More power to you: properties of a more powerful event study methodology

Abstract

PurposeThe purpose of this paper is to demonstrate with real data the enhanced statistical power of a GLS‐based event study methodology that requires the same input data as the traditional tests.Design/methodology/approachThe paper uses full sample, subsample and simulated modified sample analyses to compare the statistical power of the GLS methodology with traditional methods.FindingsThe paper finds that it is often the case that traditional tests will not reject the null when a GLS‐based test may (strongly) reject the null. The power of the former is poor.Practical implicationsThere are many published event studies where the null is not rejected. This may be because of the phenomenon being tested but it may also be because of the lack of power of traditional estimators. Hence, rerunning them with the authors' more powerful test is likely to reject some currently well‐accepted null hypotheses of no event effect, stimulating new research ideas. Moreover, as individual stocks have become more volatile, the additional power of the authors' methodology to detect abnormal performance for recent and future events becomes even more important.Originality/valueThere are more than 500 event studies in the top finance journals, which can broadly be split into two subgroups: contemporaneous shocks like changes in regulation and non‐contemporaneous events like mergers. GLS contemporaneous modeling of covariances in the former showed little efficiency gains. The paper's GLS modeling of variances for the latter demonstrates potentially huge effects. Practitioners should be skeptical of prior results accepting the null of no event effect and incorporate GLS to be confident of their future findings.