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...