Difference-in-Differences When Trends are Uncommon and Stochastic

Abstract

We extend the difference-in-differences framework by developing a new estimator for policy evaluation that is robust to violation of the common trend assumption under a general set of trend specifications. Our estimator integrates a data-driven method to proxy trends and it can be easily implemented in two steps. We also consider two statistical tests for the common trend assumption. We use our approach to examine the effects of welfare waiver programs on welfare caseloads in the US. Overall, our approach delivers more reasonable and robust results than conventional difference-in-differences approaches.