CEO overconfidence: Towards a new measure

• Published in: International review of financial analysis Vol. 84; p. 102367
• Main Authors: Hatoum, Khalil, Moussu, Christophe, Gillet, Roland
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.11.2022; Elsevier
• Subjects: Bayesian networks; CEO overconfidence; Conditional probabilities; Corporate investments; Quantitative Finance; Corporate investments; C110; CEO overconfidence; Bayesian networks; Conditional probabilities; G400
• ISSN: 1057-5219; 1873-8079
• DOI: 10.1016/j.irfa.2022.102367

Bayesian network theory is used to construct a novel probability-based measure for CEO overconfidence. This measure is estimated by studying the probabilistic correlation between CEO overconfidence and several CEO- and firm-specific determinants of overconfidence, that have been documented in the literature. Using S&P 500 firms over the period 2007–2017, we show that the established Bayesian network model has a high fitting and prediction accuracy of CEO overconfidence. This novel measure of CEO overconfidence can be used to conduct empirical studies in corporate and behavioral finance. It also provides a tool to improve decision-making in firms and corporate governance. •Bayesian network theory is used to construct a probability-based measure for CEO overconfidence based on several CEO- and firm-specific determinants of CEO overconfidence.•Probability of CEO overconfidence helps measure the frequency of overconfidence for CEOs•Probability of CEO overconfidence can be used to conduct empirical studies in corporate and behavioral finance. It also provides a tool to improve decision-making in firms and corporate governance.