On event studies and distributed-lags in two-way fixed effects models: Identification, equivalence, and generalization

Abstract

We discuss three important properties of panel data event study designs. First, assuming constant treatment effects before and/or after some event time, also known as binning, is a natural restriction, which identifies dynamic treatment effects in the absence of never-treated units. Second, event study designs with binned endpoints and distributed-lag models are numerically identical. Third, classic dummy variable event study designs can be generalized to models that account for multiple treatments of different signs and varying intensities. We demonstrate the practical relevance of our methodological points in an application studying the effects of unemployment benefit duration on job search effort.