Report No. 73973-SZ Swaziland Using Public Transfers to Reduce Extreme Poverty November 24, 2012 Human Development Department Social Protection Unit Africa Region Kingdom of Swaziland Document of the World Bank GOVERNMENT FISCAL YEAR April 1 to March 30 CURRENCY EQUIVALENTS Currency Unit: Swaziland Lilangeni (Emalangeni) (as of December, 2011) US$1: SZL7.09 Abbreviations and Acronyms ADePT Software Platform for Automated Economic Analysis AIDS Acquired Immune Deficiency Syndrome ARV Antiretroviral CCT Conditional Cash Transfer CHH Child-headed Household CSO Central Statistical Office DHS Demographic and Health Survey DPM Deputy Prime Minister’s Office DSW Department of Social Welfare ECCD Early Childhood Care and Development FAO Food and Agriculture Organization FBO Faith-based Organization FPE Free Primary Education GDP Gross Domestic Product GF Global Fund GOS Government of Swaziland HDI Human Development Index HIES Household Income and Expenditures Survey HIV Human Immunodeficiency Virus IMF International Monetary Fund MOA Ministry of Agriculture MOET Ministry of Education and Training MOEPD Ministry of Economic Planning and Development MOLGH Ministry of Local Government and Housing MOH Ministry of Health MOTAD Ministry of Tinkhundla Administration and Development MPP Micro Projects Program NCCU National Children’s Coordinating Unit NCP Neighborhood Care Point NDMA National Disaster Management Agency NERCHA National Emergency Response Council on HIV and AIDS NGO Non-governmental Organization OAG Old Age Grant OVCs Orphans and Vulnerable Children PIN Personal Identification Number PMT Proxy Means Test PG Poverty Gap RHVP Regional Hunger and Vulnerability Program SACU South African Customs Union SEDCO Small Enterprise Development Company SFP School Feeding Program SNN Social Safety Net SPG Squared Poverty Gap TB Tuberculosis UNICEF United Nations Children's Fund VAC Vulnerability Assessment Committee WHO World Health Organization WFP World Food Programme Vice President : Makhtar Diop Country Director : Asad Alam Sector Director : Ritva S. Reinikka Sector Manager : Lynne D.Sherburne-Benz Task Team Leader : Emma S. Mistiaen Principal Author : Lorraine Blank Standard Disclaimer: . This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. FOREWORD Like many other countries, Swaziland faces daunting challenges, including widespread poverty, HIV/AIDS and its consequences, and shocks caused by the global financial crisis and price increases for food and fuel. In this context, it is vitally important that Government’s efforts to protect our poorest and most vulnerable are carefully constructed to maximize their impact and cost-effectiveness. The report focuses on the social safety net, particularly cash and in-kind transfers. The safety net can play an important role in addressing poverty and vulnerability; however, the process by which the safety nets have been developed in Swaziland has produced a fragmented system that leaves many Swazis unprotected by the safety net. Improvements in efficiency and effectiveness are both necessary and possible. Building on careful and detailed analysis of a wealth of data on safety net programs and expenditures in Swaziland and other developing countries, this report addresses the various challenges in a thorough yet accessible manner. This report highlights many areas in which Swaziland is challenged in providing an adequate safety net and identifies areas where a determined effort must be made to ensure that programs are efficient and effective so that they reach the most disadvantaged. As Swaziland navigates this period of fiscal crisis, it is essential that we craft a social safety net that helps vulnerable households to weather shocks and promotes investments in human capital development of children and facilitates transitions into the labor market. The Deputy Prime Minister’s Office is hopeful that this report will be used as a starting point for all those interested in promoting an efficient, effective, and sustainable safety net in Swaziland. TABLE OF CONTENTS Executive Summary ....................................................................................................................... i CHAPTER 1: INTRODUCTION............................................................................................... 1 A. Background ........................................................................................................................................................1 B. Why Safety Nets are Important ........................................................................................................................3 C. Scope and Methodology.....................................................................................................................................4 CHAPTER 2: POVERTY AND VULNERABILITY IN SWAZILAND ............................... 6 A. Overall Trends in Poverty 2001-2010...............................................................................................................6 B. 2010 Poverty Profile.........................................................................................................................................10 C. Food Insecurity and Vulnerability..................................................................................................................15 D. Shocks and Responses......................................................................................................................................17 E. Vulnerable Groups...........................................................................................................................................18 F. Conclusions .......................................................................................................................................................24 CHAPTER 3: THE SOCIAL SAFETY NET IN SWAZILAND........................................... 26 A. The Organization of the Social Safety Net .....................................................................................................26 B. Implementation Constraints in the Existing Safety Net ...............................................................................28 C. Cross-cutting Requirements............................................................................................................................45 D. Public Expenditures on the Safety Net ...........................................................................................................51 E. Conclusions .......................................................................................................................................................56 CHAPTER 4: Targeting and the Safety Net in Swaziland .................................................... 57 A. The Rationale for Targeting............................................................................................................................57 B. Common Targeting Methods ..........................................................................................................................59 C. Targeting in the Context of Swaziland ...........................................................................................................61 D. Conclusions .......................................................................................................................................................64 CHAPTER 5: TOWARDS EFFECTIVE AND RESPONSIVE SOCIAL TRANSFERS... 66 A. A Strategic Role for the Social Safety Net......................................................................................................66 B. Towards a Social Safety Net Strategy ............................................................................................................68 C. Matching Programs to Risks and Vulnerabilities .........................................................................................68 D. Options for Sustainable Social Transfers.......................................................................................................74 E. How Targeting Should Be Approached .........................................................................................................77 F. The Way Forward............................................................................................................................................78 References ............................................................................................................................................... 81 Annex 1: HIES Methodology............................................................................................................. 87 Annex 2: HIES Poverty Tables ......................................................................................................... 88 Annex 3: Detailed Expenditure Tables ........................................................................................... 98 Annex 4: Poverty Measurements and Poverty Mapping ......................................................... 101 Annex 5: PMT Formulae and Tables ........................................................................................... 102 MAP....................................................................................................................................................... 107 TABLES Table 1: Poverty and Extreme Poverty in Swaziland, 2001 and 2010............................................ 6 Table 2: Malnutrition in Swaziland and Comparable Countries .................................................. 16 Table 3: Poverty and Shocks......................................................................................................... 17 Table 4: Poverty and Children Under the Age of 6 ..................................................................... 19 Table 5: Social Transfer Programs in Swaziland.......................................................................... 27 Table 6: Social Pensions in Low- and Middle-income Countries ................................................ 29 Table 7: Distribution of OAG by Wealth Decile, 2009/10.......................................................... 30 Table 8: Share of OAG as a Percentage of Consumption, 2009/10............................................. 30 Table 9: Summary of Department of Social Welfare Cash Transfer Programs ........................... 32 Table 10: Coverage of the OVC Education Grant Scheme, 2010 ............................................... 34 Table 11: Beneficiaries and Coverage of the School Feeding Program, 2011/12 ........................ 37 Table 12: Costs of Meals per Child per School Day and per Child School Year, 2011/12.......... 37 Table 13: Physical Structure of NCPs ......................................................................................... 38 Table 14: Agencies Supporting NCPs .......................................................................................... 41 Table 15: Administrative Costs as a Percentage of Total Costs .................................................. 46 Table 16: Cost of Making Payments to Households.................................................................... 47 Table 17: Social Safety Net Expenditures in Selected African Countries.................................... 52 Table 18: Payments to Beneficiaries during the Fiscal Crisis ...................................................... 55 Table 19: Cost Scenarios of a Child Grant .................................................................................. 69 Table 20: Scenarios for Targeting the Old Age Grant................................................................. 73 Table 21: Core Safety Net Programs ............................................................................................ 74 Table 22: Indicative Costing a Core Set of Safety Net Programs................................................. 75 Table 23: Key Safety Net Reforms in the Short and Medium Term ............................................ 79 Annex Table 24: Swaziland Poverty Measures ............................................................................ 88 Annex Table 25: Poverty by Geographic Regions ....................................................................... 89 Annex Table 26: Poverty by Age Groups..................................................................................... 90 Annex Table 27: Poverty by Household Head's Age ................................................................... 91 Annex Table 28: Poverty and Education of Head ........................................................................ 92 Annex Table 29: Poverty by Household Head's Status of Employment ...................................... 93 Annex Table 30: Poverty and Gender of Head............................................................................. 94 Annex Table 31: Poverty and Household Composition ............................................................... 95 Annex Table 32: Poverty and Children Under 18 ........................................................................ 95 Annex Table 33: Poverty and Number of Dependents by Consumption Deciles and Poverty Status............................................................................................................................................. 96 Annex Table 34: Poverty and Orphanhood .................................................................................. 97 Annex Table 35: Social Safety Net Expenditures (E Million), FY2006/07 - FY2010/11............ 98 Annex Table 36: Distribution of Social Safety Net Spending, FY2010/11................................ 100 Annex Table 37: Rural OLS Regression, Swaziland.................................................................. 104 Annex Table 38: Urban OLS Regression, Swaziland................................................................. 105 Annex Table 39: Combined OLS Regression, Swaziland .......................................................... 106 FIGURES Figure 1: Human Development Index............................................................................................. 2 Figure 2: The Incidence of Growth, 2000/01 to 2009/10 ............................................................... 7 Figure 3: Poverty Incidence by Region in 2000/01 and 2009/10 ................................................... 7 Figure 4: Extreme Poverty Incidence by Region in 2000/01 and 2009/10..................................... 8 Figure 5: Poverty Maps................................................................................................................... 9 Figure 6: Poverty in Swaziland by Location (Rural and Urban) .................................................. 10 Figure 7: Poverty Rates by Education of Household Head .......................................................... 11 Figure 8: Percentage of Employed People Aged 15-49 by Gender and Wealth Quintile............. 13 Figure 9: Percentage of Households with Access to Safe Water by Area and Region in 2009/10 ....................................................................................................................................................... 14 Figure 10: Percentage of Households Using Electricity by Area and Quintile ............................ 15 Figure 11: Vulnerable Populations (Individuals) and Yield Production (Metric Tonnage) ......... 16 Figure 12: Nutritional Status of Children by Wealth Quintile...................................................... 19 Figure 13: Net Enrollment Rates for Primary and Secondary Education by Wealth Status, 2009/10 ......................................................................................................................................... 20 Figure 14: Poverty Rate and Orphan Status.................................................................................. 21 Figure 15: Percentage of OVCs Whose Households Received Assistance by Wealth Quintile .. 21 Figure 16: The HIV+ Population (Estimated) ............................................................................. 23 Figure 17: Coverage of Student Assistance by Wealth Quintile .................................................. 35 Figure 18: Neighborhood Care Points Providing Feeding............................................................ 39 Figure 19: Social Safety Net Spending Over Time ...................................................................... 52 Figure 20: Safety Net Expenditures as a Percentage of GDP....................................................... 52 Figure 21: The Distribution of Cash Transfers ............................................................................. 53 Figure 22: Recurrent and Capital Expenditures over Time .......................................................... 54 BOXES Box 1: How Do Safety Nets Contribute to Development Policy?................................................. 3 Box 2: The Public Safety Net in Swaziland.................................................................................... 4 Box 3: Child-Headed Households ............................................................................................... 22 Box 4: Access to Antiretroviral Therapy..................................................................................... 23 Box 5: Eligibility Criteria for OVC Grants................................................................................... 50 %R[7DUJHWLQJ&DVK7UDQVIHUVLQ=DPELD௅$Q$VVHVVPHQWRI'LIIHUHQW6HOHFWLRQ&ULWHULD ... 57 Box 7: Community-based Targeting in Africa ............................................................................ 60 Box 8: The Child Grant Program and the National Information System for Social Assistance (NISSA) in Lesotho ...................................................................................................................... 63 Box 9: Conditional Cash Transfers............................................................................................... 70 Box 10: The Productive Safety Net in Ethiopia............................................................................ 71 ACKNOWLEDGEMENTS This report was produced under the guidance of the Deputy Prime Minister’s Office in Swaziland and prepared by a team led by Emma Mistiaen. The primary author was Lorraine Blank. The poverty and targeting chapters were mainly written by Jeanine Braithwaite with valuable support from Lorraine Blank, Elina Sheja, David Cates, Sanjay Choudhury, Anna Mohan, Robert Pickmans, and Ismael Yacoubou Djima. Curline Beckford provided invaluable assistance in collecting and analyzing the expenditure data. The authors are grateful for the essential support provided by Melis Guven who led the final consultations and finalization process of the report. The authors greatly appreciate the time and input provided by all of the stakeholders consulted during the preparation of this review, particularly staff in the Deputy Prime Minister’s Office, the Ministries of Finance; Education and Training; Tinkhundla Administration and Development; and the National Emergency Response Council on HIV and AIDS and stakeholders from civil society and the development partners. The authors are thankful to the Swaziland Central Statistical Organization for permission to use its survey data. The report benefited significantly from consultations on the preliminary findings and the final draft of the report with representatives of the Government of Swaziland, civil society, and development partners. We would like to thank the participants in workshops held in Swaziland in March 2012 and September 2012. World Bank staff from the Africa Social Protection Unit, including Lynne Sherburne-Benz, Carlo del Ninno, Victoria Monchuk, and Phillippe Auffret, provided comments and contributions on the preliminary findings of the report. The report was peer reviewed by Margaret Grosh and John Elder, and the team is grateful for their comments and advice as well as those from Ashish Narain, Eugenia Marinova, Marco Scuriatti, and Claus Pram Astrup. The report was prepared under the overall guidance of Lynne Sherburne- Benz, Social Protection Sector Manager. Fiona Mackintosh edited the document and Ana Lukau provided formatting and other processing support for the report. Financial support to prepare this report was provided by the World Bank (through a grant from the Poverty and Social Impact Assessment Trust Fund). The views and conclusions expressed herein are those of the authors. EXECUTIVE SUMMARY Main Messages x Swaziland implements a number of social transfer programs, but these are not well coordinated and there is a need to better define the overall priorities and objectives of the safety net. x Swaziland spends a significant amount on social safety nets (2.2 percent of GDP), but there is room to increase the efficiency and effectiveness of spending by: (i) improving coordination between programs working towards the same objective; (ii) making adjustments to the design and implementation of these programs; and (iii) removing programs that are not contributing to the overall objective of the safety net. x Social transfers could play a larger role in reducing poverty by more efficiently and effectively targeting the poorest and most vulnerable. x The existing safety net provides important support, but there are gaps, particularly with respect to extremely poor households with children and poor rural households with unemployed household members. x Swaziland should consider adopting a national child grant program and a public works program to fill existing gaps in the safety net. Two useful first steps would be to: (i) learn lessons from the upcoming cash transfer pilot for orphans and vulnerable children and (ii) pilot a public works program in rural areas. 1. INTRODUCTION 1. Persistent poverty in Swaziland, exacerbated by the devastating impact of the Human Immunodeficiency Virus/Acquired Immune Deficiency Syndrome (HIV/AIDS), the global financial crisis, external commodity price shocks (particularly for food and fuel), and the deteriorating fiscal situation, makes it difficult for the government to protect its poorest and most vulnerable citizens. It is in this context that this report examines the country’s existing social safety net programs with a particular focus on social transfers and makes some recommendations about how to make the safety net more effective, efficient, and sustainable. 2. POVERTY AND VULNERABILITY IN SWAZILAND 2. In 2009/10, 63 percent of the total Swazi population of 1.2 million was estimated to live below the poverty line. This estimate was down from 69 percent in 2000/01, which translates into a reduction in the number of poor people from 678,500 in 2000/01 to 641,000 in 2009/10. However, this decline in poverty has not been equally distributed throughout the country. Nor has extreme poverty declined significantly. Although the percentage of extreme poor households in the population declined slightly over the decade, the number of extreme poor actually increased due to population growth – from around 288,000 in 2001 to 295,000 in 2010. Swaziland has experienced relatively modest growth in recent years, but this has not benefitted the poorest segments of the population. Growth rates in consumption have been significantly higher for those in the middle part of the income distribution than for those in the bottom 30 percent who experienced a further reduction in income. i 3. The most important features of poverty in Swaziland are its pervasiveness and its homogeneity. The elderly and children are poorer than active-aged adults but not by as much as in many other medium-income countries. Seventy percent of children aged 15 and younger are poor and 71 percent of people aged 60 and over are poor. Households with no elderly or child members constitute only 9 percent of the population, but far fewer of them are poor (only 25 percent) than households that include both (78 percent). The number of children under the age of 18 in a household is strongly correlated with poverty as poverty rates increase from 45 percent for households with one child to 90 to 100 percent in households with eight or more children. The larger the household, the more likely it is that the household will be poor. Also, poverty incidence among households with six or more members is higher than the national average. 4. Poverty and extreme poverty in Swaziland are both overwhelmingly rural phenomena. The incidence of poverty is 73 percent in rural areas but only 31 percent in urban areas. Eighty-eight percent of the poor and 95 percent of the extreme poor live in rural areas, and the average consumption of the urban poor is 33 percent below the poverty line while it is 51 percent below the poverty line among the rural poor. Also, poverty is deeper in rural areas than it is in urban areas. 5. School enrollment rates and poverty status are correlated, especially at the secondary level. Fifty-six percent of children in the wealthiest quintile are enrolled in secondary school compared to fewer than 26 percent of children in the lowest quintile. Cost is the primary reason given by survey respondents for why their children do not attend school, and this is true irrespective of wealth status. Keeping children in school is crucial since poverty is associated with a lack of education in Swaziland. 6. The nutritional status of children and their poverty status are correlated. Six percent of children under the age of 5 in Swaziland are underweight. Thirty-one percent of children are stunted or too short for their age and 3 percent are wasted or too thin for their height. Children from households that are poor are much more likely to be malnourished than their wealthier counterparts. 7. Both poverty and low agricultural productivity are widespread, with noticeable fluctuations at the district level. Households living on Swazi Nation Land (75 percent of whom are poor) are less productive than commercial producers who produce on land for which they have a title deed. ii 3. EXISTING TRANSFER AND OTHER SOCIAL SAFETY NET PROGRAMS 8. The safety net system is comprised of multiple cash and in-kind transfers, social care services, active labor market programs, and community-based programs implemented by at least five ministries or departments within ministries. Agencies delivering safety net services include the Department of Social Welfare (DSW) within the Deputy Prime Minister’s Office (DPM), the National Emergency Response Council on HIV and AIDS (NERCHA), the Ministry of Education and Training (MOET), the Ministry of Health, and the National Disaster Management Agency (NDMA) in the DPM’s Office. The following table lists the programs that currently make up Swaziland’s safety net. The Public Safety Net in Swaziland (Program/Implementing Agency) Other Social Safety Net Cash Transfers In-Kind Transfers Programs Old Age Grant /DPM OVC Education Grants /DPM Community-based Programs/MPP Public Assistance /DPM School Feeding /MOET & NERCHA Active Labor Market Programs/ Military Pensions/DPM Neighborhood Care Points /NERCHA MOET, MOL, & others Other DSW Grants /DPM Food Distribution /NDMA & NERCHA Social Care Services /DPM, Young Heroes /NERCHA Supplementary Feeding /MOH NERCHA & others) Health Fee Waivers /MOH Phalala Fund /MOH Agricultural Input Subsidies /MOA 9. In FY2010/11, Swaziland spent E584.4 million (US$82.4 million) on its safety net. This figure encompasses expenditures on cash and in-kind transfers, community-based programs, active labor market programs, and social care services. Between FY2005/06 and FY2010/11, safety net expenditures increased both in real terms and as a percentage of GDP. At about 2.2 percent of GDP in 2010/11, spending on the safety net in Swaziland is higher than the average of 1 to 2 percent spent by most developing countries but is lower than the amount spent by several other Southern African countries (including Malawi, Botswana, Mauritius, and South Africa). Expenditures on cash transfers were E131.6 million (US$18.6 million) in FY2010/11 or about 23 percent of safety net expenditures, while expenditures on in-kind transfers were E403.8 million (US$57.0 million) or about 69 percent of safety net spending. Other programs received 8 percent of the safety net budget. Between FY2006/07 and FY 2010/11, expenditures on the Old Age Grant (OAG) direct transfers increased from E42.6 million to E100.3 million (US$6.1 million to US$14.2 million) or by 135 percent in nominal terms and 72 percent in real terms. Capital expenditures, which include externally funded and local capital projects, accounted for about 21 percent of total safety net spending and represented a declining share of overall safety net expenditures. 10. Old Age Grants (OAG), the largest cash transfer program in Swaziland, has the potential to have a greater impact on poverty. 1 It was introduced in 2005 to cushion the impact on the elderly of the loss of support resulting from the deaths of their adult children and 1 While the study recognizes that political economy is an important aspect of social protection, assessment of political economy of social safety nets is beyond the scope of this study. iii the increasing burden of caring for orphaned grandchildren. Qualitative evidence suggests that the OAG has had a positive impact on beneficiaries` lives but has not contributed much in terms of reducing poverty because a significant share of OAG resources go to non-poor households. The age of eligibility for the OAG is lower than that for equivalent programs in many other countries, but the value of the benefit is broadly comparable. The irregular disbursement of benefits makes it difficult for recipients to rely on the payment, which is particularly problematic at particular times in the year when beneficiaries could use some of the cash for productive purposes, for example, at the start of the planting season in rural areas. 11. Programs for children have suffered from significant implementation challenges and an unclear policy framework. The OVC (Orphans and Vulnerable Children) Education Grant Program has had difficulty tracking individual students. There is no policy on whether students should pay any fees not covered by the grant, and some children are required to pay the remaining fee while others are not. In addition, the OVC Education Grant covers only a part of the official fees and does not fund the other out-of-pocket costs of education. As a result, the cost of education remains the primary reason why families do not send their children to school. 12. Although the government has made efforts to provide social protection to the poor and vulnerable, substantial numbers of poor and vulnerable people, particularly children, remain unprotected. For example, Swaziland has no cash assistance program for extremely poor households with children. The OVC Education Grant Program reaches only about 40 percent of children in the poorest quintiles and does not cover all of the significant out-of-pocket costs of education. The OAG benefits children indirectly if they live in a household with an elderly person, but 55 percent of poor children and 56 percent of very poor children do not live with an elderly person. When the government suspended the indigent category from the Public Assistance program in response to the fiscal crisis, this meant that what little support was available for very poor households with children is no longer available. Even when it was operating, the indigent category served fewer than 1,000 beneficiaries, which represents not even 1 percent of the extreme poor. The school feeding programs provide support to all schoolchildren, but those children who cannot afford to go to school do not benefit. Also, the absence of any public works program in Swaziland means that unemployed youths and adults have no way to raise any income to smooth the consumption of their households or support them in the aftermath of shocks or crises. 13. Improvements in administrative efficiency and accountability are possible. Information on the administrative costs of safety net programs is incomplete, but the available data suggest that the administrative costs of Swaziland’s safety net programs are high relative to those in other countries. The programs rely on a variety of mechanisms to get their grants to the beneficiary households. There has been a gradual movement towards paying OAG beneficiaries electronically, but the transition has been slower than in many other countries. At the same time, the transaction costs involved in bank-deposited benefits are also high relative to other countries. Economies of scale could be realized if programs were to share certain common functions, such as targeting, registration, information management, and monitoring and evaluation. Also, having a central beneficiary registry would provide important monitoring and planning information and could help to reduce fraud and abuse. iv 14. Swaziland needs an effective safety net response to shocks and other crises. In what is known as a “countercyclical” approach, government spending on safety nets should ideally increase in times when households are facing crises or shocks to help them to weather the shock and then should decrease as the impact of the shock subsides. However, the Government of Swaziland’s response to the fiscal crisis has been to reduce expenditures on safety net programs, particularly cash transfers, by delaying payments to some OAG recipients and eliminating the element of Public Assistance that was intended to target the poor. At the same time, many shocks experienced in Swaziland are seasonal shocks that can usually be anticipated, and there is evidence from other countries that predictable cash transfers may therefore be a more appropriate response to this pattern of seasonal risks. 15. Each program has its own unique constraints, but there are some general constraints that affect all programs. The monitoring of programs’ performance and impact is almost universally weak, and there has been no rigorous impact evaluation of any safety net program. Accountability and control need to be strengthened, including more modern information systems, updated accounting systems, more modern payment mechanisms for cash benefits, and more transparent targeting systems. The introduction of public information and education campaigns, appeals mechanisms, documented policies and procedures, and regular process audits would also help to promote accountability and control. 16. The financial sustainability of several social transfer programs is questionable given their reliance on external funding. This includes the School Feeding Program, Neighborhood Care Points, and other smaller programs. About 21 percent of total safety net spending is spent through the capital budget (on both external and local capital projects), and the sustainability of this spending can only be ensured if these items can be absorbed into the recurrent budget. 4. TARGETING AND THE SAFETY NET IN SWAZILAND 17. Programs could do a better job of identifying the vulnerable. Targeted programs each have their own application procedures and eligibility criteria, and these multiple targeting mechanisms increase administrative costs. Also, some people such as the poor or children from child-headed households can find it difficult to navigate the system. The targeting criteria for many programs are subjective, which increases the likelihood of inclusion and exclusion errors. The government has developed new criteria aimed at making the identification of OVCs more transparent, but they are still very vague and open to subjective interpretation. Developing more objective and transparent targeting mechanisms could significantly reduce the problems that can result in inclusion and exclusion errors. 18. There is a need for more solid evidence about which targeting methods work best in the context of Swaziland, but the little evidence that exists indicates that the safety net programs are not particularly well targeted to the poorest. It is important for the government to start testing different targeting methods (or combinations of methods) in a systematic way in order to learn lessons relevant to the Swazi context that will enable the safety net to better reach the poorest. It is however important to note that the choice of targeting method is not as important as the effective implementation of whichever method is selected. v 19. Some countries have developed a single unified registration system covering all safety net programs, an idea which the government could consider adopting. A unified registration system or at a minimum a unified registry of beneficiaries would increase the efficiency, effectiveness, and transparency of targeting and reduce duplications and overlaps. It is worth exploring both of these options further to determine if they are feasible and desirable in the context of Swaziland. 5. TOWARDS EFFECTIVE AND RESPONSIVE SOCIAL TRANSFERS 20. Persistent poverty and inequality in Swaziland provides a strong justification for spending on social transfers. Based on our analysis throughout this report, we believe that it would make sense to target poor elderly people and poor children, as well as unemployed people in rural areas. The key attributes of poverty that we recommend that the government should focus on are low human capital development and unemployment. 21. Swaziland needs a well-articulated social safety net strategy. Swaziland currently has an uncoordinated set of safety net programs and a number of policy documents that cover some aspects of social protection, but they do not form a coherent whole. A range of social transfer programs exist in Swaziland that collectively attempt to reduce poverty and vulnerability, but gaps in the safety net remain. In particular, extremely poor children and extremely poor unemployed youths and adults are not adequately protected by the current safety net. To improve the implementation and coordination of safety net programs, there is a need to strengthen institutional capacity. 22. A well-crafted safety net should consist of a range of programs that address the critical risks and vulnerability faced by the population and that serve the dual role of protection and promotion. The analysis in this report suggests that the core safety net in Swaziland could consist of the following existing and proposed programs: Existing New Old Age Grant Child Grant (pilot initially) OVC Education Grant Public Works (pilot initially) School Feeding Program 23. These core safety net programs are affordable, but the Government of Swaziland will need to make a strong fiscal commitment to its social safety net. Using a medium cost scenario, we estimate that expenditures on the core programs would amount to about E428.6 million (US$60.5 million). This represents 4.3 percent of central government spending and about 1.6 percent of GDP. The government spent an estimated E584.4 million on the safety net in FY2010/11. With no changes in any of the other existing safety programs that are not included in the core and adding the two core programs as proposed, the new total is estimated to be E607.5 million (US$85.7 million) or 6 percent of central government spending and 2.2 percent of GDP. With only a 10 percent reduction in expenditures on the existing “non-core” programs, the proposed options would be essentially budget neutral. vi The Way Forward 24. This review of the country’s social transfers indicates that many of the elements of an appropriate safety net are present but that it ought to be possible to increase efficiency and effectiveness. Based on the analysis in this report, the safety net strategy could include the following elements: Key Safety Net Reforms in the Short and Medium Term Short-term Strategy Medium-term Strategy (1-2 Years) (3-5 Years) Program-specific Recommendations ߤ Ensure reliable budget allocation to ߤ Fully implement targeting plan the program ߤ Decide whether to increase eligibility age ߤ Ensure regular payments to beneficiaries ߤ Determine targeting criteria, develop plan for implementation, and begin implementation of plan Old Age ߤ Modernize payment systems Grant ߤ Upgrade MIS ߤ Introduce mechanism for regularly increasing benefits ߤ Assess costs and benefits of raising the age of eligibility ߤ Prohibit military pensioners from receiving the OAG ߤ Design pilot ߤ Evaluate pilot ߤ Develop systems ߤ Expand based on the findings of evaluation Child ߤ Implement pilot ߤ Secure government financing Grant ߤ Fully implement the issuing of PINs to all children ߤ Improve targeting ߤ Secure government financing ߤ Document policies and procedures ߤ Make beneficiaries of Child Grant OVC ߤ Upgrade MIS and establish link to automatically eligible for OVC Education Education Child Grant MIS Grant Grant ߤ Strengthen monitoring ߤ Fully implement the issuing of PINs to all children ߤ Strengthen monitoring and supervision ߤ Secure government financing School ߤ Implement process audits ߤ Redesign program as required based on the Feeding ߤ Evaluate the program findings of process audits and evaluations ߤ Assess feasibility ߤ Implement pilot ߤ Cost pilot ߤ Evaluate pilot Public ߤ Identify financing for pilot ߤ Secure financing for potential expansion Works ߤ Design pilot ߤ Develop systems Cross-cutting Recommendations Social ߤ Form inter-ministerial committee ߤ Core safety net programs reflected in the Safety Net ߤ Hold national consultations recurrent budget Strategy ߤ Develop SSN strategy Efficiency ߤ Modern MIS for all programs ߤ Establish central beneficiary registry and ߤ Modernize payment systems for all ߤ Merge small safety net programs and vii Account- programs remove those that are not consistent with ability ߤ Build capacity of DPM the overall safety net objective ߤ Improve accounting systems ߤ Implement information and education campaigns ߤ Establish appeals and complaints mechanisms ߤ Conduct regular process audits ߤ Document policies and procedures for all programs ߤ Ensure regular payments to beneficiaries Targeting ߤ Develop and pilot objective targeting ߤ Establish objective targeting criteria for all methodologies targeted programs Program ߤ Develop monitoring and evaluation ߤ Mobilize resources for more regular Monitorin framework poverty and labor force surveys g Crisis ߤ Pilot public works (see above) ߤ Move away from providing food aid as a Response crisis response viii CHAPTER 1: INTRODUCTION A. Background 1. Swaziland faces daunting challenges in its efforts to provide an adequate safety net for its population of 1.2 million. Persistent poverty, exacerbated by the devastating impact of HIV/AIDS, the global financial crisis, external commodity price shocks (particularly for food and fuel), and the deteriorating fiscal situation, makes it difficult for the Government of Swaziland to protect its poorest and most vulnerable citizens. Meanwhile, the informal safety net has been overwhelmed, and the need for formal safety nets will continue to increase. 2. The progress made so far in reducing poverty has not helped the very poorest. Although poverty declined from 69 percent in 2000 to 63 percent in 2010, it remains high. About 641,000 Swazis are poor, of whom approximately 295,000 are extremely poor (including about 181,000 children) and are unable to meet their daily nutritional needs. 2 Despite the overall decrease in poverty over the last decade, the share of extremely poor people in the population has remained essentially the same (30 percent in 2000 and 29 percent in 2010). Growth in consumption has been significantly higher for those households in the middle part of the income distribution, whereas those in the bottom 30 percent of the distribution experienced a further reduction in consumption. 3 In addition to poverty, every year Swazi households face the risk of other shocks, to which the poor are typically most vulnerable, including the seasonal unpredictability of the food supply, food price increases, droughts, floods, and the sickness and death of the working members of their households. 3. Social indicators highlight the magnitude of the challenges that the country faces. The United Nations’ Human Development Report ranked Swaziland very low in terms of the well-being of its citizens - at 140 out of 187 countries. 4 Moreover, its Human Development Index (HDI) score has deteriorated from 0.641 in 1995 to 0.522 in 2011. The HDI for Swaziland is now approximately equal to the average for all of Sub-Saharan Africa (Figure 1). This decline can largely be attributed to the decline in life expectancy, which stood at 59 years in the early 1990s but had dropped to 49 years by 2010. On a positive note, Swaziland’s score on the Education Index, a component of the HDI, continues to improve, with more children accessing education than in previous years. The key social protection challenge will be to ensure that safety net programs are designed in a way that will continue these education gains. 2 See Chapter 2 for a more detailed discussion of poverty and vulnerability. 3 CSO (2011c) 4 The ranking is based on a composite measure that covers three dimensions of human welfare (income, education, and health) and serves as an indicator of national well-being. 1 Figure 1: Human Development Index HDI Value Source: UNDP (2011) 4. The current economic and fiscal crises have compounded the country’s already significant challenges. Real GDP growth slowed to an estimated 0.3 percent in 2011, compared with 2.0 percent in 2010 and an average of 3.0 percent in the pre-crisis period. Inflation increased from 4.5 percent in 2010 to 5.5 percent in 2011, but this was still much lower than the peak of 12.6 percent recorded in 2008. 5 The impact of the global financial and economic crisis on the economy of Swaziland mostly manifested itself in the form of contracted demand for Swazi exports, with the manufacturing sector being particularly affected. Unemployment stood at 28.5 percent in 2010 and, if discouraged workers 6 are included, that rate increases to 40.6 percent. Unemployment has a strong rural dimension in that it is much higher (37.1 percent) in rural areas than in urban areas (16.7 percent). 5. The global financial and economic crisis has contributed to the unprecedented fall in Swaziland’s trade-related revenue from the Southern African Customs Union (SACU). Swaziland recorded fiscal surpluses in the period leading up to the global financial crisis as a result of SACU revenue transfers. Between 2008/09 and 2010/11, Swaziland’s SACU receipts fell by 56 percent from E6.0 billion to E2.6 billion, largely due to the reduction in trade. 7 The fiscal deficit is expected to have reached close to 10 percent of GDP in FY2011/12. 8 6. Sluggish economic growth and revenue shortfalls have made it difficult for the Government of Swaziland to finance its key safety net programs. Going forward it will need to ensure that it maximizes the returns to its investments in its social safety net. It is in this context that this report examines the country’s social safety net with a particular focus on social transfers and makes recommendations for making it more effective, efficient, and sustainable. 5 IMF (2011) 6 Defined as workers who have stopped looking for jobs but who want to work and are available to work. 7 Ministry of Economic Planning and Development (2011) 8 IMF (2011) 2 B. Why Safety Nets are Important 7. Social safety nets consist of non-contributory transfer programs targeted to the poor or vulnerable. A country’s safety net typically includes some combination of cash and in-kind transfers, subsidies, public works programs, 9 social care services, and community-based interventions. The safety net could and should form an important part of the overall poverty reduction strategy in the long run while also helping to reduce extreme poverty in the short run. Safety net programs help to reduce the welfare losses that households experience when shocks occur and are an important tool for reducing households’ vulnerability to a range of factors, including but not limited to poverty. Safety nets can also help households to invest in productive activities and in the development of their own human capital, both of which can ultimately increase productivity and income and in this way be a tool for promoting human development, economic growth, and sustainable poverty reduction (Box 1). Box 1: How Do Safety Nets Contribute to Development Policy? Safety nets are one part of a country’s broader poverty reduction strategy. They interact with and exist alongside social insurance, health, education, and financial services, the provision of utilities and roads, and other programs and policies that aim to reduce poverty and minimize risk. Safety net programs can play four main roles in development policy: x Safety nets redistribute income to the poorest and most vulnerable, which has an immediate impact in terms of reducing poverty and inequality. The strong consensus in most countries is that adequate provision should be made to protect the poor, though there may be different beliefs about how this should be achieved. x Safety nets can enable households to make more and better investments in their future. They do so basically by compensating for the absence of credit markets for the poor, thus allowing these households to take advantage of investment opportunities that they would otherwise miss - both in the human capital of their children and in the livelihoods of their members of working age. x Safety nets help households to manage risk. At a minimum, safety net programs help households that are facing hard times to avoid irreversible losses, allowing them to maintain the household and business assets on which their livelihoods are based and to adequately nourish and school their children. At best, they can provide an insurance element that lets households invest in alternative livelihoods that yield higher earnings. Thus, safety nets not only protect households but also promote their independence. x Safety nets allow governments to make choices that support efficiency and growth. An adequate permanent social assistance system can fulfill whatever redistributive goals the society has, freeing other sectors from the role and allowing them to concentrate on the efficient provision of services. Thus, for example, the energy sector can set its prices for efficiency, and trade policy can focus on growth rather than job protection. Short-term safety net programs can compensate those who are negatively affected by needed reforms. Source: Grosh et al (2008) 8. The Government of Swaziland has demonstrated its commitment to safety nets in a number of different ways. It has endorsed important international treaties and agreements 10 and 9 Also referred to as workfare or short-term employment programs. 10 Including the ILO Philadelphia Declaration (1944); the Universal Declaration of Human Rights (1948); the International Covenant on Economic, Social, and Cultural Rights; the Convention on the Rights of the Child 1989; the African Charter on Human and Peoples’ Rights; the SADC Charter of Fundamental Social Rights; and the Livingston Accord. 3 issued a number of local policy statements. 11 In combination, these documents stipulate that the government, in collaboration with its partners, will provide social safety nets to cushion vulnerable individuals and families. The government currently allocates approximately 2.2 percent of GDP to safety net programs and has recently introduced new programs, including the Old Age Grant and the Orphans and Vulnerable Children Education Grants. Given the magnitude of vulnerability in Swaziland, the country’s scarce fiscal resources, and the government’s stated commitment to its social safety net, it is important to ensure not only that the social safety net is efficient but also that it is relevant to the current context in Swaziland. C. Scope and Methodology 9. This review focuses on the largest social transfer programs in Swaziland. The review includes all publicly funded social safety net programs. These programs were identified by our government counterparts and other stakeholders based on the definition of social safety net programs as non-contributory programs targeted to poor or otherwise vulnerable people. These programs have different objectives, for example, some are designed to address chronic poverty or vulnerability while others aim to mitigate short-term poverty. Box 2 lists the programs that make up the social safety net. Box 2: The Public Safety Net in Swaziland Program/Implementing Agency Cash Transfers In-kind Transfers Other SSN Programs Old Age Grant /DPM OVC Education Grants /DPM Community-based Programs/MPP Public Assistance /DPM School Feeding /MOET & NERCHA Active Labor Market Programs/ Military Pensions/DPM Neighborhood Care Points /NERCHA MOET, MOL, & others Other DSW Grants /DPM Food Distribution /NDMA & NERCHA Social Care Services /DPM, Young Heroes /NERCHA Supplementary Feeding /MOH NERCHA & others) Health Fee Waivers /MOH Phalala Fund /MOH Agricultural Input Subsidies /MOA 10. The objective of this study is to identify viable ways to make the safety net more relevant and efficient through an in-depth analysis of poverty and vulnerability and of the efficacy of current safety net programs. The report focuses on publicly financed social transfers in Swaziland, including cash and in-kind transfers. This includes programs funded by either national or official international aid. Chapter 2 explores the risks faced by the Swazis, including but not limited to poverty. Chapter 3 reviews current social net programs and expenditures and analyzes the efficiency and effectiveness of social transfers. Chapter 4 analyzes ways to target safety net programs, and Chapter 5 discusses options to increase the relevance and efficiency of the safety net, particularly in light of the recent financial crisis. 11 Including the Constitution, the National Social Development Policy, the National Multi-sectoral Strategic Framework on HIV and AIDS 2009-2014, the National Children’s Policy, the National Plan of Action for Children, the Poverty Reduction Strategy, the National Food Security Policy, the Disaster Management Policy, the Comprehensive Agricultural Policy, the HIV/AIDS Policy, and the Department of Social Welfare Strategic Plan. 4 11. There are a number of limitations to our analysis. Because some programs did not supply us with expenditure information, our expenditure analysis should be considered as a close approximation but not an exact reflection of total safety net spending. In addition, it was beyond the scope of this report to analyze community-based safety net programs, active labor market programs (including second chance education and training programs targeted to the poor and vulnerable), or social care services. Therefore, the report looks only briefly at these important kinds of safety nets, but it does attempt to capture expenditures on these programs. Nor does this review cover the range of small programs implemented by non-governmental organizations (NGOs) and faith-based organizations (FBOs). It does, however, include in the expenditure review any subventions from the government budget given to those NGOs that provide safety net services. Finally, this review does not examine contributory social insurance programs, although it recognizes that social insurance forms an important part of the social protection system in Swaziland. 12. This report has drawn on data from a variety of sources. The review team held individual and group meetings with policymakers, public sector managers, program staff, and representatives of international agencies. The team reviewed Government of Swaziland Estimates of Expenditure and program-level expenditure information as well as other reports, reviews, and secondary data sources. Also, the team analyzed data from the Swaziland Household Income and Expenditure Surveys (SHIES) of 2000/01 and 2009/10 and secondary data sources for the risk and vulnerability and targeting analyses. 5 CHAPTER 2: POVERTY AND VULNERABILITY IN SWAZILAND 13. In this chapter, we examine poverty and vulnerability in Swaziland. We draw a profile of the poor and extreme poor and examine the sources of their vulnerability, including food insecurity and shocks, and their movements in and out of poverty. The chapter also provides in-depth analysis of vulnerable groups (children, elderly, and those with HIV/AIDS). The main data sources for this analysis were the two most recent household budget surveys, 12 but we also consulted a considerable amount of secondary literature. A. Overall Trends in Poverty 2001-2010 14. The Swaziland Household Income and Expenditure Survey of 2009/10 showed that 63 percent of the total population was living below the basic needs poverty line, down from 69 percent in 2000/01. This translates into a reduction in the number of poor people from 678,500 in 2000/01 to 641,000 in 2009/10. 13 There was no equivalent drop in extreme poverty over the past decade, with about 295,000 Swazis recorded as living below the extreme poverty line in 2010 (Table 1). 14 Swaziland’s inequality of consumption (its Gini coefficient is 0.425) is lower than many other African countries, which reflects its more homogeneous living standards, but income inequality is higher (0.504). 15 The decline in the level of poverty occurred both in rural areas (from 80 to 73 percent) and in urban areas (from 36 to 31 percent). 16 A description of the Household Income and Expenditure Survey (HIES) methodology is presented in Annex 1 and additional poverty measurements are presented in Annex Table 2.3. Table 1: Poverty and Extreme Poverty in Swaziland, 2001 and 2010 Extreme Number of Number of Extreme Poverty Poverty Poor Poor Year (percent) (percent) 2001 69 30 678,500 288,000 2010 63 29 641,000 295,000 Source: Authors’ calculations based on SHIES 2000/01 and 2009/10 12 Swazi Household Income and Expenditure Survey (SHIES) for 2000/01 and 2009/10 13 The poverty line used to calculate these 2009/2010 figures was determined to be E461 per month per adult equivalent in constant January 2010 currency, and the extreme poverty line was set at E215 per adult equivalent. The methodology is described by the Swazi Central Statistical Office (CSO) in two reports: CSO (2011a) and CSO (undated). Poverty was defined based on a cost of basic needs approach, and extreme poverty was defined as consumption below the cost of the food portion of the basic needs poverty line. 14 It is important to recognize that these findings are sensitive to where the poverty line is drawn. According to the CSO, setting the extreme poverty line at E180 per month per adult equivalent in 2009/2010 would have yielded the conclusion that extreme poverty in Swaziland had actually increased over the previous decade (see Annex 1). 15 This is the Gini coefficient in 2010 as calculated using per capita consumption by the World Bank’s ADePT software. The US Central Intelligence Agency reported the Gini coefficient for family income in 2001 as 0.504. https://www.cia.gov/library/publications/the-world-factbook/rankorder/2172rank.html. Income is typically much more unequally distributed than consumption. 16 CSO (2011a) 6 15. Growth has not benefitted the poorest segments of the population. Swaziland has experienced relatively modest growth, but the growth rates in consumption have been significantly higher for those in the middle part of the income distribution, whereas the bottom 30 percent of the distribution has experienced a further reduction in income (Figure 2). 17 Figure 2: The Incidence of Growth, 2000/01 to 2009/10 Source: CSO (2011 a), Figure 9, p. 16. 16. The decline in poverty was not equally distributed throughout the country. While the poverty headcount in Shiselweni declined by 14 percentage points, there was no real change in poverty in Hhohho and Lubombo (Figure 3). Extreme poverty decreased in Shiselweni, increased in Hhohho and Lubombo, and remained constant in Manzini (Figure 4). Poverty by geographic regions is also presented in Annex Table 2.4. Figure 3: Poverty Incidence by Region in 2000/01 and 2009/10 100% 80% 60% 2000/01 40% 2009/10 20% 0% Hhohho Manzini Shiselweni Lubombo Source: SHIES (CSO, 2011a), Figure 1, Table A1.1. 17 CSO (2011a) 7 Figure 4: Extreme Poverty Incidence by Region in 2000/01 and 2009/10 40% 35% 30% 25% 20% 2000/01 15% 2009/10 10% 5% 0% Hhohho Manzini Shiselweni Lubombo Swaziland Source: SHIES (CSO 2011a), Figure 3, Table A1.2. 17. A recently completed poverty map demonstrates that the four regions of Swaziland are far from homogenous and that three of the four regions have pockets of both wealth and extreme poverty. In particular, Shiselwheni is poorer than the other regions, but its population lives closer to the poverty line and it has two pockets of relatively low poverty. The other three regions have more variation, with very poor districts alternating with better-off districts (Figure 5). 18 18 CSO (2011b) 8 Figure 5: Poverty Maps Tinkhundla Po\'crty Hcadcount (in%) Source: Ca.lcularjons based on SHlES 2009/JO and Census Source: CSO (2011b), p. 11. 9 B. 2010 Poverty Profile 18. The most important feature of poverty in Swaziland other than its pervasiveness is its homogeneity in the sense that there is little difference in key poverty indicators. The elderly and children are poorer than active-aged adults but not as sharply as in many other medium-income countries (Annex Table 2.5). Poverty is lower for households with active-aged heads and heads with more education (Annex Tables 2.6 and 2.7). A more notable result is the sharply lower than average poverty rate for heads who work in the public service (Annex Table 2.8). Headship was about evenly split by gender in the 2010 Swazi population, with 50.5 percent of the population living in households with male heads and 49.5 percent in those headed by females. Female-headed households constitute 53 percent of poor households and 56 percent of extremely poor households (Annex Table 2.9). 19 19. Poverty and extreme poverty in Swaziland are overwhelmingly a rural phenomenon. The incidence of poverty is 73 percent in rural areas and 31 percent in urban areas, down from 80 percent and 36 percent respectively in 2000/2001 (Figure 6). In 2010, extreme poverty was 29 percent, but only 6 percent of urban dwellers were extremely poor compared to 33 percent of those in rural areas. Eighty-eight percent of the poor and 95 percent of the extreme poor live in rural areas. Also, the rural poor are worse off than their urban counterparts. In 2010, average consumption among the urban poor was 33 percent below the poverty line, but it was 51 percent below the poverty line among the rural poor. Figure 6: Poverty in Swaziland by Location (Rural and Urban) Source: CSO 2011a and author calculations from SHIES 2009-2010 20. Poverty rates increase as the size of the household increases. The larger the household, the more likely it is that the household will be poor. Also, poverty incidence among 19 The 2009-2010 questionnaire had many more questions than the 2000/2001 questionnaire, so it is possible to draw a full profile of the poor and extreme poor from the 2010 data. CSO (2011) made comparisons between 2000 and 2010 regarding household assets and access to services such as water, sanitation, and electricity for which data were available for both years. The tables presented in Annex 2 are from SHIES 2009-2010. 10 households with six or more members is higher than the national average (Annex Table 2.11). 20,21 21. The population in Swaziland is very young, and in general, children and the elderly are at higher risk of being poor. Nearly half of the Swazi population is under the age of 18. Seventy percent of children aged 15 and younger are poor, as are 71 percent of people aged 60 and older. This is also the pattern for extreme poverty. 22 Households with children are poorer than households with elderly members, and households with both are even poorer. The least poor households have no children and no elderly members (Annex Table 2.10), and households with elderly members are poorer (76.7 percent) than households without elderly members (56.6 percent poor). This topic is discussed further in Section E of this chapter. 22. In Swaziland, as in most countries of the world, having more education reduces the likelihood of being poor. Households with illiterate heads are 50 percent more likely to be poor than those with literate heads, with a poverty rate of 85 percent. Households with illiterate heads are twice as likely to be extremely poor, with an extreme poverty rate of nearly 50 percent. Also, households headed by someone who completed high school or a public or private college or university are much less likely to be poor (Figure 7). Figure 7: Poverty Rates by Education of Household Head Source: Authors’ calculations from SHIES 2009-2010 23. Not surprisingly, the education levels of poor people are not as high as those of their wealthier counterparts. Poor Swazis rarely finish high school (only 9 percent of the poor), and hardly any poor people receive any education beyond the secondary level. More than one in five 20 The data (SHIES 2009-10) show that poverty increases monotonically until virtually all households with 12 or more children are poor. 21 It should be emphasized that only 4 percent of the Swazi population live in households consisting of only one member, while fewer than 0.5 percent live in households with 25 members. 22 According to the SHIES, about 485,000 Swazi citizens (about 48 percent of the population) are children aged 18 or under, while about 60,000 are elderly (about 6 percent of the population). Adults comprise about 46 percent of the population. 11 extremely poor people are illiterate, compared to just one in twenty among those above the poverty line. 23 24. Unemployment is high in Swaziland, particularly in rural areas. According to preliminary results from the Swaziland Labor Force Survey 2010-11, the total unemployment rate in 2010, as defined by the International Labour Organization (ILO), was 28.5 percent, with very little regional differentiation. 24 Unemployment was higher in rural areas (37.1 percent) than in urban (16.7 percent). The average unemployment rate for young people (15-24 years of age) was 52 percent compared to 12 percent of people aged 50 to 54, and female unemployment was higher than male unemployment (31.3 percent compared to 25.7 percent). 25 25. Unemployment is associated with poverty and extreme poverty status but not as strongly as in countries with more widespread formal employment than Swaziland.26 Household heads who reported they had worked for at least one hour in the previous 30 days had a lower poverty rate (49.4 percent) than household heads who had done no work (77.4 percent). They also had a lower extreme poverty rate (19.2 percent) than household heads who had done no work (39 percent extremely poor). These numbers should however be treated with caution, since the definition of employment makes a difference in interpreting these numbers as it is not clear whether survey respondents understood the question to mean any kind of work, even unpaid work or helping out neighbors in exchange for future food. Also, agricultural employment is highly seasonal, and some poor people may have been surveyed during the off- season. 26. Being productively employed protects against income poverty, but this is not a guarantee. The likelihood of being employed increases with wealth status. However, 36 percent of men and 29 percent of women in the poorest quintile are employed (Figure 8). Again, it is not clear if respondents understood employment in the ILO sense as they may have under-reported their economic activity. This is because only slightly more than half of all jobs provide earnings throughout the year. In the agricultural sector in particular, employment is often seasonal for both men and women, whereas the non-agricultural sector is more likely to provide year-round employment. Overall, men are less likely to be employed throughout the year (57 percent) than women (75 percent). 27 23 Literacy was defined as the proportion of those over the age of 15 who were able to read and write. 24 The preliminary report on the LFS 2010/11 uses two definitions of unemployment, the “relax[ed]” and the “strict” where the latter refers to the standard ILO definition. 25 CSO (2012) 26 It should be noted that it is not clear from the SHIES questionnaire what exactly respondents thought was meant by the question “did you work at least one hour in the past 30 days?” as this might have connoted “work for pay” as opposed to work on a household garden plot or “work for food” on a neighbor’s plot. 27 CSO (2008), p. 48 12 Figure 8: Percentage of Employed People Aged 15-49 by Gender and Wealth Quintile 70% 60% 50% 40% Men 30% Women 20% 10% 0% Lowest Second Third Fourth Highest Source: CSO (2008), Swaziland Demographic and Health Survey 27. People who live on land with usage rights given by the Chief are poorer than average. Nearly three-quarters of all Swazis live on Swazi Nation Land. This land is held in trust for the nation by the King and is administered by the chiefs. Most of the people who live on Swazi Nation Land farm small plots, cultivating maize, keeping cattle, and occasionally producing a cash crop. About 75 percent of people on Swazi National Land are poor compared with only 35 percent who have rented or purchased their land. Thirty-six percent of people living on Swazi Nation Land are extremely poor. 28. The majority of Swazi households (83 percent) live in houses with a roof made of corrugated iron, and these households have a poverty rate almost equal to the overall poverty rate of 63 percent. The main wealth-based difference in roofing material is that poorer Swazis without corrugated iron roofs tend to have roofs made of grass and mud brick roofs (15 percent of the poor live under such roofs, and their poverty rate is 92 percent). Wealthier Swazis are more likely to have roofs made of asbestos, tiles, slate, or concrete. Most Swazis (71 percent) live in houses with concrete or plaster walls, with a corresponding poverty rate of 53 percent. The 29 percent of Swazis who live in homes made of traditional material are much poorer ௅ 88 percent are poor compared to the overall poverty rate of 63 percent. This pattern persists with extreme poverty, where those whose houses have walls made of traditional material have an extreme poverty rate that is more than twice as high as those whose walls are made of concrete or plaster. 29. In Swaziland, nearly all urban households (90 percent) have access to safe water compared to only 60 percent of rural households (Figure 9). 28 Access to water has increased for all income groups during the past decade. On the other hand, there has been a drop in the standard of sanitation in Swaziland’s two most populous regions, Hhohho and Manzini, while sanitation conditions have significantly improved in the Shiselweni region where the proportion of households using flush toilets more than doubled over the past decade. Currently 37.7 percent 28 Safe water refers here to water piped into the building, piped water accessible from outside the building, or water from a borehole, protected well, or protected spring (CSO, 2011a, p. 22). 13 of the population has to rely on unsafe water sources, and four-fifths of these people (81.7 percent) are poor. Also, 16 percent rely on public standpipes, 72.8 percent of whom are poor. More than one-third of the population has piped water in their homes, but only 28.5 percent of these households are poor, versus the overall poverty rate of 63 percent. Figure 9: Percentage of Households with Access to Safe Water by Area and Region in 2009/10 100% 90% 80% 70% 60% Rural 50% Urban 40% 30% Total 20% 10% 0% Hhohho Manzini Shiselweni Lubombo Source: Authors’ calculations based on CSO (2011a), Tables A3.1, A3.2, and A3.3 30. A major issue for the poor and extreme poor who lack piped water is how long it takes to get to the water source. Only 6.3 percent of the SHIES sample responded to this question, but of these, 13.7 percent had to walk more than an hour to reach their water sources, and the poverty rate for these people was around 82 to 85 percent. It appears that poorer people were more likely to have responded to this question, since the average poverty rate of respondents was 77 percent, which is higher than the overall rate of 63 percent. 31. Sanitation is a problem in Swaziland, particularly for the poor. Approximately 14.7 percent of the population has no toilet in their home and has to use the bush or the fields. Of these, 93 percent are poor and 57 percent are extremely poor. About half of all Swazis have an ordinary pit toilet, and of these, 70 percent are poor (versus the overall poverty rate of 63 percent). Predictably, it is primarily the non-poor who have flush toilets (15 percent of the population have flush toilets and only 12.8 percent are poor), and the remainder have a ventilated or improved pit toilet (17.3 percent of the population) and have basically the same poverty rate as the national average. 32. Finally, access to electricity is highly related to households’ level of income and location as two-thirds of urban households have access to electricity compared to only one- third of rural households (Figure 10). 29 Paraffin and candles are the main light sources of the poor in Swaziland, with 20.2 percent of the poor using paraffin and 60 percent of the poor using candles. There are evidently some Swazi households that are too poor even for paraffin or candles and use only wood fires to light their houses. 29 CSO (2011a) 14 Figure 10: Percentage of Households Using Electricity by Area and Quintile 90% 80% 70% 60% 50% Urban 40% Rural 30% 20% 10% 0% Lowest Second Third Fourth Highest Total Source: CSO (2011a), Table A3.14 C. Food Insecurity and Vulnerability 33. In Swaziland, the most common adverse events are related to agricultural shocks such as reoccurring droughts, erratic rainfall patterns, prolonged dry spells, and high temperatures. Drought was cited by more than 80 percent of the SHIES respondents as the major shock that they experienced in 2008. The high costs of agricultural inputs and high food prices – both of which are closely related to drought – were cited as other shocks with which households had to find ways to cope to sustain their livelihood levels. 30 34. Swaziland is not self-sufficient in food production, but thanks to a recent slight increase in agricultural production, levels of food insecurity seem to have decreased. Smallholders occupy 75 percent of the land under cultivation, but their production accounts for only 10 percent of total agricultural output. The country relies on food imports and food aid to cover the needs of the population, and despite the slight decrease in food insecurity, 161,000 people were still estimated to be food-insecure in 2009/10 (Figure 11). It is also worth mentioning that food insecurity is regional; for example, the Lubombo region has twice as many food-insecure people as Manzini. A fluctuating supply of food led to rapid hikes in food prices (especially in 2008), which put pressure on household spending. 31 However, food prices declined sharply between 2008 and 2010. 32 In the 2009/10 season, the rains started on time and were adequate throughout the season, thus allowing the timely planting of crops. However, the dry spell in late December and early January had a negative effect on crops. The implications for the 2010/11 consumption period of this weather on early crops were expected to be minimal, whereas crops planted later were expected to be compromised due to dry spells, heat waves, and sporadic localized flooding. 33 30 VAC (2009) 31 VAC (2009) 32 VAC (2010) 33 Food Security Update (2010) 15 35. Both poverty and low agricultural productivity are widespread, with noticeable fluctuations at the district level. A visual comparison of the poverty map (CSO, 2011b) and the VAC surveys (of 2009 and 2010) suggest that low agricultural productivity is related to poverty, which may be empirically determined in future research. The average harvest of a poor household was 64 units of maize, whereas non-poor households produced 99.5 units. Extremely poor households produced less maize (52 units) than households that were not extremely poor (85.9 units). Households living on Swazi Nation Land (75 percent of whom are poor) are less productive than commercial producers who produce on title deed land. A worrying trend is that smallholder farm productivity and production have been declining over time. 34 Figure 11: Vulnerable Populations (Individuals) and Yield Production (Metric Tonnage) Source: VAC (2010), Figure 10, p. 13 36. Food insecurity adversely affects nutritional status and yields poor nutritional outcomes. The Multiple Indicator Cluster Survey report produced in 2010 by the Swazi Central Statistical Office in partnership with UNICEF 35 estimated that 6 percent of children under the age of 5 in Swaziland are underweight. Thirty-one percent of children are stunted or too short for their age, and 3 percent are wasted or too thin for their height (Table 2). Children in the Shiselweni region are more likely to be underweight and stunted than other children. Table 2: Malnutrition in Swaziland and Comparable Countries Stunting Wasting Underweight (moderate (moderate (moderate and and Countries and severe) severe) severe) Year Source Swaziland 6 31 1 2010 Multiple Indicator Cluster Survey Botswana 11 31 7 2007 Family Health Survey Ethiopia 33 51 12 2005 Demographic and Health Survey Lesotho 13 39 4 2009 Demographic and Health Survey Malawi 13 47 4 2010 Demographic and Health Survey Mozambique 18 44 4 2008 Multiple Indicator Cluster Survey Namibia 17 29 8 2006/07 Demographic and Health Survey South Africa 9 24 5 2008 National Income Dynamics Study Tanzania 16 42 5 2010 Demographic and Health Survey 34 World Bank (2011) 35 CSO and UNICEF (2011) 16 Zambia 15 45 5 2007 Demographic and Health Survey Zimbabwe 10 32 3 2010/11 Demographic and Health Survey Source: UNICEF, Child-Info, at http://www.childinfo.org/undernutrition_nutritional_status.php D. Shocks and Responses 37. Almost every household in the SHIES sample (95 percent) reported that they had experienced at least one of the shocks that were listed in the survey. The poor comprised 64 percent of those who reported experiencing a shock, while the extreme poor comprised 30 percent. The major shocks experienced by the poor and extreme poor are reported in Table 3. Table 3: Poverty and Shocks % of % of the Type of Shock Poverty % of the Populati Poverty Extreme % of Headcount Poor on Headcount Poor Population Poor Extreme Poor Food Price Increase 59.0 24.3 28.3 25.8 22.4 28.3 Drought/Flood 80.1 17.1 14.7 42.2 19.0 14.7 Ag Input Price 73.6 12.6 11.7 29.7 10.7 11.7 Increase Chronic/Severe Illness 65.7 6.9 7.2 31.3 6.9 7.2 Livestock Died/Stolen 77.7 11.8 10.5 39.3 12.6 10.5 Death of Household 67.9 3.0 3.0 31.8 2.9 3.0 Head Death of Working 76.8 2.7 2.4 44.5 2.0 2.4 Household Member Total 68.8 100.0 100.0 32.6 100.0 100.0 Source: Authors’ calculations from SHIES 2009-2010 Note: The poverty and extreme poverty rates pertain only to respondents. Respondents were allowed to name one, two, or three shocks, so poverty rates do not exactly match population total rates. 38. Twenty-eight percent of the population identified food price increases as one of the three major shocks faced by their household. 36 The SHIES 2009-2010 was conducted after food and fuel prices rose sharply in 2008, so this finding is not surprising. Interestingly, only 24 percent of the poor identified food price increases as one of the three major shocks, and of those who reported experiencing this kind of shock, only 59 percent were poor (as opposed to a total of 68.8 percent poor among all respondents). This suggests that food price increases were more of a problem for non-poor households. This makes sense in that poor households consume more from their own production than from purchased sources, although this varies by livelihood and region. Fifteen percent of respondents reported experiencing drought or flood, and 80 percent of those 36 Because respondents were allowed to name one, two, or three shocks, there is some double-counting in the reported population figures. The shares here are therefore proportions of the total number of responses rather than of the whole population of 1 million people. 17 who reported this as a major shock were poor people (17 percent). The third most frequently reported shock was agricultural input price increases, reported by about 12 percent of respondents and 13 percent of the poor. This was related to the rise in food and fuel prices in 2008, which caused widespread difficulties. The fourth shock was the idiosyncratic risk of chronic or severe illness in the household, with 7.2 percent of all respondents reporting having experienced the chronic or severe illness of a household member, but only 6.9 percent of the poor. 39. Respondents named help from family members as their top strategy for coping with drought/flood, chronic/severe illness, or the death of the head or a working member. This fact suggests that some kinds of shocks mobilize family support and that extended family members are willing to draw down their savings to help the afflicted household. The one coping strategy that was reported in the case of all seven shocks was “started to save cash” although the extent of this varied depending on the shock. Respondents resorted to reducing their food consumption only if they reported one or both of the two price shocks but markedly more in response to the food price shock than the agricultural input price shock. Fortunately, only 1.4 percent of the population reported having used this negative coping strategy, of whom 83 percent were poor. E. Vulnerable Groups 40. A number of specific groups are especially vulnerable, including children, orphans, and children made vulnerable by HIV/AIDS and the elderly. These groups are examined in this section. Children who are Poor 41. Children are a major correlate of poverty. Household poverty rates increase with each additional child under the age of 18. According to SHIES 2009/10, 67.1 percent of households with children under the age of 18 are poor compared to only 28.8 percent of households with no children under the age of 18 (Annex Table 2.11). Poverty is even more strongly correlated with the presence of young children in the household, reaching up to 87 percent of households that contain five or more children (Table 4). Children are over-represented in the bottom consumption deciles and under-represented in the upper deciles (Annex Table 3.1); there are proportionately more children who are poor than who are not poor. 18 Table 4: Poverty and Children Under the Age of 6 Number of Children Under 6 / Household Poverty Rate None 47.4 One 63.3 Two 76.0 Three 83.3 Four 85.7 Five+ 87.0 Source: Authors’ calculations from 2009-2010 SHIES 42. The nutritional status of children and their poverty status are correlated. Children from households that are poor are much more likely to be malnourished than their wealthier counterparts (Figure 12). 37 Improving the nutritional status of children should be an important safety net objective. Figure 12: Nutritional Status of Children by Wealth Quintile (percentage of children below 2 and 3 standard deviations) Percentage Weight for age: -2 SD Weight for age: -3 SD Height for age: -2 SD Height for age: -3 SD Weight for height -2 SD Weight for height: -3 SD Source: Swaziland Demographic and Health Survey (CSO 2008), Table 11.1., p.143. 43. Education and poverty status are also correlated, especially at the secondary level. Eighty-one percent of children in the poorest quintile (in contrast to 87 percent in the wealthiest quintile) are enrolled in primary school. Net secondary school enrollment rates are low for all wealth groups, partly as a result of children of secondary school age still attending primary school and partly as a result of dropouts. However, 56 percent of children in the wealthiest quintile are enrolled in secondary school as compared to fewer than 26 percent of children in the poorest quintile (Figure 13). 37 CSO (2008) 19 Figure 13: Net Enrollment Rates for Primary and Secondary Education by Wealth Status, 2009/10 % Enrolled Wealth Quintile Primary Secondary Source: CSO (2011a) from the SHIES 44. Cost was the primary reason for children not attending school, irrespective of the wealth status of the survey respondents. Fewer than 2 percent of respondents (in all poverty classifications) reported having to work as being the primary reason why their children did not attend school, and almost no respondents indicated that the distance of the school from the home was a primary obstacle. Orphans and Children Made Vulnerable by HIV/AIDS 45. Twenty-three percent of children under the age of 18 are orphaned (one or both parents are dead), and 4 percent are double orphans. According to the SHIES, the number of orphans in Swaziland in 2010 was 123,000, of whom 25,000 were double orphans. 38 The National Plan of Action for Orphans and Vulnerable Children (2006-2010) estimated that for every two orphans one other child is made vulnerable due to HIV/AIDS. The prevalence of orphanhood increases with age, with 7 percent of children under the age of 5 being orphans compared to 37 percent of 15 to 17 year olds. Many children do not live with their parents even if the parents are alive. According to the Swaziland Demographic and Health Survey (DHS) of 2008, 78 percent of children under the age of 18 were not living with both parents, and 34 percent were not living with either parent. This phenomenon is more common in rural areas than in urban areas (36 percent versus 24 percent.) 46. Poverty and extreme poverty rates for (single) orphans are slightly higher than for other children. The poverty rate for single orphans (one parent dead) is 75 percent and 65 percent for double orphans. Maternal orphans seem to be slightly poorer (80 percent) than paternal orphans (76.3 percent) as can be seen in Figure 14. A similar pattern is true for extreme poverty, with 36 percent of single orphans and 26 percent of double orphans being extremely poor compared to the average of 34.1 percent for all children (Annex Table 3.2). The lower 38 Using the SHIES population basis of 1,017,406 people, we estimate that there are 26,874 maternal orphans (2.64 percent of the population), 71,251 paternal orphans (7.03 percent of the population), and 24,524 double orphans (2.41 percent of the population). Our estimates are based on population sampling weights. 20 poverty rates for double orphans than for single orphans suggest that double orphans are living in slightly better-off households, as was found by Beegle et al (2010) to be the case in other Sub- Saharan African countries. Figure 14: Poverty Rate and Orphan Status Source: Authors’ calculations from SHIES 2009-2010 47. Orphans and vulnerable children (OVCs) are mostly cared for by their extended families and are often depend on community assistance to survive, but this assistance is not readily available to the majority of OVCs. According to the 2008 DHS, only 41 percent of OVCs received any external assistance (Figure 15). The most common form of support was school assistance, which was received by one-third of OVCs, while considerably fewer received other material assistance (8 percent) or emotional support (5 percent). Figure 15: Percentage of OVCs Whose Households Received Assistance by Wealth Quintile 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Lowest Second Third Fourth Highest TOTAL Medical support Emotional support Material support School assistance At least one type of support Source: Swaziland Demographic and Health Survey, CSO (2008), Table 17.11, p.273 21 48. A particularly vulnerable group among OVCs is child-headed households (CHH), but this group accounts for less than 1 percent of all Swazi households. A recent study by Save the Children and UNICEF investigated the situation of child-headed households in drought- prone areas of Swaziland and provides insights into the situation of these children (see Box 3). Box 3: Child-Headed Households Children in child-headed households (CHH) are typically orphans, but this is not true for all CHH. Seventy-five percent of CHH are double orphans, but 35 percent of child-headed households have at least one parent alive who has abandoned the children. The most common reason the children gave for why mothers left them was her remarriage while more than half of the children did not know the reason why their father had left. Almost 90 percent of the CHH had an extended family, but “poor relationship” was given as the main reason why the children were not living with their extended family. In addition, more than 20 percent of respondents suggested that fear of losing their property was another important reason for not living with relatives. More than 20 percent of CHH had experienced loss of property since their parents had passed away, with most having been taken by the community and neighbors or paternal relatives. CHH live in deprived situations. They consume a carbohydrate-intense diet that lacks nutrients like fats and proteins that are essential to development. More than 55 percent of CHHs lack access to safe water, 37 percent lack access to proper sanitation, and only 6 percent have electricity in their homes. Ninety percent of CHHs had received at least one form of external support in the previous six months. The government was the largest source of external support (69 percent), followed by NGOs at 41 percent, and friends and neighbors at 36 percent. Source: Imai et al (2009) People with HIV/AIDS 49. Swaziland has the highest HIV prevalence in the world with 26 percent of all adults between the ages of 15 and 49 being infected. Infection is more prevalent among women (31 percent) than men (20 percent). An estimated 7,000 Swazis die each year from AIDS-related illnesses, and over half of all hospital admissions are for people living with HIV/AIDS. 39 HIV prevalence is higher in urban areas (31 percent) than in rural areas (24 percent) for people aged 15 to 49. It is not directly related to income, as the infection rates are largely the same for all income quintiles (varying from 26.4 percent in the lowest to 25.1 percent in the highest quintile). 40 However, HIV interacts with other aspects of poverty as the poor are often diagnosed late, are less likely to get treatment for sexually transmitted diseases, and are less likely to use condoms. 41 HIV/AIDS has resulted in large increases in the numbers of single-parent families, grandparent-headed families, and child-headed families. The cost of providing treatment for HIV/AIDS patients drains the country’s already scarce resources. The economic consequences of the pandemic, including declines in productivity and output and the loss of many skilled workers, have also been substantial. 39 VAC (2009) 40 CSO (2008) 41 VAC (2008) 22 50. The social and economic impact of the pandemic will continue to grow, albeit at a diminishing rate. Progress in the fight against HIV/AIDS is reflected in the reduction in the number of new infections, which declined from 20,000 in 2000 to 12,281 in 2009 with a further reduction to 11,381 by 2015 being projected. However, the number of Swazis living with HIV in 2010 was 191,147, and this is projected to increase to 214,210 by 2015 (Figure 16). AIDS- related deaths are projected to increase from 7,114 to 8,389 during the same period. 42 Access to antiretroviral therapy (ART) has increased dramatically and the number of people accessing care is projected to grow (Box 4). Figure 16: The HIV+ Population (Estimated) Total Adults Children Source: Clinton Health Access Initiative based on Swaziland HIV Estimates and Projections, June 2010 Box 4: Access to Antiretroviral Therapy The provision of antiretroviral therapy (ART) through the public sector began in December 2003. The number of people receiving ART rose from about 6,000 people at the end of 2004 (out of an estimated 45,000 in need of treatment) to 72,000 (out of an estimated 185,800 who needed treatment). By 2015, an estimated 110,605 Swazis will be in need of treatment out of an infected population of 214,210 (including both children and adults). The Government of Swaziland envisages that the coverage of ART will reach 85 percent of adults in need and 90 percent of children in need by 2014. Sources: NERCHA (private communication), World Bank (2010), and Clinton Health Access Initiative, Swaziland 51. The vulnerability of women to HIV is exacerbated by customs that leave women unable to protect themselves. Under customary law, when women get married, their legal status becomes comparable to that of a minor, and there are severe limitations on the freedom of women to own property, access finance, and make life choices. As polygamy is still common and protection against sexual assault and physical violence against women is weak, many women are unable to protect themselves against infection. 43 Poverty and food insecurity has also driven many poor women to exchange sex for food or money, which increases their risk of becoming infected with HIV. 44 42 UNAIDS online database: http://www.aidsinfoonline.org/ and NERCHA (2010) 43 NERCHA (2009a) 44 Weiser et al (2007) 23 The Elderly 52. The elderly (people aged 60 years and over) comprise only 6 percent of the Swazi population but are more likely to be poor than households with no elderly members. Households with one elderly member are more likely to be poor (77 percent) than the average for the population (63 percent), but only 66 percent of households with two elderly people are poor. About 32 percent of the population lives in households with one or more elderly members. Households with elderly members are more prevalent in rural areas (87 percent), which is also where the majority of the poor reside. Of the poor elderly, 95 percent live in rural areas. Of the non-poor elderly, 68 percent live in rural areas. 53. The elderly are socially vulnerable, particularly because of the loss of their children and extended families to death or to urban and foreign migration. As a result, the social isolation of the elderly is increasing, and elderly people commonly report that they are lonely. A study of the elderly in 2003 found that about 46 percent of them received some form of care and support from either family or neighbors but that 85 percent were responsible for the care of others. About 11 percent of the elderly were employed and 34 percent were active in subsistence farming. Over three-quarters of the elderly reported that their health was poor. Abuse of the elderly by their children was also reported. 45 F. Conclusions 54. Despite a decline in overall poverty rates in Swaziland, extreme poverty remains the same, growth has not been pro-poor, and poverty remains significantly higher in rural areas than in urban areas. Sixty-three percent of the population remains poor and 29 percent remains extremely poor. In addition, the recent modest economic growth has not benefitted the poorest people in the population. 55. There are only two major correlates of poverty ௅ rural location and the presence of dependents (not necessarily orphans) in the household. Rural Swazis are significantly more likely to be poor than their urban counterparts, and poverty is deeper in rural areas than in urban areas. While there are differences between the four regions, most of the differences in poverty are evident between the 195 districts in Swaziland. 56. Households with children and elderly members are poorer than the average household, and nearly half of the population is under the age of 18. Households with no elderly members or children constitute only 9 percent of the population, but they are much less poor (only 25 percent) than households that include both elderly members and children (78 percent poor). The number of children under the age of 18 in the household is strongly correlated with poverty as poverty rates increase from 45 percent for one-child households to 90 to 100 percent for households with eight or more children. 45 Umchumanisi Link Action Research Network (2003) Social Protection of the Elderly in Swaziland , Coordinating Assembly of NGOs (CANGO), Mbabane. 24 57. Children and the elderly are more likely than working age adults to be poor. Poverty is lowest for adults aged 45 to 49 (54 percent), while poverty rates for children under the age of 5 and the elderly aged 60 and above are 71 percent and 70 percent respectively. Although only about one-tenth of the population live in households with a head aged 35 to 39, only about one- half of such households are poor. Almost one-third of people live in households with heads aged 60 and above, and people in these households are poorer (76 to 77 percent) than the overall poverty rate of 63 percent. 58. Poverty is associated with a lack of education in Swaziland. Illiterate household heads are 50 percent more likely to be poor than literate heads, and the poverty rate falls from 84 percent of households whose heads who have less than a primary education to 7 percent for households whose heads have a public college education. 59. Housing characteristics are somewhat correlated with poverty. Swazis living on land with usage rights granted by their Chiefs are poorer than other Swazis. Extremely poor Swazis are more likely to live in homes made of traditional materials with grass and mud brick roofs and to lack access to safe water, toilets, and electricity. However, housing correlates like a lack of access to electricity, water, and sanitation primarily reflect their location in rural areas. 60. The most common shocks reported by poor households were food price increases and drought/flood. Over 20 percent of the poor and of the extreme poor reported food price increases as being the major shock that they had experienced in the previous year, while drought/flood was reported by just under 20 percent of both the poor and the extreme poor. 25 CHAPTER 3: THE SOCIAL SAFETY NET IN SWAZILAND 61. This chapter reviews the existing safety net programs in Swaziland, including their expenditures and implementation constraints. This report focuses on social transfers, including cash and in-kind transfers. Table 5 lists the main social transfer programs that are discussed in this chapter. Thereafter, public expenditures on the social safety net, particularly social transfers, are analyzed as well as the impact of the fiscal crisis on the safety net. The chapter concludes with an examination of several administrative and implementation issues that apply to all safety net programs. A. The Organization of the Social Safety Net 62. The safety net system consists of multiple cash and in-kind transfers, social care services, active labor market programs, and community-based programs implemented by at least five ministries or departments within ministries. Agencies delivering safety net services include the Department of Social Welfare (DSW) within the Deputy Prime Minister’s Office (DPM), the National Emergency Response Council on HIV and AIDS (NERCHA), the Ministry of Education and Training (MOET), the Ministry of Health (MOH), and the National Disaster Management Agency (NDMA) in the DPM’s Office. The DSW, which was formerly in the Ministry of Health but has been housed in the DPM’s Office since 2009, is responsible for Swaziland’s largest cash transfer programs and is also responsible for overseeing social care services. The DPM also has portfolio responsibility for the NDMA, the agency responsible for government-funded food distribution. NERCHA is responsible for the distribution of donor- funded food to schools for the school feeding program and to communities for the Neighborhood Care Points (NCPs), for social care services in the NCPs, and for Young Heroes, a small cash grant program. MOET implements the School Feeding Program with assistance from NERCHA and also has the responsibility for overseeing non-formal labor market programs. The MOH is responsible for administering fee waivers for health care and the Phalala Fund to cover any medical care not available in Swaziland and also implements a food by prescription program for malnourished beneficiaries of ART, TB, and prevention of mother to child transmission programs. Other agencies, such as the Micro-Projects Program and the Ministry of Labor, implement other safety net programs. The DPM has some responsibility for coordinating the safety net but lacks a broad mandate for sector-wide coordination. 26 Table 5: Social Transfer Programs in Swaziland Number of Expenditure Expenditure Implementing Targeting Beneficiaries Source(s) of EM million US$ million Benefit Provided Program Agency Target Group Mechanism 2011 Funding (FY2010/11) (FY2010/11) (E) CASH TRANSFERS Policy: Means- Old Age Grant DPM/DSW Elderly 60+ tested. In practice: 55,692 GOS 100.3 14.1 200/month Categorical Disabled poor, 4.6 0.6 Public Assistance DPM/DSW Means-tested 3,075 (disabled only) GOS 80/month indigent WWII veterans or Veterans (46) 2.9 0.4 600/month to veteran; Military Pensions DPM/DSW Categorical GOS their widows Widows (464) 300/month to widow Grants to Persons with Categorical and 0.8 0.1 Variable benefit for DPM/DSW Not available GOS Handicapped disabilities means-tested occupational equipment Grants to Fire 0.1 0.01 Variable benefit to replace DPM/DSW Fire victims Categorical Not available GOS Victims personal/household goods Categorical and NERCHA; 4.0 0.6 Young Heroes NERCHA Orphans 887 180/month community donations IN-KIND TRANSFERS Primary = 560/year; Form I-III = 1,950/year; Form OVC Education Orphans and Categorical and GOS; 239.8 33.8 DPM/DSW 118,219 IV = 2,500/year; Grant vulnerable children community EU until 3/12 Form V = 1,950/year + up to 2,000 for exam fees GOS for School Feeding Primary and high Primary = 241,637; School meal of maize meal, MOET Universal primary; GF 28.0 3.9 Program school students High school = 23,039 pulses, oil for High Neighborhood Care Categorical and 54, 850 children in GF, WFP, Food, psychosocial support, NERCHA Orphans 15.1 2.1 Points (NCPs) self-targeted 1,554 NCPs World Vision ECD Food Distribution/ Food insecure Geographic and 67,000 (Nov-Dec,2011 17.0 2.4 DPM /NDMA GOS Maize meal, pulses, oil Food for Work Pilot households self-targeted 53,000 (Jan-Mar, 2012) Categorical and Free medical care at public Health Fee Waivers MOH Elderly and indigent Not available GOS Not available Not available means-tested health facilities Malnourished beneficiaries of ART, 8,664 patients + 6 Soya blend, maize meal, Food by Prescription MOH Categorical WFP 3.5 0.5 TB, and PMTCT household members pulses, oil programs Patients requiring care Phalala Fund MOH Medical referral 450 GOS 60.0 8.5 Medical care in South Africa not available in SZ MOA/ Community and 20 households; 101 Agricultural Inputs Orphans GF 5.1 0.7 Agricultural inputs NERCHA self-targeted chiefdoms 27 B. Implementation Constraints within the Existing Safety Net 99. This section reviews the main social transfers in Swaziland. The review presents operating and financial statistics and analyzes key implementation issues and constraints. An overview of smaller social transfer and other safety net programs is also included. Cash Transfer Programs Old Age Grants 100. The Old Age Grant (OAG) was introduced by the government in 2005 with the objective of protecting the elderly. As noted by King Mswati III in his 2005 Throne Speech that launched the Old Age Grant, the need for protection largely arose as a consequence of HIV/AIDS. Elderly people were losing their younger income-earning family members to the disease and had to shoulder the burden of providing and caring for their grandchildren. The OAG is administered by the DSW and is financed by the government. The direct cost of the grant is estimated to have been E100.3 million (US$14.1 million) in FY2010/2011. OAG transfers account for almost 90 percent of all cash transfers to households. Between 2005 and 2010/11, the number of beneficiaries increased by 27 percent from about 43,900 to almost 55,700 in 2011. 101. In theory, OAG beneficiaries are not eligible if they receive an employment pension of more than E1,000 per month, but this rule has not been enforced so the OAG essentially operates as a universal social pension. All elderly citizens aged 60 and over are eligible as long as they have a national identification card. The number of beneficiaries increased from 43,830 in 2005 to 55,692 in 2011/12. About 91 percent of the population aged 60 years and over receive the OAG. 46 The current value of the Old Age Grant is E200 (US$26) per month or 43 percent of the adult equivalent poverty line and 93 percent of the adult equivalent food poverty line. 102. Qualitative evidence suggests that the OAG has had a positive impact on beneficiaries` lives by increasing their food security as a result of more frequent and better meals and the ability to purchase more food and to buy it in bulk. In addition to the direct beneficiaries of the grant, other household members also benefitted significantly, particularly with regard to their nutritional status, education, and health care. Nearly 70 percent of OAG recipients spent more on groceries than before receiving benefits and 63 percent spent more on protein, while 71 percent reported that they were “more able to afford health care.” Seventy- four percent of children in beneficiary households were enrolled in school, and the OAG was used to pay their fees. 47 46 Based on the authors’ population estimates using data from the 2010 SHIES. 47 Regional Hunger and Vulnerability Program/HelpAge International/UNICEF/(2010) 28 103. The age of eligibility for the OAG is lower than that in similar programs in many other countries. The minimum age to be eligible for the OAG is 60, which is younger than in the 20 countries studied in a recent report. 48 Moreover, only one-third of these countries offered a universal pension (Table 6). Table 6: Social Pensions in Low- and Middle-income Countries Universal (U) or Country Eligibility Age Means Tested (M) MIDDLE-INCOME Swaziland a/ 60 U Argentina 70 M Bolivia 65 U Botswana 65 U Brazil 67 M Brazil Rural 60 men; 55 women M Chile 65 M Costa Rica 65 M Mauritius 60 U Namibia 60 M South Africa 65 men; 60 women M Thailand 60 M Uruguay 70 M LOW-INCOME Lesotho b/ 70 U Moldova 62 men; 57 women M Nepal 75 U Tajikistan 63 men; 58 women M Vietnam 60 M Vietnam 90 U Source: Swaziland, authors’ calculations, all others HelpAge International (2009) Notes: a/ Universal due to fact that pensions exclusions not applied. b/ Universal with a few exceptions, primarily people who are already receiving a substantial government pension (about 4 percent of those who would otherwise be eligible) 104. The OAG is not sufficiently pro-poor – although it does not contribute much to the welfare of non-poor households, a significant share of OAG resources goes to these households. About 28 percent of OAG beneficiaries are not poor and over 11 percent of beneficiaries are from the two wealthiest deciles (Table 7). As can be seen in Table 7, the OAG is somewhat progressive, with a smaller share of benefits going to households in the wealthier deciles than to households in the poorer deciles. However, on average, the OAG accounts for 55 percent of the consumption of households in the poorest decile but only 1 percent of the consumption of households in the wealthiest decile (Table 8). We used SHIES data to estimate poverty rates if the OAG did not exist and found that the poverty rate would increase only slightly from 63 percent to 64 percent and that the extreme poverty rate would increase from 29 percent to 32 percent. 48 HelpAge International (2009) 29 Table 7: Distribution of OAG by Wealth Decile, 2009/10 Decile Decile Decile Decile Decile Decile Decile Decile Decile Decile 1 2 3 4 5 6 7 8 9 10 Total 11.0 13.2 11.5 12.3 12.5 11.5 9.2 7.7 6.8 4.4 100.0 Source: Authors’ calculations using SHIES 2009-2010 Table 8: Share of OAG as a Percentage of Consumption, 2009/10 Decile Decile Decile Decile Decile Decile Decile Decile Decile Decile 1 2 3 4 5 6 7 8 9 10 Average Monthly 88 149 197 249 305 380 487 644 935 2,351 Consumption OAG as a % of Average 55 32 22 18 16 13 7 4 2 1 Monthly Consumption Source: Authors’ calculations using SHIES 2009-2010. 105. The OAG is intended to help grandparents to shoulder the burden of caring for orphans. However, the logic of providing grants to all elderly people with the intention of helping the elderly who care for orphans is questionable. Eighty-three percent of the elderly do not live with an orphan, and about 55 percent of orphans do not live with an elderly person. 49 Moreover, the OAG discriminates against any elderly people caring for many orphans since the benefit is not differentiated between an elderly person living alone and one caring for several dependents. Among households that contain orphans, the average number of orphans is two, and about 7 percent of households caring for orphans consisted of six or more orphans. 50 106. The OAG is supposed to be disbursed quarterly, but in reality it is subject to the availability of funds and recipients are not always sure when they will actually receive the benefit. Late disbursements make it difficult for recipients to plan, especially at crucial times in the year when they could use some of the cash for productive purposes, for example, at the start of the planting season in rural areas. In fact, social workers have identified cutting back on planting as the main strategy used by the poor to cope with the recent series of late payments of the OAG grant. In addition, the value of the benefit has only rarely been increased. Although the benefits are higher in real terms than when the program began in 2005, they have not been increased since 2009. 107. The resource requirements for the OAG program have increased since the program’s inception and will continue to increase due to the increasing number of elderly Swazis. The over-60 age cohort is expected to grow by almost 30 percent (or about 18,000 people) by 2020 and will continue to grow after that. This will put increasing pressure on already scarce resources. During the same period, the working age population (who will need to finance the OAG) is projected to increase by only 18 percent. The grant is disbursed quarterly to 49 Authors’ calculations from SHIES 2009-2010 50 Regional Hunger and Vulnerability Program/HelpAge International/UNICEF (2010) 30 those beneficiaries who are paid in cash and monthly to those beneficiaries who receive their grant via direct deposit into their bank accounts. The quarterly cash payments, which account for over 80 percent of payments, are disbursed through local government (Tinkhundla) offices and social welfare offices by DSW staffs, who are accompanied by police for security reasons. Other DSW Grants 108. Public Assistance, which has been in operation for over 30 years, was one of Swaziland’s earliest efforts to establish a formal social safety net. Public Assistance targets people under the age of 60 who are both disabled and destitute who are not beneficiaries of any other grant and have no other source of income. The grant is administered by the DSW. The government finances the program, with the direct costs of the grants estimated to have been E4.6 million (US$0.6 million) in FY2010/2011. Candidates apply for assistance at one of the four regional social welfare offices. Their eligibility is established by a means test plus a determination of disability. The means test includes a home visit by a social welfare officer to assess the applicant’s need, income, and health. There were 5,075 beneficiaries in December 2011, equivalent to about 15 percent of the severely disabled 51 and a mere 0.5 percent of the poor. The value of the Public Assistance Grant is E80 (US$10) per month or 17 percent of the adult equivalent poverty line and 37 percent of the food poverty line. The grant is paid in cash and is disbursed quarterly at the same time that OAG benefits are paid. Beneficiaries can benefit for an indefinite period as there is no clear exit strategy and, thus, there has been little change in caseloads over time. 109. Military Pensions are non-contributory entitlements awarded to World War II veterans or their widows. The program is administered by the DSW and provides E600 (US$78) to veterans and E300 (US$39) to widows. The government finances the program, with the direct costs of the grants estimated to have been E2.9 million (US$0.4 million) in FY2010/2011. There were 46 veterans and 464 widows collecting this benefit as of December 2011. Benefits are paid monthly via checks collected at the regional offices of the DSW. In cases where a deceased veteran had more than one wife, the senior wife receives the benefit. When she dies, the next senior wife becomes beneficiary. Rights to the benefit pass from widow to widow until all of the widows have died. Beneficiaries may also collect an Old Age Grant and/or earn income. The DSW reports that the number of beneficiaries has declined and predicts that, based on the age of remaining beneficiaries, the program will expire in approximately 10 years. 110. The DSW provides small Grants for People with Disabilities to be used to purchase either prostheses or equipment to promote self-employment. Eligibility for these benefits is subject to DSW discretion, and there are no established criteria. The DSW also used to operate a grant program for Fire Victims, but this program has since been suspended. Finally, the DSW used to pay Child Welfare Grants to households that were fostering children, but this program was suspended upon the introduction of the OVC Education Grant. Estimated expenditures on these programs in FY2010/2011 were around E1.0 million (US$0.1 million). 51 Estimated at 2 percent of the population. 31 111. The coverage of these small DSW grant programs is very limited and there are large differences between programs in the size of the benefits provided. Even when it was operating, the indigent category of Public Assistance served fewer than 1,000 beneficiaries, which is not even 1 percent of the extreme poor. 112. The DSW cash transfer programs provide benefits that are broadly comparable with those in other countries. In a study of cash transfer programs in middle-income countries, the median value of social pension benefits as a share of the consumption of recipient households was 20 percent. 52 The OAG accounts for 17 percent of the consumption of all recipient households and 26 percent of consumption of poor recipient households. The benefit provided to disabled Public Assistance beneficiaries represents 37 percent of the extreme poverty line and is comparable with targeted programs internationally (Table 9). A study of cash transfer programs in middle-income countries in Europe, Central Asia, and Latin America found that the median value of benefits as a share of the consumption of recipient households 53 is 13 percent for family allowances and 15 percent for means-tested programs, 54 while the median value of benefits for seven conditional cash transfer programs was 13 percent. 55 In addition, the value of the benefit is uniform across households and does not take household size into account. Flat rates are not equitable in that the value of the benefit is relatively smaller for larger households than for smaller households. Table 9: Summary of Department of Social Welfare Cash Transfer Programs Benefit/ % Month of Poverty % (E) Line of Extreme Poverty Line Old Age Grant 200 43 93 Public Assistance 80 17 37 Military Grant (Veteran) 600 130 279 Military Grant (Widow) 300 65 140 Source: Authors’ compilation 113. At the same time, benefits vary widely between programs, and benefit increases are awarded in an ad hoc manner. There is no systematic mechanism that triggers benefit increases and price inflation has eroded the value of cash transfers. The OAG was last increased in 2009, and the Public Assistance benefit has not been increased in more than a decade. 114. It is also worth mentioning that the government is in the process of designing a pilot cash transfer program for OVCs. This pilot will be implemented by the Deputy Prime Minister’s Office and is co-financed by the World Bank and the EU. The pilot is expected to reach roughly 8,000 households that contain OVCs. 56 The design features of the pilot, including 52 Grosh et al (2009) 53 This is a slightly different measure than the percentage of the poverty line reported above, but data on the consumption of recipient households was not available. Nevertheless, the poverty and extreme poverty lines provide a proxy for consumption among the poor and indigent. 54 Grosh et al (2009) 55 Fiszbein and Schady (2009) 56 The number of households that the pilot will be able to reach will depend on several factors, including the level of benefit and the pace of the expansion of the pilot. 32 the definition of OVCs and the level of benefit, will be developed during 2012, and payments will not commence until the design features have been finalized, the systems have been put in place (including an MIS and a payment system), and the staff have been trained in targeting and enrollment. Young Heroes 115. Young Heroes is a cash grant program targeted to orphans and administered by a secretariat within NERCHA. A Board of Directors guides and oversees the work of the secretariat. In FY2010/11, total program costs were E4.0 million (US$0.6 million), with 58 percent of expenditures covered by NERCHA and 41 percent by donations from local and international donors. 116. Community members are responsible for identifying beneficiaries. Young Heroes staff visit applicants in their homes to verify their need, taking into account their living conditions, the numbers of elderly people and working adults in the home, and any sources of income. As of December 2011, the program had 887 beneficiaries in 497 households. The grants are E180 (US$23) per month or 39 percent of the adult equivalent poverty line and 84 percent of the adult equivalent food poverty line. Benefits are paid semi-annually through Post Offices. Monitoring officers visit schools once per year to verify that the children are enrolled and to ensure that they are not getting assistance from other sources. In addition to providing cash grants, Young Heroes pays the school fees of some of its beneficiaries and also runs HIV testing and care programs for the children and week-long camps for HIV-positive children twice per year. 117. The Young Heroes Program faces a number of implementation challenges, including significant diseconomies of scale and the high costs involved in administering donor funds. The program is so small that it cannot realize any economies of scale. In addition, the reporting requirements related to raising contributions from donors in the United States are significant. Except for some effort to exclude the beneficiaries of the OVC Education Grants from receiving education assistance provided by Young Heroes, Young Heroes operates in virtual isolation from DSW programs. Monitoring enrollment once per year does not necessarily mean that a child will attend school regularly. The program has not been evaluated and there is no information on whether Young Heroes reaches the poorest orphans and with what impact. In-kind Transfers OVC Education Grant Scheme 118. The Education Grant Scheme for OVCs (also known as the Bursary Scheme or the Capitation Grant Scheme) is targeted to orphans and vulnerable children to enable them to access education. It was introduced by the government with support from UNICEF in 2003. The grant is administered by the DSW. The Education Grant Scheme complements other government funding to schools and is intended to offset the fees charged to students. Bursaries are paid in the name of an individual child, with the beneficiaries selected by a committee comprised of school and community representatives. The government finances orphans in 33 primary schools and all students in high schools, while the EU finances vulnerable children in primary schools. A small number of students are supported by other organizations including Young Heroes, the Manzini Youth Center, and World Vision (Table 10). The EU contribution was estimated to have been E24.0 million (US$3.4 million) in FY2010/2011 with government expenditures on the program estimated at E215.8 million (US$30.8 million), including spending under the DSW OVC Education Grant and MOET (Primary and Secondary Bursaries) line items. 57 Table 10: Coverage of the OVC Education Grant Scheme, 2010 Number of Beneficiaries Government of Swaziland Primary School Orphans 42,057 High School Vulnerable 16,570 High School Orphans 28,095 European Union Primary School Vulnerable 29,484 Other Partners 2,013 Total 118,219 Source: Department of Social Welfare 119. The value of the bursaries ranges from E560 per year for primary students to E2,500 per year for Form IV students. The bursary covers approximately 65 percent of average primary school fees and somewhat less than 30 percent of average secondary school fees. 58 120. Although the cost of education is the main reason why Swazi children do not attend school, the government’s education grants reach only a small share of poor children while benefitting a significant share of the non-poor. Education assistance programs, of which the OVC Education Grants and EU Capitation Grants account for about 98 percent, reach only about 40 percent of school-age children in the poorest quintile and about 36 percent in the second and third quintiles. Yet at the same time, about 18 percent of students in the wealthiest quintile receive education support (Figure 17). 57 It should however be noted that the EU will soon be changing how it supports education in Swaziland. It will be funding a fully supply-side intervention in which it will give grants to the MOET to help schools to improve the quality of primary education. 58 Primary school fees range between E800 and E1,000. Secondary school fees range between E6,000 and E10,000. 34 Figure 17: Coverage of Student Assistance by Wealth Quintile Source: CSO (2008) 121. The OVC Education Grant covers only part of the official school fees and covers none of the other out-of-pocket costs of education, such as transportation, uniforms, and books for secondary students. Yet these out-of-pocket costs often make attending school unaffordable for poor students. It is important to note that the government has introduced initiatives that have reduced other costs that can prevent poor children from accessing primary education, including the introduction of free primary education (FPE) (which is being phased in gradually and will reach universal coverage by 2015) and free books and stationery. However, secondary school students have not benefitted from any of these initiatives. 122. The OVC program faces significant operational challenges, including the difficulty of keeping track of individual students and inadequate monitoring of the program. The lack of identification numbers for students has led to double counting of students and the presence of “ghost students” (students who are registered and paid for but who do not exist in reality). In addition, some applicants are not being notified that they have been awarded the OVC Grant, and this has led to schools receiving the fees from their parents as well as from the OVC Grant program. In addition, a number of beneficiaries receive grants from both the government and other organizations. The Auditor General has reported that, in combination, approximately 8 percent of all beneficiary students fell into one of these categories of abuse in FY2009/10, and this resulted in a loss of E6.7 million (US$0.9 million). 59 Ongoing efforts to issue identification numbers to all children and to upgrade the management information system for the program should address some of these concerns. 123. There is no policy on whether beneficiary students should pay any fees not covered by the grant, and the absence of such a policy has resulted in students from equally poor households being treated differently. Some schools exempt bursary students from paying the additional fees while others require them to pay and deny the students who do not pay either access to school or to the results of their examinations. This problem is compounded by the fact that there is also no policy on school fees so that different schools charge widely different fees. In other words, the value of the grant is different for different students. Some schools have introduced a cap on OVC numbers, which has resulted in the rationing of places. 59 Auditor General (2011) 35 124. At the school level, financial accounting for the OVC grants received by each school is weak. This problem is compounded by the fact that MOET regulations require schools to have only one bank account, and funds from all sources are deposited into one account, making it difficult to track the use of OVC grants in schools. In addition, schools keep all OVC grants received from the government, whether or not the beneficiary children actually attend school. 125. The Ministry of Education and Training began phasing in free primary education (FPE) in 2010 with the aim of extending it gradually to the whole primary school cycle by 2015. However, FPE will not eliminate all of the barriers to access faced by poor children. Even with FPE, primary students will continue to have to pay considerable out-of-pocket costs for education, while secondary students are not covered by the program at all. 126. The sustainability of the OVC Education Grant program is problematic. The EU, which has supported the primary school bursary program, will be changing how it supports education in Swaziland to a fully supply-side intervention in which it will give grants to the MOET to help schools to improve the quality of primary education. This means that the government will now be required to support students who were previously covered under the EU-funded part of the program (Capitation Grants) but who are not in a grade covered by free primary education. School Feeding Program 127. The government introduced the School Feeding Program (SFP) in 1982 in partnership with the World Food Program and with early support from UNICEF. The objectives of the program are to increase school enrollment, attendance, and retention, to increase learning, to improve children’s nutritional status, and to strengthen the capacity of local communities to support the program. The SFP is currently implemented by the School Health and Nutrition Department within MOET with technical support from the World Food Programme (WFP). MOET is responsible for the procurement of commodities for primary schools, which NERCHA then delivers. NERCHA is responsible for the procurement and distribution of commodities (financed by the Global Fund) to secondary schools. Food is delivered to schools three times per year. Save the Children assists with the distribution of food and the training of cooks. The government’s and the Global Fund’s provisions are supplemented by food from community donations or school gardens. 128. The SFP is financed from multiple sources, and it is impossible to determine the actual cost of the program. MOET assumed responsibility for financing the primary school component of the program in 2009 and estimates its contribution to the program to have been E18.8 million (US$2.7 million) in FY2010/11. Contributions from the Global Fund, which supports the secondary school component, are estimated to have been E9.2 million (US$1.3 million) in FY 2010/11. The WFP provided food for the SFP in FY 2010/11, but neither NERCHA nor the WFP accounts for SFP donations separately so we do not know the value of these WFP inputs. Students also finance the program through their annual school meal fee of E150 to E200, which is primarily used to pay the salaries of cooks, but the Ministry of Education does not collect data on the total value of these fees collected from students. Students 36 in receipt of OVC Education Grants are exempt from paying the school meal fee as the grant is expected to cover this fee. Under FPE, 60 MOET pays schools a per capita grant inclusive of the SFP fee. This means that the government’s financing of the program is both direct (via procurement of food) and indirect via OVC Education Grants and FPE Grants to schools. 129. The SFP provides a cooked lunch to primary and secondary students. In 2011/12, 790 schools (95 percent of all schools) participated, and meals were served to an estimated 264,676 students (Table 11). The meal includes cereals, pulses, and vegetable oil. In caloric terms, it represents about 30 percent of a child’s minimum food requirements. The average cost of providing the meal (including food and transport) is 8½ US cents per day or E114 (US$15.68) per year for primary students and E162 (US$21.72) per year for secondary students (Table 12). 61 Table 11: Beneficiaries and Coverage of the School Feeding Program, 2011/12 Number of Number of % Students Schools of Schools Primary Schools 241,637 572 96 High Schools 23,039 218 93 Source: Ministry of Education and Training, School Nutrition Unit Table 12: Costs of Meals per Child per School Day and per Child School Year, 2011/12 Per Day Per Year E (US$) E (US$) Primary 0.60 (0.08) 114 (15.68) Secondary 0.64 (0.09) 162 (21.72) Source: WFP (2010) 130. The SFP is an important source of food for poor children, but its long-term sustainability is in doubt. The program relies heavily on external support, and the government already faces challenges in meeting its share of the program costs. Fiscal flows for the program are already irregular, and schools do not always have enough money to pay the cooks, in which case no meals are provided. There is some anecdotal evidence that attendance drops off on these days. Irregular financial flows sometimes also result in food aid being distributed late. As described above, under the FPE initiative, the portion of school fee accounted for by the school meal will be paid by the government. The rollout of FPE will require an increasing fiscal commitment to the school meal program and further calls into question the sustainability of the program. Moreover, the long-term availability of WFP support for the secondary school part of the program is not guaranteed, which means that the government might be required to finance the entire program in the future. 131. The capacity of the government to implement the School Feeding Program will need to be strengthened. Currently only one officer in MOET is assigned to the program and works in tandem with [an equivalent person in?] NERCHA. There is limited monitoring 60 The government has been introducing free primary education on a phased basis. Grades 1-3 are now covered. The implications of FPE for the social safety net are discussed in paragraph 141, below. 61 WFP (2010) 37 capacity in MOET, and the program has never been evaluated. Standardized operational guidelines for school meals are only now being developed, and it is not clear how they will be enforced. MOET already finds it difficult to monitor how schools are implementing the SFP, so it is not clear whether schools are providing students with the required meal or are diverting this money to other uses. Neighborhood Care Points 132. Neighborhood Care Points (NCPs) are service delivery points at the community level where OVCs can come daily to receive food, psychosocial support, basic health care, and pre-primary education (only in some NCPs) and to participate in recreational activities. 62 NCPs grew out of local volunteers taking the initiative to start informal daycare activities. NCPs are a collaborative effort between NERCHA, the WFP, UNICEF, the DSW, World Vision, and other NGOs. Initially most NCPs operated in makeshift buildings or outdoors, but approximately 46 percent now operate in NERCHA-financed brick and mortar buildings (Table 13). Expenditures on NCPs were estimated to have been E15.1 million (US$2.1 million) in FY2010/2011, plus E0.4 million (US$0.1 million) for social care services. Food is provided by the WFP and World Vision, but information on the value of the commodities that they provide is not available. Table 13: Physical Structure of NCPs Number % Brick and Mortar 713 46 Stick and Mud 202 13 Shacks 229 15 Open Areas 406 26 Total 1550 100 Source: NERCHA Mapping of Neighborhood Care Points, 2010/11 133. The first NCPs were established in 2001 in 18 communities. By 2011, 54,850 children aged 5 and under were registered in 1,554 NCPs. Ninety-six percent of centers are open every day and 93 percent provide feeding (Figure 18), 83 percent provide psychosocial support services, 71 percent provide health care services, and 70 percent provide early childhood care. 63 62 NCPs serve multiple safety net functions, but the largest share of expenditures relates to the provision of food, which is why we group them with other in-kind programs. 63 NERCHA Mapping of Neighborhood Care Points, 2010/11 38 Figure 18: Neighborhood Care Points Providing Feeding Legend NCh FEEDING SCHEME * YES e NO TIHI<>UNOLA c::::JREG'OHS ··- OJS7 Source: NERCHA Mapping of Neighborhood Care Points, 2010/11 ,,. -~ 21118 39 134. The primary groups targeted by the NCP project are young children living in child-headed households, children orphaned by AIDS, and other vulnerable children who are at risk of neglect, abuse, malnutrition, or exploitation. The primary focus is on pre-school- age children, although older children also participate in NCP activities, both as beneficiaries and as contributors to help with the younger children. The children are selected by the communities in consultation with traditional leaders. In some cases, the NCP caregivers in collaboration with other community-based workers such as Rural Health Motivators take the leading role in the selection process. The National Plan of Action for Children (2011-2015) estimates that 50 percent of OVCs aged 2 to 6 years old have received services at NCPs, but it is difficult to interpret this finding since it is not clear how many OVCs in this age group live in situations that would require them to seek out food and psychosocial support at an NCP. NCPs are typically open for half a day (from 8am to 1pm) and provide breakfast and lunch – a morning bowl of corn soya blend and a hot lunch of maize, pulses, and vegetable oil. The cost of providing meals is estimated to be E0.36 (US$0.05) per breakfast and E0.47 (US$0.06) per lunch or about E164 (US$22) per child per year. 64 135. NCPs rely on a cadre of volunteer caregivers and community support, and the capacity of caregivers and support from communities both vary widely. The caregivers are usually adult women selected by the community and ratified by the Chief and Inner Council. They are usually unpaid volunteers, some of whom receive food rations as compensation. The responsibilities of the NCP caregivers include preparing and serving food, organizing and supervising play and development activities, running health promotion activities, providing training on hygiene, sanitation, and basic child self-care, taking care of minor ailments (such as diarrhea or skin infections), providing education on HIV/ AIDS prevention, and protecting children from sexual exploitation and abuse. World Vision has supported the training of these caregivers to enable them to fill their care-giving roles. A management committee comprising community members oversees the affairs of the NCP. The volunteers are coordinated and monitored by KaGogo Centers, which are responsible for coordinating the community response to HIV and AIDS in addition to overseeing NCPs. A program was introduced to encourage caregivers to enroll in a certificate program in counseling for children at the University of KwaZulu Natal, but not all caregivers have had the opportunity to participate. Stress and burnout on the part of caregivers are ongoing concerns. In addition, community support for the NCPs has been uneven at best. These design features raise questions about the viability and sustainability of NCPs. 136. NCPs provide significant benefits to children, and several assessments have indicated that OVCs enrolled in NCPs fare better in terms of welfare than their counterparts who are not enrolled in NCPs. 65 Children in communities that have NCPs have more access to food, health care, education, and overall care and support. There is even some evidence that the number of child-head households declined as a result of NCPs, possibly because they made the orphans more visible to the community (thereby attracting community support for child-headed households) and because they reduced the cost to foster families of taking care of an orphan as they provided some of the food for the orphans. However, there is no uniformity 64 WFP (2010) 65 UNICEF (2006) and Dlamini (2007) 40 in the quality of services that NCPs provide or how well they function. About 7 percent of centers do not provide meals, which is arguably the most important benefit provided by NCPs. NCPs also require a number of inputs beside food, particularly learning and recreational materials, furniture, and cooking utensils, which are not always available. Finally, young children receive psychosocial support from NCPs, but the psychosocial needs of older orphans and vulnerable children are largely ignored. The recordkeeping and community monitoring and supervision of NCPs are both uneven. 137. For NCPs to be effective, they require significant capacity, cooperation among different agencies, and sophisticated monitoring, but it is not clear that these requirements are being met. At least three ministries, multiple international agencies, and a number of international and local NGOs provide inputs for NCPs (Table 14). As the menu of services provided by NCPs expands, particularly in the area of early childhood care and development (ECCD), the requirement for agencies to communicate and cooperate effectively with each other is becoming even more critical. Table 14: Agencies Supporting NCPs Function Agency Food and Nutrition MOA, WFP, FAO, Communities Basic Health Care MOH Water, Sanitation, and Hygiene MOH and NGOs Child Protection and Safety Lihlombe Lekukhalela, NGOs Early Childhood Care and Development MOET Psychosocial Support NGOS Links to Education MOET, NGOs Life Skills Development MOET, NGOs Source: GOS (2009) 138. Like other safety net programs, the sustainability of the NCP initiative is questionable. The program is dependent on support provided by donors such as UNICEF, the WFP, and World Vision, and it is unlikely that the government or the communities themselves would have the resources to absorb the operating costs of the NCPs if and when donor support were to be withdrawn. This is a particular concern in very poor communities. In the long run, the current reliance on volunteers may also not be sustainable. Moreover, no government agency has stepped forward to claim ownership of the NCPs, and as a result, the NCP draft strategy has never been finalized. Food Distribution 139. The NDMA in the DPM is responsible for coordinating and managing disaster- related activities, including administering the National Disaster Management Fund and managing national emergency assistance commodities. In 2006, the government passed the Disaster Management Act and established the National Disaster Management Agency (NDMA). It subsequently developed a Disaster Risk Reduction Action Plan (2008-2015) and a National Disaster Management Policy (2010) in an effort to better manage disasters. The number of recipients served each year is determined annually by the Vulnerability Assessment Committee and, based on this number; the government procures food (maize, pulses, and oil) from an in- 41 country supplier (only one contract is awarded annually). The food is distributed through local government (Iinkhundla) offices in drought-affected areas. Each Inkhundla receives 1,000 50kg bags of maize meal, 200 50kg bags of beans, and cooking oil in the amount of three times 750ml per person. Food is also distributed through NGOs. 140. The food is targeted to the elderly, child-headed households, people with disabilities, and households containing a chronically ill person. The Chief’s Assistant at the Inkhundla level works with the community disaster committee to determine who is in need. Food is typically distributed between September and March with enough food for three months given at each distribution. Approximately 67,000 households received assistance between November and December 2011, and another 53,000 were scheduled to receive assistance between January and March 2012. In FY2010/11, the total costs of food aid were estimated to have been E14.0 million (US$2.0 million). 141. The NDMA faces a number of implementation constraints that interfere with the effective implementation of the program. The agency has only one professional staff member and this means that it is unable to effectively oversee and monitor the distribution of food aid. As a result, oversight is provided by Members of Parliament, a situation that increases the possibility of political interference in the distribution of food aid. There are no clear guidelines for the distribution of food within communities. Finally, transportation shortages sometimes result in the late distribution of food. 142. Also, stakeholders have questioned the continued need for food aid in the context of the regularly occurring and seasonal nature of food shortages. According to the Vulnerability Assessment Committee’s 2011 Vulnerability Assessment, there were no major shocks to markets during the period between the previous assessment in June 2010 and June 2011. The level of consumer inflation remained relatively low, and Swaziland’s trading with its major partners remained constant throughout the period under review, suggesting that there is scope for the government to move away from providing food aid. Other In-kind Social Transfer Programs 143. Supplementary Feeding Programs. The WFP has been the government’s main partner in providing support for a supplementary feeding program targeted to malnourished people living with HIV, but that program has now been turned into a Food by Prescription Program. The benefits include a corn soya blend ration for the patient and a family ration of cereals, pulses, and vegetable oil. The program was launched in 2011, and between June and December of that year it served 8,664 patients plus up to six of their household members at a cost of E3.5 million (US$0.5 million). 144. Food for Work. There is no established safety net program designed to provide short- term employment to vulnerable people, but the NDMA began a pilot food-for-work scheme in 2011. 66 Under the pilot, households that do not fall into the target groups for food distribution 66 The Ministry of Public Works encourages contractors to hire local labor. The Ministry estimates that 400 people were employed for short periods in the six contracts executed during FY2010/11. 42 are required to provide labor to work on a community project. The communities themselves determine which infrastructure projects are to be included, but there is no financing for asset creation so the work is limited to what communities can do with their own tools and resources. Guidelines specify that recipients must do 20 hours of work per week for a ration that covers six people, but there seems to be some variation in the implementation of this requirement. Communities determine who can work on the scheme, with input from the Member of Parliament (MP). In the absence of sufficient NDMA staff to monitor the program, monitoring is left to MPs. Communities also determine the frequency of the distribution of food. No information is available on how the implementation of the pilot is progressing. World Vision also supports food distribution and food-for-work programs that target the elderly, orphans, the disabled, and the chronically ill. The pilot program provided support to 20,590 people in 2011 at an estimated cost of E0.5 million (US$0.08 million). 145. Fee Waivers for Health Care. Primary health care is free for all age groups. General maternal and child health services, including antenatal care, postnatal care, immunizations, and family planning services, are also provided for free at public and missionary health facilities. In addition, people aged 60 and over and people with disabilities, tuberculosis, leprosy, or mental illnesses are exempted from medical fees, except for an E10 registration fee. People with disabilities and OVCs receive free transport. There is no policy on fee waivers for the indigent so these are currently granted on an ad hoc basis. Public assistance beneficiaries are not automatically exempt from health fees. There is no information on the number of people exempted from health fees or on the value of the fees waived. 146. Phalala Specialist Referral Fund. The Phalala Fund was established in 2001 to help poor and uninsured Swazi citizens to access specialist medical care that is not available within the country. The program is administered by the MOH under the Administration of the Specialist Medical Aid Fund Regulations and is financed from general revenues. Local clinicians in public or private clinics and hospitals provide patients with referrals to specialists in South Africa, and each referral is reviewed and approved or turned down by a committee of medical experts. The MOH provides the patient with a guarantee letter that specifies how much is available for his or her care, but in practice the MOH will pay costs in excess of the specified limits and there is no incentive for hospitals to limit costs. There were 450 beneficiaries in 2010. The Fund receives an allocation of E60 million (US$8.5 million) annually but has had cost overruns of an undetermined amount every year. 147. Subsidies for Agricultural Inputs. Since 2005, agricultural subsidies for such inputs as fertilizer, limestone, ammonia, and nitrate have been provided to households with OVCs and other vulnerable groups. The program is implemented by community volunteers with support from the Global Fund (through NERCHA and the Ministry of Agriculture and Cooperatives). Inputs are provided to 324 chiefdoms for use in community gardens (of one hectare per chiefdom) and 20 household gardens (of 0.5 hectare per household) per chiefdom. Only 101 chiefdoms received support in 2010/11 because the remaining chiefdoms had leftover inputs from the previous year. Expenditures were equal to E8 million (US$1.13 million) in FY2010/2011. There was no allocation in FY2011/12. The plan for the future is to support households only, but it is not clear where any funding for future subsidies will be available. 43 Other Social Safety Net Programs 148. Community-based Programs. Community demand-driven projects are implemented by the Micro Projects Program (MPP), a semi-independent unit of the Ministry of Economic Planning and Development (MOEPD). The MPP was established in 1988 with support from the European Commission. The projects are adopted in response to demand from communities in such areas as water, education, electricity, agriculture, health and social welfare, dams, and small bridges. The beneficiaries of community demand-driven projects must be a clearly identifiable group located in deprived or peri-urban areas. The smallest group that can benefit is normally 10 people. The MMP contribution to each project ranges from E20,000 to E1 million. Geographic targeting is used to select communities. Communities are required to contribute 25 percent of the estimated project cost either in cash or in-kind. Operation and maintenance are the responsibility of the community. A total of 60 capital projects were completed in 2010/11 with another 12 projects in progress. The total cost of these community demand-driven projects was estimated to have been E20 million (US$2.7 million) in FY2010/11. More recently, the MPP has been charged with implementing small-scale capital projects for government ministries. A total of 201 capital projects were completed in 2010 at a cost of E93.7 million (US$12.6 million), with E77.2 million spent on MOET projects, E9.0 million spent on Ministry of Tinkhundla Administration and Development projects, and 7.5 million spent on MOH projects. Community demand-driven projects now account for only about 18 percent of MPP expenditures. 149. Active Labor Market Programs. Labor market programs (for example, second chance education, non-formal skills training, labor market mediation services, and microenterprise training or credit) are provided by both government and non-government providers. The Sebenta Institute offers non-formal primary education programs in all primary school subjects plus the siSwati language and the English language. Enrollment in 2007 was just under 3,300 students. The Emlalatini Development Center provides second chance education both face-to- face and through distance learning. At the secondary level, the center follows the same curriculum as formal secondary schools and provides programs for students requiring second chance education and for those trying to improve their examination grades to increase their chances of getting into university, technical and vocational education, and/or skills development programs. In 2008, 700 secondary-level students were enrolled. In terms of adult basic training and skills development, rural education centers are the main providers of this level of education, but their enrollment has been declining since 2001. Eighteen non-formal providers receive government assistance to cover their instructors’ salaries. Collectively, these providers enroll about 800 trainees a year and tend to offer entry-level skills training programs. Two nongovernmental privately financed centers enroll about 190 trainees. 67 The government provides support for microenterprise development through the Small Enterprise Development Company (SEDCO), which provides marketing, management consultancy and training, mentoring, and legal services. During FY2010/2011, 150 people from the four regions were trained and then started businesses, which cost E1.4 million (US$0.2 million). FINCORP, a government corporation, provides individual loans to micro- and small businesses and for on- 67 Marope (2010) 44 lending by smaller lending agencies. In 2009, FINCORP had a gross loan portfolio of approximately E42.1 million (US$5.94 million), with a little over 4,500 borrowers. 150. Social Care Services. The DSW provides social care services, including both residential and non-residential care. Between October and December 2010, the DSW provided counseling services to 523 people including 124 families, 300 children, and 99 elderly people. It also provides services for the prevention and treatment of domestic violence, child abuse, and human trafficking as well as services for people with disabilities. The DSW is responsible for supervising residential care facilities in Swaziland, all of which are run by NGOs. In 2010/11 there were 1,465 children housed in 40 institutions. The National Children’s Coordinating Unit (NCCU) is responsible for coordinating and upgrading psychosocial support services, particularly those provided by the NCPs. In FY2010/11, the government supported selected NGOs that were implementing social care services to the tune of E1.1 million (US$0.2 million). 68 151. General Subsidies. There are no general subsidies in Swaziland, but the prices of a number of commodities are regulated including those for bread, petrol, liquid petroleum gas, and transport. A value added tax (VAT) was introduced on April 1, 2012 on commodities and services though a deliberate attempt was made to zero rate goods consumed by the poor. It is beyond the scope of this report to assess the likely impact of the VAT on poverty, but this information will be needed to inform the design of future safety net interventions, particularly cash transfers. C. Cross-cutting Requirements 152. In addition to the constraints identified above for individual programs, there are a number of concerns that are common across programs. These are discussed in this section. Developing a Strategic Vision for the Safety Net 153. Swaziland implements a number of social transfer programs, but the programs do not form a coherent whole. Swaziland’s social safety net programs are not well coordinated and there is a need to clearly define the overall objective of the country’s safety net. A number of policy documents exist that cover some aspects of social protection, but at present they do not form a coherent whole. The safety net strategy should spell out the government’s objectives for its safety net system and commit to allocating the resources necessary to achieve those objectives. Increasing Efficiency and Accountability 154. Information on the administrative costs of safety net programs is incomplete, but available data suggest that administrative costs are high in Swaziland relative to other 68 Including social care services provided by NCPs through the Lutsango Lwaka Ngwane program. 45 countries. According to the government’s Estimates of Expenditure, the administrative costs of the cash transfer programs run by the DSW account for approximately 15 percent of total DSW costs. However, these are actually higher than this as these estimates do not include the administrative costs incurred by the DPM in overseeing the program or the costs of office space and utilities. The non-transfer costs of the Young Heroes Program (which is partially funded by private donor contributions) account for about 45 percent of total costs not including the costs of office space and utilities or other support services provided by NERCHA but including the costs involved in fundraising. In other countries, administrative costs average 8.2 percent of total costs (Table 15). 69 Although these international comparisons are somewhat outdated and need to be treated with caution as programs operate in different contexts and because approaches to measurement may be different, these findings suggest that there is considerable scope for reducing administrative costs in Swaziland. 155. Programs use a variety of mechanisms to get cash transfers to households. The DSW uses cash and direct deposits for the OAG and checks for the Military Pension and Public Assistance (distributed at regional offices). Young Heroes delivers payments through the Post Office. Having a range of different payment mechanisms is administratively more complex than just having one as it can impede the ability of administrators to oversee and reconcile disbursements, thus reducing accountability and control. The transaction costs incurred by beneficiaries also differ depending on which payment method is used. The cash payments made by the DSW involve significant administrative costs. In addition to the costs of funding the police personnel who are involved in safeguarding the distribution of the cash, the DSW reports that between 65 and 100 staff are engaged in making payments during four months of the year (one month for each quarterly payment). Table 15: Administrative Costs as a Percentage of Total Costs Country Year Program % Cash and Near Transfersa/ Albania n.a. Ndihme Ekonomika 7.2 Armenia 2004 Family Poverty Benefit Program 2.2 Bulgaria 2006 Guaranteed Minimum Income Program 9.9 Bulgaria 2004 Child Allowances 5.6 Lithuania 2005 Social Benefit Program 6.5 Romania 2004 Guaranteed Minimum Income Program 9.8 Honduras 2003 Food Stamps For Female-headed Households 12.0 Sri Lanka 2001 Food Stamp Program 2.0 Mean 6.9 Conditional Cash Transfers Bangladesh 2002 Primary Education Stipend Program 4.0 Brazil Bolsa Familia 12.3 Colombia 2000/4 Familias en Acción 10.5 Dominican Republic 2006 Solidaridad 5.9 Ecuador 2005 Bono de Desarrollo Humano 4.1 Jamaica 2004/5 PATH 13.0 Mexico 2003 PROGRESA/ Oportunidades 6.0 69 Conditional cash transfer (CCT) programs are cash transfer programs that require beneficiaries to meet certain conditions in order to receive benefits, such as sending their children to school or taking them to health centers. CCTs and their applicability to Swaziland are discussed in more detail in Chapter 5. 46 Pakistan 2005/6 Child Support Program (pilot) 6.7 Peru 2006 Juntos 11.6 Mean 8.2 Source: Adapted from Grosh et al (2008) Note: a/ Near cash programs provide a coupon that can be used like cash but only for selected items such as food. 156. Swaziland has slowly been moving towards paying OAG beneficiaries electronically, but this transition has been slower than in some other countries. In Swaziland, only 20 percent of OAG beneficiaries currently receive their benefits through bank accounts compared to 91 percent in Columbia, 59 percent in South Africa, 34 percent in Mexico, and 15 percent in Brazil. 70 At the same time, the transaction costs involved in depositing benefits into recipients’ bank accounts are high in Swaziland compared to other countries (Table 16), although it is important to note that unit costs may be affected by the program’s size. International evidence indicates that moving to electronic payments can result in cost savings for the programs and that recipients eventually welcome the convenience of electronic payments over cash. However, it will take time for their entrenched attitudes and behavior patterns to change, and it will require clear, consistent communication to explain the advantages of the new payment method. The government should attempt to bring current costs down when and if it decides to expand electronic payments. Table 16: Cost of Making Payments to Households Number of Country Cost per Transaction a/ Payments US$ per Year Swaziland OAG $3.25 for mainstream financial account b/ 12 Swaziland Young Heroes 3% of transaction with a maximum of US$1.36 per transaction 4 Ecuador $0.3 per transaction 12 Honduras 2.5% of amount to be transferred 3 El Salvador $1.40-1.60 6 Jamaica $0.46 for checks delivered at post offices c/ 6 Brazil $0.88 for limited purpose account 12 $0.60 for mainstream financial account $6.24 for limited purpose account 6 Colombia $5.20 for cash payment $2.35 for cash payment 5 Mexico $2.84 for mainstream financial account Sources: CGAP (2012) for Mexico, Columbia, and Brazil. Information collected by authors for Ecuador, Honduras, El Salvador, and Jamaica. Notes: a/ Limited purpose accounts cannot store funds indefinitely, cannot access funds outside dedicated infrastructure, and cannot deposit additional funds. Mainstream financial accounts can store funds indefinitely, can access funds outside dedicated infrastructure, and can deposit additional funds. b/ Calculated at US$1 = E7.37 as of March 5, 2012. c/ Does not include cost of transactions via direct deposit. 157. As a small state, Swaziland cannot easily realize economies of scale under the best of circumstances. However, there is some scope for achieving economies of scale by requiring programs to share certain common functions, such as targeting, registration, information 70 CGAP (2012) 47 management, and monitoring and evaluation. For example, a central beneficiary registry would provide important monitoring and planning information and could reduce abuse. 158. Accountability and control mechanisms need to be strengthened. Accountability and control mechanisms promote more efficient use of public resources, prevent political manipulation, and build public support for programs. In the absence of a modern information system that can generate basic monitoring data (for example, on the age, gender, payment history, and collection history of recipients), it is difficult to monitor who is receiving benefits or to understand why a beneficiary has not collected his or her benefit and to plan accordingly. Updated accounting systems are also needed to reconcile payments after each payment cycle. The process of paying out cash benefits consumes a large amount of time (between 65 and 100 DSW staff devote four months per year to making payments). In addition, cash payments leave the government vulnerable to uninsured losses, such as a recent robbery of a payment station. 159. Setting simple eligibility criteria would reduce the possibility of administrators using their own discretion and the opportunity for political manipulation. Eliminating intermediaries by transferring benefits directly into the beneficiaries’ bank accounts or by using cell phones or remittance companies would reduce the opportunity for fraud. Information and education campaigns would make sure that the program’s eligibility rules, benefit levels, and beneficiary rights and obligations were clear to staff, clients, and to the general public, which would reduce error and fraud. Appeals mechanisms also enhance accountability and control, and collecting information from the public, matching data from different sources, randomly checking payments, carrying out client and external audits, and increasing sanctions and prosecutions can also help to detect and reduce fraud. 71 Documented policies and procedures governing the operational aspects of safety net programs are also needed. At the same time, the Auditor General’s Office should be congratulated on its efforts to conduct process and financial audits of key programs. 160. Program monitoring is almost universally weak, and there has been no rigorous impact evaluation of any safety net program. Most programs collect basic information, but these data are not always systematically collated. There is no monitoring and evaluation framework to track progress in implementing safety net programs or assessing what progress they make towards their objectives. An upgrading of the information system for the OVC Education Grant is in progress. An upgraded MIS for the OAG is also needed, and staff need to be trained in the use of these systems. There is also a crucial need to devise clear and measurable monitoring indicators and to develop a mechanism for systematically reporting these monitoring indicators. The lack of regular national monitoring data, particularly on poverty, living standards, and the labor force, further impedes monitoring and evaluation. The system would benefit from a monitoring framework that includes all programs as well as common key performance indicators and a standard methodology for collecting and reporting on those indicators. A simple and implementable monitoring framework that builds upon a common MIS would help to reduce duplication and overlap and could reduce monitoring costs and increase program efficiency and accountability. 71 van Stolk (2007) and Grosh and Tesliuc (2007) 48 Filling Gaps in the Social Safety Net 161. The government has made efforts to provide social protection to the poor and vulnerable. However, significant numbers of poor and vulnerable people remain unprotected, and children seem to be a particularly vulnerable group. As indicated in Chapter 2, children have an average poverty rate of about 70 percent, but Swaziland has no cash assistance program for extremely poor households with children. The OVC Education Grant Program reaches only about 40 percent of children in the poorest quintiles and does not cover the considerable out-of-pocket costs of education. The OAG pension was introduced to lighten the burden on elderly people caring for orphans, and it has provided some support to orphans who live with an elderly person. However, 55 percent of poor children and 56 percent of very poor children do not live with an elderly person. Only disabled people are now eligible for Public Assistance, and this program has the capacity to reach only about 15 percent of the severely disabled and less than 1 percent of the poor. Given that the estimated number of extreme poor in Swaziland is 295,000 and that the existing safety net reaches many people who are not poor, coverage of the extreme poor is low. 162. The School Feeding Program provides support to all schoolchildren, but children who cannot afford to go to school do not benefit. In addition, the School Feeding Program faces the same challenges as school feeding programs around the world, in that it only provides meals for about 200 days (60 percent) of the year so children do not receive the meal during weekends and school holidays. 163. The attention focused on orphans as a result of the HIV/AIDS pandemic means that less attention has been paid to children made vulnerable by poverty. The two groups are not synonymous. Thirty-six percent of orphans are very poor, but 25 percent of orphans are not poor. Eight-five percent of extremely poor children are not orphans. 72 Fifty-six percent of very poor children do not live with an elderly person and, therefore, do not have access to any cash assistance for their basic needs. 164. The removal in 2011 of the indigent category from Public Assistance means that the little support that used to be available for very poor households is no longer available. Since very poor households with children are the most likely to be extremely poor, this means than extremely poor households with children do not have access to any cash assistance to help them to meet their basic needs. 165. There are no public works programs in Swaziland that could mitigate the risks involved in different shocks. Public works programs can help unemployed youths and adults trade their labor for income to smooth their consumption and can also play an important role in supporting households during the agricultural lean season and in the aftemath of certain shocks. The government is currently piloting a small food-for-work program, but very limited information about this initiative is available. The government would need to carefully consider whether an expanded public works program is feasible given existing resource and capacity constraints. However, it is still important to note that this constitutes a major gap in the safety 72 Authors’ calculations based on SHIES 2009/10. 49 net as these type of programs have played an important role in supporting poor and vulnerable households in other countries. It might be a good idea to pilot a public works program from which to learn lessons that would guide any potential expansion of the program. Doing a Better Job of Identifying the Vulnerable 166. Means-tested programs each have their own application procedures and eligibility criteria, and these multiple targeting mechanisms increase administrative costs. Multiple application procedures also increase transaction costs for beneficiaries and make it difficult for vulnerable people like the poor or children from child-headed households to navigate the system. The OVC Education Grant, Child Welfare Grant, and Public Assistance (if the latter two are re-introduced) as well as the Young Heroes program, and fee waivers for health all use different targeting criteria and procedures although, to a large extent, the programs serve the same households. 167. Targeting criteria for programs are subjective, and this increases the likelihood of inclusion and exclusion errors. Social workers and program managers have expressed concern about the opportunity for elite capture in the community targeting of OVC grants and for political interference in the selection of the beneficiaries of means-tested programs. At the same time, eligible orphans may be denied benefits because they cannot produce a death certificate for their deceased parents, which may signify a family conflict that is putting those particular orphans in an even more vulnerable position than those who are able to produce the necessary death certificate. The government has developed new criteria aimed at making the identification of OVCs more transparent, but they are still very vague and open to subjective interpretation (Box 5). More objective and transparent targeting mechanisms would significantly reduce the problems that can lead to inclusion and exclusion errors. Box 5: Eligibility Criteria for OVC Grants x One or both parents is chronically/terminally ill (bedridden) x One or both parents has a severe disability x One or both parents has passed away and the guardian or child does not have a source of income to pay for school fees x The child has been deserted by his or her parents and their whereabouts are not known x The child is living with parents who have no source of income or with a single parent who is not employed x Parents have many children with very limited financial resources. Source: Department of Social Welfare Improving the Crisis Response 168. At present Swaziland lacks a safety net that can effectively respond to shocks and crises of different kinds. In what is known as a “counter-cyclical” approach, government spending on safety nets should ideally increase in times when households are facing crises or shocks to help them to weather the shock (as indicated in Chapter 2, food price increases and the occurrence of drought/flood are the most common risks faced by poor households) and then 50 should decrease as the impact of the shock subsides. However, the Government of Swaziland’s response to the fiscal crisis has been to reduce expenditures on safety net programs, particularly cash transfers, by delaying payments to some OAG recipients and eliminating the element of Public Assistance that was intended to target the poor. This approach has served to exacerbate vulnerability and also to increase inequality, which is already a problem in Swaziland. 169. Natural disasters and economic shocks are often accompanied by significant drops in household income and resources, which means that social safety nets should be designed to help households that become poor as a result of the shock. If well-designed, targeting mechanisms can identify important segments of society in need of assistance, such as the newly poor. 73 At the same time, there is evidence that ongoing (as opposed to emergency) cash transfers can reduce the impact of shocks and, thus, reduce the need for emergency interventions. One estimate suggests that, had Mexico’s Oportunidades cash grant program been in place during the Mexican financial crisis in 1995, the poverty headcount, poverty gap, and squared poverty gap would have been 8.8 percent, 17.1 percent, and 22.5 percent less than their actual values during the crisis. 74 170. Although many shocks in Swaziland are reoccurring seasonal shocks that are highly predictable, the government’s only real crisis response is to provide food aid. Evidence from other countries suggests that cash transfers can help to maintain households’ food consumption, preserve their investments in education and health, and enable them to avoid distress sales of their assets. The extent to which cash transfers can do this effectively in emergency situations depends on whether local markets exist that are capable of efficiently meeting the increased demand and of offering a range of food and goods that are accessible to affected populations and with low risk of inflationary pressure. Both conditions have been met in recent years, and this means that policymakers in Swaziland should consider moving away from food aid and towards cash assistance, possibly through a public works program. 171. Swaziland does not currently have a functional public works program. Public works programs are a type of safety net intervention in which recipients temporarily trade their labor for cash or in-kind payments to smooth their income during crises such as drought or recession or during the agricultural off-season. They have been used extensively in countries of all income levels around the world and specifically in the developing world as counter-cyclical interventions. Not only do they help to reduce short-term poverty but they also have the indirect effect of creating or maintaining public assets. However, since the cost of transferring US$1 of income to a poor household under a public works program is higher than in a simple cash transfer, public works programs need to be carefully designed. 75 D. Public Expenditures on the Safety Net Trends in Safety Net Spending 73 Blomquist et al (2001) 74 Davis et al (2004) 75 del Ninno et al (2009) 51 171. Swaziland spends about 2.2 percent of GDP on its social safety net, and it is important to maximize the returns from these investments. In FY2010/11, expenditures on the safety net were estimated to have been E584.4 million (US$82.4 million). 76 Between FY2005/06 and FY2010/11, expenditures increased in both nominal and real terms (Figure 19) and as a percentage of GDP (Figure 20 – also see Annex 5, Annex Table 5.1 for details. Spending on the safety net in Swaziland is higher than the average of 1 to 2 percent spent by most developing countries 77 but is lower than the amount spent by several other Southern African countries, including Malawi, Botswana, Mauritius, and South Africa (Table 20). Figure 19: Social Safety Net Spending Over Time Source: Authors’ calculations Figure 20: Safety Net Expenditures as a Percentage of GDP Source: Authors’ calculations Table 17: Social Safety Net Expenditures in Selected African Countries (% of Country/region Year GDP) Source Malawi 2003-2006 6.5 Slater and Tsoka (2007). Mauritius 2008/09 4.4 Blank (2010) South Africa 2002/03 3.2 World Bank (2012) 76 The totals reported here include spending on social transfers and other social safety net programs. The social transfers reported in Table 3.1 [which table is this?] represent a subset of these expenditures. 77 Grosh et al (2008) 52 Botswana 2010 2.2 World Bank (2012) Swaziland 2010/11 2.2 This report Mozambique 2010 1.7 World Bank (2012) Rwanda 2010/11 1.1 World Bank (2012) Kenya 2010 0.8 World Bank (2012) Burkina Faso 2005-2009 0.6 World Bank (2012) Mali 2009 0.5 World Bank (2012) Tanzania 2010 0.3 World Bank (2012) Benin 2005-2009 0.3 World Bank (2012) Zambia 2010 0.2 World Bank (2012) Source: Authors’ compilation 172. In-kind transfers account for the largest share of safety net programs, followed by cash transfers. Expenditures on cash transfers were estimated to have been E131.6 million (US$18.6 million) in FY2010/11 or about 23 percent of safety net expenditures, while expenditures on in-kind transfers were estimated to have been E403.8 million (US$57.0 million) or about 69 percent of safety net spending. In combination, other programs received 8 percent of the safety net budget (Annex 5 Table 5.2). Opportunities to rein in spending on in-kind transfers, particularly through better targeting and more efficient delivery, will need to be explored. 173. Expenditures on cash transfers are estimated to amount to E131.6 million (US$18.6 million) or 23 percent of total safety net spending. The OAG absorbs the largest share of cash transfer spending (89 percent), dwarfing all other programs in terms of fiscal allocations (Figure 21). Between FY2006/07 and FY2010/11, expenditures on OAG direct transfers increased from E42.6 million to E100.3 million (US$6.1 million to US$14.2 million), or by 135 percent in nominal terms and 72 percent in real terms. Figure 21: The Distribution of Cash Transfers Source: Authors’ calculations 174. Capital expenditures, which include externally funded and local capital projects, account for about 21 percent of total safety net spending and represent a declining share of overall safety net expenditures. Between FY2006/07 and FY2010/11, recurrent expenditures on safety net programs increased by over 100 percent in nominal terms and by 50 53 percent in real terms, primarily as a result of increased expenditures on OVC Education Grants and the Old Age Grant (OAG). During the same period, capital expenditures (funded by both the government and external donors) decreased by an estimated 57 percent in real terms (Figure 22). Figure 22: Recurrent and Capital Expenditures over Time Source: Authors’ calculations 175. The sustainability of the current social safety net is in question because many key programs depend on external funding. A number of key programs, including the SFP, the NCPs, and food distribution are either partially or totally funded by external donors. The EU’s funding for OVC bursaries ended in 2012. The WFP’s financing of the school feeding program ended in 2011, though its support will be temporarily reintroduced in 2012. Young Heroes is partially donor-funded, and this has resulted in an unpredictable income stream. Global Fund Round 11 was cancelled, and NERCHA is currently receiving resources with through an interim financing facility. Future financing flows from the Global Fund are uncertain. About 21 percent of total safety net spending comes through the capital budget (both foreign and local capital projects), but sustainability requires that key safety net programs be absorbed into the government’s recurrent budget. The Impact of the Fiscal Crisis on the Social Safety Net 176. In the face of budget constraints and because spending on cash grants is discretionary, estimated and actual expenditures on cash grants can differ widely. As the fiscal crisis deepened, fiscal allocations to cash transfer programs became unreliable. During the first three-quarters of FY2011/12, expenditures on cash grants were considerably lower than the original budget estimates. Most OAG beneficiaries and all beneficiaries of other DSW cash transfer programs received only two quarterly payments during the first three-quarters of the fiscal year. At the end of the fourth quarter, beneficiaries who had received only two payments received a double payment. 177. The response to the fiscal crisis has exacerbated existing inequalities and may increase vulnerability in the future. DSW social workers have reported that OAG recipients who have bank accounts are likely to be former formal sector workers (public or private) receiving government or private pensions, while the recipients who are paid in cash are more likely to be poor. The government’s response to the fiscal shortfalls was to pay the former group 54 of beneficiaries on time and in full, while paying the latter beneficiaries between one and two months late. The indigent category of Public Assistance was suspended, meaning that there is essentially no cash transfer program for poor households with no elderly members. Child Welfare Grants were merged with OVC grants, and grants to children in hospitals and to fire victims were suspended in FY2011/12 (Table 18). Table 18: Payments to Beneficiaries during the Fiscal Crisis Program Payments The 10,000 beneficiaries who receive their payments monthly via direct deposit to bank accounts have been paid on time and in full. Old Age Grant The 45,692 beneficiaries who are paid in cash quarterly received their full payments but were paid from one to two months late in each pay period in FY2011/12. It is therefore likely that these beneficiaries received only three instead of four payments in FY2011/12. Military Pensions All pensioners (or their widows) were paid on time and in full. Support to the indigent under Public Assistance was terminated in 2011, and 935 beneficiaries were dropped from the program. All beneficiaries now fall under the disability category. Public Assistance The 3,075 disability beneficiaries who are paid in cash quarterly received their full payments from one to two months late for each pay period in FY2011/12. It is likely that these beneficiaries received only three instead of four payments in FY2011/12. Child Welfare Grants This program was merged with the OVC Education Grant. Children in Hospitals No payments in FY2011/12 Handicapped Children No payments in FY2011/12 Training Centers for Closed in FY2011/12 Disabled HIV/AIDS and Social Safety Net Spending 178. The national HIV/AIDS response is guided by the National Multi-sectoral Strategic Framework for HIV and AIDS 2009-2014, which devotes a large share of HIV/AIDS resources to safety net programs. The framework includes impact mitigation, which it defines as programs to alleviate the negative social and economic effects on the lives of people, society, and economy of the HIV/AIDS pandemic, poverty, or income inequalities. 78 The following social safety net programs form the core of the impact mitigation strategy: (i) food and nutrition for vulnerable households and individuals; (ii) education support for OVCs; (iii) provision of psychosocial support; (iv) community capacity building; and (v) building the capacity of the DSW and the National Children’s Coordination Unit. 179. Strengthening the safety net will strengthen the HIV and AIDS response. Impact mitigation absorbs about 30 percent of all spending related to HIV and AIDS. It also accounts 78 The three other thematic areas included in the strategic framework are prevention; treatment, care, and support; and response management. 55 for 1.6 percent of GDP, and this is projected to increase to 2.3 percent of GDP by 2020. 79 This highlights again the importance of increasing the efficiency and effectiveness of the safety net. E. Conclusions 180. In assessing the effectiveness of safety net programs, we need to ask how well they help households to manage covariate and idiosyncratic risks, including chronic and transient poverty, risks that threaten the human capital development of children and adolescents, unemployment, disease, and disability. We also need to know how well they meet the needs of particularly vulnerable groups, including single-headed households, the elderly taking care of children, people with disabilities, and people affected by HIV/AIDS. One program alone cannot address all of these critical risks. What will be required is a range of programs that, in combination, can protect the poor and vulnerable during periods of macroeconomic instability, during economic downturns, and after droughts/floods. 181. This review of Swaziland’s social transfers indicates that some of the elements of an appropriate safety net are present, but the system currently consists of a range of uncoordinated programs and some critical challenges remain. These include gaps in coverage (particularly of poor children and unemployed youths and adults), leakage of benefits to the non-poor, administrative inefficiencies, a lack of effective accountability mechanisms, no system for responding to shocks and crises, a lack of adequate monitoring and evaluation of programs, and non-sustainable financing arrangements. Programs need to be better coordinated, and economies of scale and efficiency gains need to be realized, for example, by implementing common application, targeting, and MIS systems. 182. Swaziland faces an unprecedented fiscal crisis, which means that the safety net must be carefully constructed to maximize its impact and cost-effectiveness. The following chapter looks at options for targeting the safety net so that it can do a better job of reaching the needy, and the final chapter discusses sustainable social safety net reforms. 79 World Bank (2010b) The Fiscal Dimensions of HIV/AIDS in SADC Member Countries, Washington DC. 56 CHAPTER 4: Targeting and the Safety Net in Swaziland A. The Rationale for Targeting 183. When facing constrained budgets, a natural response for policymakers is to consider how to concentrate spending on the poor or otherwise vulnerable. How and whom they choose to target affects the cost and potentially the effectiveness of transfer programs in reducing poverty. In Swaziland, the government uses a variety of different targeting methods. For example, it uses universal (or near-universal) targeting in the Old Age Grant and School Feeding programs but aims for a narrower, more specific pool of beneficiaries in the OVC Education Grant Scheme and the Public Assistance Program. 184. Targeting involves two decisions ௅ whom to target (the eligible group) and how to select them (the targeting methodology). Who to target (for example, orphans, the elderly, children, or poor households) can sometimes guide the choice of targeting methodology. For example, if a government targets the elderly, data from administrative records will be needed to determine the age and therefore the eligibility of potential recipients. Deciding to target one group, such as households with poor children, does not mean that other vulnerable groups, for example orphans or the elderly, will necessarily be excluded. However, the decision about which group to target does have significant implications for the share of each group selected (Box 6). Box 6: Targeting Cash Transfers in Zambia ௅An Assessment of Different Selection Criteria In 2008, researchers analyzed the four selection methods that were being used in a poverty targeted cash transfer scheme in Zambia to see which method reached the greatest number of poor children. Similar methods are currently being used in several other programs in Africa. The first selection method targeted labor-constrained households, defined as households that contained no individuals between the ages of 15 and 60 and had a dependency ratio greater than three (3.0). This method is similar to those used in programs in Malawi and Zambia. The second method targeted households that contained vulnerable adults, defined as a woman or a man over the ages of 55 or 60 respectively, or a disabled adult. This method is currently used in the cash transfer program in Mozambique. The third method targeted poor households with children under the age of 18, a variation of which is used in the Kenyan OVC cash transfer program. The fourth selection method targeted the poorest households without taking into account any other characteristic. Eligibility under all four regimes was limited to the households in the bottom three deciles of per capita consumption. For each method, the researchers identified eligible households and ranked them from poorest to richest based on per capita consumption. The table below presents results for the total number of households and individuals reached under each selection scheme. In all cases, the share of children in recipient households was at least 40 percent and reached 48 percent under the third method, which explicitly targeted households with children. All four methods, even the first and second, which did not explicitly target children, reached a significant number of children (at least 40 percent). However, the proportion of children in the poorest decile varied significantly between methods. In the first and second methods, only 36 and 38 percent of children were from the poorest consumption decile, in contrast to the third and fourth methods where 68 and 76 percent of children were from the 57 bottom consumption decile. Orphans (defined here as any type of orphan) represented about 25 percent of all children in the households selected by the first and second methods but only about 16 percent of all children under the third and fourth methods. Hence, labor-constrained households and those containing vulnerable adults (elderly or disabled) seemed to be more likely to contain orphans than other types of households. However, the third and fourth methods reached a larger total number of orphans than the first and second. In addition, as we indicated earlier, the children reached by the third and fourth methods, including orphans, were poorer than those reached by the first and second methods. So the total number of orphans reached under the third and fourth methods was greater, and these orphans were from poorer households. Moreover, the first and second methods probably missed the poorest orphans, for example, those who did not live in labor-constrained households. Demographic Profile of Households Reached Using Different Selection Criteria Mean Children Children in Children Orphans Household Individuals 0-17 bottom Orphans as % of as % of Size Reached Reached Decile Reached Total Children 1. Labor- constrained households 6.25 440,240 197,602 71,498 47,414 44.89 23.99 2. Households with elderly or disabled 6.45 1,148,399 480,720 188,428 126,657 41.86 26.35 3. Households with children 6.65 1,812,867 878,385 600,218 143,707 48.45 16.36 4. Poorest households 6.30 1,718,402 793,598 600,218 131,063 46.18 16.52 In summary, all of the targeting schemes reached a significant number of children. However, the proportion of children in the poorest decile was highest in schemes 3 and 4, the absolute number of orphans reached was also highest, and these orphans were from the poorest households. Source: Blank and Handa (2008) 185. Given Swaziland’s high poverty levels, it would not be realistic to target all poor households so the government will need to identify priority groups, such as children, the elderly, and the unemployed. The government would achieve the greatest impact in terms of reducing poverty by targeting the extremely poor. This is the group of Swazis who live below the food poverty line and do not have enough to eat. This group needs a range of safety net assistance including cash transfers to smooth their consumption, education grants, school feeding, fee waivers for health care, and perhaps others. 186. In theory, the gains of targeting can appear to be large, but the associated costs may reduce those gains. If all of the benefits provided by a transfer program were received by the poorest people in the population, then the same amount of resources would have more impact (in terms of reducing poverty) than if they had been universally targeted. However, in practice, the full theoretical gain is never realized because targeting is never perfect and because there are costs associated with the targeting (including administrative costs and transaction costs). 80 80 Grosh et al (2008) 58 187. Which targeting method/s to choose depends on many different variables, and there is no one magic solution to the challenge of targeting. An approach that works well in one country may not work as well in another. The country’s context, community structures, institutional capacity, and budget, what data are easily available, and the specific program objective are only some of the variables that need to be taken into consideration when deciding what targeting method to use. No targeting method is perfect, and they all result in some eligible people being erroneously excluded (exclusion errors) and some ineligible people being erroneously included (inclusion errors). B. Common Targeting Methods 188. In this section, we briefly describe each of the main targeting methods used in transfer programs worldwide. 189. Geographical targeting identifies beneficiaries by their geographical location (regions, districts, or locations) based on the objective of the program. If the program’s aim is to reduce poverty, then it focuses its activities in areas with high levels of poverty. If its aim is to reduce food insecurity, then it targets food-insecure areas. This method has the benefit of being easy to implement and easy to justify as a way of limiting program coverage when it is not possible to implement a universal program. Geographical targeting makes particular sense in response to localized problems (such as drought). However, this method requires adequate data, for example, survey data to rank districts with the largest number of a targeted group (such as orphans) or a high level of poverty. The problem is that survey data is seldom disaggregated to the lowest levels (such as locations within districts). Thus, more specific geographic targeting requires so-called “poverty maps,” which combine survey data and census data to specifically identify the poorest areas of a country. 190. Means testing involves using data on the income or wealth of potential beneficiaries to determine their eligibility. This method is how most programs in developed countries are targeted. Per capita income is one measure of welfare, but it is very difficult to obtain correct data on earnings, especially in countries with large informal or subsistence economies and in rural areas. 191. Proxy means testing (PMT) is an alternative way to measure need in countries where income data are unreliable or hard to obtain. The PMT is based on national household survey data, and the method uses the ownership of assets and other indicators to estimate household income. Since data need to be collected for these proxies, it is important to select proxies that are easy to measure and verify, such as the age of household members, the size of the household, the education level of the household head, or the type of roof or floor in the home. Each variable is given a weight based on its estimated impact on household expenditure (for example, the size of a household may be more correlated with poverty than the type of roof or floor.). Once the necessary data have been collected from potential beneficiary households, the households are ranked and given a score based on those data. Households that receive a score below a certain threshold point are eligible for benefits, while households that receive a score 59 above that point are not eligible. The threshold to be used can vary even for different programs that use the same PMT. For example, the threshold for a cash grant targeted to the poor might be lower than the one for accessing an OVC Education Grant or a health fee waiver. The main challenges with the PMT method is that: (i) it is administratively demanding to collect the necessary data and (ii) it can be challenging to identify easily measurable variables that adequately distinguish the poorest from the not so poor (as many of the population may live in similar homes and have almost no assets, making it hard to find attributes that distinguish the poorest from the less poor). Therefore, the PMT approach is commonly used in combination with community-based targeting or validation. 192. Categorical targeting is based on the idea that certain groups (the elderly, children, orphans, or disabled people) are at greater risk of being poor than the general population. This form of targeting is complicated by the fact that most vulnerable group members live in households with others who can and do contribute to their consumption/welfare. Targeting based on age is often politically popular and is administratively simple unless birth certificates are not widely available or are unreliable. However, its limitations are that age is often only weakly correlated with poverty, or the age group in question may be so numerous that covering all members is not financially feasible. 193. Self-targeting requires the least data and involves almost no targeting cost. Examples include offering a low wage rate for public works employment – so only the poorest will apply – or offering inferior goods (such as yellow maize or broken rice) that families will only take if they are in extreme need, or very small packages of benefits such as fertilizer. The main challenge is that policymakers and the public may object to these extremely low levels of benefits, but to make them any higher would defeat the objective of self-targeting. 194. Community-based targeting involves using community members to identify those households who are eligible for a program. Often a group of respected elders or elected villagers are given guidelines on who should benefit from the program in question. The rationale behind this method is that only local people with knowledge of the community can identify those who fall into a specific category (for example, there may be many orphans in a village but only some may be destitute). The problems that can arise using this method are the risk of favoritism and the risk that communities will spread the benefits too thinly because they wish to avoid anyone being excluded. Community-based targeting systems are also labor-intensive to set up and support. However, evidence from Africa suggests that community-based targeting can be used successfully to identify the poor (Box 7). Box 7: Community-based Targeting in Africa The Malawi Social Cash Transfer Scheme (SCT), the Kenya Cash Transfer for Orphans and Vulnerable Children (CTOVC), and Mozambique’s Programa Subsídio de Alimentos (PSA) Program are targeted cash transfer programs that incorporate demographic eligibility criteria, as well as community-based targeting into their beneficiary selection process. The process in each countries is as follows: x In Malawi, community members form a Community Social Protection Committee that visits each household to verify its eligibility and ranks each household according to need. Results are published publicly. District officials verify that all applicants qualify. x In Kenya, districts are selected based on HIV prevalence, and community-level Location OVC Committees 60 (LOC) targets households based on the program’s eligibility criteria. The LOCs visit all households and decide who qualifies. The list is then sent to Nairobi, and enumerators return to households to further prioritize households according to a ranking system. x In Mozambique, the applicant applies to a local intermediary or permanente (who is chosen by the community and appointed by National Institute for Social Action (INAS). The permanente completes an application form for the candidate and sends the application to INAS for approval. INAS then visits the candidate to check his or her socioeconomic information. The targeting performance of these programs was assessed by examining the share of actual participants who were in the target group. For Malawi, Kenya, and Mozambique, the target group is the poorest 20 percent of the population. An assessment of the targeting performance of each program shows that each of the three programs have a targeting performance that is better than the mean of 122 programs assessed in a study of targeting outcomes conducted by Coady et al (2004). Among the subsample of cash transfer programs assessed by Coady et al. the mean score is 1.80 – both the Malawi SCT and Kenya CT-OVC have scores that are higher than this, while the Mozambique PSA has a score that is only slightly lower at 1.73. Source: Handa et al (2012) C. Targeting in the Context of Swaziland 195. Little solid evidence is available regarding the efficiency of different targeting methods in Swaziland so it is difficult to judge which method/s work better than others. In this section, we have attempted to gauge the potential and relevance of each targeting method in Swaziland. 196. Means Testing – Very few people in Swaziland currently receive Public Assistance benefits and these benefits are means-tested. After reviewing the targeting documentation and considering the small number of social workers employed in the entire safety net, we can conclude that this approach is neither systematic nor transparent. The large size of the informal sector makes it very difficult to assess household income and income itself is generally considered to be an inferior indicator of actual household welfare, especially in countries like Swaziland with large non-monetized subsistence agriculture sectors. The small number of beneficiaries served relative to need and the lack of any well documented criteria for eligibility in this method make it vulnerable to subjectivity and political interference in the decision-making process, and too few households receive benefits to make it possible to measure these errors reliably. 197. Categorical Targeting – This is one of the most common targeting methods in Swaziland, which makes sense as there are particular categorical groups in Swaziland that seem to be more vulnerable than others. First, the elderly and the disabled are a small share of the population (6 percent and around 2 percent of the population respectively, with some definite overlaps). &KLOGUHQ FRPSULVH D PXFK ODUJHU VKDUH RI WKH SRSXODWLRQ ௅ DERXW  SHUFHQW  +RZHYHU households with many children (five or more children) are noticeably poorer than households with fewer children and comprise a smaller share of the population. Orphans are not that much poorer than other children and represent a larger share of the population (10 percent) than elderly or disabled people (simulations for categorical children’s benefits with perfect targeting are presented in Chapter 5). An alternative would be to combine categorical targeting with geographic targeting (based on poverty maps), which would make it easier to reach the poorest within a certain categorical group. 61 198. Proxy Means Testing – Although it is possible to create a PMT formula using the latest SHIES data (see examples in Annex Tables 5.1, 5.2, and 5.3), the errors associated with these formulas are not insignificant. 81 It is also worth highlighting that the targeting formula derived from the SHIES is more plausible for use in urban areas than in rural areas. 82 Therefore, if policymakers wish to consider using the PMT in Swaziland, it will be necessary to develop a more reliable PMT formula for rural areas. A possible interim step might be to use data on rural livelihoods taken from other sources such as the VAC surveys to derive a better formula. Given that it can be challenging to identify variables that easily distinguish the welfare of one household from another and the fact that it is administratively demanding to implement a PMT, the government would need to carefully assess the expected costs and benefits of a PMT in order to decide whether to adopt this method. 199. Geographical Targeting – This is used in some programs, particularly food distribution and community-based projects implemented through the MPP. Geographical targeting has the potential to increase targeting efficiency in Swaziland. As mentioned in Chapter 2, the CSO recently developed poverty maps using the 2009-2010 SHIES and the 2007 population census. These maps indicate that the poverty level differs substantially between districts within the area around any given tinkhundla. In addition, other small area statistical databases are available (for example, VAC livelihoods mapping and NERCHA NCP mapping). Depending on the objective of a specific program, it may also be worth exploring how to combine information from these different sources to increase the accuracy of geographic targeting. The poverty maps produced by the CSO could be used to support the geographic targeting of safety nets. 83 These poverty maps disaggregate poverty rates into 195 districts, which show fairly sharp differentiation across the country. There are about 5,000 to 6,000 residents in each district. Using the poverty mapping technique, it would be possible to estimate which districts include the poorest populations (Annex 4). 200. Self-targeting – This could be used in Swaziland, for example, in a public works intervention. However, it is likely that such a program would be over-subscribed given the country’s extremely low employment rates and pervasive and relatively homogenous rural poverty as well as the general lack of cash in rural areas. When demand exceeds the supply in public works programs, one commonly used approach is to implement a lottery (run in a transparent manner) where registration is open to everyone who is interested in participating and the lottery decides who among all of these interested participants may participate in the program. Alternatively, applicants could be offered a limited period of participation in the program, such as a maximum of a certain number of months, to enable more beneficiaries to benefit overall. 81 We used ordinary least squares (OLS) as this is better known to policymakers than probits and logits. Probits and logits were also used and yielded similar results (which are available on request). The derivation of the PMT is discussed in Annex 4. 82 The urban formula is more plausible in that the coefficients are significant and the expected sign makes rough intuitive sense. The R-square of 36 percent is somewhat higher than many typical PMT formulations. However, only 12 percent of total number of the poor lives in urban areas. 83 CSO (2011b) 62 201. Community-based Targeting – Community-based targeting is used in the OVC Bursary Program and is a potentially efficient way to reach the poorest in Swaziland. Community members themselves are usually the most knowledgeable about those in their communities who are most needy, which means that they can play an important role in targeting. Community-based targeting is sometimes used in combination with a PMT, where community members are used to verify the list of beneficiaries created using the PMT formula. However, community-based targeting would have to be implemented more systematically than is currently the case in order for it to work effectively. 202. Many programs use a combination of two or more targeting methods in an effort to target intended beneficiaries as accurately as possible. Preliminary findings from an ongoing review of the social safety nets in 13 African countries indicate that a wide range of targeting mechanisms are used, with programs often using a combination of more than one. The methods that are most commonly used in the countries in question are geographic targeting (about 50 percent of programs), self-targeting (about 33 percent), community-based or categorical targeting (about 26 percent each), and (proxy) means tests (around 10 percent). However, 60 percent of programs combined at least two methods. 84 For example, the Lesotho Child Grant Program combines community targeting with PMT verification (Box 8), which is an approach that might be appropriate for Swaziland. However, the Government of Lesotho is still evaluating this approach, and any potential expansion will be based on lessons learned from this initial pilot. Box 8: The Child Grant Program and the National Information System for Social Assistance (NISSA) in Lesotho The Child Grants Program (CGP) is an unconditional cash transfer targeted to poor and vulnerable households in Lesotho. The primary objective of the CGP is to improve the living standards of orphans and vulnerable children (OVC) so as to reduce malnutrition, improve their health status, and increase school enrollment. In order to do so, it provides a regular transfer to poor households with children, selected through a combination of an objective proxy means test (PMT) and community validation. The targeting for the CGP follows several steps. Following community mobilization and the formation of Village Assistance Committees (VAC), a door-to-door census was conducted to collect information that would inform the first stage of selection, which was the application of a PMT to rank households in five different poverty groups (from poorest to richest, called NISSA1 to NISSA5). Some categorical filters were also added at this stage, including the requirement that only households with children between 0 and 18 be included in subsequent targeting steps. The next phase included making the lists of all households registered in the census available to the VACs and asking them to confirm that the poorest households on the list were indeed the poorest in the community. The final list of selected households consisted of those that were PMT-eligible and had been validated by the VAC. A recent report that reviewed the CGP’s targeting, including the NISSA, found that households eligible through the CGP targeting process were significantly more likely to be poor (74 percent) than those that were not eligible (43 percent). Moreover, two standard measures of targeting effectiveness are inclusion errors and exclusion errors. In the evaluation areas, the poverty rate was 50 percent, but program coverage was only 22 percent, leading to high exclusion errors; the study found that 67 percent of poor households were not included in the program. The report 84 World Bank (2012). The countries included in the draft report are Benin, Botswana, Burkina Faso, Cameroon, Ghana, Kenya, Liberia, Malawi, Mali, Mozambique, Niger, Tanzania, and Togo. 63 identified several reasons for this poor targeting performance: (i) the PMT model was estimated using nationally representative data and was not designed to reflect local differences in the poverty profile; (ii) the dataset used for the estimation of the model (HBS 2002/03) was outdated and the quality of the data was poor, with inconsistencies, outliers, and difficulties in data processing; (iii) partly as a consequence of this poor data quality, a series of discrete models rather than a continuous model was implemented to predict correlation with consumption expenditure, leading to a significant loss of information and precision in the estimation; and (iv) the micro dataset used for the estimation of the model did not contain information about key assets or income flows (such as pensions). The government has a vision for expanding the NISSA into a national system that includes all social assistance programs in Lesotho. It is currently in the process of developing a road map for this expansion and is taking into considerations the findings of the recent report. The government will carefully analyze the current NISSA and ensure that the new design is in line with the government’s strategy for social protection and social assistance programs, as well as ensuring that the design of the program allows for adequate data collection and system maintenance. Source: Luca Pellerano et al (2012) D. Conclusions 203. How well a certain targeting method works depends greatly on how well it is implemented. A 2004 study of targeting experience worldwide concluded that self-targeting through a low wage rate, geographical targeting, and means testing are the most effective. 85 The study also found that PMT, community-based targeting, and targeting of children can be successful but are generally less effective. However, the key message of the study was that the correct implementation of any targeting method is more important than the targeting method chosen, with 80 percent of the variation in targeting effectiveness being due to differences within methods and only 20 percent to the choice of method. 204. There is need for more solid evidence about which targeting methods work best in the context of Swaziland. Given the lack of necessary data, we were not able to do a thorough analysis to determine if any particular program was well targeted. Therefore, it is important for the government to start testing different methods (or combinations of methods) in a systematic way in order to learn lessons that are relevant to the Swazi context. 205. The little evidence that exists indicates that the safety net programs in Swaziland are not particularly well targeted to the poorest. In some cases, the program itself was not designed to do so (the School Feeding Program), and in other cases the targeting was not correctly implemented (the OAG). By making adjustments to some existing programs, it may be possible to target the poorest more efficiently, although this will not be feasible for all programs. For example, it may not be desirable to target the School Feeding Program to the poorest, since it would be difficult to exclude some children and not others. Geographic targeting might be a better possibility if there are areas that are much worse off than others. Other programs, such as the OAG, could be better managed to target those who are most in need. 85 Coady et al (2004) 64 206. A combination of targeting approaches would seem to make sense. We noted in Chapter 2 that the two major correlates of poverty are rural location and the existence of dependents, meaning children and the elderly. Most people in Swaziland live in rural areas (76 percent), and children under 18 comprise 45 percent of the population. Neither of these two indicators would be useful for targeting benefits to the poorest because they are too common. Therefore, one approach could be to combine categorical targeting of children with geographic targeting to the poorest areas (using the poverty maps). Census data could be used in addition to the poverty maps to produce a “dependency map” where districts with the highest numbers of children and elderly could be given priority when providing safety net benefits. Another approach might be to combine geographic targeting using poverty maps with a PMT. Under this approach, programs would provide benefits to all households (within a certain categorical target group) in the very poorest areas, while using a PMT in the areas that are not as poor. This approach has been successfully used in El Salvador and Honduras. 86 207. If the government wants to explore the possibility of using a PMT, further work will need to be done to develop a relevant PMT. The SHIES data make it clear that a PMT would not be an effective way to target the poor in rural areas, where 88 percent of the poor are located, as consumption is distributed evenly and rural living standards are homogeneous. It might be possible to use the VAC surveys in conjunction with the SHIES to devise a PMT that would work in the rural areas of Swaziland. A PMT would most likely need to be combined with an approach where the community is involved to verify the beneficiary lists. 208. Self-targeting could be used in a potential public works program but would probably have to be combined with another method. Due to the high likelihood that demand for employment in a public works program would exceed the supply of available places, a lottery approach would be a feasible way of selecting participants among those who have expressed an interest in participating. 209. Some countries have developed a single unified registration system covering all safety net programs, an idea which the government could consider adopting. A unified registration system would increase the efficiency, effectiveness, and transparency of targeting. It would streamline the application process for safety net programs and would use objective criteria, screening instruments, and procedures for approval. A less ambitious but probably more sustainable idea would be to create a unified registry of beneficiaries of all safety net programs to ensure that there are no overlaps or duplication of support among these programs. It is worth exploring both of these options further to determine if they might be feasible and desirable in the context of Swaziland. 86 Ripani and Tejerina (2008) 65 CHAPTER 5: TOWARDS EFFECTIVE AND RESPONSIVE SOCIAL TRANSFERS 210. The objective of this study was to lay out viable options for making the safety net more effective and responsive, and these are presented in this chapter. It is important to note that this report does not intend to be prescriptive but rather to inform the policy dialogue. A. A Strategic Role for the Social Safety Net 211. Persistent poverty and inequality in Swaziland provides a strong justification for spending on social transfers. Economic growth is not taking place quickly enough to benefit for a large number of Swazis and, as a result, about 30 percent of the population remains extremely poor. The persistently high poverty and inequality that prevails in Swaziland is an argument for reallocating safety net spending so that a greater share of resources goes to the poor. 212. The role of larger-scale transfers would be to reduce poverty and inequality by supporting the extremely poor while at the same time fostering investments in human and physical assets. Well-designed safety net programs can protect extremely poor households by providing them with small but reliable amounts of cash to meet their basic needs. The safety net can also be productive, in the sense that it combines transfers with investments in human capital (health, nutrition, education, and skills) and/or in improving and maintaining community assets (such as roads, land, and other physical assets). 213. From a fiscal perspective it is not possible to target all poor so it would make sense to focus safety nets expenditures on the extreme poor. In Swaziland, the extreme poor represent roughly one-third of the population, so this group might be an appropriate target group for the safety net. For the most part, these are small landholding households who are stuck in low-productivity agriculture, but this group also includes some urban dwellers. This group also includes (but is not limited to) the seasonally poor who survive on subsistence agriculture for most of the year but who fall into extreme poverty every year during the lean season. 214. Based on our analysis, we believe it would make sense to target poor elderly people and poor children, as well as unemployed people in the rural areas. Targeting poor elderly people makes sense because they are a particularly vulnerable group with few income-earning opportunities. To target poor children makes particular sense as an investment in the country’s human capital development to try to break the intergenerational cycle of poverty. Finally, a safety net targeted to unemployed people in rural areas who lack sufficient income during low season not only would provide them with protection but also act as a counter-cyclical response to different shocks or crises. 215. The most important attributes of poverty that need to be addressed are low human capital development and unemployment. Providing a safety net program targeted to children 66 would address the problem of poor households not investing enough in the health, nutrition, and education of their children, resulting in the children grow up to be low-productivity workers with little earning potential. Unemployment could be reduced by the introduction of a public works program. These safety nets must be designed so that they can be scaled up in response to large-scale shocks such as drought or food price increases and then scaled back down after the crisis is over. 216. Political leaders (as well as society as a whole) are often concerned about the sustainability of safety net programs and the risk of creating “dependency” among beneficiaries, but the Swazi government also recognizes the importance of safety nets for supporting certain groups. These groups include the growing number of orphans and the vulnerable elderly. The government also recognizes that the informal safety net has been substantially weakened by factors such as the HIV/AIDS pandemic. In addition, recent economic growth does not seem to have benefitted the poorest, and safety nets could play an important role in redistributing wealth to those most in need. 217. One of the main challenges moving forward will be to design a safety net that promotes productivity as well as supporting the most vulnerable groups. The concern about creating dependency is an argument in favor of adopting programs that promote investments in human capital as well as programs that provide temporary employment while creating productive assets. As indicated in Box 1 in Chapter 1, safety nets can enable households to take up investment opportunities that they would otherwise miss, both in the human capital of their children and in the livelihoods of the able-bodied adults. They can also help households to manage risk or at least to support themselves and adequately nourish and school their children during hard times without having to resort to selling off the household and business assets on which their livelihoods are based. Safety net cash transfers also constitute an injection of funds directly into the lowest level of the economy, where there are multiplier effects that increase demand and fuel growth opportunities at the village level. 218. Safety nets ought to be counter-cyclical, meaning that they can be scaled up in time of shocks or crisis and down again afterwards. However, so far the Government of Swaziland has taken the opposite (cyclical) approach, not only reducing funding to programs after a crisis but also resorting to delaying and/or reducing its payments of cash transfers to recipients. To have a more long-lasting impact, cash transfers need to be provided in a regular and predictable manner that beneficiaries can rely on. 219. Swaziland needs to make a strong fiscal commitment to its social safety net. This will require the government to identify a core set of programs to which it will then commit predictable and sustainable financing. It also means that the government must commit to paying benefits on time. Paying benefits late or at unpredictable times makes it impossible for beneficiaries cope effectively with risk and reduces the overall effectiveness of the safety net. 67 B. Towards a Social Safety Net Strategy 220. Swaziland needs a fully articulated social safety net strategy. The country currently has an uncoordinated set of safety net programs implemented by multiple agencies and a patchwork of policy documents that cover some aspects of social protection, but these do not form a coherent whole. 221. The safety net strategy should spell out the government’s objectives for its safety net system. It should also specify the resources and programs that will be needed to achieve those objectives and define the arrangements for implementing the strategy. The safety net strategy should include: x Priority target groups x Priority programs x A clear road map for rationalizing the safety net and improving service delivery x An accountability and control framework x A monitoring and evaluation framework x A plan for fiscal sustainability. 222. The first step towards preparing the strategy should be to establish an inter- ministerial committee with representatives from all of the ministries that administer safety net programs. This committee would oversee the preparation and implementation of the Social Safety Net Reform Strategy. At the national level, a multi-sectoral Technical Working Group for mitigating the impact of HIV/AIDS already exists, composed mainly of representatives of institutions that are actively involved in providing impact mitigation services. This Working Group could become the core of the inter-ministerial committee for safety nets. 223. The strategy must be developed with the input of all stakeholders in the sector. Therefore, the inter-ministerial committee should hold widespread consultations and discussions with the NGOs and civil society organizations that play a role in implementing social safety net programs. 224. The government will need to decide whether it would prefer to develop a comprehensive social protection strategy or a social safety net strategy. A social protection strategy would include not only social safety nets but also social insurance (pensions and other contributory benefits). C. Matching Programs to Risks and Vulnerabilities 225. Based on this review, we make the following recommendations regarding the essential elements of a safety net reform strategy. 68 Filling Gaps in the Safety Net 226. Swaziland implements a range of social transfer programs that together attempt to address the various risks faced by the population, but gaps in the safety net remain. The gaps related to the protection of extremely poor children and the unemployed are particularly significant, and there is a need to increase the ability of the safety net to respond to shocks. The question is how best to fill these gaps not only to meet the needs of those who are currently vulnerable but also to reduce their vulnerability in the future. The most critical objectives in this regard must be the reduction of poverty and the promotion of human capital. 227. Swaziland does not have a cash support program for extremely poor households, even households with children. The benefits of giving cash grants to extremely poor households with children have been widely documented in developing countries, including countries in Sub-Saharan Africa. Cash grants have many advantages, including: (i) they can immediately increase the income of these households (thus reducing overall inequality); (ii) they can also immediately increase their consumption; (iii) they increase the amount and improve the quality of the food being consumed and also improve nutritional outcomes among children; (iv) they can promote school enrollment for children; and (v) they can increase household investments in productive assets. 87 228. A grant targeted to the poorest households with children could significantly reduce poverty in Swaziland. For example, a monthly per child grant of E50 targeted to the poorest 30 percent of children would benefit about 173,207 children (including 46,611 orphans) at a cost of about E103.9 million (US$14.8 million) annually. Simulations using the recent SHIES indicate that such a program could lead to a reduction in the extreme poverty headcount from 29 percent to 20.7 percent (Table 19). A grant of E100 per month to this same group could lead to a reduction of the extreme poverty headcount to 12.9 percent but would cost E207.8 million (US$29.7 million) annually. Table 19: Cost Scenarios of a Child Grant Extreme Poverty Total Headcount Total Annual Annual Number Number After Child Transfer Transfer of of Grant Cost Cost Children Orphans (%) (E millions) (US$ millions) Baseline (No benefit) 486,751 28.8 E20 to the bottom 10 percent 60,424 16,471 28.8 14.5 2.1 E20 to the bottom 20 percent 117,736 31,817 28.8 28.3 4.0 E20 to the bottom 30 percent 173,207 46,611 25.8 41.6 5.9 E50 to the bottom 10 percent 60,424 16,471 28.8 36.3 5.2 E50 to the bottom 20 percent 117,736 31,817 27.9 70.6 10.1 E50 to the bottom 30 percent 173,207 46,611 20.7 103.9 14.8 E100 to the bottom 10 percent 60,424 16,471 27.9 72.5 10.3 E100 to the bottom 20 percent 117,736 31,817 21.2 141.3 20.2 87 See, for example, ODI (2009), Gertler et al (2006), Devereux (2002), and DFID (2006). 69 E100 to the bottom 30 percent 173,207 46,611 12.9 207.8 29.7 E200 to the bottom 10 percent 60,424 16,471 21.0 145.0 20.7 E200 to the bottom 20 percent 117,736 31,817 11.6 282.6 40.3 E200 to the bottom 30 percent 173,207 46,611 3.3 415.7 59.3 Source: Authors’ calculations using SHIES 2009-2010 229. The safety net could do a better job of promoting human capital by reallocating scarce resources from social safety net activities that do not contribute much to the welfare of the poor to interventions that would promote human capital among, for example, Swazi children. An example of such a program would be a cash transfer that gives recipients an incentive to send their children to school or take them to health facilities. Such a program could include conditions where recipients are required to fulfill the conditions or receive a penalty for non-compliance or so-called “soft” conditions where recipients are simply encouraged to comply with the conditions but are not penalized for non-compliance (see Box 9). 88 230. The Child Grant to be piloted by the Deputy Prime Minister’s Office (co-financed by the EU and the World Bank) will provide an opportunity to evaluate the best way to design and implement such a transfer. 89 Over 20 developing countries have chosen to provide such cash transfers. The pilot Child Grant in Swaziland could be the basis on which a national Child Grant could be built. Box 9: Conditional Cash Transfers Conditional cash transfers (CCTs) are now implemented in over two dozen developing countries. Almost every country in Latin America has a CCT. There are large-scale programs in Bangladesh, Indonesia, and Turkey and pilot programs in Cambodia, Malawi, Kenya, Morocco, Pakistan, and South Africa among others. CCTs make beneficiaries fulfill specific conditions in return for receiving the grant. These conditions are usually designed to promote the human capital development of their children and can include requiring parents to send their children to school on a regular basis or to take them for preventive health care visits. Some programs require mothers to participate in health education classes. Thus, CCT programs take an integrated approach to poverty reduction by balancing goals of social assistance and human capital formation. The Benefits CCTs have had positive effects on poverty and household consumption. CCTs have reduced the national headcount poverty rate by 8 percent in Ecuador and Mexico, by 4.5 percent in Jamaica, and by 3 percent in Brazil. While these effects are modest, they reflect the fact that the pre-transfer incomes of participants were well below the poverty line. As a result, although most beneficiary households are still poor, the intensity or depth of their poverty has been significantly reduced. Households that receive CCTs have been found to consume more food of higher quality than households that do not receive the transfer but have comparable overall income or consumption levels. (It is important to note that this has also been observed among recipients of unconditional cash transfer programs.) CCTs have also helped households to smooth their consumption during shocks. Evaluation studies have found increased use of preventive health care, including vaccinations, and greater use of health centers, which has resulted in reduced morbidity rates 88 “Soft” conditions can be used in an environment where monitoring the conditions would be challenging or where there may not be sufficient services available to all participants in the program. 89 The pilot is currently in the design phase, and the specific conditions (or soft conditions) to be used for the pilot will be agreed upon during the design of the project. 70 in several countries. There is also some suggestion that CCTs have contributed to increases in children’s height and improvements in their health status. In Mexico, a reduction in maternal and infant mortality has been reported. CCTs have resulted in increased school enrollment rates among participating households, particularly among the poorest children, and improved grade progression. There is emerging evidence that CCTs may lead to improved test scores among beneficiary children. There has also been a significant decline in child labor among participating households. The Cons Some argue that conditions are not necessary because demand for health and education increases with wealth so families automatically increase their consumption of these goods when provided with additional cash. Also, imposing conditions regarding school and clinic attendance can only work if these services are in sufficient supply and this is not the case in most countries. Even when supply is available, their poor quality can mean that using them will not lead to automatic increases in human capital. Another argument against conditional programs is that those who do not fulfill conditions are typically the poorest and most in need. Also, monitoring recipients’ compliance with the conditions requires administrative capacity and increases the cost of implementing the cash transfer program. Conditions reduce the net benefit of the transfer to individuals since the time costs of complying can be high and may be especially onerous for women who are often responsible for health and schooling-related decisions within the household. Therefore, some programs adopt so called “soft” conditions where recipients are encouraged to fulfill the conditions, but are not penalized for non-compliance. Sources: Fiszbein and Schady (2009), Baird et al (2009), and World Bank (2009) 231. Public works programs can play an important role in providing income support to workers experiencing temporary and chronic unemployment or underemployment. Public works programs also tend to be more politically and socially acceptable than other forms of transfers because they are perceived to finance productive investments and are not seen as a handouts. Furthermore, these programs can be scaled up and down in response to cyclical needs (due to shocks and other form of crisis) and, if well-designed, can create physical assets that also raise longer-term incomes in the areas affected and contribute to rural development (see Box 10). In Swaziland, investments in asset creation have declined as a result of the fiscal crisis, and a feasibility assessment would be needed to determine the extent of the need for the creation and/or rehabilitation of community assets and the maintenance of roads and public buildings. A pilot public works program could then be implemented from which policymakers could learn lessons about what works in Swaziland, which would then inform the design of any expansion of the public works program. Box 10: The Productive Safety Net in Ethiopia The Productive Safety Net Program (PSNP) covers about 7.6 million rural people in 290 woredas (districts) in eight regions. This is roughly 10 percent of the population and one-third of rural woredas. The main component of the program is labor-intensive public works in which the chronically food-insecure are employed on rural infrastructure projects such as road construction and maintenance, small-scale irrigation, and reforestation. The second component is an unconditional transfer of cash or food to vulnerable households with no able-bodied members who are capable of participating in public works projects. The objective of the program is to provide households with enough income (cash and/or food) to fill their food gap and thereby prevent them from having to sell their household assets while at the same time building assets in the community. The majority of beneficiaries (83 percent) participate in public works and the remaining 17 percent who are labor-constrained receive the unconditional transfer. The annual program budget is around US$347 million, 88 percent of which consists of transfers, with about 60 71 percent provided in cash. This represents approximately 1.2 percent of GDP. Roughly 80 percent of the transfers are delivered through public works and 20 percent through the unconditional transfers. The program is geographically targeted to food-insecure areas using community and administrative targeting within areas. A community committee is set up by the woreda to select potential beneficiaries. This committee reviews the beneficiary lists, amends them as necessary, and endorses them. Appeal committees exist to handle any targeting complaints. The amount transferred is the same whether a family is participating in public works or receiving the unconditional transfer. Each person is entitled to the equivalent of five days work per month for six months of the year. The payment is 3 kilograms of grain or 10 Ethiopian Birr per day (about US$0.60). The average participating household receives about US$137 per year, which is equivalent to about US$2.28 per person per month. The Ethiopian PSNP was designed as a way of shifting away from emergency food distribution. However, about 48 percent of beneficiaries receive only food, 36 percent receive a mixture of food and a cash payment, and 17 percent receive cash only. Source: Wiseman (2010) Strengthening Existing Programs 232. Any reform of the safety net will also need to take into consideration how to build on the programs that are already in place. As discussed in Chapter 3, there are a number of important in-kind and cash transfer programs in Swaziland, and several of them could continue to play an important role in the overall safety net. There is scope for making them operate more efficiently simply by addressing some of the operational challenges highlighted in this report. 233. Introducing some form of targeting to the OAG would make the program more pro-poor and free up resources for other safety net programs. 90 Even though the OAG adds little to the consumption of the non-poor, almost one-third of benefits go to non-poor households. Enforcing the currently unenforced income exclusion rule (E1, 000 per month adjusted for inflation since the program was introduced) would reduce OAG expenditures significantly. It is impossible to determine how many beneficiaries would be excluded if the income rules were strictly applied. However, targeting the OAG to the extreme poor would mean that as many as two-thirds of existing beneficiaries would be removed from the program, and this is likely to be unacceptable from a political economy standpoint. Table 20 presents estimates of the costs involved in various scenarios if the OAG were to be targeted to the elderly poor. Targeting the poorest 60 percent would exclude about 30 percent of existing beneficiaries and reduce the average consumption of excluded households by only 3.5 percent, so this is likely to be a politically acceptable reform. While it would be easy to apply the means test to recipients of the DSW’s Military Pension and civil service pensions, it would be a bigger challenge to determine the incomes of former private sector workers or worked in the informal sector. 90 While the study recognizes that political economy is an important aspect of social protection, assessment of political economy of social safety nets is beyond the scope of this study. 72 Table 20: Scenarios for Targeting the Old Age Grant Current Reduction in Household Beneficiaries Consumption - Scenario Excluded Estimated Cost Average Range % E US$ % % Target Poorest 80% 11.2 89.1 12.6 1.5 1-2 Target Poorest 60% 28.1 72.1 10.2 3.5 1-7 Target Poorest 40% 52.1 48.0 6.8 7.1 1-16 Target Poorest 30% 64.4 35.7 5.1 8.7 1-18 Source: Authors’ calculations using SHIES 2009-2010 234. Now that the government has introduced free primary education, the OVC Education Grant will not be needed for primary school students, but there will still be a need for the program for secondary school students. Several operational challenges need to be tackled urgently, including weak targeting and monitoring systems and the absence of clear operational policies and procedures at the school level. It is important to note that improving the targeting of the education grant would mean that the School Feeding Program would also be better targeted since the education grant covers the student share of the School Feeding Program. The level of the OVC benefit needs to be reconsidered, especially for secondary school students as the grant covers less than 30 percent of average secondary school fees. 235. The School Feeding Program is one of the larger safety net programs, but it is facing implementation and financing challenges. School feeding programs can have significant benefits when they are well-designed and delivered including alleviating short-term hunger, increasing school enrollments, reducing micronutrient deficiencies, and increasing community involvement in schools. 91 However, there is currently no evidence of the impact of the SFP in Swaziland on attendance, learning, and other outcomes or of how efficiently it is being implemented. Therefore, both process and impact evaluations of the SFP are needed. The capacity of the MOET to monitor and supervise the program also needs to be upgraded. As currently financed, the SFP is not sustainable. 92 Therefore, it is imperative to assess the full costs of the program to the government and to develop a plan to secure government financing for the entire SFP. The possibility of recovering a higher proportion of costs from students should also be explored. 236. NCPs provided an important emergency response for young children during the early days of the HIV/AIDS pandemic, but their role as a long-term safety net needs to be carefully considered. NCPs now provide a range of services, including feeding, medical screening, psycho-social support, and early childhood care and development. However, NCPs depend on donated food and are unsustainable from a fiscal perspective. They may also be unsustainable from a community perspective as they rely on community management, volunteers, maintenance, and donations. It is also of paramount importance to establish an institutional home for the NCPs. 91 Bundy et al (2009) 92 The secondary school part of the program is funded by the Global Fund (for commodities) and indirectly by the government via payment of school fees through the OVC Education Grant. The primary school component is funded by the government – directly via the procurement of commodities and indirectly via the payment of school fees under free primary education and through the OVC Education Grant. 73 237. Other programs should be carefully reviewed with an eye towards increasing their efficiency gains and/or terminating those that no longer seem to make sense. For example, interviewees reported leakage in the Phalala Fund due to poor oversight and ineffective cost control mechanisms. Similarly, the rationale for allowing the Military Pensions for World War II veterans to be inherited by wife after wife is questionable. The Military Pension is three times higher than the OAG, and beneficiaries can receive both grants. On equity grounds, it would make sense to limit beneficiaries to only one government-provided cash grant. Also, the government’s main response to covariate shocks is food distribution by the NDMA during the lean season. However, there is considerable international evidence that predictable cash transfers are more effective in reducing predictable annual food insecurity than emergency aid. Gains in poverty reduction are more likely if the government were to reallocate resources away from providing food to provide predictable cash-based safety nets (in the form of cash transfers or public works). D. Options for Sustainable Social Transfers 238. The government needs to be careful to ensure that any expansion of safety net spending is affordable while at the same time taking into account the potential future cost of not having an adequate safety net. The government currently spends about E584.4 million (US$82.4 million), which, if evenly distributed among the population, amounts to about US$69 per person per year. If perfectly targeted to the extreme poor, this would provide a transfer of US$279 per person per year, which is equivalent to 77 percent of the annual extreme poverty line (and equal to about US$1 per day). It is also worth highlighting that it would require about E74.6 million (US$10.5 million) in transfers to the extreme poor to cover the poverty gap, which only represents about 0.27 percent of GDP. In reality it is not possible to perfectly target the extreme poor, but these are indicative figures to illustrate that, if carefully designed, a safety net is more than affordable for Swaziland. However, it will be vital to select the most cost- effective and fiscally sustainable instruments. To the extent possible, the government should try to reallocate its existing expenditures to ensure that they are used more efficiently. 239. A well-crafted safety net should serve the dual role of protecting households against poverty and promoting their independence and should consist of a range of programs that together address the critical risks and vulnerabilities faced by the population. Based on the analysis in this report, it would make sense to have the set of core safety net programs that is shown in Table 21, which can be combined with other type of complementary programs (including, for example, the NCPs, agriculture subsidies, and active labor market programs). Table 21: Core Safety Net Programs Existing New x Old Age Grant x Child Grant (with initial pilot) x School Feeding Program x Public Works (with initial pilot) x OVC Education Grants 240. Rough calculations indicate that this set of core safety net programs would be affordable for Swaziland. Indicative costs of these core programs are provided in Table 22. It 74 is important to note that these figures are based on very rough estimations and are only intended to give indicative figures of the cost and affordability of this proposed core set of safety net programs. They should not be used for planning or budgeting purposes for which more sophisticated simulations would need to be carried out. As shown, the estimated total expenditures on the core programs would amount to about E428.6 million (US$60.5 million). This represents 4.3 percent of central government spending and about 1.6 percent of GDP. Swaziland spent an estimated E584.4 million on its safety net in FY2010/11. Taking the core programs as proposed and assuming that no changes are made in any of the other existing safety programs not included in the core, the new estimated total would be E607.5 million (US$85.7 million) or 6 percent of central government spending and 2.2 percent of GDP. With only a 10 percent reduction in expenditures on the existing “non-core” programs, the proposed options would be essentially budget neutral. Table 22: Indicative Costing a Core Set of Safety Net Programs E US$ Individual Beneficiaries (million) (million) (thousands) a/ OAG 82.9 11.7 41.6 b/ Child Grant 119.5 16.9 173.2 c/ OVC Education Grant 183.4 25.9 87.7 d/ School Feeding Program 27.9 3.9 264.7 e/ Public Works 14.9 2.1 49.0 428.6 60.5 Notes: a/Targeted to the bottom 60 percent of the elderly with a benefit equal to E200/month/elderly person, plus 15 percent operating costs. b/ Targeted to the poorest 30 percent of children, with a benefit equal to E50/month/child, plus 15 percent operating costs. c/ Assumes beneficiaries and expenditures at FY2010/11 levels. d/ Assumes continued expenditures by the DSW at same levels as in FY2011/12 (E159.5 million), plus 15 percent operating costs, but does not include EU capitation grants (E24.8 million) or MOET bursaries (E56.4 million). In absence of information on the flow of students from primary to secondary, we assume the same number of students as in 2011 with some in primary grades uncovered by FPE and some secondary grades. We also assume that any efficiency gains from better targeting are used to increase the amount of the grant, particularly for secondary students. e/ Targeted to 50 percent of extremely poor households and calculated based on extreme poverty gap of 9.8 percent with an extreme poverty line of E215 (that is E21.07 per adult equivalent per month), assuming an average of two people per household for six months and with 50 percent going to wages and 50 percent to operating costs and material. This would provide a benefit of E1,793 (US$253) per household per year. 241. It is important to note that, although NCPs are not included in the set of core programs, we feel that it will be necessary to assess the role played by NCPs in the future to answer certain questions. For example, should NCPs continue to provide meals and, if so, to which children? What other services should NCPs provide? An institutional home for the NCPs is needed, and their sustainability will need to be considered. 242. It is also important to highlight the lack of active labor market programs for vulnerable youths. Stakeholders noted that poorly educated and unemployed young people are a particularly vulnerable group in Swaziland, especially those made further vulnerable by the loss of parents and other role models. While it was beyond the scope of this report to assess the circumstances faced by young people in Swaziland or to thoroughly assess active labor market 75 programs, both assessments will be needed to inform the design of programs to fill critical gaps in youth services. 243. Strengthening administrative capacity and increasing administrative efficiency would enhance the sustainability of these core safety net programs. The DSW needs to be more proactive in finding alternatives to its current system of cash payments in order to reduce the heavy demands on staff time. An improved MIS is being developed for the OVC Education Grant, but one is also needed for the OAG. The MIS to be developed for the Child Grant could be designed in such a way that it could eventually serve as the MIS for all DSW programs, including the OAG. Recordkeeping and monitoring in programs for children are impeded by the fact that many children do not have a personal identification number (PIN). The government is in the process of issuing these numbers, but this exercise needs to be given a higher priority and all bottlenecks to giving PINs to vulnerable children should be removed. Being able to identify beneficiary children by their PINs would reduce leakage in programs run by the DSW and also enable the MOET to track all children in the school system and not just in the schools themselves. The DSW’s MIS could serve as the core of a central beneficiary registry that would establish a national targeting system for safety net programs. This would be a useful planning and management tool that would also promote accountability and control. Such a registry could be used in other kinds of safety net programs as well. Merging small programs, such as the proposed Child Grant and Young Heroes, would also reduce administrative overheads. In this theoretical harmonized program, Young Heroes would retain its revenue-generating role, and Young Heroes beneficiaries would be identified as such by the MIS to ensure accountability to donors. Making the households that receive the Child Grant automatically eligible for the OVC Education Grants and health fee waivers would also reduce the transaction costs involved in targeting. 244. None of Swaziland’s safety net programs seem to have adequate information systems, and there appears to be little emphasis on monitoring basic statistics, such as the number and characteristics of beneficiaries and the amounts of cash disbursed. Developing management information systems for key programs would make it possible to monitor whether programs are reaching their targets and achieving their objectives Each program should be able to collect and manage data on, for example, the coverage of the target group, the benefits provided, the administrative costs and the costs of the benefits delivered, and program outcomes such as school enrollment and attendance, clinic attendance and vaccination coverage, and employment and earnings outcomes for participants in active labor market programs. Because poverty and labor force surveys are implemented only infrequently, these data are not currently sufficient for monitoring the performance and outcomes of safety net programs. The capacity of the Central Statistical Office will need to be strengthened to support more regular implementation of labor force and poverty monitoring surveys. 245. To increase implementation and coordination capacity, key institutions will need to be strengthened. For example, as highlighted already, some of the key implementing ministries do not have enough staff to implement the programs efficiently, particularly the Deputy Prime Minister’s Office, which is the lead agency for social protection in the country. 76 246. As noted above, the government needs to make a strong fiscal commitment to its social safety net. This is not impossible, even in the current fiscal context. Capital expenditures, including externally funded and local capital projects, currently account for about 21 percent of social safety net spending and for a declining share of the safety net budget. This suggests that it will be fiscally feasible for the government to gradually begin to fully fund the safety net through the recurrent budget, which is essential to ensure its sustainability. At the same time, the government needs to commit to carrying out regular financial and process audits of key programs to enhance efficiency and accountability. E. How Targeting Should Be Approached 247. The existing safety net programs are not well-targeted to the neediest. Some of these existing programs were not designed to reach the poorest (or neediest), while others were designed to do so but have not been well implemented. Revising the targeting methods used in existing programs may be challenging for several reasons, including the difficulties involved in excluding some of the current beneficiaries. Nevertheless, we highly recommend doing so, particularly given the fiscal constraints facing the country and the importance of trying to reach those most in need. As already mentioned, an effort should be made to enforce the targeting of the OAG, and there seems to be general consensus about doing so, particularly since the program was specifically designed to target the needy among the elderly population. The government should also make an attempt to improve the targeting of the OVC Education Grant, ensuring that support goes to those students who are most in need of it. 248. Targeting orphans was an accepted approach in the earlier days of the HIV/AIDS pandemic, but singling out orphans is no longer the recommended approach. In the context of Swaziland, data indicate that orphans are not necessarily any poorer than children in general. International experience has shown that providing support for orphans can stigmatize them and can create jealousy among children who are needy or among non-orphans living in the same households as the orphans who are receiving benefits. As discussed earlier, singling out orphans in Swaziland means that many children who are not needy receive assistance while many extremely needy children fall outside the safety net. One approach that governments are taking in some countries is to define OVCs to include children who are poor and vulnerable but who may not necessarily be orphaned. 249. Programs should be designed to target the poorest and most vulnerable people. This might be done using a variety of different approaches depending on the objective of each program, but experience has shown that using a combination of targeting approaches makes the most sense. One method would be to combine geographic targeting (using the poverty maps) with geographic targeting. The actual number of people living in a poor area could also be taken into account in the geographic targeting phase. This could then be reinforced by using categorical targeting to direct the benefits to certain priority groups chosen by the government. Another approach might be to combine geographic targeting using poverty maps with a PMT. The combination of proxy means testing (PMT), community-based targeting, and self-targeting is also worth exploring. 77 F. The Way Forward 250. Table 23 outlines some key safety net reforms that are suggested by the findings of this report. Recommendations are phased and show that both implementation and policy improvements are possible at the individual program level. It is recommended that the Inter- Ministerial Committee consider these recommendations when developing a Social Safety Net Reform Strategy for Swaziland. Additional reforms are likely to be identified during the course of national consultations. 78 Table 23: Key Safety Net Reforms in the Short and Medium Term Short-term Strategy Medium-term Strategy (1-2 Years) (3-5 Years) Program Specific Recommendations ߤ Ensure reliable budget allocation to the program ߤ Fully implement targeting plan ߤ Ensure regular payments to beneficiaries ߤ Decide whether to increase eligibility age ߤ Determine targeting criteria, develop plan for implementation and begin implementation of plan ߤ Modernize payment systems Old Age Grant ߤ Upgrade MIS ߤ Introduce mechanism for regularly increasing benefits ߤ Assess the costs and benefits of raising the age of eligibility ߤ Prohibit Military Pension recipients from receiving the OAG ߤ Design pilot ߤ Evaluate pilot ߤ Develop systems ߤ Expand based on findings of evaluation Child Grant ߤ Implement pilot ߤ Secure government financing ߤ Fully implement the issuing of PINs to all children ߤ Improve targeting ߤ Secure government financing OVC ߤ Document policies and procedures ߤ Make beneficiaries of Child Grant automatically eligible for the Education ߤ Upgrade MIS and link it with the Child Grant MIS OVC Education Grant Grant ߤ Strengthen monitoring ߤ Fully implement the issuing of PINs to all children ߤ Strengthen monitoring and supervision ߤ Secure government financing School Feeding ߤ Implement process audits ߤ Redesign program as required based on the findings of process Program ߤ Evaluate the program audits and evaluations ߤ Assess feasibility ߤ Implement pilot ߤ Cost pilot ߤ Evaluate pilot Public Works ߤ Identify financing for pilot ߤ Secure financing for potential expansion ߤ Design pilot ߤ Develop systems Cross-cutting Recommendations Social Safety ߤ Form inter-ministerial committee ߤ Fund core safety net programs from the GOS recurrent budget Net Strategy ߤ Hold national consultations ߤ Develop SSN Strategy Efficiency and ߤ Modern MIS for all programs ߤ Establish central beneficiary registry Accountability ߤ Modernize payment systems for all programs ߤ Merge small safety net programs and remove those that are not ߤ Build the capacity of the DPM consistent with the overall safety net objective 79 ߤ Improve accounting systems ߤ Implement information and education campaigns ߤ Establish appeals and complaints mechanisms ߤ Conduct regular process audits ߤ Document policies and procedures for all programs ߤ Ensure regular payments to beneficiaries Targeting ߤ Develop and pilot objective targeting methodologies ߤ Establish objective targeting criteria for all targeted programs Program ߤ Develop monitoring and evaluation framework ߤ Identify resources to fund more regular poverty and labor force monitoring surveys Crisis ߤ Pilot public works (see above) ߤ Move away from using food aid as a crisis response Response 80 References Auditor General (2011) Performance Audit Report of the Auditor General on the Orphaned and Vulnerable Children Grant. Auditor General’s Office, Mbabane Baird, Sarah, Ephraim Chirwa, Craig McIntosh, and Berk Özler (2009). “The Short-term Impacts of a Schooling Conditional Cash Transfer Program on the Sexual Behavior of Young Women.” World Bank, Washington D.C. Beegle, Kathleen, Deon Filmer, Andrew Stokes, and Lucia Tiererova (2009) “Orphanhood and the Living Arrangements of Children in Sub-Saharan Africa.” World Development, Vol. 38, No. 12, pp. 1727-1746. Blank, Lorraine (2010). “Mauritius: Social Protection Review and Strategy.” World Bank, Washington D.C. Blank, Lorraine and Sudhanshu Handa (2008). “Social Protection in Eastern and Southern Africa.” UNICEF, Nairobi. Blomquist, John, Marijn Verhoeven, and Juan Pablo Cordoba (2001). “Social Safety Nets in Response to Crisis: Lessons and Guidelines from Asia and Latin America.” World Bank, Washington D.C. Bundy, Donald, Carmen Burbano, Margaret Grosh, Aulo Gelli, Mathew Jukes, and Lesley Drake (2009). “Rethinking School Feeding: Social Safety Nets, Child Development, and the Education Sector.” World Food Programme/World Bank, Washington D.C. CGAP (2012). “Social Cash Transfers and Financial Inclusion: Evidence from Four Countries.” Presentation. Consultative Group to Assist the Poor, Washington D.C. Coady, David, Margaret Grosh, and John Hoddinott. 2003. “Targeting Outcomes, Redux.” International Food Policy Research Institute, Washington, D.C. CSO (2008). Swaziland Demographic and Health Survey Report. Central Statistical Office, Mbabane. CSO (2011a). Poverty in a Decade of Slow Economic Growth: Swaziland in the 2000s. Central Statistical Office, Mbabane. CSO (2011b). 2010 Swaziland Poverty and MDG Maps: Region, Tinkhunlda and Major Areas. Central Statistical Office, Mbabane, June. 81 CSO (2011c). Poverty Profile: Swaziland Household Income and Expenditure Survey (SHIES) 2010. Central Statistical Office, Mbabane. CSO (2012). Preliminary Results from the Swaziland Labor Force Survey 2010/2011. Draft. Central Statistical Office, Mbabane. CSO and UNICEF (2011). Swaziland Multiple Indicator Cluster Survey 2010. Central Statistical Office, Mbabane and UNICEF, Mbabane, December. CSO and Macro International (2008). Swaziland: Demographic and Health Survey. Central Statistical Office, Mbabane, and Macro International, Calverton. Davis, Benjamin, Sudhanshu Handa, and Humberto Soto (2004). “Households, Poverty, and Policy in Times of Crisis: Mexico 1992-1996.” CEPAL Review. Vol. 82 (April):191-212. del Ninno, Carlo, Subbarao Kalanidhi and Milazzo, Annamaria. “How to Make Public Works Work: A Review of the Experiences.” World Bank, Washington D.C. Devereux, Stephen (2002). “Can Social Safety Nets Reduce Chronic Poverty?” Development Policy Review 20(5): 657-675. DFID (2006). “Using Social Transfers To Improve Human Development.” Social Protection Briefing, Department for International Development, London. Dlamini Solomon (2007). “Neighborhood Care Points in Swaziland: A Case Study.” REBA/RHVP, Mbabane. Fiszbein, Ariel and Norbert Schady (2009). “Conditional Cash Transfers: Reducing Present and Future Poverty.” World Bank, Washington D.C “Food Security Update” (2010) http://www.sadc.int/fanr/aims/rvaa/Documents/Swaziland/2010%20Swaziland%20Food%20Se curity%20Update,%20February.pdf Gertler, Paul, Sebastian Martinez, and Marta Rubio-Codina (2006). “Investing in Cash Transfers to Raise Long-term Living Standards.” World Bank Policy Research Working Paper 3994, World Bank, Washington D.C. Goldberg, Jessica (2010). “Kwacha Gonna Do? Experimental Evidence about Labor Supply in Rural Malawi.” Draft. Department of Economics and Ford School of Public Policy, University of Michigan. GOS (2010). Children’s Protection and Welfare Bill. Government of Swaziland, Mbabane. GOS (2009). “Neighborhood Care and Development Centers, Ten Year Strategic Plan: 2008- 2018.” Draft. Government of Swaziland, Mbabane.Grosh, Margaret; Carlo Del Ninno, Emil 82 Tesliuc, and Azedine Ouerghi (2008). For Protection and Promotion: The Design and Implementation of Effective Safety Nets. World Bank, Washington D.C. Grosh, Margaret and Emil Tesliuc (2007). “Reducing Fraud, Error and Corruption in Social Protection Programs, Workshop on Fraud and Error Control in Social Protection Programs.” World Bank, Washington D.C. HelpAge International (2009). “Social Pensions Database.” http://www.pension- watch.net/about-social-pensions/about-social-pensions/social-pensions-database/ HISP (2004). Project Report: A Situation Analysis of the Health Information System in Swaziland. Health Information System Program, World Bank, Washington D.C. Imai, Kumiko, Nathi Vilakati, Khetho Dlamini, Robert Fakudze, and Sibongile Hlatshwayo (2009). A Situation and Needs Assessment of Child-headed Households in Drought-prone Areas of Swaziland: Key Findings. UNICEF and Save the Children, Mbabane. IMF (2011a). “Kingdom of Swaziland: Staff Monitored Program—Staff Report; Staff Supplement.” IMF Country Report No. 11/84. International Monetary Fund, Washington D.C., April IMF (2011b). “Staff Report for the 2011 Article IV Consultation.” International Monetary Fund, Washington DC. Marope, Mmantsetsa (2010). “The Education System in Swaziland: Training and Skills Development for Shared Growth and Competitiveness.” World Bank, Washington DC Micro Projects Program (2011). Annual Report 2010/2011. Mbabane. Ministry of Economic Planning and Development (2011). “Economic Recovery Strategy for Accelerated, Inclusive and Sustainable Economic Growth.” Government of Swaziland, Mbabane. Ministry of Health and Social Welfare (2008). 11th Round of HIV Sentinel Serosurveillance in Women Attending Antenatal Care Services in Swaziland: Survey Highlights. Government of Swaziland, Mbabane. Ministry of Local Government and Housing and National Children’s Co-ordination Unit (2009). “Neighborhood Care and Development Centers: Ten Year Strategic Plan 2008-2018 Zero Draft.” Government of Swaziland, Mbabane. NERCHA (2011). Mapping of Neighborhood Care Points, 2010/11. National Emergency Council on HIV and AIDS, National Emergency Council on HIV and AIDS, Government of Swaziland, Mbabane. 83 NERCHA (2010). Swaziland Country Report, UNAIDS. National Emergency Council on HIV and AIDS, National Emergency Council on HIV and AIDS, Government of Swaziland, Mbabane. NERCHA (2009a). Swaziland HIV Prevention Response and Modes of Transmission Analysis. National Emergency Council on HIV and AIDS, Government of Swaziland, Mbabane. NERCHA (2009b). Swaziland National Multisectoral HIV and AIDS Monitoring and Evaluation Framework 2009-2014. National Emergency Response Council on HIV and AIDS, Government of Swaziland, Mbabane. NERCHA (2009c). The National Multisectoral Framework for HIV and AIDS 2009-2014. National Emergency Response Council on HIV and AIDS, Government of Swaziland, Mbabane. NERCHA (2008a). Formative Research Report: Quality, Relevance and Comprehensiveness of Impact Mitigation Services Survey (QIMS) in Swaziland. National Emergency Response Council on HIV and AIDS, Government of Swaziland, Mbabane. NERCHA (2008b). Protocol: Quality, Relevance and Comprehensiveness of Impact Mitigation Services Survey (QIMS) in Swaziland. National Emergency Response Council on HIV and AIDS, Government of Swaziland, Mbabane. NERCHA (2008c). Report on the Piloting of the Quality, Relevance and Comprehensiveness of Impact Mitigation Services Survey (QIMS) in Swaziland. National Emergency Response Council on HIV and AIDS, Government of Swaziland, Mbabane. NERCHA (2006). National Plan of Action for Orphans and Vulnerable Children 2006-2010. National Emergency Response Council on HIV and AIDS, Government of Swaziland, Mbabane. Oversees Development Institute (2009). “Child Poverty: A Role For Cash Transfers?” West and Central Africa Regional Thematic Report. UNICEF, Dakar and Oversees Development Institute, London. Pellerano, Luca, Hurrell Alex, Kardan Andrew, Barca Valentina, hove Fidelis, Beazley Rodolfo, Modise Boipuse, MacAuslan Ian, Dodd Samantha, Crawfurd, Lee. (2012). “CGP Impact Evaluation: Targeting Report.” Oxford Policy Management, Oxford, UNICEF, Masero, FAO, Masero. RHVP (2010). Swaziland Old Age Grant Impact Assessment. Regional Hunger and Vulnerability Programme, Pretoria. 84 Ripani, Laura, and Luis Tejerina (2008). “Targeting of Conditional Cash Transfer Programs in Latin America and the Caribbean.” Inter-American Development Bank, Washington D.C. Handa, Sudhanshu, Carolyn Huang, Nicola Hypher, Clarissa Teixeira, Fabio Soares, and Benjamin Davis (2012). “The Targeting Effectiveness of Social Cash Transfer Programs in Three African Countries.” The Journal of Development Effectiveness. Volume 4, Issue 1, March. Tsela, Sanelisiwe, and Helen Atieno Odido (2007). Swaziland HIV Estimates and Projections Workshop Report. National Emergency Response Council on HIV and AIDS (NERCHA), Mbabane. UNAIDS online database: http://www.aidsinfoonline.org/ UNICEF (2006). “Report on the Assessment of Neighborhood Care Points.” United Nations Children’s Fund, Mbabane. UNDP (2011). Human Development Report. United Nations Development Programme, New York. UNDP Swaziland (2007). Swaziland Human Development Report: HIV and AIDS and Culture. United Nations Development Programme, New York van Stolk, Chriss (2007). “International Benchmark of Fraud and Error in Social Security Systems.” Workshop on Fraud and Error Control in Social Protection Programs, World Bank, Washington D.C. Swaziland Vulnerability Assessment Committee (2010) Swaziland Annual Vulnerability Assessment & Analysis Report. World Food Programme and Regional Hunger and Vulnerability Programme. Swaziland Vulnerability Assessment Committee (2009) Swaziland Annual Vulnerability Assessment & Analysis Report. World Food Programme and Regional Hunger and Vulnerability Programme. Swaziland Vulnerability Assessment Committee (2008) Swaziland Annual Vulnerability Assessment & Analysis Report. World Food Programme and Regional Hunger and Vulnerability Programme. Weiser SD, Leiter K, Bangsberg DR, Butler LM, Percy-de Korte F. (2007). “Food Insecurity is Associated with High-Risk Sexual Behavior among Women in Botswana and Swaziland.” PLOS Medicine, Volume 4, Issue 10, October. 85 Wiseman, Will (2010). “A Short-term Crisis ௅ A Long-term Response: The Case of Ethiopia’s Productive Safety Net Program.” Presentation at Social Safety Net Showcase: Post-crisis Agendas, World Bank, Washington D.C. WFP (2010). “Government of Swaziland’s National School Meals Program – Cost Estimation Report.” World Food Programme, Mbabane. World Bank (2012). “Social Safety Nets in Africa – A Regional Synthesis.” Draft report, World Bank, Washington D.C., May. World Bank (2011). Swaziland “Rural Sector Review: Priorities for the Development of Smallholder Agriculture in Swaziland.” Washington D.C. World Bank (2010b) The Fiscal Dimensions of HIV/AIDS in SADC Member Countries, Washington, DC. World Bank (2009). “Building an Effective and Inclusive Social Protection System in Latin America: Diagnosis and Policy Directions.” World Bank, Washington D.C. World Bank (2008). “Interim Strategy Note: A Framework for Scaling up Support for the Kingdom of Swaziland.” Report No. 42059-SZ, World Bank, Washington D.C. Slater, R. and M. Tsoka (2007). “Social Protection in Malawi: A Status Report.” Overseas Development Institute, London. 86 Annex 1: HIES Methodology 251. The SHIES was conducted in 2009-2010 and involved interviews with 3,167 households with 14,164 individual members. The sample was random and stratified and, using the sampling weights, produced a population total of 1,017,406. The SHIES questionnaire was a standard household budget questionnaire with some customization for Swaziland, specifically questions on land usage to reflect the role of the chief and to ascertain orphan status. 252. A detailed expenditures diary and agricultural production inventory were collected, so that household consumption (including own-consumption of own-produced food) could be estimated. The dataset included the estimated basic needs poverty and extreme poverty lines. Questions were asked about the households’ housing attributes including water, sanitation, roof, floor, and wall construction, and the number of their consumer durables was tallied. Modules on education and health were included. 253. Unfortunately, the questions on employment and government transfers were ambiguous, so there is a degree of uncertainty about the estimates of unemployment and benefit incidence. 87 Annex 2: HIES Poverty Tables Annex Table 24: Swaziland Poverty Measures Poverty Poverty Squared Poverty Gap Headcount Gap (P1) (P2) Rate (P0) Data1 Data1 Data1 Monthly Poverty Line Area of residence Urban 31.1 10.3 4.7 Rural 73.1 36.9 22.3 Total 63.0 30.4 18.1 Monthly Extreme Poverty Line Area of residence Urban 5.8 1.4 0.6 Rural 36.1 12.5 6.0 Total 28.8 9.8 4.7 Note: Calculated from SHIES using ADePT software. 88 Annex Table 25: Poverty by Geographic Regions Poverty Distribution of Distribution of Headcount the Poor Population Rate Data1 Data1 Data1 Monthly Poverty Line Area of residence Urban 31.1 12.0 24.2 Rural 73.1 88.0 75.8 Region Hhohho 60.9 25.0 25.9 Manzini 57.7 31.3 34.1 Shiselweni 68.3 22.2 20.4 Lubombo 69.3 21.6 19.6 Total 63.0 100.0 100.0 Monthly Extreme Poverty Line Area of residence Urban 5.8 4.8 24.2 Rural 36.1 95.2 75.8 Region Hhohho 28.4 25.6 25.9 Manzini 25.2 29.9 34.1 Shiselweni 26.9 19.1 20.4 Lubombo 37.3 25.4 19.6 Total 28.8 100.0 100.0 Note: Calculated from SHIES using ADePT software 89 Annex Table 26: Poverty by Age Groups Poverty Distribution Headcount Distribution of Population of the Poor Rate Data1 Data1 Data1 Monthly Poverty Line Age 0-5 70.1 17.2 15.5 6-14 69.9 27.1 24.4 15-19 68.6 13.8 12.6 20-24 59.2 9.3 9.9 25-29 50.4 7.3 9.1 30-34 49.5 4.7 6.0 35-39 48.3 3.6 4.7 40-44 51.6 3.2 3.9 45-49 54.2 2.7 3.1 50-54 57.1 2.1 2.4 55-59 57.1 2.2 2.5 60-64 71.1 2.2 1.9 65+ 71.1 4.6 4.1 Total 63.0 100.0 100.0 Monthly Extreme Poverty Line Age 0-5 33.1 17.8 15.5 6-14 34.5 29.3 24.4 15-19 32.4 14.2 12.6 20-24 25.2 8.7 9.9 25-29 21.9 6.9 9.1 30-34 20.4 4.2 6.0 35-39 20.0 3.3 4.7 40-44 18.4 2.5 3.9 45-49 22.8 2.5 3.1 50-54 23.7 1.9 2.4 55-59 27.2 2.3 2.5 60-64 30.5 2.1 1.9 65+ 29.8 4.3 4.1 Total 28.8 100.0 100.0 Note: Calculated from SHIES using ADePT software 90 Annex Table 27: Poverty by Household Head's Age Poverty Distribution of Distribution of Headcount the Poor Population Rate Poverty household head's age 6-14 100.0 0.2 0.1 15-19 65.2 0.5 0.5 20-24 49.4 1.9 2.5 25-29 43.6 4.4 6.4 30-34 54.5 8.1 9.4 35-39 51.0 8.9 11.1 40-44 57.6 11.0 12.0 45-49 59.4 9.2 9.8 50-54 66.1 8.9 8.5 55-59 66.8 12.2 11.5 60-64 77.7 11.0 9.0 65+ 76.4 23.6 19.4 Total 63.0 100.0 100.0 Extreme Poverty household head's age 6-14 27.9 0.1 0.1 15-19 31.6 0.5 0.5 20-24 10.8 0.9 2.5 25-29 14.2 3.1 6.4 30-34 27.9 9.1 9.4 35-39 21.1 8.1 11.1 40-44 24.4 10.2 12.0 45-49 28.7 9.7 9.8 50-54 28.5 8.4 8.5 55-59 35.0 13.9 11.5 60-64 38.9 12.1 9.0 65+ 35.0 23.6 19.4 Total 28.8 100.0 100.0 Note: Calculated from SHIES using ADePT software 91 Annex Table 28: Poverty and Education of Head Poverty Headcount Rate Monthly Poverty Line Poverty Rate Literacy Program 84.1 Lower Primary 80.6 Higher Primary 75.6 Secondary 59.7 High School 36.0 Public College 7.2 University 10.0 Private College 11.0 Vocational 42.5 Total Reporting Education 56.3 Monthly Extreme Poverty Line Extreme Poverty Rate Literacy Program 37.7 Lower Primary 40.9 Higher Primary 34.8 Secondary 18.5 High School 12.1 Public College 0.0 University 1.1 Private College 1.4 Vocational 0.0 Total Reporting Education 22.7 92 Annex Table 29: Poverty by Household Head's Status of Employment Poverty Distribution Headcount Distribution of Population of the Poor Rate Data1 Data1 Data1 Monthly Poverty Line Employment status of the household head Employer 64.1 2.0 1.5 Self employed 65.0 43.2 33.0 Public Service 26.0 10.8 20.6 Private Sector 47.3 39.5 41.5 Family Farm/Business 68.8 2.0 1.4 Other 65.7 2.5 1.9 Total 63.0 100.0 100.0 Monthly Extreme Poverty Line Employment status of the household head Employer 23.3 1.9 1.5 Self employed 25.6 43.4 33.0 Public Service 8.0 8.4 20.6 Private Sector 18.5 39.6 41.5 Family Farm/Business 41.9 3.1 1.4 Other 37.0 3.6 1.9 Total 28.8 100.0 100.0 93 Annex Table 30: Poverty and Gender of Head Monthly Poverty Line Gender of the Poverty Percent of Poor Percent of Population household head Male 58.9 47.3 50.5 Female 67.1 52.7 49.5 Total 63 100 100 Monthly Extreme Poverty Line Gender of the household head Male 24.9 43.8 50.5 Female 32.7 56.2 49.5 Total 28.8 100 100 Note: Calculated from SHIES using ADePT software 94 Annex Table 31: Poverty and Household Composition Share in Household Composition Poor Extreme Poor Population children & elderly 78.2 37.9 32.5 no children, no elderly 25.4 5.4 8.9 no children, but elderly 45.2 15.0 1.9 no elderly, but children 60.8 27.7 56.7 Total 63.0 28.8 100.0 Source: Authors’ calculations from SHIES 2009-2010 Annex Table 32: Poverty and Children Under 18 Overall Poverty Extreme Poverty Share of Poor Poor Not Poor Poor No children under 18 28.8 4.9 7.0 93.0 One or more children under 18 67.1 32.9 31.4 68.6 Poverty Rate Childen Under 18 None 28.8 71.2 7 93 One 45.2 54.8 11.6 88.4 Two 50.1 49.9 18.9 81.1 Three 60.6 39.4 26.1 73.9 Four 64.6 35.4 31.5 68.5 Five 83.7 16.3 39.9 60.1 Six 89.9 10.1 43.4 56.6 Seven 82.3 17.7 44.5 55.5 Eight 86.4 13.6 54.2 45.8 Nine 92.4 7.6 57.8 42.2 Ten 79.2 20.8 53.3 46.7 Eleven 85.8 14.2 40 60 Twelve or More 100 0 90.8 9.2 Source: Authors’ calculations from 2009-2010 SHIES 95 Annex Table 33: Poverty and Number of Dependents by Consumption Deciles and Poverty Status Number of Number of Number of Number of Children Orphans Elderly People Decile 1 60,424 16,471 5,274 101,798 Decile 2 57,312 15,346 6,803 102,690 Decile 3 55,471 14,794 7,075 100,772 Decile 4 53,679 16,236 7,337 101,831 Decile 5 53,702 14,849 7,649 101,889 Decile 6 45,798 10,454 7,466 101,515 Decile 7 47,232 12,498 6,154 101,765 Decile 8 43,491 9,160 4,754 101,621 Decile 9 39,098 8,113 4,251 101,904 Decile 10 30,544 4,998 4,705 101,571 Total 486,751 122,919 61,468 1,017,356 Poor 341,334 91,862 43,716 640,669 Extremely Poor 165,898 44,534 18,471 292,592 Note: Decile 1 60,424 16,471 5,274 101,798 Decile 1+2 117,736 31,817 12,077 204,488 Decile 1+2+3 173,207 46,611 19,152 305,260 Source: Authors’ calculations SHIES 2009-2010. Note: Survey estimates will not match census results exactly. 96 Annex Table 34: Poverty and Orphanhood Poverty Distribution Headcount Distribution of Population of the Poor Rate Monthly Poverty Line Non-Orphans 68.6 73.1 74.8 Maternal Orphan 79.8 6.3 5.5 Paternal Orphan 76.3 16.0 14.7 Double Orphan 65.0 4.7 5.0 Total 70.1 100.0 100.0 Monthly Extreme Poverty Line Non-Orphans 24.9 73.2 74.8 Maternal Orphan 35.1 5.7 5.5 Paternal Orphan 40.3 17.4 14.7 Double Orphan 25.6 3.8 5.0 Total 34.1 100.0 100.0 Source: Authors’ calculations from the 2009-2010 SHIES Note: Poverty and extreme poverty rates pertain only to children under the age of 18. 97 Annex 3: Detailed Expenditure Tables Annex Table 35: Social Safety Net Expenditures (E Million), FY2006/07 - FY2010/11 2006/07 2007/08 2008/09 2009/10 2010/11 Programs by Category Actual Actual Actual Actual Estimated Cash Transfers 71.4 33.7 228.3 151.6 131.6 Department of Social Welfare (Admin) 20.0 10.9 35.0 20.5 18.9 Grants to Individuals 0.0 0.0 0.0 0.0 0.0 Old Age Grant 42.6 8.7 179.9 115.9 100.3 Fire Disaster 0.1 0.1 0.1 0.1 0.1 Public Assistance 3.7 4.3 6.2 4.3 4.6 Child Welfare Foster Children 1.1 5.0 1.1 2.5 0.0 Handicapped Children 0.0 0.1 0.1 0.1 0.1 Adult Handicapped - - 0.0 - - Assistance for the Disabled (Project) 0.1 - 0.5 1.5 0.8 Military Pensions 3.0 2.9 2.5 3.1 2.9 Young Heroes 1.0 1.7 3.1 3.7 4.0 In-Kind Transfers 330.4 327.7 344.7 370.9 403.8 Children in Clinics & Hospitals 0.0 - 5.5 14.8 11.2 Disaster Management Agency 100.0 100.0 54.0 34.2 17.0 Disaster Management Admin - - 0.0 2.2 3.0 DMA Food Distribution 100.0 100.0 54.0 32.0 14.0 Neighborhood Care Points 5.8 0.9 - 13.1 15.1 Agricultural Inputs 2.7 2.6 1.4 1.5 5.1 School Feeding Program 10.1 10.7 6.1 30.8 28.0 OVC Education Grant & Bursaries 67.8 80.2 117.5 167.5 239.8 OVC Education Grant 45.5 44.9 103.3 132.2 159.5 Primary Bursaries 9.9 21.1 0.0 14.7 24.8 Secondary School Bursaries 8.2 - - - 31.5 EU Capitation Grant 4.2 14.2 14.2 20.7 24.0 Phalala Fund 60.0 60.0 60.0 60.0 60.0 Prosthetics Rehabilitation Grants - - 0.0 - - WFP Commodities 83.9 73.3 100.1 48.9 27.7 Active Labor Market 16.6 12.2 23.2 33.0 26.3 Adult and Non-Formal Education 7.9 7.9 9.9 14.0 14.0 Manzini Industrial Training Centre 1.5 0.8 0.8 2.4 2.8 Nhlangano Agricultural Skills Training - 0.4 0.4 1.1 0.9 Center Siteki Industrial Training Centre - 0.3 0.3 0.6 0.5 98 2006/07 2007/08 2008/09 2009/10 2010/11 Programs by Category Actual Actual Actual Actual Estimated Other Adult Non-Formal Education 6.3 6.4 8.4 9.9 9.8 Vocational Training Centers for Disabled 5.1 4.3 5.6 10.4 2.2 SEDCO 3.7 - 7.8 8.5 10.2 Community-based 12.0 10.9 3.9 (17.7) 21.5 Micro Projects Program 12.0 10.9 3.9 (17.7) 21.5 Social Care 3.0 2.2 1.4 2.5 1.1 Caritas Orphan AID 0.5 0.4 0.4 0.4 0.4 SOS Children's Village 0.1 0.0 0.2 0.3 0.3 Voluntary Organizations 0.1 0.1 0.1 0.1 - Department of Gender and Family Issues 1.0 1.1 0.1 1.3 0.0 Lutsango Lwaka Ngwan 1.3 0.5 0.6 0.4 0.4 TOTAL SOCIAL SAFETY NET 433.5 386.8 601.5 540.2 584.4 99 Annex Table 36: Distribution of Social Safety Net Spending, FY2010/11 Program Expenditures 2010/11 Programs E Million US$ Million % of SSN Cash Transfers 131.6 18.6 22.5 Department of Social Welfare (Admin) 18.9 2.7 3.2 Grants to Individuals 0.0 0.0 0.0 Old Age Grant 100.3 14.1 17.2 Fire Disaster 0.1 0.0 0.0 Public Assistance 4.6 0.6 0.8 Child Welfare Foster Children 0.0 0.0 0.0 Handicapped Children 0.1 0.0 0.0 Adult Handicapped - 0.0 0.0 Assistance for the Disabled (Project) 0.8 0.1 0.1 Military Pensions 2.9 0.4 0.5 Young Heroes 4.0 0.6 0.7 In-Kind Transfers 403.8 57.0 69.1 Children in Clinics & Hospitals 11.2 1.6 1.9 Disaster Management Agency 17.0 2.4 2.9 Disaster Management Admin 3.0 0.4 0.5 DMA Food Distribution 14.0 2.0 2.4 Neighborhood Care Points 15.1 2.1 2.6 Agricultural Inputs 5.1 0.7 0.9 School Feeding Program 28.0 3.9 4.8 OVC Education Grant & Bursaries 239.8 33.8 41.0 OVC Education Grant 159.5 22.5 27.3 Primary Bursaries 24.8 3.5 4.2 Secondary School Bursaries 31.5 4.4 5.4 EU Capitation Grant 24.0 3.4 4.1 Phalala Fund 60.0 8.5 10.3 Prosthetics Rehabilitation Grants - 0.0 0.0 WFP Commodities 27.7 3.9 4.7 Active Labor Market 26.3 3.7 4.5 Adult and Non-Formal Education 14.0 2.0 2.4 Manzini Industrial Training Centre (M.I.T.C.) 2.8 0.4 0.5 Nhlangano Agricultural Skills Training Centre 0.1 0.1 (NASTC) 0.9 Siteki Industrial Training Centre 0.5 0.1 0.1 Other Adult Non-Formal Education 9.8 1.4 1.7 Vocational Training Centers for Disabled 2.2 0.3 0.4 SEDCO 10.2 1.4 1.7 Community-based 21.5 3.0 3.7 Micro Projects Program 21.5 3.0 3.7 Social Care 1.1 0.2 0.2 Caritas Orphan AID 0.4 0.1 0.1 SOS Children's Village 0.3 0.0 0.1 Voluntary Organizations - 0.0 0.0 Department of Gender and Family Issues 0.0 0.0 0.0 Lutsango Lwaka Ngwan 0.4 0.1 0.1 TOTAL SOCIAL SAFETY NET 584.4 82.4 100.0 100 Annex 4: Poverty Measurements and Poverty Mapping 254. Poverty is measured by the poverty incidence, gap, and severity indicators. The headcount index or poverty incidence is the simplest and most frequently used measure of poverty. It represents the fraction of individuals with consumption per adult equivalent below the poverty line. The poverty gap index indicates how poor the poor people are, in other words, how far below the poverty line their consumption is. The severity of poverty indicator is used to measure the inequality of consumption among the poor (some poor people may have consumption close to the poverty line, while some may be far from it). 255. The Central Statistical Office (CSO) recently completed a poverty map for Swaziland, based on the Housing and Population Census of 2007 and the SHIES (CSO, 2011b). The poverty mapping technique takes basic demographic data for each of 195 districts in Swaziland and imputes the welfare level of the people living in each district (CSO, 2011b). The poverty map demonstrates that the four regions of Swaziland are far from homogenous and that three of the four regions have pockets of both relative wealth and extreme poverty. 256. The report also includes Millennium Development Goal indicators for each region, tinkundla, and district. The regional pattern of poverty is basically replicated by the 20 other indicators used in the report, including school enrollment and employment patterns. 101 Annex 5: PMT Formulae and Tables 257. The following annex includes a summary of the work undertaken by the team analyzing the SHIES data for this report. The work was done in collaboration with the CSO, and this chapter aims to provide details of some of the technical aspects of how the team prepared three different PMT formulas and highlights some of the challenges that they faced. 258. The team’s multivariate analysis 93 indicated a lack of any sharp correlates with poverty status beyond rural location and the presence of children or elderly people in the household. Factors that are more difficult for social workers to verify (such as education of the household head and whether the head worked for one or more hours during the preceding month) were also significant in multivariate specifications but not as sharply as rural location and presence of dependents in the household. 259. In Annex Tables 5.1 to 5.3 below, any variables with the suffix “dum” are 0-1 dummies. Other variables are dummies for housing characteristics, such as the type of roofing. Demographic and livestock variables without suffixes are counts of variables, so that child is the number of children under 18, elderly is the number of people aged 60 or over in the household, and cattle is the number of cattle. The demographic variables are strongly correlated with household welfare, with an additional child or elderly member reducing welfare about the same amount in rural as in urban areas (Annex Table 5.1). In urban areas (Annex Table 5.2), having children in the household is sharply negative for consumption, while having an elderly household member is positively correlated with welfare, which is a paradoxical finding. 260. Some general findings from the multivariate analysis are that welfare is not correlated strongly with livestock holdings in rural areas. For example, the ordinary least squares (OLS) regression for the population in general (Annex Table 5.3) found that each cow added only SWE 0.7 to household consumption, while having an expensive roof contributed SWE 373. 261. Children and elderly members subtracted from household consumption in rural areas but not as much as using the bush or a field for a toilet or having an unsanitary water source. In urban areas, children subtracted very significantly from household consumption, but elderly members added to it (although few elderly people actually live in urban areas). 262. Some urban households reported owning livestock, presumably on rural plots owned by these urban households, or by households in more rural peri-urban settings that were classified as urban. Unfortunately, the negative coefficient estimate for cattle in the rural OLS is not plausible. In a culture where cattle are still often a bride price and a store of wealth, one would expect there to be a large positive coefficient on cattle. Anecdotal qualitative evidence from similar countries suggest that livestock are kept to store wealth and are only sold when the household needs money to pay for a bride price, a funeral, or major illness of a family member. 93 OLS, probit, and logit. Available upon request from: jeanine.braithwaite@gmail.com. 102 Such irregular expenditures do not show up in the consumption aggregate. Monthly consumption in rural areas remains low and living standards are relatively homogenous, regardless of wealth holdings. Unfortunately, although we tried many different ways of formulating the livestock variables, we could not get them to come out more positively. 263. Despite the team’s many efforts with alternative specifications, the only kind of livestock that was significantly positively correlated with rural household welfare was pigs, and then their coefficient was only a fraction of many of the other household attributes. 264. The lack of significant, larger coefficients for livestock in rural areas is a serious barrier to the creation of a politically acceptable PMT formula based only the SHIES. It would be counter-intuitive to offer a PMT without positive coefficients on cattle in rural areas. With 88 percent of the poor in rural areas and livestock being a major means by which rural households survive and hold wealth, more work is required to develop a reliable PMT for rural households. 265. The urban-only regression (Annex Table 5.2) has a low exclusion error of 4.8 percent, but leakage is higher at 25.8 percent. The combined regression is slightly better (as the estimates are more robust having been run on a larger number of households), with low exclusion errors of 5.2 percent and leakage errors of 17 percent. The rural regression, even without significant livestock estimates, also had very low errors of exclusion (5 percent) and inclusion (17.4 percent). Exclusion errors are when a poor household is found to be ineligible under the targeting formula and “leakage” is when a non-poor household is found to be eligible. These errors are not that high for social assistance systems, which suggests that the system works better for identifying poorer households (low exclusion errors) than for excluding non- needy households. 266. A combination of ad hoc adjustments and human oversight can help to increase the targeting efficiency of a PMT. Every scoring formula and every targeting mechanism is subject to type I and type II errors 267. Another way to increase the targeting efficiency of a PMT is to include an appeals mechanism in the program in question, which would enable applicants to appeal against being incorrectly excluded from the program by the PMT. A local committee is usually established to discuss such cases, but this committee can also investigate cases where households have misrepresented data or other errors have occurred. 103 Annex Table 37: Rural OLS Regression, Swaziland Source: Estimated from the SHIES 104 Annex Table 38: Urban OLS Regression, Swaziland Source: Estimated from the SHIES 105 Annex Table 39: Combined OLS Regression, Swaziland Source: Estimated from the SHIES 106 IBRD 33491 S WAZI LA N D SELECTED CITIES AND TOWNS DISTRICT CAPITALS NATIONAL CAPITAL SWAZILAND RIVERS SO UT H A FRICA To MAIN ROADS Hectorspruit RAILROADS DISTRICT BOUNDARIES Ngonini INTERNATIONAL BOUNDARIES ti To Mluma Komatipoort To Nelspruit 31°E Emlembe L (1,862 m) e Bulembu Piggs Peak b 26°S shaneni Lomahasha Tshaneni T 26°S r g Mhlume o Komati To Madlangampisi Maputo n e b b To HHOHHO s Bethal n o Ka Dake e zi k MO ZA MBIQ UE Mliba bul u a M M r ne MBABANE D u za t l bu M s Ezulwini LUBOMBO . Siteki Mhlambanyatsi Matsapha Manzini Bhunya Mz im Ny etane ph ofu Lus To MANZINI Sidvokodvo utf Bethal u Siphofaneni Mankayane Ngwempi s i Big Bend L us u tfu Sicunusa Sitobela To Mkuze Gege To d vo Volksrust 27°S n Hlatikulu 27°S o Maloma Mk Nsoko SHISELWENI Nhlangano Mgwav um e S OU T H AF R I C A Mhlosheni 0 10 20 30 Kilometers Lavumisa 0 5 10 15 20 Miles This map was produced by the Map Design Unit of The World Bank. The boundaries, colors, denominations and any other information shown on this map do not imply, on the part of The World Bank To Group, any judgment on the legal status of any territory, or any Hluhluwe endorsement or acceptance of such boundaries. To 31°E Ulundi 32°E SEPTEMBER 2004