سهامداران بزرگ، نظارت و پویایی های مالکیت: به سوی شرکت های مدیریتی خالص؟

We study ownership dynamics when the manager and the large shareholder, both risk neutral, simultaneously choose effort and monitoring level respectively to serve their non-congruent interests.We show that there is a wedge between the valuation of shares by atomistic shareholders and the large shareholder's valuation. At the Markov-perfect equilibrium, the large shareholder divests her shares. If the incongruence of their interests is mild, divestment is drastic: all her shares are sold immediately. If their interests diverge sharply, the divestment is gradual in order to prevent a sharp fall in share price. In the limit the firm becomes purely managerial.

This paper develops a dynamic model of divestment by a large shareholder of a firm where her interest and that of the manager are not perfectly congruent. In our model, all shareholders, large and small, are risk neutral and have perfectly congruent objectives; however, only the large shareholder monitors the manager, while the small investors free ride on her monitoring effort.1 We show how the degree of divergence of interests between the manager and the large shareholder affects the process of divestment. We demonstrate that when their interests diverge sharply, the divestment is gradual in order to prevent a sharp fall in share price. In the limit the firm becomes purely managerial, with a diverse ownership, and no monitoring by shareholders. This paper thus serves to highlight a mechanism that lies behind the tendency for corporate governance to move gradually from concentrated to dispersed ownership, a pattern that has been observed over more than a century in major capitalist economies (such as Great Britain and USA), and also more recently in countries such as Brazil. The key to our explanation is that the large shareholder cannot resist the temptation to sell shares when small investors' marginal benefit flow is greater than her own. While reducing her ownership (which entails a decrease in her monitoring effort) adversely affects the dividend flow to all investors, it does elicit more effort from the manager. Berle and Means (1932) pointed to the transition to dispersed ownership in US. Recent empirical work confirms this tendency.2 For UK, the same tendency was reported in Scott (1990), Franks et al. (2005), among others. Gorga (2009) documented a similar trend in Brazil from 1997 to 2002. Various reasons have been offered to explain the tendency for reduced concentration of ownership. Subrahmanyam and Titman (1999) argue that it becomes advantageous for firms to have a more dispersed ownership when informational asymmetries between insiders and external investors are less important. Roe (1996) and LaPorta et al. (1999) attribute the dispersion of ownership in US to the specific US laws and policies that discourage ownership concentration. In this paper, we explain the tendency toward dispersed ownership by modelling, on the one hand, the trade-off between the gains from monitoring by a large shareholder and those from managerial initiatives, and on the other hand, the incentives for the large shareholder to divest (gradually, in typical cases) when her marginal valuation of ownership are below the small investors' valuation of the dividend stream that would arise on the assumption that she does not divest. The former aspect was investigated in an elegant static model by Burkart et al. (1997). The latter aspect is built on the literature concerning the Coase conjecture. Coase (1972) argued that when a monopolist producing a durable good at constant marginal cost cannot commit, rational expectations by potential buyers, and his ability to sell repeatedly would result in only one possible equilibrium outcome: he can only charge the price that would prevail under perfect competition, and the market demand is satisfied instantaneously.3 In our model, where the large shareholder corresponds to the Coasian monopolist, we show that Coase's conjecture holds if the divergence of interests between the large shareholder and the manager is mild; in contrast, if this divergence is very strong, the Coase conjecture fails, and the large shareholder will divest only gradually, with share price falling slowly over time, converging only in the long run to the competitive price. There is also an intermediate case, in which at first the large shareholder undertakes a massive sale of shares, to be followed by a slow process of divestment of the remaining shares. The intuition behind our results is simple. In all cases, the divestment is caused by the fact that small shareholders perceive that, under the assumption that the large shareholder would not divest, their dividend stream per share is worth more than the large shareholder's marginal returns on a share (as she has to incur the monitoring cost). This wedge in marginal valuations implies that equilibrium must involve share trading. When the divergence of interests between the manager and the large shareholder is mild, her total instantaneous payoff (net of monitoring cost) is a strictly concave and increasing function of her fraction of ownership. Therefore, the revenue she would obtain from selling her shares at the competitive share price strictly dominates the present value of the stream of her instantaneous payoff obtained from maintaining her initial stock. Hence, her optimal policy is to sell off all her shares in one go. In the reverse case, the strong divergence of interests implies that her total instantaneous payoff is a strictly convex and increasing function of her fraction of ownership. The equilibrium share price function must in this case equal the large shareholder's capitalized marginal instantaneous payoff, which increases in her shareholding. Selling shares too quickly would cause a drastic fall in share price. So it is optimal for her to sell gradually.4 Our paper is related to a strand of the literature which deals with the dynamic process of adjustment of shareholding based on the insight into the literature on the Coase conjecture. Unlike our model specification which places emphasis on the conflict between the manager and the large shareholder, Gomes (2000) assumes that the large shareholder is also the manager of the firm. In that model, the owner–manager is playing a share-selling game against the collection of small investors. The gains from trade arise because by selling her shares, the owner–manager can diversify idiosyncratic risks with investors. The investors perceive that the owner–manager may be of one type or another. Although the owner–manager knows her type, investors know only the probability distribution of types. At each period, the owner–manager moves by choosing her new fraction of equity ownership and her effort level which is unobservable. Investors update their belief about the owner–manager's type, and they price shares in the market accordingly. Gomes shows that when outside investors face this adverse selection problem, the owner–manager's equilibrium strategy involves divesting her shares gradually over time (in contrast to the perfect information benchmark, where the owner–manager would sell all her shares in the first period). This gradualism is necessary for the entrepreneur to develop a reputation for treating minority shareholders well. Gomes's conclusion that a risk-averse owner–manager would divest shares gradually over time is re-enforced by DeMarzo and Urošević (2006) who show (in a model with moral hazard instead of adverse selection) that if moral hazard is weak enough, the large shareholder trades immediately to the competitive price-taking allocation. With strong moral hazard, however, she will adjust her stake gradually. DeMarzo and Urošević (2006) emphasize the large shareholder's trade-off between risk diversification (which calls for a small shareholding) and her incentives and ability to improve the firm's performance (which increases with her fraction of ownership of her firm). DeMarzo and Urošević assume that the utility function of the large shareholder exhibits constant absolute risk aversion. Her wealth consists of a risk-free account and risky shares in her firm. Her sale strategy is motivated by consumption smoothing and risk diversification. When she sells her shares, investors anticipate a decrease in her effort. Hence, when reducing her stake, she is likely to generate a decrease in share price.5Edelstein et al. (2010) generalize the model of DeMarzo and Urošević (2006) to a setting with multiple strategic insiders. They show that the aggregate stake of the insiders decreases gradually over time, and that the long run equilibrium aggregate stake of the insiders are greater for firms with a larger number of insiders. In contrast, in our model, all agents (the large shareholder, the small investors, and the manager) are risk neutral: their utility function is linear in income. Moreover, we focus on the divergence of interests between the large shareholder and the manager (who does not own shares): in each period, there is an agency problem occurring between the large shareholder and the manager.6 The separation of management (by the manager) and control (by the large shareholder) is a major driving force behind the dynamics of share sales. By divesting, the large shareholder can influence the time path of the equilibrium effort level of the manager, as well as the time path of her own level of monitoring of his action. In our model, the wedge between her marginal valuation of a share and that of the atomistic investor indeed arises from a strategic effect, namely Nash equilibrium managerial effort (in the game between the manager and the large shareholder) being decreasing in the fraction of shares held by the large shareholder. This strategic effect is absent in DeMarzo and Urošević (2006). While DeMarzo and Urošević (2006) obtained the result that total divestment is the ultimate outcome if control benefits are small relative to risk aversion, the mechanisms driving the dynamic process in the two models are quite different. The paper is organized as follows. Section 2 describes the basic framework, drawn from Burkart et al. (1997). Section 3 deals with the commitment benchmark. Section 4 turns to time-consistent strategies and characterizes the Markov-perfect equilibrium corresponding to different regions of the parameter space. Section 5 discusses the main contributions of our model in relation to the literature on divestment. Section 6 concludes.

We have shown that a large shareholder divests her shares because, in the absence of share trading, there would exist a wedge between her marginal returns on holding these assets and the atomistic investors' valuation of a share. This wedge arises because the atomistic investors free ride on her monitoring effort which is aimed at reducing the manager's opportunistic behavior (such as choosing projects that are more advantageous to him than to the shareholders). As she divests, the manager increases his effort, but in general this is to the detriment of the firm's profit stream (we show that W′(α)>0W′(α)>0 for View the MathML sourceα<α2⁎). This evolution toward a pure managerial firm, in which the owners do not monitor the manager, can be gradual or immediate, depending on the degree of incongruence of the manager's interest to that of the owners. Our conclusions differ significantly from the original static model of Burkart et al. (1997). While BGP correctly pointed to the conflict of interest between the large shareholder and the manager, and offered an appealing formulation of that conflict, they abstracted from rational expectations of small investors which would arise in a dynamic setting. By fully allowing for rational expectations, we obtain some sharp results. Even in the case of full commitment, so that asset sales take place only once, the optimal commitment level of stock holding by the large shareholder is different from the level derived by BGP in a static model. In the non-commitment case, we show that complete divestment is the ultimate outcome. One may wonder whether this tendency for disperse ownership would disappear if the owners can propose incentive contracts to the manager. Burkart et al. (1997) have already answered this question by showing that there nevertheless remains in that case some scope for monitoring and a negative relationship between the manager's effort and the large shareholder's stake in the firm. Our model can be enriched by extension in several directions. First, one can allow the manager to play a dynamic game. If he realizes that the large shareholder may want to divest, and the speed of divestment depends on the expected net income flow at any level of the state variable αα, he may have strong incentives to adopt strategies that condition his effort level on the state variable. Second, the impacts of dynamic incentives on the welfare of the small investors could be investigated in a model where they are not simply indifferent between alternative forms of asset holding. These are parts of our future research agenda.