The SociAl ProTecTion inDicATor for ASiA

This publication provides updates on Social Protection Indicators of 24 countries in Asia, with an analysis of 2015 data on social protection programs. It shows progress in expenditure, primarily driven by social insurance and coverage between 2009 and 2015. Spending on women has improved in several countries, yet others continued to favor the nonpoor over the poor, and men over women. The Social Protection Index— now the Social Protection Indicator—was developed by the Asian Development Bank and its partners as the first comprehensive and quantitative measure of social protection systems in Asia and the Pacific.

In 2015, the Sustainable Development Goals were adopted by United Nations member states to serve as a blueprint for achieving a better and more sustainable future for all. The provision of social protection was identified as one of the key elements of the first goal on ending poverty.
The Asian Development Bank (ADB) developed the Social Protection Indicator (SPI) in 2005 as a tool for monitoring and analyzing social protection at the country and regional levels. The SPI captures various aspects of social protection systems including expenditure, coverage, benefit levels, poverty and gender dimensions, and progress over time. Since then, the SPI has been used by the region's governments, international organizations, civil society, and researchers for assessing the quality and scope of social protection in the region.
This report presents an analysis of the 2015 data on social protection programs in 25 countries in Asia and is an update of the previous report (2016) that analyzed 2012 data. It is hoped that the findings will contribute to efforts to monitor and assess social protection, and to inform the design and delivery of current and future social protection in the region.
The SPI is even more important as ADB implements its new Strategy 2030, specifically in advancing the first operational priority (OP1) of addressing remaining poverty and reducing inequalities in Asia and the Pacific. OP1 will aim to enhance human capital and social inclusion, generate quality jobs, and reduce inequality in access to opportunities. Under OP1, ADB is committed to strengthening its support to social protection systems and services in the region by employing a differentiated approach to varying needs across countries and among vulnerable populations, and addressing the requirements for social protection throughout individual lifecycles. OP1 seeks to expand the coverage and quality of social protection systems and services. The SPI is one effort by ADB to develop robust data and a strong knowledge base in Asia and the Pacific which can support the development of social protection in the region.

SP
social protection SPI Social Protection Indicator SRI Sri Lanka TAJ Tajikistan  THA  Thailand  UMIC upper-middle-income country UZB Uzbekistan VIE Viet Nam This report presents the analysis of comprehensive 2015 data on government social protection systems in 25 countries in Asia. It updates the results of the 2012 analysis published in the Asian Development Bank (ADB) report The Social Protection Indicator: Assessing Results for Asia (ADB 2016b). It thus continues the effort to systematically track and assess the developments in social protection began by ADB and its development partners in 2005.
The report uses the Social Protection Indicator (SPI) for assessing the nature and performance of social protection within and across countries in Asia. The SPI approach highlights the level of resources invested in social protection both as a share of aggregate gross domestic product (GDP) and as a share of GDP per capita for each intended beneficiary-the latter is the SPI itself-as well as the extent of coverage, benefit size, and distribution of expenditure along poverty and gender lines. It also reviews trends between 2009 (based on ADB 2013) and 2015. The analysis of 2015 data in this report follows the methodology used in the 2016 publication and the results in both reports are compatible.
The report provides data on all types of government initiatives and thus enables one to assess social protection effectiveness as a "system." It allows an assessment of the relative significance of the major categories of social protection: social insurance, social assistance, and active labor market programs (ALMPs); and corresponding programs (e.g., pensions, health insurance, welfare assistance, child welfare, and others).

Poverty Dimension
Social protection spending in Asia favors the nonpoor over the poor (defined as the population living below the nationally defined poverty line). The expenditure on each nonpoor beneficiary amounted to 3.0% of GDP per capita, against 1.0% for the poor. The discrepancy was the highest in social insurance, where spending on the nonpoor was five times as high (2.6%) as on the poor (0.5%), given that the poor are the least represented in social insurance.
The results were more equitable in social assistance, where spending was evenly split between the poor and nonpoor at 0.4% of GDP per capita. Social assistance expenditure on the poor was higher in 13 countries and equal to that on the nonpoor in four countries. These results reflect the fact that social assistance is primarily used as a tool for supporting the poor in the region. However, social assistance spending favored the nonpoor in eight countries.
In ALMPs, the difference in spending on the poor and nonpoor was negligible (albeit with slightly pro-poor results). While public works programs specifically target the poor, skills development programs do not necessarily set income levels as an eligibility condition.

Gender Dimension
Social protection expenditure in Asia prioritized men over women, at 2.1% of GDP per capita on men and 1.9% on women. Social insurance expenditure favored men in 20 countries. Differential access to social insurance can be explained by gender discrimination in the labor force. Fewer women have access to formal wage and salaried employment, and women tend to have lower and The average social protection benefits in 23 countries in Asia (that is, excluding Bangladesh for technical reasons, as well as Myanmar) for each actual beneficiary improved from 8.9% of GDP per capita in 2009 to 11.4% in 2015. Social insurance benefits increased from 20.8% to 22.1% of GDP per capita. Ten countries improved social insurance benefits, and 13 saw them decline. Social assistance benefits for each actual beneficiary remained unchanged at 5.7% of GDP per capita, but 11 countries improved, and 12 countries decreased, their benefits.
Between 2009 and 2015, social protection spending in 24 Asian countries continued to favor the nonpoor over the poor. Social insurance in most countries continued to favor the nonpoor, but the number of countries that favored the poor in social assistance increased from nine to 13. Fourteen countries increased expenditure on the poor in social insurance, 11 in social assistance, and 5 in ALMPs-the rest did not.

1
This report presents the analysis of comprehensive 2015 data on government social protection programs in 25 countries in Asia (excluding, notably, India). It uses the Social Protection Indicator (SPI) for assessing social protection within and across countries.
This report continues the effort to systematically track and assess developments in social protection. The Asian Development Bank (ADB) and its development partners originally developed its precursor-the Social Protection Index-in 2005 as a tool to monitor social protection in Asia and the Pacific. This report updates the results of a 2012 analysis for Asia published in 2016 (ADB 2016b).
Systematic and comprehensive monitoring and evaluation are essential for improving social protection systems to enhance their capacity to support the poor and vulnerable. The SPI approach offers uniform indicators that can help governments and other stakeholders assess the effectiveness of social protection systems and to track progress over time. At country level, it assesses social protection systems as well as progress. At cross-country level, it can help trace and assess patterns in social protection. This SPI monitoring helps track the progress of target 1.3 of the Sustainable Development Goals (SDGs). 1 The analysis in this report is useful for two reasons. First, the report provides data on all types of government social protection initiatives, focusing on aggregate expenditure, coverage, benefit size, and poverty and gender dimensions. This enables one to assess social protection as a system by looking at social protection as a whole. The SPI thus complements existing national monitoring and evaluation tools, which tend to focus on individual programs rather than the overall system.
Second, the SPI analysis allows a focus on distinct categories of social protection activities. It provides disaggregated data on components of social protection systems and allows one to assess their relative significance. These components comprise three "categories"-social insurance, social assistance, and active labor market programs (ALMPs)-and include corresponding "programs," like health insurance, pensions, and welfare assistance (Table 1).
This report also presents comparisons of social protection indicators for 24 countries over time between 2009 and 2015. 2 Since its inception, the SPI methodology has been reviewed and revised to refine the way the indicators are constructed and used. SPI results for [2004][2005]2009, and 2012 were published in three reports (ADB 2006(ADB , 2013(ADB , and 2016b. The analysis of 2015 data in this report follows the methodology used in the 2016 publication, and the results in both reports are compatible.

What Is the Social Protection Indicator?
The SPI is an indicator for assessing social protection effectiveness within and across countries. Going beyond measuring social protection expenditure as a share of aggregate GDP, the SPI assesses expenditure for each intended beneficiary as a share of GDP per capita. Intended beneficiary population groups 2 See footnote 1 in the Executive Summary. are the key unit for assessing social protection effectiveness in this report-in other words, the share of the population that could qualify for benefits from a particular social protection category and program. The report distinguishes intended beneficiaries from actual beneficiaries, that is, those given support in practice. This distinction is captured by the coverage indicator, which measures the share of actual beneficiaries in the total intended beneficiary population. The SPI is a relative indicator against the average GDP per capita of each country. 3 Table 2 presents the main intended beneficiary groups for each of the programs. A detailed explanation for defining and selecting the groups is in ADB (2011).

Methodology for Deriving the Indicators
Using 2015 data, this report draws on the methodology set out in ADB (2011), which provides a detailed explanation of defining and constructing the SPI and key data sources.
The information required for the analysis falls into two general categories: basic statistics and indicators; and data on social protection expenditure and on the number of social protection beneficiaries. The following data sources were used: government statistics and reports, reports by international financial institutions and bilateral agencies, discussions and interviews with agencies responsible for social protection, and household survey data.
For compiling the indicators, the programs included refer only to national or central government programs. 4 Programs implemented by local governments, civic groups, or other nongovernment entities are not included because they are usually short term and hard to monitor. However, programs by international organizations or international funding agencies administered by the government are included.
Throughout this report, most averages are unweighted. This represents a methodological choice, signifying that indicators for each country are treated as equal in weight to those in any other country.
Results on the coverage of SPI show some countries exceeding 100% of intended beneficiaries. This simply means that in some countries, beneficiaries may receive more than one kind of social protection. For example, an elderly person may receive some form of pension while receiving health assistance benefits.
The analysis of social protection is also carried out by countries' income levels and regional groupings (Table 3). 5 4 This exercise did not monitor programs at province or state level in large countries such as Indonesia, Pakistan, and the PRC, where the SPI values are only indicative of national programs and therefore might be lower than the true figures. 5 For the 2018 fiscal year (1 July 2017-30 June 2018), the World Bank classifies countries by income as follows: low-income economies with a gross national income (GNI) per capita of $1,005 or less in 2016; lower-middle-income economies with a GNI per capita between $1,006 and $3,955; upper-middle-income economies with a GNI per capita between $3,956 and $12,235; and highincome economies with a GNI per capita of $12,236 or more. https://datahelpdesk.worldbank.org/ knowledgebase/articles/906519 (accessed 20 December 2017).

Structure of the Book
Chapter 2 starts the analysis by presenting social protection expenditure in Asia as a share of aggregate GDP and expenditure per intended beneficiary as a share of GDP per capita-the SPI. This analysis continues in Chapter 3, which examines social protection expenditure by category and program. Chapter 4 discusses coverage and benefit size, while Chapter 5 provides an assessment of the poverty and gender dimensions of social protection. Chapter 6 examines the main trends in social protection in Asia between 2009 and 2015. Chapter 7 provides the conclusions.

2
• In 2015, overall expenditure on social protection as a share of aggregate GDP in 25 countries in Asia averaged 5.3%. The SPI ranged from 12.1% (Japan) to 0.1% of GDP (Myanmar). • There was a wide variation in overall spending at country level, from 21.1% of aggregate GDP (Japan) down to 0.1% (Myanmar). • Expenditure for each intended beneficiary as a share of GDP per capita-the SPI-averaged 4.0%.
This chapter highlights government expenditure on social protection in 2015 for 25 countries in Asia. It focuses on two metrics ( Figure 1): overall expenditure as a share of aggregate GDP; and expenditure for each intended beneficiary as a share of GDP per capita-the SPI. 6 Overall social protection expenditure as a share of aggregate GDP in Asia was 5.3% in 2015 (Appendix, By income group, the share of overall social protection expenditure in GDP was the highest in the three high-income countries (HICs) at an average of 11.6% of GDP, followed by the upper-middle-income countries (UMICs) at 6.0% of GDP. Social protection expenditure averaged 3.9% of GDP in the lower-middle-income countries (LMICs) (Figure 2).
The pattern of social protection expenditure for each intended beneficiary is like that of overall expenditure. Spending averaged 7.9% of GDP per capita in HICs, 5.2% in the UMIC group, and 2.8% in LMICs ( Figure 2). All three HICs (Japan, the Republic of Korea, and Singapore) were above the Asia average of 4.0% of GDP per capita (Appendix, Table A.5). Among the 16 countries in the LMIC group, four were above the Asian average: Uzbekistan (9.1%), Azerbaijan (8.3%), the Kyrgyz Republic (5.5%), and Viet Nam (4.1%).
By regional group, East Asia was ahead of other country groups, with overall spending of 11.5% of GDP and spending for each intended beneficiary at 6.4% of GDP per capita ( Figure 2). This group includes countries with high GDP per capita: Japan and the Republic of Korea. Expenditure was high in Central and West Asia, with the overall spending of 6.8% of GDP and spending for each intended beneficiary of 5.6% of GDP per capita (Appendix , Table A.1 and Table  A.5). Expenditure in both Southeast Asia and South Asia was below the regional average. In Southeast Asia, social protection spending was 3.0% of GDP and 2.6% of GDP per capita for each intended beneficiary. This group includes highincome Singapore as well as upper-middle-income Malaysia and Thailand.
Several factors explain the performance of countries in Asia. One of these factors is a country's income. Countries with high national income spent above  A country's income does not, however, fully explain social protection spending patterns, which also reflect policy preferences (World Bank 2015), and more specifically, the social protection agenda and policy framework. These set out legal provisions and program rules that define policy programs, intended beneficiary groups, eligibility criteria, and benefit levels, alongside the corresponding level of financing. The results of this study attest to this. Countries with the highest spending as a share of GDP per capita include, for example, Georgia, the Kyrgyz Republic, and Uzbekistan (Figure 1), which have higher expenditure as a share of GDP per capita than richer countries (e.g., the PRC and the Republic of Korea) as well as other countries with similar income levels, though the absolute amounts are generally very low.
Viet Nam is another country that spends relatively generously on social protection. With a socialist-oriented market economy, it has prioritized gradual expansion of comprehensive social protection. The country has a GDP per capita of $2,107 but spent 4.1% of GDP per capita for each intended beneficiary and 6.3% of aggregate GDP. This is largely driven by spending on social insurance, underpinned by the existing policy objective to achieve universal protection for health and old age.
Similarly, the existing commitments to universal coverage in Maldives and Thailand explain the overall high expenditure on social assistance there. Maldives has expanded the coverage and benefit levels of the noncontributory Old-Age Basic Pension and Senior Citizen Allowance for citizens over 65. In Thailand, the Universal Coverage Scheme (see Box 2 on page 16) has achieved near-universal health-care coverage.
These findings suggest that even countries with lower income can enhance spending on social protection, provided that they set appropriate policy goals and develop a pertinent policy framework. Prioritizing the allocation of public resources is therefore not only a technical but also a political matter.
• Averaging 4.2% of aggregate GDP and 3.1% of GDP per capita for each intended beneficiary among 25 countries in Asia, social insurance dominated social protection spending. Contributory pensions took up a large share of social insurance spending (2.8% of aggregate GDP). • Social assistance expenditure came to 1.1% of aggregate GDP, dominated by welfare assistance. Spending on social assistance was 0.8% of GDP per capita for each intended beneficiary. • Spending on ALMPs was a paltry 0.1% of aggregate and per capita GDP, with cash for work and food for work seemingly exceeding skills development and training.
This chapter examines social protection expenditure for the three categories and their programs (Appendix , Table A.2 and Table A.3).
The expenditure pattern across Asia shows the prevalence of social insurance over social assistance, while spending on ALMPs is negligible. Spending on social insurance averaged 4.2% of aggregate GDP; social assistance, 1.1%; and ALMPs, 0.1% ( Figure 3). The dominance of social insurance was driven by pensions. Welfare assistance was the main contributor to social assistance spending.
This pattern largely holds for social protection expenditure for each intended beneficiary as a share of GDP per capita ( Figure 3). Of the 4.0% average, social insurance accounted for 3.1%; social assistance, 0.9%; and ALMPs, a negligible amount (0.05%).
Spending on social insurance prevails over social assistance and ALMPs in all income groups. HICs have the highest spending in social insurance (10.0% of GDP) and UMICs in social assistance (1.6% of GDP) (Appendix,

Social Protection Spending by Category and Program
Similarly, spending on social insurance is higher than that on social assistance and ALMPs in all regions. East Asia exhibits the highest spending of 9.8% of GDP (and 5.5% of GDP per capita) on social insurance (Appendix , Table A.1 and  Table A.5). Both East Asia and Central and West Asia have the highest spending of 1.6% of GDP on social assistance, but Central and West Asia retains the highest share of spending on each beneficiary of social assistance-1.5% of GDP per capita.

Social Insurance
In 10 countries, social insurance expenditure was higher than the Asia average of 4.2% of aggregate GDP (Appendix , Table A.2). By income group, social insurance expenditure was above the regional average, at 10.0% of aggregate GDP in the HICs and 4.5% in the UMICs, and below the regional average at 3.0% of aggregate GDP in the LMICs. This trend remains for expenditure per intended beneficiary, at 6.6%, 3.9%, and 2.2% of GDP per capita for the three groups ( Figure 4).
In the HICs, social insurance is the primary policy model for social protection. Social insurance programs such as pensions and health care play a key role in Social insurance spending is inadequate to support the majority of intended beneficiaries in countries with a large share of workers in informal employment. Over 68% of employment in Asia and the Pacific is informal (ILO 2018). 7 The existing social insurance schemes have partial coverage and mostly support a small subsection of the population. These are mainly formally employed or salaried workers who hold public sector jobs or work for particular industries or sizable private firms. As the next chapter demonstrates, 15 countries covered less than a third of intended beneficiaries.
The key challenge to extending social insurance to workers in the informal economy is posed by the difficulty to raise individual contributions toward social insurance schemes. Many informal workers have too low and/or unstable income to pay contributions; they may also be reluctant to pay due to perceived cost of application and lack of incentives (Yeates 2009, ILO 2013, Handayani 2016.
Other reasons for the lack of social insurance coverage include non-declaration of jobs, casual and temporary jobs, and jobs with hours of work or wages that fall below a required threshold.
Regionally, social insurance expenditure in East Asia (9.8% of aggregate GDP) was the highest (Figure 4; Appendix, A similar pattern is seen for expenditure for each intended beneficiary as a share of GDP per capita. East Asia had the highest spending (5.5%), followed by Central and West Asia (4.2%), Southeast Asia (2.0%), and South Asia (1.6%) (Figure 4).

Pensions
The dominance of social insurance spending over the other two categories is largely driven by pensions-defined in this report to include contributory benefits for persons aged 60 years and above. They average 2.8% of aggregate GDP. 8 Ten countries spent above the regional average, including Japan (10.8% of aggregate GDP), Uzbekistan (8.3%), the Kyrgyz Republic (7.3%), Azerbaijan (5.4%), and Mongolia (5.1%) (Appendix , Table A.2).
Several countries improved financing for social protection for older persons between 2000 and 2015, including Maldives, Mongolia, the PRC, Thailand, and Uzbekistan. They extended their contributory pension schemes and set up noncontributory social pensions to cover most of the older population, and several achieved universal pension coverage by 2017 (ILO 2017). To extend pension coverage, these countries strengthened policy frameworks underlying pension systems and developed financing arrangements that combine contributory and noncontributory mechanisms.
By income group, HICs had the highest share of expenditure (4.3% of aggregate GDP) on pensions, largely driven by Japan. Spending in the UMICs (at 2.7% of GDP) and LMICs (at 2.6% of GDP) followed a similar trend ( Figure 5). By region, East Asia spent highest, at 5.3% of aggregate GDP on pensions. Central and West Asia spent 5.0% of GDP. Southeast Asia's spending was at 1.1% of GDP, and South Asia's at 1.0% ( Figure 5). This initiative seeks to unify and standardize the existing basic pensions to ensure fair benefits for non-employed urban residents and all rural residents. The residents' pension consists of a solidarity component that is entirely financed by the government and an individual pension component financed by contributions of the beneficiary as well as some government subsidies (ILO 2015b).
In practice, the residents' pension is noncontributory for most of the current generation of pensioners. They only receive the solidarity component as they had already exceeded the pensionable age when the scheme was introduced. A further reform in 2015 removed the contribution exemption from civil servants and government employees, making their pension entitlement like that of urban workers, that is, comprising 8.0% individual and 20.0% employer contributions.

Health insurance
Health insurance refers to contributory programs that enable access to health care. The analysis of health insurance in this study also covers health insurance programs that provide tax-funded noncontributory benefits. The overall average spending on health insurance across the 25 countries in Asia is far lower than that on pensions, at 0.9% of GDP. 9 Only a handful of countries spent above the regional average, including Japan ( Box 2: health insurance in indonesia, Maldives, and Thailand A major social security reform in Indonesia consolidated the existing social insurance programs and established a foundation for a universal healthcare provision. It set up two schemes: BPJS Health, for health insurance; and BPJS Employment, focusing on social security for employees. These schemes integrate previously fragmented social insurance schemes managed by different agencies. BPJS Health began operations on 1 January 2014, BPJS Employment on 1 July 2015. Between 2012 and 2015, social insurance expenditure per intended beneficiary as a share of GDP per capita increased by more than three times, mainly because of expansion in BPJS Health.
In 2012, Maldives established universal health coverage through the Aasandha program. Aasandha covers the cost of medicines and inpatient services incurred in public health institutions. The entire population of the country is eligible for the scheme without the need to pay premium contributions. In 2014, the government expanded Aasandha by establishing Husnuvaa Aasandha, removing the existing annual cap, broadening the list of eligible expenses, and introducing subsidies for transportation costs in an emergency.
The Universal Coverage Scheme in Thailand, which was established with the adoption of the National Heath Act of 2002, covered around 48 million people in 2015, providing the majority of the population with free access to health-care services. It is tax financed and provides a comprehensive benefit package, including outpatient, inpatient, and emergency services (ILO 2016a). Insurance is given to any citizen with a 13-digit government identification number. The scheme covers the population not covered by other health schemes, including the Social Security Scheme for private sector employees and the Civil Servant Medical Benefit Scheme (ILO 2016a). It covers workers in informal employment, who were previously without access to health insurance. However, it does not reach stateless persons and migrant workers who do not have Thai citizenship.
The highest spending on health insurance as a share of aggregate GDP (3.8%) was observed in the HICs. Expenditure was below 1.0% in the UMICs (0.9%) and LMICs (0.2%). Health insurance spending in East Asia was the highest (3.4% of aggregate GDP), dominated by high-income Japan and the Republic of Korea, where health insurance is a key social protection instrument. Southeast Asia and South Asia had similar spending of 0.5% and 0.4% of aggregate GDP ( Figure 5).
Health insurance spending should be viewed in conjunction with health assistance, which covers health-care needs among population groups, such as the poor or those with specific chronic illnesses. But, as the next section shows, health assistance is less developed than health insurance.

Social Assistance
Expenditure on social assistance is similar in the HICs and UMICs at 1.5% of GDP. LMICs generally spend 0.9% of GDP, less than the regional average. The expenditure on each intended beneficiary in social assistance exhibits similar patterns across all income groups ( Figure 6). As for the regional pattern, the share of social assistance expenditure was the same for both East Asia and Central and West Asia at 1.6% of aggregate GDP ( Figure 6), above the 1.1% average for Asia. For each intended beneficiary, only the Central and West Asia region with its 1.3% of GDP per capita spending allocated more than the 0.9% of GDP per capita average for Asia.

Welfare assistance
Welfare assistance includes conditional and unconditional cash and food/in-kind transfers. Each of the social assistance programs accounted for less than 1.0% of GDP, and welfare assistance was the biggest contributor to social assistance expenditure at 0.5% of GDP (Appendix, Table A.2).
Expenditure on welfare assistance in Armenia, Azerbaijan, Georgia, and the Kyrgyz Republic were the highest in Asia. Welfare assistance programs in these countries have national coverage and play a prominent role in providing income support. Similar programs in other parts of Asia do not cover the entire population within potentially beneficiary groups, targeting only smaller population subsections or distinct geographic areas.
In recent years, many countries have made efforts to expand coverage and benefit levels of welfare assistance, exemplified in Box 3.
Spending on welfare assistance was the highest in the HICs (0.6% of GDP), where Japan and Singapore spent 0.8% of aggregate GDP. They are followed by the UMICs (0.5%) and LMICs (0.4%). Among the regions, Central and West Asia had the highest spending (0.9% of GDP). East Asia was at the level of the regional average (0.5%), and both Southeast Asia and South Asia spent 0.2% (Figure 7).

Box 3: expanding Social assistance Programs in asia
The Minimum Living Standards Guarantee program (dibao) in the People's Republic of China dominates the country's social assistance spending. It has grown into the main national income guarantee program since it was set up as a small urban program in 1993. The recipient population in urban areas climbed from 11 million in 2001 to 15 million in 2016, and rural beneficiaries increased from 3 million to 45 million over the same period (OECD 2017). Between 2010 and 2016, the average monthly assistance threshold across counties rose by more than 90% in urban areas and 150% in rural areas.
The conditional cash transfer program Indonesia Family Hope Program (PKH) has been progressively expanding its coverage. The PKH provides cash support to poor and vulnerable families to support their immediate needs; it also promotes access of their children to basic health, nutrition, and education services.

Child welfare
Child welfare refers to assistance targeted solely at children, such as school feeding programs, child allowances, and child grants. It is distinguished from welfare assistance programs that target children as part of a broader objective to improve household well-being. Spending on child welfare was, as a share of aggregate GDP, relatively high in Mongolia (1.1%; see Box 4), Uzbekistan (0.8%), and the Kyrgyz Republic (0.7%), against a regional average of 0.2% (Appendix ,  Table A.2). Spending was the same in both the HICs and LMICs at 0.3% of aggregate GDP (Figure 7). The UMICs spent 0.1% of GDP, below the regional average. East Asia spent 0.6% of aggregate GDP on child welfare (Appendix, By income group, the highest spending of 0.5% of aggregate GDP was in the UMICs ( Figure 7). This group includes Maldives (2.3%) and Thailand (0.5%), which spent above the regional average (Appendix, Table A.2). The HICs spent 0.2% and the LMICs, 0.1%. By region, South Asia had the highest spending of 0.5% of aggregate GDP, followed by East Asia (0.3%) and Southeast Asia (0.1%) (Figure 7; Appendix, Table A.2).

Box 5: Social Pensions in asia
Noncontributory social pensions in Asia are financed from government revenues and seek to provide support persons who are mostly poor and cannot make social insurance contributions. They vary considerably in their eligibility conditions and pensionable age threshold. The social pension in Viet Nam supports older people who have reached the minimum age of 80 and who do not receive contributory pensions, and people aged between 60 and 79 who are poor and living alone or living with ill old spouse and do not have children or relatives to support them. In 2015, it covered 17.3% of older people aged 60 and above, providing a maximum of D270,000 ($12) per person.
In Bangladesh, the Old-Age Allowance is means-tested. It is given to men aged 65 years and older and women aged 62 years and older who have annual income below Tk3,000 ($36). It covered 22.3% of older persons aged 60 and above in 2015, providing them with Tk500 ($6) a month (www.pensionwatch.net).
The social pension in the Philippines was introduced 2011 to provide ₱500 per month ($10) to indigent senior citizens. Initially, due to its limited budget, it covered indigent senior citizens aged 77 and above. The government lowered the age qualification to 65 in 2015.

Health assistance
Health assistance includes tax-funded benefits that enable individuals to utilize health services (as well as free medicines and sanatorium services in some countries). 10 The average share of expenditure on health assistance across the region is negligible at 0.1% of aggregate GDP (Appendix, Table A.2). Eight countries were equal or above the regional average. The highest expenditure as a share of aggregate GDP was in Georgia (0.5%); and in Azerbaijan, the Republic of Korea, and Viet Nam (all 0.4%).
Most of these countries have introduced health assistance to provide free health care to the poor and vulnerable persons. The UMICs spent 0.2%, and HICs and LMICs 0.1%, of aggregate GDP (Figure 7). Central and West Asia spent 0.2% of aggregate GDP on health assistance, while other regions spent below 0.1%.
The Government of the Lao PDR seeks to expand access to health care and to reduce out-of-pocket health-care expenses. In 2012, it established two programs that cover health-care costs for mothers, pregnant women, and children under age 5. In Cambodia, the Health Equity Funds, established in 2003, expanded coverage by almost 1 million from 2010 to 2015. These funds support the implementation of Cambodia's Second Health Strategic Plan and subsidize basic health for the poor; and finance the direct costs of health services and medications for the poor as well as transport cost and food allowance for patient caretakers during hospitalization.

Disability assistance
Disability assistance includes cash benefits for persons with disabilities (as well as prosthetics and free/discounted fares in several countries). Expenditure on disability assistance is low, with a regional average of 0.1% of aggregate GDP (Appendix , Table A.2), partly because disability in the region is often highly stigmatized and thus invisible, affecting take-up as well as knowledge of true numbers of people with disabilities.
Both HICs and UMICs spent 0.2% of aggregate GDP on disability assistance; LMICs spent a negligible figure (0.04%) (Figure 7). Central and West Asia provided 0.2% of aggregate GDP for disability assistance, but other regions spent less than 0.1% ( Figure 7).

Active Labor Market Programs
ALMPs monitored in this report generally comprise two programs: skills development and training, and cash-for-work or food-for-work programs. (They do not include technical and vocational education and training programs.) They remain the most underfunded across the whole of Asia and averaged at 0.1% of aggregate GDP. Expenditures for all income groups were small relative to social insurance and social assistance programs. The average expenditure on skills development and training (0.04%) exceeded spending on cash-or food-for-work (0.02%) (Appendix Table A.2).
Spending in eight countries was equal or exceeded this average. Countries that exhibited a relatively high level of spending among Asian countries were Singapore and Bangladesh, both spending 0.3% of aggregate GDP on ALMPs.

Box 6: Promoting inclusion of Persons with disabilities in Thailand
The universal noncontributory Disability Allowance in Thailand was established in 2007 to provide income security to people with disabilities. In 2015, 1.67 million people registered for a disability card-some 2.5% of the population. Among those who registered, some This expenditure went entirely on skills development and training in Singapore and cash or food for work in Bangladesh.

Skills development and training
Spending in six countries was equal or above the regional average for skills development and training. These include the highest spending Singapore (0.3%) as well as Bhutan, Indonesia, the PRC, the Republic of Korea, and Viet Nam, each spending 0.1% of GDP. In Singapore, skills development and training programs are key for supporting low-income earners. They consist of various "workfare" schemes including the workfare income supplement, workfare special bonus, and workfare training support, all of which are designed to supplement the earnings of low-wage workers. The workfare income supplement was set up in 2007 as a permanent scheme to supplement the wages and retirement savings of older low-wage workers as well as to encourage them to continue working.

Cash-and food-for-work program
Spending on cash or food for work in nine countries was equal or exceeded the regional average. The highest spending of 0.3% of GDP was in Bangladesh. Other countries in this group include Armenia, Cambodia, Japan, the Kyrgyz Republic, Mongolia, Nepal, the Republic of Korea, and Viet Nam. Bangladesh runs seven major ALMPs. These are public works programs that provide cash or in-kind benefits and training. The most notable is the Rural Employment and Road Maintenance Program that seeks to create employment for rural women. This program is implemented by the Local Government Engineering Department under the Ministry of Local Government, Rural Development, and Cooperatives. The program involves beneficiaries in road maintenance and in tree plantation, and provides them with training in self-employment opportunities. Between 2012 and 2015, the number of beneficiaries surged from 69,000 to 553,000.
This chapter first examines coverage, that is, the share of actual beneficiaries among intended beneficiaries. It thus reflects the number of people receiving benefits in practice. The difference between the intended and actual beneficiaries reveals the social protection gap. The chapter then discusses benefit levels of social protection in Asia, both as a share of GDP per capita and in absolute amounts, in terms of the average benefit received by each actual (not potential) beneficiary.

Coverage
• Social protection in 25 countries in Asia covers 55.1% of intended beneficiaries. Thus, nearly half of intended beneficiaries remain without support. • Social insurance has the widest coverage, at 35.2% of intended beneficiaries. Social assistance reaches only 18.4% of intended beneficiaries, as most programs target only population subgroups or distinct geographic areas. ALMP coverage is very low, at 1.5%.
Social protection coverage averaged 55.1% in Asia in 2015 ( Figure 8; Appendix, Table A.6). Social insurance dominated in Asia, both across country income groups and regions.

Social insurance
Eleven countries were above the 35.2% average for Asia ( Figure 8). These include all three HICs, upper-middle-income Maldives and Thailand, as well as lower-middle-income Philippines and Mongolia. Most countries in the region, however, supported less than a third of their intended beneficiaries, given that social insurance remains underdeveloped and fails to reach most workers in the informal economy.
Coverage and Benefit Size of Social Protection Social insurance covered more than 70% of intended beneficiaries in Japan, the Republic of Korea, and Singapore (Appendix, Table A.6). As discussed earlier, these countries have, historically, relied on social insurance as the main program for social protection and allocate high level of expenditure to finance it. In fact, they have the highest share of social insurance expenditure in Asia, spending 10.0% of aggregate GDP (and 6.6% for each intended beneficiary as a share of GDP per capita). For example, the Republic of Korea provides universal insurance for both health insurance and pensions, covering 85.3% of intended beneficiaries. It also supports a large number of workers with unemployment insurance and universal health insurance. The country's compulsory health insurance dominates social protection expenditure, covering some 97% of the population, while the remaining 3% is covered by the Medical Aid health assistance program.
By 2015, social insurance programs in Indonesia, Maldives, the Philippines, Thailand, and Viet Nam had greatly expanded their coverage, supported by significant spending, usually with explicit mechanisms to reach the poor and vulnerable, who were previously ineligible for social insurance. Box 7 discusses three of these countries.
Social insurance coverage was highest in the HICs (78.3%) (Figure 9; Appendix, Table A.6). Coverage in the UMICs was 42.3%, where Maldives, the PRC, and and Thailand achieved impressive coverage. The LMICs reached a smaller share (24.5%) than the other two groups, but it includes countries with the highest coverage in Asia (Indonesia, Mongolia, the Philippines, and Viet Nam).
Regionally, the highest coverage of social insurance was in East Asia (75.1%), given high coverage in all four countries ( Figure 9; Appendix, Table A.6). Social insurance covered 38.5% in Southeast Asia. This region also includes Cambodia, Malaysia, and Myanmar, which had coverage far below the Asian average. South Asia's coverage was 16.7%, but Maldives reached 71.4%. There was a fair degree of variation in Central and West Asia (16.0%), where the Kyrgyz Republic reached 43.2%, while coverage in the remaining countries ranged from 3.2% in Pakistan to 19.2% in Azerbaijan.

Box 7: extending Social insurance coverage in indonesia, the Philippines, and Viet nam
In Indonesia, social insurance coverage increased from 28.3% in 2009 to 51.2% of intended beneficiaries in 2015, mainly due to the expansion of health insurance under BPJS Health (see Box 2, on page 16). BPJS Health builds on the lessons learned under the Jamkesmas social health insurance to enhance the coverage of the poor and near-poor and reach out to the informal sector workers. In 2014, Jamkesmas, with Askes (covering civil servants) and Jamsostek (covering formal sector employees), was transformed into BPJS Health. The new institutional setup aims to achieve universal coverage by 2019.
The Philippines is expanding its social insurance programs, and coverage increased from 75.0% of intended beneficiaries in 2012 to 84.9% in 2015. In 2014, the government expanded PhilHealth coverage through the PhilHealth Indigent Program and enlisted 14.7 million families identified in the national poverty database Listahanan. The number of beneficiaries has also increased with inclusion of war veterans.
In Viet Nam, the rise in the level of benefits in health insurance and new payments for specific illnesses prompted a rise in expenditure from D29.4 trillion in 2012 to D49.5 trillion in 2015. During this time, health insurance coverage beneficiaries increased by 20%, from 59 million to 71 million, accompanied by an extension of subsidized health insurance to vulnerable groups. A new law on health insurance went into effect in 2015 with the aim of achieving universal health insurance. Participation is compulsory for all citizens, and the government is to fully or partly fund the health insurance contributions of vulnerable groups including veterans, ethnic minorities, children under the age of 6, and those with serious illnesses. (Currency unit: D21,670 = $1.00)

Social assistance
Coverage for social assistance was less prevalent than that for social insurance, with a regional average of 18.4% ( Figure 8). Social assistance has, as said, a residual role in HICs; it remains underfunded in most other countries, providing only partial coverage and benefits of low value. Ten countries were above the regional average (Appendix , Table A.6).
Sri Lanka had the highest coverage for social assistance of 54.7% of intended beneficiaries. Sri Lanka's coverage was well balanced along with welfare assistance, including the Samurdhi subsidy for the poor and child welfare support. Coverage in other countries of South Asia was below the regional average of 20.0% (Appendix , Table A.6).
Cambodia, Indonesia, the Philippines, and Viet Nam covered over a third of all intended beneficiaries. Cambodia was among the few countries with a relatively high level of health assistance coverage. The program includes health vouchers for poor pregnant women and Health Equity Funds that provide benefits packaged to vulnerable groups. Between 2010 and 2015, coverage by Health Equity Funds increased by nearly 1 million to 4.1 million people. Indonesia provided social assistance through a range of programs, including the PKH, scholarships for poor students, subsidized rice for poor families, and health assistance for the poor under BPJS Health. Coverage in the Philippines was largely driven by its Pantawid Pamilya conditional cash transfer program.
By income group, coverage was highest in the LMICs (20.1%) (Figure 9), driven by high coverage in Cambodia, Indonesia, the Philippines, Sri Lanka, and Viet Nam (Appendix , Table A.6). The UMICs reached 16.1% of intended beneficiaries. Azerbaijan and Georgia in this group were above the average for Asia. The result was also due to notable progress in the expansion of coverage of noncontributory pensions in Maldives and Thailand. Social assistance coverage was lowest (13.6%) in the HICs, essentially because of the primacy of social insurance.
By region, low social assistance coverage in East Asia (16.4%) was due to low coverage in high-income Japan (8.3%) and the Republic of Korea (13.0%) ( Figure 9; Appendix, Table A.6). Coverage in South Asia was 20.0% and in Central and West Asia, 16.8%. The Southeast Asia region provided the highest coverage of 24.4% of intended beneficiaries.

Active labor market programs
The overall coverage of ALMPs was a stunted 1.5% across Asia. Singapore had the highest coverage (9.5%) (Appendix , Table A.6). Other countries with relatively high coverage include Thailand (5.0%), the Republic of Korea and Indonesia (both 4.1%), and Bangladesh (3.0%). Thailand covers over 6 million beneficiaries under two programs: Skills Development Training and Public Employment Services.
Among income groups, HICs had the highest coverage, reaching 4.8% ( Figure 9). Coverage was 1.4% in the UMICs and 0.9% in the LMICs. By region, Southeast Asia covered 2.8% of intended beneficiaries ( Figure 9). This group includes countries with the highest coverage in Asia, such as Indonesia, Singapore, and Thailand. East Asia covered 2.0% of intended beneficiaries, with the PRC and the Republic of Korea showing results above the average for this region (Appendix , Table A.6). Coverage in South Asia was 0.8%, with Bangladesh posting coverage of 3.0%. Coverage in Central and West Asia was a mere 0.4% of intended beneficiaries.

Benefit Size 11
• Social protection benefits for each actual beneficiary as a share of GDP per capita among 24 countries in Asia 12 came to 11.2% ($812.73). • Social insurance had by far the largest average benefits at 21.4% of GDP per capita ($1,121.42). • More than half of the 24 Asian countries offered benefits below the regional average by monetary value. • Only a few countries combined generous benefits with adequate coverage.
This section considers social protection expenditure for each actual beneficiary relative to a country's GDP per capita as well as the monetary value of benefits (Appendix,

Social insurance
In social insurance, 10 countries offered benefits above the regional average for each actual beneficiary of 21.4% of GDP per capita ( Figure 10). Only in eight countries were benefits above the regional average on monetary value (Appendix,   There is a trade-off between financing generous benefits and extending coverage. Addressing this trade-off depends on a country's economic conditions, but also on its policy preferences. Our analyses suggest the following variation in the design of social protection policies across the region. First, some countries distributed generous social insurance benefits across most of the beneficiary population. The benefits were particularly substantial per actual beneficiary in high-income Japan ($4,743.10), Singapore ($3,449.79), and the Republic of Korea ($1,306.63). Social insurance coverage in these countries was also high.
Second, many countries provided generous benefits to a small share of the population. For example, despite substantial benefits, coverage was low in Malaysia (6.1%) and Bhutan (1.0%) (Appendix , Table A.6). This is because social insurance mainly provides an instrument for supporting workers in the formal economy.
Third, some countries offered low benefits but achieved high coverage.

Social assistance
Social assistance benefits in nine countries were above the regional average of 5.7% of GDP per capita ( Figure 10). Five countries were above the regional average on the monetary value. Social assistance benefits were highest in the HICs at 8.8% of GDP per capita, followed by the UMICs at 8.0% and LMICs at 4.1% ( Figure 11). This pattern holds when looking at the monetary amounts-$3,083.74 (HICs), $520.44 (UMICs), and $82.37 (LMICs).
Unlike social insurance, no country provided generous social assistance benefits across most of the beneficiary population, though the other three patterns are in evidence. That is, first, some countries provided high benefit amounts but retained low coverage. In Central and West Asia, despite the high average on share of GDP per capita (7.6%) relative to other regional groups, the average monetary amount was low at $222.82. In contrast, East Asia had the least amount in terms of share of GDP (6.6%), but the highest monetary value ($1,671.92). Benefits were 7.5% in South Asia and 2.8% in Southeast Asia.

Active labor market programs
In ALMPs, seven countries were above the average for Asia at 5.4% of GDP per capita ( Figure 10), an average one-fourth of that for social insurance both as a share of GDP per capita and in monetary terms (Appendix , Table A.8).
Spending was highest in the LMICs at 7.8% of GDP per capita ( Figure 11), against 3.1% in HICs and 0.7% in UMICs. 12 In monetary terms, HICs had the highest spending per actual beneficiary ($1,201.50).
Most countries combined low spending with limited coverage. For example, the majority of LMICs spent less than $100, covering less than 1.0%. A notable exception was Singapore, which spent $1,804.51 per actual beneficiary, covering 9.5% of intended beneficiaries.
Several countries had relatively generous benefits but small coverage. In particular, the average benefit provided by ALMPs in Bhutan was equivalent to 25.8% of GDP per capita with a relatively high monetary value of $695.01. Benefits in Pakistan (20.3%), Cambodia (14.9%), and Viet Nam (14.0%) were also above the average for Asia. All these countries, however, covered less than 1.0% of intended beneficiaries. As for regional patterns, South Asia had the highest level of benefits at 9.4% of GDP per capita, followed by Central and West Asia (6.1%), Southeast Asia (4.4%), and East Asia (2.8%) ( Figure 11). Monetarily, East Asia spent the most ($542.85), mainly owing to high benefit amounts in Japan.
This chapter focuses on the distribution of social protection expenditure between nonpoor and poor 13 beneficiaries as well as between male and female beneficiaries as a share of GDP per capita. These distributional results are indicative, however, because disaggregated data was often unavailable from official administrative sources, and national researchers had to rely on informed estimates from government officials, secondary data sources, and program practitioners.

Distribution of Expenditure between the Poor and Nonpoor
• Social protection spending in 25 countries in Asia favored the nonpoor over the poor (intended nonpoor beneficiaries were allocated 3.0% and intended poor beneficiaries 1.0% of GDP per capita). • Social insurance spending on the nonpoor (2.6% of GDP per capita) was far higher than on the poor (0.5%). Nearly all countries spent more on the nonpoor. • Social assistance expenditure was evenly split between the poor and nonpoor (0.4% of GDP per capita). Thirteen countries spent more on the poor. • In ALMPs, the difference in spending on the poor and nonpoor was negligible.
Overall social protection spending in 25 countries in Asia favors the nonpoor ( Figure 12). Expenditure for each intended poor beneficiary amounted to 1.0% of GDP per capita, that for the nonpoor, 3.0%.
The difference in expenditure between the poor and nonpoor is largest in social insurance and is the key determinant of the overall outcome. In all 25 countries- 13 The number of poor refers to the number of people living below the nationally defined poverty line.

Poverty and Gender Dimensions of Social Protection Spending
Georgia aside-social insurance expenditure was higher for the nonpoor (2.6% of GDP per capita) than the poor (0.5%) ( Figure 13).
The poor are concentrated in the informal economy, and the majority are not covered by existing contributory social insurance schemes (Handayani 2016). Poor informal-economy workers have too low and/or unstable income to pay insurance contributions. Even when they may have the capacity to contribute, they may be reluctant to do so because of, for example, the lack of trust in social protection institutions, the perceived mismatch between the type of benefits and their priority needs, the difficulty of application procedures, and the time required (ILO 2013). Entitlement to benefits may also require a long contribution period, discouraging many workers in the informal economy.
Spending in social assistance was more equitable than in social insurance, at 0.4% of GDP per capita for both the poor and nonpoor. Social assistance expenditure on the poor was higher than that on the nonpoor in 13 countries ( Figure 14). Given the primary role of social assistance as a safety net for the poor, one should expect the expenditure on the poor to dominate that of nonpoor. In fact, it favored the nonpoor in eight countries and where it favored the poor, spending on the poor was only slightly greater than that on the nonpoor, ranging from 0.1 to 0.6 percentage points. These distributional effects of social assistance can be explained as follows.
First, several social assistance programs in the region are universal and are aimed at specific categories of beneficiaries regardless of their income status, normally supporting both the poor and nonpoor.
Second, distributional effects of social assistance can be influenced by programs that seek to support not only the poor but also the near-poor, that is, those above the official poverty line who can slide into poverty when faced with economic shocks and adverse events. For example, local governments in the PRC have extended to the near-poor their core social assistance programs, including the dibao program, Medical Financial Assistance, Education Assistance, Housing Assistance, and Temporary Assistance. BPJS Health in Indonesia is another example.
Finally, means-tested programs are prone to inclusion of ineligible (nonpoor) and exclusion of eligible (poor) applicants. Means-tested programs intend to support population groups below a specified income level who are unable to derive sufficient livelihoods from the labor market. Accurately measuring and verifying people's income is challenging, especially in countries with a sizable informal economy. Income from agriculture or occasional labor is difficult to estimate, while other types of income, such as from remittances and self-employment in the informal sector, may not be declared or may be undervalued by applicants. This affects the extent of social assistance reach to its target population. Improvements in targeting can therefore be crucial in enhancing the pro-poor orientation of social assistance.
In ALMPs, the difference in spending on the poor and nonpoor was negligible (albeit with slightly pro-poor results). Singapore and Viet Nam favored the poor by a small margin. A possible contributing factor may be that, unlike public works, skills development programs do not necessarily set income level as an eligibility condition and so do not seek to benefit the poor exclusively.
Across income groups, the overall distribution of social protection expenditure is unequal (Figure 15). For all income groups, the overall social protection expenditure is higher for the nonpoor. All income groups spend more on contributory social insurance: the highest ratio is 1:6 among the LMICs, and the lowest 1:4 in the HICs.
For regions, too, social protection expenditure for the nonpoor is generally higher than for the poor (Figure 15). The social insurance ratio in Southeast Asia is especially pronounced: 1:11. The least pronounced gap is in Central and West Asia, but even that is at least 1:3. As among income groups, across the regions social assistance is generally more equitably distributed than social insurance.
Reforms in social insurance and social assistance are likely to enhance the pro-poor orientation of social protection. In the last decade, several countries have expanded social insurance coverage to workers in informal employment via, for example, a combination of contributory and noncontributory financing mechanisms.
Improvements in social assistance can result from accelerated efforts to improve governance of social assistance as a means of reaching to the poor and vulnerable (Babajanian 2017

Distribution of Expenditure by Gender
• Social protection spending in Asia slightly higher for men than women, at 2.1% of GDP per capita on male and 1.9% on female intended beneficiaries. • Social insurance expenditure favored men in 20 countries and women in only five. • Social assistance expenditure favored women in 11 countries, men in 4, and was equal in 10. • In ALMPs, the difference in spending along the gender dimension was negligible.
This section discusses the distribution of social protection expenditure by female and male intended beneficiaries. Such spending for each intended female beneficiary (1.9% of GDP per capita) was less than that for a male beneficiary (2.1%). Spending favored women in four countries (Armenia, the Kyrgyz Republic, Uzbekistan, and Viet Nam) and was equal for both genders in another four (Cambodia, the Lao PDR, the PRC, and Thailand) ( Figure 16).
Social insurance expenditure for men (1.7%) was higher than that for women (1.4%) (Appendix , Table A.10). Social assistance expenditure was equally distributed between women and men. Similarly, the difference in spending on men and women was negligible in ALMPs.
Social insurance expenditure favored men in 20 countries ( Figure 17). However, five countries-Armenia, the Kyrgyz Republic, the PRC, Tajikistan, and Uzbekistan-had higher social insurance spending for women, due to the Social assistance favored women in 11 countries, showed equal distribution in 10 countries, and favored men in 4 countries ( Figure 18). The more equitable results in social assistance are because several universal and means-tested cash transfer programs prioritize women as their main beneficiaries. Many programs are targeted at children as their main beneficiaries, with mothers designated as the main benefit recipients on their behalf. For example, the per capita social assistance for women in Indonesia was greater than that for men (0.4% vs 0.3%). In some countries, most social assistance programs, such as the Indonesia-PKH, Pakistan-Benazir Income Support Programme, and Philippines-Pantawid Pamilya Program, explicitly target women. In the Lao PDR, social assistance favored women and girls with the introduction in 2012 of free maternal and neonatal health care, and free health care for children under 5 years.
By income group, the social protection expenditure for each male beneficiary was higher than that for a female beneficiary in both HICs and UMICs and was equal for both genders in the LMICs (Figure 19). In social insurance, the expenditure was higher for men than for women in HICs and UMICs, and roughly equal in LMICs. In social assistance, men had a slight advantage in HICs, but women did in LMICs; UMICs had an equal distribution. In all regions apart from Central and West Asia, social protection expenditure on male beneficiaries was higher ( Figure 19). In Central and West Asia, expenditure on social insurance and social assistance was higher for women than men. In social insurance, South Asia appears to be the most unequal region, with a ratio of 1:2.2 in favor of men. In social assistance, the distribution was equal for men and women in both East Asia and South Asia, with Southeast Asia favoring men only marginally.
This chapter reviews progress (or otherwise) from 2009 to 2015 in the following: social protection spending for each intended beneficiary as a share of GDP per capita; in coverage; in benefit size; and in the distribution of such spending among the poor vs the poor and by gender, generally in 24 countries. Myanmar is excluded from this analysis because its SPI compilation started only in 2015.

Social Protection Expenditure
• Social protection expenditure for each intended beneficiary in 24 Asian countries increased from 3.4% of GDP per capita in 2009 to 4.2% in 2015. • Social insurance spending rose from 2.5% to 3.2% over the period.
Spending on social assistance and on ALMPs were unchanged, at 0.9% and 0.1% of GDP per capita. • Many countries expanded coverage and benefits, albeit not necessarily at the same time. Less positively, some countries showed reversals in expenditure, coverage, and benefits in all three categories. • Several countries increased their spending on the poor and on women, but a handful of countries saw the reverse trend.
Social protection expenditure for each intended beneficiary (as a share of GDP per capita) increased from 3.4% to 4.2% between 2009 and 2015. 14 Twentyone countries raised their expenditure, and only three countries decreased it (Figure 20).
14 The rate is 4.0% if Myanmar is included in 2015.

Changes in Social Protection Expenditure between 2009 and 2015
For social insurance, average spending on each intended beneficiary in Asia increased from 2.5% in 2009 to 3.2% in 2015 (Appendix , Table A.11), propelling the increase in overall social protection expenditure. Spending on social assistance and on ALMPs stayed unchanged, at 0.9% and 0.1% of GDP per capita.
Most of the 24 countries lifted their social insurance spending between 2009 and 2015 ( Figure 21). In nine, expenditure went up sharply-gains ranged from 1.4 percentage points to 3.0 percentage points. Still, coverage decreased in Armenia, Tajikistan, and Uzbekistan (Appendix, Table A.12 and Table A.13).
Among HICs, Japan increased its social insurance spending from 9.2% of GDP per capita in 2009 to 11.0% in 2015, increasing both coverage and benefit size (Figure 21; Appendix, Table A.11). In the PRC, spending increased from 3.0% of GDP per capita to 4.1%, due to the extension of pension coverage.
Progress was more modest in 12 countries (between 0.3 and 1.0 percentage points); nevertheless, it was substantial, especially relative to the existing levels of spending in Indonesia, Maldives, and Viet Nam. Expenditure on social insurance doubled in Maldives from 1.5% to 3.3% of GDP per capita, due to Aasandha, the country's universal health insurance program. Social insurance spending in Indonesia rose from 0.4% to 1.4%, as coverage and benefits expanded with the implementation of BPJS Health in 2014. The increase in spending from 3.3% to 3.6% in Viet Nam was built on the extension of entitlements for health insurance.
In social assistance, expenditure in nine countries improved ( Figure 22). These improvements, however, did not result in a discernible change for Asia in aggregate terms as another nine countries decreased their spending.
Azerbaijan and Georgia increased social assistance spending by 0.9 and 0.8 percentage points (Appendix, Table A.11). It was accompanied by increased coverage and benefit levels in Azerbaijan, but the increase in expenditure in Georgia mainly financed an extension in coverage (Appendix, Social insurance increased by 1.2 percentage points in the HICs, by 0.9 percentage points in the UMICs, and by 0.6 percentage points in the LMICs. Social assistance expenditure went up by 0.4 percentage points in the HICs and UMICs but declined slightly in the LMICs. Spending on ALMPs remained unchanged in all three income groups. Apart from East Asia, all regions registered an upward trend in social protection expenditure (Figure 23b). The highest increase of 1.7% was observed in Central and West Asia. Expenditure in South Asia increased by 0.7 and in Southeast Asia, by 0.4. East Asia experienced a decline of 0.2 percentage points, largely attributable to a decline of 4.7 percentage points in Mongolia.
Among the three categories, the most substantial progress was in social insurance expenditure in Central and West Asia (by 1.7 percentage points) (Appendix , Table A.11). Other regions improved social insurance spending too, but by less than 1 percentage point. Some, albeit modest, gain was recorded in social assistance in South Asia (0.2 percentage points) and Southeast Asia, but spending in the other two regions declined. Little change was observed in the ALMPs in all three categories.  Overall, social protection coverage went up, essentially due to social insurance ( Figure 24). The biggest expansion, of 7.4 percentage points, was from 2009 to 2012, followed by a gain of 2.8 percentage points from 2012 to 2015. Sixteen countries expanded their coverage, but eight decreased it (Appendix, In ALMPs, only six countries increased coverage between 2009 and 2015 (Appendix , Table A.12). Programs in 15 countries experienced retrenchment, which was underpinned by a decline in the expenditure on these programs.
Coverage of ALMPs in Singapore increased from 7.7% in 2009 to 9.5% in 2015, driven by the expansion of "workfare" schemes for low-income earners. Indonesia increased coverage from 1.3% to 4.1% by promoting skills development programs.
Other countries, such as Azerbaijan, the Lao PDR, and Tajikistan, improved active labor coverage marginally, by less than 1 percentage point. Looking at the most recent time from 2012 to 2015, Bhutan and the Lao PDR doubled their coverage. In Bhutan, this was largely driven by the expansion of the Guaranteed Employment Program designed to support the unemployed youth. Coverage increased in the Lao PDR with programs for citizenship verification and registration for migrant workers, and skills training for out-of-school youth.
Social protection coverage picked up between 2009 and 2015 in all income groups, owing to the increase in social insurance coverage (Figure 25a). The highest increase in coverage of 15.4 percentage points took place in HICs, mainly driven by the extension of social insurance coverage in the Republic of Korea and Singapore (Appendix , Table A.12). It was followed by 12.4 percentage point increase in UMICs and 8.4 in LMICs. In both income groups, social insurance coverage increased. Social assistance coverage decreased in HICs and LMICs. All three country groups experienced a drop in active labor market coverage. The most notable was a reduction in UMICs, due to sizable declines in Thailand and Malaysia.
The most notable increase in coverage among all three categories occurred in social insurance in HICs (20.7 percentage points). Social insurance coverage also improved in UMICs (by 10.9 percentage points) and in LMICs (by 10.0 percentage points). Social assistance coverage improved in UMICs (by 5.9 percentage points), with substantial progress in Georgia, the PRC, and Thailand. Social assistance coverage, however, declined in the other income groups over this period. ALMP coverage decreased in all the income groups, mostly in the UMICs (by 4.4 percentage points).
Social protection coverage improved in all regions between 2009 and 2015, apart from Central and West Asia (Figure 25b), where social insurance coverage declined by 2.7 percentage points (Appendix, Table A.12). Gains in other regions were largely due to social insurance. The highest increase in social protection coverage was in Southeast Asia (22.1 percentage points), reflecting sharply higher social insurance coverage (25.9 percentage points).
The highest progress among programs was in extending social insurance in Southeast Asia (by 25.9 percentage points). Social insurance coverage also substantially improved in South Asia (by 10.9 percentage points) and in East Asia (by 9.0 percentage points). In social assistance, South Asia had the strongest result (5.8 percentage points), due to expansion in Bangladesh and Sri Lanka. Coverage in East Asia declined, largely due to a substantial drop in Mongolia. Similarly, it decreased in Southeast Asia as four out of its eight countries reduced coverage. Coverage of ALMPs improved in Central and West Asia, but other regions experienced reduction. Social Assistance Active Labor Market Programs remained at 5.7%, but ALMPs showed some progress-from 4.8% of GDP per capita to 5.6% (Appendix , Table A.13).
Between 2009 and 2015, improvements in social insurance benefits were observed in 10 countries. The most significant progress was achieved in Armenia, Pakistan, Tajikistan, and Uzbekistan, where benefits increased by more than 20 percentage points. Benefits also improved in Georgia (by 11 percentage points) and in the Kyrgyz Republic (by 6.8 percentage points). Social insurance benefits declined in 13 countries, falling heavily in the Philippines (by 24.3 percentage points), Maldives (23.2), and Sri Lanka (12.0).
In social assistance, 11 countries improved their benefits over this period. Substantial progress was achieved in Singapore, where benefits increased by 4.7 percentage points. It was followed by Azerbaijan (3.4 percentage points), Japan (2.5), and Thailand (2.3). Despite a modest percentage increase, benefits doubled in the Lao PDR, the Philippines, and Sri Lanka. Twelve countries decreased social assistance benefits: Pakistan by 6.2 percentage points; Bhutan, 2.8; and the PRC and Mongolia, 2.4. As for ALMPs, benefits improved in 9 countries, remained the same in 2 (Georgia and the Lao PDR), and declined in 12.
Benefits increased in all income groups from 2009 to 2015 (Figure 27a). The most significant increase of 4.1 percentage points was in UMICs (Appendix ,  Table A.13). It was followed by an increase of 2.2 percentage points in LMICs and 1.1 in HICs. The improvement in HICs was due to an increase in social assistance (2.4 percentage points) and ALMP benefits (1.6 percentage points), while lower-income countries benefited from improvements in social insurance (3.2 percentage points) and ALMPs (1.9 percentage points).
The most progress was observed in social insurance benefits (3.2 percentage points) in LMICs. Social assistance benefits increased by 2.4 percentage points in HICs, as benefits improved in Japan and Singapore. Social assistance benefits in the other two income groups declined. In ALMPs, benefits in LMICs increased by 1.9 percentage points and in HICs, 1.6 percentage points.
Regional trends exhibit a mixed picture about the change in benefits from 2009 to 2015 ( Figure 27b). Overall, benefits increased by 6.7 percentage points in Central and West Asia, owing to a substantial increase in social insurance benefits in all countries of this region (Appendix , Table A.13). Southeast Asia was another region with a positive trend (1.8 percentage point increase). Benefits remained the same in East Asia but declined by 0.5 percentage points in South Asia as social insurance and social assistance benefits dropped. The rise in social insurance benefits in Central and West Asia by 18.7 percentage points represented the most substantial progress among the three social protection categories. Social insurance benefits in the other regions declined. Social assistance benefits increased in Southeast Asia (by 1.2 percentage points) but declined in the other regions. ALMPs improved benefits in South Asia (by 5.2 percentage points).
A small core of countries improved both benefits and coverage. Among 10 countries improving social insurance benefits, coverage increased in five (Azerbaijan, Indonesia, Japan, the Kyrgyz Republic, and the PRC). In social assistance, 11 countries increased benefits, but only four of them (Azerbaijan, the Philippines, Sri Lanka, and Thailand) also improved coverage. In ALMPs, only Bhutan and Singapore improved both benefits and coverage.
Most countries in Asia prioritized expanding either coverage or benefits. The prevailing trend in both social insurance and social assistance over this period was to expand coverage and reduce benefits. Social insurance benefits dropped in 13 countries 15 but coverage in all these countries increased. Among 12 countries that reduced social assistance benefits, eight improved coverage. 16

Distribution of Expenditure between the Poor and Nonpoor
• Nearly half the 24 countries increased spending on the poor in social insurance and social assistance. • Social insurance continued to favor the nonpoor, but the number of countries favoring the poor in social assistance increased from nine to 13.
Overall spending on social protection in Asia continued to favor the nonpoor over the period (Figure 28). In 2009, all 24 countries favored the nonpoor over the poor, a situation that changed little by 2015: only Georgia favored the poor and Cambodia showed an equal distribution.
Still, spending on the poor increased in 14 countries. In some countries, this improvement was substantial. For example, pro-poor expenditure increased by 1.5 percentage points in Armenia and by 1.7 percentage points in Georgia, Maldives, and Uzbekistan (Appendix , Table A.14).
Social protection spending on the nonpoor increased in 18 countries, decreased in 5, and remained the same in 1 (Appendix , Table A.14). Although the increased expenditure on the nonpoor may reflect inclusion and exclusion errors in meanstested social assistance programs, it is also likely to be the result of the extension in social insurance coverage and establishment of universal entitlements.
Social insurance spending continued to favor the nonpoor in all countries, apart from Georgia. Spending on the poor increased in 14 countries and decreased slightly in 4 (Appendix , Table A.14). Social insurance spending on the nonpoor increased in 17 countries, decreased in 4, and and remained the same in 3. The increase in spending on the poor was due to the extension of social insurance, including pensions in the PRC, BPJS Health in Indonesia, and health insurance in Maldives.
figure 28: distribution of expenditure between the Poor and the nonpoor, 24 countries in asia, 2009, 2012, and 2015 Note: For abbreviations used, refer to this publication's Abbreviations list on page ix. Source: Table A.14 in the Appendix. In social assistance, the number of countries that favored the poor over the nonpoor increased from nine to 13. Countries that reversed their spending pattern on the poor in this category include Cambodia, Indonesia, the Kyrgyz Republic, and Tajikistan. Spending on the poor increased in 11 countries, decreased in 9, and remained the same in 3. Spending on the nonpoor increased in 9 countries, decreased in 10, and remained unchanged in 5. The increase in spending in social assistance was less substantial than in social insurance and did not exceed 0.4 percentage points. The drop in spending on the poor ranged from 0.1 percentage points in the PRC to 0.8 percentage points in Mongolia.
Expenditure on the poor increased in all income groups from 2009 to 2015, by 0.4 in HICs and UMICs, and by 0.2 in LMICs ( Figure 29a). The most significant was an improvement by 1.7 percentage points in the UMICs Maldives and Georgia, largely due to increased pro-poor spending in social insurance (Appendix ,  Table A.14). The largest increase in spending on the nonpoor was observed in HICs (1.2), followed by the UMICs (0.8) and LMICs (0.1).
Among the three categories, the most significant improvement in spending on the poor was recorded in social insurance in the UMICs (0.5 percentage points), particularly due to progress in Maldives and Georgia. Social insurance spending in the other two income groups also increased, by 0.3 percentage points in both. Spending on the poor improved only marginally, below 1 percentage point in both social assistance and ALMPs in all income groups. Spending on the nonpoor especially increased in social insurance in HICs (by 0.9), followed by the other two income groups (by 0.4 in each). All three income groups increased social assistance spending on the nonpoor, by 0.4 percentage points in HICs and UMICs, and by 0.2 in LMICs, while changes in the ALMPs were marginal.
Social protection spending on the poor increased in all regions, apart from East Asia (Figure 29b). The most significant gain for the poor (0.7 percentage points) was registered in Central and West Asia, thanks to progress in social insurance (Appendix , Table A.14). Similarly, spending on the nonpoor increased in the regions, apart from East Asia. Again, Central and West Asia had the highest increase of 1 percentage point, due to increasing spending in social insurance.
The most substantial spending on the poor (0.7 percentage points) was observed in social insurance in Central and West Asia. Spending on the poor in social assistance improved marginally in South Asia (0.2 percentage points) and Southeast Asia (0.1 percentage points) but declined in the other two regions.
figure 29: nonpoor vs Poor Social Protection expenditure by income Group and region, 24 countries in asia, 2009Group and region, 24 countries in asia, , 2012Group and region, 24 countries in asia, , and 2015 Note: For abbreviations used, refer to this publication's Abbreviations list on page ix. Source: Table A.14 in the Appendix.

Distribution of Expenditure by Gender
• Social protection spending on women for each intended beneficiary as a share of GDP per capita in 24 countries in Asia was still behind that on men in 2015. • Nineteen countries increased spending on women and five reversed their spending to favor women over men.
The average social protection spending on women for Asia was still behind that on men in 2015, but the gap had narrowed (Figure 30), with spending on women climbing from 1.5% of GDP per capita in 2009 to 2.0% in 2015. Spending on men also increased, but only by 0.2 percentage points.
Progress toward greater gender-sensitive expenditure was observed at country level. All countries in the study spent more on men than women in 2009; this situation improved in 2015 as spending came to favor women in four countries (Armenia, the Kyrgyz Republic, Tajikistan, and Uzbekistan) and was equally distributed between men and women in another four (Cambodia, the Lao PDR, In social insurance, spending on men continued to prevail over that on women in Asia. However, the gap between average social insurance spending on women and men as a share of GDP per capita narrowed from 0.5 percentage points to 0.2 percentage points (Appendix, Table A.15). Sixteen countries increased spending on women during this period. Especially notable was progress in the Kyrgyz Republic and Uzbekistan, where average expenditure grew by more than 2.0 percentage points. Six countries (Bhutan, Mongolia, Pakistan, the Philippines, Sri Lanka, and Thailand) slightly decreased their spending on women.
Social assistance spending on women in 2015 prevailed in 11 countries, and it was equally distributed in 10 (Appendix , Table A.15). This represents a substantial progress from 2009, when only five countries favored women. The average social assistance spending on men and women as a share of GDP per capita did not change over this period and was equally distributed at 0.4% of GDP per capita. Social assistance spending on women increased in 9 countries, declined slightly in 5, and remained the same in 10. The increase was marginal, ranging from 0.2 percentage points in Cambodia to 0.5 percentage points in Georgia.
Spending on women increased in all income groups from 2009 to 2015 ( Figure 31a). The most significant was the increase of 0.7 percentage points in UMICs, followed by 0.6 in HICs and 0.4 in LMICs. These results were largely driven by gains in social insurance expenditure in all three income groups. Spending on men increased by 1.1 percentage points in HICs and by 0.5 percentage points in UMICs, but declined slightly in LMICs.
The most significant was the rise in expenditure on women in social insurance, by 0.5 percentage points in both HICs and UMICs and by 0.4 percentage points in LMICs (Appendix , Table A.15). In social assistance, spending on women increased by 0.2 percentage points in UMICs and by 0.1 percentage points in HICs, but declined marginally (by 0.1 percentage points) in LMICs. The highest rise in the expenditure on men was in social insurance in HICs (0.7 percentage points). Spending on men increased in social insurance in the other two groups, but not substantially. Social assistance spending on men somewhat increased in HICs (by 0.3 percentage points) and in UMICs (by 0.2 percentage points), but decreased slightly in LMICs. Spending on women showed progress in all four regions (Figure 31b). The sharpest was in Central and West Asia (1.1 percentage points), due to improvements in social insurance in most countries there (Appendix , Table A.15). Spending on men increased by 0.4 percentage points in South Asia and by 0.2 percentage points in Central and West Asia, mainly owing to an increase in social insurance spending.
The most substantial increase on spending on women (1.2 percentage points) was in social insurance in Central and West Asia. Social insurance in all the other regional groups also increased but at a slower rate. In social assistance, spending on women increased marginally, by a mere 0.1 percentage points, in both South Asia and Southeast Asia, with a lack of progress in Central and West Asia and East Asia. Social insurance spending on men increased in all regional groups, apart from East Asia. Social assistance spending on men increased slightly in South Asia, with other regions exhibiting a slight decline. In ALMPs, Southeast Asia increased spending on women by 0.1 percentage points, with very small changes in spending on men and women in other regions.
Social protection expenditure is increasing, but more is needed The analysis of data on overall national social protection expenditure suggests that mean social protection expenditure in 2015 was 5.3% of GDP and average expenditure per intended beneficiary amounted to 4.0% of GDP per capita. There are positive trends across the region, as many countries have expanded their public expenditure and made real efforts to strengthen social protection systems. This resulted in the expansion of coverage and, in some cases, benefits in different areas of social protection, including pensions, health insurance, and welfare assistance.
Despite these gains, there is an urgent need to increase public expenditure in social protection to develop comprehensive and sustainable social protection systems. This is especially pertinent for UMICs and LMICs. Greater investment would help them close existing coverage gaps to achieve the social protection target of the United Nations Sustainable Development Goals by 2030. 17 Resources are required to support expansion in coverage and benefits.

The extent of spending depends on both a country's income and its policy agenda
High levels of national income can enable greater social protection spending, as is the case in richer countries. Countries with high national income also spent above the regional average in per capita terms. HICs were well ahead of other income groups, spending an average of 11.6% of aggregate GDP, or 7.9% of GDP per capita for each intended beneficiary.
Among regional groups, East Asia had the highest expenditure (11.5% of aggregate GDP, or 6.4% of GDP per capita). Correspondingly, countries with high national income tended to offer most generous social protection benefits.
But country income does not explain the full extent or pattern of social protection spending, which reflects national policy priorities. Social protection expenditure was especially high in countries with social protection systems stipulating broad-based national social protection entitlements. In particular, all HICs spent significant resources on social insurance, which seeks to provide nationwide oldage and health protection.
These findings suggest that even countries with lower income can enhance spending on social protection, provided they set policy goals and develop a policy framework to ensure that all people can access and benefit from social protection.

A trade-off is apparent between widening coverage and offering generous benefits to all intended beneficiaries
Only a few countries have managed to distribute, for example, generous social insurance benefits across most of the intended beneficiaries, notably all three HICs (Japan, the Republic of Korea, and Singapore), with social insurance coverage of more than double the Asian average of 35.2%.
Many countries have expanded social protection coverage but still offer small benefits. Some LMICs with the highest social insurance coverage-Indonesia (51.2%), Viet Nam ( Many other countries provided generous benefits to a small share of their population, usually via social insurance, which serves mainly as an instrument for supporting workers in the formal economy. For example, social insurance benefits in Malaysia reached $6,728.14 and that in Bhutan, $1,171.82; however, coverage was low (6.1% in Malaysia and 1.0% in Bhutan). In social assistance, the actual monetary benefits were notably high in Japan ($4,125.85), Singapore ($2,865.98), and the Republic of Korea ($2,259.38). However, social assistance coverage in these countries was limited due to the prevalence of social insurance as the primary social protection program. In ALMPs, the relatively high benefits in HICs were accompanied by low coverage-4.8% of intended beneficiariesmainly because governments regard these programs as residual to social insurance.
Several countries combined small benefits and low coverage. Social insurance benefits in Cambodia and Myanmar were below $100, covering 9.0% and 1.4% of intended beneficiaries; Nepal had low benefits of $333.31 and coverage of 2.7%. Social assistance benefits in the Lao PDR, Myanmar, and Nepal were below $100 and coverage was low. In ALMPs, most countries combined small benefits with paltry coverage.

Social insurance is expanding and dominates the other two categories
Social insurance dominated expenditure in Asia across all income groups and regions, at four-fifths of social protection spending (4.2% of 5.3% as a share of aggregate GDP). It also had the widest coverage among the three social protection categories, reaching a third of intended beneficiaries. It was also largest in terms of benefits, both as a share of GDP per capita (21.4% of GDP) and its monetary value ($1,121.42). The dominance of social insurance spending was largely driven by contributory pensions.
Social insurance expenditure was higher than or near the Asian average (4.2% of GDP) in the three HICs, and in upper-middle-income PRC and Malaysia. Social insurance is the primary social protection program in HICs, supporting more than 78% of intended beneficiaries. HICs have the highest share of social insurance expenditure in Asia, spending 10.0% of aggregate GDP (and 6.6% of GDP per capita).
Between 2009 and 2015, social insurance spending climbed in all but three countries (Georgia, Mongolia, and Thailand). Average social insurance spending increased from 2.5% to 3.2%, and coverage rose by 11.6 percentage points, from 25.1% to 36.6%. Seventeen countries expanded social insurance coverage over this period. Between 2009 and 2015, social insurance benefits increased by 1.3 percentage points, from 20.8% to 22.1%. Social insurance benefits rose in 10 countries but declined in 13.
This dominance of social insurance, however, does not mean that all countries have adequate coverage for their intended beneficiaries. A third of intended beneficiaries in Asia are not reached by social insurance. Social insurance in large parts of Asia remains limited and mostly supports a small subsection of the population employed in the formal economy.

Old-age pension is expanding
Spending on contributory pensions (2.8% of aggregate GDP) accounted for the largest share of social insurance spending of 4.2% of aggregate GDP. It was high in HICs (4.3% of aggregate GDP), largely due to high expenditure in Japan (10.8% of aggregate GDP). Regionally, East Asia spent the highest (5.3% of aggregate GDP), with Mongolia (5.1%) and the PRC (3.3%) at the fore. It is followed by Central and West Asia (5% of GDP). These countries have undertaken reforms to strengthen and expand contributory programs. Notably, the PRC has also extended pension entitlements by promoting noncontributory or social pensions.
The extension of pension entitlements through social assistance has been especially prominent in Southeast Asia and South Asia, which historically had less inclusive contributory systems. The highest spending on assistance to the elderly was in the UMICs (0.5% of aggregate GDP), with especially high spending on universal Old-Age Basic Pension and Senior Citizen Allowance in Maldives (2.3%) and Old-Age Pension in Thailand (0.5%). Between 2012 and 2015, governments in many LMICs (including Bangladesh, Nepal, the Philippines, and Viet Nam) increased funding and expanded their noncontributory social pensions.

Social protection in health remains inadequate
Average spending on health insurance across Asia (0.9% of aggregate GDP) remained low. Only a handful of countries spent above this average, including high-income Japan (8.3% of GDP) and Republic of Korea (2.8%), as well as upper-middle-income Maldives (2.7%).
Between 2009 and 2015, a number of countries in Asia have supported institutional reforms and increased social expenditure to expand health insurance coverage. In fact, Maldives, the PRC, and Thailand have achieved universal pension coverage.
The average share of expenditure on health assistance in Asia was negligible at 0.1% of GDP. 18 Spending on health assistance was highest in the UMICs (0.3% of GDP). Central and West Asia had the largest spending (0.2% of GDP), with Armenia, Azerbaijan, and Georgia dominating the group.

Social assistance is improving, but modestly
Social assistance covered only 18.4% of intended beneficiaries, leaving the majority of the poor and vulnerable without support. From 2009 to 2015, social assistance spending increased in nine countries. Spending volume remained nearly unchanged in aggregate terms. From 2009 to 2015, social assistance coverage increased from 18.4% to 19.1%. Social assistance benefits remained unchanged at 5.7% from 2009 to 2015; 11 countries improved their benefit size and 12 countries decreased it.
The widest coverage was achieved in the UMICs. This group includes Azerbaijan, Georgia, and Thailand which, with lower-middle-income Armenia, covered over a quarter of intended beneficiaries. The LMICs in Southeast Asia had some of the highest social assistance coverage in Asia. This is largely due to the expansion in social assistance schemes in Cambodia, Indonesia, the Philippines, and Viet Nam, covering over a third of intended beneficiaries. Other countries that expanded social assistance coverage include Maldives, Nepal, and Sri Lanka. The low coverage in the three HICs is due to their use of social insurance as the main social protection category.
Welfare assistance is the main program for social assistance Welfare assistance was the biggest contributor to social assistance expenditure in Asia. By income group the highest spending was in HICs (0.6% of GDP) and by region in Central and West Asia (0.9% of GDP). There was a significant expansion of welfare assistance programs in Indonesia, the Philippines, and the PRC.
Further expansion of welfare assistance can help address persistent poverty and improve other socioeconomic outcomes in Asia. Evidence suggests that welfare assistance and more specifically conditional and unconditional cash transfers have an important role in supporting the poor and vulnerable.

Expenditure on child welfare and on disability assistance is negligible
Spending on child welfare programs in Asia was at a low 0.3% of GDP. Spending was highest in Japan (0.5% of GDP) and the Republic of Korea (0.4% of GDP) in the HIC group, and in Uzbekistan (0.8% of GDP) and the Kyrgyz Republic (0.7% of GDP) in the LMICs. By region, Central and West Asia had the highest spending on child welfare in Asia (0.6% of GDP). Many welfare assistance programs in the region seek address child poverty, including basic needs, nutrition, access to health and education. Still, it is important to enhance funding and coverage of programs to tackle other life cycle and social vulnerabilities, e.g., through birth grants, integrated early childhood development services, and child protection measures.
Expenditure on disability assistance was also limited, which is compounded by the fact that disability in the region is often invisible and highly stigmatized. Both HICs and UMICs spent more (0.2% of GDP) on disability assistance than LMICs. Central and West Asia provided the highest share of funding (0.2% of GDP).

Active labor market programs are underdeveloped
By income group, HICs and LMICs spent more (0.1% of aggregate GDP) than UMICs. By region, East Asia, Southeast Asia, and South Asia spent 0.1% of GDP each. At country level, the highest level of expenditure (0.3% of GDP) was in Bangladesh and Singapore. The overall coverage of ALMPs was very low, averaging 1.5% across Asia.
Yet ALMPs can play a crucial role in improving the existing skills supply and promoting inclusion in the labor market. The international experience shows that proactive skills development and training support can adjust to the rapidly growing skills requirements in the labor market. It can help reduce reliance on vulnerable jobs and enhance the quality of jobs for young people (ILO 2015a). Skills development should be made a central pillar in national planning as well as expansion in technical and vocational education and training, and improving the quality and relevance of general-track education. It is important that these programs specifically target women to reduce their dependence on vulnerable forms of work.

Progress in pro-poor spending is too slow and patchy
Social protection spending in Asia still favors the nonpoor over the poor. The discrepancy was highest in social insurance, where spending on the nonpoor for each intended beneficiary as a share of GDP per capita was five times as high as that on the poor. Only Georgia moved to a pro-poor spending pattern in social insurance between 2009 and 2015.
The results were more equitable in social assistance, where aggregate spending was evenly split between the poor and nonpoor. Social assistance expenditure on the poor prevailed over that on the nonpoor in 13 countries and was equal in four.
The number of countries that favored the poor in social assistance increased from nine to 13 between 2009 and 2015.
Fourteen countries in the study increased expenditure on the poor in social insurance-a majority-but only 11 did in social assistance and 5 in ALMPs. Both the other countries' slow progress in enhancing pro-poor expenditure and insufficient coverage of the poor should flag a key policy priority: extending coverage to support the poor who are not covered by current arrangements. This is challenging, especially in social insurance where it is hard to raise contributions from workers in the informal economy.
Social protection is becoming more gender sensitive, but needs to be more inclusive Social protection expenditure in the region prioritized men over women: 2.1% of GDP per capita on men vs 1.9% on women. Social insurance expenditure favored men in 20 countries, because the majority of women are engaged in the informal economy and are less likely to afford social insurance contributions than men. Social assistance favored women in 11 countries, showed equal results in 10, and favored men in 4.
Progress toward more gender-sensitive expenditure was seen at country level. All countries in 2015 spent more on men than women in 2009, and spending came to favor women in four countries (Armenia, the Kyrgyz Republic, Uzbekistan, and Viet Nam) and became equal in another four (Cambodia, the Lao PDR, the PRC, and Thailand). Five countries came to spend more on women in social insurance, six in social assistance (bringing the total to 11), and two in ALMPs. Sixteen countries increased spending on women in social insurance and nine countries in social assistance, but two countries (Bhutan and Mongolia) decreased spending on women in both categories. The design of social protection therefore needs to incorporate mechanisms that address gender-specific barriers to access social protection.            Weighted breadth equals to actual subprogram beneficiaries divided by the total intended beneficiaries of social protection.
Source: ADB estimates based on Social Protection Indicator country reports, 2017.