THE APPARENT PARADOX of highly compensated CEOs and poor job performance has sparked discussion on what the optimal CEO compensation contract should look like. The received wisdom is that observed compensation contracts are market optimal outcomes that address a moral hazard problem present within a principal-agent relationship. This article challenges the traditional approach by exploring two alternative models for observed CEO contracts. It concludes with recommendations on achieving better compensation decisions, namely, through reducing information asymmetry, checking conflicts of interest, and reacquainting directors with managerial incentives.

Challenging the dominant model
The long-held view is that compensation contracts observed in the market place have been designed to give strong incentives for the CEO to exert effort for the interests of the shareholders. Such a principal-agent model has been criticized along several fronts, of which I will examine two in particular. First, the assumption of variable compensation as an incentive for hard work has been compromised in consideration of the fact that considerable systematic risk underlies observed contracts. Instead, under an alternative formulation of the contract, the existence of variable compensation serves to ensure participation in a competitive market rather than to encourage greater effort. Second, the assumption of arm’s length bargaining between the principal and the agent has been challenged in view of the pivotal role that the board plays in the bargaining process. Specifically, evaluating director incentives reveals compelling reasons for the existence of compensation schemes that favour CEO interests to the detriment of shareholders. Following this line of reasoning, some scholars have posited a model of managerial power to explain observed contracts.

Ensuring participation
Empirically, little indexing of performance incentives such as stocks and stock options occurs relative to the market. Without indexing, compensation is tied to absolute rather than relative returns, and the incentive to exert effort is undermined. However, the persistence of the systematic risk can be explained under an alternative formulation. Specifically, the systematic risk in a compensation contract is a mechanism for satisfying the CEO’s participation constraint.

Assuming reasonable volatility within an industry, the spot market value of a CEO’s services rises and falls relatively frequently (since wages ought to be tied to total industry value). Consequently, the reservation wage, that is, the participation constraint, of the CEO is also volatile. To address this volatility, firms can tie more of the CEO’s compensation to market risk. While tying more of the CEO’s reservation wage to variable payment might lead to higher average pay, given the risk averseness of the CEO, there are several advantages to the approach as well. For instance, a variable reservation wage allows companies to accommodate a CEO’s shifting participation constraint while avoiding the costs of turnover from reducing the level of fixed payment. In fact, evidence exists for CEOs being rewarded for factors beyond their control in a range of industries. Consider the case of observed contracts within oil companies, where the pay of CEOs is affected by the price of oil, a variable that is beyond their control. According to this model, the relevant incentive in such a contract is ensuring participation. An increase in the price of oil, for instance, would increase the demand for oil company CEOs, either because of entry or because incumbent firms launch new projects. In response to this, the reservation wage of a CEO would increase, and this increased wage would be accommodated in a compensation structure tied to the firm’s absolute performance, rather than the relative performance with respect to other firms.

Managerial power
Yet another alternative to observed CEO compensation contracts takes as its starting point the relationship between the board of directors and the CEO. Ideally, shareholders would directly oversee management to ensure alignment of interests. However, given the high coordination costs that would be involved in having a large number of dispersed and partial owners oversee management, a board of directors is involved to represent the shareholders. Under the traditional principal-agent model, the directors are completely aligned with shareholders, and the resulting contracts in the market are therefore optimal responses. A strong case could be made for why such theoretical alignment is lacking in the real world. First, directors hold lucrative positions ($116,000 on average among the largest 1,000 companies) and receive substantial perks. As examples of such perks, directors of UAL Corporation can fly free of charge on United Airlines while directors of Starwood Hotels receive complementary stays. Opposing a CEO’s wishes substantially increases the likelihood that a director will not be renominated to the board and will lose these benefits. Second, directors benefit indirectly by being on good terms with the CEO. For instance, CEOs may provide business to companies in which the directors work, or may direct funds to charitable contributions favoured by the directors. Third, directors may land additional lucrative director positions through the help of CEOs who sit on the boards of other companies. Finally, as a board member, and often as the Chair of the board of directors, the CEO may exert significant influence in the identification and selection of board members. Such a scenario would create an inherent conflict of interest between the selected board member’s fiduciary duties and debt of the gratitude owed to the CEO.

Maximizing Compensation Subject to an Outrage Constraint
Given these considerations, some scholars propose an alternative model of managerial power to explain observed compensation contracts. They propose that management seeks to maximize compensation subject to a public outrage constraint. The difference between the managerial power and principal-agent explanations is sharp: The level of pay in the principal-agent approach is set so that the CEO receives at least his reservation utility in order to satisfy his participation constraint, and this keeps him from leaving and going to another firm. In contrast, the level of pay in the managerial power approach is set as high as possible, with the upper bound on pay determined by public relations. Several examples of how existing forms of compensation can be explained using the managerial power model. For instance, stock options and other incentive pay are used as a way of justifying high compensation at a minimum of public outrage. Since options and other incentive pay theoretically align the interests of both management and the shareholders, they can be significantly increased beyond the CEO’s reservation utility without raising many eyebrows.

Moreover, the traditional model requires that stock options be relative to the market to filter out all factors that are beyond the management’s control. In practice however, contracts are written as a factor of absolute (not relative) returns, leaving CEOs with windfall profits during booms, due to no effort (or output) on their part. CEOs also benefit during recessions by having their options “repriced” at lower exercise prices due to “no fault of their own.” As a result, managers are better off with options on absolute prices rather than those based on relative performance. Additionally, CEOs have the freedom to unwind any potential incentive effects within their compensation scheme by cashing in options relatively early and by buying derivatives that have the effect of locking in today’s stock price, thereby offsetting the incentives in option plans. Under optimal contracts, there would be no reason to camouflage CEO compensation from the public through the use of obscure perks, such as special retirement contracts, large loans, ongoing consulting, and so on. Finally, the generosity of severance arrangements are often difficult to reconcile with arm’s length bargaining. For example, when Mattel CEO Jill Barad resigned under pressure, the board forgave a $4.2 million loan, gave her an additional $3.3 million in cash, and allowed her unvested options to vest automatically. These benefits were in addition to her contractually defined severance benefits, which included a termination payment of $26.4 million and a stream of retirement benefits exceeding $700,000 a year. One can think of a multitude of reasons for a company engaging in such activities, but few if any of these are ultimately aligned with the best interests of the company and its true owners. These precious parachutes are suboptimal, unsustainable, and act as a reinforcement for whatever behaviour or condition caused the exit.

Discussion and recommendations
Theoretical principal-agent incentive contracts would optimally require that CEOs be given their minimum participation wage and that they be offered additional incentives to increase shareholder value in a sustainable way. The latter is best achieved through tying CEO incentive compensation to variables that they have control over, such as a company’s performance relative to its industry peers, and doing so in a way that makes the CEO concerned with sustainable value, such as by specifying longer payout horizons. In practice however, we observe that compensation schemes focus on absolute rather than relative performance, and have relatively short time-horizons (for instance, one to three years). This article has outlined two alternative hypotheses to explain such discrepancies. The first focuses on the portion of CEO compensation that is outside his control, and argues that the reason for the existence of such variable compensation is not to offer incentives for harder work, but to ensure CEO participation with the rise and fall of the market. This explanation would be plausible if CEO base salaries were relatively low, such that the remaining part of the CEOs reservation wages were made up for via variable compensation tied to the company’s stock price or profits. However, recent research on 589 CEOs suggests that base salaries are in fact much higher than they should be. If true, then an explanation that puts a great amount of emphasis on satisfying the participation constraint is not satisfactory.

An alternative managerial power model has also been discussed. This model essentially argues that the conditions for optimal contracts within the principal-agent model have not been satisfied. Most importantly, board directors are not perfect representatives of shareholders, as mentioned above. In fact, they have good reasons to prefer CEO interests to the detriment of shareholders. Thus, the alternative model formulates observed contracts as compensation maximizing subject to a public outrage constraint. The notion of public outrage is not well-specified, but the model is consistent with discrepancies in CEO compensation alluded to above. Specifically, it accounts for why CEOs receive excessive stock compensation tied to absolute rather than relative returns, and it also accounts for narrow-time horizons. That is, if CEOs discount future payments sufficiently highly, they will push for payments based on near-term horizons. Extending this model, one could argue that once compensation levels have been set by a number of major players, they become an important reference point for recruiting talented CEOs. As such, even well-meaning directors that attempt to accurately represent shareholder interests find it difficult to break ranks by negotiating theoretically optimal compensation contracts that are much less attractive than those observed in the market. That is, in order to attract and retain CEOs, they consciously or unconsciously give in to the dominant paradigm of compensation contracts that extract rents from the shareholders.

Even if the managerial power model feels too cynical to be true, we can use its insights when considering recommendations that would be more broadly useful as well. For instance, the model posits that substantial information asymmetry exists between the CEO and board members on the one hand, and shareholders and the press – and other information dissemination devices – at large. To remedy this, CEO compensation contracts for public companies should be made available for the public to review. In addition to laying out the details of base and variable compensation, camouflaged perks such as access to large loans, special retirement, and ongoing consulting ought to be clearly laid out as well. Board compensation must also be brought under scrutiny, with compensation and perks clearly mentioned. Given the additional information, if any public “outrage constraint” does exist, it will now be able to function more effectively, thereby reducing any managerial rents that might have previously existed. Second, to the extent that it doesn’t already exist, regulators should institute a formal process to ensure that directors on the compensation committee are independent of the CEO. By increasing information and reducing conflicts of interest, one would expect compensation contracts closer to market optimum levels. The precedent of regulatory transparency and control of market transactions already exists. It is curious that comparable enforcement has not percolated down to the compensation of CEOs who are at the helm of the companies that constitute the market.

In addition to the recommendations already mentioned, the financial crisis has also underscored the importance of having a board that understands managerial incentives. Board members should therefore reacquaint themselves with the principles behind incentive contracts. According to these principles, effective incentive contracts should be based on controllable rather than uncontrollable variables, encourage long-term rather than short-term viewpoints, have the CEO share in both upside and downside risks, and prevent the CEO from unwinding performance incentives. An example of a compensation scheme that would be consistent with such a model would be as follows: offer stock (both upside and downside risk) which vests over a number of years (long-term horizon), and is indexed to the market (focus on controllable variables via relative returns); additionally, require the CEO to always have a certain level of exposure to the stock, and ensure contractually that he does not hedge this exposure (that is, prevent the CEO from unwinding the performance incentive). If directors keep these principles top-of-mind, and make it a point to discuss how existing compensation contracts depart and/or adhere to these principles, they will improve their chances of avoiding suboptimal contracts as a result of “rational” thinking (my competitors do it, so it is logical for me to do so as well in to attract and retain talent), psychological “biases” (this is what I’ve seen recently, and it must therefore be valid), or computational complexity (the ideal solution has too many interacting variables, so I will settle for a simpler alternative). One way of having boards adhere more closely to these original principles is to require that new directors undergo compensation training periodically, or once at the very least. Such training would help moderate the force of the status quo – to the extent that it is suboptimal – and set a sound baseline for future compensation decisions.
 

Yousuf Haque is an associate at Surje & Company, based in our Toronto office.
 

Related articles

CEO Performance: Absolute vs. Relative

Is Pay for Performance Substantiated by the Data?