Monitoring and Collusion in the Family Firm Part 2

4. Model 4.1 Manager-CEO Collusion and Proxy Fights My model consists of a three-period game that explores the largest non-family shareholder’s ability to start a proxy fight and put independent directors of his choosing on the board in order to diminish the family’s ability to reduce firm profits for its own purposes. The game has three players: the manager, the family, and the largest nonfamily shareholder. The other shareholders are assumed to be passive. Since their individual ownership shares are very small, each has only a small incentive to monitor. Their action in the case of a proxy fight is important: their votes determine who wins or loses. However, I assume that their votes are determined by the costs the family and the largest shareholder take on in case of a proxy contest, which include the costs of persuading these small shareholders. In the first period, the family begins with complete control over the board and monitors the CEO. As large shareholders, they may choose to collude with the CEO at the expense of shareholder value. In the second period, however, the non-family shareholders can observe the firm’s profits as a result of the CEO’s actions in the first period and may choose to start a proxy fight in order to gain control of the board and thus prevent the family from extracting private benefits in the future. The actions of the first period are repeated in the third, except that if the shareholder has won the proxy fight, then the family has monitoring power but has lost its ability to collude. The timeline is as follows: 16 Period One: 1. Manager decides to work, shirk, or take a less profitable action that yields private benefits he can share with the family. 2. Profits are realized and observed by the manager, the family, and all shareholders. With some probability h, the largest shareholder observes a private benefit, if any is taken. Period Two: 1. The largest shareholder decides to whether to wage a proxy fight, based on first period profits and observation of private benefits. 2. Family decides to concede or fight. 3. Outcome of proxy fight determined. Period Three: 1. Manager decides to work, shirk, or collude with the family. If the outsiders have gained dominance on the board, they can prevent collusion. 2. Profits are realized and observed by all players. We assume for now that shareholders can buy and sell their shares easily at the current value, and that there is no possibility of a takeover, regardless of firm profits. The first assumption does not directly affect my simple model, but it does affect a real-life shareholder’s incentive to start a proxy fight1. It also simplifies the results: if a proxy fight is unsuccessful, the dissident shareholder usually sells his shares in the company. 1 Maug (1998) examines how stock market liquidity affects a shareholder’s incentive to monitor. As shares become easier to sell, a shareholder’s incentive to monitor falls. However, a liquid stock market also makes it easier for the shareholder to buy a bigger share in the company, which lessens free-riding. 17 This does not affect total firm value in this model, however, since we assume that these shares will be bought for their real value. Takeovers would add another discipline device to the family, making it even less likely that they would collude with management. I look briefly at this possibility in Section 4.2. There are three players in the game: the manager, the family, and the largest shareholder. They are each entitled to a certain fraction of the firm’s profits: s1 for the manager, s2 for the family, and s3 for the largest non-family shareholder. The remaining profits, s4 = 1 - (s1 + s2+ s3) are claimed by the small shareholders. The model hinges on the manager’s choices. The manager is choosing a project that will result in firm profits R if successful and 0 if unsuccessful. If he works hard, the project will succeed with probability p. However, if left to his own devices, he would prefer to shirk by running the project in another way that reduces the probability of success to p-µ1 but produces private benefits B > 0 that only he can partake in. He may also choose an action which produces lower private benefits b for him, but where these benefits also go to the controlling family. This action reduces the probability of success to p-µ2, where µ2 <µ1. The private benefits b that are shared with the family can take on value bH or bL, depending on the family’s ability to extract private benefits. Perhaps it is difficult for them to do so because they cannot act as a group or because they do not have a close relationship with the manager. In that case, b=bL and they will not collude. If b=bH then they will. The non-family shareholders do not, however, know the value of b except if they observe the private benefit being taken, in which case they know that b=bH. The 18 outside shareholders discover the private benefits, if they are taken, with probability h. They also know that the probability that b=bH in the population of family firms is λ. By monitoring the manager, the board can control his choice of action. However, monitoring is costly. For the family on the board, it costs c > 0 to monitor the manager if the manager is not a family member. If the manager is a family member, the family’s monitoring cost is 0 because the manager cannot easily conceal information from his family. For the outsiders on the board, monitoring costs k > c > 0. Monitoring is costlier for outsiders because they have less knowledge about the firm and the manager’s choice of projects. Since the outsiders are typically executives of other firms, while family members are not, their opportunity cost of serving on the board and monitoring the manager is also higher. Assumption 1 : ( p − µ1 ) s1 R +B ≥ ( p − µ 2 ) s1 R +b ≥ ps1 R In other words, the manager would most like to shirk. If he is prevented from doing so, he would rather collude with the family than to work hard. Assumption 2: B ≤ µ1R and b ≤ µ2R Shirking destroys firm value. Assumption 3: If the family and the manager colluded in Period 1 then they will collude in Period 3 as well. They can collude in Period 3 only if they have colluded in Period 1. Therefore the non-family shareholder will want to act if he observes private benefits taken in the first period. Assumption 4: It is always in the family’s interest to monitor and prevent the manager from shirking. If b=bH, the family takes private benefits, so this means that: ( p − µ1 ) s 2 R ≤ ( p − µ 2 ) s 2 R + b H − c 19 which simplifies to: c ≤ ( µ1 − µ 2 ) s 2 R + b H If b=bL, the family will make sure that the manager works hard when they monitor: ( p − µ 2 ) s 2 R + bL − c ≤ ps 2 R − c which simplifies to: and ( p − µ1 ) s 2 R ≤ ps 2 R − c c ≤ µ1 s1 R Assumption 4 shows that the family will monitor because the cost of monitoring is less than their share of the lost profit from the monitors’ shirking. Note that if the manager is chosen from within the family, it is always in the family’s interest to monitor, since c = 0. Note also that holding a smaller share in the company reduces the family’s incentive to monitor, since µ1s2R becomes smaller. Now we can solve the game by backwards induction. First, consider the payoffs of each possible outcome at Period 3. Outcome 1: If the largest outside shareholder does not start a proxy fight or if the proxy fight is unsuccessful, then the family retains control. By assumption, the family and the manager choose to collude if b=bH. Total surplus = (p-µ2)R +b - c. If b=bL then they do not collude, and total surplus = pR-c. Outcome 2: If the family remains on the board but the board is dominated by outsiders, then the family can monitor the manager but is prevented from colluding. Total surplus = pR-c. Outcome 3: If the family is completely taken off the board, then it will cost more for the board to monitor the manager. Total surplus = pR- k. These outcomes correspond to Anderson and Reeb’s finding that a board with a mix of family and outsiders maximizes firm profits. The family can collude with the manager, at the expense of shareholders, if they are not prevented from doing so by 20 outsiders on the board. However, if there are not enough family members on the board to effectively monitor, total surplus falls because it costs more for outsiders to monitor. Now, going back to the beginning of period 2, the largest shareholder considers the results of his waging a proxy fight to replace family members on the board with outsiders. Suppose that threatening a proxy fight costs the largest shareholder t, with an added cost κ if the family fights back instead of conceding to its request. The proxy contest succeeds with some probability φ. φ depends on the way the company is structured and how easy it is for the large shareholder to persuade the small shareholders to vote for his candidates for the board. If the family fights back, then they must spend cost χ. If they win, they again receive the payoff from Period 1. If they lose, they are prevented from colluding in Period 3. As monitors, they can in that case choose either to make sure that the manager works hard or to let him shirk. By Assumption 4, they are better off if he works hard, so they will monitor the manager and make sure he works. Proposition 1: The family will fight back in a proxy fight only if they have been colluding with the CEO. If the proxy fight will produce Outcome 2, then the family will fight back if (1) (1 − ϕ )bH > (1 − ϕ ) µ 2 s 2 R + χ If the proxy fight will produce Outcome 3, then the family will fight back if (2) (1 − ϕ )bH > (1 − ϕ ) µ 2 s 2 R + (1 − ϕ )c + χ Proof of Proposition 1: If the family has not been colluding with the manager, then they do not fight back in a proxy fight because their payoff in Period 3 is the same whether they win or lose, but fighting back is costly. 21 If the family has been colluding and the proxy fight will put more independents on the board, but not enough to replace the family as primary monitor, then the family will fight back if: (1 − ϕ )[( p − µ ) s R + b − c] + ϕ[ ps R − c] − χ > ps R − c, 2 2 H 2 2 which simplifies to equation 1. If the family has been colluding and the proxy fight will put so many outsiders on the board that the family no longer monitors the CEO, then the family fights back if: (1 − ϕ )[( p − µ 2 ) s 2 R + bH − c] + ϕ[ ps 2 R] − χ > ps 2 R which simplifies to equation 2. Proposition 1 simply says that the family will fight back only if the expected loss of private benefits from resisting a proxy fight is greater than the cost of fighting and the potential benefits of the new board, which will make sure that expected firm profits are higher and, if it produces Outcome 3, take away the family’s need to monitor the manager. If the private benefits are very high, the family will fight. If it is costly for them to fight back, then they will concede to the dissident’s request without fighting. Proposition 2: The largest outside shareholder threatens a proxy fight when one of the following is true: (a) He observes the family’s private benefits, the family will not fight back, and t<µ2s3R. (b) He observes the family’s private benefits, the family will fight back, and φµ2s3R>t+κ. (c) He does not observe private benefits but the project failed in Period 1 and λ * ϕµ 2 s3 R > t + κ (if the family will fight back) or λ * µ s R > t 2 3 22 (if they will concede) where (d) λ* = λ (1 − p + µ 2 ) 1 − p + λµ 2 He does not observe private benefits, the project succeeds in Period 2, and λφµ2s3R > t+κ (if the family will fight back) or λµ2s3R > t (if they will concede) Proof of Proposition 2: (a) If the outside shareholder observes the family’s private benefits, then he knows for certain that the manager has not been working hard. If the family will not fight back during a proxy battle, then by threatening a proxy fight, the outside shareholder knows that the family will allow his candidates onto the board without his having to spend к: ( p − µ 2 ) s3 R < ps3 R − t He will thus threaten a proxy fight if the costs of doing so do not outweigh the loss he suffers from the manager’s profit-reducing action. (b) As in case (a), the outside shareholder knows that the manager and the family have been colluding, so he threatens a proxy fight if: ( p − µ 2 ) s3 R < ϕps3 R + (1 − ϕ )(( p − µ 2 ) s3 R) − (t + κ ) The left-hand side is his payoff in Period 3 if he does not act. The right-hand side shows his payoff in Period 3 if he starts a proxy fight in Period 2. He will have to pay costs t+κ whether he wins or not. If he wins, which occurs with probability φ, he receives ps3R. If not, the manager and the family will collude again in Period 3. This equation simplifies to φµ2s3R>t+κ, which says that he will threaten a proxy fight only if the costs of going through with it are less than his expected gain. 23 (c) If he doesn’t observe private benefits being taken, then he knows that with probability λ, they are being taken. If the project fails in Period 1, the likelihood of collusion increases to λ*. λ* can be found with Bayes’ Rule: λ* is the probability that the family and manager have been colluding, given that the project fails in Period 1. Since the probability that the project will fail when the manager and family are colluding is 1-(pµ2), the probability of collusion is λ, and the probability of the project failing is λ (1-(pµ2)) + (1- λ)(1-p), λ* = λ (1 − p + µ 2 ) 1 − p + λµ 2 Notice that, as you would expect, λ*> λ because 1-p+ µ2 > 1-p+ λ µ2 (since λ<1). In other words, if the project fails in Period 1, then it is more likely that the manager and the family are colluding. The outside shareholder will then go through with a proxy fight if: ϕps3 R + (1 − ϕ )[λ * ( p − µ 2 ) s3 R + (1 − λ *) ps3 R] − (t + κ ) > λ * ( p − µ 2 ) s3 R + (1 − λ*) ps3 R which simplifies to: ϕλ * µ 2 s3 R > t + κ (d) If in Period 1 the project succeeds and the outside shareholder does not observe the manager or family taking private benefits, he will still expect them to be colluding with probability λ. Therefore it is in his interest to start a proxy fight if: ϕps3 R + (1 − ϕ )[λ ( p − µ 2 ) s3 R + (1 − λ ) ps3 R] − (t + κ ) > λ ( p − µ 2 ) s3 R + (1 − λ ) ps3 R Notice that if λ is large and there are many firms where private benefits are large enough to tempt the family and the CEO to collude, then the outside shareholder is more likely to start a proxy fight even if he has not directly observed any sign of collusion. 24 Proposition 3: The family will not collude with the CEO in Period 1 if: (1) λ * ϕµ 2 s 2 R > t + κ and (2) ϕµ 2 s 2 R > t + κ > λ * ϕµ 2 s 2 R and χh ≥ (2 − hϕ )(bH − µ 2 s 2 R) χ ≥ (2 − ϕ )(bH − µ 2 s 2 R) Proof of Proposition 3: First note that because λ*< 1, λ*φµ2s3R < φµ2s3R. Also, if φµ2s3R < t+ κ then the outside shareholder cannot credibly threaten to carry through a proxy fight, so the family can continue extracting its private benefits in Period 3. (1) If φµ2s3R > t+ κ, the outside shareholder will only try a proxy fight if he observes private benefits being taken, which occurs with probability h. Therefore, if the family is prepared to spend χ in then event of a proxy fight, it will not collude in Period 1 if: 2( ps 2 R − c) ≥ (1 + (1 − h))(( p − µ 2 ) s 2 R + bH − c) + h[ϕ ( ps 2 R − c) + (1 − ϕ )(( p − µ 2 ) s3 R + bH − c) − χ ] which simplifies to: χh ≥ (2 − hϕ )(bH − µ 2 s 2 R) If the family is not prepared to fight back, then it can never be prevented from colluding in Period 1. In that case, the family would lose the fight. In order for it to be worthwhile for them to not collude in the first round, it would have to be the case that: 2( ps 2 R − c) ≥ (1 + (1 − h))(( p − µ 2 ) s 2 R + bH − c) + h( ps 2 R − c) This is never true, because ps3R < (p-µ2) s3R+bH. (2) If λ*φµ2s3R > t+ κ, then the outside shareholder will start a proxy fight whether he observes private benefits or not. In that case, the family will not collude with the manager if: 2( ps 2 R − c) ≥ ϕ ( ps 2 R − c + ( p − µ 2 ) s 2 R + bH − c) + (1 − ϕ )(2( p − µ 2 ) s 2 R + bH − c) − χ or, more simply, χ ≥ (2 − ϕ )(bH − µ 2 s 2 R) 25 Proposition 3 shows that the family can be prevented from ever colluding with the CEO if their cost of fighting back once a proxy contest has been threatened is sufficiently high to outweigh their net gain from collusion. 4.2 Takeovers and Family Management We can use the three propositions to look at takeovers and the decision to hire a family member as CEO. First, consider the case of family management. Monitoring a family member is costless, so the family is indifferent between outcomes 2 and 3 of a proxy fight. Appointing a family member may also be a sure signal to the outside shareholder that the family will collude with the manager, so h is close to 1 in this case. Therefore, the family will only hire a family member if no outside shareholder will fight back, either because the costs of starting a proxy fight are too high, the probability of it succeeding is to low, or because no single outside shareholder holds a large enough stake in the firm to make it worth his while. Takeovers are an interesting case because they affect the size of each player’s shareholdings. When a company takes over, its share in the acquired company rises. As a result, its interest in the firm’s share value also increases, so it may find it in its interest to hire its own representatives to monitor the manager. The family, on the other hand, ends up with a smaller share in the firm, so it is less willing to monitor the CEO. Takeovers may also be more costly to the family than proxy fights because the family risks losing its highly-valued reputation as the company’s controlling family. It may therefore be more costly to initiate a takeover than to announce a proxy contest. 26