The Centre for the Study of Democratic Citizenship, presents:
Scott Cunningham (Baylor University)
New Developments in Differences-in-Differences
You can learn more about Scott Cunningham by clicking here.
Where and When: Thursday, November 5, at 1pm on Zoom.
RSVP: please contact Costin Ciobanu at costin.ciobanu@mail.mcgill.ca
Abstract: Historically, the differences-in-differences quasi experimental design has been one of the most if not the most popular design among all quasi-experimental papers written in economics. The most common situation has been a staggered rollout in which groups are treated at different points in time. In such situations, the panel fixed effects model with panel unit and time fixed effects (“twoway fixed effects”) has been employed because it handles such situations well. But recent work by Goodman-Bacon shows that this estimator is biased if treatment effects are dynamic. I will show the bias of OLS with twoway fixed effects under differential timing and dynamic treatment effects including a simulation that illustrates the seriousness of the bias. Time permitting, I will discuss a recent estimator by Callaway and Sant’Anna which is unbiased under the above mentioned scenario.
See more Montreal Methods Workshops events here.