| This study investigates a firm's optimal CEO firing decision and its subsequent effect on CEO compensation in a dynamic model with moral hazard and learning. Based on perceived CEO talent, I explicitly solve for the cutoff point below which a firm will fire its CEO and show that the cutoff depends on variance of assessed talent. As a result, CEO turnover depends on both CEO performance and variance of his performance. Using risk as a proxy for variance of CEO talent, I find empirically that risk, not firm's past accounting or stock price performance, has the most significant impact on CEO turnover. Further, it is the firm-specific risk, not the market risk which affects the CEO turnover decision. I also find that after controlling for the risk, there exists a trade-off between the first two moments of stock price performance in CEO turnover decision. In addition, I solve for the optimal linear compensation contract and compare pay-for-performance sensitivity (PPS) over CEO's tenure (as well as across different firms) to examine how the pressure from firing affects the pay-for-performance sensitivity. Empirical results from accounting earnings show that the pay-for-performance sensitivity is smaller for CEOs who face larger firing pressures. This is consistent with the model predictions. However, the empirical results from stock returns are not statistically significant. |
