What Is CEO Overconfidence? Evidence from Executive Assessments

Steven N. Kaplan is the Neubauer Family Professor of Entrepreneurship and Finance at the University of Chicago Booth School of Business; Morten Sorensen is Associate Professor of Finance at Dartmouth University Tuck School of Business; and Anastasia Zakolyukina is Associate Professor of Accounting at the University of Chicago Booth School of Business. This post is based on their recent paper, forthcoming in the Journal of Financial Economics.

“Unskilled and unaware of it.”

— Kruger and Dunning (1999)

Overconfidence is prevalent among corporate executives, and a number of academic studies have blamed overconfidence for distorting executives’ investment and merger decisions. Traditionally, these studies have measured overconfidence using executives’ personal option holdings, using the so-called Longholder measure that was introduced in a seminal study by Ulrike Malemendier and Geoffrey Tate. However, no direct link has been drawn between overconfidence as a psychological trait and the option-based Longholder measure. The findings thus could be confounding overconfidence with other personality traits correlated with but different from overconfidence. It is also possible that the executives’ option holdings are rational, for example capturing a response to governance constraints on executive compensation or private information, rather than reflecting overconfidence.

A challenge for research is then benchmarking extant measures of overconfidence against executive traits. Data on executive traits are hard to find. Our paper leverages uniquely rich data on personality assessments for more than 2,600 candidates for management positions from a consulting firm that specializes in assessing top management candidates ghSMART. Using these assessments, we examine which traits are behind the Longholder measure of overconfidence.

For each ghSMART candidate, we track their subsequent career to determine which candidates become the CEO of a public company. We identify 67 such candidates. Of these 67 CEOs, nine (13%) are identified as overconfident by the Longholder measure. This approach allows us to compare the personalities of CEOs classified as Longholders and non-Longholders.

We find that Longholder CEOs have significantly lower scores on a number of characteristics: having a strong network, being organized, being calm under pressure, moving fast, sticking to commitments, having strong analytical skills, being creative, having a strong work ethic, having good listening skills, and being open to criticism. CEOs identified as overconfident by the Longholder measure thus exhibit these characteristics to a lesser extent than other CEOs. Lower scores on many of these characteristics were shown to be related to overconfidence by prior psychology literature.

Because the specific characteristics measured by ghSMART are highly correlated, we use factor analysis to identify the main variation in the data. Personalities can be summarized by four factors that explain 54% of the variation across characteristics. These factors are (1) general talent, (2) execution (vs. interpersonal), (3) charisma (vs. analytical), and (4) strategic (vs. managerial). Interestingly, CEO candidates and hired CEOs score higher on the first factor (more general talent), more negatively on the second factor (more execution), more negatively on the third factor (more charismatic), and more positively on the fourth factor (more strategic).

The main finding in our study is that Longholder is significantly negatively related to the first factor, suggesting that overconfident CEOs tend to have less general talent or ability. This finding is consistent with the classic study documenting Dunning–Kruger effect that shows that lower-ability individuals tend to be more overconfident. The stronger overestimation by less competent people can occur because their lack of competence deprives them of the metacognitive ability to realize they make mistakes. As Kruger and Dunning write in their paper “Unskilled and unaware of it: How difficulties in recognizing one’s own incompetence lead to inflated self-assessments:” “When people are incompetent in the strategies they adopt to achieve success and satisfaction, they suffer a dual burden: Not only do they reach erroneous conclusions and make unfortunate choices, but their incompetence robs them of the ability to realize it” (p. 1121).

Our primary contribution is documenting executives’ traits behind the popular measure of overconfidence. Additionally, we show that, while capturing low general ability, overconfidence as measured by Longholder appears to be empirically distinct from general ability.

The complete paper is available for download here.

Both comments and trackbacks are currently closed.