Lower Salaries and No Options? On the Optimal Structure of Executive Pay

Abstract

We estimate a standard principal agent model with constant relative risk aversion and lognormal stock prices for a sample of 598 US CEOs. The model is widely used in the compensation literature, but it predicts that almost all of the CEOs in our sample should hold no stock options. Instead, CEOs should have lower base salaries and receive additional shares in their companies. For a typical value of relative risk aversion, almost half of the CEOs in our sample would be required to purchase additional stock in their companies from their private savings. The model predicts contracts that would reduce average compensation costs by 20% while providing the same incentives and the same utility to CEOs. We investigate a number of extensions and modi…cations of the standard model, but …nd none of them to be satisfactory. We conclude that the standard principal agent model typically used in the literature cannot rationalize observed contracts. One reason may be that executive pay contracts are suboptimal.JEL Classi…cation: G30, M52

Performance standards and optimal incentives

“…However, in a survey of the literature on theory and practice of managerial incentives and compensation Prendergast (1999) claims that, ''the theoretical literature has made little progress in understanding the observed (nonlinear) shape of compensation contracts, despite costs associated with nonlinearities.'' More recently, the challenge has been to explain the generalized use of convex compensation functions, such as stock options, as one way to align interests of managers and shareholders (Hemmer et al, 2000;Feltham and Wu, 2001;Hall and Murphy, 2002;Cadenillas et al, 2003;Dittmann and Maug, 2007). But, somehow surprisingly given its wide use, there are few theories about compensation contracts that incorporate standards and thresholds.…”

Reward for failure and executive compensation in institutional investors

“…Our paper is related to previous research on how other non-monotone and convex incentive schemes motivate managers to take desirable risk levels from the principal's standpoint, especially option-like schedules (Carpenter, 2000;Garcia, 2001;Goetzmann et al, 2003;Ross, 2004;Kadan and Swinkels, 2007;Feltham and Wu, 2001;Hemmer et al, 1999;Dittmann and Maug, 2007;Duan and Wei, 2005;Bolton et al, 2010;Coles et al, 2006;Hirshleifer and Suh, 1992) and bonus payment structures (Starks, 1987;Leisen, 2014). At first glance, the convexity involved in the payoff function of this class of compensation schemes should incentivize risk-taking.…”

Changes in probability distributions and the form of compensation contracts

“…First, it can explain why, for an agent with log utility, the optimal contract is concave for a lognormally distributed performance measure (Dittmann and Maug 2007), 5 while it is linear with a normally distributed performance measure (Hemmer et al 2000). Indeed, suppose that π is normally distributed.…”

“…Since the likelihood ratio of the normal distribution is also linear, it follows that the optimal contract is linear in performance (π ). In addition, for a lognormally distributed measure of performance (x = exp(π )), we can write x (x|e) = (ln(x)|e) (Dittmann and Maug 2007). Now consider (7) and (8).…”