Smoke and Mirrors: Evidence of Microfinance Impact from an Evaluation of SEWA Bank in India

This paper re-visits evidence from the impact evaluation of SEWA Bank conducted by the United States Agency for International Development (USAID) in India in 1998 and 2000 to illustrate the broader challenges of measuring the impact of microfinance. It discusses the challenges of controlling for selection bias and the role of unobservables.

The paper uses propensity score matching (PSM) and panel data techniques on existing panel data to eliminate selection bias in impact evaluations. It uses sensitivity analysis of matching results to explore their reliability. The paper also examines the impact of savings versus credit by comparing borrowers, savers, and control groups. Conclusions include:

• There are doubts about the quality of the PSM impact estimates, although the results presented by USAID cannot be contradicted;
• Sensitivity analysis suggests significant over-estimation of impact due to the presence of unobservables;
• Selection processes driven by unobservables strongly influence who becomes a participant in microfinance and cannot be fully controlled by econometric techniques;
• Panel design is problematic because it does not have a baseline which would allow a before and after comparison since the treatment groups had already joined SEWA Bank before the baseline period.