Difference-in-Differences with matching methods in leadership studies: A review and practical guide

Abstract

Difference-in-differences (DiD) model has been widely used in leadership and management research over the last two decades. The appropriateness of DiD model relies on parallel trend assumption. In practice, researchers in leadership and management studies have employed a hybrid approach that combines the matching method and the DiD estimator to reduce bias when parallel trend assumption fails. The matching method is applied to treatment and control observations of outcomes in pre-intervention periods before a DiD method is applied. Given that this hybrid approach is intuitively preferable over other approaches, we propose four critical conditions in leadership studies to justify its appropriateness. We also propose four questions by leveraging a tutorial-type format that reviews advanced DiD-type models. Since advanced DiD-type models require large T (time dimension), we utilize a highly effective but underutilized approach for small T that builds on insights behind the DiD and matching methods. We present an advanced application of the hybrid method to illustrate how it can be employed to examine whether the proposed additional disclosure rules released by the SEC in 2009 and the subsequent choice of firms’ compensation consultants lead to lower CEO pay.