SOUTH AFRICA Social Assistance Programs and Systems Review Foreword A cornerstone of the partnership between South Africa and the World Bank Group is knowledge development and exchange. The Social Assistance Programs and Systems Review for South Africa generates knowledge not only for South Africa in its endeavors to strengthen its social protection system, but also for other countries in the world to learn from one of South Africa’s most important post-Apartheid policy successes, that is, protecting the poorest segments of its population. This Social Assistance Programs and Systems Review takes stock of the current performance of South Africa’s social assistance policies, programs and systems, and their appropriateness in the face of the poverty and vulnerability profile of the country. To be effective, social protection systems need to be tailored to the social, economic, fiscal and policy context of a country, which may change over time. Poverty and inequality, including pockets of deep deprivation, remain two of South Africa’s most pressing concerns. Alongside extreme inequalities, South Africa struggles with high unemployment and low labour market participation rates. In this context, social assistance is a critical policy response and represents one of South Africa’s important successes in the post- apartheid era. South Africa’s social assistance system is an effective intervention for providing support to the poorest segments of the population. Social grants provide resources to poor households which, at the very least, significantly reduces the depth of poverty and inequality. Further, by providing regular and dependable income, they ameliorate vulnerability. This is particularly true if the effects of social grants on other outcomes such as health, education, and labour supply are considered. The system is extensive in terms of both the number of people it covers, directly and indirectly, as well as in terms of the amount of scarce resources it consumes. Approximately one in three South Africans is a direct beneficiary of a social grant, while nearly two-thirds of the population (64.0 percent) are either direct or indirect beneficiaries of the system. Evidence shows that social assistance transfers have significant positive impacts on reducing poverty and inequality in South Africa and boosting development outcomes. This review highlights, that in the medium term, there is an opportunity for the social assistance system in South Africa to link beneficiaries to other Government services and programs that help advance access to the labour market and earnings. South Africa spends more on social assistance than most other countries globally - 3.31 percent of Gross Domestic Product (GDP). Yet, social assistance is not available for a large share of the working-age members of the population and unemployment benefits are only available for those who work in the formal sector. The social assistance system may also benefit from greater integration of technology- based solutions in the application, eligibility testing, and payment processes, as well as from addressing the fragmentation of the social assistance system at the institutional level. It is my sincere hope that the analysis of this report will be helpful not only for the next steps in South Africa’s quest to strengthen social protection systems and sustain its commitment to protecting vulnerable groups, but also for other countries, especially when the COVID-19 pandemic has been testing the effectiveness of existing social assistance systems, generally. Marie Françoise Marie-Nelly World Bank Country Director 2021 The International Bank for Reconstruction and Development/ THE WORLD BANK 1818 H Street NW Washington, DC 20433 USA All rights reserved Photos: Shutterstock. i Acknowledgements This report was prepared by a team of World Bank staff and have also peer-reviewed the report. A full bibliography of the consultants in 2020. Morné Oosthuizen, University of Cape sources is included in the annex. Town and World Bank consultant, led the analysis and report writing of the document. Edits and overall oversight were The publication was supported by Melanie Jaya and Zandile provided by Victoria Monchuk and Bongisa Lekezwa. Inputs Portia Ratshitanga. and reviewer comments were provided by Melis Guven, Paolo Belli, Precious Zikhali, Ugo Gentilini, Iftikhar Malik, and Wendy The World Bank team is especially grateful to the Department of Cunningham from the World Bank. Haroon Bhorat (University of Social Development for their inputs and feedback on the report Cape Town) and Servaas van der Berg (Stellenbosch University) and the collaboration on its launch and dissemination. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW ii Table of Contents Executive Summary ...................................................................................................................................................................vii 1. Introduction ......................................................................................................................................................................1 2. South Africa’s Socio-economic Context ..........................................................................................................................3 2.1. Macro-economic Performance ................................................................................................................................................................................................................ 3 2.2. The Labour Market and Human Development ............................................................................................................................................................................ 4 2.2.1. Recent Labour Market Performance.................................................................................................................................................................................................... 4 2.3. Poverty and Inequality .................................................................................................................................................................................................................................. 8 2.4. The Impact of Covid-19 and the Nationwide Lockdown ....................................................................................................................................................10 3. Social Assistance Policy, Systems, and Programmes .................................................................................................. 12 3.1. Social Assistance in South Africa ..........................................................................................................................................................................................................12 3.1.1. An Overview of Social Protection Programmes in South Africa ....................................................................................................................................12 3.1.2. The Current Social Assistance Policy Landscape ......................................................................................................................................................................17 3.2. Social Assistance Programmes ..............................................................................................................................................................................................................19 3.2.1. Overview ...............................................................................................................................................................................................................................................................19 3.2.2. Older Persons Grant.......................................................................................................................................................................................................................................20 3.2.3. Child Support Grant ......................................................................................................................................................................................................................................20 3.2.4. Disability Grant .................................................................................................................................................................................................................................................20 3.2.5. Care Dependency Grant .............................................................................................................................................................................................................................20 3.2.6. Foster Child Grant ...........................................................................................................................................................................................................................................21 3.2.7. War Veterans Grant.........................................................................................................................................................................................................................................21 3.2.8. Grant-in-Aid .........................................................................................................................................................................................................................................................21 3.2.9. Social Relief of Distress ................................................................................................................................................................................................................................29 3.2.10. Covid-19 Social Relief of Distress Grant...........................................................................................................................................................................................22 3.2.11. Summary ...............................................................................................................................................................................................................................................................22 3.3. Resourcing for Social Assistance in South Africa ......................................................................................................................................................................27 3.4. Administration and Delivery of Social Protection and Social Assistance ................................................................................................................32 3.4.1. Institutions and coordination of social assistance...................................................................................................................................................................32 3.4.2. Delivery systems of social assistance ................................................................................................................................................................................................33 3.4.3. Administration of social assistance ....................................................................................................................................................................................................36 4. Social Assistance Programme Performance ................................................................................................................ 38 4.1. Coverage and Adequacy ............................................................................................................................................................................................................................38 4.1.1. Coverage................................................................................................................................................................................................................................................................38 4.1.2. Adequacy/Benefit Levels ...........................................................................................................................................................................................................................41 4.2. Inclusiveness.......................................................................................................................................................................................................................................................43 4.2.1. Targeting ...............................................................................................................................................................................................................................................................43 4.2.2. Benefit Incidence .............................................................................................................................................................................................................................................45 4.3. Cost Effectiveness ...........................................................................................................................................................................................................................................47 4.4. Impacts of Social Assistance in South Africa ...............................................................................................................................................................................51 4.4.1. Poverty and Inequality ................................................................................................................................................................................................................................52 4.4.2. Nutrition, Food Security, and Hunger...............................................................................................................................................................................................53 4.4.3. Education ..............................................................................................................................................................................................................................................................54 4.4.4. Health ......................................................................................................................................................................................................................................................................55 4.4.5. Labour Supply and Livelihoods ............................................................................................................................................................................................................55 iii 4.4.6. Fertility and Childbearing .........................................................................................................................................................................................................................56 4.4.7. Shock responsiveness ..................................................................................................................................................................................................................................57 5. Effectiveness of social assistance in South Africa ....................................................................................................... 61 5.1. Strengths of the social assistance system......................................................................................................................................................................................61 5.2. Shortcomings/areas of improvement of the social assistance system .....................................................................................................................61 5.2.1. Categorical programs addressing individual risk .....................................................................................................................................................................62 5.2.2. Integration and coordination .................................................................................................................................................................................................................62 5.2.3. Working-age adults .......................................................................................................................................................................................................................................62 5.2.4. Value for money – expenditure efficiency.....................................................................................................................................................................................63 5.2.5. Shock responsiveness – ability to scale up to address crises...........................................................................................................................................63 5.3. Fit of the System vs. South Africa’s Development Challenges ........................................................................................................................................63 6. Conclusions and forward look – reform options ......................................................................................................... 66 6.1. Feasibility of broader reforms ................................................................................................................................................................................................................66 6.2. Shorter-term reform options ..................................................................................................................................................................................................................66 6.2.1. Addressing cost-effectiveness and value for money .............................................................................................................................................................66 6.2.2. Strengthening outcomes and incentivising productivity and economic mobility .........................................................................................67 6.2.3. Providing coverage for the working-age adults........................................................................................................................................................................67 6.2.4. Delivery system and program level technical reforms .........................................................................................................................................................70 7. Bibliography .................................................................................................................................................................. 73 A. Appendix One ................................................................................................................................................................ 81 B. Appendix Two ................................................................................................................................................................ 83 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW iv List of Figures Figure 2.1. Real GDP Growth and GDP Per Capita, 1990-2019 ................................................................................................................................................................ 3 Figure 2.2. Labour Market Trends since 1993 ..................................................................................................................................................................................................... 4 Figure 2.3. Labour Force Participation and Unemployment Rates, 2014 and 2019 ................................................................................................................... 6 Figure 2.4. Labour Market and Socio-economic Outcomes across the Income Distribution (decile), 2017 .............................................................. 7 Figure 2.5. South Africa’s Gini Coefficient, 2006-2015 .................................................................................................................................................................................10 Figure 3.1. Social Security in South Africa ..........................................................................................................................................................................................................12 Figure 3.2. Access to School Feeding in South Africa, 2014/15 and 2018 .....................................................................................................................................19 Figure 3.3. Spending on Social Assistance as Share of GDP, 2009-2016 ..........................................................................................................................................27 Figure 3.4. Real Spending on Grants, 2006/07-2018/19 (log scale)....................................................................................................................................................29 Figure 3.5. Number of Grants, 2006/07-2018/19 ...........................................................................................................................................................................................30 Figure 3.6. Main Budget Aggregates for South Africa since 2001/02 (percent of GDP) ........................................................................................................31 Figure 3.7. Figure 3.7. Efficiency of Social Assistance Administration since 2004/05 ..............................................................................................................37 Figure 4.1. Coverage of Direct and Indirect Social Assistance Beneficiaries across Quintiles ............................................................................................39 Figure 4.2. Average Transfer Value Per Capita, Beneficiary Households Only, 2014/15 ..........................................................................................................41 Figure 4.3. Social Assistance Benefits as a Share of Total Expenditure (Adequacy of Social Assistance Benefits) across Quintiles ............42 Figure 4.4. Distribution of Social Assistance Beneficiaries Across Quintiles ..................................................................................................................................44 Figure 4.5. Distribution of Social Grant Beneficiaries Across Quintiles, 2014/15 ........................................................................................................................45 Figure 4.6. Distribution of Social Assistance Benefits Across Quintiles ............................................................................................................................................45 Figure 4.7. Simulated Poverty and Inequality Reductions (%) Associated with Social Assistance Programmes ...................................................47 Figure 4.8. Simulated Poverty Reduction (%) Associated with Social Assistance Programmes Globally....................................................................48 Figure 4.9. Simulated Inequality Reduction (%) Associated with Social Assistance Programmes Globally ..............................................................49 Figure 4.10. Benefit-Cost Ratio of All Social Assistance ..............................................................................................................................................................................50 Figure B.1. Social Grants Across the Income Distribution, 2017 ...........................................................................................................................................................84 Figure B.2. Adequacy of Social Assistance Benefits Across Geography, 2014/15 ......................................................................................................................85 v List of Tables Table 2.1. Absolute and Relative Poverty Indicators, 2006-2015 ............................................................................................................................................................ 8 Table 3.1. Social Protection Coverage since 2013 (‘000s)..........................................................................................................................................................................13 Table 3.2. South Africa’s Social Grants, 2020......................................................................................................................................................................................................23 Table 3.3. Social Assistance Spending by Programme as Percent of GDP, 2009-2016............................................................................................................28 Table 3.4. Consolidated Government Spending, 2010/11-2022/23 ..................................................................................................................................................28 Table 4.1. Social Assistance Coverage Rates (%), Direct Beneficiaries Only, 2014/15 ..............................................................................................................38 Table 4.2. Composition of Grants Received by Households across the Pre-Transfer Distribution, 2014/15 .............................................................40 Table 4.3. Relative Incidence, (%), 2014/15 ........................................................................................................................................................................................................43 Table 4.4. Distribution of Beneficiaries and Benefits Across Pre-Transfer Quintiles, 2014/15 .............................................................................................46 Table 4.5. Simulated Poverty Reductions (%) Associated with Social Assistance Programmes, 2014/15...................................................................49 Table A.1. Allocation of Grants to Individuals in the LCS 2014/15 Micro-data ............................................................................................................................82 Table B.1. Consolidated Government Spending, 2010/11-2022/23 ..................................................................................................................................................83 Table B.2. Grant Beneficiaries and Spending, 2006/07-2018/19...........................................................................................................................................................83 Table B.3. ASPIRE Estimates for Upper Middle Income Countries .......................................................................................................................................................86 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW vi Executive Summary Despite being an upper middle income country, South This report provides a review of the South African social Africa’s high inequality and the long-lasting legacies assistance system and consists of three broad thrusts. of apartheid mean that the country is faced with First, the review provides a sense of the operation of the social numerous development challenges, many of which are assistance system, the types of benefits it provides through its characteristic of countries with much lower incomes. key programmes, and the tools and administrative systems that Poverty and inequality remain two of the country’s most support its functioning. Second, it reviews the performance of pressing concerns. While money-metric poverty rates in 2015 the social assistance system in terms of coverage, targeting, were lower than those observed in 2006, there are some benefit incidence, adequacy, cost-effectiveness, and outcomes. indications that the latter part of the period saw deterioration. Third, it assesses the extent to which the system is aligned and Inequality, as measured by the Gini coefficient, fell marginally equipped to address the so-called “triple challenge” of poverty, over the same period, although it remains extreme by any inequality, and unemployment as shown by data, and reviews measure. Alongside extreme inequalities, South Africa struggles its limitations in the design, delivery systems, and institutional with high unemployment, low labour market participation coordination at different administrative levels. rates, and widespread poverty, including pockets of deep deprivation. In this context, social assistance is a critical policy The core focus of this paper is on social assistance and, response on the part of government and represents one of the specifically, the system of social grants in South Africa. important successes of the post-apartheid era. Labour market programmes are addressed in a separate forthcoming paper, while social insurance and contributory It is within this context that social assistance and social protection programmes are not included. Five key questions guide the policy is implemented in South Africa. The country’s broader analysis; these are: social security system consists of three main pillars: 1. What is the landscape of social protection and social assistance, the statutory funds, and the voluntary social assistance in South Africa, and what risks and funds. Social assistance, broadly defined, covers three sets of vulnerabilities do the policies and programmes aim to government interventions: i) social grants, the responsibility address (chapter 3)? of the Department of Social Development and administered 2. How is South Africa’s social assistance system performing by the South African Social Security Agency; ii) public works, in terms of providing adequate support to the poorest, such as the Expanded Public Works Programme, coordinated and addressing and preventing vulnerability and by the Department of Public Works and Infrastructure, and the inequality (chapter 4)? Community Work Programme, which falls under the auspices of the Department of Cooperative Governance and Traditional 3. What is the value for money, spending efficiency, and Affairs; and iii) other programmes such as the National School future fiscal sustainability of the current social assistance Nutrition Programme within the Department of Basic Education. landscape (chapters 2, 3, and 4)? The statutory funds include the Unemployment Insurance Fund 4. How well are the current social assistance programmes and the Compensation Funds, which fall under the Department aligned with South Africa’s development challenges, and of Employment and Labour, and the Road Accident Fund, which to what extent is South Africa’s social assistance system falls under the Department of Transport. Finally, the voluntary set up to mitigate the structural causes of poverty and funds are comprised of medical schemes and retirement funds, inequality and improve the economic inclusion and which are regulated by the relevant government authorities. human capital of the poorest (chapter 5)? 5. Are the current governance and coordination South Africa’s social assistance system represents a major arrangements, the level of coordination and capacity, intervention by government in addressing deprivation and integration of systems appropriate for social amongst the country’s population. The system is assistance programmes to effectively address the extensive both in terms of the number of people it covers, country’s development challenges (chapter 3)? directly and indirectly, and in terms of the amount of scarce resources it consumes. According to official data, To answer these questions, the paper brings together the number of grants paid out by government has increased a variety of data, including household survey data, from 12.02 million in 2006/07 to 17.81 million in 2018/19. Of administrative and official data, information from these, child support grants are the vast majority (12.45 million discussions with Government officials, and data from children), followed by the older persons grant (3.55 million global databases to describe and compare the South people), and the disability grant (1.05 million people). These African social assistance system with that of other three grants also dominate spending on grants: of the total of countries. R162.7 billion spent on grants in 2018/19, the older persons grant accounts for R70.6 billion, the child support grant for The first question to answer is what is the landscape of R60.6 billion, and the disability grant for R22.0 billion. Together, social protection and social assistance in South Africa these three grants account for 94 percent of total spending on and what risks and vulnerabilities do the policies and grants and nearly 96 percent of all grants. programmes aim to address? In terms of the design of vii the social protection system in South Africa, there are three The second question revolves around South Africa social main components: i) social assistance, which includes assistance system’s performance in terms of providing social grants, the public works programmes, and other adequate support to the poorest and addressing and interventions such as the National School Nutrition preventing vulnerability and inequality. By any measure, Programme; ii) the statutory funds, including the the South African social assistance system is extensive. Unemployment Insurance Fund and the Compensation Approximately one in three South Africans is a direct beneficiary Fund; and iii) the voluntary funds, such as medical schemes of a social grant, while nearly two-thirds of the population and retirement funds. While the employment-linked statutory (64.0 percent) are either direct or indirect beneficiaries of the and voluntary funds are financed through contributions by system. Transfers are equivalent to 7.3 percent of households’ employers and workers, social assistance is financed from expenditure nationally and 60 percent of expenditures in general tax revenues. quintile 1, the poorest 20 percent of the population. This is one way in which South African society demonstrates, through Social grants are by far the largest facet of the social government, the value placed on providing support to its protection system in terms of the number of people poorest and most vulnerable members. High coverage rates covered, with 17.8 million grants paid out by SASSA on a are primarily the consequence of the size of the programme of monthly basis in the 2018/19 financial year. The National School child support grants: children receiving a child support grant Nutrition Programme reaches upwards of nine million learners. represent almost one-quarter of all South Africans according to Social grants encompass eight key programmes, excluding the the Living Conditions Survey (LCS) of 2014/15. COVID-19 social relief of distress grant implemented in 2020, namely: the older persons grant, the child support grant, the The data presented demonstrates that the system disability grant, the care dependency grant, the foster child performs well in addressing both poverty and inequality. grant, the war veterans grant, grant-in-aid, and social relief Based on static simulations using data from the LCS 2014/15, of distress. The system is dominated in numerical and social grants are estimated to reduce the poverty rate by budgetary terms by the older persons, child support, and between 10.1 percentage points and 38.5 percentage points, disability grants. The grants are designed to address specific depending on the choice of official poverty line. Similarly, lifecycle and other risks, with a particular emphasis on children the post-transfer Gini coefficient (i.e. income including social (the care dependency, child support, and foster child grants) grants) is 6.7 percent lower than the pre-transfer Gini coefficient and the elderly (older persons and war veterans’ grants, and (i.e. income excluding social grants). grant-in-aid). The temporary COVID-19 social relief of distress grant was implemented to address the impact of the national These strong effects on poverty and inequality are lockdown in response to the COVID-19 pandemic, targeting the benefits of a system that is well-targeted at those working-age individuals with no income and no access to other who most need support. Coverage—including indirect forms of assistance. beneficiaries—is almost universal in the poorest pre-transfer quintile (95.2 percent) and is as high as three-quarters (74.1 The three compulsory contributory social security percent) in the third quintile. Indeed, more than half (56.1 funds—the Unemployment Insurance Fund (UIF) and the percent) of the population in the poorest pre-transfer quintile Compensation Funds, administered by the Department alone are direct grant beneficiaries, while coverage for the of Employment and Labour, and the Road Accident child support and older persons grants of the age-eligible Fund (RAF)—provide conditional income for eligible population in the bottom quintile is 86.9 percent and 96.6 individuals. The UIF provides unemployment insurance percent respectively. As a result, the poorest 60 percent of immediately after the loss of employment, including where the population account for almost 80 percent of all direct and this is the result of illness, maternity or adoption, and is the indirect grant beneficiaries, and a similar proportion of social largest of the three funds in terms of claims. The Compensation assistance benefits. Quintile 1 alone accounts for 29.8 percent Funds provide compensation for disablement or death caused of direct and indirect beneficiaries and 33.1 percent of benefits. by occupational injuries or diseases sustained or contracted by employees. However, both UIF and Compensation Fund Importantly, while grants are small in value in absolute benefits are available only to formal sector employees. The terms, the extent of inequality means that they Road Accident Fund (RAF) is funded primarily through a fuel are relatively large for a significant proportion of levy and it provides compensation in relation to road accidents. households. The average transfer per capita for beneficiary households in 2014/15 is estimated to have been only R3 279, The Expanded Public Works Programme (EPWP) and or around R273 per month. However, compared to beneficiary Community Work Programme (CWP) are key interventions households’ per capita household expenditure, this amount is targeted at the working age population, which aim to significant. Averaged across all beneficiary households, grant provide income, work experience, and training to the income is equivalent to roughly one-quarter of per capita unemployed. In 2019/20, the EPWP provided 838 000 work household expenditure. However, this figure is as high as two- opportunities or 267 000 full-time equivalent jobs, while in thirds for beneficiary households in quintile 1 and 40 percent 2018/19 the CWP provided 280 000 work opportunities. These for beneficiary households in quintile 2. programmes target the working-age population as part of government’s broader efforts to address joblessness in South Africa. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW viii South Africa’s social assistance system is therefore Nevertheless, the evidence suggests that the efficiency effective in providing support to the poorest segments of of the system has been improving. The average cost to the population. Social grants provide resources to poor pay out a grant has decreased in real terms from around R57 households that, at the very least, significantly reduce in March 2020 prices during the 2005/06-2009/10 period, to the depth of poverty and inequality. Further, by providing R36.70 in the 2019/20 financial year. Similarly, the proportion regular, dependable income, they ameliorate vulnerability. of the budgeted social assistance transfers that is allocated to This is particularly true if the effects of social grants on other administration has fallen from 7.8 percent in 2008/09 to 4.4 outcomes such as health, education, and labour supply are percent in 2019/20. considered. Between health, education, and social protection, roughly A key weakness of the system, which has been identified half of consolidated government spending is accounted for. by a number of authors, is the system’s blind spot At the same time, spending on social protection increased around working-age adults. While there are several by 3.7 percent per annum in real terms during the 2010s, programmes within the social protection system that cover which is somewhat more rapid than the rate of growth of working-age adults, each of them is limited in terms of their total spending (3.3. percent). Total spending on social grants, coverage. The only social grant accessible to working-age excluding administration costs, increased by 3.2 percent adults is the disability grant, which is predicated on disability; per annum on average in real terms between 2008/09 and and unemployment insurance and the Compensation Funds 2018/19. While the level and pace of spending growth is are only accessible to formal sector workers. In the absence not problematic on its own, the country’s fiscus has been of a jobseekers grant, the EPWP and CWP are the only under significant strain for some time. This is the result of a interventions available to the majority of working-age decade of particularly slow economic growth, diminished state adults and, while they can potentially play an important capacity, and other effects of state capture, and an inability to role in establishing a minimum level of income, their rein in spending, all of which are exacerbated by the COVID-19 current coverage is limited. As with interventions aimed pandemic. Thus, while there are not particularly pressing at children, these programmes may benefit from greater concerns regarding the long-term financial sustainability integration in order to encourage the unemployed to re- of the social assistance system on its own, it seems clear join the labour market. Indeed, there is scope for integration that government’s ability to further expand the system with labour market interventions through, for example, the will be constrained for the foreseeable future. Department of Employment and Labour to strengthen overall outcomes. The result of weak coverage of this cohort, Fourth, how well are the current social assistance however, has important implications for other social programmes aligned with South Africa’s development assistance interventions, as benefits received by children challenges, and to what extent is the social assistance and the elderly are shared with working-age adults who system set up to mitigate the structural causes of poverty have no other means of support. and inequality and improve the economic inclusion and human capital of the poorest? In answering these questions, South Africa spends more on social assistance than most we focus primarily on the triple challenges of unemployment, other countries globally: at 3.31 percent of GDP, it ranks as the poverty, and inequality. Economic growth in South Africa fourth-highest spender on the continent and tenth amongst all since the global financial crisis has been weak. Low growth countries with data. Given the competing demands in terms of has constrained job creation and hence the ability of the government spending, it is therefore important to understand economy to absorb new jobseekers into employment. This has what is the value for money, spending efficiency, and future made it difficult for households to support themselves sustainability of the current social assistance landscape? and invest in their human capital. Further, growth has been relatively capital-intensive, and where job creation has occurred From the perspective of value for money, estimates of the it has been biased towards higher skilled occupations. benefit-cost ratio for social assistance in South Africa reveal that, while the country performs around ten percent better than the Even before the national lockdown aimed at slowing average for Sub-Saharan African countries and is on par with the COVID-19 pandemic, unemployment in South Africa upper middle-income countries overall, its performance is was close to all-time highs, the narrow unemployment almost one-fifth weaker than the average for countries in Latin rate having reached 30.1 percent in the first quarter of 2020 America and the Caribbean. Given South Africa’s strong (Statistics South Africa, 2020d). The labour market is one of the performance in terms of the poverty-reducing impact arenas in which the fault lines of disadvantage and exclusion— of social assistance, the value-for-money performance is across race, gender, age, educational attainment, and location, lower than expected and suggests that the costs of South amongst others—are clearly evident. Spatially, the effects of Africa’s system are relatively high compared to other apartheid have been to locate many jobseekers far from work countries. Given the relatively sparse information in the public opportunities, with the result that transportation costs have domain on the costs associated with administering social become a significant barrier to poorer jobseekers. These spatial assistance in South Africa, understanding the cost structures distortions have been largely unaddressed and have, in some and cost drivers in different settings is an area for future research. instances, been compounded in the post-apartheid era. ix Although there is still relatively little data available on to shape the design of the system of social grants. The the longer-term effects of the COVID-19 response and evidence also suggests that, while negative impacts on labour lockdown, it is clear that the labour market has been supply may be observed, these may be explained by changes deeply impacted as employers have been forced to in household structure and by their location. This research also reduce wages or retrench workers. Along with the total highlights the importance of having regular household surveys shutdown of informal sector activity during the initial (Level 5) that collect sufficient data to explore these cross-cutting issues. lockdown, the poverty impact has certainly been substantial It is also important that SASSA and the DSD regularly publish and immediate, prompting government to announce a series performance data, not just on the numbers of grant recipients, of interventions aimed at cushioning the blow. Amongst these but on aspects of administration such as costs and modes of interventions has been the implementation of the COVID-19 payment. social relief of distress grant. South Africa is typically not affected by shocks in the way that Social assistance has a significant impact on both poverty many other countries tend to experience weather-related and inequality. Based on the Living Conditions Survey cyclical shocks. However, the COVID-19 crisis and national 2014/15 data, it was shown that social assistance significantly lockdown has arguably plunged the country into the deepest reduces poverty across a broad range of poverty lines. The economic, unemployment, and poverty crisis seen in a long impact is stronger for measures, such as the poverty gap and time. Parts of the social protection system were able to poverty gap squared, that place greater emphasis on individuals effectively scale up – social grants quickly increased furthest below the poverty line. Thus, while social grants may the benefit levels and payments from the UIF could be be insufficient to lift the poor completely out of poverty, they channelled on to furloughed or laid-off formal sector do go some way towards ameliorating the deepest poverty in workers. However, the crisis exposed other parts of the the country. system that were not able to respond quickly to the crisis. There was no effective way of identifying new shock-affected In terms of the design of the social grants, however, there people to provide them with support, whether through cash appears to be no overt consideration of or attempt to grants or food parcels. The National Integrated Social Protection align them with South Africa’s systemic development Information System (NISPIS) project should be fast-tracked to challenges, apart from poverty. Indeed, the emphasis is address the lack of central social registry. Moreover, limitations very much on the amelioration of deprivation—as illustrated by in payment withdrawal caused delays, confusion, and social the DSD’s and SASSA’s stated objectives and mandates, which crowding at pay-points. Further research could investigate mention poverty and vulnerability, but not inequality—so that alternative payment modalities which would allow recipients the impact on inequality is almost incidental. South Africa does to retrieve and use their social assistance payments closer to not make use of conditional cash transfers, which can be used to where they live and in markets where they normally shop. encourage or discourage specific behaviours such as increasing household investment in health and education, a policy choice Finally, are the current governance and coordination that aligns to government’s rights-based approach. Improving arrangements, the level of coordination and capacity, and the integration of the social protection system into a broader integration of systems appropriate for social assistance response to the underlying causes of socio-economic inequality programmes to effectively address the country’s – lack of opportunity, unequal access to and level of human development challenges? Unfortunately, integration capital, unemployment, and economic exclusion – would across programmes and government agencies and allow for the development of a package of services available to departments is not particularly strong. This represents individuals and households, especially for poor children, based a lost opportunity to build the types of synergies that on their particular situations. could lead to strong positive impacts for programmes, both individually and collectively. Such integration This is not to say that social grants do not have broader may be particularly beneficial for the child support impacts that may address key development challenges. grant, which has already been shown to have important There is a growing literature that points to broadly beneficial positive effects on human development. Setting up a impacts of social grants—either a specific grant or grants unified social registry, such as the NISPIS, linking together and generally—on a wide variety of outcomes in the areas of making interoperable a number of government databases poverty and inequality, nutrition and food security, education, will be a large step in the right direction. Given the long-term health, labour supply and livelihoods, and fertility. This body consequences of investment (or lack thereof ) in children’s of research points to the ways in which social grants have human capital, there is strong incentive to do as much as enabled poor households to invest in and build their possible to strengthen impacts. This is particularly true within human capital through improvements in educational the current fiscally constrained environment. attainment, nutrition and health, and suggest the potential for the grant system to have positive effects The South African social protection system is highly that play out intergenerationally and over the long term. capable and benefits from strong delivery systems All these positive impacts could be further strengthened for targeting, case management, data administration, if they would be made more explicit and their pursuit able and payments. However, there is room for improvement, SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW x especially in terms of coordination and integration, starting able to more effectively link social assistance beneficiaries with the interoperability of databases across government to the labour market. As noted in the beginning of this department as well as last-mile payment services. Mthethwa report, a review of active labour market programmes, (2019, p.103) notes the lack of integration of the institutional especially youth employment programmes, is conducted and administrative frameworks related to social security. At the separately. very least, this leads to duplication of work, of processes, and • Strengthen the links for social grants to other social services of function, all of which drive up the cost of the system. This via case management to facilitate households’ access and type of fragmentation and duplication is not unique to SASSA invest in the human capital of their children. and the DSD but is widespread across government. Moreover, while the payment system is highly digitised and large number It must be recognised, however, that the South African of grants are paid out on a timely basis and accounted for every government faces severe fiscal constraints that are month, beneficiaries still struggle to access funds queuing at likely to impact on the flexibility of policy to address the retailers and other pay-points month-after-month. country’s challenges in the post-COVID environment. While revenue shortfalls, rising expenditure, and rapidly In sum, to better align the social protection system, growing public debt are problems that have longer term roots, especially the social assistance system, to more effectively they have been exacerbated by the impact of the lockdown address the structural causes of socio-economic and the cost of interventions that the COVID-19 pandemic and inequality in South Africa a number of adjustments are the lockdown itself have necessitated. Given government’s suggested over the next five years. stated commitment to rein in public spending in order to stabilise public debt (National Treasury, 2020c), all ministries Strengthening delivery systems, integration, and coordination: have come under pressure to cut spending. We would argue • Continue to improve the interoperability of databases and that enforcing such cuts on social assistance would have registries in the government departments to serve as a significant negative impacts across a wide range of social registry to identify groups of vulnerable individuals potential outcomes, and that the cost would be borne and households. by those households who are least able to weather such • Strengthen the overall coordination and integration of shocks, undermining the system’s objectives of preventing social grants with system and services in other departments and alleviating poverty in both the short- and long-term. including the Departments of Basic Education, Health, Employment and Labour, Home Affairs, and Public Works That said, the analysis does suggest South Africa’s average and Infrastructure. benefit-cost ratio of social assistance is driven by relatively • Improve the last mile accessibility of social grants to quickly high costs. In this respect, the system may benefit over the and efficiently get funds to recipients by, for instance, medium-term through greater integration of technology- engaging the vast network of informal ‘spaza’ shops. based solutions in the application and payment processes, and through addressing the fragmentation Programme-level adjustments: of the social assistance system at the institutional level. With technology solutions safety and security measures need • Prioritise strengthening the quality and reach of public and to be in place to minimize fraud. non-government employment service programmes to be xi 1. Introduction South Africa is an upper middle income country (UMIC) with the limitations in the design (e.g. structure of cash grants the most sophisticated economy on the continent. In 2018, and appropriateness of program mix), delivery systems (e.g. its GDP per capita in constant 2020 US dollars was estimated interoperability of information systems for different schemes), at $7 434, which places South Africa just behind Colombia ($7 and institutional coordination at different administrative levels. 692), China ($7 753), and Botswana ($8 031), but 13 percent below the upper middle income average of $8 541 (World This paper focuses on social assistance and, specifically, Bank, 2020b). However, despite this relatively high level the system of social grants in South Africa. Labour market of income, inequality and other long-lasting legacies of programmes are assessed in a separate paper which will be apartheid mean that the country is faced with numerous forthcoming. Social insurance and contributory programs are socio-economic development challenges, many of which not included. This paper aims to provide answers to five broad are characteristic of countries with much lower incomes. questions, namely: 1. What is the landscape of social protection and These problems often contribute to and reinforce patterns social assistance in South Africa, and what risks and of disadvantage established under apartheid, serving to vulnerabilities do the policies and programs aim to perpetuate them through a quarter of a century post-apartheid address (chapter 3)? democracy. This compromises the ability of individuals and 2. How is South Africa’s social assistance system performing households to earn a living and invest in their human capital. in terms of providing adequate support to the poorest Thus, alongside extreme inequalities, South Africa struggles and addressing and preventing vulnerability and with high unemployment and widespread poverty, poor inequality (chapter 4)? education outcomes, and pockets of deep deprivation as its economy proves unable to generate rapid and 3. What is the value of money, spending efficiency, and inclusive economic growth over a sustained period future fiscal sustainability of the current social assistance of time. In this context, social assistance is a critical policy landscape (chapters 2, 3 and 4)? response on the part of government and, indeed, represents 4. How well are the current social assistance programmes one of the important successes of the post-apartheid era. aligned with South Africa’s development challenges, and to what extent is South Africa’s social assistance system Social assistance has a long history in South Africa. set up to mitigate the structural causes of poverty and Beginning in the early twentieth century, the system was inequality and improve the economic inclusion and initially implemented along racial lines with Whites as the human capital of the poorest (chapter 5)? primary beneficiaries and discriminated against other groups 5. Are the current governance and coordination in terms of both access and benefits. However, the system was arrangements, the level of coordination and capacity, gradually extended to include other race groups while the gap and integration of systems appropriate for social in benefits narrowed in the latter years of the apartheid era. assistance programmes to effectively address the Post-1994, the system has been refined and further extended country’s development challenges (chapter 3)? so that it is now accessible and well-targeted. Today, the South African social assistance system is one of the most extensive in To answer these questions, the paper brings together the world and the system has been studied globally. a variety of data, including household survey data, administrative and official data, information from However, the social assistance system in South Africa is discussions with Government officials, and data from global expensive (3.31 percent of GDP) and with a tightening databases to describe and compare the South African social fiscal environment there is increasing pressure for the assistance system with that of other countries. Chapter 2 system to become more flexible and to deliver better provides context in terms of the South African economy and its value-for money especially in protecting working-age key socio-economic challenges of unemployment, poverty, and adults who are outside the labour market. These are inequality. It also outlines some of the impacts—both measured some of the challenges, and more are discussed in subsequent and predicted—of the national lockdown in response to the chapters, which motivates this report. COVID-19 pandemic. In chapter 3, the focus turns to social assistance policy and briefly outlines the historical development This report aims to review the social assistance system of the system before detailing the current institutional set- in South Africa to first understand how it functions and what up and South Africa’s social assistance programmes (one of kinds of benefits it provides through which programs, and which being the temporary COVID-19 grant). Section 3.1 places what tools and systems it uses to do so. Second, it reviews the social assistance within a broader picture of social protection performance of the system in terms of coverage, targeting, programmes, while section 3.3 details the resourcing of social benefit incidence, adequacy, cost-effectiveness, and outcomes. assistance in South Africa, and section 3.4 focuses on delivery Third, it assesses the extent to which the system is aligned and systems and administration. In chapter 4, the performance equipped to address the so called “triple challenge” of poverty, of social assistance programmes is analysed across a variety inequality, and unemployment as shown by data. It also reviews of metrics related to adequacy, inclusiveness, and cost- 1 effectiveness. The chapter also summarises some of the existing improvement the South African social assistance system evidence on the impact of social assistance grants within could undertake in the next five years in order to better the South African context. Finally, chapter 5 summarizes the align the system address the structural causes of poverty and assessment findings and discusses the fit of the social assistance inequality in addition to providing relief and income support. system, given South Africa’s development challenges. Chapter 6 concludes and provides some policy and programme recommendations for the future. Based on the analysis, this paper provides some recommendation for what adjustments and SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 2 2. South Africa’s Socio-economic Context 2.1 Macro-economic Performance despite a period of dynamic economic growth following the end of Apartheid and up to the Global Financial Crisis (GFC). South Africa is an upper-middle income country and the South Africa’s real GDP per capita grew by 2.1 percent per second-largest economy on the African continent, just annum between 1994 and 2008. However, South Africa was behind Nigeria, based on 2019 estimates of GDP in current unable to return to pre-GFC growth rates over the next decade, US dollars (World Bank, 2020b). However, South Africa’s mostly hampered by domestic structural constraints, leading to economic growth over the past 30 years has been largely a contraction of real GDP per capita by 0.1 percent per annum unremarkable: in real Rand terms, GDP growth has averaged over 2008-2019, partly reverting progress achieved in the just 2.2 percent per annum between 1989 and 2019, while real previous decade. per capita GDP growth has averaged just 0.5 percent per annum over the same period. Based on World Bank (2020b) data, GDP Figure 2.1 presents South Africa’s economic growth performance per capita in constant 2011 PPP dollars grew by an average of since 1990, as well as the level of real GDP per capita in 0.73 percent per annum between 1990 and 2018; this is below constant 2010 prices. Recovering from a deep recession in the the average of 1.04 percent for Sub-Saharan African countries, early 1990s, South Africa’s economy had just returned to and compares poorly with peers such as Brazil (1.16 percent), growth by the time of the first democratic elections in Nigeria (1.65 percent), Colombia (1.96 percent), Botswana (2.40 1994. GDP growth averaged 1.8 percent per annum over the percent), and Thailand (3.39 percent). Indeed, South Africa’s 1990s, but real GDP per capita was 2.8 percent lower by the end growth is far below the upper middle income country average of the decade than in 1990 (R44 735 in 2000 compared to R46 of 3.72 percent per annum over this 28-year period. This is 020 in 1990). Figure 2.1. Real GDP Growth and GDP Per Capita, 1990-2019 6.0 60.0 5.0 57.5 56.5 54.3 4.0 55.0 54.9 Real GDP growth (annual, %) Real GDP per capita (R ‘00s) 3.0 52.8 52.5 2.0 50.0 1.0 47.5 0.0 45.0 -1.0 42.8 42.5 -2.0 40.0 GDP Growth Ave. GDP gr. (1990-’00): 1.8%p.a. Ave. GDP gr. (2000-’10): 3.5% p.a Ave. GDP gr. (2010-’19): 1.5% p.a GDP per capita -3.0 37.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2004 2005 2006 2007 2008 2009 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2002 2003 2010 Source: Own calculations, South African Reserve Bank (2020). Notes: Horizontal dashed lines represent the average annual GDP growth rate for each decade. GDP per capita figures are in constant 2010 prices. The global recession in 2008-09 revealed the fragility The past decade has seen the country’s growth of the South African economy and labour market, and performance continue to deteriorate as it struggled with concerns around the types of jobs that had been created policy uncertainty, political turmoil, a deepening energy during the period of rapid growth, as the economy shed crisis, and mounting evidence of deeply entrenched 900 000 jobs within three quarters (own calculations, Kerr et al., corruption and so-called ‘state capture’. While economic 2019). GDP expanded by an average of 3.5 percent per annum growth quickly rebounded after the 2009 recession, it became between 2000 and 2010, and per capita GDP ended the decade more fragile over time. With economic growth falling below the 20.3 percent higher at R53 823. rate of population growth, GDP per capita began to decline from its 2015 peak of R56 470, falling 2.7 percent to R54 906 in 2019. This is a level last seen in 2011 and confirms the 2010s 3 as a lost decade from the perspective of economic growth. Apartheid Labour Market Series (Kerr et al., 2019), which The structural weaknesses of the economy produce important harmonises the country’s labour market surveys conducted negative consequences across a wide range of socio-economic since 1993. The upper panel of the figure presents the total issues. The low-growth environment has constrained job number of employed and unemployed individuals, using both creation, which in turn has put pressure on resources the narrow and broad definitions of unemployment, while the available to households to support themselves and lower panel presents the narrow and broad unemployment invest in their own human capital. rates, as well as the employment-to-population ratio. 2.2. The Labour Market and Human On the basis of the data presented in Figure 2.2, a number of Development points should be highlighted. First, South Africa’s economy has generally been creating jobs throughout the 2.2.1. Recent Labour Market Performance post-apartheid period. The key exception is the roughly two-year period after the global financial crisis in 2008 (the The labour market plays a pivotal role in linking individuals decline observed in 2001 being due to the overestimation of and households with the economy and the fruits of economic employment in the early waves of the Labour Force Survey growth. However, its history of discrimination and exclusionary (Kerr and Wittenberg, 2019, p.3)). policy has meant that South Africa’s labour market is an arena in which inequality and disadvantage both Second, despite this growth in employment, the number manifest themselves and are replicated. As a result, labour of people who are unemployed has also increased over market policy and regulation are hotly contested as they try to time. Given changes in the measurement of unemployment, deal with apartheid’s historical legacy and present economic comparisons over time of the results from the same types of realities. surveys (i.e. between the dotted vertical lines) are more reliable than those that compare results from different surveys. This rise The transition to democracy coincided with the first in unemployment is observed for each sub-period—1994-1999, efforts at the systematic collection of household survey 2000-2007, and 2008- 2019—and across both definitions of data covering the entire population by Statistics South unemployment. The only exception is that the number of Africa. Figure 2.2 presents the trends in key labour market narrow unemployed at the end of 2007 was marginally lower indicators over the past quarter century using the Post- than in early 2000. Figure 2.2. Labour Market Trends since 1993 18.0 16.0 Employed 14.0 12.0 10.0 Millions Broad unemployed 8.0 6.0 4.0 Narrow unemployed 2.0 0.0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 4 50.0 Employment-to-population ratio 45.0 40.0 35.0 Broad unemployment rate 30.0 Percent 25.0 Narrow unemployment rate 20.0 15.0 10.0 5.0 0.0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Source: Own calculations, Kerr et al. (2019). Note: Vertical lines indicate changes in the underlying surveys: The Project for Statistics on Living Standards and Development in 1993; the October Household Surveys from 1994 to 1999; the Labour Force Surveys from 2000 to 2007; and the Quarterly Labour Force Surveys from 2008 onwards. The narrow definition of unemployment, which is the official definition, defines the unemployed as individuals aged 15-64 years who were not employed in the reference week, who actively sought work or tried to start a business during the preceding four weeks, and who were available for work in the reference week (Statistics South Africa 2008). The broad definition is identical to the narrow definition but does not require that the unemployed took active steps to find work or start a business during the preceding four weeks. Third, the 2000-2007 period is unique in that for the population aged 15 years and above; this is much lower unemployment rates fell. This period, particularly than the estimate of 58.8 percent for Africa and 59.9 percent the latter years, coincided with rapid job growth for upper-middle income countries. Only 11 countries globally that exceeded labour force growth and drove down have lower ratios. With COVID-19 impacts hitting employment unemployment. However, in general, the situation has been hard, the official unemployment rate stood at 23.3 percent one of rising unemployment rates even while employment has (excluding discouraged workers) in the second quarter of 2020 been growing, indicating labour force growth rates that are in (Statistics South Africa, 2020f ). excess of the employment growth rate. Fourth, despite the increase in employment over the past quarter century, Aggregate figures obscure important differences the employment-to-population has remained relatively between groups and it is these differences that are stable and within a narrow range of between 40 percent critical to the understanding of the socio-economic and 45 percent. This is very low in comparison with other challenges facing South Africa. Figure 2.3 presents estimates countries and is the result of relatively low participation of the labour force participation (LFPR) and unemployment rates combined with very high unemployment rates. rates for the two definitions of unemployment across a range The International Labour Organisation (2020) estimates an of demographic covariates, namely race, gender, age, and employment-to-population ratio for South Africa in 2020 of 40.0 educational attainment. 5 Figure 2.3. Labour Force Participation and Unemployment Rates, 2014 and 2019 Narrow labour force participation rate Narrow unemployment rate 100 88.6 86.3 84.7 87.6 80.2 78.1 75.0 80 68.3 72.7 74.1 70.3 69.5 64.1 67.7 68.8 59.8 64.0 65.9 58.5 59.9 63.5 56.8 54.6 59.8 58.1 60 53.8 50.1 50.2 44.5 45.4 46.6 48.8 42.8 43.4 40 32.4 31.3 35.6 31.2 36.8 26.2 29.1 26.6 29.7 27.2 27.2 27.4 24.6 24.3 22.9 24.9 22.4 23.0 23.2 24.8 19.0 17.5 20 11.913.9 12.3 12.6 18.5 7.7 7.6 9.2 7.6 7.1 5.3 0 Overall African Coloured Asian White Female 15-24yrs 25-34yrs 35-44yrs 45-54yrs 55-64yrs Primary Inc Sec Grade 12 Diploma Degress Overall African Coloured Asian White Female 15-24yrs 25-34yrs 35-44yrs 45-54yrs 55-64yrs Primary Inc Sec Grade 12 Diploma Degress Male Male Narrow labour force participation rate Narrow unemployment rate 100 90.4 87.0 85.9 88.6 82.7 89.2 89.8 89.5 84.5 78.8 74.4 77.0 77.3 80 69.2 69.3 68.8 70.0 71.7 69.5 67.5 68.9 63.6 65.7 65.2 64.9 64.1 61.2 62.7 59.9 55.6 55.9 60 53.5 48.3 48.7 45.2 43.7 46.7 43.0 40.6 38.7 39.0 42.4 38.0 38.4 39.4 37.7 36.0 34.6 40 35.5 31.3 35.5 30.4 33.2 30.1 25.2 26.8 26.4 22.6 20.6 16.8 19.9 17.0 16.6 20 15.0 9.5 9.4 9.8 6.5 0 Overall African Coloured Asian White Female 15-24yrs 25-34yrs 35-44yrs 45-54yrs 55-64yrs Primary Inc Sec Grade 12 Diploma Degress Overall African Coloured Asian White Female 15-24yrs 25-34yrs 35-44yrs 45-54yrs 55-64yrs Primary Inc Sec Grade 12 Diploma Degress Male Male 2014 2019 Source: Own calculations, Statistics South Africa (2015b, 2020c). There are clear differences across groups in terms of these two Labour market disadvantage extends beyond participation types of indicators. Labour force participation rates are and unemployment rates. For example, similar patterns of higher for men than for women, for those aged 25-54 disadvantage have been observed in terms of prevalence of years than for other working age adults, and for those low pay (Oosthuizen, 2012); informality (Bhorat et al., 2016), and with higher levels of educational attainment. While the employment volatility (Zizzamia and Ranchhod, 2019). differences between race groups in expanded LFPRs are small, they are much more pronounced for the narrow LFPR: Whites The extent of inequality in South Africa will be discussed are more likely to participate in the labour market than Africans, in more detail in section 2.3, but it is worth noting here the for example. However, at least part of this difference is explained significant disparities that exist in terms of labour market by differences in, for example, educational attainment and and other outcomes across the income distribution. Figure location (urban areas have higher participation rates, and 2.4 provides a sense of some of these inequalities: the first row higher proportions of Whites reside in urban areas). of figures relates to the labour market, the second to individual characteristics, and the third to household characteristics. Bars Unemployment rates tend to be higher for Africans and denote deciles of the per capita income distribution, with the Coloureds than for Asians and Whites. Women are more poorest decile (decile 1) the leftmost and the richest decile likely to be unemployed than men, while unemployment (decile 10) the rightmost in each panel. rates are particularly high for the youth (15-24 year-olds in particular). Higher levels of education are associated with lower unemployment rates. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 6 Figure 2.4. Labour Market and Socio-economic Outcomes across the Income Distribution (decile), 2017 Narrow Unemp. Rate (%) Expanded Unemp. Rate (%) Information (%) Employment-to-Population Ratio (%) 100.0 90.0 80.1 80.0 63.0 65.8 70.0 60.5 57.9 60.0 52.2 51.5 53.4 49.8 51.4 50.9 47.2 43.3 43.0 50.0 40.5 38.1 40.0 29.6 32.1 28.8 31.8 26.5 24.6 26.5 30.0 21.9 18.5 16.1 21.2 17.1 15.2 16.1 15.7 20.0 10.6 6.1 11.3 6.9 3.7 8.7 7.8 10.5 3.1 10.0 0.0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Years of Education (25-54 yr olds) Perceived Good Health (%) Computer Literate (%) Banked Adults (25-54 yr olds) (%) 97.9 100.0 90.7 92.4 87.9 Percent / Years 90.0 81.9 73.3 70.4 71.2 71.2 72.1 72.5 74.4 80.0 67.6 70.1 70.5 69.9 68.9 65.4 67.7 70.0 61.4 52.4 52.1 60.0 46.6 50.0 39.6 39.1 41.9 40.0 30.7 34.6 26.3 26.6 30.0 20.0 9.0 9.2 9.5 9.7 10.0 10.3 10.7 11.2 12.0 13.5 10.0 0.0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Urbanisation (%) HH Access to Electricity (%) HH Access to Flush toilet (%) HH Access to Piped water (%) 100.0 89.2 88.7 87.2 87.0 89.2 90.6 94.8 95.1 91.6 96.5 90.0 79.1 82.7 81.2 82.9 83.5 83.4 75.5 77.7 79.1 80.0 72.2 68.2 70.2 60.7 62.2 70.0 55.8 55.4 60.0 55.1 57.6 58.6 60.0 47.3 46.6 50.0 43.8 38.1 40.2 32.9 40.0 25.7 29.9 30.0 19.3 18.6 20.0 10.0 0.0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Source: Own calculations, Saldru (2018). Note: (1) The informality rate refers to the proportion of employment in each decile that is informally employed. (2) The employment-to- population ratio is the ratio of employment in a given decile to that decile’s total population. (3) Perceived good health refers to the proportion of each decile’s population that rates their health as excellent or very good. (4) Computer literate refers to the proportion of each deciles’ population aged 15 years and older that rates their computer literacy as advanced or basic. (5) Banked adults refers to the proportion of adults between the ages of 25 and 54 years who have their own bank accounts. (6) For urbanisation and household access rates, figures are calculated as the proportion of households and not of the population. Across virtually every one of the 12 indicators, there are participated in the Southern and Eastern Africa Consortium for important differences based on the position on the income Monitoring Education Quality (SACMEQ) tests (Mlachila and distribution. Unemployment rates are up to 20 times Moeletsi, 2019, p.14). higher for the poorest ten percent of the population compared to those in the richest decile. Similarly, while There are also strong differences in the proportion of the 8.7 percent of the employed in the top decile are informally population aged 15 years and above who indicate they at employed, this is true of 80.1 percent of those in the bottom least have basic computer literacy (90.7 percent in decile 10 decile. The employment-to-population ratio is around 50 compared to 26.3 percent in decile 1), and in the proportion percent in the top three deciles; this means that, on average, of adults between the ages of 25 and 54 years who have bank for every employed person in these deciles there is one other accounts (97.9 percent in decile 10 compared to 41.9 percent person who is not employed. In contrast, this proportion is in decile 1). The richest segment of the population is around 10 percent in the poorest three deciles, implying that overwhelmingly urban, while amongst the poorest for every employed person there are nine others who are not deciles the majority live in rural areas. The households in employed. which the poorest deciles reside are also substantially less likely than their wealthier counterparts to have access to electricity, Adults between the ages of 25 and 54 years in the top decile have to flush toilets, and to piped water. an average of 4.5 years more education than their counterparts in decile 1. While there are important differentials in Together, these figures bring the extent of inequalities in South Africa in terms of number of years of education, South Africa into stark relief. The poor are particularly these underestimate the true nature of inequality in disadvantaged in terms of labour market outcomes— educational outcomes. This is due to the wide variation in the this is a dominant reason why they are poor—but quality of education across the South African education system. they are also marginalised in terms of their ability to The South African education system performs poorly relative engage fully in the economy. They have around one-third to those of other countries, with Van der Berg et al. (2007, less education than their counterparts in wealthiest decile, p.854) highlighting the fact that the country is outperformed have much lower levels of computer literacy, and reside in by other African countries with far lower incomes and fewer households and communities that are under-resourced and government resources. In 2013, for example, South Africa that are often distant from employment. Kerr (2017) thus finds was ranked seventh and sixth amongst the 13 countries that that Africans’ average daily commute time in 2013 was 58.44 7 minutes, compared to 37.17 minutes on average for Whites. - Expected Years of School. In South Africa, a child who starts school at age 4 can expect to complete 10.2 years Kerr’s analysis is useful in providing a sense of how urban spatial of school by her 18th birthday. patterns, established under apartheid and largely unaddressed - Harmonized Test Scores. Students in South Africa post-1994, impact on ordinary South Africans’ daily lives and score 343 on a scale where 625 represents advanced budgets. On average, commuting costs were estimated at attainment and 300 represents minimum attainment. 15 percent and 17 percent of total income for those who - Learning-adjusted Years of School. Factoring in what used their own cars, buses, or minibus taxis for their daily children actually learn, expected years of school is only commute, and as high as 21 percent for those who used 5.6 years. multiple modes of transport (Kerr, 2017, p.334). Framing - Adult Survival Rate. Across South Africa, 69 percent commuting time as a tax on income, Kerr (2017, p.335) finds of 15-year olds will survive until age 60. This statistic is a effective tax rates of 38.5 percent for those who use multiple proxy for the range of health risks that a child born today modes of transport, and around 28 percent for users of public would experience as an adult under current conditions. transport (trains, buses, and minibus taxis) who are typically - Healthy Growth (Not Stunted Rate). 73 out of from lower income groups. 100 children are not stunted. 27 out of 100 children are stunted, and so are at risk of cognitive and physical Human development outcomes in South Africa are lower limitations that can last a lifetime. than what is expected for the country’s GDP level – especially educational attainment. The World Bank’s 2020 2.3. Poverty and Inequality Human Capital Index (HCI) notes that a child born in South Africa today will be 43 percent as productive when she grows Poverty and inequality are two of the so-called ‘triple up as she could be if she enjoyed complete education and full challenge’ that faces South Africa, and closely related to health (World Bank 2020d). This is higher than the average for the third, namely unemployment. As with unemployment, sub-Saharan Africa but lower than the average for Upper Middle there is a clear ‘hierarchy’ in South African society in terms of Income Countries (UMICs). Between 2010 and 2020, the HCI poverty. This should come as no surprise given the importance value for South Africa remained approximately the same at 0.43. of income from labour (or a lack thereof ) within total household The part that sets South Africa behind is mainly linked to the low income. Hundenborn et al. (2016), for example, show that levels of test scores and learning-adjusted years of schooling. labour income accounted for 73.0 percent of total household In South Africa, 80 percent (2016) of 10-year-olds cannot read income in 2014. and understand a simple text by the end of primary school. This is similar to the average for sub-Sahara Africa (80%) but much The measurement of poverty trends relies on the higher than the average for countries of similar income levels consistent application of poverty line across survey data (38%). Moreover, in 2017 22 percent of adolescent girls were from different points in time. This makes it difficult to find out of school, which is more than twice as high as for countries consistent estimates across the full post-apartheid period. of similar income levels (10 percent). Below are the ratings of Fortunately, Statistics South Africa (2017) have published the HCI components in 2020. poverty estimates covering the period between 2006 and - Probability of Survival to Age 5. 97 out of 100 2015 using their official set of poverty lines and per capita children born in South Africa survive to age 5. household expenditure. Table 2.1 presents some of their high- level estimates. Table 2.1. Absolute and Relative Poverty Indicators, 2006-2015 Change Unit 2006 2009 2011 2015 (2006-2015) Food poverty line Population in poverty mil 13.4 16.7 11.0 13.8 +0.4 Poverty rate % 28.4 33.5 21.4 25.2 -3.2 Poverty gap % 9.3 12.3 6.8 9.0 -0.3 Lower-bound poverty line         Population in poverty mil 24.2 23.7 18.7 21.9 -2.3 Poverty rate % 51.0 47.6 36.4 40.0 -11.0 Poverty gap % 22.2 21.0 14.3 16.6 -5.6 Upper-bound poverty line         Population in poverty mil 31.6 30.9 27.3 30.4 -1.2 Poverty rate % 66.6 62.1 53.2 55.5 -11.1 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 8 Change Unit 2006 2009 2011 2015 (2006-2015) Poverty gap % 35.6 33.5 25.5 27.7 -7.9 Share of income to bottom 40% of households % 8.2 7.3 7.5 8.3 +0.1 Source: Statistics South Africa (2017) and own calculations. Note: The levels of the three poverty lines are as follows (in April 2017 Rands): the food poverty line is set at R531 per capita per month, the lower-bound poverty line is R758 per capita per month, and the upper-bound poverty line is R1 138 per capita per month. Estimates presented here are based on per capita household expenditure. Estimates based on the Income and Expenditure Surveys of 2005/06 and 2010/11 and the Living Conditions Surveys of 2008/09 and 2014/15. Irrespective of the poverty line used, the poverty rate Africa”, with unemployment’s contribution increasing over time. was lower in 2015 than in 2006. Using the upper-bound poverty line (UBPL), the poverty rate is estimated to have been Despite gradually falling poverty rates, South African 55.5 percent in 2015, 11.1 percentage points lower than in society remains the world’s most unequal and there is no 2006. The reduction in the poverty rate using the lower-bound compelling evidence of improvement in this regard. One poverty line (LBPL) is of a similar magnitude (11.0 percentage way that this inequality is illustrated is through the proportion points down to 40.0 percent). However, the food poverty rate of income that accrues to the poor. For example, Statistics declined by just 3.2 percentage points to 25.2 percent in 2015, South Africa (2017) shows that the proportion of income indicating that much of the progress has been concentrated on accruing to the bottom 40 percent of households (which, due the population above the food poverty line but below the LBPL to the correlation between per capita income and household and UBPL. This is further confirmed by the virtually unchanged size, are home to more than 40 percent of the population) has food poverty gap (9.3 percent in 2006 and 9.0 percent in 2015) been virtually unchanged over the 2006-2015 period (see Table compared to declines in both the lower- and upper-bound 2.2). In 2015, this share was estimated to be a mere 8.3 percent. poverty gaps. The slow progress in reducing food poverty means that the population in food poverty increased by 400 Estimates of South Africa’s Gini coefficient indicate only slight 000 over the 2006-2015 period to 13.8 million. changes in inequality between 2006 and 2015 (Figure 2.5). The Gini coefficient can take on a value between zero and one, Although more recent official poverty estimates do not with zero indicating perfect equality (i.e. all individuals earn the exist due to a lack of data, the World Bank (2020d) has same income) and a value of one indicating perfect inequality projected that the upper bound poverty rate increased (i.e. one individual earns all the income). Estimated using either from 55.5 percent in 2015 to 56.5 percent in 2019. The per capita income or expenditure data, South Africa’s Gini poverty rate for 2020 is projected to rise to 58.6 percent—1.7 coefficient is extremely high. Expenditure-based estimates percentage points higher than the original projection—due to of the Gini published by Statistics South Africa (2017) range the COVID-19 pandemic and associated economic disruption. between 0.64 and 0.67 between 2006 and 2015; income-based Using the $1.90 per day poverty line, they project an increase estimates are even higher, ranging between 0.68 and 0.72. in the poverty rate from 19.0 percent in 2015 to 19.8 percent in Amongst the 160 countries with data since 2000, South Africa 2019, and 20.6 percent in 2020. ranks as the most unequal country in the world with a Gini coefficient of 0.630 (World Bank, 2020b). Poverty trends based on subjective and multidimensional poverty indicators confirm a decline in poverty, although Compared more broadly with regional medians provides a the periods are slightly different. Using three subjective stark indication of the extent to which South Africa stands out. poverty measures, Statistics South Africa (2018b) finds a decline South Africa’s Gini coefficient is around two-thirds higher in subjective poverty rates between 2009 and 2015. Similarly, than the global median of just 0.369 and 0.387 for upper the multidimensional poverty index, which includes three non- middle income countries. Inequality in South Africa is monetary dimensions of deprivation (health, education, and strongly linked to labour market inequality. Hundenborn living standards) suggests that the multidimensional poverty et al. (2016), using data from the National Income Dynamics rate fell from 17.9 percent in 2001 to 8.0 percent in 2011 Survey, estimate that income from the labour market accounted (Statistics South Africa, 2014), and further to 7.0 percent in 2016 for 90 percent of the Gini coefficient in 2014. Labour income (World Bank, 2018). Declines in multidimensional poverty are was found to have had a Gini coefficient of 0.73 compared to a consistent with positive trends in access to assets and services, value of 0.655 for total household income per capita. Further, such as water, electricity, and sanitation, and represent an lower-income households are much less likely to have important achievement of the South African government in the access to labour income, which in turn accounts for a post-apartheid era. However, the World Bank (2018) notes that smaller proportion of total income for these households “[unemployment] and education (years of schooling) remain (see, for example, Leibbrandt et al., 2016). the top two contributors to multidimensional poverty in South 9 Figure 2.5. South Africa’s Gini Coefficient, 2006-2015 Per capita expenditure Per capita income 0.80 0.72 0.70 0.69 0.70 0.67 0.68 0.64 0.65 0.64 0.60 0.50 Gini coefficient 0.40 0.30 0.20 0.10 0.00 2006 2009 2011 2015 2006 2009 2011 2015 Source: Statistics South Africa (2017). 2.4. The Impact of COVID-19 and the The lockdown and other interventions aimed at dealing Nationwide Lockdown with the pandemic will have numerous long-term effects, both large and small, for economies and societies around During the time that this report has been written, the the world. From the perspective of social assistance, some of COVID-19 outbreak had rapidly escalated to a global the key impacts in South Africa include: pandemic. The ensuing responses from government around the world—to partially or totally restrict individual movement - A sharp economic contraction and a high degree of and economic activity—have led to large scale economic economic uncertainty; disruption, massive job losses, and widespread uncertainty. In - Formal sector job losses with knock-on effects in the South Africa, a National State of Disaster was declared informal sector, and a large increase in unemployment: on 15 March 2020, and the country went into a complete it is estimated that South Africa lost 2.16 million jobs national lockdown—one of the strictest globally—on between the second quarters of 2019 and 2020, of which 26 March 2020 for three weeks, which was later extended roughly half were either in the informal sector (767 000 to five weeks. At the end of this period, the country moved to jobs lost) or domestic work (278 000 jobs lost) (Statistics what has been termed a Level Four lockdown, which is slightly South Africa, 2020e), while third quarter figures suggest less restrictive, with further easing to Level Three on 1 June 1.7 million jobs were lost year-on-year, of which 539 000 2020, and eventually down to Level 2 in July 2020, and Level 1 in job losses were in the informal sector and 165 000 in August 2020. As part of the gradual reopening, the Department domestic work (Statistics South Africa, 2020f ); of Basic Education reopened schools in phases by grade. - Rising poverty levels and incidence of hunger; - Pressure on tax revenues alongside increased demands While the lockdown has had some success in delaying the for higher government spending to both deal with the infection curve and providing time for government to prepare negative health and socio-economic effects of the crisis for the predicted rise in COVID-19 infections, it is expected and stimulate economic activity and, consequently, a to have a devastating economic impact. While National reprioritisation of the national budget; and Treasury predicted 0.9 percent growth in real GDP for - Increased prominence of the Department of Social 2020 in the 2020 Budget delivered at the end of February, Development and the South African Social Security forecasts from various institutions compiled by Bhorat et Agency (SASSA) as frontline state institutions tasked with al. (2020) range between -5.8 percent and -9.5 percent. mitigating the effects of the crisis, and increased scrutiny In October 2020 the World Economic Outlook projected GDP of their technical capacity, efficiency, and flexibility in growth for South Africa to slow down to -8.0 percent (IMF, responding. 2020). However, at this point there is very little hard data that provides an indication of the extent of the long-term impacts In terms of the poverty effects, a number of papers of lockdown phases on the economy. have begun to emerge that assess the poverty impact of the lockdown and related policy responses. In their SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 10 simulation of the impacts of social protection interventions respondents reported that their households had insufficient in response to the pandemic, Caron and Tiongson (2020) money to purchase food in April 2020, while 15 percent of of the World Bank estimate an increase in the upper adults with co-resident children reported that at least one bound poverty rate of between 1.5 and 3.9 percentage of those children had gone hungry in the preceding week, a points, depending on the length of the lockdown period situation exacerbated by the fact that the National School considered and assumptions regarding the types of Nutrition Programme was not operational from the start of the workers experiencing income shocks. Bassier et al. lockdown until early July 2020. Based on the same data, Jain et (2020) estimate a 5.2 percentage point increase in the al. (2020) estimate that between 15 percent and 30 percent of upper bound poverty rate, rising to 18.8 percentage individuals who lost their jobs between February and April 2020 points for households that include informal workers (they fell into poverty, and suggest figures of one million job losses assume a 75 percent reduction in informal worker incomes due and an additional two million dependents falling into poverty to the lockdown). In their analysis of the first wave of the NIDS- due to lockdown-related job losses. CRAM survey, Wills et al. (2020) find that 47 percent of adult 11 3. Social Assistance Policy, Systems, and Programmes In this chapter we review the social assistance landscape Social assistance represents only one component of a in South Africa, the policies and legislation that guide broader system designed to provide support to individuals the system and the programs that exist. The chapter also and households who may be in need due to a number of presents the resources and financing of social assistance given life events or risks. The contours of South Africa’s broader the high costs and limited fiscal space. Finally, the chapter social security system are illustrated in Figure 3.1. The three reviews the institutions and delivery systems that are used main components of the system are social assistance, to administer the programs and the level of capacity and the statutory funds, and the voluntary funds. Within coordination of institutions and agencies. social assistance are the social grants, the public works programmes, and other interventions such as the National 3.1. Social Assistance in South Africa School Nutrition Programme. There are three main statutory social insurance funds. Two of these programmes are linked to 3.1.1. An Overview of Social Protection Programmes in the labour market, namely the Unemployment Insurance Fund South Africa and the Compensation Funds, while the Road Accident Fund is intended to protect victims of road accidents. Finally, the voluntary funds can be divided into those that protect against health risks and those that provide retirement benefits. Figure 3.1. Social Security in South Africa Employers and Workers Financing General Tax Revenues Road Users Social Assistance Statutory Funds Voluntary Funds Social Grants Public Works Other Employment Other Health Retirement Programmes Child support Disability Older persons Expanded National Unemployment Road Accident Medical Retirement Grant-in-aid Public Works School Insurance Funds Fund Schemes funds Care depend. (Covid-19) Nutrition War veterans Community Programme Compensation Foster care Work Funds Children Working ages Older persons Other groups Intended Beneficiaries Source: Adapted from National Treasury (2010a, p.102). The various programmes are financed differently. Social funds, and as contributing members or dependents in assistance is financed from general tax revenues, while medical schemes. Older persons are covered by three the employment-linked statutory funds and voluntary funds types of social grants, as contributing members or are financed through contributions by employers and workers. dependents in medical schemes, and through retirement The Road Accident Fund is financed through a fuel levy, which funds. However, the centrality of formal sector employment to is paid for by road users. Employers and workers also indirectly ‘earning’ coverage should not be overlooked: without formal help finance social assistance through their tax contributions employment, working-age adults are only covered by social and, as road users, contribute to the Road Accident Fund. assistance if they have a disability, by the Road Accident Fund if they are involved in a road accident, and by medical schemes Support received from any of these programmes is likely to if they are a dependent of a member. Given high levels of be shared within and across households to at least some long-term unemployment in South Africa, this means that a extent. However, the lower section of the figure indicates large proportion of the working-age population have no who the intended direct beneficiaries of each of the listed access to social security. Further, contributions to voluntary programmes are. Children are directly targeted through retirement funds are contingent on employment, particularly child grants, school feeding, and as dependents in formal employment, thus limiting the number of older persons medical schemes. Working-age adults are covered by who are covered by retirement funds. Box 1 provides a brief the disability grant (and the new temporary COVID-19 historical review of social assistance in South Africa. grant), public works, the employment-linked statutory SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 12 Box 1: A Brief Overview of the Historical Development of Social Assistance in South Africa While social security has a relatively long history in South Africa, for most of that time it was characterised by discrimination, particularly discrimination along racial lines. As Van der Berg (1997) notes, “an embryonic welfare state was erected [by the apartheid state] to protect whites against various contingencies”. Specifically, racial discrimination was exercised along both the extensive and intensive margins: individuals were only eligible for access to particular programmes if they were a member of a particular race group (extensive margin), while benefit levels were often differentiated according to race (intensive margin). The earliest social assistance programmes date back to the early decades of the twentieth century. The first programme to provide support to children was implemented through the Children’s Protection Act No. 25 of 1913, with Africans being ineligible (Mthethwa, 2019). Military pensions were established in 1919 (Van der Berg, 1997). The 1928 Old-Age Pensions Act (No.22) introduced state (or social) pensions with age criteria and means-testing for Whites and Coloureds without occupational retirement insurance, with the former entitled to higher benefit amounts (Mthethwa, 2019; Van der Berg, 1997). Finally, support for the disabled was implemented in the mid-1930s—1936 for the blind, and 1937 for the disabled—but access was again initially limited to only White and Coloured individuals. The early 1940s saw further expansion of the system. In 1941, pensions were instituted for war veterans. The social pension was extended to Asians and Africans in 1944, with different benefit levels and stricter means-testing (Van der Berg, 1997). Two years later, in 1946, Asians and Africans became eligible for disability benefits, while “family allowances for large low-income families [were introduced] in 1947, but these excluded black people” (Van der Berg, 1997). According to Van der Berg (1997), “[the] levels and types of social grants were thus a result of the peculiar nature of political patronage in apartheid society”. By the 1970s and 1980s, there was a gradual move towards equalising benefits across race groups. However, fiscal considerations meant that this would have to be achieved through an erosion of benefit levels for Whites and occurred “most readily …where resistance to reducing white benefit levels was least” (Van der Berg, 1997). One example was social pensions, where benefits were equalised through “enhancing black pension benefits (by 7.3 per cent per year in real terms from 1970 to 1993) and seriously eroding real white pensions” (Van der Berg, 1997). South Africa’s first democratic government therefore inherited a system of social assistance that was relatively extensive given the country’s income levels (Woolard et al., 2010). However, the process of equalisation of benefits across race groups meant that absolute benefit levels were lowered to ensure that programmes were affordable, if not sustainable over time. Nevertheless, the extent of inequality within the country meant that even these lower benefit levels were able to make significant contributions to the resources available to destitute households. Table 3.1 provides an overview of the extent of some of industries, as well as the Government Employees Pension Fund the key programmes within the broader social protection (GEPF). It is immediately clear from Table 3.1 that the social system in South Africa, including social grants and school grants are by far the largest facet of the social protection feeding (social assistance); the Unemployment Insurance Fund system from the perspective of the number of people (UIF), Compensation Fund, and Road Accident Fund (RAF) (social covered. As of the 2018/19 financial year, 17.8 million grants— insurance); and the Expanded Public Works and Community equivalent to roughly 31 percent of the country’s population of Work Programmes (employment programmes). Details are 57.7 million (Statistics South Africa, 2018a)—were paid out by also provided for the private retirement and medical scheme SASSA on a monthly basis. Table 3.1. Social Protection Coverage since 2013 (‘000s) Financial Year 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 Social grants Recipients 15 932 16 643 16 992 17 201 17 510 17 812 .. National School Learners fed 9 132 Nutrition Prog. Unemployment Approved claims 763 708 691 651 799 771 .. Insurance Fund Compensation Fund Claims registered 311 226 129 352 155 156 .. Road Accident Fund Claims registered 147 174 189 202 272 328 .. FTE jobs 213 387 285 302 399 404 267 Expanded Public Works Work opp. 862 1 104 742 779 900 997 838 Community Work Prog. Work opp. 217 202 213 235 264 280 .. 13 Active members 1 277 1 266 1 270 1 274 1 273 1 265 .. Government Employees Pensioners 272 283 292 303 .. Pension Fund (GEPF) 391 406 Spouses 150 153 157 160 .. Calendar Year 2013 2014 2015 2016 2017 2018 2019 Active/Contrib. 10 411 10 969 11 134 11 070 11 245 .. .. Private sector retirement members funds Other members 4 845 4 973 5 306 5 574 5 700 .. .. Members 3 931 3 914 3 871 3 925 3 912 3 947 4 000 Medical schemes Dependents 4 847 4 900 4 938 4 953 4 960 4 970 4 955 Source: Council for Medical Schemes (2017, 2020); Department of Basic Education (2015); Department of Cooperative Governance (2017, 2019); Department of Labour (2013, 2016b, 2019a,b); Department of Public Works and Infrastructure (2014, 2015, 2016, 2017, 2018b, 2019a, 2020); Financial Services Board (2011, 2012, 2013a,b, 2014, 2015, 2016, 2017, 2018); Government Employees Pension Fund (2017, 2019); National Treasury (2012a, 2017b); Road Accident Fund (2013, 2016, 2019); SASSA (2019). Notes: (1) EPWP figures for 2019/20 are for the nine months ending 31 December 2019. (2) Retirement fund figures include double counting. (3) Where data has not been located, this is indicated by ‘..’. (3) Figures for 2013/14 and 2014/15 for the GEPF combine pensioners and spouses. Data on the extent of the National School Nutrition primarily through a fuel levy and it provides compensation “for Programme (NSNP) is relatively scarce, but the number of loss or damage wrongfully caused by the driving of a motor learners fed by the programme does not appear to have vehicle” (Road Accident Fund, 2016). Compensation covers loss changed much from the 9.1 million reported for 2013/14. of earning, loss of support, general damages, and medical and In its latest annual report, Department of Basic Education funeral costs for road accident victims. (2019, p.18) indicates that “over nine million learners” benefited from the programme, with almost R7.7 billion allocated to the Of these three social security funds, the UIF is the largest National School Nutrition Programme for the 2020/21 financial in terms of claims. Over the six-year period between year in the national budget (National Treasury 2020b). Based on 2013/14 and 2018/19, the UIF received between 650 000 data from the 2018 General Household Survey, it is estimated and 800  000 claims per year. Given that the period of UIF that 10.3 million school-attending children under the age of coverage is less than one year, the average number of individuals 20 at least occasionally eat food provided by the NSNP, while receiving UIF benefits is slightly lower. The Compensation Fund 8.95 million report eating this food every day (own calculations, and RAF are broadly similar in size: claims registered by the Statistics South Africa 2018c). Compensation Fund ranged between 129 000 and 352 000 per year over the six-year period, while those registered by the RAF The three compulsory contributory social security ranged between 147 000 and 328 000. The data confirm strong funds—the UIF and the Compensation Funds, growth in the number of claims registered by the RAF, with administered by the Department of Employment and claims having more than doubled over the period. However, it Labour, and the RAF—provide conditional income should be noted that by the start of the six-year period claims for eligible individuals. The UIF provides unemployment had been in decline for four years from almost 300 000 in insurance for a period of up to 365 days immediately after 2008/09. the loss of employment. This period was recently increased from 238 days through an amendment to the Unemployment The Expanded Public Works Programme (EPWP) is a key Insurance Act (Republic of South Africa, 2016). Benefits are paid intervention targeted at the working-age population, in the event of unemployment to certain employees, and for which aims to provide income, work experience, and illness, maternity, adoption and dependant benefits related to training to the unemployed (Box 2). The programme the unemployment of such employees (Department of Labour, is administered by the Department of Public Works and 2016b). Contributions of one percent of gross salary are paid by Infrastructure. Work opportunities—up to 100 days of work—are both formal private sector employees and their employers. The provided in four sectors: infrastructure, non-state (supporting Compensation Funds provide compensation for disablement employment in non-profit organisations), environment, and or death caused by occupational injuries or diseases sustained culture and social. In 2019/20, the EPWP provided 838 000 work or contracted by employees (Department of Labour, 2016a). opportunities or 267 000 full-time equivalent jobs. In terms of The Fund provides “medical care and income benefits to work opportunities, this is considerably lower than the peak workers who are injured while at work, or who develop recorded in 2014/15 of just over 1.1 million. Similarly, full-time occupational diseases”, as well as benefits for workers who are equivalent jobs in the 2019/20 financial year were a third lower fatally injured at work (Woolard et al., 2010). Importantly, the UIF than the 404 000 recorded in the preceding year. Across the and Compensation Funds only cover individuals in the formal four sectors, more than R7.7 billion was paid out to employees sector of the economy. The Road Accident Fund (RAF) is funded on EPWP projects for the nine months to the end of December 2019 (Department of Public Works and Infrastructure, 2020). SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 14 Box 2: The Expanded Public Works Program The Expanded Public Works Programme (EPWP) provides poverty and income relief through temporary work for the unemployed to carry out socially useful activities. It is one of the government’s flagship programmes aimed at drawing a significant number of unemployed South Africans into productive work and enable them to gain skills and increase their capacity to earn an income that will contribute towards the development of their communities. The EPWP is implemented across all spheres of government (national, provincial, and local) with work opportunities grouped into four productive sectors (figure below). The Department of Public Works and Infrastructure (DPW) acts as the overall coordinator with the departments of Cooperative Governance and Traditional Affairs, Environmental Affairs and Tourism, and Social Development as sector leads for the non-state, environment and culture, and social sectors respectively. In the infrastructure sector, beneficiaries are engaged in the construction, rehabilitation, and maintenance of rural and low-volume roads as well as schools and clinics. In the non-state sector, work opportunities are created through the non-profit and community organisations where participants deliver communal programmes and services. Activities include planting and cultivating food gardens at clinics, schools, churches, and household plots; home-based care; developing recreation spaces and sporting facilities; environmental rehabilitation; general maintenance work including the cleaning of schools; and, other tasks to support schools and community safety. The non-state sector encompasses the Community Works Programme (CWP), introduced in the second phase of the EPWP, whose primary objective is to create access to a minimum level of regular and predictable work opportunities for those in need. In the environment and culture sector, beneficiaries are involved in public environment management (e.g. water, parks, fire, wetlands, waste) as well as through cultural programmes (e.g. tourism, arts, crafts). In the social sector, participants undertake public social programmes including Early Childhood Development, Community Based Care, and Community Safety. Thematic Areas of the EPWP Source: Department of Public Works and Infrastructure website Since its inception in 2004, the EPWP has been successful in creating mass employment. The programme is currently in Phase IV (2019-2024). In 2017/18, 900 234 work opportunities1 were created resulting in the transfer of R10.108 billion in wages to 827 205 participants – the CWP alone reached over 280 000 beneficiaries (2018/19). A wide array of services was provided, and assets created in poor communities including home-care services, school feeding, and community gardens. Access roads were built in the infrastructure sector and community parks were beautified, while the coast was cleaned and maintained in the environmental sector. Of the 4.5 million work opportunities created during Phase III (8 million since EPSW inception), 44 percent comprised youth, 66 percent women, and 1 percent people living with disabilities. In addition, most beneficiaries were poor or historically disadvantaged in quintile 2 and 3 households, residing in poorer or high unemployment provinces and had completed less than a matric education. Employment creation in the infrastructure sector often innovation within a global context—initially entailed home- involves ramping up labour-intensity where feasible based care, particularly for those with HIV and tuberculosis, within existing government budgets for infrastructure, but later expanded to support Early Childhood Development “making these jobs technically ‘free’ in budgetary Centres and literacy training. The environment and culture terms” (TIPS, 2018, p.13). The social sector—a South African sector began with the Working for Water programme within 1 35.7% of these opportunities were created at the national sphere, 41.5at the provincial sphere and 22.8% by municipalities. It is important to note that the focus on performance against ‘work opportunity’ targets for the implementing ministries has in some cases, resulted in substitution of workers displaced from existing formal jobs. The renaming of pre-existing jobs as EPWP jobs, and the re-categorisation of voluntary workers receiving sub-market rate stipends as ‘EPWP employees’, has not necessarily contributed to additional employment (McCord, 2017). 15 the Department of Water Affairs, with a variety of additional 2018/19. environmental programmes subsequently added. In contrast to EPWP, the CWP is “designed as a community- The Community Work Programme (CWP) is a component driven model of public employment” (TIPS, 2018, p.15). of the EPWP—specifically, it is part of the non-state The programme is implemented by non-profit agencies at a sector—but differs from it in a number of key respects. variety of different sites, which cover defined geographic areas The CWP aims to provide “a social safety net and within a particular ward or group of wards, and the intention work experience for participants and promote social is that they remain active on an ongoing basis. These sites are and economic inclusion by targeting areas of high chosen to specifically “[target] the poorest and most marginal unemployment” (Department of Cooperative Governance, areas” (Philip, 2013, p.12). Importantly, the work undertaken in 2019, p.75). While both the EPWP and CWP aim to provide 100 a given site is determined by the community, rather than by days of work in a year, the CWP’s approach is to spread this government, and such work must be ‘useful work’, defined as across the entire year, providing “regular, part-time work - in “work that contributes to the public good, or improves the practice two days a week or eight days a month” (Department of quality of life in communities” (TIPS, 2018, p.17). As such, the Cooperative Governance, 2011, p.3). This approach is a response work covers a wide variety of activities, but the programme’s to the context of high unemployment in South Africa: EPWP target of labour-intensity of 70 percent—meaning that labour participants are likely to return to unemployment given costs must make up 70 percent of the cost at the site level— the economic and structural constraints to employment does limit the types of activities that are feasible. generation, and the concentrated period of full-time employment does not typically lead to fundamental In 2020, in response to the economic and employment changes in livelihood strategies. Regular income from crisis caused by the COVID-19 pandemic the Presidency the CWP, however, serves to create an income floor for launched the Employment Stimulus (Box 3) with the aim participants in the same way that a social grant does. to create 800 000 temporary employment opportunities Further, there is no forced exit from the CWP, “in recognition of across a number of government departments and the limited economic alternatives for participants under current national and local levels. The Program builds and expands on conditions” (TIPS, 2018, p.16). Work opportunities provided by the EPWP model and its timed start date for workers in January the CWP are, however, limited due to a shortage of funding and 2021 coincides with the ending of the temporary COVID-19 have grown gradually from 217 000 in 2013/14 to 280 000 in grant which provided needed relief for a large number of workers, especially informal sector workers since May 2020. Box 3: The Presidential Employment Stimulus The Presidential Employment Stimulus was launched in October 2020 and seeks to confront the impact on poverty and employment caused by the COVID-19 pandemic, as part of government’s broader economic recovery agenda. Its aim is to support livelihoods while the labour market recovers – investing in public goods and services, enhancing skills and employability, and boosting demand in the economy at the same time. The vision of the program is to build a South Africa that works – counteracting job losses and creating new opportunity for growth and renewal. The Employment Stimulus is part of the R100 billion commitment for job creation made by President Ramaphosa in April 2020. The Presidential Employment Stimulus is designed to support a spectrum of opportunities, focusing on job creation through public employment; on job retention in vulnerable sectors; on direct support to livelihood strategies; as well as on fast-tracking high-impact employment enablers. Phase 1 of the employment stimulus, which received funding of R12.6 billion from the Special Adjustment Budget, spans all 11 government departments, all provinces, all metro areas, and aims to create 800 000 temporary opportunities—most of them public employment opportunities. The majority of opportunities are in basic education (school assistants), social development (income support to the early childhood development sector), and agriculture (relief for subsistence producers). Other more traditional public works programmes are being expanded such as the EPWP’s provincial roads maintenance and environment programmes. The 11 national departments are responsible for implementing the various programmes either directly or through provincial governments. The Project Management Office (PMO) in the Presidency is responsible for overall coordination of the stimulus. The implementing departments are responsible for all program activities including targeting, recruitment, and operational management of each program. The wages set in public employment programs depend on the skill level required of the participants who vary from engineers to low-skilled youth with a matric or less. Some employment opportunities are full-time for four to six months while others, like those offered by the EPWP are part time. As of December 1, 2020, the Presidential Employment Stimulus has already recruited workers for over 400 000 opportunities which are expected to start around January 2021. Source: Extracted from Government of South Africa (2020) Retirement funds and medical schemes are voluntary is estimated that there were 11.2 million active or contributing insurance schemes, which are regulated by the state. It members of retirement funds in South Africa in 2017, the latest SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 16 year for which the Registrar of Pension Funds has published - The Probation Services Act, 1991 (Act 116 of 1991); data. The GEPF, which covers the public sector, had just shy of - The Prevention and Treatment of Drug Dependency Act, 1.3 million active members as of March 2019. It is the largest 1992 (Act 20 of 1992); pension fund in South Africa, and one of the largest on the - The Social Assistance Act, 1992 (Act 59 of 1992); continent (GEPF 2020). Medical schemes cover just under - The Non-profit Organisations Act, 1997 (Act 71 of 1997); nine million people: 4.0 million contributing members and - The Welfare Law Amendment Act, 1997 (Act 106 of 4.955 million dependents. Membership of retirement funds 1997); and medical schemes has been growing relatively slowly, - The White Paper for Social Welfare Service (1997); constrained as it is by the level of formal employment in the - The Older Persons Amendment Act, 1998 (Act of 1998); South African economy. - The White Paper on Population Policy for South Africa (1998); 3.1.2. The Current Social Assistance Policy Landscape - The Advisory Board on Social Development Act, 2001 (Act 3 of 2001); Today, the responsibility for social assistance lies with - The Social Assistance Act, 2004 (Act 13 of 2004); the Department of Social Development (DSD). The DSD - The South African Social Security Agency Act, 2004 (Act delineates their work according to two primary functions. First, 9 of 2004); and the Department is responsible for managing and overseeing - The Policy on Financial Awards to Service Providers. social security, including both social assistance and social insurance policies. Here, the objective is to “prevent and Except for the Welfare Laws Amendment Act, 1997, and alleviate poverty in the event of life cycle risks such as loss of the Advisory Board on Social Development Act, 2001, income due to unemployment, disability, old age, or death all other acts have been amended at various points in occurring” (Department of Social Development, n.d.). Second, time since the advent of democracy. Further, the Children’s the Department is tasked with providing development social Act (2005), the Older Persons Act (2006), the Prevention of welfare services in partnership with civil society organisations and Treatment for Substance Abuse Act (2008), and the 2015 and other institutions to “reduce poverty, vulnerability, White Paper on the Rights of Persons with Disabilities are also and the impact of HIV and AIDS through sustainable identified as determining the DSD’s mandate by National development programmes” (Department of Social Treasury (2020b, p.293). Development, n.d.). In terms of social assistance, specifically, the two key At its most foundational level, the DSD derives its core mandate pieces of legislation are the Social Assistance Act (No.13 from the Constitution of the Republic (Republic of South Africa, of 2004) and the South African Social Security Agency Act 1996). First, Section 27(1) states that “Everyone has the right (No.9 of 2004) (Republic of South Africa, 2004a, b). The to have access to …social security, including, if they are Social Assistance Act provides the legislative framework for the unable to support themselves and their dependants, implementation of social assistance in South Africa and, inter appropriate social assistance”. Second, social services are alia, makes provision for a national-level agency responsible explicitly referenced in Section 28(1), which states that “[every] for delivering grants. The Act also specifically provides for child has the right …to basic nutrition, shelter, basic health care the current suite of grants provided by government: the care services, and social services”. Finally, in terms of Schedule 4 of dependency grant, the child support grant, the disability grant, the Constitution, welfare services, population development, the foster child grant, grant-in-aid, the older persons grant, and disaster management are denoted as being of concurrent and the war veterans grant. It is in terms of the latter piece of national and provincial legislative competence. legislation that the South Africa Social Security Agency (SASSA) was established in 2006, as a schedule 3A public entity in terms While the Constitution establishes South Africans’ rights of the Public Finance Management Act. According to SASSA to social security, social assistance, and social services in (n.d.), “the principle aim of the Act is to make provision for the broad terms, and allocates responsibility to the national effective management, administration and payment of social and provincial spheres of government, the detailed assistance and service through the establishment of the South operationalisation of these rights occurs through various African Social Security Agency”. pieces of legislation. Thus, the DSD identifies the following pieces of legislation as comprising their legislative mandate SASSA outlines its mandate as being “to ensure (Department of Social Development, n.d.): the provision of comprehensive social security - The Aged Persons Act, 1967 (Act 81 of 1967); services against vulnerability and poverty within the - The Fund-raising Act, 1978 (Act 107 of 1978); constitutional and legislative framework” (SASSA, n.d.). - The Social Service Professions Act, 1978 (Act 110 of 1978); - The Child Care Act, 1983 (Act 74 of 1983); - The National Development Agency Act, 1998 (Act 108 of 1998); 17 The lead institution responsible for the Expanded Public CWP Reference Committees play a central role in Works Programme is the Department of Public Works bringing stakeholders together to ensure successful and Infrastructure, with the Minister seen as its “overall implementation at a given site. These committees provide champion” (Department of Public Works and Infrastructure, an advisory role only as they are not governance structures. n.d.b). According to the Department, it “derives its mandate from The committees are comprised of stakeholders from the the President’s call following the Cabinet Lekgotla held in July local community, representatives of local government, and 2006, the mandate which includes the eradication of poverty, community leaders. Membership of these committees is often unemployment, and underdevelopment” (Department of also extended to representatives of “local offices of provincial Public Works and Infrastructure, n.d.a). Within the Department government departments, such as Social Development, of Public Works and Infrastructure, the EPWP Branch is tasked Health, Education and Agriculture” (Department of Cooperative with “overall coordinating and implementing support, Governance, 2011, p.8). developing funding frameworks, providing technical support to participating public bodies and monitoring [and] evaluation” The National School Nutrition Programme is implemented (Department of Public Works and Infrastructure, n.d.b). by the Department of Basic Education, although it was initially the responsibility of the Department of Health when it However, EPWP involves government broadly, including was established (as the Primary School Nutrition Programme) all of its spheres and the state-owned enterprises and in 1994 (JET Education Services, 2016). Although the original therefore requires substantial cooperation between intention was that it would be superseded by other initiatives numerous institutions. The Department of Public Works implemented as part of the Reconstruction and Development and Infrastructure coordinates the infrastructure and non- Programme, in 2004 the Department of Education (as it was state sectors, while the environment and culture sector and called at the time) took over the programme, which became the social sector are coordinated and led by the Department known as the National School Nutrition Programme. Having of Environmental Affairs and the Department of Social initially targeted all primary schools in quintiles one, two and Development respectively. The EPWP also coordinates with three, coverage was gradually extended: in 2009, quintile one the Department of Higher Education and Training and various high schools were included, with quintile two and three high Sector Education and Training Authorities (SETA) with respect schools added in April 2010 and April 2011 respectively (JET to the training component of the programme (Department of Education Services, 2016). Public Works and Infrastructure, 2018a, p.15). The NSNP is viewed as being an educational intervention The structures responsible for the EPWP at the national in the first instance: the programme aims to “enhance the level are mirrored within each of the provinces. Provincial educational experience of the neediest primary school learners Departments of Public Works, led by Members of the Executive through promoting punctual school attendance, alleviating Councils for Public Works within provincial governments, short-term hunger, improving concentration, and contributing provide leadership at the provincial level. Within each of to general health development” (JET Education Services, these provincial departments exists an EPWP Unit; which is 2016). According the Department of Basic Education (2020) “instrumental in mobilising other provincial departments as well the programme’s objectives are “to provide nutritious meals to as municipalities within the province to perform in accordance learners so as to improve their ability to learn [and to teach] with the objectives of the EPWP” (Department of Public Works learners and parents on ways of living a healthy lifestyle, and and Infrastructure, n.d.b). promoting development of school vegetable gardens”. While the Community Work Programme forms part of the Since education is a provincial competence, the NSNP EPWP, it is coordinated by the Department of Cooperative is effectively implemented by provincial education Governance and Traditional Affairs (COGTA). However, departments, with funding provided by the national given the design of the programme it requires extensive department through the NSNP conditional grant. This intergovernmental cooperation. Thus, while the programme occurs through either a centralised or a decentralised model is managed and coordinated nationally, it is supported by the (JET Education Services, 2016). In the case of the centralised provinces and local governments. Indeed, local government model, the provincial education department appoints approval is required before a CWP site may be set up within service providers to source and deliver food to schools. In a given area. Furthermore, partnership with non-governmental the decentralised model, schools themselves are responsible organisations (NGO) is central to the CWP model: the CWP is for appointing local service providers. Food is prepared by implemented at the local level by implementing agents— volunteer food handlers (VFH), based on a ratio of 1 VFH for themselves non-profit NGOs—who partner with local non- every 200 learners; in small schools, a ratio of 1:125 is used governmental and community-based organisations. The CWP instead (Department of Basic Education, 2020). While led by the also requires close engagement with local communities, who DBE and provincial education departments, the NSNP is a multi- play a central role in terms of identifying useful work to be stakeholder programme which receives support from various performed. partners including the departments of health and agriculture, the private sector, and non-governmental organisations. As described by JET Education Services (2016), the “programme SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 18 operates at four levels: national (the DBE and partners); Further, more than 80 percent of (pre-)school-going children provincial ([provincial education departments] and partners); under the age of 20 were reported to attend a school at which district ([provincial education department] district officials a government feeding scheme operated in 2014/15. Coverage and partners); and school (principals, school management rates decline from 84.9 percent amongst the poorest 20 percent teams, [school governing bodies], NSNP Co-ordinators, NSNP of the population to 63.1 percent in quintile 4, and 40.1 percent committees, VFHs, and gardeners)”. amongst the richest 20 percent of the population. However, since targeting is done through schools, there are issues of poor As Figure 3.2 illustrates, between two-thirds and three- learners not being able to access the NSNP because they attend quarters of (pre-)school-going children aged six to 16 at quintile four and five schools (JET Education Services, 2016). least occasionally ate food provided by the NSNP in 2018. Figure 3.2. Access to School Feeding in South Africa, 2014/15 and 2018 90 84.9 81.2 80 75.8 75.7 73.0 72.5 74.1 72.8 73.2 69.9 71.2 69.0 67.6 66.6 70 63.1 58.9 50 52.7 45.8 40 40.1 Proportion (%) 30 30.1 20 10 0 1 2 3 4 5 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Pre-Transfer Quintile (2014/15) Age in years (2018) Source: Own calculations, Statistics South Africa (2015a, 2018c). Notes: (1) Figures for 2014/15 refer to the proportion of children under the age of 20 who report attending a school at which a government feeding scheme operates. (2) Figures for 2018 refer to the proportion of (pre-)school-going children who at least occasionally eat food provided by “the school feeding scheme/Government nutrition program”. 3.2. Social Assistance Programmes These eight programmes/ social grants are designed to address specific lifecycle and other risks, with particular 3.2.1. Overview emphasis on children—the care dependency, child support, and foster child grants—and the elderly—the Social assistance in South Africa currently encompasses older persons and war veterans’ grants, and the grant-in- eight key programmes, excluding the public works aid. These groups of people are those that cannot participate programs (and the 2020 Presidential Employment in the labour market and therefore face the risk of poverty. The Stimulus), school feeding, and the COVID-19 grant that system of individual (categorical) grants each targeting a specific was recently introduced in response to the economic risk-group stem out of the social and political dynamics of the effects of the pandemic and the consequent national country at the time the system was designed. All programmes lockdown. These programmes are the older persons grant, the are unconditional (do not impose any required actions or co- child support grant, the disability grant, the care dependency responsibilities of for recipients such as investments in human grant, the foster child grant, the war veterans’ grant, grant-in- capital or work-seeking) and are all means-tested based on aid, and social relief of distress. As will be shown, the system income (explicitly or based on proxies), except for the foster is dominated in numerical and budgetary terms by the child grant. The pros and cons of the overall system design older persons, child support, and disability grants. and composition is discussed further in chapters 5 and 6. The remainder of section 3.2 provides an overview of each of the eight social assistance programmes mentioned above, as well as of the COVID-19 grant. The aim has been to provide a 19 description of the purpose of the grant, the eligibility criteria, applicants are single, they must earn no more than R 52 800 per and the current grant values. annum in order to qualify for the grant; for married applicants, the earnings threshold is raised to R105 600 (SASSA, 2020c). 3.2.2. Older Persons Grant The value of the child support grant is R450 per month The older persons’ grant is a non-contributory means- (R 5 400 per annum) from 1 October 2020 making it the tested pension that was at different times in the post- lowest value grant amongst the pre-COVID-19 suite of apartheid period known by different names. The older social grants (SASSA, 2020c). The child support grant was persons’ grant is accessible from the age of 60 years, paid out to caregivers on behalf of approximately 12.7 million provided that the individual is not cared for in a state children in the 2019/20 financial year (National Treasury, 2020b, institution. To qualify for the grant individuals must be South p.294). African citizens, permanent residents or refugees, and should not be in receipt of any other social grant for themselves. In 3.2.4. Disability Grant other words, to qualify for the older persons’ grant, individuals should not be direct beneficiaries of any other grant. Income support is provided to individuals between the ages of 18 and 59 years with permanent or temporary As noted, the older persons’ grant is means tested. From disabilities through the disability grant. In order for April 2020, eligibility is restricted to age-eligible individuals individuals to qualify for the disability grant, they must provide earning less than R86 280 per annum if they are unmarried, a medical report confirming severe permanent physical or and whose assets do not exceed R 1 227 600; for those who are mental disability; this report must not be older than three married, the applicable limits are R172 560 per annum and R2 months at the time of application. Further, applicants must be 455 200 (National Treasury, 2020b, p.300). South African citizens, permanent resident or refugees; they must be resident in the country; and they may not be a direct The older persons’ grant has two benefit levels: beneficiary of any other grant. Individuals cared for in state individuals aged 60-75 years receive R1 860 per month institutions are not eligible for the disability grant. (R22 320 per annum), while those above the age of 75 receive R1 880 per month (R22 560 per annum) (SASSA, Disability grants may be permanent or temporary. 2020c). The grant is paid on a sliding scale, with higher private Temporary disability grants may be valid for between six and 12 income leading to a lower grant value. months, after which the individual would need to reapply if he or she has not returned to work (GroundUp, 2017). Permanent Approximately 3.7 million people received the older disability grants may be reviewed every 12 months to determine persons’ grant in the 2019/20 financial year, making it continued eligibility (Social Security Administration, 2019). the second largest grant in terms of the number of direct beneficiaries (National Treasury, 2020b, p.294). As with the older persons grant, the disability grant is means tested on both income and assets. Single recipients 3.2.3. Child Support Grant of the grant may not earn more than R86 280 per annum or have assets in excess of R1 227 600. For married recipients, the The child support grant provides income support to income and asset thresholds are doubled to R172 560 and R2 parents and caregivers of children under the age of 455 200 respectively (National Treasury, 2020b, p.300). 18 years and is the country’s largest social assistance programme by number of beneficiaries. Established in The disability grant is valued up to R1 860 per month 1998, the child support grant was initially available only to (R22 320 per annum), on par with the older persons’ children under the age of seven. These grants are applied for grant (SASSA, 2020c). The grant is paid on a sliding scale, by the parent or primary caregiver on behalf of the child; the with higher benefit levels for households with lower private parents and caregivers are deemed the grant recipients, while income. In the 2019/20 financial year, approximately one million the children are the beneficiaries. In order to qualify for the child disability grants were paid out, making it the third largest social support grant, the primary caregiver must be a South African grant in terms of the number of beneficiaries (National Treasury, citizen, permanent resident or refugee, and both caregiver and 2020b, p.294). child must reside in the country. Where primary caregivers are not the child’s biological parents, they are required to prove 3.2.5. Care Dependency Grant their status as primary caregiver. While there is no limit to the number of biological children for whom a parent may receive The care dependency grant is aimed at supporting the the child support grant, primary caregivers may not apply for care of children under the age of 18 years with mental the grant for more than six non-biological children. Children or physical disabilities. To be eligible, the child must have who are cared for in state institutions are not eligible for this been found to have a disability that is both permanent and grant. severe, and caregivers applying for the grant must avail the child for assessment by a medical officer. Caregivers applying The child support grant is also means tested, although for the care dependency grant are required to be South African only on self-declared (with evidence) income. Where citizens, permanent residents or refugees. Both the caregiver(s) SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 20 and the child in question must be resident within the country. War veterans’ grants provide R1 880 per month to Children who are cared for in a state institution on a permanent beneficiaries, which is R10 per month higher than the basis are not eligible for the care dependency grant. upper tier value of the older persons’ grant (SASSA, 2020c). Over a year, the war veterans’ grant is worth R22 560. Eligibility for the care dependency grant is subject to an As of the 2019/20 financial year, just 78 veterans were receiving income-based means test, except in the case of foster war veterans’ grants (National Treasury, 2020b, p.294). parents. The current means test for the grant limits eligibility to single caregivers earning no more than R223 200 per annum 3.2.8. Grant-in-Aid and double that amount (R446 400) for married caregivers (National Treasury, 2020b, p.300). The grant-in-aid is an additional benefit to beneficiaries of either the older persons, disability, or war veterans Like the older persons grant and the disability grant, the grants who require someone to provide regular value of the care dependency grant is currently set at R1 attendance for them due to their physical or mental 860 per month (R22 320 per annum) (SASSA, 2020c). In disabilities. The grant-in-aid is therefore not a standalone the 2019/20 financial year, almost 155 000 disabled children grant. Proof of the disability is required in order to access the were covered by the care dependency grant (National Treasury, grant, while those who are cared for in a state-subsidised 2020b, p.294). institution are ineligible. There are no additional eligibility criteria, although applicants would have already complied with 3.2.6. Foster Child Grant the specific criteria for whichever of the older persons, disability or war veterans’ grant they receive. For children under the age of 18 years who have been placed in foster care by the courts, the foster child grant The grant-in-aid is valued at R450 per month (R5 400 per is available to foster parents who are South African annum) from 1 October 2020 (SASSA, 2020c). Almost a quarter citizens, permanent residents or refugees. Further, both of a million (247 000) adults received the grant-in-aid in the the foster parent applying for the grant and the child in question 2019/20 financial year (National Treasury, 2020b, p.294). are required to reside within South Africa. Foster parents are no longer entitled to the grant once the child leaves their care. The 3.2.9. Social Relief of Distress foster child grant is unique amongst other South African grants in that eligibility is not conditioned on meeting a means test. Social relief of distress is “the temporary provision of assistance intended for persons in such dire need that The value of the foster child grant is currently R1 040 they are unable to meet their or their families’ most basic per month (R12 480 per annum) (SASSA, 2020c). As of the needs” (SASSA, 2020c). Beneficiaries must be South African 2019/20 financial year, just over 351 000 children were covered citizens, permanent residents or refugees, and must be resident by the foster child grant (National Treasury, 2020b, p.294). in South Africa. In order access social relief of distress, individuals must meet at least one of the following criteria: 3.2.7. War Veterans’ Grant - They are waiting for payment of an approved social grant; The war veterans’ grant is currently targeted at individuals - They have been found to be medically unfit to work for who fought in World War II or the Korean War. As such, this pay in the short-term (less than six months); grant is currently ‘ageing out’ of the system and is now by far - The household’s breadwinner has died within the South Africa’s smallest social assistance programme in terms of 12-month period preceding the application; numbers of beneficiaries. In order to be eligible for this grant, - The household’s breadwinner has been admitted to a veterans must be South African citizens or permanent residents, public or private institution for at least one month; and must be resident within the country. Further, veterans must - They have been affected by a disaster as per the Disaster be at least 60 years old (a criterion that is now non-binding in Management Act or the Fund Raising Act of 1978; or practice) or disabled, should not be a direct beneficiary of any - “Refusal of the application…will cause undue hardship” other social grant, and may not be cared for in a state institution. (SASSA, 2020c). Eligibility for the war veterans grant is subject to an Social relief of distress is approved for a maximum period income- and asset-based means test. Single applicants may of three months, although extensions for an additional not earn more than R86 280 per annum or have assets in excess three months can be made in exceptional circumstances. of R1 227 600; for married applicants the respective thresholds Individuals who receive other grants are not eligible for social are doubled to R172 560 and R2 455 200 (National Treasury, relief of distress. 2020b, p.300). This means test is consistent with that for the older persons grant, for war veterans would also be eligible. Unlike other grants, social relief of distress can take various forms. For example, it may be issued in cash as income support, but it may also take the form of food parcels. 21 In essence, then, the social relief of distress grant is - Not reside in a government-funded or -subsidised a flexible social assistance intervention that allows institution. government to deal with conventional temporary situations of need and to respond rapidly in emergency However, there is an important grey area in terms of the situations, such as natural disasters. Social relief of distress requirement to be unemployed as government has only a has, for example, been a key way through which government limited ability to confirm that an individual is unemployed. has provided support in the context of the COVID-19 lockdown. Certainly, government may be able to detect income-related Further, the COVID-19 grant discussed below is implemented as financial flows through applicants’ bank accounts, to identify social relief of distress in the form of cash income support. individuals currently paying income tax or unemployment insurance contributions, and to know whether individuals are 3.2.10 .COVID-19 Social Relief of Distress Grant receiving unemployment benefit or NSFAS stipends. However, it cannot distinguish the economically inactive from the As its name suggests, the COVID-19 Social Relief of unemployed, or from those working for cash in the informal Distress Grant is an ad hoc intervention to address the sector. The consequence at the onset of this grant was economic fallout of the national lockdown. Aiming to that the pool of potential beneficiaries of this grant was reach working-age individuals who are unable to access other massive, estimated between 8-15 million people, and forms of assistance, whether COVID-19-specific or conventional the final numbers of beneficiaries would be a function of interventions, the grant has a relatively broad set of eligibility SASSA’s ability to process applications. criteria. Specifically, applicants for the COVID- 19 grant must: - Be a citizen or permanent resident of South Africa, or a The value of the COVID-19 grant was set at R350 per refugee, special permit holder, or asylum seeker2; month for 6 months: May-October 2020, equivalent to R - Be a resident within the country; 4  200 per annum. In October 2020 grant was extended for - Be unemployed; three more months until the January 2021. In their review of - Be over the age of 18 years; the COVID-19 response, the Auditor-General (2020) reports that - Not be in receipt of any income; R4.318 billion—just over one-third of the amount budgeted - Not be in receipt of a social grant of any kind; for the six-month intervention—had been paid out to a total - Not be receiving or eligible to receive benefits from the of around 6 million approved applicants between May and Unemployment Insurance Fund (UIF); November 2020. - Not be receiving a stipend from the National Student Financial Aid Scheme (NSFAS); and 3.2.11 Summary 2 In June 2020 asylum seekers whose Section 22 visas or permits were valid as of Table 3.2 provides an overview of each of the grants discussed March 15, 2020 were also accepted to receive the COVID-19 grant in addition to special permit holders from Lesotho, Angola and Zimbabwe. above. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 22 Table 3.2. South Africa’s Social Grants, 2020 Older Child Care Social relief of Disability Foster child War veterans Grant-in-aid COVID-19 persons support dependency distress Nationality Applicant Applicant Applicant Applicant Foster parent Applicant None Applicant Applicant must be South must be South must be South must be South must be South must be South specified, but must be South must be South African citizen, African citizen, African citizen, African citizen, African citizen, African citizen, implied African citizen, African citizen, permanent permanent permanent permanent permanent permanent permanent permanent resident or resident or resident or resident or resident or resident or resident or resident, refugee refugee refugee refugee refugee refugee refugee refugee or asylum seeker, special permit holders from Lesotho, Angola or Zimbabwe Residence Resident in Resident Resident in Resident in Resident in Resident in None Resident in Resident in South Africa in South South Africa South Africa South Africa South Africa specified, but South Africa South Africa Africa (both (both caregiver (both foster implied caregiver and and child) parent(s) and child) child) 23 Age restrictions Age 60 and Child aged Aged 18-59 Child under 18 Child under 18 Age 60 and None None Above the age above 18 years or years years years above (or specified, but of 18 years younger disabled) implied (i.e. 18 years or older) Identity Identity Identity Identity Identity Identity Identity Identity None Identity confirmation document or document document or document or document document or document or document or smart ID card or smart ID smart ID card smart ID card; or smart ID smart ID card smart ID card smart ID card card; birth birth certificate card; birth certificate for for care- certificate for child dependent foster child child Documentary Confirmation Report Report Court order Documentation None, but evidence that applicant confirming confirming indicating only for specific SASSA required is the child’s disability child’s foster care cases verification of primary within three permanent status information caregiver months of date severe disability in other of application government databases Older Child Care Social relief of Disability Foster child War veterans Grant-in-aid COVID-19 persons support dependency distress Means-tested Yes Yes Yes Yes (excl. foster No Yes None No None, but parents) specified, but SASSA implied verification of information in other government databases including the South African Revenue Service (SARS) Income-based R86 280 per R52 800 R86 280 R223 200 None R86 280 per None None None means test annum (single); per annum per annum per annum annum (single); specified, but threshold R172 560 (single); (single); R172 (single); R446 R172 560 implied per annum R105 600 560 per annum 400 per annum per annum (married) per annum (married) (married) (married) (married) Asset-based R1 227 600 None R1 227 600 None None R1 227 600 None None None 24 means test (single); R2 455 (single); R2 455 (single); R2 455 specified, but threshold 200 (married) 200 (married) 200 (married) implied Beneficiary Older person Child Adult with a Child Child War veteran Adult, older Individuals or Unemployed disability person or war households adult with no veteran, with income a disability SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW Value (as of 1 R1 860 per R440 per R1 860 per R1 860 per R1 040 per R1 880 per R440 per No set value R350 per April 2020), per month for month month month month month month month for beneficiary 60-75 year- May-October olds; R1 880 2020 only but per month for in October the those over 75 grant period years was extended by 3 months until January 2021. Older Child Care Social relief of Disability Foster child War veterans Grant-in-aid COVID-19 persons support dependency distress COVID-19 Between May In May 2020, Between May Between May Between May Between May Between May N.A. N.A. adjustments and October the grant was and October and October and October and October and October 2020, the grant increased 2020, the grant 2020, the grant 2020, the grant 2020, the grant 2020, the was increased by R300 per was increased was increased was increased was increased grant was by R250 per beneficiary (i.e. by R250 per by R250 per by R250 per by R250 per increased beneficiary. per child). For beneficiary. beneficiary. beneficiary. beneficiary. by R250 per the months of beneficiary. Jun-Oct 2020, an additional R500 was paid per recipient (i.e. per caregiver, who may be receiving child support grants for multiple children). Sliding scale Yes No Yes No No Yes No No No 25 applicable Eligible if No No No No No No Only available No No receiving other if receiving grants older persons, disability or war veterans’ grant Older Child Care Social relief of Disability Foster child War veterans Grant-in-aid COVID-19 persons support dependency distress Other Not available Child must Must have Must require Range of Applicant may restrictions for more than remain in the fought in World regular specific not receive six non- care of the War II or the attendance circumstances in any income, biological or foster parent(s) Korean War by another terms of which any social non-adopted person due individuals grant, any UIF children to physical are eligible, benefits, or or mental including that any stipend disability refusal of the from NSFAS; application Applicant may would cause not qualify for undue hardship UIF benefits Ineligible Ineligible if Ineligible Ineligible if Ineligible if Ineligible Ineligible if if cared for child cared if cared for child cared cared for in a if cared for resident in a in a state for in a state in a state for in a state state institution in state- government- institution institution institution institution subsidised funded or institution -subsidised caring for institution beneficiary 26 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 3.3 Resourcing for Social Assistance in South percent). While the preceding section clearly shows that social Africa protection generally, and social assistance in particular, requires a substantial amount of resources from the state, this section South Africa is one of the biggest spenders globally on puts these figures in an international context. It is clear that social assistance as a share of GDP. The country allocates there is wide variation in the share of GDP allocated to social 3.3 percent of GDP to social assistance, the fourth-highest assistance around the world. However, the vast majority of share in Sub-Saharan Africa and the tenth-highest share countries are within the range of zero percent to four percent of of all countries for which there is data. details the level GDP as just five countries have shares over four percent. These of spending on social assistance as a share of GDP across 124 are South Sudan (10.1 percent of GDP), Lesotho (6.4 percent), countries globally. At 3.3 percent of GDP, this places the country Georgia (7.0 percent), Timor-Leste (6.5 percent), and the Ukraine higher than many of its peers, including Argentina (2.1 percent (4.4 percent). At the opposite end of the spectrum, Côte d’Ivoire of GDP), the Russian Federation (1.9 percent), Mexico (1.7 and Papua New Guinea allocate just 0.01 percent of GDP to percent), India (1.5 percent), Brazil (1.4 percent), and China (0.8 social assistance spending. Figure 3.3. Spending on Social Assistance as Share of GDP, 2009-2016 10.0 9.0 8.0 7.0 Share of GDP (%) 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Sub-Sahara Africa Eur.& Central Asia E.Asia & Pacific L. America & Carib MENA S. Asia Source: World Bank (2020a). Note: Most recent estimates for 2009-2016 for 124 countries. explores the level of resource allocation in further detail, GDP on school feeding, and 6.3 times the median share disaggregating the total by type of programme and comparing on public works. Compared with other upper-middle these with figures for Sub-Saharan Africa, upper-middle income income countries, South Africa devotes relatively large countries, and all countries for which there is data. The high proportions of GDP to social pensions and unconditional proportion of GDP allocated to social assistance in South Africa cash transfers, respectively 6.2 times and 4.7 times the is reconfirmed: South Africa’s 3.3 percent share is 3.4 times medians for this group of countries. Indeed, resource the median for Sub-Saharan African countries, 2.4 times the allocations to the social pension stand out across all three median for upper-middle income countries, and 2.9 times the country groupings, indicating the uniqueness of the older global median. Compared with other countries in Sub- persons grant within a global context. Saharan Africa, South Africa is an outlier in terms of its resource allocations to unconditional cash transfers; at 1.27 percent of GDP it is 14.1 times the regional median. The country also spends 6.5 times the median share of 27 Table 3.3. Social Assistance Spending by Programme as Percent of GDP, 2009-2016 South Sub-Saharan Africa Upper-Middle All Africa Income Median Ratio Median Ratio Median Ratio All social assistance 3.31 0.98 3.4 1.37 2.4 1.14 2.9 Unconditional cash transfers 1.27 0.09 14.1 0.27 4.7 0.17 7.5 Social pension 1.68 0.00 - 0.27 6.2 0.04 42.0 School feeding 0.13 0.02 6.5 0.00 - 0.00 - Public works 0.22 0.04 6.3 0.00 - 0.00 - Food and in-kind 0.01 0.04 0.3 0.01 1.0 0.02 0.5 Fee waivers 0.00 0.00 - 0.00 0.0 0.00 - Other social assistance - 0.00 - 0.01 - 0.01 - Source: World Bank (2020a). Notes: (1) Most recent estimates for 2009-2016 for 124 countries. Ratios are calculated as the value for South Africa divided by the relevant median. (2) Programme figures do not add to the total for all social assistance due to differing data availability within countries across programmes. The set of social assistance programmes implemented points per annum higher than the growth rate of total spending by the South African government requires a substantial (3.3 percent); it also represents somewhat faster growth than commitment of resources on an ongoing basis, even if occurred for health (3.3 percent per annum) and education (3.6 grant values are relatively low especially for the Child percent). In the case of the latter, at least part of this growth Support Grants. For the system to be sustainable over time can be linked to the phased-in implementation of free tertiary and for it to be possible to extend coverage or raise benefit education during the latter part of the period. Due to the above levels, it is critical that the state is able to raise sufficient average growth rate of spending, social protection has resources for—and allocate them to—social assistance. seen its share of consolidated government spending rise from 15.2 percent in 2010/11 to 16.2 percent in 2019/20. Social spending, broadly defined to include health, education, and social protection has generally accounted Within this context of a relatively robust rate of increase in for just under one-half of total government spending spending on social protection, illustrates how real expenditure over the past decade, or between 14 percent and 17 on social grants has increased since 2006/07. Overall spending percent of GDP. Of these three sets of expenditure, education on social grants increased by 2.9 percent per annum in real is the largest, averaging 21.0 percent of total government terms over the period, rising from R122.1 billion to R172.8 billion spending (Table 3.4). Education is followed by social protection, Rand in March 2020 prices. Amongst the current suite of which has averaged 15.4 percent of government spending over social grants, the three largest in terms of spending are the period, while health has averaged 12.4 percent. the older persons grant, the child support grant, and the disability grant. In 2018/19, R75.0 billion in March 2020 Rand Of the three types of spending, social protection spending was transferred to households through the older persons’ grant, has grown the most rapidly over the nine-year period, while a further R64.4 billion and R23.4 billion were transferred averaging 3.7 percent in real terms. This is 0.4 percentage through the child support and disability grants. Table 3.4. Consolidated Government Spending, 2010/11-2019/20 Financial Year Spending (R billions) Share (%) Total Health Education Social Health Education Social Combined Protection Protection Total Nominal R billions 2010/11 839.0 101.7 171.7 130.7 12.1 20.5 15.6 48.2 2011/12 922.0 115.1 197.4 140.4 12.5 21.4 15.2 49.1 2012/13 1 001.9 124.2 213.7 152.1 12.4 21.3 15.2 48.9 2013/14 1 095.7 133.0 230.4 170.3 12.1 21.0 15.5 48.7 2014/15 1 144.0 143.8 246.4 146.7 12.6 21.5 12.8 46.9 2015/16 1 303.2 154.8 265.1 201.7 11.9 20.3 15.5 47.7 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 28 Financial Year Spending (R billions) Share (%) Total Health Education Social Health Education Social Combined Protection Protection Total 2016/17 1 379.8 172.7 286.8 219.4 12.5 20.8 15.9 49.2 2017/18 1 479.4 188.2 304.8 235.3 12.7 20.6 15.9 49.2 2018/19 1 591.4 203.6 341.5 259.3 12.8 21.5 16.3 50.5 2019/20 1 781.2 217.4 375.7 288.4 12.2 21.1 16.2 49.5 Real R billions, March 2020 prices 2010/11 1 358.9 164.8 278.1 211.7 12.1 20.5 15.6 48.2 2011/12 1 414.8 176.6 303.0 215.5 12.5 21.4 15.2 49.1 2012/13 1 456.6 180.5 310.7 221.2 12.4 21.3 15.2 48.9 2013/14 1 505.6 182.7 316.6 234.0 12.1 21.0 15.5 48.7 2014/15 1 488.1 187.1 320.6 190.8 12.6 21.5 12.8 46.9 2015/16 1 611.9 191.5 327.9 249.5 11.9 20.3 15.5 47.7 2016/17 1 605.6 201.0 333.8 255.3 12.5 20.8 15.9 49.2 2017/18 1 644.0 209.1 338.7 261.5 12.7 20.6 15.9 49.2 2018/19 1 689.9 216.2 362.6 275.4 12.8 21.5 16.3 50.5 2019/20 1 815.5 221.6 382.9 294.0 12.2 21.1 16.2 49.5 Average annual growth rate (2010/11-2019/20) Nominal 8.7 8.8 9.1 9.2 Real 3.3 3.3 3.6 3.7 Source: National Treasury (2014a, 2015a, 2016a, 2017a, 2018a, 2019a, 2020a); Statistics South Africa (2020b). Note: Nominal expenditures deflated using average headline CPI for April to March of each year. Updated figures based on the Supplementary Budget have not yet been released. Spending as a share of GDP is presented in Table B.1 in the appendix. Figure 3.4. Real Spending on Grants, 2006/07-2018/19 (log scale) 256.0 166.9 172.8 156.0 158.7 161.6 141.7 151.1 150.6 125.7 133.3 122.1 123.7 128.0 71.4 75.0 63.8 65.7 67.9 57.0 58.8 60.5 64.0 50.2 54.7 64.4 45.4 45.1 46.1 59.9 62.1 55.4 56.9 58.5 52.7 54.4 49.1 Spending (billion) 44.8 37.6 38.8 39.7 32.0 30.5 30.2 29.3 27.9 27.3 26.7 25.6 24.4 24.4 23.7 23.2 23.2 23.4 16.0 10.6 10.6 11.0 11.2 11.2 11.0 10.8 10.7 10.4 10.2 10.0 8.5 8.0 Total Child support 4.0 Older persons Disability Other 2.0 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 Source: Own calculations, SASSA (2019); Statistics South Africa (2020b). Notes: Grants included under ‘Other’ are the care dependency grant, the foster child grant, grant in aid, social relief of distress, and the war veterans’ grant. Spending figures are deflated to March 2020 prices using average headline CPI for April to March of each year. Full details are available in Table B.2 in the Appendix. 29 Both the older persons’ grant and the child support grant distress) increased by 2.9 percent on average per annum, the have seen above average growth in real spending over number of direct grant beneficiaries (or the number of grants the period: real spending on these two grants increased paid out) increased by 3.3 percent. This indicates a decline in by an average of 4.3 percent and 4.6 percent per annum the average monetary value of grants paid out by the respectively. This is in line with the relatively strong South African government over the 12-year period and increases in beneficiaries observed for both of these is largely the result of a shift in the composition of grants grants during this period of 4.1 percent and 3.9 percent towards low value child support grants as access to this respectively (see discussion of ). In contrast, spending via the grant increased. Indeed, the data indicates that the average disability grant has declined by almost a quarter in real terms annual grant paid per beneficiary fell from R10 153 in 2006/07 over the 12 years, equivalent to an annual contraction of 2.2 to R9 676 in 2018/19 in March 2020, a decline of 4.7 percent. percent on average. This fall in spending, alongside declines However, over time the values of the major grants have gradually in the much smaller foster child grant and war veterans’ grant, increased in real terms. The older persons’ and disability grants has served to dampen the overall increase in spending over the have increased from R1 641 in April 2020 in October 1994 to period. R1 860 in April 2020, a growth rate of approximately 0.5 percent per annum in real terms. The child support grant has increased On average, though, spending on grants has not kept from R319 in April 2020 when it was implemented in July pace with the growing number of beneficiaries (): while 1998, to R450 in October 2020, equivalent to real growth of 1.4 total spending across all grants (excluding social relief of percent per annum. Figure 3.5. Number of Grants, 2006/07-2018/19 32.00 16.99 17.20 17.51 17.81 16.11 15.93 16.64 14.94 15.60 16.00 13.07 14.06 12.02 12.42 11.97 12.08 12.27 12.45 11.34 11.70 10.93 11.13 10.37 9.57 8.77 7.86 8.19 Number of Grants (millions) 8.00 4.00 3.30 3.42 3.55 3.09 3.19 2.87 2.97 2.68 2.75 2.55 2.39 2.20 2.23 2.00 1.42 1.41 1.29 1.26 1.20 1.20 1.16 Total 1.12 1.11 1.09 1.07 1.06 1.05 Child support 1.00 0.74 0.74 0.75 0.76 0.76 Older persons 0.72 0.73 0.72 0.68 0.68 0.60 0.63 Disability 0.53 Other 0.50 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 Source: Own calculations, SASSA (2019). Notes: Grants included under ‘Other’ are the care dependency grant, the foster child grant, grant in aid, and the war veterans’ grant. Full details are available in Table B.2 in the Appendix. In the 2018/19 financial year, SASSA paid out 17.81 million of the older persons grant and the child support grant. grants, up from 13.07 million a decade earlier and 14.94 million Beneficiaries of these two grants increased by 5.95 million over in 2010/11. Given the national population of 57.94 million the 12 years considered here, compared to the overall increase in mid-2018 (Statistics South Africa, 2019a), this suggests a of 5.8 million. In contrast, the number of disability grant coverage rate of 30.7 percent of the South African population. beneficiaries fell by almost 375 000, outweighing the increase An estimated 75 percent of the grant recipients are of 225 000 observed for the remaining grants. This latter women. Women are the bulk (97 percent) of the recipients of increase was primarily driven by an increase in the number of the (12 million) child support grants and they are roughly half beneficiaries of the grant-in-aid, which grew by an average of of the beneficiaries of the other social programs (approx. 12 17.5 percent per annum over the 12 years from 32 000 to 222 million beneficiaries). 000. The overall increase in the number of grant beneficiaries The data on spending and numbers of beneficiaries is clearly driven by growing numbers of beneficiaries reveal the dominance of the older persons’ grant, the SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 30 child support grant, and the disability grant within the growth performance, particularly in the decade preceding the social grants system. Together, these three grants accounted global financial crisis. However, South Africa has failed to rebuild for 94.2 percent of total spending on grants and 69.9 percent fiscal buffers since the global financial crisis. Fiscal balances of the total number of grants in the 2018/19 financial year. This have deteriorated as a result of rising expenditures, especially high proportion of spending relative to the number of grants is transfer to public corporations, the public sector wage bill, driven by the fact that the older persons’ grant and the disability and debt service payments (Figure 3.7). As a result, while the grant are two of the highest value grants, as shown in Table 3.2. country generated budget surpluses in 2006/07-2007/08, since 2009/10 it has maintained large budget deficits of between For the first 15 years of the post-apartheid era, there 3.8 percent and 6.5 percent of GDP. The projected deficit of 6.8 was increasing fiscal space available to government to percent of GDP in the 2020/21 financial year does not account roll out new programmes and expand existing ones. This for the economic fallout of the COVID-19 pandemic, with more was particularly true once government debt levels had been recent estimates putting the deficit as high as 15.7 percent of stabilised and later reduced, and coincided with the improving GDP in 2020/21. Figure 3.6. Main Budget Aggregates for South Africa since 2001/02 (percent of GDP) 40.0 36.0 32.0 28.0 24.0 20.0 Percent of GDP 16.0 12.0 8.0 4.0 0.0 -4.0 Total revenue -8.0 Total expenditure Main budget balance -12.0 Debt-service costs -16.0 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18* 2018/19* 2019/20* 2020/21* 2021/22* 2022/23* Source: National Treasury (2020a, c). Note: Data for 2017/18 and 2018/19 are preliminary outcomes; data for 2019/20 are revised estimates; and data for 2020/21 onwards are medium-term estimates; these are indicated by an asterisk (*). Estimates from 2020/21 onwards that are in grey are from the 2020 National Budget, while those in colour are from the Supplementary Budget released in June 2020. A key concern in this regard has been the increase in the reveals the need to realign the composition of spending from public sector wage bill. Real public sector salaries increased consumption towards investment and to reduce budget by around 40 percent during the past 12 years (National deficits. The consolidation plan relies principally on a significant Treasury, 2020a), with compensation of employees projected reduction of the compensation of public sector employees. to represent 32.8 percent of consolidated public spending However, it also presents conservative assumptions for social in 2020/21 (National Treasury, 2020a). At the same time, debt protection spending, which is assumed to grow by about 2.2 service costs have been the fastest rising budget category percent per annum over the next few years in nominal terms. and are expected to reach 4.8 percent of GDP in 2020, risking This holds important implications for spending on social crowding out much needed social expenditures. The projected assistance and the DSD and SASSA will need to actively pursue impact of the pandemic on government finances is clear from solutions, such as improving efficiencies, that reduce the impact the figure. Together, the expected impacts on revenues and of budget cuts on grant recipients. expenditures imply a budget deficit of 15.3 percent of GDP in 2020/21, more than twice the original estimate. Thus, the Consolidated government spending has increased by an projected budget deficits for the next three financial years are average of 8.7 percent per annum in nominal terms, or higher than any recorded since 2001/02. 3.3 percent per annum in real terms between 2010/11 and 2019/20 (; spending as a share of GDP is presented in The deteriorated fiscal situation and need for Table B.1 in the appendix). According to figures presented consolidation to restore debt sustainability implies in the national budget in February 2020, total government trade-offs in government expenditures over the next spending reached R1.78 trillion in 2019/20, which is up by few years. The 2020 Medium-Term Budget Policy Statement roughly one-third in real terms since 2010/11. 31 3.4. Administration and Delivery of Social National and local level have been moving towards an Protection and Social Assistance integrated approach to service delivery to poor and vulnerable people, but it still remains fragmented. One 3.4.1. Institutions and coordination of social assistance of NDP 13 commitment is to build uniformity across social development system and that consistency is maintained Social assistance in South Africa relies on an intricate across different spheres through the monitoring framework network across the three spheres of government, its embedded in the five-year strategies and the annual agencies, and partnership with implementers such as state- performance plans. However, there is no integrated monitoring funded institutions, Non-Governmental Organisations (NGOs), and evaluation system with national level and monitoring is Community-Based Organisations (CBOs), and Faith-Based still heavily paper based. While provincial departments monitor Organisations (FBOs) to deliver services to vulnerable people the same indicators across the five programmatic areas,5 it is and communities. As noted above, the national Department not clear how they link with national targets and indicators. of Social Development (DSD) has overall responsibility and The development of a joint data management system would accountability for provision of social assistance to reduce enable all levels to access information, but such a system is poverty, vulnerability, and the impact of HIV and AIDS guided still a long way away, although it is contained in the national by the ‘Batho Pele’ principle that places people at centre of strategy 2015-2020. service delivery (Department of Social Development, n.d.). It also has the responsibility for national legislation, the overall To an extent, the three spheres of government provide policy environment and to coordinate. similar services and programs at varying scale and impact, and there are some overlaps and duplication At the national level, social assistance is delivered and of some programs. Decentralisation brings services closer monitored through five programmatic areas in the Strategy to people in line with ‘Batho Pele’ principles and benefit poor Plan 2015-2020: i) administration manages governance risk and and vulnerable families, but it can be often confusing for monitoring and evaluation component of the department; ii) applicants to know where to go to access services. A similar social assistance is responsible for the delivery of the eight offer of programs across provinces and local districts such as grants; iii) social security policy and administration enables youth programs and extended public works may limit impact development of policy and removes barriers to access for and reach with beneficiaries double dipping across programs. beneficiaries, iv) welfare services policy puts systems in place Coordination could be improved. For instance, each province for efficient delivery of social services; and, and v) social policy registers early childhood development centres (ECDs) and issues and integrated service delivery supports community-based best practices and conditional grants, but these functions are interventions and provides the research for evidence based also provided by national DSD creating duplication of functions policy making. The baseline indicators are set every five years and added admin for ECDs, which are NPOs government in the national strategy and implemented through the Annual relies on for service delivery. Nutrition programs and relief to Performance Plans (Department of Social Development, 2015). vulnerable people are also administered at the three spheres. The other services include shelter and programs for homeless, Provincial departments follow similar programmatic old age homes, for people with disabilities and those requiring categorisation; administration, social welfare services, assistance to substance abuse, victims of gender-based children and families, restorative services and violence and the prevention thereof. development and research, with a heavier emphasis on individual, household, and community welfare programs DSD established the South African Social Security to reflect this proximity to grass root challenges (Western Agency (SASSA) mandated by the South African Social Cape Department of Social Development, 2020; Gauteng Security Act of 2004 to ensure an effective and efficient Department of Social Development, 2020). The decentralised administration, management, and payment of social provincial and municipal structures coordinate with national assistance (Department of Social Development, 2019, p.54). DSD to an extent and largely independent in program design SASSA is regulated, operationalised and reports under the and implementation. Provincial departments have additional social assistance of DSD’s five programme areas (Department policies and laws3 guided and consistent with the constitution of Social Development, 2019, p. 43). As of 2019, the agency has and national policies.4 a network of 9 provincial offices, 46 district offices, 389 local offices, 1163 service points, and 1740 pay points nationwide (SASSA, 2019a, p.12). 3 Example: Western Cape Commissioner for Children’s Act, (2/ 2019) Section 78 of the Constitution of the Western Cape, 1997, establishes the office of a provincial Commissioner for Children and provides that the Commissioner must assist the Western Cape Government in protecting and promoting the rights, needs, and the interests of children in the Province. Western Cape Strategic Plan 2020-2025. P 15. 4 Examples: The Gauteng Strategic Policy Framework on Gender Equality and Women Empowerment (Gauteng Annual Report 2019/20. Western Cape Provincial Strategy for the Provision of Child and Youth Care Centres (CYCCs) (2016) in the Western Cape Department of Social Development Strategy 2020- 5 Based on Strategic Plans, Annual Performance Plans and Annual Reports for 2025. Limpopo, Gauteng, and Western Cape. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 32 Despite the heavy footprint of offices throughout the country, The SOCPEN interfaces with the Department of Home beneficiary access to services remains a challenge to the Affairs ID system and a number of other government extent that SASSA and the department are implementing databases such as the government’s payroll system, the Project Mikondzo and the Integrated Community Registration Unemployment Insurance Fund, the National Treasury, Outreach Programme (ICROP) to improve the service delivery, and the Department of Basic Education learner database. and to increase access to social assistance to poor beneficiaries However, it does not interface with the SARS database with (Department of Social Development, 2019, p. 81). income information and also not with databases of the Department of Employment and Labour, and Public Works But there are issues related to capacity within SASSA. which have information on jobseekers and those benefitting SASSA (2019, p.88) reported a vacancy rate of 55.9 percent as from public employment programs. SOCPEN includes data for of 31 March 2019. This is a problem experienced broadly across all major social grant applicants (eligible and non-eligible) and the organisation and in critical areas. For example, vacancy rates all the payments made to them. The COVID-19 top-ups of the were reported to be 67.5 percent in Fraud and Compliance, 61.8 existing social grants are all recorded in SOCPEN except for the percent in Internal Audit and Risk Management, 65.5 percent special new COVID-19 grant. in Information and Communications Technology, and 77.8 percent in Strategy and Business Development. However, of In addition, within the DSD the Social Development the 10 477 vacant posts just 332 were funded (SASSA, 2019b, Integrated Case Management System (SDICMS) enable p.88); in other words, SASSA is only allowed to fill three percent the tracking of social services for households. DSD of these vacancies. Recent problems associated with the services that are currently covered include: Household and implementation of the COVID-19 grant serve to highlight these Community Profiling, Child Protection Register, Probation capacity constraints, which are largely linked to the current Case Management, Victims Empowerment, and Adoptions weak fiscal situation. and Register of Adoptable Children and Prospective Adoptive Parents. While the face-to-face grant application process at the 3.4.2. Delivery systems of social assistance local offices and the up to 90-day period to determine eligibility seem cumbersome and lengthy, it is necessary to collect all Applications for social grants take place in person at one the information which are needed for the system. The face- of the 360 local SASSA offices (except for the COVID-19 grant to-face meeting with local social workers is also important which has an electronic application process discussed below). for case management and ensuring households get the right Applicants should bring their 13-digit South African national kinds of social service support and counselling for their needs. ID number and supporting documents related to children, The Integrated Community Registration Outreach Programme disability, residency, assets and income, marital status, etc. Based (operating in some wards) enables poor and vulnerable people on the form filled out by the applicant at the local SASSA office it to access government services within their reach. According takes SASSA up to three months to process the application and to the SASSA annual report, the time spent processing social cross-check the documentation with the national systems to grants continues to be narrowed as 98.88 percent (1 618 503 confirm eligibility. The applicants are informed via letter about of 1 636 755) of grants were processed within 10 days, while the status of their application and, if admitted, they are added 84 percent (1 372 781 of 1 636 755) were processed within one to the database and pay lists for the programme for which they day. qualify. Biometric enrolment was tried in 2018 using an external service provider but phased out later that year due to a dispute While SOCPEN is largely a database for grant management, between SASSA and Labour Unions. there is no real functioning social registry in South Africa with the ability to link together all social services for its As noted in the grant descriptions above, all social grants citizens. However, in order to comprehensively address (except the foster child grant) are means tested in different the triple challenge of poverty, unemployment and ways using the national ID number and income or assets as inequality the National Development Plan (Vision 2030) a basis and comparing the applicants’ documentation against proposes the development of the National Integrated the national databases such as the South African Revenue Social Protection Information System (NISPIS). At the Service (SARS). beginning, in 2014, the DSD wanted to measure education outcomes for social grant recipients with the view to locate The information system used to manage the social grants and improve the wellbeing of vulnerable children through is called the Social Grant Payment System (SOCPEN) and education as an essential building block in the progress towards is mainly used for: sustainable development. Hence, efforts were made to link the • processing applications for the old age, disability, war SASSA grants databases, other social service databases of the veterans, child support, foster child and care dependency DSD, NSFAS student bursaries, with the learner databases of the grants; Department of Basic Education (DBE), the Department of Health • generating a pay file monthly for the approximately 17 information system and Home Affairs. The vision is also to add million grants; and links to the databases of Department of Rural Development, • automatically producing a list of beneficiaries due to be CoGTA, Department of Employment and Labour, Department of reassessed. Public Works and Infrastructure, and Department of Transport. NISPIS would also be accessible by provinces and districts. 33 Today the NISPIS project is well underway, although it may have facilitated the identification of food insecure has taken a back seat to other urgent priorities during households, improved the targeting, and enhanced the the COVID-19 period. A steering committee has been put effectiveness of the food parcels distributed during the in place to develop and implement the NISPIS. A thorough COVID-19 crisis. strategy and costed process of linking systems are necessary to move forward. A number of assessments of the existing social Moreover, the COVID-19 crisis also exposed that no good databases in various departments have been undertaken, and central system existed for identifying informal sector a set of recommendations have been made available for how workers who lost their income as a result of the lockdown. datasets can be made interoperable, both functionally and Because needs were urgent and a national lockdown was in technically. The NISPIS has the potential to lead to a better effect it was not possible for SASSA to accept new applications tracking, not just of social grant recipients and beneficiaries, at their local offices, which had been closed down. In record but also those who receive other kinds of social assistance such time, SASSA had to build up a new application and registration as those participating in the EPWP or the recent Employment system to handle the huge caseload of millions of applications Stimulus program. It may be possible to strengthen the support for the special COVID-19 grant using all electronic means. Using that social grant recipients get in accessing employment public announcements, in early May 2020 SASSA opened a fully services such as those provided by the Department of digital process where applicants sent in their applications and Employment and Labour or the Presidential Youth Employment supporting documentation via WhatsApp, SMS, USSD, or online. Intervention Pathway Management Network. At the end of June, social workers were also dispatched to some areas to assist applicants who had difficulty using the electronic The need for the social protection system to quickly methods. By the end of November 2020, over 9 million identify households and individuals affected by complete applications had been received and around 6 million joblessness, loss of income, and food insecurity caused had been approved.6 . To determine eligibility of applicants by the COVID-19 pandemic exposed that there was no SASSA checks the master applications with six databases: good central information system that could identify SARS, the Department of Home Affairs, UIF, SOCPEN, and the people in need. While existing social grants could top up NSFAS. Further, for eligible applicants, banking information benefits, there was no central way of knowing who the newly is checked with the National Treasury. The COVID-19 grant is affected households were. Especially urgent was the need to fully administered outside the SOCPEN system. Examples of increase food distribution and to provide support to informal how Chile and Turkey have developed social registries and sector workers who did not qualify for the UIF. interoperable databases for better managing social protection programmes are described in Box 4. In addition, the scale up of food distribution programs during the COVID-19 lockdown received a lot of criticism as media reported on food packages ending up with the wrong people and extremely long queues of hungry people at distribution points. A number of databases were consulted such as that of the school nutrition programme at the DBE, the indigent registry, SOCPEN, and the malnutrition databases of the Department of Health. However, lists were not compatible. A unified registry or interoperable information systems 6 Many applications were received but about half of them were duplications or incomplete and were therefore not considered for further processing. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 34 Box 4: Interoperability and Integrated Social Protection Information Systems: Chile & Turkey Chile’s Social Registry of Households (RSH) is one example of a highly interoperable system that combines self-reported information from citizens and real-time data exchange with numerous other administrative systems. Chile first pioneered the development of a social registration and eligibility system (Ficha CAS) in the early 1980s, with the Ficha CAS proxy-means testing system serving multiple social programs early on in its inception. The RSH built on that early experience with the Ficha CAS system and was developed in response to the concrete changes and operational needs of the Chile Solidario initiative, which links extreme poor families to numerous benefits and services with active social worker intermediation and outreach. The design of a national system for social protection addressed the lack of communication among information systems managed by different agencies for numerous programs serving the Chile Solidario initiative. The 2004 law creating the Chile Solidario System included a mandate for the creation of a Social Information Registry (RIS), combining both the Household Social Registry (RSH) with an Integrated Beneficiary Registry (RIB) that links numerous program beneficiary registries. The RSH now covers about 75 percent of the Chilean population and serves over 80 programs. Citizen interface is permanent, integrated, and dynamic: citizens can apply for over 80 social programs, update their information, and access their information online or through local offices. Self-reported information includes family composition, housing conditions, education, health, occupation, and income. Data drawn from other administrative systems include information on: taxes, social security contributions, unemployment insurance, pensions, health insurance, education, and property and vehicle ownership, and so forth. Interoperability is facilitated by a unique National ID. The RSH operates within the context of an Integrated System for Social Information (SIIS), with real-time two-way links to an Integrated Beneficiary Registry that permits coordination of both the demand for social programs (via the Social Registry) and the supply of programs (via the Integrated Beneficiary Registry). Turkey’s Integrated Social Assistance System (ISAS) also maintains real-time interoperability with numerous information systems (population registry, social security, education and health, land registry, revenue administration, agriculture, etc.). This capacity was developed to consolidate parallel social registries that were largely paper-based systems and to reduce the amount of time needed to collect appropriate paper documents and complete the processing of applications. With the improved technology, the Integrated Social Assistance Service System (ISAS), Bütünlesik, was developed within the context of a broader digital governance strategy, allowing program administrators to query in real time and online a large number of government databases to verify the status of households applying for social assistance. At present, the system gives online query access to 22 institutions and 28 databases through a web service system and is used by numerous social programs. For all social assistance programs, the initial application involves presentation of the applicants’ National ID numbers and signing of a consent form to allow institutions to review their information. A socio-economic profile is generated in ISAS by linking datasets from various institutions to the citizen’s unique national ID number. The profile is then assessed for completeness of information, inconsistencies, and potential eligibility via data exchange with numerous information systems (population registry, social security, education and health, land registry, revenues administration, agriculture, etc.). Subsequently, a social worker carries out a home visit to collect and verify information of households and their member using a standardised questionnaire (with approximately 50 questions). At present, this home visit questionnaire is still paper based, but there are plans to move to a digital interface. Once information from the home visit is digitalised, the Social Registry is available for use by 17 programs (as of 2017), including various types of income support (such as CCT, old age and disability pension), Universal Health Insurance subsidies, scholarships and other educational supports, and so forth. Source: World Bank (2017). “Social Registries for Social Assistance and Beyond: A Guidance Note & Assessment Tool. There has been a gradual shift within the social assistance boosting financial inclusion, its range of services has so far been system towards paying grants electronically, rather than limited. Postbank only accepts deposits—it does not offer loans. having people come to pay-points to collect their grants in SOCPEN and the digital payment system of the South cash. Figure 3.8 provides estimates of the share of grants paid African grant system make it possible to pay millions of through the banking system between 2004/05 and 2010/11. grants quickly and timely to the right beneficiaries every Over this period, the proportion increased from 10.8 percent month. In September/October 2020 SASSA also piloted a new to 37.5 percent, an increase of almost 250 percent.7 Today, cash send/mobile money option intended to make it easier for the vast majority of payments are made electronically unbanked recipients to obtain their grants but only to 100 000 via SASSA debit cards (Mastercards) which can be used recipients (see more below). Hence, to date very few social at any ATM and major retailers, or to the applicants’ own grant payments are made using mobile technology and bank accounts. Payments are made timely on a monthly bank payments dominate. basis during the first days of each month. A small number of beneficiaries, mainly in rural and remote areas still retrieve their Another main challenge with the social grant payment payments in person/over the counter at a physical pay-point, system is the last mile, namely the withdrawal and commonly the local Post Bank, which does not offer online or use of funds by the beneficiaries. While the first part of mobile banking. While the Post Bank has an important role in the payment delivery chain – from the government to the 7 More recent data are not available to complete the figure. beneficiaries’ accounts/debit cards – is highly digitised, the 35 last mile distribution related to how the beneficiary accesses, 1.7 were made through direct deposits to bank accounts, withdraws, and uses the funds is still a challenge in South Africa. and approximately 100 000 payments were made through First, retailers are often overwhelmed by volumes of grant the new cash send/mobile money channel. For the first time, recipients who withdraw money on paydays even though the mobile payments which can be cashed out via ATMs without SASSA card can be used as a debit card. This (even pre-COVID) a debit card were tried, although the number of people leads to long lines of people waiting at retailers on payday at the reached through that channel was much lower than expected. beginning of each month. Second, ATM machines and retailers According to SASSA, this was due to a requirement that there who accept debit cards are scarce in townships and rural zones has to be a positive link between the applicant and the mobile where many beneficiaries live. Recipients are not able to use number into which the grant would be paid. As the majority local spaza shops (small informal convenience shop) or other of applicants did not have phones registered in their name local convenience stores to withdraw cash or to make purchases the direct link could not be established. In June 2020 SASSA using the SASSA card. Instead, many recipients, especially in also reported that some COVID-19 grant payment challenges rural areas, must travel far incurring transport expenses and were encountered as the system could not pay out multiple dedicating significant time to reach urban centres to be able to payments to the same bank account or mobile number. It is access and spend the grant money. Due to these costs, many evident that challenges still remain in how social assistance beneficiaries tend to withdraw the full amount of their funds payments can be effectively delivered to people outside with up-front. limited financial inclusion, and how the social grant system can also be leveraged to enhance the financial inclusion of the Engaging the extensive network of spaza shops population. As the Presidential Employment Stimulus rolls out and other informal vendors in the digital payments wage payments to around 800 000 temporary workers in early ecosystem is an enormous opportunity to overcome the 2021 opportunities for innovation may be possible. last mile challenges while supporting commerce in poor neighbourhoods. There is a gap in SASSA’s vendor model and SASSA and DSD have invested significantly in the new in South Africa’s ecosystem for digital payments more broadly. electronic application and management system for Spaza shops and informal vendors in townships and rural zones the special COVID-19 grant. Although the COVID-19 grant are largely outside of the digital payments eco-system. Most is temporary and may only be active until January 2021, spaza shops and other informal vendors do not have electronic the investments made in the system will likely continue to card readers and cannot accept debit cards, only accepting benefit existing grant programs for processing applications cash. The relatively high cost of a point-of-sale system and the electronically. Especially, it is expected that the regular Social fees associated (approx. 3.5% of transacted value) with these Relief of Distress Grant will be able to benefit for this investment payments make cash more attractive for spaza shop owners. to help process applications electronically and making more Most spaza shops are informal, some are owned by migrant payments in cash instead of in-kind. workers, and their regulatory environment is complicated. In short, improving the last mile accessibility would reduce the 3.4.3. Administration of social assistance travelling and queueing for grants, reduce transport costs, enhance social distancing, and stimulate the local economy if There is comparatively little data available about the grant payments were made available closer to where recipient efficiency with which the South African government has households live and in places where they normally shop and been able to administer the social assistance system. trade. Importantly, engaging spaza shops in grant payments Currently, two relevant measures are published as part of the would however require that service quality assurance measures national budget documentation, namely the average cost are in place – especially for older persons. of administering social assistance per month (administrative cost per beneficiary per month), and the administration costs New mobile payments have been piloted but providing as a proportion of the social assistance transfers budget. A payments outside the formal banking system remains third measure—the share of beneficiaries receiving payments challenging. The total number of around 6 million people through the banking system—an indirect measure of paid though the COVID-19 grant is right in the middle of the administrative efficiency, was published between 2004/05 and 5-8 million that SASSA expected in the first few months of the 2010/11. The evolution of these three measures is presented grant. Around 4.07 million payments were made though the in Figure 3.8, with the current MTEF projections indicated by Post Bank (which was meant as the default option), another dotted lines. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 36 Figure 3.7. Efficiency of Social Assistance Administration since 2004/05 60.0 57.1 56.8 57.4 57.3 55.1 55.0 49.0 50.0 45.0 44.5 43.6 45.3 45.0 41.6 40.8 40.7 37.5 37.8 40.0 36.7 34.0 Percent / Rands 32.9 37 37 36 36 35 35.0 34 33 33 34 31 32 31 30 30 30.0 29 29 27 25 25.0 16.4 19 14.4 15.0 12.0 14.4 10.8 Admin. cost/beneficiary/mo. (R) 10.0 Admin. cost/beneficiary/mo. (R, real) 7.8 5.0 6.6 6.7 7.3 6.9 5.9 5.5 5.6 5.7 Admin. cost share (%) 4.6 5.2 5.2 5.4 4.8 4.4 4.4 4.0 4.2 3.9 Share paid through banks (%) 0.0 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18* 2018/19* 2019/20* 2020/21* 2021/22* 2022/23* Source: National Treasury (2008, 2009, 2010b, 2011, 2012b, 2013, 2014b, 2015b, 2016b, 2017b, 2018b, 2019b, 2020b). Note: * Data for 2017/18 and 2018/19 are preliminary outcomes; data for 2019/20 are revised estimates; and data for 2020/21 onwards are medium-term estimates. Real administration cost figures are deflated to March 2020 prices using average headline CPI for April to March of each year. In the 2019/20 financial year, the latest for which there is to administration. This indicator peaked at 7.8 percent in actual data, it is estimated that it cost an average of R36 2008/09 but has subsequently declined by just over two-fifths per month to pay each grant. While this is almost double the to 4.0 percent in 2018/19 and 4.4 percent in 2019/20. R19 estimated for 2004/05, in real terms the cost has declined significantly. Between 2005/06 and 2009/10, administration Between 2004/05 and 2010/11, the proportion of grants costs per beneficiary per month were around R57 in March paid through banks more than tripled from 10.8 percent 2020 prices; however, this fell to R49 in 2010/11 and generally to 37.5 percent. According to more recently published data continued falling thereafter, reaching a low of R32.90 in 2018/19 (SASSA 2020d), virtually all beneficiaries in July 2020 were paid before rising again to R36.70 in the following year. This means through either the South African Post Office (SAPO)/Postbank that in 2018/19, the latest year for which there is data on (8.3 million, or 72.7 percent), ACB/Banks (2.2 million, or 19.0 numbers of and expenditure on grants, the average cost of percent), or Grinrod Bank (944 000, or 8.3 percent). administering social assistance represented 4.1 percent of the value of the average grant. This is down 2.6 percentage points (or around two-fifths) from 6.7 percent in 2006/07. This decline in the average cost of administering social assistance is further reflected in the declining proportion of the social assistance transfers budget that is allocated 37 4. Social Assistance Programme Performance In this chapter we assess the performance of the social grant 4.1. Coverage and Adequacy system in terms of providing adequate support to the poorest, 4.1.1. Coverage and preventing and addressing vulnerability and inequality. The chapter also presents the impacts and outcomes of social As noted previously, official data on the number of grants assistance programs in South Africa on a number of economic paid out by SASSA suggest a coverage rate of social and social outcomes and reviews the spending efficiency and assistance of 30.7 percent of the total population. This value for money of the current social assistance system. It also is relatively close to the estimate of coverage from the Living puts South Africa’s performance in perspective against global Conditions Survey 2014/15 data in terms of direct beneficiaries.8 evidence and other UMICs. Finally, the chapter discusses the As Table 4.1 indicates, the coverage rate for all social assistance Government’s social protection response to the COVID-19 programmes is estimated at 33.1 percent in 2014/15. This slightly pandemic and its impacts. higher figure aligns with the relatively high estimate from this survey of the number of children receiving child support grants. Table 4.1. Social Assistance Coverage Rates (%), Direct Beneficiaries Only, 2014/15 Pre-Transfer Distribution Post-Transfer Distribution Total Q1 Q2 Q3 Q4 Q5 Total Q1 Q2 Q3 Q4 Q5 All social assistance 33.1 56.1 45.9 36.0 21.4 5.8 33.1 46.8 47.2 40.3 24.4 6.6 Older persons 5.8 9.1 7.1 4.8 4.9 3.1 5.8 3.2 7.5 7.9 6.8 3.6 Disability 2.2 4.0 3.0 2.2 1.5 0.5 2.2 1.3 3.3 3.8 2.2 0.6 Child support 24.1 41.0 34.6 28.2 14.6 2.1 24.1 41.3 34.9 27.3 14.9 2.3 Care dependency 0.2 0.4 0.4 0.2 0.1 0.0 0.2 0.1 0.4 0.4 0.1 0.0 Foster child 0.7 1.6 0.9 0.5 0.4 0.1 0.7 0.9 1.1 0.9 0.4 0.1 Grant-in-aid 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Other, e.g. social relief 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Age-specific coverage rates Child support (<18 yrs) 67.2 86.9 81.1 76.3 49.7 9.2 67.2 85.6 81.4 75.2 51.0 9.9 Older persons (60 yrs+) 71.8 96.6 94.6 85.5 70.6 28.3 71.8 89.0 94.1 91.0 77.8 31.7 Source: Own calculations, Statistics South Africa (2015a). Note: (1) Figures for the war veterans grant are not included as no respondents in the Living Conditions Survey 2014/15 report receiving the grant. (2) The two age-specific coverage rates refer to the proportions of the age-eligible population covered by the child support or older persons grants. The overall coverage rate is driven primarily by the child those within quintile 5 households. Here too, this difference is support grant: it is estimated that nearly one-quarter of largely the result of difference in coverage of the child support the population were direct beneficiaries of the grant in grant, which ranges from 41.3 percent in quintile 1 to just 2.3 2014/15. This is followed by the older persons grant (5.8 percent in quintile 5. The difference is even wider in terms of the percent) and the disability grant (2.2 percent). Between age-specific coverage rate: amongst children under 18 years, them, the remaining grants account for just 0.9 percentage 85.6 percent of quintile 1 children were covered compared points of the overall coverage rate. Given that each of the grants to 9.9 percent of quintile 5 children. In contrast, inter-quintile has its own age-eligibility criteria, Table 4.1 also provides age- differences in coverage rates for the older persons’ grant are specific coverage rates for the child support and older persons small: just 4.3 percentage points between the highest and grants. Coverage rates were high for both grants: 67.2 lowest quintile-specific coverage rates. percent of children under the age of 18 years received child support grants, while 71.8 percent of adults aged For the older persons’ grant, coverage at the population level 60 years and above received the older persons grant. was highest in quintile 3 (7.9 percent), while age-specific coverage was highest in quintile 2 (94.1 percent). This is linked Coverage is the highest among the poorest households.9 to the reordering of the pre-transfer income distribution due to Considering the post-transfer distribution, almost half (46.8 the magnitude of the older persons grant. Thus, for example, percent) of individuals in quintile 1 households were direct coverage rates for each of the grants was highest in quintile 1 of beneficiaries of social assistance, compared to 6.6 percent of the pre-transfer distribution. It is estimated that 56.1 percent of 8 See Appendix A for detail regarding the identification of direct grant beneficiaries in the Living Conditions Survey 2014/15 9 Throughout this report, where figures are reported across the pre- and post-transfer distributions, these are the distributions of per capita household income. The post-transfer distribution is the distribution observed in the data, while the pre-transfer distribution is a hypothetical distribution that is created by removing income from social assistance from respondents’ reported incomes. It is important to note, however, that this pre-transfer distribution is unable to account for changes in the structure of households or the patterns of household formation that would occur in the absence of social assistance transfers. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 38 the pre-transfer quintile 1 population were direct beneficiaries Given that household members also benefit indirectly from an of social assistance, with coverage falling to 45.9 percent in individual receiving a social grant, these indirect beneficiaries quintile 2 and 36.0 percent in quintile 3. Age-specific coverage can be considered to be covered. The inclusion of indirect rates indicate that the older persons’ grant was virtually universal beneficiaries within the measure nearly doubles the coverage amongst adults aged at least 60 years in the poorest 40 percent rate for all social assistance. Thus, in 2014/15, close to two- of the population, while the child support grant reached at thirds (64.0 percent) of the South African population least four out of five children within this cohort. either received a social grant or were co-resident with someone who received a grant (Figure 4.1). Figure 4.1. Coverage of Direct and Indirect Social Assistance Beneficiaries across Quintiles 100.0 95.2 90.0 86.1 85.0 85.8 80.0 77.5 74.1 70.0 65.9 64.0 64.0 59.7 60.0 54.0 Percent 52.2 54.1 50.2 50.5 50.0 48.2 45.0 42.9 39.8 40.0 32.8 34.0 34.4 30.0 19.4 South Africa (pre-transfer) 20.0 17.0 17.8 16.5 South Africa (post-transfer) 16.0 14.0 15.2 11.6 World 10.0 Upper Middle Income Sub-Saharan Africa 0.0 Overall Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Source: World Bank (2020a) and own calculations, Statistics South Africa (2015a). Notes: Data for regional averages are for the 2008-2016 period; data for South Africa are for 2014/15. The coverage of South Africa social assistance grants is from 65.9 percent in quintile 1 to 34.4 percent in quintile 5. four times the level of coverage in Sub-Saharan Africa Globally, coverage rates are highest in quintile 2, while in Sub- (16.0 percent) and around one-third higher than the Saharan Africa they are correlated with level of welfare, rising global average (48.2 percent across all quintiles); it is from 11.6 percent in quintile 1 to 19.4 percent in quintile 5. Few also substantially higher than the average coverage rate upper middle income countries can compare to South Africa amongst upper middle income countries (45.0 percent in terms of coverage rates amongst the poorest quintile of the across all quintiles). Based on estimates from the ASPIRE population. In terms of the pre-transfer distribution, the only database (World Bank 2020a) (see Table B.3 in the appendix), countries with coverage rates of over 90 percent are Belarus South Africa compares favourably with upper middle income (91.8 percent), Botswana (94.9 percent), Georgia (92.9 percent), countries such as Argentina (coverage of 19.8 percent), Brazil and Malaysia (94.2 percent), making South Africa the top (23.7 percent), China (43.8 percent), Mexico (32.5 percent), and performer on this metric. By regional and global standards, Turkey (18.0 percent). However, coverage rates are even higher therefore, not only is overall coverage of social assistance in Botswana (73.8 percent), Malaysia (82.8 percent), and the in South Africa high, but it is also so strongly targeted at Russian Federation (67.9 percent). the poorest 60 percent of the population that coverage in quintile 5 in South Africa is less than half the average Disaggregating the population by income quintile for upper middle income countries. confirms the strong progressivity in coverage rates in South Africa, with coverage of social assistance programmes Given potential demographic differences across quintiles falling from 95.2 percent in quintile 1 of the pre-transfer and the significant differences in the values of the various distribution to 15.2 percent in quintile 5. Thus, the quintile 1 grants, it is useful to analyse the extent to which the coverage rate is more than six times that in quintile 5. A similar mix of grants received at different points of the income pattern is observed for South Africa’s post-transfer distribution. distribution may differ. Table 4.2 presents a breakdown While the average upper middle income coverage rates also of grants received by the population within each of the five decline as welfare increases, the decline is more gradual, falling quintiles. According to the survey estimates, nearly three- 39 quarters (73.0 percent) of all grants received in 2014/15 were total. Together, these three grants accounted for more than 97 child support grants. A further 17.7 percent were older persons’ percent of all grants received by respondents. grants, while disability grants accounted for 6.8 percent of the Table 4.2. Composition of Grants Received by Households across the Pre-Transfer Distribution, 2014/15 Overall Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Number of grants (‘000s) Any social assistance 18 102 6 145 5 026 3 947 2 344 640 Older persons 3 178 1 002 781 528 532 336 Disability 1 232 436 330 245 163 57 Child support 13 206 4 496 3 784 3 093 1 598 235 Care dependency 121 43 39 25 11 4 Foster child 373 171 96 58 41 8 Grant-in-aid 8 3 1 1 1 2 War veterans 0 0 0 0 0 0 Other (e.g. social relief ) 5 2 1 1 2 0 Proportion (%) Any social assistance 100.0 100.0 100.0 100.0 100.0 100.0 Older persons 17.6 16.3 15.5 13.4 22.7 52.6 Disability 6.8 7.1 6.6 6.2 7.0 8.9 Child support 73.0 73.2 75.3 78.4 68.2 36.7 Care dependency 0.7 0.7 0.8 0.6 0.5 0.6 Foster child 2.1 2.8 1.9 1.5 1.7 1.3 Grant-in-aid 0.0 0.0 0.0 0.0 0.1 0.3 War veterans 0.0 0.0 0.0 0.0 0.0 0.0 Other (e.g. social relief ) 0.0 0.0 0.0 0.0 0.1 0.0 Source: Own calculations, Statistics South Africa (2015a). Note: Figures for the war veterans grant are not included as no respondents in the Living Conditions Survey 2014/15 report receiving the grant Amongst the poorest three pre-transfer quintiles, child income countries and relative to other countries in support grants account for around three-quarters of all the region. Importantly, this high rate of coverage is grants received. Even in quintile 4, nearly seven out of ten (68.2 combined with a strong focus on poorer individuals and percent) of grants are child support grants. However, in quintile households, such that coverage amongst the quintile 1 5 this proportion falls to just over one-third (36.7 percent). population in South Africa is around five times that of Instead, amongst the richest 20 percent of the population, the the quintile 5 population. This helps to ensure that a large older persons grant dominates, accounting for 52.6 percent of proportion of the benefits of the system accrues to the poor all grants received by this group. This is more than three times and is a crucial requirement for a social assistance system to the proportion observed in the poorest three quintiles (13- be able to impact efficiently on poverty and inequality. At the 17 percent) and more than twice the proportion in quintile 4 same time, even in terms of the pre-transfer distribution, (22.7 percent). The data also illustrates that the disability grant high-value grants such as the older persons’ grant and is relatively more common amongst quintile 1 and quintile the disability grant make up larger proportions of all 5 households, while the foster child grant is relatively more grants received by richer quintiles. Conversely, low-value common amongst quintile 1 households. These differences grants such as the child support grant dominate the mix are the result of a number of factors related to household of grants within the poorer quintiles. structures and differences in means tests, amongst others, and have implications for some of the patterns that will be observed 4.1.2. Adequacy/Benefit Levels below (for example, benefit incidence). In order to begin to understand the impact of social assistance Therefore, overall social assistance coverage rates in transfers, one must know how much households are receiving. South Africa are high relative to other upper middle Figure 4.2 presents the average transfer value per capita per SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 40 annum for beneficiary households only. Both direct and indirect equivalent to an average of R15 798 per household per beneficiaries are included in the calculation. On average, annum. To put these figures in perspective, the upper bound beneficiary households received social assistance poverty line in April 2015 prices was R11 904 per capita per transfers of R3 279 per capita in April 2015 prices. This is annum (R992 per capita per month). Figure 4.2. Average Transfer Value Per Capita, Beneficiary Households Only, 2014/15 5000 4455 4500 4169 4000 3743 3665 3697 Rands (April 2015 prices) 3472 3500 3279 3286 3279 2975 3000 2800 2500 2169 2000 1500 1000 500 0 Total Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Total Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Pre-Transfer Distribution Post-Transfer Distribution Source: Own calculations, Statistics South Africa (2015a). Notes: Average per capita transfer value is calculated excluding households that do not report receiving any grants. What is clear from the pre-transfer distribution is that per or consumption. Thus, the adequacy of benefits in quintile 1 is capita transfers are largest for the top quintile (R4 169 calculated as the amount of transfers received by beneficiaries per capita), followed by the poorest quintile (R3 665). within quintile 1 divided by the total income or consumption The average transfer per capita in the middle three quintiles of beneficiaries in quintile 1. The adequacy of benefits across ranges between R2 800 and R3 300. At least two factors income quintiles is presented in Figure 4.3. For South Africa, contribute to this pattern. First, as already mentioned, better-off the measure of welfare is per capita household expenditure. households tend to have fewer members, resulting in higher Overall, social assistance transfers in South Africa are per capita transfers. The second relates to the pattern of grant equivalent to 26.0 percent of beneficiaries’ expenditure. types received by the beneficiaries across the five quintiles. A This is a higher proportion than in Sub-Saharan Africa breakdown of the composition of grants across the quintiles (19.4 percent) and is almost five times the proportion in of the pre-transfer distribution was presented in Table 4.2. upper middle income countries (5.6 percent). Thus, social There, it was shown that the older persons’ grant was more assistance is relatively generous in South Africa when common within quintile 5 households than was the case in any compared to beneficiaries’ expenditure. Only one upper other quintile: 52.6 percent of grants received by the quintile middle income country has a significantly higher adequacy 5 population were older persons’ grants, compared with 17.6 rate than South Africa: Belarus, where social assistance benefits percent for the population as a whole. Overall, high value represent 42.2 percent of beneficiaries’ expenditure (see Table grants such as the older persons’ and disability grants represent B.3). Adequacy rates comparable to South Africa’s are observed a higher proportion of total grants received by quintile 5 than in Georgia (29.2 percent), the Maldives (24.8 percent), and they do in other quintiles, thereby contributing to higher Mauritius (28.8 percent). In contrast, South Africa performs average transfer values per capita for quintile 5. particularly well compared to China (2.3 percent), Colombia (5.1 percent), Malaysia (1.7 percent), Peru (6.8 percent), the Russian The absolute value of the transfers made by government Federation (6.8 percent), and Turkey (6.5 percent). are a first step in assessing their importance in supporting consumption amongst the poor in particular. The value of social assistance can be related directly to individuals’ welfare through the measure referred to as the adequacy, or generosity, of benefits. The adequacy of benefits is defined as the total transfer received by beneficiaries relative to their total welfare, with welfare an appropriate money-metric measure such as income 41 Figure 4.3. Social Assistance Benefits as a Share of Total Expenditure (Adequacy of Social Assistance Benefits) across Quintiles 110.0 99.4 100.0 90.0 80.0 76.3 70.0 66.4 Percent 60.0 49.4 50.0 46.1 43.9 42.8 40.7 40.0 33.6 34.7 32.5 30.0 26.0 25.1 24.8 26.0 19.4 19.3 19.4 18.4 17.4 20.0 14.3 South Africa 13.0 11.8 7.1 6.7 7.9 7.1 7.4 Upper Middle Income 10.0 5.6 5.6 5.1 3.4 5.6 5.4 4.9 4.9 Sub-Saharan Africa 0.0 Overall Q1 Q2 Q3 Q4 Q5 Overall Q1 Q2 Q3 Q4 Q5 Pre-Transfer Distribution Post-Transfer Distribution Source: World Bank (2020a) and own calculations, Statistics South Africa (2015a). Notes: (1) Data for regional averages are for the 2008-2016 period; data for South Africa are for 2014/15. (2) For South Africa, the welfare measure is per capita household expenditure as recorded in the survey. (3) The ASPIRE database does not include estimates for the World. The pattern of adequacy of benefits across the income percent in quintile 3 of the pre-transfer distribution. By quintile distribution in South Africa is broadly similar to the patterns 4, the adequacy rate in South Africa drops below that of Sub- observed in both upper middle income and Sub-Saharan Saharan Africa and, by quintile 5, the three adequacy rates are African countries: adequacy is highest for the poorest quintile similar. and falls consistently to the richest quintile. Sub-Saharan Africa performs particularly well in terms of adequacy of benefits Overall, the South African government made social for the poorest quintile: social assistance transfers account for assistance transfers equivalent to 7.3 percent of 99.4 percent of consumption in quintile 1 of the pre-transfer individuals’ total expenditure in 2014/15 (Table 4.3). This is distribution and 76.3 percent in the post-transfer distribution. referred to as the relative incidence of social assistance, which In South Africa, social assistance accounts for two- is defined as the proportion of social assistance transfers within thirds of expenditure for beneficiaries in quintile 1 of total expenditure. Relative incidence is very similar to adequacy the pre-transfer distribution. In contrast, amongst upper of benefits, with the key difference being that where adequacy middle income countries, this proportion is only 17.4 percent. of benefits considers only the income or expenditure of direct Adequacy in the lowest quintile of the pre-transfer distribution and indirect beneficiaries, relative incidence covers the entire is estimated at 34.0 percent in Brazil, 7.5 percent in China, and population. The two largest contributors to this figure were the 15.1 percent in Thailand, but is as high as 124.1 percent in older persons’ grant and the child support grant; transfers in Belarus, 83.0 percent in Georgia, and 63.9 percent in Mauritius. terms of these two grants accounted for 3.2 percent and 2.6 Adequacy rates are particularly high in South Africa’s percent of total expenditure respectively. The disability grant rural areas—the overall adequacy rate is 41.4 percent in accounted for a further 1.2 percent of expenditure, with the rural areas (compared to 19.5 percent in urban areas), and remaining grants accounting for around 0.3 percent. rates are highest in the Northern Cape (36.8 percent), Eastern Cape (35.7 percent), and Limpopo (35.7 percent) provinces— The figure of 7.3 percent obscures wide variation in relative indicating a severe lack of alternative income sources, such as incidence across the income distribution. Within the wage income or even subsistence agriculture, in these areas. pre-transfer distribution, social assistance transfers Estimates of adequacy by geographic location can be found in accounted for 60.7 percent of quintile 1 expenditure, Figure B.2 in the Appendix. 31.9 percent of quintile 2 expenditure, and 16.6 percent of quintile 3 expenditure. The older persons’ grant accounts While adequacy rates drop off quite quickly of higher for 25.5 percentage points of the total for quintile 1, and this quintiles, social assistance transfers are still equivalent to 40.7 contribution roughly halves from one quintile to the next, percent of beneficiaries’ expenditure in quintile 2 and 25.1 falling to 13.0 percent in quintile 2, 6.1 percent in quintile 3, 3.2 percent in quintile 4, and just 0.5 percent in quintile 5. A similar SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 42 pattern is observed for the disability grant. The child support lower at 21.6 percentage points, falling to 12.3 percentage grant’s contribution to quintile 1 relative incidence is slightly points in quintile 2, and just 0.1 percentage points in quintile 5. Table 4.3. Relative Incidence, (%), 2014/15 Overall Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Pre-Transfer Distribution All social assistance 7.3 60.7 31.9 16.6 6.4 0.7 Older persons 3.2 25.5 13.0 6.1 3.2 0.5 Disability 1.2 10.4 5.3 2.6 0.9 0.1 Child support 2.6 21.6 12.3 7.2 2.0 0.1 Care dependency 0.1 0.9 0.5 0.3 0.1 0.0 Foster child 0.2 2.3 0.8 0.4 0.1 0.0 Grant-in-aid 0.0 0.0 0.0 0.0 0.0 0.0 Other (e.g. social relief ) 0.0 0.0 0.0 0.0 0.0 0.0 Post-Transfer Distribution All social assistance 7.3 31.8 34.6 23.1 8.4 0.8 Older persons 3.2 8.4 14.0 10.1 4.5 0.6 Disability 1.2 3.0 5.8 4.6 1.4 0.1 Child support 2.6 19.2 12.9 7.3 2.2 0.1 Care dependency 0.1 0.2 0.7 0.4 0.1 0.0 Foster child 0.2 0.9 1.2 0.7 0.1 0.0 Grant-in-aid 0.0 0.0 0.0 0.0 0.0 0.0 Other (e.g. social relief ) 0.0 0.0 0.0 0.0 0.0 0.0 Source: Own calculations, Statistics South Africa (2015a). Note: Figures for the war veterans grant are not included as no respondents in the Living Conditions Survey 2014/15 report receiving the grant. These patterns are quite different in the post-transfer 4.2. Inclusiveness distribution. Relative incidence is around one-third in quintiles 1 and 2, falling to 23.1 percent in quintile 4.2.1. Targeting 3, 8.4 percent in quintile 4, and 0.8 percent in quintile 5. However, in quintile 1 of the post-transfer distribution the Most social assistance beneficiaries can be found in child support grant is the dominant contributor to the relative the lower income groups. Higher coverage rates amongst incidence estimate at 19.2 percentage points. This is more the poorest segments of South Africa’s population relative than twice the contribution of the older persons’ grant (8.4 to better-off groups translate into high proportions of social percentage points) and more than six times the contribution of assistance beneficiaries at the lower end of the income the disability grant (3.0 percentage points). However, by quintile distribution. Figure 4.4 presents the distribution of direct and 2 the older persons grant has overtaken the child support grant indirect beneficiaries across quintiles using both the pre- and (14.0 percentage points compared to 12.9 percentage points) post-transfer distributions for South Africa. In 2014/15, the and remains the largest contributor in each of the higher poorest three quintiles accounted for the lion’s share of direct quintiles. The disability grant is a particularly large contributor and indirect beneficiaries: in the pre-transfer distribution, these in quintiles 2 and 3 (5.8 percentage points and 4.6 percentage quintiles accounted for 79.5 percent of all beneficiaries, with points respectively). 29.8 percent in quintile 1 alone. In contrast, just 4.7 percent of beneficiaries were resident in quintile 5 households. A similar pattern is observed in the post-transfer distribution, although beneficiaries are slightly less concentrated in the lower quintiles. 43 Figure 4.4. Distribution of Social Assistance Beneficiaries Across Quintiles 29.8 30.0 29.3 26.9 26.8 26.5 24.8 25.0 24.2 24.4 23.2 23.2 22.5 22.2 21.3 20.8 20.0 17.7 17.8 17.5 16.9 Percent 15.8 15.3 14.6 14.6 15.0 14.1 10.0 South Africa (pre-transfer) 5.1 4.7 South Africa (post-transfer) 5.0 World Upper Middle Income Sub-Saharan Africa 0.0 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Source: World Bank (2020a), and own calculations, Statistics South Africa (2015a). Notes: (1) Data for regional averages are for the 2008-2016 period; data for South Africa are for 2014/15. (2) Beneficiaries include direct and indirect beneficiaries. South Africa performs well in the international comparison. At the programme level, there is some variation in the On average, amongst all the countries in the ASPIRE database, distribution of direct and indirect beneficiaries and, 68.1 percent of beneficiaries are found in the poorest three therefore, the proportion of beneficiaries within the quintiles; the corresponding proportion amongst upper poorest three quintiles. Figure 4.5 presents estimates of the middle income countries is 70.1 percent. However, amongst distribution of direct and indirect grant beneficiaries across Sub-Saharan African countries just 53.4 percent of social quintiles for each of the grants in 2014/15. The poorest three assistance beneficiaries are found in the bottom three quintiles. quintiles of the pre-transfer distribution accounted for nine Since the poorest three quintiles represent 60 percent of the out of ten care dependency grants (89.6 percent) and foster population, this means that the poor are under-represented child grants (89.0 percent). For the disability grant, the child amongst social assistance beneficiaries in the region. South support grant, and other social assistance (which includes Africa’s relatively high concentration of beneficiaries social relief of distress), this proportion ranged between 83 within the bottom three quintiles of the distribution percent and 88 percent. For the older persons’ grant and grant- puts it on par with the average for Latin America and the in-aid, the poorest 60 percent of the population accounted for Caribbean (80.5 percent), suggesting that the country’s almost eight out of ten grants (77.7 percent and 79.0 percent performance on this measure is not particularly unusual respectively). amongst highly unequal middle income countries. Indeed, a comparison amongst upper middle income countries reveals a number of countries that have substantially higher proportions of beneficiaries amongst the poorest 60 percent of the population, including Argentina (91.6 percent), Brazil (94.0 percent), Montenegro (93.6 percent), and Turkey (90.8 percent) (see Table B.3 in the appendix). At the other extreme, only 64.0 percent of beneficiaries in Malaysia are in the poorest 60 percent of the population, as are 66.7 percent in Romania, and 68.1 percent in the Russian Federation. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 44 Figure 4.5. Distribution of Social Grant Beneficiaries Across Quintiles, 2014/15 Any grant Older persons Disability Child Support 50 37.2 40 35.6 29.8 31.1 27.0 27.6 30 26.5 23.2 25.0 24.5 20.5 20 15.8 17.2 14.8 14.4 11.7 7.5 10 4.7 3.6 2.4 0 Percent Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Care dependency Foster child Grant-in-aid Other 50.8 52.8 50 45.7 40 31.1 31.4 30 27.1 24.4 18.9 17.6 18.5 20 17.0 16.4 10.6 12.3 7.9 8.9 10 4.0 2.5 2.1 0.0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Source: Own calculations, Statistics South Africa (2015a). Notes: (1) Beneficiaries include direct and indirect beneficiaries. (2) The LCS 2014/15 data includes no beneficiaries of the war veterans’ grant. 4.2.2. Benefit Incidence Therefore, Figure 4.6 presents the distribution of social assistance benefits in South Africa across the pre- and post-transfer The South African social assistance system performs income distributions and compares them with the distributions well relative to other countries in ensuring that a large for upper-middle income and Sub-Saharan African countries. proportion of social assistance beneficiaries come The ASPIRE database does not include average estimates for from the poorest segments of the population (as was the world. The estimates for South Africa show that close seen in Section 4.2.1). Figure 4.6 takes this a step further and to one-third of social assistance benefits accrue to the investigates the extent to which actual financial benefits poorest 20 percent of the population, and a further 26.4 accrue to beneficiaries across the income distribution. It is percent accrue to those in the second-poorest quintile. quite possible, for example, with a system that includes grants Therefore, the poorest 40 percent of the population account for of different monetary values, that poor beneficiaries are well- three-fifths (59.5 percent) of total benefits, and the poorest 60 targeted while the bulk of the benefits accrue to the rich. The percent account for four-fifths (79.3 percent). Just 6.2 percent opposite may also be true. of social assistance benefits flow to the richest quintile. Figure 4.6. Distribution of Social Assistance Benefits Across Quintiles 50.0 45.8 45.0 42.8 41.1 40.0 35.0 33.1 28.9 30.0 27.6 Percent 26.1 26.4 25.0 21.6 19.7 19.3 20.0 17.7 16.6 17.0 15.7 15.3 14.4 13.9 14.5 14.0 15.0 12.5 0.3 10.9 9.9 9.3 10.0 7.2 6.2 6.0 South Africa 5.0 Upper Middle Income Sub-Saharan Africa 0.0 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Pre-Transfer Distribution Post-Transfer Distribution Source: World Bank (2020a), and own calculations, Statistics South Africa (2015a). Notes: (1) Data for regional averages are for the 2008-2016 period; data for South Africa are for 2014/15. (2) The ASPIRE database does not include estimates for the World. 45 The figure clearly illustrates that South Africa is assistance to significantly reorder distributions. If one takes the something of an outlier amongst upper-middle income view that social assistance benefits that flow to the top quintile countries and amongst those in Sub-Saharan Africa. Both are essentially leakages, then it is clear that leakages in South groups of countries have high proportions of benefits Africa are low by international standards. Nevertheless, various flowing to the poorest quintile and to the richest quintile upper-middle income countries—all in Latin America and the in the pre-transfer distribution, creating a U-shaped Caribbean—see even smaller shares of benefits accruing to the pattern that is particularly deep for the latter. Thus, while richest quintile, including Argentina (2.6 percent), Brazil (2.0 Sub-Saharan Africa does well in ensuring that 42.8 percent of percent), and Peru (0.5 percent). benefits accrue to the poorest 20 percent of the population, more than one-quarter (27.6 percent) flow to the wealthiest Importantly, the proportion of benefits going to the quintile. This means that each of the middle three quintiles poorest three quintiles in South Africa is very similar to receive around 10 percent of total benefits each. Amongst the proportion of beneficiaries within those quintiles upper-middle income countries, the largest proportion of (79.5 percent, as highlighted in section 4.2.1). A benefits flow to the richest quintile. Targeting in terms of the comparison of the pre-transfer distribution in Figure 4.4 and proportion of benefits flowing to the poorest three deciles in Figure 4.6 reveals that is only quintile 1 and quintile 5 where the the pre-transfer distribution is particularly good in Argentina share of benefits exceeds the share of beneficiaries. For quintile (91.1 percent of benefits), Belarus (92.3 percent), Brazil (92.6 1, this is the type of pattern that one would hope to see as it percent), Costa Rica (90.0 percent), Kosovo (91.1 percent), implies relatively high per capita transfers amongst the poorest Montenegro (90.1 percent), and Peru (96.6 percent). In contrast, households. For quintile 5, this is likely related to the greater just 66.3 percent of benefits accrue to the poorest 60 percent importance of higher value grants such as the older persons’ of the population in Botswana, with a slightly lower proportion grant for these households. observed in Malaysia (60.9 percent). Detailed figures for upper- middle income countries are presented in Table B.3 in the Table 4.4 presents details on the distributions of appendix. beneficiaries and benefits across the five pre-transfer quintiles for all social assistance, as well as for the three The difference between South Africa and the two major grants, namely the older persons, child support, country groupings is even starker in the post-transfer and disability grants. Overall, the data shows that the three distribution, with nearly half of all benefits accruing to middle quintiles receive a smaller share of social assistance the top quintile (45.8 percent in Sub-Saharan Africa and benefits than their shares of total beneficiaries. Quintile 1’s 41.1 percent amongst upper-middle income countries). share of benefits is around 11 percent higher than its share While the pattern for upper-middle income countries is more of beneficiaries; for quintile 5, this rises to 32 percent. These of a J-curve, that for Sub-Saharan Africa is monotonically differences translate into lower per capita transfers for the increasing with income. Along with the inverted U-shaped middle three quintiles relative to quintile 1, which in turn has pattern for South Africa, the figure reveals the ability of social lower per capita transfers than those received in quintile 5. Table 4.4. Distribution of Beneficiaries and Benefits Across Pre-Transfer Quintiles, 2014/15 Grant Type Measure Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Beneficiaries 29.8 26.5 23.2 15.8 4.7 All social assistance Benefits 33.1 26.4 19.7 14.5 6.2 Beneficiaries 35.6 25.0 17.2 14.8 7.5 Older persons Benefits 31.6 24.6 16.6 16.8 10.4 Beneficiaries 37.2 27.0 20.5 11.7 3.6 Disability Benefits 35.5 27.2 19.6 13.2 4.6 Beneficiaries 31.1 27.6 24.5 14.4 2.4 Child support Benefits 32.7 28.3 23.8 13.0 2.1 Source: Own calculations, Statistics South Africa (2015a). Note: Includes direct and indirect beneficiaries of social assistance programmes. This general pattern is not observable for any of the three older persons’ grant, the poorest three quintiles are home to main grants, however. For the child support grant, the larger shares of beneficiaries than benefits, while the richest poorest two pre-transfer quintiles account for slightly two quintiles receive substantially larger shares of benefits higher shares of benefits than beneficiaries; in the case than their shares of beneficiaries. For the disability grant of quintile 1, the share of benefits is approximately 5 though, there is no clear pattern. Given the standardisation percent higher than the share of beneficiaries, while in of grant values, these patterns are driven by a combination quintile 2 this gap falls to half that proportion. For the of differences in household size and in the coverage rates of SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 46 direct beneficiaries in each quintile. For example, the coverage Figure 4.7 explores the impact of social assistance on the poverty rate of the child support grant in terms of direct beneficiaries headcount rate and the poverty gap, using various poverty is 41.0 percent in pre-transfer quintile 1 (see Table 4.1), roughly lines. The food, lower-bound and upper-bound poverty lines 20 times the coverage rate in quintile 5; for the older persons’ are the official poverty lines published by Statistics South Africa grant, the quintile 1 coverage rate is less than three times that (2019b) of R441, R647, and R992 per capita per month in April of quintile 5. 2015 prices. The upper bound (poverty line) of quintile 1 is the Rand value of that designates the boundary between quintiles 4.3. Cost Effectiveness 1 and 2 in the pre-transfer distribution, which is R161.71 per capita per month. It is important to note here that determining Given that one of the key aims of social assistance is to the true effect of social assistance on poverty and inequality is ameliorate poverty, this represents an important metric complicated by the fact that household formation is influenced in assessing social assistance systems. Poverty is measured by available income. In other words, many households would here using the first two of the conventional Foster-Greer- not be viable economic units in the absence of grant income Thorbecke (Foster et al., 1984) (FGT) P-alpha measures. These and would either fragment or would not have formed without measures are the poverty headcount rate (P0), the proportion it. The estimates presented in Figure 4.7 simply consider poverty of the population that falls below the poverty line; the poverty and inequality with and without social grants; they do not allow gap (P1), the average percentage shortfall in the income of the for the dissolution or re-formation of households in response to poor; and the squared poverty gap (P2). As alpha increases from changes in income. zero, so the weight attached to individuals who are furthest below the poverty line increases.10 Figure 4.7. Simulated Poverty and Inequality Reductions (%) Associated with Social Assistance Programmes 100.0 89.9 90.0 81.9 80.0 72.2 70.0 67.4 63.4 60.0 Percent 51.8 50.0 46.6 47.8 42.1 40.0 38.5 34.4 34.7 32.3 32.8 30.0 27.5 23.3 22.0 21.3 22.4 20.0 20.0 14.4 12.8 Pre-Transfer Value 10.1 10.0 8.2 6.7 Post-Transfer Valuee 3.6 Reduction 1.3 0.0 Headcount Gap Headcount Gap Headcount Gap Headcount Gap Lower-bound Food Quintile 1 Source: Own calculations, Statistics South Africa (2015a). Note: Per capita monthly poverty lines in April 2015 Rands are: R441 (food poverty line), R647 (lower-bound poverty line), and R992 (upper- bound poverty line) (Statistics South Africa, 2019b); and R161.71 (quintile 1 poverty line) (own calculations, Statistics South Africa, 2015a). The Gini coefficient is calculated using per capita household income with and without income from social grants. Social assistance significantly lowers poverty in South percent to 46.6 percent). In terms of the poverty gap, the Africa. Based on per capita household income, 46.6 percent reduction is almost one-third (32.3 percent), from 34.4 of the South African population were poor relative to the percent without social assistance to 23.3 percent. The upper-bound poverty line in 2014/15. If income from grants upper-bound poverty line is the highest of the four poverty is excluded, the poverty rate rises to 51.8 percent. lines, and as the poverty line and poverty rates fall so the Thus, social assistance is associated with a 10.1 percent poverty impact of social assistance rises. For the lower-bound decrease in the upper-bound poverty rate (from 51.8 poverty line, the impact on the headcount and poverty gap is 10 The Foster-Greer-Thorbecke indices are a family of poverty measures calculated as: where refers to the size of the population, refers to the number of poor individuals, is the poverty line, and is the income of individual . The parameter is a measure of poverty aversion: a “larger gives greater emphasis to the poorest poor” (Foster et al., 1984). 47 estimated at 22.0 percent and 47.8 percent; for the quintile 1 Figure 4.8 and Figure 4.9 present estimates for regional poverty line, these rise to 81.9 percent and 89.9 percent. and income groupings, as well as the published estimates for South Africa from the ASPIRE database (World Bank, Social assistance significantly lowers inequality measures 2020b). These estimates for South Africa are quite different from in South Africa. The figure also presents the inequality impact the estimates calculated directly from the Living Conditions of social assistance transfers, using the Gini coefficient as the Survey 2014/15 (Figure 4.7); however, having been calculated measure of inequality. Based on per capita household income, according to a standardised methodology, these estimates the Gini coefficient for South Africa in 2014/15 is estimated provide a good sense of South Africa’s performance relative to at 0.68. This represents a 6.7 percent reduction from the pre- other countries. transfer income Gini coefficient of 0.72. Figure 4.8. Simulated Poverty Reduction (%) Associated with Social Assistance Programmes Globally 12.1 High income 27.3 9.3 Upper middle income 20.2 6.5 Lower middle income 14.0 2.3 Low income 5.1 7.3 East Asian & Pacific 14.7 14.3 Europe & Central Asia 27.2 9.5 Latin America & Caribbean 19.6 4.8 Middle East & North Africa 10.0 5.6 South Asia 11.8 6.4 Sub-Saharan Africa 19.9 SOUTH AFRICA 45.7 Poverty headcount Poverty gap 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0 75.0 Percent Source: World Bank (2020b). Note: Estimates for South Africa are the estimates published in the ASPIRE database. In terms of both poverty reduction and inequality 77.6 percent, 68.4 percent, and 60.9 percent respectively. reduction, it is clear that South Africa performs well. According to these estimates, social assistance transfers are Similarly, South Africa’s social assistance system has estimated to have reduced the poverty headcount rate and a substantial impact on inequality, reducing the Gini poverty gap in South Africa by 45.7 percent and 73.4 percent coefficient by 10.5 percent. This is eight times the magnitude respectively. This is a substantially larger impact than is observed of the average impact in upper-middle income countries, and for any of the country groupings. For example, amongst upper- closer to nine times that observed for Sub-Saharan African middle income countries, the poverty rate is reduced by 9.3 countries. Amongst upper-middle income countries, the largest percent on average, while the poverty gap is reduced by 20.2 inequality reductions are observed in Belarus (31.4 percent), percent. In Sub-Saharan Africa, income from social assistance Georgia (19.1 percent), and Mauritius (13.8 percent), with South is associated with declines in these measures of 6.4 percent Africa seeing the fourth-largest reduction (see Table B.3 in the and 19.9 percent respectively. On these metrics, South Africa appendix for detailed estimates). Indeed, these four countries performs well compared to other upper-middle income along with Bulgaria (7.8 percent) and Romania (9.3 percent) countries such as Brazil (10.9 percent poverty rate reduction are outliers: in no other country does social assistance reduce and 38.4 percent poverty gap reduction), Colombia (6.5 percent the Gini coefficient by more than five percent. South Africa’s and 10.6 percent), Malaysia (6.3 percent and 13.3 percent), the performance is all the more impressive given the country’s Russian Federation (16.9 percent and 25.5 percent), and Turkey extremely high Gini coefficient. (3.1 percent and 10.4 percent). Only three upper-middle income countries come close to South Africa’s performance on these metrics: social assistance in Belarus, Georgia, and Mauritius is estimated to reduce the poverty rate by 41.6 percent, 42.6 percent, and 36.9 percent respectively, and the poverty gap by SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 48 Figure 4.9. Simulated Inequality Reduction (%) Associated with Social Assistance Programmes Globally High income 3.2 Upper middle income 1.3 Lower middle income 1.2 Low income 0.2 East Asian & Pacific 1.4 Europe & Central Asia 2.1 Latin America & Caribbean 2.1 Middle East & North Africa 0.6 South Asia 1.3 Sub-Saharan Africa 1.2 SOUTH AFRICA 10.5 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Percent Source: World Bank (2020b). Note: Estimates for South Africa are the estimates published in the ASPIRE database. Once the poverty and inequality effects are disaggregated are associated with a 46.4 percent reduction in this poverty by the type of grant, it is clear that the older persons and measure. The simulated poverty reductions for the individual child support grants are responsible for the majority of grants give an estimate of the impact of a particular grant on its the poverty and inequality reductions associated with own. Since households might receive multiple types of grants social assistance (Table 4.5). In addition to the measures and any of these might lift a household out of poverty, the presented above, the table includes the impact on the squared estimates for the grants individually do not sum to the estimate poverty gap (P2): social assistance transfers in South Africa for all social assistance. Table 4.5. Simulated Poverty Reductions (%) Associated with Social Assistance Programmes, 2014/15 Poverty Poverty Gap Headcount Poverty Gap Squared Gini Coefficient All social assistance 10.1 32.3 46.4 6.7 Older persons 5.8 15.8 23.3 2.7 Disability 2.4 7.1 10.3 1.1 Child support 4.4 16.4 26.0 2.8 Care dependency 0.3 0.7 1.1 0.1 Foster child 0.5 1.5 2.4 0.2 Grant-in-aid 0.0 0.0 0.0 0.0 Other (e.g. social relief ) 0.0 0.0 0.0 0.0 Source: Own calculations, Statistics South Africa (2015a). Note: Estimates based on upper-bound poverty line, set at R992 per capita per month in April 2015 Rands (Statistics South Africa, 2019b). In terms of the poverty headcount rate, the older persons’ furthest below the poverty line, such as the poverty gap and grant has the largest impact, reducing the poverty rate squared poverty gap, the child support grant has the largest by 5.8 percent, followed by the child support grant with impact, followed by the older persons’ grant. Similarly, the child a 4.4 percent reduction. The impact of the disability grant support grant followed by the older persons’ grant have the is considerably smaller at 2.4 percent, with the foster child largest impacts on the Gini coefficient. Schiel et al. (2014, p.20) (0.5 percent reduction) and care dependency (0.3 percent) highlight this ability of the child support grant to reduce income the only other grants with a measurable impact. However, for inequality in their study focussing on the impact of grants on poverty measures that place greater emphasis on individuals inequality, noting that “even though the child support grant 49 makes a small contribution to total income this contribution squared) of poverty are significantly higher. Similar has increased substantially over the post-apartheid period and findings are presented by Leibbrandt et al. (2010). Using data when this is combined with the fact that it is well targeted at for 1997 and 2006, Posel and Rogan (2012) find that “[w]ith the the bottom of the income distribution, it leads to …a notable receipt of social grant income in households, both the extent impact on reducing inequality”. and depth of poverty are significantly lower than they would have been had households relied only on income earned Differences in the strength of the impacts are the result through employment”. of a combination of various factors, including coverage patterns, the value of the grant, and the level of the On the inequality impact, however, the findings here poverty line itself. It is therefore interesting to see two quite stand in contrast to decompositions of the Gini coefficient different grants—the low value child support grant with high that find that grants contribute little to the level of the coverage rates, and the higher value older persons’ grant with Gini coefficient, whether positively or negatively. Van der much lower coverage rates—have relatively similar effects on Berg (2014) finds that in 2005/06 grants contributed less than poverty and inequality. Thus, it is only on a measure such as 0.2 percent (0.001 out of 0.6501) to the Gini coefficient. Similarly, the poverty rate that the high value of the older persons’ Leibbrandt et al. (2012b) show that grants’ contributions to the grant allows it to outperform the child support grant. Gini coefficient in 1993 and 2008 were 0.2 percent and 0.3 Further, the much lower coverage rate of the disability grant percent respectively. These analyses do, though, decompose weakens its eventual impact, despite being a high value grant, the post-transfer distribution, rather than doing direct while no other grants have the type of coverage that would see comparisons of the pre- and post-transfer Gini coefficients. The them have significant effects on poverty or inequality. These findings presented here are, however, consistent with those results are consistent with the findings of Beukes et al. (2016) of the World Bank (2014), which finds a significant reduction who model the poverty effects of changes to the child support in inequality once cash transfers are included within income grant’s eligibility criteria. They find that, of all their simulations, (gross market income less direct taxes, compared with gross simply doubling the value of the child support grant “resulted market income less direct taxes plus cash transfers). in the biggest decline in poverty and inequality” (Beukes et al., 2016, p.523). Finally, Figure 4.10 relates the costs associated with social assistance to the benefits they generate. Specifically, These results—that social assistance programmes reduce the figure calculates a benefit-cost ratio across countries that poverty—are not new, but confirm the continuation of relates the simulated reduction in the poverty gap (the pre- the social assistance system’s poverty-reducing impact. transfer poverty gap less the post-transfer poverty gap) to total Woolard et al. (2010), for example, show similar results using spending on social assistance. In this calculation, poverty is data for 1993, 2000, and 2008. Specifically, they show that defined to be the poorest 20 percent of the income distribution. without grants the poverty rate is marginally higher, Thus, the higher the benefit-cost ratio, the greater the benefit but the depth (poverty gap) and severity (poverty gap for a given cost. Figure 4.10. Benefit-Cost Ratio of All Social Assistance South Africa (2014) 0.34 Sub-Saharan Africa 0.31 East Asian & Pacific 0.33 Europe & Central Asia 0.28 Latin America & Caribbean 0.41 Middle East & North Africa 0.25 South Asia 0.27 High income 0.45 Upper middle income 0.30 Lower middle income 0.34 Low income 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 Poverty gap reduction for each $1 spent on Social Assistance Source: World Bank (2020b). Note: Estimates for South Africa are the estimates published in the ASPIRE database. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 50 The benefit-cost ratio is estimated at 0.34 for South Africa, which performs very well in reducing poverty, this relatively weaker places it within the third quartile of countries for which there are performance in terms of the benefit-cost ratio seems data between 2008 and 2016 (33rd out of 99 countries). While the instead to be linked to the level of costs associated with country slightly outperforms other countries in the region—the social assistance in South Africa. average benefit-cost ratio for Sub-Saharan Africa is 0.31—it lags countries in Latin America and the Caribbean by some margin. 4.4. Impacts of Social Assistance in South In fact, South Africa’s ratio is only three-quarters of the average Africa for Latin American and Caribbean countries of 0.41. In Latin America and the Caribbean, countries such as Argentina, Brazil, Given the various pressing problems around poverty and Ecuador, and Peru perform particularly well with benefit-cost deprivation facing the South African society, and given ratios of 0.525, 0.440, 0.430, and 0.636 respectively. Alongside the existence of an extensive social assistance system Kosovo (0.619), Peru is the top performer globally by this metric. that specifically aims to address these challenges, it is Nevertheless, South Africa’s benefit-cost ratio is slightly unsurprising that a large literature has developed on the above the average for all upper-middle income countries. role and impact of social assistance in addressing these Amongst these countries, South Africa performs better than issues. Importantly, in a variety of areas, there are important Botswana (0.188), Malaysia (0.237), and the Russian Federation differences in effects by gender, which have implications (0.224), and similarly to Albania (0.328), Colombia (0.378), and for policy design as well as for the ability of the social grants Georgia (0.335). system to positively impact South African society more broadly than simply through poverty reduction. This section aims to Given the strong effect of the social assistance system provide a sense of this literature and its findings regarding in reducing poverty in South Africa, the country’s the role of social grants in improving the welfare of deprived performance on the benefit-cost ratio is lower than South Africans across a wide range of dimensions. Box 5 also expected. Given that the country’s social assistance system summarizes the evidence from evaluations of the EPWP. Box 5: Outcomes of the Expanded Public Works Program Programme results of the EPWP has evolved over the four phases but have hinged around employment creation and skills development (enhancing potential to find future work including self-employment); income support and poverty alleviation in poor communities (earning an increased income and improving social security); and, development of community assets and provision of services to benefit communities. The EPWP contributes to different broader social protection functions as per the National Development Plan’s vision of a comprehensive social protection system for the country: • Employment creation and skills development: Though the EPWP is neither creating sustainable employment nor building the human capital of the unemployed, this is to be expected as public employment programs offer short-term unemployment relief and typically do not have medium- to long-term job creation effects (Kluve, 2014). Tracer studies from phase III indicate that 75% of individuals remain unemployed, with 20% employed after exiting the EPWP compared to roughly the same 65% unemployed and 19% employed before joining the program (DPW, 2019). Although the integration of skills development training has been a key EPWP innovation, its success has been limited as the training component of the programme has not been adequate to lead to the acquisition of higher skills (McCord, 2017). In addition, the shorter than anticipated duration of the programme (an average of four months) does not enable meaningful upskilling (DPW, 2019). As a result, many participants return to unemployment status upon exiting the programme. • Income support and poverty alleviation: Income transfers as wages into poor communities not only reduces poverty but is also a form of economic stimulus targeted directly at the poor. Macro-economic analysis indicates that an injection into the economy in the form of EPWP expenditure triggers a positive impact on the whole economy in terms of an increase in output, GDP, and income (DPW, 2019). The increased focus on the CWP and other sectors in the EPWP Phase III also allows this stimulus to address spatial inequality, targeting the poorest areas, and strengthening productive activities in marginalised local economies. In phase II of the program, the EPWP doubled the annual household income for the poorest group11 for the year they were working on the program. Phase III impacts are presented in the table below – both poverty and inequality diminish (DPW, 2019). While incomes received are mostly used to cover family and household expenses, particularly food and utilities, close to 35% spent some of their earnings on education (either for themselves or someone else), while only 5% of participants (compared to 11% before enrolling on the program) indicated borrowing money to live on. The EPWP has thus, positive household formation effects. However, some critics argue that the poverty alleviating effects have been minimal with EPWP wage rates significantly lower than the National Minimum Wage (DPW, 2019). 11 The majority (60%) of participants were poor and had income levels below the poverty line used in the NDP; 32% had income levels that fell below half this poverty line. The poverty line used in the NDP is R419 per capita per month, and half of this amounts to R210 per capita per month. 51 Box 5: Outcomes of the Expanded Public Works Program Poverty headcount index Poverty gap Index Gini Post-transfer indicators 0.541 0.458 0.847 Indicators without EPWP transfers 0.547 0.469 0.850 Source: Department of Public Works and Infrastructure (2019b). Notes: The simulated impact is the change in a poverty or inequality indicator due to EPWP transfer, assuming that household welfare will diminish by the full value of that transfer. Community assets and service provision: These can have transformative social development impacts and include food security, community safety, as well as improving the overall quality of life in communities. Improved psycho-social outcomes (individual and community): Unemployment contributes to a myriad of social problems, with high social costs. These include health problems, depression, alcohol and drug abuse, domestic violence, crime, and alienation from society. As such, participation through work in public employment programs can assist in mitigating these effects through building self-esteem, social networks, providing structure in individuals’ lives, and recognition of their value to their communities. To this end, the EPWP has offered hope to the unemployed who have less than a matric education and were unemployed for a long time, many of them females and young people. Thus, the EPWP has achieved significant success by attempting to address common limitations of public works programs through its innovative methods and approaches. These include creating short-term employment at a large scale, diversifying employment in various sectors, creating incentives and formal obligations for various ministries to share responsibility, and creating employment in social services (Peres, 2019). Even so, while the EPWP provides an important avenue for labour absorption and income transfers to poor households in the short- to medium-term, it was not designed as a policy instrument to address the structural nature of the unemployment crisis. Also, monitoring and evaluation of the EPWP should be strengthened through a more systematic and rigorous comparison of before and after situations to determine program impacts particularly on poverty, inequality, labour market participation, and human capital formation. 4.4.1. Poverty and Inequality Jensen (2004) and Oosthuizen (2013) for the older persons’ grant). These types of positive impacts should not come as too Social assistance and social grants explicitly aim to address much of a surprise, given the extent to which grants are targeted at least the worst deprivations of poverty and, by extension, at the lower end of the income distribution. Leibbrandt and inequality and it is therefore a key area of research interest. A Levinsohn (2011, pp.7-8) find concentration ratios for spending key challenge in terms of the grant system’s ability to on social grants to be between -0.35 and -0.44 between 1995 address poverty and inequality is that, in many instances, and 2006 (where a value of -1 indicates fully progressive and a the values of the grants are insufficient to lift whole value of one is fully regressive), making it more progressive than households out of poverty. This is especially the case any other type of spending they assessed. These estimates for for child grants. This issue is explicitly or implicitly noted by social grants are similar to those by Van der Berg et al. (2010). various authors focussing on both direct and indirect measures of poverty and deprivation (for example, Posel and Rogan, There are various ways in which to look at the poverty 2012; Zimbalist, 2017; Ngubane and Maharaj, 2018; Chakona impact of grants. An example noted above is a focus on a and Shackleton, 2019). Thus, research has not simply focussed particular sub-group targeted by a specific grant. Posel and on the poverty rate, but has also included other measures of Rogan (2012, p.111) focus on gender and find that “receipt of poverty, such as the poverty gap, that are more sensitive to social grant income may have been relatively more effective a narrowing of the distance of poor households from the in reducing particularly the depth of poverty for females and poverty line. Thus, authors tend to find relatively small poverty female-headed households”. Zimbalist (2017) explores the reductions due to grants in terms of the poverty headcount, impact of social assistance in the context of urbanisation. Using but much larger effects for the depth and severity of poverty household survey data, he finds a positive effect of social grants (see, for example, Barrientos, 2003; Armstrong and Burger, 2009; on poverty at the national level but shows that this effect is Woolard and Leibbrandt, 2010; World Bank, 2018). much stronger in rural areas, and that the magnitude of the effect increased over time. In rural areas, “the poverty-reducing Numerous authors have found positive effects on poverty contribution of grants increased by 13 percentage points for of social assistance in general (Armstrong and Burger, the headcount rate and by 19 and 20 percentage points for the 2009; Leibbrandt et al., 2010; Van der Berg et al., 2010; depth and severity of poverty” (Zimbalist, 2017, p.160)8. Woolard and Leibbrandt, 2010; Posel and Rogan, 2012; Inchauste et al., 2015). Others have found similarly positive Importantly, households that have access to social grants effects for specific social grants (for example, Barrientos (2003), are highly dependent on these grants as a source of 8 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 52 income. For example, Delany et al. (2008, p.30) show that, for income (market income less direct taxes plus cash households receiving it, the child support grant accounted for transfers), while Pasha (2016) finds positive effects of 40 percent of household income; further, 21 percent of these grants on multidimensional inequality. households’ income derived from other grants or money from government. More recently, it is estimated that 54 percent of Using dynamic methods, Hundenborn et al. (2016) find the incomes of the chronically poor comes from social grants, that social grants were important in countering increasing as does 25 percent of the incomes of the transient poor World inequality in South Africa between 1993 and 2008. They find Bank (2018, p.36). that “[poverty]-alleviating policies that resulted in an increase in government grants limited the increase in inequality [between Various studies have analysed the poverty-reducing 1993 and 2008] immensely” (Hundenborn et al., 2016, p.20). effects of grants from the broader perspective of Further, the authors find that these policies were able to multidimensional poverty. Pasha (2016), for example, uses successfully reach the poorest households. the National Income Dynamics Survey data to investigate the impact of cash grants on multidimensional poverty and finds While we have not managed to locate South African a positive impact. The author finds that, despite its low value, evidence on this issue, it is important to highlight the child support grant was “able to reduce multidimensional that social assistance transfers represent substantial poverty and inequality amongst each household”, and also injections of resources into local economies, many finds positive effects for the older persons’ grant (Pasha, 2016, of which are characterised by very little in the way of p.38, 39). economic activity. As a result, these large injections of cash have the potential to stimulate particularly rural local economies The evidence on the effect of grants on inequality is less and can achieve relatively large multiplier effects. In their review clear. Studies that perform static decompositions of inequality of the evidence, for example, the FAO (2015) show that local by income source find that grant income slightly lowers income multipliers of cash transfer programmes in Africa range (Armstrong and Burger, 2009) or marginally raises (Hundenborn between 1.25 and 2.52 (i.e. one dollar transferred through a cash et al., 2016, using 1993 data) inequality, or has no discernible transfer is able to generate between 0.25 and 1.52 additional impact (Leibbrandt et al. 2012a; Hundenborn et al. 2016, using dollars of income in the local economy). Such multiplier effects 2008 and 2014 data). suggest the potential for greater coordination between social assistance programmes and interventions aimed at supporting Agüero et al. (2007) consider the effect of government small-scale farming activities in South Africa’s rural areas. through taxes and social grants on income distribution, using the KwaZulu-Natal Income Dynamics Survey panel 4.4.2. Nutrition, Food Security, and Hunger data, and find that these reduce the Gini coefficient by 8.5 points in the 2004 data. They find, however, that the The impact of grants on nutrition and other related effect increased from 4 points in 1993 to 7 points in 1998, a outcomes, such as prevalence of hunger, are mixed. period they note as coinciding with improvements in the grant Closely related to concerns around poverty, the impacts on in terms of coverage and benefit level. They also find a negative nutrition have also received attention within the South African correlation—indicating progressivity—between grant income literature but impacts are not always discernible and appear and their measure of expenditure having excluded the effects to be dependent on the exact outcome variable used. Further, of taxation and grants, with a correlation coefficient of -0.30 effects may be mediated by the gender of the grant recipient. (Agüero et al., 2007, p.806). Duflo (2000; 2003) investigates the impact of receipt Schiel et al. (2014) focus specifically on understanding of the older persons’ grant on the nutritional status of the effect of social grants on inequality. Using the young children, using weight-for-height and height-for- Lerman and Yitzhaki decomposition of inequality by age to assess shorter and longer term impacts of grant income source, a static approach, they find that grants receipt. Using the 1993 PSLSD data, she finds that receipt of have “either a negligible effect or small equalising effect the older persons’ grant is associated with improvements in on total income inequality” (Schiel et al., 2014, p.9). the nutrition of young co-resident girls, but not for boys. Thus, However, dynamic approaches yielded different results. Their receipt of the grant was associated with improvements in “the results lead them to conclude that “social assistance awarded height-for-age z-scores of younger girls by at least 1.16 standard to the elderly has had little effect on equality …[changes] in the deviations, and the weight-for-height z-scores of girls by 1.19 targeting of the state pension have led to a small disequalising standard deviations”, roughly equivalent to the gap between effect …[while] additional social protection programs initiated South African and US girls (Duflo, 2003, pp.21-22). Importantly, in the post-apartheid era have had an equalising effect” (Schiel Duflo finds that the effect is driven by grants received by et al., 2014, p.20). women. The 2000 analysis covers only height-for-age and the size of the effect is closer to half the gap between South African Inchauste et al. (2015, p.29) finds that the Gini coefficient and US girls (Duflo, 2000, p.398). falls from 0.750 to 0.694 when comparing net market income (market income less direct taxes) to disposable Based on data on 290 households in the Agincourt 53 demographic surveillance area, Case and Menendez but in a number of these studies, there are important (2007) investigate the impact of the older persons’ grant differences according to the gender of grant recipient on adult and child nutrition. They find an approximately 20 and, indeed, according to the gender of the child. percentage point reduction in the incidence of adults missing meals in households that have access to the grant (Case Case et al. (2005) find a positive relationship between and Menendez, 2007, p.160). They also find that, by raising receipt of the child support grant and school enrolment household income, the older persons grant also reduces the using longitudinal data collected as part of the Africa incidence of hunger amongst children (Case and Menendez, Centre for Health and Population Studies’ demographic 2007, p.161). surveillance system in KwaZulu-Natal. At the time, the child support grant was only available to children under Using data from the National Income Dynamics Survey, 7 years. The authors found that receipt of the grant was Coetzee (2013) finds a positive effect of the child support associated with an increase in the school enrolment rate of 8.1 grant on households’ food expenditure, and on children’s percentage points for six-year-olds, and 1.8 percentage points height-for-age. However, she notes that, while “these for seven-year-olds (Case et al., 2005, p.479). This finding, they estimates do provide some evidence of the positive effect of argue, is particularly important given that child support grant the CSG on the lives of children, the estimates are small and beneficiaries live in poorer households and their caregivers have do not provide clear evidence that the transfers received by lower levels of education than is the case for non-beneficiary caregivers are spent mainly on improving the well-being of children. The presence of an older persons’ grant within the children” (Coetzee, 2013, p.448). household was also associated with an increase of 5 centimetres in a child’s height for age, equivalent to approximately one Waidler and Devereux (2019), using data from the standard deviation (Case, 2001, p.11). National Income Dynamics Survey, focus on three indicators of food security, namely total expenditure on In their analysis of Agincourt data, Case and Menendez food, dietary diversity, and body mass index. They find, (2007, p.162) find evidence of a positive impact of the for example, that while the older persons’ grant is associated older persons’ grant on school enrolment for girls, with improvements in the dietary diversity index, this is not the with girls living with recipients of the grant being ten case for the child support grant; they further find no significant percentage points more likely to be enrolled in school relationship with total household expenditure on food for either than their same-aged counterparts in similarly wealthy grant (Waidler and Devereux, 2019, p.693). While the authors households. Further, they find that this relationship is linked find no effect on the body mass index for the older persons’ to female pensioners specifically: “[all] else held constant, girls grant, they do find some evidence of a positive effect for the living with a woman receiving the pension are 14 percentage child support grant (Waidler and Devereux, 2019, p.693). points more likely to be enrolled in school” (Case and Menendez, 2007, p.162). No positive effect was found for boys, however. Based on their quantitative and qualitative data collected from 554 women aged 15-49 years in Richards Receipt of the child support grant has been found to Bay, Dundee, and Harrismith in a study looking at the be positively associated with progress through the role of social grants and the consumption of wild foods schooling system (Coetzee, 2013). Similarly, income from as options to address hunger, Chakona and Shackleton the older persons’ grant has been found to positively impact (2019, p.92) argue that social grants have “no significant rates of school attendance, particularly for girls (Samson et influence” on household food security. The study, however, al., 2001, as cited by Leibbrandt et al., 2010). Pension income does not assess the potential impact in a multivariate setting, is found to positively impact total years of schooling for co- nor does it account for the fact that, while household food resident children aged 6-19 years, with the effect larger for security is a function of household welfare levels, grant receipt girls, although for children aged 13-19 years the effect differs is both a function and a determinant of household welfare by gender of the pensioner: education is increased for boys levels. Instead, the argument rests on comparing the means and decreased for girls if the recipient is a male, while female of the chosen measures of food security between households recipients have little effect on either (Hamoudi and Thomas, receiving social grants and those not receiving social grants. 2005, as cited in Leibbrandt et al., 2010). Findings from the qualitative data did, however, highlight the fact that grant recipient households were highly dependent on More recently, Standish-White and Finn (2015) find that grants as a source of income and the perception of grant values the older persons’ grant impacts co-resident children’s being very low (Chakona and Shackleton, 2019, p.89). education positively, irrespective of the child’s gender. Using National Income Dynamics Survey data, the 4.4.3. Education authors find that, on average, in households with female pensioners, girls obtained an additional 0.6 years of A number of papers have focussed on the relationship between education and boys an additional 0.4 years compared receipt of different grants by households and educational with their counterparts in other households. While a outcomes of school-aged household members. The impact of similar effect is not found for male pensioners, they were found social grants on education outcomes tend to be positive, to provide other benefits: “[girls] living with a male pensioner SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 54 miss 1.3 fewer days of school each month on average” (Standish- in sexual partnerships, thereby reducing the opportunity for White and Finn, 2015, p.24). [intimate partner violence]”, which may in turn reduce the risk of HIV infection (Kilburn et al., 2018, p.47). 4.4.4. Health Finally, Eyal and Burns (2018) investigate the impact of Case (2001) analyse data collected in a 1999 survey of cash transfers from the perspective of mental health 300 households in the Langeberg Health District in the and, in particular, the intergenerational transmission Western Cape to investigate the relationship between of depression. Using panel data from the National Income receipt of the older persons’ grant and health outcomes. Dynamics Survey, the authors find large and significant The authors find a positive relationship for the health outcomes protective effects of receipt of cash transfers for teenagers’ of adult household members, including the recipients of the mental health. Specifically, their results suggest that “CSG older persons’ grant, that differed according to whether or receipt reduces parental depression transmission by forty-five not the households pooled income: where they did not, grant or sixty-seven percent, for maternal or paternal depression recipients reported significantly better health status than other respectively”, and that rates of transmission are also lower adult household members, but this difference disappeared where households are in receipt of other social grants (Eyal and where there was pooling. Burns, 2018, p.44). A 2013 set of case studies by Knight et al. (2013) found 4.4.5. Labour Supply and Livelihoods that the disability grant was important in terms of enabling households to care for members undergoing One of the key concerns around social assistance generally is anti-retroviral treatment for HIV. Thus, income from the its perceived negative impact on the willingness of recipients disability grant helped to compensate for the sick member’s to work and establish sustainable sources of income from lost labour income and, at the same time, bolstered their ability work. This is an issue for direct beneficiaries (i.e. the elderly to care for them until such time as they were able again to themselves) and for indirect beneficiaries (i.e. co-resident contribute to the household (Knight et al., 2013, p.145). working age adults). One of the key complicating issues is the way in which households form and reform in response to Using data from the Agincourt Health and Socio- changes in the financial resources available. This is particularly Demographic Surveillance System site in rural important when analysing labour supply in South Africa, given Mpumalanga, Pettifor et al. (2016) investigate whether that labour migration—and therefore household exit—is an cash transfers might be able to reduce the likelihood of young important strategy amongst work seekers. The literature women contracting HIV using a randomised controlled trial in has found both negative and positive impacts of social which approximately 2 500 young women, aged 13 to 20 years, grants on labour supply, although more recent studies and their parents or guardians were enrolled over a period of up have tended to find positive impacts. to three years. Half of the young women received a cash transfer of R100 per month, while their parent or guardian received One of the earliest papers to investigate the impact R200 per month, conditional on an 80 percent attendance of social grants on labour supply in South Africa is rate at school. The authors find “no significant effect of a cash that by Bertrand et al. (2003). The authors make use of the transfer conditional on school attendance on HIV incidence 1993 PSLSD data to analyse the labour supply response of in young women”, nor do they find a positive effect on school working-age African adults aged 16-50 years to the presence attendance although they note that attendance rates were of an older persons’ grant within the household and find very high (95 percent) in both study groups (Pettifor et al., 2016, significant negative effects on labour supply. Specifically, men’s pp.e983-e984). However, they do find that school attendance employment rates and hours of work decline in response to was associated with a lower risk of contracting HIV, whether or an increase in pension income, while for women the impact is not the individual received the cash transfer (Pettifor et al., 2016, smaller and limited to working hours. The authors further find p.e983). that the effect is larger where the individual receiving the older persons’ grant is a woman (Bertrand et al., 2003, p.43). Some research has explored the relationship between cash transfers and gender-based violence. Kilburn et al. Jensen (2004) use the 1993 Project for Statistics on (2018), for example, consider the effect of a conditional cash Living Standards and Development (PSLSD) to analyse transfer on young women’s risk of physical intimate partner potential crowding out of private income by public violence, using the same Agincourt study as Pettifor et al. (2016). transfers. They also find evidence that receipt of the older The authors find that receipt of the transfer was associated with persons’ grant is associated with lower household earnings. reduced risk of physical violence, but not for sexual violence, However, the estimated effect is small: “receiving a pension of and had positive effects on sexual debut, having a sexual 370 rand reduces home income by about 22 rand, which is less partner in the preceding 12 months and the number of sexual than 3 percent of the average household income” (Jensen, 2004, partners in the preceding 12 months (Kilburn et al., 2018). p.108). Further, the authors do not find evidence to support the Further, the authors find that the reduction in risk of intimate idea that pension receipt affects household composition. partner violence “is due in part to girls choosing not to engage 55 Ranchhod’s (2006) analysis of Labour Force Survey data of the child support grant on labour market outcomes focuses on the labour supply of elderly Africans. He finds amongst African mothers between the ages of 20 and that receipt of the older persons’ grant has a significant effect on 45 years who are co-resident with their children. They their labour supply: a reduction of 8.4 percent for elderly men find that receipt of the child support grant is associated with and 12.6 percent for elderly women. He further finds a positive a higher likelihood of labour force participation and, amongst impact on the likelihood of the elderly being employed in jobs those who were economically active, a higher probability of where they have full control over the number of hours worked, employment. which also implies a reduction in their labour supply. Sinyolo et al. (2019) explore the issue from the angle of Eyal and Keswell (2008) replicate the approach by smallholder farmers and their engagement with local Bertrand et al. (2003) using later Labour Force Survey data markets. Using data they collected in four districts and, while they find that receipt of an older persons’ grant is within KwaZulu-Natal from 774 smallholder farmers, negatively related to labour supply, the effect is much smaller. the authors explore the effects of grant receipt on the The authors suggest that this may be due to the earlier study probability and level of market participation. Based having picked up the effects of the equalisation of the value of on their results, they conclude that “social grants undermine the pension across race groups. smallholder incentives to produce a marketable surplus or sell their agricultural produce”, decreasing both the probability and By including migrant household members in defining level of participation in the market for maize (Sinyolo et al., households, Ardington et al. (2009) find that the presence 2019, p.466). of a recipient of an older persons’ grant raises labour supply amongst working-age household members. Using longitudinal Lovo (2011), however, found that receipt of the older data on roughly 100 000 individuals in KwaZulu-Natal, the persons’ grant had a positive impact on farming authors find that, in households where at least one member households’ technical efficiency amongst a sample of 549 receives an older persons’ grant, the likelihood of employment farming households from the third wave of the KwaZulu- amongst prime working-age adults increases by approximately Natal Income Dynamics Survey, and allowed households three percent (Ardington et al., 2009, p.32). to increase their involvement in both on- and off-farm productive activities. Grants may also impact on labour supply indirectly. In their examination of the persistence of unemployment 4.4.6. Fertility and Childbearing in South Africa, Klasen and Woolard (2009) argue that access to resources plays an important part in Amongst all the social grants provided by the South determining where the unemployed locate themselves. African government, the child support grant is perhaps Without access to unemployment insurance, the unemployed the most contentious within the public discourse. must rely on private support networks and, since familial Given that children are the direct beneficiaries of the grants, support is often located in rural areas, the unemployed are some have argued that the grant incentivises childbearing, drawn to economically distant areas where job search is difficult particularly amongst young and teenage women. A strong and employment opportunities relatively scarce. Thus, by at counterargument, however, is that the low value of the child least partly funding this type of familial support, older persons’ support grant means that it is unable to cover the costs grants may contribute to longer unemployment spells. associated with raising a child and therefore provides little incentive to fall pregnant. Overall, the literature does not The dynamic nature of household formation is explored find any positive impact of social grants on childbearing. further by Ranchhod (2017), who investigates the effect of a loss of pension income on household composition Rosenberg et al. (2015) analyse data from the Agincourt Health and the employment of household members using and Socio-Demographic Surveillance System in Mpumalanga, panel data from the Quarterly Labour Force Survey. Loss as well as data from the Africa Centre Demographic Information of pension income is indeed found to impact on households System, to investigate the impact of grant receipt—specifically in important ways, including out-migration of school-aged the child support grant—on second pregnancy. Instead of children and young, typically non-employed adults. Households providing incentive to fall pregnant, the data points to the gain older adult members and, at the same time, an increase opposite effect: “[time] to second pregnancy was significantly in the probability of older adults being employed is observed. longer among CSG recipients compared to non-recipients at As a result of these changes, young adults that remain within both the 25th…and 50th percentiles” (Rosenberg et al., 2015, the household are one-fifth more likely to be employed than p.7). Further, they found no evidence that losing access to the before, while there is an increase in the amount of time older child support grant for the first child was associated with a adults allocate to productive activities in the market and the second pregnancy and, importantly, that these results hold true home (Ranchhod, 2017, pp.12-13). for both younger and older women (Rosenberg et al., 2015, p.8). Eyal and Woolard (2011) use a decade’s worth of national Ngubane and Maharaj (2018) approach the issue household survey data to explore the impact of receipt qualitatively, conducting in-depth interviews with 15 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 56 African mothers aged 18-24 years in a rural area in four national food distribution NPOs distributed 151 276 KwaZulu-Natal. Only mothers who were in receipt of a child food parcels valued at almost R56 million; community- and support grant were selected for participation. Consistent across faith-based organisations distributed another 69 000 parcels all participants was the assertion that the pregnancy was valued at almost R27.5 million; and, in a partnership with the unplanned and unmotivated by access to the child support South African Council of Churches, 23 500 food vouchers were grant. Further, the interviews confirmed the small value of the distributed. By mid-June, the Fund expects to have reached child support grant relative to expenses, but that “the grant over 300 000 households. benefited children especially in the absence of financial support from their fathers” (Ngubane and Maharaj, 2018, p.7). Implementing a brand new grant under conditions of a national lockdown was always going to be a challenge. New grants are Similarly, Makiwane (2010) does not find any evidence introduced infrequently and, as noted above, the lockdown of a link between the child support grant and trends in imposed important constraints in terms of enrolling new teenage fertility. applicants into the system and paying them. In many respects, SASSA therefore found itself in uncharted territory 4.4.7. Shock responsiveness and under massive pressure to rapidly implement a programme at scale using untested approaches and The social assistance response to the lockdown-induced technologies. The speed with which the crisis unfolded and humanitarian crisis has been impressive in scale. its immense scale exposed important weaknesses within Government announced a boost of R300 per beneficiary for the the system, ranging from issues around design, to technical child support grant for May 2020, to be followed by an increase capacity to rapidly roll out a grant at scale, to communication of R500 per month per recipient (i.e. caregiver) for the child around the new grant. support grant in the following months, as well as an increase of R250 per beneficiary per month for all other grants. In addition, In terms of design, it became clear very early in the the R350 per month COVID-19 grant was introduced. In total, process that the current suite of social grants, while depending on take-up rates and the speed of rollout, these effective at reaching the poorest members of South interventions have been estimated to cost as much as R45 Africa society, would be insufficient to address the billion over a six-month period (Bhorat et al., 2020). Importantly, fallout associated with the lockdown. In particular, a this intervention is just one component of a much broader large portion of informal sector workers—who would policy response. have suffered almost complete loss of income during the lockdown—are not co-resident with grant recipients In the context of the lockdown, conventional social relief (Bhorat et al., 2020). However, it was also clear that government of distress interventions have played an important role, had no real way of identifying these individuals through any even considering the challenges associated with trying single database at its disposal. Eligibility for the COVID-19 grant is to reach the kind of scale that the situation required. Data instead determined by cross-referencing multiple government recently published by SASSA (2020d) indicate a massive rollout databases belonging to SASSA, the Unemployment Insurance of additional social relief of distress. In March 202012, SASSA Fund, the South African Revenue Service, the Government reported 10 762 beneficiaries receiving social relief of distress; a Employees Pensions fund, and the National Student Financial month later this had more than doubled to 26 619 beneficiaries Aid Scheme. It should be noted that the challenge in estimating and averaged nearly 23 000 beneficiaries each month over and determining the number of people who may be eligible the May-July period. According to the Auditor-General (2020), for the grant also had a lot to do with budget limitations and SASSA distributed 146 963 food parcels, valued at almost R177 care had to be taken not to over-promise on the number of million, to applicants between the end of March 2020 and estimated beneficiaries. 11 May 2020, with SRD applications made from 11 May 2020 diverted through the COVID-19 grant application process. Other design-related problems include SASSA’s underestimation of the extent to which a purely The Solidarity Fund—a public benefit organisation electronic application process would represent a critical established in the wake of the outbreak of the pandemic barrier to access and initial eligibility criteria that were to “support the national health response, contribute to simply not feasible in the context of the lockdown. humanitarian relief efforts and mobilise South Africans Reliance on an electronic system was necessitated by the in the fight against COVID-19” (Solidarity Fund 2020b)— lockdown regulations, but it also represents an efficient means details the broader relief effort in terms of food relief. of receiving very large numbers of applications. More recently, According to the Fund, 59 811 food parcels were distributed despite the relaxation of the lockdown, SASSA is still only through DSD’s Community Nutrition and Development taking electronic applications while the appeals process is also Centres, financed through a R20 million contribution from the electronic. Social workers were dispatched to communities DSD and R23.5 million from the Solidarity Fund. However, this to help people prepare and submit their COVID-19 electronic was only one of four pillars in the Solidarity Fund’s approach: grant applications. 12 The SASSA (2020d) publication refers to March 2019, but this seems to be an error. 57 SASSA’s technical capacity to implement the COVID-19 social grants system occurred at the beginning of this period: grant has also been put to the test. While relatively between 2002/03 and 2003/04, the number of grants grew little information has been made public as to the progress from 5.8 million to 7.9 million, an increase of just over 2.1 million with rolling out the grant, it is clear that there have been or 36.7 percent. Approximately 1.2 million COVID-19 grants significant problems in terms of capacity to process the were paid out in the (just less than) six weeks between the number of applications, despite the electronic application opening for applications on 11 May 2020 and the 18 June 2020 process. However, building the whole system in 30 days was media release, and in November 2020 over 6 million people a mammoth undertaking. A SASSA media statement from 25 had received the grant. This is equivalent to processing and May 2020 indicated that they had tested the payment system paying out 10.4 million new grants in a 12-month period; this is on a sample of ten beneficiaries, with one failing due to an error a remarkable pace and would represent roughly five times the in the submitted bank details. According to SASSA the ideal largest annual increase in grant recipients observed since 2002. time between application to payment was seven days. In the beginning, processing took much longer as the system was The process has also been plagued with unclear developed at the same time and the number of applicants was communication from SASSA and government more very high. Further technical capacity constraints are evident in broadly. Perhaps most glaring was the fact that the initial the problems experienced in implementing the increases to announcement by the President of the intervention the existing grants, with reports of some recipients being paid did not match the programme that was eventually twice, while others were not paid at all. implemented. Specifically, the initial announcement spoke of higher payments per child support grant beneficiary, To put these figures in context, it is worth looking back while the implementation has been in terms of child support at data detailing the expansion of the grant system over grant recipients (i.e. caregivers). This has a material impact the past 20 years. An extended time series of detailed annual for households depending on the number of children and data on grant beneficiaries by programme is difficult to locate, primary caregivers, and in terms of the progressivity of the but based on estimates from National Treasury (2007, 2009) programme (Box 6, Bhorat et al., 2020). There has also been poor and SASSA (2019) it is possible to construct a series going back and conflicting communication with respect to the required to the 2002/03 financial year. The most rapid expansion of the documentation for applications. Box 6: Take-up of the COVID-19 Grant Given the timing of the implementation of the COVID-19 grant evidence on the patterns of coverage, take-up or impact of the grant has only recently begun to emerge. A key source of these emerging data is the National Income Dynamics Study – Coronavirus Rapid Mobile Survey (NIDS-CRAM), a panel survey of South African individuals that derives its sample from Wave 5 of the National Income Dynamics Study (NIDS). The NIDS-CRAM survey is planned to include five waves during 2020 and 2021, with the second wave having already been conducted, and the data is considered to be “broadly nationally-representative” (www.cramsurvey.org). Using the second wave data from NIDS-CRAM, Köhler and Bhorat (2020) analyse, amongst other things, take-up and coverage of the COVID-19 grant: We estimate that as of the time of the NIDS-CRAM Wave 2 survey in July and August 2020, of the 11.33 million individuals who reported applying for the grant, 4.32 million (nearly two in every five, or 38.1%) were successful. The remaining 7 million individuals either report a pending (4.35 million, or 38.5%) or rejected (2.65 million, or 23.4%) application. However, application for and receipt of the grant appears to have been relatively pro-poor: most individuals who applied for the grant, and were successful in their application, are in the middle and lower parts of the June 2020 household income distribution [see figure below]. Conditional on applying, 23% of individuals (1.4 million) in the poorest quintile of households were successful, in contrast to 4.5% (250 000) in the richest quintile. Close to 90% of individuals in this latter group never applied, in contrast to nearly one in every two individuals in the poorest quintile. Up to the richest quintile, pending applications do not vary considerably across the distribution, although individuals in the poorest quintile of households were more likely than others to experience this outcome (17.64%, or 1.1 million individuals). SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 58 Application Status for COVID-19 SRD Grant across the Distribution, June 2020 Source: Reproduced from Köhler & Bhorat (2020). Receipt of COVID-19 SRD Grant across the Distribution, June 2020 Source: Reproduced from Köhler & Bhorat (2020). Although the above findings refer to application (successful or not) of the grant at the time of the survey, we can also analyse variation in actual receipt in June 2020. [The figure above] presents the distribution of personal receipt of the COVID-19 SRD grant, in June 2020, across the June 2020 household income distribution. Our aforementioned finding holds: in both absolute and relative terms, individuals who live in poorer households were more likely than others to receive the grant. About 11.5% of individuals (or 720 000) who live in the poorest 20% of households received the grant in June 2020. This is in sharp contrast to the 3.3% (184 000) who live in the richest 20% of households. In other words, for every person who lived in quintile 5 households and received the grant in June, nearly four who lived in quintile 1 households received the grant. Household-level receipt was also progressive, as indicated in Figure 6: of the 7.9 million individuals who co-resided at least one household member who received the COVID-19 SRD grant in June, about three in every five (59.5%) live in the poorest 40% of households, as opposed to 5.6% who live in the richest quintile of households. Source: Köhler and Bhorat (2020, pp.15-16) Finally, there has been a clear tension between accuracy have sky-rocketed to over R81 million by October”. The delay in of targeting and speed of rollout. While targeting of grants payments was directly blamed on the application verification is important in terms of ensuring that those who most need process in SASSA’s June media release (SASSA, 2020a). Clearly, support receive it, it has come at the cost of time. In their SASSA is required to appropriately screen applicants, but the apology for the slow rollout, SASSA (2020b) highlighted how lack of integration of systems across departments combined they “have eliminated a number of undeserving applicants with an approach that arguably over-emphasises eliminating and this has saved the Fiscus close to R14 million which could leakages appears to be contributing to the slow response. 59 Certainly, a saving of R81 million is substantial, but it pales in to share data so as to screen applicants. This may potentially contrast to the total estimated cost of COVID-19 grant and, provide a basis—and precedent—for further cooperation indeed, to the desperate need for support. and integration as part of the NISPIS project. Importantly, the implementation of the COVID-19 grant may be breaking Despite these problems, there have been important through some of the resistance to the idea of expanding social successes. These include the relatively quick agreement assistance to better support working-age cohorts through, for across various government departments and institutions example, some type of basic income grant or jobseeker grant. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 60 5. Effectiveness of social assistance in South Africa In this chapter we summarize the strengths and areas gap squared, that place greater emphasis on individuals for improvement of the social assistance system and furthest below the poverty line. Thus, while social grants may discuss how well it is aligned with the country’s broader be insufficient to lift the poor completely out of poverty, they development challenges. While recognising that it is not do go a long way towards ameliorating the deepest poverty in the role and mandate of social protection or social assistance the country. Additionally, the beneficial impact of the grant to address the underlying social and structural challenges in system extends far beyond money-metric poverty and a society, but rather to protect its vulnerable groups from the inequality as illustrated by the numerous studies that consequences, we do ask the question “to what extent is South have found important broader positive effects of grant Africa’s social assistance system geared to mitigate and reduce receipt on various aspects of human development. Such the structural causes of poverty and inequality and to improve findings suggest the potential for the grant system to economic inclusion, upward mobility, and human capital have positive effects that play out intergenerationally investment of the poorest”. and over the long-term. However, these benefits could be multiplied if the social assistance programs could be 5.1. Strengths of the social assistance system intentionally and explicitly reformed towards nudging households’ investments in human capital. The social assistance system has a number of important strengths. First, the social grant programmes are These strong effects on poverty and inequality are the extensive. Approximately one in three South Africans is benefits of a system that is well-targeted at those who a direct beneficiary of a social grant. This is primarily due most need support. Coverage is almost universal in the to the number of beneficiaries of the child support grant— poorest pre-transfer quintile (95.2 percent), while it remains children receiving a child support grant represent almost one- around three-quarters (74.1 percent) in the third quintile. As a quarter of South Africans according to the Living Conditions result, the poorest 60 percent of the population account for Survey 2014/15 data. More than two-thirds of children under almost 80 percent of all direct and indirect grant beneficiaries, the age of 18 years received child support grants, while over 70 and a similar proportion of social assistance benefits. Quintile percent of adults aged 60 years and above received the older 1 alone accounts for 29.8 percent of direct and indirect persons’ grant. On average, coverage of South Africa’s social beneficiaries and 33.1 percent of benefits. assistance grants is substantially higher than for other upper- middle income countries. Importantly, despite the fact that grant values are low in absolute terms, the extent of inequality means that Second, the system is well targeted. The ability of the they are relatively large for a significant proportion of system to provide benefits to those in need is critical households. The average transfer per capita for beneficiary if it is to have the intended beneficial impact in terms households in 2014/15 is estimated to have been only R3 279, of reducing poverty. More than half (56.1 percent) of the or around R273 per month. However, compared to beneficiary population in the poorest pre-transfer quintile are direct households’ per capita household expenditure, this amount is beneficiaries of the grant system, while coverage for the significant. Averaged across all beneficiary households, grant child support and older persons grants of the age-eligible income is equivalent to roughly one-quarter of per capita population in the bottom quintile is 86.9 percent and 96.6 household expenditure. However, this figure is as high as two- percent respectively. The poorest 40 percent (pre-transfer) of thirds for beneficiary households in quintile 1 and two-fifths for the population account for 56 percent of beneficiaries and beneficiary households in quintile 2. almost 60 percent of the total benefits. In some sense, one might expect that these proportions of beneficiaries and In summary, South Africa’s social assistance system is benefits could be even higher, ensuring a more concentrated effective in providing support to the poorest segments focus on those at the lower end of the income distribution. of the population and significantly reduce the depth However, the country’s extreme level of inequality means that of poverty and inequality. Further, by providing regular, even households in the middle of the distribution have low dependable income, they ameliorate vulnerability. This is incomes and require social assistance. particularly true if the effects of social grants on other outcomes such as health, education, and labour supply are considered. Third, the system has a significant impact on both poverty and inequality. Based on the Living Conditions 5.2. Shortcomings/areas for improvement of Survey 2014/15 data, it was shown that social assistance the social assistance system significantly reduces poverty across a broad range of poverty lines. Based on simulations using the LCS 2014/15 data, it was At the same time, the social assistance system does have estimated that social grants reduced the poverty headcount some weaknesses from the perspective of this research. rate by between 10.1 percentage points and 81.9 percentage These weaknesses, or better yet, areas for improvement, points depending on the poverty line used. Similarly, the presently limit the ability of the system to address the structural Gini coefficient was reduced by 6.7 percent. The impact is causes of poverty and inequality beyond just providing relief. stronger for measures, such as the poverty gap and poverty 61 5.2.1. Categorical programs addressing individual risk functions, all of which drive up the cost of the system. This type of fragmentation and duplication is not unique to SASSA and First, although the system of social grants reaches the main the DSD but is widespread across government. vulnerable groups, an inherent weakness with social assistance systems built around lifecycle risks and 5.2.3. Working-age adults composed of programs aimed for particular categorical groups (although means-tested to target the poor) is A third area for improvement—and one that has been that they tend to lack measures to assist households to identified by a number of authors—is the system’s blind transition out of poverty, and may have limitations in spot around working-age adults, who have no access to providing coordinated support to poor households over social assistance unless they have a disability. While there time. In the South African case, households are only supported are some programmes within the social protection system that by social grants if they include a child, a person with a disability, cover working-age adults, each of them is limited in terms of or an elderly person. But in South Africa households are formed their coverage. The only social grant accessible to working-age around income and decisions on household composition and adults is the disability grant, which is predicated on disability; labour force participation may be endogenous choices as their and unemployment insurance and the Compensation Funds needs change over time. On the other hand, grants aimed for are only accessible to formal sector workers. During the individual groups of the population tend to be relatively fixed COVID-19 crisis, the government introduced some temporary and leave policy makers with limited flexibility to support relief programs targeted at working adults; It is estimated that households dynamically and promote mobility towards the COVID-19 SDR grant, which is meant to close in January productive inclusion and better access to the labour market 2021, provided benefits to around 6 million people, while the and self-sustainability. We delve on this broader system design Employment Stimulus may only, with its best effort, create challenge and its opportunities more in the next section. up to a million work opportunities, the target being 800 000. However, these programs are meant to be phased out as 5.2.2. Integration and coordination the current crisis ends. Hence, the working-age population represent a key gap in the system in terms of coverage. Indeed, Second, integration across programmes, government as Altman et al. (2014, p.349) note, “there is no social assistance levels and departments is not particularly strong. This aimed at able-bodied working-age young people”. represents a lost opportunity to build the types of synergies that could lead to strong positive impacts for While programs such as basic income guarantees, universal programmes, both individually and collectively. Such basic incomes, or job-seeker grants have been discussed and integration may be particularly beneficial for the child support debated from time to time, the EPWP and CWP are the only grant, which has already been shown to have important positive interventions available to the majority of working-age effects on human development. The delivery systems behind the adults. While they can potentially play an important role South African social protection programs are technically highly in establishing a minimum level of income, their current capable and benefits from strong systems for targeting, case coverage is limited. Also, they do not benefit from systematic management, data administration, and payments. But there is rigorous impact evaluation to fully know their impacts of room for improvement, especially in terms of coordination and poverty reduction and sustainable job creation (see Box 5). The integration, starting with delineation of responsibility for the recent Presidential Employment Stimulus initiative is building three levels of government, the interoperability of databases upon and scaling up these same approaches exponentially across government department, as well as last-mile payment in an attempt to provide work opportunities to those who services. This lack of interoperability was one of the weaknesses have been left idle by the COVID-19 crisis and to the millions in terms of the COVID-19 response as discussed below. who were already chronically unemployed before the crisis. As with interventions aimed at children, the EPWP and CWP Setting up a unified social registry, such as the NISPIS, programmes may benefit from greater integration in order and linking together and making interoperable a number to link the unemployed back into the labour market. Indeed, of government databases would be a significant step in there is scope for integration with labour market interventions the right direction. Given the long-term consequences of through the Department of Employment and Labour or the investment (or lack thereof ) in children’s human capital, there Pathway Management Network, for example, to strengthen should be strong incentives to do as much as possible to overall outcomes. strengthen impacts. This is particularly true within the current fiscally constrained environment. Similarly, as discussed below, The result of weak coverage of the working-age adult there is scope to better integrate the public employment cohort has important implications for other social programs to provide beneficiaries with a wider set of support and assistance interventions, as benefits received by children job-seeker services. However, the governance improvements and the elderly are shared with working-age adults in the and integration ought to go beyond ICT platforms. Mthethwa same households who have no other means of support. (2019, p.103) notes the lack of integration of the institutional While many recipients of child support grants are working-age and administrative frameworks related to social security. At the adults (97 percent are women), these grants are meant for the very least, this leads to duplication of work, of processes, and of children. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 62 What makes the gap for the working-age adults particularly 5.2.5. Shock responsiveness – ability to scale up to glaring is the fact that the social security system is not address crises designed or equipped to provide comprehensive protection against unemployment, leading Van der Berg Finally, there is space to improve how the social (1997) to argue that unemployment is the “major contingency assistance system can address major covariant shocks against which the social security system provides no proper such as the economic consequences of the COVID-19 protection”. The unemployment insurance system pandemic. South Africa is typically not severely affected by caters only to those who have been employed in the shocks the way that many other countries tend to experience formal sector; considering the high levels of long- weather-related cyclical shocks, for example. But the COVID-19 term unemployment and that close to one-fifth of the crisis and national lockdown significantly lowered the country employed are located in the informal sector, this leaves into the deepest economic, unemployment, and poverty crisis the vast majority of the unemployed without any direct seen in a long time. Parts of the social protection system access to government support. could effectively be scaled up—social grants quickly increased the benefit levels and payments from the UIF 5.2.4. Value for money – expenditure efficiency could be channelled on to furloughed or laid-off formal sector workers. However, the crisis exposed that other A fourth area for improvement is the value for money, parts of the system were not ready to respond quickly to spending efficiency, and future fiscal sustainability of the crisis. There was no effective way of identifying new shock- the current social assistance system. From the perspective affected people to provide them with support—cash grants or of value for money, estimates of the benefit-cost ratio for food parcels. Yet, the current crisis may offer the opportunity social assistance in South Africa reveal that, while the country to undertake the needed reforms and bring the system of performs around ten percent better than the average for Sub- social assistance to “the next level”, towards the design and Saharan African countries and is on par with upper-middle implementation features of countries of more advanced level income countries overall, its performance is almost one-fifth of development. weaker than the average for countries in Latin America and the Caribbean. Given South Africa’s strong performance In the next chapter, chapter 6, we discuss possible in terms of the poverty-reducing impact of social reform options to address the above noted areas assistance, the value-for-money performance is lower for improvement in the context of both the political than expected and suggests that the issue lies with the environment related to efficiency reforms and the system’s cost which is high compared to other countries. continuously tightening fiscal space in the aftermath of the COVID-19 crisis. The overall challenge is that further improvements in the social assistance system may have to be implemented First, however, we take a wider look at the fit of the social in a “zero” additional budget context, given the dire fiscal assistance system for addressing South Africa’s broader situation in the country, documented in chapters 2 and 3. As development challenges. While recognising that it is not already noted, social assistance consumes a significant amount the role and mandate of social protection or social assistance of resources. Between health, education, and social protection, to address the underlying social and structural challenges in roughly half of consolidated government spending is a society, but rather to protect its vulnerable groups from the accounted for. At the same time, spending on social protection consequences, we do ask the question “to what extent is South increased by 3.7 percent per annum in real terms during the Africa’s social assistance system geared to mitigate and reduce 2010s, which is somewhat more rapid than the rate of growth the structural causes of poverty and inequality and to improve of total spending (3.3 percent). Total spending on social grants, economic inclusion, upward mobility, and human capital excluding administration costs, increased by 3.2 percent investment of the poorest”. per annum on average in real terms between 2008/09 and 2018/19. While the level and pace of spending growth is 5.3. Fit of the System vs. South Africa’s not problematic on its own, the country’s fiscus has been Development Challenges under significant strain for some time. This is the result of a decade of slow growth, diminished state capacity and other In the National Development Plan (NDP), the National effects of state capture, and an inability to rein in spending, and Planning Commission (2011) identifies the eradication will be exacerbated by the COVID-19 pandemic. Thus, while of poverty and the reduction of inequality as two there are not particularly pressing concerns regarding the broad policy objectives, both of which are identified long-term financial sustainability of the social assistance as deeply intertwined with the country’s shortage of system, it seems clear that government’s ability to formal employment. Consequently, poverty, inequality, and further expand the same system will be constrained for unemployment have come to be seen as South Africa’s “triple the foreseeable future. challenge” and feature prominently in the contextualisation of the NDP’s chapter on social protection. 63 As was discussed in chapter 2, South Africa has numerous provision is made through the care dependency grant and the development challenges that are the outcome of policy- grant-in-aid where disability amongst children and the elderly induced distortions and chronic exclusion of the majority requires full-time caregivers. of the population under apartheid. These challenges are manifested in diverse, yet often intersecting arenas, such As already highlighted, the system of social grants as the labour market, human development, the quality of has played a critical role as a relief against the human capital, poverty, and inequality. At their most basic, structural problems of poverty, inequality, and chronic however, unemployment, poverty and inequality are at the unemployment. One can even go on as to say that without core of the country’s various development challenges. Aside this system of social grants it would have been much more from the extent to which these problems reinforce each other, difficult to maintain social peace in the post-apartheid era. addressing them is made more complicated by the overlapping disadvantages associated with race, gender, age, and location, However, in terms of the design of the social grants, amongst others. there appears to be no overt consideration of or attempt of designing the system of social grants to address the While South Africa’s social assistance system was largely chronic exclusion of the majority of citizens, which is at developed under minority rule to cater primarily to the the root of South Africa socio-economic challenges. This needs of the country’s White population, it has been may be a missed opportunity given the poor Human Capital expanded during the democratic era and consequently Index in South Africa—where a child born today is only 43 plays a pivotal role in addressing chronic and deep percent as productive when she grows up as she could be if deprivation. Of the country’s various development challenges, she enjoyed complete education and full health (World Bank it is poverty which is the system’s primary focus. 2020d). Indeed, the emphasis is very much on the amelioration of deprivation—as illustrated by the DSD’s and SASSA’s As it is currently designed, the social assistance system stated objectives and mandates, which mention poverty is geared towards key lifecycle risks. Thus, virtually all and vulnerability, but not inequality—so that the impact on programmes are targeted to children and the elderly, inequality of the South Africa system of social assistance is who are at greatest risk of poverty due to their inability to almost incidental. For example, as previously noted, South participate in the labour force. Of the suite of seven grants Africa does not make use of conditional cash transfers, which (excluding the COVID-19 grant and social relief of distress), three can be used to encourage specific behaviours such as increased are only accessible to the elderly, while three are only accessible investment in health and education, a policy choice that to children. Given that the system is very successful in targeting aligns to government’s rights-based approach. Improving the the poorest in South African society (as shown in chapter 4), integration of the social protection system, through the NISPIS it is not surprising that it performs well in terms of alleviating project for example, into a broader response to the underlying poverty. Both the child support grant and the older causes of socio-economic inequality—lack of opportunity, persons’ grant make substantial contributions towards unequal access to and level of human capital, unemployment, reducing extreme poverty amongst their respective and economic exclusion—would allow for the development of target populations, as well as their households. a package of services available to individuals and households, especially for poor children, based on their particular situations. The HIV/Aids pandemic has presented significant additional risks to children, particularly through the This is not to say that social grants do not have broader potential for orphan hood. Here, the foster child grant is impacts that may address key development challenges. available to address this risk. Further, the stipulation that child As noted above, there is a growing literature that points to support grants can be paid to the child’s primary caregiver, broadly beneficial impacts of social grants—either a specific rather than necessarily the parent, introduces additional grant, or grants generally—on a wide variety of outcomes flexibility into the system in the context of the many different in the areas of poverty and inequality, nutrition and food patterns of household formation that exist in South Africa. Thus, security, education, health, labour supply and livelihoods, and this stipulation responds to the South African context in terms fertility. This body of research points to the ways in which of varying household structures and facilitates the ability of the social grants have enabled poor households to invest in system to automatically respond to changing circumstances. and build their human capital through improvements in educational attainment, nutrition and health. The The final risk that is addressed by the social assistance evidence also suggests that, while negative impacts on labour system is that of disability. Amongst working-age adults, supply may be observed, these may be explained by changes disability may compromise their ability to be and remain in household structure and by their location. This research also gainfully employed; amongst children and the elderly, disability highlights the importance of having regular household surveys may require households to devote additional scarce resources to that collect sufficient data to explore these cross-cutting issues. their care. In both instances, disability poses a risk to individuals Yet, in the case of South Africa, these “positive spill-overs” of the and households that they may fall into poverty. Thus, South system of social grants on human capital or economic inclusion Africa’s social assistance system addresses disability amongst are almost “incidental”, in the sense that they are not the result the working-age population through the disability grant, while of explicit design features or policies. More and better results SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 64 may be possible with more explicit program features to support to ECD centres is not yet widespread and where households households to invest in human capital, for instance when it and caregivers play a large role in young children’s cognitive comes to early childhood development (ECD) where access development. 65 6. Conclusions and forward look – reform options South Africa’s social assistance system represents a major 6.1. Feasibility of broader reforms intervention by government in addressing deprivation amongst the country’s population. The system is It must be recognised that the South African government extensive both in terms of the number of people it faces severe fiscal constraints that are likely to impact covers and in terms of the amount of scarce resources on the flexibility of policy to address the country’s it consumes. According to the LCS 2014/15 data, nearly two- challenges in the post-COVID environment. While revenue thirds of the population (64.0 percent) are covered, directly shortfalls, rising expenditure, and rapidly growing public debt or indirectly, by the system, while data from National Treasury are problems that have longer term roots, they have been indicate that social assistance grants in 2020 cost R156.0 billion exacerbated by the impact of the lockdown and the cost of in March 2020 prices. Put differently, it is estimated that social interventions that the COVID-19 pandemic and the lockdown assistance transfers are equivalent to 7.3 percent of households’ itself have necessitated. Given the government’s stated expenditure nationally, and around 60 percent of the total commitment to rein in public spending in order to stabilise expenditure for the poorest quintile. This is one way in which public debt (National Treasury, 2020c), government ministries South African society demonstrates, through government, the have been required to cut spending. value placed on providing support to its poorest and most vulnerable members. We would argue, as is evident by the analysis in the preceding chapters which points to the importance of the social assistance In South Africa the social assistance system has been used programs in preventing further poverty and inequality, that to “substitute” and mitigate the absence or weakness or enforcing such cuts on social assistance would have lack of inclusivity of other mechanisms such as contributory significant negative impacts across a wide range of pensions utilised in other countries to respond to the same risks potential outcomes and that the cost would be borne (risk of poverty in old age, for example, or risk of disability), and by those households who are least able to weather such to provide relief against the structural problems of poverty, shocks, undermining the system’s objectives of preventing and inequality, and chronic unemployment discussed in chapter 2. alleviating poverty in both the short- and long-term. Moreover, we think that eroding the support for the poorest groups of the There is always room for improvement in a social populations, those historically excluded from participating in assistance system as long as poverty remains a problem the economic growth is also not politically feasible. The South in society. Sometimes such improvements derive from African social assistance system is a pillar of the social contract adjusting the rules of existing programmes or in the form between the state and the people and enjoys strong political of higher benefit levels or alterations to improve benefit support. incidence. In other instances, improvements in the efficiency of the system are required. And sometimes yet, new programmes On the other hand, any major scale up of the existing are required to reach previously uncovered populations. grant system in the current fiscal situation seems difficult and would require further investigation of alternatives. A number of adjustments are suggested over the next Savings and resources may have to be found in efficiency five years to better align the social protection system, improvements and in further sharpening the pro-poor focus especially social assistance, to more effectively mitigate of some of the grants. The dilemma of the future of the South the structural causes of socio-economic inequality, Africa’s social assistance system rests in the opposing pull of improve its cost- and administrative-effectiveness, these two forces: the limited political appetite for cost-saving and the ability to protect working-age informal sector reforms and the need to consolidate expenditures to avoid workers. This chapter lays out some simulations to serve as further deepening of the macro-economic crisis and debt examples of possible program-level effectiveness gains.13 It also burden. Feasible options for broader reform hence need reflects on the feasibility of broad reforms and proposes some to balance political will and the need to contain costs. shorter-term options which policy makers could consider for the future of social assistance in South Africa. 6.2. Shorter-term reform options 6.2.1. Addressing cost-effectiveness and value for money As discussed above, the system of categorical grants could benefit from introducing more flexibilities. South Africa’s expansive social assistance system with one grant for each vulnerable group is in no doubt expensive at 3.31 percent 13 For both simulations, it is important to remember that the simulations are static. They do not account for changes in household formation that may result from a changed distribution of resources. Further, they do not account for changes in individual behaviour in response to changed incentives. It is also assumed that there is full income sharing within households. In other words, any additional income (or loss of income) is shared equally amongst household members. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 66 of GDP and may not provide the tools for policy makers to poorer households in lagging regions. Moreover, in Rwanda, adjust parameters over time to contain costs while maximizing Ethiopia, and Madagascar mothers and caregivers are being outcomes. However, there is no clear evidence to suggest that trained to manage home-based and community ECD centres. broader reform to the set of programs (e.g. consolidation of This modality has helped in highlighting the role of cash programs, introducing conditional cash transfers) would lead grants beyond income support/amelioration of deprivation, to to drastic efficiency gains, stronger impacts or reduced costs, also addressing other stubborn human capital development or even gain any political traction. But it may be useful to challenges such as stunting and low cognitive development speculate as to how the overall package of programs during early years, especially among the poorer families. It also could transition to be more productive and outcome- allowed governments to start making other adjustments in the focused over time. Some more technical reform options that overall safety net programs to link up with diverse development should be considered are discussed below. objectives. 6.2.2. Strengthening outcomes and incentivising Moreover, supporting households to engage more productive inclusion and economic mobility actively in job-searching, training, and develop small productive activities has shown positive impacts around At its core the social assistance system is focussed on the globe. This may be an option especially for households providing relief and income, rather than on attempting which are not labour constrained. With the Presidency to systematically address structural development developing the new Pathway Manager Network as a digital challenges and underlying causes of poverty and platform for job-seekers to get coaching, job-readiness training, inequality either alone or in concert with other profiling and matching, there is an opportunity to link working- government interventions. For example, South Africa age adult social grant recipients/beneficiaries (for households does not make use of conditional cash transfers to receiving the child support grant, the COVID-19 special SDR attempt to change behaviour, since conditionalities do grant, or who participate in public employment programs). not align well with South Africa’s rights-based approach and may lead, for example, to children being punished twice 6.2.3. Providing coverage for the working-age adults where caregivers are non-compliant. There are however potential ways to improve coordination of social service Since unemployment insurance is only available to provision for poor children, such as through case management those who were previously formally employed, this systems and usage of an integrated registries (e.g. NISPIS) and excludes everyone employed in the informal sector, as management systems which can track services provided to well as individuals informally employed within formal children as well as the development and achievement of the sector firms. While these are not particularly problematic children benefitting from social services. features of unemployment insurance in developed economies, South Africa’s dire unemployment problem—a combination One way to improve expenditure efficiency is to strengthen of high unemployment rates and the prevalence of long outcomes without significantly increasing costs. We ask the unemployment spells—means that relatively few are able question: would it be possible to enhance the untapped to access unemployment benefits. Indeed, these same potential of social grants to address other long-standing characteristics of unemployment in South Africa pose an development challenges by slightly adjusting programme immense challenge from the perspective of cost and financial designs and implementation arrangements? This may sustainability to extending social assistance coverage to the involve building in more explicit support services and working-age segment of the population. family sessions for households with children to break the intergenerational cycle of poverty by investing more If the gap in the system is to be closed, it would require strongly in the development of the children. Social service either significant scaling up of existing public works programme coordination across government departments interventions, which is currently being started through and case management would be important (see discussion on the Employment Stimulus or the introduction of a new integrated service databases below). type of grant, or both. The cost of implementing such a new grant would be considerable and would require careful analysis Access to primary education in South Africa is almost and investigation of alternatives. Box 7 presents a simulation universal but the quality of schooling is low compared which builds on the COVID-19 grant and can be seen as an to comparable countries and learning outcomes are application example of a basic income guarantee. Specifically, below UMIC averages (see presentation of HCI results in this exercise simulates a ‘jobseekers’ grant’, targeted at the chapter 2). However, access to early childhood centres is not unemployed. The implementation of such a grant may provide yet available to a large number of young children and may be opportunities to link jobseekers into other programmes, such encouraged through the child support grant. Also, in several as the Department of Employment and Labour’s Employment countries including in the Sahel, in Rwanda, Jamaica, and Services of South Africa (ESSA), which may constitute a package Madagascar, providing a package of accompanying measures of support for the unemployed. Indeed, such integration with (family sessions on better parenting, childcare, nutrition, and other services may be important in terms of ensuring that only children’s cognitive stimulation) together with the grant have active jobseekers are eligible for the grant, thereby containing shown positive impacts on social outcomes. Especially for costs to some extent. 67 Box 7: A Jobseekers’ Grant The simulation presented here provides an example of a jobseekers’ grant which aims to give support to unemployed working- age adults who would otherwise have no direct means of support. However, instead of this grant being simply about income support for the unemployed, it could be integrated within a broader set of interventions that are explicitly focussed on getting the unemployed into work. Grant beneficiaries would be able to benefit from services such as job matching, career counselling, job- readiness training, training on life skills, and accessing apprenticeships, amongst other services. Thus, a well-designed jobseekers’ grant would have as its primary objective to link the unemployed back into gainful employment and strengthening their ability to remain employed. For the simulation, the value of the grant is set at R350 per month in April 2019 prices similar to the special COVID-19 grant. Given that the idea is that the grant be targeted at those actively seeking employment, eligibility is restricted to the narrow unemployed over the age of 18 and under the age of 60. This need for beneficiaries to prove their eligibility (being job-seeking and within the age rage) is precisely what distinguishes this from a basic income guarantee which is commonly discussed in South African media. Narrow unemployment assumes active job search, and this ties into the idea of linking the grant to an employment service programme. Individuals receiving the disability grant or income from the UIF are also excluded. For this simulation, answers to the following questions are sought: - What is the total cost of the new grant example? - How are the benefits and beneficiaries distributed? - How does the grant impact on poverty and inequality? Simulated Cost and Distributional Impact of a Jobseekers Grant Decile Annual Cost (R bil) Benefits Beneficiaries 2017 Prices Apr 2020 Prices (%) (‘000s) (%) Decile 1 2.5 2.7 16.9 649.7 16.9 Decile 2 2.0 2.2 13.6 522.9 13.6 Decile 3 2.2 2.4 14.8 570.8 14.8 Decile 4 1.6 1.8 11.0 423.8 11.0 Decile 5 1.5 1.7 10.4 399.8 10.4 Decile 6 1.3 1.5 9.0 348.1 9.0 Decile 7 1.4 1.5 9.3 358.4 9.3 Decile 8 1.2 1.4 8.4 322.5 8.4 Decile 9 0.7 0.7 4.5 174.7 4.5 Decile 10 0.3 0.3 2.1 81.3 2.1 TOTAL 14.9 16.2 100.0 3 852.1 100.0 Source: Own calculations, Saldru (2018). Note: Deciles refer to the post-transfer income distribution. Eligibility for the jobseekers’ grant is modelled as: (1) narrow unemployed; (2) over the age of 18, but younger than 60 years; (3) receives no income from the UIF; and (4) does not receive a disability grant. For the purposes of the simulation, a 100 percent take-up rate is assumed. At R350 per month, the cost of a jobseekers’ grant is significant. Based on an estimate of almost 3.9 million beneficiaries, the cost is estimated at R14.9 billion per annum in 2017 prices or R16.2 billion in April 2020 prices. Of this R16.2 billion amount, R7.3 billion in benefits—or 45.3 percent of total benefits—accrues to the poorest 30 percent of the population. A further 39.7 percent of benefits accrues to the middle 40 percent of the population (deciles 4 through 7). Given that the value of the proposed grant does not vary across individuals, individual deciles’ shares of benefits correspond to their shares of beneficiaries. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 68 Box 7: A Jobseekers’ Grant Given the grant’s relatively small value and the relatively small number of beneficiaries—a jobseekers’ grant value would be somewhat lower than the child support grant, but it has less than one-third as many direct beneficiaries— the impact on poverty and inequality is relatively small (table below). However, as a complement to already existing social assistance grants, a jobseekers’ grant’s main objective may not be poverty reduction, rather it may be to increase those who actively seek employment and increase labour supply. Nevertheless, the proposed grant does reduce both poverty (irrespective of the poverty line used) and inequality. Using the upper-bound poverty line, the poverty rate is estimated to decline by 0.9 percentage points to 42.4 percent. This 0.9 percentage point reduction compares quite favourably with the reduction (using a different dataset) attributed to the child support grant presented in the table below. The magnitude of the reduction is similar across all three poverty measures and for both poverty lines. The Gini coefficient falls slightly from 66.3 to 65.8 with the introduction of the grant. Simulated Poverty and Inequality Impact of a Jobseekers Grant Poverty Rate Poverty Gap Poverty Gap Gini Squared Upper-bound poverty line Baseline 43.2 19.7 11.3 66.3 With proposed grant 42.4 18.7 10.4 65.8 … Change from baseline -0.9 -1.0 -0.9 -0.5 Lower-bound poverty line Baseline 29.3 11.1 5.6 66.3 With proposed grant 28.0 10.0 4.8 65.8 … Change from baseline -1.3 -1.1 -0.8 -0.5 Source: Own calculations, Saldru (2018). The simulations indicate that the cost of a jobseekers’ grant is considerable. Its R16.2 billion price tag would probably make it the fourth largest grant in terms of spending and represents approximately 9.4 percent of total spending on social grants in the 2018/19 financial year. However, the cost of such a grant may be bigger than shown in numbers, because of perverse incentives created in the labour market. Those currently not searching for jobs (the discouraged jobseekers) may change their behaviour to be identified as searching unemployed, and low-income informal workers may also decide to rather search for jobs in order to qualify, thus reducing informal activity. In contrast, the Presidency’s recently launched Employment Stimulus received a budget of R12.6 billion for Phase 1 to create 800 000 temporary employment opportunities for around six months. While the jobseekers’ grant has a beneficial impact on both poverty and inequality, it is perhaps the integration of the grant with other interventions that offers the greatest potential for impact. By linking grant recipients to the Employment Stimulus, the Department of Employment and Labour, to the SETAs and, where appropriate, to the broader education and training system, it may be possible to facilitate economic participation and support livelihoods. Public works programmes have typically been measurable positive effects on the probability of employment established to deal with unemployment of a cyclical or post-participation. frictional (i.e. short-term) nature. However, South Africa’s unemployment is structural and long-term: according to More palatable, perhaps, may be the scaling up of the data published by Statistics South Africa (2020d), 72 percent CWP, given its targeting of the poorest communities and the of the unemployed are ‘long-term unemployed’, having been potential multiplier effects from specific types of investments unemployed for more than one year. The EPWP and CWP within these communities, although here too there are represent key interventions to address poverty for working- significant fiscal implications for interventions that do not age adults, providing up to 100 days of work to working-age significantly reduce unemployment and whose longer-term participants. One of the advantages of the CWP—mentioned effects on employment have not been rigorously assessed. by Philip (2013, p.15)—over the EPWP is the sustained part- This is in fact the direction taken by the Employment Stimulus time employment that it offers, which establishes an earnings recently introduced by the Office of the President (see Box 3). floor and mimics the effect of a social grant. While it seems clear that participation in the EPWP would have some beneficial Regardless of other considerations (e.g. fiscal constraints, impact on individual- and household-level poverty during the implementation challenges) one fact remains: a better period of participation, it is not clear what the impact of the evaluation of what has worked and not worked under the programme is post-participation, or whether participation has existing EPWP and CWP programmes would have created 69 a better evidence basis for this new Employment Stimulus agencies at central and local level responsible for cash initiative. Going forward, therefore, greater emphasis needs transfers, health, education, and nutrition services. to be placed on measuring, monitoring, and evaluating the impacts of these interventions. Unfortunately, evaluations of The usefulness of this type of coordination and these programmes are scarce (see Box 5). The main monitoring information sharing across departments has surfaced and evaluation reports focus mainly on the ‘outputs’ of the in 2020 with the outbreak of the global COVID-19 programmes and opinions of participating individuals and pandemic. South Africa’s social assistance system has not organisation, but do not try to measure the actual impact of often had to deal with large-scale immediate crises, such the programme for individuals and their households, either as widespread droughts: the country is relatively urbanised, during or after their participation in the programme. Prioritising with smallholder agriculture representing a small proportion strengthening the quality and reach of public and non- of household income (especially given low participation in government employment service programmes to be able to markets), while market-oriented agriculture is dominated by a more effectively link social assistance beneficiaries to the labour relatively small number of large-scale commercial farmers. This market would be important. As noted in the beginning of this means that such disasters are addressed through mechanisms report, a review of active labour market programmes, especially other than social assistance. Thus, the COVID-19 pandemic youth employment programmes, is conducted by the World and the national lockdown implemented by government Bank separately. poses an unprecedented challenge to the ability of the social assistance system to respond rapidly and at scale. Another alternative involves reform of unemployment insurance by raising benefits or extending the period As another example, while the payment system is highly of coverage (as was recently done), or by extending digitised and large number of grants are paid out on a participation in the system to informal workers. Once timely basis and accounted for every month, beneficiaries again, irrespective of the policy choice, it should be based on still struggle to access funds queuing at retailers and clear evidence of impact in order to achieve the objectives of other pay-points month-after-month. Limitations in poverty reduction, employment creation, and ensuring that as payment withdrawal cause delays, confusion, and social many South Africans as is feasible are covered by the system. crowding at pay-points. Further research could look into alternative payment modalities which would allow recipients 6.2.4. Delivery system and program level technical to retrieve and use their social assistance payments closer reforms to where they live and in markets where thy normally shop. Improving the last mile accessibility of social grants by, for Reducing the administrative and delivery costs of instance, engaging the vast network of informal spaza shops implementing social assistance programs may reduce or other informal vendors would be important provided service expenditures. As demonstrated above in chapter 3, while quality assurance measures are in place. administration costs have been reduced over the years, cost- efficiency is low. The need to strengthen program integration, Another way to increase outcomes without significantly interoperability, and reduce overheads could help improve the raising costs is to attempt to adjust the technical value-for-money of the social assistance system. parameters of programs—such as the benefit incidence— to attempt to skew outcomes towards the group with Except for the NISPIS project, there has been limited collaboration lowest outcomes. A second simulation (Box 8) shows that and sharing of information across government departments there is room to improve programme benefit incidence toward in a way that might create synergies and amplify impact. The the lower quintiles especially for the child support grant. This DSD and SASSA do not normally coordinate with relevant reform would strengthen impact on poverty without increasing departments—such as the Department of Employment and cost. The simulation involves a refinement of the child support Labour, the Department of Basic Education, or the Department grant that introduces a rough sliding scale, providing slightly of Health—to create a package of services or interventions larger benefits to children in the poorest households and to address broader challenges related to the labour market slightly lower benefits to children in better-off households. The or human capital development. An interoperable social analysis in chapter 4 above shows that the amount of the child registry as envisioned for NISPIS could serve the basis grant is relatively low, and impact could be improved if the for more integrated service delivery, especially related grant was increased, especially for the poorest. The idea with to investments in human capital of vulnerable children this simulation is to improve the targeting of benefits to those and their school-to-work transitions. However, the issue children (and their households) who most need the additional is not just one of dataset interoperability, but rather one income. Further, the proposal acknowledges the fact that of a different and better governance systems and more children in the upper part of the income distribution require effective inclusive leadership. Institutions ought to be less support to escape poverty. reorganised so that they are functionally better aligned to the development challenges they are confronting. For Unlike the jobseekers’ grant discussed above, a example, making a dent on early childhood development differentiated child support grant would in principle not would require much better coordination among several increase costs, though administrative costs may rise. Also, SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 70 such differentiation requires a fine-tuning of the means test that may be beyond the capacity of the administrative system. As a substantial part of the recipients of the child support grant are in the informal sector, where incomes are difficult to monitor and wages change frequently, it would be extremely difficult to maintain such a system. Box 8: A Differentiated Child Support Grant The objective of this simulated adjustment of the child support grant is to boost the benefit levels for children in the poorest households and to lower the levels for those in better-off households. This has a precedent in the sliding scale of the older persons’ grant, which sees the grant value reduced depending on the level of recipients’ income from other sources. While this proposal would allow for greater support to be directed to those most in need, it does come with an additional administrative burden. Certainly, it would seem that regular means-testing would be required to ensure that beneficiaries are awarded the correct level of benefit, particularly given the fact that caregivers may only have an incentive to report income reductions as opposed to rises in incomes. For the simulation, the design is kept very simple. For children in the poorest 20 percent of the population, the value of the child support grant is increased by 25 percent; for those in the richest 50 percent of the population, the value of the grant is reduced by 30 percent; however, children in deciles 3 through 5 see no change to the grant value. These numbers are somewhat arbitrary but are informed by the fact that child support grant recipients are concentrated amongst the poorest deciles and by the desire to keep the overall budget envelope unchanged (or to at least keep the budget at a similar level). This means that this simulation represents a redistribution of spending across deciles but within the current budget envelope. The simulation uses the Living Conditions Survey 2014/15 data (Statistics South Africa, 2015a). For this simulation, answers to the following questions are sought: - What is the total cost of the proposal relative to the original cost of the grant? - How does the distribution of benefits change and how does it compare to the distribution of beneficiaries? - How does the impact on poverty and inequality change? The table below details the cost and distributional impact of the differentiated child support grant. The baseline cost of the grant— i.e. the total value of child support grant income observed in the survey data—is R41.7 billion for the year in April 2015 prices. As noted, the intention with the simulation is to keep the overall cost very similar and this has been achieved with the proposed grant costing R41.9 billion for the year. Simulated Cost and Distributional Impact of a Differentiated Child Support Grant Decile Annual Cost (R bn, Apr 2015 prices) Share of Benefits (%) Beneficiaries Baseline Proposed Change Baseline Proposed Change (% share) Decile 1 7.2 9.0 1.8 17.3 21.6 4.2 17.9 Decile 2 6.4 8.0 1.6 15.4 19.2 3.8 16.2 Decile 3 6.2 6.2 0.0 14.8 14.7 -0.1 15.2 Decile 4 5.7 5.7 0.0 13.6 13.5 -0.1 13.5 Decile 5 5.4 5.4 0.0 13.0 13.0 -0.1 12.9 Decile 6 4.5 3.1 -1.3 10.7 7.5 -3.3 10.5 Decile 7 3.6 2.5 -1.1 8.6 6.0 -2.6 7.9 Decile 8 1.9 1.3 -0.6 4.4 3.1 -1.3 4.2 Decile 9 0.7 0.5 -0.2 1.7 1.2 -0.5 1.4 Decile 10 0.2 0.1 -0.1 0.4 0.3 -0.1 0.3 TOTAL 41.7 41.9 0.2 100.0 100.0 0.0 100.0 Source: Own calculations, Statistics South Africa (2015a). Note: Deciles refer to the pre-transfer distribution of the population. 71 Box 8: A Differentiated Child Support Grant It is clear that the differentiated grant alters the distribution of benefits quite significantly. In total, the poorest 20 percent of the population receive R17.0 billion for the year under the simulation, compared to 13.6 billion under the prevailing structure, an increase of R3.4 billion. In contrast, the top half of the distribution sees the annual value of transfers received in terms of the child support grant fall from R10.8 billion to R7.6 billion. In relative terms, the poorest 20 percent of the population account for 34.0 percent of beneficiaries and see their share of benefits rise from 32.7 percent of the total to 40.8 percent under the proposed grant. Deciles 3 through 5 maintain their share of benefits under the change, which are roughly equivalent to their share of beneficiaries. Consequently, the top half of the distribution sees its share of benefits decline from 25.9 percent to 18.1 percent, while it accounts for 24.4 percent of beneficiaries. The table below provides a sense of the impact of the simulated change on poverty and inequality. The impact on poverty is assessed in terms of both the upper- and lower-bound poverty lines, and across three conventional poverty measures. By allocating larger benefits to the poorest grant recipients, the revised child grant has a stronger poverty-reducing impact than the child support grant in its current format. Further, this impact is observed across all three poverty measures. Simulated Poverty and Inequality Impact of a Differentiated Child Support Grant Poverty Rate Poverty Gap Poverty Gap Gini Squared Upper-bound poverty line Baseline (excluding CSG) 48.2 27.4 19.3 69.3 With conventional CSG 46.6 23.3 14.6 67.4 … Change from baseline -1.6 -4.1 -4.7 -1.9 With proposed CSG 45.9 22.4 13.6 67.2 … Change from baseline -2.3 -5.0 -5.7 -2.1 Lower-bound poverty line Baseline (excluding CSG) 36.4 19.4 13.3 69.3 With conventional CSG 32.8 14.4 8.2 67.4 … Change from baseline -3.6 -5.0 -5.1 -1.9 With proposed CSG 32.1 13.3 7.2 67.2 … Change from baseline -4.3 -6.0 -6.0 -2.1 Source: Own calculations, Statistics South Africa (2015a). Using the upper-bound poverty line, it is estimated that the proposed grant would lower the poverty rate by 2.3 percentage points compared to the baseline (post-transfer income but excluding child support grant income). This is almost one and a half times the effect of the standard child support grant (1.6 percentage points). The difference in impacts is smaller for the poverty gap and poverty gap squared, but in both instances the proposed grant is more effective in reducing poverty. The simulated grant also has a slightly stronger inequality-reducing effect, lowering the Gini coefficient by 2.1 points from 69.3 to 67.2. This simulation suggests that there is scope for fine-tuning the child support grant to achieve a stronger impact in terms of poverty reduction. By providing different levels of support according to need, more resources are directed to the poorest households, which is particularly beneficial in reducing the poverty gap and poverty gap squared, measures that are more sensitive to particularly poor individuals. One potential way around the administrative burden associated with updating beneficiaries’ information for regular means-testing may be to use other administrative data, such as the child’s school quintile. This would, though, be dependent on the properly integrated social registry that has been highlighted already. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 72 7. Bibliography AGÜERO, J., CARTER, M.R., and MAY, J., 2007, ‘Poverty and Research Unit, University of Cape Town, Retrieved from http:// inequality in the first decade of South Africa’s democracy: What www.dpru.uct.ac.za. can be learnt from panel data from KwaZulu-Natal?’ Journal of African Economies, 16(5): 782–812. CARON, L. and TIONGSON, E., 2020, ‘COVID-19 and the Labour Market in South Africa: Impact on Poverty and Inequality.’ ALTMAN, M., MOKOMANE, Z., and WRIGHT, G., 2014, Unpublished paper prepared for the World Bank. Revised 16 ‘Social security for young people amidst high poverty and May 2020. unemployment: Some policy options for South Africa.’ Development Southern Africa, 31(2): 347–362, doi:10.1080/037 CASE, A., 2001, ‘Does Money Protect Health Status? Evidence 6835X.2013.873346. from South African Pensions.’ Working Paper 8495, National Bureau of Economic Research, doi:10.3386/w8495, Retrieved ARDINGTON, C., CASE, A., and HOSEGOOD, V., 2009, ‘Labour from http://www.nber.org/papers/w8495. supply responses to large social transfers: Longitudinal evidence from South Africa.’ American Economic Journal: CASE, A., HOSEGOOD, V., and LUND, F., 2005, ‘The reach and Applied Economics, 1(1): 22–48. impact of Child Support Grants: evidence from KwaZulu- Natal.’ Development Southern Africa, 22(5): 467–482, ARMSTRONG, P. and BURGER, C., 2009, ‘Poverty, Inequality and doi:10.1080/03768350500322925. the Role of Social Grants: An Analysis using Decomposition Techniques.’ Stellenbosch Economic Working Papers No. 15/09/, CASE, A. and MENENDEZ, A., 2007, ‘Does money empower the Retrieved from http://www.ekon.sun.ac.za/wpapers/2009/ elderly? Evidence from the Agincourt demographic surveillance wp152009/. site, South Africa.’ Scandinavian Journal of Public Health, 35(Suppl 69): 157–164. AUDITOR-GENERAL, 2020. ‘First Special Report on the Financial Management of Government’s Covid-19 Initiatives’. Republic of CHAKONA, G. and SHACKLETON, C.M., 2019, ‘Food insecurity in South Africa. Retrieved from: http://www.agsa.co.za. South Africa: To what extent can social grants and consumption of wild foods eradicate hunger?’ World Development BARRIENTOS, A., 2003, ‘What is the impact of non-contributory Perspectives, 13(C): 87–94. pensions on poverty? Estimates from Brazil and South Africa.’ CPRC Working Paper No. 33. Chronic Poverty Research Centre. COETZEE, M., 2013, ‘Finding the Benefits: Estimating the impact of the South African child support grant.’ South African Journal BASSIER, I., BUDLENDER, J., ZIZZAMIA, R., LEIBBRANDT, M., and of Economics, 81(3): 427–450. RANCHHOD, V., 2020, ‘Locked down and locked out: Repurposing social assistance as emergency relief to informal workers.’ COUNCIL FOR MEDICAL SCHEMES, 2017, ‘Quarterly Reports for SALDRU Working Paper Number 261, version 2. Southern Africa the Period ending 30 September 2016.’ Retrieved from https:// Labour and Development Research Unit, University of Cape www.medicalschemes.com/Publications.aspx. Town, Retrieved from http://www.saldru.uct.ac.za. , 2020, ‘Quarterly Reports for the Period ending 30 BERTRAND, M., MULLAINATHAN, S., and MILLER, D., 2003, September 2019.’ Retrieved from https://www.medicalschemes. ‘Public Policy and Extended Families: Evidence from Pensions com/Publications.aspx. in South Africa.’ The World Bank Economic Review, 17(1): 27–50, doi:10.1093/wber/lhg014. DELANY, A., ISMAIL, Z., GRAHAM, L., and RAMKISSOON, Y., 2008, Review of the child support grant: Uses, implementation and BEUKES, R., JANSEN, A., MOSES, M., and YU, D., 2016, ‘Exploring obstacles. United Nations Children’s Fund, Johannesburg, South the Eligibility Criteria of the Child Support Grant and its Impact Africa. on Poverty.’ Social Indicators Research, 134(2): 511–529, doi:10.1007/s11205-016-1433-z. DEPARTMENT OF BASIC EDUCATION, 2015, National School Nutrition Programme (NSNP) 2013/14 Annual Report. BHORAT, H., KÖHLER, T., OOSTHUIZEN, M., STANWIX, B., Department of Basic Education, Republic of South Africa, STEENKAMP, F., and THORNTON, A., 2020, ‘The Economics of Pretoria, Retrieved from http://www.education.gov.za. Covid-19 in South Africa: Early Impressions.’ DPRU Working Paper No. 202004. Development Policy Research Unit, University of , 2019, Annual Report 2018/2019. Department of Basic Cape Town, Retrieved from http://www.dpru.uct.ac.za. Education, Republic of South Africa, Pretoria, Retrieved from http://www.education.gov.za. BHORAT, H., LILENSTEIN, K., OOSTHUIZEN, M., and THORNTON, A., 2016, ‘Vulnerability in Employment: Evidence from South , 2020, Department of Basic Education Website. Africa.’ DPRU Working Paper No. 201604. Development Policy Retrieved from http://www.education.gov.za. 73 DEPARTMENT OF COOPERATIVE GOVERNANCE, 2011, , 2018a, Annual Report 2017/18. Department of Public Communities at Work: Community Work Programme Works and Infrastructure, Republic of South Africa, Pretoria, 2010/2011. Department of Cooperative Governance, Pretoria, Retrieved from http://www.publicworks.gov.za. South Africa. , 2018b, Expanded Public Works Programme (EPWP) , 2017, Annual Report 2016-2017. Department of Quarterly Report: Report for the period 1 April - 31 December Cooperative Governance, Pretoria, Retrieved from http://www. financial year 2017/18. Department of Public Works and cogta.gov.za. Infrastructure, Republic of South Africa, Pretoria, Retrieved from http://www.epwp.gov.za. , 2019, 2018/19 Annual Report. Department of Cooperative Governance, Pretoria, Retrieved from http://www. , 2019a, Expanded Public Works Programme (EPWP) cogta.gov.za. Quarterly Report: Report for the period 1 April - 31 December financial year 2018/19. Department of Public Works and DEPARTMENT OF LABOUR, 2013, Annual Report of the Infrastructure, Republic of South Africa, Pretoria, Retrieved from Compensation Fund for the year ended 31 March 2013. http://www.epwp.gov.za. Department of Labour, Republic of South Africa, Pretoria, Retrieved from http://www.labour.gov.za. , 2019b, Expanded Public Works Programme Phase 3 – Impact Evaluation Study. http://www.epwp.gov.za. , 2016a, Annual Report of the Compensation Fund 2015/2016. Department of Labour, Republic of South Africa, , 2020, Expanded Public Works Programme (EPWP) Pretoria. Quarterly Report: Report for the period 1 April - 31 December financial year 2019/20. Department of Public Works and , 2016b, Unemployment Insurance Fund Annual Infrastructure, Republic of South Africa, Pretoria. Report 2015/16. Department of Labour, Republic of South Africa, Pretoria. , n.d.a, ‘Department of Public Works and Infrastructure website.’ Retrieved from http://www.publicworks.gov.za. , 2019a, Annual Report: Department of Labour 2018/19. Department of Labour, Republic of South Africa, , n.d.b, ‘Expanded Public Works Programme website.’ Pretoria, Retrieved from http://www.labour.gov.za. Retrieved from http://www.epwp.gov.za. , 2019b, Unemployment Insurance Fund Annual DEPARTMENT OF SOCIAL DEVELOPMENT, n.d., ‘Department of Report 2018/19. Department of Labour, Republic of South Social Development website.’ Retrieved from http://www.dsd. Africa, Pretoria, Retrieved from http://www.labour.gov.za. gov.za. DEPARTMENT OF PUBLIC WORKS AND INFRASTRUCTURE, , 2015. Social Development Strategic Plan 2015- 2014, Expanded Public Works Programme (EPWP) Quarterly 2020. Republic of South Africa, Pretoria. Report: Report for the period 1 April - 31 December financial year 2013/14. Department of Public Works and Infrastructure, , 2019. Annual Performance Plan 2019/20. Republic of South Africa, Pretoria, Retrieved from http://www. Republic of South Africa, Pretoria. Retrieved from http:// epwp.gov.za. www.dsd.gov.za. , 2015, Expanded Public Works Programme (EPWP) DUFLO, E., 2000, ‘Child health and household resources in South Quarterly Report: Report for the period 1 April - 31 December Africa: Evidence from the old age pension program.’ American financial year 2014/15. Department of Public Works and Economic Review, 90(2): 393–398. Infrastructure, Republic of South Africa, Pretoria, Retrieved from http://www.epwp.gov.za. , 2003, ‘Grandmother and granddaughters: Old-age pensions and intrahousehold allocation in South Africa.’ The , 2016, Expanded Public Works Programme (EPWP) World Bank Economic Review, 17(1): 1–25. Quarterly Report: Report for the period 1 April - 31 December financial year 2015/16. Department of Public Works and ECONOMIC POLICY RESEARCH INSTITUTE (EPRI), 2015. Infrastructure, Republic of South Africa, Pretoria, Retrieved from Implementation Evaluation of EPWP in the Social Sector: Phase http://www.epwp.gov.za. Two (2009/10-2013/14). Department of Social Development and Department of Planning, Monitoring and Evaluation, , 2017, Expanded Public Works Programme (EPWP) Republic of South Africa. Retrieved from https://evaluations. Quarterly Report: Report for the period 1 April - 31 December dpme.gov.za/evaluations/508 financial year 2016/17. Department of Public Works and Infrastructure, Republic of South Africa, Pretoria, Retrieved from EYAL, K. and BURNS, J., 2018, ‘The parent trap: Cash transfers http://www.epwp.gov.za. and the intergenerational transmission of depression in South Africa.’ SALDRU Working Paper Number 165, version 2. Southern SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 74 Africa Labour and Development Research Unit, University of , 2019. ‘2018/2019 Annual Report’. Government Cape Town, Retrieved from http://www.opensaldru.uct.ac.za/ Employees Pension Fund, Republic of South Africa. Retrieved handle/11090/814. from https://www.gepf.gov.za/annual-reports/. EYAL, K. and KESWELL, M., 2008, ‘Identifying Pure-Income , 2020. ‘Government Employees Pension Fund Website’. Effects in an Empirical Model of Labour Supply: the case of the Retrieved from http://www.gepf.gov.za/. South African Social Pension.’ SALDRU Working Paper Number 19. Southern Africa Labour and Development Research GOVERNMENT OF SOUTH AFRICA, 2020. ‘Building a Society That Unit, University of Cape Town, Retrieved from http://www. Works: Public Investment in a Mass Employment Strategy to opensaldru.uct.ac.za/handle/11090/863. Build a New Economy’. EYAL, K. and WOOLARD, I., 2011, ‘Throwing the Book at the CSG.’ GROUNDUP, 2017, ‘Everything you need to know about social SALDRU Working Paper Number 53. Southern Africa Labour and grants.’ GroundUp Website, as updated, Retrieved from https:// Development Research Unit, University of Cape Town, Retrieved www.groundup.org.za/article/everything-you-need-know- from http://www.opensaldru.uct.ac.za/handle/11090/64. about-social-grants_820/. FAO, 2015. The State of Food and Agriculture: Social Protection HUNDENBORN, J., LEIBBRANDT, M., and WOOLARD, I., 2016, and Agriculture: Breaking the cycle of rural poverty. Food and ‘Drivers of Inequality in South Africa.’ SALDRU Working Paper Agriculture Organization of the United Nations, Rome. Number 194. Cape Town: SALDRU, University of Cape Town, Retrieved from http://opensaldru.uct.ac.za/handle/11090/853. FINANCIAL SERVICES BOARD, 2011, ‘Registrar of Pension Funds Annual Report 2009.’ Retrieved from http://www.fsca.co.za. INCHAUSTE, G., LUSTIG, N., MABOSHE, M., PURFILED, C., and WOOLARD, I., 2015, ‘The Distributional Impact of Fiscal Policy , 2012, ‘Registrar of Pension Funds Annual Report 2010.’ in South Africa.’ Policy Research Working Paper No. 7194. Retrieved from http://www.fsca.co.za. Washington, DC: World Bank Group, Retrieved from http:// documents.worldbank.org/curated/en/606161524230637651/ , 2013a, ‘Registrar of Pension Funds Annual Report The-distributional-impact-of-fiscal-policy-in-South-Africa. 2011.’ Retrieved from http://www.fsca.co.za. INTERNATIONAL LABOUR ORGANISATION, 2020, ‘ILO Estimates , 2013b, ‘Registrar of Pension Funds Annual Report and Projections.’ ILO Data Explorer. ILOSTAT Online database, 2012.’ Retrieved from http://www.fsca.co.za. November 2019, Retrieved from https://ilostat.ilo.org/. , 2014, ‘Registrar of Pension Funds Annual Report 2013.’ INTERNATIONAL MONETARY FUND, 2020. World Economic Retrieved from http://www.fsca.co.za. Outlook, October, 2020. , 2015, ‘Registrar of Pension Funds Annual Report 2014.’ JAIN, R., BUDLENDER, J., ZIZZAMIA, R., & BASSIER, I., 2020. ‘The Retrieved from http://www.fsca.co.za. labour market and poverty impacts of Covid-19 in South Africa’. NIDS-CRAM Wave 1 Working Paper No 5. National Income , 2016, ‘Registrar of Pension Funds Annual Report 2015.’ Dynamics Study (NIDS) – Coronavirus Rapid Mobile Survey Retrieved from http://www.fsca.co.za. (CRAM), July. Retrieved from: https://cramsurvey.org/reports/ , 2017, ‘Registrar of Pension Funds Annual Report 2016.’ JENSEN, R.T., 2004, ‘Do private transfers “displace” the benefits of Retrieved from http://www.fsca.co.za. public transfers? Evidence from South Africa.’ Journal of Public Economics, 88(1-2): 89–112, doi:10.1016/s0047-2727(02)00085- , 2018, ‘Registrar of Pension Funds Annual Report 2017.’ 3. Retrieved from http://www.fsca.co.za. JET EDUCATION SERVICES, 2016. Report on the Implementation FOSTER, J., GREER, J., and THORBECKE, E., 1984, ‘A class of Evaluation of the National School Nutrition Programme. decomposable poverty measures.’ Econometrica, 52(3): 761– Department of Planning, Monitoring and Evaluation and the 766. Department of Basic Education: Pretoria. GAUTENG DEPARTMENT OF SOCIAL DEVELOPMENT, 2020. KERR, A., 2017, ‘Tax(i)ing the Poor? Commuting Costs in South Gauteng Department of Social Development Annual Report African Cities.’ South African Journal of Economics, 85(3): 321– 2019/20. Republic of South Africa, Johannesburg. 340, ISSN 0038-2280, doi:10.1111/saje.12161. GOVERNMENT EMPLOYEES PENSION FUND (GEPF), 2017. KERR, A., LAM, D., and WITTENBERG, M., 2019, ‘Post-Apartheid ‘2017 Annual Report’. Government Employees Pension Fund, Labour Market Series 1993-2019.’ Dataset, Version 3.3. Cape Republic of South Africa. Retrieved from https://www.gepf.gov. Town: DataFirst [producer and distributor], doi:10.25828/gtr1- za/annual-reports/. 8r20. 75 KERR, A. and WITTENBERG, M., 2019, ‘Earnings and employment efficiency: Evidence from South Africa.’ American Journal of microdata in South Africa.’ WIDER Working Paper 2019/47. UNU- Agricultural Economics, 93(5): 1391–1405. WIDER, Helsinki, doi:10.35188/UNU-WIDER/2019/681-4. MAKIWANE, M., 2010, ‘The Child Support Grant and teenage KILBURN, K.N., PETTIFOR, A., EDWARDS, J.K., SELIN, A., TWINE, R., childbearing in South Africa.’Development Southern Africa, 27(2): MACPHAIL, C., WAGNER, R., HUGHES, J.P., WANG, J., AND KAHN, 193–204, doi:https://doi.org/10.1080/03768351003740498. K., 2018. ‘Conditional cash transfers and the reduction in partner violence for young women: an investigation of causal pathways MCCORD, A., 2017: ‘The Role of Public Works in Addressing using evidence from a randomized experiment in South Africa’. Poverty: Lessons from Recent Developments in Public Works Journal of the International AIDS Society 21(S1):e25043: 47-54. Programming’, in Lawson, D.; Ado-Kofie L.; and Hulme D. https://doi.org/10.1002/jia2.25043. (ed.), What works for Africa’s Poorest? Programmes and Policies for the Extreme Poor. KLASEN, S. and WOOLARD, I., 2009, ‘Surviving Unemployment Without State Support: Unemployment and Household MLACHILA, M. and MOELETSI, T., 2019, ‘Struggling to Make the Formation in South Africa.’ Journal of African Economies, 18(1): Grade: A Review of the Causes and Consequences of the Weak 1–51. Outcomes of South Africa’s Education System.’ IMF Working Paper WP/19/47. International Monetary Fund: Washington, KLUVE, J., 2014: ‘Youth Labour Market Interventions’, IZA World D.C., Retrieved from https://www.imf.org/en/Publications. of Labour 106. MTHETHWA, R., 2019, ‘Challenges and dilemmas in KNIGHT, L., HOSEGOOD, V., and TIMÆUS, I.M., 2013, ‘The South implementing the child support grant policy in South Africa.’ African disability grant: Influence on HIV treatment outcomes International Journal of Management Practice, 12(1): 94–108. and household well-being in KwaZulu-Natal.’ Development Southern Africa, 30(1): 135–147, doi:10.1080/037683 NATIONAL TREASURY, 2007, Estimates of National Expenditure 5X.2013.755767. 2007. National Treasury, Republic of South Africa, Pretoria. KÖHLER, T., & BHORAT, H., 2020. ‘Social assistance during South , 2008, Estimates of National Expenditure 2008. Africa’s national lockdown: Examining the COVID-19 grant, National Treasury, Republic of South Africa, Pretoria. changes to the Child Support Grant, and post-October policy options’. NIDS-CRAM Wave 2 Working Paper No. 9. National , 2009, Estimates of National Expenditure 2009. Income Dynamics Study (NIDS) – Coronavirus Rapid Mobile National Treasury, Republic of South Africa, Pretoria. Survey (CRAM), September. Retrieved from: https://cramsurvey. org/reports/. , 2010a, Budget Review 2010. National Treasury, Republic of South Africa, Pretoria. LEIBBRANDT, M., FINN, A., and OOSTHUIZEN, M., 2016, ‘Poverty, Inequality, and Prices in Post-Apartheid South Africa.’ In: C. Arndt, , 2010b, Estimates of National Expenditure 2010. A. McKay, and F. Tarp (eds.), Growth and poverty in sub-Saharan National Treasury, Republic of South Africa, Pretoria. Africa, pp. 393–417, UNU-Wider Studies in Development Economics, Oxford University Press, Oxford, United Kingdom. , 2011, Estimates of National Expenditure 2011. National Treasury, Republic of South Africa, Pretoria. LEIBBRANDT, M., FINN, A., and WOOLARD, I., 2012a, ‘Describing and decomposing post-apartheid income inequality in South , 2012a, ‘Enabling a better income in retirement.’ Africa.’ Development Southern Africa, 29(1): pp.19–34. Technical Discussion Paper B for public comment. National Treasury: Pretoria, Retrieved from http://www.treasury.gov.za. , 2012b, ‘Post-Apartheid Changes in South African Inequality.’ In: J. Gornick and M. Jäntti (eds.), Income Inequality: , 2012b, Estimates of National Expenditure 2012. Economic Disparities and the Middle Class in Affluent Countries, National Treasury, Republic of South Africa, Pretoria. Stanford University Press, Palo Alto. , 2013, Estimates of National Expenditure 2013. LEIBBRANDT, M. and LEVINSOHN, J., 2011, ‘Fifteen years on: National Treasury, Republic of South Africa, Pretoria. Household incomes in South Africa.’ National Bureau of Economic Research Working Paper No. 16661. , 2014a, Budget Review 2014. National Treasury, Republic of South Africa, Pretoria, Retrieved from http://www. LEIBBRANDT, M., WOOLARD, I., FINN, A., and ARGENT, J., 2010, treasury.gov.za. ‘Trends in South African Income Distribution and Poverty since the Fall of Apartheid.’ OECD Social, Employment and Migration , 2014b, Estimates of National Expenditure 2014. Working Papers No. 101, OECD Publishing. National Treasury, Republic of South Africa, Pretoria. LOVO, S., 2011, ‘Pension transfers and farm household technical , 2015a, Budget Review 2015. National Treasury, SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 76 Republic of South Africa, Pretoria, Retrieved from http://www. Poverty in South Africa.’ Discussion Papers No. 208. Courant treasury.gov.za. Research Centre ‘Poverty, Equity and Growth in Developing and Transition Countries: Statistical Methods and Empirical , 2015b, Estimates of National Expenditure 2015. Analysis’, Goettingen, Germany, Retrieved from http://www.uni- National Treasury, Republic of South Africa, Pretoria. goettingen.de/en/67061.html. , 2016a, Budget Review 2016. National Treasury, Peres, A., 2019, South Africa’s Expanded Public Works Republic of South Africa, Pretoria, Retrieved from http://www. Programme: Innovations and challenges, Social Protection for treasury.gov.za. Employment Series. , 2016b, Estimates of National Expenditure 2016. PETTIFOR, A., MACPHAIL, C., HUGHES, J.P., SELIN, A., WANG, J., National Treasury, Republic of South Africa, Pretoria. GÓMEZ-OLIVÉ, F.X., ESHLEMAN, S.H., WAGNER, R.G., MABUZA, W., KHOZA, N., SUCHINDRAN, C., MOKOENA, I., TWINE, R., ANDREW, , 2017a, Budget Review 2017. National Treasury, P., TOWNLEY, E., LAEYENDECKER, O., AGYEI, Y., TOLLMAN, S., & Republic of South Africa, Pretoria, Retrieved from http://www. KAHN, K., 2016. ‘The effect of a conditional cash transfer on HIV treasury.gov.za. incidence in young women in rural South Africa (HPTN 068): a phase 3, randomised controlled trial’. Lancet Global Health 4, , 2017b, Estimates of National Expenditure 2017. pp.e978-88. http://dx.doi.org/10.1016/S2214-109X(16)30253-4. National Treasury, Republic of South Africa, Pretoria. PHILIP, K., 2013, ‘The Community Work Programme: Building , 2018a, Budget Review 2018. National Treasury, a Society that Works.’ Employment Working Paper No. Republic of South Africa, Pretoria. 149. International Labour Office, Employment Sector: Geneva, Retrieved from https://www.ilo.org/public/libdoc/ , 2018b, Estimates of National Expenditure 2018. ilo/2013/113B09_183_engl.pdf. National Treasury, Republic of South Africa, Pretoria. POSEL, D. and ROGAN, M., 2012, ‘Gendered trends in poverty in , 2019a, Budget Review 2019. National Treasury, the post-apartheid period, 1997-2006.’ Development Southern Republic of South Africa, Pretoria. Africa, 29(1): 97–113, doi:10.1080/0376835X.2012.645645. , 2019b, Estimates of National Expenditure 2019. RANCHHOD, V., 2006, ‘The effect of the South African old age National Treasury, Republic of South Africa, Pretoria. pension on labour supply of the elderly.’ South African Journal of Economics, 74(4): 725–744. , 2020a, Budget Review 2020. National Treasury, Republic of South Africa, Pretoria. , 2017, ‘Household responses to the cessation of grant income: The case of South Africa’s Old Age Pension.’ SALDRU , 2020b, Estimates of National Expenditure 2020. Working Paper Number 213. Southern Africa Labour and National Treasury, Republic of South Africa, Pretoria. Development Research Unit, University of Cape Town, Retrieved from http://www.opensaldru.uct.ac.za/handle/11090/897. , 2020c, Supplementary Budget Review 2020. National Treasury, Republic of South Africa, Pretoria. REPUBLIC OF SOUTH AFRICA, 1996, ‘Constitution of the Republic of South Africa, No. 108 of 1996.’ Retrieved from https://www. NGUBANE, N. and MAHARAJ, P., 2018, ‘Childbearing in the gov.za/documents/constitution-republic-south-africa-0. Context of the Child Support Grant in a Rural Area in South Africa.’ SAGE Open, 8(4): 1–8. , 2004a, ‘Social Assistance Act, No. 13 of 2004.’ Retrieved from https://www.gov.za/documents/social-assistance-act. OOSTHUIZEN, M., 2012, ‘Low pay in South Africa.’ International Labour Review, 151(3): 173–192. , 2004b, ‘South African Social Security Agency Act, No. 9 of 2004.’ Retrieved from https://www.gov.za/documents/south- , 2013, ‘Inflation Inequality in South Africa.’ DPRU african-social-security-agency-act. Working Paper No. 13/158, Development Policy Research Unit, University of Cape Town, Retrieved from http://www.dpru.uct. , 2016, ‘Unemployment Insurance Act, No. 10 of ac.za. 2016.’ Retrieved from https://www.gov.za/documents/ unemployment-insurance-amendment-act-10-2016-english- , 2017, ‘Social Assistance in South Africa.’ Unpublished tshivenda-19-jan-2017-0000. input paper for the World Bank report entitled “Overcoming Poverty and Inequality in South Africa: An Assessment of ROAD ACCIDENT FUND, 2013, Integrated Annual Report 2012. Drivers, Constraints and Opportunities”. Road Accident Fund, Republic of South Africa, Centurion, South Africa, Retrieved from http://www.raf.co.za/Media-Center/ PASHA, A., 2016, ‘Impact of Cash Grants on Multidimensional Pages/Annual-Reports.aspx. 77 , 2016, Annual Report 2015/16. Road Accident Fund, SOCIAL SECURITY ADMINISTRATION, 2019, Social Security Republic of South Africa, Centurion, South Africa, Retrieved Programs throughout the World: Africa, 2019. SSA Publication from http://www.raf.co.za. No. 13-11803, Social Security Administration, United States Government, Washington, DC. , 2019, Annual Report 2018/19. Road Accident Fund, Republic of South Africa, Centurion, South Africa, Retrieved SOLIDARITY FUND, 2020a, Solidarity Fund Food Relief from http://www.raf.co.za/. Programme Summary Report: June 2020. Retrieved from: http://www.solidarityfund.co.za. ROSENBERG, M., PETTIFOR, A., NGUYEN, N., WESTREICH, D., BOR, J., BÄRNIGHAUSEN, T., MEE, P., TWINE, R., TOLLMAN, S., and KAHN, , 2020b, Solidarity Fund Website. Retrieved from: K., 2015, ‘Relationship between Receipt of a Social Protection http://www.solidarityfund.co.za. Grant for a Child and Second Pregnancy Rates among South African Women: A Cohort Study.’ PLOS ONE, 10(9): e0137352, SOUTH AFRICAN RESERVE BANK, 2020, ‘Quarterly Bulletin Online ISSN 1932-6203, doi:10.1371/journal.pone.0137352. Statistical Query.’ Online database, Retrieved from http://www. reservebank.co.za/. SALDRU, 2018, ‘National Income Dynamics Study 2017, Wave 5.’ Version 1 [dataset]. Pretoria: Department of Planning, SPECIAL COVID GRANT WORKING GROUP, 2020, ‘Economic Monitoring and Evaluation [funding agency]. Cape Town: Mitigation and Relief Measures for COVID-19: A Proposal for a Southern Africa Labour and Development Research Unit Special COVID Grant.’ Policy Proposal. 17 April 2020., Retrieved [implementer], doi:10.25828/fw3h-v708, Retrieved from www. from https://covid19economicideas.org/wp-content/ datafirst.uct.ac.za. uploads/2020/04/Special-Covid-19-Grant-Proposal-full- version-17-April-2020.pdf. SASSA, 2019a. Annual Performance Plan 2019/20. Retrieved from https://www.sassa.gov.za. STANDISH-WHITE, J. and FINN, A., 2015, ‘Unconditional cash transfers and children’s educational outcomes: Evidence from , 2019b, Annual Report 2018/19. South African Social the old-age pension programme in South Africa.’ SALDRU Security Agency, Republic of South Africa, Pretoria. Working Paper Number 147. Southern Africa Labour and Development Research Unit, Cape Town, Retrieved from http:// , 2020a, ‘SASSA Declines Unqualifying COVID-19 Grant opensaldru.uct.ac.za/handle/11090/780. Applications.’ South African Social Security Agency (SASSA) media release published on 18 June 2020. STATISTICS SOUTH AFRICA, 2008. Concepts and definitions used in the Quarterly Labour Force Survey. Report No. 02-11-01, , 2020b, ‘SASSA Media Statement: Level 3 Lockdown August. Statistics South Africa, Pretoria. Announcement Enables Faster Payment Of Covid-19 Grants.’ Media statement released on 25 May 2020, Retrieved from , 2014, The South African MPI: Creating a https://www.sassa.gov.za/Pages/Media-Statement.aspx. multidimensional poverty index using census data. Report No. 03-10-08, Statistics South Africa, Pretoria. , 2020c, ‘You and Your Grants 2020/21.’ South African Social Security Agency (SASSA), Retrieved from https://www. , 2015a, ‘Living Conditions Survey 2014/15.’ Statistics sassa.gov.za/Pages/Grant-Booklets.aspx. South Africa, Pretoria. Dataset, Retrieved from www.statssa.gov. za. , 2020d, ‘Fourth Statistical Report: Payment System. Period: July 2020’. South African Social Security Agency (SASSA), , 2015b, ‘Quarterly Labour Force Survey (2014Q4).’ Department: Monitoring and Evaluation. Retrieved from http:// Dataset. Pretoria: Statistics South Africa. www.sassa.gov.za. , 2017, Poverty trends in South Africa: An examination , n.d., ‘South African Social Security Agency website.’ of absolute poverty between 2006 and 2015. Report No. 03-10- Retrieved from http://www.sassa.gov.za. 06, Statistics South Africa, Pretoria. SCHIEL, R., LEIBBRANDT, M., and LAM, D., 2014, ‘Assessing the , 2018a, ‘Mid-Year Population Estimates, 2018.’ Statistical impact of social grants on inequality: A South African case Release P0302. Pretoria: Statistics South Africa, Retrieved from study.’ WIDER Working Paper 2014/160. Helsinki: UNU-WIDER, http://www.statssa.gov.za. Retrieved from https://www.wider.unu.edu/sites/default/files/ wp2014-160.pdf. , 2018b, Subjective Poverty in South Africa: Findings from the Living Conditions Surveys, 2008/2009-2014/2015. SINYOLO, S., MUDHARA, M., and WALE, E., 2019, ‘The role of social Report No. 03-10-01, Statistics South Africa, Pretoria. grants on commercialization among smallholder farmers in South Africa: Evidence from a continuous treatment approach.’ , 2018c, ‘General Household Survey 2018.’ Statistics South Agribusiness, 35(3): 457–470, doi:10.1002/agr.21601. Africa, Pretoria. Dataset, Retrieved from www.statssa.gov.za. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 78 , 2019a, ‘Mid-Year Population Estimates, 2019.’ Statistical and equity effects of social grants in South Africa.’ Stellenbosch Release P0302. Pretoria: Statistics South Africa, Retrieved from Economic Working Papers No. 15/10. Stellenbosch, South Africa, http://www.statssa.gov.za. Retrieved from https://www.ekon.sun.ac.za/wpapers/2010/ wp152010. , 2019b, ‘National Poverty Lines 2019.’ Statistical Release P0310.1. Pretoria: Statistics South Africa, Retrieved from http:// WAIDLER, J. and DEVEREUX, S., 2019, ‘Social grants, remittances, www.statssa.gov.za. and food security: does the source of income matter?’ Food Security, 11(3:16): 679–702. , 2020a, ‘Business impact survey of the COVID-19 pandemic in South Africa.’ Report No. 00-80-01. Pretoria: WESTERN CAPE DEPARTMENT OF SOCIAL DEVELOPMENT, Statistics South Africa. 2020. Western Cape Strategic Plan 2020-2025. Republic of South Africa, Cape Town. , 2020b, ‘CPI History.’ Time series data published by Statistics South Africa, Retrieved from http://www.statssa.gov. , 2020. Annual Performance Plan 2020/21. za. Republic of South Africa, Cape Town. , 2020c, ‘Quarterly Labour Force Survey (2019Q4).’ WILLS, G., PATEL, L., VAN DER BERG, S., AND MPETA, B., 2020. Dataset. Pretoria: Statistics South Africa. ‘Household resource flows and food poverty during South Africa’s lockdown: Short-term policy implications for three , 2020d, ‘Quarterly Labour Force Survey (2020Q1).’ channels of social protection’. NIDS-CRAM Wave 1 Report 12. Statistical Release P0211. Statistics South Africa: Pretoria, National Income Dynamics Study (NIDS) – Coronavirus Rapid Retrieved from http://www.statssa.gov.za. Mobile Survey (CRAM), July. Retrieved from: https://cramsurvey. org/reports/ , 2020e, ‘Quarterly Labour Force Survey (2020Q2).’ Statistical Release P0211. Statistics South Africa: Pretoria, WOOLARD, I., HARTTGEN, K., and KLASEN, S., 2010, ‘The Evolution Retrieved from http://www.statssa.gov.za. and Impact of Social Security in South Africa.’ In: Conference on “Promoting Resilience through Social Protection in Sub- , 2020f, ‘Quarterly Labour Force Survey (2020Q3).’ Saharan Africa”, Dakar, Senegal, 28-30 June 2010, Retrieved from Statistical Release P0211. Statistics South Africa: Pretoria, http://erd.eui.eu/media/BackgroundPapers/Woolard-Harttgen- Retrieved from http://www.statssa.gov.za. Klasen.pdf. TIPS, 2018, ‘Community Work Programme.’ Background WOOLARD, I. and LEIBBRANDT, M., 2010, ‘The evolution and Materials published by Trade and Industrial Policy Strategies impact of unconditional cash transfers in South Africa.’ Southern (TIPS), Retrieved from https://www.tips.org.za/research- Africa Labour and Development Research Unit Working Paper archive/inequality-and-economic-inclusion/community- No. 51. University of Cape Town, Retrieved from http://www. work-programme/item/3481-community-work-programme- saldru.uct.ac.za/. background-materials. WORLD BANK, 2014, South Africa Economic Update: Fiscal VAN DER BERG, S., 1997, ‘South African social security under Policy and Redistribution in an Unequal Society. World Bank, apartheid and beyond.’ Development Southern Africa, 14(4): Washington, DC. 481–503. , 2018, Overcoming Poverty and Inequality in South , 2001, ‘Redistribution through the budget: Public Africa: An Assessment of Drivers, Constraints and Opportunities. expenditure incidence in South Africa 1993-1997.’ Social The World Bank, Washington, DC. Dynamics, 27(1): 140–164. , 2020a, ‘ASPIRE: The Atlas of Social Protection , 2007, ‘Apartheid’s Enduring Legacy: Inequalities in Indicators of Resilience and Equity.’ Online database., Retrieved Education.’ Journal of African Economies, 16(5): 849–880. from http://datatopics.worldbank.org/aspire/. , 2014, ‘Inequality, poverty and prospects for , 2020b, ‘World DataBank.’ Retrieved from http:// redistribution.’ Development Southern Africa, 31(2): 197–218, do databank.worldbank.org/. i:10.1080/0376835X.2013.871196. , 2020c, ‘Draft: Country Partnership Framework for VAN DER BERG, S., BURGER, R., BURGER, R., LOUW, M., and YU, South Africa’. D., 2007, ‘A Series of National Accounts-Consistent Estimates of Poverty and Inequality in South Africa.’ Stellenbosch Economic , 2020d, ‘The Human Capital Project’. The World Bank, Working Papers No. 09/07. Washington, DC. VAN DER BERG, S., SIEBRITS, K., and LEKEZWA, B., 2010, ‘Efficiency ZIMBALIST, Z., 2017, ‘Analysing post-apartheid poverty trends by 79 geo-type, 1997-2012: The understated role of urbanisation and volatility in South Africa using NIDS: 2008-2017.’ NIDS Discussion social grants.’ Development Southern Africa, 34(2): 151–167, doi Paper 2019/13, version 1. Southern Africa Labour and :10.1080/0376835X.2016.1259989. Development Research Unit, University of Cape Town, Retrieved from http://www.nids.uct.ac.za. ZIZZAMIA, R. and RANCHHOD, V., 2019, ‘Measuring employment SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 80 A. Appendix One The Living Conditions Survey 2014/15 Microdata There are thus two options for imputing grant access at the individual level in this dataset. The first option uses The Living Conditions Survey 2014/15 (Statistics South the access data from section 4 but requires the allocation of Africa, 2015a) is a key source of data for the analysis child grants to individual children in isolation of any further in this report. While the survey collects extensive data on, information, and raises significant issues around the distribution amongst other things, incomes and receipt of social grants, of multiple types of grants across multiple children within a it presents particular challenges with respect to the analysis household. The second option approaches the challenge from of social grants. This is primarily because the two parts of the the income side: by using the level of income relative to the survey that collect information about the receipt of social known value of each of the grants, it is possible to determine grants do not necessarily align with each other. approximately how many grants of each type are received. These need to then be allocated to individual children, although In section 4 of the questionnaire, respondents are asked for the purposes of this report it is not important exactly which to indicate whether they receive any social grants. Those child receives which grant. that answer in the affirmative are then asked to indicate which grant they receive. This section is complicated by two issues. Given that the second approach uses more information First, children under the age of 18 do not answer the initial to make the allocation, it is the preferred option here. In question directly; instead, where children under 18 years receive making the allocation it is assumed that children receive only a grant, the information is linked to the caregiver. In other one grant and that the reported income refers to receipt of a words, where a child receives a child support grant, they would grant for the full 12 months (i.e. not two grants received for 6 answer “No” to the question whether they receive a social grant, months). The steps followed were: while their caregiver would answer “Yes” and would indicate 1. Construct household-level annual grant income that they receive a child support grant. Thus, it is possible that variables for each of the child grants, consolidating all the information pertaining to multiple children is consolidated income for each grant across all household members. under a single caregiver. Second, there is insufficient information 2. Divide the household-level annual grant income to accurately disentangle these consolidated responses, or to variables by 12 and by the value of that grant itself, to even identify these consolidated responses. This is because estimate the number of grants. the household roster identifies individuals’ relationships to the household head, but not to each other. Further, when 3. Based on the number of children within each household, respondents are asked to indicate which grants they receive, determine how many receive each type of grant (e.g. they are not asked how many of these grants they receive. within each household, the number of child support grants divided by total number of child grants multiplied This problem impacts all of the child-related grants, namely the by the number of children). When these numbers are child support, care dependency, and foster child grants, which aggregated again at the household level, they should together account for 73 percent of all grants in the 2018/19 not exceed the number of children. In the case of only financial year (see Table B.2). Further, it is particularly difficult two households, the number of grants is estimated to where households receive multiple different types of child be greater (by one) than the number of children; for grants. these households, the number of child support grants is reduced by one. In section 24 of the questionnaire, respondents are asked 4. Allocation to specific children can be done in various to provide detail on the income received in the 12 months ways. For this research, children are ranked from youngest preceding the survey period. They are further asked to detail to oldest and grants allocated sequentially. If there income received during the survey period through the diary are x child support grants, y care dependency grants, that is administered as part of the survey. Here, grant income is and z foster child grants in a household, the youngest recorded against the adult grant recipient or the adult caregiver, x children are allocated child support grants, the next and it is therefore not possible to directly tie a particular child y youngest children are allocated care dependency grant to a particular child. However, this data does provide a grants, and the next z children are allocated foster child sense of the number of children within a household for whom grants. This sequence is arbitrary and inconsequential to grants are received. It should also be noted that cross-checking the analysis presented, since grant access is not analysed this information against the access information from section by age. 4 reveals inconsistencies, which are not entirely unexpected 5. Finally, based on these allocations, it is then possible to given the different reference periods. allocate the household-level income for each grant to individual children. 81 Table A.1 presents the estimates of direct grant beneficiaries the estimates generated from a pre-release version of the LCS used as a basis for the analysis in this report, and compares 2014/15 microdata and used by Oosthuizen (2017). them to official numbers of grants paid out by SASSA and to Table A.1. Allocation of Grants to Individuals in the LCS 2014/15 Microdata SASSA (2014/15) Previous Report Derived from Income Questions Older persons 3 087 3 204 3 178 Disability 1 113 1 165 1 232 Child support 11 703 13 275 13 206 Care dependency 127 127 121 Foster child 500 464 373 Grant-in-aid 113 79 8 Other, e.g. social relief 255 5 TOTAL 16 643 18 569 18 124 Total excl. other 16 643 18 314 18 119 Source: Own calculations, Oosthuizen (2017); SASSA (2019); Statistics South Africa (2015a). Notes: Figures from the Oosthuizen (2017) report were derived from a pre-release version of the LCS 2014/15 microdata by the World Bank. As is the case for the data used by Oosthuizen (2017), the allocation used here provides broadly similar estimates to those method described above yields substantially more child used previously. What it does mean, however, is that estimates support grants than are officially reported by SASSA (2019). It of coverage based on the microdata will be somewhat higher also yields fewer foster child grants. However, in general, the than the true value. SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 82 B. Appendix Two Table B.1. Consolidated Government Spending, 2010/11-2022/23 Financial Year GDP Consolidated Government Spending as Share of GDP (%) Total Health Education Social Combined (Nominal R Protection Total billions) 2010/11 2 749.5 30.5 3.7 6.2 4.8 14.7 2011/12 3 080.9 29.9 3.7 6.4 4.6 14.7 2012/13 3 327.6 30.1 3.7 6.4 4.6 14.7 2013/14 3 624.3 30.2 3.7 6.4 4.7 14.7 2014/15 3 867.9 29.6 3.7 6.4 3.8 13.9 2015/16 4 127.0 31.6 3.8 6.4 4.9 15.1 2016/17 4 419.4 31.2 3.9 6.5 5.0 15.4 2017/18 4 698.7 31.5 4.0 6.5 5.0 15.5 2018/19 4 921.5 32.3 4.1 6.9 5.3 16.3 2019/20 5 157.3 34.5 4.2 7.3 5.6 17.1 2020/21 5 428.2 34.6 4.1 7.1 5.8 17.0 2021/22 5 759.0 34.0 4.1 7.1 5.6 16.9 2022/23 6 126.3 33.5 4.1 7.0 5.7 16.7 Source: Own calculations, National Treasury (2014a, 2015a, 2016a, 2017a, 2018a, 2019a, 2020a). Note: Budgeted expenditure for the MTEF (2020/21-2022/23) presented in the 2020 National Budget. Table B.2. Grant Beneficiaries and Spending, 2006/07-2018/19 Care Child Disability Foster Grant-in- Older Social War Vete- TOTAL Depen- Support Child Aid Persons Relief of rans dency Distress Nominal Spending (R billions) 2006/07 1.0 17.6 14.3 2.9 3.0 21.2 0.0 0.0 57.0 2007/08 1.1 19.6 15.3 3.4 3.0 22.8 0.1 0.0 62.5 2008/09 1.3 22.3 16.5 3.9 3.0 25.9 0.6 0.0 70.7 2009/10 1.4 26.7 16.6 4.4 3.0 29.8 0.2 0.0 79.3 2010/11 1.6 30.3 16.8 4.6 3.0 33.8 0.2 0.0 87.5 2011/12 1.7 34.3 17.4 5.0 3.0 37.1 0.2 0.0 96.0 2012/13 1.9 38.1 17.6 5.3 3.0 40.5 0.2 0.0 103.9 2013/14 2.0 39.6 17.8 5.3 3.0 44.1 0.5 0.0 109.6 2014/15 2.2 43.7 18.7 5.4 3.0 49.0 0.5 0.0 120.0 2015/16 2.4 47.3 19.2 5.4 3.0 53.1 0.4 0.0 128.3 2016/17 2.6 51.5 19.9 5.3 3.0 58.3 0.6 0.0 138.9 2017/18 2.8 55.9 20.9 5.0 3.0 64.2 0.5 0.0 150.2 2018/19 3.1 60.6 22.0 5.1 3.0 70.6 0.4 0.0 162.7 Real Spending (R billions, March 2020 prices) 2006/07 2.2 37.6 30.5 6.1 3 45.4 0.1 0.1 122.1 2007/08 2.2 38.8 30.2 6.8 3 45.1 0.2 0 123.7 2008/09 2.3 39.7 29.3 7 3 46.1 1.1 0 125.7 2009/10 2.4 44.8 27.9 7.5 3 50.2 0.3 0 133.3 2010/11 2.6 49.1 27.3 7.5 3 54.7 0.3 0 141.7 2011/12 2.7 52.7 26.7 7.7 3 57 0.3 0 147.3 83 Care Child Disability Foster Grant-in- Older Social War Vete- TOTAL Depen- Support Child Aid Persons Relief of rans dency Distress 2012/13 2.7 55.4 25.6 7.8 3 58.8 0.3 0 151.1 2013/14 2.7 54.4 24.4 7.3 3 60.5 0.7 0 150.6 2014/15 2.9 56.9 24.4 7 3 63.8 0.6 0 156 2015/16 3 58.5 23.7 6.7 3 65.7 0.5 0 158.7 2016/17 3 59.9 23.2 6.2 3 67.9 0.7 0 161.6 2017/18 3.2 62.1 23.2 5.5 3 71.4 0.5 0 166.9 2018/19 3.3 64.4 23.4 5.4 3 75 0.4 0 172.8 Beneficiaries (thousands) 2006/07 99 7 864 1 423 401 32 2 195 2.3 12 015 2007/08 102 8 190 1 408 454 37 2 230 1.9 12 424 2008/09 107 8 765 1 287 475 46 2 391 1.5 13 072 2009/10 111 9 570 1 264 511 53 2 547 1.2 14 057 2010/11 112 10 372 1 201 513 58 2 679 1 14 936 2011/12 115 10 928 1 198 537 66 2 751 0.8 15 596 2012/13 120 11 342 1 164 532 74 2 873 0.6 16 106 2013/14 121 11 126 1 120 512 83 2 970 0.4 15 932 2014/15 127 11 703 1 113 500 113 3 087 0.3 16 643 2015/16 131 11 973 1 086 470 138 3 194 0.2 16 992 2016/17 145 12 081 1 067 440 164 3 302 0.2 17 201 2017/18 147 12 269 1 062 416 192 3 423 0.1 17 510 2018/19 150 12 452 1 048 386 222 3 553 0.1 17 812 Source: Own calculations, SASSA (2019); Statistics South Africa (2020b). Notes: Grants included under ‘Other’ are the care dependency grant, the foster child grant, grant in aid, social relief of distress, and the war veterans’ grant. Spending figures are deflated to March 2020 prices using average headline CPI for April to March of each year. The number of beneficiaries of social relief of distress is not reported by SASSA. Figure B.1. Social Grants Across the Income Distribution, 2017 Source: Own calculations, Saldru (2018). SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 84 Figure B.2. Adequacy of Social Assistance Benefits Across Geography, 2014/15 Source: Own calculations, Statistics South Africa (2015a). 85 Table B.3. ASPIRE Estimates for Upper-Middle Income Countries Country Spending Social Assistance Coverage (% of Population) on Social Assistance Year Pre-Transfer Distribution Post-Transfer Distribution Year % of GDP Overall Q1 Q2 Q3 Q4 Q5 Overall Q1 Q2 Q3 Q4 Q5 Albania 2014 1.57 2012 19.1 31.3 20.4 19.8 13.2 10.6 19.1 27.7 22.8 19.8 15.1 10.0 Argentina 2015 2.05 2013 19.8 49.2 25.9 15.4 6.9 1.4 19.8 45.8 30.2 14.7 6.6 1.5 Armenia 2014 1.37 2014 28.4 45.7 32.8 23.8 21.9 17.9 28.4 40.1 31.9 27.0 23.3 19.6 Azerbaijan 2014 0.84 2015 26.9 38.6 24.8 25.7 23.1 22.5 26.9 33.4 25.1 26.7 24.9 24.5 Belarus 2015 3.06 2016 54.1 91.8 59.6 46.6 39.2 33.3 54.1 64.9 56.7 52.2 50.5 46.1 Belize .. 2009 16.3 20.1 20.3 12.7 16.7 11.6 16.3 18.5 21.1 13.6 16.3 11.9 Bosnia & Herzegovina 2010, ‘11 3.89 2015 17.4 30.7 21.5 14.6 11.8 8.6 17.4 23.2 23.2 16.5 13.3 11.0 Botswana 2014–16 1.66 2009 73.8 94.9 87.8 79.4 66.6 40.3 73.8 91.6 87.3 80.2 67.5 42.3 Brazil 2015 1.35 2015 23.7 64.2 31.6 15.7 5.5 1.6 23.7 58.5 33.6 17.3 7.3 1.9 Bulgaria 2014 1.39 2007 39.5 61.5 45.8 35.6 33.9 20.9 39.5 57.6 41.6 38.9 35.4 24.2 China 2014 0.76 2013 43.8 61.0 54.5 42.1 32.8 28.7 43.8 65.0 53.3 40.2 31.2 29.3 Colombia 2015 3.01 2014 59.3 83.3 77.0 64.1 47.3 24.5 59.3 81.4 78.5 64.5 47.8 24.1 86 Costa Rica 2013 0.74 2014 45.8 79.4 62.6 47.5 29.4 10.3 45.8 77.6 63.6 48.3 29.6 10.1 Dominica .. 2002 8.0 15.8 11.0 8.9 3.0 1.6 8.0 10.6 13.6 10.0 2.9 3.0 Dominican Rep. 2015 1.18 2014 30.0 44.7 38.9 30.5 24.6 11.1 30.0 40.6 39.6 31.3 26.5 11.8 Ecuador 2010, ‘15 1.49 2016 67.1 87.1 81.4 71.8 58.6 36.3 67.1 86.7 82.0 72.0 59.3 35.3 Fiji 2015 1.14 2008 9.6 16.5 7.9 7.4 6.9 9.0 9.6 11.2 9.0 9.4 8.1 10.0 Gabon 2014 0.20 2005 44.7 48.2 54.2 50.2 42.5 28.4 44.7 48.8 54.0 48.8 43.6 28.3 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW Georgia 2013 6.99 2011 64.6 92.9 76.3 62.8 47.1 44.1 64.6 77.8 67.8 66.7 58.2 52.8 Guatemala 2013 0.19 2014 59.1 71.5 73.0 67.1 55.7 28.2 59.1 71.6 73.1 66.9 55.8 28.1 Iraq 2012–13 2.56 2012 75.8 87.1 81.7 77.5 72.0 60.5 75.8 86.1 81.7 77.2 72.4 61.5 Jamaica .. 2010 55.0 73.9 73.4 52.3 44.1 31.4 55.0 72.1 74.5 51.7 45.1 31.5 Jordan 2009 0.68 2010 65.7 86.4 78.1 72.2 61.3 30.5 65.7 83.3 76.8 75.4 61.1 31.8 Kazakhstan 2014 1.62 2015 30.6 53.9 35.9 26.5 21.5 15.2 30.6 44.6 34.8 29.2 25.0 19.5 Kosovo .. 2013 14.1 42.7 10.3 7.5 4.4 5.5 14.1 35.4 13.6 9.2 5.5 6.6 Lebanon 2013 1.04 2004 4.8 3.2 4.6 6.1 5.7 4.6 4.8 3.2 4.6 6.1 5.7 4.6 Malaysia 2013 0.72 2008 82.8 94.2 88.1 82.8 78.4 70.5 82.8 93.8 87.9 82.7 78.6 71.0 Maldives 2010–11 1.21 2009 13.5 15.4 15.1 12.7 13.3 11.4 13.5 14.0 15.9 12.4 13.1 12.2 Mauritius 2014–15 3.46 2012 44.9 83.5 51.1 36.0 27.8 25.9 44.9 51.8 46.8 43.1 42.5 40.0 Mexico 2015 1.67 2014 32.5 61.2 39.0 30.3 20.1 12.1 32.5 53.8 39.3 32.7 23.4 13.4 Country Spending Social Assistance Coverage (% of Population) on Social Assistance Year Pre-Transfer Distribution Post-Transfer Distribution Year % of GDP Overall Q1 Q2 Q3 Q4 Q5 Overall Q1 Q2 Q3 Q4 Q5 Montenegro 2013 1.76 2014 8.1 24.8 5.6 7.4 1.3 1.3 8.1 22.6 6.4 4.1 4.6 2.6 Namibia 2014 3.19 2009 15.2 26.1 21.9 16.1 9.0 2.8 15.2 26.1 21.9 16.1 9.0 2.8 Paraguay .. 2014 55.8 82.6 72.7 58.1 44.9 20.6 55.8 81.7 72.8 58.1 45.8 20.4 Peru 2015 1.43 2014 56.1 88.0 76.1 59.2 42.0 15.1 56.1 88.0 76.4 59.8 41.5 14.8 Romania 2014 1.06 2012 61.8 74.7 67.0 64.6 60.3 42.6 61.8 82.9 69.3 62.7 52.5 41.8 Russian Fed. 2015 1.89 2016 67.9 85.3 76.7 69.1 59.2 49.1 67.9 78.9 73.1 68.2 64.8 54.5 Serbia 2013 1.96 2015 13.4 30.7 16.1 9.6 6.4 4.1 13.4 29.5 13.2 10.7 8.1 5.6 South Africa (ASPIRE) 2015 3.31 2014 78.6 99.6 95.1 85.3 69.4 43.6 78.6 96.1 93.0 86.8 72.4 44.9 South Africa (own calc.) .. 2014/15 64.0 95.2 85.0 74.1 50.5 15.2 64.0 86.1 85.8 77.5 54.0 16.5 Sri Lanka 2013–15 0.66 2012 26.2 48.3 33.3 23.9 16.5 9.0 26.2 46.7 34.1 24.3 16.7 9.2 Thailand 2010–11 0.47 2013 59.3 83.7 72.4 63.1 45.5 31.8 59.3 81.6 72.3 63.4 47.0 32.4 Turkey 2013 1.14 2016 18.0 49.1 22.4 10.0 5.9 2.4 18.0 51.3 20.1 10.9 5.5 2.1 87 Venezuela .. 2006 4.7 5.0 6.1 5.6 4.1 2.9 4.7 5.0 6.1 5.6 4.1 2.9 Table B.3. ASPIRE Estimates for Upper-Middle Income Countries (cont.) Country Beneficiary Incidence (% of Beneficiaries) Pre-Transfer Distribution Post-Transfer Distribution Year Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Albania 2012 32.8 21.4 20.7 13.9 11.2 29.0 23.9 20.7 15.9 10.5 Argentina 2013 49.8 26.2 15.6 7.0 1.4 46.3 30.5 14.9 6.7 1.5 Armenia 2014 32.1 23.1 16.7 15.4 12.6 28.2 22.5 19.0 16.4 13.8 Azerbaijan 2015 28.7 18.4 19.1 17.1 16.7 24.8 18.7 19.8 18.5 18.2 Belarus 2016 33.9 22.1 17.2 14.5 12.3 24.0 21.0 19.3 18.7 17.1 Belize 2009 24.6 25.0 15.6 20.5 14.3 22.7 25.8 16.8 20.1 14.6 Bosnia & Herzegovina 2015 35.2 24.7 16.7 13.6 9.8 26.6 26.6 18.9 15.2 12.6 Botswana 2009 25.7 23.8 21.5 18.1 10.9 24.8 23.7 21.7 18.3 11.5 Brazil 2015 54.1 26.7 13.2 4.7 1.3 49.4 28.3 14.6 6.1 1.6 Bulgaria 2007 31.1 23.2 18.0 17.1 10.6 29.1 21.0 19.7 17.9 12.3 China 2013 27.8 24.9 19.2 15.0 13.1 29.7 24.3 18.4 14.2 13.4 Colombia 2014 28.1 26.0 21.7 16.0 8.3 27.5 26.5 21.8 16.1 8.2 Costa Rica 2014 34.6 27.3 20.7 12.8 4.5 33.8 27.8 21.1 12.9 4.4 Dominica 2002 39.4 27.1 22.1 7.5 3.9 26.5 33.8 24.7 7.4 7.7 Dominican Rep. 2014 29.9 26.0 20.3 16.4 7.4 27.1 26.4 20.9 17.7 7.9 88 Ecuador 2016 25.9 24.3 21.4 17.5 10.8 25.9 24.4 21.5 17.7 10.5 Fiji 2008 34.5 16.6 15.5 14.3 19.1 23.4 18.9 19.7 17.0 20.9 Gabon 2005 21.6 24.2 22.5 19.0 12.7 21.8 24.2 21.8 19.5 12.7 Georgia 2011 28.7 23.6 19.4 14.6 13.6 24.0 21.0 20.6 18.0 16.4 Guatemala 2014 24.2 24.7 22.7 18.9 9.5 24.2 24.8 22.6 18.9 9.5 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW Iraq 2012 23.0 21.6 20.4 19.0 16.0 22.7 21.6 20.4 19.1 16.2 Jamaica 2010 26.8 26.6 19.1 16.1 11.4 26.0 27.3 18.8 16.4 11.5 Jordan 2010 26.3 23.8 22.0 18.7 9.3 25.4 23.4 23.0 18.6 9.7 Kazakhstan 2015 35.2 23.5 17.3 14.1 9.9 29.1 22.8 19.1 16.3 12.7 Kosovo 2013 60.6 14.7 10.6 6.2 7.9 50.3 19.3 13.2 7.8 9.4 Lebanon 2004 13.0 19.1 25.3 23.6 19.0 13.0 19.1 25.3 23.6 19.0 Malaysia 2008 22.8 21.3 20.0 18.9 17.0 22.7 21.2 20.0 19.0 17.2 Maldives 2009 19.2 22.7 19.7 20.5 17.9 20.2 23.2 18.6 19.8 18.3 Mauritius 2012 37.2 22.8 16.0 12.4 11.6 23.1 20.9 19.2 19.0 17.9 Mexico 2014 37.6 24.0 18.6 12.4 7.4 33.1 24.1 20.1 14.4 8.3 Montenegro 2014 61.5 13.7 18.5 3.3 3.1 56.1 15.9 10.0 11.6 6.4 Namibia 2009 34.4 28.9 21.2 11.8 3.7 34.4 28.9 21.2 11.8 3.7 Paraguay 2014 29.6 26.1 20.8 16.1 7.4 29.3 26.1 20.8 16.4 7.3 Country Beneficiary Incidence (% of Beneficiaries) Pre-Transfer Distribution Post-Transfer Distribution Year Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Peru 2014 31.4 27.1 21.1 15.0 5.4 31.4 27.2 21.3 14.8 5.3 Romania 2012 24.1 21.7 20.9 19.5 13.8 26.8 22.4 20.3 17.0 13.5 Russian Fed. 2016 25.1 22.6 20.4 17.4 14.5 23.2 21.5 20.1 19.1 16.1 Serbia 2015 45.8 24.1 14.4 9.6 6.1 44.1 19.7 15.9 12.0 8.3 South Africa (ASPIRE) 2014 25.3 24.2 21.7 17.7 11.1 24.4 23.7 22.1 18.4 11.4 South Africa (own calc.) 2014/15 29.8 26.5 23.2 15.8 4.7 26.9 26.8 24.2 16.9 5.1 Sri Lanka 2012 36.9 25.4 18.3 12.6 6.9 35.7 26.0 18.5 12.7 7.0 Thailand 2013 28.2 24.4 21.3 15.3 10.7 27.5 24.4 21.4 15.8 10.9 Turkey 2016 54.6 25.0 11.2 6.6 2.6 57.1 22.4 12.1 6.1 2.4 Venezuela 2006 21.1 25.7 23.8 17.3 12.2 21.1 25.7 23.8 17.3 12.2 89 Table B.3. ASPIRE Estimates for Upper-Middle Income Countries (cont.) Country Benefit Incidence (% of Benefits) Pre-Transfer Distribution Post-Transfer Distribution Year Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Albania 2012 39.3 18.2 18.7 12.3 11.4 27.6 18.3 21.2 19.2 13.7 Argentina 2013 54.4 23.0 13.6 6.4 2.6 46.4 28.2 14.8 7.1 3.5 Armenia 2014 50.9 18.6 12.3 9.8 8.4 32.3 22.0 17.3 16.2 12.2 Azerbaijan 2015 57.0 13.6 9.7 9.9 9.8 17.4 11.6 19.1 19.0 32.9 Belarus 2016 76.9 10.5 4.9 3.5 4.2 27.1 23.3 18.1 14.9 16.6 Belize 2009 32.6 14.7 14.7 14.6 23.4 18.7 20.6 10.4 19.8 30.4 Bosnia & Herzegovina 2015 44.3 18.9 12.8 12.8 11.2 16.5 20.0 19.5 19.3 24.7 Botswana 2009 32.8 18.4 15.2 17.3 16.4 12.8 15.6 18.7 24.2 28.6 Brazil 2015 57.0 22.9 12.6 5.5 2.0 31.6 26.2 24.8 13.5 4.0 Bulgaria 2007 38.0 23.7 14.9 14.0 9.3 28.1 20.2 19.3 17.3 15.1 China 2013 33.0 23.0 16.8 14.2 13.0 24.2 23.7 19.2 16.4 16.4 Colombia 2014 40.8 26.9 17.8 8.4 6.0 33.1 28.9 20.5 11.4 6.2 Costa Rica 2014 50.0 24.2 15.8 8.1 1.9 31.1 31.7 21.9 12.4 2.9 Dominica 2002 12.8 14.7 34.7 11.6 26.2 2.4 17.5 19.2 13.7 47.2 Dominican Rep. 2014 27.8 24.7 20.7 17.7 9.1 22.5 24.9 21.7 20.2 10.6 90 Ecuador 2016 46.9 22.8 16.2 10.7 3.4 35.7 26.6 19.4 13.8 4.4 Fiji 2008 57.0 10.1 8.0 8.5 16.5 13.7 11.2 24.1 17.6 33.4 Gabon 2005 48.1 17.2 19.1 8.7 6.8 4.7 8.7 13.7 31.0 42.0 Georgia 2011 38.8 19.5 16.6 12.7 12.4 20.4 19.4 20.6 20.6 19.0 Guatemala 2011 37.6 25.0 19.1 11.2 7.1 30.5 25.8 20.0 13.7 10.0 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW Iraq 2012 34.3 15.9 14.6 16.4 18.9 12.4 14.4 17.6 20.1 35.5 Jamaica 2010 45.7 28.7 15.3 7.4 2.9 41.7 28.2 16.8 9.2 4.1 Jordan 2010 47.7 14.4 13.2 11.3 13.5 22.7 17.0 17.7 20.4 22.2 Kazakhstan 2015 39.1 19.4 14.9 14.2 12.4 21.5 20.0 19.7 19.5 19.3 Kosovo 2013 72.2 11.6 7.4 3.9 5.0 50.6 22.5 13.3 4.8 8.8 Lebanon .. .. .. .. .. .. .. .. .. .. Malaysia 2008 25.5 18.5 16.9 17.0 22.1 20.8 18.5 17.8 18.0 24.9 Maldives 2009 22.2 18.9 17.4 17.7 23.8 19.8 19.8 18.3 17.2 24.9 Mauritius 2012 45.6 18.5 13.3 10.6 11.9 16.5 18.7 19.3 21.3 24.2 Mexico 2014 42.1 23.1 17.0 10.9 6.8 28.4 24.7 21.7 15.6 9.7 Montenegro 2014 55.5 11.6 23.1 7.1 2.8 43.1 17.7 10.9 17.7 10.6 Namibia .. .. .. .. .. .. .. .. .. .. Paraguay 2014 49.8 24.0 13.1 10.6 2.5 25.4 33.8 22.7 14.8 3.3 Country Benefit Incidence (% of Benefits) Pre-Transfer Distribution Post-Transfer Distribution Year Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Peru 2014 65.2 22.6 8.9 2.8 0.5 60.7 25.6 9.9 3.2 0.5 Romania 2012 41.3 20.3 15.9 12.4 10.1 24.0 20.9 17.7 16.4 21.0 Russian Fed. 2016 29.7 22.8 20.5 15.5 11.5 12.7 22.3 24.0 22.2 18.8 Serbia 2015 53.2 20.5 12.2 7.3 6.8 34.5 16.2 16.8 14.2 18.3 South Africa (ASPIRE) 2014 40.0 23.0 16.9 12.1 8.0 25.5 25.0 22.6 16.5 10.5 South Africa (own calc.) 2014/15 33.1 26.4 19.7 14.5 6.2 17.7 28.2 27.6 19.3 7.2 Sri Lanka 2012 46.0 19.8 14.6 11.7 7.8 30.0 22.0 19.0 12.2 16.8 Thailand 2013 35.0 21.9 17.2 14.1 11.8 24.4 23.4 19.5 16.5 16.3 Turkey 2016 52.1 22.5 13.1 7.3 5.0 43.9 24.2 17.6 8.2 6.1 Venezuela .. .. .. .. .. .. .. .. .. .. 91 Table B.3. ASPIRE Estimates for Upper-Middle Income Countries (cont.) Country Adequacy (Social Assistance Benefits as a Share of Total Expenditure) (%) Pre-Transfer Distribution Post-Transfer Distribution Year Overall Q1 Q2 Q3 Q4 Q5 Overall Q1 Q2 Q3 Q4 Q5 Albania 2012 6.0 12.7 6.3 5.2 3.9 2.9 6.0 10.9 5.8 6.0 5.3 3.6 Argentina 2013 11.3 22.3 9.3 6.6 4.7 5.3 11.3 21.9 10.0 7.2 5.4 6.8 Armenia 2014 17.0 31.5 17.4 13.1 9.9 6.6 17.0 32.4 21.1 16.3 12.9 8.0 Azerbaijan 2015 6.1 20.4 5.7 3.4 3.0 1.9 6.1 8.1 5.0 6.6 5.4 5.9 Belarus 2016 42.2 124.1 24.6 12.7 8.6 7.8 42.2 92.3 61.0 41.7 28.6 22.4 Belize 2009 8.6 27.6 9.0 12.8 8.9 3.9 8.6 23.4 12.0 9.6 12.2 4.8 Bosnia & Herzegovina 2015 13.8 31.4 13.3 10.2 9.1 6.4 13.8 19.4 14.1 14.3 13.0 11.6 Botswana 2009 9.5 47.5 16.3 9.2 7.0 3.6 9.5 22.5 14.3 11.2 9.9 6.1 Brazil 2015 17.3 34.0 13.8 10.0 8.0 3.8 17.3 24.5 15.7 18.0 15.5 6.7 Bulgaria 2007 13.0 31.0 16.1 10.1 7.9 5.4 13.0 26.8 15.7 12.3 9.5 7.7 China 2013 2.3 7.5 3.9 2.4 1.7 0.8 2.3 6.0 4.0 2.8 2.0 0.9 Colombia 2014 5.1 13.2 5.9 3.8 2.5 1.5 5.1 12.4 6.2 4.3 3.1 1.6 Costa Rica 2014 13.3 26.4 12.2 8.3 6.3 4.2 13.3 22.9 15.1 11.1 7.9 5.6 Dominica 2002 21.0 47.0 20.0 26.8 12.9 16.8 21.0 31.2 25.4 16.7 19.3 22.1 Dominican Rep. 2014 5.3 11.1 6.8 5.0 3.7 2.3 5.3 10.7 6.9 5.3 4.0 2.6 92 Ecuador 2016 12.6 27.0 11.9 8.8 6.8 3.9 12.6 26.3 14.2 10.0 7.7 4.5 Fiji 2008 14.0 61.9 16.2 10.0 7.8 4.5 14.0 30.3 18.7 25.8 14.5 8.5 Gabon 2005 19.6 66.6 25.1 17.8 6.2 5.6 19.6 11.2 17.9 19.7 23.1 19.4 Georgia 2011 29.2 83.0 36.5 26.2 19.5 10.9 29.2 68.4 43.4 33.1 27.4 14.5 Guatemala 2011 3.4 11.1 4.8 3.0 1.7 0.9 3.4 9.9 5.0 3.2 2.0 1.2 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW Iraq 2012 2.8 9.1 3.0 2.2 2.0 1.5 2.8 3.4 2.8 2.7 2.4 2.8 Jamaica 2010 4.6 9.3 4.7 3.0 2.0 1.7 4.6 9.6 4.5 3.4 2.2 2.2 Jordan 2010 4.0 13.0 3.1 2.4 1.8 2.7 4.0 6.9 3.8 3.1 3.2 4.2 Kazakhstan 2015 12.6 21.1 12.3 10.4 9.8 7.6 12.6 15.8 13.5 12.9 12.0 9.9 Kosovo 2013 13.0 24.2 9.7 6.8 4.9 3.0 13.0 22.9 15.7 10.7 5.3 4.6 Lebanon .. .. .. .. .. .. .. .. .. .. .. .. Malaysia 2008 1.7 7.8 3.2 2.0 1.3 0.8 1.7 6.5 3.2 2.1 1.4 0.9 Maldives 2009 24.8 85.3 38.2 27.8 19.2 13.7 24.8 75.6 37.6 30.6 19.4 14.1 Mauritius 2012 28.8 63.9 31.8 24.3 18.7 11.2 28.8 54.9 42.2 35.1 28.6 16.8 Mexico 2014 13.9 29.7 14.5 10.6 7.8 5.8 13.9 26.0 16.9 13.0 9.7 7.1 Montenegro 2014 22.1 31.1 18.0 19.0 19.2 5.5 22.1 28.0 25.7 19.1 21.7 11.6 Namibia .. .. .. .. .. .. .. .. .. .. .. .. Paraguay 2014 14.2 24.2 13.5 8.5 7.7 9.3 14.2 18.2 17.9 12.7 9.0 10.3 Country Adequacy (Social Assistance Benefits as a Share of Total Expenditure) (%) Pre-Transfer Distribution Post-Transfer Distribution Year Overall Q1 Q2 Q3 Q4 Q5 Overall Q1 Q2 Q3 Q4 Q5 Peru 2014 8.0 12.3 6.0 4.1 2.8 2.2 8.0 12.5 6.4 4.3 2.9 2.2 Romania 2012 10.9 30.8 13.5 8.7 5.7 4.3 10.9 19.7 13.1 9.5 8.1 8.5 Russian Fed. 2016 6.8 19.8 10.3 7.3 4.6 2.4 6.8 10.5 11.1 9.0 6.3 3.6 Serbia 2015 22.2 41.4 19.6 15.1 11.0 9.6 22.2 31.3 20.0 19.9 17.3 19.7 South Africa (ASPIRE) 2014 29.4 125.3 50.0 27.0 15.1 6.7 29.4 114.0 58.9 36.2 19.5 8.1 South Africa (own calc.) 2014/15 26.0 66.4 40.7 25.1 14.3 6.7 26.0 42.8 46.1 33.6 18.4 7.4 Sri Lanka 2012 3.7 8.5 3.5 2.7 2.2 1.4 3.7 5.8 3.8 3.4 2.3 3.0 Thailand 2013 6.2 15.1 8.2 5.7 4.3 2.5 6.2 13.1 8.9 6.3 4.8 3.2 Turkey 2016 6.5 10.8 5.8 5.1 3.4 2.7 6.5 8.8 6.7 6.1 3.8 3.5 Venezuela .. .. .. .. .. .. .. .. .. .. .. .. 93 Table B.3. ASPIRE Estimates for Upper-Middle Income Countries (cont.) Country Social Assistance Impact on Poverty & Inequality Benefit-Cost Ratio Reduction (%) Reduction (%) Year Year Year Ratio Poverty Rate Poverty Gap Gini Albania 2012 4.6 11.0 2012 1.6 2012 0.328 Argentina 2013 6.5 16.7 2013 2.1 2013 0.525 Armenia 2014 11.8 28.7 2014 4.2 2014 0.434 Azerbaijan 2015 6.3 15.5 2015 2.1 2015 0.343 Belarus 2016 41.6 77.6 2016 31.4 2016 0.434 Belize 2009 0.7 2.2 2009 0.2 2009 0.259 Bosnia & Herzegovina 2015 10.1 17.0 2015 2.6 2015 0.281 Botswana 2009 20.0 38.4 2009 3.9 2009 0.188 Brazil 2015 10.9 23.5 2015 2.8 2015 0.440 Bulgaria 2007 19.2 34.9 2007 7.8 2007 0.384 China 2013 5.0 10.0 2013 1.1 2013 0.286 Colombia 2014 6.5 10.6 2014 0.9 2014 0.378 Costa Rica 2014 8.9 16.9 2014 1.9 2014 0.421 Dominica 2002 4.4 4.8 2002 0.5 2002 0.072 Dominican Rep. 2014 6.1 10.2 2014 1.3 2014 0.257 94 Ecuador 2016 6.9 15.6 2016 2.0 2016 0.430 Fiji 2008 5.7 11.2 2008 1.0 2008 0.211 Gabon 2005 0.7 1.6 2005 0.1 2005 0.139 Georgia 2011 42.6 68.4 2011 19.1 2011 0.335 Guatemala 2011 1.5 4.1 2011 0.4 2011 0.329 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW Iraq 2012 7.9 14.8 2012 1.6 2012 0.212 Jamaica 2010 10.2 9.4 2010 1.2 2010 0.401 Jordan 2010 10.4 24.8 2010 3.0 2010 0.354 Kazakhstan 2015 17.0 32.3 2015 4.9 2015 0.327 Kosovo 2013 7.5 21.8 2013 3.8 2013 0.619 Lebanon .. .. .. .. Malaysia 2008 6.3 13.3 2008 1.3 2008 0.237 Maldives 2009 11.7 27.9 2009 3.9 2009 0.287 Mauritius 2012 36.9 60.9 2012 13.8 2012 0.319 Mexico 2014 13.6 25.3 2014 3.2 2014 0.376 Montenegro 2014 3.9 23.1 2014 2.6 2014 0.527 Namibia .. .. .. .. Paraguay 2014 5.5 8.8 2014 1.0 2014 0.381 Country Social Assistance Impact on Poverty & Inequality Benefit-Cost Ratio Reduction (%) Reduction (%) Year Year Year Ratio Poverty Rate Poverty Gap Gini Peru 2014 2.6 7.7 2014 0.8 2014 0.636 Romania 2012 23.1 43.5 2012 9.3 2012 0.359 Russian Fed. 2016 16.9 25.5 2016 4.8 2016 0.224 Serbia 2015 9.9 27.7 2015 4.6 2015 0.471 South Africa (ASPIRE) 2014 45.7 73.4 2014 10.5 2014 0.338 South Africa (own calc.) .. .. .. .. Sri Lanka 2012 4.1 9.1 2012 0.9 2012 0.324 Thailand 2013 11.5 21.1 2013 2.6 2013 0.300 Turkey 2016 3.1 10.4 2016 1.1 2014 0.434 Venezuela .. .. .. .. 95 SOUTH AFRICA: SOCIAL ASSISTANCE PROGRAMS AND SYSTEM REVIEW 96