I develop a theory of stock-based compensation contracts for the chief executive officers (CEOs) of firms and confront the theoretical predictions with recent CEO compensation data. The model characterizes the optimal contract for a CEO whose reputation evolves as signals of the executive's ability are observed by shareholders. Using various proxies for CEO reputation, I show a positive and economically meaningful relationship between stock-based pay-sensitivities and CEO reputation. The findings are robust to controls for CEO age, firm size, the dollar variability of the stock returns, and industry effects.

Principal-agent theory suggests that a manager should be paid relative to a benchmark that removes the effect of market or sector performance on the firm's own performance. Recently, it has been argued that such indexation is not observed in the data because executives can set pay in their own interests; that is, they can enjoy “pay for luck” as well as “pay for performance.” We first show that this argument is incomplete. The positive expected return on stock markets reflects compensation for bearing systematic risk. If executives’ pay is tied to market movements, they can only expect to receive the market-determined return for risk-bearing. This argument, however, assumes that executive pay is tied to bad luck as well as to good luck. If executives can truly influence the setting of their pay, they will seek to have their performance benchmarked only when it is in their interest, namely, when the benchmark has fallen. Using industry benchmarks, we find significantly less pay for luck when luck is down (in which case, pay for luck would reduce compensation) than when it is up. These empirical results are robust to a variety of alternative hypotheses and robustness checks, and they suggest that the average executive loses 25–45% less pay from bad luck than is gained from good luck.