Weak Governance by Informed Active Shareholders

Figure 2 illustrates an example of how each component in Equation ( 10) varies with the active shareholder’s initial ownership, |$X_$|⁠ . In this example, the takeover policy has a negative expected return without the active shareholder’s influence on deal outcome, and thus we see that as |$X_$| grows, the total loss due to the intrinsic value of the takeover policy increases linearly, represented by the dotted line in the figure. Second, the active shareholder’s trading profits remain constant when his initial holding |$X_$| is below the threshold and they start declining linearly afterwards, represented by the dash-dash line. Third, when the active shareholder’s initial holding is below the threshold, he optimally sells all his shares following a low signal and hence his influence power on the final outcome of the deal is zero. This means that his benefits from voting are also zero in this range. As |$X_$| goes beyond the threshold, the benefit of voting increases quadratically with |$X_$|⁠ , which is represented by the dashed line.

Decomposition of expected gains to the active shareholder

This figure shows the decomposition of net benefits to the active shareholder if he promotes the takeover policy ex ante. The net benefits can be decomposed into three parts: the intrinsic value of M†A, the trading profits, and the benefits from influencing the deal outcome. The figure illustrates how the dollar value of each component varies with the active shareholder’s initial ownership |$X_$| assuming that the active shareholder follows his optimal trading and voting strategy.

The red, solid line shows the combined profits, or the total expected profits to the active shareholder if a takeover policy is implemented. The active shareholder will support a takeover policy if the total expected profits from trading and voting are positive and vote against the policy if they are negative.

One key implication of our model is that, the informed, active shareholder’s interest is not fully aligned with the interest of passive shareholders because of his ability to collect information and trade. 5 This ability endogenously changes the active shareholder’s preference for the ex ante takeover policy even when the takeover policy has a negative expected return to the firm.

We demonstrate this point more clearly by comparing the total expected gains to both types of shareholders when the firm implements a takeover policy. For ease of comparison, we create a hypothetical uninformed, passive shareholder who owns the same amount of initial shares as the active shareholder. However, we assume that this uninformed, passive shareholder does not trade or vote, and the active shareholder still follows his optimal trading and voting strategy as in the equilibrium. In this case, the expected gains to the passive uninformed shareholder are:

Note that |$\varDelta Y_$| in the equation above denote the |$\varDelta Y_$| defined for the active shareholder. Comparing Equation ( 12) with Equation ( 10), we see that the expected gains to the uninformed, passive shareholder capture both the intrinsic value of the takeover policy and the potential benefits from the active shareholder’s ex post influencing. His gains, however, do not contain any trading profits. The active shareholder’s trading profits, therefore, create a wedge between the preference of the informed, active shareholder and that of the uninformed, passive shareholder.

Figure 3 demonstrates an example of this wedge or disagreement between the two shareholders. In this figure, disagreement arises when the informed, active shareholder’s expected gains (plotted by the red line) have a different sign from the uninformed, passive shareholder’s expected gains (plotted by the blue line). In this example, we find that such disagreement occurs in two shaded areas.

Disagreement between the active shareholder and passive shareholders

This figure illustrates the expected profits to an informed, active shareholder (red line) and those to an uninformed, passive shareholder (blue line) when the firm implements a takeover policy. The shaded area represents the area in which the informed, active shareholder’s preference for the takeover policy conflicts with the uninformed, passive shareholder’s preference for the takeover policy.

In the first shaded area, where |$X_$| is relatively low, the red line is above zero, whereas the blue line is below zero, so the informed, active shareholder prefers to promote the takeover policy, whereas this policy is harmful to the uninformed, passive shareholder (and also to the firm). This disagreement occurs because, in this area, the informed, active shareholder earns positive trading profits from the takeover policy and such trading profits are unavailable to the uninformed, passive shareholder. The trading profits induce the active shareholder to support the takeover policy even if it destroys firm value and hurts the uninformed, passive shareholder. A simple numerical example helps demonstrate the intuition: assume that the takeover policy has an intrinsic return of -5%, and the active shareholder’s ex post influence on deal outcome increases the expected return by 3%. The active shareholder earns an expected trading profit of 4%. In this case, the total expected gain for the uninformed, passive shareholder is |$-$| 2% ( ⁠|$-$| 2% = |$-$| 5% |$+$| 3%), and the total expected gain for the informed, active shareholder is 2% (2% = |$-$| 5% |$+$| 3% |$+$| 4%). As a result, the informed, active shareholder prefers to implement the takeover policy that hurts the uninformed, passive shareholder.

In the second shaded area, where |$X_$| is relatively high, we observe the opposite situation. Here, the red line is below zero, whereas the blue line is above zero, so the informed, active shareholder opposes the takeover policy, whereas this policy creates value for the uninformed, passive shareholder. This disagreement occurs because of a free-riding problem. In our model, the active shareholder’s ex post influence is not free, even if we do not have an explicit cost of influence in the model. To gain sufficient influence power, the active shareholder has to retain enough shares after his trading, which requires him to sell less or even buy more shares in the face of a bad deal. This reduces his expected trading profits and may even turn the expected trading profits to negative when his initial ownership is high, as we show in Figure 2. The decreased expected trading profits, therefore, can be viewed as an endogenous cost of influencing. While only the active shareholder bears this cost of influencing, the benefits of his influencing are shared with the uninformed, passive shareholders, creating a free-riding problem: in some circumstances, the active shareholder does not want to implement the takeover policy because it generates a negative expected gain for him while the passive shareholder hopes to implement the policy and free-ride the active shareholder’s ex post influence effort. A simple numerical example helps demonstrate the intuition: assume that the takeover policy has an intrinsic return of -5%, and the active shareholder’s ex post influence increases the expected return by 6%. The active shareholder’s expected trading profit is -2%. In this case, the total expected gain for the uninformed, passive shareholder is 1% (1% = -5% + 6%), and the total expected gain for the informed, active shareholder is -1% (-1% = -5% + 6% - 2%). As a result, the informed, active shareholder, ex ante, opposes the takeover policy that benefits the uninformed, passive shareholder.

In sum, our model highlights two settings in which an informed, active shareholder and the uninformed, passive shareholder may endogenously disagree on whether to implement the takeover policy ex ante. First, when the takeover policy has a negative expected return, the active shareholder may promote the firm to implement this policy due to his expected trading profits. Second, when the takeover policy has a positive expected return with the active shareholder’s ex post influencing, he may oppose this policy ex ante because of the free-riding problem. In these situations, giving the active shareholder greater influence power may not maximize firm value. In what follows, we characterize how firm value varies with the active shareholder’s influence power per unit of ownership, |$\xi$|⁠ . We demonstrate that there exist settings in which empowering the informed, active shareholder can lead to weaker or even bad governance.

2.2 Firm value and influence power

In the model, the active shareholder’s influence power per unit of ownership, |$\xi$|⁠ , affects firm value through two channels. First, |$\xi$| affects the active shareholder’s chance of blocking a bad deal ex post. Through this channel, giving the active shareholder greater influence power (i.e., a larger |$\xi$|⁠ ) enhances firm value if he chooses to influence ex post. Second, |$\xi$| affects the active shareholder’s chance of determining the takeover policy ex ante. Through this channel, giving the active shareholder greater influence power may enhance or destroy firm value, depending on whether the active shareholder’s preference for the takeover policy conflicts with the passive shareholder’s preference (and thus firm value), as we discussed in detail above. The overall effect of the active shareholder’s influence power on firm value, therefore, is determined jointly by his ex ante preference for the takeover policy and his ex post decision on whether to retain shares and influence deal outcome (e.g., vote against a bad deal). To perform the analyses below, we consider four possible equilibrium regions. Note that for any given set of parameters, only one equilibrium region is realized as the model solution. Different equilibrium regions may emerge as the model solution for different parameter sets.

We label the four regions as S-V (support policy ex ante and vote ex post), S-NV (support policy ex ante and not vote ex post), NS-V (not support policy ex ante and vote ex post), and NS-NV (not support policy ex ante and not vote ex post), in which S (NS) means the active shareholder promotes (opposes) the takeover policy ex ante and V (NV) means the active shareholder retains some shares to vote (sells all shares and does not vote) against a bad deal ex post. Note that even if the active shareholder opposes the takeover policy ex ante (i.e., NS), he succeeds only with a certain probability. So the firm could possibly pursue a takeover with the active shareholder’s ex ante objection, and in this case, the active shareholder still needs to make decisions on his trading and voting ex post. NS-V and NS-NV characterize these situations.

Below we will focus on the setting in which the takeover policy has a negative intrinsic return, because this is the setting where the informed, active shareholder and the uninformed, passive shareholder may disagree on their preference for the takeover policy ex ante. Recall that this is the setting that gives rise to the novel model implications we wish to highlight in the paper. It is straightforward to show that, when the takeover policy has a positive intrinsic return, the two types of shareholders always agree with each other on their preference for the takeover policy and therefore giving the active shareholder greater influence power always improves corporate governance and enhances firm value. 6

Given the above, we can now make the following assumption for the remaining analyses below.

The takeover policy has a negative intrinsic return to the firm: |$\theta V_+(1-\theta)V_-V_

It is worth noting that Assumption 6 does not suggest that the expected return of the takeover policy is always negative. This is because the intrinsic return does not include the active shareholder’s influence on deal outcome ex post. It is possible that a takeover policy has a negative intrinsic return but a positive expected return, with the active shareholder’s influence ex post.

2.2.1 S-NV region

If the active shareholder has a small initial stake (i.e., |$X_<0>\leq\frac<2\lambda>(V_-V_)$|⁠ ) and his expected gains from the takeover policy is positive (i.e., |$E\left(\Pi_\right)>0$| in Equation ( 7)), then he uses his influence power to promote the takeover policy ex ante, and he sells all his shares or even takes a short position if he receives a low signal and hence have no influence power to block bad deals ex post. In this case, we get the S-NV equilibrium. In this equilibrium, we can write the ex ante value of the firm at |$t=0$| as

Equation ( 13) decomposes firm value into two components. The first component captures the firm’s value under the manager’s full control. Without any influence from the active shareholder, the manager implements the takeover policy with a probability |$z$| and this has the expected value of |$\theta V_+(1-\theta)V_$|⁠ , and with a probability |$1-z$|⁠ , she prefers that the firm not pursue takeovers and firm value remains at |$V_$|⁠ . The second component captures the value derived from the active shareholder’s influence on the takeover policy. The term |$(1-z)\xi X_$| represents the probability that the manager does not want to implement the takeover policy but the active shareholder successfully overrules her decision, and |$\theta V_+(1-\theta)V_-V_$| represents the expected return from the takeover policy. In this region, the active shareholder does not influence deal outcome ex post, so the expected return from the takeover policy is simply equal to the intrinsic return of the takeover policy, which is assumed to be negative in Assumption 6.

Proposition 1 summarizes how firm value varies with the active shareholder’s influence power per unit of ownership, |$\xi$|⁠ .

When |$X_\leq\frac<2\lambda>(V_-V_)$| and |$E\left(\Pi_\right)>0$| holds in Equation ( 7), the active shareholder promotes the takeover policy ex ante and sells all his shares and does not vote against a bad deal ex post. In this setting, giving the active shareholder greater influence power always reduces the ex ante value of the firm.

See Section A.4 in the Appendix. ■

Proposition 1 highlights the simplest setting in which a more influential active shareholder can weaken corporate governance and lower firm value. This happens because in this S-NV region, the active shareholder promotes a value-destroying takeover policy ex ante but does not use his influence power to block bad deals ex post. As a result, his influence power unambiguously destroys firm value. Given his preference for the value-destroying takeover policy ex ante, he does not correct a manager’s bad decision. In some circumstances, he even overturns a manager’s good decision, thereby aggravating the overinvestment problem.

2.2.2 NS-NV region

Equation ( 14) decomposes the value of the firm into two components. The first component captures the firm’s value under the manager’s full control, as we discussed above. The second component captures the value derived from the active shareholder’s influence on the takeover policy ex ante. Here |$z\xi X_$| represents the probability that the manager proposes to implement the takeover policy and the active shareholder successfully overturns her decision. |$V_-\theta V_-(1-\theta)V_$| represents the value created by the active shareholder overruling the manager’s bad decision ex ante.

Proposition 2 summarizes how firm value varies with the active shareholder’s influence power per unit of ownership, |$\xi$|⁠ .

See Section A.5 in the Appendix. ■

Proposition 2 highlights the simplest setting in which a more influential active shareholder can strengthen corporate governance and increase firm value. This happens because in this NS-NV region, the active shareholder opposes the firm’s decision to implement a value-destroying takeover policy, so his influence power can be used to overturn the manager’s bad decision ex ante (i.e., the overinvestment problem). The active shareholder, however, does not create more value for the firm ex post if the firm eventually pursues a takeover, because he sells all his shares if he observes a low signal, and, hence, he will have no influence power left to block bad deals ex post.

The above two equilibrium regions represent the settings in which the active shareholder finds it optimal not to use his influence power to vote ex post. Below we analyze the two regions in which he finds it optimal to retain some shares and vote ex post. Here we obtain more nuanced results on how influence power can affect firm value.

2.2.3 S-V region

We next turn to the region in which the active shareholder prefers promoting the takeover policy ex ante but also retain a positive ownership stake to vote ex post if he observes a bad deal (i.e., the S-V region). The region obtains when the active shareholder has a relatively large initial stake (i.e., |$X_>\frac<2\lambda>(V_-V_)$|⁠ ) and his expected gains from the takeover policy are positive (i.e., |$E\left(\Pi_\right)>0$| in Equation ( 8)).

Equation ( 15) decomposes firm value into three components. The first component captures the firm’s value under the manager’s full control, and the second component captures the value derived from the active shareholder’s influence on the takeover policy ex ante. Equations ( 13) and ( 14) touch on these two components. The last component captures the value derived from the active shareholder’s influence on the ex post outcome of the deal. Specifically, |$z+(1-z)\xi X_$| is the probability that the takeover policy is implemented, |$1-\theta$| is the probability that a bad deal is identified (the active shareholder’s ex post influence creates value only when the deal is bad), |$\xi(X_+X_)$| is the probability that the active shareholder is able to block the bad deal, and |$V_-V_$| is the gain from blocking a bad deal.

The second and third components in Equation ( 15) highlight the cost and benefit of giving the active shareholder greater power to influence corporate decisions, which is the main tension highlighted in our paper. On the one hand, giving the active shareholder more influence power increases the chance that the firm implements a takeover policy that could be value destroying on average (even with the active shareholder’s ex post influence); on the other hand, giving the active shareholder more influence power also increases the chance that a bad deal will be blocked ex post and thus increases the expected return to the takeover policy. The overall effect of the active shareholder’s influence power depends on which component dominates, which in turn depends on the level of the influence power |$\xi$|⁠ , as summarized next in Proposition 3.

When |$X_\geq\frac<2\lambda>(V_-V_)$| and |$E\left(\Pi_\right)>0$| holds in Equation ( 8), the active shareholder promotes the takeover policy ex ante and retains a positive ownership stake to vote against a bad deal ex post. In this setting, the relation between firm value and the active shareholder’s influence power can be nonmonotonic. Specifically, there exists a cutoff level |$\xi_$| such that firm value is decreasing in |$\xi$| for |$\xi\leq\xi_$| and increasing in |$\xi$| for |$\xi>\xi_$|⁠ .

See Section A.6 in the Appendix. ■

This result obtains because in this region, the benefit of the active shareholder’s influence power is quadratic in |$\xi$| while the cost is linear in it. Hence, when the level of |$\xi$| is low, the cost may overweigh the benefit, leading to a negative relation between firm value and influence power. However, when influence power is sufficiently large, the benefit always outweigh the cost, and, thus, the relation turns positive.

2.2.4 NS-V region

Last, we turn to the region in which the active shareholder opposes the takeover policy ex ante, but if the takeover policy is implemented, he retains a positive ownership stake and votes against bad deals ex post (i.e., the NS-V region). This region obtains when the active shareholder has a relatively large initial stake (i.e., |$X_>\frac<2\lambda>(V_-V_)$|⁠ ) and his expected gains from the takeover policy are negative (i.e., |$E\left(\Pi_\right)<0$| in Equation ( 8)).

Similar to Equation ( 15), Equation ( 16) decomposes firm value into three components. The first component captures the firm’s value under the manager’s full control, and the second component captures the value derived from the active shareholder’s influence on the takeover policy ex ante. The last component captures the expected value derived from the active shareholder’s influence on deal outcome ex post. Specifically, |$z(1-\xi X_)$| is the probability that the firm implements the takeover policy, |$1-\theta$| is the probability that a bad deal is identified, |$\xi(X_+X_)$| is the probability that the active shareholder is able to block the bad deal, and |$V_-V_$| is the gain from blocking a bad deal.

Comparing Equation ( 16) with Equation ( 14), we notice that firm value is always higher in this NS-V region than in the NS-NV region because the third component is always positive. But, interestingly, this does not suggest that firm value in this NS-V region always increases with |$\xi$| as it does in the NS-NV region. To see this, note that although the second component is always increasing in |$\xi$| as in the NS-NV region, the third component, which is unique to this NS-V region is nonmonotonic in |$\xi$|⁠ . This is because the expected value created by the active shareholder through his influence on deal outcome depends on two factors: (1) the probability that a takeover policy is implemented and thus he may need to influence ex post (i.e., |$z(1-\xi X_)$|⁠ ) and (2) the expected value he can create if he influences (i.e., |$(1-\theta)\xi(X_+X_)(V_-V_)$|⁠ ). We see that |$\xi$| affects these two factors in the opposite direction. On the one hand, giving the active shareholder greater influence power reduces the probability that the firm will pursue a takeover, because the active shareholder opposes the takeover policy ex ante in this region. This force reduces the probability that he creates value by influencing. On the other hand, giving him greater influence power increases his chance of blocking a bad deal ex post and thus increases the expected value he can create through his influence. When the first factor dominates, the third component can be decreasing in |$\xi$|⁠ .

Proposition 4 summarizes the relation between firm value and the active shareholder’s influence power in this region.

See Section A.7 in the Appendix. ■

Proposition 4 suggests that in this NS-V region, it is more likely that firm value increases with the active shareholder’s influence power, but the opposite can also happen for some parameter sets. In particular, when the active shareholders’ initial ownership is high and when he has high influence power, giving him even more influence power may hurt firm value. This intriguing situation happens when the active shareholder’s influence on deal outcome turns the expected return of the takeover policy positive, and yet the active shareholder still opposes the takeover policy ex ante. In this case, giving the active shareholder greater influence power reduces the firm’s probability of implementing a value-enhancing takeover policy and thus may hurt firm value. If the takeover policy has a positive expected return, then why does the active shareholder oppose it ex ante? This is because of the free-riding problem we analyzed at the end of Section 2.1.3.

It is worth noting that each of the above propositions only holds in its respective region. It is possible that changing |$\xi$| may also cause a switch of regions. In this case, the model prediction has to combine propositions in different regions. For example, Proposition 4 only holds in the NS-V region. That means, when one increases |$\xi$| beyond |$\xi_$|⁠ , firm value may initially decrease with |$\xi$|⁠ . But once |$\xi$| becomes large enough, the active shareholder’s ex ante preference for the takeover policy may switch from NS to S, and the region will switch from NS-V to S-V accordingly. And in this case, we showed in Proposition 3 that increasing |$\xi$| would eventually enhance firm value.

2.3 Model Extensions

In this section we present several extensions of the baseline model and investigate how these extensions affect our main results concerning the way in which the active shareholder’s influence power relates to firm value.

2.3.1 Strategic manager

In the baseline model, we have assumed that the manager has a preference for takeovers by modeling an exogenous parameter |$z$| which captures the probability that the manager supports the policy of engaging in takeovers ex ante. In this section, we consider an extension in which the manager’s preference for takeovers is endogenous (i.e., we endogenize the parameter |$z$|⁠ ) and study how the effect of the active shareholder’s influence power |$\xi$| on firm value changes.

First, because the active shareholder’s trading and governance decisions are made after the manager decides on pursuing a takeover strategy, the active shareholder’s optimal trading quantity and ex ante and ex post governance actions do not depend on the manager’s preference for takeovers. Thus, endogenizing the manager’s decision will not change the boundary of the four regions we identified in our baseline model (i.e., NS-NV, S-NV, NS-V, and S-V).

However, the manager’s preference for pursuing acquisitions may depend on the active shareholders ex post trading and governance decisions and hence may affect how firm value varies with the active shareholder’s influence power |$\xi$|⁠ .

To keep the analysis simple, we assume that the manager only cares about her private benefits of control which she derives only if a takeover deal is completed. Specifically, let |$B$| denote the utility derived by the manager from a completed takeover and assume that initially this variable is given by |$\tilde$|⁠ , which takes a value from a commonly known distribution with the support of |$[-L,+L]$|⁠ . 7 A positive value means that the manager is biased towards over investing in takeovers (i.e., empire-building motive), and a negative value implies that the manager has a bias against takeovers (i.e., seeks the quiet life). In addition, we assume that if the firm decides to pursue takeovers, then identifying an actual target requires the manager to spend a (utility) search cost of |$C$|⁠ .

Thus, the manager’s expected gains from promoting a takeover may depend on the active shareholder’s ex post action, because these actions affect the likelihood of deal completion.

Given the above, we analyze the manager’s optimal choice under two scenarios: (1) the equilibrium region is NS-NV or S-NV and hence the manager anticipates the informed active shareholder will sell all his shares in the face of a bad deal), and (2) the equilibrium region is NS-V or S-V, and, hence, the manager anticipates the informed active shareholder will maintain a positive equity stake and vote against bad deals. These two scenarios depend on whether or not |$X_\leq\frac<2\lambda>\left(V_-V_\right)$|⁠ , as analyzed in the baseline model.

Our main findings, formally proved in the Online Appendix, are as follows:

Scenario 1: In Lemma OA.1 we show that in the regions in which the informed active shareholder does not vote ex post, endogenizing managerial preference for takeovers does not generate a disciplinary effect, and hence the manager’s decision on whether or not to propose a takeover policy is not affected by |$\xi$|⁠ . For this reason, it is also the case that in these regions, the effect of |$\xi$| on firm value is the same as in the baseline model.

Scenario 2: In Lemma OA.2, we will show that in the regions in which the informed active shareholder votes ex post, increasing his influence power has an additional governance effect that decreases the ex ante likelihood that the manager promotes the takeover policy. This setting captures the intuition that due to the ex post ability of the active shareholder to vote down a bad takeover, the manager will be less likely to receive her private benefits, and hence will have a lower ex ante incentive to propose pursuing takeovers.

The remaining question then is how a change in influence power affects firm value in these regions (i.e., where |$X_\geq\frac<2\lambda>(V_-V_)$| ). Our formal analysis of how firm value varies with |$\xi$| in the S-V and NS-V regions demonstrates that it is still the case that firm value may decrease with |$\xi$|⁠ , but also that endogenizing the manager’s decision creates an additional disciplinary effect of |$\frac$|⁠ . This additional effect enhances the governance role of the active shareholder. Section A.1 of the Online Appendix formally shows this.

2.3.2 Effect of short-selling pressure

2.3.2.1 Optimal trading quantity

In our baseline model, we assume that the active shareholder cannot influence the probability of deal completion when he sells short. Previous studies have suggested that short selling by institutional investors may generate price pressure that sometimes diminishes the likelihood of deal completion. We explore this mechanism in this subsection and examine how it affects the model’s implications.

To do so, we add a function |$s(X)=min\$| to the model such that |$s(X)$| is deducted from the probability of deal completion. |$s(X)$| therefore reduces the likelihood of deal completion when |$X0$| captures the pressure of short selling. Function |$s(X)$| is analogous to function |$f(X)=max\$| which captures the effect of active shareholder’s influence on deal outcome via his ex post voting.

In this extension of the model, both positive and negative holdings by the active shareholder may affect the probability of completion of a bad deal. This extension therefore allows for a nonmonotonic relation between deal completion rate and active shareholder holding (i.e., an inverted V shape centered on zero). Note that the parameter |$s$| can be different from the parameter |$\xi$|⁠ , so we allow for asymmetric effects of positive holdings (through influencing) and negative holdings (through short-selling pressure) on deal outcomes.

where |$g(X)=f(X)$| if |$X>0$| and |$g(X)=s(X)$| if |$X\leq0$|⁠ . It is worth noting that |$f(X)$| and |$s(X)$| cannot be nonzero simultaneously.

The extended model can be solved similar to the baseline model. As we show more formally in Lemmas OA.3a and OA.3b of the Online Appendix, we have that when |$X_\geq\frac<2\lambda>\left(V_-V_\right)$| the model results remain as before. But when |$X_\leq\frac<2\lambda>\left(V_-V_\right)$| we show that the optimal trading is such that for |$S=L$| the active shareholder trades |$X_=-\frac<2\lambda>\left(V_-V_\right)+\triangle X_^$| and |$\triangle X_^\equiv\frac<(-s)(V_-V_)><\lambda>\frac<[X_-\frac<2\lambda>\left(V_-V_\right)]><[1-(-s)\frac<\lambda>]>$|⁠ .

Thus, when |$0\leq X_<2\lambda>\left(V_-V_\right)$| the active shareholder short sells shares if his signal is low , and given that his short-selling pressure results in a reduced probability of the bad deal being completed, the optimal trading quantity is different in the two models.

Finally, we further demonstrate in Corollaries OA.1 and OA.2 of the Online Appendix that |$\triangle X_^$| is positive and is decreasing in |$X_$|⁠ , increasing in |$s$|⁠ , and increasing in |$V_-V_$|⁠ . Hence, the ability to influence deal outcome via short selling induces the active shareholder to sell less aggressively if he receives a low signal, and that in turn affects the probability of deal completion.

2.3.2.2 Ex ante preference for the takeover policy

Given the active shareholder’s optimal trading quantity and his influence over deal outcome (through short-selling pressure or explicit voting) derived above, we can compute his expected gains from a takeover policy. The expected gains determine whether or not the active shareholder supports the firm’s decision to pursue takeovers ex ante in this extended model.

What we find from this extension is as follows. Let |$E\left(\Pi_^\right)$| denote the active shareholder’s expected gains in the extended model and |$E\left(\Pi_^\right)$| denote his expected gains in the baseline model. Given the above we find that when |$0\leq X_<2\lambda>\left(V_-V_\right)$|⁠ , then |$E\left(\Pi_^\right)\left(V_-V_\right)$|⁠ , then |$E\left(\Pi_^\right)=E\left(\Pi_^\right)$|⁠ . Lemmas OA.4 and OA.5 of the Online Appendix provide a formal discussion and proof.

We, therefore, conclude that in the region of |$0\leq X_<2\lambda>\left(V_-V_\right)$|⁠ , his preference for the takeover policy is reduced in the extended model compared with that in the baseline model, because |$E\left(\Pi_^\right)

2.3.2.3 Firm value and influence power

Finally, an analysis of how firm value varies with the active shareholder’s influence power in the presence of short-selling pressure (Proposition OA.1.) shows that nothing changes for |$X_\geq\frac<2\lambda>(V_-V_)$|⁠ . However, when |$0\leq X_<\frac<2\lambda>(V_-V_)$| (Proposition OA.2.), we find three regions based on the parameter |$s$|⁠ .

In particular, we show that if |$E\left(\Pi_^\right)\mid_>0$| (i.e., our baseline assumption) then there exists two cutoffs for the marginal effect of short selling |$s_^$| and |$s_^$| such that firm value decreases with |$\xi$| for |$s$|⁠ , firm value increases with |$\xi$| for |$min\$|⁠ , and firm value decreases with |$\xi$| for |$s>max\$|⁠ .

This means that when short-selling pressure has a relatively small effect on the probability of deal completion (i.e., |$s^,s_^\>$|⁠ ), then the active shareholder will still promote the takeover policy ex ante and an increase in his influence power will still lead to a reduction in firm value, as in the baseline model. This result is well expected, because the extended model’s implications are continuous with respect to the parameter |$s$|⁠ , and it nests the baseline model when |$s=0$|⁠ . Once the short-selling effect, |$s$|⁠ , is in the intermediate range ( |$min\^,s_^\>$| ), the model generates two possible scenarios, both of which lead to a positive relation between the active shareholder’s influence power |$\xi$| and firm value.

When the effect of short selling is very large ( |$s>max\^,s_^\>$| ), we find that firm value decreases with the active shareholder’s influence power again. The reason for this third region is similar to the result in Proposition 4 in which firm value was decreasing in |$\xi$| for |$\xi>\xi_$|⁠ . The intuition is that, when |$s$| is very large, the active shareholder’s expected gains from the takeover policy may become negative, 8 and, thus, he would oppose the takeover policy ex ante. The problem is that following a bad signal he will still short sell, and his short-selling pressure will reduce the likelihood of the bad deal being completed. This will turn the expected value of the takeover policy to be positive for dispersed shareholders, but will also imply a negative trading profit for the active shareholder. As a result, he opposes it ex ante. In this case, giving the active shareholder more influence power merely reduces the likelihood that the value-enhancing takeover policy be implemented. His influence power plays no role ex post (because he short sells and thus does not vote ex post), so the combined effect is therefore negative.

2.3.3 The probability of being informed

The baseline model suggests that the most misaligned active shareholder is one with a zero initial equity stake (e.g., a hedge fund). This results is due to the simplifying assumption that the active shareholder is always fully informed. In this extension of the model we consider the posibility that the probability of becoming informed is a function of the initial position and demonstrate that a larger ownership stake (e.g., a mutual fund) can increase the level of misalignment.

In particular, we assume that the active shareholder becomes informed with a probability |$\psi(X_)$| and remains uninformed with a complement probability |$1-\psi(X_)$|⁠ . The function |$\psi(X_)$| is assumed to be weakly increasing in |$X_$| and this captures the intuition that a larger equity stake increases the informational advantage of the active shareholder. To simplify the analysis, we further assume that all agents in the model know whether or not the active shareholder becomes informed and that he trades and/or influences only when he is informed.

The focus of our analysis is a comparative static of an active shareholder with a zero stake to one with a positive stake, we consider the parameter region in which |$0\leq X_<2\lambda>\left(V_-V_\right)$|⁠ .

Given the active shareholder’s expected gains from takeovers when informed and uninformed we can recompute the active shareholder’s expected gain from takeovers in the extended model as

The second term is his expected gains from becoming informed and trading. The first term is decreasing in |$X_<0>$|⁠ , and the second term is increasing in |$X_<0>$|⁠ , so adding |$\psi(X_<0>)$| into the baseline model generates richer interpretations of the effects of initial ownership |$X_<0>$| on the active shareholder’s ex ante preference for the takeover policy. More specifically, the active shareholder now pursues takeovers if |$E\left(\Pi_\right)>0$|⁠ , that is,

\psi(X_)>\frac<\left[V_-\left(\theta V_+(1-\theta)V_\right)\right]><4\lambda>(V_-V_)^>X_.\label \end$$ Now define the right-hand side of Equation ( 17) as

Our model predicts that the distortion to the active shareholder’s ex ante governance decision is determined by the value of |$\psi(X_)-h(X_)$|⁠ . More specifically, if |$\psi(X_)-h(X_)>0$|⁠ , the active shareholder’s ex ante governance decision is distorted; otherwise, his ex ante governance decision is aligned with dispersed shareholders. Moreover, when |$\psi(X_)$| is differentiable, the derivative of |$\psi(X_)-h(X_)$| with respect to |$X_$| (i.e., |$\frac-c$|⁠ ) determines how this distortion changes with |$X_$|⁠ .

In Lemma OA.6, we prove that the distortion to the active shareholder’s ex ante governance decision will be increasing in |$X_$| whenever |$\frac>c$|⁠ , where |$c=\frac<\left[V_-\left(\theta V_+(1-\theta)V_\right)\right]><4\lambda>(V_-V_)^>$|⁠ . Thus, the model applies to mutual funds who hold positive equity stakes.

3. Conclusion

Previous studies have documented that informed, active shareholders may improve corporate governance through active monitoring, passive trading, or both. In this paper, we present a model of corporate takeovers in which a manager may sometimes propose a takeover policy that harms shareholders. In this setting, we demonstrate that an informed, active shareholder’s joint ability to influence corporate decisions and trade on his private information, may sometimes weaken corporate governance. We further show that giving this shareholder greater power to influence corporate decisions, even if it is ex post beneficial to all shareholders, may lower the ex ante value of the firm.

Overall, our paper highlights one important reason informed, active shareholders, such as institutional investors, may make distorted governance decisions that can be harmful to the firm. Our paper adds to the corporate governance debate on whether shareholder empowerment necessarily leads to better governance and highlights settings in which the answer to this is negative.

Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web site next to the link to the final published paper online.

Acknowledgement

We are very grateful to the editor Francesca Cornelli and two anonymous referees for their helpful comments and suggestions. We thank Eliezer Fich, Jarrad Harford, Lixin Huang, Doron Levit, and Andrey Malenko; conference participants at the 2016 Cass Mergers and Acquisitions Research Centre Conference, Northern Finance Association Meeting; and seminar participants at Cheung Kong Graduate School of Business, Indiana University, and University of South Florida for very helpful comments. Supplementary data can be found on The Review of Financial Studies web site.

A. Model Proofs

A.1 Proof of Lemmas 2a and 2b

The first-order condition (F.O.C.) of Equation ( 3) characterizes the optimal trading strategy. We can express the F.O.C. as

$$\begin f(X+X_)V_+(1-f(X+X_))V_-P_(X)+\frac(V_-V_)(X+X_)-\lambda X=0.\label \end$$

If |$S=H$|⁠ , the active shareholder supports the takeover ex post, and hence |$V_=V_=V_$|⁠ . After writing the market maker’s price function as in Equation ( 1), we get the F.O.C. as

$$\begin V_-P_-2\lambda X_+\lambda E(X)=0\label \end$$ and hence that $$\begin X_=\frac<2\lambda>\left(V_-P_\right)+\fracE(X).\label \end$$

If |$S=L$|⁠ , the active shareholder votes against the bad acquisition and tries to block the deal if he can. In this case, |$V_=V_$| and |$V_=V_$|⁠ , and the F.O.C. depends on whether or not |$X_+X_>0$| . Lemmas 2a and 2b present the results based on this key assumption.

Note that the solution to |$X_$|⁠ , presented in Equation ( A.3), applies to both Lemma 2a and 2b, but the solution to |$X_$| differs in these two lemmas, which we prove below separately.

To prove Lemma 2a, we notice that when |$X_<0>+X_\leq0$|⁠ , the active shareholder will sell all his shares or even take a short position if |$S=L$|⁠ , and therefore he has no influence power left after his trades, and thus |$f(X_+X_<0>)=0$| and |$\frac)>=0$|⁠ . The F.O.C., when |$S=L$|⁠ , simplifies to

$$\begin V_-P_-2\lambda X_+\lambda E(X)=0\label \end$$ $$\begin X_=\frac<2\lambda>\left(V_-P_\right)+\fracE(X).\label \end$$

The active shareholder has no power to influence deal outcome ex post when |$S=L$| and the market maker correctly anticipates this in equilibrium, so we have that

$$\begin P_=\left[\theta V_+(1-\theta)V_\right].\label \end$$ Substituting Equations ( A.3), ( A.5) and ( A.6) into the definition of |$E(X)$|⁠ , we get $$\begin E(X)=\theta X_+(1-\theta)X_=\frac<2\lambda>[\theta V_+(1-\theta)V_-P_+\lambda E(X)].\label \end$$

Thus, we get that |$E(X)=\fracE(X)$|⁠ , and the only fixed point solution is that |$E(X)=0.$|

Finally, substituting Equations ( A.6) and ( A.7) back to Equations ( A.3) and ( A.5), we get $$\begin X_ & =\frac<2\lambda>\left(V_-V_\right),\\ X_ & =-\frac<2\lambda>\left(V_-V_\right)\!. \end$$

Lastly, we verify that the condition |$X_+X_\leq0$| holds if and only if |$X_\leq\frac<2\lambda>(V_-V_)$|⁠ . Lemma 2a follows.

To prove Lemma 2b, we note that when |$X_<0>+X_>0$|⁠ , the active shareholder would have remaining influence power after his trades when |$S=L$|⁠ , and thus |$f(X_+X_<0>)=\xi(X_+X_<0>)$| and |$\frac)>=\xi$|⁠ . The F.O.C., when |$S=L$|⁠ , becomes

$$\begin \xi(X_+X_)V_+(1-\xi(X_+X_))V_-P_(X_)+\xi(V_-V_)(X_+X_)-\lambda X_=0.\label \end$$

Again, note that when |$S=L$|⁠ , the active shareholder would vote against the bad deal ex post, so |$V_=V_$| and |$V_=V_$| in Equation A.8, and we rewrite the F.O.C. as