Revisiting the Effect of Food Aid on Conflict: A Methodological Caution

• Development Workshop

Date and Location

Monday, May 15, 2017, 4:10 PM - 5:30 PM
ARE Library Conference Room, 4101 Social Sciences and Humanities

Abstract

A popular identification strategy in non-experimental panel data uses an instrumental variable constructed by interacting an endogenous time series or spatial variable with an endogenous exposure variable to generate identifying variation through assumptions similar to those of a Differences-in-Differences estimator. Revisiting a celebrated study linking food aid and conflict shows that this strategy is susceptible to bias due to spurious trends. Through re-randomization simulations and Monte Carlo analysis we find that the strategy identifies a spurious relationship, even when the true effect could be non-causal or causal in the opposite direction. This provides a caution about relying on similar strategies.