COVID-19 and Social Protection in Asia and the Pacific: Projected Costs for 2020–2030

Against the backdrop of the COVID-19 pandemic, this working paper applies a new costing model to analyze the costs associated with social protection in 30 Asia and Pacific countries. Using the Social Protection Reform Simulation (SPRS20) model, the authors seek to estimate the cost of delivering standardized social protection packages through the emergency (2020), recovery/transition (2021–2023), and the remaining duration for achieving the Sustainable Development Goals (2024–2030). This paper anticipates further demand and need for social protection for considerably large sections of the population. It aims to foster further research and discussion on social protection in Asia and the Pacific and aid countries in strategizing for the future.


eXeCUtIVe sUMMARY
The widespread social and economic strain imposed by the coronavirus disease (COVID-19) pandemic continues to affect countries around the world. Countries in Asia and the Pacific have been particularly hit hard and the effects of the pandemic stand to undo progress made in alleviating poverty and improving well-being in the region. Social protection systems play a critical role in mitigating risks and supporting vulnerable groups through these unprecedented times. The COVID-19 pandemic triggered a social protection response globally of previously unseen magnitude, but countries still struggle to provide adequate support to their citizens as the pandemic continues to unfold. COVID-19 has further exposed gaps and challenges in the social protection landscape across countries, providing an important opportunity for policy makers to re-evaluate investments and strategies.
Data indicate that even prior to the onset of the COVID-19 crisis, the global community was lagging to meet its policy commitments in line with International Labour Organization (ILO) Recommendation No. 202 on social protection floors; and targets 1.3 and 3.8 on social protection and universal health coverage under the United Nations Sustainable Development Goals (SDGs). Prior to COVID-19, only 45.0% of the global population was reportedly covered by at least one social protection benefit, and only 29.0% was covered by comprehensive social security systems addressing life-cycle vulnerabilities. Latest available estimates under the Social Protection Indicator of the Asian Development Bank (ADB) suggests that effective coverage was extended to just 55.1% and 31.2%, respectively, of intended beneficiaries in Asia and the Pacific prior to the onset of the pandemic. These indicators imply that a considerable portion of the population, globally and regionally, were left without adequate access to social protection when COVID-19 hit. ADB's Social Protection Indicator data also estimates an average social protection spending (pre-pandemic) amounting to 5.3% of gross domestic product (GDP) for Asian countries and 6.0% for Pacific countries.
Against the backdrop of the COVID-19 pandemic, this report applies a new costing model-the Social Protection Reform Simulation (SPRS20)-to analyze the costs associated with delivering standardized social protection packages through the emergency (2020), recovery/transition (2021)(2022)(2023), and the remaining duration of the SDG targets (2024)(2025)(2026)(2027)(2028)(2029)(2030) for 30 Asia and Pacific countries. For the emergency and recovery/transition phases, these packages have been designed to replicate regional trends in social protection response to COVID-19 and thus represent additional strain to the system in the wake of the pandemic. Through the emergency phase, such costs are assumed to be fully financed by the government. In the recovery/transition phase, it is assumed that government contribution will gradually taper off, to be replaced by contribution financing by 2024. The SPRS20 tool and the results presented in this report do not aim to provide the cost of the actual measures introduced, but to estimate the cost of an adequate response for particularly vulnerable population groups, at different stages of the pandemic. It seeks to add to the resource base available to Asia and Pacific countries to facilitate further planning and expansion of national social protection systems.
The emergency package is estimated to cost approximately 3.0% of GDP on average across the 30 countries studied, accounting for almost 15.0% of government revenues in 2020. Costs in this phase are estimated to be driven by the emergency wage subsidy offered to vulnerably employed individuals (1.0% of GDP), followed by the health insurance contribution waiver for uninsured individuals (0.6% of GDP) and cash transfers for poor households (0.5% of GDP). Except for cash transfers, these measures are new to most of the countries and therefore the costs of the emergency package are predominantly topping up existing expenditure on social protection. The package is estimated to be particularly expensive in low-income countries (6.6% of GDP), where per capita GDP is close to the national poverty line, driving high benefit costs accompanied by large sections of population qualifying for emergency benefits. By subregion, the package is estimated to be most expensive among countries in Central and West Asia (3.9% of GDP), primarily due to large variations in costs among the countries included in this subregion, including low-income countries considered in this study.
The recovery/transition package (2021)(2022)(2023) aims to provide continuous recovery support for the most vulnerable individuals and households and facilitate activation in the labor market, while slowly restoring the role of contribution financing over time. The transition from the emergency to the recovery/transition package is thus accompanied by several changes, including the (i) elimination of the rather expensive wage subsidy for the vulnerably employed and unemployment assistance for informal sector workers offered through the emergency phase; (ii) extension of benefit durations up to 1 year for many programs, from 6 months in the emergency phase; (iii) gradual tapering of government contributions especially for social insurance programs; and (iv) introduction of activation measures like skills training and public works and/or employment guarantee. The package is estimated to cost approximately 3.4% of GDP in 2021 and gradually decrease to 3.2% in 2023. By 2023, government contributions are estimated to taper down to 84.0% of total costs with the remaining 16.0% financed through contributions. Costs in 2023 are expected to be driven by cash transfers for poor households (1.0% of GDP), followed by the health insurance contribution waiver for the uninsured (0.8% of GDP) and food assistance (0.3% of GDP). Social insurance measures and the additional activation measures introduced in this phase account for a cumulative total of 0.6% and 0.3% of GDP, respectively, in 2023. The recovery/transition package continues to be particularly expensive in low-income countries (7.9% of GDP in 2023) and countries in Central and West Asia (4.4% of GDP in 2023).
Women are estimated to benefit the most from health insurance benefits (both formal sector health insurance waiver and waiver for the uninsured) and sickness benefits offered through the emergency and recovery/transition phases, accounting for 40%-50% of the total beneficiaries for these programs in the region. In comparison, labor market programs like the wage subsidy offered in the emergency phase and the activation measures introduced in the recovery/transition phase are estimated to account for just 30%-35% of female beneficiaries over these phases. These estimates indicate that targeting of emergency measures based on employment status or industry may not be favorable for women considering current employment trends. Benefits are estimated to be equally distributed among urban and rural beneficiaries through the emergency and recovery/transition phases. All emergency measures, except for poverty-targeted household social assistance, are estimated to account for a mere 10%-16% of poor among total beneficiaries for such programs in the region-indicating that the poor may benefit only marginally from emergency measures not specifically targeted at them.
From 2024 to 2030, SPRS20 models a much more comprehensive social protection package that aims to deliver the least adequate social protection floors in line with ILO Recommendation 202 and social protection-related SDG targets in the 30 countries. The package is thus designed to provide access to essential health care, including maternity care; basic income security for children-providing access to nutrition, education, and other necessary goods and services; income security for persons of active age but unable to earn sufficient income due to sickness, unemployment, maternity, or disability; and basic income security for the elderly. Estimates drawn in this phase, therefore, represent total costs to the system, assuming the package covers the sum of all social protection programs in a country. Programs modeled in this phase are assumed to gradually mature over time and achieve full coverage by 2030.
x Executive Summary The social protection floors package is estimated to cost approximately 4.1% of GDP on average across the 30 countries in 2024 and increase over time to 5.9% in 2030. In 2030, the package is estimated to be particularly expensive in several countries like the Kyrgyz Republic (10.7%), Timor-Leste (10.6%), and most of the Pacific countries. On average, child benefits (1.2% of GDP) are estimated to be the most expensive program at the end of the projection period in 2030, followed by the universal old-age assistance (0.8%). By subregion, the package is estimated to be particularly expensive among the Pacific countries, at an average cost of 6.5% of GDP in 2024 and reaching 9.0% in 2030, consistently above the regional average through this phase. This is estimated to be driven by significantly high costs incurred in these countries for child benefits, old-age, and disability (both formal sector insurance and universal assistance) benefits. In comparison, the package is estimated to be least expensive in South Asian countries throughout the social protection floors phase, registering an average cost of 3.0% of GDP in 2030. The low costs in South Asia may be explained by faster GDP growth rates in countries in the region compared to the cost of the package.
Close to 50% of all beneficiaries in the social protection floors phase (except for household-level programs like cash transfers and food assistance) are estimated to be women and girls. Social insurance programs in this phase also demonstrate a favorable gender impact. Women are estimated to be particularly favored in old-age benefits, both formal sector pensions and tax-financed universal assistance. On the other hand, projected employment trends indicate that active labor market program (ALMP) measures disproportionately favor men over women, with just over 30% female beneficiaries. These estimates provide hints on programs where gender dimensions need to be particularly emphasized in order to ensure equitable distribution of benefits to a country's population. Benefits are once again estimated to be equally distributed among urban and rural beneficiaries through the social protection floors phase. However, most social assistance measures (except for poverty-targeted household assistance) and ALMP measures in the social protection floors phase are estimated to benefit the nonpoor more than the poor, with the poor accounting for less than 18% of total beneficiaries for such programs in the region.
The current version of the SPRS20 and the results discussed in this report have several limitations and adopt a range of assumptions to help estimate overall costs. For example, programs modeled by SPRS20 assume "perfect" targeting, with no inclusion or exclusion errors. While this is an oversimplification of reality, this assumption helps in producing cost estimates given all intended beneficiaries (as per user-specified parameters) receive benefits. Further, SPRS20 adopts a program-based modeling approach and as a result, considerable overlap is expected between beneficiary groups of different programs. For this reason, beneficiary disaggregation and coverage are discussed in this report at the program level across individual countries, regions, and subregions. Other program-specific assumptions are discussed in detail under different subsections of the report. The current version of SPRS20 also does not account for dynamic poverty modeling, and instead uses a standard cut-off (based on regional program examples as reference) applied universally across countries for poverty-targeted programs. These limitations will be addressed in upcoming versions of the model.
It is tempting to compare the cost estimates discussed in this report with current or pre-COVID-19 social protection spending; however, this is not a straightforward exercise. The social protection floors package modeled here, for example, is an ideal package that assumes full coverage without errors in targeting. In current spending, on the other hand, these errors of inclusion and exclusion are imminent and whereas large numbers of people will receive less than they should, some will receive more. These inefficiencies are not included in the simulations undertaken in this report. Therefore, comparing these cost estimates with current expenditure should be done with great caution.
Following the spike in public spending in 2020 due to the emergency measures introduced in response to the COVID-19 pandemic, due to the subsequent drop in tax revenues, and given the essential nature of social protection provisions, there will be a need for increased domestic resource mobilization going forward. The incremental social protection financing needs estimated in this report also makes it necessary for countries to explore options for making the required fiscal space available for social protection. Research further indicates that countries transition into a higher growth path and, in part, higher social spending once tax revenues reach 15.0% of GDP. Countries in Asia and the Pacific have been gradually improving their tax-to-GDP ratio in the past 2 decades, with many reporting tax ratios well above the 15.0% threshold. This demonstrates high potential for increased social spending in the region. Similarly, other options to increase available fiscal space and overall social spending in countries have to be explored in depth.
The affordability of social protection is also closely linked to the expected economic growth in countries over the coming years. As highlighted in the results discussed in this report, the cost of delivering adequate social protection coverage is estimated to be particularly high in several countries, including low-income countries where resource availability is often limited with a heavy dependency on donor financing. Global mechanisms to sustainably support such requirements, like the recent call for the establishment of a Global Social Protection Fund as put forth by the Global Coalition for Social Protection Floors, also warrant critical consideration in the future. For such considerations and to foster further planning, SPRS20 provides a user-friendly platform to help model potential policy packages and estimate associated costs to adequately expand social protection coverage in Asia and the Pacific.

IntRoDUCtIon
As 2021 comes to a close, the COVID-19 pandemic continues to spread at varying speeds across the globe. As of July 2021, the World Health Organization (WHO) reports over 188 million confirmed cases of COVID-19 globally with over 4 million deaths (WHO 2021). In addition to the catastrophic health implications, the pandemic continues to impose unprecedented social and economic turmoil in countries around the world. An estimated 94% of the world's workforce was reported to be living in countries with workplace closure measures in place (ILO 2020). Although the more stringent lockdown and closure measures have now widely been limited to the most severely affected areas and sectors in many countries, the impact of these measures continues to affect the global economy and the well-being of individuals. As the pandemic continues to spread, many countries have slowed down reopening measures, with some reinstating partial or full lockdowns.
In its October 2020 update, the International Monetary Fund (IMF) projected global growth at -4.4% for 2020, almost 0.8 percentage point above the June 2020 projections with implications that the global economy was slowly climbing out of the depths it had plunged to in April. The economic recovery, which seems to have started in the third quarter of 2020, was expected to gradually strengthen over the coming year with global growth projected at 5.2% in 2021. In comparison with 2019, the expected increase in global gross domestic product (GDP) over 2020-2021 was projected to be a mere 0.6 percentage point (IMF 2020a). A more recent update from the IMF (April 2021) pegs global growth at -3.3% for 2020 and estimates an improved recovery rate of 6% for 2021 (IMF 2021).
Recovery, however, is projected to be gradual, uneven, and for the most part uncertain, particularly in emerging markets and developing economies where cases continue to rise, resulting in further loss of output. Like most regions around the world, Asia and the Pacific was also heavily hit by the social and economic implications of COVID-19. The IMF's Regional Economic Outlook for Asia and the Pacific 1 (IMF 2020b) estimated a -1.7% growth rate in emerging markets and developing economies in Asia for 2020, while among the Pacific island countries and small states, it was estimated at -7.5%. Recovery in the region, however, is anticipated at a much faster rate in comparison to the rest of the world. Economic growth rate in 2021 was projected at 8.0% for emerging markets and developing economies in the region, and 4.2% for Pacific islands and small states.
Governments around the world face an array of challenges: having to contain the public health crisis while adequately responding to the economic and social impact of the pandemic; and at the same time facilitating sustainable economic and social recovery. Given the current crisis, the role of social protection systems is more significant than before and form an indispensable part of the coordinated policy response in countries, ensuring effective access to health care while supporting jobs and income security for those most affected. 1 Based on last available update dated October 2020, as of July 2021.
Social protection systems represent a critical pillar of the overall policy response, increasing resilience in individuals and helping to address poverty and unemployment while enabling economic and social stability needed for recovery (ILO 2020a). Further, the economic implications of a timely social protection response through a crisis have been well established in literature. Studies indicate that every $1 invested in building people's resilience in countries can result in savings of up to $3 in humanitarian aid (SPACE 2020). Evidence emerging from across the world also indicates that countries with effective health and social protection systems were better prepared to respond to the COVID-19 crisis, but still struggle due to the unprecedented scale of the pandemic.
In addition, low-income households are expected to be most affected by the adverse effects of the pandemic on the global economy, thus striking a severe blow to the progress made in reducing extreme poverty since the 1990s (IMF 2020c). Globally, an estimated 90 million people were expected to fall into extreme deprivation in 2020 (IMF 2020a). Asia and the Pacific is home to a disproportionately large share of those considered particularly vulnerable to the negative consequences of the COVID-19 pandemic. In many parts of the region, large sections of the population still lack access to basic infrastructure and services (UNESCAP 2019).
Countries around the world, including those in Asia and the Pacific, initiated numerous social protection measures in response to the pandemic. The nature and scope of responses covered several functions of social protection with unemployment protection, income protection, and special allowances accounting for more than 50% of measures introduced. Of such measures, a majority were noncontributory, a significant portion of which accounted for by new programs or benefits. On the other hand, contributory measures introduced mainly took the shape of spending adjustments for existing schemes and programs (ILO 2020b). In most countries, measures introduced in response to the pandemic were time bound and linked to the duration of lockdown and containment measures.
However, available data indicate that even prior to the onset of the COVID-19 crisis, the global community was lagging to meet its policy commitments in line with ILO Recommendation No. 202 on social protection floors; and targets 1.3 and 3.8 on social protection and universal health coverage under the Sustainable Development Goals (SDGs) (ILO 2020c). Prior to COVID-19, only 45% of the global population was covered by at least one social protection benefit and a mere 29% was covered by comprehensive social security systems addressing life cycle vulnerabilities. In Asia and the Pacific, this figure was even lower with just 38.9% of the population covered by at least one social protection benefit (ILO 2017). These large and persistent gaps in the global social protection ecosystem have been linked to significant financing gaps in countries, further worsened by the immediate needs imposed by the COVID-19 pandemic and the associated erosion of government resources, diminishing tax and social insurance revenues (ILO 2020c). Given this scenario, the pandemic also presents an important opportunity for policy makers to re-evaluate social protection systems in their respective countries and consider possibilities of sustaining the current momentum, at least partially in years to come.
In the wake of such an unprecedented global scenario, this report presents the results of a detailed costing exercise undertaken for 30 countries in Asia and the Pacific, to estimate the cost of delivering a set of standardized social protection packages through the emergency phase (2020), recovery/transition phase (2021)(2022)(2023), and the remaining duration of the SDG targets (2024)(2025)(2026)(2027)(2028)(2029)(2030). The report uses the Social Protection Reform Simulation (SPRS20) model, which is a newly developed, customizable social protection costing tool designed to support social protection planning and policy making for the region through and beyond the aftermaths of the COVID-19 pandemic. This report and the SPRS20 model further aim to simulate research and discussions on social protection in Asia and the Pacific countries, while moving toward the targets of the SDG agenda. COVID-19 and Social Protection in Asia and the Pacific The forthcoming sections of the report are structured as follows: Section 2 discusses the broad implications of the COVID-19 pandemic for Asia and the Pacific. Section 3 introduces the SPRS20 model, its methodological approach, scope of the current costing exercise, and details of various simulated social protection packages. Section 4 presents the results and findings of the costing exercise, followed by concluding remarks in section 5.

CoVID-19 CRIsIs AnD Its IMPLICAtIons FoR AsIA AnD tHe PACIFIC
The COVID-19 pandemic and the associated health, social, and economic turmoil is reported to have dire consequences for Asia and the Pacific. As of July 2021, Asia and the Pacific accounted for nearly 40 million confirmed cases of COVID-19, the third-highest number of confirmed cases by region, next to the Americas and Europe, respectively. Among Asia and Pacific countries, India has the highest number of cases with over 31 million confirmed cases, followed by Indonesia with over 2.7 million cases and the Philippines with over 1.4 million cases (WHO 2021).
Like many parts of the world, Asia and the Pacific is also on a slow and gradual path to economic and social recovery after the initial outbreak in early-to mid-2020. Overall, economic activity in the region was expected to contract by almost -2.2% in 2020 and to further grow by 6.9% in 2021-almost 0.3 percentage point higher compared with previous projections. Emerging markets and developing economies in the region exhibit slightly better-off trends, with a projected economic contraction of -1.7% in 2020, followed by an 8% growth in 2021. The Pacific islands and other small states exhibit slightly different trends, with a projected economic contraction of -7.5% in 2020, followed by a 4.2% recovery rate in 2021 (IMF 2020b). 2 Several countries in the region had eased their lockdown and containment measures on account of suppressing the initial outbreak. Patterns of easing measures vary across countries depending on the stage of the pandemic. Some countries, however, reopened long before infection rates fell and subsequently experienced a further increase in cases. Many countries are adopting a sequential reopening strategy, opening lower-risk regions or sectors initially and reimposing localized lockdowns and restrictive measures as needed to contain any new clusters and further spread (IMF 2020b).
As mentioned, the demographic composition of Asia and the Pacific reflects large concentrations of population considered particularly vulnerable to the social and economic implications of the COVID-19 pandemic. This includes those already marginalized by structural barriers, and gender and wealth inequalities even before the onset of the pandemic; among them, the elderly, women and girls, children, persons with disabilities, caregivers, and migrant and informal sector workers (UNDRR 2020).
Asia and Pacific countries are home to a rapidly aging population with projections indicating that one in four individuals in the region will be 60 years or older by 2050 (UNESCAP 2017). The elderly and persons with underlying health conditions become particularly susceptible under the current scenario due to 2 Based on IMF's Regional Economic Outlook for Asia and the Pacific, last update October 2020.
weakened immune systems. Similarly, lack of access to necessary medication and health care facilities can critically endanger pregnant women and persons with ongoing illnesses. The region is also home to over 690 million persons with disabilities who have more immediate health care needs. Restrictions imposed by the pandemic limit access to services, particularly quality health care, which makes it more difficult for such individuals (UNDRR 2020).
The labor market in Asia and the Pacific is characterized by high levels of informality, with informal workers accounting for nearly 60% of all non-farm employment in the region. Moreover, informal workers are reported to be twice as likely as formal workers to belong to poor households, disproportionally bearing the brunt of the social and economic impact of the pandemic (IMF 2020d). Sixty-five percent of working women in Asia and the Pacific are employed in the informal sector and depend on daily wages for survival. Along with other informal sector workers, they represent a huge share of the population who are least likely to be able to practice preventive measures such as physical distancing and self-isolation (UNDRR 2020).
Furthermore, women account for over two-thirds of the global workforce in the health and social sector, placing them at the forefront of the pandemic response and thus vulnerable to both direct and indirect risks. In Southeast Asia and the Western Pacific, women account for over 75% of nurses in medical facilities. The spread of COVID-19 has also exacerbated the situation of unpaid domestic and care workers in the region, with women accounting for majority of such work. Gender-based violence has also been on the rise since the onset of the pandemic. Lockdowns and quarantine measures are forcing women and girls to be confined with their abusers in many countries. Countries like India, Indonesia, and Singapore reported staggering increases in women's helpline traffic since the onset of the pandemic (UN Women 2020).
Children represent yet another demographic group that is disproportionality affected by the socioeconomic implications of COVID-19. Prior to COVID-19, an estimated 47 million children younger than 5 years old were moderately or severely wasted. Majority of them were reported to be residing in Sub-Saharan Africa and South Asia. Estimates indicate that an additional 6.7 million children under the age of 5 could have fallen victim to wasting in 2020 alone in the wake of the pandemic, with over half of them residing in South Asia (UNICEF 2020a).
The COVID-19 pandemic has also disproportionately impacted youth employment in Asia and the Pacific. Young people in the region were already facing numerous constraints in their access to decent work, with youth unemployment rates reaching nearly 13.8% in 2019, compared with 3% among adults. At the onset of the crisis, close to 50% of the working youth in the region were employed in the most critically hit sectors: wholesale, retail trade and repair, manufacturing, rental and business services, and accommodation and food services, with an overrepresentation of women in three of the four abovementioned sectors (ILO and ADB 2020).
Asia and the Pacific is also home to over 60% of the global urban population and around 65% of the global slum population. With many of the major cities in the region reporting high congestion, measures like physical distancing are often extremely difficult to practice. More than half the region's population live in rural areas, most of whom are still engaged in agriculture, with considerably less access to adequate sanitation, health services, education, internet and communication technology, social protection, and public infrastructure (FAO 2020). COVID-19 and Social Protection in Asia and the Pacific In addition, COVID-19 has delivered a severe blow to poverty alleviation in developing Asia.
In the absence of COVID-19, poverty incidence in the region was estimated to decline in line with developments over the past 2 decades. This would have meant close to 114 million poor individuals as defined using the $1.90 poverty line and 734 million as defined using the $3.20 poverty line at the end of 2020. However, growth forecasts undertaken by the Asian Development Bank (ADB) and other stakeholders indicate that the number of poor in the region is likely to increase to 192 million using the $1.90 poverty line and 896 million using the $3.20 poverty line. These estimates point to either an additional 78 million or 162 million poor in the region, depending on the poverty line used-reversing poverty alleviation achieved over the past 3 to 4 years (ADB 2020a).
With such inherent vulnerabilities affecting large sections of the population in Asia and the Pacific, social protection plays a critical role in the region's policy response to address existing and new vulnerabilities for individuals. Based on pre-pandemic ADB Social Protection Indicator aggregations, average social protection expenditure as a share of GDP in Asian countries was estimated at 5.3%, while for Pacific countries it was estimated at 6%, with effective coverage extended to 55.1% and 31.2% of intended beneficiaries, respectively, for Asian and Pacific countries (ADB 2019a(ADB , 2019b.
Meaning that at the onset of the pandemic, significantly large sections of the region's population were left with limited or no social protection coverage.

tHe soCIAL PRoteCtIon ReFoRM sIMULAtIon MoDeL
Social Protection Reform Simulation (SPRS20) is a comprehensive social protection costing tool developed in the wake of the COVID-19 pandemic to help countries estimate the additional costs associated with delivering adequate social protection coverage. The model further allows users to estimate the cost of delivering a customizable social protection floor, in line with social protection-related SDG targets up to 2030, while factoring for the economic and social recovery anticipated from the COVID-19 shock. The SPRS20 model provides users with a list of social protection programs that can be included in a country's social protection package and further allows for a range of program-level customizations, including the level of benefits, benefit adjustment mechanisms, benefit durations, program maturity periods, target populations, and other criteria. The costing estimates produced by the model can assist policy makers in making informed decisions on social protection programming in their respective countries.

methodological overview and modeling Assumptions
The SPRS20 model derives its data from publicly available data sources. The following data sources have been used: • UN (https://population.un.org/wpp/Download/Standard/Interpolated/).
The source data feeds into three modules. The first is a population and labor force characteristics module. This provides the number of eligible beneficiaries for the various social protection programs. The module includes historic trends and projections. The population and labor force projections are derived from UN and ILOSTAT. For the employment and unemployment projections, a satellite module is used. This satellite module simulates an unemployment rate projection path, starting from its actual value in the base year gradually converging to a realistic target rate toward the end of the projection period, and then calculates the corresponding employment-to-population ratio through the projection period.
The second main module is an economic module which contains historic trends for GDP, inflation, wages, interest rates, etc. This module also contains projections for these variables which are based on the following three "anchoring" assumptions: • GDP per capita growth 2015-2019 is extrapolated linearly to 2020-2030.
• 2015-2018 labor productivity growth is extrapolated linearly to 2020-2030 and it is assumed that this will drive growth in real wages.
The other economic indicators follow (mathematically) from these three anchors.
The economic module feeds into the labor market satellite module. First, a counterfactual employment growth scenario is calculated supposing that no COVID-19 shock had happened. Then the fall in employment (and rise in unemployment) in 2020 and 2021 is calculated using the impact of the crisis on GDP growth for 2020 and 2021 and an employment/GDP growth elasticity, derived from the historic (2010-2019) trends. COVID-19 increases unemployment rates in the short to medium term, but the assumption is that unemployment rates in the long term will resume back to normal.
The third main module is a public finance (government finance) module. In the current version of the SPRS20 model, this is still a work in progress. The module contains historic data but not projections. This will be addressed in forthcoming versions of the model.
The three modules feed into the social protection program module which is the heart of the SPRS20 model. This module contains more than 20 different options for social protection programs: contributory, noncontributory, cash, in-kind, health insurance, and active labor market programs (ALMPs). The model further contains a dashboard 3 which allows users to change parameter settings for the various programs and view consolidated country results. This includes options to select poverty-line thresholds (national/ international), various phase in and/or program maturity periods, benefit amounts, benefit durations and benefit adjustment mechanisms (earnings or inflation), eligibility conditions, and administrative costs involved. Other factors, including formality rate in countries, are derived from respective national 3 The SPRS20 dashboard is a work in progress. It enables users to view consolidated results by country, subregion, and program, and will be made available with upcoming versions of the report. statistics, with provisions to allow for further manual calibration of these factors, enabling users to model different labor market scenarios as necessary. Finally, the model contains a tables and graphs module which shows the outputs from the modeling exercise. Figure 1 provides an overview of the model architecture. Cost estimates are underpinned by the overarching assumption that programs are perfectly targeted with no inclusion or exclusion errors. While this is an oversimplification of reality, the assumption helps in producing cost estimates given all intended beneficiaries (as per user-specified parameters) receive benefits. Further, SPRS20 adopts a program-based modeling approach and, as a result, considerable overlap is expected between beneficiary groups of different programs. While this limits the model's capacity to effectively estimate overall social protection beneficiary size for countries, it does allow for effective estimates at a program level locally and across the region. Other program-level assumptions will be discussed in forthcoming subsections. The current version of SPRS20 also does not account for dynamic poverty modeling, and instead used a standard cut-off (based on regional program examples as reference) applied universally across countries for poverty-targeted programs. This limitation will be addressed further in upcoming versions of the model.

Scope of Current Costing exercise
For the costing exercise undertaken here, SPRS20 adheres to ADB's long-term containment scenario (ADB 2020b) reflecting lockdown and containment measures for a period of 6 months in countries. The model has been further calibrated to reflect three distinct phases in a country's social protection ecosystem considering the COVID-19 pandemic: the 2020 COVID-19 emergency phase, the 2021-2023 recovery/transition phase, and the remaining duration of the SDG targets-the social protection floors phase (2024)(2025)(2026)(2027)(2028)(2029)(2030). Each of these phases is underpinned by a customized social protection package consisting of a combination of program choices classified further under social insurance, social assistance, and ALMPs, corresponding to the objectives of the respective phases.
For the emergency (2020) and recovery/transition (2021-2023) phases, SPRS20 has been calibrated to produce estimates of additional costs to the social protection systems induced by COVID-19. The social protection packages and the associated costing estimates produced for these phases are therefore not a representation of the cumulative social protection systems and costs in these countries, rather an estimate of the additional burden induced by the COVID-19 pandemic in the respective years.
Globally, the COVID-19 pandemic triggered a social protection response of unforeseen magnitude, with countries investing heavily to provide adequate health, income, and livelihood support for their citizens. The nature and choice of programs however vary from country to country. To keep the model's program and parametric choices close to real-time measures introduced in Asia and the Pacific, a broad scoping exercise was undertaken to plot trends in such response. 4 The analysis reveals details of program and parametric choices across the region including benefit amounts, duration of emergency programs, and various eligibility criteria adopted. While data availability on the public domain for these factors have severe limitations, the analysis does help grasp a broader trend of the social protection response in the region. Observations from this analysis have thus been used as a reference to design and cost a comparable set of measures in the emergency and recovery/transition phases for the 30 countries included in this report.
On the other hand, for the period 2024-2030, SPRS20 models a comprehensive social protection package envisioned as a sum of all measures in the country, designed to include at the least measures as envisioned under the Social Protection Floors Recommendation No. 202 of the ILO (ILO 2012). The package also takes into consideration that in the wake of the vulnerabilities exposed by the COVID-19 pandemic, countries may choose to maintain a range/version of programs/benefit extensions/coverage extensions through the emergency and recovery phases. Costing estimates produced for this period thus represent the estimated total social protection costs in these countries.

Social Protection Packages Simulated by SPrS20
The 2020 CoViD-19 emergency Package The COVID-19 emergency package is modeled as a short-term crisis response taking hints from actual program choices and trends observed in the social protection response to COVID-19 among countries in Asia and the Pacific in the year 2020. These measures are assumed to be in addition to existing social protection programs in countries. Table 1 provides an overview of the 2020 COVID-19 emergency package, and its associated program and parametric choices.
Regional trends in social insurance response to COVID-19 indicate three predominant forms of interventions: contribution waivers, sickness benefits, and unemployment benefits, accounting for close to 33% of social insurance measures introduced in the region. In light of these trends, the 2020 COVID-19 emergency package models the following social insurance measures targeted at formal sector workers: (i) a health insurance contribution waiver for a period of 6 months at a benefit level modeled using Indonesia's fee waiver program as a regional benchmark, but with country-specific data; COVID-19 and Social Protection in Asia and the Pacific (ii) a sickness benefit for 3 months at 60% of average earnings; and (iii) an unemployment insurance benefit for 6 months at 60% of average earnings. For the health insurance contribution waiver, benefit levels have been modeled based on the average cost of delivering a health package per capita as a cost-recovering contribution. Administrative costs for social insurance programs have been set at 5% of total program cost, assuming that countries can capitalize on existing social insurance delivery mechanisms to administer benefits.
Observed trends in social assistance response include programs offering health and sickness assistance, unemployment assistance, cash transfers, and food assistance. Such measures were primarily targeted at individuals and households belonging to vulnerable categories such as the poor and informal sector workers. In light of these trends, social assistance measures under the 2020 COVID-19 emergency package have been modeled to include (i) a health insurance contribution waiver for uninsured population offered for a period of 6 months at a benefit level modeled using Indonesia's fee waiver program as a regional benchmark, but with country-specific data; (ii) a set of COVID-19-related sickness assistance (paid sick leave, for 3 months; and reimbursements for medical treatment) offered to both formal and informal sector workers testing positive for the virus, at benefit levels of 60% of average earnings for paid sick leaves and a per capita lump-sum benefit calculated using WHO country data for medical treatments; (iii) an unemployment assistance benefit for informal sector workers for a period of 6 months at a benefit level of 100% of the national poverty line; and (iv) a cash transfer and a food assistance program for poor and vulnerable households most susceptible to the social and economic shocks of COVID-19, both offered for a period of 6 months. 5 The household-level benefits are modeled using Indonesia's cash transfer and food assistance program (Rastra) as regional benchmarks. Benefit levels for the household cash transfers are set relative to national individual poverty lines and household sizes, such that a 4-to 5-person household receives exactly the amount of the individual poverty line, with larger households receiving more and smaller households receiving less. The value of the food assistance program has been determined in a similar manner (varying with household size) at approximately one-third of the cash benefit level. Administrative costs have been set at 10% of total program costs for all social assistance programs except for COVID-19-related reimbursements for medical care and the household food assistance, which, owing to foreseeable logistical challenges, have been allocated a higher administrative cost of 20% of overall program cost.
Observed regional trends indicate that wage subsidies account for the largest share in the number of ALMPs introduced in Asia and the Pacific countries as a response to the pandemic. Other programs, including skills development, training, and public works programs, have only been introduced in a few countries, where they account for a significantly smaller portion of the ALMP expenditure. Mirroring this trend, the 2020 COVID-19 emergency package models a wage subsidy as the predominant labor market program to provide immediate relief to employers and employees in the formal sector in industries most severely hit by the pandemic (Notes in Table 1 lists heavily affected sectors). Benefit levels have been set at 20% of average earnings for a duration of 6 months through the lockdown and containment periods at an administrative cost of 5% of total program costs. 5 Global data on the duration of emergency social assistance measures, particularly cash transfers, indicate an average duration of 3.3 months as of September 2020. However, considering the extended nature of the crisis, many countries, including those in Asia and the Pacific like Azerbaijan, Bhutan, Indonesia, and Uzbekistan further extended the duration of measures (Gentilini et al. 2020), with several others announcing intentions to do so. In lieu of these observations, social assistance measures in the emergency package have been modeled for a duration of 6 months in 2020, anticipating such extensions in more countries through the critical period in 2020, and in line with ADB's long-term (6 months) containment scenario adopted for this report. 2. It is assumed that children are up to the age of 17 and the elderly start at 60.
3. It is assumed that in countries where insurance for these contingencies already exists, benefits will be channeled through the existing social insurance administration at lower administration costs.
4. The heavily affected industry sectors considered for the wage subsidy are mining, manufacturing, transport, wholesale and retail, accommodation, and food services. a Modeled after the Indonesian fee waiver, but with country-specific data inputs.
b Modeled using Indonesia's cash transfer program as a reference. Cash transfer levels are set relative to the poverty line and household size, with larger households receiving higher benefit levels.
c Modeled using Indonesia's Rastra program as a reference. The value of food assistance is again determined relative to household size.
Source: Authors' elaboration based on program and parametric choices adopted for current costing exercise. COVID-19 and Social Protection in Asia and the Pacific The emergency package is administered by SPRS20 for 2020 alone. As reflected in the previous section, the social protection programs modeled under this package primarily take the shape of emergency measures mirroring observed regional trends in social protection response. From 2021, in line with the social and economic recovery projected for the region, the social protection packages modeled by SPRS20 will adopt slightly modified program/parametric choices as will be discussed in forthcoming sections.

The 2021-2023 recovery/Transition Package
The period 2021-2023 is assumed to facilitate a gradual recovery scenario for countries in the region. The social protection package modeled by SPRS20 for this phase continues to represent costs in addition to that of existing national social protection programs, but under the assumption that programs modeled in this period, specifically social insurance programs for the formal sector, will be accompanied by a phase-out plan to account for a gradual transition from government financing to contribution financing. Table 2 provides an overview of measures under the recovery/transition package.  Notes: Social insurance programs commencing in 2021 will be accompanied by a phase-out plan to account for the gradual transition from government financing to contribution financing. Government contributions will thus reduce to 75% in 2021, 25% in 2022 and finally 0% in 2023-transitioning to a fully contribution-financed model from 2024. a Modeled after the Indonesian fee waiver, but with country-specific data inputs.
Source: Authors' elaboration based on program and parametric choices adopted for current costing exercise.
Program choices for social insurance remain without change in the recovery/transition package, providing a health insurance contribution waiver, sickness benefit, and unemployment insurance for formal sector workers. Benefit levels remain the same as in the emergency package for all three programs and they are assumed to only gradually mature over this phase. Benefit durations are reduced to 3 months a year for both the sickness benefit and unemployment insurance. The waiver for the formal sector health insurance scheme, on the other hand, is offered for 12 months a year in this phase, with tapering government contributions over time, and is phased out by 2024. Administrative costs for social insurance programs are maintained at 5% of total program expenditure.
With regard to social assistance measures, the health insurance contribution waiver offered to the uninsured population is maintained in the recovery/transition phase; however, the program is tapered down to cover 62% of informal sector workers and their families by 2024 and is accompanied by the assumption that 25% of the health cost is paid by the individual as out-of-pocket expenses after direct government subsidies are subtracted from per capita health expenditure. Reimbursements for COVID-19-related medical treatment are also maintained in this phase. However, the paid sick leave offered to formal and informal sector workers, and the emergency unemployment assistance offered to informal sector workers in the emergency phase, are dropped in the recovery/transition phase. The model also includes a rather strong assumption that there will be a gradual yet steady increase in the formality rates in countries over the years as a consequence of demonstrated benefits of formal sector social protection coverage and government efforts to foster and encourage such participation. Cash transfer and food assistance continue to remain in this phase at the same benefit levels as in the emergency phase. However, both these measures are extended for up to 12 months a year in this phase, aimed at providing consistent recovery support for vulnerable households. Administrative costs for the respective social assistance measures are again maintained without change from the emergency phase.
Transitioning from the emergency phase to the recovery phase, the choice of ALMPs encounters the biggest change. While wage subsidies formed the predominant labor market program in the emergency phase, during the recovery phase, priority is placed on activation measures aimed at stimulating employment and employability of the vulnerable labor force in the country. Through the recovery/ transition phase, this is facilitated by a skills training and intermediation program for urban unemployed youth and workers in hard-hit industry sectors for a duration of 10 months a year; and a public works/ employment guarantee program for informal sector workers in both urban and rural areas offering employment for up to 100 days a year. Benefit levels are set at 10% of average earnings for the skills training program, and 150% of the national individual poverty line for the public works program at an administrative cost of 40% of overall program costs for each.

Social Protection Floors Package for 2024-2030
After the recovery/transition phase, SPRS20 administers a revised social protection package from 2024 up to 2030 with the objective of delivering sustainable social protection floors in countries, building on and complementing the recovery/transition package. While the emergency and recovery/transition packages are aimed as short-duration additional measures complementing existing social protection systems in a country, the social protection floors package is designed as a comprehensive blanket of measures representing the sum of all social protection programs in a country for the remaining duration of the SDG targets. The programs in this phase aim to cover life-cycle vulnerabilities for the most vulnerable population after the immediate foreseeable impact of the COVID-19 pandemic.
In line with the ILO Social Protection Floors Recommendation, 2012 (No. 202) (ILO 2012), the social protection floors package is designed to provide, at the least, (i) access to essential health care including maternity care; (ii) basic income security for children, providing access to nutrition, education, and other necessary goods and services; (iii) income security for persons in active age but unable to earn sufficient income due to sickness, unemployment, maternity, or disability; and (iv) basic income security for the elderly. These policy priorities have been reflected in the social protection floors package through a combination of social insurance, social assistance, and ALMPs as shown in Table 3.
Social insurance programs modeled under this package are primarily underpinned by the assumption that all programs will move to a fully contribution-financed approach without government subsidies during this period. Social insurance programs are targeted at formal sector workers and include health insurance offered for 12 months a year at a benefit level calculated with country-specific information; sickness and unemployment benefits offered for up to 3 months in a year at 60% of average earnings; a maternity benefit for women in the formal sector for a period of 3 months a year at 100% of average earnings; and old-age (survivor and disability) pension and a disability benefit both offered at 50% of average earnings for 12 months a year.
The social protection floors package offers a wider variety of noncontributory social assistance programs in line with Recommendation 202 of the ILO. For the period 2024 to 2030, this includes (i) a health insurance contribution waiver for uninsured population continued from previous phases but assumed to stabilize at 50% of informal sector workers and their families from 2025 onward; (ii) a child benefit for children up to the age of 17, at a benefit level of 20% of the national poverty line for 12 months in a year; (iii) a maternity benefit for informal sector workers and their families at a benefit level of 100% of national individual poverty line for 4 months in a year; (iv) a cash transfer for poor households at a benefit level set relative to national individual poverty lines (on average, 170% of national individual poverty lines) and household sizes for 12 months in a year; and (v) universal old-age benefit and disability benefit for those not covered under formal social insurance schemes at 100% of the national poverty line for 12 months in a year. The child benefit in the package is an individual-level benefit envisioned as a top-up for all households with children, possibly eligible for other social assistance benefits as well, and hence has a much smaller benefit level in comparison to other programs.  Modeled after the Indonesian fee waiver, but with country-specific data inputs.
Source: Authors' elaboration based on program and parametric choices adopted for current costing exercise.
Labor market programs under the social protection floors package remain without change from that of the recovery/transition package, offering skills training and public works/employment guarantee programs at the same parameter levels. All programs are assumed to be gradually phased in to achieve full coverage at the end of the social protection floors phase.

ResULts AnD FInDInGs
This section presents the results of the costing exercise undertaken, segregated into two parts: (i) estimates of COVID-19-related social protection costs incurred through the emergency (2020) and recovery/transition (2021-2023) phases; and (ii) overall cost estimates of delivering a comprehensive social protection floor for the remaining duration of the SDG targets (2024)(2025)(2026)(2027)(2028)(2029)(2030). The results are further disaggregated by income groups, subregions, and individual programs to help elaborate on estimated social protection cost patterns for different countries and population groups in Asia and the Pacific. The costing analysis was undertaken for a set of 30 countries in the region which included two low-income countries, 11 upper middle-income (UMI) countries, and 17 lower middle-income (LMI) countries. Costs have been expressed as a percentage of GDP and government share (in percentage of GDP). For the emergency phase in 2020, costs are further expressed as a percentage of overall government revenue to help estimate the burden of social protection costs on government revenue through this critical period. 6 It should be noted that coverage evaluation based on the present costing exercise is subject to some level of double counting, considering the overlap of eligible beneficiaries for 6 Current version of SPRS20 does not include projections of government revenue for 2022-2030. This will be included in upcoming versions of the model. different programs simulated. Considering this limitation, coverage is discussed here at a program level to help estimate access to programs for different population groups. Detailed costing results for different countries/subregions/income groups and beneficiary disaggregation for selected programs that run in more than two phases (by urban/rural, poverty status, and gender) 7 can be found in the Appendix.

emergency (2020) and recovery/Transition (2021-2023) Phases
The COVID-19 emergency package modeled for 2020 is estimated to cost approximately 3.02% of GDP on average for the 30 countries studied. In the emergency phase, costs are assumed to be completely government-financed, estimated to account, on average, for 14.3% of government revenue across these countries. The emergency package is estimated to be particularly expensive in several countries including the Kyrgyz Republic (5.9% of GDP), Cambodia (4.3% of GDP), Timor-Leste (4.1% of GDP), and Solomon Islands (4.8% of GDP). Figure 2 plots such costs for a selected set of countries studied. Source: Authors' illustration based on simulation results generated by SPRS20.
In Cambodia and the Kyrgyz Republic, high costs are driven primarily by large shares of uninsured and vulnerably employed population, making programs targeted at such individuals particularly expensive. COVID-19 and Social Protection in Asia and the Pacific This also explains why costs are estimated to drop in these countries over time as some of these programs are eliminated in the recovery/transition package. Costs in Solomon Islands and Timor-Leste, on the other hand, are expected to be driven by rather large contributions from unemployment assistance and cash transfers.
On average, costs in the emergency phase are primarily led by the wage subsidy for vulnerably employed individuals, accounting for 1.02% of GDP and catering to a total of over 713 million beneficiaries across the 30 countries. This is followed by the health insurance contribution waiver for the uninsured population at an average cost of 0.6% of GDP, reaching over 2.3 billion beneficiaries across the 30 countries; and cash transfers accounting for 0.5% of GDP and reaching a total of 141 million poor households across these countries.
As we move to the recovery/transition phase, total average costs are estimated to initially increase in 2021, in comparison with the emergency phase, to 3.4% of GDP and then gradually decrease over time to 3.2% of GDP in 2023. There are several factors to be considered while interpreting these results: (i) several large emergency programs like the wage subsidy for the vulnerably employed population and the unemployment assistance for informal sector workers are eliminated in the recovery/transition phase; (ii) the benefit duration of several programs, including the health insurance contribution waivers, cash transfers, and food assistance which are carried forward from the emergency phase, are extended for up to a year in the recovery/transition phase; (iii) government contributions, primarily for social insurance programs, are gradually phased out through the recovery/transition phase to be replaced instead by contribution financing; and (iv) new programs (activation measures) like skills training and public works/ employment guarantee are introduced in the recovery/transition phase. Government contributions are thus estimated to reduce over time, accounting on average for approximately 85% of total costs (2.7% of GDP) in 2023. Figure 3 plots the average costs as a share of GDP across the 30 countries for the emergency and recovery/transition phases, including program-wise contributions and costs to the government (dotted red line). Health insurance contribution waiver (SI and SA) Cost to the government (in GDP) ALMP = active labor market program, GDP = gross domestic product, SA = social assistance, SI = social insurance.
Source: Authors' illustration based on simulation results generated by SPRS20.
At the beginning of the recovery/transition phase in 2021, costs are estimated to be driven by the health insurance contribution waiver for the uninsured, accounting on average for 1.16% of GDP, followed by cash transfers at 1.15% of GDP and food assistance at 0.3% of GDP. At the end of the recovery/transition phase in 2023, cash transfers dominate costs on average at 1.02% of GDP, catering to a total of over 144 million poor households across the 30 countries, followed by health insurance contribution waiver for the uninsured at 0.8% of GDP and catering to a total of 1.8 billion beneficiaries across these countries. Food assistance is the third most expensive program in 2023 at an average cost of 0.3% of GDP, catering to over 239 million households in the 30 countries. Labor market programs (activation measures) introduced in this phase collectively account for 0.33% of GDP by 2023.

By income groups
In the 11 upper middle-income (UMI) countries studied, the emergency package is estimated to cost approximately 2.4% of GDP, accounting for 10.9% of government revenue on average in 2020.
In comparison, the package is estimated to cost approximately 3.0% of GDP and close to 16% of government revenue on average for the 17 lower middle-income (LMI) countries studied. Only two low-income countries were included in the costing exercise, registering an average cost of 6.6% of GDP and representing close to 23.0% of government revenue in 2020, the highest among all income groups ( Figure 4). The high costs in low-income countries may be associated with several factors, including a per capita GDP that is fairly close to the national poverty line, leading to high per capita benefit costs accompanied by high poverty, unemployment, and informality rates-qualifying large shares of the population as beneficiaries of emergency assistance in such countries. Source: Authors' illustration based on simulation results generated by SPRS20. COVID-19 and Social Protection in Asia and the Pacific Programs driving costs vary across income groups. For upper and lower middle-income countries, the wage subsidy (UMI: 0.9%, LMI: 1.1% of GDP); cash transfers (UMI: 0.5%, LMI: 0.6% of GDP); and the health insurance contribution waiver for the uninsured (UMI: 0.4%, LMI: 0.5% of GDP) are estimated to be the main drivers of costs in the emergency phase. In low-income countries, costs are led by the health insurance contribution waiver for the uninsured (2.3% of GDP), followed by the unemployment assistance for informal sector workers (2.2%) and cash transfers (1.0%).
Moving to the recovery/transition phase, the pattern of costs across income groups follows similar trends in the overall Asia and Pacific region, increasing initially in 2021 and gradually reducing over time. Low-income countries are estimated to incur the largest change, where costs are estimated to increase by almost 2 percentage points in 2021 compared to 2020, driven primarily by the contribution waiver for the uninsured (4.2% of GDP), cash transfers (2%), and the public works/employment guarantee program (0.9%). By 2023, the recovery package is estimated to cost 2.8% of GDP in UMI countries (government contributions: 2.1%, contribution financing: 0.7%); 2.9% of GDP in LMI countries (government contributions: 2.5%, contribution financing: 0.4%); and 7.9% of GDP in low-income countries (government contributions: 7.0%, contribution financing: 0.9%). Social assistance continues to be the most expensive program category through the recover/transition phase for all income groups. By 2023, however, cost of several social insurance measures, such as health insurance for formal sector workers, are expected to rise significantly, particularly in UMI countries. In low-income countries, labor market programs additionally show a steady increase in costs over this period, with the public works/ employment guarantee program contributing significantly to such costs.

By Subregion
Overall, the emergency package is estimated to be particularly expensive in Central and West Asian countries, registering an average cost of 3.9% of GDP and representing close to 15% of government revenue in 2020 ( Figure 5). The subregion is estimated to incur the highest costs through the recovery/ transition phase as well, registering a cost of 4.4% of GDP (government contributions: 3.6%, contribution financing: 0.8%) in 2023. The higher costs may be explained by looking into the income composition of countries studied within this subregion, which includes four UMI countries, two LMI countries, and both the low-income countries studied. The subregional average is thus escalated considering the large variations between these income groups. Costs in this subregion are estimated to be driven by the contribution waiver for the uninsured (1.1% of GDP), wage subsidy (1.01%), and unemployment assistance for informal workers (0.7%) in 2020. In comparison, the emergency package is estimated to be least expensive in East Asian countries at 2.3% of GDP. Wage subsidies (1.2% of GDP), the contribution waiver (0.4%), and cash transfers (0.2%) are estimated to be the most expensive programs in East Asia through the emergency phase.
In contrast to overall Asia and Pacific trends, where costs are estimated to initially increase in 2021 in comparison to 2020 and then eventually drop, costs in East and South Asian countries are estimated to drop throughout the recovery/transition phase. This may be associated with the rather large contribution of the wage subsidy for vulnerably employed being eliminated in 2021, accompanied by rather strong GDP recovery rates estimated in these subregions. At the end of the recovery/transition phase, East Asia continues to register the lowest subregional costs at 2.08% of GDP (government contributions: 1.4%, contribution financing: 0.6%). Source: Authors' illustration based on simulation results generated by SPRS20.

By gender, Poverty Status, and urban/rural Classification of Beneficiaries
While the overall impact of the social protection packages by gender, poverty status, and urban/rural classification are difficult to estimate due to overlap in beneficiary groups, estimations of reach may be aggregated at the program level (Table 4). In the emergency and recovery/transition phases, women are estimated to benefit most from the health insurance benefits and sickness benefit, as close to 50% of all beneficiaries are women across all income groups. In comparison, 37% of beneficiaries of wage subsidies for the vulnerably employed in the region are women. The skills training and intermediation program also indicates similar distributions. Similarly, unemployment benefits for both formal and informal workers account for about 30%-35% of female beneficiaries in the region for such programs. This trend may be attributed to low female labor force participation rates in the region. These estimations also indicate that industry sector or employment status-based targeting approaches may not sufficiently reach women in such crisis scenarios.
Disaggregation by poverty status and urban/rural classification are further influenced by specific program-based assumptions and eligibility considerations in the model. The wage subsidy, which is the most expensive program in the emergency phase, is estimated to cater to almost equal shares urban and rural beneficiaries across Asia and the Pacific. However, the program is estimated to favor the nonpoor over the poor, with a mere 10% of total beneficiaries of the program estimated to be poor. Similarly, the health insurance contribution waiver for the uninsured appears to be equally distributed among urban and rural beneficiaries, but again favors the nonpoor, with just 16% poor beneficiaries across the region. The poor appear to benefit most from these programs in low-income countries, where close to 50% of total beneficiaries are estimated to be poor. Cash and food assistance programs are particularly targeted at poor households and cater almost equally to urban and rural beneficiaries. The skills training and intermediation program again accounts for just 11% of poor beneficiaries. These estimates indicate that the poor may benefit only marginally from emergency measures in these phases, other than those directly targeted at such individuals or households.  Source: Authors' elaboration based on simulation results generated by SPRS20.

Cost of Comprehensive Social Protection Floors, 2024-2030
Costing estimates produced by SPRS20 for the period 2024 to 2030 represent the total cost of delivering a comprehensive social protection package, including floors in line with SDG target 1.3 and ILO Recommendation 202. In 2024, the social protection floors package modeled is estimated to cost on average 4.1% of GDP in the 30 countries studied, with costs to the government estimated at 3.1% of GDP and the remaining 1% financed through contributions. Figure 6 presents total costs in selected countries as a share of GDP for specific years in the social protection floors phase. In 2024, the package is estimated to be particularly expensive in several countries, including Solomon Islands (11.0%), Tonga (6.7%), Vanuatu (7.1%), the Kyrgyz Republic (5.9%), and Armenia (5.7%) registering costs considerably above the regional average for this year.
On average, cost of the social protection floor package is estimated to increase over time in the 30 countries as the programs mature and achieve full coverage (Figure 7).  Source: Authors' illustration based on simulation results generated by SPRS20.
Costs are estimated to peak in the years 2027 and 2028 (6.2% of GDP), and further decrease to reach 5.9% of GDP (government contributions: 3.6%; contribution financing: 2.3%) at the end of the projection period in 2030. This trend may be explained by the underlying modeling assumption that programs in the social protection floors phase are gradually phased in, accompanied by the denominator effect of a faster GDP growth in comparison to overall cost of the social protection package through this phase. By 2030, the package is estimated to be particularly expensive in several countries like the Kyrgyz Republic (10.7%), Timor-Leste (10.6%), and most of the Pacific countries. In the Kyrgyz Republic and Timor-Leste, these costs are driven by rather large contributions added by the child benefit and old-age benefits. The Pacific island countries, in addition, are estimated to incur comparatively high costs throughout this phase; factors driving such costs will be discussed further in forthcoming subsections.
In 2024, the cash transfer (0.9% of GDP), the health insurance contribution waiver for the uninsured (0.7%), and the universal old-age assistance (0.5%) are estimated to be the most expensive programs on average across the 30 countries-all social assistance measures (Figure 8). At the end of the projection period in 2030, costs are estimated to be driven by a combination of social assistance and social insurance measures, although social assistance continues to contribute the largest share (61%) of total costs. The child benefit (1.2% of GDP), formal sector pensions and universal old-age assistance (0.8% each), and the formal sector health insurance program (0.7%) are estimated to be the most expensive programs in 2030. Child benefits are estimated to reach a total of over 1.1 billion children across the 30 countries in 2030, while old-age benefits, both formal sector pensions and universal assistance, are estimated to reach over 360 million elderly. Labor market programs, on the other hand, together contribute a cost of just 0.28% of GDP on average in 2024, further reducing to 0.04% in 2030. Health insurance contribution waiver (SA)

Short-term SI benefits (sickness maternity and unemployment)
Health insurance (SI)

Cost to the government (in GDP)
Long-term SI benefits (old age survivors and disability) ALMP = active labor market program, GDP = gross domestic product, SA = social assistance, SI = social insurance.
Source: Authors' illustration based on simulation results generated by SPRS20.

By income group
By 2030, the social protection floors package is estimated to cost approximately 5.6% of GDP on average (government contributions: 2.5%, contribution financing: 3.1% of GDP) in UMI countries. In LMI countries, the package is estimated at 5.4% of GDP (government contributions: 3.7%, contribution financing: 1.7% of GDP). The package is estimated to be particularly expensive in low-income countries, registering an average cost of approximately 12.4% of GDP (government contribution: 9.5%, contribution financing: 2.9% of GDP) in 2030 (Figure 9). The ratio of government financing to contribution financing remains significantly high in low-income countries, where close to 85% (2024) to 75% (2030) of total costs on average are estimated to be borne by the government. This may again be associated with a low per capita GDP which is fairly close to the national poverty line, accompanied by high informality and poverty rates contributing to large shares of beneficiaries for government-financed social assistance measures, in combination with a rather slow overall GDP growth in comparison to the size of beneficiary groups in these countries. Source: Authors' illustration based on simulation results generated by SPRS20.
Programs driving costs vary across income groups. In 2030, for UMI countries, the formal sector old-age pension (1.1% of GDP) is estimated to be the most expensive program, followed by the health insurance for formal sector workers (0.7%) and the child benefit (0.7%). Overall, social insurance measures are estimated to dominate costs in UMI countries, accounting for close to 54% of total costs in 2030.
In comparison, costs in LMI countries are led by the child benefit (1.3% of GDP), followed by universal old-age assistance (0.9%) and formal sector old-age pension (0.6%). Social assistance programs account for the largest share (68%) of total costs in LMI countries in 2030. In low-income countries, costs are again led by social assistance measures accounting for almost 76% for total costs. The child benefit (3.7% of GDP), the health insurance contribution waiver for the uninsured (1.8%), and the universal old-age assistance (1.6%) are estimated to be the most expensive programs in low-income countries in 2030.

By Subregion
The cost of the social protection floors package is estimated to be highest among the Pacific countries studied, registering an average of 6.5% of GDP (government contributions: 5.1%, contribution financing: 1.4% of GDP) in 2024 and reaching 9.02% of GDP (government contributions: 6.02%, contribution financing: 3%) in 2030, consistently above the Asia and the Pacific average through this phase ( Figure 10). In 2024, this high average is driven primarily by Solomon Islands with 11.02% of GDP and Vanuatu with 7.1% of GDP, both LMI countries. Toward the end of the projection period, however, costs are estimated to significantly escalate for all six countries, with Fiji registering the lowest cost in 2030 at 5.5% of GDP. On average, costs in the Pacific countries in 2030 are estimated to be driven by the child benefit (2.2% of GDP), followed by the universal old-age assistance (1.2%) and the formal sector old-age pension (1.1%). Several countries in this subregion, including Papua New Guinea (9.7% of GDP), Samoa (10.7%), and Solomon Islands (13.2%), are estimated to incur costs well above the Asia and the Pacific average in 2030. This is primarily driven by significantly high costs incurred in these countries for child benefits, old age, and disability (both formal sector insurance and universal assistance) benefits. Source: Authors' illustration based on simulation results generated by SPRS20.
In comparison, the social protection floors package is estimated to be least expensive among South Asian countries, registering an average cost of 2.5% of GDP in 2024 (government contributions: 1.9%, contribution financing: 0.8%), and reaching 3.0% of GDP in 2030 (government contributions: 1.9%, contribution financing: 1.1%)-considerably lower than the Asia and Pacific average for these years. This cost may be explained by the overall high GDP growth rates estimated in the subregion in this period in comparison to the total cost of the package. The universal old-age assistance (0.6% of GDP) is estimated to be the most expensive program in South Asia in 2030, followed by the child benefit (0.5%) and formal sector old-age pension (0.4%). Costs through the social protection floors phase are primarily driven by social assistance measures in most subregions studied, except in East Asia where social insurance measures contribute the largest share (up to 74% in 2030) of total costs throughout this phase. This is largely due to the high costs incurred by the formal sector old-age pension in East Asian countries, which in 2030 is estimated to cost up to 2.0% of GDP on average.

By gender Poverty Status and urban/rural Classification of Beneficiaries
By the end of the projection period in 2030, the maternity benefit for formal sector workers (social insurance) is estimated to reach over 50 million pregnant women and new mothers, while the maternity benefit for the informal sector (social assistance) is estimated to reach over 69 million beneficiaries across the 30 countries. Close to 50% of social assistance beneficiaries in the region for different programs (except for household-level programs like cash transfers and food assistance) are estimated to be women and girls. Social insurance programs also demonstrate a favorable gender impact except for the formal sector sickness benefit, which caters to just over 9 million women compared with 15 million men. Old-age benefits, both social insurance and tax-financed universal programs, are estimated have larger shares of female beneficiaries in comparison to males, reaching a total of over 192 million and 188 million female beneficiaries, respectively, in the region by 2030. ALMP measures, on the other hand, are estimated to disproportionately favor men, with just over 30% of female beneficiaries. These estimates may provide hints on programs where the gender dimension needs to be particularly emphasized in different countries in order to ensure equitable distribution of benefits among the population.
Disaggregation based on poverty status and urban/rural classification are again subject to parametric eligibility set across programs in the social protection floors phase. Social insurance programs are strictly limited to urban nonpoor. At the end of the projection period in 2030, East Asia and South Asia account for the largest shares of formal sector maternity program beneficiaries-close to 36% each of the total beneficiaries for the programs in the 30 countries. Most other formal sector social insurance programs are estimated to have their largest share of beneficiaries in East Asia driven primarily by the PRC.
Social assistance programs in the social protection floors package cater to a wide range of beneficiary groups, including those left outside the formal health insurance schemes, pregnant women, children, elderly, and persons with disabilities. Beneficiaries for such programs are generally concentrated in South Asia, accounting for over 40% of total beneficiaries in the Asia and Pacific region owing to large contributions from countries like India. Social assistance measures, including household-level programs, are estimated to have almost equal shares of urban and rural beneficiaries across the countries studied. The health insurance contribution waiver for the uninsured is estimated to perform comparatively better in this regard, particularly in Pacific and South Asian countries where close to 80% and 65%, respectively, of all beneficiaries for the program are rural. However, with respect to poverty status, all social assistance programs (with the exception of poverty-targeted household assistance) appear to perform in favor of the nonpoor, overall catering to an estimated 10%-18% poor among total beneficiaries for such programs in the region. Social assistance programs are estimated to better benefit the poor in low-income countries and Pacific countries, where between 35%-50% of beneficiaries are poor in comparison with other income groups and subregions.
With respect to labor market programs, the skills training and intermediation program has been particularly targeted at the young urban unemployed. Overall, the training program is estimated to have a very low poverty impact catering to just 11% poor among total beneficiaries for the program in the region. The poor are estimated to benefit most from the program in Central and West Asian countries, where close to 30% of beneficiaries are poor in comparison to Southeast Asia and East Asia where a mere 11% and 1.74%, respectively, of beneficiaries are poor. Similarly, the public works program in this phase has been particularly targeted at poor individuals. Overall public works programs are estimated to perform better for the rural poor, with over 52% rural beneficiaries in the region. The rural poor appear to particularly benefit in Pacific countries, where they account for over 76% of total beneficiaries for this program, in comparison with 40% in East Asian countries.

ConCLUsIons
While the COVID-19 pandemic delivered a severe blow to achievements toward the SDG targets, including poverty alleviation and social protection, it also serves as an important lesson in the urgency of strengthening comprehensive national social protection systems in countries around the world. To support this endeavor in Asia and the Pacific, the estimates presented in this report plot an approximation of incremental financial needs in 30 countries in the region to deliver a set of measures to facilitate emergency assistance (2020), recovery support (2021-2023), and prolonged comprehensive social protection floors for the remaining duration of the SDG targets (2024-2030). The results presented here provide an estimate of the pattern of individual program costs; their contribution to overall expenditure; and beneficiary distribution for these programs across gender, urban/rural, and poverty status.
The social protection packages administered in this report provide an example of the level of customization and disaggregated results offered by the SPRS20 model. While for the purpose of this report, a uniform set of measures have been included in each social protection package administered, the model allows for customizations at a country level based on national strategies and priorities. Such estimates could serve as references for social protection planning in the next few years in these countries.
It is tempting to compare the cost estimates discussed in this report with current or pre-COVID-19 social protection spending; however, this is not a straightforward exercise. The social protection floors package modeled here, for example, is an ideal package that assumes full coverage without targeting errors. In current spending, on the other hand, these errors of inclusion and exclusion are imminent and whereas large numbers of people will receive less than they should, some will receive more. These inefficiencies are not included in the simulations undertaken in this report. Therefore, comparing these cost estimates with current expenditure should be done with great caution.
The incremental social protection costs estimated in these countries further need to be met with sustainable resource allocation, mobilization, and political will at the national level. With large informal sectors and low tax bases, low-income countries and other developing economies often find themselves relying heavily on general government revenues to finance social protection benefits, in comparison to more advanced economies that rely more on contribution financing (Coady 2018). Building adequate fiscal space to cater to the incrementing social protection costs in countries thus forms a critical step in the path toward achieving the SDG targets.
Literature on social protection discuss several options to increase fiscal space availability for social protection, including taxation, reducing illicit financial flows and corruption, public expenditure reprioritization, and foreign aid. Taxation, for example, remains the main source for public finance in most countries, accounting for over 50% of total government revenues in almost every country (Ortiz et al. 2019). Studies further indicate that countries move to a higher growth path and in part, higher social spending, once tax revenues reach about 15% of GDP (Gaspar, Jaramillo, and Wingender 2016). COVID-19 and Social Protection in Asia and the Pacific However, globally, many low-income countries and emerging market economics are reported to have tax ratios below this threshold (Coady 2018). In developing Asia, most countries have been systematically increasing their tax-to-GDP ratio in recent years, with the exception of a few like Malaysia (12.5%) and Bhutan (12.3%), which reported a decline in tax-to-GDP ratio between 2017 and 2018. Others like Kazakhstan (16.8%), Mongolia (24.0%), the Philippines (18.2%), Thailand (17.5%), Solomon Islands (30.4%), and Fiji (23.7%) report ratios well above the 15% threshold (OECD 2020), and, in turn, conveying greater potential for increased social protection spending. Further, a close evaluation of tax performance in developing economies reveals ample scope in these countries for raising tax revenues, especially when it is complemented by a strong political will (Ortiz et al. 2019). Increasing tax compliance, restructuring tax rates, and improving tax system efficiency may thus serve as critical tools to improve fiscal space availability and increased social protection spending in countries. Similarly, other fiscal space options relevant to the economic, social, and political landscape of individual countries need to be explored in depth.
In addition, the expected economic growth in countries over the coming years plays a critical role in determining the affordability of extended social protection coverage and expenditures. As the results discussed in section 4 elaborate, the cost of social protection packages at different stages is estimated to be highest among low-income and LMI countries, where resource availability is often limited with a heavy dependency on donor financing and international aid. These estimations, along with the global vision for inclusive sustainable development, highlight the necessity to make adequate resources available to low-income and LMI countries to extend adequate support to their citizens. These further justify the recent call for the establishment of a Global Social Protection Fund as put forth by the Global Coalition for Social Protection Floors (2021).
Further research is also warranted toward policy frameworks and administrative mechanisms supporting social protection in countries. Often, different components of social protection are responsibilities of different line ministries, local governance units, and other social partners within countries. In developing economies, this forms a rather uncoordinated social protection landscape with roles and responsibilities often not well-defined, resulting in gaps, duplication of efforts, and efficiency losses (UNDG Asia Pacific 2016). A coherent social protection landscape is thus essential to ensure that national efforts translate effectively for different beneficiary groups within the country.
This report and the SPRS20 tool adds to the resource base available to countries to effectively facilitate such planning, and the realization of national and regional social protection targets in the future.  Source: Authors' elaboration based on simulation results generated by SPRS20. Source: Authors' elaboration based on simulation results generated by SPRS20.