Smoke and Mirrors

Abstract

Microfinance has been on the development agenda for more than 30 years, heralded as the wondrous tool that reduces poverty and empowers women (Hulme and Mosley, 1996; Rutherford, 2001; Morduch and Haley, 2002; Khandker, 1998). Doubts, however, have recently been raised about the success of microfinance (Dichter and Harper, 2007; Banerjee et al, 2009; Roodman and Morduch, 2009; Karlan and Zinman, 2009; Bateman and Chang, 2009). Given this context, this paper re-examines the microfinance impact evaluation of SEWA Bank conducted by the United States Agency for International Development (USAID) in India in 1998 and 2000. The USAID panel and a new cross-section data set are analysed using propensity score matching (PSM) and panel data techniques to address selection bias. Sensitivity analysis of the matching results is used to explore their reliability. Various sub-group comparisons between borrowers, savers and controls are also conducted to shed some light on the impact of savings versus credit. The paper concludes that doubts remain about the quality of the impact estimates obtained through advanced econometric techniques. Direct observation and the outcome of sensitivity analysis of the PSM analysis suggest that the application of PSM and differences-in-differences (DID) to these observational data were probably unable to account for selection on unobservables1 .