Kostas Katsanevas
The Problem
In publicly traded firms one of the most hotly contested issues is executive remuneration. Fear of concentrated and consolidated power goes back to the early 19th century when Thomas Jefferson, a staunch critic of corporations announced his hopes of crushing “in its birth the aristocracy of moneyed corporations” (Nichols, 2010). This rather old and multifaceted problem stems from a fickle balance between shareholder interest and executive performance. This has made it to national debate, as a matter of controversy, especially since average worker pay has stagnated in recent times, further perpetuating income inequality (Holmberg & Schmitt, 2014).
Nonetheless, academic understanding of executive payment structures has, since the early 1970s, heavily relied on basic economic assumptions, namely that agents are fully rational, self-interested and motivated merely by financial means. This has created a misleading realization - in order to encourage executives to perform in the interests of the shareholders, companies must offer performance based compensation, usually in the form of stock-options, which has only facilitated in inflating remuneration.
This on its own is a problem, as it signals a large transfer of wealth from shareholders to the management due to the fact that stock-options created out of thin air are not free. Organizations would be wiser to raise worker wages or reinvest it in the company. Nevertheless, the real risk lies in short-term performance goals and how these act as incentives for top executives to diverge from shareholder interest. The 2004 Fannie Mae scandal is a notable example of executive fraud and a direct result of payment compensation motivated by wrong incentives (Day, 2006).
This ill-informed pattern adopted by many large firms worldwide, does not take into account nonpecuniary factors that could play a larger role in executive behavior. If the question then is how to align CEO performance with shareholder interest in lieu of misinformed incentives, behavioral economics might have the answer. According to a seminal piece by Alexander Pepper, highly paid executives are less homoeconomicus than previously believed (Pepper, 2015), but before the importance of behavioral principles is discussed, lets digress for greater context.
The Proposal
The year is 1991 and presidential candidate Bill Clinton proposes new policy condemning highly paid executives. At the time like all normal businesses, corporations were allowed to deduct all compensation rewarded to their top management as an expense. In an attempt to stop large remunerations, Clinton proposed reform in tax code, preventing companies from writing off anything over $1 million under executive salary – unless they hit specified performance goals (Epstein & Javers, 2006). The idea had an academic backbone, one founded in Jensen & Murphy’s influential piece, Performance Pay and Top-Management Incentives, which in short argued that the way CEOs were being paid was wrong. CEO pay they reasoned, should be based on performance (Jensen & Murphy, 1990). This got a lot of attention, potential campaign-winning kind of attention.
Bill Clinton, spoiler alert, won the elections making good on his promise in 1993 and compensation committees in top firms immediately began to work on ways they could change pay structure to tie it to performance, the solution was stock-options. To say that Clinton’s policy backfired would be an understatement, as the law’s first decade witnessed average CEO compensation in Standard & Poor’s stock index companies soar from $3.7 million to $9.1 million, an increase of 146% (Bebchuk & Grinstein, 2005). Like this, one comes to realize that the problem lies in the way that CEOs were being paid not how much they made.
Theories of the Firm
The literature on theories of the firm is vast, all of which depend on rational economic principles: tournament theory (Edward P. & Sherwin, 1981), institutional theory (Balkin, 2008) and human capital theory (Combs & Skills, 2003) just to name a few. The most prominent answer on deconstructing and understanding organizations, agent behavior and executive compensation is agency theory (Tosi & Gomez-Mejia, 1989). Agency theory is a supposition that explains the relationship between shareholders and executives, addressing conflicts of interests between the two (Jensen M. , 1983).
It follows all the aforementioned requirements of any rational choice model. Most importantly, it presupposes that executives “are risk neutral, because they can balance their portfolios, risk averse, because the potential wealth effects of the employment relationship are significant, [that their] utility is positively contingent on pecuniary incentives and negatively contingent on effort, and that time preferences are calculated mathematically according to an exponential discount function” (Pepper & Gore, 2012). This proved to be an elegant model to work with, but its predictive power is not all its made out to be. Studies by Frydman & Saks among others have time and again found little evidence of a significant link between executive pay and performance (2010). Enter behavioral insights for the rescue.
A 1998 theoretical study proposed the first behavioral modification to agency theory. Wiseman & Gomez-Mejia proposed a behavioral agency model (BAM) of executive risk taking that combines elements of the principle (shareholder) and agent (executive) relationship with reference points, problem framing and other postulates of Kahneman and Tversky’s Prospect Theory (Wiseman & Gomez-Mejia, 1998). First it emphasizes that maximizing agent performance is key in the principle-agent relationship and that monetary rewards might conversely cause a reduction in agent motivation. Second, behavioral agency theory is concerned with risk and uncertainty, arguing that agents are primarily loss averse.
This suggests that below their reference point, agents will be loss averse making them increasingly more likely to take short-term risk, and above it, less likely to do so. Thirdly, BAM assumes agents discount time using a hyperbolic discount function rather than an exponential one, implying that future rewards are heavily discounted, leaving room for preference reversals. Finally, the fourth modification takes into account the agents’ perception of equitable compensation. Their motivations it argues, fluctuate based on market norms and peer referents, a phenomenon dubbed as “inequity aversion” by Fehr & Schmidt (1999). Fortunately, this isn’t just theory as an empirical study conducted by Alexander Pepper found agent behavior to be aligned with Wiseman & Gomez-Mejia’s approach.
Pepper, a professor of management practice at the LSE, surveyed 756 senior global executives, asking them questions taken straight out of behavioral economic and psychological literature, to better understand the relationship between reward and motivation (Pepper, Gore, & Crossman, 2013). Four major points stand out from this research. First, most pecuniary incentivized theories suggest top executives are risk neutral whereas Pepper found them to be risk averse, 30% more on average. Secondly, it turns out that money isn’t everything, CEOs would rather give up 30% of their pay for a more satisfying role in their job. Pepper’s third notable finding suggests that executives discount long-term rewards at a rate just over 30%, making them heavy time discounters. Finally, to everyone’s surprise, fairness is an important aspect for executives in their work. That is to say, CEOs are more concerned with how much they make in relation to their co-workers than as an absolute measure. Could it be that this elite group of people are, well, people?
Redesigning Compensation
Traditionally, behavioral approaches target cognitive processes, the kinds that form our biases. If the goal is to have executives make better informed decisions, ones that align with shareholder long-term goals, it would be wise to consider behavioral agency theory in conjunction with Pepper’s empirical findings. Such a theoretical background supported by empirical results reasonably point towards a behaviorally informed policy in redesigning executive payment structures. Taken all together, three cognitive biases found in BAM and Pepper, can be used to create more successful compensation packages: loss aversion, time discounting and perceptions of equity.
If agents calculate gains and losses according to Kahneman & Tversky’s “S-shaped” value function (1979) as BAM predicts, their perceived impact of a loss looms greater than its equivalent gain. Traditionally, compensation packages reward bonuses to executives when they reach a specified short-term goal (Jachimowicz, 2013). Alternatively, incentives should be framed as losses in terms of a charge to be enforced if the CEO does not reach that target. Simply by reframing extrinsic motivation as a forgone loss, CEOs will now be more incentivized to meet short-term goals set by their shareholders once again aligning performance with interest.
Secondly, as discussed above time can have a strong influence over preferences. Research has shown that individuals tend to discount the future a lot, preferring $100 today over $120 tomorrow exhibiting a bias for the present (Laibson, 1997). Paralleling Prospect Theory, hyperbolic discounting is a descriptive approach that changes slightly the actual discount factors so as to make discounting initially fast, then have it slow down as time periods pass. Hyperbolic discounting naturally captures the short term impatience effect. More to the point, hyperbolic discounting, the kind of time dilation experienced by our top executives, lowers discount rates when such events are separated by a larger time frame. That is to say, preference of $100 over $120 hyperbolically decreases as these two choices scurry further into the future.
Imagine a compensation package that exploits hyperbolic discounting. Once an executive has forgone a loss by reaching a shareholder specified short-term goal, an extra bonus should be put in a mutual fund that cannot be accessed right away. This in turn will make executives more long-term oriented which is great for shareholder long-term goals. The UK has implemented this in the form of long-term incentive plans (LITPs) for senior executives, although I am not convinced that they had hyperbolic discounting in mind at the time. LITPs were first recommended by the Greenbury report (1995) as an alternative to stock-options. Since, they have taken the form of “award of deferred shares that vest over a 3 year period conditional upon the satisfactory achievement of a number of financial performance targets” (Pepper, Gore, & Crossman, 2013).
A third modification to current compensation packages uses a combination of the last two insights made in both the predictions of behavioral agency theory and Pepper’s research: agents’ perception of equitable compensation and that more money might not lead to higher performance. Once these two ideas are digested they intuitively translate to one policy: compensation packages should give CEOs the opportunity to give up a portion of their bonus, which in this scenario is paid into a mutual fund, to co-workers. To make this even more effective, opting-out should become the default.
Public policies in various countries in Europe take advantage of the effort put in when opting out of something like organ donation. In the presumed-consent states like Austria, where individuals are donors unless they register not to be, the effective rate of consent is 99.98%, relative to opt-in states like Denmark which dwindle at a mere 4.25% (Johnson & Goldstein, 2003). Nevertheless, as far as BAM is concerned CEOs are averse to inequity while Pepper’s study showed that executives are willing to give up to 30% of their pay for additional satisfaction in their work.
This is also supported by the notion of reciprocation. Take the concept of trust, an abstract, innately human construct that Kenneth Arrow himself argued is the basic social cohesive, the lubricant of social systems at many levels, from friendships and marriages to business partnerships (Arrow, 1974). In a trust game there are two participants, one decides how much money to endow the other, while the other then watches that money multiply by a certain amount, then the endowed must decide how much to return back to the endower.
Contrary to rational choice models, experiments have rarely reached perfect subgame Nash Equilibrium, meaning that almost always, the endowed returned some amount of money illustrating that human beings tend to reciprocate (Cesarini et al., 2008). It is reasonable to assume then, that executives will reciprocate and give portions up to 30% of their bonuses to their co-workers.
Conclusion & Discussion
For clarity’s sake, a recap. Current executive remuneration packages benefit only senior executives often times to the disadvantage of the corporations themselves as stock-option bonuses can be costly to shareholders and the company as a whole. Bill Clinton’s 1993 tax reform had good intentions and was backed by classic agency theory rhetoric whose predictions rely only on rational choice principles. Behavioral modifications to agency theory by Wiseman & Gomez-Mejia have found empirical grounding, which in turn leads to the presumption that CEOs might not be perfectly rational, self-interested super-beings.
A revamping of the design of executive compensation packages as envisioned by behavioral agency theory in conjunction with Alexander Pepper’s research takes advantage of loss aversion, time discounting and perceptions of equity, which could prove effective in changing agent behavior subsequently balancing executive performance and shareholder long-term interests.
Recent critiques of behaviorally informed policies argue that any manipulation of the subconscious via changes in choice architecture, use of defaults, of reframing incentives and the like are not actually what they claim to be – liberty preserving (Sunstein & Thaler, 2008). The idea of using people’s biases against them, to some, sounds like the end of free will. However, this idea is a philosophical one, as anyone could look to Hobbes’s social contract (1651) as the first transfer of rights and in turn free will, to a greater entity, an institution that is meant to protect us from each other and ourselves.
That the state by nature is paternalistic is commonsense, otherwise it ceases to exist. Additionally, the idea that behaviorally informed policy extends paternalism is an argument, but then again, that is what society has set up government for. The question then becomes not that its morally wrong if the state use BE to change behavior – this is what it was designed to do: find new and improved ways to protect us in our image, as we mustn’t forget, it is but a projection of us. The question should be how to redesign government altogether so as to keep it, while doing away with paternalism as a whole and I am not convinced this is all too possible.