Is Post-Matching DiD an Appropriate Method in Empirical Studies? The Research Design Matters

Abstract

This study clarifies the model that should be applied when investigating the impact of an event on outcome variable for a perfectly-matched sample. Empirically, DiD regression is a widely used approach in analyzing exogenous shocks with imperfectly matched or unmatched sample. Results from the coefficient and graphic analyses show that only dummies that indicate the post-treatment effect shall be considered if exact matching in individual characteristics is conducted beforehand. The results can provide clear guidance for social science studies such as finance and economics.

Keywords: Matching Method, Difference-in-Difference (DiD), Endogeneity, Parallel Trend Assumption

JEL Classification: C81, C50, G15, G30