Share

Agnes R. Quisumbing, Bob Baulch and Neha Kumar (2011), IFPRI Discussion Paper No. 01077, March.

This paper provides an overview of a research project that assessed the long-term impact of three antipoverty interventions in Bangladesh. The analysis of the long-term impact on outcome, including positive gender outcomes, shows that the divergence from short-term impact evaluation can arise from factors such as the characteristics of the technology introduced, targeting methods applied as well as the modalities of implementation used.

Bangladesh is one of the least developed countries of the world with a large proportion of the population living below the poverty line in 2006. Gender discrimination too is strong is Bangladesh, as a result of which “there is a strong gender dimension to poverty in Bangladesh”. Despite significant reduction in poverty and improvement in several human development indicators in Bangladesh since the 1990s, the current rate of poverty in Bangladesh continues to be extremely high. Poverty reduction is therefore the central challenge of the government and several important programmes to reduce individual and household poverty have been implemented by the government. Many of these antipoverty interventions have contributed to poverty reduction as well as reduced gender gaps in Bangladesh. However, most impact evaluations have focused on assessing the short-term impacts. In order to fill this gap, a recent research project assesses the long-term impact of anti-poverty interventions in Bangladesh.

The present paper, authored by Agnes R. Quisumbing et al, presents an overview of the research project, in particular looking at three specific interventions —the introduction of new agricultural technologies (improved vegetable, group level and individual fish ponds), educational transfers, and microfinance. In this context, the paper discusses the conceptual framework, methodology applied as well as the empirical methods used for the evaluation of long-term impacts.

Discussing the conceptual framework of the research project, the authors point out that short-term impact evaluation of antipoverty programmes, though valuable, can also be misleading. There can be several reasons that make the timing of the evaluation important. In the case of programmes that require behavioural changes of both the beneficiaries and the service providers, short-term impact evaluation can be particularly inadequate. For example, introduction of an intervention invariably requires training of service providers or cannot be introduced simultaneously because of lack of enough manpower. Also, there are spillover effects that arise from learning and adoption by non-beneficiaries or later adopters. Given these factors, evaluation conducted too early might show partial or no impact.

Regarding the techniques used for assessing long-term impact, the project uses mixed-method approach, i.e. uses both quantitative and qualitative methods for impact evaluation. While both the methods were used for evaluating the long-term impact of the first two interventions, for microfinance programmes, on the other hand, mainly qualitative technique based on life-history was employed. The quantitative approach used in the project mainly relies on assessing the impact of a programme by comparing it against a counterfactual. The idea is to assess “what would have happened to beneficiaries in the absence of the intervention or treatment”. This involves comparing the impact on participants (treatment group) with similar group of people who are not participants of the programme (control or comparison group). Different studies in the research project use different sets of comparison and treatment groups, such as: comparison between early and later adopters, or between NGO members and non-NGO members.

Based on these estimation techniques, the long-term impact evaluations of the three types of interventions bring out several interesting results. Regarding the introduction of the new agricultural technologies, the study reveals that the characteristics of technology and the modalities used for implementing the programmes matter in determining the long-term impact. For example, in the case of the improved vegetable technology, contrary to significant positive short-term impact on household outcomes, long-term-impacts are minimal. On the other hand, in the case of introduction of individual polyculture fish technologies, early adopters gained considerably more in terms of increases in total and food consumption, when compared to later adopters. The divergences in long-term impact could arise from the fact that the “owners of the individual fishponds were able to capture the gains from the technology, which requires larger capital investments and had a longer pay-back period”. The new vegetable technology on the other hand, was relatively easier to adopt for later adopters.

The issue of implementation modalities becomes apparent when the impact of new agricultural technologies on men’s and women’s asset accumulation is compared. The study shows that women’s assets increases more when technologies are disseminated through women’s groups, even though this by itself does not eliminate the initial gender disparities in asset ownership.  A comparison of the individual and group fishpond sites (using the same polyculture fish technology), reveals that dissemination through women’s groups are more successful in reducing gender asset inequality in the long term, than in the case of individual fish ponds, where disparities increased between husbands and wives actually increased.

Evaluation of the Primary Education Stipend (PES), a cash transfer programme to encourage poor parents to send their children to school, shows that it has some positive impact at the individual level. However, overall the impact of the programme is not very significant because of the declining value of the stipend and poor targeting.

On the issue of the cost-effectiveness of the different interventions in attaining a given outcome, the authors show that the positive impacts on monetary measures of well-being, including asset accumulation, are evident only in the case of educational transfers programme and the individual fishponds programme. The group fishponds and improved vegetables programmes on the other hand, had lower monetary measures of well-being, but much better individual level child nutrition impacts. These “mixed results of comparing impacts on monetary versus nonmonetary measures of well-being, illustrate the dilemmas of using cost-effectiveness to rank antipoverty interventions”.

Based on these findings, the authors conclude by making certain recommendations for evaluating impact of antipoverty interventions. They point out that for choosing the right interventions it is important that long-term impacts are also evaluated. Also, given that targeting mechanisms and implementation modalities have important implications for the ability of interventions to reach the poor, better targeting method needs to be used for reaching the poor. They also recommend that impact evaluation of antipoverty programmes must take into account intrahousehold impacts, rather than just focusing on the household as the unit of analysis.