CHAPTER 1

Explaining the Social Safety Net’s Data Landscape

INTRODUCTION

What are social protection and social safety net (SSN) interventions? How does this book classify SSN programs? What is the Atlas of Social Protection: Indicators of Resilience and Equity (ASPIRE) database? How does the ASPIRE team collect data and ensure data quality? What are the limitations of the administrative and household survey data used in this book? How is the performance of SSN programs measured? This chapter aims to answer these questions, and, by doing so, lays out the landscape for understanding the book in its entirety.

WHAT ARE SOCIAL PROTECTION AND SOCIAL SAFETY NET INTERVENTIONS? WHAT IS THE ASPIRE CLASSIFICATION OF SOCIAL SAFETY NET PROGRAMS?

Social protection and labor (SPL) interventions are well recognized for promoting resilience, equity, and opportunity. The World Bank 2012–2022 Social Protection and Labor Strategy: Resilience, Equity, and Opportunity argues that SPL systems, policies, and instruments help individuals and societies manage risk and volatility and protect them from poverty and destitution (World Bank 2012). Equity is enhanced through instruments that help protect against destitution and promote equality of opportunity. Resilience is promoted through programs that minimize the negative effect of economic shocks and natural disasters on individuals and families. Opportunity is enhanced through policies and instruments that contribute to building human capital and facilitate access to jobs and investments in livelihoods.

SPL instruments generally fall into the following three categories:

1. Social safety net (SSN)/social assistance (SA) programs are noncontributory interventions designed to help individuals and households cope with chronic poverty, destitution, and vulnerability. SSN/SA programs target the poor and vulnerable. Examples include unconditional and conditional cash transfers, noncontributory social pensions, food and in-kind transfers, school feeding programs, public works, and fee waivers (see table 1.1).

2. Social insurance programs are contributory interventions that are designed to help individuals manage sudden changes in income because of old age, sickness, disability, or natural disaster. Individuals pay insurance premiums to be eligible for coverage or contribute a percentage of their earnings to a mandatory insurance scheme. Examples include contributory old-age, survivor, and disability pensions; sick leave and maternity/paternity benefits; and health insurance coverage.

3. Labor market programs can be contributory or noncontributory programs and are designed to help protect individuals against loss of income from unemployment (passive labor market policies) or help individuals acquire skills and connect them to labor markets (active labor market policies). Unemployment insurance and early retirement incentives are examples of passive labor market policies, whereas training, employment intermediation services, and wage subsidies are examples of active policies.

TABLE 1.1 Social Protection and Labor Market Intervention Areas

Social protection and labor programs	Objectives	Types of programs
Social safety nets/social assistance (noncontributory)	Reduce poverty and inequality	• Unconditional cash transfers • Conditional cash transfers • Social pensions • Food and in-kind transfers • School feeding programs • Public works • Fee waivers and targeted subsidies • Other interventions (social services)
Social insurance (contributory)	Ensure adequate living standards in the face of shocks and life changes	• Contributory old-age, survivor, and disability pensions • Sick leave • Maternity/paternity benefits • Health insurance coverage • Other types of insurance
Labor market programs (contributory and noncontributory)	Improve chances of employment and earnings; smooth income during unemployment	• Active labor market programs (training, employment intermediation services, wage subsidies) • Passive labor market programs (unemployment insurance, early retirement incentives)

Source: World Bank 2012.

Note: ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

This book, in its empirical analysis, focuses on the state of SSN/SA programs. This focus reflects an increased use of social safety net instruments as well as the need to capture up-to-date data and to assess the status of SSN programs globally. The advance of available data resources positions the ASPIRE team well to analyze SSN/SA programs. In addition, by focusing on the SSN/SA programs (as a subset of SPL programs), this book provides continuity with the two previous books on the state of social safety nets (2014 and 2015).

This book also extends to the broader SPL agenda. The “Special Topics” section considers old-age social pensions, which are linked to the social insurance agenda, and adaptive social protection, which can be achieved through both safety nets and contributory/insurance programs. Ultimately, the policy issues related to social safety nets, social insurance, and labor market agendas are closely connected.

There is clear demand for a tool that helps monitor the scope, performance, and effect of SPL programs in countries worldwide. The Social Protection and Jobs Global Practice of the World Bank Group is committed to developing and continuously updating a comprehensive set of comparable and accessible indicators to help measure the performance of SSN/SA (as well as broader SPL) programs.

The Social Protection and Jobs Global Practice has created a user-friendly benchmarking tool that continuously updates key SSN/SPL indicators: ASPIRE. This portal serves as a one-stop shop for SPL indicators for both World Bank staff and external practitioners. ASPIRE links directly to the World Bank Group Databank to provide users with tools to search the database and to generate customized tables and charts. In addition, the portal includes related survey information from the World Bank Microdata Library.

For cross-country comparability, this book follows the ASPIRE harmonized classification of SSN/SA programs. ASPIRE groups SSN/SA programs into eight harmonized categories on the basis of program objectives (see table 1.2 and appendix A). This classification is applied to each country in the database to generate comparable program expenditure and performance indicators. Whereas table 1.2 reflects various types of social pensions as captured by the ASPIRE database, this book’s Special Topics section focuses exclusively on old-age social pensions, which facilitates a clean comparison of this instrument across countries.

TABLE 1.2 ASPIRE Classification of Social Safety Net Programs

Program category	Program subcategory
Unconditional cash transfersa	Poverty-targeted cash transfers, last-resort programs
	Family, children, orphan allowance, including orphans and vulnerable children benefits
	Noncontributory funeral grants, burial allowances
	Emergency cash support, including support to refugees and returning migrants
	Public charity, including zakāt
Conditional cash transfersb	Conditional cash transfers
Social pensions (noncontributory)c	Old-age social pensions
	Disability benefits
	War veteran benefits
	Survivorship benefits
Food and in-kind transfers	Food stamps, rations, vouchers
	Nutrition programs (therapeutic, supplementary feeding)
	School supplies (free textbooks, uniforms)
	In-kind/nonfood emergency support
	Other in-kind transfers
School feeding	School feeding programs
Public works, workfare, and direct job creation	Cash-for-work
	Food-for-work, including food-for-training, food-for-assets
Fee waivers and targeted subsidies	Health insurance exemptions, reduced medical fees
	Education fee waivers
	Food subsidies
	Housing subsidies and allowances
	Utility and electricity subsidies and allowances
	Agricultural-inputs subsidies
	Transportation benefits
Other social assistance	Scholarships, education benefits
	Social services, transfers for caregivers (care for children, youth, family, working-age, disabled, and older persons)
	Tax exemptions

Source: ASPIRE database.

Note:

a. Conditional cash transfer programs aim to reduce poverty by making welfare programs conditional upon actions by the beneficiary. The government (or an implementing agency) transfers the money only to those households or persons (beneficiaries) that meet certain criteria in the form of actions, such as enrolling children in public schools, getting regular check-ups at the doctor’s office, receiving vaccinations, or the like. Conditional cash transfer programs seek to help the current generation in poverty and to break the cycle of poverty for the next generation by developing human capital.

b. Unconditional cash transfer programs do not require beneficiaries to perform any specific actions to be eligible for the benefit. However, these programs may require benficiaries to meet certain criteria or have a certain status to be eligible; for example, for a poverty-targeted benefit, a household must be below a poverty threshold.

c. Social pensions here encompass various types of social pensions, such as old-age pensions, disability benefits, and survivorship benefits, whereas chapter 4 focuses exclusively on old-age social pensions.

ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

WHAT IS THE ASPIRE DATABASE? HOW DOES THE ASPIRE TEAM COLLECT AND ENSURE QUALITY OF THE DATA? WHAT ARE THE LIMITATIONS OF THE DATA?

The ASPIRE database has achieved significant scale and has become the World Bank’s premier compilation of performance indicators for social protection and labor programs. ASPIRE has two main sources of data: administrative data, from which program expenditures and number of beneficiaries are derived; and household survey data, which are used to estimate the coverage, benefit incidence, benefit levels, and poverty/inequality impact of SPL programs. The two data sources complement each other, and thus provide a more comprehensive view of SPL program performance around the world.

The ASPIRE work program supports continuous improvement in the quality, comparability, and availability of SPL/SSN data to facilitate SPL benchmarking and inform policies. As of November of 2017, the ASPIRE database included administrative spending data for 124 developing and transition countries and economies (see appendix D) and administrative data on the number of beneficiaries of the largest programs for 142 countries (see appendix C).¹ This book uses only the most recent subset of the ASPIRE data—specifically, the most recent year of data available per country. These data constitute the basis for the analysis in chapter 2. For the performance analysis presented in chapter 3, the book uses the most recent household survey data from 96 countries (see appendix B).² The examples of monthly benefit levels per household for selected programs are presented in appendix E. Appendix F presents key performance indicators. Appendix G takes stock of old-age social pensions around the world (which are discussed in chapter 4). The full list of countries found in the household and administrative data used in this book is presented, by country income group, in table 1.3. Basic characteristics of these countries, such as the total country population and gross national income GDP per capita, can be found in appendix H.

TABLE 1.3 Countries with the Household and Administrative Data Used in This Book, by Country Income Group

Source: ASPIRE database.

Note: Economies are divided among income groups according to 2016 gross national income per capita, calculated using the World Bank Atlas method. The groups are as follows: low-income, US$1,005 or less; lower-middle-income, US$1,006–3,955; upper-middle-income, US$3,956–12,235; and high-income, US$12,236 or more. See appendix H for gross national income per capita statistics for individual countries. ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity.

World Bank staff and in-country consultants collect and harmonize the administrative data using standardized terms of reference, data templates, and classifications. Publicly available government statistics, such as annual program budget expenditures, are the primary source of administrative data. In the case of donor-funded programs, the program budget provided by the donor is also considered a primary source of data. Other information and data received from program and sector officials, as well as existing analysis such as public expenditure reviews, constitute the secondary source of administrative data.

Although the SSN classification of programs facilitates cross-country comparison, it does not necessarily imply an easy and clean-cut differentiation of programs. As mentioned, ASPIRE classifies and aggregates individual SSN/SA programs into eight categories, largely on the basis of the objective and nature of each program. However, in practice, program objectives often tend to overlap, blurring the line between classifications. For example, although a cash transfer program may not have explicit eligibility conditions (making it an unconditional cash transfer), it may have strong uptake incentive mechanisms or soft conditions that influence decisions on how households spend the transfer, making it in principle a conditional cash transfer (Daidone and others 2015).

Available program-level administrative spending data currently covers 124 countries representing 80 percent of the world’s population. Updates are available for 28 countries through 2016; for 42 countries through 2015; and for 41 countries through either 2013 or 2014. The year of reference for the remaining countries in the database ranges from 2010 to 2013, except for Bhutan, Jordan, Marshall Islands, and Vanuatu. For these four countries, only total SSN spending is available from secondary sources, and the reference year is 2009. Countries with data points before 2009 are considered outdated and are not included in the analysis. A complete summary of spending indicators disaggregated by program categories can be found in appendix D.³ The program-level administrative data facilitates a granular look at country-level spending on social safety nets/social assistance. Furthermore, by comparing global spending trends and patterns, the spending profiles and program portfolios of countries and regions can be benchmarked.

The presence on the ground of the larger ASPIRE and Social Protection and Jobs Global Practice teams, including consultants, facilitates information flows that help improve the quality of administrative data. The engagement in the country helps establish a dialogue with government counterparts and assists in gathering the required information/data, verifying classification, and checking quality. Continuous improvement to the data is ensured by close collaboration between the ASPIRE central team, the ASPIRE focal points in the regions, and the World Bank Social Protection and Jobs Global Practice staff at the country level, who have extensive program knowledge.

When estimating the amount a country spends on SSN programs, the book uses the latest-available-year approach. For the list of all active programs, the latest year for which updates are available (as mentioned) is considered the reference year for which expenditures are tallied. Unfortunately, for some countries, spending information for the latest year is not available for all active programs, but in many cases prior-year information on spending is available. In such cases, this prior-year spending (relative to the same-year gross domestic product, GDP) is used. In sorting out the data, the focus is always on updating the largest programs in terms of beneficiary numbers and spending amounts. The analysis of spending (see chapter 2) also distinguishes the inclusion/exclusion of the health fee waivers in total SSN spending whenever possible.⁴

Performance indicators are estimated using nationally representative household surveys. As of November of 2017, the ASPIRE database included 309 household surveys, corresponding to 123 developing countries. The book uses only the latest year for each country and only if the data are from at least 2008; under this criterion, 20 countries were excluded. In addition, several countries whose surveys did not have SSN information were excluded: Cambodia (2013), Mali (2009), Myanmar (2009), Samoa (2008), Togo (2011), and Tonga (2009).⁵ As a result, the performance indicators are based on 96 countries (see appendix B for a full list of the household surveys used).

The ASPIRE team carefully reviews different household surveys to identify relevant SPL program information. Typically, the surveys include household income and expenditure surveys; household budget surveys; living standards measurement surveys; integrated, multipurpose, and socioeconomic surveys; or any other survey that is nationally representative and captures information on social protection and labor programs. In some cases, this work also leads to recommendations made to government counterparts on how the design of the survey instrument/module can be changed to better capture SPL programs.

Individual variables are generated for each SSN program captured in the survey; they are then grouped into the eight harmonized program categories.⁶ The performance indicators are generated using these harmonized program categories. These indicators, in turn, can be disaggregated by quintiles of welfare before and after transfers, extreme poverty status (defined as US$1.90/day in terms of purchasing power parity, PPP), and rural/urban populations. Household weights are used to expand results to the total population of each country.

For cross-country comparability, all monetary variables are expressed in 2011 prices and daily PPP in U.S. dollars. This also facilitates the PPP US$1.90/day poverty-line metric to determine the poverty status for each country/survey. Note that 2011 is used as a base year because this is the year when the most recent comprehensive global price statistics were collected as a part of the International Comparison Program.⁷ The consumption or income aggregates used to rank households by their welfare distribution are validated by the World Bank’s regional poverty teams.

There are important considerations to keep in mind when going through the performance analysis. First, this analysis is limited to the programs captured in the household surveys. Most household surveys capture only a fraction of the programs administered in a given country. Thus the data do not always include a comprehensive list of programs (which are likely to appear in administrative data) implemented in each country. Accordingly, coverage indicators are underestimated with respect to overall social spending. To illustrate this point, the ASPIRE team conducted a matching exercise, looking at program overlap between the administrative data and household surveys for several countries (see table 1.4). A few key messages emerge from this exercise:

TABLE 1.4 Matching of Administrative and Household Survey Data for Social Safety Net Programs for Selected Countries/Economies

Source: ASPIRE team calculations, 2017.

Note: For Vietnam, out of 58 programs, 21 are under Decree 136 (also called Program 136). Hence, Program 136 is a breakdown into 21 programs. ASPIRE = Atlas of Social Protection: Indicators of Resilience and Equity; HH = household; SSN = social safety net.

1. There is generally little overlap between administrative and household data. In the sample of counties (see table 1.4), on average only about 20 percent of programs can be found in both administrative and household survey data; for some countries, the matching rate is less than 10 percent.

2. Household surveys tend to capture larger programs, although only in part. On average, the matching programs capture about 50 percent of the SSN budget, as seen by summing up the budget for the matching programs on the basis of administrative budget data.

3. Every country case is unique. There are significant variations across countries in terms of how many programs are captured in the household survey (as a percentage of the total number of programs) and what percent of the total budget they account for in the administrative data. For example, in Chile, 14 out of 135 programs (10 percent) match, accounting for 30 percent of SSN programs’ total budget; in Romania, 10 out of 65 (15 percent) match, accounting for 96 percent of the total budget; and in South Africa, 6 out of 16 (40 percent) match, accounting for 85 percent of the total budget (see table 1.4).

Also, some surveys collect information only on program participation without including transfer amounts. In such cases, only coverage and beneficiary incidence indicators can be estimated. Last, because household surveys differ in the method for collecting SPL information across countries, the quality of the information varies. For example, some surveys collect information on social programs mixed with private transfers, making it difficult to isolate individual SPL programs. Despite these limitations, household surveys have unique advantages (see box 1.1) and are the sole source for calculating most performance indicators presented in this book.

BOX 1.1 Leveraging Household Survey Data to Monitor and Measure Social Protection and Labor Program Performance

Household surveys have great potential as instruments to monitor and assess the performance of social protection and labor (SPL) programs. However, not all countries use household surveys to estimate SPL program trends or generate basic performance indicators (such as coverage, benefit level, benefits and beneficiary incidence, and effects on poverty). A main factor behind the low use of household data is the inadequate SPL information captured in most national household surveys. Thus there is a need to improve data collection and the quality of SPL information in household surveys to better inform social policy.

Why are household surveys necessary to measure SPL performance? Household surveys are the only source of information regarding potential beneficiaries and the basis for ex ante simulations for policy reform. By including information representative of the total population, household surveys allow the identification of populations that, because of their characteristics, may be eligible for an SPL program (for example, the poor, disabled, and unemployed). Ex ante assessments can be conducted for policy reforms by simulating the effect of a newly introduced program or parameter-adjusted existing programs. In addition, ex post assessments facilitate evaluating whether SPL programs are reaching intended objectives. The availability of total household income or consumption in household surveys also enables analysis of the distributional effects of SPL programs and their effects on poverty and inequality.

How can household surveys be leveraged to become key instruments for monitoring and evaluating social policy? Leite et al. (forthcoming) propose a series of recommendations to improve the collection of SPL information in survey instruments, including the following:

1. Review existing SPL programs in the country. Obtaining a full list of programs and their specifications (for example, target population, benefit level, frequency of payments, and program size) will help make the list of SPL programs in the questionnaire more complete and better formulate the survey questions to capture adequate information. In addition, information about the program size will help evaluate whether a program is large enough to be captured by the sample frame or if oversampling is needed.

2. Identify and coordinate with key partners. Coordination between policy makers, program implementers, and National Statistical Office officials is crucial to design a good set of questions and sampling frame. The survey’s representation of programs can be imprecise if the sample does not overlap with areas where the programs are implemented.

3. Design the best format to collect SPL program information. Whenever possible, survey questions should be specific for each program, keeping answers at the individual level if the programs are provided to the individual (and not to the household). In addition, recording the value of the benefit (or an estimated value) makes performance analysis richer because monetary-based indicators (such as benefits incidence, benefit size, and effects on poverty and inequality) can be estimated. Moreover, different collection formats can be explored, such as designing modules specific to social safety net programs and/or placing questions in sector-specific modules, given that SPL programs tend to be multisectorial.

Source: Leite et al., forthcoming.

Despite data limitations, the global SPL landscape today is much more accurate than even a few years ago because of advances in the identification, capture, and harmonization of ASPIRE data. This global accumulation of knowledge is reflected in this book, which builds on more extensive data (more countries/programs captured) and adds sophistication of analysis relative to the 2015 book on the state of social safety nets (Honorati, Gentilini, and Yemtsov 2015). Furthermore, for the first time, this book attempts to look at SSN programs over time when the data allow intertemporal comparisons.

HOW IS THE PERFORMANCE OF SOCIAL SAFETY NET PROGRAMS MEASURED?

Performance measurement is the process of collecting, analyzing, and/or reporting information regarding the performance of an individual, group, organization, system, or program component. The objective is to determine if the results/outputs align with the intention or the intended achievement. Performance measurement estimates the parameters under which programs are reaching the targeted results. By measuring performance, decisions can be made and interventions carried out to improve programs.

The analysis of SSN programs presented in this book relies on a number of key terms/parameters. These parameters include spending/budget, number of beneficiaries, coverage, beneficiary/benefit incidence, benefit size/adequacy, and poverty/inequality impact. This book focuses on the core performance indicators found in the ASPIRE database. Accordingly, the effects of SSN programs in such areas as health or education outcomes, saving behavior, labor supply, fertility, and migration are not considered; these effects can be measured only through rigorous impact evaluations.8 The main contribution of this book is to present a set of comparable core indicators for many programs/countries, allowing a global picture of social safety nets to evolve.

“Spending” indicates the program budget

In most cases, the costs of benefits provided account for most spending. While most programs have administrative costs (which are the costs of running/implementing the program), those are rarely available and/or cannot be separated from the amount spent on benefits.

“Number of beneficiaries” is simply how many people or households benefit from the program

Usually, this information is available in the administrative data. The intricacy here is related to what the beneficiary unit is. Many programs are targeted at households as a beneficiary unit. In this case, it is often assumed that all household members benefit from the program/benefit, and hence the number of individual beneficiaries is simply the number of individuals living in the beneficiary households. For some programs, such as conditional cash transfers, very often a subset of the household members is assumed to benefit directly from the program (for example, children who get vaccinations or go to school). In this case, the book simply uses the number of direct beneficiaries provided (through primary and secondary data, as described earlier), without any further calculations. For individual-level benefits (for example, old-age social pensions), the number of direct beneficiaries is reported (even though, indirectly, all household members may benefit from a household member receiving a benefit). In any case, in the administrative data, the original beneficiary units are always reported (see appendix C). For the household-level benefits, the number of recipient households and the number of individuals living in those households are reported.

“Coverage” indicates the absolute number of program beneficiaries or percentage of the population or a given population group that benefits from a given SSN program

Coverage is important because it indicates the size of the program “blanket” in both absolute and relative terms. In the ideal world, the number of beneficiaries from the administrative data could be closely matched by coverage (of the same program) from the household survey data (using population weights). However, as table 1.4 demonstrates, this ideal is elusive. For the purposes of the performance analysis (chapter 3), this book evaluates coverage relying on the household survey data. This approach is taken because it would be helpful to know how various population groups (for example, poor versus nonpoor) are covered by the same program (that can be found in the household survey). This level of analysis is simply not possible using administrative data. Coverage, in combination with benefit size/adequacy, is very often related to the program’s impact.

“Beneficiary/benefit incidence” shows which segment of the population receives the program benefits

The beneficiary/benefit incidence can indicate what percentage of the total number of beneficiaries/total amount of benefits go to the poorest quintile of the welfare distribution. The calculation of this indicator requires the use of household survey data that include the welfare indicator. Moreover, the household survey needs to have the information about the SSN programs for which the benefit incidence is being assessed. Thus, the data demands are very high when it comes to estimating this parameter.

“Benefit level” indicates the amount of the benefit, whereas “benefit adequacy” is a measure of the relative benefit level

The main purpose of estimating benefit adequacy is to get some idea of to what extent the benefit size is small or large in comparison to a benchmark (for example, average income/consumption in a country, poverty line, minimum subsistence level, minimum wage, per capita GDP). The impact evaluation literature (cited later in this book) often finds that fragmented/small benefits fall short of achieving desired developmental effects.

“Poverty/inequality impact” reveals the distributional effects of the benefit

Regarding poverty impact, two indicators are often looked at: percentage reduction in the poverty headcount (prevalence) as a result of the benefit; and percentage reduction in the poverty depth (distance to the poverty line). The cost–benefit ratio can also be calculated. It indicates how much money, in U.S. dollars, it costs to reduce a poverty gap by US$1. As empirical evidence around the world suggests, many SSN benefits help poor people become less poor (that is, reduce the poverty gap/depth) rather than graduate entirely from poverty. Many SSN benefits also often help make societies more equal. This is estimated empirically by looking at the reduction in the measures of welfare (income/consumption) inequality, such as the Gini coefficient.

NOTES

1. Appendix C presents information available in the ASPIRE database on the biggest programs (in terms of numbers of beneficiaries) in 142 countries by aggregate program categories. Countries differ significantly in the number of SSN programs operating in the country, ranging from fewer than 10–15 (such as in Bolivia, Croatia, or Timor-Leste), to more than 50 programs (such as in Burkina Faso, Chile, or Vietnam). Thus, for some countries with a large number of programs, appendix C does not present the full picture of coverage or versatility of programs and should be treated with caution.

2. Data availability here refers to the most recent data available to the ASPIRE team. In some cases, more recent household survey data may be available for a given country, but these data have not been properly processed yet, or the welfare aggregate has not yet been derived, rendering the data unusable for calculating the performance indicators.

3. To calculate total spending as a percentage of GDP, program-level spending is divided by GDP using the GDP data from the corresponding year. In this chapter and in appendix D, the World Development Indicators database (July 2017 version) is used for all GDP data except for Timor-Leste, which uses the World Economic Outlook database (April 2017 version). The World Bank income group classification as of July of 2017 is used.

4. To make clear which categories of spending are presented in chapter 2, the figures and notes to each relevant figure indicate whether the data cover total SSN spending (including health fee waivers) or “core” SSN spending (excluding health fee waivers). This technique could potentially be used with other categories, such as educational fee waivers.

5. For the household survey data to be included in the analysis: (i) household surveys need to be nationally representative; (ii) they need to include information on social protection; (iii) there is a clearly defined welfare aggregate (either income or consumption). On the basis of these criteria, the household surveys for Azerbaijan and Lesotho, for example, were not used in the analysis. In the case of Azerbaijan, the survey is a nonrandom sample of the applicants to Targeted Social Assistance; in the case of Lesotho, it is the only country in the sample where the asset index (rather than consumption or income) is used for welfare rankings.

6. SSN/SA includes eight harmonized program categories, whereas a broader SPL includes 12 harmonized program categories. See appendix A for further details.

7. See http://siteresources.worldbank.org/ICPEXT/Resources/ICP_2011.html/.

8. Chapter 3 reviews some of these studies on the role of SSN in enhancing productive inclusion.

REFERENCES

ASPIRE (Atlas of Social Protection: Indicators of Resilience and Equity). 2017. “Data Sources and Methodology.” Database, World Bank, Washington, DC. http://datatopics.worldbank.org/aspire/~/documentation/.

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

Honorati, M., U. Gentilini, and R. Yemtsov. 2015. The State of Social Safety Nets 2015. Washington, DC: World Bank.

Leite, P., C. Rodríguez Alas, and V. Reboul. Forthcoming. “Measuring Social Protection and Labor Programs through Household Surveys.” Policy Research Working Paper, World Bank, Washington, DC.

World Bank. 2012. “Resilience, Equity, and Opportunity: The World Bank Social Protection Strategy 2012–2022.” World Bank, Washington, DC.