جبران خسارت و جذب نیرو: دانشگاه های خصوصی در مقابل شرکت های خصوصی

This paper attempts to shed light on the continuing debate regarding executive compensation by comparing the income of S&P 500 CEOs with that of the presidents of elite private universities. The results reveal that university presidents are paid only a fraction of what CEOs are paid less than 5% in 2000. Nonetheless, universities are able to attract leaders with qualifications and accomplishments equivalent to that of the most distinguished CEOs. Furthermore, university presidents appear to be willing to work as hard and as much in the interests of their constituents as corporate CEOs despite the lack of any meaningful incentive clauses in their contracts. These results suggest that the standard principal-agent model used in evaluating compensation needs to be extended significantly before it can be applied to situations in which a few select people are recruited for highly paid jobs that offer the chance to lead major institutions.

In his extensive review article, Murphy (1999) observes that by 1998 CEO compensation had become a political “hot button” and had attracted widespread scholarly interest. Little did he know that the interest was about to accelerate sharply. The huge compensation packages paid to executives prior to the collapse of companies, such as Enron and Global Crossings, have led to unprecedented attention being focused on executive compensation. Even the President of the United States has weighed in on the subject. The popular press routinely portrays CEOs as grossly overpaid. However, the scholarly literature demonstrates that reaching such a conclusion is fraught with complexities. Given the massive resources controlled by major corporations, bad or improperly incentivized, leaders can cost firms billions of dollars. In addition, as Lazear and Rosen (1981) originally explain and as Martin (2001) emphasizes, the pay packages of company leaders may serve functions beyond compensating and incentivizing the CEO. They can function as tournament prizes that provide incentives for employees throughout the organization who are in a position to compete for the prize, not just the executive who obtains it. Despite the extensive research on CEO compensation, Murphy (1999) points out that some of the most basic questions have not been adequately addressed. In particular, there is little evidence to document that increases in financial incentives lead CEOs to work harder, smarter, and more in the interest of shareholders. More importantly, it remains unknown whether companies could find and recruit people who would work as hard, smart, and effectively if they were to cut CEO compensation by a factor of two or more. A fundamental empirical roadblock to answering these questions is developing an appropriate baseline for assessing the level of CEO compensation. One approach, popularized by Crystal (1991), is to compare CEO compensation with that of staff employees at the same company. By this standard, CEO compensation has risen dramatically. Murphy (1999) reports that in 1970 the average S&P 500 CEO made about 30 times more than the average production worker. By 1996, that factor had increased to 90 times, excluding compensation received from exercising stock options. If option exercises are included, the factor rises to 210 times. Another approach, taken by Abowd and Bognanno (1995), Cosh and Hughes (1997), Kaplan (1995), and Zhou (2000), among others, is to use international data. Although the specific conclusions of the international studies depend on the sample period and the comparison countries, every study finds that US CEOs are the most highly compensated both in absolute terms and relative to staff employees. While comparisons to staff workers and international CEOs show that US CEOs are very highly compensated, it does not demonstrate that they are “too” highly paid. It is possible that circumstances unique to the American economy and American companies lead to the higher level of compensation. For instance, size and total compensation are related in virtually every country, and American companies tend to be larger in terms of market capitalization. In addition, Abowd and Bognanno (1995) stress that international comparisons can be affected by taxes, purchasing power parity, and public benefits. Nonetheless, after adjusting for these factors they still find the US CEOs the most highly compensated. From a less positive perspective, it is also possible that American corporations are particularly susceptible to what Bertrand and Mullainathan (2000) call “skimming”1. The skimming hypothesis holds that the separation of ownership and control allows CEOs to gain effective control of the pay setting process. Both because of entrenchment, such as packing the board with supporters, and because of the complexity of the pay process, many CEOs are effectively able to set their own pay with little oversight from investors. Skimming may be more likely to occur at American companies because of the lack of large shareholders and reduced governmental oversight. To provide a new source of empirical evidence, this paper compares data on the compensation of S&P 500 CEOs to the compensation of the presidents of America's leading private research universities. America's private research universities are among the most complex and most productive institutions in our society. They are responsible for training the nation's gifted young people. The research they produce is fundamental to basic understanding across an enormous variety of disciplines. That research has also been a critical element in promoting the country's economic growth. Many of the innovations that have led the revolutions in computing and biotechnology have their foundations in university research. Furthermore, major research universities typically operate extensive medical centers that provide state of the art medical care, as well as producing innovative medical research. In addition, universities run hundreds of specialized centers dedicated to widely diverse intellectual goals, many of which are controversial. To fund these diverse activities, the budgets of major universities run into the billions of dollars. Endowments at the elite schools exceed US$10 billion. The scope of university activities is rivaled by the difficulty of managing them. Added to the normal problems that arise when managing any large, diverse enterprise, are a host of university specific issues. For instance, there are constant public and political pressures. Not only are the political views of the basic constituents: faculty, students, and alumni often divergent, but they typically are held with great conviction by highly intelligent, successful people who are used to getting their way. Furthermore, the large contingent of young people on campus tends to make universities centers for protest and political unrest. In addition, systems such as faculty tenure add another layer of complexity to managing a university. The option of dealing with conflict by simply terminating recalcitrant employees that is available in most private corporations is not possible in a university context. The faculty cannot be fired, and the students and alumni are “customers”. Finally, universities do not have a clear objective function equivalent to “maximize shareholder wealth”. As a result, dispute frequently arises as to the role of the university, which adds to management difficulties. Overall, there is a little dispute that leading an elite private university is one of the most difficult jobs in America, comparable in complexity to managing a major corporation. It is not surprising, therefore, that the qualifications necessary to become president of a research university are unusually stringent. Presidents generally must be distinguished in both scholarship and administration. To satisfy the faculty, an advanced degree and a prominent record of publication is required. To satisfy the alumni, successful involvement in business or government is seen as a prerequisite. Finally, to do the job, highly developed people skills and administrative skills are necessary. Once on the job, the hours and pressures are brutal. Past Stanford President, Gerhardt Casper, once said that a successful university president must have three qualities—stamina, stamina, and stamina. During his presidency at Harvard, Neil Rudenstine became so exhausted that he had to be hospitalized. Rarely does a president of an elite private university serves for more than 10 years. The job is simply too taxing. Nonetheless, the success of the university depends on the performance of the president. The ability of the president to raise funds, attract faculty, and keep the political peace among dozens of other functions, determines the relative success of competing schools. In short, elite private universities require people with talent, skill, and a record of accomplishment comparable to that of the people chosen to run America's major corporations. This raises two basic empirical questions related to compensation. Are private universities able to attract leaders with qualifications comparable to that of major corporate leaders? If so, how do they compensate and recruit such people? To answer these questions, the remainder of the paper is organized as follows. The next section briefly reviews the principal-agent model, which serves as the theoretical basis on which much of the CEO compensation literature is based. The goal of that review is to highlight some issues that the principal-agent model fails to address that make data on the compensation of university presidents particularly interesting. Section 3 presents empirical data on the compensation of private university presidents in comparison to corporate CEOs. The final section discusses the implications of the empirical data and presents the conclusions.

The results presented in the previous section show a dramatic difference between the compensation of S&P 500 CEOs and the compensation of the presidents of America's leading private universities. The difference is so large, a factor of more than 33 in the year 2000, that it swamps concerns that might arise about how compensation is measured or how the sample is chosen. The pay of university presidents and corporate CEOs is not even in the same ballpark. It may be argued that universities do not pay as much as corporations because they cannot. It is possible that constraints, such as the nonprofit nature of the organization or the power of certain stakeholders, such as the faculty, sharply limits the compensation that a university president can receive. But that is not the issue. The point here is that despite these constraints, or whatever other factors limit the compensation that universities can offer their presidents, the universities have been able to attract remarkably accomplished people who have performed their duties with great energy and diligence. A detailed study of how universities are able to attract people of apparently equal quality to CEOs while paying less that 5% of the compensation that CEOs received is beyond the scope of this paper. Nonetheless, conversations with university recruiters reveal that they stress many of the factors emphasized by organizational theorists. To make the discussion of these factors more precise, assume that the utility function of potential presidents and CEOs is of the form Ui(c, x), where c denotes consumption and x denotes the nonpecuniary benefits of the job discussed earlier. The superscript, i, reflects the fact that the utility derived from a given level of consumption and nonpecuniary benefits varies across individuals. For university recruiting, the critical issue is the behavior of the marginal utility of consumption at very high levels of c; in particular, what is the cross-sectional variance of the marginal utility of c, relative to the marginal utility of x, for various combinations of c and x. Universities operate under the apparent assumption that at a salary of US$500,000 (in year 2000 dollars) plus benefits and perks, there exist highly qualified individuals whose marginal utility of consumption is low enough relative to the marginal utility of the job's nonpecuniary benefits, that universities can compete with the corporate sector for their services. On this basis, universities specifically search for people who value the nonpecuniary aspects of the job, such as the chance to make a social contribution and the opportunity to lead a great institution. They also search for people who are motivated to achieve by other factors in addition to monetary compensation. This is consistent with the fact that university compensation contracts are not designed to “incentivize” the president. Despite the fact that the compensation is largely fixed, independent of both effort and results, the universities expect their presidents will put forth a “best effort”. The supposition is that there are people in the applicant pool with a sense of personal pride and character that motivates them to work hard, and in the interests of their constituents, even with a fixed salary contract. In terms of utility analysis, this is equivalent to saying that recognition for a job well done is an element of x. The desire to “consume” this recognition is apparently, in the eyes of universities, sufficient incentive for certain individuals. The assumption underlying university recruiting is that if that applicant pool is large enough, qualified people who place a high relative value on x can be found. The success they have had in recruiting presidents indicates that this assumption is well founded. The question remains as to what extent the university experience can be carried over to corporations. Universities may be particularly well suited to attracting individuals who value the nonpecuniary aspects of their product, namely research and education. Furthermore, universities draw their presidents from people in academic life. Such people presumably made the decision at some earlier time to pursue a career with lower pecuniary compensation in return for the nonpecuniary benefits. People who enter commerce, on the other hand, presumably value pecuniary compensation more highly on average. Nonetheless, major corporations have huge pools from which they can select their leaders. Even if searches were limited to internal candidates, S&P 500 companies employ approximately 50,000 people on average. Presumably, that pool is sufficiently large that highly trained, skilled, and motivated leaders could be located even if the current level of compensation were cut in half, particularly if the search placed added emphasis on finding people who valued the nonpecuniary benefits of the job. Overall, the data presented here add anecdotal support to the skimming theory of Bertrand and Mullainathan (2001). The differential between CEO compensation and university president compensation is so large that it is hard to imagine that a gap of that magnitude is required for companies to attract qualified leaders. It appears instead that CEOs, perhaps with the consent of boards that are often dominated by other CEOs, are playing an important role in setting his or her own compensation. The ability of universities to attract accomplished leaders, while paying only a small fraction of what corporations, does suggest that boards could do more to maximize shareholder wealth. The university data also cast light on the continuing debate regarding whether attempting to align executive compensation with shareholder wealth actually causes executives to work harder, more intelligently, or more in the interests of shareholders. The empirical literature on this issue including Leonard (1990), Abowd (1990), and Kahn and Sherer (1990) is mixed. Furthermore, the literature is difficult to interpret because the linkage between specific actions taken by the chief executive and changes in shareholder wealth is so weak.6 If stock returns are the measure of performance, as in Leonard and Abowd, then correlations between prior executive compensation and subsequent stock returns tend to be low because so many other factors influence stock returns. Furthermore, the current level of the stock price presumably already reflects the CEO's expected actions so that stock returns will be affected only by unexpected decisions. If direct evaluations are used to measure management performance, as in Kahn and Sherer, then the problem of evaluating the evaluations arises. The university data suggest that the focus placed on incentive alignment by the agency theory model may be exaggerated. Leaders of major institutions are a highly select group of people who, even in the case of the lower paid university presidents, are highly compensated compared to members of society at large. Given a declining marginal utility of consumption, for such individuals the nonpecuniary aspects of their job appear to be an important part of the total compensation. Unfortunately, to this point, there is little research on how CEO decision-making and work habits are influenced by variation in compensation of any type. This should be a fertile area for future work because the behavior of university presidents suggests that marginal compensation is a relatively unimportant determinant of their behavior on the job. In fact, previous Stanford President Gerhardt Casper said he never worked harder than during his years as Stanford's President. At his retirement, furthermore, the university lauded the remarkable energy and enthusiasm he brought to the job. All of this occurred despite the lack of incentive clauses in his contract. It is possible that the tournament model of Lazear and Rosen (1981) could be extended to explain at least some of the findings. Suppose that people who enter academic life do so in part because of the nonpecuniary benefits offered by the life style and the opportunity to teach and do research. In that case, their motivation is likely to be largely orthogonal to the opportunity to become president of a university. Put in another way, paying the President of Harvard more is unlikely to cause a young Physics professor to do better research. In the corporate context, on the other hand, it may be more reasonable to assume that wealth and power are more important motivating forces, and that from the start many young executives have their eye on the prize of becoming a CEO. In that situation, offering a very large prize for winning the tournament could well provide stronger incentives throughout an organization. A major deficiency of the tournament theory is that it does not explain the contingent nature of CEO compensation. Most tournaments, such as those in golf and tennis, simply have large cash prizes. If the goal of CEO compensation is to motivate managers coming up through the organization, it is unclear why adding contingent risk to the prize is helpful. Of course, it could be argued that compensation serves two functions: it is a prize for future CEOs and an incentive contract for the current CEO. More subtly, option-based compensation could also perform a political function. Suppose it is the board's goal to pay huge expected compensation to the CEO, either because it is designed to function as a tournament prize or simply because the CEO has effectively captured the board. In either case, very large cash compensation is likely to cause more political problems for the firm in dealing with stakeholders such as shareholders and employees than a contract that is, at least allegedly, tied to performance. For instance, in its survey of executive compensation, Business Week (2002) reported that using their measure of pay, the compensation of Henry Silverman, CEO of Cendant, jumped 64% to US$251 million in 2001. Business Week goes on to say, “Shareholders didn't mind because their returns rose 104 percent.”7 Such a statement makes no sense from a value maximization perspective. Shareholders should always mind if the CEO is overpaid, independent of their returns in any given year. If a CEO, comparable to Silverman, could have been had for US$50 million, it would have added US$200 million to shareholder's wealth. It is also worth noting that the 104% return in 2001 represented, in part, a recovery of the stock from depressed levels associated with shareholder litigation. For the 5-year period from 1996 to 2001, the company whose Beta is near one, underperformed the S&P 500 Index by more than 30%. Thus, Silverman's experience adds further support to Bertrand and Mullainathan's (2001) conclusion that in many circumstances option-based pay provides extraordinary compensation for luck when stock prices jump, without an offsetting payback of compensation when prices fall. Unfortunately, the political economics of option-based compensation is not well understood. Whereas, hard data are available on the amount of compensation granted, information on the motivation for granting such compensation is much more murky. Therefore, this remains an exciting, but difficult, area for future research. In summary, the results reported in this paper point to two basic conclusions. First, elite private universities are able to attract presidents with capabilities, skills, and experience equal to that of top corporate CEOs despite the fact that they currently offer less than 5% of the average CEO's compensation. This adds to the growing literature, which suggests that boards are not doing all they can to maximize shareholder wealth when recruiting and compensating CEOs. Second, the agency theory model commonly applied in compensation analysis has some significant shortcomings when applied to the most senior positions. When recruiters have the ability to search a large pool of potential applicants for a position of visibility, power and leadership that, even in the worst case, will be highly compensated, factors come into play that the agency theory model ignores. Among these factors are the nonpecuniary benefits of job and the personal characteristics of the competing candidates. The success of universities in recruiting presidents who typically have served with great energy, distinction, and dedication, even without “incentivizing” contracts, demonstrates that these other factors are important.