DiD

Question 1

What are the assumptions behind the two-way fixed effects estimator in difference-in-differences?

Answer 1

We need there to be: parallel trends, no anticipation effects, common treatment timing, and homogeneous treatment effects.

Question 2

Describe the limitations of using difference-in-differences estimation when dealing with more than two groups or time periods.

Answer 2

When there are more than two groups or time periods we need extra assumptions for TWFE to recover the late. Specifically, we need homogeneous treatment effects when there are more groups, and also common timing if there are extra time periods.

Question 3

In what situations is the difference-in-differences approach more powerful than the before-and-after approach for estimating the causal effect of a treatment?

Answer 3

The before and after approach works well when there is no time trend. When outcomes of interest are changing over time then DiD will work better.

Question 4

What are the two set of controls we can choose from in the Callaway and Sant’Anna DiD procedure?

Answer 4

We can choose the set of control “cells” that is group-by-time sets of observations by choosing among the set of observations which are not yet treated (treated groups in pre-treatment periods) or those
which are never treated (groups that are never treated).