Vulnerable Group Feeding (VGF), designed to provide one or more months of food rations to a selected number of households in a period of distress, is implemented by the Directorate of Relief under the Ministry of Disaster Management and Relief. Resources are allocated to the program by the Ministry of Finance when thought necessary. VGF started as a small program in the mid-1990s in response to a drought, but it was rapidly expanded, with supplementary food aid from WFP, to provide relief assistance after the 1998 floods.7 Over 460 thousand metric tons of food grains were distributed across 51 flood-affected districts, making VGF the second largest food-assistance program that year.
Under the program, food is distributed only among thanas that are affected by disaster. Among the selected thanas, two-thirds of the food is distributed according to population and the remaining one-third is allotted to thanas rated as having very high or high food insecurity on the Government/WFP resource allocation map for food-assisted development.8 Within thanas, the households to receive food are selected by local communities. Each UP chairman sends a list of potential beneficiaries (compiled from recommendations of ward members, with 50 percent of the beneficiaries taken from female ward member lists) to a thana-level committee including the TNO, UP chairmen, and other local government officials, which makes the final selection. Government selection criteria are defined to include: (i) households in which the head earns less than 300 Tk in a normal month; (ii) households with no agricultural land or productive assets; and (iii) households headed by day-laborers. Priority is given to households that have been physically dislocated by the disaster and to female-headed households.
Each household on the list is issued a ration card to receive food for a few months. The Ministry decides how much they receive and for how long, depending on the extent of the disaster. In practice, it is not always possible to restrict food distribution to selected households when other needy households are also present at the UP center at the time of distribution. In such cases, the distributors sometimes have to reduce the amount given to each card holder in order to increase the total number of beneficiaries.
Gratuitous Relief (GR), designed to provide emergency relief to disaster victims, is also implemented by the Directorate of Relief under the Ministry of Disaster Management and Relief. While GR is a relatively small program, it is the main way in which the Government provides immediate, short-term relief to disaster areas. Union Disaster Committees (UDCs), comprising the Union Parishad Chairman, local elites, and local government officials, collect information on the number of households affected and the extent of the disaster. This information is compiled at the thana-level by the TNO, and relayed to the Ministry by the District Deputy Commissioners. Based on these reports, the Ministry determines the type of assistance (cash, food, blankets, building materials, and so on), geographic allocations, and beneficiary entitlements. The UDC selects the households to receive assistance.
2.5 Rural Maintenance Program
The Rural Maintenance Program (RMP), while similar to FFW programs, is smaller than most (50,000 metric tons of wheat equivalent per year) and differs in one important respect: it pays wages in cash, not in kind. It is implemented by local government institutions (Union Parishads, under the Local Government Division of the Ministry of Local Government, Rural Development, and Cooperatives) with the assistance of CARE-Bangladesh. Financial support comes from the proceeds of monetized wheat aid from CIDA and from the Government’s own resources. Originally started in 1983 as a pilot project with the dual objective of maintaining rural earthen roads as well as providing employment and training to destitute women, RMP has now been extended to approximately 4,100 union councils throughout Bangladesh.9In each union, 20 kilometers of rural roads that have been built with funds from the IFFD-FFW program are maintained by a roads maintenance association (RMA) comprising 10 women from each union council. The association is responsible as a group for maintaining a given stretch of road every fortnight, with the quality of the work monitored by a member of the union parishad.
In each union, the union parishad, in collaboration with a Project Implementation Committee (comprising village members) is responsible for overseeing the beneficiary selection process. The women in each RMA are selected by lot from a list of 50 eligible women in the union; the criteria for determining eligibility include being divorced, separated, or otherwise destitute; in addition, the women must be between 18 and 35 years of age, and should be physically fit enough to do the work. Participating women are enrolled in a four-year program of work and training during which they receive 43 Tk. per day as wages, nearly one-fourth percent of which is automatically retained in an escrow savings account on their behalf. During the last year of the project cycle, women receive more intensive training in basic business management to maximize potential returns to savings following their eventual graduation from the project.10
3.Scope for Better Regional Targeting?
Of the various government programs described above, only the VGD and RD programs – and to a limited extent the FFE – attempt regional targeting. In the case of the FFE, allocations of resources to the various thanas are made in proportion to their population.11 However, within each thana, 2 to 3 unions that are economically backward and have a low literacy rate are selected for the program. In the case of the VGD and RD programs, allocations to each thana are made on the basis of the Resource Allocation Map developed jointly in 1995 by WFP and the Planning Commission. The thanas are classified as having very high, high, moderate or low food insecurity, using a composite index based on the following criteria: (i) incidence of natural disasters, (ii) food-grain deficit/surplus, (iii) agricultural wage rate, (iv) proportion of households not owning land, (v) proportion of unemployed persons, (vi) proportion of widowed, divorced, and separated women, and (vii) proportion of literate women. Data for deriving this composite index are obtained from a variety of sources, including the 1991 Population Census, data from the Disaster Management Bureau, and the Agriculture Statistics wing of the BBS.
While there clearly is scope for improving upon the proxy criteria used under this system to identify regions with higher poverty, the VGD and RD are amongst the few programs that at least attempt to target resources explicitly towards regions with greater need. Analyzing targeting performance for the FFE, Galasso and Ravallion (2000) find that most of the overall pro-poor targeting performance was due to pro-poor targeting within villages. Geographic targeting, across unions, contributed less to overall targeting performance than intra-village targeting. Therefore, they conclude that the Government’s desire to assure broad geographic coverage of the program constrains considerably overall targeting effectiveness of the program.
Because of constraints of political economy or otherwise, most other programs are nation-wide in coverage, with little or no effort directed to target regions with higher concentration of the poor. However, as the World Bank’s Poverty Assessment (1997) for Bangladesh and a number of other studies (e.g., Ravallion and Wodon, 1997) have shown, poverty in Bangladesh varies quite considerably from one region to another. Thus, a priori, there appears to be considerable scope for improvement as far as more “poverty or need-focused” allocation of resources across regions is concerned. Consider, for example, the Food-for-Education program: if one of the main objectives of this program is indeed to offer incentives to parents to send their children to school, the striking variation in school enrollment rates across different parts of the country suggests that the program could be made much more effective if more resources were directed to the areas with lowest enrollment. Similarly, data from the 2000 HIES shows that there is considerable variation in living standards across the country, which could more explicitly be taken into account when making allocations for some of the other programs.
The Research Department at the World Bank (see Hentschel, et al., 2000, Elbers et al, 2000) has been developing a technique that combines household survey and census data to estimate consumption-based welfare indicators for small geographic areas such as districts and thanas —i.e., a poverty map. A poverty map can highlight statistically reliable differences in local poverty levels at a high resolution that is not possible to gauge with national level statistics. Currently, poverty maps using this method are used in countries including Nicaragua, Panama, and Guatemala for targeting transfers to poor areas. Since Bangladesh has recently completed both a census exercise and a nationally representative household survey, there could be high payoffs to developing a poverty map for the purposes of improving geographic targeting of food-assistance or other such targeted programs.
