Majority support of shareholders, monitoring incentive, and dividend policy

Majority support of shareholders, monitoring incentive, and dividend policy

Journal of Corporate Finance 30 (2015) 1–10 Contents lists available at ScienceDirect Journal of Corporate Finance journal homepage: www.elsevier.co...

401KB Sizes 0 Downloads 41 Views

Journal of Corporate Finance 30 (2015) 1–10

Contents lists available at ScienceDirect

Journal of Corporate Finance journal homepage: www.elsevier.com/locate/jcorpfin

Majority support of shareholders, monitoring incentive, and dividend policy Naoya Mori a,⁎, Naoshi Ikeda b,1 a b

Nihon University, College of Commerce, 5-2-1 #429 Kinuta, Setagaya-ku, Tokyo 157-8570, Japan Tokyo Institute of Technology, Department of Industrial Engineering and Management, 2-12-1 W9-51 Ookayama, Meguro-ku, Tokyo 152-8552, Japan

a r t i c l e

i n f o

Article history: Received 5 January 2014 Received in revised form 12 October 2014 Accepted 13 October 2014 Available online 28 October 2014 JEL classification: G35 G32 G34

a b s t r a c t As an alternative version of the side-payment model, this paper presents a demonstration of how the necessity of winning majority support of shareholders influences the relation between a blockholder's monitoring incentive and a firm's dividend policy. When dividend-averse individuals collectively hold a majority stake in a dispersed ownership structure, a dividend-seeking blockholder might be compelled to propose lower dividends than the tax-optimum to dominate the zero-dividend proposal. Under such circumstances, the blockholder has an incentive to provide unprofitable monitoring activity as long as the private benefits of tax-saving are greater than the pecuniary loss from the monitoring activity. © 2014 Elsevier B.V. All rights reserved.

Keywords: Ownership structure Corporate governance Tax preferences

1. Introduction The existence of a blockholder can be an effective corporate governance mechanism for resolving free-rider problems among shareholders. Grossman and Hart (1980) argue that a small investor has insufficient incentive to monitor the firm managers' discretion because collecting monitoring costs on personal accounts is usually impossible. Therefore, investors with small shareholdings have a strong incentive to free-ride the benefits of other shareholders' monitoring activities in a dispersed ownership structure. In contrast, holders of sufficiently large stakes find it profitable to discipline managers to enhance the firm's value.2 Individual investors with small shareholdings and corporate and institutional investors with large shareholdings are mutually complementary because these two players are expected to have different capabilities. Individuals as a group typically have a majority stake in a dispersed ownership structure. For that reason, they can force the firm to distribute favored amounts of dividends. Furthermore, they can remove managers for lack of concern. Nevertheless, they have insufficient capability of evaluating managers' performance properly. In contrast, corporations and institutions have excellent monitoring capability and hold sufficiently large stakes to collect monitoring costs on personal accounts. Nevertheless, it is impossible for them to affect the payout policy without depending on individuals' voting power in a dispersed ownership structure. ⁎ Corresponding author. Tel.: +81 3 3749 6833; fax: +81 3 3417 1681. E-mail addresses: [email protected] (N. Mori), [email protected] (N. Ikeda). 1 Tel.: +81 3 5734 3606. 2 According to Shleifer and Vishny (1997), a minority shareholder whose stake in a firm is 10–20% has sufficient incentive to gather information and to act as a monitor of management. Admati et al. (1994) and Maug (1998) argue that investors with large shareholdings tend to engage in monitoring activities in spite of the fact that investors with small shareholdings enjoy the benefits as free-riders.

http://dx.doi.org/10.1016/j.jcorpfin.2014.10.015 0929-1199/© 2014 Elsevier B.V. All rights reserved.

2

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

In this line of argument, two important models are useful to examine the relation between dividend policy and monitoring incentives of corporate and institutional investors. Shleifer and Vishny (1986) hypothesize that small individual investors compel the firm to distribute dividends as inducements to large corporate investors to become owners of the firm as monitors. Similarly, Allen et al. (2000) present a model including the assumption that institutional investors who are attracted to dividend-paying stocks mitigate agency conflicts between shareholders and managers. Within these models, based on corporate/institutional investors' tax preferences, positive dividends serve as a side payment from individuals who delegate monitoring tasks to the blockholder. Benefits from monitoring activities are distributed among all shareholders. However, these models incorporate the assumption that individual investors are homogeneous. Shleifer and Vishny (1986) demonstrate that positive dividend policy is acceptable as long as the individuals' tax burden, considered collectively, is less than the added value attributable to a blockholder's monitoring service. Allen et al. (2000) assume two groups of investors (i.e., individuals and institutions), and consider dividend policy from the perspective of two representative investors. In each group, all investors are identical. In reality, individual investors are heterogeneous in the sense that they have different marginal tax rates, although their taxoptimal dividends are unanimously zero. Because of the difference of tax status, as our model reveals, the dividend level at which an individual investor switches her approval from the zero-dividend proposal to the monitoring-based dividend proposal is not identical to the dividend level at which another individual investor switches. Given a proposal with a certain level of dividend payment, some individuals support this, but other individuals do not. To control dividend policy, a dividend-seeking blockholder must win the majority support of shareholders. Earlier models have not examined the effects of internal conflicts among individual investors on the size of monitoring activity provided to a firm. Shleifer and Vishny (1986) regard the added value as exogenous, and do not derive the optimization condition of a blockholder's monitoring service. Although Allen et al. (2000) demonstrate the optimum, they do not consider the influence that the necessity of winning the majority support of shareholders exercises on it. Eckbo and Verma (1994) show that shareholders with large holdings use voting power on dividend policy in an attempt to impose costs on other shareholders. The level of dividend payments varies with the relative voting power of different shareholders (corporations/institutions versus manager–owners in their analysis) who have opposite preferences (i.e., higher dividends versus lower dividends). Nevertheless, their concern is not the blockholder's monitoring incentive. This paper presents an alternative version of the side-payment model of dividend policy. Our concern is to examine how the necessity of obtaining majority support of shareholders influences the relation between a blockholder's monitoring incentive and the level of dividend payment. The model development presented in this paper reveals that the optimal level of monitoring activity depends not only on the blockholder's monitoring efficiency, but also on the tax status of the median shareholder, who casts the deciding vote on the blockholder's dividend proposal. First, the model presented herein formulates that the winning proposal in a majority-rule voting contest determines the level of dividend payment. Assuming conditions of dispersed ownership, a dividend-seeking blockholder competes against the zerodividend proposal for tax-savings benefits. To increase the number of votes they receive from dividend-averse individual investors, the blockholder might be compelled to propose lower dividends than the tax-optimum level. Such a concession engenders lower dividend payments under the assumption that individual investors as a group have a majority stake in the firm. Second, we examine the blockholder's incentive compatibility condition. It makes no sense for a blockholder to compromise on the dividend level if accepting the zero-dividend proposal becomes the better alternative. On such an occasion, the blockholder loses an incentive to offer monitoring services for the benefit of all shareholders. An inefficient blockholder gives up controlling dividend policy, but an efficient blockholder dominates the zero-dividend proposal. Third, we derive the maximization condition of a blockholder's after-tax return from monitoring activity and dividend policy. The analysis demonstrates that monitoring activity not only adds market value to the firm; it also alters individual investors' voting behavior, and thereby influences a blockholder's tax burden. The dividend-seeking blockholder has an incentive to provide unprofitable monitoring activity if the private benefits of tax-saving are greater than the pecuniary loss from the monitoring activity. Therefore, the optimal degree of monitoring activity becomes greater. The remainder of this paper is organized as follows. Section 2 presents the model structure. Section 3 describes development of the model. Section 4 shows some theoretical and empirical implications of our model for future studies. Finally, Section 5 concludes this paper with a summary of these arguments. 2. Model structure We assume that investors hold all types of stocks to benefit fully from risk diversification. Generally speaking, a portfolio with fewer assets implies a higher-risk portfolio. The complete exclusion of sub-optimal dividend-paying stocks therefore engenders ineffective diversification. Risk-averse investors would be expected to assign priority to risk-sharing rather than single-mindedly pursuing tax-saving opportunities. This tendency is expected to be the central cause of insufficient clientele effects of dividends.3 By assumption, heterogeneous shareholders, who are expected to have different tax preferences, are expected to hold shares in a firm. Therefore, internal disagreement would emerge around the issue of the dividend payment level. In such a situation, investors might pursue minimization of tax burden to the greatest degree possible. Nevertheless, preventing the waste of resources attributable to inefficient corporate governance might pose a more severe problem than tax saving. Our concern is to consider the effects of 3 For example, Richardson et al. (1986) investigate trading volume around the announcement date of changes in dividend policy. They conclude that shareholder clientele adjustments are small. Eckbo and Verma (1994) present evidence of stable shareholder clientele over time using data of Canadian firms.

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

3

internal conflicts between individual investors on the blockholder's monitoring incentive and the level of dividend payments. To concentrate on this issue, we are not concerned with the effect of uncertainty. In addition, debt financing is assumed away. Investor i's tax burden fluctuates in response to the dividend payment change. Under many prevailing tax codes, marginal tax rates on dividends tiD are higher for individual investors, but they are lower for corporate investors. Conversely, marginal tax rates on capital gains tiG are usually higher for corporate investors, but are lower for individual investors. Based on static tax-clientele framework (e.g. Allen and Michaely, 2003), individual investors (i.e. tDi N tGi) prefer lower dividends whereas corporate investors (i.e. tDi b tGi) prefer higher dividends.4 Taxed institutions belong either to the category of individuals or to the category of corporations.5 As Shleifer and Vishny (1986) have done, this study analyzes a set of circumstances in which only one blockholder can undertake a monitoring function in a firm. Concretely, this model incorporates a dividend-seeking blockholder x (i.e. tDx b tGx) competing against the zero-dividend proposal in a majority-rule voting contest. The case of a dividend-averse blockholder (i.e. tDx N tGx) or a tax-exempt blockholder's (i.e. tDx = tGx = 0) undertaking is discussed later in Section 4. Within the model presented herein, individual investors collectively have a majority stake in a dispersed ownership structure by assumption, which implies that a single corporate or institutional investor cannot control the firm's dividend policy without asking for individual investors' approval. Regarding tax-savings, dividend-averse individuals (i.e. tDi N tGi) unanimously support the zero-dividend policy. In this context, dividend-seeking blockholder x (i.e. tDx b tGx) is burdened with higher taxes. It is rational for blockholder x to offer monitoring services for the benefit of all shareholders, provided that individuals compromise on the level of dividend payment. Even if individuals as a group command a majority vote, they might find it reasonable to tailor positive dividend payments to satisfy blockholder x's preferences instead of their own tax-minimizing zero-dividend policy. This setting is identical to that of the central ideas presented by Shleifer and Vishny (1986) and by Allen et al. (2000). The model developed in this paper comprises three decisions. First, all shareholders determine the firm's dividend policy in a majority-rule voting contest. Second, a blockholder chooses the level of monitoring activity in a firm. Third, the blockholder decides whether it is favorable to undertake a monitoring function or not. To determine the optimal level of monitoring activity, the dividend level at which the blockholder wins a voting contest is necessary information. Similarly, the dividend level at which the blockholder gives up undertaking a monitoring function is necessary. Therefore, our analysis proceeds first by consideration of a majority-rule voting contest. The algebraic symbols used for this study are presented below. G Mx α v(Mx) γ Dx θ tD tG R

accrued capital gain in a firm with no monitoring activities, G N 0 level of monitoring activity in the firm provided by blockholder x, Mx = 0 or Mx N 1; blockholder x's efficiency in monitoring activity, 0 b α b 1 / 2 market value added to the firm by blockholder x, v(Mx) = (1 / α)Mαx marginal cost of monitoring activity, γ N 0; monitoring-based dividend proposal by blockholder x in the firm, 0 ≤ Dx ≤ G + v(Mx) percentage of investor's shareholdings in the firm, 0 b θ b 1 marginal tax rate on dividend income, 0 b tD b 1 marginal tax rate on accrued capital gain, 0 b tG b 1 after-tax return, R N 0

3. Model In Section 3.1, we formulate that a dividend-seeking blockholder competes against the zero-dividend proposal in a majority-rule voting contest. In Section 3.2, we consider the blockholder's incentive compatibility condition, with a demonstration that the firm distributes positive dividends in general if the dividend-seeking blockholder takes a monitoring position. In Section 3.3, we assess the process by which the optimal level of monitoring activity is derived.

3.1. Voting contest The model presented herein formulates that the winning proposal in a majority-rule voting contest determines the level of the dividend payment. Each investor chooses between the monitoring-based dividend proposal D = Dx by a blockholder x and the zero-dividend proposal D = 0 as the initial consensus of dividend-averse individual investors. Similarly to that described by Allen et al. (2000), the Mx unit of the monitoring activity provided by blockholder x adds market value of v(Mx) = (1 / α)Mαx to the firm, where 0 b α b 1 / 2 (i.e., v′(Mx) N 0, v″(Mx) b 0). The optimization problem of monitoring service 4 Within an intertemporal setting, Mori (2010) demonstrates that corporate investors prefer consumption-optimal dividends that are consistent with their liquidity needs because excessive payment causes intertemporal double taxation on reinvested dividend income. Based on this view, corporate investors do not necessarily prefer higher dividends. Nevertheless, it is not our purpose to argue this point here. Therefore, we assume that corporate investors simply prefer higher dividends based on a traditional tax-preference view. 5 Institutional investors do not face the same tax status. See, for example, Desai and Jin (2011) for several methods of categorization. According to them, about 40% of institutions were dividend-averse during 1981–1997.

4

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

is examined later in Section 3.3. For the moment, the level of monitoring activity Mx is regarded as an exogenously assigned parameter. When the firm adopts a zero-dividend policy (i.e. D = 0), the after-tax return of individual investor i (i.e. tiD N tiG) in this case is R0i, as expressed in Eq. (1). By assumption, the zero-dividend proposal has no monitoring activity (i.e. v(0) = 0).   i i i R0 ¼ 1−t G θ G

ð1Þ

In contrast, when the firm adopts a monitoring-based dividend policy (i.e. D = Dx), individual investor i's (i.e. tiD N tiG) after-tax return is Rim, as shown in Eq. (2).6 The added value (1 / α)Mαx attributable to a blockholder x's monitoring service is a shared benefit for all shareholders. In addition, individual investor i's extra tax burden is (tiD − tiG)θiDx.         1 i i i i i i α Mx − t D −t G θ Dx Rm ¼ 1−t G θ G þ α

ð2Þ

Shareholders, each of whom is assumed to have one vote for each share held, support a policy that proposes a higher after-tax return. It is noteworthy that individual investors are heterogeneous because of different marginal tax rates. If Rim N R0i, then individual investor i votes for a monitoring-based dividend proposal (i.e. D = Dx). In contrast, if Rim b R0i, then individual investor i votes for the zero-dividend proposal (i.e. D = 0). Although the better course of winning for blockholder x is to receive D = G + (1 / α)Mαx , such a favorable plan is not usually equivalent to a policy that gains a majority vote. As a concession, lowering the level of dividend proposal increases the number of individual investors i (i.e. tiD N tiG), whose status switches to Rim N R0i. When arranging all shareholders according to the level at which they switch their votes, the shareholder in the middle casts the deciding vote. That is to say, blockholder x's victory hinges upon the median shareholder's vote. Individual investors in these analyses have a majority stake as a group. Therefore, the median shareholder is a dividend-averse individual investor. The condition of the median shareholder's (i.e. i = μ) support for blockholder x's dividend proposal (i.e. D = Dx) is expressed in Eq. (3) as follows. Given the monitoring value (1 / α)Mαx , Dμx is defined as the dividend level at which blockholder x seizes the median shareholder μ's vote from the zero-dividend proposal, as Fig. 1 shows. It depends on the magnitude of blockholder x's monitoring activity and the median shareholder μ's tax status.  μ

μ

Rm ≥R0 ⇔Dx ≤

  1 Mx α  α μ ≡ Dx μ μ t D −t G μ

1−t G

ð3Þ

Given monitoring value (1 / α)Mαx attributable to a blockholder x's monitoring service, the after-tax return Rμm of median shareholder μ (i.e. t μD N tGμ ) is negatively related to blockholder x's dividend proposal Dx. If blockholder x proposes Dμx, then median shareholder μ switches her vote to the monitoring-based dividend proposal (i.e. D = Dx) from the zero-dividend proposal (i.e. D = 0). Consequently, the monitoring-based dividend proposal Dw x which can just dominate the zero-dividend proposal is expressed in Eq. (4) as follows: the winning point. For Dxμ b G + (1 / α)Mαx , if blockholder x proposes Dxμ, then the median shareholder μ switches μ μ α her vote to the monitoring-based proposal (i.e. Dw x = Dx ). In contrast, if Dx ≥ G + (1 / α)Mx , then blockholder x wins the voting contest α by proposing x's tax-optimal dividend with no concession [i.e. Dw x = G + (1 / α)Mx ].     1 w μ α Mx Dx ¼ min Dx ; G þ α

ð4Þ

Proposition 1. To win a voting contest against the zero-dividend proposal (i.e. D = 0), dividend-seeking blockholder x (i.e. tDx b tGx) must μ μ propose the monitoring-based dividend proposal (i.e. D = Dw x ) by which blockholder x seizes median shareholder μ's (i.e. t D N t G) vote from the zero-dividend proposal. This analysis is an application of the famous two-party model of public choice. According to the median voter theorem in political competition (e.g., Black, 1948 and Downs, 1957), a dominant position exists at the median in majority-rule voting on a single issue if all voters' preferences are single-peaked. In general, the possibility exists that multiple-peaked preferences engender impossibility in identifying a dominant position, which is the voting paradox.7 The voting paradox does not arise in the model presented herein because all shareholders' preferences are single-peaked. The additional value coming from monitoring services is constant by assumption. In addition, this is majority-rule voting on a single issue. Therefore, it can be modeled as a simple one-dimensional choice. 6 When a firm distributes positive dividends Dx as a whole, investor i's dividend income is (1 − tiD)θiDx on an after-tax basis, whereas ex-dividend capital gains are (1 − tiG)θi{G + (1 / α)Mαx − Dx} on an after-tax basis. Rearranging these two components, we obtain Eq. (2). 7 See, for example, Hirshleifer et al. (2005) for a brief discussion of the voting paradox.

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

5



1−

1

0

0 Fig. 1. Dividend level at which blockholder x seizes median shareholder μ's vote from the zero-dividend proposal.

3.2. Incentive compatibility We are now ready to consider blockholder x's (i.e. tDx b tGx) incentive compatibility condition. Of course, it makes no sense for blockholder x to compromise on the dividend level if it becomes more favorable to accept the zero-dividend proposal D = 0. Under such circumstances, blockholder x loses incentive to offer monitoring services for the benefit of all shareholders. If blockholder x accepts the zero-dividend proposal (i.e. D = 0), then blockholder x's after-tax return R0x is expressed in Eq. (5) as follows. By assumption, the zero-dividend proposal is with no monitoring activity (i.e. Mx = 0). Although blockholder x usually bears a heavier excess tax burden, the monitoring cost is reduced.  x x x R0 ¼ 1−t G θ G

ð5Þ

However, if blockholder x undertakes a monitoring function (i.e. D = Dx), then blockholder x's after-tax return Rxm is expressed in Eq. (6) as follows. The monitoring activity provided by blockholder x adds market value of (1 / α)Mαx . The added value is a shared benefit for all shareholders. In contrast, the monitoring cost γMx is the expenditure on personal account only for blockholder x. The tax burden is eased by virtue of the blockholder's control over dividend policy.      x 1 x x x x x α M x −γMx þ t G −t D θ Dx Rm ¼ 1−t G θ G þ α

ð6Þ

Therefore, if R xm ≥ R0 x , then blockholder x wishes to undertake a monitoring function. Conversely, if Rxm b R0 x , then blockholder x loses incentive to control dividend policy. The incentive compatibility condition of blockholder x is expressed in Eq. (7) as follows. Here, Dgx is defined as the dividend level proposed by blockholder x when blockholder x gives up undertaking a monitoring function, which demonstrates that the giving-up point depends on monitoring activity and blockholder x's tax status.    1 γM x − 1−t xG θx Mx α  α x x g x ≡ Dx Rm ≥R0 ⇔Dx ≥ t G −t xD θx

ð7Þ

The current analysis suggests that seemingly unprofitable monitoring activities also make sense as long as the private benefits of tax-saving are greater than the pecuniary loss from the monitoring activity. As Fig. 2 portrays, if γMx ≥ (1 − txG)θx(1 / α)Mαx , then Rxm ≷ R0x. Therefore, blockholder x might take a monitoring position even if the monitoring cost is higher than the added value on personal account. In contrast, if γMx b (1 − txG)θx(1 / α)Mαx , then Rxm N R0x. Such an efficient blockholder x always undertakes the monitoring activity, and necessarily dominates the zero-dividend proposal. In this case, monitoring activities are profitable. Fig. 2 presents the case of γMx ≥ (1 − txG)θx(1 / α)Mαx exclusively (i.e. Dgx N 0). Given monitoring value (1 / α)Mαx and monitoring cost γMx, the after-tax return Rxm of blockholder x (i.e. txD b txG) decreases as the dividend proposal Dx ≥ 0 decreases. Alternatively, blockholder x can enjoy after-tax return R0x by accepting the zero-dividend proposal D0 = 0. If Rxm b R0x, then blockholder x loses incentive to undertake a monitoring function.

6

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

(1 −

)



)

1

0

(

0 Fig. 2. Dividend level at which blockholder x decides to accept the zero-dividend proposal.

On deciding the firm's dividend policy D, it is reasonable for managers to be loyal to the dominant proposal because shareholders in the winning party have sufficient power to remove managers for a lack of concern if the firm does not adopt the proposal. Based on Eqs. (3), (4), and (7), the firm's dividend policy D is selected as expressed in Eq. (8).  D¼

w

Dx 0

if if

w

g

Dx NDx w g Dx bDx

ð8Þ

g w If Dw x N Dx, then the incentive compatibility condition of blockholder x holds. In this case, when blockholder x proposes Dx , the w w g median shareholder μ switches her vote to the monitoring-based proposal (i.e. D = Dx ). However, if Dx b Dx, then the incentive compatibility condition does not hold. In this case, blockholder x accepts the zero-dividend proposal (i.e. D = 0). When a blockholder gives up undertaking a monitoring function, the existence of large shareholdings cannot be an effective mechanism for the resolution of free-rider problems among shareholders. g x x Proposition 2. If the incentive compatibility condition holds (i.e. Dw x N Dx), then dividend-seeking blockholder x (i.e. tD b tG) has an w incentive to undertake a monitoring function. Therefore, the monitoring-based dividend policy (i.e. D = Dx ) is adopted. Otherwise, the zero-dividend policy (i.e. D = 0), which minimizes the tax burden of dividend-averse investors (i.e. tiD N tiG), is adopted.

3.3. Optimal monitoring activity Although we have regarded the added value (1 / α)Mαx as an exogenously given parameter in Sections 3.1 and 3.2, this section describes derivation of the optimum monitoring activity Mx⁎. As explained already in Section 3.1, we assume that the Mx unit of the monitoring activity provided by blockholder x adds market value of v(M x ) = (1 / α)Mαx to the firm, where 0 b α b 1 / 2 (i.e., v′(Mx) N 0, v″(Mx) b 0). That is to say, α denotes blockholder x's efficiency in monitoring activity. This assumption is the same as that used by Allen et al. (2000).8 Unfortunately, if we allow the range of monitoring activity be 0 b Mx b 1, then the market value added v(Mx) = (1 / α)Mαx becomes a monotonically decreasing exponential function of blockholder x's efficiency in monitoring activity α. To avoid this and make the issue tractable, we assume that Mx N 1 in the case of blockholder x's undertaking a monitoring function (i.e. ∂v(Mx) / ∂α N 0), whereas Mx = 0 in the case of no monitoring services (i.e. ∂v(Mx)/∂α = 0). Blockholder x is assumed to maximize the after-tax return Rxm subject to Eqs. (3), (4), and (7), as shown in Eq. (9). In contrast to the formulation presented by Allen et al. (2000), the newly added third term represents the private benefit from tax saving when undertaking a monitoring function. The magnitude of the after-tax return depends not only on monitoring activity but also on the dividend proposal.      x 1 x x x x w x α M x −γMx þ t G −t D θ Dx ≡ Rm max 1−t G θ G þ Mx α

ð9Þ

Differentiating Eq. (9) with respect to Mx, we obtain the first-order maximization condition of the after-tax return, as presented in Eq. (10) as follows (i.e. ∂Rxm / ∂Mx = 0). The first term of the left-side is designated as direct effects because the monitoring activity 8 Strictly speaking, in the original formulation presented in Allen et al. (2000), the level of monitoring activity in the aggregate is an increasing function of the proportion of multiple blockholders. This paper differs from theirs in terms of the assumption of a single monitor as in Shleifer and Vishny (1986). The proportion θx is not of multiple blockholders, but of a single blockholder.

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

(1 −

−1

) +(

7



)



(1 −

)

−1

0 Fig. 3. Optimal monitoring activity.

directly enhances the after-tax return Rxm. The second term is designated as indirect effects because the monitoring activity alters dividend policy, and thereby indirectly influences the after-tax return Rxm. 

w x x x α−1 x x ∂D 1−t G θ Mx þ t G −t D θ  x ¼ γ ∂Mx

ð10Þ

When the left-side of Eq. (10) is smaller than the right-side at any Mx, the optimal level of monitoring activity must be zero as a g corner solution (i.e. Mx⁎ = 0). Otherwise it is positive as an interior solution (i.e. Mx⁎ N 1). As described in Eq. (8), if Dw x b Dx, then blockholder x accepts the zero-dividend proposal (i.e. D = 0) and provides no monitoring activity (i.e. Mx⁎ = 0). Nevertheless, we specifically examine the case in which the optimal level of monitoring activity is positive hereinafter (i.e. M⁎ x N 1). Solving Eq. (10), we obtain the optimal level as shown in Fig. 3 and Eq. (11) as follows. In this case, blockholder x's optimal monitoring 9 activity is an increasing function of blockholder x's monitoring efficiency (i.e. ∂M⁎ x / ∂α N 0). 

Mx ¼

 

1 1−α w x x x ∂D x 1−t G þ t G −t D  x θ =γ ∂M x

ð11Þ

⁎ Fig. 3 presents the case of ∂Dw x / ∂Mx N 0. The optimal level of monitoring activity Mx is obtained at the point where the marginal benefit of monitoring activity, (1 − txG)θxMxα−1 + (txG − txD)θx{∂Dw x / ∂Mx}, equals the marginal cost of monitoring activity γ. Within ⁎ and Mx⁎, the direct effects of additional monitoring effort produce marginal losses of γ − (1 − txG)θxMxα−1. the area between M⁎ x ⁎ is not a reasonable solution because indirect effects (txG − txD)θx{∂Dw Nevertheless, selecting M⁎ x x / ∂Mx} are not used completely. Based on Eq. (4), if Dμx b G + (1 / α)Mαx , then blockholder x must lower the level of the dividend proposal to dominate the zeroμ dividend proposal (i.e. Dw x = Dx). The increased monitoring activity raises the dividend level at which blockholder x seizes median sharex x holder μ's vote from the zero dividend proposal (i.e. ∂Dw x / ∂Mx N 0). Subsequently, this action reduces blockholder x's (i.e. tD b tG) tax burden. The optimal level of monitoring activity is the following. 

Mx ¼

"( 

μ x x x 1−t G 1−t G þ t G −t D μ μ t D −t G

!)

# x

θ =γ

1 1−α

μ

if Dx bG þ

  1 α Mx α

ð12Þ

In this case, the monitoring-based dividend proposal Dw x which can just dominate the zero-dividend proposal is rewritten from Eqs. (4) to (13) as follows. w

Dx ¼

Dw x

μ

1−t G μ

μ

t D −t G



!) #α  "( μ 1−α  x  1 x x 1−t G x μ θ =γ ≡ Dx 1−t G þ t G −t D μ μ α t D −t G   1 μ α if Dx bG þ Mx α

ð13Þ

However, if Dμx ≥ G + (1 / α)Mαx , then blockholder x wins the voting contest by proposing full dividends with no concession [i.e. = G + (1 / α)Mαx ], as demonstrated in Eq. (4). Therefore, the marginal effect of increased monitoring activity on the dividend

9 x Within our model, there exists no possibility that ∂M⁎x / ∂α b 0. If {(1 − txG) + (txG − txD)(∂Dw x / ∂Mx)}θ is smaller than γ within Eq. (11), then the left-side of Eq. (10) is always smaller than the right side because Mxα−1 ≤ 1 under the assumption that 0 b α b 1 / 2. In this case, no monitoring service should be provided (i.e. M⁎x = 0). Therefore, ∂M⁎x / ∂α = 0.

8

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

level at which blockholder x seizes median shareholder μ's vote from the zero dividend proposal is zero (i.e. ∂Dw x / ∂Mx = 0). The optimal level of monitoring activity is the following.  Mx

( ) 1 1−t xG θx 1−α ¼ γ

μ

if Dx ≥G þ

  1 α Mx α

ð14Þ

In this case, the monitoring-based dividend proposal Dw x which can just dominate the zero-dividend proposal is rewritten from Eqs. (4) to (15) as follows. w

Dx ¼ G þ

)α  ( 1−t xG θx 1−α 1 γ α

μ

if Dx ≥G þ

  1 α Mx α

ð15Þ

Proposition 3. As long as the indirect effects of monitoring activity [i.e. (txG − txD)θx{∂Dw x / ∂Mx}] reduces the tax burden sufficiently, dividend-seeking blockholder x (i.e. txD b txG) has an incentive to exert additional monitoring effort in an attempt to affect firm dividend policy, even if the direct effects of monitoring activity (i.e. (1 − txG)θxMxα−1) produce a marginal loss. In short, a dividend-seeking corporation or institution with higher ability in monitoring activity is more likely to exercise greater monitoring activity Mx in an attempt to raise the winning point Dw x . Comparative static analysis reveals that blockholder x's optimal monitoring activity is an increasing function of (i) blockholder x's efficiency in monitoring activity in the case of blockholder x's unx ⁎ dertaking a monitoring function (i.e. ∂M⁎ x / ∂α N 0), (ii) the percentage of blockholder x's shareholdings in firm (i.e. ∂Mx / ∂θ N 0), x x ⁎ and (iii) the tax differential for dividend-seeking blockholder x (i.e. ∂Mx / ∂(tG − tD) N 0). However, blockholder x's optimal monitoring activity is a decreasing function of (iv) the marginal cost of monitoring activity (i.e. ∂M⁎ x / ∂γ b 0), and (v) the tax differential for μ μ dividend-averse median shareholder μ (i.e. ∂M⁎ x / ∂(tD − tG) b 0). 4. Implications When a blockholder must make a concession to win a voting contest [i.e. Dμx b G + (1 / α)Mαx ], the necessity of obtaining majority support reduces individuals' respective tax burdens because the blockholder must lower the dividend proposal. Benefits from this are shared by individuals at the expense of the blockholder. Reduced dividends imply reduced side payments from individuals to the blockholder. Furthermore, the necessity of obtaining majority support elicits additional monitoring effort because the blockholder attempts to raise the level of the dividend proposal. Benefits from this effect are free-ridden by individuals at the expense of the blockholder. However, individuals bear additional expenses of increased tax burdens. Additional dividends serve as additional side payments from individuals to the blockholder. The model presented in this paper provides several empirical implications. First, the changes in ownership structure or marginal tax rates are expected to exercise influences on dividend policy. In the case of Eq. (13), blockholder x's winning point is a decreasing μ μ function of dividend-averse median shareholder μ's tax differential (i.e. ∂Dw x / ∂(tD − tG) b 0). Under a system of progressive taxation, higher income tax rates are levied on investors in the higher income group. Therefore, we predict that the level of dividend payments is lower in firms that have many investors in higher income brackets.10 Nevertheless, testing this prediction directly would be extremely difficult because the shareholders' income distribution for each stock is unobservable.11 Second, dividend policy depends on the degree of the blockholder's monitoring efficiency. Based on Eqs. (13) and (15), a dividendseeking corporation or institution with higher ability in monitoring activity seizes the median shareholder's vote at a higher winning point (i.e. ∂Dw x / ∂α N 0). Therefore, we predict that a firm recruiting an efficient dividend-seeking monitor is more likely to distribute higher dividends. According to Chen et al. (2007), independent institutions with long-term investment horizons are expected to have superior monitoring abilities. Gaspar et al. (2013) report that institutions making long-term investments compel firms to pay more dividends. Apparently, our model provides a theoretical explanation for this phenomenon. Additionally, we predict that a firm with a high managerial stake distributes little or no dividends. In the model developed for these analyses, we assume that a blockholder's monitoring activity mitigates agency conflicts between shareholders and managers, and therefore increases the market value. However, if the firm has no potentially severe agency problems, then it has little room for improvement of its value from the blockholder's monitoring services. The less severe the agency conflict becomes, the lower the g blockholder's efficiency in monitoring activity α becomes. In this case, the possibility of satisfying condition Dw x b Dx and paying no g dividend is increased because the level of (1 / α)Mαx is low. Even if the condition Dw x N Dx is satisfied, the dividend payments adopted by the firm with less severe agency conflicts are expected to be low (i.e. Dw x is low). Empirical results show a negative association 10 For example, Dong et al. (2005) conduct a questionnaire targeted at individual investors in the Netherlands and report that the high income group has less preference for dividends under that system of progressive taxation. 11 Chetty and Saez (2005) find increased dividend payments after the U.S. tax reform of 2003, which reduced marginal tax rates on dividend income. Korkeamaki et al. (2010) examine the effect of the dividend tax increase around the 2004 tax reform in Finland, and show that dividend payment decreases across all firms after the reform. These empirical facts are consistent with our prediction, although they are not firm level evidence.

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

9

between managerial ownership and the dividend payment (e.g., Ang et al., 2000; Eckbo and Verma, 1994; Jensen et al., 1992; and Short et al., 2002). Third, we predict that a corporation or institution with a larger tax differential is more likely to dominate the zero-dividend prog posal (i.e. Dw x N Dx). Based on Eq. (13), dividend-seeking blockholder x's winning-point is an increasing function of the tax differential x x (i.e. ∂Dw x / ∂(tG − tD) N 0). In contrast, based on Eq. (7), blockholder x's giving-up point is a decreasing function of the tax differential g x x (i.e. ∂Dx / ∂(tG − tD) b 0). This implication is the most characteristic of the current model. In a new light, our model might provide fresh insight into the dividend controversy arena. Fourth, the relation between a dividend-seeking blockholder's stake and the level of dividend payments is expected to be positive. The winning-point Dw x is an increasing function of the percentage of dividend-seeking blockholder x's shareholdings in the firm (i.e. x ∂Dw x / ∂θ N 0) because the optimal level of monitoring activity is increased with an increase in the blockholder's stake. Some empirical studies provide evidence that the level of corporate/institutional ownership is positively associated with dividend payments (e.g., Eckbo and Verma, 1994; Moh'd et al., 1995; and Short et al., 2002). Our model seems consistent with this empirical evidence. Fifth, asymmetric information between managers and investors might influence dividend policy. For example, we predict that a blockholder of a firm in R&D intensive industry confronts a higher marginal cost of monitoring activity because of more severe asymw metric information. The winning-point Dw x is a decreasing function of the marginal cost of monitoring activity (i.e. ∂Dx / ∂γ b 0). Some empirical reports describe that R&D-intensive firms are likely to pay lower dividends, but hold more cash than non R&D-intensive firms in an attempt to finance growth opportunities using internal funds (e.g., Bah and Dumontier, 2001; Bates et al., 2009; and Gugler, 2003). Our prediction does not contradict these facts, although the reasons for lower dividends differ. Some empirical studies have revealed that corporate and institutional ownership does not induce firms to increase dividends. Grinstein and Michaely (2005) report that institutional investors tend to prefer low dividends to high dividends overall. Barclay et al. (2009) report that corporations and institutions are not attracted to high-dividend paying stocks. Desai and Jin (2011) find no evidence that corporations affect firm dividend policy, although they do find that managers alter their dividend policy to serve the interests of dividend-averse institutions. As described in this paper, a dividend-seeking blockholder's monitoring incentive is an increasing function of a firm's dividend payment because a corporate or a dividend-seeking institution in a traditional tax-preference framework is assumed to prefer higher dividends monotonously (e.g. Allen and Michaely, 2003). Within such a one-date framework, it might be puzzling that dividendseeking investors do not cause firms to increase dividend payments, in contrast to the intuitive case of dividend reduction tailored for dividend-averse investors. Grinstein and Michaely (2005) and Barclay et al. (2009) argue that their evidence is difficult to reconcile with the monitoring-based theory of dividend policy (e.g., Allen et al., 2000 and Shleifer and Vishny, 1986). Alternatively, within a two-date framework, Mori (2010) demonstrates that dividend-seeking investors prefer consumptionoptimal dividends to avoid intertemporal double taxation on reinvested dividends. Based on this view, the dividend-seeking investor's preference for low dividend payment (e.g. Barclay et al., 2009; Desai and Jin, 2011; and Grinstein and Michaely, 2005) is insufficient evidence to reject the monitoring-based theory of dividend policy. Although the current paper is not concerned with intertemporal aspects of investor preferences, the possibility exists that not only the necessity of winning majority support of shareholders but also a dividend-seeking blockholder's liquidity needs engender lower dividend payments. From the perspective of dividend-averse individuals, a monitor might be recruited to solve the free-rider problem without using a high dividend payment as an inducement. As explained in Section 2, institutional investors of several types are used, although the model developed in this paper exclusively considers a dividend-seeking blockholder (i.e. txD b txG).12 For dividend-averse individuals (i.e. tiD N tiG), recruiting a dividend-averse blockholder (i.e. txD N txG), who also prefers zero dividends, or a tax-exempt institution (i.e. txD = txG = 0), that is indifferent to the level of dividend payment, is apparently a more favorable alternative. This adds a higher threshold for a dividend-seeking blockholder to succeed in undertaking a monitoring function in a firm. Conversely speaking, a positive dividend policy can be interpreted as a result of a dividend-seeking blockholder's advantage in monitoring ability over the other two types. One might infer that the consequence of prudent-man rules is a much simpler explanation for distributing positive dividends by firms. Some institutional investors must invest in dividend-paying stocks because of their fiduciary duties. Brav and Heaton (1998) find that institutions tend to stop holding dividend-omitting stocks because they are sensitive to laws governing their actions. According to a survey by Brav et al. (2005), many financial managers pay dividends to attract institutions subject to prudent-man rules. However, the foregoing argument leaves unanswered the question of how firms determine dividend policy. Institutions might be motivated to hold dividend-paying stocks for non-tax reasons, but a convincing criterion for determining the level of dividend payment can be provided by tax reasons, as this paper demonstrates. 5. Conclusion According to Shleifer and Vishny (1986) and Allen et al. (2000), the existence of a corporate or institutional blockholder can be an effective mechanism for mitigating agency conflicts between shareholders and managers in a dispersed ownership structure. Based on a corporate or institutional investor's tax preference, positive dividends serve as a side payment from small individual investors who delegate monitoring tasks to a dividend-seeking blockholder. 12 For example, mutual funds and investment advisors are usually regarded as dividend-averse (i.e. txD N txG) because they pass tax burdens through to their underlying shareholders, who are typically individual investors. In contrast, banks and insurance companies are usually categorized as dividend-seeking (i.e. txD b txG) because their own investment can benefit from a dividend tax deduction.

10

N. Mori, N. Ikeda / Journal of Corporate Finance 30 (2015) 1–10

However, existing models do not consider how the necessity of winning majority support of shareholders influences the relation between a blockholder's monitoring incentive and a firm's dividend policy. In reality, when some individuals support a positive dividend policy, other individuals will prefer zero-dividend policy because they have different division points between these two alternatives. In this sense, they should be regarded as heterogeneous in spite of the fact that their tax-optimal dividends are unanimously zero. For undertaking a monitoring function in a firm, a blockholder must win the majority support of shareholders who have various marginal tax rates. This setting differs from that of previous models. This paper presents an alternative version of the side-payment model of dividend policy. When dividend-averse individuals collectively share a majority stake in a dispersed ownership structure, a dividend-seeking corporation or institution might be compelled to propose low dividends to dominate the zero-dividend proposal. Unless the blockholder loses the incentive to undertake a monitoring function, the firm gives approval to the monitoring-based dividend proposal that obtains the majority support of shareholders. Under such circumstances, the blockholder finds it reasonable to raise the division point at which the median shareholder decides to support. As long as the tax burden is reduced sufficiently, the blockholder has an incentive to provide unprofitable monitoring activities in an attempt to affect firm dividend policy, even if such additional monitoring effort yields marginal losses. The results obtained from our model present important implications for financial management and corporate governance. The optimal level of monitoring activity depends not only on a blockholder's efficiency in monitoring activity, but also on the median shareholder's tax status. Benefits from additional monitoring effort are free-ridden by individuals. However, additional dividends serve as additional side payments from individuals to the blockholder. Consequently, the mutually complementary relation between individuals as a group and a dividend-seeking blockholder is broader than previously recognized. Finally, we outline some important directions for future research. First, the current model can be modified to demonstrate that multiple blockholders compete for controlling positions related to dividend policies in a firm. Second, the model presented herein does not address the manner in which firms use share repurchases with dividend payment. This line of extension presents interesting avenues leading to a richer analysis of payout policy. Third, empirical tests are reserved for future investigations. As Allen and Michaely (2003) report, whether investors with large shareholdings are attracted to dividend-paying stocks or not remains as a matter that has yet to be tested empirically. Acknowledgements Authors wish to thank the editor (Jeffry Netter) and an anonymous referee. For helpful comments and suggestions, we also thank Toshiki Honda, Nobuyuki Isagawa, Katsumasa Nishide, Kazuhiko Ohashi, Wataru Ohta, Takashi Shibata, Masahiko Sugie and seminar participants at Doshisha University, Hitotsubashi University, Osaka University, Tokyo Metropolitan University, and Yokohama National University. Financial support from Nihon University (803265-2012) and language help from Brad Fast are gratefully acknowledged. References Admati, A., Pfleiderer, P., Zechner, J., 1994. Large shareholder activism, risk sharing and financial market equilibrium. J. Polit. Econ. 102, 1097–1130. Allen, F., Michaely, R., 2003. Payout policy. In: Constantinides, G., Harris, M., Stulz, R. (Eds.), Handbook of the Economics of Finance vol. 1A. Elsevier, North-Holland, pp. 337–429. Allen, F., Bernardo, A., Welch, I., 2000. A theory of dividends based on tax clienteles. J. Financ. 55, 2499–2536. Ang, J., Cole, R., Lin, J., 2000. Agency costs and ownership structure. J. Financ. 55, 81–106. Bah, R., Dumontier, P., 2001. R&D intensity and corporate financial policy: some international evidence. J. Bus. Financ. Account. 28, 671–692. Barclay, M., Holderness, C., Sheehan, D., 2009. Dividends and corporate shareholders. Rev. Financ. Stud. 22, 2423–2455. Bates, T., Kahle, K., Stulz, R., 2009. Why do U.S. firms hold so much more cash than they used to? J. Financ. 64, 1985–2021. Black, D., 1948. On the rationale of group decision-making. J. Polit. Econ. 56, 23–34. Brav, A., Heaton, J., 1998. Did ERISA prudent man rule change the pricing of dividend omitting firms? Working Paper. Duke University. Brav, A., Graham, J., Harvey, C., Michaely, R., 2005. Payout policy in the 21st century. J. Financ. Econ. 77, 483–527. Chen, X., Harford, J., Li, K., 2007. Monitoring: which institutions matter? J. Financ. Econ. 86, 279–305. Chetty, R., Saez, E., 2005. Dividend taxes and corporate behavior: evidence from the 2003 dividend tax cut. Q. J. Econ. 120, 791–833. Desai, M., Jin, L., 2011. Institutional tax clienteles and payout policy. J. Financ. Econ. 100, 68–84. Dong, M., Robinson, C., Veld, C., 2005. Why individual investors want dividends. J. Corp. Financ. 12, 121–158. Downs, A., 1957. An economic theory of political action in democracy. J. Polit. Econ. 65, 135–150. Eckbo, B., Verma, S., 1994. Managerial shareownership, voting power, and cash dividend policy. J. Corp. Financ. 1, 33–62. Gaspar, J., Massa, M., Matos, P., Patgiri, R., Rehman, Z., 2013. Payout policy choices and shareholder investment horizons. Review of Finance 17, 261–320. Grinstein, Y., Michaely, R., 2005. Institutional holdings and payout policy. J. Financ. 60, 1389–1426. Grossman, S., Hart, O., 1980. Takeover bids, free rider problem and the theory of the corporation. Bell J. Econ. 11, 42–64. Gugler, K., 2003. Corporate governance, dividend payout policy, and the interrelation between dividends, R&D, and capital investment. J. Bank. Financ. 27, 1297–1321. Hirshleifer, J., Glazer, A., Hirshleifer, D., 2005. Price Theory and Applications: Decisions, Markets, and Information, 7th edition. Cambridge Univ. Press. Jensen, G., Solberg, D., Zorn, T., 1992. Simultaneous determination of insider ownership, debt, and dividend policies. J. Financ. Quant. Anal. 27, 247–263. Korkeamaki, T., Liljeblom, E., Pasternack, D., 2010. Tax reform and payout policy: do shareholder clienteles or payout policy adjust? J. Corp. Financ. 16, 572–587. Maug, E., 1998. Large shareholders as monitors: is there a trade-off between liquidity and control. J. Financ. 53, 65–98. Moh'd, M., Perry, L., Rimbey, J., 1995. An investigation of the dynamic relationship between agency theory and dividend policy. Financ. Rev. 30, 367–385. Mori, N., 2010. Tax clientele effects of dividends under intertemporal consumption choices. J. Bank. Financ. 34, 1089–1097. Richardson, G., Sefcik, S., Thompson, R., 1986. A test of dividend irrelevance using volume reactions to a change in dividend policy. J. Financ. Econ. 17, 313–333. Shleifer, A., Vishny, R., 1986. Large shareholders and corporate control. J. Polit. Econ. 94, 461–488. Shleifer, A., Vishny, R., 1997. A survey of corporate governance. J. Financ. 52, 737–783. Short, H., Zhang, H., Keasey, K., 2002. The link between dividend policy and institutional ownership. J. Corp. Financ. 8, 105–122.