SOCIAL PROTECTION DISCUSSION PAPER No. 2508 | MARCH 2025 State of Social Protection Report 2025 The 2-Billion-Person Challenge Background Paper #1 Mind the Gap: Coverage, Adequacy and Financing Gaps in Social Protection for the Extreme Poor and the Poorest Quintile Emil Daniel Tesliuc Ana Sofia Martinez Cordova © 2025 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Telephone: +1 (202) 473 1000; Internet: www.worldbank.org. This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. RIGHTS AND PERMISSIONS The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: +1 (202) 522 2625; e-mail: pubrights@worldbank.org. Abstract Mind the Gap evaluates the progress in strengthening social protection and labor systems in emerging and developing economies (EDEs) in three areas: coverage, adequacy of benefits for the poor, and level of financing. The paper explores policy options to reduce the coverage and income gap for the world’s poorest. Over the past decade, social protection coverage in the average EDEs has increased by 10 percentage points, from 41 percent around 2010 to approximately 51 percent in 2022. In absolute terms, as of 2022, 4.7 billion out of 6.3 billion people in low- and middle-income countries were covered by social protection, while 1.6 billion were not covered at all. Moreover, about 0.4 billion people from the poorest quintile received insufficient social protection support. Overall, 2 billion people in low- and middle- income countries remain uncovered, or inadequately covered while poor, by social protection. Apart from coverage, Mind the Gap also evaluates the extent to which EDEs are progressing toward Universal Social Protection across various policy dimensions: benefit adequacy, financing levels, and targeting accuracy of social assistance programs. The findings indicate variable adequacy in benefit levels, significant government spending but persistent financing gaps, and generally modest pro-poor targeting with potential for improvement. These challenges are more pronounced in low-income countries. The paper also assesses the impact of social assistance programs on reducing extreme and relative poverty gaps, quantifies the annual costs needed to address these income shortfalls in EDEs, and explores options for closing these gaps, such as enhancing the efficiency of social assistance transfers and expanding fiscal space. JEL Codes: H53, I38, J28, O15 Keywords: social protection system, coverage gap, social protection spending, adequacy gap, social protection targeting. 2 Acknowledgments This paper was prepared by the Social Protection Global Department at the World Bank as a part of its flagship report ‘The State of Social Protection 2025 - The 2-Billion-Person Challenge: Strengthening Social Protection and Labor for a World in Transition’. This paper was authored by Emil Daniel Tesliuc and Ana Sofia Martinez Cordova with contributions from Maria Belen Fonteñez, Muhsine Senart, Ingrid Mujica, Johanna Estefania Andrango Brito, Xuejiao Xu, and Claudia P. Rodriguez. The paper was made possible thanks to the ASPIRE database which served as the primary data source for the analyses. The authors would also like to thank Zurab Sajaia and Sergiy Radyakin for the fruitful collaboration to enrich ADePT functionalities to expand the level of analysis and automate the production of the indicators presented in the report. The team is also deeply grateful for the guidance provided by the peer reviewers: Aline Coudouel, David Coady (IMF), Carolina Diaz-Bonilla, Phillippe Leite, Anita Schwarz, Joana Silva, and Ruslan Yemtsov at the concept stage, and Brooks Evans (IMF), Ugo Gentilini, Ruth Hill, Harry Edmund Moroz, and Joana Silva at the decision stage. The background paper was produced under the guidance of Iffath Sharif (Global Director, Social Protection), Michal Rutkowski (former Global Director, Social Protection), and Loli Arribas-Baños (Practice Manager, Social Protection). Lastly, the team benefited from valuable support from Matthew Naumann (editor), Fiona Mackintosh (acquisitions editor), Agnes Nderakindo Mganga (Program Assistant) and Alexandra Humme (Senior External Affairs Officer). 3 Table of contents Abstract ................................................................................................................................................... 2 Acknowledgments................................................................................................................................... 3 Introduction ............................................................................................................................................ 7 I. Context: Population and Poverty in EDEs...................................................................................... 10 II. Coverage: Noteworthy Progress but Many Outstanding Gaps, Including of the Poorest Households and Extreme Poor Populations, Particularly in LICs .......................................................... 14 IIA. The Coverage Gap: The Missing 2 Billion .................................................................................... 14 IIB. The Coverage Gap: Cross-country Performance ......................................................................... 27 IIC. Progress in Closing the Beneficiaries’ Coverage Gap Over the Last Decade (circa 2010 versus circa 2022) ......................................................................................................................................... 32 III. Benefit Levels: Variable and Insufficient to Fill the Income Shortfall of the Extreme Poor or for Those in the Poorest Quintile for Most Social Protection Programs, in the Majority of EDEs ............. 36 IV. Spending: A Notable Area of Government Spending but Inadequate to Fill the Coverage and Adequacy Gaps...................................................................................................................................... 43 V. Targeting of Social Protection: Zooming in on the Equity of Social Assistance Allocation ............ 54 VI. Closing the Extreme Poverty Gap and Reducing the Relative Poverty Gap: Increasing Efficiency, Augmenting Domestic Financing, and Mobilizing International Finance............................. 61 VIA. Closing the Extreme Poverty Gap in EDEs ................................................................................. 65 VIB. Reducing the Relative Poverty Gap............................................................................................ 70 VIC. A Call for Action: Ensuring Adequate Financing and Strengthening the Pro-poor Equity of Social Protection Systems ................................................................................................................. 74 References............................................................................................................................................. 76 Annex 1: ASPIRE: Key Concepts, Definitions, and Data ......................................................................... 80 Annex 2: Assumptions of a technically and politically feasible social assistance program mix where welfare targeted programs deliver a targeting performance at the pro-poor targeting frontier ......... 85 Figures Figure II.1. 2 Billion People in LICs and MICs Are Missed or Inadequately Covered by Social Protection .............................................................................................................................................................. 20 Figure II.2. People in LICs and MICs who Remain Uncovered or Inadequately Covered by Social Protection.............................................................................................................................................. 21 Figure II.3. The Coverage Gap Increases Substantially Once China and India Are Excluded................. 22 4 Figure II.4. Distribution of the 2 Billion Missed, or Inadequately Covered if Poor, by Social Protection by Quintiles ........................................................................................................................................... 23 Figure II.5. The Coverage Gap Rises to Almost Half for Those in the Poorest Quintile ......................... 24 Figure II.6. In LICs, Coverage Gaps Reach 98 Percent among the Extreme Poor .................................. 25 Figure II.7. The ‘New’ HICs Have Almost Closed the Coverage Gaps of the Poorest Quintile .............. 26 Figure II.8. Proportion of the Total, Poorest Quintile, and Extreme Poor Population Receiving Various Types of Social Protection Benefit, by the Country’s Level of Economic Development (percentages) 28 Figure II.9. Proportion of the Total, Poorest Quintile, and Extreme Poor Population Receiving Various Types of Social Protection Benefit, by World Bank Group Region (percentages) ................................. 30 Figure II.10. Proportion of the Total, Poorest Quintile, and Extreme Poor Population Receiving Various Types of Social Protection Benefit, by Region and the Country’s Level of Economic Development (percentages) ........................................................................................................................................ 31 Figure II.11. Expansion of Access to Social Protection Benefits over 2010–2022................................. 33 Figure II.12. Expansion of Social Assistance Coverage as the Main Driver for the Increase in Social Protection Coverage from 2010 to 2022 ............................................................................................... 34 Figure II.13. Receipt of Social Insurance Benefits (from Contributory Pensions or Other Social Insurance Programs) between 2010 and 2022 ..................................................................................... 35 Figure III.1. Adequacy Ratio of Social Protection Benefits Received by the Total, Poorest Quintile, and Extreme Poor Population, by Social Protection Type ............................................................................ 37 Figure III.2. Adequacy Ratio of Social Protection Benefits Received by the Total, Poorest Quintile, and Extreme Poor Population, by Country Income Level ............................................................................ 38 Figure III.3.Adequacy Ratio of Social Protection by Income Level for the Total Population, the Poorest Quintile, and the Extreme Poor ............................................................................................................ 39 Figure III.4. Adequacy Ratio of Social Protection by Region for Total Population, the Poorest Quintile, and the Extreme Poor ........................................................................................................................... 40 Figure III.5. Social Protection Benefits Make Up a Large Share of the Pre-transfer Consumption of Households in Extreme Poverty and in the Poorest Quintile—But Not in LICs or in Countries with High Extreme Poverty Rates .......................................................................................................................... 42 Figure IV.1. Social Protection Spending as Percentage of GDP, Unweighted Average, 2022 ................ 45 Figure IV.2. Absolute Level of Social Protection Spending (US$ 2017 PPP) Per Capita, Unweighted Average, 2022 ....................................................................................................................................... 46 Figure IV.3. Heterogeneity in the Share of Social Protection Spending in GDP across EDEs, Unweighted Average, 2022 ................................................................................................................... 48 Figure IV.4. Absolute Level of Social Protection Spending (US$ 2017 PPP) Per Capita across EDEs, Unweighted Average, 2022 ................................................................................................................... 49 Figure IV.5. Proportion of Total Government Spending on Social Protection, by Social Protection Area, 2022 ...................................................................................................................................................... 50 Figure IV.6. Proportion of Total Government Expenditure Spent on Social Protection Correlates with the Revenue-Generating Capacity of the Country ................................................................................ 52 Figure IV.7. Countries with Low Social Assistance Spending Levels Have Low Levels of Coverage of the Extreme Poor and the Poorest Quintile ................................................................................................ 53 5 Figure IV.8. Countries with Low Social Assistance Spending Levels Also Have Low Adequacy Ratios .. 53 Figure V.1. Distribution of Beneficiaries by Quintile, Social Protection Area, and Country Income Group .................................................................................................................................................... 56 Figure V.2. Proportion of Social Assistance Beneficiaries from the Poorest Quintile, by Country Income Group ....................................................................................................................................... 57 Figure V.3. EDEs Could Significantly Improve the Pro-poor Allocation of Their Social Assistance Spending: Average Share of Beneficiaries from Best-Performing Social Assistance Program, by Country Income Group.......................................................................................................................... 59 Figure VI.1. Social Assistance Closes One-Third of the Income Shortfall and Could Reduce It Further with Improved Pro-poor Targeting ........................................................................................................ 64 Figure VI.2. Contribution of Social Assistance Spending to the Reduction of the Income Shortfall of the Extreme Poor, Total and by Countries Grouped by Their Extreme Poverty Rate............................ 67 Figure VI.3. Reduction of the Income Shortfall of the Extreme Poor Due to Social Assistance, Potential Reduction with Greater Efficiency, and the Additional Fiscal Space Needed ....................................... 67 Figure VI.4. Reduction of the Income Shortfall of the Extreme Poor Due to Social Assistance, Potential Reduction with Greater Efficiency, and the Additional Fiscal Space Needed, as % of GDP .................. 68 Figure VI.5. Differences in the Daily Per Capita Consumption of the Population in the Poorest Four Quintiles, in 10 Countries with Very High Levels of Extreme Poverty................................................... 70 Figure VI.6. Contribution of Social Assistance Spending to the Reduction in the Income Shortfall of the Poorest Quintile, Total and by Countries, Grouped by Income Level ............................................. 71 Figure VI.7. Reduction of the Income Shortfall of the Poorest Quintile Due to Social Assistance, Potential Reduction with Greater Efficiency, and the Additional Fiscal Space Needed ........................ 72 Figure VI.8. Reduction of the Income Shortfall of the Poorest Quintile Due to Social Assistance, Potential Reduction with Greater Efficiency, and the Additional Fiscal Space Needed, as % of GDP ... 73 Tables Table I.1 Distribution of 153 EDEs by Income Level and Region ........................................................... 11 Table I.2 Distribution of 73 EDEs with Extreme Poor Population by Income Level and Region ............ 12 Table I.3 Distribution of the Population from the Poorest Quintile in 153 EDEs by Income Level and Region ................................................................................................................................................... 13 Table II.1 Methodology to Estimate the Coverage of Social Protection Beneficiaries and Contributors .............................................................................................................................................................. 19 Table IV.1. Total Number of Programs 2022, by Social Protection Area, Income Group, and Region... 44 Table A.I. 1. ASPIRE Program Classification ........................................................................................... 81 Table A.I. 2. Proportion of All EDEs Included in This Paper ................................................................... 83 Table A.II. 1. Benefit Distribution for LICs, MICs, and HICs, Including Administrative Cost .................. 87 Table A.II. 2. Pro-poor Benefit Distribution by Quintile for LICs, MICs, and HICs (%) ........................... 88 6 Introduction The COVID-19 years, 2019–2021, have seen the largest scale-up of social protection in history, underlining the need for social protection like never before, sparking widespread calls to ‘build back better’ after the crisis and giving new life to the universal social protection (USP) agenda.1 USP refers to a nationally defined system of integrated policies and programs that provide equitable access to all people, protecting them throughout their lives against poverty and risks to their livelihoods and well-being (USP2030 2019; World Bank Group 2022). USP recognizes that no individual program can protect people from the many different risks they are likely to face throughout their lifetimes, from poverty to ill health, income loss, unemployment, and old age. Therefore, comprehensive and effective coverage can only be provided by a suite of programs that span the domains of social insurance, labor and economic inclusion, and social assistance and care, ensuring that all people are effectively protected at all points in the lifecycle (World Bank Group, 2022). Given the major fiscal, political, and operational challenges inherent in universalizing social protection systems, the World Bank advocates the progressive realization of USP. This means prioritizing the poorest and most vulnerable for social protection—particularly for social assistance—from the outset, to ensure that they are the first to benefit from efforts to enhance provision. While national governments and international agencies alike are increasingly recognizing the need to accelerate efforts to achieve USP, realizing this vision requires us to know where we are, what has been achieved so far, and where the outstanding gaps lie. This requires stakeholders to come together at the national and international level to maximize data collection efforts. The World Social Protection Report of the International Labour Organization (ILO, 2024) draws on rich and detailed administrative data to estimate coverage, 1 For example, see the calls to action published by the Social Protection Interagency Cooperation Board (SPIAC-B), a partnership of 25 intergovernmental agencies and 10 government bodies, and by the Global Partnership for Universal Social Protection (USP 2030), a global partnership including governments, international and regional organizations and civil society organizations, among others. Analysis of the social protection crisis response and implications include Bastagli and Lowe (2021); Beazley, Marzi, and Steller (2021); Beazley, Bischler, and Doyle (2021); Gentilini (2022); Hammad, Bacil, and Soares (2021); ILO (2021b); IPC-IG (2021); and many others. 7 adequacy, and expenditure on social protection worldwide, while the Organisation for Economic Co-operation and Development (OECD) and Eurostat provide data on social protection expenditure, redistributive impact, and design parameters for their respective member states. This paper is intended to complement these valuable stock takes by drawing on nationally representative household survey data, alongside program-level administrative data, to provide a detailed exploration of social protection in emerging and developing economies (EDEs)—the focus of the World Bank’s operations. The data in this paper are mainly based on the Atlas of Social Protection: Indicators of Resilience and Equity (ASPIRE), the World Bank’s premier compilation of social protection and labor (SPL) indicators, gathered from both administrative program-level data and nationally representative household surveys. Administrative program-level data are used to calculate social protection expenditure as a percentage of gross domestic product (GDP). Nationally representative household surveys are used to calculate performance indicators, including program coverage, incidence of beneficiaries and benefits, level of transfers and simulated impacts on poverty, and inequality reduction due to transfers. For the first time, this paper uses an expanded and improved version of the dataset, titled ASPIRE 2.0, which covers not only social assistance programs but also labor market and contributory pensions. In addition, ASPIRE 2.0 now enables indicators from administrative data to be linked with indicators from household survey data at the program level, substantially expanding the scope for analysis. Annex 1 describes the main indicators covered by ASPIRE, and the paper’s coverage of EDEs based on both administrative and household survey data. This paper, a background paper for the 2025 State of Social Protection Report ‘The 2-Billion- Person Challenge: Strengthening Social Protection and Labor for a World in Transition’, assesses the progress toward USP and the first objective of the 2030 Agenda for Sustainable Development: to end poverty in all its forms everywhere (Goal 1). The analysis focuses on quantifying the progress toward the first three targets: (a) implementing nationally 8 appropriate social protection systems and measures for all, including floors, and, by 2030, achieving substantial coverage of the poor and the vulnerable (Target 1.3); (b) eliminating extreme poverty (Target 1.1); and (c) halving national poverty rates (Target 1.2). For Target 1.2, the paper estimates a related but slightly different indicator: what is the annual cost of closing the income gap for the 20 percent poorest. The paper examines the outcomes achieved by the entire social protection system of each EDE in expanding social protection coverage, ensuring adequate benefits particularly for the impoverished, mobilizing financing, targeting the poor with social assistance programs, and reducing the income gap of the poor. It investigates the overall impact resulting from the mobilization of the entire social protection system, as well as its individual components or policy areas: social assistance and social services, labor market programs, and social insurance programs. Other background papers provide a deeper analysis for social assistance (Okamura, Iyengar, and Andrews 2025), labor market programs (Carranza, Morgandi, and Sverdlin 2025), and pensions (Reyes et al. 2025), examining the performance of various types of programs within their respective policy areas. The paper proceeds as follows. The first section sets the stage for the paper, by presenting both the population and poverty in EDEs around 2022. The second section quantifies the share of population covered and missed by social protection (IIA), examines the cross-country performance in closing the coverage gap among EDEs (IIB), and quantifies the progress achieved from circa 2010 to circa 2022 in expanding social protection coverage2. The next three sections examine how far EDEs are from the progressive realization of USP across other policy dimensions: the adequacy of benefits (section III), the level of financing (section IV), and the targeting accuracy of social assistance programs (section V). In doing so, it finds evidence of enhanced but still insufficient population coverage (with 2 billion out of 6.3 billion 2 “Circa 2022” refers to data from 2022 or the most recent available year within the period 2015 to 2022. “Circa 2010” includes data from 2010 or any year available between the period 2006 to 2014. 9 people not covered by social protection or inadequately covered while poor, in low-income countries [LICs] and middle-income countries [MICs]), including a vast ‘missed middle’; variable adequacy of benefit levels; notable government spending but continuing major financing gaps; and overall modest pro-poor targeting with clear room for improvement. The sixth and last section of the paper quantifies the annual costs required to close the extreme poverty and relative poverty income shortfalls in EDEs and highlights potential avenues for closing these gaps. I. Context: Population and Poverty in EDEs This paper focuses on 153 EDEs. These include all low- and middle-income countries (LICs and MICs: 130 countries) and a few high-income countries (HICs: 23 countries). The paper focuses on the 130 LICs and MICs where challenges and needs are significant. It also analyzes 23 HICs that have recently cooperated or are currently cooperating with the World Bank. Therefore, the HIC category in this paper does not include all high-income countries but only 23 countries that have recently attained this status. These countries are tracked in ASPIRE for continuity and serve as useful comparators for LICs and MICs. The paper also breaks down the EDEs regionally; these regions differ from geographical regions and align with the World Bank’s regional units. Since many of the social protection indicators in the paper represent population shares (such as poverty rate, coverage, and beneficiaries’ incidence), the paper begins by presenting key population statistics. The EDEs analyzed in this paper account for four-fifths of the population of the planet, where social protection challenges and needs are the greatest. Around 2022,3 EDEs made up 82 percent of the global population, representing 6.6 billion out of 8 billion people. Of these, 3 Most of the paper presents distributional performance indicators for social protection programs (coverage, benefit adequacy, incidence of beneficiaries or of benefits, simulated impact on poverty) derived from household surveys. Of the 153 countries in ASPIRE, we have used survey information for 111 countries, from 2022 or the most recent survey wave available. Most of these surveys are recent, from 2018 to 2022. Population aggregates are generated using the survey distributional performance indicators and the population of the survey year and are referred to in the paper as estimates for “circa 2022”. 10 6.3 billion people reside in 130 LICs and MICs and 0.3 billion in the new HICs (Table I.1). By income level, 9 percent of the population lives in LICs, 86 percent in MICs, and 5 percent in the ‘new’ HICs. Regionally, the largest areas include East Asia and Pacific (including China), South Asia (including India), and Sub-Saharan Africa (which comprises 48, representing one- third of the EDEs). Table I.1 Distribution of 153 EDEs by Income Level and Region Share of total population Income level/Region No. of countries Population in millions (%) Total 153 6,610 100 Low income 25 605 9 Lower middle income 52 2,931 44 Upper middle income 53 2,761 42 High income 23 314 5 East Asia and Pacific 23 2,078 31 Europe and Central Asia 30 493 7 Latin America and Caribbean 30 634 10 Middle East and North Africa 14 429 6 South Asia 8 1,888 29 Sub-Saharan Africa 48 1,088 16 Source: World Development Indicators and ASPIRE. In line with the concept of progressive realization of the USP, the paper quantifies the proportion of the population living in households not covered by social protection systems, with a special focus on the poor. Ideally, the entire population should be protected in the event of a shock affecting their livelihoods. However, certain segments of the population are at greater risk than others: specifically, those classified as extreme poor (individuals living on less than US$2.15 in 2017 purchasing power parity [PPP] per capita per day) and the poorest quintile (the poorest 20 percent). This section outlines the extent of poverty in EDEs, providing context for subsequent analyses and policy recommendations. The discussion begins with extreme poverty, which is a key focus of the Sustainable Development Goal (SDG) commitments. 11 Circa 2022, the number of extreme poor in the 153 EDEs was estimated to about 510 million people. Of these, 73 countries with extreme poverty of 2 percent or more account for 496 million extreme poor, or 97 percent of the total for EDEs (Table I.2). The other 78 countries have either eliminated or reduced extreme poverty to very low levels. Two countries, Zimbabwe and Guyana, are not included in the estimation for lack of data. About 93 percent of the extreme poor are from LICs and LMICs. By region, most extreme poverty is concentrated in Sub-Saharan Africa (78 percent) and South Asia (10 percent). Table I.2 Distribution of 73 EDEs with Extreme Poor Population by Income Level and Region Share of extreme No. of Population in Extreme population Income level/Region poor population countries millions in millions (%) Total 73 3,811 496 100 Low income 22 557 248 50 Lower middle income 36 2,562 215 43 Upper middle income 15 692 33 7 High income 0 — — 0 East Asia and Pacific 9 456 19 4 Europe and Central Asia 5 53 2 0 Latin America and Caribbean 10 366 18 4 Middle East and North Africa 3 51 20 4 South Asia 4 1,835 51 10 Sub-Saharan Africa 42 1,051 386 78 Source: Original calculations for this publication based on World Bank Poverty and Inequality Platform (PIP). Recent years have interrupted decades of efforts to eliminate extreme poverty. Before the COVID-19 pandemic, significant progress had been made, with approximately 1 billion people lifted out of extreme poverty between 1990 and 2014. In the five years immediately preceding the pandemic, the reduction in poverty had slowed considerably but continued at a rate of about 0.6 percentage points per year. However, the COVID-19 pandemic caused an increase in extreme poverty for the first time in a generation, with an additional 71 million people living in extreme poverty in 2020 compared to 2019—offsetting nearly three years of progress. This setback, combined with overlapping climate, conflict, and economic challenges, resulted in 510 million people living in extreme poverty in EDEs as of 2022, representing 8 percent of their population. 12 Concerted efforts are needed to enable Target 1.1. of eliminating extreme poverty to be achieved—in which the progressive realization of USP can play a central role. Social protection can play a key part in eliminating extreme poverty, based on the extensive evidence of such impacts in wide-ranging contexts (Banerjee et al. 2024; Bastagli, Hagen-Zanker, and Sturge 2016; Lustig 2018; UNDESA 2018). New data on the extent to which social protection is currently contributing to poverty reduction and extreme poverty elimination are presented throughout this paper, along with proposals for maximizing its potential poverty-reducing impacts—thus accelerating progress toward the progressive realization of USP and Target 1.3. The paper also examines relative poverty in EDEs, defined as the population in the poorest quintile, as redistributive policies such as social protection, and especially social assistance, prioritize this group. The size of this group is 1.3 billion people (Table I.3). The distribution of this group by the income level of the country or region is identical with that of the total population (Table I.1). Table I.3 Distribution of the Population from the Poorest Quintile in 153 EDEs by Income Level and Region Share of total population Income level/Region No. of countries Population in millions (%) Total 153 1,322 100 Low income 25 121 9 Lower middle income 52 586 44 Upper middle income 53 552 42 High income 23 63 5 East Asia and Pacific 23 416 31 Europe and Central Asia 30 99 7 Latin America and Caribbean 30 127 10 Middle East and North Africa 14 86 6 South Asia 8 378 29 Sub-Saharan Africa 48 218 16 Source: World Development Indicators and ASPIRE. 13 II. Coverage: Noteworthy Progress but Many Outstanding Gaps, Including of the Poorest Households and Extreme Poor Populations, Particularly in LICs This section quantifies the share of population covered and respectively missed by social protection systems in EDEs, with a special focus on the LICs and MICs, which account for almost all the coverage gaps in 2022. Subsection IIA presents the absolute numbers of people covered and respectively missed by social protection systems in EDEs. Subsection IIB examines the cross-country performance in closing the coverage gap among EDEs, identifying the type of countries and regions where the coverage challenges are the greatest. Subsection IIC quantifies the progress achieved from circa 2010 to circa 2022 in expanding social protection coverage. IIA. The Coverage Gap: The Missing 2 Billion This section provides a comprehensive quantification of the social protection coverage gap around 2022 in 153 countries, current or past clients of the World Bank. These countries include all LICs (25) and MICs (105) as well as a group of HICs that are or have been cooperating with the World Bank in the recent past (23 ‘new’ HICs). The population covered includes beneficiaries of social assistance, social insurance, and labor market programs as well as those who contribute to social insurance schemes. Household survey microdata is utilized to measure the share of population living in households that contribute to or receive at least one social protection benefit. Survey-based estimates of coverage and adequacy offer valuable insights that complement existing program-level data from administrative sources. These estimates enhance the understanding of the socioeconomic profile of beneficiaries, such as their gender, location, or poverty status, and consider resource sharing within households. Motivation and focus countries. USP provision for a country’s population allows for optimal social risk management (Packard et al. 2019). A system where most of the population contributes to social insurance schemes achieves effective risk pooling and risk sharing while protecting the population against social risks of loss of earnings during old age, disability, 14 death of a breadwinner, or unemployment. USP is also critical for the support of the poor and vulnerable. SDG Target 1.3, critical for reaching SDG goal 1 of ending poverty in all its forms everywhere, aims to achieve substantial coverage of the poor and the vulnerable through nationally appropriate social protection systems and measures for all, including floors, by 2030. The empirical question addressed in this section is how big the coverage gap is in LICs and MICs. The section focuses on the LICs and MICs, the group of 130 countries where 81 percent of the planet’s population lived as of 2022 and where most of the coverage challenge remains. The section presents, selectively, estimates of the coverage gap for a group of HICs (23) that are aspirational peers for the upper-middle-income countries (UMICs). These 23 countries have graduated to HIC status relatively recently and are included in the target group of ASPIRE countries as they cooperate with the World Bank or have cooperated in the recent past. 1. Measuring global social protection coverage in LICS and MICs: Methodological considerations There are multiple ways to measure the coverage of the social protection system, that is, estimating the share of the population that is covered by the system and the share that is missed.4 Coverage estimates can be population-weighted or simple averages of the social protection coverage in each country. This section reports population-weighted averages, to generate a headline number of total population covered by the social protection system and those missed. The rest of the paper reports simple averages, as it examines the cross-country performance. As the policy decision on the extent and generosity of the social protection 4 For example, ILO uses administrative data on (mostly) individuals receiving different types of social protection interventions to arrive at a measure of coverage. By this definition, in a household with 4 members where only 2 receive individual-based benefits (like pensions or child allowances), 50 percent of the household is covered. The complementary measures estimated by the World Bank focus on households, with the underlying assumption that all income sources are shared among members. For the same family, the World Bank measures will show 100 percent coverage. By definition, the World Bank measures show a higher level of coverage and a smaller share of missed individuals than the ILO ones. 15 system is determined by each country, the rest of the report uses simple country averages to give equal weight to each country. Coverage can be estimated at individual or household level. Some global estimates of the share of world population not covered by social protection programs are based on administrative data on receipt of social protection benefits or contribution to social insurance programs, at individual level (ILO, 2024). These indicators are informative and simple to interpret, especially when focused on homogenous programs with individual-level benefits and when expressed relative to a specific target group (for example, share of persons with disabilities covered by disability programs or unemployed receiving active labor market programs [ALMPs] or passive labor market programs [PLMPs]). However, individual-level coverage works well at the program level but there are issues when aggregated into groups of programs or for programs where the assistance unit is the family or households. Coverage estimates based on individual-level data can lead to double counting, when the same person receives more than one benefit, or a household collects multiple benefits. When aggregated for the total population, the caseload of different programs needs to be expressed in the same assistance unit (typically household) and deduplicated. This is relatively straightforward with household survey data. Moreover, individual-level administrative data will not help if social protection benefits are sufficient, or adequate, to protect a household from poverty; such an assessment requires information on both social protection benefits received and the income or consumption of that household. For the first time, this report quantifies the global coverage gap in LICs and MICs at the household level, based on recent household surveys that track the welfare of the household, the receipt of social protection benefits, and contribution to social insurance schemes. This allows quantification of the cumulative support received by a household from multiple social protection programs, eliminating the risk of double counting inherent in administrative data. The coverage statistic has the following meaning: it counts all individuals 16 from households connected to the social protection system as covered. This is the population living in households where at least one individual is identified and typically paid or provided with goods and services through social protection programs or contributes to social insurance schemes. In the event of a shock affecting the livelihood of that household, the social protection system has the information and delivery mechanism to protect them. The coverage gap measures the population living in households that are not connected to the social protection system in any way. This population group is at a high risk if affected by shocks with economic consequences. Within the group, the poor population that is not connected to the social protection system is at an even greater risk, as they do not have assets or jobs with decent pay to self-insure amid shock. The report quantifies the group of poor not covered by the social protection system, a key priority for the horizontal expansion of the system according to the progressive realization of the USP. Also, for the first time, the report estimates whether the social protection support received by poor households is adequate or not. The report compares the cumulative social protection support received by households with relative poverty (the poorest quintile) and extreme poverty, thus revealing whether the social protection support is adequate or not for the households.5 This allows quantification of the adequacy of social protection benefits at the level of the social protection system for the poor. The threshold used to classify the social protection system support as adequate is whether social protection benefits account for at least 20 percent of the respective poverty line (the income that separates the poorest quintile from the upper level of the income distribution, or the extreme poverty line, of US$2.15 in 2017 PPP). This threshold represents about half the adequacy gap of the poorest quintile in LICs and MICs; thus, it represents a substantial reduction of the income gap of the poor. This 5 To estimate adequacy of programs without monetary values, we used a matrix with average adequacies by type of social protection program category and area, calculated for the different quintiles of the welfare distribution and for the different income level groups (low income, lower middle income, upper middle income, high income). This matrix was based on observed adequacies of programs with monetary values in our sample. We input these adequacies and calculate the cumulative social protection support received by households to later compare them with the relative and the absolute poverty. 17 group of poor not adequately covered is another key priority for the vertical expansion of the social protection system in LICs and MICs. These global estimates use a unique dataset of country-level distributional performance indicators of social protection programs, ASPIRE (see Annex 1) and cover all 130 LICs and MICs. The estimate has strong empirical foundations, as a large share of the indicator is measured, and only a small portion of it is estimated. For 61 countries accounting for 48 percent of the total LMIC population, all indicators are derived from the most recent household survey; for another 43 countries accounting for 48 percent of the LICs and MIC population where the household surveys capture the receipt of benefits but not contributions to social insurance schemes, the coverage indicator relies on the share of population receiving social protection benefits, adjusted upward with the average increase due to social insurance contributors of the respective country income group. For the remaining 26 countries accounting for 3 percent of the LICs and MIC population, the report assumes distributions similar to the average of the country income grouping, adjusted for the individual coverage of the contributors to the main social insurance schemes where available. The estimates are also recent. For about 90 percent of the LICs and MIC population, the estimations are based on surveys carried out between 2018 and 2022, with almost half of the population surveyed since 2020. The methodology used to measure and estimate global coverage in LICs and MICs is described in Table II.1. For each country, the report estimates the distribution of social protection coverage for the total population, for rural and urban areas, for each quintile, and for extreme poverty. 18 Table II.1 Methodology to Estimate the Coverage of Social Protection Beneficiaries and Contributors Country groups6 No. of Share of LICs Methodology LICs and and MICs MICs population (%) A. Surveys with 61 48 Estimate based on the information on social protection data on beneficiaries and contributors from the from the latest beneficiaries and household survey available in ASPIRE (the comprehensive contributors indicator of coverage) B. Surveys with 43 48 Estimate based on the information on social protection data on social beneficiaries from the latest household survey (the narrow protection indicator of coverage), total and by groups, increased with beneficiaries only a correction factor matrix (X or Y). Correction factor X = ratio of the comprehensive/narrow coverage of the Sample A countries, total and by groups, and country income level7 Correction factor Y = ratio of the comprehensive/narrow coverage for countries in Sample A with narrow (beneficiaries) coverage higher than 75%, total and by groups. C. Administrative 7 1 Estimate based on the multiplication of data on the • A vector of social protection coverage of beneficiaries individual and contributors, for the respective income country coverage of the grouping (LICs, LMICs, UMICs, and new HICs) from social insurance Sample A countries with contributors • A correction factor from administrative data as the ratio of (social insurance contributors’ coverage of the country) / (the average contributors’ coverage for the country’s income group. D. No data 19 2 Estimate equal to the average coverage of beneficiaries and contributors for the respective country’s income group. Total 130 100 Note: Own elaboration for this paper. 2. Two billion people missed by the social protection system or inadequately covered while poor As of 2022, out of 6.3 billion people living in the 130 LICs and MICs, 2 billion were missed or inadequately covered by the social protection systems if poor (Figure II.1). Of these, 1.6 6 The groups are based on the availability of recent household survey data to measure the coverage gap. 7We used the data from the 64 countries in Sample A to calculate a matrix of correction factors as the ratio of (coverage of contributors and beneficiaries) / (coverage of beneficiaries) for the following social protection coverage indicators: total population, rural-urban, 5 quintiles, and extreme poverty, for each of the 4 income groups (LICs, LMICs, UMICs, HICs). The social protection beneficiary coverage for each of the 47 countries (total and subgroups) was increased by a vector of correction factors derived from Sample A countries, corresponding to the income group to which the country belongs. 19 billion were people living in households without any connection to the social protection system: none of their members received a social protection benefit or contributed to a social insurance scheme. Another 400 million were people living in relative poverty—in the poorest quintile of each of the 130 LICs and MICs—receiving low, inadequate benefits. These households, which are often part of the lowest income groups, do not have sufficient support to overcome poverty, cope with economic shocks and crises, or pursue improved opportunities in the labor market during a period of rapid global change. Consistent with the strategy of progressive realization of the USP, the 2 billion missed people represents the immediate challenge faced by LICs and MICs, and the international development community. Figure II.1. 2 Billion People in LICs and MICs Are Missed or Inadequately Covered by Social Protection A. Billions of people in LICs and MICs by income level B. Share of population in LICs and MICs by income level and region 100 11 10 18 6 26 20 21 90 26 28 6 Share of total population 80 4 8 8 6 70 6 5 29 30 5 8 11 71 60 16 78 22 27 50 40 72 30 59 54 54 60 52 47 45 5 20 3 5 3 10 16 19 0 Low-income East Asia & Pacific South Asia Europe & Central Asia Upper-middle income Latin America & Caribbean Sub-Saharan Africa Middle East & North Africa Total Lower-middle income Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data from 2022 or most recent available survey (https://www.worldbank.org/aspire). Note: Figure shows coverage of direct and indirect beneficiaries and social insurance contributors in 130 low- and middle- income countries. For sample size, see Table I.1. Q1 = first (poorest) quintile; SI= social insurance; SP = social protection. While the gap remains substantially higher in LICs, in absolute terms, more people are not covered in MICs (Figure II.1). The share of population missed by the social protection system 20 is strongly related to the country’s level of economic development. The most severe gaps are found in LICs where an average of 77 percent of the population receives no social protection benefit and an additional 4 percent receive inadequate benefits. In contrast, the social protection systems in UMICs miss only 10 percent of their populations and provide inadequate coverage to only 6 percent. However, the picture changes radically when coverage gaps are measured in absolute terms (in other words, in millions of people). The absolute number of people not covered by social protection is substantially higher in MICs than in LICs (1.2 billion versus 500 million), which reflects the reality that around 90 percent of the people now live in MICs and only 10 percent in LICs. The majority of the 2 billion people who are not covered are located in Sub-Saharan Africa, South Asia, and East Asia regions (Figure II.2.). In relative terms, most regions fail to cover 10–26 percent of their total population and provide inadequate coverage to 4–8 percent of the population (Figure II.1. b). Sub-Saharan Africa is notable, with 71 percent of the region’s population not covered, while 5 percent receive inadequate coverage. Figure II.2. People in LICs and MICs who Remain Uncovered or Inadequately Covered by Social Protection Source: Original map for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Coverage consists of direct and indirect beneficiaries and social insurance contributors. Map is based on 130 low- and middle-income countries. For sample size see Table I.1. M = million. 21 Figure II.3. The Coverage Gap Increases Substantially Once China and India Are Excluded A. All low- and middle-income countries B. Low and middle-income countries excluding China and India 26% 52% 6% 16% SP beneficiary SI contributor, not SP beneficiary SP beneficiary covered inadequately, Q1 Missed Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data for 2022 or most recent available survey (https://www.worldbank.org/aspire). Note: Figure shows coverage of direct and indirect beneficiaries and social insurance contributors in 128 low- and middle- income countries. For sample size see Table I.1. Q1 = first (poorest) quintile, generated using pretransfer welfare; SI= social insurance; SP = social protection. Coverage gaps become much more pronounced once China and India, the world’s two most populous countries, are excluded, rising from 32 to 49 percent (Figure II.3). The largest increase is in the share of population that is missed by social protection systems, which increases from 26 to 42 percent. This signals the existence of a large gap in social protection coverage in many LICs and MICs. The averages reported for Europe and Central Asia and Latin America and the Caribbean regions do not include 21 countries which are ‘new’ HICs, whose coverage is as high as in China and India. If these are included, the share missed by social protection will be reduced in these two regions as well. The population missed by the social protection systems of LICs and MICs is uniformly distributed across quintiles, which suggests room for improvements for the progressive realization of the USP, by first addressing the poorest. Each quintile has 0.31–0.33 billion missed people. The poorest quintile also has 0.35 billion people with inadequate social protection support. This category does not exist for the upper quintiles. The data also suggest 22 that the share of people in households that only contribute to social insurance and do not receive social protection benefits increases across quintiles. Figure II.4. Distribution of the 2 Billion Missed, or Inadequately Covered if Poor, by Social Protection by Quintiles 1.40 1.26 1.26 1.26 1.26 1.26 1.20 0.32 0.33 0.33 0.33 0.31 Billions of people 1.00 - - - - 0.15 0.20 0.80 0.35 0.25 0.35 0.60 0.09 0.40 0.79 0.73 0.68 0.61 0.50 0.20 - Q1 Q2 Q3 Q4 Q5 SP beneficiary SI contributor, not SP beneficiary SP beneficiary covered inadequately, Q1 Missed Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data for 2022 or most recent available survey (https://www.worldbank.org/aspire). Note: Figure shows coverage of direct and indirect beneficiaries and social insurance contributors in 130 low- and middle- income countries. For sample size see Table I.1. Q1 = first quintile (poorest); Q2 = second quintile; Q3 = third quintile; Q4 = fourth quintile; Q5 = fifth quintile (richest), using pretransfer welfare; SI= social insurance; SP = social protection. 3. Half-empty glass for the poorest quintile: 53 percent of the poor are missed or inadequately covered by social protection Social protection systems also fail to provide adequate coverage to nearly half of all people living in the poorest quintile in the 130 LICs and MICs. The coverage and adequacy gap are significantly higher for the people in the poorest quintile in LICs and MICs. About 53 percent of the 1.26 billion people from the poorest quintile are either not covered (25 percent) or covered inadequately (28 percent) (Figure II.5). Compared to the global coverage (Figure II.1.), the share of people not covered or inadequately covered in the poorest quintile is uniformly higher. The largest gaps are in LICs (97 percent) and in Sub-Saharan Africa (94 percent). The share of those missed by the social protection system falls significantly with the income level 23 of the country, while the share of the population in relative poverty and covered inadequately is significant in all country settings (from LICs to UMICs). Figure II.5. The Coverage Gap Rises to Almost Half for Those in the Poorest Quintile A. Billions of people in LICs and MICs in the poorest B. Share of total population in LICs and MICs in the poorest quintile, by quintile, by income level income level and regions Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data for 2022 or most recent available survey (https://www.worldbank.org/aspire). Note: Figure shows coverage of direct and indirect beneficiaries and social insurance contributors in 130 low- and middle- income countries. For sample size see Table I.3. Q1 = first (poorest) quintile, generated using pretransfer welfare; SI= social insurance; SP = social protection. 4. In countries with significant levels of extreme poverty, 87 percent are missed or inadequately covered by social protection In 73 EDEs with extreme poverty rate of over 2 percent,8 87 percent of the extreme poor are missed or are inadequately covered by social protection (Figure II.6.). This share rises to 98 8The estimate covers 73 countries with extreme poverty rates equal to or higher than 2 percent of the population. A total of 80 EDEs with no or low extreme poverty rates are not included in the analysis, given the lack of reliable survey estimates in countries where extreme poverty is below this threshold. These 80 countries accounted for about 3 percent of the total number of extreme poor in EDEs in the period analyzed in this report. 24 percent in LICs. The 73 countries with levels of extreme poverty of above 2 percent comprise almost half a million people who are either missed or insufficiently protected. In absolute numbers, the largest number of extreme poor are in LICs (0.25 billion) and LMICs (0.21 billion). In relative terms, the most severe coverage gaps exist in LICs (98 percent) and Sub-Saharan Africa (97 percent). As the income level of the country increases, the share of the extreme poor population missed by the social protection system falls, especially in UMICs. However, the share of those with inadequate coverage increases in these countries, signaling a problem of adequacy of benefits in the 15 UMICs that have significant extreme poverty rates. Figure II.6. In LICs, Coverage Gaps Reach 98 Percent among the Extreme Poor A. Billions of people in LICs and MICs in extreme poverty, B. Share of total population in LICs and MICs in extreme by income level poverty, by income level and regions 100 0.50 90 19 16 Share of popualtion in extreme poverty 0.50 80 39 37 42 Billions of people in extreme poverty 70 60 57 21 0.40 68 60 43 2 79 0.34 98% 83 14 50 0.30 26 15 27 0.25 0.21 40 2 0.20 30 21 5 1 61 0.01 0.01 35 0.20 0.13 0.01 20 19 0.10 0.001 3 36 30 34 30 0.10 0.01 10 2 1 15 1 18 0.003 0.003 0.04 0.03 10 1 16 3 0.05 0.04 0.01 0 5 2 0.03 Latin America Total South Asia East Asia & Lower-middle Low-income Central Asia Middle East & Upper-middle Sub-Saharan 0.00 & Caribbean North Africa Europe & Pacific income income Africa Total Low-income Lower- Upper- middle middle income income Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data for 2022 or most recent available survey (https://www.worldbank.org/aspire). Note: Estimates are based on observations from 73 countries that have extreme poverty rates of 2 percent of the population or greater. Figure shows coverage of direct and indirect beneficiaries and social insurance contributors. For sample size see Table I.2. SI= social insurance; SP = social protection; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). 25 5. HICs have succeeded in reducing the coverage gap to low levels A group of HICs who recently graduated from MIC status have succeeded in eliminating extreme poverty and covering almost the entire population from the poorest quintile. ASPIRE tracks the size, composition, evolution, and distributional performance indicators not only for LICs and MICs but also for 23 HICs with whom the World Bank cooperates or has cooperated. Typically, these are countries that have recently graduated to HIC status, that is, ‘new’ HICs. They are a useful comparator for LICs and MICs. This country group has succeeded in eliminating or bringing extreme poverty to very low levels. Moreover, as illustrated in Figure II.7., the new HICs have almost closed the coverage gap for the poorest quintile of their populations: only 6 percent of the population of these 23 HICs are missed by social protection, with another 6 percent not adequately covered. In these countries, USP has become a reality. Figure II.7. The ‘New’ HICs Have Almost Closed the Coverage Gaps of the Poorest Quintile A. Billions of people in low/middle and high-income B. Share of total population in low/middle and high-income countries, by income level countries in extreme poor, by income level and regions 7 100 6 6.3 90 6 26 Share of total population 6 80 16 1.6 2 5 70 6 billion Billion of people 0.4 60 4 16 1.0 50 3 40 0.02 71 2 30 3.3 52 0.05 20 1 0.22 0.31 10 0 0 Low and Middle High income Low and Middle High income income income Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data for 2022 or most recent available survey (https://www.worldbank.org/aspire). Note: Figure shows coverage of direct and indirect beneficiaries and social insurance contributors in 153 low- and middle- income and high-income countries. For sample size see Table I.1. Q1 = first (poorest) quintile, generated using pretransfer welfare; SI= social insurance; SP = social protection, XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). 26 IIB. The Coverage Gap: Cross-country Performance While absolute, population-weighted measures of social protection coverage provide insights into the magnitude of worldwide challenges, they are often influenced by a few large countries. Therefore, these trends may not be very effective in identifying common policy challenges across different nations. To gain a better understanding of the shared factors contributing to significant coverage gaps in LICs and MICs, this section will examine the performance and challenges faced by the individual country systems. Consequently, the figures presented in this section and the rest of the paper are cross-country averages, not population-weighted averages (used in Section IIA), to ensure that the analysis of the social protection systems of smaller countries and of systems in more populous countries are given the same importance. Another methodological distinction between the coverage figures reported in this and the subsequent sections, compared to those presented in Section IIA, is the scope of social protection coverage considered. Section IIA employs a comprehensive definition of social protection coverage, which includes both the receipt of benefits and contributions to social insurance. In contrast, the remainder of the paper presents figures based on a narrower definition of social protection coverage that considers only the receipt of benefits. In addition, the rest of the paper reports on all ASPIRE countries, including the ‘new’ HICs. Based on household surveys from 2022 or the most recent available for 73 EDEs,9 on average half of the total population in EDEs currently access social protection benefits, resulting in a coverage gap of 49 percent. As shown in Figure II.8., 51 percent of the population in EDEs received, directly or indirectly, a social protection transfer during 2017–2022.10 Despite 9 All statistics on the distributional performance of social protection programs are derived from the ASPIRE database of the World Bank and correspond to 73 emerging and developing countries with a recent household survey (carried out between 2017 and 2022). These household surveys have been used by the respective countries to track, among other things, the level of poverty as defined by the national poverty lines and by the World Bank against international poverty lines. The 73 countries account for 46 percent of the population of emerging and developing countries. 10 Coverage is calculated using household survey data, by dividing the number of individuals who live in a household receiving at least one social protection benefit by the total number of individuals in the population. 27 methodological differences, the World Bank data indicate a significant and persistent coverage gap of similar magnitude to the ILO’s latest estimates that 47 percent of the global population were effectively covered by at least one social protection benefit (ILO, 2021a). Social protection coverage increases, on average, with a country’s economic development and is therefore lower in regions with weaker levels of economic development. Overall, access to social protection benefits is only 25 percent in LICs and 42 percent in LMICs, compared to 61 percent in UMICs and 82 percent in the new HICs (Figure II.8.). While coverage is higher on average across all countries for the poorest quintile and the extreme poor population, the increase is smaller in LICs (only 3 percentage points difference) compared to LMICs (11 percentage points), UMICs (21 percentage points), and HICs (15 percentage points). Figure II.8. Proportion of the Total, Poorest Quintile, and Extreme Poor Population Receiving Various Types of Social Protection Benefit, by the Country’s Level of Economic Development (percentages) 97 98 100 89 88 90 82 82 83 Coverage (% of population) 80 75 70 65 66 66 70 62 62 61 58 56 60 51 53 53 48 50 42 45 42 40 41 42 40 35 32 28 27 28 26 30 24 25 25 22 22 19 20 11 8 8 6 6 10 5 5 4 3 3 5 5 4 4 22 12 12 1 0 Total Q1 XP Total Q1 XP Total Q1 XP Total Q1 XP Total Q1 XP Total Low-income countries Lower-middle-income Upper-middle-income High-income countries countries countries All social protection All social assistance All social insurance All labor market Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Figure shows coverage of both direct and indirect beneficiaries for 2022 or most recent year for which data are available and is based on 73 observations, which include 67 low- and middle-income countries (LIC:12, LMIC:24, UMIC: 31) and 6 high- income countries monitored by ASPIRE. For XP the sample is reduced to 69 observations – Belarus does not have XP, and data on XP are not available for Zimbabwe, Bosnia and Herzegovina and Tunisia. Coverage is determined as follows: (number of individuals in the total population, poorest quintile, or living in extreme poverty who live in a household where at least one member receives the transfer) / (number of individuals in the total population or group). Although it captures the largest programs, the figure may underestimate total social protection coverage, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross-country averages. Q1 = poorest 20 percent; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). 28 The average coverage gap of the poorest quintile is about 35 percent, smaller than the gap for the total population of low-, middle-, and high-income countries.11 Among the bottom 20, around 65 percent of the population reside in households where at least one person receives benefits. This higher coverage is principally due to the targeting of social assistance programs for the poor. As Figure II.8.Figure II.8. illustrates, non-contributory social assistance programs are the largest contributor to social protection receipt, at 58 percent, followed by contributory social insurance programs at 22 percent, and labor market programs at 5 percent.12 Meanwhile, the gap in coverage of the extreme poor is slightly lower still, at 30 percent, with 70 percent of the extreme poor receiving a social protection benefit (Figure II.8.). As in the case of relative poverty, social assistance is the major channel for social protection receipt among the extreme poor (62 percent), followed by social insurance (24 percent) and labor market programs (4 percent), with some overlap between the two types of social protection programs. There is substantial regional variation in the coverage of social protection. For example, in Sub-Saharan Africa, social protection coverage is less than half the level of Europe and Central Asia, received by only 28 percent of the population, with large gaps even for the poorest quintile and for the extreme poor population (Figure II.9.). 11 The quintile distribution is based on per capita consumption or income of the household, before receiving any social protection transfers (50 percent of transfers from social insurance and the labor market and 100 percent of social assistance transfers). 12 The total coverage is less than the sum of these programs because some households receive multiple benefits. 29 Figure II.9. Proportion of the Total, Poorest Quintile, and Extreme Poor Population Receiving Various Types of Social Protection Benefit, by World Bank Group Region (percentages) 98 100 91 88 85 90 84 82 81 82 79 80 Coverage (% of population) 70 71 73 70 70 69 70 65 66 66 65 62 62 61 58 57 57 58 60 54 51 51 49 53 50 46 44 42 43 43 41 44 42 40 39 40 33 35 36 35 33 30 22 24 19 20 17 17 20 12 10 10 7 6 6 10 5 5 4 2 3 4 5 6 5 2 43 4 44 35 46 1 00 00 00 0 Total Q1 XP Total Q1 XP Total Q1 XP Total Q1 XP Total Q1 XP Total Q1 XP Total Q1 XP Total Europe & Central Latin America & East Asia & Pacific South Asia Sub-Saharan Africa Middle East & Asia Caribbean North Africa All social protection All social assistance All social insurance All labor market Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Figure shows coverage of both direct and indirect beneficiaries for 2022 or most recent year for which data are available and is based on 73 observations, which include 67 low- and middle-income countries (LIC:12, LMIC:24, UMIC: 31) and 6 high- income countries monitored by ASPIRE. For XP the sample is reduced to 69 observations. Although it captures the largest programs, the figure may underestimate total social protection coverage, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross-country averages. Q1 = poorest 20 percent; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). There is some overlap between recipients of various social protection instruments, with the types and number of different instruments increasing with the income level of the country. Most countries provide a combination of social protection programs to serve different categories of the population and to address different types of risk for the same population group. Some of these programs overlap at the household level. Household surveys are the main data source to account for the overlap, enabling the calculation of total coverage without double counting beneficiaries of different programs. Consistent with the goal of progressive realization of the USP, we look at the mix of programs and their overlap that cover the extreme poor and the poorest quintile in EDEs (Figure II.10.). Two main stylized facts emerge: (a) social assistance programs are the backbone of the social protection benefit receipt across all EDEs, and (b) the overlap between the different types of social protection programs—social 30 assistance, labor market programs, or social insurance programs—increases with the level of economic development of the country. For example, among LICs there is almost no overlap, with almost all social protection benefits received from social assistance programs (23 percent of a total of 25 percent). Social insurance alone accounts for an additional 4 percentage points in LMICs, 16 in UMICs, and 14 in HICs. As countries’ per capita income grows, the share of households receiving social protection benefits from multiple social protection instruments (social assistance, social insurance, labor market programs) rises, from 2 percent among LMICs to 15 percent for UMICs and 29 percent for HICs. In lower-income regions, social protection receipt is almost exclusively from social assistance programs—for example, with no or very little overlap between the social protection thematic areas in Sub-Saharan Africa. Figure II.10. Proportion of the Total, Poorest Quintile, and Extreme Poor Population Receiving Various Types of Social Protection Benefit, by Region and the Country’s Level of Economic Development (percentages) 97 98 0.1 98 0.2 100 0.3 89 88 0.3 0.6 0.4 90 0.3 82 82 82 84 2 82 0.3 0.3 3 34 29 Coverage (% of population) 80 19 37 9 4 1 70 0.3 15 25 11 31 69 70 65 0.2 65 0.6 18 0.2 2 61 61 0.3 29 62 0.2 3 12 16 0.4 57 1 58 1 60 10 2 0.2 53 0.2 10 66 10 0.5 51 10 17 15 1 14 26 3 11 0.3 1 10 1 13 8 50 10 0.3 0.2 42 4 1 0.2 36 13 43 14 44 43 7 14 1 0.2 36 40 1 1 16 5 2 2 35 3 3 10 4 72 28 29 67 11 2 30 25 28 59 2 1 54 54 46 48 49 49 20 45 47 0.3 47 47 41 41 38 39 40 38 33 1 35 34 32 31 27 27 26 30 10 23 18 0 Total Total Total Total Total Total Total Total Total Total XP Q1 Q1 Q1 XP Q1 XP Q1 XP Q1 XP Q1 XP Q1 XP XP Q1 XP Q1 XP Total Low-income Lower- Upper- High-income Europe & Latin East Asia & South Asia Sub-Saharan countries middle- middle- countries Central Asia America & Pacific Africa income income Caribbean countries countries Only social assistance Only social insurance Only labor market Social assistance and others Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Figure shows coverage of both direct and indirect beneficiaries for 2022 or most recent year for which data are available and is based on 73 observations, which include 67 low- and middle-income countries (LIC:12, LMIC:24, UMIC: 31) and 6 high- income countries monitored by ASPIRE. For XP the sample is reduced to 69 observations, Middle East and North Africa (n=3) is not included in the graph because the number of observations is too small, but these are included in total numbers. Although it captures the largest programs, the figure may underestimate total social protection coverage, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple 31 cross-country averages. Q1 = poorest 20 percent; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). IIC. Progress in Closing the Beneficiaries’ Coverage Gap Over the Last Decade (circa 2010 versus circa 2022) This section takes a broader view and tracks the progress of the average EDEs in increasing access to social protection benefits in the decade before the onset of the COVID-19 shock. This information is critical for quantifying the progressive realization of the USP. The empirical base includes 73 EDEs with comparable surveys implemented around 2010 and during or before 2022. Social protection coverage is defined as the share of population living in households where at least one member receives benefits from social assistance, social insurance, and labor market programs. Coverage indicators, aggregated for countries classified by their level or income or by region, are not population weighted, thus giving an equal share to each country irrespective of their size. The 73 countries with comparable surveys represent 58 percentage of the total population of EDEs in 2022. Cross-country data over time show substantial progress in expanding social protection coverage over the past decade, across EDEs of all income levels. As shown in Figure II.11., in our household survey sample, the share of households receiving social protection benefits grew from, on average, 41 percent around 2010 to 51 percent 12 years later. Expansion occurred in all groups of economies but was most pronounced in low-income economies, where coverage more than doubled (albeit from a low base, rising from 11 percent to 25 percent). Within the population, social protection continues to be progressively distributed on average, with benefit receipt highest among the poorest quintile and decreasing as household wealth increases. For the poorest quintile, social protection coverage increased by an average of 11 percentage points over the decade, from 54 percent to 65 percent. For households living in extreme poverty, average coverage increased from 59 percent to 70 percent. 32 Figure II.11. Expansion of Access to Social Protection Benefits over 2010–2022 97 98 100 92 93 89 90 82 82 82 Coverage (% of population) 77 80 70 73 70 65 62 59 61 54 56 60 51 53 50 41 42 43 36 40 28 28 28 30 25 20 11 11 11 10 0 XP Total XP XP XP XP Total Total Total Total Q1 Q1 Q1 Q1 Q1 Total Low-income countries Lower-middle-income Upper-middle-income High-income countries countries countries Circa 2010 Circa 2022 Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: “Circa 2022” refers to data from 2022 or the most recent available year within the period 2015 to 2022. “Circa 2010” includes data from 2010 or any year available between the period 2006 to 2014. Figure shows coverage of both direct and indirect beneficiaries and is based on 73 observations, which include 67 low- and middle-income countries (LIC:12, LMIC:24, UMIC: 31) and 6 high-income countries monitored by ASPIRE. For XP the sample is reduced to 69 observations. Although it captures the largest programs, the figure may underestimate total social protection coverage, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross- country averages. Q1 = poorest 20 percent; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). Nevertheless, progress is slow: at current rates of expansion, covering the poor—let alone achieving universal coverage—will take several decades. If the expansion of the last decade were to continue at the same rate, access to social protection benefits among those living in extreme poverty and among the poorest 20 percent would increase by only around 1 percent per year. At this pace, it would take until 2043 for those living in extreme poverty to be fully covered and until 2045 for the poorest 20 percent of households in LICs and MICs to be covered. That would be a decade and a half behind the target year for achieving the SDGs. Social assistance was and remains the social protection area that contributes the most to closing coverage gaps for the poorest quintile and the extreme poor (Figure II.12.). Social assistance programs account for, on average, 80 percent of total access to social protection 33 benefits. For the extreme poor, social assistance accounts for 85 percent of total social protection coverage as well as for the poorest quintile. In LICs, coverage of social assistance programs doubled among the total population (from 12 percent to 22 percent), increased by 2.2 times among the poorest quintile, and more than doubled for the extreme poor. However, the coverage gap in LICs remains substantial. This highlights the need for both EDE governments and international financial institutions (IFIs) to maintain and expand their efforts to close this gap, especially for the poorest quintile and the extreme poor, to bring the progressive expansion of USP closer to fruition by 2030. Figure II.12. Expansion of Social Assistance Coverage as the Main Driver for the Increase in Social Protection Coverage from 2010 to 2022 100 90 80 75 71 % of population 66 63 66 66 66 70 62 58 56 60 53 53 53 53 53 47 45 50 42 40 41 42 40 33 35 27 25 25 30 22 20 12 12 12 10 0 Q1 (68) Q1 (10) Q1 (22) Q1 (30) XP (64) Q1 (6) Total (68) Total (10) XP (10) Total (22) XP (20) Total (30) XP (28) Total (6) XP (6) Emerging and Low-income countries Lower-middle-income Upper-middle-income High-income Developing Economies countries countries countries Circa 2010 Circa 2022 Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: “Circa 2022” refers to data from 2022 or the most recent available year within the period 2015 to 2022. “Circa 2010” includes data from 2010 or any year available between the period 2006 to 2014. Figure shows coverage of both direct and indirect beneficiaries and is based on 68 observations, which include 62 low- and middle-income countries (LIC: 10, LMIC: 22, UMIC:30) and 6 high-income countries monitored by ASPIRE. For XP the sample is reduced to 69 observations. Although it captures the largest programs, the figure may underestimate total social protection coverage, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross- country averages. Q1 = poorest 20 percent; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). 34 In contrast to social assistance, access to social insurance benefits saw little growth in the decade before the COVID-19 shock (Figure II.13.). Overall, social insurance benefits are received by 19 percent of the total population in EDEs, 22 percent of the population from the poorest quintile, and 24 percent of the extreme poor population. Social insurance receipt is significant in UMICs and HICs, where the share of formal employment is large (over 40 percent), but is almost absent from LICs and low in LMICs. This paper does not provide an estimate of the decadal chance in the coverage of labor market programs (ALMPs and PLMPs), due to the uneven inclusion of such programs in household surveys.13 Figure II.13. Receipt of Social Insurance Benefits (from Contributory Pensions or Other Social Insurance Programs) between 2010 and 2022 100 90 80 % of population 70 60 48 50 42 41 41 39 32 35 35 40 26 27 29 22 24 23 30 1719 19 21 20 78 68 811 10 22 11 11 0 Total (27) Total (57) Total (8) Q1 (8) Total (16) Q1 (57) Q1 (16) Q1 (27) Total (6) Q1 (6) XP (54) XP (8) XP (15) XP (25) XP (6) Emerging and Low-income countries Lower-middle-income Upper-middle-income High-income Developing Economies countries countries countries Circa 2010 Circa 2022 Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: “Circa 2022” refers to data from 2022 or the most recent available year within the period 2015 to 2022. “Circa 2010” includes data from 2010 or any year available between the period 2006 to 2014. Figure shows coverage of both direct and indirect beneficiaries and is based on 57 observations, which include 51 low- and middle-income countries (LIC: 8, LMIC: 16, UMIC:27) and 6 high-income countries monitored by ASPIRE. For XP the sample is reduced to 64 observations. Although it captures the largest programs, the figure may underestimate total social protection coverage, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross- country averages. Q1 = poorest 20 percent; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). 13 Only 16 out of 73 countries report labor market programs in their surveys and there is an under-representation of the large or moderately sized labor market programs even in these surveys. 35 III. Benefit Levels: Variable and Insufficient to Fill the Income Shortfall of the Extreme Poor or for Those in the Poorest Quintile for Most Social Protection Programs, in the Majority of EDEs Alongside coverage, the adequacy of social protection benefits is a key factor for determining the extent to which appropriate social protection systems and measures have been established (SDG Target 1.3) and social protection’s contribution to extreme poverty and poverty reduction (Targets 1.1 and Targets 1.2). This section uses the most recent household surveys available for 2017–2022 to examine the benefit level of social protection programs relative to the welfare (meaning either the consumption or income level) of recipient households.14 The ratio of social protection benefits received by beneficiary households to their total welfare is also known as the adequacy ratio.15 The benefit level— and therefore the adequacy ratio—of different social protection programs varies based on their differing objectives as well as diverse contextual factors, notably budget constraints. Social protection programs have different objectives, from full income replacement (for example, for contributory pensions, unemployment benefits, public works, or social pensions) to filling of specific consumption or income gaps, often combined with behavioral nudges, complementary services, or social inclusion packages (for social assistance programs). By design, income-replacement benefits tend to have a higher benefit level than programs that aim only to fill specific consumption gaps. However, even if social assistance programs are not designed to provide full replacement of income for the average household in the population, they are often intended to alleviate income shortfalls for the extreme poor or poorest quintile in the population. The success of such programs in reducing (extreme or relative) poverty 14 The sample surveys used in this section are the same as those mentioned in Section IIB (73 countries), conditional on availability of benefit adequacy data. Not all 73 surveys collect information of the benefit values: 60 countries have information to calculate benefit adequacy for total population and the population in the poorest quintile; 48 countries have information to calculate benefit adequacy for the extreme poor for total social protection. These numbers drop further for the thematic areas. 15 The adequacy ratio is defined as the percentage that transfer amounts received by beneficiaries of a group contribute to the total welfare of the beneficiaries of that group. The population income distribution (for example, the poorest quintile or wealthiest quintile) is calculated based on the post-transfer welfare aggregate. 36 therefore depends partly on how much they contribute to the income shortfall required to eliminate poverty (the ratio of per capita consumption of the household relative to the poverty line in question). Unsurprisingly, given the differing objectives outlined above, recent household survey data from EDEs show that income replacement programs deliver the most substantial benefit levels (Figure III.1.). On average, social insurance programs account for 33 percent of the beneficiary household’s welfare. Labor market programs and social assistance transfers contribute far less, accounting for 10–11 percent of the welfare of their beneficiaries. In relative terms, the adequacy ratio is higher for the extreme poor (80 percent for social insurance programs, 42 percent for labor market programs, and 37 percent for social assistance programs) and for the poorest quintile (48 percent for social insurance programs, 19 percent for labor market programs, and 20 percent for social assistance programs). This higher adequacy ratio for the extreme poor and the poorest quintile is driven by the fact that their consumption or income levels are inherently lower than those of the wealthier population. Figure III.1. Adequacy Ratio of Social Protection Benefits Received by the Total, Poorest Quintile, and Extreme Poor Population, by Social Protection Type 100 % of beneficiaries' welfare 80 68 58 60 49 46 40 32 36 31 27 22 18 15 20 11 0 Total Social Protection Social Assistance Social Insurance Labor Market Programs Total population Poorest quintile Extreme poor Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). 37 Note: Figure is based on 60 observations, which include 55 low- and middle-income countries (LIC: 9, LMIC: 18, UMIC:28) and 5 high-income countries monitored by ASPIRE. Countries with extreme poverty incidence close to zero have been removed from the extreme poverty sample. Adequacy is calculated 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). Although it captures the largest programs, the figure may underestimate total social protection adequacy of benefit, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross-country averages. Q1 = poorest quintile, generated using post-transfer welfare; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). The adequacy ratio of social protection benefits increases, on average, with the country’s income level (Figure III.2.). On average, the adequacy of benefits in UMICs is three times higher than for LICs (32 percent and 11 percent, respectively). In the poorest quintile, and mainly among the extreme poor, adequacy is higher in large part because of the low consumption of the population in this group and not because the level of benefit is, in absolute terms, higher. Figure III.2. Adequacy Ratio of Social Protection Benefits Received by the Total, Poorest Quintile, and Extreme Poor Population, by Country Income Level 100 90 % of beneficiaries' welfare 80 67 70 57 60 50 46 43 38 37 40 32 34 27 25 26 30 24 23 16 19 20 10 0 Total Low-income countriesLower-middle-income Upper-middle-income High-income countries countries countries Total population Poorest quintile Extreme poor Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Figure is based on 60 observations, which include 55 low- and middle-income countries (LIC: 9, LMIC: 18, UMIC:28) and 5 high-income countries monitored by ASPIRE. Countries with extreme poverty incidence close to zero have been removed from the extreme poverty sample. Although it captures the largest programs, the figure may underestimate total social protection adequacy of benefit, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross-country averages. Q1 = poorest quintile, generated using post- transfer welfare; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). 38 The positive correlation of the adequacy ratio with the type of social protection program (income replacement or not), the country’s income level, and the household poverty level is also observed when information is fully disaggregated across these dimensions (Figure III.3.). The adequacy ratio increases monotonically with the income of the country, it is higher for social insurance compared to labor market or social assistance programs, and it is higher among the poorest quintile and extreme poor compared to the total population. Figure III.3.Adequacy Ratio of Social Protection by Income Level for the Total Population, the Poorest Quintile, and the Extreme Poor 100 100 100 100 % of beneficiaries welfare (income or 90 77 80 68 67 70 60 consumption) 58 56 57 60 53 49 48 49 47 47 46 50 41 40 43 40 38 36 37 40 32 31 30 34 27 25 26 26 26 30 22 21 24 23 23 23 18 19 18 15 16 15 20 11 11 11 14 14 14 11 10 6 5 5 10 0 0 0 3 0 Total Q1 XP Total Q1 XP Total Q1 XP Total Q1 XP Total Q1 XP Total Low-income Lower-middle- Upper-middle- High-income countries income countries income countries countries All social protection All social assistance All social insurance All labor market Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Figure is based on 60 observations, which include 55 low- and middle-income countries (LIC: 9, LMIC: 18, UMIC:28) and 5 high-income countries monitored by ASPIRE. Countries with extreme poverty incidence close to zero have been removed from the extreme poverty sample. Although it captures the largest programs, the figure may underestimate total social protection adequacy of benefit, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross-country averages. Q1 = poorest quintile, generated using post- transfer welfare; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). 39 Across regions, the adequacy ratio of social protection benefits is highest in Europe and Central Asia (40 percent), followed by East Asia and the Pacific, and relatively modest in the other regions (Figure III.4.). Among the poorest quintile and extreme poor, benefit adequacy is mainly higher than for the total population in Europe and Central Asia (55 percent and 156 percent compared to 40 percent); the differences across the other regions are lower. Figure III.4. Adequacy Ratio of Social Protection by Region for Total Population, the Poorest Quintile, and the Extreme Poor 123 % of beneficiaries' welfare 120 100 80 55 60 46 46 50 41 39 35 34 40 27 32 27 29 23 21 20 22 24 20 10 9 0 0 Total East Asia & Europe & Latin America Middle East & Sub-Saharan South Asia Pacific Central Asia & Caribbean North Africa Africa Total population Poorest quintile Extreme poor Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Figure is based on 60 observations, which include 55 low- and middle-income countries (LIC: 9, LMIC: 18, UMIC:28) and 5 high-income countries monitored by ASPIRE. Countries with extreme poverty incidence close to zero have been removed from the extreme poverty sample. Although it captures the largest programs, the figure may underestimate total social protection adequacy of benefit, because household surveys may not include all programs that exist in each economy. Aggregated indicators have been calculated using simple cross-country averages. Q1 = poorest quintile, generated using post- transfer welfare; XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). When considering the adequacy ratio of social protection benefits relative to the income shortfall required to eliminate extreme poverty, current benefit levels contribute substantially overall to reducing extreme poverty for those covered, but they are not sufficient in contexts where extreme poverty is prevalent (Figure III.5.A). While it is difficult to assess the adequacy of social protection benefits against the multiple objectives of these programs, this paper aims to quantify the contribution of social protection benefits to 40 reducing the welfare (income or consumption) shortfall 16 of extreme poor recipients. The 14F average consumption of people living in extreme poverty is 40 percent below the international extreme poverty line of US$2.15 per day in EDEs. On average, social protection transfers in EDEs reduce the pre-transfer shortfall of recipients by one-quarter (or by 10 percentage points). The higher the prevalence of extreme poverty, the higher the shortfall. However, the contribution of social protection to reducing the gap in these contexts is marginal, oscillating between 0 and 3 percentage points in countries with extreme poverty prevalence of between 2 and 40 percent. By contrast, for countries with ‘no’ extreme poor (meaning those with less than 2 percent prevalence of extreme poverty), the contribution of social protection transfers to reducing the income shortfall is higher as benefit amounts tend to be more generous, reducing the gap by almost 40 percent (from −42 percent to −27 percent). 16 We use income shortfall as a general term to measure either the consumption or income shortfall of the individuals in extreme poverty or from the poorest quintile. The income shortfall measures the relative gap in welfare of beneficiaries compared to the (extreme) poverty line before the receipt of social protection transfers. The contribution of social protection transfers to reducing the income shortfall measures the transfer amount relative to the (extreme) poverty line. The difference between the two is the net income shortfall: that is, the increase in benefits (and budget) needed to bring the individuals to the (extreme) poverty line (to eliminate extreme poverty), expressed as a share of the (extreme) poverty line. The same formula is applied in the subsequent paragraph in relation to the relative poverty line (the welfare threshold of the poorest 20 percent of the population). 41 Figure III.5. Social Protection Benefits Make Up a Large Share of the Pre-transfer Consumption of Households in Extreme Poverty and in the Poorest Quintile—But Not in LICs or in Countries with High Extreme Poverty Rates A. Reduction in the income shortfall of individuals in B. Reduction in the income shortfall of individuals in poorest quintile after the receipt of social protection extreme poverty after the receipt of social protection benefits benefits 100 100 % of relative poverty line 80 80 60 60 28 % OF POVERTY LINE 40 16 18 6 40 15 20 1 10 0 20 1 0 -20 0 -40 -20 -27 -27 -26 -31 -25 -27 -30 -60 -43 -45 -40 -80 -58 -30 -31 -30 -27 -60 -40 -42 -100 -50 -49 -80 Total Low Lower Upper High -100 income middle middle income income income Total 2-20 20-40 No XP Shortfall in relation to relative poverty line Shortfall relative to extreme poverty line Average SP contribution to reducing shortfall Average SP contribution to reducing shortfall Net shortfall Net shortfall Source: Original calculation using ASPIRE household survey data. https://www.worldbank.org/aspire Note: This sample includes countries for which ASPIRE has recent survey data (Total:73, LIC:15, LMIC: 23, UMIC: 29 and HIC:6). The extreme poor panel excludes Somalia because of lack of availability of an extreme poverty rate (Total: 72, 2-20%: 19, 20- 40%:5, No XP: 48). Aggregated indicators are calculated using simple averages of country-level shortfalls and adequacy by income groups. Shortfall respective to the poverty lines is determined as follows: (mean consumption before transfer – poverty line) / poverty line. Average social protection contribution to reducing shortfall: (mean consumption after transfer – mean consumption before transfer) / poverty line. Mean consumption before transfer is calculated as mean consumption after transfer – (100 percent of social assistance transfers + 50 percent of social insurance and labor market programs transfers). This figure underestimates income shortfall and social protection contribution because household surveys do not include all programs that exist in each country. Similarly, the contribution of social protection transfers to reducing the income shortfall of the poorest quintile for all EDEs is quite significant, halving the pre-transfer income shortfall of the poorest quintile but with far fewer impacts in LICs (Error! Reference source not found.Figure III.5.B). On average, the per capita pre-transfer income of individuals from the poorest 20 percent of the population is 43 percent below the respective poverty line. The average contribution of social protection transfers to reducing the income shortfall is 37 percent (or 16 percentage points), leaving a net income shortfall (after the receipt of social protection transfers) of 27 percent. This contribution, however, varies greatly across countries based on their income level. First, the higher the income level of a country, the bigger the shortfall, partly because higher-income countries offer more generous benefits. The income 42 shortfall goes from −27 percent for LICs, to −36 percent for LMICS, to −44 percent for UMICs. Second, the average social protection contribution to reducing this shortfall increases with the income level of the country, from a very low 1 percent in LICs (highlighting the low level of benefits in these countries) to a high of 16 percent in UMICs. The net relative shortfall has less variance but remains significant: 26 percent in LICs, 25 percent in LMICs, and 28 percent in UMICs. IV. Spending: A Notable Area of Government Spending but Inadequate to Fill the Coverage and Adequacy Gaps The gaps in coverage and adequacy discussed in the above section largely reflect the levels and allocation of social protection financing. This section provides a picture of the size and composition of social protection expenditure in EDEs in 2022 using detailed social protection administrative data from 1,388 programs in 72 countries (Table IV.1.). Of the social protection programs analyzed, 21 percent correspond to labor market programs, 71 percent to social assistance, and 8 percent to social insurance.17 Regarding country income groups, almost half the data are focused on UMICs, followed by LMICs and HICs, while social protection programs implemented by LICs make up 7 percent of total programs in the sample. By region, social protection programs in Europe and Central Asia, Latin America and the Caribbean, and Sub- Saharan Africa account for almost 77 percent of total programs, while the remaining 23 percent are programs in Middle East and North Africa, East Asia and the Pacific, and South Asia. 17 Note that the ASPIRE administrative data only includes contributory pensions within social insurance programs. This implies that there might be differences in spending with other sources that collect Other social insurance (occupational injury benefits, paid sick leave benefits, contributory health, and maternity/paternity leave benefits), such as EUROSTAT and the International Labor Organization, among others. 43 Table IV.1. Total Number of Programs 2022, by Social Protection Area, Income Group, and Region Labor Social Social Social market assistance insurance protection programs By income group Low income (n=9) 20 61 16 97 Lower-middle income (n=21) 80 251 44 375 Upper-middle income (n=29) 117 456 38 611 High income (n=13) 78 209 18 305 By region Sub-Saharan Africa (n=18) 57 152 30 239 East Asia and Pacific (n=8) 20 79 11 110 Europe and Central Asia (n=23) 113 390 21 524 Latin America and Caribbean (n=10) 74 217 18 309 Middle East and North Africa (n=8) 12 76 24 112 South Asia (n=5) 19 63 12 94 Total (n=72) 295 977 116 1,388 Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). Note: Figure is based on a total of 72 observations, which include 59 low- and middle-income countries and 13 high-income countries monitored by ASPIRE. Data correspond to 2022. Aggregated indicators have been calculated using simple cross- country averages. Although social protection is an important category of public expenditure, addressing the gaps in coverage and adequacy described in Section II requires far greater investment levels, particularly in LICs. In 2022,16FEDEs spent on average 5.3 percent of GDP on social protection programs (Figure IV.1.). The share of GDP allocated to social protection is strongly correlated with the level of economic development of the country (as measured by GDP per capita). The share of GDP spent on social protection was 5.2 times higher in HICs than in LICs. For example, within HICs, Slovenia, Bulgaria, and the Czech Republic spent 12.2 percent, 11.3 percent, and 9.7 percent of their GDP, respectively, on social protection programs, whereas the figure was less than 1 percent in some LICs, such as the Democratic Republic of Congo, Ethiopia, and Liberia. Regions with a high concentration of LICs—such as Sub-Saharan Africa and South Asia—tend to have lower relative spending on social protection. The difference between the proportion of GDP spent by HICs and LICs was most pronounced for labor market programs (10.9 times higher) and social insurance programs (8.6 times higher), linked to differences in labor market formality rates and, consequently, of 44 associated formal employment-based social protection programs. Overall, the largest area of spending was on social insurance programs (3.7 percent of GDP), followed by social assistance (1.5 percent) and labor market programs (0.2 percent). However, countries with larger shares of formal employment in total employment (which are typically higher with higher income levels) have substantially higher levels of spending on social insurance and unemployment insurance programs. Except for Chile and Seychelles—which tend to invest a disproportionately large share of their spending on social assistance programs—a less pronounced difference is evident between HICs and LICs for social assistance spending. Even so, the proportion of GDP spent on social assistance was, on average, 1.9 times higher for HICs than LICs and LMICs. Not surprisingly, countries like Ethiopia and the Democratic Republic of Congo, among others, concentrate, on average, more than 85 percent of their social protection expenditure on social assistance programs. Figure IV.1. Social Protection Spending as Percentage of GDP, Unweighted Average, 2022 9 8.3 8.4 8 6.4 6.6 7 6.1 Percentage of GDP 6 5.3 5 6.2 6.3 4.2 3.7 4.1 4 3.8 5.0 3.7 2.5 3 2.7 1.6 2.7 1.7 2 1.2 0.7 2.0 1.7 0.9 1.9 2.0 1 1.5 1.3 1.6 0.8 1.0 0.8 1.2 0 0.2 0.0 0.0 0.2 0.3 0.0 0.2 0.3 0.3 0.0 0.1 Total Low- Lower- Upper- High- Sub- East Asia Europe & Latin Middle South Asia income middle- middle- income Saharan & Pacific Central America & East & income income Africa Asia Caribbean North Africa LM SA SI Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). Note: Figure is based on a total of 72 observations, which include 59 low- and middle-income countries and 13 high-income countries monitored by ASPIRE. Data correspond to 2022. Aggregated indicators have been calculated using simple cross- country averages. For sample size, as well as methodology for estimations, please refer to Tesliuc and Fonteñez (2025). Unlike 45 that for social assistance, social insurance expenditure does not represent government spending alone but also includes contributions made by pension system participants during their working years. LM = labor market programs, SA = social assistance; and SI = social insurance. Differences in the absolute level of social protection spending are substantially larger, with HICs spending 151 times more than LICs on social insurance, 152 times more on labor market programs and 30 times more on social assistance, in absolute terms. Figure IV.2. presents average real spending per capita expressed in the same level of purchasing power (US$ 2017 PPP). By this measure, in 2022 EDEs have spent on average, slightly more than US$1,000 (in 2017 PPP) per capita per year—more precisely US$1,002. While the proportions between social insurance, social assistance, and labor market programs are largely similar to Figure IV.1., the differences in spending between LIC, MICs, and HICs are much larger. Average annual per capita spending peaks at US$2,608 PPP in HICs, compared to only US$13 PPP in LICs, about 151 times higher. While the differences in real expenditure are more extreme between LICs and HICs, there are also notable differences when comparing groups that are closer to one another in terms of economic development. For example, LMICs spend approximately 16 times more in real expenditure on social protection than LICs and approximately 10 times less than HICs. UMICs also show less per capita spending than HICs, although in a much smaller magnitude, mainly due to the level of spending in Türkiye, Montenegro, Ukraine, Brazil, and Argentina, which, on average, spent more than US$2,150 (in 2017 PPP). Figure IV.2. Absolute Level of Social Protection Spending (US$ 2017 PPP) Per Capita, Unweighted Average, 2022 3,000 2,608 in constant USD PPP $2017 2,500 1,938 2,000 2,010 1,500 1,106 1,133 1,002 1,508 869 1,000 701 738 691 723 281 626 500 146 488 210 30 495 211 331 356 392 244 61 185 239 95 0 34 13 16 1 66 3 37 103 84 2 27 74 49 4 112 2 Total Low- Lower- Upper- High- Sub- East Asia & Europe & Latin Middle South Asia income middle- middle- income Saharan Pacific Central America & East & income income Africa Asia Caribbean North Africa LM SA SI Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). Note: Figure is based on a total of 72 observations, which include 59 low- and middle-income countries and 13 high-income countries monitored by ASPIRE. Data correspond to 2022. Aggregated indicators have been calculated using simple cross- country averages. For sample size, as well as methodology for estimations, please refer to Tesliuc and Fonteñez (2025). Unlike 46 that for social assistance, social insurance expenditure does not represent government spending alone but also includes contributions made by pension system participants during their working years. LM = labor market programs, SA = social assistance; and SI = social insurance. Even within relatively homogenous groups of countries (for example, ranked by per capita gross national income [GNI] in LICs, LMICs, UMICs, and HICs), there are large differences in both the share of social protection spending in GDP (Figure IV.3.Error! Reference source not found.Error! Reference source not found.) and absolute per capita spending (Figure IV.4.). Guinea-Bissau18, a LIC, spends a slightly larger share of its GDP on social protection than the average country in LMICs. Ukraine spends the highest proportion of GDP on social protection within the UMICs, partly to protect its population from the effects of Russian’s invasion and protect livelihoods and partly because of the country’s historical legacy (social protection spending was higher than comparators even before the invasion). 18 For Guinea-Bissau, the spending primarily pertains to donor-financed initiatives. 47 Figure IV.3. Heterogeneity in the Share of Social Protection Spending in GDP across EDEs, Unweighted Average, 2022 Congo, Democratic Republic of Ethiopia Liberia LIC Togo LM SA SI Central African Republic Guinea-Bissau Nigeria Lao People's Democratic Republic Ghana Cote d'Ivoire Bangladesh Pakistan Cameroon LMIC Benin Philippines Sri Lanka Tajikistan Egypt, Arab Republic of Uzbekistan Kyrgyz Republic Tunisia Indonesia Gabon Peru Moldova Kazakhstan Maldives Tonga Iran, Islamic Republic of Dominican Republic Azerbaijan Grenada Kosovo UMIC Mexico Ecuador China Colombia Georgia Iraq Albania Argentina Turkiye Mongolia Brazil Montenegro Ukraine Panama Seychelles Chile Lithuania Latvia HIC Estonia Hungary Slovak Republic Czech Republic Bulgaria Slovenia 0 2 4 6 8 10 12 14 16 Percentage of GDP Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). Note: Data correspond to 2022. For methodology for estimations, please refer to Tesliuc and Fonteñez (2025). Unlike that for social assistance, social insurance expenditure does not represent government spending alone but also includes contributions made by pension system participants during their working years. LM = labor market programs, SA = social assistance; and SI = social insurance. 48 Figure IV.4. Absolute Level of Social Protection Spending (US$ 2017 PPP) Per Capita across EDEs, Unweighted Average, 2022 Congo, Democratic Republic of Liberia Ethiopia LIC Central African Republic LM SA SI Togo Guinea-Bissau Nigeria Lao People's Democratic Republic Ghana Cameroon Pakistan Bangladesh Benin LMIC Cote d'Ivoire Tajikistan Philippines Sri Lanka Kyrgyz Republic Uzbekistan Egypt, Arab Republic of Tunisia Indonesia Tonga Peru Moldova Gabon Maldives Kazakhstan Iran, Islamic Republic of Kosovo Grenada Ecuador Dominican Republic UMIC Azerbaijan Mexico Albania Georgia China Iraq Colombia Mongolia Argentina Brazil Ukraine Montenegro Turkiye Panama Chile Seychelles Latvia Slovak Republic HIC Lithuania Bulgaria Estonia Hungary Czech Republic Slovenia 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 5,000 in constant USD PPP $2017 Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). Note: Data correspond to 2022. For methodology for estimations, please refer to Tesliuc and Fonteñez (2025). Unlike that for social assistance, social insurance expenditure does not represent government spending alone but also includes contributions made by pension system participants during their working years. LM = labor market programs, SA = social assistance; and SI = social insurance. 49 In the vast majority of EDEs, social protection is an important spending area. Social protection is an important area of government spending in most EDEs, especially in HICs and MICs (Figure IV.5.). On average it represents 7.6 percent of government spending in LICs, 12.5 percent in LMICs, 19.3 percent in UMICs, and 22.0 percent in HICs. For example, while social protection expenditure makes up 36.8 percent of total spending in Türkiye, it is less than 5 percent in Lao People’s Democratic Republic, Nigeria, and Ghana, among other countries. Figure IV.5. Proportion of Total Government Spending on Social Protection, by Social Protection Area, 2022 35 Share of government spending 30 25 0 19 33 10 20 7 2 20 25 9 10 18 20 16 15 10 24 9 14 14 16 24 13 17 18 25 10 15 25 16 17 18 19 10 11 0 12 14 17 17 12 9 4 15 5 8 6 7 3 8 12 11 11 5 1 9 5 8 1 3 0 1 7 9 7 5 6 5 1 3 3 4 6 7 4 3 3 0 2 4 3 3 5 4 4 4 6 6 6 4 6 4 5 4 3 5 3 4 4 3 4 3 2 0 2 2 1 1 0 0 0 1 0 0 0 0 0 2 0 1 3 0 0 2 0 2 0 0 0 0 0 0 0 0 2 1 0 3 0 0 0 0 0 0 1 1 0 1 3 0 0 0 2 0 0 1 0 1 1 0 0 0 0 0 1 0 2 3 1 0 0 2 Kosovo Togo Central African Republic Sri Lanka Uzbekistan Tunisia Gabon Azerbaijan Grenada Albania Lithuania Czechia Seychelles Bulgaria Pakistan Bangladesh Tajikistan Peru China Liberia Ethiopia Benin Congo, Democratic Republic of Guinea-Bissau Cameroon Argentina Montenegro Chile Philippines Kyrgyz Republic Egypt, Arab Republic of Maldives Ecuador Mexico Iran, Islamic Republic of Ghana Tonga Slovak Republic Cote d'Ivoire Iraq Dominican Republic Ukraine Georgia Panama Latvia Nigeria Lao People's Democratic Republic Indonesia Moldova Kazakhstan Colombia Brazil Mongolia Turkiye Hungary Estonia Slovenia LIC LM SA SI Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). Note: Data correspond to 2022. Unlike that for social assistance, social insurance expenditure does not represent government spending alone but also includes contributions made by pension system participants during their working years. LM = labor market programs, SA = social assistance; and SI = social insurance. Most LICs and LMICs are constrained by low fiscal collection and low fiscal space, limiting their spending on social protection, which then directly hinders their ability to provide adequate and effective coverage of poor and vulnerable populations. As shown in Figure IV.6.Error! Reference source not found., there is a direct correlation between the revenue- generating capacity of a country and the share of government spending that is allocated to social protection. This in turn, affects the social protection system’s capacity to effectively 50 protect the population, in general, and the most vulnerable, in particular. As presented in Figure IV.7. and Figure IV.8., there is a strong correlation between per capita real spending on social assistance and the coverage and adequacy of that assistance for the poorest quintile and the extreme poor. Countries that spend more resources through their social assistance systems cover larger proportions of the individuals in the poorest quintile and the extreme poor (Figure IV.7., panels A and B). Conversely, countries with low levels of coverage are associated with smaller levels of per capita spending in social assistance programs. The same correlations are observed in relation to the adequacy of social assistance for the relative and extreme poor (Figure IV.8., panels A and B). This, together with the fact that LICs and LMICs tend to provide less generous benefits than higher income countriesError! Reference source not found., reinforces the idea that much more investment in social assistance in particular is needed to make progress toward SDG targets to eliminate extreme poverty and reduce national poverty rates. Given the fiscal restrictions that countries face, significant resource mobilization efforts will be needed to support EDEs in reducing this shortfall (as discussed further in Section V). 51 Figure IV.6. Proportion of Total Government Expenditure Spent on Social Protection Correlates with the Revenue-Generating Capacity of the Country Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). Note: Figure is based on data for up to 72 countries (with the number of observations for each income level varying based on data availability), which include 59 low- and middle-income countries and 13 high-income countries monitored by ASPIRE. CI= confidence interval; HICs = high-income countries; LICs = low-income countries; LM = labor market; LMICs = lower-middle-income countries; SA= social assistance; UMICs = upper-middle-income countries. 52 Figure IV.7. Countries with Low Social Assistance Spending Levels Have Low Levels of Coverage of the Extreme Poor and the Poorest Quintile A. Social assistance per capita spending and coverage B. Social assistance per capita spending and - poorest quintile coverage - extreme poor Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). Note: Figure is based on data for up to 72 countries (with the number of observations for each income level varying based on data availability), which include 59 low- and middle-income countries and 13 high-income countries monitored by ASPIRE. CI= confidence interval; LM = labor market; SA= social assistance. Figure IV.8. Countries with Low Social Assistance Spending Levels Also Have Low Adequacy Ratios A. Social assistance per capita spending and B. Social assistance per capita spending and adequacy ratio for poorest quantile adequacy ratio for the extreme poor Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) administrative data (https://www.worldbank.org/aspire). 53 Note: Figure is based on data for up to 72 countries (with the number of observations for each income level varying based on data availability), which include 59 low- and middle-income countries and 13 high-income countries monitored by ASPIRE. CI= confidence interval; LM = labor market; SA= social assistance. V. Targeting of Social Protection: Zooming in on the Equity of Social Assistance Allocation The previous section highlights the importance of sufficient financing of social protection— and social assistance in particular—to achieve adequate and effective coverage, of the population as a whole and of the extreme poor and poorest quintile as the first priority, based on the principle of progressive realization of USP. This section considers the extent to which social protection benefits overall, and social assistance in particular, are currently allocated toward the extreme poor and the poorest quintile. This analysis highlights how efficiently social protection resources are being allocated to achieve SDG zero poverty Targets 1.1, 1.2, and 1.3. Examining the social protection system as a whole, the pro-poor allocation of social protection benefits is moderate at best and virtually nonexistent for LICs. On average, only 25 percent of the direct and indirect recipients of social protection programs are in the poorest quintile (Error! Reference source not found..), while in quintiles two, three, and four, the share of beneficiaries is closer to their share in the population (that is, similar to random allocation). The richest quintile accounts for 16 percent of the total number of beneficiaries. There is almost no indication of pro-poor targeting in LICs’ programs overall, and the pro-poor leaning of social protection program allocation increases moderately from LMICs to UMICs and HICs. The pro-poor leaning of the social protection sector as a whole is moderate because it combines social assistance, social insurance, and labor market programs; the degree of targeting is typically higher for social assistance (which often has poverty reduction as one of its objectives) than for labor market and social insurance programs (which typically do not have an explicit poverty alleviation objective). 54 While social insurance tends to benefit wealthier households, social assistance has, on average,19 a moderate pro-poor leaning, with substantial room for improvement, particularly in lower-income countries (Figure V.1.). On average, slightly over half (52 percent) of social insurance recipients belong to the wealthiest 40 percent of the population, whereas an equivalent proportion (54 percent) of social assistance recipients are in the poorest 40 percent of the population. Social insurance therefore tends to benefit wealthier households, while social assistance recipients tend to belong to the poorest quintiles. This is largely due to the presence of poverty-targeted programs in the social assistance program mix, though overall targeting profile of social assistance is strongly influenced by categorical or geographical targeted programs, which tend to have larger caseloads, higher rates of inclusion of non-poor households, and lower coverage of poor households. On average, around 30 percent of social assistance program recipients are in the poorest quintile—that is, 50 percent higher than a random selection of beneficiaries. However, nearly the same proportion (27 percent) belong to the richest two quintiles, illustrating clear room for improvements in targeting efficiency. Furthermore, the pro-poor allocation of social assistance is weaker in lower-income countries, with the poorest quintile receiving only 27 percent of social assistance in LICs, compared to a high of 34 percent in UMICs. 19The average targeting performance considers the total program mix, including, for example, geographical and categorical schemes, as well as poverty-targeted programs, in the social assistance category. 55 Figure V.1. Distribution of Beneficiaries by Quintile, Social Protection Area, and Country Income Group 100% 15 13 14 12 16 12 9 13 15 14 13 90% 20 16 16 18 30 24 29 15 14 33 80% 16 16 18 15 % of beneficiaries 17 18 18 18 46 19 19 20 70% 18 19 19 21 60% 20 19 19 22 20 19 19 21 20 21 22 21 19 24 23 23 50% 28 25 18 40% 24 24 24 18 22 17 22 22 20 23 27 20 23 17 23 23 30% 18 21 15 17 16 20% 16 34 13 26 25 28 24 30 27 29 25 10 25 10% 23 19 22 21 19 21 15 13 15 0% 7 Total Total Total Total Low-income countries Upper-middle-income countries High-income countries Low-income countries Upper-middle-income countries High-income countries Low-income countries Upper-middle-income countries High-income countries Low-income countries Upper-middle-income countries High-income countries Lower-middle-income countries Lower-middle-income countries Lower-middle-income countries Lower-middle-income countries Social Protection Social Assistance Social Insurance Labor Market Programs Q1 Q2 Q3 Q4 Q5 Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Figure shows distribution of direct and indirect beneficiaries across quintiles for 2022 or most recent year for which data are available, and is based on a total of 73 observations, which include 67 low- and middle-income countries (LIC:12, LMIC:24, UMIC: 31) and 6 high-income countries monitored by ASPIRE. Aggregated indicators have been calculated using simple cross-country averages. Q1 = first quintile (poorest); Q2 = second quintile; Q3 = third quintile; Q4 = fourth quintile; Q5 = fifth quintile (richest). These averages hide notable country-level variation, including the strong pro-poor allocation achieved by the best performing countries. Figure V.2. unpacks this variation among countries at different levels of economic development (grouped by their level of per capita income). Two trends are apparent: (a) the pro-poor leaning increases from LICs to UMICs and (b) within each group of countries, there is a robust 25 percent of cases that achieve a significantly higher level of pro-poor targeting (the upper quartile of countries in each box-and-whiskers plot). If this is considered the highest feasible pro-poor allocation level for each group of countries, it suggests that LICs could aspire to allocate about 30 percent of average program slots to the poorest quintile (across all social assistance programs, with some 56 more targeted and some less targeted to the poor); this figure could reach 35–45 percent in LMICs and 40–55 percent in UMICs. Figure V.2. Proportion of Social Assistance Beneficiaries from the Poorest Quintile, by Country Income Group Source: Original figure for this publication using Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Beneficiary incidence for direct and indirect beneficiaries for most recent survey between 2017 and 2022. Zooming from country-level averages to program-level data helps unpack the performance of the best programs; variation in performance across country income levels; and what is feasible, good performance, or the ‘pro-poor targeting frontier’. While the average share of social assistance program beneficiaries belonging to the poorest quintile is only 30 percent in EDEs, there is significant variation in pro-poor allocation across programs within a country (depending on their intended target group and method) and across countries (depending on their level of economic development). EDEs could significantly improve the pro-poor allocation of their social assistance programming by increasing their investments in well-targeted programs and improving the pro-poor targeting of these programs to levels achieved by their peers. Analysis of ASPIRE social assistance program-level data (444 programs) shows that 54–88 percent of 57 beneficiaries from the 10 percent best-performing programs are from the poorest quintile (Figure V.3.). The pro-poor targeting accuracy of the top 10 percent best-performing programs increases with the level of economic development of the country: from 35–41 percent of beneficiaries from the poorest quintile for LICs, to 50–77 percent for LMICs, 54–78 percent for UMICs, and 63–88 percent for HICs. Across countries with different income levels, the variation in targeting accuracy for the poorest quintile is related to the level of labor market formality and the administrative capacity of the country. In UMICs and HICs, a higher share of earnings is in the formal sector, hence observable to the state, which makes hybrid- and means-testing feasible and methods associated with higher targeting accuracy (Grosh, Leite, Wai-Poi, and Tesliuc 2022) more feasible. In LICs, most incomes are in the informal sector and not verifiable, which restrict the choice of targeting methods to categorical, geographical, or proxy-means testing, methods with intrinsically higher levels of non-poor inclusion error. The level of targeting performance achieved by the best 10 percent of social assistance programs within each country income group can be seen as representing a robust estimate of the extent of pro-poor targeting accuracy that is feasible for a given country setting or the ‘pro-poor targeting frontier’. Compared to the average targeting profile observed in Figure V.1., all countries can improve their targeting accuracy to the poorest quintile by 55–75 percent on average. Improving the pro-poor allocation of the social assistance spending could significantly dent the coverage gaps in all EDEs, from LICs to UMICs. 58 Figure V.3. EDEs Could Significantly Improve the Pro-poor Allocation of Their Social Assistance Spending: Average Share of Beneficiaries from Best-Performing Social Assistance Program, by Country Income Group 100 Share of beneficiaries from the poorest quintile 90 88 80 78 77 70 covered by a SA program 60 63 54 50 50 41 40 41 35 34 30 29 20 22 21 22 14 14 10 5 5 0 1 1 Low income Lower middle income Upper middle income High income Countries ranked by their income level Minimum 10th percentile Median 90th percentile Max Source: Original calculation using ASPIRE household survey data. https://www.worldbank.org/aspire Note: The indicator measures the share of beneficiaries from the poorest quintile achieved by a social assistance program, where the programs are ranked from the lowest to the highest share of Q1 beneficiaries reached by each program. Figures shown are for the minimum, the 10th percentile, median, the 90th percentile, and best-performing social assistance programs. The yellow line shows the pro-poor targeting frontier, the share of beneficiaries from Q1 achieved by the 90th best-performing program, by countries’ income groups. Moving from the current level of average pro-poor targeting to a higher one could face political economy difficulties in many countries, but more equitable targeting is technically possible. And with improved targeting of social assistance spending, most EDEs could close or reduce the coverage gap in extreme poverty and the income shortfall of the extreme poor and poorest quintile. To mitigate the eventual political opposition to a better targeted social assistance system— one that delivers a higher share of support for the poorest quintile and the extreme poor— this paper evaluates a technically feasible scenario which accounts for many real-world difficulties associated with targeting. This scenario, spelled out in detail in Annex 2, uses the following assumptions about the degree of pro-poorness of different social assistance programs that make up the program mix: 59 • The social assistance program mix uses a combination of categorically targeted and welfare targeted programs. To achieve their social assistance objectives, most countries use a diversified portfolio of social assistance programs. • Categorical programs account for half of coverage and one-quarter of spending. Specifically, we assume that one-quarter of social assistance spending is allocated to categorical targeted programs. In line with the empirical findings, we assume a lower average benefit for these programs. A 1:3 benefit ratio for categorically targeted programs implies that the coverage of categorical programs is equal to that for welfare targeted ones. • About three-quarters of total spending is earmarked for poverty-targeted programs, the targeting performance of which varies with the income level of the country. The incidence of beneficiaries of these programs is at the lower margin of the targeting accuracy frontier: that is, equal to the 90th percentile best targeted programs that the country’s income group is currently achieving. Specifically, we assume that a welfare targeted program or programs in LICs will correctly identify 35 percent of the beneficiaries from the (pre-transfer) poorest quintile, 54 percent in MICs, and 65 percent in HICs. These parameters are in line with the inclusion and exclusion errors observed in the top 10 percent of programs. • About 5 percent of all social assistance spending goes to administrative costs. This share is in line with the share of administrative costs for mature social assistance programs for which administrative functions are financed—but in an efficient way (there is little waste in administration). The next section uses this scenario to quantify the poverty-reduction impact that each EDE could achieve if it earmarked more resources for welfare targeted programs and improved the targeting of these programs up to the pro-poor targeting frontier. 60 VI. Closing the Extreme Poverty Gap and Reducing the Relative Poverty Gap: Increasing Efficiency, Augmenting Domestic Financing, and Mobilizing International Finance To achieve SDG 1 (Goal 1) and its related targets faster, EDEs need faster growth, pro-poor growth, and stronger pro-poor policies. Among the latter, social protection—pro-poor social assistance in particular—could make an important dent in the number of extreme poor with existing budgetary allocations. Why this is important? First, because budgets remain tight in most EDEs, the road to USP needs to be progressive by prioritizing elimination of extreme poverty—especially in LICs— and significantly reducing relative poverty (for example, the income shortfall of the poorest quintile); this will require greater efficiency of social assistance. Second, we need to know how big a challenge this is and how the poverty gaps can be closed. For countries where extreme poverty is too high and cannot be eliminated by 2030, this section considers intermediate targets consistent with the principle of progressive realization of USP. Third, for all EDEs, we need to identify and quantify the proximate solutions to achieving Targets 1.1, 1.2, and 1.3: improving the efficiency of social assistance spending, expanding the fiscal space for pro-poor social assistance through international solidarity and complementary domestic policies that stimulate growth, and maintaining macroeconomic stability. This section (a) examines the current contribution of social assistance programs to reducing the extreme and relative poverty gap; (b) quantifies the gaps left uncovered, after the receipt of social assistance transfers; and (c) examines the share of this gap that can be covered by improving the efficiency of social assistance transfers and the part that will require additional fiscal space. The estimated volume of resources needed to fill the income shortfall of the extreme poor and reduce it for those in the poorest quintile in 2022 covers all EDEs for which there are estimates of extreme poverty and relative poverty at the time of writing: 133 out of the 152 EDEs. From the full range of social protection programs, this section focuses only on the 61 poverty alleviation effects of social assistance programs, as most of these programs have poverty reduction as their main or secondary objective. The estimates are adjusted to account for the fact that not all social assistance programs are intended to reduce poverty, or they can do so only imperfectly, incurring targeting errors. The main motivation for this exercise is the realization, documented in previous sections, that achieving USP in LICs and LMICs— confronted with insufficient fiscal space and significant coverage, adequacy, and efficiency gaps—is feasible if done progressively, starting with the poorest. We assume a progressive expansion of the social protection system that aims, over the short to medium term, to cover first the extreme poor or—in countries where over 20 percent of the population live in extreme poverty—at least the poorest quintile. The main question is how far are EDEs— particularly LICs and LMICs—from achieving this intermediate goal? As shown earlier, on aggregate for all EDEs, the total social assistance spending is large enough to eradicate extreme poverty and sustainably halve the income shortfall of the poorest quintile. Around 2020, total social assistance spending for cash transfer programs was estimated at US$1,263 billion in 2017 PPP, five and a half times higher than the aggregate income shortfall of the extreme poor before social assistance transfer receipt (US$227 billion) and 35 percent higher than income shortfall of the poorest 20 percent in these countries before social assistance transfer receipt (US$939 billion). However, there is a mismatch in resources across countries (with LICs having large deficits) and within countries (only a fraction of this spending accrues to the extreme poor as well as to the poorest 20 percent of the population). Achieving SDG 1 to eliminate extreme poverty and achieve substantial coverage of the poor and vulnerable is fiscally well within reach. The decisions for putting these measures into practice will require careful consideration of the political economy in each context (for example, Grosh, Leite, Wai-Poi, and Tesliuc 2022; ODI-UNICEF 2020). The path toward USP is not only about more resources but also about more efficient and strategic use of existing 62 resources. Based on the estimated income shortfall that needs to be filled to lift all households out of extreme poverty, current social assistance budgets could significantly reduce or even eliminate extreme poverty in most EDEs20 if more directly targeted toward households living in poverty, though additional international support would be required for the highest-poverty contexts. In 2022 in EDEs, the additional cost of eradicating extreme poverty was US$137 billion in 2017 PPP per year, and the corresponding cost of eliminating the income shortfall of the poorest quintile was US$581 billion per year (Figure VI.1.). This cost is additional to the social assistance budgets that already go to the extreme poor (US$90 billion) and the poorest quintile (US$358 billion). Combined, these figures amount to the total income shortfall required to lift—and keep—all households out of extreme poverty (around US$227 billion per year) and the poorest quintile out of relative poverty (around US$939 billion per year). 20In EDEs with extreme poverty headcount below 10 percent. As of 2022, 86 of the 133 EDEs with estimates of extreme poverty have a headcount of between 0 and 10 percent. In these countries, the current level of social assistance spending could eliminate extreme poverty if better targeted. 63 Figure VI.1. Social Assistance Closes One-Third of the Income Shortfall and Could Reduce It Further with Improved Pro-poor Targeting 1,000 939 Income shortfall before social protection 900 transfers, US$, billions 2017 PPP 800 311 700 581 600 500 270 400 300 227 200 120 137 358 100 17 90 - Extreme poor Poorest quintile Already covered by SA More efficient SA spending Remaining income shortfall Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: Figure shows how much social assistance (SA) spending at current levels reduces the income shortfall, that is, the income needed to lift—and keep—all households out of extreme and relative poverty (poorest quintile); the additional potential reduction in the income shortfall that can be achieved through greater efficiency in SA spending; and the additional fiscal resources needed to bridge the remaining income shortfall or spending gap. More effective SA spending assumes an average targeting efficiency equal to the targeting performance of the 10 most pro-poor SA programs within each country income group. The figure is based on a total of 133 observations, which include 114 low- and middle-income countries (LIC: 19, LMIC: 49, UMIC: 46) and 19 high-income countries monitored by ASPIRE. Extreme poor (those living on less than $2.15 a day at purchasing power parity). Improving the efficiency of social assistance transfers using the scenario described in Section V could further reduce the income shortfall of the extreme poor and the poorest quintile. The total income shortfall of the extreme poor in EDEs after the receipt of current social protection transfers, estimated at US$137 billion (expressed in 2017 PPP), could be reduced to US$120 million through a pro-poor allocation of existing social assistance budgets. The total income shortfall of the poorest quintile in EDEs after the receipt of current social protection transfers, estimated at US$581 billion, could also be reduced to US$311 billion through a pro-poor allocation of existing social assistance budgets. 64 Closing the extreme poverty gap and reducing the poverty gap among the population in the poorest quintile require different strategies. While improving the efficiency of social assistance and expanding the fiscal space for pro-poor programs is an important part of the solution, tackling extreme poverty requires a focus on LICs and LMICs, while reducing relative poverty is a broader agenda. Extreme poverty is concentrated in LICs and some LMICs as well as in a few populous UMICs like Brazil. Relative poverty affects, by definition, all EDEs. The next subsections will cost the two gaps, identify the share of the gaps currently covered by social assistance, how much could be covered by more efficient deployment of social assistance spending, and how much will require additional fiscal space. We will also identify different groups of countries, their initial conditions, and the solution that help address the gaps. The next subsections examine the feasibility of closing the income shortfall for different types of countries, grouped by their level of extreme poverty in 2022 (Section VIA) or their level of income (Section VIB). The final subsection concludes with a call to action to improve financing to eliminate extreme poverty (Target 1.1), halve relative poverty (Target 1.2), and achieve full coverage at least for the poor and vulnerable (Target 1.3), the first step of a much longer journey toward progressively realizing USP. VIA. Closing the Extreme Poverty Gap in EDEs In 2022, the annual income shortfall of the extreme poor before social assistance transfers was about US$227 billion in 2017 PPP (Figure VI.1.). Of this total, social assistance spending already covers (that is, reduces the income shortfall) about US$90 billion, or 40 percent of the shortfall. After social assistance spending, the total income shortfall of the extreme poor remains at US$137 billion in 2017 PPP. More pro-poor social assistance spending could reduce the income shortfall of the extreme poor even further, by an additional US$17 billion, leaving an annual residual of about US$120 billion in 2017 PPP not covered. To close the residual 65 income shortfall, EDEs overall would need an annual allocation of US$120 billion PPP as additional fiscal space, from domestic or international resources. However, as outlined throughout this section, the extent of additional resources needed varies significantly between countries, depending on their extreme poverty rates and current social assistance budgets. For 62 countries with zero or very low extreme poverty, the income shortfall of the extreme poor after social assistance transfers is already close to zero. These countries should maintain their level of social assistance spending and coverage and adequacy for the extreme poor. Without social assistance spending, some of their population would fall into extreme poverty. These countries include all HICs and many UMICs. The 39 countries with extreme poverty rates between 2 and 20 percent have a relatively small income shortfall after current social assistance spending, which could almost be closed by improving the pro-poor efficiency of current social assistance spending. The annual aggregate income shortfall of the extreme poor in these countries amount to US$15 billion in 2017 PPP (about 11 percent of the total income shortfall of extreme poor in the 133 EDEs for which we have data, of US$137 billion in 2017 PPP). Feasible improvements in targeting of social assistance budget halve this shortfall to about US$7 billion in 2017 PPP per year. The fiscal space required to eliminate the income shortfall could be accommodated by a modest 6 percent expansion of the social assistance budget. 66 Figure VI.2. Contribution of Social Assistance Spending to the Reduction of the Income Shortfall of the Extreme Poor, Total and by Countries Grouped by Their Extreme Poverty Rate A. Shortfall in US$ PPP B. Shortfall and social assistance spending as percentage of GDP 100 14.0 12.0 90 81 12.0 80 10.0 Billons in USD PPP 70 60 54 54 8.0 % GDP 50 0 2 39 6.0 40 78 7 35 30 8 4.0 52 20 6 2.0 1.6 1.7 1.5 1.7 1.4 10 24 0.6 12 0.3 0.1 0.2 - 2 2 - No XP 2-20 20-40 40-100 Total No XP 2-20 20-40 40-100 More efficient SA spending Remaining income shortfall Income shortfall as % of GDP Total income shortfall SA expenditure as % of GDP Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: The figure is based on a total of 133 observations (No XP: 62, 20-20% XP rate: 39, 20-40% XP rate: 20, 40-100% XP rate:12). XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). Figure VI.3. Reduction of the Income Shortfall of the Extreme Poor Due to Social Assistance, Potential Reduction with Greater Efficiency, and the Additional Fiscal Space Needed 100 0.2 90 3 17 % of total income shortfall 80 70 53 20 65 60 50 96 97 40 8 30 62 12 20 40 10 23 2 2 - Total No XP 2-20 20-40 40-100 Already covered by SA More efficient SA spending Remaining income shortfall Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: The figure is based on a total of 133 observations (No XP: 62, 20-20% XP rate: 39, 20-40% XP rate: 20, 40-100% XP rate:12). XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). 67 A third group comprises 20 countries with extreme poverty rates between 20 and 40 percent, with an overall annual income shortfall of US$41 billion after social assistance spending. The shortfall could be reduced by US$6 billion, to US$35 billion if social assistance spending were more pro-poor. The fiscal space required to eliminate this income shortfall would require almost doubling the current social assistance budgets (an increase by about 83 percent, from US$42 billion to US$77 billion). Such an expansion may not be feasible in many of these countries. Instead, the countries could focus first on reducing the income shortfall of the poorest 20 percent. Figure VI.4. Reduction of the Income Shortfall of the Extreme Poor Due to Social Assistance, Potential Reduction with Greater Efficiency, and the Additional Fiscal Space Needed, as % of GDP 12.0 12.0 10.0 8.0 % of GDP 6.0 11.6 4.0 0.03 0.03 0.2 1.7 0.3 2.0 0.2 0.0002 0.1 0.02 0.3 0.003 1.1 0.1 0.2 - 0.1 0.1 0.4 0.2 Total No XP 2-20 20-40 40-100 Already covered by SA More efficient SA spending Remaining income shortfall Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: The figure is based on a total of 133 observations (No XP: 62, 20-20% XP rate: 39, 20-40% XP rate: 20, 40-100% XP rate:12). XP = extreme poor (those living on less than $2.15 a day at purchasing power parity). Finally, the fourth group comprises 12 countries with very high poverty rates (40–81 percent) that have large income shortfalls, 20 times larger than their current social assistance budgets. These countries also have low fiscal tax collection. For this group of LICs, the objective of eliminating extreme poverty is not achievable through more efficient social 68 assistance policies alone nor by expanding the level of domestic budgetary resources for the sector. For this group of countries, elimination of extreme poverty remains a medium-term objective, which will require a combination of policies to support growth, macroeconomic stability, and larger and more efficient social assistance spending, as well as international support. Over the medium term, these countries could first focus on reducing the income shortfall of the poorest 20 percent of their population. While all those living in extreme poverty in the 32 countries in the fourth and fifth groups experience severe hardship, the population in the poorest quintile is substantially more deprived. Figure VI.5. illustrates the daily per capita consumption (in US$ 2017 PPP) for the populations in 10 such countries, for four income quintiles. Compared to the average per capita consumption of the population in the middle quintile, those in the poorest quintile consume only 39 percent and those in the second poorest quintile only 68 percent of median consumption of the country. Using the existing scarce social assistance resources to reduce the income shortfall of the poorest group is an achievable goal, with a combination of improved efficiency of existing spending and modest increases in fiscal space, as illustrated in the next subsection. 69 Figure VI.5. Differences in the Daily Per Capita Consumption of the Population in the Poorest Four Quintiles, in 10 Countries with Very High Levels of Extreme Poverty 7.0 6.6 Daily per capita consumption in US$ 2017 PPP 6.0 5.2 4.9 5.0 4.1 3.9 4.0 4.0 3.6 3.3 2.8 2.8 2.9 3.0 2.4 2.5 2.4 2.4 2.2 1.9 2.0 2.0 2.1 1.7 1.5 1.6 1.5 1.7 2.0 1.4 1.1 1.1 1.2 1.2 1.4 0.9 1.1 1.0 1.1 0.6 0.8 0.7 0.6 1.0 0.3 - Malawi Mozambique Chad Lesotho Angola Nigeria Uganda Niger Congo, Dem. Rep. Central African Republic Mean consumption Q1 Mean consumption Q2 Mean consumption Q3 Mean consumption Q4 Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). VIB. Reducing the Relative Poverty Gap In most EDEs, a significant proportion of social assistance programs are intended to reduce the deprivation of the poorest. The definition of poorest varies from country to country, but a widely used measure of relative poverty is the population in the poorest quintile. Many poverty-alleviation social assistance programs focus on groups around this size. On aggregate, social assistance spending across EDEs (US$1,263 billion in 2017 PPP) is already in theory enough to cover the income shortfall of the poorest quintile of the population before social assistance transfers (US$939 billion in 2017 PPP). Indeed, three- quarters of EDEs’ social assistance spending should already be enough to close the income shortfall of the poorest quintile (Figure VI.6., panel A). The total budget spent in each group of countries ranked by per capita income holds good for UMICs and HICs. Meanwhile, LICs would need 50 percent more social assistance spending, perfectly targeted toward the 20 percent 70 poorest, to bring all this population up to the relative poverty line of the poorest quintile. For LMICs, aggregated spending is only 10 percent higher than the income shortfall. There are, however, marked differences across country income categories in the overall size and pro- poor efficiency of their current social assistance budgets and in the potential improvements in allocation, which makes the scenario of eliminating the income shortfall for poorest quintile feasible with existing or slightly expanded domestic budgets in many MICs, but less so in LICs (Figure VI.6., panel B). Figure VI.6. Contribution of Social Assistance Spending to the Reduction in the Income Shortfall of the Poorest Quintile, Total and by Countries, Grouped by Income Level A. Shortfall in US$ PPP B. Shortfall and social assistance spending as percent of GDP 700 4.5 4.1 4.0 600 553 3.5 Billions of USD PPP 500 196 3.0 400 % OF GDP 2.5 300 1.9 142 2.0 1.7 207 1.6 1.4 200 161 1.5 1.2 1.1 1.1 88 13 1.0 217 62 1.0 0.8 100 1 66 2 18 87 53 0.5 0 15 Low income Lower Upper High income 0.0 middle middle Total Low Lower Upper High income income income middle middle income income income More efficient SA spending Remaining income shortfall Total income shortfall Total income shortfall SA expending as % of GDP Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: The figure is based on a total of 133 observations, which include 114 low- and middle-income (LIC: 19, LMIC: 49, UMIC: 46) countries and 19 high-income countries monitored by ASPIRE Social assistance spending plays an important role in reducing the relative poverty gap of the population from the poorest quintile and could play an even bigger role in reducing this gap if spending could be deployed in a more pro-poor fashion (Figure VI.7.). About 38 percent of the relative income shortfall of the poorest quintile before the receipt of social assistance 71 transfers is already filled by current social assistance transfers. Improvements in targeting could reduce the shortfall by a further US$270 billion—or by about 29 percent—which would reduce the shortfall to only 33 percent of its pre-social assistance spending levels. This aspirational scenario of ‘improved targeting’ follows the assumptions outlined at the end of Section V: each country earmarks three-quarters of social assistance spending for welfare- targeted programs (with the rest financing other categorical programs); overall coverage of the population from targeted and categorical programs is approximately equal; around 5 percent of social assistance budgets are absorbed by administrative costs; and leakage will continue to exist at the level of the top 10 percent best-performing programs today. Figure VI.7. Reduction of the Income Shortfall of the Poorest Quintile Due to Social Assistance, Potential Reduction with Greater Efficiency, and the Additional Fiscal Space Needed 100% 8% 90% % of total income shortfall 80% 33% 35% 42% 70% 38% 60% 83% 50% 29% 26% 40% 32% 30% 54% 20% 38% 39% 6% 26% 10% 11% 0% Total Low income Lower middle Upper middle High income income income Already covered by SA More efficient SA spending Remaining income shortfall Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: The figure is based on a total of 133 observations, which include 114 low- and middle-income (LIC: 19, LMIC: 49, UMIC: 46) countries and 19 high-income countries monitored by ASPIRE. Most EDEs, except LICs, could reduce the income shortfall of the population in the poorest quintile by more than half through these improvements in the targeting of their current social assistance budgets and thus achieve in the short term a modified version of SDG 1, Target 1.2 (halving national poverty rates) (Figure VI.7. and Figure VI.8.). The share of pre- transfer income covered by social assistance grows with the income level of the country: it 72 accounts for 11 percent in LICs and increases monotonically to 56 percent in HICs. Indeed, when analyzing social assistance’s ability to halve the poorest quintile’s income shortfall, HICs have already achieved this. In LMICs and UMICs, the potential improvement in efficiency could more than halve the poorest quintile’s income shortfall over the short term as well. It is only in LICs that improvements in pro-poor efficiency will not be enough to halve the income shortfall of the poorest quintile. For this group of countries, governments and IFIs will need to work together to go beyond the improvements already identified in the optimal targeting scenario in Section V. This will likely need to include both additional international support and innovations that could more dramatically enhance targeting efficiency, such as the digital identification and delivery tools used by some LICs during the COVID-19 crisis (Okamura, Iyengar, and Andrews 2025). Figure VI.8. Reduction of the Income Shortfall of the Poorest Quintile Due to Social Assistance, Potential Reduction with Greater Efficiency, and the Additional Fiscal Space Needed, as % of GDP 2.00 1.9 1.7 0.15 1.50 1.2 0.74 % of GDP 1.1 1.0 1.00 0.40 1.40 0.41 0.43 0.35 0.29 0.50 1.03 0.32 0.46 0.10 0.45 0.19 0.26 0.00 Total Low income Lower middle Upper middle High income income income Already covered by SA More efficient SA spending Remaining income shortfall Source: Original figure for this publication based on Atlas of Social Protection Indicators of Resilience and Equity (ASPIRE) household survey data (https://www.worldbank.org/aspire). Note: The figure is based on a total of 133 observations, which include 114 low- and middle-income (LIC: 19, LMIC: 49, UMIC: 46) countries and 19 high-income countries monitored by ASPIRE. 73 VIC. A Call for Action: Ensuring Adequate Financing and Strengthening the Pro-poor Equity of Social Protection Systems This section documents the important role played by social assistance spending in reducing the income shortfall of both the relative and extreme poor across all EDEs and the large gains that could be reached through improved efficiency of social assistance spending in most countries, except 32 LICs and LMICs with higher extreme poverty rates (of 20–81 percent). As shown, in most UMICs and a majority of LMICs, existing levels of social assistance spending used in a more pro-poor fashion could achieve this goal. Countries like Brazil, India, Nigeria, Tanzania, Kenya and Zambia could accelerate the use of redistributive policies such as pro- poor social assistance programs to alleviate or even eradicate extreme poverty. India has demonstrated the effectiveness of such pro-poor social assistance programs for poverty alleviation in recent years. This finding brings an optimistic message: such policies, complementing economic growth and macroeconomic stability, could reduce or even eliminate two-thirds of extreme poverty. For LICs, however, the story is more nuanced. For these countries with low fiscal revenues and low social assistance spending, international support and stronger growth will be particularly critical. This paper therefore suggests that to eliminate extreme poverty and reduce national poverty, an agenda should be tailored by country context and anchored in realistic and feasible parametric reforms of the social assistance budgets. The focus on social assistance is warranted because of two factors: many social assistance programs have poverty or vulnerability reduction as a primary or secondary objective, and social assistance is the main social protection area that reaches the extreme poor and the poorest quintile. For the group of countries with high extreme poverty, the paper acknowledges the need for broad-based policies supporting pro-poor growth, macroeconomic stability, and pro-poor investments, as well as the prioritization of social assistance resources toward the poorest quintile. Especially for the 12 countries with extreme poverty rates of over 40 percent in 2022, the paper envisages a strong role for the international community in complementing domestic 74 resources. For the group of countries where extreme poverty is between 10 percent and 20 percent of the population, the paper envisages efficiency improvements in social assistance spending that could eliminate extreme poverty and halve the income shortfall of the poorest quintile. 75 References Banerjee, Abhijit, Rema Hanna, Benjamin A. Olken, and Diana Sverdlin Lisker. 2024. “Social Protection in the Developing World.” NBER Working Paper 32382. https://www.nber.org/papers/w32382. Bastagli, Francesca, Jessica Hagen-Zanker, and Georgina Sturge. 2016. “Cash Transfers: What Does the Evidence Say?” Overseas Development Institute, London. https://odi.org/en/publications/cash-transfers-what-does-the-evidence-say-a- rigorous-review-of-impacts-and-the-role-of-design-and-implementation-features/. Bastagli, Francesca, and Christina Lowe. 2021. “Social Protection Response to COVID-19 and Beyond: Emerging Evidence and Learning for Future Crises.” Overseas Development Institute, London. https://odi.org/en/publications/social-protection-response-to- covid-19-and-beyond-emerging-evidence-and-learning-for-future-crises/. Beazley, R., M. Marzi, and R. Steller. 2021. “Drivers of Timely and Large-Scale Cash Responses to COVID-19: What Does the Data Say?” SPACE. London and Bonn. Beazley, Rodolfo, Jana Bischler, and Alexandra Doyle. 2021. “Towards Shock-Responsive Social Protection: Lessons from the COVID-19 Response in Six Countries.” Oxford Policy Management, Oxford. https://www.opml.co.uk/files/Publications/A2241- maintains/maintains-towards-shock-responsive-social-protection-synthesis- report.pdf. Carranza, Eliana, Matteo Morgandi, and Diana Sverdlin. 2025. “Optimizing Labor Market Programs and Strengthening Delivery Systems for Impact and Scale.” World Bank, Washington, DC. Gentilini, Ugo. 2022. “Cash Transfers in Pandemic Times Evidence, Practices, and Implications from the Largest Scale Up in History.” World Bank, Washington, DC. 76 https://documents1.worldbank.org/curated/en/099800007112236655/pdf/P176585 05ca3820930a254018e229a30bf8.pdf. Grosh, Margaret, Phillippe Leite, Matthew Wai-Poi, and Emil Tesliuc. 2022. Revisiting Targeting in Social Assistance: A New Look at Old Dilemmas. Human Development Perspectives. Washington, DC: World Bank. http://hdl.handle.net/10986/37228. Hammad, M., F. Bacil, and F. Veras Soares. 2021. “Next Practices: Innovations in the COVID-19 Social Protection Responses and Beyond.” International Policy Centre for Inclusive Growth, Brasilia. https://ipcig.org/sites/default/files/pub/en/RR59_Executive_Summ ary_Next_Practices_Innovations_in_the_COVID_19_IPC_UNDP.pdf. ILO (International Labour Organization). 2021a. World Social Protection Report 2020-22: Social Protection at the Crossroads - In Pursuit of a Better Future. International Labour Organization, Geneva. https://www.ilo.org/global/publications/books/WCMS_81757 2/lang--en/index.htm. ILO. 2021b. Social Protection Responses to COVID-19 Crisis around the World. International Labour Organization, Geneva. https://www.social- protection.org/gimi/ShowWiki.action?id=3417. ILO. 2024. World Social Protection Report 2024-26: Universal Social Protection for Climate Action and a Just Transition. International Labour Organization, Geneva. https://www.ilo.org/publications/flagship-reports/world-social-protection-report- 2024-26-universal-social-protection-climate. IPC-IG (International Policy Centre for Inclusive Growth). 2021. “Social Protection Responses to COVID-19 in the Global South: Online Dashboard.” International Policy Centre for Inclusive Growth, Brasilia. 77 Lustig, Nora. 2018. "Commitment to Equity Handbook: Estimating the Impact of Fiscal Policy on Inequality and Poverty.” Brookings Institution Press, Washington. https://commitmentoequity.org/publications-ceq-handbook. ODI (Overseas Development Institute) and UNICEF (United Nations Children’s Fund). 2020. “Universal Child Benefits: Policy Issues and Options.” Overseas Development Institute, London and UNICEF, New York. https://odi.org/en/publications/universal-child- benefits-policy-issues-and-options/. Okamura, Yuko, Tim Ohlenburg, and Emil Tesliuc. 2024. “Scaling Up Social Assistance Where Data Is Scarce - Opportunities and Limits of Novel Data and AI.” World Bank, Washington, DC. http://hdl.handle.net/10986/41553. Okamura, Yuko, Hrishikesh TMM Iyengar, and Colin Andrews. 2025. “Wake-up Call for Social Assistance? Unfinished Mission to Reach the Poor and Beyond.” World Bank, Washington, DC. Oxfam. 2020. “Shelter from the Storm: The Global Need for Universal Social Protection in Times of COVID-19”. Oxford. https://www.oxfam.org/en/research/shelter-storm- global-need-universal-social-protection-times-covid-19. Packard, Truman G., Ugo Gentilini, Margaret Ellen Grosh, Philip B. O’Keefe, Robert J. Palacios, David A. Robalino, and Indhira Vanessa. 2019. “Protecting All: Risk Sharing for a Diverse and Diversifying World of Work.” (English). World Bank, Washington, DC. http://documents.worldbank.org/curated/en/997741568048792164/Protecting-All- Risk-Sharing-for-a-Diverse-and-Diversifying-World-of-Work. Reyes, Hartley, Gonzalo Javier, Gustavo Demarco, and Miglena Abels. 2025. Riding the Demographic Wave: Pensions and Retirement Income in an Aging World. World Bank, Washington, DC. 78 SPIAC-B (Social Protection Interagency Cooperation Board). 2021. “Key COVID-19 Lessons and Recommendations for Working Towards Universal Social Protection.” SPIAC-B Joint Statement, October 2021. https://www.social- protection.org/gimi/Media.action?id=18656. Tesliuc, Emil, Fonteñez, Maria Belen. 2025. Adaptive Social Protection Agenda: Lessons from Responses to the COVID-19 Shock. World Bank, Washington D.C. UNDESA (United Nations Department of Economic and Social Affairs). 2018. The Report on the World Social Situation 2018: Promoting Inclusion through Social Protection. United Nations Department of Economic and Social Affairs, New York. https://www.un.org/development/desa/dspd/world-social-report/2018-2.html. UNDESA. 2023. World Social Report 2023: Leaving No One Behind in an Ageing World. United Nations Department of Economic and Social Affairs, New York. https://www.un.org/development/desa/pd/content/launch-world-social-report- 2023. USP2030 (Global Partnership for Universal Social Protection). 2019. “Together to Achieve Universal Social Protection (USP) by 2030: A Call to Action.” https://usp2030.org/wp- content/uploads/calltoaction_en.pdf. World Bank Group. 2022. “Charting a Course Towards Universal Social Protection: Resilience, Equity, and Opportunity for All.” World Bank Group, Washington, DC. http://hdl.handle.net/10986/38031. 79 Annex 1: ASPIRE: Key Concepts, Definitions, and Data The paper focuses on social protection policies and programs to improve equity, opportunity, and resilience: • Equity: to reduce poverty and inequality, promote equality of opportunity, and address exclusion—especially for women, girls, and persons with disabilities—primarily through social assistance programs. • Opportunity: to improve access to productive work, facilitate the transition from school to work, and build skills along extended working lives, primarily through labor market programs. • Resilience: to provide protection and insurance against shocks and build capacity to manage them, including against natural disasters, pandemics, climate change, economic and financial crises, conflict or forced displacement; to expand social insurance, pensions, and unemployment insurance to support informal workers, the elderly, and persons with disabilities and provide effective long-term care, primarily through social and employment insurance programs. Within social protection policy, the paper focuses on the main social protection programs in World Bank client economies: programs of large or moderate size by level of spending and/or number of beneficiaries served, based on administrative and household survey data. The level of analysis is on the social protection system and its thematic areas: social insurance, labor market programs, and social assistance programs. Table A.I. 1. lists the social protection programs and how they are aggregated into categories and thematic areas in ASPIRE. ASPIRE data come from two main sources: administrative and household survey data. • Nationally representative household surveys are used to calculate performance indicators of SPL programs for 111 countries (to estimate the global coverage gap in Section IIA) and to track performance of the SPL systems from circa 2010 to circa 2022 80 for 73 countries. Performance indicators include program coverage, incidence of beneficiaries and benefits, level of transfers, and simulated impacts of SPL public transfers on poverty and inequality reduction. • Administrative program-level data are used to track the level, composition, and evolution of social protection spending and beneficiaries for all large and medium-size programs in EDEs. The current collection used in the paper covers 72 countries for 2017–2022. Table A.I. 1. ASPIRE Program Classification SPL AREA PROGRAM CATEGORY PROGRAM SUB-CATEGORY Old age pensions (all schemes, national, civil servants, veterans, other special) Survivors’ pensions (all schemes, national, civil Contributory pensions servants, veterans, other special) SOCIAL Disability pensions (all schemes, national, civil INSURANCE servants, veterans, other special) Occupational injury benefits Paid sick leave benefits Other social insurance21 Health Maternity/paternity benefits Training (vocational, life skills, cash for training) Employment incentives/wage subsidies Employment measures for persons with disabilities Labor market policy measures Entrepreneurship support/startup incentives (cash (active labor market programs) and in-kind grants, microcredit) Labor market services and intermediation through LABOR MARKET public employment services Other active labor market programs Out-of-work income maintenance (unemployment Labor market policy support benefits, contributory) (passive labor market programs) Out-of-work income maintenance (unemployment benefits, non-contributory) Poverty-targeted cash transfers, last resort programs SOCIAL Unconditional cash transfers Family/child/orphan allowances (including orphan and ASSISTANCE vulnerable children benefits) 21 Note that ASPIRE Administrative data does not include this type of programs. 81 SPL AREA PROGRAM CATEGORY PROGRAM SUB-CATEGORY Non-contributory funeral grants, burial allowances Emergency cash support (including support to refugees/returning migrants) Public charity, including zakat Conditional cash transfers Conditional cash transfers Old-age social pensions Disability benefits/war victims non-contributory- Social pensions (non-contributory) related benefits Survivorship Food stamps, rations, and vouchers Food distribution programs Nutritional programs (therapeutic, supplementary Food and in-kind transfers feeding, and persons living with HIV) In kind/non-food support (education supplies, free textbooks and uniforms) School feeding School feeding Cash for work Public works, workfare, and direct job creation Food for work (including food for training, food for assets, and so on) Health insurance exemptions and reduced medical fees Education fee waivers Food subsidies Fee waivers and subsidies Housing subsidies and allowances Utility and electricity subsidies and allowances Subsidies for agricultural inputs Scholarships/education benefits Other social assistance Social care services, transfers for care givers Items left out from above categories Period covered by the paper. Social protection administrative data cover 2017–2022. Household survey data use 2022 or the year with the most recent household survey—what the paper refers as ‘circa 2022’. To track the performance of the social protection systems over the last decade, we use a subset of 73 countries with surveys from circa 2010 and circa 2022. Regional/country focus of the paper: EDEs. The paper examines 153 economies included in the category of EDEs. The EDEs are a heterogenous group and can be subdivided by level of 82 development (LICs, MICs, ‘new’ HICs) as well as regional groups (as used in the operational work of the World Bank). The paper draws on detailed recent household survey data for 73 of these economies and detailed program-level administrative data for 76 economies. The only exceptions are the estimates of global social protection coverage in Section IIA, where we use surveys from 111 countries, including a few surveys older than 2018. Table A.I. 2. lists the share of number of countries, population, and GDP for each of the samples used in the paper. Overall, we have information from administrative or survey data for 127 countries; these countries represent 83 percent of the total number of countries, 97 percent of the total population of EDEs in 2022, and 99 percent of their GDP in 2022. Table A.I. 2. Proportion of All EDEs Included in This Paper No. of EDEs With survey data Total no yes no 26 39 65 With administrative data yes 16 72 88 Total 42 111 153 Share of EDEs With survey data Total no yes no 17 25 42 With administrative data yes 10 47 58 Total 27 73 100 Share of population of EDEsa With survey data Total no yes no 3 14 17 With administrative data yes 2 81 83 Total 5 95 100 Share of GDP of EDEsb With survey data Total no yes no 1 13 15 With administrative data yes 5 80 85 Total 6 94 100 Note: a. EDEs’ population in 2022. b. EDEs’ GDP in US$ 2017 PPP in 2022. Data sources: mainly ASPIRE, plus other sources. ASPIRE data are harmonized, that is, spending and other monetary data are presented in comparable purchasing power (US$ 2017 PPP). Program coverage data are presented in terms of direct and indirect beneficiaries unless 83 explicitly mentioned, and programs are grouped into homogenous categories and subcategories. 84 Annex 2: Assumptions of a technically and politically feasible social assistance program mix where welfare targeted programs deliver a targeting performance at the pro-poor targeting frontier This paper provides an up-to-date estimate of the annual financing gaps required to eliminate the income shortfall of the extreme poor and the poorest quintile in the 133 EDEs. For 73 countries with recent household surveys, the section provides direct estimates;21 for the remaining 60 countries, we provide indirect estimates. For each EDE, we have quantified three key statistics: • Annual income shortfall of the poor before and after the receipt of social assistance programs. The difference between the two is the contribution of social assistance spending in reducing the respective gap. We use 2022 estimates of the most recent income shortfall of the extreme poor (XP) and of the poorest quintile (relatively poor or RP) after the receipt of social protection transfers from the Poverty and Inequality Platform (DEC). These gaps are then expressed as the total annual income shortfall of the extreme or relatively poor in US$ 2017 PPP and as a share of GDP.  We also calculate the income shortfall of the XP and RP before (in the absence of) social assistance transfers, by estimating the share of SA spending that goes to the XP or RP (the contribution of social assistance programs to the reduction of the poverty gaps is estimated by multiplying the spending on social assistance programs in 2022 or the most recent year (from administrative data) with the share of social assistance beneficiaries for extreme or relatively poor (from household survey data).  For countries without a recent household survey, we use the average beneficiaries’ incidence for the respective income group (LICs, LMICs, UMICs, or HICs). • Potential reduction of the annual income shortfall of the poor from improving the efficiency of the social assistance spending. On average, about 27 percent of the social assistance beneficiaries belong to the poorest quintile and 15 percent are extreme poor. This section assumes different level of improvement to the pro- 85 poorness of SA spending based on what has been achieved across the EDEs for well- targeted programs and the level of income of a country (LICs, LMICs, UMICs, or HICs). • The residual poverty gap that will require additional budget allocation. This variable quantifies the poverty gaps not covered by a more pro-poor allocation of the existing social assistance spending.  It measures the additional volume of resources needed— from domestic resources and, in the case of some LICs, from domestic and international resources—to eliminate extreme or relative poverty. Data limitations: Access to representative household surveys that track both household welfare and the receipt of social protection programs is lagging. For 111 countries, the analyses use the most recent household survey available in ASPIRE; the other 42 countries are estimated. Improving the pro-poor efficiency of social assistance transfers. The simulations presented in this section use an ambitious but feasible scenario of pro-poor social assistance spending, in line with what was achieved across the EDEs for well-targeted programs. They are anchored in targeting performance of real-world well-performing programs and explicitly account for inherent targeting errors. The assumed scenario also considers that targeting accuracy varies across countries with different income levels (lower in LICs than in MICs), with the program size and program target group. Not all social assistance programming is assumed to be disbursed via poverty- or welfare-targeted programs; one-quarter of the social assistance spending is assumed to finance categorical targeted programs. Finally, the scenario accounts for administrative costs, albeit the level accounted for correspond to efficient programs: programs that are at scale, finance all delivery service functions adequately, and are not wasteful (Grosh, Leite, Wai-Poi, and Tesliuc 2022). Specifically, this analysis uses the following assumptions for improving pro-poor efficiency of social assistance programs to reducing the income shortfall for extreme poor and relatively poor: 86 • Empirical data show that the 90th percentile best targeted social assistance program has a beneficiaries’ targeting accuracy in reaching the poorest quintile of 35 percent in LICs, 52 percent in MICs, and 63 percent in the new HICs; see Figure V.3. We have parametrized this performance in Table A.II.1, where the assumed benefits’ targeting accuracy in reaching the poorest quintile was set to 36 percent in LICs (25 percent in Q1 relative to 70 percent of total transfers for targeted programs); 54 percent in MICs (38 percent in Q1 relative to 70 percent of total transfers for targeted programs) and 64 percent in HICs (45 percent in Q1 relative to 70 percent of total transfers for targeted programs). We used these efficiency parameters (36, 54 and 64) to calculate the potential increase in targeting accuracy of the social assistance spending for LICs, MICs and new HICs. • Based on these assumptions, we have estimated the social assistance expenditure distribution for targeted and categorical programs by quintile (and decile for Q1) for countries according to their income level, in Table A.II. 1., panels A, B, and C. Table A.II. 1. Benefit Distribution for LICs, MICs, and HICs, Including Administrative Cost A. Low income Administrative   D1 D2 Q2 Q3 Q4 Q5 Total Cost Poverty targeted 15 10 20 10 8 7 4 74 Categorical 5 5 5 5 3 2 1 26 Total 20 15 25 15 11 9 5 100 B. Middle income Administrative   D1 D2 Q2 Q3 Q4 Q5 Total Cost Poverty targeted 22 16 20 6 4 2 4 74 Categorical 5 5 5 5 3 2 1 26 Total 27 21 25 11 7 4 5 100 C. High income Administrative   D1 D2 Q2 Q3 Q4 Q5 Cost Total Poverty targeted 25 20 20 2 2 1 4 74 Categorical 5 5 5 5 3 2 1 26 Total 30 25 25 7 5 3 5 100 87 • The total targeting accuracy for total social assistance spending is calculated in Table A.II. 2, panels A, B, and C, as the combined accuracy of categorical and poverty targeted programs. This accuracy varies with the coverage of the program or, for our simulations, with the extreme poverty headcount. For instance, for Angola which is an LMIC with an extreme poverty incidence, in 2022, of 35 percent, the efficiency assumption we use corresponds to the one in Table A.II. 2., Panel B Q2 of 73 percent. Table A.II. 2. Pro-poor Cumulative Benefit Distribution by Quintile for LICs, MICs, and HICs (%) A. Low income   D1 D2 Q2 Q3 Q4 Q5 Poverty targeted 20 34 61 74 85 95 Categorical 19 38 58 77 88 96 Total 20 35 60 75 86 95 B. Middle income   D1 D2 Q2 Q3 Q4 Q5 Poverty targeted 30 51 78 86 92 95 Categorical 19 38 58 77 88 96 Total 27 48 73 84 91 95 C. High income   D1 D2 Q2 Q3 Q4 Q5 Poverty targeted 34 61 88 91 93 95 Categorical 19 38 58 77 88 96 Total 30 55 80 87 92 95 88 Social Protection & Jobs Discussion Paper Series Titles 2025 No. Title 2508 State of Social Protection Report 2025: The 2-Billion-Person Challenge. Background Paper #1: Mind the Gap: Coverage, Adequacy and Financing Gaps in Social Protection for the Extreme Poor and the Poorest Quintile 2507 Service Integration and Case Management for People on the Move: A Review of Selected International Practices 2506 Impact of Climate Change and the Green Transition on Human Capital: A Review of the Evidence from Europe and Central Asia 2505 A slippery slope: the opportunities and risks of digital approaches and technology in Social Protection Systems 2504 De Jure and De Facto Coverage of Parental Benefits in Nepal 2503 Awareness, Access, and Perceptions around Parental benefits among Urban Argentinians 2502 Regulating Markets So More People Find Better Jobs 2501 São Tomé and Príncipe Unpacking Migration Dynamics: Critical Issues and Policy Recommendations To view Social Protection & Jobs Discussion Papers published prior to 2021, please visit www.worldbank.org/sp. ABSTRACT Mind the Gap evaluates the progress in strengthening social protection and labor systems in emerging and developing economies (EDEs) in three areas: coverage, adequacy of benefits for the poor, and level of financing. The paper explores policy options to reduce the coverage and income gap for the world’s poorest. Over the past decade, social protection coverage in the average EDEs has increased by 10 percentage points, from 41 percent around 2010 to approximately 51 percent in 2022. In absolute terms, as of 2022, 4.7 billion out of 6.3 billion people in low- and middle-income countries were covered by social protection, while 1.6 billion were not covered at all. Moreover, about 0.4 billion people from the poorest quintile received insufficient social protection support. Overall, 2 billion people in low- and middle-income countries remain uncovered, or inadequately covered while poor, by social protection. Apart from coverage, Mind the Gap also evaluates the extent to which EDEs are progressing toward Universal Social Protection across various policy dimensions: benefit adequacy, financing levels, and targeting accuracy of social assistance programs. The findings indicate variable adequacy in benefit levels, significant government spending but persistent financing gaps, and generally modest pro-poor targeting with potential for improvement. These challenges are more pronounced in low-income countries. The paper also assesses the impact of social assistance programs on reducing extreme and relative poverty gaps, quantifies the annual costs needed to address these income shortfalls in EDEs, and explores options for closing these gaps, such as enhancing the efficiency of social assistance transfers and expanding fiscal space. JEL CODES H53, I38, J28, O15 KEYWORDS Social protection system, coverage gap, social protection spending, adequacy gap, social protection targeting. ABOUT THIS SERIES Social Protection & Jobs Discussion Papers are published to communicate the results of The World Bank’s work to the development community with the least possible delay. This paper therefore has not been prepared in accordance with the procedures appropriate for formally edited texts. For more information, please visit us online at www.worldbank.org/socialprotection