Bayes’ Factors for Event-Study: The Case of A Single Security
27 Pages Posted: 5 Aug 2020
Date Written: January 15, 2016
Abstract
Event Studies have been a subject matter of interest in corporate finance for quite sometime. While the frequentist literature of event-study is fairly matured, the same cannot be said about its Bayesian counter-parts. This article derives the Bayes’ Factors for the hypotheses testing problems of event study, in the context of a single security, involving both its abnormal as well as cumulative abnormal returns. The priors needed for the computation of the Bayes’ Factors, are constructed using the estimation period data. The developed methodologies are illustrated to study the behavior of the Volkswagen stock returns since the recent news, as well as the phenomenon of derivatives introduction in the Indian stock market. The study also conducts a simulation experiment involving the returns of all the stocks listed in India since 1998, to compare the performances of the existing frequentist methods with the proposed Bayesian ones. It is found that the Bayes’ Factor based tests are the best specified ones, with powers better or comparable to the non-parametric ones, but somewhat lesser than the corresponding t-tests.
Keywords: Event Study, Bayesian Analysis, Econometrics
Suggested Citation: Suggested Citation