Report No. 15437-MAI Malawi Human Resources and Poverty Profile and Priorities for Action March 19, 1996 Human Resources Division Southern Africa Department Africa Region Document of the World Bank * 0 0 0 0 0 0 0 e e 0 0 0 0 0 0 0 0 0 0 0 a 0 0- 0 a e e a IM a Ia v im Human Resources and Poverty Profile and Priorities for Action November 1995 This report was prepared for the Government of Malawi by a team from the Human Resources Division, Southern I w The World Bank Africa Department of the World Bank. l Southern Africa Department ACKNOWLEDGMENTS This report has been prepared at the request of the Government of Malawi as the World Bank's input for their Poverty Alleviation Program. The analysis builds on the existing literature and on the insights of many people with whom the Bank has had extensive con- sultations during various visits to Malawi since October 1994. The report is based on data sets made available by the National Statistical Office in Malawi to the World Bank team in October 1994. The scope of the report was reviewed by both government personnel and the research community in Malawi. An earlier draft of the report was presented at several workshops with policy makers, researchers and donors in Malawi in July 1995. This report was written by a team including Helena Ribe (team leader), Ingeborg Astrid Kleppe, Jeff Alwang and Trina Haque. Jeff Alwang conducted the analysis on rural smallholders and Florencia Castro-Leal conducted the analysis on education. Additional work was prepared by Ingeborg A. Kleppe and Venanzio Vetla (human resources, popula- tion, and nutrition), and by Jeff Alwang, Trina Haque, and Paul Siegel (poverty measure- ment, rural poverty and literature review). Trina Haque and Meera Venkataraman devel- oped the statistical methodology for analyzing the HESSEA data. Jeff Alwang and William S. Brown developed the statistical methodology for analyzing the NSSA data. Fiona Macintosh and Leo Demesmaker edited the report. Vinnette Gordon provided secretarial support. The team thanks: S. Kakhobwe (PAP); A. Gomani, M. Kutengule B. Lodh, A. Chulu (EP&D); W Chilowa, S. Khaila (CSR); C. Mataya, M. Zeller (Bunda College); L.F. Golosi, T. Konyani, C. Machinjili (NSO); R. Ayoade (MoAG); E. Chisala, N. Hahn (UNICEF); P. Peters (HIID); and representatives from Chancellors College and FEWS for their valu- able comments. The team also received useful comments from the following World Bank staff: D. Bruns, R. Grawe, B. Kafka, K. Kostermans, H. Schaeffer, M. Salim, and G. Tidrick. The peer reviewers were: K. Krumm, J. van Holst Pellekaan, and P. Lanjouw. Maps appearing in this report are for the convenience of readers. The denominations used and boundaries shown do not imply any judgement on the legal status of any territory or any endorsement or acceptance of such boundaries. A summarized version of this report may be obtained from: Internal Documents Unit HB1-151 Ext. 34641 GLOSSARY ADD Agricultural Development Division ADMARC Agricultural Development and Marketing Corporation AE Adult Equivalent AIDS Acquired Immune Deficiency Syndrome COL Cost of Living Index CPIs Consumer Price Index CSR Centre for Social Research DHS Demographic and Health Survey EA Enumeration Area EP&D Ministry of Economic Planning and Development FEWS Famine Early Warning System GDP Gross Domestic Product GNP Gross National Product HA Hectare HESSEA Household Expenditure and Small-Scale Economic Activities HIID Harvard Institute for Inrernational Development HIV Human Immunodeficiencv Virus IMF International Monetary Fund Kg Kilograms LC Lower Cutoff Line MCH Maternal and Child Health MDHS Malawi Demographic and Health Survey MK Malawiani Kwacha MoAG Ministry of Agriculture MOE Ministry of Educationi NSO National Statistical Office NSSA National Sample Survey of Agriculture PAP Poverty Alleviation Program RDP Rural Development Project SDA Social Dimensionis of Adjustment UC Upper Cutoff Line UNICEF United Nations Children's Fund m ~~~~~~~~Malawi * Profile and Priorities for Action Priority Indicators of Poverty Population Total population ..................... 11 million Health Population growth rate .... 3.3% annually Mortality rates Population share: Rural .................... 90% Urban ...........8% Infant mortality rate 1988-1992 ... 134.3 BOMAs .. 2% 1978-1982 ... 136.4 (Sources: HESSEA 1990/1991, WDR) Child mortality rate 1988-1992 ... 114.9 1978-1982 ... 140.8 Population Classified Maternal mortality rate 1986-1992 ... 620.0 as "Poor" Infant mortality number of cnidren dying before their Percent, By Rural Development Project Area first year per 1.000 live oirths (Incomes below 40th percentile) Child mortality = number of ch dren dying between 12 and 59 months per 1,000 livF. births Maternal mortn itv y maternal morta ity rate d vided by general ferti ty rate per 100.000 live births. Nutritional status of children under 5 years old Stunted growth ........... 49% 4-25% Underweight ........... 27% 26-35% (Source: MDHS 1992) 36-45% 46-60% F.4 Missing or incomplete data (Source: NSSA 1992/1993) Indicators Education Enrollment Net primary Quality of primary education enrollment rate for girls 50 Students/ 'Net enrollment is defined as the official number of primary permanent classroom ratio ................... 422 school-age pupils laken as a percentage of total school- age population. Students/ qualified teacher ratio .................... 131 (Source: MOE 1994/1995) Public spending on primary education Unit costs: universily student/primary student ........ 103/1 (Source: HESSEA 1990/17991) Poverty and Inequality Distribution of Incomes Income for rural smailholders for rural smaliholders (annual per capita) 7 Malawi Kwacha 80% below .......... 241 40% below .......... 117 CD 20% below ..... ..... 54 (Source: NSSA 1992/17993) 0-I , 0 300 800 4600 Landholding size Income (MK/AE/Year) for rural smaliholders Source: (NSSA 1992/1993) Less than 0.5 Ha .......... 41% 0.05-1.0 Ha ........... 31% Gini-coefficients Less than 1.0 Ha .......... 72% All Malawi ............ 62 (Source: NSSA 1992/1993) Rural smallholders ............ 57 i | ~~~~~~~~~~~~~~~~~~Malawi - Profile and Priorities for Action Contents Introduction ....................................................................1 1.1 Review of literature .............................................................. 1 1.2 Content and structure .............................................................. 2 Data sources .............................................................. 2 Profile of Human Resources for Malawi ................................................................... 5 2.1 A very young population .............................................................. 5 Population doubled in thirty years .................................... .......................... 5 Half the population is under 15 years old .............................................................. 5 Rural households have more children per adult ................................................... .... 6 2.2 Poor health and malnutrition .............................................................. 7 Child malnutrition is widespread ................................. ............................. 7 lnfant and child mortality are high .............................................................. 7 Child mortality is more prevalent in the Central region ................... ....................... 8 Maternal mortality is high .............................................................. 8 HIV and AIDS are spreading ...................... ........................................ 8 Sanitation is poor and there is little infrastructure ................................. .................. 8 2.3 An overstretched educational system ................................. ............................. 10 School enrollment .............................................................. 10 Important gains in school attendance .............................................................. 10 Enrollment rates are low and the poorest children are least likely to be in school ... 11 The poorest girls are least likely to go to school ......................................11............... 1 Enrollment rates are higher in the Northern region ............................................... 12 There is considerable age/grade mismatch ............................................................. 12 Most primary graduates never attend higher education .................. ....................... 13 Recent changes in enrollment .............................................................. 13 Important inequities remain in primary education ........................... ..................... 13 Public spending on education .................... .......................................... 14 Public spending on primary education increased, but bias against primary education persists ......................... ..................................... 15 Public spending in education has become more pro-poor ...................................... 16 Services and quality have deteriorated as enrollments have surged ........... .............. 17 2.4 Human resources development in an international context ................. ....................... 18 Living conditions are among the poorest in the world although public spending is high .............................................................. 18 2.5 The state of human resources .............................................................. 20 Profile of Poverty for Malawi ................................................................... 23 3.1 Identifying the poor ................................................................... 23 Establishing a meaningful poverty line .................................................................. 23 The household is the unit of analysis ................................................................... 25 3.2 The countrywide picture of poverty ................................................................... 26 Incomes and expenditures suggest widespread deprivation .............. ...................... 26 Most of the poor live in rural areas ................................................................... 26 Contents I ii Poverty is most prevalent and severe in rural areas ............................ ..................... 27 The prevalence of poverty varies across and within regions ............... ..................... 27 Major cities have less poverty ......................... ..................................... 28 Poverty prevalence in BOMAs is closer to rural than city levels ........... ............... 28 3.3 Characteristics of poor households .............................................................. 29 No education among household heads is mainly a rural problem ........... ............... 29 Poor households have the highest dependency ratios ............................................. 29 Although most poorer households are male-headed, female-headed households are more likely to be poor ........................................ 30 Most female heads of households are divorced, widowed or single ........... .............. 31 Female-headed households who receive regular cash remittances are poorer .......... 32 3.4 The geography of rural poverty .............................................................. 32 The most populated ADDs in the Southern and Central regions have the most severe poverty .............................................................. 32 Mzuzu is poorer than other ADDs in the North .................................. ................. 33 Smallholders in Kasungu are much better-off than other smallholders in Malawi .............................................................. 34 The highest severity and densities of poor households are at the Mozambique border .................... .......................................... 35 There are poverty pockets in the North .............................................................. 36 Scope for geographical targeting .............................................................. 36 3.5 Main factors associated with poverty in rural Malawi ............................ ..................... 36 Landholdings and assets .............................................................. 37 Poorer households have least access to land under cultivation ............. ................... 37 Poverty prevalence and cultivated landholdings show the same geographic pattern .............................................................. 37 Households with more members have less land per person ............... ..................... 38 Female-headed households are more likely to farm small areas ............ .................. 38 Cultivated land size alone does not predict household income .............................. 38 Poorer smallholders use few inputs and lack assets ................................................. 39 Female-headed poorer households have the fewest productive assets ...................... 41 Crop profile .............................................................. 41 Local maize is the main crop for the poorest smallholders ..................................... 41 Poorest smallholders do not farm burley tobacco ................................................... 43 Poorer households are net purchasers of maize ........................................... ........... 43 3.6 Limited income from off-farm sources .............................................................. 43 Off-farm employment .............................................................. 44 Smaliholders earn little off-farm income .............................................................. 44 Off-farm employment provides a higher share of income for poorer households ... 44 Sources of income vary across regions .............................................................. 45 Female household heads have less off-farm employment than male heads ............. 46 Medium-sized farmers are a mixed group .............................................................. 47 Livestock .............................................................. 47 Livestock ownership is positively related to income ............................................... 47 Livestock ownership varies by geography and income ........................................... 47 3.7 Main poverty determinants in Malawi .............................................................. 48 Landholding size and location and smallholder incomes ....................................... 48 3.8 The state of poverty in Malawi .............................................................. 49 iii | Malawi - Profile and Priorities for Action Implementing a Strategy to Reduce Poverty in Malawi ......................................... 55 Setting priorities .............................................................. 55 4.1 Developing human resources .............................................................. 56 Expanding access and reducing inequities in the social sectors ............. .................. 56 Increasing effectiveness and quality of social services ...................... ....................... 56 4.2 Improving rural livelihoods ............................................................... 57 Ongoing reforms are necessary but insufficient for the poorest .............................. 57 Economic reforms should be broadened to reach more smallholders .......... ........... 58 Examining land policies ............................................................... 58 Increasing off-farm incomes ...................... ........................................ 59 4.3 Safety net interventions for the poorest .............................................................. 59 Supporting transfer programs for the poorest ........................................................ 59 Distributing free inputs ............................................................... 60 Improving poverty monitoring for better targeting ................................................ 60 4.4 Agenda for further study .............................................................. 60 Methodological Notes .............................................................. 63 Tables for Chapter 2: Profile of Human Resouces for Malawi ........................................................ 71 Tables and Figures for Chapter 3: A Poverty Profile for Malawi ..................................................... 73 References .............................................................. 87 Tables Table 1.1 Review of literature: Summary of findings and questions for further study .3 Table 2.1 Household size and relationship structure by location .6 Table 2.2 Malnutrition indicators in Malawi and neighboring countries .7 Table 2.3 Child mortality and life expectancy for selected African countries .8 Table 2.4 Infant and child mortality in Malawi: the past 15 years .9 Table 2.5 Regional variation in child mortality .9 Table 2.6 Water source, sanitation and flooring for rural and urban households .9 Table 2.7 Gross enrollment rates by quintiles and gender, 1990/199 1.1. Table 2.8 Gross enrollment rates in primary school by region and residence, 1990/1991 .12 Table 2.9 School enrollment: age/grade matching, 1990/1991 .12 Table 2.10 Gross primary enrollment rates by quintiles and gender, 1990/1991 and 1994/1995. 13 Table 2.11 Cross-country comparisons of education financing per student .15 Table 2.12 Public education spending on poorest and richest quintiles .16 Table 2.13 Quality indicators in primary schools, 1992/1993 and 1994/1995 .17 Table 2.14 Aggregate statiscics for Malawi compared to similar groups of countries .19 iv Contents I i Table 3.1 Poverty prevalence for rural smallholders according to different poverty lines .......................... 25 Table 3.2 Households by national and rural-urban location ............................................................... 26 Table 3.3 Poverty indices by location, using upper cutoff, percentages .................................................... 27 Table 3.4 Educational status of household head by gender and location, percentages .............................. 29 Table 3.5 Dependency ratio by income group ............................................................... 30 Table 3.6 Household composition by gender of household head ............................................................. 30 Table 3.7 Poverty prevalence among male- and female-headed households ............................................. 31 Table 3.8 Poverty in female-headed households by marital status of household head ............. .................. 3 1 Table 3.9 Dependency ratio for female-headed households by marital status ........................................... 3 1 Table 3.10 Percentage of poor female-headed households receiving cash allowanices by marital status ....... 32 Table 3.11 Percentage distribution of income group and ADD ............................................................... 33 Table 3.12 Poverty indices by ADD using the 40th percentile income cutoff ............................. .............. 33 Table 3.13 Percentage distribution of cultivated areas by poverty group ................................................... 37 Table 3.14 Percentage of households in each ADD in each cultivated area class ........................................ 38 Table 3.15 Distribution of cultivated areas and income deciles ............................................................... 38 Table 3.16 Proportion of smallholder households using purchased inputs by income groups .................... 39 Table 3.17 Percentage of households with access to credit by area cultivated, income group and ADD ..... 40 Table 3.18 Percentage of ownership of productive assets by cultivated area and income group .................. 40 Table 3.19 Farm characteristics by gender of household head ............................................................... 41 Table 3.20 Percentage of area under cultivation with different crops by size of cultivated area .......... ......... 4 1 Table 3.21 Percentage shares of total cultivated area allocated to crops by poverty group .......................... 42 Table 3.22 Shares of household maize requirements met by own production ............................................ 4'2 Table 3.23 Percentage of smallholder household income from different sources by income group ......... .... 44 Table 3.24 Shares of income received from different sources by ADD ....................................................... 45 Table 3.25 Percentage distribution of income sources by headship and income group ............................... 46 Table 3. 26 Regression analysis: Determinants of smallholder incomes in rural Malawi ............................. 48 Table Al. I Sample by stratum, number of households, and sampling weights ........................................... 65 Table A 1.2 Items included in computing the cost of living indices ............................................................ 65 Table Al.3 Head and spouse earnings and wages from SDA Module A ..................................................... 68 Table Al.4 Coding and mean wages from SDA Module A ..................................................... .......... 69 Table A2.1 Comparisons of fertility indicators between Malawi and neighboring countries .......................71 Table A2.2 Net and gross enrollment rates in primary education, 1990/1991, by quintiles and gender ..... 71 Table A2.3 Net enrollment rates in primary education 1990/1991. by region, residence, and gender ........ 72 Table A2.4 Net enrollment rates in secondary education 1990/1991. by quintile and gender .................... 72 Table A2.5 Gross enrollment rates in primary education, 1990/1991 and 1994/1995, by region and residence ..................................................... 72 Table A3.1 Educational status of household heads by gender and cutoff line, rural Malawi ....................... 73 Table A3.2 Smallholder pover ry by c haracteristic of head ...................................................... 74 Table A3.3 Percentage of households by gender of household head ...................................................... 74 Table A3.4 Percentage of households below the two expenditure cutoff lines by gender of household head ............................................................... 74 Table A3.5 Distribution of female-headed households across regions ......................................................... 75 Table A3.6 Marital status of head of the household by gender ............................................................... 75 Table A3.7 Female-headed households below the 40 percent cutoff line by age group and marital status.. 75 Table A3.8 Areas cultivated by income decile ................................................................ 75 Table A3.9 Farm size by gender of household head ............................................................... 75 Table A3.10 Poverty indices, in percents by ADD, different cutoffs ........................................................... 76 Table A 3.11 Average livestock owned by income decile .76 Table A3.12 Average livestock ownership by poverty cutoff and ADD ....................................... 77 v | Malawi * Profile and Priorities for Action Figures Figure 2.1 Distribution of population across age groups ..................................................................6 Figure 2.2 Education level attained by adults in two age groups, 1990/1991 ............................................ 10 Figure 2.3 Growth in primary enrollments from 1992/1993 to 1994/1995 by region .............................. 14 Figure 3.1 Cumulative smallholder income distribution ................................................................. 24 Figure 3.2 Distribution of incomes for rural smaliholders, per capita ....................................................... 25 Figure 3.3 Lorenz curve for national distribution of expenditures ............................................................. 26 Figure 3.4 Distribution of total household population and poor households by region ............................. 27 Figure 3.5 Percentages of households below 20th and 40th percentiles in BOMAs/Cities ........................ 28 Figure 3.6 Mean annual household incomes and Gini-coefficients for rural smallholders by ADD ........... 33 Figure 3.7 Population classified "poor" below 20th income percentile by rural development project ........ 34 Figure 3.8 Poverty gap 40th percentile by rural development project ........................................................ 35 Figure A2.1 Rural primary enrollments in 1992/1993 and 1994/1995 by regions ...................................... 72 Figure A3.1 Cumulative distribution of smallholder income ................................................................. 77 Figure A3.2 Population classified "poorest" below 20th income percentile by Rural Development Project.. 78 Figure A3.3 Area planted per adult equivalent in household by Rural Development Project ....................... 79 Figure A3.4 Hybrid maize yields by Rural Development Project ................................................................. 80 Figure A3.5 Fertizer usage by Rural Development Project ................................................................. 81 Figure A3.6 Share of hybrid maize in total area planted by Rural Development Project .............................. 82 Figure A3.7 Share of burley tobacco in total area planted by Rural Development Project ............................ 83 Figure A3.8 Persons per hospital by administrative division ................................................................. 84 Figure A3.9 Persons per well by administrative division ................................................................. 85 Figure A3. 10 Number of students per teacher by administrative division .86 Text Boxes Box 3.1 Food security and landholding size ...................................................... 43 Box 3.2 Socioeconomic classification of the population ...................................................... 51 Box Al.l Primary-level enrollment algorithm, by socioeconomic groups, in 1994/1995 ......................... 64 Introduction Introduction Reducing poverty is a central policy objective of the new The economy is overwhelmingly agricultural. Three Government of Malawi which initiated the Poverty Al- broad categories of the population are engaged in agri- leviation Programme. This created a new imperative to culture-owners of large farms (the estate sector), increase understanding of the magnitude and of the mul- smallholders (people who own their own small farms) tiple dimensions of poverty in the country. And this is and agricultural laborers (who work mainly on estates or the objective of the Profile of Human Resources and Pov- on the more prosperous smallholdings). Approximately erty presented in this report. Such knowledge can help 90 percent of the inhabitants of rural Malawi are to guide policy and investment priorities and inform the smallholders, and their main income comes from their design of programs intended to improve living condi- landholdings. tions and increase incomes of the people in Malawi. A Studies focusing on human resources agree that the greater understanding of the magnitude and the profile need for extensive investments in the social sectors is of poverty will also make it easier to implement a moni- urgent. Health and social indicators are among the poor- roringsystem to evaluate the effects of programs and track est in Africa, and Malawi has one of the lowest life ex- the progress of key indicators of poverty. pectancies in the world. Child and maternal health are documented at a critically low level. Education is scarce and most adult Malawians are illiterate. Therefore pro- vision of priority human development services such as primary health care, basic education, nutrition, water and 1.1 Review of literature sanitation is viewed as one of the most cost-effective ways of reducing poverty. A number of studies conducted since the late 1980s There is consensus that past policies have conI- have examined human resources and poverty in strained productivity of the huge rural smallholder Malawi.' A summary of the key findings of these stud- sector and resulted in very low incomes and food in- ies and of issues that need to be addressed is shown in security for these households. The policies of the pre- Table 1.1. The summary also uses inputs from policy- vious government created a dual economy by trans- makers and from the research community in Malawi. ferring customary land to the estate owners and by The studies document the main patterns of human giving them the sole right to grow burley tobacco, resources development and poverty . Their main short- Malawi's most profitable crop. This left most coming is that they cannot draw on a country-wide smallholder families with plots of land too small to data set providing information on household expen- support families and forced them to sell their labor diture and incomes. to the estates. 2 | Malawi * Profile and Priorities for Action Efforts to provide smaliholders with improved Data sources agricultural technologies and to persuade them to di- versify their crops have largely failed. Most This report uses data from three household surveys smallholders continued to grow only local maize and, conducted in the early 1990s: the HESSEA (1990/ as their agricultural productrivitty stagnated, subsis- 1991), the NSSA (1992/1993), and the DHS (1992). tence households were forced to buy more food. There The analysis presented in this report is the first at- is little off-farm work in rural areas and real wages for tempt to analyze poverty in Malawi using nationwide unskilled workers have not risen. Men had to seek data on household expenditures and income. The work elsewhere, sending remittances to support their main sources are: families. * The 1990-1991 Household Expenditure and Small-Scale The findings from this profile should provide EconomicActivities Surve (HESSEA) conducted by the greater detail about the characteristics of the poor and National Statistical Office of Malawi, is used to create a about the main factors associated with poverty, thus abLtt hnnfoaoameasure of household expenditures for Malawi on a na- answering some of the questions presented in Table tional basis. The sample consists of 6,000 households. 1. 1. In doing so these findings should guide policy TheHESSEA-the onlysource of informationoncon- and investment priorities to reduce poverty. The re- sumtio -ba e ousehol re withonational cov- sumption-based household welfare with national cover- port is also uiarended Lo identifv tOpICS for further port~~~~~~~~~~ isas neddt2dnif oisfrfrhr ae-makes it possible to analyze both urban and rural study by Malawian researchers and thus help define are-as es it also ntains iormation on surce * ~~~~~~~~~~~~~~areas. The HESSEA also contains information on sources the policy agenda for povertv reduction. ... of income, demographics, economic activities, and some social indicators. The 1992/1993 National Sample Survey ofAgriculture (NSSA), conducted by the National Statistical Office 1.2 Content and structure of Malawi, is used to create a measure of household income for rural smallholders, who are by far the larg- The report starts with a profile of human resources est group of poor people in Malawi. The NSSA sample and comparisons between aggregate social and eco- consists of observations from some 12,000 nomic indicators for Malawi and those for other simi- smallholders and includes information on household lar groups of countries. Second, household survey data characteristics, demographics, labor supply, farm la- are used to build the poverty profile to assess the preva- bor demand, agricultural practices, and livestock own- lence, depth, and severity of poverty across urban and ership and on changes in stocks, asset ownership, and rural areas and to show the extent of income inequal- the earnings of the household head and spouse. A itv. Third, household survey data are used to show shortcoming of the NSSA in analyzing rural poverty the relationship between poverty and geographic lo- is that the sample frame does not include estate ten- cation, household demograplhics, asset ownership, ants or estate owners.2 access to infrastructure and public goods, linkages to markets, sources of income, and other factors. Finally, * The 1992 Demographic and Health Survey (DHS) was the summarized findings and messages of the report carried out by Macro International Inc., in collabora- are used to develop priorities for a poverty reduction tion with the National Statistical Office of Malawi, strategy in Malawi. on a nationwide sample of 5,300 households. Very similar DHS surveys were carried out in several coun- tries in Sub-Saharan Africa in the late 1980s and early 1990s. These have made it possible to make compari- sons across countries. The DHS contains informa- tion on household demographics, assets, sanitation, and child and maternal health. Introduction 3 Table 1.1 Review of llterature: Summary of findings and questions for further study Topic Findings Questions for further study Human resources Health Infant and child mortality rates are -How is access to and use of health facilities among the highest in the world distributed across regions? Women suffer more than men from -How are differences in health outcomes across (seasonal) malnutrition regions explained? Education Malawi has worse education indicators - What is the variation in enrollment rates across than the average for Sub-Saharan Africa gender, ages, regions, and rural/urban location? Education among the poor is much - What is the composition, incidence, and higher in urban than in rural areas efficiency of public spending in the education sector? - What are the determinants of the household demand for education and health for their members? Household characteristics Female-headed households are more - What is the relationship between poverty and likely to be poor than male-headed the characteristics of the household head? households - Do female-headed households engage in Female-headed households have less different types of income-earning activities than resources to earn off-farm incomes and male-headed households, and to what extent is are more dependent on remittances this related to poverty? Large family size and high dependency Is gender of the household head useful for ratios are associated with poverty targeting purposes? Regional and urban poverty Poverty is most prevalent in the Southern -Are there important differences in poverty Region among regions. ADDs and RDPs? Urban poverty is characterized by sub- -What are the extent and characteristics of urban standard housing, low levels of poverty? education, and limited employment opportunities Rural poverty Agricultural production Higher yield practices have not been - How much can agricultural intensification adopted by the poorest farmers reduce poverty? Maize is the predominant crop among - How important to smaliholders are the prices of smaliholders and access to purchased inputs? Expansion in the estate sector has - How will changes in the relative prices of inputs caused increased deprivation among and outputs affect the poor? smaliholders - How has the smaliholder burley program affected the poor? - How to put in place drought mitigation programs? Landholding size Landholding size is an important but not - What is the relationship between household size, perfect indicator of poverty landholding size, and the poverty status of the Most farmers cultivate very small land household? plots - What is the relative importance of land and labor constraints to smaliholder producers? - What is the relationship between land tenure status and nutrition in the households? Income sources Poor households receive higher - Do non-agricultural income-earning proportions of their income from off-farm opportunities contribute to the incomes of poor sources households? Self-employment income has decreased - How do household income patterns vary among the poorest in recent years across geographical areas and sociological characteristics (such as gender), and the poverty status of the household? - Do remittances bring households out of poverty? 4 | Malawi * Profile and Priorities for Action Chapter 1 Notes I These include Malawi: Growthl Through Poverty Reduction, World Bank (1990); Situation Analysis ofPoverty in Malawi, Government of Malawi and United Nations (1993); Poverty in Malawi: A Review ofLiterature, Centre for Social Research (1994); Poverty Profile of Rural Households in Malawi: A Summary of Recent Findings, Simler and Quisumbing included in the Agricultural Sector Memorandum, World Bank (1994); Beyond Hunger. Ignorance, and Disease: A Poverty Impact Assessmentfor Malawi, M. Kostner, et al. (1994); Policy Reform and Poverty in Malawi. A Survey ofa Decade ofExperience, Sahn et al. 1989; studies on urban poverty at the Centre for Social Research. and studies on food security in rural Malawi by Pauline Peters. 2 Estate tenants and their dependents numbered 586,000 (in 1989), or roughly 7 percent of Malawi's total population (Jaffee, Mkandawire, and Bertoli, 1989). Casual workers on estates are included, however, as these workers are technically considered to be smallholders. The families of permanent estate workers may be included, depending largely on whether they reside on the estate. Profile of Human Resources for Malawi 1 5 Profile of Human Resources for Malawi People in Malawi face more difficult circumstances than stress on scarce land resources. The population den- people in most countries. Malawi does not have adequate sity at the time of the 1987 census was estimated to resources to secure minimum standards of health, edu- be 85 people per square km-and it has increased by cation, and nutrition for its rapidly growing population. about 15 people during the past 8 years. This situa- This section analyzes human resources in Malawi in- tion is certainly worse for the poor - land is distrib- cluding such indicators as demography, health, nutri- uted very unequally. tion, and education. Half the population is under 15 years old 2.1 A very young population The shape of the population pyramid for Malawi shows a very large proportion of children. Almost half (47 percent) of the population are under 15 years of Population doubled in thirty years age. Figure 2.1 shows the population age profile and illustrates the large number of young children rela- Between 1964, the year of the country's independence, tive to the adult population. The age structure of the and 1994, Malawi's population increased from 4 mil- population in Malawi indicates that institutions pro- lion to about 11 million. If population growth re- viding nutrition, health and education face an over- mains at the present level (3.3 percent per annum), whelming task. the population in Malawi will double in the next 20 The population dependency ratio is very high and years. The average woman bears almost seven chil- increasing. l The 1977 census shows 97 dependants dren in her lifetime. Access to and the use of modern for every 100 adults of working age, the 1987 census contraceptive methods is very low, which makes it diffi- shows 101 dependants, and the 1992 MDHS shows cult for Malawian women to control their fertility even a dependency ratio of 1.06, meaning that Malawi now though there is evidence that their desired fertility level has 106 dependants per 100 adults of working age. It is lower than the actual (Annex 2: Table A2.1). is therefore common for Malawian children to be The worst impact of the rapid population growth working before the age of 15 thereby reducing the is likely to be felt by the poorest households since actual dependency ratio. However, economic activi- they tend to live far away from social services and ties compete with school attendance and probably are they are the last to be reached if services are not ex- related to the high drop-out rates for primary educa- panded. The high population growth places great tion in Malawi. 61 6 ~~~~~~~~~~~~~~~~~~~~Malawi * Profile and Priorities for Action Rural households have more children per adult The most common relationship structure is tvo re- lated adults: a man and a woman. Only one adult Table 2.1 resident is more frequent in rural households, while Household size and relationship urban households have a higher proportion of adult structure by location residents. Urban households also tend to have more memnbers than rural households, but the dependency Rural Urban Total ratio in urban households is lower than in rural house- Percent of household holds, 0.90 compared to 1.08 (Table 2.1). One fac- population 92 8 100 Mean household size tot contributing to this may be that fertility rates are (Number of persons) 4.4 4.8 4.5 lower in urban areas. Foster children are also more Percentage of prevalent in urban households. Another factor could households with: be the presence of rural migrants who have joined One adult 19 13 18 the households of relatives in urban areas. Two related adults: Consequently, the average urban household has -of opposite sex 45 42 44 -of same sex 4 5 4 more potential providers than the average rural house- Three or more hold. Urban households may therefore be less likely related adults 28 32 29 to be poor since there are more people to provide for With foster children 19 24 20 dependent household members. On the other hand, Other 4 9 5 in poor urban households, the large number of adults Source: MDHS 992 may mean that housing is overcrowded. Figure 2.1 Distribution of population across age groups Age Age 80+ EN 80+ 75-79 II 75-79 70-74 Males EU Females 70-74 65-69 EZi 65-69 60-64 60-64 55-59 r _ 55-59 50-54 L111_50-54 45-49 45-49 40-44 I 40-44 35-39 E 35-39 30-34 30-3d 25-29 25-29 20-24 20-24 15-19 15-19 10-14 10-14 5-9 _ _:: _ 5-9 0-4 0-4 10 8 6 4 2 0 2 4 6 8 10 Percentage of population in each age cohort Source: MDHS 1992 Profile of Human Resources for Malawi 2.2 Poor health and to the continuing poor living conditions and the perma- nence of the risk factors that cause mortality during preg- malnutrition nanic and the first vear of life. The reductionl in child mortalitv is probably related Malawi has a very high tinder-five mortality rate. Al- to the high vaccination rate, which has prevented though the rate declined durinig the last decade, it is the immunizable diseases suchi as measles. According to DHS Worst amongy the COun1tries iln Eastern and Southern Af- the vacciniationl coverage for childrell tinder the age of rica for which data are available. l ife expectancy is among two in 9 9)2 was: 88 percent vaccinatced against measles, the lowest ini the world. 'This is due to illiteracy precari- 88 pci-ceit received all rhree doses of polio vaccine, and otis living conditionis such as food insecurity', and limited percenit received a BCG vaccination. Nevertheless, access to saniration, health care and other social seirvices. even wirh this very high immunnization coverage, almost one in evel-V four Malawian children dies before reach- Child malnutrition is widespread ing the fiftlh birthday. These exceptionally high rates of intf'ant and child mortality are caused by high rates of Malnitr-ition, a major problem in Malawi, is caused by malniLtrition, infectiouis diseases, and malaria. The poor diets, short birth intervals, and inappropriate feed- MDHSi reported a prevalence' of acute respiratory in- ing practices. StuLnting, which is an indicaror- of long- fectiois, fever, and diarrhea among children tinder the terml malnutrition, is more prevalent among childrenl in age of five of aboLit 1 5 percent. 40 percent, and 22 per- Malawi than in neighboring countries (Table 2.2). Nearly cent respectively. one in every two childrei tinder five is short for his/her Mother's edLication is strongly associaced with un- age and onie in four is uinderweighlt. der-five mor-tality. Childreni born to mothers whio had ilo educatiCon were twvice as likely to die before their fifth Infant and child birthday compared with children born to mothiers who mortality are high had been educated to the secondary, level (DHS, 1992). Risk factors associated with under-Five mortality included Although during the last decade there has been a decline having a mothel- who was very young or very old; birth in child mortality (childrell who die betweeni one and intervals shorter than 24 months; birth order, with the five years of age), infant mortality (children who die firstborn child more likely to die than children fourth or during: thieir first vear of age) has remainied high (Tables later in the birth order; small size at birth; and little uti- 2.3 and 2.4). The high infant mortality is probablv due lizationi of health services by the mother during preg- Table 2.2 Malnutrition indicators in Malawi and neighboring countries Malawi Zambia Zimbabwe Tanzania Kenya Namibia Indicators 1992 1992 1994 1992 1993 1992 Percentage of children under 5 -Underweight 27 25 16 29 22 26 -Stunted 49 40 21 47 33 28 Underweight =Proportion of children <-2 Standard Deviations from the median standard Weight for Age Stunting= Proportion of children <-2 Standard Deviations from the median standard Height for Age Sources. The Demographic and Heath Survey Reports for the individaLc countries by Macro InternationaL Inc. 8~~~~~~~~ MaOi*Poie and Priorities for Action nancy' and delivery' 620 maternal deaths per 100(00() births. The high fertil- In the context of neighboring countries, the mortal- itr rate and the tendeicy tfor short inter-vals between births itv' rates for Mialawi aire startling. The infant morrality do niot allowx wvomien in Malawi to recover betwveen births, rare is 30 percent highier- thani in neighboring Zambia whichi increases the risk for complicated preginanicies. A which also has a relatively high int- int mortality rate. high prevalence of malnlutrition amiong Malawianwomen Malawi lias the lowest lif expecrancy-44 vears-of the of childbearing age is an additrional factor affecting couLitries in the region for whitch data are available, materinal mortalitY rates. Child mortality is more HIV and AIDS are spreading prevalent in the Central region The 225,000 estiiated cases of AID)S pUt a heavy bur- \Within Malawi child mortality rate is highest in the den on the already very poor Malawi. T Fllifetimile hos- Central region-50 percent higher than in the other re- pital cost of between X2(00 to $90() eqtials I to 4 years of gions (ETable 2.5). Tle reasons for til's are probably the wages. Homle-based cart, a chcaper alternative to hospi- higher prevalence of riskl factors in this region, including ral care, is an ordeal for imany families because of the short birth intervals, low birth weight and lower utilizationi lack of potable water and latrines. Bv the year 2000, an of MCH services. Feer ominti in thet Central regioll used esritimated 2 million people will be infected with HIIV anteniatal and deliveryv care, andl this cotld also be a proxy and about 355,000 children will he orphianed. This will for low utilization of MCIH scivices after birth. Women in impose high costs o1 hotuseholds and on1 society ill gen- the Central iregion also had miore short-spaced pregnan- eral. In tirban areas, the prevalence of the virlius asnion cies, which are associated with a higher risk for child woImlenl attending antenatal care clinics is cuIrenitly esti- mortality, mated to bec at .0 percenit. Tihe Spread of thIe VilrIS is caIsingll, increalses in borh chil3d and adult mor-taliry rates. Maternal mortality is high Sanitation is poor and TIhe i low utilization of healthi ser-vices duiring chilld deliv- there is little infrastructure ery (Annex 2: Table A2- 1) anid the poor conditions dur- ing preg natncy are the main CaUsCS of the high maternal Over half oft'rhe population obtain their water fronm mortality rate, which was estimated by the MDHIS to be unsafe sources (Table 2.6). A large majority of hotise- Table 2.3 Child mortality and life expectancy for selected African countries Malawi Zambia Zimbabwe Tanzania Kenya Botswana Namibia Indicators 1992 DHS 1992 DHS 1994 DHS 1992 DHS 1993 DHS 1988 DHS 1992 DHS Infant mortality 134.3 107.2 52.8 91.6 61.7 37.4 56.6 Child mortality 114.9 93.6 25.8 54.6 36.7 16.0 28 1 Under-5 mortality 233.8 190.7 77.2 141.2 96.1 52.7 83.2 Life expectancy 44 48 60 51 59 68 59 Infant Mortality Rate= Nurnber of children dying before their first year per 1,000 live births Child Mortality Rate= Number of children dying aged between 12 and 59 months per 1,000 live births Under-five Mortality Rate=Number of children dying aged under 5 years per 1,000 live births Sources: Life expectoncy dota are for 1992. The Africa Development Indicators 1994-95 The World Bank 1995. Other indicators are from the Demographic and Hea/th Survey Reports. from Macro Internationai. Inc. Profile of Human Resources for Malawi 9 holds live in dwellings with floors made of packed no sanitation facility. This increases the spread of in- earth. Electricity is practically noni-existent in rural fections. especially in crowded households. In 34 per- areas, and even in urbani areas only 20 percent of cent of households, three to four people sleep in the houselholds have access to electricity. same room: about 10 percent of households have five While almost everybody in urban areas has ac- or more people per sleeping room. cess co a latrinie, one-tlird of rural households have Table 2.4 Infant and child mortality in Malawi: the past 15 years Infant Child Under-five Period mortality mortality mortality 1988-92 134.3 114.9 233.8 1983-87 137.5 126.1 246.3 1978-82 136.4 140.8 258.0 Source: 1992 MDHS Table 2.5 Regional variation in child mortality Percent of children Prevalence of Prevalence of born with a antenatal care deliveries Infant Child Under-five birth interval from health in health Region _ mortality mortality mortality below 24 months personnel facilities Northern 120.7 92.3 201.9 16.6 92.9 67.5 Central 130.2 151 0 261.6 24.0 86.3 51.2 Southern 144.3 100.1 230.0 19.2 92.0 56.3 Malawi 135.7 120.1 239.5 21.0 89.7 55.5 Source: MDHS 1992, Macro International, Inc. Table 2.6 Water source, sanitation and flooring for rural and urban households Percentage of households: Rural Urban Total -using an unsafe source for drinking water 58 9 53 -with no sanifation facility 31 3 28 -with mud flooring 93 44 87 Source: MDHS 1992 1 0 Malawi * Profile and Priorities for Action . nche analvsis in this section focuises mainl1 on pri- educational overshe m nmarv education as this educational level is most relevanlt educational system for sedticinig poverty. The analysis dr;ws oni thc ]990/ 1991 1 E SSEA data and is tipdated uising the most re- alawi's CdUtcatiorn system is hampered by problemils of cent available data firomii the MN'inistrV of Education.' por access, hill repetition and drop-out rates, poor inlfaStrturIeLIC , and inequality. lo redr-ess this sitiatil School enrollment Ilbc nicwv ( overnincit of NIalavwi made eduIcation par- rictclarkl primary edLiCatio1, its top prior0ity. In 1994 Important gains in school aftendance school ees fol- Pmill-larV educLation weir eliminiiated and public spending on1 eduication1 was inicreased sharpiv 'There has been a substalnial increase in the slhare of mainlIV O filnalnce the wenLty thOuLsISand ceachers who were yoIng adLJlts receiving soime pri mary edtLctLion, co(in- rc(ruLited to mIeeIt the expected surge in enirollmeit. picinig primilar school, and eveni attaiinigi higher levels .Malavian houiosehlolds responded to this new priorirv and of eductionl compared to the share of ot(ld-e adults wlho the nunmbebr of children en rolled at the primiarv level reached those levels in the past. Tlbe noinmther of jumniped byN 61 percenit froni 1 .8 million to near three Malawviars whio have never attenided school has beein sig- uillion students for the 1 994/1995 school year. 'ificain tlv redLuced AboLut one in thitee MaNlaiaw S cur- Detspite the recenlt impr(ovements. there are still se- rently aged betweeen 16 and 35 Cears old have necver- at- riotis problems and severe iCneqLialties in pulblic pri imar tenided school, compared to one in two oftthose ated i5 edlcta1ti011. TIheret is a dearthi of eveni miilinial essnttial years or older. The sharc of the populartioll aged betWeenv1 teachirig materials and hulmliani resouL-ces necessary to 1( and(t 35 years old whio have attained all educational edtrcate childi-n [Ithere are significant disparities ill par- level of at least Standards V to VIlil has doubIed conii- tiipantion and achilevemlielit a;inI ng socroeconolilic pared to that for the popuLatioll over 35 years of age. groups arid regionts and by genider. The qualitv of edu- It is also clear from compar-ing these tLwo age tgrotips cation, already severely comproniised by large class sizes that disparities between 111en1 ard women ill educational even befoere therecelt surge in primary enrollment, has attainmenit have declited, althouglh the pcenteIt;lgv of detetriorated t'lurtile.- women who have nevei- attenided school remains higher thaln the comparable perceiltage of miieni. Ariioiig iiiales, the share of those wlho have never attenided school has declinted fromi onie in three for Figure 2.2 the older group tO one in five for the Education level attained by adults vouLiger group (FiguL )' 2_2) Amrnong f'e- in two age groups, 1990/1991 niales this share has declined from tv( in three for the oldker gl-OlIp to Ies.s than, Onle h{' * alVE 3s Inarb m m 16-3s APalS U1 two for the von nlger group. hbus, genl- der disparities are still substantial among aduilts aged 16 to 35, but inequities be- tween males and females are declininig over :3 1)- _ time. Thle largest Increase over' ti1ine In school attenldanice in Standards V to VIII was among woisv t'r . Onie out of every L -Ci -3 - Sd. I _ I | fotir womilenl ill thel Voting1. ' tO tlF -,:,Stla I IV S'c VV S-cnnt-lrs Ed-[oll Sld l-IV 51C1 FVVI- Sreached at least Standardks \' to V 11 Males Females whichi is three times thlC nlulimber of tlc - older- grotp of femalets (otel 9 percenit) Source: HESSEA (1990/19 91) r s. __ __ _ ~~~~~~~~~~~~~~~~~wlio reachled t1hlS level prCVIOlY.si Profile of Human Resources for Malawi I 11 Enrollment rates are low and the among the poorest children: two in three drop out poorest children are least compared to less thani one in two among the richest likely to be in school children. Thus the poor are less likely to enroll in Stan- Only about half of all children six to thirteen years of dards V to VIII (Table 2.7). age were enrolled in primary schools in 1990/1991 and childreni in the lower expenditure quintiles are The poorest girls are munc less likely to be in school. Net primary enroll- least likely to go to school ment rates are more than 40 percentage points lower in the poorest quintile than in the richest (Aninex 2: Gender disparities in net enrollmelit rates at the pri- Table A2.2). mary' level do not appcar to be signiflicant (Anilex 2 The gross primary enrollment rate is substantially 'lable A2.3), but the gross rate is systematically lowelr higher thani the net rate arid this difference is more for girls than for boys across all the expenditUle marked for the poorest than for the richest quintile. quiitiles. This is an indication that girls aged 13 or Gross enrollment rates in primary school for the two older (captired in the gross rate) are more likely to poorest quintiles are about double the net enrollment dr-op out of primarv school than boys aged 1 3 or older rates for the same groups, confirming that late entry (Table 2.7). Thus it is crucial to put in place incci- and repetition occur particularly amonig lower-income tives for eariv enroililenti antd reteitioni of girls in groups. Thus, the poor are more li kely to dr-op out or priniar' education. to leave and returi repeatedly, decreasing their chanices Genider disparities in gross enrollmenit rates for of completing the primary cycle. primarv education are considerably hilgher in the Gross enrollment rates are much lower for Stan- South than in the North or( Cential re^,ions (Annie7x dards V to Vill than for Standards I to [V. The de- 2 T'able A1.3). Box's in the rulal Southi hiave anl over- cline in gross enrollietili rates betweeni the first- aild 111 gross e2nrollmnent ratc about 25 percetit higlher than the second-half of primary educationi is sharpest the rare for girls, xvhile in the rural North and Ceii- Table 2.7 Gross enrollment rates by quintiles and gender, 1990/1991 Household Gross Enrollment Rates Gross Enrollment Rates Expenditure Standards I to IV Standards V to VilI Quintile _ Boys Girls National Boys Girls National I - Poorest 96 70 82 36 27 32 11 1108 98 104 53 38 45 Ill 116 116 116 55 42 48 IV 132 115 123 76 58 68 V - Richest 133 151 142 88 68 77 All 1114 102 108 56 43 50 Primary education in Malawi takes a minimum of 8 years and it goes from Standard I to Standard VIII Gross enrollment rate in Standards I to IV All children enrolled in Standards I to IV as % of 6-9 year old population, Gross enrollment rate In Standards V to VIII. All children enrolled In Standards V to VIII as s/c of 10-13 year old ocoulation, Income quintiles are created by classifving every individual from the poorest to the richest and then dividing the population ins groups containing 20 percent of all individuals The poorest quintile represents the poorest 20 percent of the population Source. HESSEA 1990/199/ 1 2 | Malawi * Profile and Priorities for Action tral regions, the gross rates are about the same for There is considerable boys and girls. age/grade mismatch Enrollment rates are higher in the Late entry is very common, and older students are Northern region more likely to be enrolled at every educational level than are students of the correct age group. The age/ Children in the Northern region are the most likely grade matching analysis shows that repetition is also to be, and to remain, enrolled throughout all years of widespread. primary school (Table 2.8). Children in the rural About66percentofallstudentsenrolledinStan- Center have higher gross enrollment rates in Stan- dards l-IV are older than nine years ofage, and arouLnd dards I to IV than children in the rural South. 75 percent of all students enrolled in Standards V- However, enrollment rates in Standards V to VIII VIII are older than 14 years of age (Table 2.9). In are the same for children in the two regions. This in- Forms I-IV in secondary school, 75 percent of all stu- dicates that in both areas many children do not en- dents enrolled are older than 18 vears of age. A re- roll in school after Standard 1V Educational programs cent study found that the average age in Standard I is for the rural Center and the rtural South ought to ap- 10 vears of age and in Standard VIII. it is 22 years of proach two different sources of low completion. Pri- age." The main disadvantage of attending primary mary schools in the rural Center need to increase their school at an older age is that these children are often ability to retain children in school. In the rural South, needed to help to support the household. This be- enrollnients in primary schools need to increase alto- comes a disincentive for them to complete that level gether. Table 2.8 Gross enrollment rates in primary school by region and residence, 1990/1991 Household Gross Enrollment Rates Gross Enrollment Rates Expenditure Standards I to IV Standards V to VIII Quintile Rural Urban Total Rural Urban Total North 139 150 140 94 106 95 Center 103 138 107 38 85 44 South 98 146 103 38 86 44 Ali 104 143 108 45 87 50 Source: HESSEA (1990/1991) Table 2.9 School enrollment: age/grade matching, 1990/1991 Age group Students Percentage of total enrolled in level Level enrolled 5 or less 6-9 10-13 14-15 16-17 18 or over Total Standard I-IV 993,377 2 34 44 12 5 4 100 Standard V-VIII 407,305 0 2 25 25 24 25 100 Forms l-IlV 41,876 0 0 4 9 18 69 100 Forms III-IV* 17,839 0 0 0 2 8 89 100 Forms V-VI 145 0 0 0 13 11 76 100 Tertiary- 6,991 0 0 0 0 0 100 100 Includes M.C.D.E. (Malawi College of Distance Education) students. Tertiary = Primary Teacher Training, Technical Training and University. Source: HESSEA (7990/1991) Profile of Human Resources for Malawi 1 3 of education and then proceed to higher educational level students in Malawi, includini, studenits attend- levels. ing Primilary Teachcr Trainiiig. Technical Training, and Universitv. In 1994/1995, these enrollments repre- Most primary graduates never se nt a iere 0.3 percent of the total number of enr-oll- attend higher education minits at all levels of eduIcation. The niumber of students conmpletling each cycle anld Recent changes in enrollment enrolling in the next is extremely low. In 1990/1 991, only 18 percent of the children enrolled in primiary The responsc from houiseholds and students itn Malawi school completed the full eight years of primary edu- to the elimination of school fees and the ipriority giVen cation. In addition, since secondary schools offer only by the new government to prim1ary educationi has been a limited number of places, most childreni never at- overwhelming. This section analyzes the distribLution tend secondary school. Only one out of every 100 of the increase in enrollmenit across incomc groups, children entering the primary cycle is admitted to the gender and reg0ions using recenit data from the Minis- secondary cycle. The overall net enrollmelit rate in try of Education. secondary schools is 2.2 percent, and the gross rate is about 10 percent. Given the recent surge in enroll- Important inequities remain in ments and the deterioration in the quality of primary primary education schools, it is possible that a proportionally smaller number of children who complete primary schiooling Estimated gross primarv enrollimenit rates by income wiI attend secondary school in the near future. groups indicate that the same clhildren who were identi- Practically no children from the two poorest fied as beiing disadvantaged in 1990/1991 are llkely to quiintiles attend secondary school (Annex 2: fable remain disadvantaged in 1994/1995 (Table 2.10). Ma- A2.4). Even in the highest expenditure quintile, only jor increases in gross eniroililmeit rates occurred in the 8 percent of children aged 14 to 17 years old are en- 1994/1995 school year at everv incomile lcvel; however, rolled in secondary school. half as mianiy childreni fromil the poorest quiltiile are eu- It is also very rare for Malawians to attend ter- rolled in pirimary school as from the richest quiltile. The tiary education. There are only about 7,500 tertiary- gross enirollmenit rate foir the poorest quiltile is 74 per- Table 2.10 Gross primary enrollment rates by quintiles and gender, 1990/1991 and 1994/1995 Household 1990/1991 1994/1995' expenditure quintiles Boys Girls National Boys Girls National I - Poorest 65 51 58 100 69 74 11 83 69 76 117 88 102 Ill 88 83 86 118 98 114 IV 104 89 97 134 104 131 V - Richest 113 106 110 134 120 133 All 86 75 81 121 96 108 The numerator for the gross enrollment rates in primary education by quintiles in 1994/5 s total enrollrents estimated on tne loasis of 1994/5 regional enrollrnents provided by MOE (1995) and taking the HESSEA survey (1 9Q0/1) as our baseline. This computation involves using a simple mathematical algorithm in which total primary enrollments in each quintile In 1994/1995 are obtained yv applying toe regional rate ot growth in primary enrollments between 1990/1991 anc 19Q4/1995 to the regional composition of prirriary enrollments in trhat same quintile in 1900/19Q1 For further information see Annex 1. The dencronator for tne gross enrollment rates ri primary education by quinties in 19Q4/1995 uses a 3.3'. population growth rate for the6to 13 year-nd poplat on group. taking tne HESSEA suivey- (990/1991)as our baseline. Sources HESSEA (1990/1991). MOE (1993)g MOE (1995) 14 | Malawi * Profile and Priorities forAction cent, while it is 133 percent for the richest quintile. rates in the rural Center and rural South are around 100 Gender disparities in gross enrollment rates have percent, while the rate in the urban North is much increased for all income groups-girls in the poorest higher-I67 percent (Annex 2: Table A2.5). quintile have a gross enrollment rate of only 69 percent. 'The surge in primary enrollments in 1994/1995 has almost one-third lo-,ver than boys. put tremendous stress on schools where classrooms are Estimated gross primary enrollment rates by region badly equipped and where educational materials are indicate that, although the regional disparities are nar- scarce. The quality of primary education is more likely rowing, the same regions thar were relatively disadvan- to have deteriorated in the rural Central and Southern taged in 1990/1991 are still disadvantaged in 1994/1995. regions which are disproportionately affected by the lack Growth rates in primary school enrollments have been of education resources. Since the rural Central and South- highest in those regions that had lagged behind in net ern regions contain the majority of poor students, these enrollment rates in 1 990/1991. Between 1 992/1993 and should become the priority areas for allocating public 1994/1995, primary enrollment in the rural Central re- education spending as this should improve the access to gion increased by 70 percent, and in the rural Southern primary schooling for the poor. region, by 80 percent (Figure 2.3). By contrast, enroll- ments in the rest of the country increased by less than Public spending on education 30 percent. Enrollments in the rural Center and South increased The recent increase in the budget share allocated to pri- most in 1994/1995. In the rural South, enrollments mar' education is a very positive development in the surged from 700,000 to 1.2 million (Figure 2.3). Al- process of increasing investments in humiian capital and though regional disparities have narrowed with the re- improving equity. Tfhe priority now is to restructre the cent rise in enrollments, the rural Center and rural South composition of budgetary allocations within the sector still lag behind in primary enrollments, which are par- to continue improving equity, and to raise the qualit of ticularly low in these regions. Gross primary enrollment service delivery. Figure 2.3 Growth in primary enrollments from 19921993 to 1994/1995 by region Rate 90 80 70 60 50 40 30 20 10 0 North Center South All North Center South RAl Urban Rural Urban Rural Source: MOE (1993); MOE (1995)1 Profile of Human Resources for Malawi I 1 5 Public spending on primary education This situation has improved considerably, but re- increased, but bias against primary mains inequitable. At present, the cost of one sec- education persists ondary-level student in Malawi could finance four Because of the policy shift in 1994/1995. public recur- students at the primary-level, while the cost of one rent spending on educationi has more than tripled in real tertiary-level student is IIow 71 times more than the terms between the 1990/1991 and 1994/1995 fiscal cost of one primary-level student (Table 2.11). Uni- years, from NMK 267.5 million in 1990/1991 to MK versity education is still very costly otn a per stLident 899.7 million in 1994/1995, expressed in constant 1995 basis: a university student in Malawi is 103 times more Kwachas.l Primary-level spending has increased even costly thatn a primary-level student. The much greater more: it surged by 4.3 times in real terms during the cost per student of higher and university education same period and its share hias increased from 45 percent in Malawi is largely because there are very few stmt- to 71 percent of total spending on education. As a re- dents at this level. Progression rates from one educa- suit, public spending per studenta ar the primary level tional level to the next are extremely low, so that verv more than doubled in real terms between the 1990/1991 few students are admitted into tertiary education. and 1994/1995 school years, even after the sharp increase International comparisons of education statistics in enrollment. Per student spending in 1990/1991 was illustrate the relatively small numbers ofstudents go- MK85, and this increased to MK220 in 1994/1995. ex- ing beyond primary education in Malawi although a pressed in constant 1995 Kwachas.' large share of GDP is dedicated to the education sec- The bias against primarv education in the alloca- tor as a whole. In Madagascar and Lao P.D.R., coun- tion of resources within the sector as a whole is demon- tries of comparable income levels, a secondary stl- strated by estimating how many primlarv students could dent is four times as expensive as a primary student, be financed by the cost of one studetnt in either second- the same as for Mlalawi in 1994/1995 (Table 2.11). 1 ary or tertiary education. In 1990/1991, a secondary- However, the contrast at the tertiary level is striking. level student in Malawi was seven titnes as expensive as a In Malawi, tertiary students are about 70 times as primary-tevel stLdent, a tertiary-level studenlt was 97 expensive as primary students, while in Madagascar times as expensive, and a universiry studenit was 165 times and Lao P.D.R., they' are only about 20 times as costly. as expensive. ' In Kenya and Guinea, countries with a higher per Table 2.11 Cross-country comparisons of education financing per student Primary student equivalent cost' GNP/capita Education Country (1992 US$) Primary Secondary Tertiary University (%GDP) Malawi (1994/1995) 220 1 4 71 103 7.5 Madagascar (1993/1994) 230 1 4 22 22 2.6 Lao PD.R. (1992/1993) 230 1 4 24 - 3.6 Kenya (1992/1993) 310 1 3 41 41 9.2 Guinea (1994/1995) 510 1 2 35 58 2.0 South Afnca (1993/1994) 2,670 1 2 5 5 7.3 Mexico (1992/1993) 2,976 1 2 5 5 4.5 t The primary student equivalent cost (subsidy) expresses the number of primary school students that could be financed per student in higher levels. Sources: Castro-Leal (1995a, 1995b, 1994); Castro-Le1o and Dayton (1994); Doyton (1995a. 1995b) 16 Malawi * Profile and Priorities for Action capita income than Malawi, tertiary-level students are Public spending in education has about 40 timiies as expensivc as primarv students. In become more pro-poor South Africa and Mexico. COtiitlries witih a larger num- ber of studenits going bevond primary iand a much The edtication sector in Malawi in 1990/1991 had one higher per capita incomrie lvel, secondary educationi of the worst distributions of public education subsidies per student is onl' tw ice as expernsive as primaryv while in Sub-Saharani Africa. But because of the recent surge tertiary studenits are only five times as costly. in primary enrollments, the increase in government These results indicate that cven though a larger spending on the primary education sector, and the bud- share of puiblic money is being spent per student on getary reallocations within the education sector, the distri- the early levels of edLIcationi in Malawi. completion bution of public education spending across income groups and progression rates rem;ain low at this level. Pri- has improved considerably. mary students receive loxv qual i ty services even thoughI The share of public education resources received by Malawi allocates 70( percelmt of public educationi re- the poor increased substantially berween 1990/1991 and soLirces to this level and a larger share of its GDP goes 1994/1995. In 1994/1995, the poorest income quintile to the educaltion sector thani in some other cO11ntries. received 16 percent of all public education spending com- In 1994/1 995. Malawi spenlt 7.5 percent of its GDP) pared to 10 percentin 1990/1991 (Table 2.12).Theshare on educaItion, while Madagascar spent 3.6 percenit. going to the richest income quintile in 1994/1995 was Lao P.D.R. spent 3.6 percent aind Guinea spent onlv' only 25 percent compared to 38 percent in 1990/1991. 2 percent. The Malawi example shows that an effective way to in- Table 2.12 Public education spending on poorest and richest quintiles Education spending benefiting: the poorest 20% the richest 20% of the population of the population Country Year (percentage share) (percentage share) Malawi 1994/1995' 16 25 1990/1991 10 38 Cote d'lvoire 1993 10 38 Ghana 1992 16 21 Kenya 1992/1993 17 21 Madagascar 1993 9 44 South Afrca 1993 14 40 Tarnzani 1993 13 23 The distributon of all Qublic eoucation spend ng by quintiles in 1994/5 involves estimating total enrollments in 1994/5 by quintiles and by educational leve At the pr mary level total enrollrnents in 1994/5 by quintiles are derived by using a simple mathematical algorithm by app ying the regional rate of growth in primary enro Iments between 1990/1 and 1994/5 to the regional composition ot primary enrollments in that same quintile in 1990/1. The same distr bution of enroilments across quintiles at the secondary and tertiary levels observed in 1990/1 is maintained in 1994/5. For furtner information see Annex 1. Note: It public education spending was equally aistributed across popu ation quintiles, tne poorest and richest quintiles (every number in the table) would receive a 20 percentage share of spending Sources: Dayton (1995); Demery (1995); Bernier. Chao, and Demery (1994), Castro-Leol (1995a, 1995b); Demery and Verghis (1994); World Bank (1995). Profile of Human Resources for Malawi I ~~ 17 crease the equity of public spending in education is to inl- schools, few new classr 40- . shelter, needed to sustain an adult Malawian at a MK 172/year basic level. This line implies an annual income per E 80- adult equivalent of MK 151.* j MK 151/year The 1990 Reference Line was used in the World 20-_ . .. .. . . ,/ . Bank report 'Malawi-Growth Through Poverty .7. MK 98 /year Reduction" (1990) and implies an annual income 10 per adult equivalent of MK 172 (US$40). X-axis is on a log scale____ 0- x *MK are for the year of the NSSA survey, 1992/ o- I 1' I I I 1993 10 20 50 100 200 500 1000 _- Income (MK /Adult Equivalent / Year) Source: NSSA (I 992/1993) Profile of Poverty for Maai}25 range of the illustrative poverty lines defined, there is low priate as almost all the determinants of poverty affect the variability in household incomes. Thus, a change in the entire household. To make comparisons across households, value of the cutoff leads to a less than proportional change taking into account variations in their composition among in the measured prevalence of poverry.' age groups, incomes and expenditures were converted us- ing adult equivalency scales to produce a figure calculated The household is the unit of analysis at per adult basis. Tro compare across houiseholds with different incomes In the analyses presented here, incomes (using the NSSA and costs of living in different areas ofthe country, incomes data) and expenditures (using the HESSE-A data) are com- and expendituLres were adjusted to include estimates of own- puted and presented at the level of the household rather account productioni and consumption and of imputed ex- than the individual. The methodological details are pre- penditures on rental housing and to accotint for cost-of- sented in Annex 1. The focus on the household is appro- living differenices between areas. Table 3.1 Poverty prevalence for rural smallholders according to different poverty lines Percent of Annual income population per capita Poverty line -Percentile 80 % below ..... .. MK 241 60 % below .................... MK 213 1990 Reference Line ..... 54' percentile 50% below.................MK 158 --- Basic Needs Line . 43d percentile 40% below ....MK 1 17 Calorie Needs Line ...... 30th, percentile 20% below ....... MK 54 Source; NSSA 1992/1993 Figure 3.2 Distribution of incomes for rural smallholders, per capita 7- 0 300 800 4600 Income (MK/AE/Year) Source: NSSA 1992/1993 26 Malawi * Profile and Priorities for Action 3.2 The ounie piturebution of expenditures shows high inequality (Figure 3.3), 3.2oThe povertrywede pectureas does the Gini coefficient which is 0.62-the highest of poverty level of inequality for any of the 13 African countries for whiclh data are available.' Incomes and expenditures A similar pattern is observed for rural smaliholders. suggest widespread deprivation Their mean annual household incomes were 312 MK/ AE (US$66) and the median was 188 MK (US$39). The Incomes and expenditures in Malawi are both very low, bottom half of the distribuition of smaliholder house- and are distributed very unequally.) The national mean holds received 1 5 percent of the income, while the top expenditure level per household was US$189 and the 10 percent received 35 percent of the income. Eighty median (50h percentile) expenditure level was US$104 percent of smallholder households have incomes lower (at 1990-91 exchange rates). The large difference between than 500 MKlAE/year. about US$ 1 00/AE/year at July the mean and the median indicates a skewed distribu- 1993 exchange rates." The Gini coefficient for rural tio. T ioal Lorenz curve for the naional distri- smiallholder incomes is 0.57, which is also extremely high. These data confirimi that rural poverty is deeper and more severe than aggregate GNP figures indicate. Figure 3.3 Lorenz curve for national Most of the poor live in rural areas distribution of expenditures 1.0 - Poverty in Malawi is predominantly rural. Roughly 90 0.9 - percenit of the Malawian people live in rural areas, and W 0.8 - / / their share of the poor is slightly larger than their share of the poptilation-94 percent of the households with ~ 0.7 /expenditLres belov the 40"1 percentile live in rural areas J 06 - / (Table 3.2). 0.s - By contrast. urban households are under-repre- @ 0.4 - / / sented amonig thie poor.' Only 4 percent of the house- °- 03 / / holds below the 40' percentile are found in urban 0.3- a / /areas and only about 2 percent in regional BOMAs, , 0.2 - / / and together these areas contain only about 6 percent 0.1 - / of the households below the 20"h percentile (Table oo 2 0 / 3.2 2). Mean expenditures in the cities are more than 0 20 40 60 80 100 double those in rural or BOMAs households. Urban Percentage of Households househholds accouL1t for about 17 percent of all house- - ourc HESSEA dcoto hold expenditures. while constituting only 8 percent Source: HESSEA dayta Table 3.2 Households by national and rural-urban location Percent of Percent of all households Percent of all households Location household population below the 20th percentile below the 40th percentile Rural 90 95 Urban 8 4 4 BOMAs 2 2 2 Total 100 100 100 Note: Expenditures are all adjusted for differences n costs of living between urban and rural areas. Source: HESSEA 1990/1991 Profiie of Poverty for Malawi | - 27 of all households, which illustrates the skew of in- (see, for example. Malawi-Growrh Through Poverty Re- come in favor of the urban areas. duction, 1990; and the Situation Analysis of Povertrx in Malawi, 1993). Poverty is most prevalent and severe in rural areas The prevalence of poverty varies across and within regions The prevalence of poverty in rural areas is about double that in urban areas (Table 3.3). The prevalence of pov- Fifty-one pcrcent of all households are located in the erty in BOlMAs is close to that in rural areas. Souti hern Region, but this region contains a dispropor- The poverty gap (the percentage shortfall in expen- tionately high share of the poorer households in Malawi ditures of households below the poverty line or cutoff) (66 percent). The number of poorer households in the for rural areas is about two and a half times larger in Central and Northern Regions is lower than their share rural areas than in urban areas. In fact, 95 percent of the of the poputlation as a whole, indicating a lower preva- total Malawian poverty gap (thze shortfall below the pov- lence of- poverty there. The Central region, for example, erty Iline) is found in rLtral areas. Rural poverty is the contains 39 percent of all households but only 27 percent most pressing problem faced by Malawian policymakers. of poorer households (Figure 3.4). This is in contrast to the These findings are the same for the poorest 20 per- Central region's poor performance on health indicators. cent of the households and are conisistenit with estimates The percentage of households below each expen- reported in the literature using seconcary' and othrc data diture CLutoff varies significanitly between rural and Table 3.3 Poverty indices by location, using the 40th percentile, percentages Location _ Prevalence Poverty gap Severity Percent of total poverty gap Rijral 42 17 1 0 95 Urban/Cities 20 7 4 4 BOMAs 36 14 7 2 Notes: Four poverty indices are presented in this tabie. The Poverty prevaience (P(=,) is the percentage of total households that fatl below a given poverty cutoff The Poverty gap (P,=I) Is the Percentage shortfall in the incomes of poor households below the poverty cutoff multiplied by the prevalence. The poverty gap is an indicator of the depth of poverty-larger numbers mean deeper poverty. The Severity index (P=2) is the sum of the squared percentcge aeviations below the cutoff of the incomes or expenditures of poor households. Because the deviation is squared, relatively more weight is placed on the poorest. This measure is therefore sensitive to the distributior of the incomnes/expenditures of the poor The percentage of the total gap is the sum of the total shortfall in incomes below the cutoff bv area (rural, urban, etc.) divided by the total national shortfali. Source: HESSEA 1990/1991 Figure 3.4 Distribution of total household population and poor households by region Distribution of all households Distribution of poorer households Northern Northern 10% 70/ Southern Cen Southern Cenra 59%S66% 2H%. Source: HESSEA 1990/1991 28 | Malawi * Profile and Priorities for Action urban areas within each region (especially in the ing on which cutoff is used. Blantyre has a higher Southern part of Malawi) and also between regions. prevalence of poorest households than Mzuzu, but Whether the focus is on the poorer or the poorest Mzu-zUi has a higher prevalence of poorer households. households, there are more poor people relative to the Zomba has the lowest prevalence of poverty of any population in Southern rural areas. This finding is city usinig either cutoff (see Figure 3.5). consistent with patterns found in previous studies. The prevalence of poorer and poorest households in Poverty prevalence in BOMAs both Central and Northern rural areas is roughly half is closer to rural than city levels those in the Southern rural area. In the Northern and Central Regions, BOMAs Major cities have less poverty have higher prevalences of poverty than the rural areas in these regions. It may be that BOMAs at- Maj;or cities tend to have the lowest prevalences of tract destiture people from the ruiral areas who can poverty, except for Lilongwe where the prevalence of only afford to travel to nearby urban centers, where poorer and poorest households is slightly higher than the economy may not be strong enough to sup- in Central and Northern rural areas. The prevalence port them. The ranking of BOMAs according to of households below the 20th percentile in Lilongwe poverty prevalence also changes with the different is more than twice as high as in any other city, and poverty lines. Northern BOMAs have the lowest the prevalence below the 40th percentile is 50 percent prevalence of households below the 20th percen- higher than in any other city. The high levels of poverty tile, while Southern BOMAs have the lowest preva- in urbani Lilongwe deserve special attention. Efforts to lence of households below the 40th percentile. reduce poverty in Lilongwe should include generating BOMAs in the Central Region have higher preva- employment opportunities for low-skilled workers. lence of houselholds below either cutoff than other The relative ranking of the other urban areas ac- regions, indicating that poverty in the central cording to poverty prevalence changes slightly depend- BOMAs is deep and severe. Figure 3.5 Percentages of households below 20th and 40th percentiles in BOMAs/Cities 50 o <20% cutoff 40 * <40% cutoff 30 20 10 0 Northern Central Southern Mzuzu Lilongwe Zomba Blantyre BOMAs Cities Source. HESSEA 7 990/1991 Profile of Poverty for Maai|29 3.3 Charac-teristics the population in general. Education is related to the Charactersstics of poor poverty status of the household (see Annex 3: Table A3. 1). households However, the correlation is not too strong given that a large proportion of adults in Malawi lack educationi. Evidence from previous studies suggests that certain char- Sixty-nine percent of female heads of households below acteristics of a household are good predictors of its pov- the 40 percentile cutoff line have never attended school, erty status. This section presents an analysis of those char- compared to 55 percent of the female heads of house- acteristics, some of which relate specifically to the head holds above this cutoff. This trend is the same for male of the household, to provide an insight into the factors household heads who have no education, as the preva- that influence the chances of a household being poor. lence of poverty is 13 percent lower for those above the 40 percent cutoff compared to those below this cutoff. No education among household heads is mainly a rural problem Poor households have the highest dependency ratios Female heads of households in rural Malawi are particu- larly likely to lack education-63 percent have never at- The poorest households in Malawi have the highest av- tended school whereas 31 percent of male heads of rural erage household dependency ratio (1.36). Households households have no education. By contrast, close to 20 above the 40th percentile have more potential providers percent of urban female household heads and 7 percent than dependants and the dependency ratio for this group of urban male heads of households have no formal edu- is lower than one (Table 3.5). cation (Table 3.4). The analysis of household composition shows that The urban/rural contrast is evident at all levels. Close the average household dependency ratio is higher in fe- to one-third of urban female households heads have at- male-headed households than in male-headed in both tended or completed secondary school, while this is the rural and urban areas (Table 3.6). The average age of case for only one percent of the women who head a rural female household heads is lower than for male heads of household. The same pattern is also valid for the male households, which is an indication that these households heads of households. contain more dependent children. The likelihood of finding a household head with no Although female-headed households have more education among poorer households is greater than in young dependants to provide for per adult household Thble 3.4 Educational status of household head by gender and location, percentages Education status Heads of rural Heads of urban Heads of all of head of the households households households household Female Male Female Male Female Male No education 62 31 22 7 60 28 Literacy training 6 3 2 1 5 3 Std l-IV 21 30 16 15 21 28 Std V-VIII 11 29 28 35 11 30 Some secondary 1 4 15 16 2 6 Secondary finished 0 0 12 16 1 4 University 0 0 1 1 0 0 Source: HESSEA 1990/1991 30 - Malawi * Profile and Priorities for Action member, they have usually less income. Simler (1994c) the chance of a female-headed household being among finds that female-headed households have per capita ex- the poorest 40 percent is one in two (Annex 3: Tables penditure levels that are only three-quarters of those of A3.2 and A3.4). male-headed households. As shown elsewhere in this re- In general, the HESSEA data show that rural pov- port, female household heads have also fewer resources erty is more prevalent than urban poverty, and the dif- for generating on-farm and off-farm income to provide ference is particularly clear in the case of female-headed for their families. Thus, many female household heads households. Most (95 percent) female-headed households need to increase their incomes to support their depen- are rural and the percentage of female-headed households dent children, but they do not have the means to do so. that fall below the 40th percentile cutoff is 52 percent Female-headed households are, therefore, more vulner- compared to 22 percent in urban areas (Annex 3: Table able. If the female household head gets ill or loses her A3.4). The equivalents for male-headed households are source of income, these households are more likely to 38 percent of rural and 20 percent of urban households. have temporary or permanent declines in consumption. This is confirmed by data from the NSSA household survey (Table 3.7). Although most poorer households Rural female-headed households also account for are male-headed, female-headed a much larger share of the national poverty gap than households are more likely to be poor their population share would predict. Female-headed households constitute 30 percent of all rural Male-headed households constitute the majority- smallholder households but represent 42 percent of 75 percent-of all households in Malawi and the chance the rural poverty gap using the 40"h percentile income of a male-headed household being among the poorest cutoff (see Annex 3: Table A.3.2). The gap and sever- 40 percent of households is one in three. By comparison ity indices also show that rural poverty is deeper in Thble 3.5 Dependency ratio by income group' Below 20th percentile Below 40th percentile Above 40th percentile Average household size 5.07 4.83 4.12 Average dependency ratio 1.36 1.19 0.92 This is the ratio between household members aged 15 to 64 and household members under 15 and over 65 years of age. Source: HESSEA 1990/1991 Thble 3.6 Household composition by gender of household head Rural Urban National Measures of Household Female- Male- Female- Male- Female- Male- Composition headed headed headed headed headed headed Mean dependency ratio 1.43 0.96 1.24 0.84 1.42 0.94 Average number of household 1.88 2.57 2.36 2.74 1.9 2.58 members 15 years and over Proportion of hh members 0.57 0.63 0.59 0.65 0.57 0.63 15 years or older Average household size 3.8 4.7 4.5 4.9 3.8 4.7 Source: HESSEA 1990/1991 Profile of Poverty for Malawi 1 31 households headed by females. than households with a female head who is married or The Southern rural region contains the largest single (Table 3.8). Households with a female head who number of female-headed households and is also the is single are the least likely to fall in this group. One can region with the highest prevalence of poor female- speculate that single female household heads are also the headed households. A regional decomposition of least likely to have families to provide for since these household headship (Annex 3: Table A3.5) shows that households have a lower dependency ratio than other about 60 percent of female-headed households in the female-headed households (Table A3.9). This group of Southern region are poorer compared to roughly one- female-headed households may be a little better-off. third of the female-headed households in the North- The prevalences of poverty among households with ern and Central rural areas. a female head of child-bearing age (between 20-35 years) are much higher for the divorced and widowed than for Most female heads of households are the married and single. (Annex 3: Table A3.7). The ma- divorced, widowed or single jority of households headed by an older woman are among the poorest 40 percent of the population irre- Almost all male household heads (95 percent) are mar- spective of marital status. Simler and Quisumbing (1992) ried. By contrast, only one-third of all female heads of find that household incomes, expenditures, and poverty households are married, another third are divorced, 24 status do not differ significantly by the marital status of poercent are widowed, and 12 percent are single. Thus, the female household head. This analysis, however, in- most female household heads do not have a spouse. dicates that marital status and age explain some of the Households headed by divorced or widowed women variation in prevalence of poverty among female-headed are more likely to fall below the 40th percentile cutoff households. Table 3.7 Poverty prevalence among male- and female-headed households Percent of households Location/source Male-headed Female-headed Percent of households below the 40th percentile - national (HESSEA) 37 50 - among rural smaliholders (NSSA) 35 50 Percentage of the national poverty gap for 58 42 rural smaliholders (NSSA) Source: HESSEA 1990/1991, NSSA 1992/1993 Table 3.8 Table 3.9 Poverty in female-headed households Dependency ratio by marital status of household head for female-headed households by marital status Martal status Marital Average Poverty status Married Single Divorced Widowed status dependency ratio Below 20th percentile 26 29 28 28 Married 1.68 Below 40th percentile 48 43 53 52 Single 1 .10 Divorced 1.41 Widowed 1.21 Source: HESSEA 1990/1991 Source: HESSEA 1990/1991 32 32~~~~~~~ Malawi * Profile and Priorities for Action Female-headed households who receive regular cash 3.4 The geography of rural remittances are poorer poverty Twenty-five percent of all female-headed households regularly receive cash remittances from friends or rela- Poverty in Malawi is predominantly rural. Because pov- tives who were not living in the household at the time of erty and access to basic services are not uniformly dis- the survey. Only 5 percent of male hotiseholds receive tributed across the country, targeting resources and in- such remittances. terventions to areas where the prevalence and severity of Contrary to what one might expect, the preva- poverty are higher may be an effective approach to re- lence of poverty is higher for the households who re- ducing poverty. Interventions such as the provision of ceive cash remittances than for those who do not, basic education, health services, water and feeder roads except for the widowed who for the most part are older can lay the foundation for long-term, broad-based eco- women (Table 3.10). The difference is particularly nomic growth that will provide opportunities for income- clear for households headed by a married woman- generation for all. Providing such infrastructure through 69 percent among those who receive cash remittances labor-intensive public works in poor areas will also reduce fall below the 40th percent cutoff compared to 48 poverty in the short-term by providing employment op- percent for those who do not receive remittances. This portunities. In order to inform decisions about which areas may mean that if a household is already poor, cash of the country may benefit most from targeted interven- remiiittanices are likely to be too small to raise the tions, this section examines where the poor are located.' household out of poverty. An indication of this is also found in Peters' research (1992). She argues that fe- The most populated male-headed households dependenit on remittances ADDs in the Southern from husbands are becoming poorer as wage-work has and Central regions have become more scarce. Because the majority of widows the most severe poverty are older than 50 years of age (64 percent), it is likely Rural incomes vary dramatically between ADDs. The that they are dependanits who receive cash remittances densely-populated areas in the southern parts of the coun- from their children or relatives. try have the lowest mean incomes. As the very high Gini This findinig is consistent with the finding of this coefficients indicate, substantial inequality exists within report that the poorest households are those most de- ADDs. Kasungti ADD is an exception: it is the pendent on off-farm inlcomile. Many smaliholders in wealthiest ADD and it also has less inequality than Malawi have too little land to sustain their families. In- others (Figure 3.6). come from sources other thani the farm is therefore more There is marked variation in the prevalence of pov- common in the poorest households. erty by ADD. Three ADDs, Blantyre, Machinga, and Table 3.10 Percentage of poor female-headed households receiving cash allowances by marital status Marital status Reception of Married Single Divorced Widowed remittances by No Regular No Regular No Regular No Regular expenditure level allowances allowances allowances allowances allowances allowances allowances allowances Households below 20th percentile cutoff 27 32 33 42 35 40 35 28 Households below 40th percentile cutoff 48 69 47 50 61 66 62 55 Source: HESSEA 1990/1991 Profile of Poverty for Malawi 1 33 Lilongwe, have the largest populations and account for ent ADDs. Mzuzu ADD has a prevalence slightly 66 percent of the sample population (Table 3.11). More- higher than that of the country as a whole, while over, 76 percent of the total poverty gap for the rural Kasungu, which is Mzuzu's southern neighbor, has smallholder population (the sum of deviations in total the lowest prevalence of poverty in the country. De- smallholder household income below the 40th percen- spite having relatively high prevalences of poor house- tile) is found in these three ADDs. Along with Shire holds, Mzuzu and Shire Valley account for only 7 and Valley, these three densely-populated ADDs have 8 percent respectively of the aggregate rural prevalences of poverty above the national average and, smallholder poverty gap. therefore, account for higher pro- portions of all poor households than _ of the total population. Blantyre ADD, for example. contains 25 per- Figure 3.6 cent of the sample household popu- Mean annual household incomes and Gini-coefficients cen of th apehtshlou for rural smailholders by ADD lation, vet it contains 32 percent of the households below the 40th per- centile cutoff. Gini-coefficients: Blantyre Blantyre .51 Mzuzu is poorer than Shire Valley Shire Valley .53 other ADDs in the . Lilongwe .52 North Lilongwe Machinga .55 Machinga Mzuzu 554 o ~~~~~~~~~~~~~~~Salima .56 ADDs in the North have relatively < Mzuzu Karonga .58 small populations and lower than Salima Kasungu .46 average prevalences of poverty, Karonga which explains their low shares of national poverty (see Table 3.12). Kasungu Within the Northern region, however, there are substantial dif- 0 100 200 300 400 500 600 700 800 900 ferences in the prevalence of Income (MK/AE/Year) poorer households among differ- Source: NSSA 1992/1993 Table 3.11 Table 3.12 Percentage distribution of Poverty indices by ADD using the income group and ADD 40th percentile income cutoff Percentage Percentage of Percentage of household households below of the national ADD Prevalence Poverty gap Severity ADD population the 401h percentile poverty gap Karonga 30% 14% 8% Karonga 3 3 2 Mzuzu 42 22 15 Mzuzu 7 7 8 Kasungu 9 4 2 Kasungu 11 3 2 Salima 9 4 2 Salima 6 5 5 Salima 32 16 10 Lilongwe 20 22 22 Lilongwe 44 22 14 Machinga 21 22 23 Machinga 42 22 14 Blantyre 25 31 31 Blantyre 49 25 16 Shire Valley 7 7 7 Shire Vclley 42 22 15 Total 100 100 100 Source: NSSA 1992/1993 Source: NSSA 1992/1993 34 Malawi * Profile and Priorities for Action Smallholders in Figure 3.7 Kasungu are much better..off Percent population classified "Poor" than other below 40th percentile smalnholders in Malawi by rural development project Kasungu, Karonga, and Salima ADDs have the highest mean smallholder incomes (Figure 3.6). Kasungu is in a class of its own with a smallholder house- hold income level more than twice as high as the income level in the poorest ADDs. The other poverty measures (prevalence, gap, and severity) are also con- sistently lower in Kasungu than in the other ADDs. The very low prevalence, depth, and severity of poverty found in the Kasungu ADD deserve further study. Kasungu has the highest av- erage landholding size and the lowest frequency of very small landholdings among ADDs. 5 J2Ha 6 1 0 Source: NSSA 1992/1993 38 | Malawi - Profile and Priorities for Action Kasungu this reclassification had not yet led to signifi- of the poorest income group. Since average house- cantlysmallerlandholdingsintheremainingsmallholder hold size is higher for lower income groups, the sector. However, without more recent data on landhold- difference in cultivated area per adult equivalent ings in this ADD, no firm conclusion can be drawn. across income groups is even greater than the dif- ference in cultivated area. The average area per Households with more members adult equivalent is about 6.7 times greater in the 10th have less land per person compared to the first decile (Annex 3: Table A3.8). Even though average household size is posi- Poverty is closely associated with small cultivated tively associated with land under cultivation, areas. The average area per household in the 10th smallholders with more land have higher land ar- income decile is close to five times the average area eas per adult equivalent than smallholders with less land. Thus, land constraints facing the Table 3.14 poorest households are even more se- vere than the raw data on1 area culti- Percentage of households in each ADD vated indicate. in each cultivated area class Cultivated area category Female-headed households ADD < .5Ha .5-1 Ha 1-2Ha >,2 Ha Total are more likely to farm Karonga 47 30 20 3 100 small areas Mzuzu 32 29 27 13 100 Kasungu 12 27 41 20 100 Female-headed households are much more Salima 47 30 17 6 100 Lilongwe 32 37 24 7 100 ~~~~~likely than male-headed households (51 Machinga 41 36 20 3 100 percent and 37 percent respectively) to Blantyre 61 27 11 2 100 cultivate less than 0.5 hectares of land (An- Shire Valley 49 26 19 6 100 nex 3: Table A3.9), the category with the All Malawi 41 31 21 6 100 lowest household incomes. Male-headed Note: Numbers may not total to -_______ households are about twice as likely as fe- 100 because of rounding. Source: NSSA 1992/1993 male-headed households to cultivate more than one hectare. Female-headed house- holds thus cultivate less land than do those sable 3.15 households headed by males. Distribution of cultivated areas and income deciles Income Cultivated area categories Cultivated land size alone deciles < 0.5 Ha 0.5 - 1 Ha 1 - 2 Ha > 2 Ha Total does not predict household 1 81 (19) 15 (5) 4 (2) 0 (0) 100 income 2 67(16) 26 (8) 7 (3) 0 (1) 100 3 58 (14) 31 (10) 9 (4) 2 (2) 100 Although there is a strong association be- 4 48 (11) 36 (12) 15 (7) 1 (2) 100 tween the area ofland under cultivation and 5 39 (10) 39 (13) 20 (I10) 2 (4) 100 . .. 5 39(10) 39 (13) 20 ( 10) 2 (4) 100 income, it is not possible to equate having 6 36 (8) 38 (12) 23 (11) 3 (5) 100 little land with being poor. Typically, house- 7 30 (7) 36 (12) 29 (14) 5 (8) 100 8 22 (6) 31 (10) 33 (16) 10 (16) 100 holds with smaller landholdings are in the 9 20 (5) 33 (11) 33 (16) 13 (21) 100 lower-income deciles, while those house- 10 14 (3) 23 (7) 38 (18) 26 (40) 100 holds with larger landholdings are in the Total 100 100 100 100 higher-income deciles. Households with Source: NSSA 1992/1993 large areas of land under cultivation are al- most exclusively in upper-income groups; Profile of Poverty for Malawi F1 ~ ~ ~~ 39 only 7 percent of families cultivating more than two hect- smallholders used credit, but only 16 and 15 percent ares are found in the lower 50 percent of the income distri- of the poorer and poorest households received credit bution. (Table 3.16). Some households with small areas of land under culti- There is a negative association between the adop- vation are in the upper three income deciles - 14 percent tion of improved agricultural technologies and poverty. of families cultivate less than 0.5 hectares and 20 percent of Poorer households are less likely than other households families cultivate less than one hectare (Table 3.15). These to use purchased inputs. In most areas where modern few households are able to generate relativelv high incomes inputs are widely used, the prevalences of poor house- even though they have small landholdings under cultiva- holds are lower. There are generally higher yields of hy- tion. because they also have paid work outside the farm. brid maize, and chemical fertilizers are more commonly This finding is relevant to the design of programs to used in areas with lower prevalences of poverry (Annex target the poor, as the variable "cultivated area" is not 3: Figures A.3.4 and A.3.5). This finding could indicate always a good indicator of poverty. Potential errors of two different things. First, it could be concluded that inclusion (a non-poor household may be misclassified as access to purchased inputs raises farm productiviry sig- poor) could occur if the variable "cultivated area" alone nificantly and thus widespread input use should be pro- is used to target poverry programs because some house- moted as a means of lowering poverty. An alternative holds with little land are not poor. explanation is that inputs are marketed and also adopted only in those areas with significant agronomic potential Poorer smaliholders use and thus input use is caused by high incomes. few inputs and lack assets In some areas of the country input use is low yet poverty is also low. For example, few smallholders in the In general, Malawian smallholders own few productive Karonga RDP use chemical fertilizer, (probably because assets, have only small plots of land, and do not use mod- of geographic isolation). There is also a relatively low ern inputs or credit. Poorer smallholder households are prevalence of poverty in Karonga. In the Rumphi RDP less likely to use modern inputs, oxen traction, or hired yields and input use are high, yet poverty is widespread. labor than smallholder households as a whole. Poor These descriptive associations between input use and smallholders have little access to credit. According to poverty shed little light on the causal relationships be- the NSSA, approximately 22 percent of Malawian tween input use and poverty outcomes. Without more Table 3.16 Proportion of smaliholder households using purchased inputs by income groups Percentage of households: All below the below the Input ___ Malawi 20th percentile 40th percentile Fertilizer (% using) 42.0 31.0 33.0 Pesticides (% using) 3.0 1.0 1.0 Hired Labor (% hiring) 37.0 29.0 29.0 Credit (% using) 22.0 15.0 16.0 Oxen (% using) 3.0 1.0 1.0 Mean Cultivated Area (Ha) 0.807 0.388 0.488 Mean Per Adult Eq (Ha/AE) 0.224 0.091 0.118 Source: NSSA 1992/1993 40 | Malawi - Profile and Priorities for Action informatioin on such causal relationships, it is difficult cess to purchased inputs)"3 is partly explained by the to make policy prescriprions. For example, it is impos- link between extent of cultivated area and use of credit. sible to know whether a program to increase the access For every ADD, there is a monotonically positive re- of the poorest hoLIscholds to fertilizers will reduce pov- lationship between the cultivated area and the per- ertv. IPoorer households might face fundamental con- centage of households using credit (Table 3.17). straints to their capacity to benefir from fertilizer use, Poorer households cultivate smaller areas and have less and their limited use of fertilizer might simply be a mani- access to credit and inputs that would enhance the festation of this. Additional analysis is needed to inves- productivity of the land and labor than do other tigate some of these issues of causality. households. The relationship between poverty and limited ac- Smallholders in the most geographically-isolated cess to credit (and hence, given the institutional ar- ADDs such as Shire Valley and Karonga use substan- rangements that existed at the time of the survey, ac- tially less fertilizer than average. Transport costs may Table 3.17 Percentage of households with access to credit by area cultivated, income group and ADD Cultivated Household income group below: area category: 40th 20th ADD All Malawi < 0.5 Ha 0.5 - 1 Ha 1 - 2 Ha > 2 Ha percentile percentile All Malawi 23 8 24 36 48 16 15 Karonga 12 4 13 24 51 8 7 Mzuzu 32 11 36 44 58 25 27 Kasungu 36 11 31 39 50 14 12 Salima 20 8 23 40 42 11 7 Lilongwe 36 17 39 50 56 27 26 Machinga 15 5 17 28 44 13 11 Blantyre 11 7 17 22 17 11 11 Shire Valley 12 3 10 29 36 5 4 Source: NSSA 1992/1993 Table 3.18 Percentage of ownership of productive assets by area cultivated and income group Cultivated Household income group below: area category: 40th 20th Item All Malawi < 0.5 Ha 0.5 - 1 Ha 1 - 2 Ha > 2 Ha percentile percentile Hoes 95 94 97 97 97 94 93 Ploughs 2 4 2 3 10 1 1 Sprayers 2 1 1 3 6 1 1 Water cans 23 1 2 22 37 53 16 14 Wheelbarrows 1 1 1 1 2 1 1 Oxen 3 1 2 4 14 1 1 Ridgers 1 0 1 2 6 1 1 Bicycles 18 11 17 25 38 12 12 Source: NSSA 1992/1993 Profile of Poverty for Malawi |~ ~ 41 make it prohibitively expensive to deliver inputs there. Crop profile Low input use might also be attributed to credit dis- tribution problems. Resource-poor smaliholders Local maize is the main crop for the (those with less than 0.5 hectares) in Karonga and poorest smaliholders Shire Valley are about one-fifth as likely to use credit as resource-poor smallholders in Lilongwe ADD, Malawian smallholder production is concentrated in maize, which has better access to credit institutions. but other crops are important. Cropping patterns vary by Another dimension of the poverty-agriculture re- geographical location, with cassava grown in all regions lationship is illustrated by examining patterns of own- around the lakeshore, and rice grown in localized areas. In ership of productive assets by class of cultivated area Karonga and Salima RDPs, fairly large areas are devoted to and poverty group. Most Malawian smallholders own rice cultivation; minor smallholder cultivation of rice is only a hoe (Table 3.18). A few also own a water can found in other RDPs. Somechiliesandvegetablesaregrown and a bicycle. Households with larger cultivated ar- in RDPs close to urban centers in all regions. Maize, beans, eas have more assets to use in farming. and groundnuts are the principal food crops in the Central region. Tobacco is the largest occupier of smallholder land Female-headed poorer households (after maize) in the Central region. Beans and some millet have the fewest productive assets are grown in addition to maize in areas of the Northern region. In the South, millet and sorghum are widely planted, Female-headed households have the fewest resources of all smallholder households. Female-headed households cultivate the smallest areas, have least access to credit, Table 3.19 and are the least likely to own productive assets (Table Farm characteristics by gender of 3.19). These households face large obstacles to generat- household head ing income through agriculture. The reasons for this find- Farm characteristics Male- headed Female- headed ing are not well understood. It would be necessary to do Area Cultivated (Ha) 0.897 0.631 an analysis of off-farm activities for these households, Area/Adult Equivalent 0.231 0.206 particularly to find out whether the household receives Adult Equivalents 4.42 3.77 remittances from a spouse working elsewhere, before it Share of Hybrid Maize (% land) 22 16 1~~~~~ Share of Maize (% land) 78 83 would be possible to deterninle how other sources of Share of Maize Toband) 7 8 Share of Burley Tobacco (%/) 3 1 income affect their overall earnings. Hired Labor (% hiring land) 40 29 Another important question is how land and labor Fertilizer (% using) 46 30 productivity varies between male- and female-headed Pesticides (% using) 3 1 households. Lower input use by female-headed house- holds (such as fertilizer, labor, and credit) might Source: NSSA 19921993 be due to differences in land quality. If this lower use is caused by lower derived demand for in- ptits (caused in turn by low land quality) rather Table 3.20 than by constrained access to markets (or ration- Percentage of area under cultivation with ingorotherarket imperf . th a dif- different crops by size of cultivated area ing or ocher miarKet imperfections), theni a dif- ferent set of policy implications arises. The issue Cultivated Burley Improved Local of determinants of inptit use by female-headed area tobacco maize total maize households needs more investigation before spe- Less than 0.5 Ha 1 26 63 cific policv recommendations can be made. 0.5- 1 Ha 3 28 61 1-2Ha 5 30 56 More than 2 Ha 7 34 49 Source: NSSA 1992/1993 42 | Malawi * Profile and Priorities for Action along with some beans, soybeans and groundnuts. Cotton do households with incomes above the cutoffs. Female- is produced in many RDPs in the Southern and Central headed households have lower percentages of land devoted regions. Kasunlgu ADD. the area of highest income, also to hybrid maize production and higher percentages of land has the highest concentration of maize and tobacco. Out- devoted to local maize. side these two crops, very little is planted by smallholders Crop profiles show that the ADDs with the highest in the ADD. numbers and concentrations of poorer and poorest house- Maize is the most widespread smallholder crop, and holds (Blantyre, Machinga, and Lilongwe) have the high- the cropping pattern varies with the size of cultivated area . est proportions of smallholder-cultivated areas allocated to T he larger smallholders cultivate a greater portion of their maize (Annex 3: Figure A.3.4). These ADDs also have rela- land to improved hybrid varieties of maize (Table 3.20). tively high proportions of land area allocated to hybrid and Poorer smallholders devote larger shares of land to maize composite maize varieties. The adoption of these varieties and larger proportions of maize areas to local maize than does not, in itself, guarantee increases in yields and incomes because of the limited use of fertilizer and because of poor soil conditions. Table 3.21 In some areas, agricultural Percentage shares of total cultivated area allocated practices and yields are not closely to crops by poverty group linked to poverty. Those RDPs Households with incomes where this is the case should be All below the below the Crops Malawi 40h percentile 20h percentile examined in detail to find out Burley tobacco 3 0 0 which types of income-generating Hybrid & composite maize 28 26 26 activities are common in those ar- Hybrid maize 20 19 20 eas. Similarly, areas with high Local maize 48 57 58 Local maize/total maize 60 64 65 yields or widespread use of mod- ern inputs that also have high lev- Source: NSSA 1992/1993 els of poverty should be examined. Table 3.22 Shares of household maize requirements met by own production Household income status 40' percentile cutoff 20th percentile cutoff ADD All Malawi Below Above Below Above All Malawi 1.48 0.64 2.03 0.39 1.74 Karonga 0.83 0.46 0.99 0.33 0.90 Mzuzu 1.77 0.64 2.57 0.43 2.14 Kasungu 2.69 0.61 2.91 0.38 2.76 Salima 1.50 0.61 1.93 0.36 1.71 Lilongwe 1.82 0.94 2.55 0.65 2.17 Machinga 1.40 0.69 1.88 0.48 1.66 Blantyre 0.88 0.44 1.30 0.28 1.08 Shire Valley 0.91 0.36 1.29 0.23 1.08 Note: This table was created by dividing total household maize production by total maize requirements based on adult-equivalent household size. Requirements were assumed to be 200 kg per adult equivalent. The "All Malawi' column refers to all smoliholders, and the overall self-sufficiency ratio of 1.48 is not for the entire Malawian population: urban dwellers and estate tenants are not included. Source: NSSA 1992/1993 Profile of Poverty for Malawi 43 The Thiwi RDP in the Central region is one place The poorest smallholders have not been able to directly where high agricultural yields have apparently not re- take advantage of the new policy allowing smallholders to duced the prevalence of poorer households. produce burley tobacco.'4 There is, however, evidence that increased smallholder burley production is indirectly re- Poorest smaliholders do not farm ducing poverty by generating greater economic activity in burley tobacco areas where burley production is common. Areas of the country where burley production by smallholders is most Burley tobacco is the most profitable smallholder crop in common are also those areas with the lowest prevalences of Malawi. In the past, smallholders were not allowed to grow poverty. Kasungu ADD has by far the largest share of land burley tobacco: burley quotas, which were allocated only devoted to burley production and has consistently low lev- to estate producers, have historically been a major source of els of poverty. There appear to be substantial spillovers from economic rents. This began to change in the early 1990s the cultivation of burley tobacco that create income-gener- when burley quotas began to be given to smallholders. The ating possibilities both on and off the farm. Some of these NSSA provides some evidence about the impact of this spillovers are evident in the higher use of fertilizer in Kasungu change on poverty, although the full impact had not been where 69 percent of smallholders use fertilizers. Fven the felt by 1992/1993 when the survey was fielded. poor in Kasungu are more likely to apply fertilizer than are The NSSA data show that poor smallholders were not other Malawians; 46 percent of the hoLiseholds below the yet producing burley tobacco despite the relaxation of the 40th percentile cutoff in Kasungu use fertilizer. quota system (Table 3.21). To some extent this may be be- cause a formerly poor household that began to plant burley Poorer households are tobacco, and thus to earn a higher income, would be less net purchasers of maize poor at the time of the survey. However, few smallholders in the smallest cultivated areas grew burley tobacco at the Poorer households produce, on average, only about 64 per- time of the survey. Burley production was more prevalent cent of their maize requirements, while the poorest house- (even on a per-hectare basis) among households with large holds (those below the 20th percentile) produce less than cultivated areas (seeTable 3.20). Female-headed households 40 percent of their requirements. In some districts (Karonga, devoted, on average, only about 25 percent as much land Blantyre, and Shire Valley), the poorest households pro- to burley tobacco production as male-headed households. duce about 25 percent of their requiremenits (Table 3.22). These households must rely on1 off-farm work to be able to purchase their maize requirements. Even under ideal conditions, the land that the Box 3.1 smallholders in the lower three income deciles can cultivate Food security and landholding size is not sufficient to provide sufficient food for the house- Roughly 200 kg of maize are needed annually to hold (see Box 3.1). meet the calorie needs of an adult Malawian. Assume that all land is planted with hybrid maize and that the yield is 2,500 kg/ha. If input costs (fertilizer and pesticides) are financed out of the production of maize, there will be approximately 1,510 kg/ha of 3.6 Limited income from maize surplus available to consume after paying off-farm sources input costs (at August 1994 input prices). A typical hectare of land will thus produce sufficient maize Few previous studies of poverty in Malawi have ex- for 7.55 adult equivalents (1510/200) under ideol amined the types of off-farm employment that exist conditions. Therefore, to meet maize consumption in rural areas and determined how they vary by pov- needs out of own production, each adult needs 0.132 hectares. erty status and other factors. By using the NSSA, which allow examining types of off-farm employment, 44 44 -1 ~~~~~~~~~~~~~~~~~~Malawi * Profile and Priorities for Action Peters, for example, discusses income sources broken sources outside agriculture may have been underesti- down into broad categories such as on-farm non-crop mated by the NSSA for several reasons. First, the agricultural income, self-employment, and agricul- 1992/1993 NSSA took place in the year following tural wages but does not systematically analyze types the most severe drought in recent memory. It is likely of employment (for example, fishing or tin-smithing) that cash was scarce in rural areas and, as a result, off the farm. Simler and Quisumbing (1994) state that there were fewer income-earning opportunities than an investigation of these income sources is a priority usual. Few people had spare cash to hire workers, so research area, as these forms of off-farm income tend paid agricultural work was less plentiful than normal. to be of critical importance to poor smallholders. This might have created a lower than normal propor- tion of smallholder income earned off the farm. Sec- Off-farm employment ond, yields are usually higher than normal following a drought. Because of these higher yields, incomes Smaliholders earn from agriculture were higher than expected. This fac- liftle off-farm income tor might also have led to lower than normal propor- tions of income earned off the farm. Third, while ag- Agriculture accounts for about 73 percent of total ricultural income was measured very accurately in the smallholder incomes in the sample. The second larg- NSSA, data on off-farm income was gathered using est income source is paid agricultural labor (ganyu) respondents' own recollections of the situation four which accounts for 7 percent of the total (Table 3.23). months later. Many authors agree that, when income The third and fourth largest sources are other non- is self-reported, it tends to be tinder-estimated unless specified activities and professional work."' Other extreme care is taken in collecting the information sources provide very little income. (see Annex 1). The proportion of income earned off the farm (about 27 percent) estimated using the NSSA data is Off-farm employment slightly lower than the proportions (about 35 percent) provides a higher share of found in the Simler and Quisumbing (1994) and Pe- income for poorer households ters (1992) studies. The proportion of income from Shares of total income from own-farm sources are much lower for households Table 3.23 below the 40th percentile than for Percentage of smallholder household income households above (Table 3.23). All from different sources by income group Malawian smallholders receive about 73 Households below percent of their income from own-farm All the20^ the40th sources, while 68 and 61 percent come Activity Malawi percentile percentile from own-farm sources for the house- On-farm income 73 61 68 holds below the upper and lower cutoffs Paid agricultural employment 8 11 10 respectively. This finding is consistent Fishing 2 2 2 with other studies (notablv Simler and Beer brewing 2 6 4 Q Basket weaving 1 3 2 Quisumbing. 1994 and Peters, 1992). Brick making 1 2 2 Since the poor have smaller landholdings Charcoal making 0 0 0 than households above the 40th percen- Professional income 6 3 3 tile and since in manv cases this land is Other income 7 14 1 1 insufficient to ensure even a marginal ex- So_rce. N_SA 1992/1993 istence, thev must rely on off-farm incomes Source: NSSA 1992t1993 to survive. Profile of Poverty for Malawi 4 As expected, members of poorer households are more Hlouseholds above the 40th percentile have dif- likely to sell their labor to other farms (ganyu) and their ferent patterns in their sources of income than do shares of income from these sources are higher than the poorer households. Ploorer households earn much Malawian average (about 10 percent of income for the lower shares of their incomes from professional em- poor versus 8 percent for households above the cutoff). plovment and higher shares from off-farm employ- Ganyu labor is clearly an important source of income ment in agriculture (ganyu). These results confirm for the poorest. two things: people with professional employment are Poorer smailholders also receive higher income among the few relatively well-off people in rural ar- shares from beer brewing, basket weaving, and other eas, and agricultural laborers are in a very disadvan- sources than do households above the income cutoff. taged position. In the ADDs with the highest densities and The employment of a head of household or a prevalences of poverty (Blantyre, Machinga, Lilongwe, spouse in a professional occupationi (see Annex 2 for and Shire Valley) the poor tend to receive even higher a description of how this variable was created) is clearly shares from these sources than do other households. and strongly related to income. Onlv I 1 and 3 per- For example, beer brewing accounts for about 5 per- cent of households whose head is employed as a pro- cent of the income of poorer households and only I fessional are in the poorer and poorest categories re- percent of the income of households above the 40th spectivelv. For smallholder households where the percentile in the Lilongwe and Shire Valley ADDs. spouse of the head is employed as a professional, the Income from fishing as a share of total income prevalences are 7 and 2 percent respectively. does not vary much according to the poverty status of the household. However, there are differences in Sources of income the relationship between income status and shares vary across regions earned through fishing by ADD. In ADDs where fish- ing can be expected to be lucrative (in the lakeshore The percentage of total household income derived areas-see Table 3.24), shares of income from fishing from different employment activities varies substan- are significantly higher for households above the pov- tially among areas of the country. On-farm erty cutoff. For example, in Salima and Machinga, productioni's share of total income varies significantly, shares from fishing are 4 and 6 percent of the incomes constituting between 65 percent of smallholder in- of households above the 40th percentile, while they come in Blantyre and 91 percent in KasungLu (Table are 1.5 and 4 percent for households below the cutoff. 3.24). On-farm sources tend to constitute the high- Table 3.24 Shares of income received from different sources by ADD Income/Activity Karonga Mzuzu Kasungu Salima Lilongwe Machinga Blantyre Shire Valley On-farm 79 79 91 77 77 67 65 70 Paid agricultural employment 2 4 3 5 7 10 11 7 Fishing 3 2 2 3 1 5 1 2 Beer brewing 2 3 3 1 3 1 3 3 Basket weaving 1 1 - 1 1 1 2 1 Brick making 1 2 2 1 2 1 1 1 Charcoal making 0 0 0 0 1 0 0 0 Professional 7 5 3 5 4 6 10 7 Other income 5 5 3 6 5 9 8 9 Source: NSSA 1992/1993 46 46 -1 ~~~~~~~~~~~~~~~~~~Malawi * Profile and Priorities for Action est shares of total income in the land-abundant North- likely to be poorer. ern region. Kasungu's economv is clearlv driven bv on-farm Female household heads have less agricultural production. The income shares of all other off-farm employment than male heads activities except beer brewing and brick making are well below Malawian averages, and the share of pro- The ability to sell labor in the market appears to be fessional income is about 64 percent of the level of an important difference between male and female the next lowest AD)D (Iilongwe). Smallholders in heads of household. Off-farm labor supply constraints KasuLngu do not seem to engage in many economic mav contribute to female-headed households having activities other than farming their own land, and they lower incomes. Detailed studies of female time allo- also do not appear to sell significant quantities of their cation may help to reveal which constraints are the labor to estate owners or to other smallholers. most binding: one would expect that time spent gath- In contrast, in the Southern region of Malawi, ering water and fuelwood and pounding maize may sales of labor to other farms (ganyu) are substantially be high in total, thus creating labor shortage in the higher than in other areas of the country. The South- household. ern region, because of its intense land pressure, con- While income from off-farm sources is a small tains many near-landless workers engaged in farm share of total income for poorer households, it is an production activities for estates and other even lower proportion for female-headed households. smallholders. As expected, shares of income from own- More than half of all female heads describe their em- farm production are lowest in these land-scarce areas. ployment status for the year as self-employed own- Fishing, brick making, and other activities also account worker: fewer than 10 percent are employees show regional variations. Fishing is obviously local- or wage laborers. A quarter of female heads indicate ized to areas around Lake Malawi. In areas where fish- that performing domestic duties is their usual em- ing is likely to be good, particularly in the Salima and ployment status. These patterns contrast with male Machinga ADDs, poorer households generate smaller heads of household. They report similar levels of self- shares of their total incomes through Fishing than do employment as own-account workers, but more of other households. Fishing is positively associated with them are wage labors or employees (28 percent) than econiomic status in such areas. In areas with less abin- female heads (8 percent), and fewer of them perform dant marinle resotirces, hoLIseholds that fish are more domestic duties (6 percent compared to 25 percent Table 3.25 Percentage distribution of income sources by headship and income group Percent of Female-headed households: Percent of Male-headed households: Predominant all female- below the below the all-male below the below the income headed 40th 20th headed 40th 20th source households percentile percentile households percentile percentile Salary/wage 8 7 8 25 24 22 Profits 13 14 14 19 22 25 Non-cash income 0 0 0 1 1 1 Regular allowances 9 12 12 1 2 3 Other cash 7 10 13 5 11 15 Net rents 9 14 22 4 7 11 Own-account income 54 44 31 46 35 23 Source: HESSEA 1990/1991 Profile of Poverty for Malawi 47 for female heads). Household members who are un- average number of heads owned increases as the income employed or who only perform domestic duties are deciles increase. For example, the average number of more likely to be found in the very poorest house- poultry owned by households in the first two income holds (Table 3.25). deciles is about half the number owned by the upper decile. Cattle ownership is three times higher in the high- Medium-sized farmers est decile than in the lowest decile. Wealthier households are a mixed group own more goats than poorer households, but the differ- ence is not as pronounced as for poultry and cattle. Pig Better-off smallholder households are a small group and sheep ownership is constant across deciles. that falls into two main categories. First, there are farmers who produce burley tobacco and surplus Livestock ownership varies by maize. They have benefited from policy changes that geography and income permit smallholders to grow burley tobacco and are likely to increase their maize production as a result of Much of the difference in owniership patterns by income the liberalization of maize prices. Second, there are status is explained by the geography of ownierslip and some high-income, medium-sized households who the geographical distribution of income. Karonga, also receive income from professional employment Mzuzu, and Kasungu are ADDs where poverty levels elsewhere. A third group of smallholders appear to be are relatively low and more livestock tend to be ownied low productivity producers. This latter group requil-es (Annex 3: Fable A.3.12). Lilongwe, Machinga, and further study to identify whether their low produc- Blantvre, ADDs with high levels of poverty, contain tivity is due to low producer prices or poor manage- many sniallholders who have small cultivated areas and ment practices. who, as expected, own few poultry and cattle. When geographical location is taken into account, Livestock there is a much weaker relationship between income and livestock ownership. Poultry ownership is positively cor- In certain places outside the principal maize and to- related with income in all ADDs. Goat, pig. and sheep bacco growing areas of the country, many poor fami- ownership has no correlation with income within ADDs. lies base their livelihood on livestock rather than on Cattle owniership is also not correlated with income crops and off-farm income. In computing the income within ADDs, except in the Karonga ADD where there measure from the NSSA data, livestock sales and in- is a negative correlation. comes from livestock production were not included These results show that the ownership of livestock because it would have been difficult to value this in- (especially of cattle and poultry) by smallholders is closelv come (see Annex I for more details). However, the related to the area of cultivable land they ow n and that livestock module of the NSSA allows us to examine omitting revenues from livestock ownership from the the relationship between livestock ownership and in- income measure did not introduce major biases into the comes (this comparison is shown in Annex 3: Tables analysis. The lack of correlation between livestock and A3.10 and A3.1 1). income within ADDs means that the levels of income measured by the NSSA are probably lower than they Livestock ownership is positively should be (because they exclude the value of livestock), related to income but the overall ordering of incomes is unaffected if live- stock revenues are included. This ordering of incomes is Malawian smallholders generally own few livestock, but critical to the poverty analysis. livestock ownership is positively correlated with income The omission of livestock in the income measure (Annex 3: Table A.3.1 1). For most types of livestock, the tends to exaggerate differences in incomes in rural 48 48 -1 ~~~~~~~~~~~~~~~~~~Malawi C Profile and Priorities for Action Malawi, as differences in livestock ownership from low- Landholding size and location and to high-income deciles are not as great as the differences smaliholder incomes in incomes across these deciles. Of course, the fact that relatively few livestock are owned means that livestock Land under cultivation has a strong positive impact incomes are verv small, as sales of, for example, eggs or on smallholder income. A I percent increase in avail- milk are likely to be minuscule. Thus, any bias intro- able land will increase household income by 0.6 per- duced into the NSSA analysis by excltiding these incomes cent. If, for example, a smallholder family with 0.5 is probably small. hectares is provided another 0.2 hectares to cultivate, their income per capita should rise by about 40 per- cent. According to the results, a household must have at least 0.54 hectares to generate income necessary to reach the 40h percentile income cutoff. ' 3.7 Main poverty Incomes in Karonga, Kasungu, and Salima are, determinants in Malawi holding all other factors constant, significantly greater than those in Shire Valley (the comparison category). Regression analysis allows us to simultaneously ana- In the other ADDs, incomes are lower than Shire Val- lyze the impact of several factors on household in- ley, although the difference is not statistically come (TIable 3.26). significant. Table 3. 26 Regression analysis: Determinants of smallholder incomes in rural Malawi Variable Estimate t-statistic Standard Error Intercept 5.63 87.72 0.06 Cultivated land 0.72 18.82 0.04 Adult equivalents -0.18 -26.69 0.01 Dependency -0.13 -3.22 0.04 Head's age -0.00 -2.76 0.00 Head primary education 0.01 0.40 0.03 Head secondary Education 0.10 2.62 0.04 Head's gender (male=0. female=l) -0.24 -8.39 0.03 Fertilizer' 0.24 0.86 0.28 ADD-level binary variables Karonga 0.38 4,73 0.08 Mzuzu -0.16 -1.23 0.13 Kasungu 0.46 2,76 0.17 Salima 0.31 3.07 0.10 Lilongwe -0.29 -1.83 0.16 Machinga -0,06 -0.63 0.10 Blantyre 0.06 0.63 0.09 Rz .340 - - N 10,663 - - Dependent Variable = Log (Adult Equivalent Income) Endogerous variable Instruments used include: presence of oxen in enumeration area, use of fertiizer in enumeration area, and adoption of improved maize varieties in enumeration area. All instr.ments are binary variaoles Profile of Poverty for Maai149 Female-headed households have significantly 3.8 The state of poverty in lower incomes per adult equivalent, even controlling Malawi for other factors such as landholding, education, and age. When all variables are controlled, the results sug- gest that female-headed households have incomes The profile describes the geography of poverty in about 79 percent of those of identical male-headed Malawi and analyzes some of the correlates of pov- households. This result is consistent with the mes- erty. The results are broadly consistent with findings sage in the rest of the profile and indicates that in- from previous studies, but the profile is the first to come-generating opportunities for female heads of use national data to quantify rural/urban poverty dis- households are limited. The age of the head of the tribution. The nationwide data make it possible to household is negatively associated with household in- quantify many previously hypothetical relationships come.S and to provide a more detailed examination of many Both family size and dependency ratio are nega- critical issues. tively related to income per adult equivalent, mean- The first major finding of the profile is that ing that per adult incomes are smaller in larger house- Malawian poverty is widespread and deep. The broad holds. Even controlling for the dependency ratio, large picture is one of extremely low standards of living family size is associated with lower income per adult and substantial inequality throughout the country and equivalent. This result suggests that family labor for especially in rural areas. Moreover, income inequality agriculture does not constrain income, and that op- is very high which, given the low incomes in Malawi, portunities outside agriculture are limited. indicates extreme deprivation at the lower end of the Lower levels of education of the head (literacy distribution. training or Standard 1-4) have no significant impact Several themes emerge from the HESSEA analy- on income, but higher levels have a strong positive sis. First, patterns of poverty reflect, to a large degree, impact (the comparison group is those household population distributions. Poverty is more prevalent heads with no education). The results indicate that in rural areas, and within rural areas, it is more preva- households headed by someone with higher levels of lent in the Southern region. Second, among cities, education have incomes about 10 percent higher than Lilongwe is most afflicted by poverty. Urban poverty those headed by someone with no education. As a does not follow the rural pattern, in that cities in the means of comparison, secondary education has the Central and Northern regions contain more exten- same impact on income as a 17 percent increase in sive poverty than Blantyre and Zomba in the South. land under cultivation. Third, BOMAs are more like rural areas than urban The tIse \not use of fertilizer does not emerge as a areas. Finally, the observed patterns of the prevalence significant determinant of income in this analysis. of poverty are relatively insensitive to the choice of However, this result should be interpreted with cau- poverty line, although the ranking of cities does de- tion. The instruments used to identify this variable pend on which cutoff point is chosen. may not have explained a significant proportion of In rural areas, substantial variations were found its variability. A model of agricultural productivity may in the distribution of the poor. Intra-rural inequality be needed to fully capture the effects of fertilizer use. is remarkably high, and pockets of poverty exist in 50 -1 Malawi * Profile and Priorities for Action many areas of the country. Consistent with other stud- less access to formal credit than do other smallholders. ies (and with conventional wisdom), the Southern These findings provide evidence that, despite govern- region was found to have the highest concentrations ment efforts to promote agricultural intensification, of poor households, the highest prevalences of pov- the poorest households have not been able to partici- erty, and the deepest poverty. Within the Southern pate in this process. Much of the inability to inten- region, however, there are substantial variations in the sify can be linked to landholding sizes; the small land- prevalence of poverty, and deep poverty exists within holdings of the poor may increase the risk associated other regions of the country. The areas along the with adopting hybrids and make it less likely that the Mozambique border deserve further scrutiny to see poor would receive credit. Poor households have also whether the reduction in refugee population has re- not benefited from increased opportunities to pro- duced poverty. duce and market burley tobacco or other high-value The rural economy in Malawi has historically crops. Such results imply that, if agricultural intensi- been dominated by agriculture. Poor smallholders fication is to have a role in reducing poverty, con- receive most of their income from own-farm agricul- certed steps need to be taken to ensure that the poor tural production: off-farm sources of income are lim- are actively involved in the process. ited and are closely linked to agriculture. For this rea- Households in the lower socioeconomic group son, much of the profile focuses on the relationship own very few productive assets; have less than one among agriculture, landholding sizes, and poverty. hectare of land; have no livestock; most are headed Although rural poverty is associated with small by a woman who has never attended school; the house landholdings it is not exclusively tied to landholding has a mud floor and roof of grass; these households size. Among households with extremely small land- lack sanitation and have no safe water source (see holdings there is a high prevalence of poverty. How- Box 3.2). ever, there are exceptions; mainly households earning The lowest socioeconomic group had the least in- off-farm income from professional occupations. On come and had the highest levels of malnutrition. Nine- the other hand, virtually all households with access teen percent of the households in the best-off group to more than two hectares of land are not poor, even fall below the 40th percentile cutoff compared to 56 in the Northern area of the country where land qual- of the households in the worst-off group. Among chil- ity is suspected to be lower. Off-farm opportunities dren under five, the prevalence of underweight in- are confined to a few activities, most of which do not creased from 14 percent in the better-off group to 35 provide returns as high as own-farm production. percent in the worst-off group. The prevalence of There is a strong association between the agricul- stunting increased from 33 percent in the better-off tural practices and the poverty status of the house- group to 54 percent in the worst-off group. The preva- holds. The poor are less likely to plant hybrids, are lence of wasting increased from 3 percent in the bet- less likely to use fertilizer and pesticides, and have ter-off group to 13 percent in the worst-off group. Profile of Poverty for Malawi 51 Box 3.2 Socioeconomic classification of the population Cluster analysis0 was used to classify socioeconomic groups. The most important variables distinguishing a socioeconomic group were education, gender and marital status of the head of the household, landholding size. livestock ownership, and household possessions such as radio and bicycle.The analysis identified 6 different socioecomic groups of which the best and worst are presented here. Characteristics of the richest socioeconomic group: Head of the household: 73 % had gone to school, 92% were married, 84% were male, 40% had received farm training, 61% had a traditional or non-traditional position; Household possessions: 76% had a radio and 70% had a bicycle; Agricultural possessions: 63% had one or more hectares of land, 53% owned cattle, 82% owned goats or sheep, 20% owned an oxcart and 10% owned a plough; Agricultural production: 20% cultivated cash crops and 89%/. used fertilizers: House variables: 97% had a concrete floor, 99% had a roof made of material other than grass, 20% had electricity: Access to health services: 71 % were within 5 km of a health unit; Sanitation: 95% had access to a latrine and 65% had access to safe drinking water. Characteristics of the poorest socioeconomic group: Head of the household: Mostly women, 24% had gone to school, 37% were divorced, 8% had received farm training, 17% had a traditional or a non-traditional position; Household possessions: 5% had a radio and 1% had a bicycle; Agricultural possessions: Everybody had less than one acre of land, no-one possessed livestock, oxcarts or ploughs; Agricultural production: 3 % cultivated cash crops, 16% used fertilizers; House variables: 100% had a mud floor, 96% had a roof of grass and 96% had no electricity; Access to health services: 42% were within 5 km of a health unit; Sanitation: 24% had access to a latrine and 16% had access to safe drinking water. OA combination of homogeneity and cluster analysis was applied to the NSSA 1992/1993 and the MDHS 1992 data to classify the households into socioeconomic groups. The analysis is described in the paper Socioeconomic Classification of the Malawi Rural Population, Venanzio Vella, AF1HR, The World Bank, October 1995. 52 Chapter 3 Notes A measure of the sensitivity may be obtained by calculating an elasticity, represenring the percentage change in the prevalence given a 1 percent change in the poverty line. In the case of the NSSA, there is a change of 34 MK/AE /year between the 30th and 40th income percentiles. This difference represents 41 percent of the mean income for these households. A 41 percent change in the poverty line (from the 30th to the 40th percentile) thus leads to a 33 percent change in the number of rural smallholder households classified as poor - a low elasticity (.81) indicating relative insensitivity. The implied elasticities of the prevalence of poverty with respect to measured income or expenditures (see similar results for the HESSEA in Annex 2) are well below those found in many developing countries, where elasticicies of 2 are common (see Ravallion. 1992; and Ravallion, Datt, and van de Walle, 1991). 2 The analysis of the national distribution is based on the HESSEA survey and refers to household expenditure data. The analvsis that refers to rural smallholders uses NSSA data on household incomes. The two surveys have data from different years and use different designs. Consequently the levels of expenditures from the HESSEA and the levels of incomes from the NSSA are not directly comparable. However, the distribution of incomes and expenditures from che two surveys is very similar. See Annex I information on the two surveys. 3 For example, Gini coefficients for Botswana = 0.54, for South Africa = 0.55, for Tanzania =0.59. for Zambia = 0.44, and for Zimbabwe = 0.57 (Chen. Datt and Ravallion, 1994) The rural inequality evident in these data and demonstrated in Figures 3.1 and 3.2 is confirmed by the Lorenz curve in Annex 3: Figure A3. 1. The four majot cities of Blantyre, Lilongwe, Zomba, and Mzuau are designated as 'urban' according to the NSO-established convention. Results for other urban areas will be described by the term 'BOMAs', also as per NSO norms. 6 The NSSA is the primary data source used in the following rural analyses because it contains a wide range of information on the agricultural and other economic activities of smallholders. Where appropriate, the NSSA analysis is supplemented with HESSEA data. 7 The eight Agricultural Development Divisions (ADDs) in Malawi were established as a means of administering agricultural programs. h RDPs are smaller administrative units than ADDs. There are 30 RDPs in rural Malawi. Because the existing administrative structure for agricultural programs is organized through RDPs, it is useful to analyze poverty by RDP ' The map of the poverty gap index using the 20th income percentile as a cutoff (Figure A3.2) is virtually identical to Figure 3.6 which uses the 40"h percentile. n As the Food Security and Nutrition Module of the NSSA is coded, however, some of these factors can be examined more closely. Figures 3.6 and 3.7 were constructed using the 40th income percentiles for the NSSA. Maps constructed using the 20th percentile are virtually identical and are shown in Annex 3 (Figure A.3.2). 2 Casual estate workers would generally be considered to be smaliholders. They are likely to grow gardens and would thus be included in the NSSA. The only rural worker category to be systematically excluded from the NSSA is estate tenants. ' During the period when the NSSA was conducted, most fertilizer was distributed on credit through smallholder farmers clubs, although some fertilizer was available on a cash basis through other sources, notably ADMARC depots. It might be argued that many of the smallholders in the larger cultivated area categories were poor prior to the introduction of the burley program and that the program may thus have significant impact on the reduction of poverty. However, an analysis of agriculmtiral income among smallholders in the largest cultivated area category (those with more than t vo hectares under cultivation) who grow burlev tobacco shows that burlev tobacco, which occupies on average 0.632 hectares for this group. accounts for about half of their total income. If the income from burley is removed from cheir total income, the mean income of this group remains above 490 MK/AE/year, which is above the 90th percentile of income in rural Malawi. On this basis, it is not possible to conclude that the growing of burley tobacco has lifted these well-endowed households out of poverty. Profile of Poverty for 53aw s Professional work includes work in forestry and quarrying, transport and equipment manufacturing, wholesale and retail trade, transport, financial and business services, education, medical, motion pictures and repairing services (including tin-smithing and shoe repair). 6 The regression model presented in Table 3.26 was tested for the assumptions of homoskedasticity, linearity, and normality. The model exhibits no strong violation of any of these assumptions. The tests employed are detailed in Spanos (1986), and the following procedures were adopted. First, the reduced forms for the simultaneous system were estimated. For each (of two) equations, a barrage of misspecification tests was applied. Since none of the assumptionls were violated, the simultaneous model was estinmated. Tests of overidentifying restrictions were then conducted, using a Bassman test. Holding all other variables at their mean values. Ih A quadratic specification was also estimated to account for curvature in the relationship, but the quadratic terms were never significant and were dropped. 54 -1 Malawi * Profile and Priorities for Action A Strategy to Reduce Poverty in Maai155 Implementing a Strategy to Reduce Poverty in Malawi This profile confirms that the condition of humani geted interventions for the poor are another critical resources in Malawi is dismal and that poverty is wide- element of the poverty reduction strategy. spread and severe. Health and education indicators are among the worst in the world: there has been little Setting priorities or no improvemenit in the severely high child mortal- ity rates during the past 15 years; maternal and child A robust agricultural sector is a prerequisite for more malnutrition are high; and the incomiies of most rural broadly-based growth and poverty reduction. How- Malawians are not sufficient to support even mini- ever, there is a limit to the extent to which agricul- mal needs. Access to land and inputs is so limited ture alone can reduce poverty. In the long term, that most smallholders have little hope of moving out growth will have to come about from outside agricul- of poverty solely by working their farms. The gener- ture. It is unlikely that resource-poor farmers, even ally low incomes combined with marked income in- with new technologies or the ability to produce equality point to severe structural problems which are higher-value crops, will be able to escape poverty by made worse by the high population growth rate. using their own land. Small landholdings combined TIhese findings send a clear and urgent message: with high rates of dependency create conditions that povertv reduction should be the focus of all policyx make food self-sufficiency impossible. It will be nec- tormulation. The new government of Malawi made essary to improve opportunities off-farm and to in- reducing poverty its top priority and initiated the Pov- crease the returns to off-farm labor by investing in erty Alleviation Program. But the scale of the prob- rural infrastructure, among other things. lem requires broader measures and more forceful Broadly-based growth and poverty reduction also implementation than have taken place so far. The require more effective investments in people. In edu- magnitude and severity of poverty in Malawi call for cation and health, the goal should be to improve ac- broad reforms across all sectors, for example. to de- cess and quality in rural areas and to target expendi- regulate the economy further, to lower transaction tures to the more densely-populated and poorer re- costs for small producers and consumers, to re-exam- gions. To facilitate their integration with the rest of ine land policies and to reorient ptiblic spending. Tar- thle economy, safety net programs (such as public 56 | Malawi * Profile and Priorities for Action works) should be put in place to reach the poorest communication campaign and by directing more fam- who live in isolated areas without basic services. In ily planning services to the rural areas. the short term, the poor must be helped to adjust to market forces such as changes in pricing policies, Expanding access and reducing (which may adversely affect smallholders who are not inequities in the social sectors net sellers of food grains), while they develop coping strategies and until they can benefit from the growth Equitable access to health, family planning and edu- impact on the economy overall. cation could be greatly improved by implementing Policies to reduce poverty in Malawi will require three policies. First, a larger portion of government substantial efforts in every sector. Three priorities are resources could be allocated to primary services within clear and demand immediate attention: developing each sector and everything possible should be done human resources, improving rural livelihoods and cre- to increase the efficiency of resources spent in the ating safety nets for the most vulnerable. higher level of services. In education, this could be done, for example, by attracting and retaining a larger number of students at the tertiary levels. Second, pub- lic resources should be targeted to ensure that girls attend school and that they at least complete the full 4.1 Developing human primary school cycle. Third, a larger portion of re- resources sources should be spent in the rural Central and rural Southern regions which contain most of the poor and Malawi's low social indicators are a consequence of have the worst education indicators. Fourth, health inadequate rural infrastructure and low quality of ser- and family planning would be reinforced by strength- vice delivery-both of which are worse for the poor. ening rural services delivery, improving the availabil- School enrollment is lowest for the children in the ity of drugs and contraceptive supplies and placing lower income levels and they benefit least from pub- and training community health workers. lic spending on education. Enrollment rates at all lev- els are lower and drop-out rates are higher for girls. Increasing effectiveness The poorest girls are least likely to go to primary and quality of social services school and enrollments for girls are lower than for boys at every income level and in all regions. Girls The quality of primary education, already poor, has aged 13 or older are more likely to drop out of pri- declined further as enrollments have surged. Most of mary school than boys in the same age group. There the recent increase in public primary spending was are also inequiries in access by region: children living allocated to finance the salaries of newly-recruited in the Southern and in the Cencral regions are less teachers. In the future it will be desirable to establish likely to attend school. Progress is hampered by the a better balance between wage and non-wage budget- high population growth rate which places a heavy ary allocations, to increase the availability of class- burden on households and on institutions. room supplies and teaching materials, and to train The Government has shown its commitment to teachers. improving human resources by eliminating primary For the health sector, improving the quality of school fees and by sharply increasing primary educa- basic services is as critical as increasing coverage. The tion spending. However, the severity of the situation significantly low health indicators in Malawi are requires even stronger efforts to sustain and extend strongly associated with lack of access to primary recent accomplishments in education and to extend health care services. Primary health care is crucial for comparable efforts to health and nutrition. The preventing transmittable diseases, for providing ma- Government's population policy should be strength- ternal, pre- and post-natal, infant and childhood care, ened by launching an information, education and and for avoiding high cost hospitalization that will A Strategy to Reduce Poverty in Molawi r 57 be needed later if prevenltive care is lackinig. IThe health Ongoing reforms are necessary sector in Malawi nieeds to reallocatc public spending but insufficient for the poorest to rural primary--level care facilities, to gaini the great- est returins for society as a whole. Several ongoing reform programs address the issues Most people in Malawi obtain their water fi-omii of increasing smallholder incomes from agriculture: unisafe sources: one third have no sanitation facility. extendiig smallholder burley tobacco production, lib- lnvestmenits in rural sanitation and access to clean eralizing maize (and other) markers and promoting water are crucial for the national health. SLuch invest- agricultural diversification. These all rely on increased ments will be- most eftective if they support commu- market access and improved incentives to improve nitv- driven initiatives that wvill elIsure more appro- incomes. priare choices of technology, and beniefit fronm coIml- First, the smaliholder burley tobacco program is munity involveiietit througIlout the desigin, organi- intended to improve the incomes of smaliholders. As zation and maintenlanice of the improved of 1992/1993, smallholders had started to participate infrastruet ltir-e in the program but most had not been able to do so. (Incomes in Kasungu, the ADD whiere most smaliholder burley tobacco is produced, are signifi- cantly higher than in the rest of the counltry, even bthougil few land-poor sinaJlholders in Kasungu pro- 4.2 Improving rural duce tobacco). Since burley tobacco is probably the livelihoods highest-value crop in Malawi, the program ought to be expanded and all quotas on smallholder burley policies including over-regularion ofthe econory'iand tobacco should be removed. Obstacles that exclude biases that favored the estate sector have contributed smallholders from the burlev' marketing system should to the overwhelming poverty and inecqualitv in be lifted immediately. However, the poorest Malawvi. Ongoin1g rcform-1s such as libcralizaition ofag- smallholders are unlikely to plant burley tobacco be- riCUItUral prodUCer priCeS, thc sm;allholdcr burley pro- cause of persisting constraints such as lack of cash to gramii. anid the liberalization of agriCultoUral input purchiase inputs and limnited ability to manage pro- markers have initiated structural changes intenided to duction. proniote lIong-rUn broad-based growth antd beniefit Second. recent policies have moved towards lib- poot Mklalawians. cralizing maize prices: this is likely to translate into a Current efforts to liberalize the ecnonloy do not real increase in the relative price of maize. Higher go fat enough in remilovinig constraints faced by' producer prices should encourage maize production smallholder-s aid in establishing a level playing field and raise the incomes of households able to produce for all N'lalawvians. Fur[lthe deregulaltioll of rCstriction1s net surpluses of maize, but higher prices will lower on crop p roductioni and marketinig, and of agricul- the incomes of most smallholders who are net putr- tural input markets, higher investment in rural infra- chasers of maize. Given the land and asset conistraints structure, deregulation of transport services, and re- docume-nted in this profile, it is doubtful that higher form of land policies are essetiaLl for significanit pov- maize production will be able to eliminate maize defi- ertv reduction. Some interventionis will be needed so cits for the poorest producers who will suffer in the that the poor can respond to the opportunities emerg- short run until they develop new coping and off-farm ing from economilc liberalizationi. For the morc iso- strategies. This calls for complementary interventions lated and vulnlerable poor, transfer programs may' be to facilitate adoption of hybrid varieties, diversify own- the only' was to improve living conditions, farm crops, and generate off-farnm income, for example, 58 1 Malawi * Profile and Priorities for Action lV sLipportinLg puIblic work procran11is. marketing anitd input supplies. It is likely that the private I hird, aclicUirur1-al divet-sification has been promoted seCtol Will fill many of these gaps. but many' of the poor as a potc [tial Olt ofi i rducin g potverty. So i;ll-, few studl- may' remain uLiserved. Simplitfying licensing and regis- Ies pOVidCevidnIIhC-Ce a gd theI tpC s ofcrops display- tration procedures foI tradcrs should ease smaliholder log sulficiCiII proh tabiiNi, bLti Opportlunities for expan- access to inpuIts and lower thlir costs. However, high oni into suniOxvc -rr s tnrou ndnn, soybvcais, zadhigher- t ransaction costs coild, for exam-le, preveit private trad- valul-ed vc Ltabics SlthOUld be cxplored. I'h, ar-iCulturnal [r tromi servi ng snall-scale prodLCeS With small amoults reSeachI-Ch andcI tXn.s it iol ssStci lSihouild PFoliotcrol-op di- of maiirkcted StirplLs. v f lrsitatioll a:lid Cxiln\ioll servics shoild be reformicd InvCsCIlCIILts in infrastuLIctUre to improve access to to delivL- aIppropli Iic lessa(ges ot sila Il-scaler prodLC- loCal markets will lhelp nial.ke rile econiomily more efficient rrs. Iveii so. hLe Lxi ctelyll sun 1 I ttldholctidings oft'te pooi by iii ini milzig market [t'ai tires and will enable sniall-scale will 1lecssari ik liii oxiv -1ani.1 iletotnl-ealinilg potcli- j)iSt 1rnc-rs too ad jOust more siiiootlitly to changing condi- tial. Accorditigi; rese;arch ald exttision mnight be ori- tiOIls. The sucCess of liberalization dcpends critically on ented to i; od rtrop pr-odulcti' i thai would allowv ho use- propcrly fuitt ioniig fac tor aiid prlodtuct miarkets. Witlh- hol(d nicnhbcrs t wo iork ol't-lt rn Wit ni iot CatiLsi ne;a- toit in1vCstIIii Its in i fraIstFLueturr, rent-seekinog behavior yev minipactlson rtilc otti i ri in p tl tietioii aid iLefficiency will persist. Strong i lstitUtiolal or regu- \WhIi IcLsstai i(ed iiipiCIlleiltlaioll ofthiecse aenictiCltur-al latory bilascs have created distortions and market rents. policy ircttlriis iLs essential to itric ease iei lmtimiOIts of nianv '[I'he highly-centralized tr anstport svstemin has effectively of tie h nial puot trri imIpact i ofsuch Prograiils on the .Ii iiimiatedcsitu Il-scale hauli11g activities, small-scalepro- in.let iilcs u lihe putt ti est su litidellrs will otr he very' large. cssi,ig, and dcccci tralized our small-scale storage. 0'~~~~_ lheC p)OFrcs ;IIL so (.01SciILM11Cd b\ Sna1 land lholdings ( 01n1iLIIitCics, 'ild'VIIilals anid tilnIIS, w'tll SLupport anld lack oloutlilrasscts that Htl-ti inCt0ini iS nCeCSSarI' froni thc puLblic sector, should invest in basic marketing t(t Ilit t Lieiill mOiii uOVc'Ir V. his inudlrscorcs thelic inited iifraStrLucture stiCezias marketplaces and basic marketing scope that agriture alomir iMis to reduce! poverty in lu- services. Facilities lor weigiing, graditig and processing ral al.u1i. Since imICICIased pr tductiOIl of food aIid cash are also essential. Processiig faCilities such as haimermills croips by sil;lihlloldil-S ; alnlimtb h 'Ieiv\cd as the oil' iearalls dirteelY breiiet feniales, one of thie imiost vulinerable to red tiC rillil uixmcts. it iS ailso ticc eSSIa r tO support glrups of tihe poorI bv removing a minajoi tinie coInstraint. i \'esCtll its lt I fo Isi ol f't I fll II I t-o\ t. Simall-scale processing wvill allow rural areas ro retain milore valuie-added and wvill create forward- and backward- Economic reforms should linked ilnltiplitr effects. 'ilhis infrastructure could be be broadened to reach provided throughI a conipetitive biddiiig process for con- more smaliholders StlUCttol or chroctgI loiaiis for constriactioli to private atttil ts. \V1 ICl i , i.-1i progress lih s eet t11 ,,1.ia it) liberal izinig thle M'ali,viai ccomlt o-v. sign i i,cant obstacles reniaiili. Tramis- Examining land policies port. especi.il 'I in rural aleas, is stLiil 1hi>'l regulated. anid proli bI i IlSo) t i tratle ill rl LI areas COn ti ll te tOt ha1m- [he poverty' prot)lle revealecd a strong relatioriship beween per proi rcss. AS a result, i t',alltsitioll Cosrs 1tOt siiiaii-scal [lie size oftland under cultivation and income. Poor ru- pio'ld tieris ild consmliers .tr \c\ higli. Remnovitig pl'u- ral hItuseholds lend tO CLultiVateC smalll ar eas of customiiary tectititS ruO the dotille.stic irIIISOlptt i idtlsrr by ci1lli tt- land. 'T'h total land availlble for snialiholders has been ing~ nifltilthilti tariff's shoUldl tLiiitt Ic agrIeCi I tu-cal di- shrinikitig hecause of ruc convetsioni of customary land vesiicit ii- IiI ilorCase p it Ill trgi iS for rr_1ral prod Iucels to leaschold Lind and becaLIse utOpoptilatiol growth. In tild illtiCJcSC the pi)ti chi;si i` pi>VUl oe L OOr COInSliILerS. addition, tliere is itiucli Cvideclc of dectlitig soil ftrtil- Slb1t,i ice li[c,i VC li/t tilt 5\i!! V 'liitiliillt tliail 'tt1iC its ts s1iiilholders have resoitted 10 LitIstistailiiable soil sCrV me:s iii iw prvXILiCt. 1), It N,lellik seoe-. Stich 1i at;tS 11ae illll ig pracrJies. Mlannv siiallholders' landholdilgs can- A Strategy to Reduce Poverty in_MaoaAvxjw | 59 not support their minimal household food requirements large labor forcc of tenants and perilmantit wvorkers on or generate cash income, and the costs (in terms of time esLatcs, these opportunities may' hc limited. and cash outlays) of obtaining forest produCts for fuel. (Growth in off-Carnm agricultural and non-agticultti.-al agricultural production, and housing. havc Increased. inc1OleL is IS Ctssarv for iniprllving smnallholder welfare. In- On the other hand, thtiec is evidence that Soime es- tcgration of thc simalilolder ecoLno0m into niarkets where cate lanid is underutilized hbCCIUS of paSt policy biases pricesare comiipetitivelx determined will help) to r-eduLCe the that favored estates and granted privilc,es based on land- suibsistence orientation of mIucIh sm;allholcler farmiling and holding size. AboLit 20 percenit of Malawis' land area is provide incentives Ioi- prodliction of nore high-valuecrops. public land. As a result of the Cxtretme prcssLrtrs on lanid Investmeints in feeder toads, storao e facilitics and mlarkets, in Malawi, the distribution and util ,iontof the land cotipltd with furticr dcregulalltioll or markets, be cs- resource base has become a major issuc. 'The i ncrcasin senrial f5 rIovxth in the noI-ari i-secto. ilncideilnce of e ncroachimients by smnal holdcrs oni estate In the long"cer rArM, rhc reretcsr souLrcc of off;-farm in- anid lpublic lands is evidence of this problem. CoeICs will inCvitLably ComIe froil Otutside the a;gricultural A comprehensive review of land policy is Uirgenl-` sectror. As ilcomle,s ilncrc:ise, the demand 01-it a widel ralw needed. Forced lanid redistribution is neithetr politically ofgcoodsandservicesshortldcincreasicor-respond gly. lb is nor economically an optioIn. Policies and laws are ncede(d demand will stinmliaie a arlrctv of sniall and tncdituni-scalc to protcct property rights, encOIIIrag iiotIc efficielnt risc rur-al cintcrprises to supplk these goods and srxti es of land and support an aIctive land market that allo0ws the most productivc Farmers to bc cngaged in agricul- tural prodttction. 'To facilitate the land policy review atid set the basis for furrire reforimis it is important to stop the conversion 4.3 Safety net interventions of custolizarv lanld to estates wvitioti r tei agreemnicit or for the poorest all customilary holders. Existing lanld titles shouild he ex- amined. To facilitate a more activ. Iand market, rents Poverty is so pervasive in N,Ialawi that 1 gly-i ernl emlls- need to he raised on estate land and their collection) shottld Cd i'CS ShOrt d be toupled 'vi ilh short-term i t0111Cni t anls- be entfor-ced. Iand studies will lielp identify any fetr mnechanisims to alleviate the tiiost urgent problems underutilized agricultural land. The land policy review of the poor. and reform woutld need to inclrtdIe blroader natural re- soUrcej issues s acsi dS deforestation and soil t'-osiOI. Supporting transfer Malawi is a land-constrained cotIntrv witch an ex- programs for the poorest tremelv high rate of poptilatiotn growtLh, and railn-fed agriculture withla siigle croppilng sason. In light ofthese A n erffective anld wxell-rtargeted saiferv iie rthat iiitii- facts it is clear that sustainable pox\ertN reductioll will galtCs seastol aI and c(i Lronilc povxrtv, xbil le at the satile depend. to a large degree. oni rural Malawi.ans finding tilme cotitribuititig to lol g-tC enh goals. is esstiarill to a inconle-generating activities outside agricrltut-e. sustainiable poveCrtV redutieoinl straitegv. A labor-i teCi - siclh ic xvoiks program is .1 ll attrac-rtive sah; le nt Increasing off-farm incomes optioil. if it is tar,geted It priority ireas and poor hostisholds. lltIle slort- trtlil stLICI a prliopltr xxill At presetit off-farill opporttitities ate cotliledc to a sriall il Irt Ile itt tri tii)tlal leeC(1s of thl e Ooo a tull( i provide set ofactivities. ltlensificariol and diversification ofag- thtelil vi rltil ICch neeided resoSt reCs t(o p. t f!o cfritical ricttilral prodtictiotI by smallholders withi largcr hoId- tri.ll i il puis. ings and bv estates might provide llxx cWmplotlilelit op- Fxper-ietlres I--oill S,if`CtV DOlt plrg-ra1ills, particullarly porFtLIlities ai cause CJUoS e o InHpxxaid pr1-CsUrc ot xIxaes. tli(lse st.ted ifter the dr ough its, iced to bC poo(led However. with irapid poLpIlatioll gioxx'Wl and i relatiely aiim1 assimiilated. A coot-dintiated ;approitc-h is essenitial 60 Malawi * Profile and Priorities for Action for better design of safety net operations, and for Improving poverty avoiding duplication. Given the capacity constraints monitoring for better targeting on the Government, it is necessary that NGOs and the broader civil society be involved in formulationi Regular and reliable imonitoring svstreins are required for and implementation. Decentralized administration more efficient targetilg. The recent initiative bv the would help to take decisionrmaking structures closer Government to coordinate the poverty monitoring and to the people. This would make for faster responses analyziig activities and to identify kcy povertv indica- to project, program and community needs. tors that can be monitored regilarly will provide intfor- The patterns foulid in this profile can be used to martioni abouit the effect of the programs on1 lie mnost target such programs. Lilongwe ADD near the vulnerable groups. The ample data available in Malawi Mozambique border and Rumphi/N. Mizimba in On poverty, edcLcation, food insecurity and health out- Mzuzu ADD are two examples of high-poverty areas comiles can provide intforimoationi crucial to the formula- with potential for long-run growth. These areas, which tion of effective programs. Giving suLpport now to the are reasonably close to urban centers or other rural research community in Malawi would help to iden- areas with low poverty, could benefit from transport tify the most vuliierable groups. speciFy their needs, infrastructure created through labor-intensive public establish how they can best be assisted and guide over- works. Such infrastructure would increase market in- all policy dlirectioli. tegration and allow the poor to participate in the emerging opportunities. Other areas with severe pov- erty but lower economic potential, such as in the Shire Valley, might be targeted for pure transfer programs, comiibined with efforts to facilitate labor mobilitv. 4.4 Agenda for further study Distributing free inputs This profile has not covered issues ielated to urbani pov- ertv or to estate tenanlts and permanienit workers. More In an attempt to improve household nutrition and re- study is needed to gain deeper understanding of such duce vulnerabilitv among smallholders, free input dis- issues aiid to plopose solutioLIs. tribution programs have recently been implemented. WX"e lack such systematic data on estate tenanits and SSmalllolders in areas worst affected by droughts in 1993/ permanenit estate workers and their dependanits whio 1994 and 1994/1995 have received an input package of constitute about 1 0 percent of Malawians.T here is some enough hybrid seed and fertilizer to plant about 0.2 hect- evidence that tenanits and workers on larger more estab- are of land. Over 800,000 smallholder households were lished estates are less poor thani their couliter parts found targeted for this in 1994 and about 600,000 in 1995. on smaller newly-establishied estates. Thle latter tend to However, this program might not be the mosc ecofiomi- have lower incorTies and less access to water and sanica- callv efficient use of scarce funds and might lead to an tion facilities and social services. Iulrther study is needed undesirable dependency on free inputs. Self-targeting to explore how liberalization policies will provide incen- public works programs that provide cash or inputs or tives to estates to intensifl and diversify their econiomic food for work are a preferred option because these gen- activities. TIhe result miay be a higher demand for low- erate short-terimi income benefits that smallholders can skill low-wage labor. On the othel hand, there is some rise to finance inputs (or food purchases), and long-term hope that estate owners might lease or sell some land to benefits such as improved physical infrastructure with- tenants and permanent workers, who could theni purstie out perpetuating dependency on pure welfare transfers. labor-intenisive crop activities. The preferred long-term strategy is to phase-out firee Unemployment and inadcquate housinig are urgent input distributioi programs. However, in the short term, problems for the urbani poor. The impact of the well-coniceived programs that include extensioni support dlownsizing of che formal seccor on urban eniploviient, for production will continLue to be appropriate. deregulation of the in inormal sectoi, and detcrnemili- A Strategy to Reduce Poverty in Malawi l6 tion of minimum wage levels are relevant research high drop-out rates at the primary level? Analysis also topics. The urban poor are likely to be particularly suggests that there is much to be explored on the rela- vulnerable to changes in food prices. Earlier studies tionship between household headship and poverty. We on urban poverty suggest how to target the most vul- know that female-headed households tend to be poorer nerable urban poor with interventions such as food than male-headed households, but there are variations subsidies and soft credits. within the group of female headship still unexplored. This report generates unanswered questions. How How are cash remittances important at different stages can we explain the regional variation in child and mater- in the family life-cycle? What are the regional differences nal health? Why does the Central region perform worse in the influence of cash remittances? on child and maternal health than the Southern and Not all subjects for further research can be listed here. Northern regions? Regional differences in the school It is hoped that the research community in Malawi will enrollment rate remain unanswered. Do cultural or other have the resources to follow-up on some of these ques- sociological factors explain these differences? Are there tions. The newly-established Poverty Monitoring and variations in school enrollment across districts within Analysis System could be instrumental to such further regions? What are the critical factors associated with girls exploration and subsequent analysis. 62 Malawi * Profile and Priorities for Action Annex 1 63 Methodological Notes Two primary data sources have been used in this report, 287 for seconidarv education and MK 3,816 for tertiary The 1990-1991 Household Expendirtire and Small-Scale educarion is allocated to all pupils and studenits ernlolled Economic Activities (HESSEA) suivey, and the 1992/ (MIOE, 1993). 1993 National Sample Survey of Agriculture (NSSA). Per capita primary educatioin costs for 1994/1995 Both SuIVeyS were conduicted by thc Central Statiscical cann1or be calculated in as straightfOrward a maniner. In Office of Malawi. The data were made available on 1'994/19995 no houselhold survey data were ivailable, and diskettes. informatlioni on primary-level enr-olliments by socioeco- The HESSEA data provide inforimiationi On house- nolimic group had to be derived on the basis of eniroll- hold expenditures and consum11ption on a nationxwide nients by regions provided by the IMOE ( 1 995) raking sample: the NSSA data ae-C ttsd tO comIlpute 11ouselhold HESSA data as the baseline. The same per stideLt cost, incomiles for rural smallliolders only. 'he procedure to MK 220 for primar-y education, is allocated to all eni- comIpute the expetiditute and inlcoImec data is thc focus rolled (MIE, 1995). of this note. Both data sets also containi inf'orimiationi On The algorithim (Box Al 1 ) describes the procedLurc social indicators, to obtain primarv-lcvel enrolliments by sociocconiomilic groups, or q uintiles, in 1994/1995. Computation of education spending Per- capita seconidarv and tertiarv educLationi costs for using the 1990-1991 HESSEA 199()4/1995 USe projected school enrollmilenlts across so- cioecconoiilc groups. Enrol[limenits at the secondary and 'This annex describes th C use of Mlvalawian data co pro- tertiayv levels acrcoss socioeconfom(1ic grOUpS CeeliCe.- vide estimates of per capita costs across five socioeco- lated on the basis of the overiall change in enrollments nomic group.s by educationlal leVel foC 1990/1991 and between 19(0/1991 and 1 994/1995.usilng MOE statis- 1994/19'5. This analysis divxides the population into five tics. HESSEA provided the baseline information On socioeconolimic groups, constructed bv ranking each in- school enrollment. 'I'he share of secondary cnirlolinelltS dividual fiom the poorest to the richest on the basis of' to the total population incrcascd from .0.78 to 0.88 per- aditilt equivalent expendiurLe. Then the populJtion is cent and the share of terriyrv enrollments to the total divided into fivc groups, (guiintiles) each containing tx- populationi inlcreased from 0.05 to 0.08 prCicilt between actlv 20 perceit of all individuals. The methodology used 199')(/1')91 and 1 994/1 995. The same distribution of is knowin as Bencfit Incidence Analysis and it brings to- cilrollmienits across q uiniiles at the secondary and ter- gether iniformationi on school cnrollmencts and infiaorimia- tiarV levcls obse rved in 1 990/ 1991 is mailnained in 1 ')4i tion on per stUrtlr ptiblic costs. 1 995. I'Ihe saimtie per studeCnt cost by educational level, Per capita education costs in 1990/199)1 are esti- MK 909 for secondarv edLucationi aind MNK 1 5.52.3 For mated on the basis of enrollments by educational level tertiarv' education, is allocated to all enrol lcd reported in the 1990/19))1 HESSEA suirvey. The samc (MOE, 1 )95) perl- stident costr MK 40 for primary edtication, MK 64 Malawi * Profile and PrioritiesforAction Construction of household living. For the HESSEA, the NSO collected prices for a expenditures from large range of goods for the survey year, from each of the the 1990-1991 HESSEA four major cities and six markets in each of the three rural regions (Central, Southern and Northern). For the The HESSEA collected information on both cash and purpose of this profile, the NSO provided prices for food own-account consumption. The methodology for record- items and housing items for the four cities and each of ing the two components of expenditures was different, the rural regions. however. Cash expenditures were recorded on a daily The NSO also provided expenditure weights for diary basis by all household members 10 years or older these items as derived from the expenditures recorded for 28 consecutive days (HESSEA 4: Diary of Daily Ex- by households. The shares reported for rural areas are penditures). Own-account production was derived from based on cash as well as own-source consumption. The a single recall for a monthly estimate (HESSEA shares for urban areas are based only on cash expendi- Section K). tures reported by households; urban shares provided were The sample design involved stratifying the country also for high-income households only (spending >MK into 10 strata, four of which were the cities of Blantyre, 1000/month). To remove the bias introduced in the ur- Lilongwe, Mzuzu and the municipality of Zomba. The ban data by the shares reflecting wealthier household six other strata were urban and rural areas. consumption patterns, the individual item shares were adjusted by applying aggregate shares for food items and Price indices and housing costs reported by the NSO (in the relevant CPIs) cost-of-living deflation for each city, for low-income households (spending 2 Ha 8 (86) 3 (14) 100 3 .518 .128 ___ Total 100 100 4 .638 .161 _ _ 5 .684 .178 Source: NSSA 1992/1993 6 .790 .212 7 .943 .250 8 1.060 .284 9 1.161 .343 10 1.609 .516 Source: NSSA 1992/1993 76 Malawi * Profile and Priorities for Action Table A3.10 Poverty indices, in percents by ADD, different cutoffs Prevalence (%) Poverty Gap (a= 1) Severity (a=2) % National Poverty Gap Below Below Below Below Below Below Below Below the 40th the 20th the 40th the 20th the 40th the 20th the 40th the 20th ADD percentile percentile percentile percentile percentile percentile percentile percentile Karonga 12.0 29.6 4.5 13.4 2.6 8.0 1.5 2.2 Mzuzu 20.7 40.3 9.8 21.1 6.6 14 2 6.6 6.8 Kasungu 2.5 8.5 1.3 3.6 1.0 2.3 1.5 1.9 Salima 14.4 32.6 7.9 16.5 5.6 11.2 4.6 4.7 Lilongwe 21.3 43.0 10.9 22.5 8.3 15.6 21.3 21.1 Machinga 23.2 43.7 12.3 23.7 9.4 16.8 25.8 24.1 Blantyre 26.3 51.3 18.6 26.3 8.9 18.6 31.8 32.1 Shire Valley 22.5 41.7 10.7 22.6 6.6 15.2 6.9 7.1 Notes: NSSA Household income dota. Source: NSSA 1992/1993 Table A 3.11 Average livestock owned by income decile Livestock Income decile Type 1 2 3 4 5 6 7 8 9 10 Poultry 4.8 5.2 4.5 5.2 4.9 5.4 7.0 7.8 7.2 9.7 Cattle .5 .5 .6 .5 .6 .5 .7 .6 .8 1.7 Sheep .1 .1 .1 .1 .1 .1 0.0 .1 .1 .2 Goats 1.0 1.0 1.1 1.2 1.0 1.0 1.2 1.4 1.6 1.8 Pigs .2 .2 .2 .2 .3 .3 .3 .3 .3 .4 Source: NSSA 1992/1993 Annex 3 1 77 Figure A3.1 Cumulative distribution of smaliholder income 1.0 / 0.9 0.8 8) E 0.7 - 0 0.6 0 1 a) 0.45 0) 0.4 0 1 010 0 1 2 3 4 5 6 7 8 9 1 0 Percentage of Households X-axis is on log scale Source: NSSA 1992/1993 Table A3.12 Average livestock ownership by poverty cutoff and ADD Livestock Type and Income Cutoff Karonga Mzuzu Kasungu Salima Lilongwe Machinga Blantyre Shire Valley Poultry <40 PCT 8.2 11.2 7.2 3.8 5.0 4.8 3.0 5.4 >40 PCT 11.5 13.0 8.9 7.7 6.3 5.8 4.2 6.8 Cattle <40PCT 4.7 1.6 1.6 .4 .5 .2 .2 1.2 >40 PCT 2.5 1.3 1.6 .6 .8 .3 .2 .9 Sheep <40 PCT .3 .4 0 0 0.0 .1 0.0 .1 >40 PCT .1 .3 .1 .1 0.0 1 0.0 .0 Goats <40 PCT 1.3 1.2 1.4 1.0 1.3 .8 .8 1.8 >40 PCT .6 1.0 1.9 1.6 1.9 .8 .8 1.8 Pigs <40 PCT 1.7 .4 .4 .3 .4 0.0 .1 .4 >40 PCT .3 .8 .4 .2 .5 .0 1 .7 Source: NSSA 1992/1993 78 Malawi * Profile and Priorities for Action Figure A3.2 Population classified "poorest" below 20th Income percentile by Rural Development Project Percent N.A1 A/1P1 4i MBA T A~~~~~ ~~~~~~~~~~~~S{R HH K NX KAWINCA ! g::==== ^ l | D>-A_Otj~~~~~~~~~PHA' eE Annex 3 79 Figure A3.3 Area planted per adult equivalent in household by Rural Development Project !L: YI F S } t {,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i ' ' .jIF ;tiqex 80 Malawi Profile and Priorities for Action Figure A3.4 Hybrid maize yields by Rural Development Project ,;~~~~~~~~~~~~~~~~~~~~~~~. . . . .. . . KG,/HA A~~~~~~KW4 .. ....... .. . i.9 .g~~~~ ~ --- ------ 70 CO C Q)~~~~~~I IL It,~~~~~~~~~~~~~~~~~~~~, Li n 1 Nt H 4l c 0 a) o C o .. . : :- . .... . -..... - . :::7:v--77 ''-'-,=--' E.w~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i *! dU4'_ - E IL CNJ cO co) C CO x O .s --U . S U . , E ° i L_ *i .Z a' . 0 U) Z5 2 t~~~~~~~~~W co U) CO x~~~~ (D~~~~~I 16 TO~ ~ ~ _ ___-__ _______ .1 Ia. c 0 0 'O Mo G) p 0~~~~~~~~~r -0 o 3 t o~~~. 'U BBr-l O~ si \t \ oo~~~~~. References 87 References Alwang, J. and P.B. Siegel (1994). "Rural Poverty Assessmcnt, Vol. II1 of Zambia Poverty Assessment". Report No. 12985-ZA, Human Resources Division, Southeril Aftrica Department. Washington, DC: The World Bank. Becker, H. (1990). "Labor Input Decisions of Subsistence Farns Households in Southern Malawi." Journal ofAgricultural Economics. 41: 162-171. Bernier, R., S. Chao and L. Demery (I 994). Ghana: The Incidence of Social Spending. 1988-91. Background paper to: World Bank. Poverty Assessment. Blackie, M.J. (1994). Random Thoughts on a Strategy in Support of Smallholder Agriculture in Malawi, mimeo. Lilongwe, Malawi: Rockefeller Foundation. Bose, S.R. and 1. Livingstone (1993). The Labour Market ,ind WVages Policy in Malanl'i, report prepared for the Government of Malawi, Lilongwe, Malawi. Bulla, G.M. (1989) Input Demand and Supply Responst Behavior of'allauian Smallholder Farmers. Unpublished M.S. Thesis. University of Illinois, Urbana, {L.. Carr, S. (1994). Labor Datafor Selected Crops. Handwritten notes. Zomba, Malawi. Castaneda, T. (1992). Combating Poverty: Innovativ'e Social Reforlms in Chile during the 1980s. California: ICS Press. Castro-Leal, F. (1995a). Malaui: Public Education Spending and the Poor. Background paper to: World Bank. Poverty Assessment. . (1995b). South Africa: The Incidence of Public Expenditures in Health and Education. Background paper to: World Bank Poverty Assessmenit. . (1994). Lao PD.R.: Social Services and the Poor. Background paper to: World Bank. Poverty Assessment. Castro-Leal, F. and J. Dayton. (1994). Mexico: Public Education Spending and the Poor. Background paper to: World Bank. Poverty Assessimient. Centre for Social Research (1994). Poverty in Mlalawi: A Reviewu of Literature. (presented by L.A.H Msukwa, S.E. Thomas-Konyani, and N.S. Bamusi) Mimeo, Zomba, Malawi: Centre for Social Research. Chen, S., G. Datt and M. Ravallion (1994) "Is Poverty lncreasing in the Developing World?" in Re'ieu of Income and Wealth. Series 40, No. 4:359-376. Conroy, A.C. (1993) The Economics of'Smallholder Maize Production in Malawi with Reference to the Marketfor Seed and Fertilizer. Unpublished Ph.D. Dissertationi. University of Manchester, UK. Conroy, A.C. (1994). Factors Affecting Demandfor Hybrid Seed and Fertilizer by SmalIholder Households in Malawi and Prospects for Re-establishing the Farlier Base of lnput Use, prepared for Rockefeller Foundation. Lilongwe, Malawi. Dayton, J. (1995a). Madagascar: Improving the Humnan Resources of the Poor. Background paper to: World Bank. Poverty Assessment. _-. (1995b). Guinea: The Incidence of Public Expenditures in Health and Education. Background paper to: World Bank. Public Expenditure Review. Dayton, J. and L. Demery (1994). Kenya: Public Health Expenditure and the Rural loor. Background paper to: World Bank. Poverty Assessnmenit. Demery, L. and M. Verghis (1994). Kenyla: The Incidecnce of Education Expcndituri'. Background paper to: World Bank. Poverty Assessmlienrt. 88 |1 Malawi * Profile and Priorities for Action Donovan. \X'.G. ( 19)94). .alauzi: Economnic Reform and Agricultural Strategy. Technical Dcpartmncit, Af'rica Region, ATES Working Paper No. 10, Washington, DC: The WVorld Bank. Due, l.N. (1 988). "Intoa-Household Gendcer Issues in Farming Systems in Tarnzania, Zambia, and Malawi." in Gender Issuecs in Farnming .Sstems Research and Extension. edited Ibv S.V. Poalrs, M. Schmink, an(d A. Spring. Boulder, CO.: Westview Press. Gladdwin, C. 1-l. (1 9921). "G(endered Impacts of Fertilizer Subsidy Removal Programs in Malaw i anld Carneroo n.' Agricultural Economics. 714 1-I53. Kennedc, F.. and P Peters (1992 ). "Household Food Securiry and Child Nutrition: The Interaction of Incomic and Gender of Hlouschold Head." World Development. 20: I 077-1 083. Kostuer, M,.. E.E Ngalande Banda, et al. ( 1'994). Beyond Hunger, Ignorance. and Disease: i PoRterrtr Impact Asniesomentfor A2Ldai, 1i men. leach, iM. and N. Marsland (1'994). ADDF0OD Beneficiary Assessment Stiudy 1994 Llsing Patrtipi/tnto y) Rural Appraisal AMetbods. Ilanninlg )Division, Ministry of Agriculture alld Livestock Development, Lilongwc, Malawi. Iere, 17. arid S.C(. KabLn ( 199'4). An Analysis of the Indicators of'Poverty in Malawi, Ag)ricu/tual Pl;o lnni,s Ana/sis and Runra Dev!lopmnent, University Departments of Agricultural Economics in Southern Africa. Vol. 4 No. 2. Government of Mialawi ( 1 993). Situation Analysis of Poverty in Alalaivi. Lilongwe, Malawi: (Governmienrt of NMalawi and United Nations. Government of M1alawi 1 9 993). C,'uide to Agrit ultural Production in AMalawvi 1993-1994. Agricultural Communication Braniclh. Ililongwe, Mfalawi: Mfinistry of Agriculture andi Livestock [)cvelopment. Government of iMalawi (I 994a). Annual Revieuw of 'Producer Prices for Smaliholder Agii-si itural (Commnodities and Inpuets and Price Proposals for the 1994/1995 C'ropping Sc.t,sn. Pl.anninlg L)ivision, Lilongwe, MNalawi: Ministry of Agriculture and livestock Development. Governorent of Nlala\vi (1 994b). Sinallholder (;rois 'iafrgins for Selected Crops. Planning Division, Lilongw e, Malawi: Ministry of'Agriculture and Livestock Development. Governrment of Malawi (I 994c). The Agricultural and Livestock Development Strategy Action lain, tn mimco. Lilongwe. M\alawi: Ministry- of Agriculture and Livestock Devclopmern rt. Mtawali, K.M. (1993). "Malawi," (:'h. in A. Valdes and K. Muir-Leresche, eds, Agricultural Policy Refbrms and Regional M'arket Integration in Afalaawi Zambia. and Zimbabv ue. WY\lashingrton, D.C.: I FPRI Nothale, D.W. ( 19801). Labor t.se in Stoal/holder Agriculture in Mfalawi: A Critical Analsyis of'Labor-!' e l)ata from Tuelve Surrey Areas. Unpubtished Ph.D. D)issertarion. Univcrsity of Wales, UK. P'eters, P. E. ( 199 2). Monitoring the F/fr,cts of 'Grain MJarket Liberalization on the Income, 'oocd Secarittr and Nutrition of RuRi-al Houeseholds in Zomba South, Alalawi. Final Report to US AID/Malawi. Cambridge. MA.: Hlarvard Institute for International I )evelopincrn. Peters, PE . ( 1 993). Alaize and Rerlet in the Income and Food Security Strategies of ,Smallholder Families in the, Southern Region of Malau.si, Cambridge. MA.: Harvard I nsti tu te for International Development. Pet crs. 1P F. ( 1 994). Household Food Security and Drought in the Southern Region Situation Report, 1994'. Camnbridge, MA: Harvard Institute for International Development. References 1 89 Peters, P.E. and M.G. Herrera (1989). Cash Cropping, Food Security and Nutrition: The Effects of Agricultural Commercialization Among Smaliholders in Malawi. USAID/ HIID Study. Cambridge. MA: Harvard Institute for International Development. Ravallion., M. (1992). Poverty Comparisons: A G'uide to Concepts and Methods. Washington, DC: The World Bank. Living Standards Measurement Survey Working Paper. Roe, C. (1990). The Plight of the I Urban Poor in Malauwi: Results From a Baseline Study. Zomba, Malawi: Centre for Social Research. Roe. G. and W. Chilowa (1989). A Report on the First Stage ofa Study on the Effects of Macro-Economic Adjustment Policies on Urban Poor Households. Zoomba: Centre for Social Research. Sahn. D.E., J. Arulpragasam, and L. Merid (1990). Policy Reform and Poverty in Malawi. A Survey of a Decade of Experience. Ithaca, NY: CFNMP Policy Paper 7. SDA Statistical Analysis Advisor (1993). The National Sample Survey of Agriculture (NSSA): A Field Report. Monitoring and Evaluation Section, Planning Division MOA. Simler, K. (19 94). Agricultural Market Liberalization and Technological Innovation in Alalawi: Modeling Responses and Outcomes in the Smallholder Subsector. , Malawi Agricultural Sector Mlemorandum Working Paper No. 12, Washington, DC: The World Bank. Simler, K. (I 994a). Agricultural Policy and Technology Options in Malawi: Modeling Responses an2d Outcomes in the SmalIholder Subsector. Working Paper 49, Cornell Food and Nutrition Policy Program. Simler, K. (1994b). Sources of Growth in AMfalawi: Iast Trends and Future Prospects, Malawi Agricultural Sector NMemorandum, Working Paper No. 1, World Bank. Simler, K. (1994c). Coping strategiesfor smzilaholder households and scopefor targeted interv entions to improve food security, Working Paper No. I 0, World Bank. Simler, K. and A. Quisumbing (1994). Poverty Profile of Rural Households in Malawi: A Summary of Recent Findings, Malawi Agricultural Sector Memorandum Working Paper No. 3, World Bank, Washington. Simler, K. and A. Quisumbing (1994). Poverty Profile of Rural Households in Malauwi: A Summary of Recent Findings, Malawi Agricultural Sector Memorandum Working Paper No. 3, Washington, DC: The World Bank. Smale. M. et al. (1991). Chimanga M(ha Makola, Hyhrids, and Composites: An Analysis of Farmers'Adoption of Maize Technology in Malawi, 1989-91. CIMMYT Economics Working Paper 91/04. Mexico City: CIMMYTI.. Smale, NM., R.E. Just, and H.D. Leathers (1994). Land Allocation in lliYVAdoption Miodels: An investiLgation of'Alternatite Evplanationi. American Journal of Agricultural Economics. 76: 535-546. Spring, A. (1990). Profiles of Men and Women SmalIholder Farmers in AMalawi. IIn African Food Systemis in Crisis, Part 7uwo: Contending with Change, edited by R.Huss- Ashmore and S.H. Katz, New York : Gordon and Breach Science Publishers. Spring, A. (1995). Agricultural Development and Gender Issues in Malauwi. New York: University Press of America. van der Wa]le, D. (I1994). Vietnam: Social Services, Safott Nets and The IPoor. In: World Bank. Poverty Assessment. . (1993). Indonesia: Wh'/)o Benefits from Public SSpending? In: WVorld Bank. Poverty Assessmcnt. Vella, V. (1995). Socioeconomic Classification of 'the Mahlai Rural Population, Human Resources Division, Southern Africa Department. Washington. DC: The World Bank. 90 | Malawi * Proflle and Prlorles for AcHon Werner, J. (1987). Labor Requirement and Distribution for Smaliholder Crops. Liwonde, Malawi: Liwonde Agricultural Development Division Adaptive Research. World Bank (1989). Malawi: Food Security Report. Report No. 8151-MAI, Southern Africa Department. World Bank (1990). Malawi: Growth Through Poverty Reduction. Report No. 8140-MAI, Southern Africa Department. World Bank (1993). Uruguay Poverty Assessment: Public Social Expenditures and their Impact on Income Distribution. World Bank (1994a). Malawi Social Action Fund: Project Operational Manual. World Bank (1994). Colombia Poverty Assessment: Poverty in Colombia. World Bank (1994b). Project Completion Report: Malawi Agricultural Sector Adjustment Credit. Report No. 88/93 CP-MLW 23 PCR. World Bank (1 994c). Malawi Agricultural Sector Memorandum: Strategy Options in the 1990s. Report No. 12805 MAI. Southern Africa Department, Agriculture and Environment Division. Government of Malawi, (1993). Malawi Market Prices (FEWS), Ministry of Agriculture, Planning Division, Agro-economic Survey, Lilongwe, Malawi. Government of Malawi, (1994). Guide to Agricultural Production in Malawi 1993-1994, Ministry of Agriculture, Lilongwe, Malawi. National Statistical Office (NSO) (1994). The 1992/1993 National Sample Survey of Agriculture, mimeo, Zomba, Malawi, NSO. Peters, Pauline. Personal Communication, (February 1995) Demographic and Health Surveys Malawi Demographic and Health Survey (1992). Malawi National Statistical Office and Macro International Inc. January 1994. Zambia Demographic and Health Survey (1992). University of Zambia, Zambia Central Statistical Office and Macro International Inc. March 1993. Zimbabwe Demographic and Health Survey (1994). Zimbabwe Central Statistical Office and Macro International Inc. January 1995. Tanzania Demographic and Health Survey (1991/1992). Tanzania Bureau of Statistics, Planning Commission and Macro International Inc. June 1993. Kenya Demographic and Health Survey (1993). Kenya National Council for Population and Development, Central Bureau of Statistics, Office of the Vice President and Ministry of Planning and National Development and Macro International Inc. May 1994. Botswana Family Health Survey H (1988). Botswana Central Statistics Office, Ministry of Finance and Development Planning, Family Health Division Ministry of Health and Macro International Inc. August 1989. Namibia Demographic and Health Survey (1992). Namibia Ministry of Health and Social Services and Macro International Inc. May 1993. The Africa Development Indicators (1994-95), The World Bank 1995. *4 Sd t,f"' ¾ ;1N4 4 4 N4Nr - 4 / 4Na4 .$;4p%N\ ,4N - 4. . 4 Nt ! 4 '4 44, N '4NN tN IMAGING  Report No 15437 MM Type SR