Duration Analysis of CEO Turnovers Using Proportional Hazard Model

To analyze CEO turnovers in large U.S. corporations over the period 1981 to 1990, this paper adopts a minimum chi-square estimation of proportional hazard model. A simple specification test of the proportional hazard assumption is also used. Empirical results indicate that elderly CEOs are more likely to be turned over (retirement effect) and worse-than-average CEOs face a lot higher turnover risk (disciplinary effect). Interestingly, performance is found to have non-proportional effects on CEO turnovers across tenure periods. At an earlier tenure as CEO, only good performance matters, increasing the chance of survival. On the other hand. at a later tenure. only bad performance makes a difference, enhancing the possibility of turnovers.