Questioning Performance Pay
Performance pay is tricky. At a very basic level, it challenges the notion that corporate managers, as fiduciaries of the firm, should “renounce all thought of self” as it places their self-interest at the forefront of the decisions they make on behalf of the corporation. Performance pay is designed with the hope that it will align those managers’ personal interest with the goal of shareholder wealth maximization as we concede that we cannot simply trust managers to selflessly pursue the interests of others. Indeed, it may do more harm than good to the extent it gives managers both the permission and the means necessary to profit personally from corporate success without suffering in the face of corporate failure.
Last week, at the National Business Law Scholars’ Conference in Columbus, Ohio, I heard Michael Dorff present his book, Indispensable and Other Myths: The True Story of CEO Pay, forthcoming from the University of California Press this spring. In it, he argues that performance pay for CEOs is not effective to enhance firms’ values. This is true, he claims, in part because CEOs generally do not strongly influence corporate return and in part because performance pay is not effective to improve the performance of creative or analytical tasks, that is, exactly the kinds of tasks we expect CEOs to carry out. He cites numerous empirical and psychological studies to support his thesis that performance pay is ineffective at best and harmful at worst.
Scholars have long debated whether performance-based pay is the result of efficient contracting or is evidence of self-interested managerial power. The empirical evidence is mixed. Even the psychology studies Dorff relies on can be challenged by studies reaching conflicting findings (for instance, in her book about introversion called Quiet, Susan Cain cites studies suggesting that many extroverts, such as many corporate CEOs, are strongly motivated by monetary rewards while introverts, such as many academics, tend to be more motivated by fear of punishment). That we have not found a satisfactory empirical answer to these questions does not at all suggest that we should stop looking. The findings Dorff presents make a compelling case for rethinking executive compensation entirely.
In a paper that was focused specifically on the efficacy of inside debt compensation, such as deferred compensation and pension benefits, Brian Galle and I argued that increasing the complexity of executive pay packages is likely to make them less effective motivators. The harder it is for an executive to “do the math” to figure out how a given decision will affect compensation, the more difficult it is for that compensation outcome to affect the decision. In such instances, executives may lean on unreliable heuristics or choose one element of pay, such as stock options, to try to maximize. Focusing on one element of a finely honed group of incentives will render the incentives meant to balance against that one element meaningless. Studying the effect of complex pay packages on executive behavior would be yet another interesting endeavor for those seeking to answer the questions performance pay poses.
Some parts of Dorff’s argument ring particularly true. Many of his criticisms of pay using fixed options echo the complaints Warren Buffett has against it (as explained in The Essays of Warren Buffett, discussed in a symposium on this blog last month). Both Dorff and Buffett point out that one-size-fits-all performance pay will neither inspire nor discipline all managers. Buffett makes a particular effort to ensure that performance-based pay for managers of Berkshire’s subsidiaries corresponds closely to elements of firm value that the individual manager can control herself. Berkshire does not compensate its executives with stock, rather, it pays cash bonuses that the executives may then use to purchase stock. Both Dorff and Buffett (as well as Lynn Stout) fear that options compensation encourages cheating by managers in order to push quarterly earnings reports to meet expectations or to perform other maneuvers to keep the stock price artificially high. Dorff also points out that options compensation does not control for market- or industry-related fluctuations. While I agree that that is a serious problem with any compensation that relies on stock price alone for its value, that seems easy enough to fix by simply rewarding managers for their performance relative to industry- or market-wide performance.
While I share much of Dorff’s skepticism about the efficacy of performance pay, particularly options pay, there might be good reasons not to throw the baby out with the bathwater just yet. While most CEOs may not necessarily touch every part of the company the way Steve Jobs did or Warren Buffett does, they do a significant and important job for the company. If we are going to fire a CEO when the company underperforms, it should not be unreasonable to reward her when it does well. Setting high goals for a CEO does increase effort, which may lead to better results for the firm. The key is choosing the right goals, of course, and making sure to focus on aspects of business performance that the CEO can control or at least heavily influence.
One suggestion for better-calibrated performance pay might be to use less predictable metrics, or focus on more than one factor, with no one factor being particularly dominant. Then, the board or compensation committee (captured though they are) could exercise discretion in deciding how well the CEO did and how best to reward or punish that performance through compensation. The CEO would not be able to game the system, because it would not necessarily be clear ex ante that one element of performance would trump all others when bonus time comes. If stock price is to be one metric the board may use in awarding a discretionary bonus, the stock’s performance for the prior year should be weighted to account for industry- or market-wide shifts. Many professionals are paid using discretionary bonuses, including lawyers. Each firm’s board or compensation committee should understand how to tell whether its CEO is doing a good job. That information, even though it may vary by firm, should be central to designing a predictable, yet not manipulable, performance compensation scheme.
Though I was treated to a preview in the brief presentation, I have not yet read Dorff’s book. It will be published in the spring. I look forward to studying his arguments more completely. He may make indeed make a compelling case to discard performance pay entirely in the book. I’m not sure we (firms, shareholders, academics) are all the way there yet, but I think there is certainly room to move from our heavy dependence on fixed options (or pay with any securities) in order to incentivize and reward CEOs more effectively than we do now.