A Rationale for Hiring Irrationally Overconfident Managers

Abstract

The viability of managerial overconfidence is perplexing since it has been shown to lead managers to erroneous and costly decisions. This chapter addresses this issue by exploring the impact of managerial overconfidence on managerial effort, executive compensation, and the welfare of stockholders and managers. Overconfidence affects managerial effort directly and indirectly. The direct effect is that the optimal effort chosen by managers is positively related to their level of overconfidence. The indirect impact is through the influence on stockholders’ choices of contract parameters. Thus, managerial overconfidence helps mitigate the well-known conflict of interest between managers and stockholders that induces managers to exert effort levels that are lower than the socially optimal levels. We construct a measure of the combined welfare of managers and stockholders and show that it is positively related to managerial overconfidence, thus providing an explanation to the persistence of this bias.

We thank Darius Palia, Orly Sade, and seminar participants at Rutgers University and Bocconi University for comments and suggestions. The financial support of The Sanger Family Chair for Banking and Risk Management, The Galanter Fund, The Mordecai Zagagi Fund, the Whitcomb Center for Research in Financial Services, and The School of Accounting, The Hebrew University are gratefully acknowledged.

Appendix A: Calibration of the Parameters of the Base Case Calculation

We refer to Hek (1999) and Bitler et al. (2005) for the choice of the parameters and the shape of utility functions that depend also on leisure.¹⁸ Hek finds the power utility useful in explaining technological change. Bitler et al. (2005) find that behavior of entrepreneurs can be explained by this type of utility function. We start with our base case U(I, Y) =  − I^α(Y₀ − Y)^β, where α =  − 1 and β = − 0.50. We choose α =  − 1 (a measure of constant relative measure of risk aversion of 2) since this is the parameter that has been widely used in studies of overconfidence: Malmendier and Tate (2005a, b). We examine however also other parameters as robustness checks.¹⁹

We assume that the function μ(Y) is given by μ(Y) = Ln(μ₀ + Y/2) − σ²/2. By the known properties of the Lognormal distribution, the mean and variance of this distribution equal ${\mathrm{e}}^{\left[\mu (Y)+0.5{\sigma}^2\right]}$ and $\left[{\mathrm{e}}^{\left[2\mu (Y)+{\sigma}^2\right]}\left({e}^{\sigma^2}-1\right)\right]$, respectively. The parameters μ₀ and σ were selected so that, in the absence of managerial extra-effort, the expected cash flows per share are 45,000 and the coefficient of variation (i.e., the ratio of the standard deviation to the expected value) of the cash flow per share is 0.3. Since the expected cash flows serve as numeraire, the ratio of the standard deviation of the cash flows per share to their expected value is a surrogate for the standard deviation of stock returns.