 |
Home | Previous Page | |
|
||
 |
|
|
|
||
 |
|
|
|
|
|||||||||||||||
|
|
|||||||||||||||
|
|
Speech by SEC Staff:
|
This paper was prepared for the Distinguished Speaker Address at the Conference on "Agency Problems and Conflicts of Interest in Financial Intermediaries" sponsored by the Federal Reserve Bank of New York, the Dice Center for Research in Financial Economics of the Ohio State University and the Journal of Financial Economics on December 3, 2004 in Columbus, Ohio. I gratefully appreciate helpful discussions with Cindy Alexander, Eli Berkovitch, Jonathan Glover, Ronen Israel, Allan Meltzer, Oded Sarig, Jonathan Sokobin, and Per Stromberg and the comments of participants at informal presentations at the InterDisciplinary Center (Herzliya, Israel), the Washington, D.C. Alumni Chapter of Carnegie Mellon's Tepper School of Business and the Office of Economic Analysis of the Securities and Exchange Commission. However, these individuals are not responsible for any of the views, interpretations or errors herein. The Securities and Exchange Commission disclaims responsibility for any private publication or statement of any SEC employee or Commissioner. This presentation expresses the author's views and does not necessarily reflect those of the Commission, the Commissioners, or other members of the staff. |
The issue of executive compensation and specifically whether senior executives are paid too much and the nature of the impact of the economic incentives confronting senior management have become highly contentious in recent years. The economic analysis of principal-agent relationships and contracts sheds much light on this issue and will be the focus of my presentation. The purpose of my talk is to explore what we know from principal-agent theory about executive compensation and to identify some puzzles and problems that would benefit from more detailed exploration by thoughtful scholars. I see this as of potential interest and relevance not only to the academic community, but to the regulators of the financial markets.
Underlying this context is the reality that the income and wealth levels of senior executives are very large compared to those of most of the population. Furthermore, these have grown in relative terms over the last several decades.1 Consequently, in the context of the broader society at least some of the attention on executive compensation may be motivated by concerns about fairness and even envy of the very successful individuals who obtain these positions. There also are important efficiency issues that underlie the interest in executive compensation-after all, if executives are paid "too much" (either on an absolute basis or too much incentive-based pay) economic inefficiencies may result from the distortion of decisions undertaken on behalf of major firms.
Of course, the executive positions in question tend to be highly desirable. Indeed, there are many aspiring candidates for the posts (even internal to many of the organizations seeking leaders), which raises the question of whether the firms need to pay so much-especially given that there appears to be an "excess supply" of candidates for an individual position. The individuals in question are not perfect substitutes-in the language of economics there is considerable heterogeneity in the package of skills that the candidates bring to these positions. This in turn raises the question of whether there are sufficient ex ante identifiable differentials in performance among prospective senior executives to justify the observed magnitudes of compensation. The issue is whether the truly superior individual can be identified ex ante based upon their relative track records and experiences.
The ultimate outcomes experienced by a firm are highly variable. Given that efficient compensation has the senior management sharing in the performance of the firm (a theme we'll say more about later), the executive's compensation itself will be highly variable. Of course, much of this variability may be outside the direct control of the executive (e.g., reflecting market-oriented variability). In light of the nature of the compensation programs utilized, such as options and restricted stock, the senior executive's payoffs are highly variable and skewed, motivating some of the comments about the fairness of executive compensation. In the face of extremely positive outcomes-and without externally verifiable evidence of the executive's contribution--this again reinforces the fairness and envy issues. The form of the compensation programs widely used also raises the question about the absence of relative benchmarks within the compensation structure (e.g., adjusting for the ex post performance of other firms in the industry or just the market as a whole).2
One of the classic applications of "agency theory" in economics is to a manager (the agent) who operates a firm on behalf of its owners (the principal). Both the level and form of managerial compensation have been the subject of considerable interest. Before considering the form of compensation I think it is helpful to address why the level of compensation should be high. Principal-agent theory helps identify forces that can lead to higher compensation than one would otherwise expect.
Clearly, high compensation (without addressing the issue of whether the compensation is too high) is necessary to attract talented individuals, who typically possess outstanding alternative opportunities. There are several more subtle aspects of the level of compensation. For example, a high compensation level is a way for the employer to signal that the incoming senior executive will have sufficient control over resources to be able to perform his job in an appropriate fashion. This type of control is difficult to verify or contract upon ex ante, but the level of compensation (which itself is contractible) may be an effective "signal" as the willingness of the firm to offer high compensation may be optimal only if the manager has sufficient control over internal firm resources.3
In understanding the senior executive's compensation, his alternative opportunities also are important. Indeed, a generic alternative for the executive is to consume leisure. Economists typically view leisure as a "normal" good, i.e., one whose demand is increasing in the individual's wealth (and analogously, the disutility of effort/work rises with wealth). Hence, to induce an executive to accept a position with significant responsibilities requires compensation that increases in his wealth. This complements but differs from the more traditional rationale for persistence in the compensation of highly talented executives, namely that one learns about productivity over time-as the market learns more about the skills of relatively successful executives their relative compensation should rise.4
The observation that sophisticated executives with higher wealth have higher value for alternative uses of their time applies not only to allocating their time to leisure, but also to allocating it towards the management of their financial assets. This latter interpretation is particularly relevant for some senior executives in the financial services arena, whose skills would be especially germane to that task. To manage this specific problem, one well-known Wall Street partnership effectively required partners to loan much of their financial wealth to the firm for a preferred return. While this provided a source of capital in a capital intensive business, I think that the more basic interpretations were in managing potential allocation distortions in both project choice and time effort (between the firm and its partners), keeping the executive's "eye on the ball" in building the franchise rather than in managing personal financial wealth. Indeed, this type of restriction on their own portfolio management can induce key partners to retire after they have accumulated sufficient capital (perhaps after several years as partner), especially given inherent risk aversion by these partners. Of course, an interesting question in its own right is how to encourage optimal retirement.
From an agency theoretic perspective another important source of high pay levels for senior executives is that these positions are "prizes," typically internally allocated to the most successful performers.5 This induces considerable effort among those competing for these prizes (the high effort of "associates" at large metropolitan law firms who are competing to join the partnership is along these lines). A "tournament" structure can even induce efficient allocation of effort (see Lazear and Rosen (1981) and Green and Stokey (1983)) at lower levels in the organization and under restrictive conditions even perform as well as a direct contractual solution.6
Finally, an interesting observation about the connection between the level of executive pay and the incentive of the executive to take risk has its roots in the comparative static analysis of a portfolio theoretic model. It is well known in portfolio theory that the incentive to take risk can vary with wealth. Consider an investor who is deciding what amount of his assets to invest in a risky security versus a risk-free asset. Of course, if the investor had constant absolute risk-averse (exponential) preferences, the dollar demand for the risky asset would be independent of wealth. The investor's willingness to bear risk will increase with wealth as long as the investor has decreasing absolute risk-averse preferences. By paying the executive more, his incentives to take more risk on behalf of the firm are increased (a wealth effect) under decreasing absolute risk aversion (the usual case). In effect, the risk aversion of the manager can be overcome by paying the executive a lot (!) as well as by using instruments such as options that directly increase the incentive to bear risk (a substitution effect).
An important reason that executive pay levels are often viewed as very high is that the assessment is performed on an ex post basis and focuses upon managers with high realized compensation. This in turn raises the question of why incentive pay is so crucial for senior executives.
One answer is that it is extremely important to incent naturally risk-averse executives (with limited wealth) to undertake valuable risky projects for the firm. Otherwise, the executive will be relatively more risk averse about their actions on behalf of the firm than "diversified" shareholders would prefer. The executive's compensation structure typically includes such features as restricted stock, options and bonuses to increase the executive's willingness to bear risk and overcome the manager's inherent risk aversion. These types of features overcome the executive's natural risk aversion relative to the shareholders of the firms and the capital market as a whole. In addition, from a signaling perspective managers are anxious to signal their ability (type) by showing interest in incentive compensation. The pressures from a signaling equilibrium can force the executive to largely accept variable contingent compensation (otherwise, he is not sufficiently confident about his ability to be a plausible occupant of these positions!), despite his natural risk aversion. This discussion emphasizes both the moral hazard and adverse selection rationales for performance-based compensation.
There is much at stake in managerial decisions. In fact, Jensen and Murphy (1990) argue that because the CEO receives only a small proportion of the value added to the firm ($3.25 per $1,000), there is not the optimal incentive.7 However, this ignores the risk aversion of senior executives and their limited wealth relative to the capital market as a whole, which limits the amount of risk that the senior executives should bear in an optimal solution (see Haubrich's critique (1994) of Jensen and Murphy (1990)).8 To what extent is it efficient for senior executives to have a significant stake in improvements in the value of the firm, when they are inherently quite risk averse to the marginal value of the firm? I would suggest that the issue is not whether the executive receives the full incremental value of the firm (which is not realistic or economically efficient given the executive's risk aversion and limited wealth), but whether marginal incentives are substantial for the executive (relative to his own rather than the firm's wealth). In fact, the interesting analysis in Aggrawal and Samwick (1999) shows that how pay-performance sensitivity varies with the riskiness of the firm is consistent with managerial risk aversion. The marginal incentive identified in Jensen and Murphy (1990) is considerable, especially given the limited wealth of the CEO compared to the capital market, and that the CEO is only the leader of the management team, i.e., others on the team also need incentive compensation.9 The CEO is simply the leader of a hierarchy or team (e.g., Holmstrom (1982)), but the incentive issues may extend beyond the leader of management.
An issue that has engaged a lot of popular attention is whether executive pay and especially the incentive portion of the compensation are too high. While moral hazard and adverse selection arguments highlight the optimality of incentive-based compensation designs, the observed levels can be too high as well as too low. In its simplest terms, a function with a unique interior maximum increases until the maximum is reached and then decreases at higher levels of the underlying variable. The optimality of positive incentives does not imply that observed incentives are too low and that more incentives would be better. In fact, there are both good and bad consequences of additional incentive compensation and the trade-off among these determines the optimal solution.
While I have emphasized the potential value of significant incentives, there also are important reasons for incentive pay to not be too large as executives can be over-incentivized. For example, the actual compensation is paid out of the firm's resources (this is analogous to it being costly to the firm to pay an executive too much). The incentive compensation can lead for the firm's executives to have too much incentive to take additional risk despite a lack of improvement in the value of the firm. Another potential adverse effect of too much incentive pay is that the executives may have large incentives to manipulate the short-run value of the firm. In fact, a recent paper by Bergstresser, Desai and Rauh (2004) provides important cross-sectional information about short-term adjustments in the assumed rate of return on pension assets and executive compensation programs.
Of course, even if there are excessive incentives in executive pay in the United States today (or especially during the dot.com era), it can be optimal to increase incentive compensation in other contexts (such as for European executives or for United States executives in other eras). This is clearly a context-specific assessment.
Options are a key component of executive compensation. One striking feature of these programs is the discreteness of vesting dates and option exercise dates. The option grants tend to occur infrequently (e.g., annually or quarterly). This seems to be rather puzzling. Why is that an efficient form of compensation, for example, as compared to a more continuous set of vesting dates, option exercise dates and option exercise prices? Given that relevant economic decisions are being made more frequently (continuously?), it is hard to rationalize compensation that is so discontinuous. Discontinuous compensation is vulnerable to manipulation, without obvious advantages over a smooth compensation profile. The nature of managerial risk aversion reinforces this point and there is not an obvious incentive benefit to the discontinuous compensation structure. In many contexts it is optimal to impose risk on a risk-averse manager due to incentive benefits, but that type of rationale does not seem to be an obvious explanation for infrequent and lumpy option grants.10 An interesting quantitative question that this suggests is how much the firm could reduce the executive's compensation, while producing the same incentive benefits or the same expected utility for the manager.
In fact, the discussion above emphasizes that discontinuous or spiky compensation may suggest a design flaw in a variety of agency contexts. For example, a salesman often receives discontinuous compensation based upon whether he reaches a periodic quota. If the salesman perceives that the likelihood of hitting the threshold is too low, then the salesman may lack suitable marginal incentives. By extension I wonder if even discontinuous marginal incentives can be problematic; this intuition seems like the economic motivation underlying smooth pasting conditions in "free boundary" option exercise problems.
This discussion also indirectly speaks to a central aspect of many options programs in practice. If the option moves too far out of the money, the firm will sometimes reset the exercise price by granting new replacement options ("reload" options). While this is often criticized and suggests that the original grant understated the intended compensation, it may be necessary to provide the desired marginal incentives.11 In understanding the incentive structure I think it is useful for firms to focus upon incentives that are "renegotiation proof." This would represent a substantial shift from current practice.
A final aspect of executive compensation that deserves much more attention is the role of board of directors and the extent to which a board mitigates existing incentive problems and to what extent it creates incentive problems of its own. The role of the board is fundamental, but perhaps not adequately emphasized.12 The board hires the Chief Executive Officer (CEO) and is responsible for managing succession. Yet there is typically a lot of interaction between the board and the CEO. The CEO often tries to influence the composition of the board. This discussion emphasizes that "conflicts of interest" may be crucial. An interesting academic analysis of the importance of conflict of interest in the determination of CEO compensation is given by Bebchuk and Fried (2004).
The board sets the compensation for senior management, including the CEO, by a Compensation Committee. "Benchmarking" is often used, though the approach does not seemed designed to produce a lot of effective information. Interestingly, the board is self-perpetuating (with new members selected by a Nominating Committee) and the auditor reports (in part) to the board (Audit Committee).
The incentives of the board are important. Unfortunately, board members are often disengaged.13 This could be either a consequence or cause of low compensation. The fiduciary responsibility of board members leads to some "sticks." Are there sufficient "carrots" as well? I think that both positive and negative incentives are important. It is striking how little compensation is offered to board members relative to senior executives; just to illustrate as a mathematical exercise, a board member may spend about 1/20 of the time of the CEO on firm business, but receives far less than 1/20 of the compensation. This seems to me to be one of the most fundamental puzzles in management compensation. Board members often have comparatively far less at stake, unlike the high-powered incentives for the CEO at the helm. Perhaps from this perspective, the lack of attention by some board members is not very surprising. One way to formulate the issue is to ask, why is the division of compensation between the board and key executives so skewed? The skewing in compensation suggests that there is little responsibility in being a board member, the position is very attractive, there are many substitutes for the prospective board member or that the non-pecuniary benefits (such as networking) of being a board member are considerable. But there are many issues about the tradeoffs with respect to board compensation. These arise with respect to both the level of compensation and the implied incentives.
What is the right tradeoff with respect to the setting of board compensation? The answer depends, in part, on the nature of the board's contribution-as a resource for the CEO or as an independent agent of investors. These can conflict, although they are not mutually exclusive and may even be complementary. The balance that is achieved depends upon the level of compensation and incentives.
If it is desirable to have a relatively detached/independent board, for example, the current type of compensation may be appropriate. Because of the role of the CEO in selection or retention of board members, there is a natural reluctance of the board to "rock the boat," if the board positions are desirable and well paid. Of course, this is an argument against paying board members too much. In particular, one downside to higher board compensation could be a reduction in board independence (independence itself may be desired, as long as the board is motivated by the shareholder's interests; this in turn emphasizes the importance of identifying the role of the board and how that should be influenced by senior management). An alternative approach for recruiting engaged board members is to select large stockholders who are strategic (rather than passive) investors. This could at least help mitigate some of the public goods problems that are central in corporate governance.
Board members are very dependent upon the information that they receive from the management team and the outside auditor. This raises the issue of how can one ensure that the outside auditor is sufficiently forthcoming with the board. Consequently, it is important to ask good questions. For example, one former board member told me he would ask the auditors what did they discuss with the management team (or among themselves!) that they did not discuss with the board? Indeed, if the board were too adversarial with respect to the management team, management's incentives to communicate to the board would be greatly reduced. This points to a delicate aspect of trying to divorce the CEO from the board selection process-it is important to ensure that the executives are sufficiently forthcoming with their own boards. The contrast between real and formal authority highlighted by Aghion and Tirole (1997) helps illustrate that a less than completely independent board can be optimal in some settings-to the extent that the CEO has the discretion to make most decisions, he will have strong incentives to invest in obtaining the relevant information to make good decisions. Harris and Raviv (2004) emphasize the importance of communication and combining information by the board and senior management in a setting in which each has a partial information advantage and some decisions are delegated to the board to control agency concerns. The communication aspects and the interaction between the board and the CEO point to some of the subtlety in structuring the incentives for effective board behavior.
While my presentation suggests solution to a number of the following issues, I think that there are a number of important matters that are deserving of further study. The broad issue concerns whether there are adequate safeguards in the setting of executive compensation so that agency conflicts are adequately mitigated.
Aggarwal, R. and A. Samwick, 1999, "The Other Side of the Trade-Off: The Impact of Risk on Executive Compensation," Journal of Political Economy 107, 65-105.
Aghion, P. and J. Tirole, 1997, "Formal and Real Authority in Organizations," Journal of Political Economy 105, 1-29.
Bebchuk, L. and J. Fried, 2004, Pay without Performance: The Unfilled Promise of Executive Compensation, Harvard University Press, Cambridge, MA.
Bergstresser, D., M. Desai, and J. Rauh, 2004, "Earnings Manipulation, Pension Assumptions and Managerial Investment Decisions," working paper, Harvard Business School.
Chaochharia, V. and Y. Grinstein, 2004, "The Transformation of U.S. Corporate Boards-Recent Evidence," working paper, Cornell University.
Fich, E. and A. Shivdasani, 2004, "Are Busy Boards Effective Monitors?", European Corporate Governance Institute Working Paper 55/2004.
Gibbons, R. and K. Murphy, 1992, "Optimal Incentive Contracts in the Presence of Career Concerns: Theory and Evidence," Journal of Political Economy 100, 468-505.
Gibbons, R. and K. Murphy, 1990, "Relative Performance Evaluation for Chief Executive Officers," Industrial and Labor Relations Review 43, 30S-51S.
Green, J. and N. Stokey, 1983, "A Comparison of Tournaments and Contracting," Journal of Political Economy 91, 349-364.
Haubrich, J., 1994, "Risk Aversion, Performance Pay and the Principal-Agent Problem," Journal of Political Economy 102, 258-276.
Harris, M. and B. Holmstrom, 1983, "A Theory of Wage Dynamics," Review of Economic Studies 49, 315-333.
Harris, M. and A. Raviv, 2004, "A Theory of Board Control and Size," working paper, University of Chicago and Northwestern University.
Holmstrom, B., 1982, "Moral Hazard in Teams," Bell Journal of Economics 13, 324-340.
Jensen, M. and K. Murphy, 1990, "Performance Pay and Top-Management Incentives," Journal of Political Economy 98, 225-264.
Lazear, E. and S. Rosen, 1981, "Rank-Order Tournaments as Optimum Labor Contracts," Journal of Political Economy 89, 841-864.
Margiotta, M. and R. Miller, 2000, "Managerial Compensation and the Cost of Moral Hazard," International Economic Review 41, 669-719.
Merton, R., 1973, "Theory of Rational Option Pricing," Bell Journal of Economics and Management Science 4, 141-183.
http://www.sec.gov/news/speech/spch120304cs.htm
| Home | Previous Page | Modified: 12/03/2004 |