Using and Interpreting Fixed Effects Models

Abstract

Fixed effects are ubiquitous in financial economics studies, but many researchers have a limited understanding of how they function. This manuscript explains how fixed effects can eliminate omitted variable biases and affect standard errors, and discusses common pitfalls in using fixed effect regressions. I especially focus on how fixed effect groups (e.g., firms) that have little or no variation in X can confound coefficient estimates and interpretations, and I provide guidance on how to identify and avoid said confounds. I emphasize that FE can be a powerful tool for improving identification but can also introduce important problems of their own. Better understanding these issues will help researchers make better choices about how to design fixed effects models and carefully interpret the results thereof.

Keywords: research methods; econometrics; fixed effects; financial economics; accounting

JEL Classification: G00; M4; C58