Implicit Contracts and Dominant Shareholders

Transcription

1 Implicit Contracts and Dominant Shareholders José Guedes José Brito Universidade Católica Portuguesa - FCEE Palma de Cima, Lisboa, Portugal T: Fax: jcg@cee.ucp.pt; gbrito@cee.ucp.pt November 2004

2 Implicit Contracts and Dominant Shareholders (JEL CC: G30) Abstract This paper presents an eiciency argument that contributes to understand why corporate governance structures with a dominant shareholder are so prevalent in so many countries around the world. In an environment where outsiders cannot accurately monitor the perormance o transactions made between irms and stakeholders, the existence o a controlling shareholder who is an insider to the irm s management allows or eicient contracting with stakeholders. Firms controlled by outside shareholders cannot sustain relationships with stakeholders, because managers o such irms have an incentive to alsely claim that transactions with stakeholders have produced an outcome that is unavorable to the irm, and misappropriate the cash lows associated with the true outcome. To avoid being expropriated, the controlling party will ire the manager and it will reuse to make good on costly obligations toward the stakeholder, in response to the announcement o an unavorable outcome. However, because outsiders cannot observe the transaction s true outcome, punitive actions by outside shareholders will occur even when the manager truthully reports an unavorable outcome. Since these misguided disciplinary actions reduce the ex-ante value o transactions to stakeholders, stakeholders only accept doing business with irms controlled by outside investors i the requency o misguided disciplinary actions is not too high. Thus, where outside shareholders cannot target accurately disciplinary actions to opportunistic managers, insider control is required or eicient contracting with stakeholders. This eiciency beneit o insider control should be taken into account i one wants to explain the prevalence o irms eaturing dominant shareholders with a hands-on approach to their irms management. 2

3 I. INTRODUCTION According to La Porta et al. (1999) the dominant ownership and governance structure around the world is that o a controlling shareholder typically a amily who owns control rights well in excess o his cash-low rights and actively participates in management. For example, the authors report that all o the top 20 largest traded irms rom Argentina, Greece, Austria, Hong Kong, Portugal, Israel and Belgium have a shareholder who controls at least 20% o the votes. It is well known that the coexistence o a controlling shareholder and outside shareholders within the same company gives rise to an important agency conlict. The agency conlict arises because the party who runs the irm enjoys private beneits o control, which are unavailable to outside shareholders. Bebchuk, Kraakman and Triantis (2000) examine the agency costs arising rom such ownership structures which they call controlled-minority structures or CMS or short - and conclude that they are potentially large. Their analysis also indicates that the agency costs o CMS are likely to be larger in countries where outside investors cannot easily monitor corporate behavior or cannot easily obtain redress against opportunistic actions undertaken by corporate insiders through the legal and the judiciary system. As reerred by Bebchuk, Kraakman and Triantis (2000), the prevalence o CMS irms presents us with a conundrum. Why are CMS so common in spite o their large agency costs? And why are CMS so common particularly in those countries where their agency costs are likely to be bigger? A simple equilibrium argument suggests there should exist some countervailing eiciency beneits associated with CMS or otherwise they should be driven out o existence by irms adopting improved governance and 3

4 ownership structures. In addition, such eiciency beneits ought to be positively associated with the same actors that contribute to magniy the agency costs o CMS. In this paper we put orward a theory that highlights one eiciency beneit resulting rom assigning control to corporate insiders. 1 We argue that where irm value derives largely rom transactions made with stakeholders and the perormance o such transactions are hard to monitor by outsiders, insider control emerges endogenously as an eicient solution to a contracting problem. Furthermore, the eiciency o insider control goes up as the ability o outsiders to gauge transaction perormance declines. This is a desirable property i one wants to explain the CMS puzzle, since the agency costs surrounding insider control are also bigger where the transparency o irms business dealings is poor. I the perormance o transactions made between the irm and stakeholders is only imperectly observed by outsiders, ormal contracts are useless since they cannot be externally enorced by courts. That implies that a substantial portion o the value created by irms becomes dependent on implicit contracts made with stakeholders such as workers, suppliers, customers, business partners and government oicials. A contract is implicit i its enorcement depends on market mechanisms say the desire to preserve a reputation or an established relationship rather than the enorcement o the law. One eature o many transactions with stakeholders is that their payos are only determined ex-post, ater the terms o the transaction have been agreed upon. An investment made by a supplier in relationship speciic capital, an insurance policy granted 1 Khanna and Palepu (2000) suggest another beneit o CMS. They argue that CMS are eicient in emerging markets, i control is held within a large and well-diversiied business group. In these economies, business groups are able to internally replicate the unctions that are commonly provided by stand-alone 4

5 to a worker against uncertain productivity, or yet a service contract or a product warranty sold to a customer are examples o transactions in which the ultimate outcome is uncertain at the outset o the relationship. Oten the outcomes on such transactions are unavorable to the irm. When that happens irms have an incentive to break-up their implicit contracts, without regarding the adverse eect o such an action on stakeholders. A rational stakeholder who anticipates such an opportunistic behavior will be reluctant to enter into an implicit contract with the irm. Thus unless there is some enorcement mechanism that coerces irms to uphold their poorly perorming implicit contracts, many advantageous business transactions with stakeholders will ail to occur. We examine how governance structures inluence the eiciency o implicit contracting in an environment where irms interact repeatedly with a limited number o stakeholders. First, we show that irms controlled by a inside shareholder who closely monitors the management, will rerain rom breaking-up poorly perorming implicit contracts, as the controlling shareholder is aware o the potentially disruptive eect o such actions on valuable business relationships with stakeholders. Hence, i a substantial portion o irm value stems rom very proitable relationships with a limited number o business partners, a governance structure eaturing an inside controlling shareholder is eective in capturing such value or the irm s owners. There is however, a downside. When shares are also held by outside shareholders, the manager is prone to alsely report losses on implicit contracts and divert to the controlling shareholder unds or assets that belong to the irm. This incentive is exacerbated i the controlling shareholder controls the irm through the ownership o shares with concentrated voting rights or through irms in advanced economies. Firms unailiated with business groups ind it either impossible or costly to obtain access to such unctions. 5

6 pyramidal holdings and thereore, holds ew cash-low rights. 2 In sum, although irms controlled by inside shareholders are able to sustain valuable relationships with stakeholders, most o the beneit arising rom such relationships is appropriated by the controlling shareholder and thereore, shares held by outsiders are worth little. Secondly, we consider a irm that is vulnerable to a hostile takeover launched by outside investors. We show that in that case takeovers cannot discipline opportunistic managers without simultaneously disrupting relationships with stakeholders. To understand this result consider a irm, which has all its shares held by outside investors. Since the outcome o implicit contracts is not observable by outsiders, managers have an incentive to report outcomes characterized by losses against stakeholders and appropriate or themselves unds that belong to the irm. Thus outside investors will never believe managers claims on the perormance o implicit contracts, even when those claims are true 3. I allowed, outside investors will launch a takeover to prevent the manager rom withdrawing unds rom the irm to pay to stakeholders. Rational stakeholders anticipate the possibility o being expropriated and reuse to enter into implicit contracts. 4 2 A alse report o an unavorable deal outcome does not necessarily entail outright cash thet by the controlling shareholder. For example, the controlling shareholder can buy personal loyalty rom workers, suppliers, business partners and politicians by granting them rents while claiming to outside shareholders that such payments are due to them under past agreements. The loyalty o stakeholders is a valuable asset that the controlling shareholder can redeploy outside the irm when and where it best suits his interests. 3 The problem is similar to that presented by Stulz (1990) in his model o managerial discretion. In Stulz the manager always claim that cash low is low because he likes to invest and hence wants to raise as much capital as possible. Shareholders never believe manager s claim and reuse to provide unds. 4 The relationship between hostile takeovers and implicit contracts was examined by Shleier and Summers (1988). They argue that the wave o hostile takeovers that occurred in the US in the 1980s led to a breach o trust between irms and their stakeholders. According to these authors the premia paid by acquirers or shares o target companies can be accounted by the expropriation o rents enjoyed by stakeholders o target companies such as workers and suppliers. These rents however, consisted o a air payment that was due to them under implicit contracts they had previously made with the irm. Thus, takeovers caused a breach o trust and made stakeholders reluctant to enter into new implicit contracts. Shleier and Summers however, do not tackle one important issue. In the context o their analysis hostile takeovers cause two eects o opposite sign on the value o shares o target irms. On the one hand there is a positive eect resulting rom 6

7 The analysis suggests that in countries where irms business is done through private transactions, which cannot be externally monitored, insider control emerges endogenously to guarantee eicient contracting with stakeholders. Controlling parties pay or their private beneits ex-ante when they gain control o the irm. Consequently, large blocks o shares, shares with concentrated voting rights or yet shares o holding companies located at the top o pyramidal corporate structures will command high valuations relative to shares which are traded or investment purposes only. The remaining o the paper is organized as ollows. Section II presents the model and summarizes its major results. The implications o the model are discussed in section III. Section IV examines how a controlling shareholder inds external cash to und growth opportunities without oregoing control. Section V addresses the risk diversiication problem aced by controlling shareholders. Section VI concludes. II. THE MODEL II.1. Characterization o business transactions with stakeholders In this section we characterize the business environment in which irms operate, taking a special ocus on the description o transactions mediated through implicit contracts. Initially, we consider a manager-owned irm 5 and examine how such irm engages in transactions with stakeholders. The manager-owned irm exempliies an ownership structure in which corporate control is securely held by a party who is an insider to the irm s management decisions. Although more complex ownership regimes the expropriation o rents rom stakeholders; on the other hand there is a negative eect caused by the increased reluctance o stakeholders to enter into new and mutually advantageous - implicit contracts with the irm. The interaction between the two eects is an issue let unresolved by the authors. 7

8 eaturing secure control by insiders could be considered, a manager-owned irm illustrates in a simple ashion and without loss o generality - how insider control sustains eicient contracting with stakeholders. In the ollowing sections we study the impact o allocating corporate control to outsiders on the transaction eiciency o implicit contracts. In every period irms have a chance to make a business deal with a stakeholder (we will reer to the stakeholder as the deal s counterpart), which pays o at some later date. Although the payos are initially uncertain, the deal is mutually advantageous on an ex-ante basis, conditional on both parties upholding the deal. The payos depend on the realization o an exogenous state variable that takes either a high value (upstate) or a low value (downstate). The realization o the state variable occurs ater the deal has been agreed upon and is only observed by the contracting parties. Since the state variable is not observable by outsiders, the irm and the stakeholder cannot write a state contingent contract enorceable by the courts. Thereore neither party can be orced to stick to the deal i it turns out to yield an unavorable outcome. For simplicity and without loss o generality - we assume that the upstate and the downstate occur with equal probability and that it s never optimal or the counterpart to break up the deal. In contrast, it s optimal or the irm to renege the deal in the downstate. Such a breakup however, entails a loss to the counterpart. It is assumed that the loss is equal or smaller than the gain obtained by the irm so that an ex-post eicient renegotiation o the deal cannot prevent the irm rom breaking up in the downstate. The payos rom the deal are detailed in 5 By a manager-owned irm we reer to a irm in which all cash-low and control rights are held by the manager. 8

9 Table 1 and are assumed to be common knowledge to the irm s manager and to the counterpart. Upstate (50% probability) Counterpart (c) Uphold the deal Break up the deal Firm Uphold the deal (π u, π u c ) (0,0) () Break up the deal (0,0) (0,0) Downstate (50% probability) Counterpart (c) Uphold the deal Break up the deal Firm Uphold the deal (π d, π d c ) (0,ν) () Break up the deal (0,ν) (0,ν) Table 1 Payos rom a business transaction made between the irm and a stakeholder (counterpart) The preceding discussion implies the ollowing restrictions on the payos contained in Table 1: u d i. π + π > 0 (the deal has a strictly positive NPV or the irm, conditional on both parties upholding the deal); u d ii. π + π 0 c c (the deal has a non-negative NPV or the counterpart, conditional on both parties upholding the deal); u d iii. π 0, π > v (the counterpart never gains by breaking up the deal); c c d iv. π < 0 (the irm gains by breaking up the deal in the downstate); d d v. π > π (an eicient ex-post renegotiation o the deal cannot prevent the irm c v rom breaking up the deal in the downstate); u vi. π + v < 0 c (the NPV o the deal or the counterpart, conditional on the irm breaking up in the downstate, is negative). 9

10 One illustration o this set up is that o a irm which hires a contractor to build a actory. The cost o building the actory is initially unknown. The contractor pays or the construction costs rom his own pocket while the actory is being built and bills the irm or the total expenses (plus some additional markup) when the job is over. In this example the upstate corresponds to the scenario in which the construction costs turn out to be lower than expected whereas the downstate corresponds to a scenario o cost overrun. In the scenario o a cost overrun the irm may reuse to take ownership o the actory and reject the bill presented by the counterpart, causing a substantial inancial loss to the contractor. Another example is the case in which a irm is granted by a government oicial monopoly rights to service a market in exchange or a bribe (or in exchange or a promise o uture employment). In the upstate the irm makes proits rom servicing the market whereas in the downstate it makes losses. I the irm walks away rom the contract in the downstate it puts the government oicial in a tight political spot or ailing to guarantee service to the market (and perhaps subjecting the government oicial to a criminal investigation). The obvious issue that arises in such business transactions is how to induce irms to stick to deals. In other words, one needs to ind a mechanism to enorce implicit contracts made between the irm and stakeholders. We address this issue by assuming that a irm that breaks up a deal damages its ability to clinch uture deals with the same counterpart. Under this assumption, a manager-owned irm will comply with its current deal i the present value o uture business lost as a result o a break up exceeds the shortterm gain obtained by abandoning the deal in the downstate. This condition is likely to be 10

11 satisied i irms and counterparties interact repeatedly over time so that there is always a non-negligible probability o irms and counterparties encountering each other again in the uture or new business transactions. Suppose that irms have, in every period, a constant probability, p, o meeting a particular counterpart. Furthermore, irms that ail to engage in a transaction with a counterpart in any given period orego the possibility o doing business with another counterpart in that same period. Finally, assume that counterparties permanently cut o business relationships with irms that break up a deal with them. In this environment, a manager-owned irm will stick to a deal in the downstate i d u d p π < 0.5( π + π ), r where r is the periodic discount rate. We assume that this inequality holds so that manager-owned irms are able to contract eiciently with stakeholders. In the rest o the paper we ocus on a irm dealing with a single stakeholder, who is periodically available or a business transaction with the irm (so that p = 1). The ocus on such special case entails no loss o generality but reduces the complexity o the analysis. The manager-owned irm will then stick to a deal in the downstate i the ollowing inequality holds: u d 0.5( π + π ) d > π (1) r II.2. Firms vulnerable to hostile transers o control rom outside investors We now examine whether a irm, whose control is held by outside investors that do not know the perormance o transactions with stakeholders, is still able to contract eiciently with stakeholders via implicit contracts. Consider a irm that is entirely 11

12 inanced by traditional one share-one vote equity. The irm is run by a manager who owns a minority equity stake, α. All other shares are held by outside investors. The manager s task is to engage in business transactions with the irm s single stakeholder (counterpart) on behal o the irm. Speciically, the manager directly negotiates deals with the counterpart and observes privately jointly with the counterpart - the outcomes o such deals. Ater learning the outcome, the manager announces publicly to shareholders a deal outcome. Since the true outcome is private inormation shared only by the manager and the counterpart, the manager may alsely report the downstate and jointly appropriate with the counterpart the beneits o misrepresenting the outcome. However, i the manager reports the downstate, a raider may launch a hostile tender-oer aimed at orcing a break up o the deal and, in doing so, saving the irm the amount d π, the negative pay-o associated with upholding the contract in the downstate. A take-over takes place i the market value o the irm under control o the raider (net o a transaction cost, k ) exceeds the market value o the irm under the incumbent manager. The transaction cost, k, represents a reduced orm parameter capturing the various types o costs that the raider must bear to assemble a controlling block o shares. Because the model encompasses repeated business dealings between the irm and stakeholders, the value o the irm under control o the raider relects the immediate savings resulting rom a break up o the current deal plus the potentially disruptive eect 12

13 o the raid on uture deals available to the irm. 6 To help the raider decide whether or not to launch a takeover, a public signal o the true deal outcome is observed Z = ρi + u, [ ] 0 Eu=, 0 ρ 1 (2) where u is a random variable with cumulative density unction Pr ( u< U) =Φ ( U) and I is an indicator variable that takes the value 1 i the upstate has occurred and 0 otherwise. The public signal may represent ree inormation that becomes spontaneously available to the market or may relect the result o an audit to the irm s activities. It is assumed that the public signal is observed only ater the manager has announced an outcome so that the manager cannot condition her reporting policy on the value o the public signal. Ater gaining control, the raider breaks up the deal and ires the incumbent manager. He then hires a new manager and sells her an equity stake α. Finally, he sells all his remaining shares to outside investors and exits the irm. This assumption guarantees that the ownership and the control structure o the irm at the end o the period is exactly the same as the one prevailing at the beginning o the period, independently o whether a takeover has occurred in the intervening period. At the beginning o the ollowing period the irm is oered the opportunity to make a new business deal with the stakeholder and a resh cycle is started. The sequence o events is summarized in Figure 1. 6 We chose to adopt a very simple takeover model since our ocus is on the eects o control challenges launched by outside investors on eicient contracting with stakeholders. Our results hold or more complicated takeover models. 13

14 Date t Date t+1 the manager and the counterpart agree on a business deal the deal s outcome is privately observed by the manager and the counterpart the manager announces a deal outcome to outside shareholders a public signal correlated with the true outcome is observed i the public signal suggests that the downstate has been alsely announced, a raider launches a take-over to break up the deal the payos on the deal are due i a takeover has occurred, the raider hires a new manager and sells his shares to outside investors the manager and the counterpart agree on a new business deal Figure 1 Sequence o events II.2.I. The raider's problem In gauging the beneits and costs o taking over the irm ollowing an announcement o the downstate, the raider must consider the potential negative impact o breaking up the current deal on uture deals available to the irm. Initially, when the stakeholder engages in a business transaction with the manager, he is unsure whether the deal will be upheld ollowing an announcement o the downstate, since he knows that a successul takeover will cause a deal break up. Consequently, the counterpart will only agree to make a deal i he expects a low probability o takeover. The raider observes the public signal, Z, correlated with the true deal outcome. The problem aced by the raider is to set a takeover rule based on the observed value o Z. We assume that the raider ollows a rule with a very simple structure: ollowing an announcement o the downstate, he launches a takeover i and only i the value o the public signal is above a threshold, Z. The raider s problem then, is to ind such threshold. For example, he knows that the threshold cannot be set too low or otherwise the ex-ante value o deals to the stakeholder will be negative. 14

15 A sel-consistent rule has to satisy the ollowing inequalities: V raider V > 0, i manager Z > Z (3) V raider V 0, i manager Z Z (4) where V and V are, respectively, the market value o the irm under the raider raider manager and under the incumbent manager. I the counterpart continues making business deals ollowing the takeover then V V = π k (5) d raider manager In this case the successul takeover has impact only on the current cash-lows: the transaction costs incurred by the raider and the immediate savings resulting rom breaking up the current deal. I however, the takeover jeopardizes the irm s ability to make new deals, then u d 0.5( π + π d ) Vraider Vmanager = π k (6) r where the last term represents the present value o all uture business deals lost as a result o the takeover. We assume that d π > k, so that it's optimal or the raider to launch a takeover when the downstate is reported, as long as that does not disrupt the relationship u d 0.5( π + π ) d with the stakeholder. Since, rom (1), > π, then the raider will choose the r lowest possible value o Z that does not jeopardize the relationship with the stakeholder. II.2.II. The manager's problem The manager secretly splits up with the counterpart the unds that she diverts away rom the irm. I the manager misreports the downstate and there is no subsequent 15

16 take-over, the amount o unds available to be split up between the manager and the u d counterpart is equal to π π. On the other hand, i a takeover occurs ollowing a alse announcement o the downstate, the amount o unds available or diversion is equal to zero since the raider, once in control o the irm, orces the break up o the current deal. We assume that the manager keeps or hersel a ixed percentage λ o the unds available or diversion. 7 The decision aced by the manager is whether to lie or tell the truth about the deal's outcome ater privately observing the upstate. 8 Denote ϕ t as the probability o the manager reporting the upstate truthully at date t, W t as the manager's wealth at date t and W U t as the manager s wealth at date t conditional on observing the upstate. The value o W U t can be written as: [ ] u EWt+ 1 no takeover at t Wt U = ϕt απ r [ no takeover at t] no takeover at t u d d EWt+ 1 + ( 1 ϕt) Pr λ downstate is alsely reported ( π π ) + απ r [ takeover at t] takeover at t EWt+ 1 + ( 1 ϕt ) Pr downstate is alsely reported 1 + r (7) 7 The division o diversion unds between the manager and the stakeholder could be endogenously derived on the basis o relative bargaining power. Making λ endogenously however, would make the model more complex without changing its main results. 8 The manager always reports the downstate ater observing it since, by assumption, there are no nonpecuniary beneits o control. We make this assumption to keep the model simple and ocused on the role o implicit contracts as a vehicle or the misappropriation o company unds. 16

17 The manager chooses the level o ϕ t that maximizes Wt U. It ollows rom (2) that: takeover at t Pr = Pr ( Z > Z I = 1) = Pr ( u > Z ρ ) = 1 Φ ( Z ρ ) downstate is alsely reported (8) and no takeover at t = Φ ρ (9) Pr downstate is alsely reported ( Z ) Substituting (8) and (9) into the objective unction, taking the derivative with respect to ϕ t and noting that the terms [ t 1 takeover at t] EW + and [ ] EW t + 1 no takeover at t are independent o ϕ t, we get dwt u u d d = απ +Φ( Z ρ ) λ ( π π ) απ + dϕ t ( Z ρ ) [ ] EW [ ] EWt+ 1 no takeover at t t+ 1 takeover at t + 1 Φ 1+ r (10) This derivative is independent o ϕ t which implies that the manager's optimal reporting strategy, ϕ t, is o the bang-bang type (i.e., ϕ t is either zero or one). The manager is more likely to report the downstate truthully i (i) the gain in expected uture wealth rom keeping the job is large, (ii) her ownership stake is large, (iii) the probability o takeover is high, (iv) the raction o the unds diverted rom the irm that she appropriates or hersel is small and (v) the public signal is highly correlated with the deal s true outcome. We turn now to the relationship between the manager's optimal reporting strategy and the takeover rule adopted by the raider. Inspection o (10) reveals that the manager can always be induced to report deal outcomes truthully by setting Z suiciently low. 17

18 To understand this result note that as, becomes positive; conversely, as, Z ( Z ρ ) Z + ( Z ρ ) Φ 0 and expression (10) Φ 1 and expression (10) becomes negative (as long α < λ ). To put it into words, the manager always tells the truth i she is certain that a downstate announcement triggers a takeover; conversely, the manager always lies i she is convinced that there is no takeover ollowing a report o the downstate. Another key eature o expression (10) is that it is monotonically decreasing in Z. That implies that there exists a maximum value o Z (which we denote max Z ), that's necessary and suicient to induce the manager to tell the truth. In sum, the manager will only report the upstate truthully i Z < Z. max II.2.III. The counterpart s problem A stakeholder who is evaluating a possible business deal with the irm will consider the likelihood o a takeover ollowing a downstate announcement. The ex-ante value o the business deal to the counterpart is equal to NPV ( Z ) u = 0.5ϕ π + counterpart t c ( ( )) no takeover u u d ϕt Z Pr πc + ( 1 λ)( π π ) + downstate is alsely reported no takeover d + 0.5Pr π c + downstate is truthully reported takeover + 0.5Pr v downstate is truthully reported (11) 18

19 From (2) it ollows that: no takeover Pr = Pr ( Z < Z I = 0) = Pr( u < Z ) =Φ Z downstate is truthully reported ( ) (12) takeover Pr = Pr ( Z > Z I = 0) = Pr( u > Z ) = 1 Φ Z downstate is truthully reported ( ) (13) Substituting (8), (9), (12) and (13) into expression (11) yields { ( u ) ( d ϕ π )( π )} NPV = 0.5 Z + v +Φ Z + v + counterpart t c c {( ( )) ( ) ( )( ) } u u d ϕt Z Z ρ πc λ π π Φ + 1 (14) The counterpart will enter into a deal only i (14) is non-negative. Denote min Z as the lowest value o Z at which the NPV o the deal to the counterpart is equal to zero. Since or Z < Z the NPV is negative, the counterpart will only accept making deals i min Z Z. min II.2.IV. Equilibrium To study the model s equilibria one needs to characterize the belies held by the counterpart vis-à-vis the takeover rule ollowed by raiders, and also how these belies change in ace o contradictory evidence. To keep the solution simple we assume that the counterpart initially believes that the raider's takeover rule satisies Z Z min. This belie is held as long as it is not contradicted by evidence; i however, a raider launches a takeover when Z min < Z, the initial belie is shattered and the counterpart will permanently reuse to make more deals with the irm. 19

20 Under these assumptions the equilibrium o the model is determined by the relationship between the maximum threshold that is required to induce managers to report outcomes truthully (i.e., break-even (i.e., min Z ). max Z ) and the minimum threshold that allows counterparties to Consider irst the case where Z max Z min. Here there exist many takeover rules (i.e., all values o Z satisying max min Z Z Z ) that induce managers to report outcomes truthully and simultaneously allow the counterpart to earn, on average, positive proits on its business transactions with the irm. Among all easible takeover rules however, only the rule Z = Z min is sel-consistent. This rule produces an equilibrium with the ollowing characteristics: (i) Ater the downstate is reported the raider launches a takeover i and only i Z min Z ; (ii) (iii) Managers always report deals outcomes truthully; The ex-ante value o deals to the counterpart is equal to zero so that the counterpart accepts continuing the relationship with the irm; (iv) The value o the irm is equal to V min ( ) u d 0.5 π + π Φ Z =. r Hence, in this equilibrium the irm is able to contract eiciently with its stakeholder despite the ongoing takeover activity. The more interesting equilibrium occurs when Z max min < Z. In this case the level o takeover activity that is required to induce the manager to report deals outcomes truthully makes the ex-ante value o deals to the counterpart negative. That implies that 20

21 business deals between the irm and the stakeholder cannot be sustained in equilibrium and thus the value o the irm collapses to zero. The two equilibria are illustrated in igure 2. The equilibrium which sustains implicit contracts is presented in Panel B. It eatures a more precise signal o the deal's true outcome, thus making it possible or takeovers to be targeted more accurately towards managers who alsely report the downstate. The counterpart is able to break-even despite the on-going takeover activity because takeovers are unlikely ater a truthul report o the downstate. The equilibrium without implicit contracts is presented in Panel A. Here, the irm cannot contract eiciently with the stakeholder because takeovers are too blunt to eectively discipline managers. A noisy signal makes it diicult or the raider to distinguish between a alse and a truthul report o the downstate. To induce managers to be honest, raiders make too many mistakes targeting irms that truthully report the downstate. With so many misguided takeovers the stakeholder cannot breakeven and consequently, preers to discontinue his relationship with the irm. 21

22 ϕ(z ) Panel A Public signal Z has a low correlation with the deal s true outcome (i.e., ρ is small) ϕ(z ) Panel B Public signal Z has a high correlation with the deal s true outcome (i.e., ρ is large) 1 Z Max 1 Z Max dw/dϕ Z Z dw/dϕ NPV counterpart NPV counterpart Z Z Z Min Z Min Figure 2 The two equilibria o the model In panel A the public signal is noisy, so that in equilibrium the irm cannot sustain business deals and irm value collapses to zero. In Panel B the public signal o the deal s true outcome is precise, so that in equilibrium the irm sustains business deals and the manager reports deals outcomes truthully; in u d min 0.5 π + π Φ( Z ) this case, the value o the irm is equal to V =. r II. 2. V. A numerical example A numerical example helps to illustrate the model. Consider a business deal whereby the manager periodically requests a supplier to make a relationship-speciic investment that is beneicial to both parties. The manager s request to the supplier is made against the promise o ull reimbursement o the investment cost, which is initially uncertain. In the upstate the investment cost borne by the supplier is equal to zero; in the 22

23 downstate it is equal 10. As a result o the investment the irm obtains a once-and-or-all proit o 6 whereas the supplier obtains a once-and-or-all proit o 2. Accordingly, the payo to the irm rom a deal that is upheld by both parties is π = 6 0= 6 u in the upstate and π = 6 10= 4 d in the downstate; the payo to the supplier is equal to 2 in u d either state (i.e., π = π = 2 ). c c I a takeover occurs ollowing an announcement o the downstate, the new management reuses to pay the investment cost claimed by the supplier. Since the supplier cannot use the courts to enorce his contract with the irm, he has no choice but to accept the deal break-up and try to recoup as much as possible o the investment cost. Assuming that he can recoup 70% o the investment, his payo ollowing a takeover is equal to 0 in the upstate and equal to v = ( 1 0.7) ( 10) = 3 in the downstate. 9 Next we determine the amount o unds available or diversion ollowing a alse report o the downstate. I there is no takeover, the manager withdraws rom the irm an amount equal to the supplier s investment cost in the downstate. Since the real investment cost borne by the supplier is zero, the manager and the supplier can jointly divert away rom the irm an amount equal to 10. On the other hand, i a takeover occurs no unds are available or diversion. Assume that the manager s ownership stake in the irm is equal to 10% (i.e., α = 0.1) and that the supplier has no bargaining power over the division o the diverted unds (i.e., λ = 1). Regarding the public signal, Z, let the random variable, u, be 9 Note that the chosen parameters satisy all the conditions enumerated in section II.1: The ex-ante value o the deal to both parties is positive; the supplier never gains by breaking up the deal; the irm gains by breaking up in the downstate; and inally, an eicient ex-post renegotiation o the deal in the downstate cannot prevent the irm rom breaking up the deal. 23

24 distributed uniormly between 1 and 1. Under this assumption, Pr 1+ x 2 1 x > = Φ =. 2 ( u < x) =Φ ( x) = and Pr ( u x) 1 ( x) We can now investigate the model s equilibria. The maximum threshold o the takeover rule that gives the manager the inducement to report outcomes truthully (i.e., max Z ), is obtained by setting expression (9) equal to zero and solving it or Z : dw t 1+ Z ρ = ( 0.1) 6 + ( 0.1) 4 ( 6 + 4) + dϕt 2 [ no takeover at t ] EW [ takeover at t] 1 Z + ρ EWt+ 1 t+ 1 + = 0 (16) 2 1+ r We know that the manager s optimal reporting strategy in the current period is o the bang-bang type i.e., ater observing the upstate, the manager either tells the truth and reports the upstate or lies and reports the downstate. Since we are looking or a solution where the current outcome is reported truthully, to guarantee time consistency such solution must encompass truthully reporting o deal outcomes in every uture period. In that case [ ] EW [ ] EW no takeover at t takeover at t = 0 (17) t+ 1 t+ 1 since, by assumption, the manager enjoys no non-pecuniary beneits rom control. Substituting (17) in (16) yields: dw t 1+ Z ρ = [ ] = 0 dϕt 2 (18) 24

25 which has the solution Z max = ρ Consequently, in order to induce the manager to reveal the outcomes on implicit contracts truthully the raider must launch a takeover in response to a report o the downstate i and only i the public signal exceeds ρ The ex-ante value o the deal to the counterpart as a unction o the raider s takeover rule is given by expression (15). Substituting the assumed parameter values yields ( Z ) 1+ 5 NPVcounterpart = 0.5 ϕt ( Z ) ϕt ( Z ) ( 1+ Z ρ) + 2 (19) The two equilibria o the model are easy to characterize. For the good equilibrium to prevail the supplier must achieve a positive NPV when the raider s takeover rule is equal to ρ (and so the upstate is reported truthully); that is ( ) NPV Z Z 5 = ( 1+ ρ 0.875) 0 2 = = counterpart = ρ 0.875; ϕt = ρ (20) which is satisied i ρ When ρ < the bad equilibria prevails. III. Discussion The model has a number o important parameters that inluence the type o equilibrium that prevails. We have already seen that a noisy public signal (i.e., a low ρ ) makes it harder or irms to sustain relationships with stakeholders. That suggests that regulations or policies aimed at narrowing the inormational gap between management and outside shareholders such as an improvement in the disclosure o inancial 25

26 perormance as recommended by the OECD 10 - will help irms vulnerable to hostile transers o control secure relationships with stakeholders, thus diminishing the importance o controlling shareholders in corporate governance structures. The ownership stake o the manager also plays an important role. An increase in α reduces the manager's beneit rom misrepresenting deals outcomes. That in turn raises Z max thus helping to sustain business deals. One can interpret these eects under the light o Fluck s (1998) model o outside equity. Fluck shows that outside equity inancing is sustainable i managers and investors interact repeatedly over time and managerial actions are observable by outsiders. We depart rom Fluck s ramework in assuming that managerial actions are only imperectly observable. To sustain external equity our model requires the manager to hold a cashlow claim, so that her interests are more closely aligned with those o outside investors. Speciically, the model requires the manager s cash low claim to be above a minimum threshold or her to report outcomes on implicit contracts truthully; moreover, such threshold is negatively related to the precision o the public inormation available about managerial dealings with stakeholders. Another actor is captured by the parameter k, which measures the amount o transaction costs incurred by the raider when launching a hostile acquisition. Transaction costs may simply consist o costs associated with putting together a tender oer such as legal costs and ees paid to investment bankers. They may also consist o indirect costs stemming rom the ree-rider problem that occurs when the ownership o shares among 10 OECD- Principles o Corporate Governance (1999). The OECD recommends that disclosure should include inormation on material issues regarding key executives, employees and other stakeholders; that inormation about publicly traded irm should be prepared, audited, and disclosed in accordance with high standards o accounting, inancial and non-inancial disclosure, and audit; that an annual audit should be 26

27 outside investors is widely dispersed. Yet another source o transaction costs are the costs o overcoming takeover amendments enshrined in corporate charters - such as poison pills and supermajority rules - designed to make hostile takeovers expensive to the acquirer. The transaction cost parameter determines the net gain to the raider rom taking over the irm. With low transaction costs it's proitable to take over a irm announcing losses on implicit contracts, provided that the takeover does not cause stakeholders to withhold uture dealings with the irm. With large transaction costs takeovers become unproitable thereby giving the manager eective control o the irm. This last case corresponds to that o a irm eaturing a controlling shareholder who holds a small cash low claim. Under such an ownership and governance structure the irm contracts eiciently with stakeholders but the controlling party expropriates outside shareholders. The role o ownership concentration in the model has an interesting parallel with that suggested by Burkart, Gromb and Panunzi (1997). These authors present a model where ownership concentration acilitates the intervention o outside shareholders in the irm s management. The upside o increased intervention is to limit managerial moral hazard over project choice, i.e., the manager is curtailed in her ability to choose projects with high private beneits and low cash low. The downside is to reduce the incentive held by the manager to search or proitable projects. Hence, a higher ownership concentration reduces managerial moral hazard but also reduces the value o the projects available to the irm. I we view the relationship between the manager and the irm as an implicit contract whereby the irm promises the manager to give him discretion over conducted by an independent auditor in order to provide an external and objective assurance on the way inancial statements have been prepared and presented. 27

28 project choice in exchange or a commitment to invest eort in project research, then we can argue that excessive shareholder intervention disrupts the irm s ability to sustain implicit contracts with management. Their model however, diers rom ours in an important respect. In their model a irm controlled by an inside shareholder selects projects solely on the basis o cash low - so that the private beneits generated by projects are completely disregarded by the controlling shareholder. Hence, such a irm cannot sustain an implicit contract with its management. In our model the same controlling shareholder considers in his choice o project the private beneits accruing to the manager rom each available alternative, since he knows that ignoring the manager s preerences weakens her motivation to search or new projects. This distinction between the two models leads to opposite implications regarding the eect o ownership concentration. Whereas in Burkart et al. an increase in ownership concentration reduces the irm s ability to sustain implicit contracts with its management, in our model it will produce the opposite eect as long as higher concentration shits the balance o power rom outside to inside shareholders. The two models also yield opposite implication with respect to the role o inormational asymmetries between controlling shareholders and management. Narrowing inormational asymmetries helps implicit contracting in our model and hinders it in their model. At a deeper level, the issues raised in this paper delve into the relationship between the nature o the irm s assets and the nature o the irm s inancing arrangements. A irm which needs to engage in private transactions with third parties to create value cannot easily deal with arm s length inancing, be it external equity or risky 28

29 debt. That means that we will observe a limited usage o arm s length inancing whenever implicit contracts are important to mediate business transactions. La Porta et al. (1997) document that the legal environment is a key determinant o the size and extent o a country s capital market. In particular, they ind that countries whose law originates rom the French civil code and in which the quality o law enorcement is poor have weak capital markets. In the same vein, Demirguç-Kunt and Maksimovic (1998) report that companies rom countries which score poorly on an index o respect or legal norms, use little long-term external debt and equity to und their growth. Our model suggests an explanation or such patterns o inance. Because in these countries it is ineicient or irms to do business through explicit contracts, transactions with stakeholders migrate to the ramework o implicit contracts. When that happens inside inancing emerge endogenously as a means to provide or secure business relationships. IV. Funding growth with external equity We have seen that irms which rely on implicit contracts need a controlling shareholder to contract eiciently with stakeholders. One downside o having a controlling shareholder is that it makes it harder or the irm to obtain external equity, since outside investors ear being expropriated. I the irm needs cash to inance investment opportunities the reluctance o investors in providing unds can be very detrimental or irm value. 11 Is there a way or the controlling shareholder to raise money by selling shares to outsiders? 11 Demiguç-Kunt and Maksimovic (1998) report that irms rom countries with a low rating or compliance with legal norms tend to grow slower and to obtain less external inancing. These are countries where implicit contracts are likely to be important. 29

30 There are several possible answers to this question. Firms with growth opportunities can raise unds rom outside investors gradually over time, coming back to the market on an on-going basis until they have unded all available investment opportunities. Under this arrangement the controlling shareholder rerains rom expropriating outside shareholders because she knows she will need them later on. 12 This strategy o stage inancing however, is inherently unstable in a rational expectations ramework, because the incentive o the controlling shareholders to distribute unds to outsiders weakens over time as the amount o ununded growth opportunities diminishes. 13 Another solution or the controlling shareholder is to sell the entire irm to an investor who is suiciently wealthy to und all the irm s growth opportunities. This argument suggests that in countries where implicit contracts are important we should observe a higher concentration o corporate control, since irms with good growth prospects but short o cash will generally be acquired by bigger and more established companies, instead o being loated in the capital market. LaPorta et.al. (1999) ind that in countries such as Sweden, Belgium, Greece, Portugal and Israel controlling shareholders (i.e., amilies and banks) control, on average, more than two o those countries 20 largest companies, a number that is much higher than the corresponding igures or the US or the UK. They interpret this result as evidence o very signiicant control o productive resources by the largest controlling shareholders. The level o 12 Bullow and Rogo (1989) use a similar argument in the context o sovereign borrowing. 13 According to this view the crashes in equity prices occurred in recent years in emerging economies can perhaps be explained by a sudden downgrade in investors expectations about the growth prospects o local companies. Prices ell dramatically because investors realized that the deterioration o business opportunities caused controlling shareholders to abandon their normal restraint regarding the expropriation o outside investors. An alternative explanation is suggested by Johnson, Boone, Breach and Friedman 30

31 concentration o control o corporate resources is likely to be even higher among less developed nations. Yet another answer is or the controlling shareholder to sell a minority stake to a inancier, invite him to join the board o directors and give him the power to cause trouble to the irm. For example, the inancier could be oered the power to block managerial decisions through a requirement o qualiied majority voting on a number o issues o strategic importance to the irm. I the irm engages in implicit contracts involving illegal dealings with stakeholders then the inancier can hurt the irm simply by blowing the whistle. What s important here is to give the inancier the ability to retaliate i he eels he is being expropriated. Since the inancier sits in the board o directors he observes the outcomes on implicit contracts and hence, knows exactly when the controlling shareholder is misappropriating corporate resources. In a similar vein, Bennedsen and Wolenzon (2000) develop a model or closely held corporations where an owner-entrepreneur, in need o external capital, commits to a low level o ineicient extraction o private beneits o control by choosing an ownership structure eaturing multiple large shareholders. Such structure orces shareholders to talk to each other and to act in a concerted ashion, which helps to internalize the consequences o ineicient diversion o unds. V. Risk diversiication Even i the irm has all the cash it needs to und growth opportunities the controlling shareholder will still be interested in selling a minority equity stake just to (2000). These authors argue that the deterioration o the economic prospects o East Asian irms reduced the opportunity cost o diverting unds rom legitimate investment activities to controlling shareholders. 31