Social Safety Nets in Bangladesh. Shaikh S. Ahmed

Transcription

1 Social Safety Nets in Bangladesh Shaikh S. Ahmed October 2007

2 I. INTRODUCTION 1. The objective of this paper is to address issues relating to social safety net programs in Bangladesh. Although such programs have gained momentum in the country, at least the following factors need to be addressed to improve the state of social safety nets: a) The focus of the safety nets in Bangladesh needs to shift to new and emerging challenges. These challenges include rapid urbanization, globalization, export orientation and structural changes (e.g. layoffs, increase in prices, etc.). The national policy should also emphasize programs for mitigating ex-ante risks, all in a financially sustainable way. A well-designed policy for social protection programs should address various gaps including safety nets for urban areas and link programs to human development. b) Bangladesh needs a quick social safety nets response system to disaster management. This is because the country is prone to natural calamities and responses at present are reactive to these catastrophes. There is no quick social safety net response mechanism with respect to these recurring natural calamities. Safety net programs should also address chronic poverty and help poor people escape poverty. c) Food transfer programs to be replaced by cash transfer programs which have cheaper cost of delivery. This is evidenced that the cost of transferring food is higher than transferring cash benefits and the beneficiaries mostly prefer cash to kind. d) While the Government of Bangladesh has been increasing spending in social safety nets over time, these initiatives need to target the poorest regions and individuals better. Though targeting efficiency is not performing very poor at present, this could possibly improve if objective targeting criteria are mixed with community verification. Better targeting mechanism could also help increase both coverage and level of benefits for the poor. e) Implementation and administration of the programs also need improvement. This is especially important with respect to coordination among ministries, consolidation of various programs over time, improving delivery mechanisms by transferring directly to the beneficiaries, and strengthening local government capacity. 2. Rest of the paper is organized into eight sections providing analysis in support of the issues mentioned above. Section two discusses the state of poverty in Bangladesh and section three provides an overview of the main social protection programs. Section four covers financial allocations and related issue of regional disparities. Section five discusses performance of such programs. Specific issues such as targeting, leakages and preference to cash over food are covered in section six. Section seven highlights the need for better program administration. Specific conclusions are drawn in the final section. II. THE STATE OF POVERTY IN BANGLADESH 3. In recent years, there has been a substantial reduction in the number of people living under both the upper and lower poverty lines. The national poverty rate has declined at a rate of roughly one percentage point per annum over the last 15 years (see Table 1 below). Based on the upper poverty line, 40% of the national population is poor, which amounts to about 56 million 1

3 people. While this figure represents a considerable improvement over previous levels, it is still extremely high by international standards and reflects the widespread poverty and vulnerability in Bangladesh. It also draws attention to the fact that several Millennium Development Goals (MDGs) are yet to be achieved. Table 1: Poverty Distribution among Rural and Urban Populations Year Upper Poverty Line Lower Poverty Line Rural Urban National Rural Urban National Source: BBS, Household Income and Expenditure Survey, various years. 4. With rural poverty remaining higher, the plight of the poor urban population is often ignored. Based on the last four censuses, urban population in Bangladesh is growing, and is expected to keep its pace in the future. According to the Population Census 2001, 23.1% of the population was living in urban areas. 34.3% of the population of Dhaka is urban, followed by Chittagong, Khulna, Rajshahi, Barisal and Sylhet. The total urban population in Dhaka is currently around 12 million and is projected to grow to 20 million in 2020, making it the world s third largest city. Among the three divisions, where 20% or more of the population is urban, the poverty situation remains uncomfortable. The average percentage of extreme (food) poverty is 20% and overall poverty is 40% in these three cities. Khulna has the highest poverty headcount. This is depicted in the following graph: Poverty and Extreme Poverty Headcount, Urban Bangladesh % individuals living in poverty Extreme (food) Poverty Poverty 50 Dhaka Chittagong Khulna Spatial analysis, in a recent study of the World Bank (2007), shows an estimated 1,925 slums in the Dhaka City Corporation area. Based on the LGED Survey of Slums (2005), among the poorest quintile, only 9% of households have a sewage line and 27% obtain water through piped supply. Furthermore, the study shows that only 43 of the slums are within 100 meters of a public toilet. An estimated 7,600 households live in slums that are within 50 meters of the river and are at frequent risk of being flooded. 5. The urban poor are substantially less educated, and have lower school attendance rates than the non-poor. Only 40% of household heads among the poor have achieved more than 5 years of schooling, and only 6% have had more than 10 years (HIES). In four sample slums of the LGED study, more than half of the household members are under the age of 19 (56%). Literacy rates for workers aged 15 and older are only 52% among poor male workers and 33% for female workers. The recent DHS (2004) survey finds that, on average, 16.7% of poor 2

4 children suffer severe stunting. Stunting occurs among 17.4% for rural poor children and 13.5% for urban poor children. Nationally, 33.2% of women are malnourished, with 35.7% in rural and 24.6% urban localities. The urban situation aggravates when the lowest quintile of wealth is considered. 30.6% of urban children of the lowest quintile suffer from severe stunting which is higher than 24.9% of rural children who are severely stunted. In the lowest quintile, 46.2% of rural women and 44.7% of urban women suffer malnutrition. 6. The urban poor are not specifically targeted by any of the existing social safety net programs. In the Government s policy matrix for poverty reduction, the issue of improving living environments in slums is mentioned without any specific strategy or action plan. It is globally observed that conditional cash transfers are useful in overcoming urban poverty. Safety nets in Bangladesh may also have to ensure that the poor receive a basic minimum assistance but can also graduate from poverty, using innovative approaches such as conditional cash transfers. III. OVERVIEW OF SAFETY NET PROGRAMS 7. Bangladesh has a wide spectrum of social safety net programs. There are 30 specifically designed social safety net programs directly operated by the Government of Bangladesh. In addition to these, there are 15 funds to provide further assistance to improve the overall condition of the poor. Appendices A and B list these programs including their purposes, targeting criteria, financing, nature of benefit and budgetary allocation. Among the existing 30 programs, there are 8 unconditional programs of which 4 are food transfers and 4 are cash transfers. 2 of these food based programs are generally used in times of natural disasters or seasonal downturns. There are also 10 conditional programs, comprising of 7 cash transfers and 3 food transfers. Bangladesh also has 5 credit schemes and 3 conditional subsidy programs in addition to 4 funds. Several programs have been discontinued over time or replaced by newer programs, e.g. the Women Support Center or the Rural Maintenance Program. 8. At least a number of 13 ministries are engaged in the planning and implementation of these projects. These include: Ministry of Social Welfare; Ministry of Women and Child Affairs; Ministry of Liberation Affairs; Ministry of Food and Disaster Management; Ministry of Local Government, Ministry of Agriculture; Ministry of Fisheries and Livestock; Ministry of Chittagong Hill Tracts Affairs; Ministry of Education; Ministry of Primary and Mass Education; Ministry of Finance; Ministry of Health and Family Planning; Ministry of Commerce; and Ministry of Labor and Manpower. In addition to these ministries, the Bangladesh Bank and Palli Karma-Sahayak Foundation (PKSF) are also operating social safety net programs. 9. Involvement of multiple ministries cause considerable overlap in programs and administration which stretch the administrative and resource capacity of the country. For example, the Ministry of Women and Children Affairs offers the Maternity Allowance for the Poor Lactating Mothers and the Ministry of Health offers Maternal Health Vouchers. Various education stipend programs are run by three different ministries. There are similar cases when emergency or disaster related programs are considered. This implies that there is very little coordination among the ministries in planning, targeting and implementing the various social protection programs in the country. 10. Most of the programs are ex-post coping initiatives, with few being ex-ante preventative programs. High impetus on ex-post programs represents a weak social safety net policy in Bangladesh. This is because ex-post programs allow risks to occur, whereas 3

5 efficient ex-ante policy calls for protection before risks appear. Most countries at Bangladesh s level of income do not have ex-ante programs given the small size of the formal sector and limited administrative capacity. In general these programs are limited to the formal sector (civil service pensions, etc.). However, the extensive coverage of microfinance in Bangladesh provides an interesting base from which to explore ways of expanding coverage to the informal sector, which the government is now planning. As new programs are needed to address ex-ante risk management, focus could be given to various insurance policies, such as expanded pensions, health insurance, unemployment insurance, catastrophic risk and natural calamities insurance, etc. 11. Existing social safety net programs do not provide a sufficient basis to cope with the magnitude of extreme poverty. Table 2 shows that on average, 39% of eligible households are covered by different programs, therefore about 60% of eligible households are yet to benefit from a suitable program. This is elaborated further in Table 3. Table 2: Opinion of Respondents about the Number of Families under the Program and Number of Eligible families by Program Type Program Type No. of Families Under the Program No. of Eligible families % Covered Asrayan Housing FEE RMP Old Age Allowance Widowed Allowance VGD All Programs Source: BIDS, Impact Evaluation of Social Safety Nets Initiatives, Final Report, April Table 3: Number of Hardcore Poor covered by different Safety Net Programs Program Type No. of Households Estimated Population Average Family Size Covered Covered Asrayan 34, ,854 Housing 28, ,330 FEE 41, ,880 RMP 2,020, ,517,762 Old Age Allowance 500, ,500,000 Widowed Allowance 403, ,813,995 VGD 201, ,687 All Programs 3,230, ,208,508 Source: BIDS, Impact Evaluation of Social Safety Nets Initiatives, Final Report, April IV. FINANCING OVERALL AND REGIONAL ALLOCATION 12. Historically, expenditure for social safety net has been less than 1% of GDP till the late 1990s. However, this spending has been increasing in recent years, taking advantage of an annual economic growth of about 5%. Social safety net expenditure was 1.21% of GDP during and about 1.60% of GDP for the fiscal year. There were only a few programs until the mid-1990s, and these were designed only for income support, but from the second half of the 1990s new program types have been initiated almost every year. 4

6 Program 13. The Government of Bangladesh has increased budgetary allocation and identified various types of vulnerability in striving to improve social safety net during the current fiscal year. For the fiscal year , a total of Taka 8, crore has been allocated. This value constitutes of Taka 6, crore in the non-development budget and Taka 1, crore in the development budget, which are 14.38% of the non-development budget and 5.54% of the development budget. The total value for social protection constitutes 10.58% of the total budget which is 1.60% of GDP. Compared to the previous year, this is 0.40% point-to-point increase. Such an increase in allocation is a direct outcome of the recent democratic form of government in Bangladesh. The government has been reflecting on the political economy of the country and various international conventions for about a decade where the issue for poverty alleviation received special attention. The increasing and stable rate of economic growth and the demand for equitable distribution has also played an important role in the budgetary increase for social safety net. 14. Social safety net remains an important focus for the government. As a trend, though almost all programs have been receiving incremental financial allocation, some of these programs show phenomenal periodic progress with respect to coverage. Table 4 below provides break down of a sample of major programs. There are increases in the size of benefits per beneficiary as well. For example, depicted in the graph below, the coverage and benefit for the Old Age Allowance has increased more than four times in 10 years. The value for such programs however has not increased much in real terms. But the size of allowances for these programs is small, for example Taka 220 or 30kg of wheat per person. This allowance does not reflect the required minimum cost of living. If the wholesale price per kilogram of wheat is Taka 15 (FPMU, Oct 06), the value of a monthly amount of 30kg of VGD wheat per beneficiary is Taka 450. This is more than double the amount of any unconditional cash transfer program s monthly benefit. The government is moving away from this differential policy into a policy of more cash transfer programs than food transfer programs. For example, Food For Education program (FFE) has been completely discontinued and currently Cash For Education is in operation. Similarly the Cash For Work is gradually replacing the Food For Work (FFW) program. Table 4: Social Safety Net Programs with Incremental Financial Allocations Old Age Allowance 0 1,250 5,000 5,000 5,000 5,000 7,500 18,000 26,037 31,500 38,400 44,880 Fund for Rehabilitation of Acid Burnt Women & Physically Handicapped Allowance for InsolventRetarded Disabled Persons Capitation Grant for Non-Government Orphanages Subsistence Allowances for Government Orphanages Educational Stipend for the Disabled Students Allowance for Widow & Distressed Women Vulnerable Group Development (VGD) Maternity Allowance for the Poor Mother Primary Education Stipend Project (PESP) ?? 4,350 2,500 6,600 1,500 2,500 2,500 2, , ,500 4,000 5, ,000 1,050 1,110 1,150 1,210 1,400 1,600 1,700 1,900 3, ,111 1,111 1,111 1,245 1,242 1,185 1,255 1,235 1,635 1,590 1, ??? 2,500 2,500 4,000 9,000 11,880 13,500 15,600 21, , ,089 2,280 2,360 2,430 2, , ,095 32,325 36,544 40, , ,200 10,000 60,000 43,395 52,000 44,331 46,800 46,800 Reach out Education Children Project ,788 10,489 11,200 Test Relief (TR) 10,500 13, ,753 11,194 14,523 15,396 15,016 23,562 24,878 25,831 30,282 Gratuitous Relief (GR) 11,050 6, ,733 2,914 2,905 3,079 5,256 10,053 10,614 11,021 12,920 Vulnerable Group Feeding (VGF) 19,966 21, ,960 21,850 21,785 23,094 22, ,080 24,878 43,050 80,

7 Food for Work (FFW) (Khabikha) ,023 37,367 37,346 Cash for Work (Kabita) ,200 Fund for Mitigation of Risk of Natural Disaster (pre & post) Agricultural Subsidy and Fuel Price Support for Marginal Farmers Honorarium for Insolvent Freedom Fighters Allowance for Injured Freedom Fighters ,000 60, , , ,500 2,880 3,000 3,000 3,600 4,200 6,000 7, ,451 1,625 1,625 1,625 1,625 1,644 1,519 1,819 2,730 Maternal Health Voucher Scheme ,510 1,760 Area-Based Community Nutrition ,530 15,800 Special Fund for Employment Generation for the Extreme Poor Fund for Reduction of Temporary Unemployment ,000 6,882 5,000 10, Housing Program for Homeless ,000 3, ,000 Fund for Disaster Affected Marginal ,000 Farmer and Poultry Farm Owner Source: Ministry of Finance. Growth of Old-Age Allowance Allowance in Taka per Beneficiary (Nominal and Real) Year Per Beneficiary (Tk.) Real Per Beneficiary No. Beneficiary 1,800,000 1,600,000 1,400,000 1,200,000 1,000, , , , ,000 0 Total Number of Beneficiary Source: Ministry of Finance, Annual Budget, various years. 15. Programs have little or no correspondence to regional variations in poverty. Among the six divisions, Sylhet has done significantly better in reducing the poverty incidence than Khulna, Barisal and Rajshahi divisions recently. As observed in the Table 5 below, the incidence of poverty in Khulna increased over time and remained stagnant in Barisal. However, the distribution of safety net programs by region does not reflect this point since Sylhet has the highest share of beneficiary households and participation, whereas Khulna deserves highest allocation. If there was any correlation between poverty and program participation, distribution and participation would have been skewed towards Khulna and Barisal divisions. Table 5: Poverty Headcount, Distribution of Beneficiary Households, and Participation, by Division Poverty Headcount Distribution of Beneficiary Households Participation Division Total Rural Urban 2005 Dhaka Barisal

8 Chittagong Khulna Rajshahi Sylhet National Source: BBS, Household Income and Expenditure Survey, 2000 and This discrepancy is further visible in Tables 6 through Table 8 below. There is significant variation in participation among the 6 divisions. From among the poorest 10% of households, about 41% participants are in Sylhet followed by another 28% from Chittagong. Nationally, this accounts for about 21% of the people in the lowest decile. Table 6: Participation by Division and Wealth Status in the Targeted Programs by Income Group Division Lowest Lowest Quintile 2 Quintile 3 Quintile 4 Quintile 5 Total Quintile Decile Barisal Chittagong Dhaka Khulna Rajshahi Sylhet Total Source: BBS, Household Income and Expenditure Survey, Out of 13.02% of households receiving social safety net benefits, 15.54% are rural households followed by 5.45% urban households. Sylhet receives the largest share of benefits by 24.31% for rural beneficiaries and 11.25% for urban beneficiaries. In contrast, Khulna receives the lowest by 11.03% and 4.23% respectively for rural and urban households. This is elaborated in Table 7: Table 7: Distribution of Households Receiving Benefits from Social Safety Net Programs, 2005 Division % of Households Receiving Benefit Total Rural Urban Barisal Chittagong Dhaka Khulna Rajshahi Sylhet National Source: BBS, Preliminary Report on Household Income and Expenditure Survey, The highest number of beneficiaries is in the VGF program which is 33.34% of the total Sylhet having the greatest number of VGF beneficiaries. This is followed by the VGD program which provides assistance to 17.11% households, with Dhaka having the most VGD beneficiaries (see Table 8 below). This data indicates that programs have certain political dimensions when Table 7 is also taken into consideration. Table 8: Distribution of Households by Program and Division,

9 Program Total Barisal Chittagong Dhaka Khulna Rajshahi Sylhet VGD IFS FFWCFW TR VGF GR FFECFE RMP Old Age FF Other % All HH Source: BBS, Preliminary Report on Household Income and Expenditure Survey, It is therefore imperative that a clear strategy is needed, whereby programs should focus regions on the basis of relative poverty of the region. The logical outcome of such a strategy would be that the poorer regions will benefit more from these programs. V. PERFORMANCE OF THE PROGRAMS 20. Programs other than education look at income poverty and target the poor, e.g. VGD, VGF, Old Age Allowance, etc. In the case of educational programs, however, the targeting system sets performance as its main criteria. For example, in the PESP program, school attendance is a condition for receiving the stipend. Therefore, children from hardcore poor households face severe difficulties in participating in the program. Such issues challenge the effectiveness of current targeting mechanisms. 21. The Government acknowledges that all targeted programs do not reach the hardcore poor. Microfinance programs are examples in this case. It is also that a large part of the total budget allocation is termed as social safety net. But if we separate out the values of 8 microfinance and loans and 4 funds from Appendix A and 15 funds in Appendix B, the size of direct social safety net allocation drops significantly. 22. Generally, targeting mechanisms are homogeneous among all programs where benefits are transferred to individual beneficiaries. The basic targeting methodology starts with a general guideline prepared by the implementing ministry. A certain quota is allocated in the guideline, which includes the total number of beneficiaries, the number of beneficiaries per union, and the amount of transfer per beneficiary. The quota also caps the number of male and female beneficiaries per union or ward. The guideline mentions certain targeting criteria according to the theme of the program. Usually these criteria include income level, asset and household structure, and demographic features. Based on such criteria, union parishads, in consultation with other local agencies and community, identify beneficiaries. 23. Three sets of targeting criteria are presented in the following Table 9. None of these targeting approaches require that all the criteria have to be fulfilled by an individual, but in case of most selected beneficiaries, only one of these criteria can be observed. Surveys have shown fulfillment of all criteria by an individual beneficiary is rare. Table 10 shows three programs along with their beneficiary selection criteria and number of sample complying with the given criteria. 8

10 Table 9: Comparative Selection Criteria IGVGD PESP RMP Criteria: 1. Destitute woman headed family 1. Households with not more than 15 (destitute means widowed, acres of land. separated from husband and 2. Monthly household income less divorced). than Tk Principal occupation of the 3. Dependent upon seasonal wage household head is day-labor. employment. 3. Family of low income 4. Women of reproductive (18-49) age. professionals (such as: fishing, pottery, blacksmithing, weaving, 5. Day labor or temporary worker. and cobbling). 6. Lack of productive assets. 4. Landless or households that own 0.50 acres of land (marginal or share-cropper). Prerequisites for selection: 1. VGD member must be a permanent resident of the area. 2. Women of reproductive age. 3. Not associated with any NGO. 4. Did not get VGD card before. 5. One card for one family. Priorities: 1. Female-headed household. 2. Those who are mentally and physically capable of undertaking IGA activities. Source: Project Proposal, Income Generation for Vulnerable Group Development Program (IGVGD), January 2003 December 2004, BRAC. Source: Project Summery, Stipend for Primary Education Project, Financial Year, Department of Primary and Mass Education, Government of the People s Republic of Bangladesh, July 2002 (Bangla Version). 1. Must be a permanent woman resident of the union. 2. Must be either destitute, widowed, divorced, separated from husband (not less than one year), who are household head but does not have alternative means of income. 3. Must be between years of age. 4. Those women who have higher number of dependent will be given priority (dependant means children less than twelve years of age, disabled children and husband and dependant parents). 5. Must have the physical and mental capacity to do laborious works. 6. Those who have been working for the RMP for less than one years and never been fired and capable of satisfying the above criteria will be automatically selected. 7. Not a member of VGD or any other targeted program. Source: Road Maintenance Management Training Manual, RMP, Care-Bangladesh, December 2001, Third Edition, Third Reprint, June 2003 (Bangla Version). Note: This program has recently been replaced by a similar program named REOPA. Table 10: Fulfilling Selection Criteria IGVGD Criteria PESP Criteria RMP Criteria 1. Households with not more than 0.15 decimals of land 2. Monthly household income less than Tk Destitute woman headed family Widowed Must be a permanent woman resident of the union. 2. Does not have alternative means of income and: 3. Seasonal wage employment 7.30 Separated from husband 0.20 Widowed Women of reproductive age Divorced 0.50 Divorced Day labor or temporary worker Principal occupation is day-labor. 7.7 Separated Lack of productive assets Family of low income professions Household head Landless or owns 0.50 acre of land Must be between years Permanent resident of the area Has dependents: 2. Women of reproductive age Children less than twelve years of age 3. Not associated with any NGO Disabled children Did not get VGD card before One card for one family (in this case more than one card) Female headed household Disabled husband and dependant parents 5. Physical and mental capacity to laborious works. 6. Working for the RMP for less than one year. 7. Not a member of VGD or any other targeted program

11 2. Mentally and physically capable of undertaking IGA activities Lowest Quintile 2nd Quintile 3rd Quintile 4th Quintile Top Quintile Source: Ahmed S.S. (2005), Delivery Mechanism of Cash Transfer Programs to the Poor in Bangladesh, Social Protection Discussion Paper 0520, The World Bank. 24. Criteria used for targeting are useful. It is observed that 84% of beneficiaries do not own any land. At the same time 93% non-beneficiaries do not own any land. The proportion of beneficiaries having no land varies form 58% to 98 for Housing Fund and VGD respectively. Therefore, only landholding is not a sufficient criterion, especially in reaching the hardcore poor. This is presented in Table 11. Table 11: Success of Safety Nets Programs in Targeting the Poor: Land Base Assessment % of Households % of Households % of Households with landownership with landownership Type of Program with land above 0.01 acre above 0.50 acre Mean size of land (decimals) Project Control Project Control Project Control Project Control Asrayan Housing FEE RMP Old Age Allowance Widowed Allowance VGD All Programs Source: BIDS, Impact Evaluation of Social Safety Nets Initiatives, Final Report, April Bangladesh s safety net programs tend to be modestly pro-poor with an average of 34% of beneficiaries in the first quintiles, and 60% of beneficiaries up to the second quintiles, according to the HIES, However this is not very impressive by international standards, and as such, improving targeting mechanisms is the foremost challenge for the government s growing social protection initiatives. 26. The outreach of social protection programs is far below the requirement in Bangladesh. According to the HIES survey, a total of 13% of households are beneficiaries in one of the many social protection programs. This is further manifested in Table 12. From among the poorest 10% of households, about 23% participate in some form of safety net program. Locality-wise, this is 26% rural households, 19% urban households, and 7% metropolitan households. Table 12: Share of Households participating in at least one SSNP by Location, Program and Wealth Status Non- Targeted Rural Urban Metropolitan Total Non- Non- Non- Targeted Pension Total Targeted Pension Total Targeted Pension Total Targeted Pension Total Targeted Targeted Targeted Total Lowest 10% Source: BBS, Household Income and Expenditure Survey,

12 27. About 34% of beneficiaries in all safety nets programs come from the lowest quintile. In fact, other than Cash for Education and Freedom Fighters Allowance where these programs concentrate more on the second and third quintiles, more than one-third of the beneficiaries belong to the lowest quintile. A total of 59% of beneficiaries belong to the lowest two quintiles, whereas nearly 21% of beneficiaries are in the top two quintiles. Therefore, targeting efficiency is modest by international standards, as presented in Table 13 below: Table 13: Cumulative Distribution of Targeted Programs in various Countries Quintile Bait-ul-Mal (Pakistan) PRAF (Honduras) PROGRESA (Mexico) SUF (Chile) Family Poverty Benefit (Armenia) All Targeted Programs in Bangladesh* * The average of all targeted programs such as VGD, VGF, Old Age Pensions, FoodCash for Work, Test Relief, and others; it excludes Cash for Education and Freedom Fighter Allowance which are considered non-targeted programs. Source: World Bank (2007). 28. People from the lowest quintile have higher participation. This is presented in Table 14. One-third of beneficiaries surveyed by HIES 2005 were VGF beneficiaries. More than 50% of beneficiaries belong to the bottom two quintiles. At the same time the top two quintiles have about 20% of the beneficiaries. Table 14: Performance of Programs in Targeting Poor Households Program Lowest 2nd 3rd 4th Top Quintile Quintile Quintile Quintile Quintile VGD TR VGF Old Age Pension Total (targeted) Total (non-targeted) Source: BBS, Household Income and Expenditure Survey, As observed above, the targeting efficiency has mixed indication. In a recent World Bank technical note (2007b) it is observed that the performance of Bangladesh s current community-based beneficiary selection system can be enhanced in combination with the proxy means test (PMT). Simulations show that the estimated PMT can predict the poor with good accuracy and that the proposed combination of community-based screening and selection and PMT could greatly outperform the targeting efficiency of the country s current programs. In scaling-up this new targeting system, the note also recommends the use of geographical targeting in which allocations to local areas are made in proportion to the estimated number of poor households in that area. 29. This mixed methods approach works in three steps: (1) nominations by ward committee members; (2) PMT ranking and selection; and (3) community approval of final lists and appeals. In urban areas, because of the weaker local knowledge, a slightly different process is needed. In this case, PMT will be enhanced with local accountability mechanisms. A simple pre-application screening is also proposed to limit the number of applications that need to be processed. The process would also have three steps: (1) local outreach; (2) data collection for PMT and ranking of applicants; and (3) approval of final lists by ward members. 11

13 30. Table 19 shows the efficiency of the PMT selection criteria by itself, which never was introduced in the country, and a mixed approach where current community-based criteria and PMT is jointly introduced. It is observed that its little better to when PMT alone is used. But the pure PMT simulation is a somewhat unrealistic benchmark because in practice many poor never apply and are therefore not assessed by the PMT, even in countries with much higher coverage than Bangladesh. Therefore, outreach campaigns and geographic targeting are needed calling for a mixed targeting approach. Table 19: Simulated Distribution of Beneficiaries: PMT and Mixed Methods (at 10% cutoff) Rural Urban Decile PMT alone Ward comm. nomination based on firm criteria (no electricity; < 0.5 acre of land; 4 rooms or less) followed by PMT PMT alone Ward comm. nomination based on firm criteria (3 rooms or less; one fan or less in house; household head has primary education or below) followed by PMT Households Covered (%) VI. LEAKAGES, COST OF DELIVERY AND THE PREFERENCE TO CASH OVER FOOD 31. There is mixed evidence of leakage from food and cash transfer programs. The VGD, VGF and the FFE accounted for 40% of all food transfer programs in Analyses pointed to high levels of leakage from the system. This is presented in the following Table 15: Table 15: Estimates of Leakage from VGD, VGF and FFE Program 2000 HIES-based survey estimates (metric tons) 95% Confidence interval for estimate Program off-take for FY (metric tons) Survey estimate as % of total program allocation (confidence intervals) VGD 99,978 [72,894, 127,061] 216,675 [34% 59%] VGF 70,760 [44,251, 97,267] 149,138 [30% 65%] FFE 49,951 [27,192, 72,710] 285,973 [10% 25%] Source: World Bank (2002(c)); Program Off-take: Bangladesh Food Grain Digest, World Food Program, Dhaka, Bangladesh. 32. Similarly, leakages in the FFW program have been estimated to be 26% (World Bank, 2003). Ahmed (2004) also shows fairly high levels of leakage from the VGD and the PESP, a cash transfer program depending on the criteria used (Table 16). A PERC report (GoB, 2003) shows that a large part of the budgetary allocations for the Female Secondary Stipend program (about 20-40%) do not reach the beneficiaries and are perhaps appropriated by the schools and other intermediaries. 12

14 Table 16: Comparative Leakage from Different Programs Leakage VGD PESP RMP A. Losses due to delivery of amount less than stipulated and other unauthorized expenses (per taka) B. Losses due to beneficiaries not meeting (i) at least one targeting criteria (ii) at least three targeting criteria Total Losses. A + B (i) A + B (ii) Source: Ahmed S.S. (2005), Delivery Mechanism of Cash Transfer Programs to the Poor in Bangladesh, Social Protection Discussion Paper 0520, The World Bank. 33. Not all studies agree with these numbers representing leakage. Ahmed, Del Ninno and Chowdhury s evaluation of the FFE program (in Dorosh, Del Ninno and Shahabuddin, 2004), shows that the total leakage from the FFE program was significantly lower in the range of 16-20%; Tietjens (2003) estimates that the leakage in the Female Secondary Stipend program is between 10-12%; and similarly, Del Ninno, Smith and Roy (in Dorosh, Del Ninno and Shahabuddin, 2004) show that leakages from the VGF declined during, and in the aftermath, of the 1998 floods. 34. Leakages from programs show a strong correlation with the number of intermediaries in the transfer process. While different data sources point to varying estimates of leakage, the prevalence of this problem cannot be ignored. VGD and other food transfer programs depend on the public food distribution system, with food being loaded and unloaded at a number of points before finally being delivered to beneficiaries. PESP also involves a large number of intermediaries involved in selecting students, disbursing stipends and monitoring the program. Initial evaluations suggest that it has leakage of over 20%. A distinguishing characteristic of the RMP, which has an estimated leakage of less than 5%, is that it appears to have few intermediaries, or at least none that are able to intervene in the cash payment process at any point. 35. The delivery of social assistance programs could be outsourced to agents operating at the local level. There are two factors to consider when outsourcing. First, the organization would have to be provided with resources some form of management fee to cover the cost of their involvement. Second, an agent would need a secure contract with clear performance criteria and performance measuring procedures in place. Since in all cases where cash is transferred, a nationalized commercial bank or a specialized bank is used as delivery mechanism, efficiency of outsourcing can be higher. 36. The cost of delivering cash is cheaper than the cost of delivering food to the beneficiary. A recent study by Ahmed et al. (2007) finds that the cost of delivering Taka 1.00 worth of food is, on an average, Taka The cost of delivering Taka 1.00 in cash benefit, however, is virtually zero. 13

15 Cost of Delivery Paisa Rice Flour All food Cash Benefit Source: Ahmed et al. (2007) 37. The HIES 2005 survey also found that beneficiaries prefer cash over kind. This is irrespective of the income group or program type. About 75% of total beneficiaries from all groups and 68% of the bottom quintile prefer cash over kind. This is presented in Table 17. Table 17: Beneficiaries Preference to Cash over Kind Choice Lowest 2 nd 3rd 4th Top Quintile Quintile Quintile Quintile Quintile Total Does Not Prefer Cash Prefers Cash Total Source: BBS, Household Income and Expenditure Survey, Beneficiaries consider cash assistance very helpful. As shown in Table 18 below, 61.4% of the beneficiaries from the lowest decile perceive the programs are very helpful even though the size of the benefit covers only a part of the average household consumption. Table 18: Participants Perception of the Assistance received through Targeted Programs Quintile Very Moderately Not Helpful Helpful Helpful Helpful Lowest Quintile nd Quintile rd Quintile th Quintile Top Quintile Total Lowest 10% Source: BBS, Household Income and Expenditure Survey, VII. THE NEED FOR BETTER PROGRAM ADMINISTRATION 39. Safety net programs in Bangladesh are administered by numerous agencies, including many arms of government, non-governmental organizations, as well as international bilateral and multilateral partners. While the involvement of a number of agencies in social protection provides a high visibility for this sector, it may also become a hurdle when there is a need to coordinate and navigate through many different agencies to achieve an objective or undertake a task. This issue is evident from the chart below, which shows the complication in completing the entire process of offering the VGD social safety net program. 14

16 VGD Funds Flow Source: Ahmed (2005). 40. It is evident from the above discussion that there are regional disparities, failures to address urban poverty, inappropriate targeting mechanisms and substantial leakages in the current social protection programs in Bangladesh. The primary cause is the overall implementation strategy. Almost everything related to social safety net programs is designed by the central government and there is virtually no connectivity with the local governments in relation to decision making in these processes. 41. Safety net programs in Bangladesh are administered by numerous agencies, including many arms of government, non-governmental organizations, and international bilateral and multilateral partners. According to the budget , 57% (56% in ) of the budget is officially allocated towards poverty reduction, and there is not a single ministry which does not have a role in poverty reduction. Additionally, there are eight ministries which have specific social protection programs mandated by the budget. This highlights the fact that various ministries provide social protection independently without any inter-ministerial coordination. This also results in high administrative expenses when they are pooled together. The patchwork of social protection initiatives, through at least 30 known programs, raise concern that these programs are not considering an administratively viable social safety net mechanism since it is difficult to determine the accountability and responsibility of the implementing agencies. 42. Decentralizing program implementation can be highly rewarding. One of the oldest rural local government institutions in Bangladesh is the union parishad. The parishad is elected by the local people from among the local people. There are about 4,500 unions throughout the country. Similarly, municipalities and city corporations elect ward commissioners. There are 6 15

17 city corporations and 225 municipalities in the country. Elected officials intimately know their respective locality and tend to be more innovative. These local representatives can better identify poor local people, assess their needs to help them overcome their poverty, and closely monitor the execution of programs. Therefore, devolving the administration of social protection programs to local government could improve service delivery and social accountability. VIII. CONCLUDING REMARKS 43. From the analysis presented above, the following findings emerged: A regime shift in the safety net system towards emerging dimensions is needed. At present, too many safety net programs are in operation at present. The origin of the main programs relates to food shortages during the early stages of the economy. However, the economy has changed significantly over the course of time including the nation in the recent period becoming near food sufficient and achieving higher rates of economic growth. There are a number of opportunities in Bangladesh for establishing ex-ante social safety net initiatives. Some of the measures that may be considered include building on the massive microfinance network to provide social protection for the poorest, and establishing micro-insurance and micro-pension programs for those just above the poorest. New approaches to urban safety nets and safety ladders are needed. For example, school enrollment rates in metropolitan areas for the poorest quintiles are worse than those in rural areas. Little has also been done for the poor children, and new dimensions for protecting poor children require immediate attention. A better targeting approach is essential and should be implemented for greater outreach. The same group of poor and vulnerable persons is targeted by different programs causing a repetition of programs for similar risk types. This complicates the task of targeting and implementing such programs. Instead of spreading allocation among various independent risk dimensions, a cohesive universal targeting mechanism can be implemented. This will reduce the operating cost and improve the efficient distribution of social protection programs. Entrusting local government institutions in executing safety net programs shows a lot of promise, since local government institutions are ultimately involved in all programs. If Union Parishads are appointed as the major implementing institutions, efficiency of the programs can be improved. Bangladesh is poised to gradually move from social safety net programs only to social security programs as well. This is because of opportunities emanating from continuing economic growth and increasingly positive democratic political environment. Every effort should be made now to ensure smooth unfolding of a comprehensive social protection system in Bangladesh. 16

18 Appendix A Bangladesh Safety Net Programs: Objective and Beneficiary Selection Criteria Name of the Program Purpose Targeting Criteria 1. Vulnerable Group Feeding (VGF) 2. Gratuitous Relief (GR) 3. Vulnerable Group Development (VGD) 4. Community Nutrition Program 5. Allowance to the Widowed, Deserted and Destitute Women 6. Honorarium Program for the Insolvent Freedom Fighters 7. Old Age Allowances To provide food and other emergency assistance to disaster victims To provide food and other short term emergency assistance to disaster victims 1. Developing life skills for women through training, motivating savings and providing scope for availing credit 2. Build social awareness on disaster management and nutrition through training in groups To assists women and children for their nutrition needs in 168 upazillas Minimizing the problems faced by distressed women Livelihood support to the poor freedom fighters through cash transfers Livelihood support to the elderly poor Financed by and Implementing Ministry UNCONDITIONAL TRANSFERS: FOOD 1. Disaster and calamity DPs victims MFDM 1. Disaster and calamity victims 1. Households with not more than 15 acres of land 2. Households with income less than Tk. 300 dependent upon seasonal wage employment 3. Women of reproductive (18-49) age 4. Day labor or temporary worker 5. Households with little or no productive assets 1. Poor women and children in community center DPs MFDM WFP EC CIDA MWCA World Bank UNCONDITIONAL TRANSFERS: CASH 1. Women either widowed, deserted or destitute 2. The number of beneficiaries identified on MWCA the basis of the category of union 1. Freedom fighter's identity verifiable in cross section of references 2. Freedom fighters with income less than Tk per year 3. Disabled or partially disabled or landless or unemployed or none in the family to depend upon 4. Landless 5. Not beneficiary of other programs 1. At least 65 years of age 2. Income equal to Tk Must have worked in formal sector 4. Number of beneficiaries is determined on the basis of category union 5. 50% of the beneficiaries are women and the other 50% is men MFFA MSW Nature of Benefit Planned Coverage Delivery Mechanism 10KGs rice for 8 months 5,000,000 persons Food (Usually wheat) 10KGs rice variable period 6,400,000 Union Parishad 30KG wheat and training 200,000 MT Wheat Union parishad and NGO Budget Allocation [06-07] Taka in Crores (400,000 MT) (867.00) (64,000 MT) [119.00] (366.00)? [85.30] Tk. 220 per month 825,000 Bank Tk. 600 per month 100,000 Bank Tk. 220 per month 1,700,00 Bank (156.00) (60.00) (380.00) 8. Allowance for Retarded Disabled Persons 9. Food For Works (FFW) (and Cash for Work) To provide subsistence allowance to fully retarded persons 1. Employment generation for the poor in slack season 2. Developing and maintaining rural 1. People who are mentally or physically challenged or handicapped MSW CONDITIONAL TRANSFERS: FOOD 1. People who are functionally landless ADB 2. People who lack WFP productive resources Tk. 220 per month 200,000 Bank Quantity of Work Done 2,247,000 man month Food (Usually wheat) (40.00) [233.00] 2.36 million MT food [1.25 million MT food]. This is outside 17