Does Microfinance Help the Ultrapoor Cope with Seasonal Shocks? Evidence from Seasonal Famine (Monga) in Bangladesh

This paper uses a unique data set on 150,000 ultrapoor households in Bangladesh to analyze whether microfinance helps cope with aggregate shocks such as seasonal famine. To address selection on unobservables, we use two alternative approaches: an instrumental variables strategy that exploits a natural nonlinearity arising from the interaction of self-selection by households with the MFI screening, and the `Minimum-Bias Bias-Corrected' estimator due to Millimet and Tchernis (2010) that corrects for endogeneity bias without exclusion restrictions. The empirical results suggest that microfinance improves food security of a relatively better-off household, but may not have any robust effect on the food security of the poorest of the poor. Evidence is strong that microfinance helps households to avoid distress sale of labor, and reduces the probability of short term migration in search of work during Monga.