97963 NOTE 1 10 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection Reaching the poor and vulnerable: Key considerations in designing targeting systems1 Many countries develop systems to identify the poor and vulnerable in order to direct program resources to these areas and groups, often to ensure their inclusion in social service provision or community development. Determining eligibility typically depends on the overall program objectives, patterns of poverty, vulnerability and exclusion, social norms, and fiscal and political feasibility. Tar- geted programs use a range of tools to prioritize specific areas and people, in keeping with data availability and existing administrative capacity. Depending on these policy choices, effective delivery sys- tems for beneficiary registration, management, monitoring, and re- certification need to be developed, with clear assignment of roles and responsibilities at different levels. Many countries have done so in a coordinated fashion, so the same system can be used to support eligibility verification for multiple programs. 1. This Note was prepared by Puja Vasudeva Dutta and Yuko Okamura (World Bank). Comments were provided by Mariana Infante- Villarroel, Inge Stokkel, Philip O’Keefe, Jehan Arulpragasam Carlo del Ninno and Christian Bodewig. Any comments and questions can be addressed to pdutta@worldbank.org. The team is grateful to the Rapid Social Response program and its five donors the Russian Federation, Sweden, Norway, the United Kingdom and Australia for funding this assessment 2 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection Reaching the poor and vulnerable: Key considerations in designing targeting systems 3 1. Overview Many countries have redistributive policies that concentrate program resources on specific ar- eas and groups, particularly as part of reform packages aimed at promoting more inclusive growth. Developing an effective mechanism to do so is a complex and sometimes costly task, but one that can contribute to improving the quality and poverty impact of social protection and com- munity development programs. Well-administered and effectively targeted programs can assume an even greater significance during times of crisis when the quickest response is often to scale up existing programs. Global experience suggests that, even in programs delivering universal benefits or services, some element of targeting may be useful in ensuring the inclusion of groups that face resource or other constraints in accessing these services. For instance, many countries have cash transfer (CT) programs targeted to poor households to encourage children to enroll and stay in school or to utilize health and nutrition services (World Bank, 2013). The decision to provide benefits to specific areas, groups, or households ideally reflects an in- formed debate on the desirability, feasibility, and acceptability of targeting in the particular country context. Depending on the patterns of poverty, vulnerability, and exclusion in the country, the overall program objectives, administrative capacity, and social and political imperatives, tar- geting may or may not be the appropriate policy choice. In most circumstances, however, global evidence suggests targeting can often increase poverty impact by focusing resources on those most in need and facilitate inclusion by enabling marginalized groups to access social services af- fordably (Coady et al., 2004; Grosh et al., 2009). But not always: much depends on the availability of good-quality data and implementation capacity. This note lays out the key considerations in making targeting choices, based on international experience (see Annex 1 for a summary of selected examples). The next section summarizes the main elements of a targeting system. Section 3 lays out the considerations in determining eligibil- ity, including decisions to focus on specific geographic areas or people. Section 4 describes the range of targeting tools used to do so. Section 5 reflects on the nature of institutional arrange- ments and delivery systems to operationalize these choices. This experience can provide valuable insights for Myanmar as program funding increasingly emphasizes the inclusion of the poor and vulnerable. A companion Note examines the experience of targeting in Myanmar.2 Taken together, these notes aim to identify feasible options for Myanmar, based on the innovative ideas and solu- tions emerging from global experience. 2. See ‘Reaching the poor and vulnerable in Myanmar: Lessons from a social protection and poverty reduction perspective’ 4 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection 2. Key elements of a targeting system There are several elements of a targeting system (Castañeda and Lindert, 2005; Grosh et al., 2008; Zapatero and Lopez, 2015), including: • Clearly defined eligibility rules depending on policy choices about who should be made eli- gible for support and program choices about geographic focus and/or individual/household eligibility. Appropriate mix of targeting tools, with an attempt to harmonize across programs and minimize duplication of costs and efforts where feasible. See Section 3; • Clearly defined institutional roles and responsibilities with respect to defining eligibility cri- teria, beneficiary registration, enrolment, and management. See Section 4; • Robust delivery systems including a) systems for data collection, household registration, and beneficiary database management (regardless of choice of targeting tool), combined with out- reach and communication efforts to minimize exclusion; and b) systems for monitoring and fraud control, grievance redressal and recertification. See Section 4. 3. Determining eligibility 3.1. The choice of targeting methods3 Various methods are used to prioritize specific areas and people (see Box 1). In practice, the choice of which method to use depends on patterns of poverty and vulnerability, specific program design, social norms and existing administrative capacity. Considerations of budgetary implica- tions as well as political and administrative feasibility are also often paramount for government in determining eligibility. Some methods confer eligibility on broad categories of people, such as all those living in desig- nated areas (geographic targeting) or those meeting a certain criterion such as on age or disabil- ity (categorical targeting). This is appropriate when clear spatial and/or demographic patterns in the distribution of poverty, vulnerability, and exclusion make it desirable and feasible to focus on specific areas or specific groups. • In many countries, geographic targeting is used for focusing resources in specific areas, large- ly because of its administrative simplicity, its relatively few data demands, and it is unlikely to result in stigmatization (as all households in an area are eligible). This approach is most appropriate when there are wide disparities across regions, when administration capacity is limited, or when transfer amounts are too low to make individual assessment methods sensi- ble. Geographic targeting is also more appropriate for programs delivering public or non-ex- cludable goods or services (see below). In particular, geographic targeting of post-disaster re- 3. This section draws on Grosh et al. (2008) and World Bank (2010a). Reaching the poor and vulnerable: Key considerations in designing targeting systems 5 lief has been very successful, largely because of the high concentration of target households and individuals in a well-defined geographical area. Good geographic targeting requires the use of high quality, up-to-date, and sufficiently disaggregated information on poverty and/or other related indicators (e.g. malnutrition, school enrolment, food insecurity, etc.). • Countries with wide disparities among groups in terms of living standards and human devel- opment outcomes may choose to focus on the groups that might otherwise be excluded or left behind. Categorical targeting is relatively administratively simple, has few data demands, is relatively low-cost, and carries the appeal of universality. This approach works best when a specific individual characteristic is highly correlated with poverty (or the indicator of choice) and is easily verifiable. However, if there are wide disparities within groups, other approaches are needed to track those most in need of assistance. For example, not all old people are poor. While some countries have social pensions for all old people above a certain age (e.g., Nepal, Sri Lanka, Thailand, and Vietnam), others overlay additional poverty and vulnerability criteria to focus only on those most in need. For instance, India’s social pensions focus on poor old people, who have no other sources of support. Box 1: Targeting methods Geographic targeting: Geographic targeting: People who live in the designated areas are eligible and those who live elsewhere are not Categorical targeting: Typically based on age (e.g. children and elderly) or on some other individual characteristic (e.g. disability) commonly perceived to be associated with a higher likelihood of being poor, vulnerable, or socially excluded Community-based targeting: A group of community members or leaders to decide whom in the com- munity should benefit Poverty scorecard/proxy means tests: Based on fairly easy-to-observe household characteristics be- lieved to be correlated with household welfare status Means tests: Based on information on households’ income and/or wealth. Not yet used in Myanmar Self-targeting: A program design feature that encourages greater take-up by the poor relative to the non-poor or the level of benefits is higher for the poor. Source: Grosh et al. (2008). Other methods assess eligibility for each individual or household based on welfare status and vulnerability. These include means tests (i.e. using reported income that may or may not be veri- fied), community-based targeting (i.e. relying on the local knowledge and perceptions of house- hold welfare among communities), and proxy means tests or poverty scorecards (i.e. using a set of observable multidimensional household characteristics that are highly correlated with poverty). • Means tests are the most demanding in terms of data requirements and administrative ca- pacity. These are typically used in countries with largely formal, highly monetized economies, high administrative capacity, and well-documented, reliable, and verifiable income data sys- tems (e.g. the US and OECD countries). • Proxy means tests (or poverty score cards)are typically used in countries with a high degree of informality and where household incomes are not readily quantifiable or measurable. This includes, for example, Chile, Colombia, India, Mexico, Pakistan, the Philippines, and Tur- 6 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection key, among others. This approach is most appropriate in the presence of reasonable adminis- trative capacity, for programs meant to address chronic poverty in stable situations and where they are used to target a single program with large benefits or to target several programs so as to maximize the return for a fixed overhead. Often, this is combined with community valida- tion to resolve disputed cases and to incorporate eligible families.4 • Community-based targeting relies on local information on household welfare and vulnera- bility.5 This might correctly identify the poor at the community level and lead to greater own- ership and accountability. However, the costs of facilitation may be high, and outcomes may be prone to elite capture and may exacerbate existing patterns of exclusion. In addition, compa- rability across space is problematic since these are measures of relative welfare within com- munities. This approach works best where communities are cohesive and for programs that cannot support large-scale administrative structures. Finally, self-targeting is used in conjunction with program design features that discourage par- ticipation of the non-needy. For instance, most public works programs (PWPs) rely on self-target- ing through the inherent work requirement and by setting the program wage lower than the mar- ket wage rate, making the program less desirable for the non-poor (Subbararo et al., 2013). However, in some cases where underemployment and vulnerability is high, PWPs use community- based targeting to ensure the poorest and most vulnerable are provided employment (see for example, Ethiopia’s Productive Safety Net Program (PSNP) and cash and food for work programs implemented by development partners in Myanmar). Many programs use a combination of methods. Typically, geographic targeting is used as a first stage to focus on the poorest and most vulnerable areas, followed by different methods to select households and/or individuals in a second stage.6 Using multiple methods can complement weak- ness in each method, therefore improving targeting outcomes (i.e., minimizing exclusion and in- clusion errors). A key challenge for the program and policy makers has been to identify the best targeting methods based on available data and country context. 4. Experience from other countries suggests community validation often results in the inclusion of households deemed vulnerable by the community. Rarely do community meetings result in the exclusion of households in the initial beneficiary list as this can be seen as divisive. El Salvador’s conditional cash transfer (CCT), for example, allows non-governmental organizations (NGOs) to challenge the inclusion of non-needy people, for program implementers to investigate (Grosh et al., 2008). Many other countries (e.g., Tanzania, Ghana. Niger, Cambodia, Indonesia, and Afghanistan) are also exploring a combination of community based targeting and proxy means testing (Leite, 2014). 5. In addition to poverty status, indicators of food insecurity are typically used to identify vulnerable households affected by shocks, for example in World Food Program (WFP) programs (Leite, 2014). 6. For examples, the following programs use geographic targeting and additional method(s) to select program beneficiaries. Mex- ico’s CCT and Kenya’s Orphans and Vulnerable Children program use geographic targeting and PMT; Brazil’s CCT use geographic targeting and means texting; Tanzania uses geographic targeting, community based targeting, and PMT (Leite 2014). Reaching the poor and vulnerable: Key considerations in designing targeting systems 7 3.2 Additional considerations in defining eligibility Program objectives Programs that provide community-level benefits tend to have broad eligibility criteria,but in some cases, also identify the poor or other marginalized groups to ensure their inclusion in over- all community development. Typically, these focus on relatively underdeveloped areas (e.g. as captured by measures of poverty, inadequate infrastructure, or social services, etc.). In some cas- es, these also target benefits to individual/households to ensure their specific needs are ad- dressed, particularly in the aftermath of disasters. For example, communities receiving block grants under Indonesia’s PNPM program have utilized these community funds to provide support for poor households as part of disaster response and recovery.7 Similarly, programs designed to promote equal access to health and education services by ad- dressing demand-side constraints tend to focus on specific groups that lack the resources to participate or might otherwise be excluded. These programs often complement supply-side ini- tiatives by governments to strengthen the delivery of health, nutrition, and education services for all. For instance, some countries (such as Cambodia, Bangladesh, Pakistan, and Turkey) have used stipends to reduce gender disparities in education; others (such as Chile, the Philippines, and Viet- nam) use stipends and cash transfers to reduce disparities between the poor and non-poor (Fisz- bein and Schady, 2009). In addition, user fee waivers for the use of health facilities and non-con- tributory social insurance programs also aim to provide financial protection to the poor (and near-poor in some cases) (World Bank, 2014). Programs that specifically address issues of poverty and food and income insecurity typically define eligibility narrowly and focus on groups and households most in need of support. This includes, for example, cash and in-kind transfers, and PWPs. Most cash and in-kind transfers tend to be either categorically targeted, for example social pensions for all older people above a certain age as noted above or poverty-targeted, for example transfers to poor households (as in Brazil’s Bolsa Família, India’s Public Distribution System, and the Philippines’s Pantawid Paimilyang Pilipi- no Program– see Annex 1). Finally, programs operating in two specific contexts – post-conflict settings and in the aftermath of disasters – are worth examining separately. The former often cover all households in the given area as part of the rehabilitation and peace-building effort. However, in many cases, additional support is provided to particular groups (e.g. ex-combatants) to encourage them to participate in the larger program. There are also several examples of programs focused exclusively on these groups and designed specifically to promote their reinsertion and reintegration (see, e.g., pro- grams for young ex-combatants in Burundi, Nepal, and Sierra Leone; Andrews and Kryeziu, 2013; Holmes, 2009, UNDP, 2012). With respect to the latter, immediately after a disaster, emergency relief programs tend to cover all households in the geographic area affected. However, in the re- covery phase and subsequent risk reduction phases, programs are typically scaled down to target 7. See Note on ‘Social protection delivery through community-driven development platforms: International experience and key con- siderations for Myanmar’. 8 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection the poorest and most affected households in the recovery phase (e.g. Pakistan earthquake pro- gram).8 Social norms The question of who should be provided support is deeply rooted in social norms and gender dynamics, at the national and local levels. For instance, many countries have social protection programs targeted to groups perceived as vulnerable or marginalized, such as old people, persons with disabilities, single-women headed households, and ethnic minorities, among others. In oth- ers, only poor households are considered eligible for support. Gender considerations need to in- form policy choices, particularly in contexts where gender disparities translate into conflicting options. For instance women and children can face important barriers to access services but intra- family gender roles may pose difficulties when targeting women as recipients of support. At the local level, acceptance of targeted support can vary across communities for various rea- sons. For instance, Indonesia’s Raskin program provides rice for the poor, but in practice rice has often been shared among the community, either at village level or by households themselves. This has been because many communities perceive everyone as poor and distribute the rice equally to avoid social conflict (SMERU, 2008). In general, prioritizing specific individuals and households is technical feasible and culturally appropriate in communities where there is some degree of in- equality and not everyone is considered poor, and where social trust is relatively high so that identifying those in need of support is not contentious. Political support and fiscal space In most countries, these considerations of who should be eligible for support have to be weighed against the need to build political support for programs and the availability of budgetary re- sources. For instance, while rigorous geographic targeting can help concentrate support to those most in need, political pressures for wider geographic coverage may lead to less effective out- comes. In order to garner political support, some programs may need to be implemented in a large number of locations, not just in those areas most in need. This dilutes the efficacy of geo- graphic targeting and often makes it necessary to add additional mechanisms for identifying ben- eficiary individuals/households.9 At the same time, programs benefiting broad categories of peo- ple can sometimes garner more broad-based support than narrowly targeted programs can. However, financial constraints may necessitate careful prioritization and sequencing to focus re- sources on those most in need within these groups, at least initially. Budgets for social protection programs are generally limited, especially in low-income countries, and there are trade-offs be- tween the number of beneficiaries, the geographic spread, and benefit levels. 8. See Note on ‘Social protection for disaster risk management: Opportunities for Myanmar’ and Note on ‘Framework for develop- ment of social protection systems’. 9. This is also a function of the quality of data, in terms of level of disaggregation: the higher the level of aggregation, the greater the possibility of heterogeneity within the unit. For instance, provincial poverty estimates would mask disparities in living standards across districts or villages and targeting resources at selected provinces would exclude needy villages in non-needy provinces. Reaching the poor and vulnerable: Key considerations in designing targeting systems 9 4 Institutional arrangements and delivery systems Depending on the policy choices above, systems to reach the eligible population need to be developed (often building on existing systems) and clear institutional roles and responsibilities need to be assigned across administrative levels and between government and non-government actors. The main delivery systems required include the following: • Data collection processes for enrolling households and individuals, including outreach and communication: These can include a door-to-door census to collect information from house- holds at the place of residence, an on-demand application process where individuals can pro- vide information at registration centers, or community meetings (for community-based target- ing). In some cases, this is a single stage process – people are enrolled in the program and beneficiary lists are generated at the time of data collection. In others, this is a two stage pro- cess – first, data is collected to determine eligibility for a single or multiple programs, and second, different programs follow separate processes to enroll beneficiaries on the basis of which program beneficiary lists are generated. • Beneficiary database management, monitoring and verification, grievance redress, and re- certification: This includes mechanisms to verify and monitor information used to select households, such as automated checks against administrative databases, citizen oversight, random audits, social worker visits, etc. Regular recertification is also important to reflect changes in household status owing to the transient nature of poverty. These systems are required regardless of the policy choices about geographic or individual/ household focus or of the targeting tool chosen. For instance, a CT program requires a list of ben- eficiaries for making payments and monitoring program performance, regardless of whether it is targeted to the poorest households or to all households in selected areas. The same applies to a PWP in selected areas that allows the poor to self-select into the program. The key considerations in setting up these systems include the following: a) transparency (peo- ple should be aware of their entitlements and the targeting mechanism for them to perceive the system as fair); b) dynamism (the system should be able to record changes in status and update eligibility accordingly); c) outreach (to ensure those eligible are not excluded); d) cost efficiency (administrative and social costs of the system relative to the benefits of ensuring limited resources reach the intended beneficiaries); and e) administrative feasibility and cultural acceptability (in alignment with current capacity at all administrative levels, and appropriate in the local context). This leads to the question of who should perform which function. There is no single ideal model with respect to institutional roles and responsibilities (see Table 1). For instance, decentralized design of the overall system can increase local involvement in social policy and be more suited to local preferences, while decentralized implementation can reap the benefit of local knowledge and empower local authorities to identify and include the poor. On the other hand, there are economies of scale in designing centralized systems; these foster comparability across space, re- 10 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection duce the risk of manipulation by local authorities, and ensure better quality control and consis- tency of the household database. Much depends on existing institutional structures with respect to program implementation and service provision (Castañeda and Lindert, 2005). Table 1 : Institutional arrangements for targeting Brazil, Chile Mexico Philippines Vietnam US Design of system Centralized Centralized Centralized Centralized Decentralized Data collection Decentralized Centralized Centralized Decentralized Decentralized (local authorities) (federally managed) Database Centralized Centralized Centralized Decentralized Decentralized management (national database) (national database) (national database) (no national (no national database) database) Source: Castañeda and Lindert (2005): MOLISA (2010); GoP (2013). It is critical is to ensure roles are clearly and appropriately defined. The assignment needs to be aligned with the overall mandate and capacity in terms of staffing, skills, material inputs, and fi- nancial resources, at each level.10 The appropriate assignment needs be determined through thor- ough deliberation of existing structures and mandates of government agencies and stakeholders. Once roles and responsibilities are assigned, effective mechanisms for information-sharing and coordination are required. In this context, it is essential to have strong champions who can facili- tate this cross-sectoral coordination, involving the key users of the targeting system. These users are usually the implementing agencies of social programs and have strong buy-in to use the target- ing system to select their beneficiaries. Finally, while most countries rely on program-specific approaches to identify beneficiaries, some countries have put in place integrated systems that can support various programs to verify eligi- bility for support. Some countries (e.g. Cambodia, Chile, Vietnam) first put in place a targeting system, with the stated intention of supporting multiple programs. Others (e.g. India, the Philip- pines) developed program-specific targeting systems, which, when found effective, were scaled up to verify eligibility for other programs. There are important economies of scale that can be ex- ploited by developing such core systems. In many countries, a common information base for geographic targeting is managed by a gov- ernment agency. The national statistics office or/and planning ministries in several countries con- duct population and other censuses and surveys on a regular basis. Based on these, statistics on poverty (e.g. the share and the number of poor people), social indicators (e.g. enrolment and dropout rates, malnutrition incidence, food insecurity), infrastructure development (e.g. road connectivity, access to markets), etc. are generated and released for public use. These statistics, particularly if available at disaggregated levels, are the primary source of information for geo- graphic targeting of poor areas (i.e. districts, township, city/municipality). This reduces duplication 10. See Note on ‘Institutional landscape for implementation and financing of social protection programs: Towards effective service delivery in Myanmar’. Reaching the poor and vulnerable: Key considerations in designing targeting systems 11 of efforts in obtaining information on where the poor and vulnerable live. In addition, some coun- tries identify a common set of ‘priority areas’ that are targeted by national programs, either for targeting of resources exclusively to these areas or to guide a phased rollout of national programs. Household targeting systems are more complex, with wide variation on how responsibilities are assigned to various actors. Several countries in Eastern Europe, Central Asia, and Latin America have introduced an integrated household registry with a backbone targeting method (see Annex 1). East and South Asian countries like the India, Pakistan, the Philippines, and Vietnam are at dif- ferent stages in moving to a more integrated approach. At a minimum, a household registry includes basic demographic information (for categorical targeting) as well as additional characteristics (for poverty targeting as relevant) which can in- form who are eligible for the programs. This helps reduce unnecessary and costly duplication of efforts on the part of the program implementers and potential confusion on the part of communi- ties and potential beneficiaries. Unique household and individual identifiers would also help track beneficiaries across programs to obtain a more comprehensive picture of overall support received and prevent errors and fraud. The backbone targeting method is often a means or proxy means test (i.e. poverty targeting), but this does not necessarily imply the imposition of a single targeting tool for all programs. For instance, India uses a common targeting tool (similar to a proxy means test) to determine poverty status for many government programs, but other methods co-exist, depending on program design and context: social pensions to poor older persons, persons with disabilities, and widows combine categorical targeting with the common poverty targeting tool, while the PWP uses self-selection (World Bank, 2010b). Some countries also adopt complimentary approaches, using both commu- nity based targeting and PMT. For examples, Tanzania uses geographic targeting, community based targeting, and PMT (Leite 2014). 5 Summary In summary, several governments have chosen to focus resources on poor and vulnerable areas and people, using a range of different approaches. The experience documented in this note re- flects pragmatic choices on whether or not to target the poor and vulnerable, and, if yes, how. The key considerations in determining eligibility criteria include the overall program objectives, the underlying patterns of poverty, vulnerability and exclusion, national and local social norms, and political, fiscal and administrative feasibility. Next, tool(s) to reach the eligible population need to be determined. These choices are typically driven by program design, technical feasibility in terms of data available, and existing administra- tive capacity, cost efficiency, and political feasibility. In many countries, geographic targeting is often used for focusing resources in specific areas – sometimes in isolation but more often supple- mented by categorical or household- or individual-based targeting. The latter require more re- 12 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection sources, notably trained facilitators for community-based targeting and fine-tuned data and suf- ficient administrative capacity for categorical or means or proxy means testing. Finally, these choices need to be operationalized through appropriate, transparent, and cost- effective systems and institutional arrangements. Several countries have sought to reduce frag- mentation and duplication of effort by developing integrated systems that can support various programs to verify eligibility for support. In developing a common system, it is critical to have a champion and to assign clear roles to partners. Common information base for geographic target- ing is more straightforward, and typically led by government statistical offices – it can be as simple as socioeconomic indicators with geographic disaggregates, to as sophisticated as poverty maps with highly disaggregated poverty estimates. On the contrary, household and/or individual target- ing system is far more complex and there is no single ideal model. These design and implementation choices need to be made carefully based on the consider- ations noted above, and with careful experimentation and evaluation, given the specific country context. Moving from program-specific to integrated systems often takes several years to a de- cade, and institutional arrangements need to evolve over time. The literature provides useful and detailed check lists on what governments need to consider in designing and operationalizing such systems (Castañeda and Lindert, 2005; Grosh et al., 2008; Zapatero and Lopez, 2015). This experience provides useful insights for Myanmar in considering options for strengthening geographic and household targeting of its recently introduced programs. Identifying poor and vulnerable areas and people needs to become more evidence-based, effective, and systematic, with a more harmonized approach across programs. In this context, Myanmar can take advantage of the solutions developed globally. A companion Note documents the targeting experience in Myanmar in order to provide inputs on potential ways forward to reach poor and vulnerable areas and people in Myanmar more effectively. Reaching the poor and vulnerable: Key considerations in designing targeting systems 13 References Alatas, V., Banerjee, A., Hanna, R., Olken, BA., and Tobias, J. (2010) ‘Targeting the Poor: Evidence from a Field Experiment in Indonesia’. Working Paper 15980. Cambridge, MA: NBER. Andrews, C., and Kryeziu, A. (2013) ‘Public Works and the Jobs Agenda: Pathways for Social Cohe- sion?’ Background Paper for the World Development Report 2013. Castañeda, T. and Lindert, K. (2005) ‘Designing and Implementing Household Targeting Systems: Lessons from Latin American and the United States’. Social Protection Discussion Paper 0526. Washington, DC: World Bank. Coady, D., Grosh, M., and Hoddinott, J. (2004) 'Targeting of Transfers in Developing Countries: Re- view of Lessons and Experience'. Washington, DC: World Bank. del Ninno, C. and Mills, B. (2015) ‘Safety Nets in Africa: Effective Mechanism to Reach the Poor and Most Vulnerable’. Washington, DC: World Bank. Fiszbein, A. and Schady, N. with Ferreira, F., Grosh, M., Kelleher, N., Olinto, P., and Skoufias, E. (2009) ‘Conditional Cash Transfers: Reducing Present and Future Poverty’. Washington, DC: World Bank. GoP (Government of Philippines) (2013) ‘ListahananOperational Guidelines’. Manila: Department of Social Welfare and Development. Grosh, M., del Ninno, C., Tesliuc, E., and Ouerghi, A. (2008) ‘For Protection and Promotion: The Design and Implementation of Effective Safety Nets’. Washington, DC: World Bank. Holmes, R. (2009) ‘Cash Transfers in Post-Conflict Contexts’. Project Briefing 32. London: ODI. Leite, P. (2014) ‘Effective Targeting for the Poor and Vulnerable’. Social Protection and Labor Tech- nical Note. June 2013 Number 6, Washington, DC: World Bank. MOLISA (Ministry of Labor, Invalids, and Social Affairs) (2010) ‘Guidelines for the 2010 Poverty Census’. Hanoi: MOLISA. Palma, J. and Urzúa, R. (2005) ‘Anti-Poverty Policies and Citizenry: The “Chile Solidario” Experi- ence’. Policy Paper 12. Paris: UNESCO. SMERU (2008) ‘The Effectiveness of the Raskin Program’. Research Report. Jakarta: SMERU Re- search Institute. Subbarao, K., del Ninno, C., Andrews, C., and Rodríguez-Alas, C. (2013) 'Public Works as a Safety Net: Design, Evidence, and Implementation'. Washington, DC: World Bank. Sudamo Sumarto (2015) ‘Indonesia’s Unified Database (UDB): Rationalizing a Fragmented System’. 14 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection Workshop on Integrated Data and Information Management for Social Protection: Bridging the Gaps between Theory and Practice, Jakarta, 11-12 March. UNDP (UN Development Programme) (2012) ‘Réintégration socio‐économique des rapatriés du camp de Mtabila: Document de programme’. Bujumbura: UNDP. Villalobos, V.S. (2012) ‘Monitoring and Information Systems – The Case of Chile’s Integrated Social Information System’. Presentation to World Bank, Washington, DC, 6 September. World Bank (2010a) ‘Regional Study on Targeting Systems and Practices’. New Delhi: South Asia, Human Development Unit, World Bank. World Bank (2010b) ‘Social Protection for a Changing India’. Volume II. New Delhi: World Bank. World Bank (2012) ‘Targeting the Poor Households in Indonesia’. Jakarta and Washington, DC: World Bank. World Bank (2013) 'State of Social Safety Nets 2014'. Washington, DC: World Bank. World Bank (2014) ‘Going Universal: How 24 Countries Are Implementing Universal Health Cover- age from the Bottom Up’. Washington, DC: World Bank. Zapatero,E., and LópezA. V. (upcoming) ‘Discussion Paper: Operationalizing Targeting’ Social Pro- tection Discussion Paper. Washington, DC: World Bank.   Reaching the poor and vulnerable: Key considerations in designing targeting systems 15 Annex 1: Selected international experience Chile Chile is known for having a highly sophisticated social protection system. The expansion of its so- cial protection and poverty alleviation program started as a part of a ‘growth with equity’ strategy in 1990. The system has evolved gradually over the past 25 years, from a package of traditional subsidies to a coherent family-based package of interventions. Chile Solidario (CS), introduced in 2002, was among the first initiatives to offer families a package of necessary interventions instead of designing separate programs for specific groups (Palma and Urzúa, 2005). This further evolved into the Ethical Family Income Program (IEF) in 2012, adopting variable transfers depending on household socioeconomic status. A core element of the social protection system is the single registry and an integrated manage- ment information system (IMIS), known in Chile as the Registry of Social Information (RIS) and the Integrated System for Social Information (SIIS), respectively. These are often referred to as the most advanced example of integrated data management in the world. The RIS consists of informa- tion on vulnerable households – that is, both potential and actual program beneficiaries. After collecting household information, data are entered into the IMIS to estimate the welfare level of each household based on a proxy means test (PMT). Consequently, the RSI assigns unique identi- fiers to individuals, thus serving multiple programs that can have different eligibility criteria and units of intervention (e.g. household vs. family). The database is also linked to the civil registry. Nearly 80 percent of the national population is included in the registry (Villalobos, 2012). By 2012, 60 social programs and 56 social services/benefits were using the registry information to select their beneficiaries, and about 40 public and private agencies were exchanging beneficiary infor- mation. CS covered about two-thirds of the poor population and more than double the number of the extreme poor as of 2009.11 The number of CS beneficiary families had nearly doubled in 2012, to 600,000. IEF has further expanded coverage, with a stronger focus on the exit strategy through the higher monetary incentives linked to achievements in the areas of education and employ- ment. The CS blueprint was drawn up under a central government initiative (of the National Budget Of- fice and the Ministry of Finance), with municipalities the executors, including delivery of pro- grams/services to beneficiary families and collection of family data for the RIS. The promulgation of several laws demonstrates Chile’s political commitment to a strong social protection system, including identifying and managing program beneficiaries. It explains the op- erational principles, scope, and modes, and in particular establishes those benefits that must be assigned to people by right. This differentiates the system from traditional ways of assigning ben- efits through applications and waiting lists.12 11. See Note on ‘Institutional landscape for implementation and financing of social protection programs: Towards effective service delivery in Myanmar’. 12. http://www.chilesolidario.gob.cl/sist/sist1.php 16 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection Several lessons emerge from Chile’s experience in developing this system, including a) separation of roles: the administrator that plans and manages the system should not be sectoral or the ex- ecutor of programs; b) design and management need to be done centrally, whereas implementa- tion is decentralized; c) coordination and sharing information are a part of cultural change, in that they overcome personal and institutional resistance to do so; 4) the system needs to be developed in phases, to allow for coordination between modules and to build acceptance; and 5) a unique personal identifier, if it does not exist, is vital. Indonesia Indonesia has committed itself over the past decade to reforming and integrating social assistance programs as part of its poverty reduction strategy, coupled with fuel subsidy reform. Continuous reform has decreased the fuel subsidy over time, leading to reallocation of its saving towards so- cial assistance that is more targeted to the poor and vulnerable. The National Team for the Accel- eration of Poverty Reduction (TNP2K) in the Office of the Vice-President (established in 2010) manages the Unified Database (UDB) for Social Protection Programs. The Central Bureau of Statis- tics (BPS) also plays an instrumental role, collecting and developing information for geographic and household targeting. During this reform process, one of the biggest challenges has been how to reach poor and vulner- able households, which were the most affected by cuts to the general subsidy. Multiple poverty reduction programs already existed, which used different criteria and data sources to determine beneficiaries. There were also discrepancies between design and actual implementation: in prac- tice, for the most part village leaders nominated potential beneficiaries in the absence of a clear basis of information. Consequently, programs were badly targeted, and there was duplication of effort and a lack of consistency. The first effort at consolidating the list of poor households (Pendattan Sosial Ekonomi) started in 2005, when the reduction in fuel subsidies led the government to develop a temporary uncondi- tional cash transfer (BLT) aimed at all poor households. The list was updated in 2008, with signifi- cant improvements introduced in 2011 (called Pendataan Program Perlindungan Sosial (PPPS) 11). PPPS11 covered a larger number of households as well as a broader range of demographic infor- mation for more rigorous household targeting (World Bank, 2012). The latest database (updated every three years) contains 25 million interviewed households (96 million individuals, 40 percent of the population) selected based on the Census, triangulated with other data sources,13 program listing, and local knowledge. UDB employs PMT approaches to estimate the welfare level of inter- viewed households, based on socioeconomic information that is highly correlated with poverty status (based on an analysis of household survey data). A study was conducted to compare the targeting performance of three approaches, notably community-based targeting, PMT, and the hybrid. Findings suggest PMT performs better in identifying the poor particularly near thresholds,whereas community-based strategies do as well in capturing the very poor and result in higher satisfaction (Alatas et al., 2010). 13. Village potential statistics 2010, Susenas 2010, and 2008 Pendataan Program Perlindungan Social. Reaching the poor and vulnerable: Key considerations in designing targeting systems 17 The UDB has become the only source of data for poverty targeting in the country, and is currently used by five major national programs – BLT, health insurance (Jamkesma), the rice subsidy (Raskin), the scholarship program (BSM), and the CCT (PKH) – to select beneficiaries. In addition, data have been shared with 350 local governments for their own planning and program purposes. The cost of building the UDB is estimated to be around 1 percent of the combined annual budget of three main programs (BLT, Jamkesmas, and Raskin) (World Bank, 2012). The annual cost of updating data is around USD 0.60 per registered household (Sudamo Sumarto, 2015). The generation of an information base for geographic and household targeting is built on the strong capacity of the BPS, in terms of both collecting data and generating statistics. The BPS con- ducts a set of regular surveys and produces regular official statistics, including poverty maps that provide geographic information on which sub-districts (kecamatan) are in greater need of assis- tance. The design and development of the household targeting system relied heavily on these existing surveys/statistics, including identification of households to be surveyed drawing on PMT models to estimate their consumption. Indonesia is taking a further step to integrate the government database across ministries. For ex- ample, three-quarters of households in the UDB have been already matched with the citizen iden- tity number managed by the Ministry of Home Affairs, enabling the design of social security cards. Efforts continue towards e-government to ensure the interoperability of government sector sys- tems, including an IMIS for the social protection sector. Philippines Under the social protection reform agenda, the Philippines launched the CCT (the Pantawid Paim- ilyang Pilipino Program) in 2008. Implemented by the Department of Social Welfare and Develop- ment (DSWD), the program is implemented as a concerted effort with other departments, the Departments of Health and Education in particular, as well as in close coordination with local gov- ernment units, which supports its implementation. Through a series of expansions and evolution, Pantawid Paimilya has become one of the flagship programs in the Philippines, shifting its prac- tices and norms to prioritize and serve the poor in the country. Pantawid Pamilya was launched as a program to reach out to the poorest of the poor, notably the poorest households residing in the poorest areas in the country. The program was rolled out in phases, starting from the poorest geographic areas and expanding to national coverage. It first started in provinces and municipalities with the highest to higher poverty incidence, which the Philippines Statistics Authority regularly generates as official statistics. In identifying beneficiaries, a household assessment was conducted using a simple questionnaire, in order to estimate eco- nomic situation based on demographic and socioeconomic characteristics, such as education lev- el, assets, housing materials, and access to basic services (i.e. the PMT method). In 2009, successful implementation of Pantawid Pamilya and household assessment prompted the government to institutionalize and adopt the targeting system (Listahanan) as a mechanism to identify and prioritize beneficiaries of social assistance programs of the government. The National Household Targeting Office was officially established within the DSWD in 2009. About 11 million households were surveyed during the first round of Listahanan in 2008-2009, out of which 5.2 mil- 18 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection lion households were identified as poor based on the PMT. Listahanan updates the list of the poor every four years; the second round of household assessment is currently underway, with a budget allocation of about PHP2 billion (USD 45 million). Listahanan data has been widely shared with government agencies, local governments, non-gov- ernmental organizations (NGOs), and international organizations, which use the information of the poor to plan and implement their programs. As of 2014, seven national government programs select their beneficiaries using Listahanan, including Pantawid Pamilya and the PhilHealth indigent program, which cover 80 percent and 100 percent of poor households identified under Listahan- an, respectively. In sum, the government of the Philippines has successfully rolled out major national social protec- tion programs (i.e. Pantawid Pamilya and the PhilHealth indigent program) to reach the poor. Lis- tahanan has been a critical instrument for its transparent and objective identification of poor beneficiary households. Previously, numerous small programs were implemented in a fragmented manner, including the process and criteria for beneficiary selection, which were being vulnerable to governance and political risks. Currently, seven national programs use Listahanan to select its beneficiaries, and Listahanan data have been shared with 1,200 entities, including local govern- ments, civil society, NGOs, and development partners. The government continues with efforts to reach out to those not covered by Listahanan despite being poor (so-called exclusion errors) through improving the design and implementation of Lis- tahanan and linking it to other social protection programs. For example, the DSWD extends vari- ous interventions to poor and vulnerable populations, including homeless and street families and indigenous peoples in remote areas. For the PhilHealth indigent program, the poor identified by Listahanan are covered by the national budget; local governments can fund additional beneficia- ries verified as poor based on means. Reaching the poor and vulnerable: Key considerations in designing targeting systems 19 20 Building Resilience, Equity and Opportunity in Myanmar: The Role of Social Protection 'Reaching the poor and vulnerable: Key considerations in designing targeting systems' is the tenth note in the series Building Resilience, Equity and Opportunity in Myanmar: the Role of Social Protection. All notes are available at www.worldbank.org/myanmar. Myanmar Social Protection Notes Series The note – ‘Building resilience, equity, and opportunity in Myanmar: The role of social protection’ – provides an overview of the technical notes in the se- ries. These include: 1. Risks and vulnerabilities along the lifecycle: Role for social protection in Myanmar 2. Framework for the development of social protection systems: Lessons from international experience 3. Inventory of social protection programs in Myanmar 4. The experience of public works programs in Myanmar: Lessons from a social protection and poverty reduction perspective 5. The experience of cash transfers in Myanmar: Lessons from a social protection and poverty reduction perspective 6. Social protection for disaster risk management: Opportunities for Myanmar 7. Strengthening social security provision in Myanmar 8. Institutional landscape for implementation and financing of social protection programs: Towards effective service delivery in Myanmar 9. Social protection delivery through community-driven development platforms: International experience and key considerations for Myanmar 10. Reaching the poor and vulnerable: Key considerations in designing targeting systems 11. Reaching the poor and vulnerable in Myanmar: Lessons from a social protection and poverty reduction perspective 12. Developing scalable and transparent benefit payment systems in Myanmar World Bank Office Yangon 57 Pyay Road, Corner of Shwe Hinthar Road, 6 1/2 Mile, Hlaing Township, Yangon Republic of the Union of Myanmar. Tel: +95 1 654824 www.worldbank.org/myanmar www.facebook.com/WorldBankMyanmar