Does social capital build women's assets? The long-term impacts of group-based and individual dissemination of agricultural technology in Bangladesh

Abstract

This paper investigates the long-term impact of agricultural technologies, disseminated using different implementation modalities on men's and women's asset accumulation in rural Bangladesh. Panel data spanning a 10-year period are used to examine the effects of the adoption of new vegetable varieties and polyculture fish pond management technologies on household resource allocation, incomes, and nutrition. A difference-in-differences model combined with nearest-neighbour matching is used to compare changes in husbands and wives' assets within the same household. The results show women's assets increase more relative to men's when technologies are disseminated through women's groups, suggesting that implementation modalities are important in determining the gendered impact of new technologies. These findings are robust to controls for unobserved household-level characteristics. These results suggest that social capital, as embodied through women's groups, not only serves as a substitute for physical assets in the short run, but helps to build up women's asset portfolios in the long run.

Keywords:

• gender
• social capital
• Bangladesh

Acknowledgements

This research was funded by the CGIAR Systemwide Program on Collective Action and Property Rights (CAPRi) and the Department for International Development through the Economic and Social Research Council of the United Kingdom: ‘What Development Interventions Work? The Long-term Impact and Cost Effectiveness of Anti–Poverty Interventions in Bangladesh' (RES-167-25-0361). It builds on collaborative work funded by the Chronic Poverty Research Centre, HarvestPlus, and the World Bank. The authors acknowledge helpful discussions with Akhter Ahmed, Bob Baulch, Peter Davis, Ruth Meinzen-Dick, Wahid Quabili, Md Zahidul Hassan, and Md Zobair, and comments from three anonymous referees and participants at a CAPRi Workshop on Gender and Collective Action in Dhaka. All errors and omissions are the authors' own.

Notes

1. This description draws from Hallman et al. (2007), and recent field visits by the authors.

2. We also explored the comparison used in Bouis et al.'s (1998) and Hallman et al.'s (2007) short-term impact evaluation, comparing NGO members or programme members who had adopted the technology and those who did not yet have access to the technology; see Kumar and Quisumbing (2010).

3. Data not presented here also show that the size and composition of the households is similar across the three groups, although the percentage of women older than 55 years among the non-NGO households was almost double the corresponding number for NGO members, at 7 per cent.

4. 100 decimals  =  1 acre (or 0.4047 of a hectare).

5. When we matched only on household characteristics, very few treatment effects were significant, in contrast to the results reported here, based on individual matching. This suggests that, when examining intrahousehold impacts, it is important to control also for individual characteristics, not just characteristics of the household.

6. We use nnmatch in Stata10 to estimate our matching estimators (Abadie et al. 2004). Note that if the intervention was rolled out at the same time to all NGO members in all of the villages in the catchment area, this approach would not be feasible as it would not be possible to construct a statistically robust comparison group. However, because of resource constraints (mainly due to limited implementation capacity at the initial stages of technology dissemination), programme access was rationed.

7. We thank Sajeda Amin and colleagues at the Population Council for this insight.

8. During a field visit to Jessore in August 2008, a woman who had begun fish cultivation with a group pond mentioned that she had used the proceeds from the pond to buy land in her own name, and had started an individual fish pond on the newly acquired land.