Some notes on executive compensation

Over at PolicyMic, my friend Dana Teppert has an excellent post about executive compensation. She notes that executive compensation appears to be ‘spiraling out of control’ and this is a problem. And I think most of us would intuitively agree, both that executive compensation has become untethered and that this is a problem. CEOs are enjoying consistent wage-hikes that do not appear to be driven by any improvement in their own productivity or their companies’ performance; U.S. CEOs are consistently paid much more than their European counterparts and there’s no obvious reason why this would be the case if the ‘market’ for CEO-level talent were truly competitive. Morally, we are horrified by a distribution in which our society devotes 1,000 times as much resources to compensating the efforts of a $25-million-a-year CEO as to providing for the family of a $12.50-an-hour laborer who works full time. Many of us are also concerned that growing income inequality could tear at the social fabric.

How should we think about executive compensation and what, if anything, should we do about it? As is my wont, I want to step away from the outrageous facts and approach this question with basic theory. If we think we need ‘solutions’ to ever-higher executive compensation we need to understand the underlying dynamics of the system that is generating it.


In the perfectly competitive labor market of Econ 101, everyone gets paid approximately her/his ‘marginal product.’ The reason is simple: If your efforts can add $800,000 to a company’s revenues, then some company should be more than happy to pay $600,000 a year to retain you, thereby adding $200,000 to its bottom line; then, in turn, one of its competitors should instead offer you $700,000 a year, thereby adding $100,000 to its bottom line; then another competitor should then be willing to offer you $750,000, because that still adds $50,000 to its bottom line, etc. No company should be willing to offer you $800,500 a year, or even $800,001, because this would hurt its bottom line. So in the ideal market of Econ 101, compensation should never ‘spiral out of control’—any firm that overpaid for employees, including CEOs, would hurt its profits and get beaten by its competitors until it ‘exited’ the market.

(I should note right now, up front, that there’s no necessary equivalence between the positive economic concept of ‘marginal product’ and the moral concept of desert. For example, when people suffer terrible accidents, their economic ‘marginal products’ can be reduced to zero, but we all agree we should provide for them because, well, they’re human. If I were unlucky enough to be born a slow learner with low motivation, my ‘marginal product’ might be $15,000 a year, but it still might be the right thing for society to redistribute some income to me to boost my happiness and security anyway. Similarly, if a CEO can add $20 million to her company’s revenues, that doesn’t mean it is ethically right for her to take all $20 million home untaxed.  If artificial intelligence advances so much that robots can replace almost all human labor, then almost all of us will have a near-zero marginal product; but I would hope we will find ways to redistribute income such that everyone will nonetheless be better off than today. A person’s ‘marginal product’ is influenced by luck and broad, unpredictable social and technological changes over which we have no control, so it does not amount to moral desert. This is worth noting up front, because it generates a lot of confusion when economists accidentally write as if asking ‘what is this person’s marginal product?’ amounts to asking ‘what does this person deserve?’)

But despite all that, ‘marginal product’ is useful for thinking about labor markets and CEO compensation. If we knew that a CEO’s ‘marginal product’ was $20 million a year, that wouldn’t prove that it was ethically ideal for him to take home all of that money untaxed, but it would mean that paying him some $19 million was at least instrumentally rational toward the goal of increasing the company’s profits. It would mean that shareholders were increasing their earnings by hiring him and he was not actively destroying value for anyone. And it would mean that his pay was tethered to some actual metric of the value he was producing and so wouldn’t arbitrarily spiral out of control.

So are CEOs getting paid their marginal products? I have no idea; I don’t think anybody knows. But let’s assume that CEOs are getting paid much more than their marginal products. Why would this be the case? How could reality differ from the ideal Econ 101 market described above? I see three main ways:

(1) Imperfect information and risk-aversion: I have no idea what a CEO’s marginal product is and neither do the boards of directors and compensation committees who sign off on their pay. I also have no idea what skills are actually needed to, say, develop strategies for and run a snack-brands company or how we would even begin to develop metrics for any of these things. In the academic literature, they say that senior executive talent is “unobservable.” Given all this uncertainty, what do boards of directors focus on when ‘hiring’ CEOs? My sense is that they, like most of us, first and foremost want to avoid royally screwing up. They don’t really shop around aggressively for the ‘best deal’ on a prospective CEO. Instead, they’ll want to get somebody who was already CEO of another, similar company and who didn’t make too much of a mess of it. And since lots of boards want this relatively small set of people (those who have already done okay jobs as CEOs of other companies), they can demand very high pay. Since boards are nervous about rattling investors with public personality clashes, they won’t protest too much when these CEOs ask for higher and higher pay. Hiring expensive CEOs can thus be seen as a sort of conservative CYA strategy, rather than maximizing net-present value. Hiring an unproven wunderkind 29-year-old CEO on the cheap might be the highest NPV decision; but if the decision happens to go wrong, the board will be publicly embarrassed. If you instead hire someone with CEO experience and pay her a salary comparable to her peers, then who could blame you? Note also that there’s a self-reinforcing feedback loop here: Insofar as boards mostly only hire people who have already been CEOs or senior executives, fewer people gain CEO experience, strengthening the market position of CEOs.

(2) Social incentives of board members: Directors’ incentives are imperfectly aligned with the interests of shareholders. That’s because board members have social relationships with managers and so there are personal ‘costs’ associated with pushing back at managers, even when this would be in the best interest of shareholders. For example, board members like it when company managers offer them cushy consulting contracts; these contracts also conveniently make them less inclined to alienate management. Board members often run their own companies, with their own boards of directors, which include friends and family of the CEOs whose boards they sit on. If you’re a board member, clawing back senior management’s pay would only increase you dividends a little bit, realistically, but it could seriously damage connections and networks that will be essential to the rest of your career. So, I scratch your back and you scratch my back, obviously.

(3) The classic problem of distributed costs (to shareholders) vs. concentrated benefits (to managers): Finally, each individual shareholder has both limited abilities and limited incentives to push back against the back-scratching collusion between directors and managers, because executive pay isn’t an overwhelming component of shares’ earnings. The highest paid CEO of a public company in 2012 was Larry Ellison, who took home $96.2 million. Oracle’s net income in the same year was approximately $10 billion. So even if shareholders were to cut Ellison’s pay in half, they would only see a .5% increase in their earnings for the year. That isn’t much money and staging a shareholder rebellion is costly, in terms of time and legal fees, so no shareholder has an incentive to be the ‘first mover’ to roll back Ellison’s pay. Meanwhile, since Ellison himself takes home all of his compensation, he has highly concentrated interest in developing good rhetorical justifications for and legal defenses of his pay.


Those three factors above help explain the spiraling costs of CEO pay. In sum: Because it’s really hard to know what qualities/skills a CEO needs and what a CEO is really worth (the value of long-term strategic decisions does not become apparent until much later in time), boards of directors play it safe by hiring experienced and costly CEOs and paying them at the high range of their ‘peer group’; CEOs keep asking for more pay and, because there are social ‘costs’ to the board of alienating management and the costs of over-compensation are spread out among highly-distributed shareholders, the directors and owners of the company don’t have a strong incentive to say no, and so the ‘peer group’ compensation metrics rise in a positive feedback loop.

If this diagnosis is correct, it suggests some preliminary prescriptions, beyond just wage-caps and higher taxes (which managers are adept at avoiding, with loopholes, getting alternative forms of compensation, etc.). First, we need better research on what CEOs are actually worth, how they add value to a company, and how we can evaluate and account for the impact of their decisions, skills, and expertise on firm value. Business academics have lots of great work to do here!

Second, the problem of back-scratching between managers and directors and the problem of distributed costs to shareholders vs. concentrated benefits to management could only be solved with some truly creative corporate governance innovation. Here’s a very preliminary sketch of an idea I had: Firms could create a special class of ‘C-shares,’ which would total, say, 10% of all company shares and would be identical to regular common stock except that all executive-compensation costs would have to be directly expensed from the earnings to these shares. The C-shareholders would then be entitled to elect a compensation committee, whose members’ identities would not be disclosed to management. Thus, C-shareholders would constitute a much more concentrated constituency for reasonable executive pay and compensation committees would not need to worry about the ‘social costs’ of pushing back at big-eyed managers. And since executive pay is not in fact an overwhelming component of total company expenses, the C-shareholders’ incentives would still be fairly well aligned with those of other shareholders—the C-shareholders, for example, would not want to vote compensation so low that management would quit in protest, or lack incentives to work hard, or have trouble attracting talent, or do anything else that would detract strongly from company profits, because profits would still be the major component of C-shares’ values. Such C-shares would even create arbitrage opportunities for activist investors, who could accumulate shares in companies with overpaid CEOs with the specific goal of voting down the next pay package.


I think the idea above is fun and worth thinking about and developing. And shareholders have every right, and are well-advised, to do what they can to make managers act in their interests. But I’ll close this post by noting that it’s not clear to me that executive compensation is actually a pressing moral public-policy concern. On an emotional level, of course, I resent CEOs who make thousands of times more money than I do. But morally and rationally, my goal shouldn’t be to restrain the rich per se; it should be to help ordinary people, particularly the poorest. And it’s not clear that CEO pay can realistically be called a major cause of poverty and want in the U.S., nor would cutting back CEO pay be an actual solution to the problems of the poor. By my calculations, based off of the data here, the top-100 highest-paid CEOs of public companies in 2012 took home a combined $2.2 billion. Let’s assume that the universe of domestic exorbitant CEO pay that we would like to roll back is equal to 100 times this sum, or $220 billion. The U.S. population is 315 million. So even if we were to cut out all of this exorbitant pay and give it directly to the ‘bottom half’ of people in the U.S. without destroying any value (which is all, of course, unrealistic), this would amount to about $1400 per person ($220,000,000,000/157,000,000). That’s not nothing, but it’s not very much either. We could generate the same kind of relief for ordinary people with a lot of other smart policies that would actually have a chance of being implemented: Like, e.g., destroying licensing cartels that increase the costs of taxi rides home and all kinds of other ordinary goods; reforming health-care to make everyone more cost-sensitive and to reduce doctors’ liability expenses; imposing higher equity-capital requirements on banks that could prevent future financial crises and recessions; allowing denser development at urban cores to decrease the cost of housing, etc.

Are the social costs of high CEO pay enough to justify my baroque corporate-governance idea, which would be a pain to implement, particularly given that there is so much ‘low-hanging fruit’ in the policy domains I’ve just mentioned? I’m not so sure.