POLICY NOTE 1 Adequacy of Coverage, Targeting and Benefits of Social Protection Programs BACKGROUND To maximize the impact of social protection (SP) programs, it is critical to: (1) align the allocation of resources with national policies and strategies; (2) design programs that are able to reach their intended beneficiaries with adequate coverage and level of benefits; (3) ensure efficiency in programming and financing; and (4) ensure efficiency in the actual delivery of programs at the local level. This policy note analyses point (2), the adequacy of SP programs in Bangladesh, and highlights areas where challenges remain and proposes policy reforms to address those (others are covered in other policy notes and in the main report ). It focuses on the three elements of the design of social protection (SP) programs that are critical to maximum impacts on poverty and vulnerability: (1) the extent to which programs have the “right” size and coverage; (2) the extent to which programs effectively reach or target the “right” beneficiaries; and (3) the extent to which programs provide these beneficiaries with the “right” benefits. The optimal combination of these elements depends on programs’ policy objectives, the needs or demands of target population groups, the types of beneficiaries or programs, and available resources. 1 The Policy Note series discusses key thematic findings and recommendations outlined in the ‘Bangladesh Social Protection Public Expenditure Review (PER)’ report. The series includes notes on the Allocative efficiency: Better Align Social Protection Resources with National Policy; Adequacy of Coverage, Targeting and Benefits of Social Protection Programs; Efficiency in the Programming and Financing of Social Protection; and Efficiency in the delivery of Social Protection Benefits and Services to Beneficiaries. 1 Adequacy of Coverage, Targeting and Benefits of Social Protection Programs POLICY NOTE COVERAGE In Bangladesh, regular increases in social protection budget have beneficiaries from EGPP increased 1.5 times (figure 1). The number typically prioritized increases in coverage over increases in benefit of beneficiaries of a few programs have likely already reached the size or amounts. The share of households benefiting from social size of their target population, for instance, EGPP and DA. assistance programs more than doubled from 12 percent in 2005 In fact, if these were perfectly targeted to the poor, some would to 25 percent in 2010, before going at a slower pace to reach nearly cover all the poor (e.g. Vulnerable Group Development (VGD) or 28 percent of households by 2016. A few large programs, including Primary Education Stipend Program (PESP)) and a few would reach allowances or cash-based programs (Old Age Allowance (OAA), 2/3 of the poor (VGF and GR). Beyond these few programs, coverage Widow Allowance (WA), and Test Relief (TR)) as well as food-based is typically very small and variable. While some of the smallest programs (Vulnerable Group Feeding (VGF), Gratuitous Relief (GR) programs are in a “piloting” phase, which is essential to test new and school feeding program (SFP)), and stipends drive the patterns. approaches and methodologies, many are simply operated on a Over the past eight years, for instance, the OAA and WA increased very small scale, contributing to fragmentation. their number of beneficiaries six times; while the number of Old Age Allowance Allowance for Widow, Destitute and Deserted women Employment Generation Program for the Poorest Allowance for Financially Insolvent Disabled 5.0 4.5 4.0 4.40 3.50 3.5 3.0 2.72 Millions 2.48 2.48 2.5 2.0 1.70 1.5 1.27 1.55 0.92 1.01 1.11 1.0 0.97 0.63 0.5 0.75 0.40 0.29 0.0 FY2011-12 FY2012-13 FY2013-14 FY2014-15 FY2015-16 FY2016-17 FY2017-18 FY2018-19 FY2019-20 Figure 1: Number of beneficiaries of key programs over time Source: Bangladesh national budget archives (budget speeches) and administrative data from MoSW and MoDMR. 2 POLICY NOTE TARGETING Scope for improved geographic targeting On average, about 46 percent of the total SP expenditure are 100% focused on rural areas, leaving the country ill-equipped to face increased urbanization. In particular, labor market programs are 90% Household only beginning to emerge in urban areas. While a rural focus partly covered by SP reflects higher poverty in rural areas, urbanization calls for increased 80% focus on urban programming. Coverage reflects past focus on rural Household not areas, with 35.7 percent of rural households benefitting from at 70% covered least one social assistance program, versus 10.9 percent of urban Percentage of 60% households (coverage is higher than poverty rates in rural areas, the poor opposite is true in urban contexts) (figure 2). A greater focus on urban 50% areas also calls for different or adjusted programs to account for the 26.4% features of urban poverty, significantly different from those of rural 40% poverty. 18.9% 30% Ideally, coverage should follow poverty rates, with areas with higher poverty benefiting from a stronger coverage. In practice, in 2016, 20% some regions were “undercovered” (e.g. Mymensingh, with poverty 35.7% of 32.8 percent and coverage of only 27.7 percent) while other 10% 10.9% experienced “overcoverage” (e.g. Barisal with a poverty incidence of 26.5 percent and coverage of 59.9 percent) (table 1). Re-balancing 0% Urban Rural these, with programs allocating funds partly as a function of poverty prevalence, would help improve the efficiency and equity of the national social protection system. Figure 2: Social protection coverage in urban and rural areas Source: World Bank, 2021. Table 1: Coverage of main social assistance programs and poverty, by region In terms of poverty targeting, programs are on average progressive, but significant ‘inclusion’ % of individuals Incidence of poverty and ‘exclusion’ errors persist Regions receiving a safety using upper poverty (division) On average, a cumulative 36 percent of core social assistance benefits net benefit line (%) accrue to the bottom 30 percent in rural areas. At the other end of Barisal 59.9 26.5 the distribution, the 10 percent richest only receive 6.4 percent of Chittagong 18.0 18.4 benefits. Some programs outperform others in their ability to reach Dhaka 12.8 16.0 the poor and vulnerable. Food support and allowances appear to be the best performer, followed by maternity allowances. Both Khulna 42.8 27.5 ‘inclusion’ and ‘exclusion’ errors need to be addressed to improve Mymensingh 27.7 32.8 social protection’s impact on poverty. Estimations using MICS data Rajshahi 37.7 28.9 for 2019 suggest ‘inclusion’ errors of 51 percent for the allowance programs, 54 percent for food support programs, and 62 percent Rangpur 45.2 47.2 for maternity allowances in rural areas (49, 45 and 66 percent Sylhet 27.9 16.2 respectively in urban areas). Source: Sen, 2020, based on HIES 2016. 3 Adequacy of Coverage, Targeting and Benefits of Social Protection Programs POLICY NOTE Absence of a mechanism for regular review The National Household Database (NHD), of program ‘eligibility criteria’ often hinders Bangladesh’s social registry, provides a unique better targeting opportunity to improve performance For some programs, improving eligibility criteria would help better While each program defines its own criteria and thresholds, they target those most in need of support. Estimations based on HIES tend to use relatively similar indicators. Around the world, social 2016 shows that while 68 percent of OAA beneficiaries met the registries have been used by many types of programs in the process ‘demographic’ criteria, only 12 percent met the ‘economic’ criteria of identification of their beneficiaries. Typically, social registries (20 percent did not meet any criteria). This could be partly due to support one of the steps followed to identify beneficiaries, and difficulties in verifying these criteria (many elderly have no proof can be combined with other elements, such as community-based of age, income is hard to verify, and land ownership is not always elements, as well as caseworker assessments, statistical profiling, observable). It could also suggest that the ‘economic’ criteria have medical assessments, or disability assessments. The NHD compiles become obsolete, so that local committees implicitly use a higher information for all households in Bangladesh (approximately 35 threshold. Similar analysis for EGPP shows that a greater share meets million). When fully operational, the NHD can also be harnessed the ‘economic’ eligibility criteria, but fewer meet the ‘demographic’ during the shocks and emergencies, to identify households in need criteria (age 18-60). Anecdotal information on EGPP indicates that of support who are not already enrolled in regular social assistance older men (and working age women) tend to work on the program programs. as working age males commute further for better paid work. BENEFITS Relatively small benefits limit the impacts of Slow pace of food-to-cash conversion reduces many programs cost-effectiveness With a focus on increasing coverage, the benefits provided by many The share of support in the form of food transfers declined; programs have stagnated or remain very low. For instance, while but absolute amounts have remained relatively constant over the number of beneficiaries continued to increase, the monthly time. This slow conversion has important consequences for amounts for the OAA and the WA were last raised in FY2016-17 by beneficiaries, who tend to prefer to receive cash rather than food, about BDT 100 respectively and the daily rate for the EGPP was last because of the fungibility of cash, as well as the poor resale value increased in FY2013-14.Most benefits represent only 1 to 3 percent of food (and uncertainty on quality). In addition, food-based of total household income for the poor or extreme poor (with the programs can be more expensive, because of costs associated exception of TR, OAA and WA, which represent 9, 7 and 7 percent with storage, transportation, and delivery; which reduces of the poor’s average income respectively).vi Low benefits limit the their cost-effectiveness (administrative costs of food transfer impact of many programs on poverty: in 2016, among beneficiaries programs typically are almost 2.5 times higher than those of cash of any of seven core programs, programs are responsible for a transfer programs). decrease in the extreme poverty headcount from 22.2 percent to 19.8 percent (only 2.4 percent). Missing opportunity to promote behavioral Improving on targeting could yield important impacts. In rural changes through program interventions areas, assuming no increase in the budget allocated to programs, Most programs only provide cash, missing an opportunity allocating all benefits which currently go to non-poor households to to provide their beneficiaries with interventions that further poor ones would close about 24 percent of the gap for allowances, awareness and knowledge or encourage improved practices. but would more than close the gap for food support and maternity Increasingly, policymakers around the world are applying insights allowance.vii THE WORLD BANK 4 POLICY NOTE from behavioral science to enrich their programs with activities Lack of a “system’s approach” limit that promote lasting behavioral changes in order to better address households to combine interventions for challenges such as healthy eating, educational attainment, financial maximum impact over their life cycle inclusion, or climate-smart investments. Bangladesh should also promote this practice. Many programs have interactions with Currently, many programs in Bangladesh explicitly (or in practice) households, either through group activities (information session, rule out those who already benefit from other programs. As a public work sites, counseling, training, etc.), and these provide an result, only 1 and 3 percent of EGPP and Maternal Health Voucher opportunity to communicate directly with the target population Scheme (MHVS) beneficiaries also received support from an (as already done by the Maternity Allowance or Income Support NGO. Estimates from MICS show that 84 percent of all beneficiary Program for the Poorest-JAWTNO).This is particularly relevant as households in rural areas only receive one benefit (88 percent in Bangladesh expands its Mother and Child Benefit Program, and urban areas). While limiting combination might be reasonable is planning to enrich some of its key allowances programs with for some programs (which have similar design/function), it might complementary measures. be counterproductive in other cases. For instance, programs that address specific event (e.g. a disaster/seasonal shocks) typically have a limited duration, and are complementary to longer-term support. Systematically excluding those who benefit from any other program can have negative impacts on their ability to cope with a shock, such as the one resulting from COVID-19. More generally, global evidence shows how some interventions are complementary, and together can boost impacts. RECOMMENDATIONS • To increase impact on poverty, reform key programs by focusing on the size of benefits or on expansion, depending on the program. For some programs, it is critical to revisit the amounts transferred to ensure they can have a meaningful impact on poverty (slowing the expansion of their coverage) and take into account the evolution of prices. For others, expanding coverage is the priority, especially for labor market programs where large gaps are observed. Finally, for programs to yield lasting behavioral changes, they can use insights from behavioral science to enrich their interventions with accompanying measures and better address core development challenges – around economic inclusion as well as child development, for instance. • Rebalance geographic allocations between rural and urban areas, with programs tailored to urban poverty. Bangladesh Bureau of Statistics’ (BBS) poverty maps can be used to allocate resources by region, districts, or even upazilas. This is already the case for EGPP, and for the OAA/WA expansion to the 112 poorest upazilas, as part of the Government’s COVID-19 response. Other large programs could consider using this tool. Since poverty patterns, vulnerabilities, and constraints are likely different, there is also a need to adapt existing programs (or tailor urban modalities), then scale up interventions to support growing poor urban population. • Review and revise eligibility criteria to boost programs impacts and ensure greater focus on the poor. Many programs would benefit from revising their eligibility criteria, in particular economic eligibility criteria, to increase their focus on the poor and to reflect changes in poverty patterns and overall increases in incomes and living costs. This could be done using proxies for income or poverty, or by adding indicators that capture the heterogeneity of the poor and vulnerable. It is also critical to define criteria that can help identify households who are not poor in normal times, but are vulnerable to shocks, such as the COVID-19 crisis (e.g. self-employed or informal wage worker; or those with income just above the national poverty line). This would allow for their quick identification and support in times of shocks. Allowances for the Financially Insolvent Disabled 5 Adequacy of Coverage, Targeting and Benefits of Social Protection Programs POLICY NOTE • Use a social registry, such as the National Household Database (NHD), to improve targeting at a reduced cost. By merging the data collection process for multiple programs, a registry can reduce costs. Meanwhile, each program can draw from the registry’s information using its own criteria, as part of their beneficiary selection process. One of the appeals of a social registry is also the fact that it already includes households who currently do not benefit from programs, but who could be in need of support after particular shocks. For the NHD to contribute to gains in efficiency and impact, it needs to have an ‘easy to use’ operational data sharing protocol and to regularly update the information on households registered in the database. • Move towards a system’s approach, where households can combine programs according to their needs. This can help maximize impacts, as it tailors support to specific situation. It can build on the NHD as a coordination mechanism (to avoid duplication and seek synergies). The NHD, if regularly updated and audited, can also help regularly reassess beneficiaries’ needs and conditions (this is commonly known as recertification). A system’s approach also calls for some programs to have clear exit strategies to maximize the reach and performance of such programs, with households “navigating” between programs throughout their life cycle. i World Bank, 2021. “Bangladesh Social Protection Public Expenditure Review (PER)”. World Bank Group, Social Protection & Jobs, Dhaka, Bangladesh. ii World Bank, 2021. iii Using an asset index as a proxy for consumption and a poverty rate of 26.7 percent for rural areas and 19.3 percent in urban areas. Estimates reported in World Bank, 2021. iv It could also reflect the fact that some elderly recently passed away, with some household reporting having no elderly, but having received OAA during the previous 12 months (when their elderly member was still alive). v Indonesia’s Unified Data Base (UDB), Turkey’s Integrated Social Assistance System (ISAS), Philippines’ Listahanan, Colombia’s System of Identification for Social Subsidies Beneficiaries (SISBEN) or Pakistan’s National Socio-Economic Register (NSER) and the Dominican Republic’s Sistema Unico de Beneficiarios (SIUBEN) cover about 40-85 percent of their population respectively. vi World Bank, 2021. vii World Bank, 2021. World Bank Office Dhaka World Bank, 2021. “Bangladesh This Policy Note has been prepared Plot- E-32, Agargaon, Sher-e-Bangla Nagar Social Protection Public by Mostafa Amir Sabbih and Aline Coudouel, Social Protection & Jobs Dhaka-1207, Bangladesh Expenditure Review (PER)”. World Global Practice, World Bank Tel: 880-2-5566-7777, Fax: 880-2-5566-7778 Bank Group, Dhaka, Bangladesh. www.worldbank.org/bangladesh © World Bank. 6