CEO Turnover: Governance, Games and Real Options
The decision a Board of Directors (a board) makes to dismiss or retain its CEO is one of extreme importance in its role of representing shareholder interests and maximising shareholder value. This thesis presents three independent but highly related studies pertaining to the dynamics between a board and its selected CEO in deciding to retain or replace an incumbent manager. The first study presents a theoretical model of CEO turnover that is examined in order to develop new empirical testable predictions. The model employs a learning process for perceived CEO ability that offers new insight into the dynamics of the problem. We find empirical support for the theoretical predictions that: (1) if a CEO sends high noise performance signals to the board relative to the pool of possible replacements, the probability of turnover will be less sensitive early on in the manager’s tenure and more sensitive later due to the learning process; (2) the probability of turnover for a CEO who has a lower level of initial uncertainty relative to a pool of possible replacements will be less sensitive to performance because the CEO will need to be considered high quality to get the position in the first place; and (3) there is empirical evidence to support the notion that ongoing volatility in the board’s estimate of a manager’s ability plays a role in the updating process of ability assessment by a board. Recent empirical work has indicated that board-induced CEO turnover is a function of industry business cycles and not just relative performance evaluation. The literature notes that this could be because: (1) CEOs may optimally be rewarded or punished for peer group performance if a CEO’s actions affect peer performance; (2) boards receive more, or better information in industry downturns than they do during booms; or (3) boards misattribute industry performance to CEO ability. The literature concludes (3) largely due to the results not being sensitive to CEO tenure, where high tenure CEOs should have proven themselves in good and bad times. There is however no theoretical framework to help interpret these empirical findings and we consider conclusions incomplete. It is well established in the macroeconomic literature that downturns are highly correlated with increased levels of uncertainty, and as a result firm behaviour is impacted. The second study in this thesis presents a model of board-induced CEO turnover that allows analysis under two stochastic state variables: (1) perceived managerial ability; and (2) precision of the perceived ability. We use the constructed model to show that, following shocks that increase uncertainty, the probability of turnover for high tenure CEOs may be higher or lower than low tenure CEOs depending on the board’s estimate of CEO ability. This casts doubt on conclusions made from the findings of the empirical literature. The final study presented by this thesis is my job market paper. It presents a new game of performance-induced CEO turnover that analyses CEO turnover decisions in a context where the CEO and the board both have meaningful options. We show that under certain conditions a CEO has the incentive to lock in a high level of perceived ability, through good firm performance, and exercise their option to leave for other roles and increased prestige. This creates an upper and lower threshold for performance-induced CEO turnover. The lower threshold relates to the board’s option to terminate a low-quality CEO and the upper threshold relates to a high-quality CEO’s option to leave the firm. The upper threshold creates a credible threat for the board that affects its decision making. We define two types of threats: (1) the persistent threat, where the firm is one where the incumbent and any replacement will have an upper threshold; and (2) a unique threat, where only the incumbent has the upper threshold and replacements are taken from a pool of candidates who do not have an upper threshold. We estimate that both threat types have a negative impact on firm value. Empirically we find that consistent with the theory, the probability of a turnover event increases following positive relative accounting performance for small firms and firms with young CEO’s, indicating upper threshold constraints for these two groups.