CEO Overconfidence and M&A Performance: Comparing Measures of Overconfidence

Abstract

In this research the five-item CEO overconfidence proxy of Schrand & Zechman (2012) is compared with the option-based Longholder and Holder 67 CEO overconfidence proxies of Malmendier & Tate (2008) in an M&A context. This is done on a sample that attempts to replicate the sample of Malmendier & Tate (2008, 2015). The results indicate that the five-item overconfidence variable of Schrand & Zechman (2012) is not similar to the Longholder and Holder 67 variables of Malmendier & Tate (2008). The correlation coefficient is extremely low and the coefficient for the five-item overconfidence variable in an OLS regression with the Cumulative Abnormal Return of the acquiror’s stock around an M&A deal announcement as dependent variable, differs significantly from those for the Longholder and Holder 67 variables. This leads to the conclusion that the five-item overconfidence variable of Schrand & Zechman (2012) should not be used to measure CEO overconfidence in the context of M&A and that the potential data availability and variable construction benefits of the proxy cannot be realized. Additional tests provisionally indicate that the five-item overconfidence variable produces different results as the Longholder and Holder 67 variables in other firm contexts as well.