CHAPTER 3

Analyzing the Performance of Social Safety Net Programs

This chapter analyzes the performance of social safety net (SSN)/social assistance (SA) programs using four indicators: coverage, beneficiary incidence, level of benefits, and impact on poverty and inequality. To set the stage, the first section of this chapter presents an overview of the coverage of social protection and labor (SPL) programs in general that encompasses SSN/SA programs, social insurance, and labor market programs. Within this framework, this chapter then analyzes the role of SSN programs as a main SPL instrument. Performance indicators provide answers to very important development questions. Coverage indicates what percentage of the total population or a specific population group benefits from SSN programs. Among all possible population groups, the poor have the greatest need for social protection and are particularly the focus of SSNs aimed at assisting the poor. Beneficiary incidence sheds light on how the total number of beneficiaries are distributed along the welfare distribution of the population. The level of benefits indicates the proportion of the benefits with respect to the household’s total income or consumption. Impact on poverty and inequality shows the reduction in the poverty headcount, poverty gap, and inequality (as measured by the Gini index) because of the SPL transfers.

The performance indicators presented in this chapter represent only a first step at monitoring SSN program performance and do not replace results from impact evaluations. Performance indicators from household survey data help monitor SSN programs’ performance over time as information from household surveys becomes available. Given that significant resources are being invested by governments in the implementation of SSN programs, it is important to continuously monitor the effectiveness of these programs and inform social policy.

Performance indicators derived from household surveys assess the effect of the transfers on the welfare of beneficiaries (in terms of income or consumption) and their distributional effects. Impact evaluations, on the other hand, are designed to measure a broader specific set of outcomes, such as the effects on beneficiaries’ level of consumption, production, labor supply, human capital, and risk management (see highlight 1 at the end of this chapter). The effects attributed to the program are evaluated by using a counterfactual to determine the potential outcomes for the beneficiaries in the absence of the program. However, impact evaluations are not conducted very frequently and are not available for all programs; accordingly, household surveys play a crucial role in monitoring programs, given that they are systematically conducted across years and may include a larger set of programs for which impact evaluations are not available. Therefore, performance indicators from household surveys and impact evaluations provide a complementary picture of what social protection programs are achieving.

This chapter first presents evidence on coverage, then on beneficiary incidence, followed by benefit size, and impacts on poverty and inequality. In addition, a highlight focuses on productive outcomes of SSN programs in Sub-Saharan Africa based on impact evaluations. The analysis uses a subset of the most recent household surveys (2008–16) from the Atlas of Social Protection: Indicators of Resilience and Equity (ASPIRE) database, corresponding to 96 countries with information about SSN programs. The full sample of countries is used for coverage and beneficiary incidence indicators; however, for assessing benefits level and impacts on poverty and inequality, only 79 countries are used, for which the monetary value of the transfers is provided in the household surveys. This chapter also provides country-level key performance indicators of SPL programs in appendix F.1 and key performance indicators solely for SSN/SA programs in appendix F.2.

WHO IS COVERED BY SOCIAL PROTECTION AND LABOR PROGRAMS?

Coverage is expressed as the percentage of the population receiving a given type of SPL program. In this analysis, coverage includes direct and indirect beneficiaries (all household members where at least one member receives a benefit). The analysis first presents the global picture of SPL coverage by region and country income group, and then zooms in on country-level coverage rates by the type of SSN program. In discussing the coverage of the poor, the poor are defined as individuals who belong to the bottom 20 percent of the welfare distribution (in terms of household total income or consumption per capita). In all subsequent figures and in appendix F, the pretransfer welfare indicator is used to rank households, except for the indicator that expresses the social transfers as a share of total beneficiary welfare, which includes transfers.

The analysis reveals that SPL programs cover on average 44 percent of the total population. SPL programs cover more than half the population in East Asia and Pacific, Europe and Central Asia, Latin American and the Caribbean, and Middle East North Africa (see figure 3.1).¹ In terms of coverage of the poorest, Europe and Central Asia has the highest coverage of individuals in the poorest quintile (86 percent), followed by Latin American and the Caribbean (76 percent). In all regions, a higher percentage of the poorest quintile (compared to the total population) is covered.

FIGURE 3.1 Share of Total Population and the Poorest Quintile That Receives Any Social Protection and Labor Programs, as Captured in Household Surveys, by Region

Source: ASPIRE database.

Note: The total number of countries per region included in the analysis appears in parentheses. Aggregated indicators are calculated using simple averages of country-level social protection and labor coverage rates across regions. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total social protection and labor coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

The coverage rates presented in the first three figures (see figures 3.1–3.3) are particularly high because they include all types of SPL programs, not only SSN/SA programs. In other words, the figures include any type of social insurance (old-age pension and other social security), active and passive labor market programs, and SSNs (unconditional cash transfers [UCTs], conditional cash transfers [CCTs], social pensions, public works, fee waivers and targeted subsidies, school feeding, in-kind transfers, and other SA programs).² In addition, other methodological factors drive this high coverage. Specifically, the calculation includes direct and indirect beneficiaries, and the poorest quintile is estimated using pre-transfer welfare.

In a sample of countries capturing all country income groups, the average SPL coverage rate of the poorest quintile is 56 percent. The coverage of SPL programs is highly correlated with the countries’ level of income. Figure 3.2 shows that high- and upper-middle-income countries cover 97 percent and 77 percent of the poorest quintile, respectively. In contrast, lower-middle- and low-income countries cover 54 and 19 percent of the poorest quintile, respectively. These coverage figures should be interpreted with caution because coverage rates derived from household surveys are likely to be underestimated.³ As a response to observed coverage gaps for the poor, such initiatives as universal social protection (USP) have emerged (see box 3.1).

FIGURE 3.2 Share of Total Population and the Poorest Quintile That Receives Any Social Protection and Labor Programs, as Captured in Household Surveys, by Country Income Group

Source: ASPIRE database.

Note: The total number of countries per country income group included in the analysis appears in parentheses. Aggregated indicators are calculated using simple averages of country-level social protection and labor coverage rates across country income groups. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total social protection and labor coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pre-transfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

BOX 3.1 Universal Social Protection

The Universal Social Protection (USP) Initiative was launched by the World Bank Group, the International Labour Organisation, and other bilateral and multilateral partners in September 2016 to support the Sustainable Development Goal (SDG) agenda for social protection.^a The USP Initiative aims to join the efforts of the international agencies, donors, and governments in providing social protection for all people in need. Access to adequate social protection is central to ending poverty and boosting shared prosperity. The poorest countries continue to have enormous coverage/adequacy gaps, as the empirical evidence presented in this book clearly suggests.

Countries have many options and pathways to achieve universal social protection. Some countries have opted for an explicit universal coverage of specific population groups (Botswana, Timor-Leste), whereas others have used a more gradual and progressive approach to building up coverage (Brazil, Thailand). Some countries have the principles of universalism (universal rights) embedded in their national constitutions (Bolivia, South Africa), whereas others have pursued those principles without constitutional provisions (Swaziland, Uruguay). Universal social protection is most commonly started with (universal) old-age pensions (see chapter 4), but some countries have opted to make disability, maternity/paternity, and/or child benefits universal (Argentina, Nepal). There are publicly financed child benefit social pensions for all (Mongolia, Namibia) and minimum pensions for those who do not have a contributory pension, ensuring universality (Azerbaijan, China). Some countries strategically use transfers for the poor and vulnerable who could fall further behind (Brazil, Chile, Fiji, and Georgia).

The implementation of the USP Framework emphasizes both depth and breadth of coverage, or vertical and horizontal expansion. The depth of coverage is defined as areas of protection and can include income security, access to insurance and saving instruments, access to essential health care services, and other social services or levels of support (or adequacy). Expansion in the vertical sense means providing more protection to the same covered groups. In terms of horizontal expansion, there are different population groups with respect to the stage in the life cycle (children, working age, and elderly) or level of income (poor, vulnerable, middle class, rich). The evidence suggests that countries tend to gradually expand coverage both vertically and horizontally. The degree of coverage of the poor is highly correlated with the degree of coverage of the general population.

a. See http://www.ilo.org/global/topics/social-security/WCMS_378991/lang--en/index.htm.

In terms of the coverage of the poor, low-income countries lag in all three areas of social protection. Figure 3.3 shows that social insurance programs are more prevalent in high-income countries, covering 60 percent of the poorest quintile; in contrast, in low-income countries only 2 percent of the poorest quintile is covered by this program type. SSN/SA programs account for most SPL program coverage of the poor in all country income groups. Yet, high-income countries report the highest coverage of the poor by SSN programs (76 percent), compared with only 18 percent in low-income countries. Labor market programs cover the poor at a rate of 2 percent in low-income countries and 8 percent in high-income countries.4 SSN programs therefore play a pivotal role in achieving social protection coverage of the poor.⁵ The rest of this chapter focuses on analyzing the performance of SSN instruments in the countries included in the ASPIRE database.

FIGURE 3.3 Share of Poorest Quintile That Receives Any Social Protection and Labor Program, as Captured in Household Surveys, by Type of Social Protection and Labor Area and Country Income Group

Source: ASPIRE database.

Note: The total number of countries per country income group included in the analysis appears in parentheses. Aggregated indicators are calculated using simple averages of country-level coverage rates for social insurance, social assistance, and labor market programs, across country income groups. Indicators do not count for overlap among programs types (people receiving more than one program); therefore, the sum of percentages by type of program may add up to more than 100 percent. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total social protection and labor coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

WHICH TYPES OF SOCIAL SAFETY NET PROGRAMS COVER THE POOR?

Different countries focus on different SSN instruments. There is no one-size-fits-all approach to SSN/SA programs. These noncontributory programs address different issues and target different population groups based on needs and vulnerabilities. Countries generally adopt a combination of SSN/SA programs based on their social policy objectives. This section analyzes the extent to which different SSN/SA programs cover individuals in the poorest quintile. The analysis details the number of countries reporting each program typology in the 96 household surveys included in the ASPIRE database and presents the coverage of the poor (direct and indirect beneficiaries) by these instruments.

To facilitate analysis and cross-country comparisons, the programs are grouped into eight standard SSN categories. Therefore, the coverage indicator corresponds to the aggregated program category and not necessarily to an individual program. For example, Latvia includes 10 programs under UCTs, which embrace the means-tested Guaranteed Minimal Income Allowance, together with more universal child and family benefits. On the other hand, Belize includes only one program, the social welfare transfer, under UCTs. Complete documentation of the programs that are included in each SSN category, per country, is available in the ASPIRE online portal.6 In the discussion that follows, coverage of the poor is presented by each SSN program category, illustrating different patterns in the use of specific SSN interventions, the degree of variation in coverage rates across countries, and the benchmarking of country results against global program averages.⁷

UCTs constitute some of the most popular safety net tools and are included in most household surveys in all regions. They cover 23 percent of the poorest quintile, on average. In the ASPIRE database, UCT programs are reported in the household surveys of 63 countries, compared with 103 countries where the administrative data report having at least one program in this category.⁸ The most UCTs are found in the household surveys for Europe and Central Asia and Sub-Saharan Africa (20 and 17 countries, respectively). Figure 3.4 shows the distribution of UCT programs covering from 0.6 to almost 100 percent of the poorest quintile. Among the programs that achieve almost 100 percent coverage of the poor are the Child Money Program in Mongolia,⁹ and the social transfers in Malaysia (94 percent coverage of the poor), which may reflect the performance of the Bantuan Rakyat 1 Program, the country’s flagship cash transfer program for the poor.10 In Europe and Central Asia, Russian Federation’s cash transfer programs show the largest coverage of the poor (79 percent).

FIGURE 3.4 Share of the Poorest Quintile That Receives Unconditional Cash Transfer Programs, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 63); Europe and Central Asia (n = 20); Sub-Saharan Africa (n = 17); Latin America and the Caribbean (n = 9); Middle East and North Africa (n = 7); East Asia and Pacific (n = 6); and South Asia (n = 4). Unconditional cash transfers include any of the following: poverty alleviation and emergency programs, guaranteed minimum-income programs, and universal or poverty-targeted child and family allowances. They do not include social pensions or targeted subsidies in cash. The average coverage of unconditional cash transfers is estimated as the simple average of these programs’ coverage rates across countries. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

CCTs typically aim to reduce poverty and increase human capital by requiring beneficiaries to comply with conditions such as school attendance and health checkups. The average coverage of the poorest quintile by CCTs in the sample of surveys is 40 percent. Pioneered by Brazil and Mexico in the late 1990s, CCTs spread to other countries in the region and worldwide. Yet, few household surveys outside of Latin America and the Caribbean capture CCT information. Only 19 countries in ASPIRE include information on CCT programs in their household surveys (of which 16 are in Latin America and the Caribbean), compared with 64 programs observed in the administrative database. This instrument covers from 2.4 to 75 percent of the poor (see figure 3.5). Asignaciones Familiares in Uruguay has the largest coverage of the poor (75 percent), followed by Bonos Juancito Pinto and Juana Azurduy in Bolivia (73 percent), and Prospera in Mexico (63 percent). The CCT with large coverage outside Latin America and the Caribbean is the Pantawid Pamilyang Pilipino Program (4Ps) in the Philippines, which covers 60 percent of the poor (see figure 3.5), demonstrating its focus on ensuring coverage of the poorest.

FIGURE 3.5 Share of the Poorest Quintile That Receives Conditional Cash Transfer Programs, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries per region included in the analysis is as follows: total (n = 19); Latin America and the Caribbean (n = 16); East Asia and Pacific (n = 2); South Asia (n = 1); Europe and Central Asia (n = 0); Middle East and North Africa (n = 0); and Sub-Saharan Africa (n = 0). Conditional cash transfer programs include the following: Argentina 2013: Asignación Universal por Hijo. Bangladesh 2010: Maternity allowance; Program for the Poor Lactating; Stipend for Primary Students (MOPMED); Stipend for Dropout Students; Stipend for Secondary and Higher Secondary/Female Student. Bolivia 2012: Bono Juancito Pinto; and Bono Juana Azurduy. Brazil 2015: Bolsa Família. Chile 2013: Subsidio Familiar (SUF); Bono de Protección Familiar y de Egreso; Bono por control del niño sano; Bono por asistencia escolar; and Bono por logro escolar. Colombia 2014: Familias en Acción. Costa Rica 2014: Avancemos. Dominican Republic 2014: Solidaridad Program and other transfers. Ecuador 2016: Bono de Desarrollo Humano. Guatemala 2014: Programa Mi Bono Seguro. Honduras 2013: Asignaciones Familiares– Bonos PRAF and otro tipo de bonos. Jamaica 2010: Program of Advancement Through Health and Education (PATH)—child 0–71 months; 6–17 years; and pregnant and lactating women. Mexico 2012: Oportunidades. Panama 2014: Red de Oportunidades. Paraguay 2011: Tekopora. Peru 2014: Programa Juntos. Philippines 2015: Pantawid Pamilyang Pilipino Program (4Ps). Timor Leste 2011: Bolsa da Mae. Uruguay 2012: Asignaciones Familiares. The average coverage of conditional cash transfers is estimated as the simple average of these programs’ coverage rates across countries. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

For individuals who do not have access to social insurance benefits, social pensions aim to overcome loss of income because of old age, disability, or death of the bread winner. In the sample of countries, social pensions cover, on average, 20 percent of the poorest quintile. Social pensions presented here include noncontributory disability and survivor pensions, and thus represent a broader category than old-age social pensions (featured in chapter 4). Only 36 countries in the ASPIRE household survey database capture any form of social pensions, compared with 75 countries in the administrative database. Most surveys with social pension information are found in Europe and Central Asia (n = 13) and Latin America and the Caribbean (n = 10). In the sample, social pensions cover between 0.6 and 81 percent of individuals in the poorest quintile (see figure 3.6). Georgia has the highest coverage of the poorest quintile because of its universal old-age social pensions. A few countries in Africa have extensive coverage of the poorest quintile, such as Mauritius (79 percent) and South Africa (62 percent). In all these countries, formal social insurance has low coverage and social pensions constitute the main form of social protection for the elderly. Thailand’s social pensions have high coverage of the lowest quintile (58 percent), also driven by providing support to the elderly and disabled, taking into account their living arrangements within extended families.¹¹ On the opposite extreme, some countries in Europe and Central Asia have extended social insurance systems; therefore, social pensions cover only those who do not benefit from social insurance, which constitute rather narrow population groups. Hence the coverage is low (Latvia, Montenegro, Russian Federation, and Serbia).

FIGURE 3.6 Share of the Poorest Quintile That Receives Social Pensions, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 36); Europe and Central Asia (n = 13); Latin America and the Caribbean (n = 10); Sub-Saharan Africa (n = 7); South Asia (n = 4); East Asia and Pacific (n = 2); and Middle East and North Africa (n = 0). Social pensions include any of the following: noncontributory old-age pensions; disability pensions; and survivor pensions. Social pensions average coverage is the simple average of social pensions coverage rates across countries. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

Public works programs typically condition the transfer on participating in a community project/activity. Very few public works programs are captured in the sample of household surveys, and their coverage of the poorest quintile is limited, at 11 percent. Public works are implemented for many reasons, such as to provide employment of last resort or to mitigate covariate and idiosyncratic shocks. Public works programs include cash-, food-, and inputs-for-work (Andrews et al. 2012). They are more often implemented in Sub-Saharan Africa and South Asia, although respective information/data are not often captured in household surveys. Only 9 countries in ASPIRE have specific information about public works in the household surveys, as compared with 96 countries in the administrative database. In these countries, public works cover between 1 and 27 percent of the poorest quintile (figure 3.7); the largest coverage rates are observed for the MGNREG program in India (27 percent), a flagship national social safety net program with a history going back decades; and MASAF in Malawi (21 percent), the social fund program that has become the cornerstone of the national SSN system.

FIGURE 3.7 Share of the Poorest Quintile That Receives Public Works, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 9); Sub-Saharan Africa (n = 4); Latin America and the Caribbean (n = 2); South Asia (n = 3); East Asia and Pacific (n = 0); Europe and Central Asia (n = 0); and Middle East and North Africa (n = 0). Public works programs included: Afghanistan 2011: Cash-for-work programs; food-for-work programs; or income-generating program/projects. Argentina 2013: Plan de Empleo. Ethiopia 2010: Productive Safety Net Program (PSNP). India 2011: MGNREG and other public works. Malawi 2013: MASAF; PWP; Inputs-for-work program. Mexico 2012: Programa de Empleo Temporal (PET). Nepal 2010: Rural Community Infrastructure Works Program (RCIW) and other food-for-work and cash for work programs. Niger 2014: Public works. Rwanda 2013: Public works from the Vision 2020 Umurenge Program. Public works average coverage is the simple average of public works coverage rates across countries. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

Fee waivers and targeted subsidies typically subsidize services or provide access to low-priced food staples to the poor. They are common but generally provide limited coverage of the poorest quintile—13 percent, on average, in the sample of countries. However, because this category is not easily collected by household surveys, this average is most likely a considerable underestimate. Services under this category usually relate to education, health, housing, transportation, or utilities. When a beneficiary is exempt from payment for such services and the cost is borne by the government program, such fee waivers provide conditional support for the targeted group using a specific service.¹² Out of 82 countries with information on fee waivers and targeted subsidies in the administrative database, only 22 are observed in the household survey data. Among the regions, Europe and Central Asia (n = 10 countries) and Latin America and the Caribbean countries (n = 8 countries) capture this typology of programs in the surveys most often. The program covers between 0.4 and 56 percent of the poor (see figure 3.8). Coverage rates for targeted subsidies and fee waivers in Europe and Central Asia tend to be smaller because these programs focus only on a subset of the poor, but tend to include several different forms of benefits (for example, subsidized housing; and fee waivers for kindergartens, health care, public transportation).

FIGURE 3.8 Share of the Poorest Quintile That Receives Fee Waivers and Targeted Subsidies, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 22); Europe and Central Asia (n = 10); Latin America and the Caribbean (n = 8); Sub-Saharan Africa (n = 2); East Asia and Pacific (n = 2); Middle East and North Africa (n = 0); and South Asia (n = 0). Fee waivers and targeted subsidies include any of the following: energy products; education; utilities; housing or transportation fees waivers to specific households; or such fees discounted below the market cost. They do not include health fee waivers/subsidies, except for Zimbabwe. Fee waivers and targeted subsidies program coverage is the simple average of these programs’ coverage rates across countries. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

School feeding programs provide meals to students generally in poor and food-insecure areas, with the aim of improving nutrition, health, and educational outcomes. In the sample, these programs are found, on average, to benefit a significant share of the poor—37 percent. Even though school feeding is a common safety net program, it is not always captured in household surveys. Of the 117 countries reporting school feeding in administrative data, only 26 countries with this program are found in the household survey data. Of these 26 countries, 15 are in Latin America and the Caribbean. Figure 3.9 shows that coverage varies across countries from 2.21 to 86 percent of the poor. Such variation is more likely to reflect the differences in survey designs and efforts to capture this form of benefit than real differences across countries. The programs with the largest coverage of the poor, among the countries with adequate data, are in Botswana (86 percent), Bolivia (73 percent), El Salvador (69 percent), Nicaragua (67 percent), and Honduras and Panama (66 percent).

FIGURE 3.9 Share of the Poorest Quintile That Receives School Feeding Programs, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 26); Latin America and the Caribbean (n = 15); Sub-Saharan Africa (n = 5); Europe and Central Asia (n = 3); Middle East and North Africa (n = 1); East Asia and Pacific (n = 1); and South Asia (n = 1). School feeding programs encompass any type of meals or food items provided at school. School feeding average coverage is the simple average of school feeding coverage rates across countries. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

In-kind transfers consist of food rations, clothes, school supplies, shelter, fertilizers, seeds, agricultural tools or animals, and building materials, among others. They are a very common SSN instrument, and in the sample cover, on average, 27 percent of the poorest quintile. Their objectives are usually to provide food security, improve nutrition, increase agricultural productivity, and deliver emergency relief. Forty-five countries in the ASPIRE household survey database capture information on in-kind transfers (see figure 3.10), compared with 90 countries in the administrative database. About one-third of the surveys (14 out of 45) reporting in-kind transfers are in countries in Sub-Saharan Africa, where they typically consist of programs to promote agricultural productivity or emergency relief. However, programs report larger coverage of the poor in Latin America and the Caribbean and Middle East and North Africa than in Sub-Saharan Africa. Supplemental food programs for children, pregnant and nursing women, and the elderly are common in Latin America and the Caribbean, as well as school supplies and uniforms. In a few countries, these programs cover a high percentage of the poor, including Peru (84 percent), Ecuador (74 percent), and El Salvador and Paraguay (70 percent). Coverage in Peru is particularly large because the in-kind category encompasses seven in-kind programs captured in the survey, including nutritional programs, school supplies, uniforms, and shoes and laptops to school children. In Middle East and North Africa, in-kind transfers take the form of food and in-kind aid. They cover about 81 percent of the poor in Iraq and 56 percent in the Arab Republic of Egypt (see figure 3.10). In East Asia and Pacific, Indonesia’s Rastra Program (rice subsidies) has the largest coverage of the poor (71 percent).¹³

FIGURE 3.10 Share of the Poorest Quintile That Receives In-Kind Transfers, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 45); Sub-Saharan Africa (n = 14); Latin America and the Caribbean (n = 11); Europe and Central Asia (n = 9); Middle East and North Africa (n = 6); East Asia and Pacific (n = 3); and South Asia (n = 2). In-kind transfers include any of the following: food aid; agricultural inputs; clothes; school supplies; and building materials. In-kind transfers average coverage is the simple average of in-kind transfer coverage rates across countries. Coverage is determined as follows: (number of individuals in the total population or poorest quintile who live in a household where at least one member receives the transfer)/(number of individuals in the total population). This figure underestimates total coverage because household surveys do not include all programs that exist in each country. The poorest quintile is calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

WHAT IS THE BENEFICIARY INCIDENCE OF VARIOUS SOCIAL SAFETY NET INSTRUMENTS?

Beneficiary incidence indicates to what extent a given population group benefits from a program. For this analysis, individuals are ranked according to their position in the welfare distribution, based on quintiles of per capita pretransfer income or consumption; the proportion of program beneficiaries belonging to each quintile is presented. Another corresponding indicator is benefits incidence, which shows the proportion of program benefits transferred to individuals in each quintile.14 A program is considered propoor if more than 20 percent of its total beneficiaries belong to the bottom 20 percent of the distribution (or if more than 40 percent of its total beneficiaries belong to the bottom 40 percent of the distribution).¹⁵ Beneficiary incidence and benefits incidence help determine which population groups are benefiting from the program, and thus are useful indicators to analyze the performance of SPL programs. Propoor beneficiary and benefits incidence is the only way to ensure that a program within a given budget achieves greater impact in terms of poverty reduction (Yemtsov et al., forthcoming).

The beneficiary incidence analysis conducted by type of SSN instrument reveals that, on average, all types of SSN programs tend to be propoor or favor the poor and near-poor. That is, a higher percentage of beneficiaries belong to the first and second poorest quintiles. This is illustrated in figure 3.11, where the lines representing each SSN instrument show a similar downward slope. CCTs generally show a more propoor distribution compared with the other SSN instruments, which is not surprising because these programs typically target poor households. Figure 3.11 shows that, among the observed programs, 45 percent of CCT beneficiaries are in the poorest quintile on average, while only 4 percent are in the richest quintile. Between 33 and 37 percent of beneficiaries of the other SSN instruments, on average, belong to the poorest quintile, which indicates that those instruments are still propoor.

FIGURE 3.11 Global Distribution of Beneficiaries by Type of Social Safety Net Instrument, as Captured in Household Surveys, by Quintile of Pretransfer Welfare

Source: ASPIRE database.

Note: The total number of countries where the social safety net instrument is captured in household surveys is as follows: unconditional cash transfers (n = 63), conditional cash transfers (n = 19); social pensions (n = 36); public works (n = 9); fee waivers and targeted subsidies (n = 22); school feeding (n = 26); and in-kind transfers (n = 45). Beneficiaries’ incidence is calculated as follows: (number of direct and indirect beneficiaries [people who live in a household where at least one member receives the transfer] in a given quintile)/(total number of direct and indirect beneficiaries). The sum of percentages across quintiles per given instrument equals 100 percent. Aggregated indicators are calculated using simple averages of program instrument beneficiaries’ incidence rates across countries. Quintiles are calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; Q = quintile.

The analysis of beneficiary incidence by SSN instrument across countries shows the presence of highly propoor programs in each program category. Figures 3.12 through 3.18 illustrate that despite wide variation in beneficiary distribution across countries, most of them favor the poor and near-poor, with more than 20 percent of the total beneficiaries belonging to the poorest quintile. However, for every SSN category, there are also examples of programs that are proportionally distributed or even favor the rich more than the poor and the middle class. It is not always possible to draw general conclusions about the distribution of beneficiaries and benefits of a specific program without knowing detailed information on the program’s design, eligibility criteria, and implementation. Some programs may not be propoor by design; for example, they may not be addressed specifically to the poor but to the general population (that is, in the case of universal programs). Or their eligibility criteria may be categorical (for example, in terms of disability, ethnicity, and war victims) and not means-tested. In those cases, beneficiaries meeting the categorical requirements may not belong to the poor.

FIGURE 3.12 Distribution of Unconditional Cash Transfer Beneficiaries, as Captured in Household Surveys, by Quintile of Pretransfer Welfare

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 63); Sub-Saharan Africa (n = 17); Europe and Central Asia (n = 20); Latin America and the Caribbean (n = 9); Middle East and North Africa (n = 7); East Asia and Pacific (n = 6); and South Asia (n = 4). Unconditional cash transfers include any of the following: poverty alleviation and emergency programs; guaranteed minimum-income programs; and universal or poverty-targeted child and family allowances. They do not include social pensions or targeted subsidies in cash. Beneficiaries’ incidence is calculated as follows: (number of direct and indirect beneficiaries [people who live in a household where at least one member receives the transfer] in a given quintile)/(total number of direct and indirect beneficiaries). The sum of percentages across quintiles per given instrument equals 100 percent. Quintiles are calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; Q = quintile.

UCTs are characterized by a wide range of beneficiary incidence across countries, but on average, 37 percent of UCT beneficiaries belong to the poorest quintile and 23 to the second-poorest quintile (figure 3.12). One reason for this large variation of results across countries is the fact that ASPIRE—in some countries—aggregates many types of UCTs, and the respective programs may have very different objectives and eligibility criteria. However, for other countries, this category captures only one program, making it easier to interpret the observed beneficiary incidence results. Universal programs have an even distribution. For instance, for the Child Money Program in Mongolia, the participation of all five quintiles is close to 20 percent (see figure 3.12). In contrast, the programs in Kosovo and Montenegro are much more focused on the poorest quintile, which accounts for more than 70 percent of the total UCT beneficiaries.¹⁶ At the same time, in Uruguay, only 7 percent of the beneficiaries of Prima por Hogar Constituido (Transfer for Constituted Household) belong to the poorest quintile. This program is provided only to public servants who are married or have dependents whose monthly gross salary is less than the sum of two national minimum wages. This explains why the program is not propoor by design.

In general, CCTS are, as has been mentioned already, more propoor than other SSN program types. Among all 19 CCT programs included in the ASPIRE database, the average beneficiary incidence is 45 percent for the poorest quintile and 26 percent for the second-poorest quintile. Panama’s conditional cash transfer program, Red de Oportunidades, has an incidence of participants from the poorest quintile of 75 percent. More than 65 percent of the beneficiaries of Programa Juntos in Peru, belong to the poorest 20 percent. Timor-Leste’s Bolsa da Mae has the lowest beneficiary incidence rate for the poorest quintile, at 21 percent (see figure 3.13).

FIGURE 3.13 Distribution of Conditional Cash Transfer Beneficiaries, as Captured in Household Surveys, by Quintile of Pretransfer Welfare

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 19); Latin America and the Caribbean (n = 16); East Asia and Pacific (n = 2); South Asia (n = 1); Europe and Central Asia (n = 0); Middle East and North Africa (n = 0); and Sub-Saharan Africa (n = 0). Conditional cash transfer programs include the following: Argentina 2013: Asignación Universal por Hijo. Bangladesh 2010: Maternity allowance, Program for the Poor Lactating, Stipend for Primary Students (MOPMED), Stipend for Drop Out Students, Stipend for Secondary and Higher Secondary/Female Student. Bolivia 2012: Bono Juancito Pinto and Bono Juana Azurduy. Brazil 2015: Bolsa Família. Chile 2013: Subsidio Familiar (SUF), Bono de Protección Familiar y de Egreso, Bono por control del niño sano, Bono por asistencia escolar and Bono por logro escolar. Colombia 2014: Familias en Acción. Costa Rica 2014: Avancemos. Dominican Republic 2014: Solidaridad Program and other transfers. Ecuador 2016: Bono de Desarrollo Humano. Guatemala 2014: Programa Mi Bono Seguro. Honduras 2013: Asignaciones Familiares–Bonos PRAF and otro tipo de bonos. Jamaica 2010: Program of Advancement Through Health and Education (PATH)—child (0–71 months); 6–17 years; and pregnant and lactating women. Mexico 2012: Oportunidades. Panama 2014: Red de Oportunidades. Paraguay 2011: Tekopora. Peru 2014: Programa Juntos. Philippines 2015: Pantawid Pamilyang Pilipino Program (4Ps). Timor-Leste 2011: Bolsa da Mae. Uruguay 2012: Asignaciones Familiares. Beneficiaries’ incidence is calculated as follows: (number of direct and indirect beneficiaries [people who live in a household where at least one member receives the transfer] in a given quintile)/(total number of direct and indirect beneficiaries). The sum of percentages across quintiles per given instrument equals 100 percent. Quintiles are calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; Q = quintile.

Social pensions also have a very propoor distribution of beneficiaries. An average of 35 percent of beneficiaries belong to the poorest quintile and 22 percent to the second-poorest quintile. The State Social Maintenance Benefit in Latvia has the highest proportion of beneficiaries belonging to the poorest quintile (59 percent). In contrast, Programa Adulto Mayor in Guatemala and Bono por Tercera Edad in Honduras have only 8 and 9 percent of their beneficiaries coming from the poorest quintile, respectively (see figure 3.14). Part of the propoor performance shown by some countries may be related to the way the pretransfer indicator is constructed. If social pensions cover a sizable part of the poor and their benefit level is high, most beneficiaries may tend to depend on them and hence group in the lowest quintile of the welfare distribution once such transfers are removed. On the other hand, social pensions in Honduras tend to cluster further up the income distribution, reflecting this instrument’s bias in coverage toward richer areas of the country.

FIGURE 3.14 Distribution of Social Pensions Beneficiaries, as Captured in Household Surveys, by Quintile of Pretransfer Welfare

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 36); Europe and Central Asia (n = 13); Latin America and the Caribbean (n = 10); Sub-Saharan Africa (n = 7); South Asia (n = 4); East Asia and Pacific (n = 2); and Middle East and North Africa (n = 0). Social pensions include any of the following: noncontributory old-age pensions; disability pensions; and survivor pensions. Beneficiaries’ incidence is calculated as follows: (number of direct and indirect beneficiaries [people who live in a household where at least one member receives the transfer] in a given quintile)/(total number of direct and indirect beneficiaries). The sum of percentages across quintiles per given instrument equals 100 percent. Quintiles are calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; Q = quintile.

The beneficiary incidence of public works for the poorest quintile is 33 percent, on average. Yet, when looking at the two bottom quintiles, all 10 public works programs analyzed have a beneficiary incidence rate of at least 45 percent for the poorest 40 percent, which still makes it a somewhat propoor instrument. Public works are limited only to households with able-bodied, unemployed members who are willing to work. Because many public works rely on self-selection and are oversubscribed, especially in low-income countries (more people want to work in these programs than they have employment positions), there is a lot of sharing and capture of the program by the not so poor. Mexico’s Programa de Empleo Temporal shows the most propoor distribution, with 49 percent of beneficiaries coming from the poorest quintile (see figure 3.15).

FIGURE 3.15 Distribution of Public Works Beneficiaries, as Captured in Household Surveys, by Quintile of Pretransfer Welfare

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 9); Sub-Saharan Africa (n = 4); Latin America and the Caribbean (n = 2); South Asia (n = 3); East Asia and Pacific (n = 0); Europe and Central Asia (n = 0); and Middle East and North Africa (n = 0). Public works programs include the following: Afghanistan 2011: Cash-for-work, food-for-work programs, or income-generating program/projects. Argentina 2013: Plan de Empleo. Ethiopia 2010: Productive Safety Net Program (PSNP). India 2011: MGNREG and other public works. Malawi 2013: MASAF, PWP, Inputs-for-work program. Mexico 2012: Programa de Empleo Temporal (PET). Nepal 2010: Rural Community Infrastructure Works Program (RCIW) and other food-for-work and cash for work programs. Niger 2014: Public works. Rwanda 2013: Public works from the Vision 2020 Umurenge Program. Beneficiaries’ incidence is calculated as follows: (number of direct and indirect beneficiaries [people who live in a household where at least one member receives the transfer] in a given quintile)/(total number of direct and indirect beneficiaries). The sum of percentages across quintiles per given instrument equals 100 percent. Quintiles are calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; Q = quintile.

Fee waivers and targeted subsidies have a somewhat flatter beneficiary incidence along the welfare distribution, compared with other instruments. On average, 33 percent of their beneficiaries belong to the poorest quintile and 24 percent to the second-poorest quintile. In Panama, most of the beneficiaries (82 percent) of food supplements and agricultural subsidies belong to the poorest quintile, followed by Vietnam, where nearly three-quarters (73 percent) of beneficiaries of housing, petroleum, and kerosene subsidies and tuition fee exceptions are in the poorest quintile. In South Africa, by contrast, the rich capture most of these benefits (see figure 3.16).

FIGURE 3.16 Distribution of Fee Waivers and Targeted Subsidies Beneficiaries, as Captured in Household Surveys, by Quintile of Pretransfer Welfare

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 22); Europe and Central Asia (n = 10); Latin America and the Caribbean (n = 8); Sub-Saharan Africa (n = 2); East Asia and Pacific (n = 2); Middle East and North Africa (n = 0); and South Asia (n = 0). Fee waivers and targeted subsidies include any of the following: energy products; education; utilities; housing or transportation fees waivers to specific households; or such fees discounted below the market cost. They do not include health benefits or subsidies, except for Zimbabwe. Beneficiaries’ incidence is calculated as follows: (number of direct and indirect beneficiaries [people who live in a household where at least one member receives the transfer] in a given quintile)/(total number of direct and indirect beneficiaries). The sum of percentages across quintiles per given instrument equals 100 percent. Quintiles are calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; Q = quintile.

Beneficiary incidence of school feeding programs varies substantially by country. On average, 34 percent of beneficiaries of school feeding programs belong to the poorest quintile and 24 percent to the second-poorest quintile (see figure 3.17). The most propoor programs are found in the Slovak Republic and Latvia, where 82 and 73 percent of the beneficiaries are drawn from the poorest quintile, respectively. All these programs are rather narrow in coverage, and stand out from a typical universal school feeding program, with different objectives that are often not focused on alleviating poverty. In addition, in universal programs with high coverage, the incidence of school feeding will depend on the access to schooling across quintiles, which may be skewed in favor of the nonpoor, and is not a design feature of the program itself.

FIGURE 3.17 Distribution of School Feeding Beneficiaries, as Captured in Household Surveys, by Quintile of Pretransfer Welfare

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 26); Latin America and the Caribbean (n = 15); Sub-Saharan Africa (n = 5); Europe and Central Asia (n = 3); Middle East and North Africa (n = 1); East Asia and Pacific (n = 1); and South Asia (n = 1). School feeding programs encompass any type of meals or food items provided at school. Beneficiaries’ incidence is calculated as follows: (number of direct and indirect beneficiaries [people who live in a household where at least one member receives the transfer] in a given quintile)/(total number of direct and indirect beneficiaries). The sum of percentages across quintiles per given instrument equals 100 percent. Quintiles are calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; Q = quintile.

In-kind transfers are generally quite propoor. In the sample of observed programs, 34 percent of beneficiaries of in-kind transfers belong to the poorest quintile and 23 percent to the second-poorest quintile. Food aid in Djibouti and food and nutritional programs in Uruguay have especially propoor distributions, with 71 and 70 percent of the recipients belonging to the poorest quintile, respectively (see figure 3.18).¹⁷

FIGURE 3.18 Distribution of In-Kind Transfer Beneficiaries, as Captured in Household Surveys, by Quintile of Pretransfer Welfare

Source: ASPIRE database.

Note: The number of countries per region is as follows: total (n = 45); Sub-Saharan Africa (n = 14); Latin America and the Caribbean (n = 11); Europe and Central Asia (n = 9); Middle East and North Africa (n = 6); East Asia and Pacific (n = 3); and South Asia (n = 2). In-kind transfers include any of the following: food aid; agricultural inputs; clothes; school supplies; and building materials. Beneficiaries’ incidence is calculated as follows: (number of direct and indirect beneficiaries [people who live in a household where at least one member receives the transfer] in a given quintile)/(total number of direct and indirect beneficiaries). The sum of percentages across quintiles per given instrument equals 100 percent. Quintiles are calculated using per capita pretransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; Q = quintile.

WHAT ARE THE BENEFIT LEVELS OF SOCIAL SAFETY NET PROGRAMS?

The level of benefits is measured by two indicators: per capita average transfer (monetary value) and share of benefits with respect to per capita household income or consumption (adequacy ratio).18 The per-capita average transfer is an absolute measure of benefit size and can be compared with social minimums, such as the poverty line or minimum wage. Benefits as a share of household welfare (income or consumption), on the other hand, are a relative measure that allows the importance of the transfers in proportion to household per capita welfare to be assessed.

The level of benefits is set to achieve program objectives within budget constraints. There are no standard rules to set benefits levels, given that they need to be calibrated to fulfill program objectives and meet budget constraints. Contributory old-age pensions, for example, are based on the amount of contributions individuals make during their active working life. In poverty reduction programs, the size of the transfers may be calibrated to reduce the poverty gap of the target population; programs that aim to address food security will set their benefits to meet nutritional needs.¹⁹ For example, Kenya’s Hunger Safety Net Program sets transfer levels based on the five-year average price of cereals, whereas Zambia’s Social Cash Transfer Program sets its benefits close to the price of a 50-pound bag of maize (corn) monthly, which would allow a household to eat a second meal each day (Schüring 2010; Garcia and Moore 2012).

The size of the benefit is a determining factor to achieve positive impacts on household well-being. The way program administrators set benefit levels varies across countries; some programs use flat benefits, whereas others adjust benefits based on household size, number of dependents, and so forth. Flat benefits raise the issue that the per capita transfer will decrease with household size and program impacts will vary across beneficiary households. Inflation is another factor that can erode the real value of the transfers over time, unless regular adjustment mechanisms are applied. In Kenya’s Cash Transfer Program for Orphans and Vulnerable Children, for example, the value of the transfer decreased by almost 60 percent because of inflation between 2007 to 2011 (Daidone et al. 2016). Therefore, having flexibility to adjust benefit levels is important to maximize positive impacts on household well-being, granted there is budget space to do so.

How large should a transfer be to achieve meaningful impacts? The higher the share of the transfers with respect to household welfare, the greater the impacts on poverty and inequality are likely to be.20 Given budget constraints, however, larger transfers may imply fewer beneficiaries. There are concerns that larger transfer values can create disincentives to work. However, most impact evaluations have found that transfers in general do not reduce labor supply, but they do influence the allocation of labor and time. Therefore, determining the size of the transfers is usually a delicate balance between the benefit amount needed to achieve objectives, needs for program coverage, and the available budget. Given these considerations, the evidence gathered from impact evaluations suggests that most successful cash transfers programs, for example, transfer at least 20 percent of household consumption to beneficiaries (Handa et al. 2013).

Empirical analysis indicates that the benefit level expressed as a share of beneficiary welfare among recipients varies greatly across SPL areas. Figures 3.19 and 3.20 show that on average the benefit level for social insurance programs is greater than the benefit level for SSN programs. This is expected because social insurance programs are designed to replace beneficiaries’ working earnings. The global average for social insurance programs as a share of beneficiary welfare is 32 percent for the total population, whereas the share of SSN benefits is only 10 percent (see figure 3.19).²¹

FIGURE 3.19 Social Protection and Labor Transfer Value Captured in Household Surveys, as a Share of Beneficiaries’ Posttransfer Welfare among the Total Population

Source: ASPIRE database.

Note: The number of countries per country income group with monetary values for social assistance is as follows: total (n = 79), high-income countries (n = 5), upper-middle-income countries (n = 30), lower-middle-income countries (n = 30), and low-income countries (n = 14). The number of countries with monetary values for social insurance is as follows: total (n = 79), high-income countries (n = 6), upper-middle-income countries (n = 28); lower-middle-income countries (n = 29), and low-income countries (n = 16). Transfers as a share of a beneficiary’s welfare can be generated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. Labor market programs were not included because they encompass mostly active labor market programs for which only participatory variables (vs. monetary) are observed in the surveys. The sample of countries that include monetary variables (mostly for unemployment insurance) is too small to derive any meaningful conclusion (n = 18). The share of transfers is determined as follows: (transfer amount received by all direct and indirect beneficiaries in a population group)/(total welfare aggregate of the direct and indirect beneficiaries in that population group). Aggregated indicators are calculated using simple averages of country-level social assistance and social insurances transfers’ shares, across country income groups. ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

The proportion of social insurance benefits with respect to beneficiary welfare, for the total population, increases with income, from 18 percent in low-income countries to 49 percent in high-income countries. This is not the case for SSN programs, for which the relative level of benefits differs less accros country income groups; SSN benefits as share of beneficiary welfare is 7 percent for high-income countries, for example, whereas for low-income countries and upper-middle-income countries it is 11 percent (see figure 3.19). The relatively high level of SSN benefits for low-income countries could also reflect the fact that many of the SSN programs in those countries are donor/externally funded. However, the available data do not make it possible to test this hypothesis.

SPL transfers make up a significant proportion of the welfare of individuals in the poorest quintile. For the surveys included in ASPIRE, the average share of the transfers in the welfare of the poorest quintile is 46 percent for social insurance and 19 percent for SSNs (see figure 3.20), compared with 32 and 10 percent observed in the total population (see figures 3.19). For the poor, the share of social insurance benefits is still higher than the share of SSNs, but there is not a clear correlation between the magnitude of this share and country income groups. For example, social insurance makes up 48 percent of beneficiary welfare in low- and upper-middle-income countries, higher than the share observed for lower-middle-income countries (39 percent). Likewise, SSN programs make up 22 percent of beneficiary welfare in upper-middle-income countries, which is higher than the share in lower-middle- and high-income countries (18 percent) (see figure 3.20).

FIGURE 3.20 Social Protection and Labor Transfer Value Captured in Household Surveys, as a Share of Beneficiaries’ Posttransfer Welfare among the Poorest Quintile

Source: ASPIRE database.

Note: The number of countries per country income group with monetary values for social assistance is as follows: total (n = 79), high-income countries (n = 5), upper-middle-income countries (n = 30), lower-middle-income countries (n = 30), and low-income countries (n = 14). The number of countries with monetary values for social insurance is as follows: total (n = 79), high-income countries (n = 6), upper-middle-income countries (n = 28), lower-middle-income countries (n = 29), and low-income countries (n = 16). Transfers as a share of a beneficiary’s welfare can be generated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. Labor market programs were not included because they encompass mostly active labor market programs for which only participatory variables (vs. monetary) are observed in the surveys. The sample of countries that include monetary variables (mostly for unemployment insurance) is too small to derive any meaningful conclusion (n = 18). The share of transfers is determined as follows: (transfer amount received by all direct and indirect beneficiaries in a population group)/(total welfare aggregate of the direct and indirect beneficiaries in that population group). Aggregated indicators are calculated using simple averages of country-level social assistance and social insurance transfers’ shares, across country income groups. The poorest quintile is calculated using per capita posttransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

On average, SSN transfers account for 19 percent of the welfare of the poorest quintile. However, transfer levels vary greatly across SSN instruments and across countries. These differences reflect, in part, different program objectives and the degree of transfer values captured in household surveys. Figures 3.21 to 3.25 illustrate the proportion of SSN benefits with respect to beneficiary welfare by type of SSN instrument and by country.²² Social pensions and CCTs tend to make up, on average, a higher proportion of beneficiary welfare—while the share is much lower for public works, fee waivers, and targeted subsidies. However, this indicator varies greatly across countries within each SSN instrument.

FIGURE 3.21 Unconditional Cash Transfer Value Captured in Household Surveys, as a Share of Beneficiaries’ Posttransfer Welfare among the Poorest Quintile

Source: ASPIRE database.

Note: The number of countries per region with monetary values for unconditional cash transfers is as follows: total (n = 52); Europe and Central Asia (n = 20); Sub-Saharan Africa (n = 10); Latin America and the Caribbean (n = 8); East Asia and Pacific (n = 5); Middle East and North Africa (n = 5); and South Asia (n = 4). Unconditional cash transfers include any of the following: poverty alleviation and emergency programs; guaranteed minimum-income programs; and universal or poverty-targeted child and family allowances. They do not include social pensions or targeted subsidies in cash. Transfers as a share of a beneficiary’s welfare can be generated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. The share of transfers is determined as follows: (transfer amount received by all direct and indirect beneficiaries in a population group)/(total welfare aggregate of the direct and indirect beneficiaries in that population group). Unconditional cash transfer average share is the simple average of unconditional cash transfers’ shares across countries. The poorest quintile is calculated using per capita posttransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

FIGURE 3.22 Conditional Cash Transfer Value Captured in Household Surveys, as a Share of Beneficiaries’ Posttransfer Welfare among the Poorest Quintile

Source: ASPIRE database.

Note: The number of countries per region with monetary values for conditional cash transfers is as follows: total (n = 17), Latin America and the Caribbean (n = 14), East Asia and Pacific (n = 2), South Asia (n = 1), Sub-Saharan Africa (n = 0), Middle East and North Africa (n = 0), and Europe and Central Asia (n = 0). Conditional cash transfer programs include the following: Argentina 2013: Asignación Universal por Hijo. Bangladesh 2010: Maternity allowance, Program for the Poor Lactating, Stipend for Primary Students (MOPMED), Stipend for drop out students, Stipend for Secondary and Higher Secondary/Female Student. Bolivia 2012: Bono Juancito Pinto and Bono Juana Azurduy. Brazil 2015: Bolsa Família. Chile 2013: Subsidio Familiar (SUF), Bono de Protección Familiar y de Egreso, Bono por control del niño sano, Bono por asistencia escolar and Bono por logro escolar. Colombia 2014: Familias en Acción. Costa Rica 2014: Avancemos. Dominican Republic 2014: Solidaridad Program and other transfers. Ecuador 2016: Bono de Desarrollo Humano. Honduras 2013: Asignaciones Familiares–Bonos PRAF and otro tipo de bonos. Jamaica 2010: Program of Advancement Through Health and Education (PATH)—Child 0–71 months, 6–17 years and pregnant and lactating women. Mexico 2012: Oportunidades. Panama 2014: Red de Oportunidades. Paraguay 2011: Tekopora. Peru 2014: Programa Juntos. Philippines 2015: Pantawid Pamilyang Pilipino Program (4Ps). Timor-Leste 2011: Bolsa da Mae. Transfers as a share of a beneficiary’s welfare can be generated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. The share of transfers is determined as follows: (transfer amount received by all direct and indirect beneficiaries in a population group)/(total welfare aggregate of the direct and indirect beneficiaries in that population group). Conditional cash transfer average share is the simple average of conditional cash transfers’ shares across countries. The poorest quintile is calculated using per capita posttransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

FIGURE 3.23 Social Pensions’ Value Captured in Household Surveys, as a Share of Beneficiaries’ Posttransfer Welfare among the Poorest Quintile

Source: ASPIRE database.

Note: The number of countries per region with monetary values for social pensions is as follows: total (n = 30), Europe and Central Asia (n = 13), Latin America and the Caribbean (n = 8), South Asia (n = 4), Sub-Saharan Africa (n = 4), East Asia and Pacific (n = 1), and Middle East and North Africa (n = 0). Social pensions include any of the following: noncontributory old-age pensions; disability pensions; and survivor pensions. Transfers as a share of a beneficiary’s welfare can be generated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. The share of transfers is determined as follows: (transfer amount received by all direct and indirect beneficiaries in a population group)/(total welfare aggregate of the direct and indirect beneficiaries in that population group). Social pensions average share is the simple average of social pensions’ shares across countries. The poorest quintile is calculated using per capita posttransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

FIGURE 3.24 Public Works’ Value Captured in Household Surveys as a Share of Beneficiaries’ Posttransfer Welfare among the Poorest Quintile

Source: ASPIRE database.

Note: The number of countries per region with monetary values for public works is as follows: total (n = 4), Sub-Saharan Africa (n = 3), Latin America and the Caribbean (n = 1), East Asia and Pacific (n = 0), Europe and Central Asia (n = 0), Middle East and North Africa (n = 0), and South Asia (n = 0). Public works programs include the following: Mexico 2012: Programa de Empleo Temporal. Malawi 2013: MASAF, PWP, Inputs for Work Program. Niger 2014: Public works. Rwanda 2013: Public works from the Vision 2020 Umurenge Program. Transfers as a share of a beneficiary’s welfare can be generated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. The share of transfers is determined as follows: (transfer amount received by all direct and indirect beneficiaries in a population group)/(total welfare aggregate of the direct and indirect beneficiaries in that population group). Public works average share is the simple average of public works’ shares across countries. The poorest quintile is calculated using per capita posttransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

FIGURE 3.25 Fee Waivers and Targeted Subsidies Value Captured in Household Surveys, as a Share of Beneficiaries’ Posttransfer Welfare among the Poorest Quintile

Source: ASPIRE database.

Note: The number of countries per region with monetary values for few waivers and targeted subsidies is as follows: total (n = 12); Europe and Central Asia (n = 8); Latin America and the Caribbean (n = 2); East Asia and Pacific (n = 1); Sub-Saharan Africa (n = 1); Middle East and North Africa (n = 0); and South Asia (n = 0). Fee waivers and targeted subsidies include any of the following: food, energy products, education, utilities, housing or transportation fees waivers to specific households, or discounted below the market cost. They do not include health benefits or subsidies. Transfers as a share of a beneficiary’s welfare can be generated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. The share of transfers is determined as follows: (transfer amount received by all direct and indirect beneficiaries in a population group)/(total welfare aggregate of the direct and indirect beneficiaries in that population group). Fee waivers and targeted subsidies average share is the simple average of these programs’ shares across countries. The poorest quintile is calculated using per capita posttransfer welfare (income or consumption). ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

On average, UCT transfers as a share of beneficiary welfare for the poorest quintile amount to 19 percent. To estimate this indicator, 52 surveys with monetary values were used out of 63 surveys with UCT information and out of a total of 79 surveys with monetary data for SSNs (see figure 3.21). As mentioned, in the case of UCTs, this indicator could be imprecise because various cash transfers implemented with different objectives may be aggregated: for example, poverty alleviation programs and universal family allowances. However, in a few countries, a single program is included in the UCT typology; thus, results can be attributed to that particular program.

Some UCT programs with a poverty alleviation objective tend to have a higher benefit level. For example, UCT programs make up the largest share of the welfare of the poor (49 percent) in Georgia and Rwanda (figure 3.21). Several programs are aggregated into the UCT categories for Georgia, but the results are mainly driven by the Targeted Social Assistance (TSA) program.²³ In the case of Rwanda, the result corresponds to the Direct Support from the Vision 2020 Umurenge Programme (VUP), the only UCT captured in the household survey. On the other side of the spectrum, very little can be ascertained about Sudan, where the household survey includes only a single question about general government assistance, without specifying what type of assistance it may be.

On average, CCTs account for a 16 percent share of the beneficiary welfare of the poor. Only 17 of 21 surveys with CCT information include a transfer value in the household survey that could be used to calculate this indicator. Bolivia’s Bono Juancito Pinto and Bono Juana Azurduy, combined, make up the largest share of the beneficiary welfare (36 percent) and Honduras’ Bonos PRAF is second (32 percent). The combined share of CCTs aggregated for Bangladesh is only 2.2 percent, the lowest of all included countries (see figure 3.22).

Social pensions make up a higher proportion of the welfare of the poor compared with other SSN instruments: 27 percent, on average. This finding is expected because somewhat like contributory pensions, social pensions are designed to address the lack of earnings because of old age and disability. Of 37 surveys with social pension information, only 30 include transfer values that make it possible to calculate the indicator. As shown in figure 3.23, the noncontributory old-age pension, disability grants, and the war veterans pension in South Africa combined make up the largest proportion of beneficiary welfare of the included countries (77 percent), and Brazil’s Benefício de Prestação Continuada follows (66 percent).

Public works programs have one of the lowest shares among SSN programs with respect to beneficiary welfare of the poorest quintile (7 percent). However, little can be concluded from the public works average indicator because only 4 out of 10 surveys with public works information include monetary values (see figure 3.24). This indicator cannot be estimated for some flagship public works programs (for example, the Productive Safety Net Program in Ethiopia or MGNREG in India), because they do not have monetary values in household surveys.

Fee waivers and targeted subsidies have the lowest share of beneficiary welfare among SSN programs: 6.7 percent, on average—much lower than the SSN global average of 19 percent (see figure 3.25). This finding is not surprising because fee waivers and targeted subsidies included under this category are typically aimed at helping the poor offset the cost of some services rather than support main earnings. In addition, the values captured in surveys and used to build this indicator are only a subset of this type of benefits: merely subsidies paid in cash. Many of these benefits are, by design, in the form of fee waivers, and thus monetary values are not available in the surveys. This indicator was calculated using only 12 out of 24 surveys with fee waivers and targeted subsidy information.

WHAT ARE THE POVERTY AND INEQUALITY IMPACTS OF SOCIAL SAFETY NET PROGRAMS?

This section provides an analysis of the ability of SSN monetary transfers to reduce poverty and inequality. The reduction in poverty and inequality is simulated by comparing beneficiaries’ welfare recorded in the survey before and after SSN transfers. Using information from surveys, the analysis computes what household per capita income or consumption would be without SSN transfers (see box 3.2). In other words, the transfer value is subtracted from the observed welfare (posttransfer welfare) to determine who would fall into poverty (using the absolute or relative poverty lines) if the transfer is eliminated (pretransfer welfare). The proportion of individuals who are lifted out of poverty as a direct effect of the transfer is estimated and illustrated in figure 3.26, along with the poverty gap reduction.²⁴

BOX 3.2 Measuring the Impact of Social Protection and Labor Programs

Household surveys provide a unique opportunity to measure the impact of the social protection and labor (SPL) programs on poverty or inequality because they contain information on the household aggregated income or consumption (welfare). This information makes it possible to determine, in each country, who the poor are. To allow international comparability, the Atlas of Social Protection: Indicators of Resilience and Equity (ASPIRE) database adopts two measures of poverty: the international absolute poverty line of US$1.90 a day per capita in purchasing power parity (PPP) terms; and a relative poverty line set at the bottom 20 percent of the pretransfer welfare distribution.

This chapter measures impacts on poverty and inequality by comparing households’ per capita welfare before and after the transfers. For example, assume that the threshold for the bottom 20 percent in a given country is US$2.50 PPP per capita per day. The survey reports that a household has a per capita income (or consumption) of US$3.00 PPP a day, which includes a transfer of US$1.00 PPP a day. The US$3.00 PPP welfare level constitutes the posttransfer welfare for that household. Then the analysis subtracts the transfer from that welfare (3.00 — 1.00 = 2.00). The US$2.00 PPP a day amount constitutes the pretransfer welfare. Because US$2.00 is lower than the quintile threshold of US$2.50, it can be determined that this household was lifted out of poverty thanks to the transfers. By replicating this calculation for the universe of households/individuals receiving transfers and applying the population expansion factors, the book is able to estimate the direct poverty and inequality reduction rates from the transfers for any given country.

This chapter uses three indicators to measure impacts on poverty and inequality: poverty headcount, poverty gap, and inequality reductions (in percent or relative terms). Poverty headcount provides the number of people living under the poverty line in a given country. The poverty headcount reduction is the percentage reduction in poverty because of the transfer. However, not all program beneficiaries become nonpoor after receiving a transfer; this depends on their position under the poverty line and if the transfer helped the beneficiary reach or surpass the poverty line. For those who do not overcome poverty, the simple headcount does not provide information on how much “less poor” beneficiaries are after the transfer. For this reason, the poverty gap is an important indicator to estimate the depth of poverty or how far below, on average, the welfare of poor individuals is from the poverty line. The poverty gap reduction thus provides information on the ability of SPL programs to bridge this gap. The Gini coefficient measures the inequality among values of the welfare distribution. Inequality reduction thus provides the percentage reduction in the Gini index because of the SPL transfer (Yemtsov et al., forthcoming).

FIGURE 3.26 World Reductions in Poverty from Social Safety Net Transfers, as Captured in Household Surveys, as a Share of Pretransfer Indicator Levels, by Relative and Absolute Poverty Lines

Source: ASPIRE database.

Note: The number of countries per region with monetary values for social safety nets is as follows: world (n = 79); Sub-Saharan Africa (n = 23); Europe and Central Asia (n = 20); Latin America and the Caribbean (n = 16); East Asia and Pacific (n = 10); Middle East and North Africa (n = 6); and South Asia (n = 4). This figure uses a relative measure of poverty defined as the poorest 20 percent of the welfare distribution (income or consumption) and absolute measure of poverty defined as US$1.90 PPP per day. Impacts on poverty and inequality can be estimated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. Percentages of poverty and inequality reduction are calculated as follows: (poverty headcount pre-transfer – poverty headcount posttransfer)/(poverty headcount pretransfer). Same calculations apply for poverty gap percentage reductions. Aggregated indicators are calculated using simple averages of country-level percentage reductions of the indicator across country income groups. The reductions in poverty are underestimated because ASPIRE does not include data for every single country in the country income groups, and even for a given country the survey does not include all existing social safety net programs or provide monetary values for them. For example, India and the impact of its flagship program MNREGA are not included in the calculation because only participatory information is available for the program. ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; PPP = purchasing power parity.

On the basis of the information observed in household surveys, the analysis shows that SSN transfers are making a substantial contribution in the fight against poverty. Whether an absolute poverty line (measured as US$1.90 per capita per day in purchasing power parity [PPP] terms) or a relative poverty line (measured in terms of the poorest 20 percent) is used, the analysis suggests that individuals are escaping poverty or decreasing their depth of poverty because of the SSN transfers.25 For the 79 countries that have monetary information, transfers reduce the incidence of absolute poverty (US$1.90 PPP per day) by 36 percent, whereas relative poverty (the bottom 20 percent) is reduced by 8 percent (see figure 3.26).

On average, SSN transfers are reducing the poverty gap more than the poverty headcount. In other words, even if SSN transfers are not lifting the poor and near-poor above the poverty line, they significantly reduce the poverty gap. As shown in figure 3.26, SSN transfers reduce the absolute poverty gap by 45 percent and the relative poverty gap by 16 percent. These results are remarkable considering that these figures are underestimated because household surveys do not capture the whole universe of SSN programs implemented in those countries. Therefore, it can be inferred that the real impacts are likely to be even larger.

The reductions in the poverty headcount, poverty gap, and inequality by SSN transfers are observed in all country income groups. In the sample of 79 surveys, the relative poverty headcount is reduced by 8 percent, the poverty gap by 16 percent, and the Gini inequality index by 2 percent (see figure 3.27). Across country income groups, the average reduction in relative poverty headcount is only 2 percent in low-income countries, 7 percent in lower-middle-income countries, 11 percent in upper-middle-income countries, and 15 percent in high-income countries. In terms of the poverty gap, the average reduction is 3 percent for low-income countries, 14 percent for lower-middle-income countries, 21 percent for upper-middle-income countries, and 30 percent for high-income countries. The Gini inequality index is less affected by SSN transfers, but reductions are still observed, ranging from 0.2 percent in low-income countries to 5 percent in high-income countries.

FIGURE 3.27 Reductions in Poverty and Inequality from Social Safety Net Transfers, as Captured in Household Surveys, as a Share of Pretransfer Indicator Levels, by Country Income Group Using Relative Poverty Line

Source: ASPIRE database.

Note: The total number of countries per country income group included in the analysis appears in parentheses. This figure uses a relative measure of poverty defined as the poorest 20 percent of the welfare distribution (income or consumption). Impacts on poverty and inequality can be estimated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. Percentages of poverty and inequality reduction are calculated as follows: (poverty headcount pretransfer – poverty headcount posttransfer)/(poverty headcount pretransfer). The same calculations apply for the Gini index and poverty gap percentage reductions. Aggregated indicators are calculated using simple averages of country-level percentage reduction of the indicator across country income groups. The reductions in poverty and inequality are underestimated because ASPIRE does not include data for every single country in the country income groups, and even for a given country the survey does not include all existing social safety net programs or provide monetary values for them. ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

The lower reduction in poverty and inequality observed for low-income countries is likely driven by several factors. First, fewer low-income countries have recent household survey data available compared with other country income groups. Second, of the 22 low-income country surveys included in ASPIRE, only 14 surveys include monetary variables for SSN programs. Third, many low-income country surveys neither capture SSN-specific program information nor include the universe of programs that exist in the country. And fourth, less than 10 percent of the global population live in low-income countries; therefore, the number of individuals moving out of poverty (in percent terms) is lower than in other country income groups.²⁶ See appendix F.3 for a list of poverty and inequality reductions from SSN programs by country.

WHAT FACTORS AFFECT THE IMPACT OF SOCIAL SAFETY NET TRANSFERS ON POVERTY AND INEQUALITY?

The extent to which SSN transfers have an impact on poverty and inequality depends on factors such as program coverage, transfer level, and the beneficiary/benefit incidence. Policy makers need to pay attention to the interaction of these factors when designing policies to reduce poverty/inequality. Figures 3.28 and 3.29 explore the reductions in poverty and inequality achieved by each country, given their degree of coverage of the poor and benefits levels.

The analysis reveals, in general, that very high coverage levels paired with high benefit levels lead to higher outcomes in poverty and inequality reduction. For example, Georgia and South Africa display the highest poverty headcount reduction using the poorest quintile as the poverty measure (see figure 3.28). Georgia’s combination of high SSN coverage (93 percent of the poorest quintile) and high level of benefits (SSN transfers constitute 68 percent of the poor’s welfare) leads to the highest poverty headcount reduction, nearly 43 percent. The universal old-age social pension drives these results because it covers 81 percent of the poorest quintile and constitutes 56 percent of their total welfare. The five programs included under cash transfers programs for Georgia—including the Targeted Social Assistance program—also cover a high percentage of the poorest quintile (46 percent) and make up 49 percent of the beneficiary welfare. Likewise, South Africa also shows high coverage and benefit levels for the poor (96 and 72 percent, respectively), leading to a poverty headcount reduction of 40 percent. In the South African survey, family and other allowances are among the SSN benefits with high coverage of the poor; however, social pensions constitute higher shares of beneficiary welfare. In both countries, the cost of SSN programs expressed as a percentage of GDP is rather large (7 percent for Georgia and 3.3 percent for South Africa).

FIGURE 3.28 Poverty Headcount Reduction from Coverage and Level of Social Assistance Benefits for the Poorest Quintile, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries with monetary values for social safety nets is as follows: world (n = 79), high-income countries (n = 5), upper-middle-income countries (n = 30), lower-middle-income countries (n = 30), and low-income countries (n = 14). This figure uses a relative measure of poverty defined as the poorest 20 percent of the welfare distribution (income or consumption). Poverty headcount reductions can be estimated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure is smaller than the one used to estimate coverage and beneficiary incidence. Percentages of poverty headcount reduction are calculated as follows: (poverty headcount pretransfer – poverty headcount posttransfer)/(poverty headcount pretransfer). Poverty headcount reductions are underestimated because ASPIRE does not include data for every single country in the country income groups, and even for a given country the survey may not include all existing social safety net programs or provide monetary values for those programs. ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

Conversely, very low coverage levels paired with low levels of benefits lead to negligible results in reducing poverty and inequality. For example, figure 3.28 shows how Armenia, with a combination of lower SSN coverage of the poor and benefits level (46 and 32 percent, respectively), achieves a more modest reduction in the poverty headcount (12 percent). SSNs in Liberia, with a much more modest combination of coverage of the poor and benefit level (10 and 17 percent, respectively), achieve a small poverty headcount reduction (2.5 percent). The estimated poverty headcount reduction for Chad is almost negligible (0.1 percent) because scholarships are the only SSN program captured in the survey. These scholarships have very low coverage of the poor (0.2 percent), although their contribution to beneficiary welfare is 23 percent.

In terms of the poverty gap reduction, the same interplay between coverage and benefit size is observed. In third place behind Georgia and South Africa, Mauritius also shows high coverage and benefit levels for the poorest quintile (84 percent and 55 percent, respectively), leading to a poverty gap reduction of 61 percent. The survey includes monetary information for eight programs, of which the noncontributory basic retirement pension has the highest coverage of the poor. Mauritius also reports high social spending according the administrative database (3.5 percent of GDP). In Poland, SSN programs captured in the household surveys report a relatively high level of coverage of the poor (65 percent), but the transfer as a share of beneficiary welfare is smaller (27 percent), producing a more modest estimated poverty gap reduction (42 percent). In Montenegro, SSN programs provide modest coverage of the poor (25 percent) and represent a modest share of beneficiary welfare (28 percent); thus, Montenegro achieves a modest poverty gap reduction (23 percent). In the Maldives, the household survey includes only two SSN programs: unspecified government transfers and social pensions. Even though social pensions have a high benefit level, driving the Maldives’ SSN average benefit to 76 percent of the beneficiary welfare, the coverage of both types of programs is small (15 percent) and thus the poverty gap reduction is modest (28 percent). In Burkina Faso, the poverty gap reduction is 0.1 percent, mostly because the survey does not capture a monetary value for government transfers, meaning that the estimation rests only on information on scholarships and other general transfers (figure 3.29).

FIGURE 3.29 Poverty Gap Reduction from Coverage and Level of Social Assistance Benefits for the Poorest Quintile, as Captured in Household Surveys

Source: ASPIRE database.

Note: The number of countries with monetary values for social safety nets is as follows: world (n = 79); high-income countries (n = 5); upper-middle-income countries (n = 30); lower-middle-income countries (n = 30); and low-income countries (n = 14). This figure uses a relative measure of poverty defined as the poorest 20 percent of the welfare distribution (in terms of income or consumption). Poverty gap reductions can be estimated only if monetary values are recorded in the household survey; for this reason, the sample of countries used in this figure (n = 79) is smaller than the one used to estimate coverage and beneficiary incidence (n = 96). Percentages of poverty gap reduction are calculated as follows: (poverty gap pretransfer – poverty gap post transfer)/(poverty gap pretransfer). Poverty gap reductions are underestimated because ASPIRE does not include data for every single country in the country income groups, and even for a given country, the survey may not include all existing social safety net programs or provide monetary values for those programs. ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

NOTES

1. SPL programs encompass social safety nets/social assistance, social insurance, and labor market programs.

2. For example, the household survey for El Salvador includes programs under the three areas of SPL. In this case, even though the most iconic programs in the country such as the CCT and social pensions are not adequately captured in the household survey, the in-kind transfers programs as well as the school feeding program drive the coverage of SSN programs up to 53 percent of the total population and 72 percent of the poorest quintile.

3. A source of potential bias is the fact that household surveys do not collect information from all social protection programs implemented in a country, but big flagship programs are likely to be captured.

4. A factor influencing this low coverage is that very little information on labor market interventions is captured in most household income and expenditure surveys; thus, it is difficult to draw meaningful conclusions regarding this program type.

5. The terms “social safety nets” and “social assistance” are used interchangeably in this book.

6. See http://datatopics.worldbank.org/aspire/~/documentation.

7. Aggregated variables with too few observations (less than 0.5 percent of the survey sample) were excluded because they are not representative to derive meaningful results. For example, the Livelihood Empowerment Against Poverty (LEAP), Ghana’s UCT program, was excluded under this criterion.

8. Under ASPIRE, UCTs encompass interventions such as poverty alleviation or emergency programs, guaranteed minimum-income programs, and universal or poverty-targeted child and family allowances; UCTs do not include social pensions, public works, or targeted subsidies paid in cash.

9. The Child Money Program was introduced in Mongolia as a targeted program in 2005 and became universal in July of 2006. Although the program was discontinued temporarily, it was reintroduced in October 2012, with a benefit level of US$14.72 per month per child (Yeung and Howes 2015).

10. The BR1M program was launched in 2012 to provide a single transfer to the poor. Households earning below RM3,000 monthly receive RM1,200; households earning between RM3,001 amd RM4,000 monthly receive RM900; and single individuals, 21 years and older, earning less than RM2,000 per month, receive RM450 (BR1M’s official website: https://ebr1m.hasil.gov.my). The Malaysian Household Income Survey does not identify the name of the cash transfer program in the questionnaire. However, it is likely capturing the BR1M program.

11. In many countries, elderly people tend to live within larger extended multigenerational families, whose members benefit indirectly from social pensions. Moreover, these pensions provide sizable benefits, and because the measure of welfare used in this book is pretransfer, many of the recipient families of social pensions tend to concentrate in the lowest quintile once this benefit is subtracted. This helps explain the high coverage of all the poor by social pensions in some countries featured in figure 3.6.

12. For this analysis, universal health schemes and universal general price subsidies are not included (most fuel subsidies or price support schemes for food are excluded). Instead, for commodities, only those subsidies that are targeted to specific individuals and families are included. One form of targeted subsidy is a lifeline (lower) tariff for a limited quantity of electricity, which is available only for eligible groups (as in Albania or the Russian Federation).

13. The Rastra Program, formerly called Beras untuk Rakyat Miskin (Raskin), is a rice subsidy introduced in 1998 as an emergency food security program. It delivers rice for purchase at subsidized prices and prioritizes poor and near-poor households (World Bank 2012). Even though the program reports the highest coverage of the poor, the benefit has been diluted to 4 kg per household on average from the promised 15 kg. In light of this, the government is proceeding with significant reforms.

14. Benefits incidence is not included in this analysis because information on the transfer value is needed to estimate it. Because not all the surveys include the monetary value of the transfers, the report prefers to use a larger sample of surveys by analyzing only the distribution of beneficiaries. If the value of the transfers is similar across beneficiaries, the distribution of beneficiaries and benefit across income groups will be similar.

15. Whether a program is propoor can be determined by the Coady-Grosh-Hoddinott (CGH) indicator, which divides the share of the beneficiaries/benefits belonging to a certain population group (for example, the bottom 20 percent) by the population share of this group. If the result is higher than 1, the program is considered progressive toward this group. For instance, if the bottom 20 percent of the population accounts for 40 percent of beneficiaries, this indicator will be equal to 2, indicating that the program focuses on the poor. In the case of a universal program where benefits are equally distributed, the CGH indicator will be 1.

16. It is particularly interesting to observe that some programs with similar design features and institutional structures—such as the Family Material Support (MOP) and Family Allowances Program in Montenegro and Serbia, which have the same legacy of the Federal Republic of Yugoslavia, and preserve some core features—have quite different distributions of beneficiaries. That implies that implementation matters a lot in how well the program is capable of targeting the poor.

17. In the case of Djibouti, these food distribution programs focus on rural areas where over half the population lives in poverty, according to the survey.

18. Transfers as a share of beneficiaries’ welfare is estimated as the amount of transfers received by a quintile divided by the total income or consumption of beneficiaries in that quintile.

19. The poverty gap is the distance below the poverty line and the average income of the poor. It is typically expressed as the percentage shortfall in income of the poor with respect to the poverty line.

20. At the same time, the greater is the concern about potential disincentives to work.

21. To assess benefit levels, information on the actual value of the transfers is needed. Therefore, only 79 out of 96 surveys available in ASPIRE were analyzed because not all surveys include monetary information.

22. This analysis was done only for the types of programs that involve a monetary transfer. Thus, the indicator was not calculated for school feeding programs nor for in-kind transfers. The sample of countries is also different per SSN instrument, compared with the sample used for coverage and beneficiary incidence, depending on the availability of data about the transfer value in the survey for each of the SSN instruments analyzed.

23. These programs are Internally Displaced Persons Assistance, social assistance to multichildren families, social assistance to orphans, social assistance to children with disabilities, and Targeted Social Assistance (TSA).

24. Therefore, this is a first-level approximation of the direct impact of the monetary transfers observed in the surveys. It also does not assess any medium- or long-term effects of SSN transfers on their specific objectives, which are better measured by impact evaluations.

25. The absolute poverty line of US$1.90 PPP a day may not constitute a meaningful standard for many high-income countries that have a very small population, or may not even have anybody living under that threshold. In fact, countries such as Belarus, Croatia, Lithuania, Montenegro, and Romania are able to fully eradicate their poverty headcount at US$1.90 PPP a day, according to information provided by household surveys (see appendix F).

26. There are 3 billion people living in lower-middle-income countries and only 659 million in low-income countries. In lower-middle-income countries, 500 million people live below the US$1.90 PPP a day threshold, which is more than 75 percent of the total population of low-income countries. In India alone, 285 million people are living on less than US$1.90 PPP a day, as well as 120 million in Nigeria, and 49 million in China. In other words, most of the world’s poor live outside low-income countries.

REFERENCES

Andrews, C., C. Del Ninno, C. Rodriguez Alas, and K. Subbarao. 2012. Public Works as a Safety Net: Design, Evidence, and Implementation. Directions in Development. Washington, DC: World Bank.

ASPIRE (Atlas of Social Protection: Indicators of Resilience and Equity). 2017. Database, World Bank, Washington, DC. http://datatopics.worldbank.org/aspire/.

Daidone, S., S. Asfaw, B. Davis, S. Handa, and P. Winters. 2016. “The Household and Individual-Level Economic Impacts of Cash Transfer Programmes in Sub-Saharan Africa. Synthesis Report.” Food and Agricultural Organization of the United Nations, Rome.

Garcia, M., and C. Moore. 2012. The Cash Dividend: The Rise of Cash Transfer Programs in Sub-Saharan Africa. Directions in Development. Washington, DC: World Bank.

Handa, S., M. Park, R. Osei Darko, I. Osei-Akoto, B. Davis, and S. Daidone. 2013. “Livelihood Empowerment against Poverty Program: Impact Evaluation.” Carolina Population Center, University of North Carolina at Chapel Hill.

Schüring, E. 2010. “Strings Attached or Loose Ends? The Role of Conditionality in Zambia’s Social Cash Transfer Scheme.” Maastricht Graduate School of Governance, Maastricht, Netherlands.

World Bank. 2012. “Raskin Subsidized Rice Delivery. Social Assistance Program and Public Expenditure Review No. 3.” World Bank, Washington, DC.

Yemtsov, R. M. Honorati, B. Evans, Z. Sajaia, and M. Lokshin. Forthcoming. “An Analytical Approach to Assessing Social Protection Effectiveness: Concepts and Applications.” World Bank, Washington, DC.

Yeung, Y., and S. Howes. 2015. “Resources-to-Cash: A Cautionary Tale from Mongolia.” Development Policy Centre Discussion Paper #42, Crawford School of Public Policy, The Australian National University, Canberra., Australia.

HIGHLIGHT 1: Productive Outcomes of Social Safety Net Programs: Evidence from Impact Evaluations in Sub-Saharan Africa

A defining characteristic of the poor in Sub-Saharan Africa is they are trapped in low-productivity or low-paying jobs. An analysis using a sample of 28 low-income-country household surveys from the Atlas of Social Protection: Indicators of Resilience and Equity (ASPIRE) found that 73 percent of individuals in the poorest 20 percent are employed, of which 69 percent work in agriculture.1 Of those employed, 49 percent are self-employed, whereas 30 percent are unpaid workers. Furthermore, many of the poor live in a context of poorly functioning or nonexistent labor and insurance/credit markets that affect household economic decisions (Handa et al. 2017). In this context, cash transfers may help households overcome these market failures and may enable them to spend more and to make productive investments. This in turn may generate household-level multiplier effects (Daidone et al. 2016), as well as spillover and income multiplier effects in local communities (Daidone et al. 2015; Thome et al. 2016).

This highlight summarizes new research on the broader impacts of unconditional cash transfers, particularly regarding economic impacts and productive inclusion in the context of Sub-Saharan Africa. The discussion draws largely on papers that synthesize and analyze the results from evaluations in eight Sub-Saharan African countries as part of The Transfer Project² and the From Protection to Production (PtoP) Project.^3,4 The narrative is complemented with the results of a metadata analysis of the impact evaluations available for 20 programs in 10 African countries (Ralston, Andrews, and Hsiao 2017); as well as a systematic review of 56 cash transfers programs (including unconditional cash transfers, conditional cash transfers, noncontributory social pensions, and enterprise grants) in low- and middle-income countries from 2000 to 2015 (Bastagli et al. 2016).

Findings from these evaluations consistently show statistically positive impacts of cash transfer programs on productive outcomes such as crop production, productive investments, employment, and more effective risk-coping mechanisms. The findings include income-multiplier effects for beneficiary and nonbeneficiary households and local economies. They debunk misconceptions like cash transfers being handouts that promote dependency and induce wasteful consumption. On the contrary, cash transfers have economic impacts beyond their intended objectives that strengthen beneficiary livelihoods and, if complemented with other instruments, can be leveraged to produce long-lasting benefits for beneficiaries, paving their way out of poverty.

WHAT ARE THE ECONOMIC IMPACTS OF CASH TRANSFERS ON BENEFICIARIES AND THEIR HOUSEHOLDS?

Household Production

Impact evaluations indicate that cash transfers increase household crop production, lead to changes in types of crops cultivated, and increase consumption and sales of homegrown production. The evaluation of programs in seven Sub-Saharan African countries found significant impacts in all these areas, even though their magnitude varied across countries (Daidone et al. 2016). Crop production increased in Zambia and Lesotho. The value of the overall production in Zambia almost doubled, boosting postprogram per capita consumption to a level 25 percent higher than the transfer itself (Davis et al. 2016). In Ethiopia, Malawi, and Zimbabwe, transfers led to changes in the types of crops cultivated. In Kenya and Malawi-Mchinji, the cash transfer increased consumption of the home-grown production (Davis et al. 2016).

Consumption and Productive Investments

Robust evidence indicates that households use transfers for basic needs and productive investments, dispelling myths of profligate spending. A meta-analysis conducted on impacts for seven African countries found that on average, household consumption increases by US$0.74 for each US$1 transferred (Ralston, Andrews, and Hsiao 2017). The magnitude of this impact varies across countries, with the largest impact experienced by programs targeted at the poor: the Malawi Social Cash Transfer Program reports the largest consumption impact, at 179 percent of the transfer value. In addition, a review of 19 programs and 11 studies using data from Africa, Asia, and Latin America found no evidence that transfers, conditional or unconditional, increase the use of alcohol and tobacco (Evans and Popova 2014).

The review by Bastagli et al. (2016) found that cash transfers improved household consumption in 25 of 35 studies looking at this effect. Their review also found an increase in livestock ownership/purchase, agricultural assets and inputs, and savings (although not for all programs or for all types of livestock, assets, or inputs). The Transfer Project found that five out of eight programs had significant impacts on the increase in livestock ownership. The effect of households investing in diverse types of animals was large in Malawi and Zambia, whereas more limited effects were observed in Kenya, Lesotho, and Zimbabwe, where small livestock were acquired. Impacts were not found in Ghana. The meta-data analysis by Ralston, Andrews, and Hsiao (2017) confirms these findings by estimating a combined average increase of 34 percent in livestock ownership across programs in seven countries (four were significant).

Most programs had significant impacts on the purchase/use of agricultural inputs such as seeds, fertilizer, and pesticides, although the magnitude of these impacts varied across countries. In terms of agricultural assets (for example, axes, hoes, picks, and other tools), positive impacts were observed in Ethiopia, Malawi, Zambia, and Zimbabwe. Impacts were not observed in Kenya, Lesotho, and Ghana. Even though all impacts were not always significant, all countries reported positive significant results for population subgroups, type of animal, or asset (Daidone et al. 2016; Davis et al. 2016; Handa et al. 2017).

Labor and Time Allocation

Impact evaluations generally show no evidence that cash transfers reduce the labor supply. In the African context, transfers have an impact on household decision making on labor allocation and time use, in terms of switching between different income-generating activities or between labor, domestic tasks, or leisure (Handa et al. 2017). In other words, cash transfers give households flexibility to allocate time and labor to these activities, leading to a switch from casual agricultural labor to on-farm labor (David et al. 2016). As a female beneficiary in Malawi expressed with respect to doing casual labor or ganyu: “I used to be a slave to ganyu but now I’m a bit free” (Barca et al. 2015). In the context of informal rural labor markets, casual wage labor is a last-resort activity that households use to survive or when liquid funds are scarce. Therefore, for beneficiaries to increase work and time on their own farms is a preferred activity and a sign of improved economic conditions. These findings are consistent with other studies that have concluded that transfers do not reduce the labor supply or create dependency. For example, Banerjee et al. (2015), after conducting seven randomized control trials of cash transfers programs in six countries (Honduras, Indonesia, Mexico, Morocco, Nicaragua, and the Philippines), found no evidence that cash transfers have impacts on either the propensity to work or the overall number of hours worked.

Risk Management and Coping Strategies

Cash transfers, impact evaluations find, allow beneficiary households to manage risk more effectively by diversifying income-generating activities, increasing savings, and reducing detrimental coping strategies. In Zambia, the share of beneficiary households running nonfarm enterprises increased 16 percentage points, and the businesses reported 1.4 more months in operation, compared with the control group, after receiving cash transfers. In Zimbabwe, there was an increase of 5 percentage points in the share of households operating these businesses and an increase of 5 percentage points in the share reporting profits. Other countries do not report significant results (Daidone et al. 2016). In Ghana, a qualitative evaluation finds evidence of increased petty trading of small amounts of kerosene and the sale of cooked food (Barca et al. 2015).

Cash transfers often lead to increasing savings. Ghana and Zambia reported increases in savings of 11 and 24 percentage points, respectively. In Ethiopia, Ghana, and Malawi transfers led to a reduction of loans and debt repayments (Daidone et al. 2016). In Zambia and Zimbabwe, transfers contributed to an increase in households’ creditworthiness; however, households were still risk averse and reluctant to take on new credit (Davis et al. 2016). Ralston, Andrews, and Hsiao (2017) found that beneficiary households are 20 percentage points more likely to save compared with the control group; this translates into an average increase of 92 percent in the number of households setting aside savings.

Studies also report a decrease in negative risk-coping strategies, such as distress sales of assets, begging, eating less, or putting children to work. Ethiopia, Lesotho, and Malawi reported less begging and changes in eating habits. Beneficiary households were less likely to take children out of school in almost all the countries analyzed by Davis et al. (2016) and Daidone et al. (2016). As an elderly beneficiary in Ethiopia said: “Hunger pushed me to do this [beg]. Since I started to receive the cash transfer I no longer have to. I feel happier” (Barca et al. 2015).

Qualitative evaluations also show that cash transfers programs increase social capital. They help beneficiaries reenter social networks, strengthening informal social protection systems and risk-sharing management arrangements (Davis et al. 2016). The very poor faced fewer stigmas, participated more fully in the community, and supported other households or institutions because of the transfers.

Spillover Effects in Local Economies

Cash transfer programs not only bring economic impacts to beneficiary individuals and households but also can create spillover effects that benefit nonbeneficiary households and local economies. When beneficiaries spend transfers, cash is injected into local markets, potentially creating income multipliers. That is, for each US$1 transferred to a beneficiary household, the total income of the local economy may increase by more than US$1. These spillovers are difficult to measure by experimental methods because they are second-order impacts that are diffused over a population greater than the beneficiary population (Thome et al. 2016). However, researchers developed the Local Economy-Wide Impact Evaluation (LEWIE) to simulate the effect of cash transfers in local economies. LEWIE measures the impact of cash transfers on the production activities of beneficiary and nonbeneficiary households, how these effects change during programs scaled up, and the reasons these effects happen (Taylor 2012).5

In Ethiopia, Ghana, Kenya, Lesotho, Malawi, Zambia, and Zimbabwe, cash transfers have significant spillovers in the local economy and nominal income multipliers. Using the LEWIE model, income multipliers range from 1.27 in Malawi to 2.52 in Ethiopia (Hintalo region). The income spillover (multiplier minus 1) indicates that for every US$1 transferred to the beneficiary household, the local economy gains an additional US$0.27–1.25 (Thome et al. 2016).

After considering potential inflation, simulations find that the real income multiplier, even though lower than the nominal multiplier, is still greater than 1 for all seven countries. If producers in the local economy face constraints to increase production in response to the higher demand of goods led by the transfers, this may put upward pressure on prices and lower income multipliers. However, even after accounting for inflation, income multipliers are still greater than 1, ranging from 1.08 in Kenya (Nyanza Province) to 1.81 in Ethiopia (Hintalo). The elasticity of the supply of local goods largely drives the differences between real and nominal multipliers. In economies that can easily increase the supply of goods, the price increase is small and nominal and real multipliers are similar (Thome et al. 2016).

NOTES

1. ASPIRE generates country context indicators using 129 surveys standardized by the I2D2 Project (International Income Distribution Database).

2. The Transfer Project is a joint effort of the United Nations Children’s Fund (UNICEF), Save the Children, and the University of North Carolina to support the implementation of impact evaluations of cash transfer programs in Sub-Saharan Africa. It focuses on the broad range of impacts of government-run cash transfer programs in Sub-Saharan Africa.

3. The From Protection to Production (PtoP) Project is part of the Transfer Project and focuses on exploring the linkages between social protection, agriculture, and rural development; http://www.fao.org/economic/ptop/home/en/.

4. Countries and cash transfer programs included under the PtoP Project include the following: Ethiopia: Tigray Social Cash Transfer Pilot Program (SCTPP); Ghana: Livelihood Empowerment Against Poverty Program (LEAP); Kenya: Cash Transfer Program for Orphans and Vulnerable Children (CT-OVC); Lesotho: Child Grants Program (CGP); Malawi: Social Cash Transfer (SCT); Zambia: Program Child Grant Program (CGP); and Zimbabwe: Harmonized Social Cash Transfer Program (HSCT).

5. To do this, LEWIE models link agricultural household models into a general-equilibrium model of the local economy, which is a treated village (Thome et al. 2016).

REFERENCES

ASPIRE (Atlas of Social Protection: Indicators of Resilience and Equity). 2017. Database, World Bank, Washington, DC. http://datatopics.worldbank.org/aspire/.

Banerjee, A., R. Hanna, G. Kreindler, and B. A. Olken. 2015. “Debunking the Stereotype of the Lazy Welfare Recipient: Evidence from Cash Transfer Programs Worldwide.” Working Paper No. 308, Center for International Development, Harvard University, Cambridge, MA.

Barca, V., S. Brook, J. Holland, M. Otulana, and P. Pozarny. 2015. “Qualitative Research and Analyses of the Economic Impacts of Cash Transfer Programmes in Sub-Saharan Africa. Synthesis Report.” Food and Agricultural Organization of the United Nations, Rome.

Bastagli, F., J. Hagen-Zanker, L. Harman, V. Barca, G. Sturge, and T. Schmidt. 2016. “Cash Transfers: What Does the Evidence Say? A Rigorous Review of Programme Impact and the Role of Design and Implementation Features.” Overseas Development Institute, London. http://www.odi.org/projects/2797-social-protection-literature-review-poverty-impact.

Daidone, S., L. Pellerano, S. Handa, and B. Davis. 2015. “Is Graduation from Social Safety Nets Possible? Evidence from Sub-Saharan Africa.” IDS Bulletin 46 (2): 93–102.

Daidone, S., S. Asfaw, B. Davis, S. Handa, and P. Winters. 2016. “The Household and Individual-Level Economic Impacts of Cash Transfer Programmes in Sub-Saharan Africa. Synthesis Report.” Food and Agricultural Organization of the United Nations, Rome.

Davis, B., S. Handa, N. Hypher, N. Winder Rossi, P. Winters, and J. Yablonski, eds. 2016. From Evidence to Action: The Story of Cash Transfers and Impact Evaluation in Sub-Saharan Africa. Oxford, UK: Oxford University Press.

Evans, D., and A. Popova. 2014. “Cash Transfers and Temptation Goods: A Review of Global Evidence.” Policy Research Working Paper 6886, World Bank, Washington, DC.

Handa, S., S. Daidone, A. Peterman, B. Davis, A. Pereira, T. Palermo, J. Yablonski. 2017. “Myth Busting? Confronting Six Common Perceptions about Unconditional Cash Transfers as a Poverty Reduction Strategy in Africa.” Office of Research–Innocenti Working Paper 2017-11, UNICEF (United Nations Children’s Fund).

Ralston, L., C. Andrews, and A. Hsiao. 2017. “A Meta-Analysis of Safety Net Programs in Africa.” Working Paper, World Bank, Washington, DC.

Taylor, J. E. 2012. “A Methodology for Local Economy-Wide Impact Evaluation (LEWIE) of Cash Transfers.” Food and Agricultural Organization of the United Nations, Rome.

Thome, K., J. Taylor, M. Filipski, B. Davis, and S. Handa. 2016. “The Local Economy Impacts of Social Cash Transfers: A Comparative Analysis of Seven Sub-Saharan Countries.” Food and Agricultural Organization of the United Nations, Rome.