Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Учебный год 22-23 / Firms, Contracts, and Financial Structure

.pdf
Скачиваний:
2
Добавлен:
14.12.2022
Размер:
2.37 Mб
Скачать

200

II. UNDERSTANDING FINANCIAL STRUCTURE

making an offer at (just above) sByR for the B shares. The incumbent cannot resist this offer because he would incur a large capital loss by doing so. On the other hand, if yR<yI, the incumbent retains control, because the rival cannot afford to bid sByI for class B and this would be required to deter the incumbent.

Hence, in case 3 the superior management team always wins control regardless of the security-voting structure, because neither party's private benefit is large enough to offset any disadvantage in public benefit.229 Although this conclusion is fairly obvious, it confirms the idea that private benefits are a crucial ingredient in determining optimal security-voting structure.

The above analysis can be summarized as follows. One share–one vote (more generally, vB>1/2 sB=1) dominates all other security-voting structures if either the incumbent's or the rival's private benefits are insignificant. If both private benefits are insignificant, one share–one vote is neither better nor worse than other structures.

However, there is a fourth case, where one share–one vote is not generally optimal.

Case 4: bI, bRare both significant in relation toyI, yR. If both bI and bR are significant, a departure from one share–one vote may be desirable. Such a departure can increase the intensity of competition between the rival and incumbent and permit the extraction of some of the rival's private benefits.

The example of Section 1 can be used to illustrate this. Suppose yI=200, yR=300, bI=51, bR=3. Consider the dual-class structure in the example (sA=1/2, vA=0, sB=1/2, vB=1). Under this structure, the competition for control takes place over the class B shares. The rival is prepared to pay up to 153 for them (1/2(300)+3), whereas the incumbent is prepared to pay only 151 (1/2(200)+51). Thus, the rival will win with an offer of 152, say. The class A shareholders are left holding shares worth 150 (half the public value under the rival). So the total value of the company is 302.

Consider next one share–one vote. Now the rival will win control by making an offer for all the shares at 301, say. Shareholders

229 There is one situation in which this conclusion does not hold and it occurs when class B consists of pure votes (s =0). This structure is inferior to all others in case 3,

however, and so will be disregarded.

B

 

8. VOTING RIGHTS IN A PUBLIC COMPANY

201

will tender because they are being offered more than the post-takeover value of the company under either the rival or the incumbent. (If the rival offers less than 300, shareholders who expect the bid to succeed will not tender, preferring to remain as minority shareholders.) Moreover, the incumbent cannot resist this offer since the most he can afford to pay for 100 per cent of the shares is 251, comprising the public value under his management plus his private benefit. Hence under one share–one vote the rival will win and the value of the company will be 301.

To put it very simply, shareholders benefit when the rival and incumbent compete over products for which they have a similar willingness to pay. In the example, a class consisting of 50 per cent of the dividends and 100 per cent of the votes qualifies better for this than shares and votes bundled together in the same proportion. Note, however, that the example is quite fragile. If the incumbent's private benefit is 54 rather than 51, the incumbent will be able to outbid the rival for the votes, and the value of the company will be only 200. (The rival will not bid at all.) In contrast, under one share–one vote, the rival would win and the company would be worth 301.

How important is case 4 relative to the other case? For some kinds of companies, private benefits may be large and case 4 may be significant. Examples might be newspapers (where the ability to influence public opinion yields significant (non-monetary) private benefits) and sports teams and entertainment companies (where being associated with a winning team or mixing with famous people yields high (non-monetary) private benefits). However, for many other companies private benefits are likely to be less important. One reason for this is that corporate law makes it difficult for a controlling party to realize significant (monetary) private benefits. The corporation's directors have a fiduciary duty to all shareholders, and overt diversion of wealth to the controlling party violates this duty.230 Of course, the courts cannot always be relied on to ensure that the minority is protected, and the level of protection may vary greatly from one country to another or from one legal jurisdiction to another. Hence, the initial owner who is designing the security-voting structure must take into account the possibility that bR, bI are large. However, for a

230 For discussions of fiduciary duty, see Clark (1986) and Macey (1992). For an economic analysis, see Barca and Felli (1992).

202

II. UNDERSTANDING FINANCIAL STRUCTURE

reasonably wide class of companies, it may be unlikely that both are large at the same time (and that both are in the form of money, i.e. that both can be used to purchase votes). For such companies, cases 1–3 are the relevant ones. (For a further discussion of this point, see Section 4.)231

Proposition 1, which follows directly from the analysis above, is stated without proof.

PROPOSITION 1. Suppose the case where the incumbent's and the rival's private benefits are both significant occurs with (vanishingly) small probability relative to the case where just one private benefit is significant. Then one share–one vote (more generally, a structure with vBsB=1) maximizes the date 0 market value of the company's securities.232

3.Extensions

This section discusses two extensions of the model.

Restricted Offers

Proposition 1 on the optimality of one share–one vote generalizes to the case of restricted offers. (With a restricted offer, a bidder offers to buy a fraction of the shares of a class for some price and prorates equally if more shares are tendered.) It is not difficult to see why this is. (The following is a sketch of the argument; for details, see Grossman and Hart (1988) and Harris and Raviv (1988, 1989).)

231As noted, the level of protection of minority shareholders is not the same in all countries. Hence case 4 may be empirically more relevant in some countries than in others. This fact may explain why departures from one share–one vote are more common in some countries than in others. See Zingales (1994).

232It is worth noting that one share–one vote also leads to a socially optimal outcome (i.e. an outcome that maximizes public plus private value, y +b ). Moreover, this is true even if both the incumbent's and rival's private benefits can be significant. The reason is that a management team's willingness to pay for a single class of voting equity is given by (y +b ), and therefore under one share–one vote the corporate control contest will be won by the team with the higher total (i.e. public plus private) value.

8. VOTING RIGHTS IN A PUBLIC COMPANY

203

Suppose there are two classes with dividend and vote entitlements given by sA, vA and sB, vB, respectively, where vB>½, and sA+sB=vA+vB=1. Continue to assume that only one bidder has a significant private benefit. Consider a rival with a significant private benefit but with a lower public value than the incumbent's: yR<yI. (A similar argument applies if the incumbent has a significant private benefit and yI<yR.)233 Then it is clear from the logic of Section 2 that the rival will obtain control—that is, 50 per cent of the votes—by minimizing the fraction of the company's dividend stream she purchases. The reason is that she makes a capital loss of yIyR on this dividend stream. There are two cases. If sB/ vB≤1≤sA/vA, then the class B shares are rich in votes relative to dividends and the cheapest way to obtain control is to make an offer for a fraction λ of the class B shares (at a price of just above sByI per 100 per cent), where λvB=½. This way, the rival's capital loss is(8.1)

with equality if and only if sA/vA=sB/vB=1. On the other hand, if sB/vB≥1≥sA/vA, then the class A shares are rich in votes relative to dividends and the cheapest way for R to get control is to make an offer for all the class A shares (at a price just above sAyI per 100 per cent) and a fraction μ of the class B shares (at a price just above sByI per 100 per cent) where vAvB=½. This way the rival's capital loss is(8.2)

with equality if and only if sA/vA=sB/vB=1.

However, the argument of Section 2 shows that the initial owner who designs the security-voting structure wants to maximize the rival's capital loss, L. Combining the inequalities in (8.1) and (8.2) shows that one share—one vote, that is, sA/vA=sB/vB=1, is the (unique) security-voting structure that does this.234 That is,

233 If yR yI (and bR bI ), security-voting structure is irrelevant, just as in the case of no restricted offers (see § 2). 234 Readers may wish to refer to n. 17 here.

204

II. UNDERSTANDING FINANCIAL STRUCTURE

with one share—one vote the rival is forced to buy up 50 per cent of the dividend stream to obtain 50 per cent of the votes, whereas under any other security-voting structure she will buy less.

Although one share—one vote continues to be optimal in the presence of restricted offers, one share—one vote no longer fully protects shareholder property rights. In particular, an inferior rival may be able to get control (or a superior rival may be prevented from obtaining control).

To see this, suppose yI=200, bI 0, yR=180, bR=15. Then the rival can get control by making an unconditional offer for (just over) 50 per cent of the shares at a price of just over 100. (The bidder prorates equally if more than 50 per cent of the shares are tendered.) In the absence of a counter-offer from the incumbent, it is a dominant strategy for a small shareholder to tender to the rival (since the rival is offering a premium and the shares will be worth only 90 if she wins), and so the rival wins. This is in spite of the fact that in the aggregate the shareholders lose by the take-over: they obtain a slight premium on 50 per cent of their shares, but the rest are worth only 90. The rival makes a profit, since her private benefit exceeds her capital loss of ½(200)−½(180)=10 on the shares she purchases.

Could the incumbent block this bid? The answer is no. Given that the incumbent does not have a significant private benefit, he cannot offer a premium for the shares, and so he may as well offer to buy them all at 200. However, given this offer and the rival's offer, it is not a rational expectations equilibrium for the rival to lose, since, if it is thought that the rival will lose, her offer will not be prorated and so shareholders will obtain a higher return by tendering to her. Thus, the only equilibrium has the rival winning; i.e. the incumbent fails to block.

Similar examples can be constructed showing that restricted offers may prevent a superior rival from getting control.235

235One way to make it more difficult for an inferior rival to get control through a restricted offer is for the company to adopt a super-majority voting rule; that is, a rival is required to obtain a fraction α of the votes to displace incumbent management, where α>½. In fact, if the only goal were to prevent inferior rivals from obtaining control, it would be optimal to set α=1. However, this would make it very easy for an incumbent with a private benefit to resist a superior rival. For a discussion of optimal plurality levels, see Grossman and Hart (1988) and Harris and Raviv (1988, 1989).

8. VOTING RIGHTS IN A PUBLIC COMPANY

205

More Complex Security-Voting Structures

So far the optimal security-voting structure has been analysed under the assumption that all claims to dividends are proportional, i.e. that each security is characterized by a share s of profit. More general structures are, however, possible. For example, security i could be entitled to some nonlinear share fi(y) of total profit y. Harris and Raviv (1989) show that, if the functions fi(y) are restricted to be non-decreasing, then (as long as private benefits are not simultaneously large) an optimal security-voting structure consists of (any amount of non-voting) riskless debt and a single class of shares with votes attached. Since one share—one vote, with no debt, is a special case of this, one may conclude that (monotonic) nonlinear securities do not add anything.

The intuition behind the riskless debt result is as follows. A claimant holding riskless debt is unaffected by a control transaction, and so the winner of the control contest imposes no externality on such a claimant. (It is supposed that the rival creates enough public value so that the debt remains riskless.) Therefore an efficient outcome is achieved by giving all the votes to claimants other than the riskless debt-holders. In fact, if riskless debt-holders are given votes, this can lead to an inefficient outcome, since a bidder may be able to obtain control by buying their votes instead of the votes of the variable claim-holders, who are actually affected by the outcome. (Harris and Raviv (1989) call this the ‘no cheap votes’ principle.)

The result that riskless debt-holders should not have votes accords with observation. However, it is not clear that the analysis can explain why risky debt-holders also typically do not have votes.

Vote Selling

It has been assumed throughout that it is illegal or infeasible for a bidder to unbundle a security-voting structure by making an offer for the votes or proxies of shareholders in which the bidder pays only for the votes and gets none of the dividend claims. In particular, I have not allowed a bidder to make an offer to shareholders of the following form: in exchange for each share tendered, a shareholder will receive a small amount of cash together

206

II. UNDERSTANDING FINANCIAL STRUCTURE

with a share of a new company. The new company, whose assets will consist entirely of the tendered shares and votes of the original company, will pass through all dividends of the original company, but the bidder will maintain voting control of the new company (that is, the bidder will be able to vote the original company's shares). If offers like this were allowed, it is not difficult to show that the outcome of a control contest would be as if the company had a dualclass structure in which one class had all the votes and none of the dividends, and the other class had all the dividends and none of the votes.

Thus, if votes can be unbundled from shares, the security-voting structure does not matter. Note that one share—one vote is neither better nor worse than any other structure under these conditions. In practice, it may be hard to unbundle votes from shares. For example, Easterbrook and Fischel (1983) argue that a public market in votes separated from shares is illegal in the USA.236 To the extent that this is true, the earlier results apply and one share—one vote dominates all other security-voting structures.

4.Conclusions

This chapter has argued that a company's security-voting structure can have an important influence on whether the company will be taken over. It has been shown that, if a competition between bidders with significant private benefits is unlikely, the optimal choice of security-voting structure is one share—one vote.

There is a link between the material in this chapter and that in earlier chapters. Chapter 3 gave some reasons why residual control rights (in the form of votes) and residual income rights (in the form of dividends) should be bundled together. In that chapter, however, it was supposed that the holder of residual control rights and the holder of residual income rights could bargain costlessly. The analysis of this chapter provides further justification for the linking of voting rights and income rights, but for the case where large numbers make bargaining impossible and

236However, there may be ‘private’ ways to separate votes from shares through the use of voting trusts, standstill agreements, stock pyramids, and the like. On this, see DeAngelo and DeAngelo (1985).

8. VOTING RIGHTS IN A PUBLIC COMPANY

207

where shareholders face free-rider and collective action problems.

It is interesting to relate the results of this chapter to the empirical evidence on departures from one share—one vote. DeAngelo and DeAngelo (1985: 39) identified 78 companies publicly traded on the American Stock Exchange and over the counter that had classes of securities with different voting rights (out of a universe of thousands of companies).237 They found that, in most of the companies that deviated from one share—one vote, incumbent managers had enough votes so that a change in control was impossible without their approval. That is, the deviation from one share—one vote did not correspond to a situation in which widely held securities had different effective voting rights; instead, it corresponded to a situation in which the incumbent had all the effective votes necessary to maintain control. Moreover, in many of the cases the incumbent represented a family.238

The evidence suggests that the forces responsible for a deviation from one share—one vote may not be those discussed in case 4 of the model of this chapter, where the incumbent's and rival's private benefits are both significant, so much as those analysed in the Aghion—Bolton model of Chapter 5. As noted earlier, this chapter has supposed that management's preferences are ‘unimportant’ relative to those of investors and do not receive any weight in the choice of capital structure. However, if this is not the case—and family-run companies may be a leading example where it is not—then it might be efficient to allocate control rights to managers in order to allow them to enjoy their private benefits, or to motivate them to undertake relationship-specific investments.239 Alternatively, the initial owner might ‘sell’ the private benefits to a large investor, along with voting control, so that the investor can consume the private benefits without risk of expropriation. In fact, if private benefits are reliably significant, it would perhaps be surprising if one of the above scenarios did not arise; that is, it would be surprising if the company remained

237Note that until recently it has been a requirement for listing on the New York Stock Exchange that a company have a one share—one vote security structure. Thus, the authors had to look to other exchanges to find deviations.

238Zingales (1994) finds an even more concentrated pattern of insider ownership in dual-class companies in Italy.

239For a similar idea, see Laffont and Tirole (1988).

208

II. UNDERSTANDING FINANCIAL STRUCTURE

widely held. But this means that case 4, where there are no large shareholders and yet where there is a high probability of a competition between bidders with significant private benefits at some future date, is not a leading one.

In future work, it would be desirable to incorporate the entrepreneurial incentives and private benefits of Chapter 5 into the model of this chapter. It would be interesting to see whether the optimal security-voting structure consists of a single class of shares that are widely held, together with a class of superior voting stock held by insiders. Such a result would appear to be consistent with the empirical evidence.

Finally, it is useful to relate the results of this chapter to the recent US policy debate on whether a company should be required to adopt one share—one vote as a condition for listing on various stock exchanges. The chapter has argued that an initial owner has an incentive to choose a value-maximizing security-voting structure since he bears the full consequences of his actions through the effect on the prices of the company's securities. It has been shown that one share—one vote is often, but not always, optimal. Given this, there seems little reason to put restrictions on the security-voting structures of a new company; such restrictions will simply raise the cost of capital.

However, deviations from one share—one vote often occur at a later stage in a company's life, when the company's shareholders are dispersed. At this point, management—unlike an initial owner—does not have the right incentive to choose a value-maximizing security-voting structure since the consequences are borne by existing shareholders rather than by management. There is thus a suspicion that ex post changes may be carried out by management to entrench itself and may be value-reducing.240

Of course, changes in security-voting structure are sometimes justified as new events occur that were not anticipated when the original corporate chapter was written. Thus, some mechanism for change should probably be in place. However, in order to reduce the possibility of abuse, it makes sense to put the burden of proof on those who want to make the change. In particular,

240In fact, management may even be able to get shareholders to approve a change that makes them worse off by persuading them to accept inferior voting stock in place of superior voting stock through a ‘coercive’ exchange offer (see Ruback 1988).

8. VOTING RIGHTS IN A PUBLIC COMPANY

209

there should be protection for shareholders who oppose the transaction. One possibility is to allow a (relatively small) number of shareholders to veto the transaction. Another possibility is to give dissenting shareholders appraisal rights; that is, if the change in security-voting structure goes through, a shareholder could ask to be bought out at the ‘appraised’ value of the assets, in the absence of the change.