Attention conservation notice: Over 2400 words on economic mechanism design, corporate governance and finance by somebody who has no practical or academic qualifications in any of these subjects, capped with a deliberately inflammatory comparison of high-level corporate executives to tribalist thugs.
Seeing that Rakesh Khurana's excellent Searching for a Corporate Savior has come out in paperback (here's chapter 1), coupled with reading Miller's Managerial Dilemmas (rightly pushed here by Henry Farrell) and Lowenstein's Origins of the Crash (reviewed here by Krugman) has me thinking about corporate governance, and how it might be improved. In today's installment, I'm going to assume that "share-holder value" is the end-all and be-all of the corporation, and propose a way in which it might be more effectively secured. In the next installment, I will attempt to combine this with an argument that something very like the collective ownership of the most important means of production will enhance the efficiency of the market system. But today I'll leave relations to the mode of production alone.
The basic problem with making corporate capitalism work is the separation of ownership and control of large corporations. The people who own the company, officially, are its shareholders. The people who actually control the corporation, on a day-to-day and year-to-year basis, are its managers, headed by its chief executive officer. The managers may or may not own stock, but even if they do, it's generally a very small proportion of the total. The stock of any individual corporation is highly dispersed, typically among people (and institutions) which own stock in many corporations. Thus the fortunes of any one company are a comparatively small matter for a given investor, but a very large one for a manager there, giving the latter much more incentive to understand and manipulate the situation. Moreover, even if investors wanted to understand the company they've invested in, the managers are, by the very nature of their jobs, better-placed to do so --- they'll have more information, more opportunities to do things without being detected, and more influence. One thus has a classic problem of institutional design: how do the structures of information and incentives within the corporation cause the managers to act in the interests of the shareholders?
The short answer is that they don't, as is shown by even a casual acquaintance with life inside the modern corporation, never mind books like the ones I mentioned at the start. The basic mechanism for making management work for the shareholders is that the shareholders' representatives, the board of directors, gets to pick the CEO, and can throw out ones they no longer find to their liking. If the shareholders were perfectly informed about the company, they would know exactly how much it should be worth, and fire any management team which let it get below that value. But, and this is the basic problem once again, the management is better informed about the company than are its nominal owners, and in fact are in a position to deceive the owners about that worth. Moreover, as Khurana (among others) documents in detail, this mechanism has been subverted by management, which gets to control the information and options presented to the directors, and indeed basically picks the membership of the board. (I believe this is what hackers would call a "social engineering" attack, though I speak under correction.) Still, this subversion is not total, so if the management team screws up in a truly spectacular fashion, and gets caught, it can expect to be thrown out.
This is not altogether satisfactory (at least not if you're a shareholder), and two fixes have been suggested. One is the idea that there is, or should be, a "market for corporate control". If, after all, you think a company is being mis-managed, and you buy enough of shares in it, you can install a new management team. It'll cost you a lot of money, but, by hypothesis, the company will now be making more money, and you'll be getting a large share of it, so you could come out ahead in net-present-value terms. This being so, if you can make a good case, you should be able to arrange for financing. Thus the theory of the hostile takeover. Moreover, the threat that this would happen, disrupting their power and perquisites, should serve to set a definite limit to the management's abuses --- roughly speaking, they should feel safe only if the net present cost of their incompetence and abuse of position is less than the transaction costs of putting together a hostile takeover to remove them. This doesn't seem to be terribly effective, though, if only because management has become very good at finding ways to raise those transaction costs. (This is nicely discussed by Miller, and somewhat less by Lowenstein; I can't recall if it comes up in Khurana but it probably does.)
The other idea is to compensate managers in a way that links their pay to how well shareholders do. The preferred way to accomplish this, to date, has been to pay managers with grants of stock options. Basically, an option is a legal instrument which gives its owner the right, within a specified time, to buy one share of the company's stock at a fixed price, regardless of what the actual market price of the stock is. (There are a million wrinkles, starting with whether the option can be exercised at any time up to the expiration date ["American"], or only at the expiration date ["European"], etc., etc. If you care, let me recommend Shiryaev's Essentials of Stochastic Finance.) If the stock price goes over the price fixed in the option, you can make money by exercising the option, buying the stock at the low price, selling at the high, and pocketing the difference. If the stock price is below the "strike" price, exercising the option is not profitable, but you lose nothing by waiting and hoping the price will go up, and anyway you might be able to sell the option to someone who fancies its chances in the future. In the case of executive compensation, it's usual for the strike price of the options issued to be very close to the actual share price at that time. The idea is that if, later, the share price goes up, managers make a lot of money, so they're going to be strongly motivated to make that share price go up.
There are a couple of problems with this scheme. The basic one is that the share price doesn't have to stay up. Say that, as the CEO of the Bactra Consulting Group, I get issued a stack of options with a strike price of \$10 when the share price is \$10, and you similarly buy a bunch of BCG shares. If the stock then goes to \$100, I make \$90 per option, and you have capital gains of \$90 per share. So far, so good. The catch is that I make just as much money, in cash, whether the share price is \$100 for a day or forever, whereas your profits are all on paper, and you have a definite preference for the "forever" outcome rather than the "day". In fact, if you own enough shares, trying to realize your capital gains will substantially reduce them, through a combination of transaction costs and market impact. Long-term investors, such as institutions, are actually more interested in dividends than in short-term capital gains, but paying executives in stock options makes them care about the latter rather than the former. In theory, of course, the price of share is supposed to be the best estimate, on the basis of all available information, of the net present value of future earnings, but nobody is so wedded to the efficient market hypothesis as to suppose that this is exactly true, at all times. (Really, nobody ought to be wedded to the efficient market hypothesis at all, but I won't go into that again.)
The other problems with paying executives in stock options is that it provides them with a one-sided gamble, and makes them overly fond of risk; these are related. If the stock price goes above the strike price, that's good for option holders, but if the stock price goes below it, they don't lose anything. In my example, if the share price falls from \$10 to \$9, I lose nothing, while you take a capital loss of \$1 a share. Now suppose I, as CEO, confront the choice between two possible courses of action. One of them is certain to make the company a fairly small amount of money; the other has a high chance of losing everything we spend on it, but a tiny chance of making us a mint. The investor, who is exposed to both the profits and the loses, is going to tend to want me to chose the first option. I, on the other hand, have no reason to care about the potential loses, so I'm going to tend to chose the second. In fact, I have even more reason to like risk, because if the share price is below the strike price of my options, I can still sell my options to other people. The value of the options is going to depend on the odds that the share price will ever rise above the strike price. Clearly, the more volatile the share price, the more likely this is to happen, so option price rises with stock volatility, which means, ceteris paribus, executives paid in options will do what they can to make the share price fluctuate. (For more on this point, see the huge literature beginning with Black and Scholes.) So, even leaving aside the shameless accounting gimmickry (bless you, Senator Lieberman!), compensation in options is a sub-optimal way of aligning the interests of managers and shareholders. Which is where we come in.
Recall that the basic issues are ones of information and incentives: the management is more knowledgeable than the owners about what the company is worth, i.e., what kind of earnings it could be generating, and not motivated to maximize those in any case. It seems to me that a very simple device could, at one stroke, eliminate both of these.
The solution, in a word, is that the position of CEO should be for sale.
Here is how I envision it working. A fixed number of restricted shares of the company will be held in trust for the office of the CEO. They will pay the holder of that office the same dividends as common stock, but carry no voting rights. The CEO will receive no compensation other than those dividends and (let's be nice) health care. The board of directors invites bids for the position of CEO, awarding the job strictly to the highest bidder. Bidding goes on for one month, or until no new bids have been received for two weeks. The winning bid must be paid to the company, in full and in cash, before the new CEO can begin work and drawing his pay. Bids are invited again after five years. (Obviously, the numbers are pulled from the air and merely illustrative.)
Informationally, the CEO now has a real incentive to reveal how much revenue they think the company is capable of producing over the next five years. (I don't think it's fair to ask anyone to guess beyond that.) Moreover, it's easy to see whether their company met or failed to meet their expectations, which helps gauge just how good a guesser they are. Since bidding is competitive, there really is a market for corporate control, and I'd imagine it'd be substantially easier to raise funds for such bids than for hostile takeovers. The CEO's incentives are simply those a large shareholder, only without even the walk-away option of selling the shares; they are firmly committed. (This eliminates the problem of cheap talk.)
Now, it's true that nobody will bid exactly what they think the company can make, because, if they did, they wouldn't make any money. But a major problem with the current corporate arrangements is, of course, that the executives are ridiculously over-paid. By this I don't mean that they contribute nothing to the company, but that their payment is thoroughly out of line with their compensation. I defy anyone to tell me, with a straight face, that a CEO will do twice as much for the company at \$10 million a year as at \$5 million; even that they will do twice as much at \$10 million as at \$1 million. After all, \$1 million is a lot of money, on top of the pleasure which people who boast of being driven control-freaks will undeniably take from working hard at bossing others around. But if applicant A puts in a bid where they will obviously make at least \$10 million a year, I have enough faith in the forces of greed and rivalry to think that somebody will put in a bid where they get to be in charge for only \$5 million, and pretty soon we're talking about reasonable compensation.
Two possible refinements would seem to strengthen the basic idea, but I'm less sure of them. The first is to allow outsiders (but not the incumbent) to submit a bid at any time, which would signal the re-opening of bidding. This would do even more to keep management on its toes, but there may be some wrinkles on it I'm missing. The other refinement is to replace "CEO" by "senior management team" in the above, so that it the people running the company form a cohesive group of collaborators, rather than viciously competitive rivals (or at least it's less likely). How the team divides up the dividend income is their affair (and good luck to them; but of that, more next time).
Shifting to this system from the current one carries absolutely no benefits whatsoever for CEOs. This is the point. As Khurana documents, CEOs and directors have managed to replicate the whole syndrome Popper describes in The Open Society and Its Enemies as "arrested tribalism". This combines a closed circle of privilege with a cult of leaders and leadership, a primitive faith in visions, in the magical powers and supreme worth of the right kind of person, all wrapped up in a crippling inability to see beyond the interests or experiences of the privileged circle. This system is at least odious to the values and institutions of an open society and a democratic culture, if not actively threatening to them --- back in the day, those ideas about leadership went under the label of Führerprinzip. As a partisan of the open society, I want to see the whole edifice crushed, and market forces are very good at blowing this kind of cozily rotten thing apart. Of course, as a good Popperian, I don't believe in holistic social engineering, but rather the piece-meal kind, the reform of existing institutions and the introduction of new ones. This is why I am merely proposing a reform of the market in corporate control, not anarcho-syndicalism.
Posted at August 19, 2004 23:05 | permanent link