When Good Balance Goes Bad: a Discussion of Common Pitfalls When Using Entropy Balancing

Abstract

In many research settings, entropy balancing (Hainmueller 2012) is a convenient statistical method for reweighting a control sample to achieve covariate balance with a treatment sample of interest. Despite this, researchers raise several concerns and questions not specifically addressed by Hainmueller (2012). We discuss a handful of the most common issues that arise in empirical-archival accounting research settings. We demonstrate these issues and propose empirical solutions, using the setting of estimating the Big-N audit fee premium as an empirical illustration.