Two-Way Fixed Effects

Overview

In this module, we look at the use of two-way fixed effects to estimate generalized difference-in-differences models when treatment timing varies across treated units. We’ll see that the mis-specification and weighting issues that we discussed in the context of continuous variation in treatment intensity are also relevant here: two-way fixed effects estimates can be severely biased when treatment timing is staggered and treatment effects on individual units are not constant over time.

Readings

Simple Diagnostics for Two-Way Fixed Effects

Optional Readings

Difference-in-differences with variation in treatment timing

Lecture Slides

A handout version of the lecture slides is available here.

Empirical Exercise

The in-class activity as a do file or pdf.

The empirical exercise as a do file or pdf.

A web version of the empirical exercise is available here.