The replacement of a CEO is one of the most important responsibilities of corporate boards. The most common theoretical underpinning of CEO replacement is related to CEO ability: The corporate board learns about the ability of its CEO from firm performance and other sources of information. If the board’s assessment of the CEO’s ability falls below, say, the expected ability of a replacement candidate, then the board may dismiss the CEO. More recent studies associate CEO replacement with firms’ strategy changes. The board looks for a new CEO either to change the strategic direction of the firm or to implement a new strategy.  In a recent paper, I develop a model to examine the following question: Abstracting from differential CEO abilities and strategy changes, what are other determinants of CEO turnover?

Let’s consider a setting in which the CEO’s initial strategy decisions affect the firm’s future cash flows, potentially beyond the CEO’s own tenure.  Initially, the CEO can expend effort to select or propose a strategy (henceforth: project); later, the CEO in charge needs to implement the project. The board commits to a CEO replacement policy ex-ante, which specifies the conditions under which the incumbent CEO will be replaced.  In case of dismissal, the board will hire a new CEO to see through the project’s implementation. The newly hired CEO will be constrained in the sense that she inherits the predecessor’s project.  To avoid trivial reasons for CEO replacement, I assume that all (potential) CEOs are equally qualified, and they are protected by limited liability.

A common issue in corporate governance is the information asymmetry between the CEO and the board. The board is less familiar with the firm’s daily operations than the CEO is, so it is reasonable to assume that the CEO is privately informed about the quality of a project that he has developed, and the board has to rely on the CEO’s report to assess it.  The higher the quality of the project, the easier its implementation. This joint presence of private information (about the project quality) and the later moral hazard problem leads to the following misreporting incentive of the CEO: The CEO with a high-quality project will have an incentive to underreport the project quality so as to convince the board that ultimate success is more difficult and thus requires a higher bonus for implementation.  To ensure truthful reporting, the board would need to offer the CEO a menu of contracts that rewards the CEO a lower fixed wage and a higher bonus (“steeper incentive pay”) after the CEO reports the project is of high-quality. The reason is that the CEO with a high-quality project is more likely to succeed and hence find a steeper incentive pay more attractive. In contrast, the CEO with a low-quality project would pick a contract that gives her a higher fixed wage but a lower bonus.

If the board commits to retain the CEO (even after the CEO reports that the project is of low-quality), then there is a natural lower bound on the incentive pay – the incentive pay rewarded to the CEO reporting a low-quality project has to be steep enough to motivate the later implementation, and the incentive pay rewarded to the CEO reporting a high-quality project has to be even steeper to induce truthful reporting.  At the same time, the limited liability constraint imposes a lower bound on the fixed wage. With both components – the fixed wage and incentive pay – bounded from below, the CEO with a high-quality project may be left with excessive rents in the sense that the rents are more than sufficient to motivate the selection of the initial project.

If, instead, the board adopts a performance-contingent CEO replacement policy that replaces the CEO after the CEO reports the project is of low-quality, then the reward structure is different: Now there is no need to reward this CEO any incentive pay, because she will depart and receive only the fixed compensation in the form of severance pay. This in turn relaxes the lower bound imposed on the incentive pay rewarded to the CEO reporting a high-quality project (to ensure truthful reporting).  This relaxation on the incentive pay may reduce the excessive rents earned by the CEO with a high-quality project.

Therefore, even though replacing the CEO is always ex post inefficient (since no potential CEO is better than the existing CEO), it can be beneficial ex ante to replace the CEO after poor performance, because then the board doesn’t have to pay the CEO with a high-quality project excessive rents to induce her to reveal that the project is an easy success. That is, instead of “information-based entrenchment” as suggested by the prior literature (Laux 2008, Inderst and Mueller 2010), I show a countervailing effect that  the CEO’s private information (combined with the later moral hazard problem) may lead to her dismissal more often than the ex-post efficient benchmark. This result helps explain Taylor (2010)’s claim base on his empirical findings  (though insignificant) that  “boards in large firms actually fire more CEOs than is optimal for shareholders ex post.”

The model generates a number of empirical predictions regarding the sensitivity of CEO turnover to firm performance, severance pay, and CEO compensation. In particular, the model predicts that firms’ life cycle or industry characteristics are important determinants of the sensitivity of CEO turnover to performance. For start-up firms, or innovative industries, in which project-selection is relatively more important,  CEO turnover is predicted to be less sensitive to performance than for mature firms or  less innovative industries. In addition, my paper predicts that financially healthier firms will have lower sensitivity of CEO turnover to performance and offer more generous severance packages to their executives. These predictions are generated purely from an optimal contracting point of view, even though they may seem to be consistent with the “managerial power” view that boards are captured by powerful executives.

This post comes to us from Professor Xiaojing Meng at NYU’s Leonard N. Stern School of Business.  It is based on her recent paper, “Information, Incentives and CEO Replacement,” available here.