Proxy Variables in Empirical Corporate Finance
Proxy variables are ubiquitous in empirical research. This paper provides a cautionary tale on the use of proxy variables via an in-depth look at one specific setting: value creation in corporate takeovers, a core subfield of the empirical corporate finance literature. An established fact in that literature is that bidder and target size are major determinants of bidder announcement returns in corporate acquisitions. The leading explanations in the literature attribute this to size proxying for underlying value drivers. A simple but powerful test casts doubt on size-as-proxy explanations: correlations between size and bidder announcement returns change sign across economically significant subsets of the takeover universe. A scaling framework in which size magnifies per-dollar net present value parsimoniously explains these sign-changes and yields additional testable predictions we confirm in the data. Our results advocate a fundamental shift in thinking about the role of size in takeovers: size should not be interpreted as a proxy, but “size is size.” More broadly, our paper highlights the dangers of too uncritically accepting explanations for economic phenomena based on very indirect proxies. We show how additional informative heterogeneity in the data, obtained via simple methods including subsample analysis and quantile regressions, can set the bar for questionable proxy explanations higher
Dr Chau Chak Wing Building
