While all the targeted food assistance programs in Bangladesh aim to help the poor, they vary considerably in their specific objectives, and consequently in their target populations, timing of transfers, and assistance packages. The objectives can be broadly classified into two types. First, programs such as the VGF and GR have a purely ‘relief’ objective, designed to provide assistance to poor households that have been physically dislocated or otherwise adversely affected by disasters. Second, programs such as the VGD and FFE help individuals graduate out of poverty by building human capital skills and income generation skills. This section summarize some of the key findings with regard to the development impact of Bangladesh’s food assistance programs drawing upon existing literature and analysis of the 2000 HIES data.
Disaster relief and short-term coping objectives: How effective a program is in providing short-term relief depends on how quickly resources can be mobilized in response to a disaster, and delivered to people who are adversely affected. The recent massive flood in 1998 provides a good context to examine the role played by the two main relief programs – GR and VGF – in assisting flood affected households. Studies by FMRSP-IFPRI suggest that while direct distribution through these programs was quite small, as compared to private imports, the programs played an invaluable role by targeting relief to those in need. Immediate short-term relief after the flood was mainly provided by the GR. The program was rapidly expanded (with some assistance from bilateral donors), and relief efforts began in August through provision of 20,400 metric tons of rice, and another 30,800 metric tons in September, to flood-affected thanas. Overall, more than 74,000 metric tons of wheat and rice were distributed in 1998-99. del Ninno and Dorosh (1999) found that this assistance was well targeted to flood-affected households. Only 11.4 percent of GR recipients were not directly exposed to floods. However, the amount of transfer provided to beneficiaries was small: on average, households received only 10 Tk per month (over a four-month period), which was much smaller than receipts from private transfers or other government and NGO programs during the same period. In addition, the program did not achieve widespread coverage: overall, 69 percent of flood-affected households did not receive GR transfers.
In contrast, the VGF program targeted resources to all areas (both flooded and non-flood affected areas) of the country. As a result, while it was not effectively targeted according to flood exposure, it was successful at reaching the poor. Approximately 25 percent of VGF recipients were not directly exposed to the flood but households in the three lowest expenditure quintiles received an estimated 75 percent of the food grains distributed though the program. VGF assistance was an important component of the total assistance (government, NGO, and private transfers) received by the poor: for the poorest quintile, VGF transfers constituted a substantial 35 percent of total transfers received (del Ninno and Dorosh, 1999; del Ninno and Roy, October 1999).
Another context in which programs aim to provide short-term relief is in response to seasonal variations in food prices and employment opportunities in rural areas. Traditionally, there have been two lean seasons in Bangladesh: the January—March period, and the September—October period that precedes the aman rice harvest. Precisely to help households cope with the first lean season, the FFW programs provide employment opportunities and wages in kind (wheat) during the first three to four months of the year. In an early study, Osmani and Chowdhury (1983) found that FFW wages did lead to an increase in food grain consumption among beneficiaries.
Over the last two decades, the stress of the first season has dissipated somewhat due to the rapid increase in boro rice and wheat cultivation, which have both dampened consumer price pressures and created employment opportunities (Dorosh and Haggblade, 1995). However, the second lean season is still believed to be acute, when food prices peak and employment opportunities are scarce. Table 10 examines temporal and regional variation in unit-values (expenditures divided by quantity) and per capita consumption of rice and wheat to assess the extent to which this pattern is borne out by the 2000 HIES data.
The regression coefficients in Columns (1) and (2) show that there is some seasonal variation in prices of rice and wheat, but there is no peak evident during the September to November months. On average, rice prices were highest between March and May, a result that is mainly driven by higher prices in March and April. Wheat prices were highest at the beginning of the year, following which they declined by 8 percent between March and May, and continued to decline through the year. In addition, while there is some seasonal variation, it is not as large as the variation in prices across divisions and sectors.
Notes: The tables reports coefficients from a regression of the logarithms of unit values and quantities on division, urban, and month dummies. The figures in brackets are absolute t-values. ns means not significantly different from zero. Dhaka division, Rural sector, and Dec-Feb months are omitted categories.
Could it be that even though prices do not show a peak during the second lean season, lack of earning opportunities limits food consumption during this period? The results in Table 10 suggest that rice consumption was stable throughout the year, with no significant temporal variation (Column 3). Wheat consumption was highest during the first lean season which is not surprising since it coincides with both the wheat harvest and high off-takes from the PFDS channels (mainly through FFW). The same patterns are true when we look at different quintiles separately; there are no peaks or troughs in per capita rice consumption even for the poorest households in Bangladesh.22 However, focusing on ‘vulnerable’ groups, shows that not all households are able to protect consumption from seasonal variation. In particular, daily wage laborers in rural areas are found to consume about 10 percent less rice in the September to November lean season, as compared to other months (Column 5). These households also have strong temporal variation in wheat consumption, unlike other occupational classes.
Taken together, the results in Table 10 are heartening to the extent that they suggest that households in Bangladesh are now better able, on average, to shield their consumption of the staple cereal from fluctuations in prices. At the same time, the results re-emphasize that targeted interventions—across regions, in particular months, and towards particularly vulnerable groups—can play an important role in helping the poor smooth out fluctuations in consumption.
Development Objectives: FFE and VGD are the two main targeted food-assistance programs that have explicit development objectives. Previous evaluations of the FFE program suggest that the numbers of pupils enrolled at participating schools has increased (Ahmed and Billah, 1994; BIDS, 1997), but there is conflicting evidence on whether this is simply because pupils are transferring from non-FFE to FFE schools. Ravallion and Wodon (2000) using data from the 1995-96 HES found a significant and sizable impact of the program on school enrollments. By their estimates, the program increased the probability of going to school by a little over 20 percent for both boys and girls. Preliminary investigation of the 2000 HIES using the same approach as Ravallion and Wodon suggests that gains in enrollment have been sustained. However, among the sample of households that have children currently enrolled in school, participation in FFE has not had any significant impact on number of days enrolled children attended school.
Another recent study by Ahmed and del Ninno (2001) also examines the impact of FFE on a range of outcomes. It finds that enrollment and attendance rates are higher in FFE schools, and drop-out rates tend to be lower. However, FFE schools have higher student-to-teacher ratios and more crowded classrooms, as a result of which the quality of education is poorer. This points to the need to include complementary financial and technical assistance in order to ensure that student achievement also increases concurrently with higher enrollments. The study also finds that the recent switch to a system of distribution of food through private dealers has been far from satisfactory: beneficiaries now incur higher transaction costs in collecting food transfers and system leakage has worsened.
Evidence on the development impact of the VGD program is more limited. The few studies that have been done in general report that the program is targeted to the poor and is perceived to be fair at the local level (e.g. del Ninno, 1998; Sen, 2001). Comparing the socio-economic status of beneficiaries one year after the joining the program and two years after leaving the program—i.e. across a three-year interval—an evaluation conducted by WFP (1997) found that beneficiaries of the VGD program owned more assets (e.g., beds, tin-roofs, clothing), had more savings, and increased access to credit after the program. Importantly, the beneficiaries perceived improvements in their decision-making ability within the family, in the health of their family, and in incomes. However, it is interesting to note that study provides a mixed picture about whether these impacts are a result of the food transfer and savings components, or as a result of the training imparted. Comparing the performance of UPVGD and IGVGD beneficiaries (the former do not receive training as part of the VGD package) does not show any clearly discernable differences across the two groups of participants, though of-course this could also be due to the relatively small sample of the two types of beneficiaries.
