Empirical Analysis on Multiple Mergers of US Banks

Abstract

We use logistic analysis to predict the probability of making non-programmed merger in a data sample of 45 US banks. Non-programmed merger is the merger that happens next to the subject merger but has at least three years apart from the subject merger. We apply logistic regression of the occurrence of the non-programmed merger on main characteristics of the subject merger. We first examine the effects of each of three explanatory variables, which are firstly abnormal return around the approved date, secondly hubris management hidden in the subject merger, and thirdly the value of asset acquired, on the dependent variable. We then try to find the best prediction model by controlling some variables both confounding and rescaling. Our final prediction model shows that the probability of making a next merger at least three year after the subject merger will significantly decrease if there is abnormal return realized in the subject merger. On the other hand, using event study methodology to search for the abnormal return of the acquirer's stock price around the approved date, we prove that the information of FDIC s' merger decision is not totally confidential to public and has significant impact on the stock price of the acquirer