the demographic composition of a household, occupation of the household head, education levels, health and nutrition characteristics, quality of housing, asset ownership, and access to key infrastructure, to build a poverty profile for Malawi's poor in 2005. The chapter also briefly examines subjective assessments of well-being, and the extent to which such assessments are related to income and non-income measures of poverty. Then it investigates the key determinants of poverty through a multi-variate analysis THE CHARACTERISTICS OF POOR HOUSEHOLDS IN 2005: THE POVERTY PROFILE 2. The poverty profile seeks to determine which household characteristics are highly correlated with poverty, and what types of households are very likely to be poor. To this end, this section will explore the demographic composition of households as well as key differences in education, health, water and sanitation, housing quality, durable assets, land holdings, and livestock. The chapter builds on the "Profile of Poverty in Malawi, 1998: Poverty analysis of the Malawi Integrated Household Survey, 1997-98" produced using the IHS1 household data by the then National Economic Council (NEC 2000).1 It is also useful to focus on the characteristics of the ultra-poor, to inform policy interventions to alleviate extreme poverty. As highlighted in this chapter, and also in Chapter Three, most of the key characteristics remain the same, even though the profile of the ultra-poor is not fully identical as the poor's. Demographic characteristics of poor households: household size and dependency ratio 3. Malawi has a very young and rapidly growing population. Malawi's total population in 2005 was estimated at 12.3 million, of which about 60 percent is under the age of 20 and about 75 percent is under the age of 30. The total population is expected to increase rapidly over the next few decades, reaching around 20 million by 2025 (Box 2.1). 4. The rapidly growing population is a key driver of Malawi's persistent poverty. Poor households in Malawi are generally larger than non-poor households, averaging 5.4 members compared to an average of 3.8 members in non-poor households (Table 2.1). When looking at average household size by income decile, the relationship is even clearer--households in the poorest decile are more than twice as large as households in the richest decile (6.3 versus 2.9 members). Urban and rural households are similar in size, and male-headed households tend to 1Hereafter referred to as Poverty Profile 1998. The Poverty Profile 1998 cannot be used here for absolute comparisons because, as was explained in Chapter One, the methodologies used to calculate poverty are different. However, it is possible to look at the patterns reported in the Poverty Profile 1998 to determine if there are changes between the two surveys. 21 be larger than female headed households. Average households size is slightly higher in the North and Central regions (4.8 and 4.7 members, respectively) compared to the South (4.3 members). Box 2.1: Malawi's population demographic projections to 2025 Demographic breakdown of Malawi's population by age and gender, in the years 2005 and 2025 Population in year 2005 Population in year 2025 75-79 75-79 Female Female 60-64 Male 60-64 Male 45-49 45-49 e e Ag Ag 30-34 30-34 15-19 15-19 0-4 0-4 2 1 0 1 2 2 1 0 1 2 Million Million Source: National Statistical office 5. Poor households tend to have a larger dependency ratio than non-poor households,2 with, on average, a dependency ratio of 1.4 compared to non-poor households whose dependency ratio averages 0.8 (Table 2.1). When viewed by decile, the dependency ratio shows a steady decrease as households become less poor; the poorest decile has a dependency ratio (1.6) which is four times as large as the ratio for households in the richest decile (0.4). The difference is particularly large for poor households headed by females (Figure 2.1) Dependency ratios are similar for urban and rural households, though rural households have slightly higher dependency ratios. Table 2.1: Household size, number of children, and dependency ratio by poverty status, 2005 Household Size Number of Children Dependency ratio Overall Overall Overall Female Headed Overall 4.5 2.1 1.1 1.4 Non-Poor 3.8 1.5 0.8 1.0 Poor 5.4 2.8 1.4 1.8 Source: National Statistical Office, IHS2 6. The differences in dependency ratio are largely driven by the number of children in the household. In Malawi, there are on average 2.1 children (aged 0 to 14 years) per household. Male headed households consistently have more children than female headed households. The 2The dependency ratio is defined in the standard way, as the ratio of the total number of persons in the household outside the economically active population (children under the age of 15 and adults above 65 years of age) to the total number of prime-age adults. 22 average number of children is highest in rural areas and in poor households: the average number of children in the poorest decile (3.5) is four times that of the richest decile (0.9). In line with these observations, while children make up 49.9 percent of the total population of Malawi, they account for 53.4 percent of the poor population. In other words, more than half of the poor in Malawi are children (Figure 2.2). Figure 2.1: Household size, number of children, and dependency ratio, by decile, 2005 Houshold Size and Number of Children Dependency ratio 7 2.5 6 Household size Overall Number of children 2.0 5 Female-headed 4 1.5 3 1.0 2 0.5 1 0 0.0 orest 2 3 4 5 6 7 8 9 orest 2 3 4 5 6 7 8 9 Po Richest Po Richest Note: Children are those between 0 and 14 years of age Source: National Statistical Office, IHS2 Figure 2.2: Demographic composition of poverty in Malawi, 2005 Population poor and non-poor by age group 100+ 90-94 80-84 Poor Non-poor 70-74 60-64 50-54 40-44 30-34 20-24 10-14 0-4 1500000 1000000 500000 0 500000 1000000 1500000 Population Source: National Statistical Office, IHS2 Key characteristics of the household head: gender, age and level of education 7. Poverty rates are higher in female-headed households in both urban and rural areas (Figure 2.3). About 51 percent of the people who live in male-headed households are poor, while 58 percent of people living in female-headed households are poor.3 People in households 3Note that, while female headed households are disproportionately poor, the majority of the poor live in male- headed households. This is because the large majority of households in Malawi are male-headed (77 percent). 23 headed by older members consume less per capita per day than those with younger household heads, up to ages 44-49, when the relationship flattens off (Figure 2.4). Figure 2.3: Population Poverty Rates by Figure 2.4: Population poverty rates by Sex of Household Head and Residence age group of household head Poverty rates by gender and Poverty by age of household head residence of the household head 70 70 60 60 61 50 50 58 55 age 40 age 51 40 entc 30 entc 30 erP 32 erP20 20 24 10 10 0 0 0 <2 24 29 34 39 44 49 54 59 64 65+ Male Female Male Female Male Female 20- 25- 30- 35- 40- 45- 50- 55- 60- M alawi Rural Urban Age of the Household Head Source: National Statistical Office, IHS2 Source: National Statistical Office, IHS2 8. Overall, the average level of education of all household heads is low. About 28 percent of household heads have no education, and 55 percent have only primary education. These averages mask wide differences between urban and rural households, and across gender, however. In rural areas, 30 percent of household heads have no education compared to 9 percent in urban areas. On the other hand, as much as 47 percent of the population in urban areas has completed secondary education or higher Figure 2.5: Education of Household Heads by wealth decile (percent) 100% University or Training College 80% Secondary 60% Senior Primary 40% Junior Primary 20% None 0% orest 2 3 4 5 6 7 8 9 t Po Riches Notes: Junior Primary is Standards 1-4, Senior Primary is Standards 5-8, Secondary is Forms 1-6. Source: National Statistical Office, IHS2 24 9. As shown in Figure 2.5, there is a high correlation between poverty and the level of education. Almost three-quarters of the household heads in the poorest decile have less than complete primary education compared to only 20 percent of the household heads in the richest decile. On the other hand, virtually all the household heads that received a degree at University or Training College belong to the top two deciles. The lack of education is much greater for female household heads, with 50 percent of the female heads without any education. 10. In sum, poor households tend to be larger than non-poor, have higher dependency ratios, and a greater number of children. They also tend to be headed by persons with little or no education. Female headed households are disproportionately poor. These demographic characteristics of poor households are similar to those identified in the 1998 Poverty Profile. Occupation of the Household Head4 11. Very few household heads work at a wage or salaried job that can be easily identified by occupational classifications. Hence, this analysis focuses on responses of household heads to questions about how they spent hours in the last 7 days. 5 Figure 2.6: Occupation of the Household Head Occupation of the household head 100% 90% Wage w ork only 80% Household enterprise only 70% 60% More than one activity 50% Household farm or fishing only 40% Ganyu only 30% Job, but no w ork in last 7 days 20% No job 10% 0% Malaw i Poor Non-Poor Urban Rural North Central South Overall Note: See Footnote 21 for more information about these categories Source: National Statistical Office, IHS2 12. Figure 2.6 depicts a breakdown by occupation of the head of household. Most household heads report working only on their household farm or fishing activity (38 percent). As expected, this is more common in rural areas, reaching a peak of 55 percent in the North Region. Female household heads are also more likely to work solely on the household farm (45 percent, not 4If we use the International Labor Organization (ILO) standard definition of employment, 96 percent of all household heads are employed. This figure masks the actual employment situation of heads of households. To identify the employed according to the ILO definition, the respondent: (a) worked at least one hour in the last 7 days or (b) the respondent had a job to return to if they did not work in the last 7 days. 5These categories are taken from questions regarding the number of hours worked during the past 7 days on (i) household agricultural activities or fishing; (ii) non-agricultural or non-fishing household business for self; (iii) non- agricultural or non-fishing household business for other household member; (iv) casual, part-time or ganyu labor; or (v) work for a wage, salary, commission or payment in kind (excluding ganyu); or (vi) the respondent had a job to return to if they did not work in the last 7 days. 25 shown). The second largest group is formed by those who report working in multiple jobs (26 percent). Most of these individuals are farmers who work at additional jobs (94 percent of the total). The third largest group of household heads works solely at a waged or salaried job (11 percent). These wage workers are found predominantly in urban areas, where they account for 35 percent of all urban household heads. In rural areas this proportion is about 8 percent. Finally, a few heads of households work solely in a household enterprise owned by themselves or other members of the household (8 percent). This is more common in urban areas (16 percent) than in rural areas (7 percent). Of the household heads who own a household enterprise, 50 percent are in Wholesale and Retail Marketing and 26 percent are in Manufacturing. 13. In sum, the occupation of the household heads from poor households is characterized by a larger reliance solely on the household farming or fishing activity, and by a lower likelihood of working on a wage or salaried job, and of working in a household enterprise. This implies a heavy reliance on agriculture for employment and/or subsistence farming. Education and the poor: school attendance, and enrollment in primary, secondary, and tertiary education, and literacy rate School Attendance 14. As shown in Figure 2.7, at all ages, poor children are less likely to be attending school than their non-poor peers. This gap is largest at young ages. For example, for children ages 5-6, those from non-poor households are 40 percent more likely to be attending school than poor children. School attendance for both poor and non-poor children increases up to about age 12, and then attendance rates start to decline. After controlling for other differences, female-headed households spend a larger share of their total budget on education, suggesting that female heads are significantly more likely to send children to school (see Annex 2A) Figure 2.7: School attendance by poverty status 100% 90% gnid 80% 70% Malawi tentat 60% Poor 50% Non-Poor cenreP 40% 30% 20% 5 6 7 8 9 10 11 12 13 14 15 Age Source: National Statistical Office, IHS2 26 Enrollment in primary education6 15. The primary education Gross Enrolment Rate (GER) provides an indicator of the capacity of the primary education system.7 Consistent with previous estimates, the GER in Malawi is quite high at 108 percent (Figure 2.8). It should be noted that a high ratio does not necessarily indicate a successful education system, since the large number of pupils could be the result of grade repetition, as well as overage and underage enrolments.8 The rate for boys is higher than BOX 2.2: CHILD LABOR IN MALAWI Working children have less opportunity to attend school and are more susceptible than adults to bad work environments, such as low or no pay, poor working conditions, and physical abuse. The 2005 IHS2 collected information on the work activities of children age 5-14. They were asked a series of questions about whether they were doing any kind of work, whether they did unpaid family work on the farm or in a family business, and whether they regularly helped with household chores. As in many developing countries, child labor is common in Malawi. About 35 percent of children work either for the family business or farm, or work for a non-relative (paid or unpaid), or spend 4 or more hours a day doing household chores. Overall, older children and children in rural areas are more likely to be working. Girls are more likely than boys to do domestic work. The table below shows that approximately 3 percent of children age 5-14 work for persons who are not members of their household (paid or unpaid). Among children who help around the house with household chores, 40 percent of children do these chores for an average of less than 4 hours per day and 4 percent work for 4 or more hours per day. Percentage of children who are currently working by type of employment and selected background characteristics in 2005 Domestic work Currently doing work Background Work for non- Less than 4 4 hours or on family farm or Currently characteristics household member hours per day more per day family business working (A) (B) (C) (A+B+C) Children (age 5-14) 2.7 39.7 4.0 28.4 35.0 Age 5-9 0.8 27.0 1.8 14.6 17.2 10-14 5.0 54.2 6.3 44.4 55.7 Sex Male 3.0 27.0 1.7 30.4 35.1 Female 2.5 52.4 6.1 26.6 35.2 Residence Urban 1.0 40.3 8.5 5.2 14.7 Rural 2.9 39.6 3.4 31.0 37.3 Region North 1.4 32.7 2.8 37.1 41.3 Center 3.8 42.7 2.9 30.9 37.6 South 2.0 38.6 5.3 23.7 31.0 Poverty Status Non-Poor 2.4 41.8 4.2 25.8 32.4 Poor 2.9 38.4 3.8 30.0 36.7 Source: National Statistical Office, IHS2 Notes: Work for non-household member is any child who worked for 0-60 hours for ganyu or for a wage. Domestic work includes cooking, doing laundry, cleaning house, collecting water, and collecting firewood or other fuel materials. Work on family farm or business is any child who worked for 0-60 hours on household agricultural activities, on own non-agricultural or non-fishing household business, or on non-agricultural or non-fishing household business of any household member. 6The IHS2 questionnaire asked attendance and not enrollment so these figures are actually attendance and not enrollment. The rates are calculated for children attending school in the 2004 and 2005 academic years combined. 7The primary GER is defined as the total enrolment in primary education regardless of age, expressed as a percentage of the official school age population (6 to 13 years). 8For a discussion, see World Bank (2004). 27 for girls, and the rate for urban areas is higher than for rural areas. However, these differences are not very large. The largest gap is between poor and non-poor households. In fact a boy from a poor household is less likely to be in school than a girl from a non-poor household. 16. Similarly, the primary education Net Enrolment Rate (NER)9 is higher for children from non-poor households (84 percent) than their peers in poor households (75 percent). Surprisingly, the girls have slightly higher NER than boys (79 percent and 77 percent, respectively), suggesting that there is no bias against girls enrolment in primary school. On a less positive note, the higher NER also reflects the fact that girls leave school earlier than boys, resulting in less completed years of schooling. 17. The results in the Poverty Profile 1998 show very similar NER, but much larger GER. However, because we do not know exactly how the rates were calculated for the 1998 poverty profile, comparisons must be made with caution. 18. There are large differences in NER and GER across the three rural regions (Figure 2.8). The North Region has the largest NERs and GERs of all the rural regions, with the Central Region and South Region being roughly equivalent. The differences in regional enrolment rates were already highlighted in the 1998 Poverty Profile, and also in numerous studies on education in Malawi. The reason for these differences remains unclear, however. Figure 2.8: Primary Gross Enrollment Rate in 2004-2005 (percent) Primary Gross Enrolment Rate Primary Net Enrolment Rate 130 110 120 100 110 90 100 80 90 70 80 60 70 50 i r i law Poo Poor ral rth er th or ral er Ru Poor uth Ma Non- Urban No Cent Sou Malaw Po Ru North Non- Urban Cent So Boys Girls Total Boys Girls Total Note: Includes all students attending Standard 1 through 8 in 2004 and 2005. Source: National Statistical Office, IHS2 19. In Malawi, about 25 percent of school age children from poor households do not enroll in primary education. According to IHS2 respondents, lack of money is the major reason for 9The NER is defined as the percentage of the official children population of primary school age that is enrolled in primary school. The NER excludes overage students in an attempt to capture accurately the system's cover and internal efficiency. It does not solve the problem completely since some children fall outside the official age simply because of late or early entry rather than grade repetition. Simply put, the NER reflects the percent of children of official primary school age who actually attend primary school. 28 failing to enroll (about 41 percent of children who never enrolled).10 A further 10 percent report that the parents prevented the child from going to school or that their help was needed at home. 20. Further, while the primary enrollment rates are high by regional standards, very few children actually complete primary school (World Bank 2004). The 2002 DHS EdData survey indicates that the grade 1 drop out rate was 8 percent for males and 9 percent for females. Similarly, the drop out rate for grade 8 was 20 percent for males and 21 percent for female. As a result, in 2002 only 60 percent of primary school students who entered grade 1 could be expected to reach grade 5, with or without repetition, and only 39 percent of those who entered grade 1 could be expected to reach grade 8 (DHS EdData 2004). 21. According to IHS2 respondents, again the cost of schooling is the major cause for the high rates of drop out in primary education. As many as 49 percent of students report lack of money for fees and uniforms as the major cause of primary school drop out. Reported lack of interest in continuing education is also common (24 percent). Early marriage or pregnancy account for 9 percent of drop outs. A further 6 percent reports that the parents forced the child to stop and that their help was needed at home. Enrollment in secondary education11 22. Although 66 percent of children of secondary school age (14 to 19 years of age) are attending school, only 22 percent were attending secondary level school (while the balance were attending primary school, which is one of the causes of the high Primary Gross Enrollment Rate). In fact secondary gross and net enrollment rates are low (at 22 percent and 15 percent, respectively). The disparity in secondary enrollment rates is very large between poor and non- poor students and between urban and rural students. Three times as many non-poor students as poor students are enrolled in secondary education and boys and girls from the richest decile are 10 times as likely to attend secondary school compared to those in the poorest decile (Figure 2.9). Three times as many urban students as rural students are enrolled in secondary education. Differences in GER across rural regions are also substantial, with the North Region far above the South and Central regions. The regional differences become very small when looking at NER, indicating that there may be more repetition of grades in the North Region. 23. According to IHS2 respondents, lack of money is by far the most common reason for not continuing to secondary education (58 percent). Early marriage and pregnancy are also common (15 percent), and lack of interest (13 percent). 24. Secondary Net and Gross Enrollment Rates are not presented in the Poverty Profile 1998, but the information that is provided suggests that only a very small percentages of the population attended secondary school. 10While tuition fees have been eliminated in Malawi's public primary schools, parents still require money for each child's attendance to meet the costs of the uniforms, the school's development fund, maintenance of teacher's house and school blocks, school report production, and the construction of pit latrines. 11As for primary education, the IHS2 questionnaire asked attendance and not enrollment so these figures are actually attendance and not enrollment. The rates are calculated for children attending school in the 2004 and 2005 academic years combined. 29 Figure 2.9: Secondary Gross Enrollment Rate by Wealth Decile (richest and poorest), 2004-2005 (percent) Secondary Gross Enrolment Rate Secondary Net Enrolment Rate 80 80 70 Overall 70 Overall 60 Urban 60 Urban 50 Central 50 Central 40 South 40 South 30 North 30 North 20 20 10 10 0 0 Poorest Richest Poorest Richest Source: National Statistical Office, IHS2 Enrollment in tertiary education 25. Enrolment in tertiary education is very small (less than 0.1 percent of Malawi's population) and is associated almost exclusively with the households from the richest decile.12 Of those enrolled in tertiary education, the vast majority live in urban areas. Literacy Rate 26. The adult literacy rate is defined as the percentage of individuals aged 15 years and older who can, with understanding, both read and write a short, simple statement about their everyday life. As shown in Figure 2.10, the national adult literacy rate is low at 64 percent, and is substantially higher among males (76 percent) than females (53 percent). Further, the likelihood of being literate is higher for individuals residing in urban areas, and is also higher in non-poor households. The results by decile again highlight the relationship between poverty and education. Overall, only about half of the adults in the poorest decile are literate, compared to 87 percent of adults in the richest decile. The gender gap in literacy decreases steadily with income, but even in the richest decile the literacy rate for females is 10 percent lower than males. 27. The youth literacy rate (calculated on individuals aged between 15 and 24 years) is 76 percent, well above the adult literacy rate. Though a rural-urban gap in literacy remains, there is much less disparity between young females and males, with 97 literate female youth to every 100 male youth. The youth from non-poor households are more likely to be literate (83 percent) than 12The IHS2 reports few students currently in tertiary education, and few individuals with a tertiary education. Less than 60 respondents reported being enrolled in University or Training College in 2004 or 2005. Overall, less than 1 percent of the respondents report having attended University or Training College. Hence the data must be used with caution because of the very small population. 30 from poor households (70 percent), though this gap too has closed substantially when compared to the adult literacy rates. Figure 2.10: Adult Literacy Rate (percent) Adult Literacy Rate Adult Literacy Rate 100 100 on 90 90 ati 80 oni at 80 70 pul 70 popul 60 po 60 of 50 of 50 age 40 40 Overall entc 30 30 20 Female erP entagec 20 10 Male 0 erP 10 0 Malaw i Poor Non-Poor Urban Rural 2 3 4 5 6 7 8 9 t Males Total Females orest Po Riches Note: Literacy rate is defined as individuals who can read and write in their mother tongue or in English. Adults include all individuals aged 15 years and older. Source: National Statistical Office, IHS2 Health and the poor: morbidity, births attended by skilled health personnel, chronic health problems, and child malnutrition Morbidity 28. Almost 26 percent of the population reported suffering an illness or injury in the last two weeks (prior to answering the IHS2 survey). Interestingly the non-poor population is more likely to report an illness (28 percent) than the poor population (24 percent). This is mainly true in rural areas. However, it should be emphasized that this is self-reported illness, and the poor may have higher thresholds before classifying themselves as ill. Most respondents reporting an illness said that they had a fever or malaria (30 percent overall, 39 percent urban and 30 percent rural). The second most common illness is lower respiratory complaints (16 percent overall, 14 percent urban and 16 percent rural). In all, fewer urban dwellers (16 percent) than rural dwellers (27 percent) reported suffering illness or injury. The Poverty Profile 1998 found similar results; more non-poor than poor reported suffering an illness or injury, and the most common illness reported was fever. Chronic Health Problems 29. The proportion of individuals suffering from chronic illnesses is highest among the non- poor (Figure 2.11). As chronic illness is self-reported and not necessarily assessed objectively by a medical practitioner, it may not capture a person's actual health status.13 The results also 13Reporting of chronic conditions may be higher among those who have access to health services (and have a heightened perception of their health status). Conversely, low income people may not be able to afford to stop 31 show that for both poor and non-poor, more women report having a chronic illness than men, and that there are more chronically ill individuals in the rural areas than in the urban areas. By decile, the proportion of respondents reporting chronic illness is approximately 10 percent for each decile. Interestingly in the bottom 5 income deciles, much less urban respondents report having a chronic illness compared to rural respondents. However, this relationship is inverted in the richest two deciles, and as many as one-third of urban respondents in the richest decile report suffering from a chronic illness. Births Attended by Skilled Health Personnel 30. Births attended by skilled health personnel are defined as births that were attended by a doctor, clinical officer and nurse or mid-wife.14 Births by women from non-poor households are more likely to be attended by skilled health personnel (compared to births from poor households). The results also show a marked difference by decile. While only about half of women in the first decile used skilled health personnel, almost three-fourths of women in the highest decile used skilled health personnel. The results of the survey also indicate that there is a large difference between births in urban (85 percent) and rural (56 percent) areas being attended by skilled health personnel. With a few exceptions, at least 80 percent of women in urban areas in every decile used skilled health personnel while fewer than two-thirds of rural women, regardless of region, used skilled health personnel. The difference might be due to a higher rate of births attended by traditional births attendants in rural areas. It might also be due to the greater distance to health centers and the lower number of qualifies staff in rural health centers. In rural areas, the Central Region shows the least use of skilled health personnel in all deciles. Figure 2.11: Proportion of Persons Reporting Having a Chronic Illness Respondents reporting a chronic illness Respondents Reporting Chronic Illness 14 35 12 Males 30 Overall Females 10 25 Urban 20 Rural entc 8 entc erP 15 6 erP 10 4 5 2 0 0 2 3 4 5 6 7 8 9 t Malaw i Poor Non-poor Urban Rural orest Po Riches Source: National Statistical Office, IHS2 working to seek treatment, leading them to under-report health. Of all those who self-reported chronic ill health, approximately 60 percent said that they had been diagnosed by a medical professional, with 40 percent indicating that the illness had been diagnosed by a traditional healer, household member or self-diagnosed. In urban areas, 77 percent of respondents reported they were diagnosed by a medical professional, compared to 58 percent in rural areas. 14Traditional birth attendants (TBA) are not included in the definition of skilled health personnel because information on whether or not they were trained was not collected. 32 Caloric Intake and Child Nutrition Status 31. The average per capita availability of calories in 2005 was estimated at 2,366 kcal per day using the IHS2 data (Figure 2.12). The relatively low level of average per capita calorie availability in Malawi is compounded by an unequal distribution. The average per capita availability is 13 percent higher in urban areas than in rural areas. Urban dwellers have consistently higher caloric consumption than their rural counterparts in all regions. Looking at the regional composition, the figures are consistently higher in the Central region. Although it could be expected that poor households have lower levels of caloric consumption compared with better-off individuals, the magnitude of the difference is disconcerting: poor individuals consume on average 58 percent of the calories of their better-off counterparts. 32. The nutritional well-being of young children directly and indirectly contributes to the country's development. Child nutrition status is calculated using the height and weight measurements for children aged 6 to 59 months in the sample households of the IHS2. Prevalence of underweight children is the percentage of children under five years of age who are classified as undernourished according to the anthropometric index of nutritional status called weight for age.15 The results in Figure 2.13 show that 18.1 percent of children below five years of age are underweight. The results indicate there is not much difference in the prevalence of malnutrition in urban and rural areas. This indicates that the problem of malnutrition is spread nation wide. Figure 2.12: Caloric Intake (daily kcal per capita) and Child Nutritional Status (percent underweight) Average per capita daily caloric intake Percentage of children underweight 3500 25 3000 day 20 2500 perat 2000 entc 15 apic 1500 erP10 per 1000 alck 5 500 0 0 e h r NationalUrban ral rth uth or or ral rth ral Ru No Centr So Po po Poo poor Non NationalUrban Ru No Cent Sout Non- Notes: Moderate underweight are those children more than 2 standard deviations from the mean. Children include those children between 6 and 59 months of age. Source: National Statistical Office, IHS2 15Stunted children are those children with a low ratio of height for age. This indicates long-term or chronic malnutrition. Wasting children are those with low weight for height resulting from acute malnutrition, as in a situation of famine. Underweight children are those with low weight for age which is a combination effect of wasting and stunting. See Chapter Four for additional details. 33 33. Of particular interest is the apparent inconsistency in Figure 2.12 between calorie intake and nutritional status. This is most clearly seen in Central region, which has both the most calories per capita and the most underweight children. Higher calorie intake does not necessarily translate into lower levels of malnutrition, due to a host of reasons that will be discussed in Chapter Four. Housing: overall quality of the dwellings, sanitation, water, cooking fuel and lighting fuel Quality of the dwellings 34. A housing quality index has been created by combining information on the various aspects of the dwelling: housing tenure, construction materials, outer walls, roofing materials, flooring materials, number of rooms, presence of electricity in the dwelling, presence of improved drinking water, type of toilet facilities, and method of garbage disposal.16 35. On average the overall quality score for all dwellings was 5.4, with urban dwellings receiving 7.4 points and rural households receiving 5.2 points. Slightly more than half of the dwellings in Malawi are of medium quality. This holds for poor and non-poor households in both urban and rural areas. In total, only 10 percent of the dwellings are classified as good quality in Malawi, and the majority of these are in urban areas where there is more access to services. The North Region has better dwellings than any of the other rural areas, while the Central Region has the worst (not shown). Figure 2.13: Housing Quality by Wealth Decile and Location (percent) Rural Urban 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% orest 2 3 4 5 6 7 8 9 t orest 2 3 4 5 6 7 8 9 Po Low Medium Good Riches Po Low Medium Good Richest Source: National Statistical Office, IHS2 16The information for the housing index can be found in Module G of the IHS2 household questionnaire. Housing quality can be difficult to measure because of the general nature of the information collected in the questionnaire. For example, information is collected on the materials that make up the roof of the dwelling, but there is no assessment of the quality or condition of the roofing materials at the time of the interview. Points were assigned to each component to indicate their quality and summed for the household. The assignment of points for quality was based on the ranking used in the questionnaire. The total points per household ranged from 1 to 12. This was divided so that households with 1 to 4 points were designated as low quality, with 5 to 8 points as medium quality and with 9 to 12 points as good quality. 34 36. Non-poor households have a lower share of low quality dwellings and a larger share of good quality dwellings, than poor households. As expected, in urban areas, the share of good quality housing increases rapidly with wealth. Within rural households, however, the improvement in quality appears less marked and the shares of housing quality remain fairly constant across the bottom eight deciles (Figure 2.13). Presence of Improved Sanitation 37. Overall, 64 percent of the population has improved sanitation.17 The proportions are higher in non-poor households (71 percent), urban households (80 percent) and households with a male head (67 percent). Access to improved sanitation increases by wealth decile with about half of the population having improved sanitation in the lowest decile, compared to 80 percent in the highest decile. The North Region has consistently less access, with only about half of the population at each decile having improved sanitation. In the Central and South Regions, access improves from half in the lowest decile to 80 percent in the highest decile (not shown). Presence of Improved Water 38. A household's source of drinking water is important because potentially fatal and preventable diseases are prevalent in unprotected water sources. Figure 2.14 shows that while there is not much difference between the poor and non­poor in accessing improved water,18 the proportion is higher in urban than in rural areas. Two-thirds of the population in the lowest decile has access to improved water versus 78 percent in the highest decile. In fact, access to improved water increases by decile in urban areas, but stays constant in rural areas. Almost the entire urban population in the highest decile has access to an improved water source. For rural areas, the Central Region has the worst access while the South Region has the best (not shown). Figure 2.14: Proportion of Population with Improved Water Source reta 100 Access to Improved Water W 90 100 90 vedorp 80 80 70 70 Im htiw 60 60 50 noi 50 40 40 Overall latupoPfot 30 Urban 30 20 Rural 20 10 0 10 t cenreP 2 3 4 5 6 7 8 9 0 orest Poor Non- Urban Rural Male Female Po Riches Poor Head Head Source: National Statistical Office, IHS2 17Improved sanitation has been defined as households who reported having flush toilets, Ventilated Improved Pit Latrines (VIP) or traditional latrines with a roof. 18Improved water sources are defined as having the main source of water: piped into dwelling, piped outside dwelling (personal), communal stand pipe, personal hand pump, or communal hand pump. 35 Cooking Fuel 39. On average, about 90 percent of the population uses firewood, purchased or gathered, as their main source of cooking fuel. In rural areas, virtually every household uses firewood for cooking regardless of wealth (Figure 2.15). Only ten percent of the rural households in the richest decile use charcoal. In urban areas, fewer households use firewood as the income increases, from 90 percent in the poorest decile to only 20 percent in the richest decile. The other main fuel sources for urban households are charcoal (48 percent on average) and electricity (12 percent on average). The high use of firewood as cooking fuel probably contributes to the reporting of lower respiratory illness as the second largest reported illness (see above). Figure 2.15: Households Cooking with Firewood by Wealth Decile (percent) Households Cooking with Firewood 100 90 80 70 60 50 40 Overall 30 20 Urban 10 Rural 0 rest 2 3 4 5 6 7 8 9 Poo Richest Source: National Statistical Office, IHS2 Figure 2.16: Household Lighting Fuel by Residence and Wealth Decile (percent) Rural Households Lighting Fuel Urban Households Lighting Fuel 100% 100% 90% 80% 80% 70% 60% 60% 50% 40% 40% 30% 20% 20% 10% 0% 0% rest 2 3 4 5 6 7 8 9 9 t Poo Richest orest 2 3 4 5 6 7 8 Po Riches Paraffin Firew ood Electricity Other Paraffin Firew ood Electricity Other Notes: Other includes grass, gas, battery, and candles. Source: National Statistical Office, IHS2 36 Lighting Fuel 40. On average, 85 percent of households use paraffin for lighting, 4 percent use firewood, and 11 percent use other fuels. Use of paraffin is relatively constant by decile in rural areas, but decreases rapidly as income increases in urban areas. Only 59 percent of the households in the richest decile use paraffin, and this decreases to only 25 percent in urban areas. The other households are using "other" fuels, primarily electricity. Electricity is used in 6 percent of households overall for lighting, but for 33 percent of urban households. BOX 2.3: BIOMASS AVAILABILITY AND POVERTY IN MALAWI In Malawi, 90 percent of the poor live in rural areas and share space with forests and shrublands, and over 90 percent of the total energy demand in Malawi is met with biomass. As many as 96 percent of the poor use fuelwood as cooking fuel, of whom 55 percent collect from unfarmed community areas. As such, it is important to explore the extent to which poverty and forest degradation are interlinked. Scarcity could affect household wellbeing either directly, affecting income (from any limited sales of fuelwood and from labor re-allocation) or through its impact on household health or leisure. Annex 2B presents a regression model to study this issue in detail, using biomass as a proxy for fuelwood. The main findings are summarized below. Our study estimated biomass availability in Malawi from satellite data. Malawi's forests are mainly in the North region, with 41 percent of the country's biomass, or 159 cubic meters, followed by the Central region with 38 percent of biomass. The South region has the least. Moreover, forest cover in all regions has decreased significantly, and particularly in the North, since1990 when the Government conducted a Biomass Assessment. Our analysis found that on average, a Malawian household spends the equivalent of MK2,558 per capita, per year on fuelwood, or about 12 percent of total annual consumption expenditure. 84 percent of all individuals who collected fuelwood were women. On average, active women spent 1 ½ hours on firewood collection, with little difference in collection times between regions and between rich and poor households. The study suggests that average rural household consumption expenditure in Malawi declines after biomass reaches 26 cubic meters per hectare. 72 percent of rural households are in areas with biomass levels lower than this threshold. Thus, most of the rural poor would benefit if average biomasss per hectare almost doubles. Though significant, the average effect of biomass on rural per capita consumption expenditure is, but small, however. We found that a 10 percent increase in current levels of biomass per hectare is associated with approximately 0.1 percent higher annual per capital consumption expenditure. This figure is twice as large in the south, where welfare would increase by 0.2 percent. Biomass scarcity also has a significant but small effect on the number of hours spent collecting fuel wood: a decrease in biomass of one cubic meter results in a one minute increase in collection time. The small size of these effects suggests that households are using effective adjustment strategies to minimize their welfare loss from biomass scarcity.* Households use a range of strategies to cope with scarcity: planting trees, increasing the time allocated to collection while performing simultaneous tasks, sharing fires for cooking, improving fire management practices, preparing fewer meals or faster cooking foods, and using lower quality fuelwood. These results may help explain why past efforts to increase fuelwood availability by encouraging community tree planting have been unsuccessful. Though Malawians perceive scarcity, given various constraints, households may choose not to use scarce land and labor to plant fuelwood tree crops. Thus a clearer understanding of household responses to scarcity may help policy makers target poverty eradication and biomass conservation strategies more effectively in Malawi. *Some caveats apply to these findings: the measure of biomass used is crude, and a more refined measure might help explain the biomass-welfare relationship better. Our measure of household welfare does not include other benefits such as biodiversity conservation and water catchment areas, which do not directly translate into higher annual household income. In addition, collection times were measured in half-hour units, which might be too large to capture time savings accrued by women. 37 Household assets: ownership of durable goods, livestock, land and labor Durable Goods 41. Households with tangible assets can use those assets to improve their welfare, both by using the asset to help the household to work more efficiently and therefore increase income, or through the ability to sell off the assets when the household experiences a shock or there is a downturn in the economy. The IHS2 household questionnaire includes information that can be used to determine the amount of assets that is owned by a household. It includes common durable goods including household appliances, and farm implements, as well as information on land ownership and the existence of livestock. Figure 2.17 shows the distribution of selected assets. Figure 2.17: Household Ownership of Selected Assets (percent of households) 10 70 9 60 8 7 50 Malaw i 6 40 Non-Poor 5 Poor 4 30 Urban 3 20 Rural 2 10 1 0 0 Oxcart Wheelbarrow Handsprayer Bicycle Radio Source: National Statistical Office, IHS2 42. Bicycles are one of the major means of transportation in Malawi. They are used to transport goods and people. Overall, only about one-third of all households in Malawi own bicycles. Ownership is lower in urban areas (20 percent) than rural areas (38 percent), which may be the result of more transportation options in urban areas. Ownership of useful farm implements such as an oxcart, wheelbarrow or handsprayer, is rare, with less than 3 percent of all households own these assets. For oxcarts, the North region and Central region show higher percentages of ownership than South region households. More households in the Central region have a handsprayer than households in either the North region or the South. Livestock ownership 43. The concept of Tropical Livestock Units (TLU) provides a convenient method for quantifying a wide range of different livestock types and sizes in a standardized manner.19 In the IHS2, information was collected about cattle, oxen, goats, sheep, pigs, chickens and other 19A TLU is a common unit used for describing livestock numbers of different species; this unit expresses the total amount of livestock present as a single value regardless of the specific composition. This is achieved by assigning conversion factors to different species to reflect their relative value. 38 poultry, and other non-specified livestock. Malawi has very low livestock ownership by regional standards. In 2005 the average TLUs in Malawi was 0.53 per household, or around 0.12 per capita (Figure 2.18). The non-poor have more TLUs (0.61) than the poor (0.43) and rural households have higher TLUs (0.53) than urban households (0.37). Households in the North region have the highest level of TLUs (0.96), a level which is three times higher than the South. 44. As expected, the TLUs increase by decile. There are a few exceptions, however. In urban households, there is no clear pattern except that the highest decile has the highest value of TLUs. This probably indicates that few urban households have the space needed to maintain livestock unless the household is wealthy enough to have land in rural areas as well. In the North region which has the highest average overall TLUs, the households in the 5th through 7th deciles have the largest TLU values, and the values for the higher deciles actually decrease. Figure 2.18: Average Livestock Ownership in Malawi (TLU/household by household type and residence), and in Southern Africa (TLU/100 people) Household livestock ownership Average livestock ownership in Southern by status and residence Africa (TLU/100 people), 2000-2002 1.0 e 250 0.9 191 0.8 200 peopl 157 0.7 150 0.6 ndexi 100/ 105 0.5 ex 100 79 LUT 0.4 45 0.3 ndi 50 25 0.2 LUT 9 8 0.1 0 0.0 i a e lawi mbia Poor or ral al otho bique Po uth Les Malaw Na Zambi babw Ma North entr Non- Urban Ru C So Botswana Mozam Swaziland Zim Note: Tropical Livestock Units conversion factors: oxen=1.0; cattle=0.7; small ruminants (goats and sheep)=0.10; pigs=0.20; poultry=0.01; rabbits=0.01; turkeys=0.10 Source: National Statistical Office, IHS2; and FAO GLiPHA (2003) Land Holdings20 45. In any country as highly agricultural as Malawi, ownership of land will play an important role in determining levels of poverty.21 Land holdings in Malawi are small: on average, excluding the landless, households have 1.2 hectares of land.22 When looking at land per capita, the average holdings are 0.32 hectares of land (Figure 2.19). Plot size per capita is highest in the North region where it reaches 0.41 per capita, while in the South and Central regions it is 0.29 and 0.33 hectares per capita respectively. Holdings of land per capita are almost twice as high for the non-poor households (0.40 hectares) than in poor households (0.23 hectares). Further, in 20 The data presented in this section exclude the landless households. 21 About 15 percent of households have no land holdings, largely in urban areas. 22 This calculation excludes landless households. Land categories include rain-fed plots, dimba plots, tree plots, plots rented out to others and uncultivated plots. Dimba gardens are pieces of land which due to proximity to some source of water (river or stream) retain moisture for most of the year, and can therefore be cultivated during the dry season. 39 the poorest decile land holdings are as low as 0.17 hectares per capita on average. Per capita land holdings increase as expected by decile, but even in the highest deciles, the average overall per capita land holding is only 0.53 hectares. 46. As can be seen in Figure 2.19, most of the per capita land holdings are in rain-fed land (Figure 2.19). Only few households have access to dimba plots. Interestingly some of the smallholder land remains uncultivated, especially in the top 20 percent of the population and in the North region. Figure 2.19: Land Holdings: Average Hectares of Land Per Capita 0.45 0.60 0.40 Uncultivated plots 0.35 0.50 Rented out plots 0.30 Tree plots 0.40 Dimba Plots 0.25 Rain-fed Plots 0.20 0.30 0.15 0.20 0.10 0.10 0.05 0.00 0.00 Malaw i Non- Poor North Central South 2 3 4 5 6 7 8 9 t Poor Poorest Riches Note: Excludes landless households. Source: National Statistical Office, IHS2 Labor availability and use 47. The IHS2 survey provides detailed statistics on time use patterns in Malawi. Defining work broadly to include income-generating activities (including work on the household farm) as well as main household chores (including fetching firewood and water), we examined typical labor supply patterns. Details of the methodology used in this analysis are provided in Wodon and Beegle (2005). 48. Details of the distribution of hours of work according to the type of work performed, and the gender, age and location of the individual are provided in Annex 2C. The results indicate that rural individuals work longer hours than urban individuals, and women work more than men. The mean working time year-round nationally is 36.4 hours per week for the adult population (above 15 years of age) and a much lower 8.5 hours for children. In rural areas, where 88 percent of the population lives, the mean values are slightly higher. 49. As expected, adult men spend more time on the labor market than adult women, essentially because of a larger average amount of time given to salaried work, as well as casual, part-time and ganyu work and non-agricultural business-related work. On the other hand, the differences between adult men and women in terms of the time spent on agricultural work are more limited on average (all values in the tables include zero values). As for domestic work, it is performed mostly by women, and the same holds for the collection of wood and water. In total, 40 BOX 2.4: LABOR SHORTAGES DESPITE UNDEREMPLOYMENT? SEASONALITY IN TIME USE The concept of time poverty can be used to measure the share of the population that works very long hours, and can therefore be considered as time-poor. In their paper on Guinea, Bardasi and Wodon (2005), consider a time poverty line of about 70 hours per week. A similar threshold has been used in our analysis to measure the share of the population working at least 70 hours per week. In rural areas, on an annual basis 5.2 percent of the adult male population works more than 70 hours per week, while the proportion is 10.3 percent for women (Annex 2C). Interestingly, there is no clear seasonal pattern in the share of the population working more than 70 hours per week, suggesting that the overall increase in working hours observed around December-January is likely to be provided by those household members that have a reserve of time at their disposal rather than by those who already work the most. While a small share of the population in Malawi can be considered as time poor according to the data in Annex 2C, a larger share can be considered as underemployed, at least in the case of men. On an annual basis, 15.6 percent of adult males work less than 10 hours per week, and this proportion peaks to more than 20 percent in some months. For women, the proportion working less than 10 hours per week is much smaller. Importantly, we do see the impact of seasonality in this measure of underemployment, since the proportion of adults working less than 10 hours per week is lowest again in December. The corresponding data for children suggest a much larger share with a small burden of work, but also some cases apparently of very high workload. Understanding the implications of these patterns will require additional analyses, but the results suggest that the precious few endowments of poor households (labor and land) may not be utilized in the most efficient way, or at least, it can be argued that there are serious constraints to the generation of higher earnings for households, despite the presence of underemployment for most of the year. Poverty reduction strategies would need to take into account the strong seasonal dimensions to labor supply to be effective. Source: Wodon and Beegle (2005) the mean and median working hours for women are about 10 hours above the corresponding values for men at the national level. 50. The results in Figure 2.20 highlight the presence of strong seasonality in time use. For the adult population, the average level of working hours is peaking in December-January, which is the busy part of the cropping season (see discussion in Chapter Seven). At that time, the adult population works on average more than five hours per week above the annual mean. The seasonal differential in working hours is largest for the individuals who belong to the poorest quintile of the distribution of consumption per capita. In rural areas, the additional workload in December versus the annual average amounts to close to 10 hours in the poorest quintile (see Chapter Seven). December is also the busiest month of the year for children (not shown). 51. As we expect gender and seasonality issues are more pronounced in rural households. The gender differences are even larger, such that for women the additional workload in December versus the annual average is 11 hours for the median, and 11.6 hours for the mean. The workloads for children are much lower, but girls do work longer hours than boys, again mainly due to a higher burden from domestic work as well as water collection. 52. Generally, labor in Malawi is assumed to be in surplus supply, with extensive under- employment. However, low mean hours in income-generating activities mask the existence of labor shortages at the peak of the cropping season. This seasonality in labor supply can have potentially large negative impacts on the ability of households to make the most of their 41 endowments such as land as well as their labor. The IHS2 data highlights the extent to which the seasonality in the demand for labor is leading to both underemployment and labor shortages (Box 2.4). Figure 2.20: Average weekly time spent working (adults, age 15 plus), by wealth quintile Seasonality in labor use 50 Poorest quintile 45 Total eekwr Richest quintile 40 pe s 35 ourH 30 25 Mar- Apr- May- Jun- Jul- Aug- Sep- Oct- Nov- Dec- Jan- Feb- Mar- 04 04 04 04 04 04 04 04 04 04 05 05 05 Source: National Statistical Office, IHS2 Household Enterprises 53. While not strictly household assets in the sense used above, household enterprises also provide a means to generate additional income. Only about one-third of the households in Malawi have household enterprises. There is no information in the IHS2 to explain why so few households have enterprises. But it is possible to infer that obtaining capital to start a business is difficult (Box 2.5). Of the households owning enterprises, 85 percent are found in rural areas. In both urban and rural areas, the majority of household enterprises are retail businesses (69 percent and 47 percent respectively). Manufacturing is the second largest category, and accounts for 13 percent of urban household businesses and 28 percent of rural household businesses. 54. The majority (63 percent) of enterprises are found in non-poor households, supporting the analysis in the Poverty Profile 1998, which stated that having a household enterprise is an important factor associated with higher welfare status.23 Again, retail and manufacturing were the major categories for the household enterprises. The Poverty Profile 1998 further speculated that the type of manufacturing that rural households engaged in was handicraft production which would be a seasonal activity undertaken outside of the cropping season. 23It is important to recognize as well that households may also be driven to diversification as a result of distress. If agriculture fails due to harvest failure, for example, households may turn to residual non-agricultural activities as a survival strategy. 42 BOX 2.5: ACCESS TO CREDIT (*) Only 12 percent of households reported obtaining credit in the 12 months prior to the IHS2 survey. For those who did not attempt to borrow, the largest reasons given (23 percent) was that the respondent didn't know any lenders. This was followed by believing they would be refused (12 percent) and getting a loan was too much trouble for what it's worth (11 percent). Recipients of loans have a higher education level than the general population. For those households that did obtain loans, only 30 percent used the loan for business start-up capital: 37 percent in urban areas and 29 percent in rural areas. In rural areas, an equal percentage used the loan to purchase agricultural inputs for food crops (28 percent). In urban areas, 37 percent used the loans to purchase non-farm inputs. Reasons for obtaining credit (percent) Malawi Urban Rural Purchase land 0.5 1.1 0.4 Purchase agricultural inputs for food crops 26.7 16.1 27.7 Purchase inputs for tobacco 18.1 2.2 19.5 Purchase inputs for other cash crops 4.3 2.2 4.5 Business start-up capital 30.1 37.5 29.5 Purchase non-farm inputs 17.9 36.8 16.2 Other 2.4 4.1 2.2 Source: National Statistical Office, IHS2 Note: (*) The IHS2 data does not allow a distinction between household that do not have credit due to refusal or lack of access to financial institutions (or intermediaries), and household that do not have credit due to lack of demand (i.e. no interest in taking a loan). Gender dimensions in labor and income earnings24 55. Gender-based differences in access to resources and bargaining power in Malawi suggest significant disparities in welfare between men and women. The previous sections have discussed various income and non-income dimension of poverty and have highlighted the existence of gender disparities. Here we focus on two additional gender dimensions of poverty: labor and income earnings. The poverty status of individual household members cannot be directly calculated from the information in the IHS2, because data about the distribution of consumption within the households is not collected. As such, as far as income is concerned, we can only infer the effect of an individual's gender on his or her income welfare from differences between male and female headed households. Agriculture and land holdings 56. Though there is a link between poverty and the size of land holdings, there are no significant differences between male and female-headed households in terms of the size of land holdings. Nevertheless, there is a disproportionately higher rate of poor among female- compared 24Annex 2A pools together information on gender inequities in Malawi. It discusses inequities in access to resources (notably in education, access to credit, and participation in income generating activities); the division of labor between women and men both in the labor market and in the domestic sphere; the empowerment of women both with regards to decision making; the victimization of women as measured by the prevalence of violence in general and especially domestic violence. 43 to male-headed households within small landholding sizes, due to the gender differences below. Further, widows have a much higher incidence of poverty than widowers. This could be attributed to property grabbing by relatives from the husband's side of the family, a widespread but undocumented phenomenon in Malawi. 57. Approximately 90 percent of all Malawian households can be labeled farming households, but broken down by gender, 95 percent of female-headed households farm compared to 88 percent of male-headed households. However, there are significant differences in crops cultivated by women and men, and in decisions made about agricultural tasks. Regardless of household size, female-headed households grow crops for home consumption to a greater extent than male-headed households, who are more likely to cultivate at least some cash crops. The most important cash crop in Malawi is tobacco, and this predominantly `male' crop, is grown in 19 percent of male headed households compared to just 7 percent of female ones. Moreover, for food crops such as maize, male-headed households are more likely than female-headed households to utilize higher yielding hybrid strains that require fertilizer for sale, rather than the lower yielding, seed-bearing strains chosen by women for domestic use. 58. While women hold decision-making power in female-headed households, in male-headed households there is a clear division: to the extent that women are involved in decisions about inputs and planting, their role is largely limited to crops that do not require fertilizer application, and where seeds are recycled. They make these decisions about 50 percent of the time, compared to just 10 percent of decisions where fertilizer is applied. For cash crops like burley tobacco, cotton and vegetables that require purchasing more inputs (fertilizer, seeds and pesticides), men make almost all decisions. 59. Provision of extension services to households is likewise skewed: only 8 percent of female- headed farming households obtained such advice compared to 14 percent of male-headed farming households. Based on the decision-making patterns above, it can be presumed that within a household, agricultural advice provided to men is not always passed on to their wives, furthering this gender gap. Labor and income earning activities 60. There is a clear disparity in the use of time between men and women. Women work longer hours than men. However, they spend considerably less time on income generating activities (17 hours per week compared to 27 hours for men). The difference is made up in domestic chores, which men devote just 3½ hours to a woman's 24½ hours per week. Actually this disparity is likely to be even higher because it does not include child care and tending for the sick, which are traditionally female tasks. Much of the domestic work includes heavy labor such as fetching firewood and water (taking up 1½ hrs and 1 ¼ hours each day, respectively). The extra female burden also extends to girls, especially after age 10. They spend 16 hours a week on household chores compared to 10 hours for their male peers. This burden has a negative effect on girls education: among dropouts, 37 percent of girls cited the need to work at home as the reason, compared to 23 percent of boys. 44 61. Wage employment is not widespread in Malawi's economy, but there are gender gaps both in remuneration for the same type of job, and for the types of jobs performed by men and women. The median daily wage for women was MK78, as compared to MK124 for men. For the highest paid and highest skilled jobs, men and women are remunerated roughly the same. At lower wage levels, however, women are paid less for working the same number of hours on the same task as men, notably in production activities, where women are paid MK45 compared to MK120 for men, and for laborers, where women are paid 48 compared to 70 for men.25 Part of this disparity can be attributed to different levels of education. These average figures mask a further disadvantage for women because of the seasonal nature of income generating opportunities during cropping time. One can expect that female-headed households depending on ganyu agricultural labor will be particularly exposed to food shortage and poverty, because of the lack of alternative opportunities the rest of the year. Moreover, the productivity of single farming women is reduced if they engage in ganyu to obtain some cash, rather than spending sufficient time in their own fields at cropping time, further increasing their vulnerability. 62. Overall, men were more likely than women to receive credit, though women were more than men likely to receive loans less than MK1000. The larger the loan, the likelihood that the recipient is a woman decreases. It should be noted that this finding may also be due to lower demand for credit by women (for instance because women may not have the same opportunities as men to open an enterprise).26 There are clear differences in the use for credit by gender. While women are most likely to use their loan to start up a non-agricultural business (more than 50 percent of women), men, on the other hand, were more likely to use credit for inputs for agricultural production, in particular for tobacco production. 63. Approximately 10 percent of women owned and managed their own enterprises, compared to 16 percent of men. Women spend on average 20 hours per week on their enterprise, compared to 29 hours per week for men, and women tend to generate less profit than men (MK160 per day compared to MK280). These differences might be the result of women spending less time, and thus accruing less skill and opportunity for further investment on their business. It could also be related to the lower education level of women than men, and from the types of enterprise. However, most enterprises for both men and women fall into `unspecified retail', so this effect could not be measured. One other explanation for low profitability among women enterprises may be that they tend to be risk averse and invest in low capital intensive businesses. Chirwa (2005) finds that most female small enterprises are mainly in food processing and beer brewing, and less common in non- food manufacturing and high skill enterprises. Access to roads, transport, and distance from markets 64. The IHS2 community questionnaire contains a few questions regarding access to roads and transport services, and distance from markets. Overall level of access to roads and transport 25These statistics on wage differentials need to be cautiously qualified as they are just averages which are based on very few observations, and also because these averages do not control for other aspects that lead to such differences. For example, there is need to distinguish between ganyu and salaried employment, and also the educational levels, skills, and level of experience have to be taken into account. 26As indicated in Box 2.6, the IHS2 data does not allow a distinction between household who do not have credit due to refusal or lack of access, and household that do not have credit due to lack of demand. 45 services is low. As expected, urban communities report much higher levels and quality of access. Many more urban roads are tar or asphalt compared to rural roads. One third of urban roads are tar/asphalt compared to only 13 percent of rural roads. Rural communities on average are located 20 kms from a tarmac road, and this distance is higher at about 40 kms on average in the North region (Figure 2.21). Figure 2.21: Distance to nearest tar/asphalt road and number of months road is passable Average distance to nearest tar/asphalt Average months nearest road is passable road (km) by minibus or lorry 45 12 40 10 35 30 8 25 6 20 15 4 10 2 5 0 0 Malawi Urban North Central South Malawi Urban North Central South Mini-bus Lorry Source: National Statistical Office, IHS2 65. Roads in urban communities are passable for most of the year (10 or 11 months, depending on type of vehicle) while roads in rural areas are impassable up to four months in the year. In the North region, on average roads are passable by minibus for only 5 months in the whole year. In the South region, on average less than 8 months. Clearly these averages hide greater variation within each region, and highlight the fact that many communities are extremely isolated from the rest of the country. 66. Almost two-thirds of urban communities have a bus stage in the community, while only about 40 percent of rural communities have bus stages. And the nearest bus stage in rural areas averages about 7 kms distant. 67. Using Geographic Information System (GIS) information about the road network in Malawi, we produced a variable that would express the distance of the household from the nearest Boma (district administrative center) or trading center. Our estimate, provide an indication of the remoteness of each community, taking into account the distance and the different types of roads which connect it to the nearest trading center (Figure 2.22).27 In practice, travel times will vary depending on household access to means of transportation. 27The variable is constructed by assigning travel speeds of 70 km/hour on primary road, 30 km/h on secondary road, 15 km/h on tertiary road, 10 km/h on sub-tier road, and 4 km/h on walking path. The estimate of remoteness is common to all households in a given community (IHS2 enumeration area). 46 Figure 2.22: Estimated household travel time to nearest trading center by wealth deciles Market access by decile: estimated average travel time to the nearest trading center 70 60 50 40 nutes Mi30 20 10 0 orest 2 3 4 5 6 7 8 9 t Po Riches Source: National Statistical Office, IHS2, and GIS information on road network Access to communications 68. The ability to communicate with communities and individuals outside of one's own community is limited in Malawi. Less than one percent of households have a working landline telephone and only about three percent of households have someone in the household who has a cell phone (Figure 2.23). As expected the percentages are much higher for urban than rural households and much higher for non-poor than poor households. In fact poor households basically have no telephone access. Overall, only 0.2 percent of rural households have a landline and 0.9 percent of rural households have a cellular phone. Virtually all phone owners, either landline or cellular, are in the highest two deciles, and are predominantly urban. 69. The IHS2 community questionnaire solicited information about the presence of a telephone service, either public or private, in the community. Three quarters of all rural communities have to travel more than 2 kms to find a place to make a telephone call, ranging from 67 percent of communities in the South region to over 80 percent in the Central region. Figure 2.23: Proportion of households with telephones (percent of households) Access to telephone Access to telephone by decile 20 22 18 20 16 Landline 18 Landline 16 14 Cellular 14 Cellular 12 12 10 10 8 8 6 6 4 4 2 2 0 0 2 3 4 5 6 7 8 9 t Overall Poor Non-Poor Urban Rural orest Po Riches Source: National Statistical Office, IHS2 47 BOX 2.6. SUBJECTIVE VERSUS OBJECTIVE POVERTY Quality of life, happiness and well-being are broad, multi-dimensional concepts that include not only material achievements but also other aspects, such as health, respect of others, employment, and having children. A special section of the IHS2 includes questions about the subjective wellbeing of each household, to check if objective economic indicators (income or expenditure) fall short on fully assessing satisfaction with life.* Not surprisingly, the actual level of per capita consumption expenditure is a strong determinant of subjective perceptions about consumption adequacy: objectively better off households also feel richer. And the subjective perception of the minimum income level is remarkably consistent with the per capita consumption poverty line we estimated in Chapter one. One of the questions in the survey asks for the household's own assessment of the "poverty line", by asking the minimum income question (MIQ) in the following form: "What income level do you personally consider to be absolutely minimal ­ below which you could not make ends meet?" Though there was considerable deviation in the responses (Std. dev. = 18110), on average, households perceive the minimum income needed to meet food and non-food needs to be MK16,600 per year--very close to the poverty line of MK16,165. A range of variables predicts both objective and subjective poverty in the same way. These include land cultivated, number of heads of livestock, amount of durable assets, the amount of ganyu performed, and the type of dwelling. Income poor households are more likely to report expenditure on food and clothing as inadequate: approximately 75 percent of households from the lowest deciles of expenditure distribution consider their expenditure on food as inadequate, compared to 35 percent in the top deciles. Similarly, poor households are much more likely to categorize their expenditures on clothing as inadequate compared to better off households. In contrast, the share of households who felt that their health expenditures were inadequate stays almost constant up to the 60th percentile of expenditure distribution. In other areas, there is more of a discrepancy between objective and subjective poverty. While subjective perceptions of economic well being are closely tied to consumption expenditure, the relationship is not fully proportional to a household's income. Controlling for other factors, larger households are more likely to be monetarily poor, but are less likely to feel poor. Even though larger households, on average, have less income, they do not associate household size with poverty, but rather, derive a sense of belonging, support and care that feeds positively into their perception of well being. Similarly, households with a larger share of adult females report higher levels of well being. Conversely, though single adult households have poverty levels among the lowest of all households, at least 40 percent of them perceived consumption inadequacy in all categories. Households that own an enterprise, and households that are headed by an individual with a diploma or degree are less likely to feel poor than a household of similar means without these attributes. Other significant factors explaining differences between subjective and objective poverty are polygamy, female household headship (both polygamous and female-headed households feel better off), tenurial type, and language group. Interestingly, perceptions of consumption adequacy also vary geographically. Rural households rank their wellbeing lower than similar households from urban areas. But within rural areas there is also variation: though the rural Southern region is both objectively and subjectively the poorest, the rural Northern region, which objectively is second poorest, has lower perceptions of poverty than the Central region. In other words, people in the North region feel richer than people in the Central region, even though the opposite is true when we measure actual consumption. As we have seen, this can partly be explained by a higher availability of social services in the North. *The findings reported in this box draw on a report for DFID by IDS using the IHS2 data, "Vulnerability to Chronic Poverty and Malnutrition in Malawi," by Devereux et al. 2006, as well as our own analysis which is explained in greater detail in Annex 2D. 48 THE DETERMINANTS OF POVERTY IN MALAWI IN 2005 70. The poverty profile completed above is a descriptive tool that provides key information on the correlates of poverty, by comparing the poverty status of a particular household or individual to selected characteristics of that household or individual. Though insightful, such a bivariate exercise is limited in its usefulness because it shows how poverty levels are correlated to one characteristic at a time, and in so doing, tends to simplify complex relationships. The `determinants' of poverty analysis goes beyond the simple bivariate poverty profile to consider the correlates of poverty in a multi-variate context. 71. There have been two previous attempts to model the determinants of household welfare in Malawi. The first attempt was done by modeling the determinants of smallholder incomes in rural Malawi using the 1992-93 National Sample Survey of Agriculture (NSSA) data (World Bank 1995, pp. 48-49). However, the model used by the NSSA was limited to rural smallholder households, and considered income levels rather than the consumption-based household welfare indicator used here.28 The second study is "The Determinants of Poverty in Malawi, 1998", based on the analysis of the 1998 IHS1 survey data and carried out by the NEC, NSO and IFPRI in 2001 (NEC et al, 2001).29 Here, we update the work carried out in 2001 using the 2005 IHS2 data. Our model relies strongly on the earlier effort to estimate the determinants of poverty in Malawi using 1998 household data (NEC et al., 2001).30 Whenever possible, we compare our results with the findings of the 1998 Determinants of Poverty study. Modeling the determinants of poverty 72. Our approach to assessing the determinants of poverty in Malawi is based on modeling the natural logarithm of per capita consumption of survey households. In other words, our choice of dependent variable, that is our household welfare indicator, is the logarithm of total annual per capita consumption and expenditure reported by a survey household. The model can be specified as follows: [2.1] ln cj =xj + j where cj is total annual per capita consumption of household j in Malawi Kwacha (MK); xj is a set of exogenous household characteristics or other determinants, and j is a random error term. 73. The set of explanatory variables that are hypothesized to determine of consumption includes household and community characteristics. We avoid using variables that may determine living standards but also be simultaneously determined by current income (endogenous variables). Our objective is to select regressors whose values are determined outside the current economic system of the household, but which determine the level of 28In this earlier study, eight household variables, plus Agricultural Development Division fixed-effect variables, make up the final model. The most important determinants of smallholder incomes were found to be the amount of cultivated land (positive), household size in adult equivalents (negative), and gender of household head (negative if female). 29Hereafter referred to as Determinants of Poverty in 1998. 30This model is also documented in Mukherjee and Benson (2003). 49 household welfare (exogenous variables).31 Our selection of potential determinants is guided by the results of the poverty profile presented in the previous sections, as well as by those variables known to be of considerable interest to Malawian policy makers. 74. One essential point to note is that we do not determine causality here through this analysis. Rather, we build our model of consumption expenditure based on an understanding of economic theory, and we select variables that economic theory says are likely to be exogenous. We then quantify and interpret the relationship as causal. Thus, our causality hypotheses are guided by economic theory. The most our empirical model can do is test this body of theory. 75. In order to minimize problems associated with `omitted variable bias' we adopt a very broad specification of the consumption model, including as large a number of variables as possible (which we can ignore ex-post, if the statistical analysis indicates that they have no significant impact on consumption). It should be noted, however, that this approach towards model selection does not entirely free us from worries about omitted variable bias. As a result the degree to which this analysis can shed light on the determinants of living standards should not be overstated. 76. The set of regressors, or independent variables, that we chose as possible determinants of poverty in Malawi are listed in full in Annex 2E. Broadly, they may be categorized as follows: 77. Demographic: These variables aim to capture the basic demographic characteristics of the household, including the sex of the household head, the age of the household head, whether the household head is a widow, the total size of the household, the number of children. 78. Education: We included measures to capture the highest educational attainment of the household head. Specifically we distinguish between households whose heads has some primary schooling, or has completed primary, or has some post primary schooling. 79. Employment and occupation: In this category we sought to capture the effects of the distribution of different sorts of occupation at the household level. The variables used include whether the household head is engaged in formal wage employment, and/or whether the household runs a non-farm enterprise. 80. Agriculture: We also included variables to account for whether the household farms had any rain-fed plots, the total per capita landholdings of rain-fed land held by the household, 31For instance, the educational level of the head of household is an exogenous variable when examining household welfare, since it is determined by actions that are unrelated to the welfare level of the current household of which he or she is the head. The education level of the household head is likely to be an outcome of the past welfare status of his or her parent's household rather than of the current welfare status of the household. In contrast, the quality of roof under which the household sleeps is an endogenous variable when examining household welfare. It is only households with higher welfare levels that one would expect to have metal roofs. That a household has a metal roof is directly a function of its current welfare status, i.e., roof type depends on the level of household welfare. Other endogenous variables that are likely to be an outcome of current household living standards (as measured via consumption levels) include the possession of durable goods by household members, dwelling characteristics, current school attendance of children in the household, and so on. 50 whether the household has a dimba plot, and whether the household grew tobacco cultivation (in the last cropping season). 81. Community characteristics and access to services at the community level: We included variables to examine the impact of the existence of a regular bus service to/from the community, and we also controlled for the presence of a health clinic and bank in the community. We also accounted for different access to markets by including a dummy if the household is in a Boma (District administrative center) or Trading center, and checked for the presence of an ADMARC market and a daily market. We use the GIS-based access-to-market variable described earlier to express the distance of the household from the nearest Boma or trading center (in several categories: >20-30mins, >30-45mins, >45-60mins, and >60mins). Finally we included a dummy for the presence of a tarmac/asphalt road in the community. 82. Regional fixed effects variables: Regional dummies have been include to captured fixed effects based on the 3 main regions of the country (North, Central and South), and urban. Results of the analysis 83. Detailed results of the estimated regressions are presented in Annex 2E. The regression uses log of per capita consumption, and consequently the coefficients of the regression can be interpreted as partial effects measured in percentage terms. These results are depicted in Figure 2.24 and summarized below. For all estimated coefficients that are statistically significant, the figure shows the percentage impact on per capita consumption of a change in each household or community characteristic considered in the regressions. Whenever possible, the discussion also highlights any differences from the results of the earlier determinants of poverty study which used the 1998 IHS1 survey (NEC et al, 2001). 84. The results confirm that female headed households are substantially poorer than male- headed households. Holding all other variables constant, a female-headed household has 14 percent less consumption per capita than a male-headed household. This result is very significant in all rural areas, but does not appear to hold in urban areas. This contrasts with the findings of the 1998 poverty profile, which highlighted a "puzzling result" that male-headed households appeared to be poorer than female-headed households in the South region (NEC et al. 2001, page 21). 85. Households whose head is aged between 26 and 45 years appear to be richer by around 7 to 9 percent (compared to household heads aged 18-25). At other ages, the age of the household head is not significant, except for those household heads 56 or more years of age in most rural areas and in urban areas. When the head is 66 or more years of age, per capita consumption decreases by 8 percent overall (again compared to household heads aged 18-25). Unlike the findings of the 1998 determinants of poverty study, we find that this relationship is more pronounced in urban areas. 86. Households headed by a widow appear to be better off by 6 percent on average. The impact is larger and significant in the North region, but appears not significant in other regions. The difference may be due to the fact that in the North, following the death of the spouse, 51 independently of the gender, the widow generally has the choice to keep control of the household`s land holdings and assets. In the other regions, however, control of the assets depends on the gender of the surviving spouse. Figure 2.24: The correlates of poverty in Malawi in 2005 (percentage change effect) Female household head Age of household head: 26-35 years Age of household head: 36-45 years Age of household head: 56-65 years Age of household head: 66+ years Widowed household head Household size Number of children 0-4 Number of children 5-10 Number of children 11-14 Highest education: some primary Highest education: completed primary Highest education: post primary Household head has wage/salary Household has a non-farm enterprise Ln total hectares of rainfed plots Household had any dimba plot Household head grew tobacco in last season EA is a Boma or Trading center Travel to nearest boma: >30-45mins Travel to nearest boma: >45-60mins Travel to nearest boma: >60mins Tarmac/asphalt road in community Health clinic in community ADMARC market in the community North region Central region Urban -50% -40% -30% -20% -10% 0% 10% 20% 30% 40% 50% <== Poorer Richer ==> Note: OLS regression on log of per capita consumption. Only statistically significant results at 10% or lower are shown. Omitted categories are: age of household head less than 26 years, education of head is zero, travel to the nearest Boma is less than 20 minutes, regional dummy for the South region. Each bar in the graph can be interpreted as the percent change in per capita consumption associated with a unit change in that variable. Source: Own calculations based on IHS2 52 87. In line with the poverty profile analysis, household size has a highly significant negative correlation with per capita consumption. As household size increases, per capita consumption decreases by almost 28 percent. Household size has the least negative effect in the North region where increases in household size decrease per capita consumption by 22 percent. The household size squared is shown to be significant and positive, which suggests the possibility of some economies of scale of household welfare derived from increasing household size. 88. The number of children is negatively related to consumption in the household. The effect of children 0 to 4 years of age in the household is the strongest, with the overall results showing an 8 percent decrease in per capita consumption. The effect is strongest in urban households where the presence of children 0 to 4 decreases per capita consumption by 14 percent. Interestingly, in rural areas children older than 10 years do not affect consumption negatively, possibly because they start to contribute early to productive household activities. 89. Consistent with the results of the 1998 study (as well as numerous studies on the determinants of poverty in other countries), we find that the education of the head of the household is positively related to consumption and is highly significant. As the education of the household head increases, the coefficient also increases. For example, overall, having some primary education adds 5 percent to per capita consumption, having completed primary school adds 12 percent to per capita consumption and having more than primary adds 40 percent to per capita consumption. As expected, the impact of post primary education is larger in urban areas, where there is more demand for such training. 90. Participation in wage occupation is strongly associated with higher per capita consumption, by about 12 percent on average, and much higher in rural areas where such opportunities are rare. Similarly, households that have a non-farm enterprise are strongly associated with higher expenditures by about 14 percent on average, and much higher in the North region (22 percent). 91. In terms of agricultural activities, household per capita consumption increases by 8 percent with each additional hectare of rain-fed land. Given that landholdings of rain-fed land are approximately 1 hectare on average, this suggests that even doubling of landholdings would increase incomes by only 8 percent approximately. Per capita consumption of households that own a dimba plot is higher by 7 percent on average, highlighting the benefits of access to irrigated land that can be cultivate in the dry season. Households that grow tobacco (the most common cash crop) tend to have higher per capita consumption by 9 percent on average, reflecting the gains from the participation in cash crops production. 92. Distance from markets is an important determinant of poverty. Households located in a Boma32 or Trading Center on average have per capita consumption higher by 15 percent. As expected, this effect is much larger in the North region, which is the most remote and least connected region, where per capita consumption in a Boma or trading center is higher by 24 percent. In the Central and South regions, the effect on per capita consumption is about 9 percent and 6 percent respectively. 32District administrative headquarters. 53 93. If the household is located more than 30 minutes away from the Boma, the household's level of consumption per capita will be lower by at least 10 percent. Again this effect is greater in the North region. There are two puzzling results, however. Firstly, the negative impact of distance appears to be greater for households which are located between 30 and 45 minutes away from the Boma, than for households which are more than 45 minutes away. Secondly, distance from the Boma appears to have a positive impact on consumption per capita in the Central region. These puzzling results may reflect a problem with the construction of the access variable. 94. A tarmac road in the community has a significant positive effect on per capita consumption. Overall, it increases per capita consumption by 13 percent. The largest positive effect is seen in the urban areas and in the Central region where per capita consumption is increased by almost 44 percent. However, in the North region, a tarmac road in the community decreases per capita consumption by 20 percent. This result seems counter-intuitive especially in light of the positive results for the other regions and is difficult to explain. As noted above, if the household is located in a Boma in the North region, the effect is the strongest on per capita consumption. This result may be an artifact of the data due to the fact that of the households in the North region, only 11 percent have tarmac roads in the community, leading to a very small number of relevant observations on which to make inference. 95. Having a health clinic in the community is associated with 7 percent higher consumption per capita on average, although this effect is not significant in urban areas (and has a negative sign).33 Note that this relationship may be due to the fact that clinics are placed in relatively richer villages, rather than because of a positive impact of clinics on spending. Our regression analysis cannot distinguish between these two effects. 96. The presence of an ADMARC market in the community is associated with lower levels of consumption per capita by about 4 percent. Hence, the results indicate that the presence of ADMARC tends to be associated with higher levels of poverty. It is important to note that ADMARC had not been functional for the in the 4 years prior to the survey. ADMARC had not been purchasing maize due to its cash flow problems and had participated in selling maize only on occasional basis (as part of government financed food security initiatives). The impact of ADMARC's lack of regular operation, coupled with its negative effect on the presence of private traders, appears to be perpetuating poverty. 97. The fixed effects variables in the multi-variate model tell us that the North is the most disadvantaged area to be located. In interpreting these coefficients, however, it is important to realize that, when estimating the regional-effect coefficients, the model (`controls for' and) removes the effect of all other explanatory variables. In other words, the results indicates that all other characteristics being the same, a household resident in the North is likely to be poorer than 33 This is the only result connected to the health clinic that is either negative or not significant. Sixty-four percent of all urban communities have no health clinic and 100 percent of poor urban communities have no health clinic. Once again, the negative effect is probably not so much the presence of the service that is driving the effect rather it is the poverty of the communities that is being reflected in the results. 54 a household living in other regions; on the other hand, that all other characteristics being the same, a household located in urban areas or in the Central region is likely to be richer.34 34In reality, when assessing the level of welfare associated with living in a given region, we observe the impact of the many variables affecting the level of welfare, such as differences in the levels of human capital and the natural resource base in the different regions. Consequently, the regional coefficients estimated by the model do not coincide with the overall welfare conditions observed in the various regions (which have been described in the poverty profile), since the latter include the impact of differences in the other determinants. In fact, the results of the poverty profile showed that the overall incidence of poverty is higher in the South region and is lowest in the Central region. 55 CHAPTER 3: RISK AND VULNERABILITY TO SHOCKS IN MALAWI INTRODUCTION 1. Poverty is not a static concept. It includes a stochastic dimension. That is to say, it is always somewhat uncertain who will become poor, and when, and accordingly, eradicating poverty entails both the identification of who are the poor today, as well as ways to protect households that may become poor in the future. As such, a sound poverty reduction strategy must incorporate two distinct elements: (i) poverty alleviation programs to mitigate the adverse effects of current poverty, and (ii) poverty prevention programs that reduce the risk of becoming poor. 2. This chapter reviews the concepts of income poverty in Malawi, making a distinction between chronic and transient poverty, and the roles of risk and strategies to cope with shocks. Given the methodological difficulties and data limitations we do not try to measure chronic poverty and economic vulnerability directly. Rather, the approach is to examine: (i) who the ultra-poor are (the poorest households with arguably the highest probability of being chronically poor), (ii) the sources of risk in Malawi, (iii) the ways in which risk contributes to Malawi's high and persistent levels of income and non-income poverty, and (iv) what actions households take to cope with risk and shocks, ex ante and ex post. 3. The chapter is structured as follows. The next section highlights the role of risk and shocks as a cause of poverty, and the dynamic nature of poverty in Malawi. The following section reviews the prevalence and characteristics of risk and shocks in Malawi. We then discuss the concepts of chronic and transient poverty, and highlight some of the broad characteristics of members of these groups. The final section discusses the main coping strategies adopted by households. RISKS AND MOVEMENTS IN AND OUT OF POVERTY Risks and shocks as a cause of poverty 4. Individuals and households confront a number of barriers that cause many to fall into poverty, and undermine attempts to escape poverty it. Some of these barriers are predictable and known, perhaps linked to past and present policies, institutions, and structural features of the economy. Others are linked to adventitious shocks and unexpected adverse events that impact on individuals, on households, and on the wider community. Shocks such as illnesses, injuries, deaths, employment losses, crop failures, thefts, and droughts can be major set-backs to households, keeping them poor or, for the non-poor, pushing them into poverty. 5. It follows that among the poor, some fraction are living in poverty as the result of fluctuations in economic status attributed to adverse shocks. As such, risks and shocks are important determinants of poverty dynamics and growth, and it is important to understand the nature and frequency of shocks and, in particular, the varying coping strategies used to deal with these shocks. Risk affects the ability of households to sustain assets and endowments, as well as 56 BOX 3.1: KEY CONCEPTS OF VULNERABILITY, RISK AND SHOCK Risks are potentially dangerous events that are likely to cause economic loss or damage when they occur, while shocks are the actual occurrence of a risk. Although poor households may be more likely than the non-poor to be exposed to risk (for lack of ex ante options like insurance and income diversification), there may be some risks which are commonly and widely distributed within Malawi across socio-economic groups. Vulnerability implies the susceptibility of individuals, households or communities to the negative impact of events or shocks (for a review of the concept of vulnerability, see, among others, Hoogeveen et al. 2004). Shocks are often classified by the extent to which they co-vary within communities versus the extent to which they are idiosyncratic. Using these two extremes, we can think of various shocks as being arranged along a continuum. Covariate shocks such as drought and floods are those that simultaneously affect a large number of households typically in close geographic proximity to one another. Idiosyncratic shocks consist of fairly household-specific problems or crises, such as serious illness or unemployment of a household head. Of course, the same type of shock may be more or less covariate or idiosyncratic depending on the details of time and place. For example, adult morbidity or mortality can be idiosyncratic in the case of rare illness or covariate in case of epidemics. HIV/AIDS may affect several households within one family, which is some sense makes it covariate, in the sense that households may not be able to receive assistance from their main traditional network of support (e.g. family members) since those households have also suffered from the event. While idiosyncratic shocks can be singularly devastating, covariate events can be even more difficult to cope with as households may not be able to gain assistance from traditional support networks, other households, which are also affected. In a severe drought year, for example, small cultivators will not only lose their own crop, but will also find less work in other's fields. In these situations, ganyu may be an important coping mechanism for idiosyncratic shocks but not covariate events as its availability depends on the community-wide economic situation. Even when shocks are covariate, they may affect households differentially. The poor may already farm marginal lands which are more sensitive to rainfall deviations, or more prone to flooding. Families whose household members are already malnourished due to poverty may suffer more illnesses, deaths, and disabilities from a community-wide shock than wealthier and healthier households. Vulnerability can vary geographically, depending on the nature of risks and the resources available locally (within communities or district governments), but can also vary across types of individuals and households. By identifying vulnerable groups, those that are deserving of special assistance can be targeted effectively.(*) Vulnerability can imply lower consumption and increased poverty, but it can also have very important long-run implications is vulnerability is managed by compromising future income earning potential, such as selling off productive assets, reducing human capital investments (such as child schooling), and avoiding new investment opportunities. The ability of households to reduce or prevent vulnerability depends on three broad areas. The first is severity and frequency of risks facing households. The second concerns the level of the household resources which can include capital (financial assets as well as physical capital such as land and livestock). The third regards access to social networks (family, friends, neighbors, community associations, markets, etc.) and public programs.(**) All of these factors influence the ex ante and ex post coping strategies adopted by households, as well as the overall impact of negative shocks on households. Attempts have been made in some settings to use information from household surveys to calculate an indicator of vulnerability, similar to the poverty headcount measure. However, in practice, measuring vulnerability--the high exposure to risk combined with limited capacity to manage risks--is difficult to quantify. We rarely know the full set of risks that the household faces, what strategies and resources they can use to manage these risks, and what would be the expected economic or other losses in the event of an insured shock. Notes: (*) Often the term "vulnerable groups" is used to describe individuals or households characterized by exceptionally low levels of income or high levels of poverty (those in a state of being helpless, weak or otherwise "excluded" groups). The identification of these vulnerable groups can then be used to target the poor by serving as a proxy for detailed income data which is usually costly or unavailable for the entire population. This is not equivalent to a definition of "vulnerability" defined as households with high risk exposure and limited coping strategies. (**) The second and third groups are broadly consistent with the five types of capital outlined in a livelihoods framework: natural, physical, financial, human, and social. 57 the transformation of assets into incomes (activities). Understanding how risks are managed by households, communities and the public sector in Malawi can open a window to understanding critical underlying processes that have contributed to the country's high poverty levels and stagnant growth levels. 6. As will be discussed later in this chapter, risk can cause poverty and failing growth through both ex ante risk-avoidance strategies (behavioral impact), and ex post coping with the impact of shocks. Households with uninsured risk may adopt ex ante coping strategies such as avoiding profitable but risky opportunities. That is, they avoid high-risk, high-return investments in areas considered to be drivers of economic growth, such as new crop varieties or new cash crops. Instead, households enter into low risk, low-return activities or invest in low risk assets, which can then result in a poverty trap (see, for example, Dercon, 2002, among others). 7. For many households affected by a shock, some ex post strategies adopted to cope in the short-run may have detrimental long-term implications. If households draw down critical productive assets, they may end up perpetually trapped in poverty. Other forms of coping may have intergenerational effects, such as pulling children out of school and reducing meals, which then compromises the health of children and reduces future productivity, thus making them vulnerable to poverty traps (see, for example, Alderman et al., 2001, Hoddinott and Kinsey, 2001). Using long-term panel data tracking children into adulthood, Alderman et al. (2006) show that drought-induced early childhood malnutrition has long-term consequences. Beegle et al. (2006) find a long-run impact of parental deaths on education and health in adulthood. Shocks can also lead to more severe forms of coping such as prostitution. 8. By extension, risks and shocks can have significant impacts on overall economic growth. Dercon (2004) shows that uninsured risk among households in Ethiopia, mainly rainfall shocks, have substantial impact on consumption growth which persist for many years, and translate into lower growth rates overall. How exposure to risk affects economic growth is a key issue in development. A recent article quantifies both the ex ante and ex post effects of risk using long- running panel data for rural households in Zimbabwe (Elbers, Gunning, and Kinsey, 2007).55 The key finding is that risk substantially reduces growth in this particular setting: the mean capital stock in the sample is (in expectation) 46 percent lower than in the absence of risk. About two-thirds of the impact of risk is due to the ex ante effect (that is, the behavioral response to risk). These results suggest that policy interventions that reduce exposure to shocks or that help households manage risk could be much more effective than is commonly thought. The dynamic dimension of poverty 9. Absolute poverty remains high in Malawi, with nearly 6.4 million Malawians (52 percent of the population) living below the national poverty line, and about one-fifth of the population living in ultra-poverty. Poverty is persistent and self-perpetuating, but far from static, as we might expect given the pervasive nature of risk and vulnerability. When compared to the IHS1 data for 1998, some districts experienced a reduction in poverty, while others saw increases in poverty: although the overall levels of poverty remain stagnant, the ranking of districts has 55The study uses a simulation-based econometric methodology to estimate the structural form of a micro model of household investment decisions under risk. 58 changed. Even at local levels, there are important poverty dynamics, with some households moving out of poverty while others fall into poverty. 10. The best measure of living standards dynamics are constructed from panels surveys of households, where economic status is measured at two or more points in time. One such panel study tracked 291 rural households, finding that almost two-thirds of the poorest 20 percent of households in 1998 had moved into a higher income quintile by 2002 (Sharma et al. 2002).56 By implication, since overall poverty has remained steady, two-thirds of those who were not so poor in 1998 had fallen into the poorest quintile in 2002. Preliminary findings from the Moving Out of Poverty ongoing World Bank study also confirm these findings (see Box 3.2). BOX 3.2: MOVING OUT OF POVERTY - UNDERSTANDING GROWTH FROM THE BOTTOM-UP In 2005 the World Bank started a worldwide, multi-country study on how households and communities move in and out of poverty. Malawi is one of the case study countries, with the work conducted by the Center for Social Research of the University of Malawi, and IFPRI, in collaboration with World Bank staff. The policy focus of the Malawi study is to examine the impact of changes in access to basic infrastructure (markets, roads, health services, water, schools, etc.) for household economic growth. The study is based on predominantly qualitative analysis of data from an appositely designed questionnaire sampling 15 communities across Malawi, and also some quantitative analysis using data from the five Complementary Panel Surveys (follow up panel surveys on the IHS1). Data collection was completed in 2005, but the data analysis is still in progress and a draft report is expected in Summer 2006. Preliminary results are consistent with the findings of this poverty assessment. Even though poverty levels have not changed much over the past decade, there has been a large amount of movement of households in and out of poverty (see Figures below). Limited access to inputs is generally regarded as a major constraint to moving out of poverty. Cycles of hunger, mainly resulting from poor weather and low use of fertilizer, are also perceived to trap people in poverty. In this context, communities where access to markets and services has improved, have prospered. Also, areas actively serviced by NGOs have prospered over the years. At the household level, factors leading to downward mobility include natural disasters, distress sales of livestock/assets, HIV/AIDS and chronic illnesses, death of spouse (particularly a husband), and alcoholism. Factors leading to upward mobility include possession of livestock or assets, crop diversification, participation in cash cropping, venturing into small-scale businesses, building up savings, having multiple sources of income, and remittances from working children/relatives. Household movements in and out of poverty Changes in self-reported over the past decade (subjective) poverty 70 over the past decade Dropped 80 ds 62 60 61 Never moved ehol 50 Moved Up 43 60 us 40 35 35 ho 35 entc 29 31 40 of 30 27 21 erP 20 12 20 10 11 umberN 0 0 Poor (n=441) M oderately Better-Off Total (N=658) Poor (n=151) (n=66) Now 10 Years Ago Households status in 2005 Poor Moderate Better-Off Source: Preliminary findings from 2006 "Moving Out of Poverty" study (World Bank, 2007) 56It is presumed that some share of the economic mobility is actually reflecting measurement error in consumption which will result in over-estimates of transition rates. 59 11. Households in the IHS2 also report that living standards are quite dynamic and can vary across years. One-third of all households reported that their economic well-being had not changed in the last year, whereas the remaining two-thirds reported a change; 25 percent of all households reported an improvement in the last year and 43 percent reported a decline in well- being. 12. Another way to view the dynamic nature of poverty is to examine the response of poverty rates for a given change in consumption levels for all households (a distribution-neutral change in consumption). Figure 3.1 charts this relationship for Malawi. If every household experienced an increase in consumption of 20 percent, the poverty rate would fall from 52 percent to 40 percent of the population. On the other hand, a decrease in consumption of 20 percent would be associated with two-thirds of the population living in poverty (66 percent poverty rate). This emphasizes the point that the poverty level in the country is very sensitive to changes in consumption, modeled here under simplistic conditions. Another way of looking at the same concept is to note that there are millions of Malawians who live just above the poverty line threshold, and could be forced into poverty by even slight misfortune. Figure 3.1: Cumulative Density Function of per capita consumption associated with consumption increases/decreases of 20 percent 1 .8 ecp =