The Good, the Bad, and the Lucky: CEO Pay and Skill

In sports, the best-paid players are also the most skilled. The same can’t always be said about business. Are business CEOs highly paid because they are skilled, because they capture the board of directors, because they set their own pay, or because they are just plain lucky?


Very few business topics are as hotly contested as the salaries of chief executive officers of public firms. The amount that CEOs are paid and the structure of their pay is frequently debated in the popular press, television programs, proposed legislation, political campaigns, magazine cover stories, and academic research. Outrage over CEO pay has forced important changes at major firms: Jack Welch of General Electric was forced to give back part of his pay when investors complained it was excessive, and Richard Grasso of the New York Stock Exchange was forced to resign altogether when the details of his pay were revealed.

One obvious reason for the interest in CEO pay is its striking increase. In 1992, the average CEO of an S&P 500 firm earned $2.7 million. By its peak in 2000, average pay for these CEOs had grown to over $14 million—an increase of more than 400 percent. The increase in CEO pay is even more striking in relative terms. Twelve years ago, CEOs at major U.S. corporations were paid 82 times the average earnings of a blue collar worker; last year they were paid more than 400 times the average earnings of a blue collar worker. This huge increase in executive compensation has been especially controversial because CEOs are sometimes paid large sums even as the firm’s results deteriorate; CEOs at WorldCom, Tyco, and Enron collected over $100 million on average in the year before scandals broke at the firms or the firm collapsed.

These facts and the spectacular governance failures at important firms have caused many to conclude that the process for setting CEO pay and, more generally, the governance of public firms is badly broken. Critics conclude that CEOs are overpaid because they have too much influence over the board of directors who should be monitoring them on the shareholders’ behalf, and too much influence over the committee that sets their pay. And that independent directors and consultants hired to advise the board have relatively little interest in safeguarding shareholder interests. In this view, CEO pay is the product of badly functioning corporate governance. Such arguments also suggest that cases of excessive CEO pay reflect a systematic social problem of “fat-cat” CEOs skimming money at shareholders’ expense.

Others are more sanguine, arguing that the process for determining CEO pay works generally well, and that the problems arise from a few bad apples. Some argue that any problems with CEO pay have not erased the comparative advantages of the U.S. system, given the relatively good performance of the U.S. economy. Others argue that any pay distortions reflect the perceived impact of accounting and tax rules rather than governance flaws. In this view, compensation problems reflect breakdowns in particular firms, but do not indicate  a general problem in compensation or in public firm governance. 

Which version of CEO pay is accurate? Prior attempts to distinguish between these competing views examined whether CEO pay changed as the firm’s performance changed. The idea is that CEO pay should be linked to changes in a firm’s value in order to align managers’ interests with shareholders’ interests. However, lots of things affect a firm’s value, including industry and market trends that the CEO has no influence over. One shouldn’t confuse CEO brains with a bull market.

Surprisingly, for all the research on CEO pay, there is little evidence on the basic question: Is CEO pay related to CEO skill? Are highly paid CEOs better than their more poorly paid peers? That question is not only important in evaluating the appropriateness of CEO pay levels, but also in evaluating whether the governance of public firms is badly broken. If pay and skill are related, high salaries will not necessarily be evidence that fatcat CEOs have captured the board. Just as sports teams may decide to pay high salaries to attract or retain valuable players, boards may also pay more for especially talented managers. If, on the other hand, CEO pay and skill are unrelated, the process for setting CEO pay is likely to be badly broken, and highly paid CEOs may be overpaid, indicating more general problems with the governance of public firms.


Tackling the Question

To examine whether CEO pay and skill are linked, we introduced a new measure of skill. The intuition behind our measure of skill is straightforward: firms run by good CEOs should consistently do better than firms run by bad CEOs. If a firm has been performing poorly relative to its peers, a skilled CEO will consistently be more likely to reverse the firm’s fortunes, while a bad CEO will be more likely to continue the poor performance.

If, on the other hand, a firm has been performing well relative to its peers, a good CEO will consistently be more likely to continue the good performance, while a bad CEO will increase the chance of a bad outcome. Thus, using our definition of skill, good CEOs will reverse poor performance and continue positive performance. Bad CEOs will continue poor performance and reverse positive performance. And lucky CEOs are highly paid, but perform no differently from their lower-paid peers.

This methodology has a number of significant advantages over prior methods. We control for the firm’s past performance to be sure that we compare CEOs with other CEOs in the same position. Because a firm’s opportunities may be a function of past performance, we avoid comparing a CEO who is in charge of a firm that performed well in the past with another CEO who is in charge of a firm that performed poorly.

One reason that there has not been research on the relationship between CEO pay and skill is that it is very difficult to factor out the effect of the CEO. Our methodology allows us to better isolate the effect of the CEO for reasons we explain in the full paper. Another advantage of our methodology is that we can make some initial conclusions about overall pay levels, something generally difficult to do. We conclude that highly paid CEOs are overpaid if their performance is consistently worse than their lower-paid peers, because in this case there is no justification for the higher pay.

Our basic research strategy was straightforward. In a well-functioning labor market, skilled CEOs should earn more because they have more valuable outside opportunities and will earn more rewards. Therefore, we first measured the pay of every CEO for whom we could get pay data. We measured their total pay in year one, including cash, bonus, and the value of any options granted. We then looked at the firms’ performance in later years to see if the pattern was consistent with the idea that highly paid CEOs are more skilled. We defined high pay as those CEOs paid more than the median in each industry.


CEO Skill and Pay

We found evidence that highly paid CEOs are in fact more skilled when firms are small or when the CEO has relatively greater ability to affect the firm’s performance, due to reduced regulatory

or industrial constraints on managerial discretion. This link between pay and skill is especially strong if there is a blockholder (an individual or firm that owns a significant block of stock) with the incentive to monitor management.

By contrast, we found that highly paid CEOs who operate in large firms subject to environmental constraints perform worse than their more poorly paid peers. These CEOs are more likely to continue poor performance, and, surprisingly, even to reverse good performance. This negative relation between pay and skill is exacerbated in the absence of a large shareholder to monitor management. This suggests that in large firms without blockholders or where the CEO has relatively less ability to affect the outcome of the firm’s performance, there is less reason to pay a high wage. Firms that nevertheless pay high wages in these instances may also suffer from governance problems that produce poor performance.

In addition, we found that a new CEO who receives higher pay than the departing CEO is more likely to reverse prior poor performance relative to CEOs who are paid similarly or lower than the departing CEO, if the pay package has high incentive pay. Strikingly, if the highly paid new CEOs pay lacks incentive pay, the CEO is more likely to continue prior poor performance.

Finally, we show that being able to spot a skilled CEO is very valuable. We created equal-weighted stock portfolios that hold firms with highly paid CEOs and short firms with poorly paid CEOs. When pay and skill are related, we found that such a portfolio generates an annualized abnormal return of 8 percent between 1994 and 2001. When pay and skill are unrelated, there is no such abnormal return. The mean return of firms in which pay and CEO performance are linked exceeds the mean return of firms in which pay and CEO performance are not linked by almost 8 percent.