12 1.2.2 Monetary Incentives and b
1.2. Managerial Incentives: An Overview 21 explicit and implicit incentives, history has taught us that even the existing control mechanisms do not suffice to prevent misbehavior. 1.2.2 Monetary Incentives Let us first return to the managerial compensation problem and exposit it in more detail than was done in the introduction to the chapter. The compensation package.28 A typical top executive receives compensation in three ways: salary, bonus, and stock-based incentives (stock, stock options). The salary is a fixed amount (although revised over time partly on the basis of past performance). The risky bonus and stock-based compensations are the two incentive components of the package.29 They are meant to induce managers to internalize the owners’ interests. Stock-based incentives, the bulk of the incentive component, have long been used to incentivize U.S. managers. The compensation of executives in Germany or in Japan has traditionally been less tied to stock prices (which does not mean that the latter are irrelevant for the provision of managerial incentives, as we later observe). Everywhere, though, there has been a dramatic increase in equity-based pay, especially stock options. 28. See, for example, Smith and Watts (1982) and Baker et al. (1988) for more detailed discussions of compensation packages. 29. More precisely, earnings-related compensation includes bonus and performance plans. Bonus plans yield short-term rewards tied to the firm’s yearly performance. Rewards associated with performance plans (which are less frequent and less substantial than bonus plans) are contingent on earnings targets over three to five years. Many managerial contracts specify that part or all of the bonus payments can be transformed into stock options (or sometimes into phantom shares), either at the executive’s discretion or by the compensation committee. (Phantom shares are units of value that correspond to an equivalent number of shares of stock. Phantom stock plans credit the executive with shares and pay her the cash value of these shares at the end of a prespecified time period.) This operation amounts to transforming a safe income (the earned bonus) into a risky one tied to future performance. Stock-related compensation includes stock options or stock appreciation rights, and restricted or phantom stock plans. Stock options and stock appreciation rights are more popular than restricted or phantom stock plans, which put restrictions on sale: in 1980, only 14 of the largest 100 U.S. corporations had a restricted stock plan as opposed to 83 for option plans. Few had phantom stock plans, and in about half the cases these plans were part of a bonus plan, and were therefore conditional on the executive’s voluntarily deferring his bonus. Stock appreciation rights are similar to stock options and are meant to reduce the transaction costs associated with exercising options and selling shares. For example, in the United States, the sensitivity of top executives pay to shareholder returns has increased tenfold between the early 1980s and late 1990s (see, for example, Hall and Liebman 1998; Hall 2000). Needless to say, these compensation packages create an incentive to pursue profit-maximization only if the managers are not able to undo their incentives by selling the corresponding stakes to a third party. Indeed, third parties would in general love to offer, at a premium, insurance to the managers at the expense of the owners, who can no longer count on the incentives provided by the compensation package they designed. As a matter of fact, compensation package agreements make it difficult for managers to undo their position in the firm through open or secret trading. Open sales are limited for example by minimum-holding requirements while secret trading is considered insider trading.30 There are, however, some loopholes that allow managers to undo some of their exposure to the firm’s profitability through less strictly regulated financial instruments, such as equity swaps and collars.31 30. Securities and Exchange Commission (SEC) rules in the United States constrain insider trading and short selling. 31. An interesting article by Bettis et al. (1999) documents the extent of these side deals. Equity swaps and collars (among other similar instruments) are private contracts between a corporate insider (officer or director) and a counterparty (usually a bank). In an equity swap, the insider exchanges the future returns on her stock for the cash attached to another financial instrument, such as the stock market index. A collar involves the simultaneous purchase of a put option and sale of a call option on the firm’s shares. The put provides the insider with insurance against firm’s stock price decreases, and the call option reduces the insider’s revenue from a price increase. In the United States, the SEC, in two rulings in 1994 and 1996, mandated reporting of swaps and collars. Bettis et al. argue that the reporting requirements have remained ambiguous and that they have not much constrained their use by insiders (despite the general rules on insider trading that prohibit insiders from shorting their firm’s stock or from trading without disclosing their private information). Swaps and collars raise two issues. First, they may enable insiders to benefit from private information. Indeed, Bettis et al. show that insiders strategically time the purchase of these instruments. Swap and collar transactions occur after firms substantially outperform their benchmarks (by a margin of 40% in 250 trading days), and are followed by no abnormal returns in the 120 trading days after the transaction. Second, they provide insurance to the insiders and undo some of their exposure to the firm’s profitability and thereby undo some of their incentives that stocks and stock options were supported to create. Bettis et al. estimate that 30% of shares held by top executives and board members in their sample are covered by equity swaps and collars
must go band in band wit es that
22 1. Corporate Governance While there is a widespread consensus in favor of some linkage between pay and performance, it is also widely recognized that performance measurement is quite imperfect. Bonus plans are based on accounting data, which creates the incentive to manipulate such data, making performance measurement systematically biased. As we discuss in Chapter 7, profits can be shifted backward and forward in time with relative ease. Equity-based compensation is less affected by this problem provided that the manager cannot sell rapidly, since stock prices in principle reflect the present discounted value of future profits. But stock prices are subject to exogenous factors creating volatility. Nevertheless, compensation committees must use existing performance measures, however imperfect, when designing compensation packages for the firm’s executives. Bonuses and shareholdings: substitutes or complements? As we saw, it is customary to distinguish between two types of monetary compensation: bonuses are defined by current profit, that is, accounting data, while stocks and stock options are based on the value of shares, that is, on market data. The articulation between these two types of rewards matters. One could easily believe that, because they are both incentive schemes, bonuses and stock options are substitutes. An increase in a manager’s bonus could then be compensated by a reduction in managerial shareholdings. This, however, misses the point that bonuses and stock options serve two different and complementary purposes.32 A bonus-based compensation package creates a strong incentive for a manager to privilege the short term over the long term. A manager trades off shortand long-term profits when confronting subcontracting, marketing, maintenance, and investment decisions. An increase in her bonus increases her preference for current profit and can create an imbalance in incentives. This imbalance would be aggravated by a reduction in stock-based incentives, which are meant to encourage management to take a long-term perspective. Bonuses and stock options therefore tend to be complements. An increase in short-term incentives must go hand in hand with 32. This discussion is drawn from Holmström and Tirole (1993). an increase in long-term incentives, in order to keep a proper balance between short- and long-term objectives. The compensation base. It is well-known that managerial compensation should not be based on factors that are outside the control of the manager.33 One implication of this idea is that managerial compensation should be immunized against shocks such as fluctuations in exchange rate, interest rate, or price of raw materials that the manager has no control over. This can be achieved, for example, by indexing managerial compensation to the relevant variables; in practice, though, this is often achieved more indirectly and only partially through corporate risk management, a practice that tends to insulate the firm from some types of aggregate risks through insurance-like contracts such as exchange rate or interest rate swaps (see Chapter 5 for some other benefits of risk management). Another implication of the point that managerial compensation should be unaffected by the realization of exogenous shocks is relative performance evaluation (also called “yardstick competition”). The idea is that one can use the performance of firms facing similar shocks, e.g., firms in the same industry facing the same cost and demand shocks, in order to obtain information about the uncontrollable shocks faced by the managers. For example, the compensation of the CEO of General Motors can be made dependent on the performance of Ford and Chrysler, with a better performance of the competitors being associated with a lower compensation for the executive. Managers are then rewarded as a function of their relative performance in their peer group rather than on the basis of their absolute performance (see Holmström 1982a).34 There is some controversy about the extent of implicit 33. The formal version of this point is Holmström’s (1979) sufficient statistic result according to which optimal compensation packages are contingent on a sufficient statistic about the manager’s unobserved actions. See Section 3.2.5 for more details. 34. A cost of relative-performance-evaluation schemes is that they can generate distorted incentives, such as the tendency to herding; for example, herding has been observed for bank managers (perhaps more due to implicit rather than explicit incentives), as it is sometimes better to be wrong with the rest of the pack than to be right alone. As Keynes (1936, Chapter 12) said, “Worldly wisdom teaches that it is better for reputation to fail conventionally than to succeed unconventionally
rike pric The e th
1.2. Managerial Incentives: An Overview 23 relative performance evaluation (see, for example, Baker et al. 1988; Gibbons and Murphy 1990), but it is fairly clear that relative performance evaluation is not widely used in explicit incentive schemes (in particular, managerial stock ownership). Bertrand and Mullainathan (2001) provide evidence that there is often too little filtering in CEO compensation packages, and that CEOs are consequently rewarded for “luck.” For example, in the oil industry, pay changes and changes in the price of crude oil correlate quite well, even though the world oil price is largely beyond the control of any given firm; interestingly, CEOs are not always punished for bad luck, that is, there is an asymmetry in the exposure to shocks beyond the CEO’s control. Bertrand and Mullainathan also demonstrate a similar pattern for the sensitivity of CEO compensation to industryspecific exchange rates for firms in the traded goods sector and to mean industry performance. They conclude that, roughly, “CEO pay is as sensitive to a lucky dollar as to a general dollar,” suggesting that compensation contracts are poorly designed. As Bertrand and Mullainathan note, it might be that, even though oil prices, exchange rates, and industry conditions are beyond the control of managers, investors would like them to forecast these properly so as to better tailor production and investment to their anticipated evolution, in which case it might be efficient to create an exposure of CEO compensation to “luck.” Bertrand and Mullainathan, however, show that better-governed firms pay their CEOs less for luck; for example, an additional large shareholder on the board reduces CEO pay for luck by between 23 and 33%. This evidence suggests that the boards in general and the compensation committees in particular often comprise too many friends of the CEOs (see also Bertrand and Mullainathan 2000), who then de facto get to set their executive pay. We now turn to why they often gain when exposed to “luck”: their compensation package tends to be convex, with large exposure in the upper tail and little in the lower tail. Straight shares or stock options? Another aspect of the design of incentive compensation is the (non)linearity of the reward as a function of performance. Managers may be offered stock options, i.e., the right to purchase at specified dates stocks at some “exercise price” or “strike price.”35 These are call options. The options are valueless if the realized market price ends up being below the exercise price, and are worth the difference between the market price and the exercise price otherwise. In contrast, managerial holdings of straight shares let the manager internalize shareholder value over the whole range of market prices, and not only in the upper range above the exercise price. Should managers be rewarded through straight shares or through stock options?36 Given that managers rarely have a personal wealth to start with and are protected by limited liability or, due to risk aversion,37 insist on a base income, stock options seem a more appropriate instrument. Straight shares provide management with a rent even when their performance is poor, while stock options do not. In Figure 1.1(a), the managerial reward when the exercise or strike price is PS and the stock price is P at the exercise date is max(0, P − PS) for the option; it would be P for a straight share. Put another way, for a given expected cost of the managerial incentive package for the owners, the latter can provide managers with stronger incentives by using stock options. This feature explains the popularity of stock options. Stock options, on the other hand, have some drawbacks. Suppose that a manager is given stock options to be (possibly) exercised after two years on the job; and that this manager learns after one year that the firm faces an adverse shock (on which the exercise price of the options is not indexed), so that “under normal management” it becomes unlikely that the market price will exceed the strike price at the exercise date. The manager’s option is then “under water” or “out of the money” and has little value unless the firm performs remarkably well during the remaining year. This may encourage management to take substantial risks in order to increase the 35. In the United States, stock option plans, when granted, are most often at-the-money options. 36. As elsewhere in this book, we ignore tax considerations. Needless to say, these may play a role. For example, in the United States (and at the time of writing, accounting rules are likely to change in the near future), stock options grants, unlike stock grants, create no accounting expense for the firm. 37. There is a large literature on hedging by risk-averse agents (see, for example, Anderson and Danthine 1980, 1981)
in the money (say
24 1. Corporate Governance Share Stock option Average compensation under a stock option Market price P tomorrow Safe strategy Average compensation with a straight share (a) (b) PL PH PS Option 1 (in the money) Option 2 (out of the money) Risky strategy P P 2 P S 1 S Incentive compensation Incentive compensation Figure 1.1 Straight shares and stock options. (a) Expected rents (PL: low price (option “out of the money”); PS: strike price; PH: high price (option “in the money”)). (b) Risk preferences under a stock option. value of her stock options. (In Chapter 7, we observe that such “gambling for resurrection” is also likely to occur under implicit/career-concern incentives, namely, when a poorly performing manager is afraid of losing her job.) This situation is represented in Figure 1.1(b) by stock option 2 with high strike price PS 2 . That figure depicts two possible distributions (densities) for the realized price P depending on whether a safe or a risky strategy is selected. The value of this out-of-the money option is then much higher under a risky strategy than under a safe one.38 The manager’s benefit from gambling 38. Whether the manager is better off under the risky strategy depends on her risk aversion. However, if (a) the manager is risk neutral or mildly risk averse and (b) the risky strategy is a mean-preserving spread or more generally increases risk without reducing the mean too much relative to the safe strategy, then the manager will prefer the risky strategy. is much lower when the option is in the money (say, at strike price PS 1 in the figure).39 Another issue with “underwater options” relates to their credibility. Once the options are out of the money, they either induce top management to leave or create low or perverse incentives, as we just saw. They may be repriced (the exercise price is adjusted downward) or new options may be granted.40 To some extent, such ex post adjustments undermine ex ante incentives by refraining from punishing management for poor performance.41 In contrast, when the option is largely “in the money,” that is, when it looks quite likely that the market price will exceed the exercise price, a stock option has a similar incentive impact as a straight share but provides management with a lower rent, namely, the difference between market and exercise price rather than the full market price. The question of the efficient mix of options and stocks is still unsettled. Unsurprisingly, while stock options remain very popular, some companies, such as DaimlerChrysler, Deutsche Telekom, and Microsoft, have abandoned them, usually to replace them by stocks (as in the case of Microsoft). The executive compensation controversy. There has been a trend in executive compensation towards higher compensation as well as stronger performance linkages. This trend has resulted in a public outcry. Yet some have argued that the performance linkage is insufficient. In a paper whose inferences created controversy, Jensen and Murphy (1990) found a low sensitivity of CEO compensation to firm performance (see also Murphy 1985, 1999). Looking at a sample of the CEOs of the 250 39. In the figure, option 1 is almost a straight stock in that it is very unlikely that the option turns out to be valueless. 40. Consider, for example, Ascend Communications (New York Times, July 15, 1998, D1). In 1998, its stock price fell from $80 to $23 within four months. The managerial stock options had strike prices ranging up to $114 per share. The strike price was reduced twice during that period for different kinds of options (to $35 a share and to $24.50, respectively). 41. At least, if the initial options were structured properly. If repricing only reflects general market trends (after all, more than half of the stock options were out of the money in 2002), repricing may be less objectionable (although the initial package is still objectionable, to the extent that it would have rewarded management for luck). For theories of renegotiation of managerial compensation and its impact on moral hazard, see Fudenberg and Tirole (1990) and Hermalin and Katz (1991). See also Chapter 5
enal 1.2.3 Implicit Incentives mned about an ha
1.2. Managerial Incentives: An Overview 25 largest publicly traded American firms, they found that (a) the median public corporation CEO holds 0.25% of his/her firm’s equity and (b) a $1,000 increase in shareholder wealth corresponds on average to a $3.25 increase in total CEO compensation (stock and stock options, increase in this and next year’s salary, change in expected dismissal penalties). This sounds tiny. Suppose that your grocer kept 0.3 cents out of any extra $1 in net profit, and gave 99.7 cents to other people. One might imagine that the grocer would start eating the apples on the fruit stand. Jensen and Murphy argue that CEO incentives not to waste shareholder value are too small. Jensen and Murphy’s conclusion sparked some controversy, though. First, managerial risk aversion and the concomitant diminishing marginal utility of income implies that strong management incentives are costly to the firm’s owners. Indeed, Haubrich (1994) shows that the low pay–performance sensitivity pointed out by Jensen and Murphy is consistent with relatively low levels of managerial risk aversion, such as an index of relative risk aversion of about 5. Intuitively, changes in the value of large companies can have a very large impact on CEO performance-based compensation even for low sensitivity levels. Second, the CEO is only one of many employees in the firm. And so, despite the key executive responsibilities of the CEO, other parties have an important impact on firm performance. Put another way, overall performance results from the combined effort and talent of the CEO, other top executives, engineers, marketers, and blue-collar workers, not to mention the board of directors, suppliers, distributors, and other “external” parties. In the economic jargon, the joint performance creates a “moral hazard in teams,” in which many parties concur to a common final outcome. Ignoring risk aversion, the only way to properly incentivize all these parties is to promise each $1,000 any time the firm’s value increases by $1,000. This is unrealistic, if anything because the payoff must be shared with the financiers.42 Third, the work of Hall and 42. Suppose a “source” (i.e., an outside financier) brings (n−1) thousand dollars to the firm for any $1,000 increase in firm value, so that the n parties responsible for the firm’s overall performance receive $1,000 each. First, this financing source would be likely not to be able Liebman (1998) cited earlier, using a more recent dataset (1980 to 1994), points to a substantial increase in performance-based compensation, which made Jensen and Murphy’s estimates somewhat obsolete. They find that the mean (median) change in CEO wealth is $25 ($5.30) per $1,000 increase in firm value. 1.2.3 Implicit Incentives Managers are naturally concerned about keeping their job. Poor performance may induce the board to remove the CEO and the group of top executives. The board either voluntarily fires the manager, or, often, does so under the implicit or explicit pressure of shareholders observing a low stock price or a low profit. Poor performance may also generate a takeover or a proxy fight, or else may drive a fragile firm into bankruptcy and reorganization. Finally, there is evidence that the fraction of independent directors rises after poor performance, so that top management is on a tighter leash if it keeps its position (Hermalin and Weisbach 1988). As we will see, there is substantial normative appeal for these observations: efficient contracting indeed usually requires that poor performance makes it less likely that managers keep their position (Chapters 6, 7, and 11), more likely that they be starved of liquidity (Chapter 5), and more likely that they surrender control rights or that control rights be reshuffled among investors towards ones who are less congruent with management, i.e., debtholders (Chapter 10). There is a fair amount of evidence that executive turnover in the United States is correlated with poor performance, using either stock or accounting data (see Kojima (1997, p. 63) and Subramanian et al. (2002) for a list of relevant articles). The sensitivity of CEO removal to performance is higher for firms with more outside directors (Weisbach 1988) and smaller in firms run by founders (Morck et al. 1989). Thus, a tight external monitoring and a less complacent board are conducive to managerial turnover after a poor performance. to break even, since the n insiders would be unable to pay out money in the case of poor performance. Second, the n insiders could collude against the source (e.g., borrow one dollar to receive n dollars from the source)