That’s the title of my latest article, coauthored with Dan Millimet, and which just came out in the spring issue of the Journal of Economic Perspectives. Here’s the abstract:
We discuss Yair Mundlak’s (1927–2015) contribution to econometrics through the lens of the fixed effects estimator. We set the stage by discussing Mundlak’s life and his seminal 1961 article in the Journal of Farm Economics, showing how it was looking at the right application—the study of agricultural productivity, which had hitherto been thought to be marred by the presence of management bias—that led Mundlak to use the fixed effects estimator. After discussing Mundlak’s contribution, we briefly discuss the historical economic and statistical contexts in which he made that contribution. We then highlight the dialogue that took place between the proponents of fixed versus random effects and discuss how Mundlak settled the debate in his 1978 Econometrica article. We conclude by discussing how, between fixed and random effects, the fixed effects estimator won the day, becoming the de facto estimator of choice among applied economists because of the Credibility Revolution, culminating in the popularity nowadays of difference-in-differences designs and of two-way fixed effects estimators.
I have learned a lot while working on this article. Like many people, I thought Mundlak himself had developed the fixed effects estimator. But that turned out not to be true: The estimator was already available; what Mundlak did which forever associated his name with the fixed effects estimator was to find the right application for it. Having recently published an article in which my coauthors and I present the first application of an estimator to economics, this was very encouraging.