Large Shareholders and Dividends: Game Theoretic Analysis of Shareholder Power

Transcription

1 Large Shareholders and Dividends: Game Theoretic Analysis of Shareholder Power Xiaoying Chen a, 1, Amit K. Sinha b a Department of Finance, College of Business Administration, California State University, Long Beach, 1250 Bellflower Boulevard, Long Beach, CA b College of Business, Indiana State University, Terre Haute, IN Tel: ; fax: address: xchen@csulb.edu (X. Chen).

2 Large Shareholders and Dividends: Game Theoretic Analysis of Shareholder Power Abstract Using the Banzhaf index as a proxy for voting power, this paper investigates the importance of large shareholders in influencing financial policy regarding cash dividends. Our empirical findings indicate that this potential association is more significant when the impact of large shareholders on corporate management is measured by their voting power rather than by the traditional measurement, size of their equity ownership. Keywords: Corporate finance; Dividend payout; Shareholder influence; Voting power 1

3 Large Shareholders and Dividends: Game Theoretic Analysis of Shareholder Power A firm s dividend policy refers to the financial decisions the firm makes about whether to pay shareholders dividends, how large these dividends should be, and how frequently and in what form dividends should be distributed. Although shareholders may play a role, their representatives on the board of directors determine the dividend policy. Small investors normally do not have sufficient incentives to monitor the board, while large shareholders, in contrast, may potentially accumulate their control rights over the board (Shleifer and Vishny, 1986). For this reason, the interests of such large shareholders tend to steer a firm s dividend policy. This paper extends the analysis of the impact made by large shareholders on dividend payout policy. In particular, responding to the problem of measuring shareholder control power by the size of shareholdings in prior research, we advocate the use of the shareholder voting power index derived from the theory of cooperative games to analyze this corporate policy choice. In addition, we adopt a cleaned dataset of large shareholders in publicly traded corporations in the U.S. As documented in Dlugosz, Fahlenbrach, Gompers and Metrick (2006), despite the common use of large shareholder data in financial studies, there is no clean off-the-shelf database available to facilitate research. They uncover the fact that even the currently most widely used ownership database, the Compact Disclosure (CD) of Standard & Poor s, has mistakes and biases such as overlaps and erroneous consideration of preferred stocks. Therefore, they propose a consistent set of solutions to fix these problems and finally provide a clean database freely accessible to all researchers ( 2

4 This enables us to investigate the relationship between the dynamics of dividend payout and shareholder power and to gain a better understanding. To achieve our objective, we limit the scope of our research to the following questions. Do large shareholders affect payout policy? Does the answer vary across three well-established dividend models: the Full Adjustment Model (FAM), the Partial Adjustment Model (FAM) (Lintner, 1956), and the Earning Trend Model (ETM) (Fama and Babiak, 1968)? Lastly, does the answer vary if the control power of large shareholders is represented in different empirical methods such as size of stock holdings and voting power index? Our findings indicate that as far as large shareholders are concerned, voting power may be more important than equity ownership in influencing payoff polices. 1. Literature In a world without market perfections, dividend policy is irrelevant to the value of a firm. It doesn t make any difference how the firm s profit is distributed, whether paid to the owners or retained in the business for reinvestment (Miller and Modigliani, 1961). Various theories have proposed some factors for the fact that many firms do pay dividends. The agency theory of dividend policy claims that dividend payments act to reduce agency costs between managers and shareholders by forcing a firm to pay out excess profit. By doing this, managers are prevented from consuming the excess profits as perquisites or wasting them on unwise projects (Jensen, 1986). For future additional funding, managers are forced to seek out external capital markets. These markets will, in turn, closely monitor the firm s management and agency problem (Rozeff, 1982; 3

5 Easterbrook, 1984). The signaling theory of dividend policy suggests that managers use dividend payments to convey information to investors about the firm s expected earnings (Bhattacharya, 1979, 1980; Miller and Rock, 1985). Due to information asymmetry, shareholders don t have access to a complete set of information about a firm s expected cash flows. Therefore, the firm s managers, as insiders, would like to establish credible signals conveying their firm s prospects to the shareholders. Dividend payout becomes one of the most efficient signals. As this is costly, poor performers cannot simply follow. In spite of the agency theory and signaling theory, we also see some real-world factors affecting the levels of dividend payout, such as tax and new equity flotation costs. Empirically the relationship between shareholders and payouts has been investigated by Zeckhauser and Pound (1990), Gugler and Yurtoglu (2003), Goergen, Renneboog, and Silva (2005), Moh d, Perry, and Rimbey (1995), Eckbo and Verma (1994), and Short, Zhang, and Keasey (2002), among others. Zeckhauser and Pound (1990) conclude that in the United States, ownership concentration and payout policy may not be considered substitute monitoring devices, as they do not find significant differences in payout ratios with or without large block shareholders. Gugler and Yurtoglu (2003), on the other hand, find that for German firms, the power of the second largest equity holder increases payout, although the power of the largest shareholder may reduce payout. Moh d, et al. (1995) find that for U.S. firms, as ownership of the firms becomes more dispersed, payout increases. 4

6 2. Methodology Although shareholder power is crucial to the analysis of such financial decisions, the actual level of shareholder influence over board decisions is barely available publicly. In prior literature, the size of stocks controlled by shareholders is often included in empirical models for the analysis of shareholder control. However, the size of stock holdings is only a crude proxy for the influence power of a particular shareholder. The main problem is that it ignores how stocks are distributed among other shareholders. For instance, a block representing 10% of shares in a firm which has a widely dispersed ownership structure might obtain dominant control, while a block of 15% in a firm which has less dispersed ownership distribution may not give its holder significant power. Hence, a block of even 5% ownership may become important to the system of corporate governance in countries without large concentrations of share ownership such as the U.S. As we know, shareholders make collective decisions by voting. Normally, votes are granted to shareholders based upon the size of their shareholdings, and one share represents one vote. It is necessary to make a strong distinction between a shareholder s voting weight, representing the size of their shareholding, and voting power, representing their ability to change--- or the probability of changing--- the outcome of a decision with their vote. Assuming a firm follows a simple majority-voting rule, that is, any proposal will be approved if the votes cast by shareholders are over 50 percent, and, for simplicity, assuming there are only three shareholders in this firm and they hold 49, 49, and 2 votes: Thus, the voting weights of these shareholders are 49, 49 and 2 percent respectively. How many chances does each shareholder have to change the voting outcome? The answer is equal, because each shareholder wins the voting only if at least one of the other 5

7 two shareholders casts votes in the same way as they does. Thus, the normalized voting powers of these shareholders are equally 1/3. This demonstrates that it is the relative rather than the absolute influence power of a given shareholder that determines his/her ability to influence the firm s policies (Crespi and Renneboog, 2003). Therefore, we use the Banzhaf index derived from the co-operative game theory to represent the ability of large shareholders to influence the voting outcome of payout policy. The Banzhaf index was introduced for the purpose of analyzing block voting systems: the probability that a citizen's vote will change the block's "decision," and the probability that the block's votes will change the outcome of the election. Renneboog and Trojanowski (2005) point out that this mechanism captures the power to influence policy and it would be the appropriate instrument for measuring shareholder voting power, since by their nature, payout choices are policy issues. This index has been popularly adopted in recent empirical studies to examine the shareholder voting power system in the U.K., Spain, and France (Leech, 2001; Leech and Manjon, 2003; Bloch and Kremp, 1999). To the best of our knowledge, this is the first study to apply the game theory-based index to corporate payout policies in the U.S. The models we use to investigate the relationship between large shareholders and dividend policy are based upon the FAM, the PAM, and the ETM. The developments of these models are discussed in length in Short, Zhang and Keasey (2002). The FAM assumes that a firm has a target dividend payout ratio r, and then the firm adjusts the changes in dividends (D) in response to variations in firm earnings (E). For firm i at time t, this relation will be functioned as: ( ) μ, t Di, t 1 = α + r Ei, t Ei, t 1 + (FAM) (6) D i i, t 6

8 The PAM assumes that the target dividend D * of a firm, for any given year t, is related to the earnings at t, through the desired payout ratio r, then: * Di, t = r Ei, t (7) Considering that the firm, in any given year, will only adjust partially to its target dividend level, the model then becomes: D ( ) D = a + c D D (8) * i, t i, t 1 i, t i, t 1 Substituting equation 7 in equation 8, we get D ( r ) Di, t 1 = a + c Ei, t Di, t (PAM) (9) i, t 1 In this equation, the resistance of managers to reduce dividends is represented by the constant a, while coefficient c represents the speed of adjustment to the new target. The ETM is a modified partial adjustment model. It assumes that the earnings in a particular year are based on trend factor j, such that E ( j) E = Ei, t 1 + j Ei, t 1 = + i, t (10) i, t 1 1 Since target dividends are given by: D r E * i, t = i, t After substitution to (10), we get ( c) c μ, t Di, t 1 = α + cr Ei, t + rj 1 Ei, t 1 Di, t 1 + (ETM) (11) D i i, t In order to investigate the link between large shareholders and dividend policy, these models are modified by including interactions of the shareholder variables (e.g., Banzhaf index and size of equity of a large shareholder). We also include a vector of variables (firm size, leverage, Tobin s Q, year, and industry) related to a firm s dividend policy. 7

9 Now assuming that payouts of firms (cash dividend or stock repurchase) would be influenced by large shareholders, then adjusting the models above for such influences results in the following models. ( ) + ( ) shareholder μ Di, t Di, t 1 = α + r Ei, t Ei, t 1 r I Ei, t Ei, t 1 * + (12), t Di, t 1 = α + cr Ei, t + c r I Ei, t * shareholder c Di, t 1 + μ (13) D i i, t ( 1 c) + r j ( 1 c) shareholder c + μ Di, t Di, t 1 = α + cr Ei, t + rj Ei, t 1 Ei, t 1 * I (14) In these equations, the influence power of a specific shareholder (shareholder) is measured by either his/her equity ownership (the size of stockholdings, ownership) or his/her voting power (the Banzhaf index, voting power), while D is cash dividend payout. i, t D i, t 1 i, t 3. Data We start with a sample from Dlugosz et al. (2006). The authors collect data from original proxy statements for the largest 1,500 U.S. companies from 1996 to The SEC requires that proxy statements list all investors owning more than 5% of a company s common stock. The data in the sample includes company name, large shareholders, industry, size of ownership, and shareholder identity (officer, director, or outsider). After merging the initial sample with COMPUSTAT to extract variables such as total assets, sales, liabilities, dividends and stock repurchase, we obtain 5,495 firmyears for 1,494 unique firms. We focus on the 4,792 firm-years with positive earnings (EBIT), covering 1,326 unique firms. Table 1 summarizes the sample characteristics of all 1,494 firms in our sample. The average book value is 7.66 billion dollars. The average earnings, basically the earnings 8

10 before interest and taxes (EBIT), are million dollars. Shareholders (common and preference shareholders) are paid million dollars in dividends on average from 1996 to The average Tobin s Q and the average leverage of the sample firms are 2.12% and 46% respectively. For firm-years with positive earnings, as also shown in Table 1, apparently more cash dividends are compensated (70.20 million). [Please insert Table 1 about here] Table 2 shows the average dividend payout ratios in each year of our sample period for all firms and for firms with positive earnings. Payout ratios are defined as the dividend payouts divided by earnings. From this table, we can notice that dividend payout is a relatively stable financial practice from 1997 to For positive EBIT firms, the average dividend payout ratio is between 13.55% and 14.67%. For all firms, the average dividend payout ratio is between 13.55% and 14.67%. What this tells us is that firms pay out one-sixth of their earnings in dividends and reinvest the other portion of earnings back into their business. Although a stable dividend policy is in the best interests of a firm and its stockholders, in 1998, many firms, especially negative EBIT firms, undertook a dividend cut policy. The average payout ratios drop to 8.63% and 13.55% respectively, for all firms and negative EBIT firms in But in 2001, the payout ratios increase greatly and firms pay a high percentage of their earnings as dividends. The payout ratios are as high as 19.32% and 27.52% for all firms and positive EBIT firms. [Please insert Table 2 about here] Table 3 provides information about the distribution of stock ownership and voting power for positive EBIT firm-years. Panel A illustrates the distribution of stock 9

11 ownership held across different classes of shareholders. In Dlugosz et al. (2006), block shareholders are classified into five categories. They are: (1) officer blockholders, (2) non-officer director blockholders, (3) affiliated blockholders, either individuals or trusts, who are likely influenced by officer or director (4) ESOPs (Employee Share Ownership Plans) related blockholders and (5) outside blockholders. Even though any given large shareholder in our sample owns at least 5% of a given firm s common stock, some firms may not have all five types of shareholders, so the average ownership may possibly be lower than 5%. Among all, outside blockholders normally control the largest portion of total shares outstanding (18.68%). Panel B presents the size of stock holdings and the control power of the top 3 largest shareholders, as well as their identity. The largest, 2 nd largest, and 3 rd largest shareholders hold 14.35%, 6.57%, and 3.40% of the outstanding stock. As suggested by Renneboog and Trojanowski (2005), we consider a one-stage voting game, where each large shareholder is treated as a separate player and we compute the corresponding Banzhaf index. The empirical results in Renneboog and Trojanowski (2005) reject the hypothesis that large shareholders form a coalition and that the coalition s votes will change the outcome. They suggest that, rather than forming typebased coalitions and participating in the voting game, large shareholders may achieve their payout policy goals in their own interest. As shown in Panel B, the voting powers distributed among large shareholders are 0.74, 0.11, and 0.10 for the largest three shareholders. For an international comparison, these values are lower than the Banzhaf indices of the largest three shareholders for British firms ( ), 0.65, 0.14, and 0.13, responding to the more widely dispersed ownership in the U.K. Considering the 10

12 dominant power controlled by the largest shareholder in our sample, in the remaining analysis, we focus on the impact of a firm s largest shareholder on its payout policy in 4,792 firm-years with positive earnings (EBIT), covering 1,326 unique firms. In Panel B, we also report the identities of the largest shareholder. As shown, most of the largest shareholders are outsiders (3,254), followed by officer shareholders (513). [Please insert Table 3 about here] 4. Empirical results Table 4 provides the regression results for our three models, examining the impact of the largest shareholder on dividend payout dynamics. The influences of shareholders over the board are measured by equity ownership and voting power separately and then the results of the six regressions are presented accordingly. [Please insert Table 4 about here] Several findings about control factors are consistent across using proxies of equity ownership and voting power. First, two models (PAM and ETM) indicate that firm size plays a significant role in dividend dynamics. Positive coefficients indicate that larger firms are more likely to engage in compensating shareholders with cash dividends. Secondly, though statistically insignificant, the effect of leverage to dividend payout is consistently negative and the effect of Tobin s Q is consistently positive. The signs of the coefficients are consistent with the reasoning that while creditors discourage dividend payout to shareholders, a firm may increase dividends when its market value grows faster than the book value of its total assets. Thirdly, both industry and year effects are not significant for dividend dynamics. This implies that cash dividend provides a more stable 11

13 financial compensation to shareholders compared to other forms of compensation such as stock repurchase. Managers may be forced to pay out dividends even in poor operation years. The decision of stock repurchase, however, is more sensitive to a firm s profitability and timing to the market when firm stock is undervalued. Significant coefficients of variables in dividend models such as Earnings (E i,t ), Lagged dividend (D i,t-1 ), Lagged earnings (E i,t-1 ), and Earnings difference (E i,t E i,t-1 ) are aligned in the expected signs ( cr>0, -c<0, rj(1-c)>0, r>0). As far as the factors of earnings are concerned, the results indicate that firms consider more dividends with an increase in Earnings (E i,t ) and Lagged earnings (E i,t-1 ). The negative coefficients of D i,t-1 indicate a reversal adjustment to target payout ratio, implying that if dividends were aggressively conducted in the previous year, a lower compensation is expected in the current year. Comparing the dynamics of the interactions between earnings and ownership to the interactions between earnings and voting power on dividend policy, we find that the use of ownership as a proxy for shareholder influence over the board is responsible for some insignificant results in prior research. Stock holding size is a crude proxy for the strength of a particular shareholder and it ignores the relative control power of the shareholder. As shown in Table 4, when dividend dynamics are considered, the interaction variable between earnings in the current year and voting power (Earnings*voting power) is positively significant according to PAM, while according to ETM the interaction variable between voting power and the previous year s earnings (Earnings(t-1)*voting power) is positive and significant. By contrast, the interaction variables for earnings and ownership have no significant coefficients, implying that the largest shareholder is irrelevant to a 12

14 firm s payout decision. As we see, using the Bazhaf index to estimate a specific shareholder s control and voting power among all shareholders, we find that the presence of large shareholders reinforces the relationship between earnings and dividend payout. The largest shareholder asks for more dividend payout given the same level of firm earnings. Lastly, we observe a poor performance (negative R-squared) of the FAM, which assumes that a firm s desired payout target ratio reflects the changes in dividends (D) to the changes in firm earnings (E) and managers new dividend policy is based upon the consideration of whether current earnings E(t) have outperformed previous earnings E(t- 1). The empirical data is more aligned with the explanations of the PAM and the ETM. That is, managers new dividend policy reflects two factors: current earnings E(t) and previous dividend D(t-1). 5. Conclusions Although a plethora of articles have investigated issues related to dividend policy, ranging from its irrelevance (Miller and Modigliani, 1961) to the choice between the method of compensation (Allen, Bernardo and Welch, 2000), relatively few papers have investigated the relative importance of voting rights and equity ownership in determining the choice of dividend payout, particularly in U.S. firms. This paper thus contributes to finance literature by investigating whether large shareholders affect payout policy through their influence over the board. We find that large shareholders assert their power over the board to affect firm dividend policies. The computed voting power of a large 13

15 shareholder may be more important to reflect shareholder control than the size of the equity holding in business studies. 14

16 References Allen, F., Antonio, E.B. and Welch, I. (2000) Theory of Dividends Based on Tax Clienteles. Journal of Finance, 55: Bhattacharya, S. (1979) Imperfect Information, Dividend Policy and the Bird in the Hand Fallacy. Bell Journal of Economics, 10: Bhattacharya, S. (1980) Nondissipative Signaling Structure and Dividend Policy. Quarterly Journal of Economics, 95: Bloch, L. and Kremp, E.M. (1999) Ownership and Voting Power in France. FEEM Working Paper; Crespi, R. and Renneboog, L. Corporate Monitoring by Shareholder Coalitions in the UK. ECGI Discussion Paper no. 12/2003, ECGI, Brussels. Dlugosz, J., Fahlenbrach, R., Gompers, P. and Metrick, A. (2006) Large Blocks of Stock: Prevalence, Size and Measurement. Journal of Corporate Finance, 12: Easterbrook, F.H. (1984) Two Agency-Cost Explanations of Dividends. American Economic Review, 74: Eckbo, B.E, and Verma, S. (1994) Managerial Share Ownership, Voting Power, and Cash Dividend Policy. Journal of Corporate Finance, 1: Fama, E.F. and Babiak, W. (1968) Dividend Analysis: An Empirical Analysis. Journal of the American Statistical Association, 63: Goergen, M., Renneboog, L. and Correia da Silva, L. (2005) When Do German Firms Change Their Dividends? Journal of Corporate Finance, 11: Gugler, K. and Yurtoglu, B.B. (2003) Corporate Governance and Dividend Pay-Out Policy in Germany. European Economic Review, 47: Jensen, M.C. (1986) Agency Costs of Free Cash Flow, Corporate Finance and Takeovers. AEA Papers and Proceedings, 76: Leech, D. (2001) Shareholder Voting Power and Corporate Governance: A Study of Large British Companies. Nordic Journal of Political Economy, 27: Leech, D. and Manjon, M. (2003) Corporate Governance and Game Theoretic Analyses of Shareholder Power: The Case of Spain. Applied Economics, 35:

17 Linter, J. (1956) Distribution of Incomes of Corporations Among Dividends, Retained Earnings and Taxes. American Economic Review, 46: Miller, M, and Modigliani, F. (1961) Dividend Policy, Growth, and the Valuation of Shares. Journal of Business, 34: Miller, M., and Rock, K. (1985) Dividend Policy and Asymmetric Information. Journal of Finance, 40: Moh d, M.A., Perry, L.G. and Rimbey, J.N. (1995) An Investigation of the Dynamic Relationship Between Agency Theory and Dividend Policy. Financial Review, 30: Renneboog, L. and Trojanowski, G. Control Structures and Payout Policy. ECGI Finance Working Paper No. 80/2005. ECGI, Brussels. Rozeff, M.S. (1982) Growth, Beta and Agency Costs as Determinants of Dividend Payout Ratios. Journal of Financial Research, 5: Shapley, L.S. and Shubik, M. A. (1954) Method for Evaluating the Distribution of Power in a Committee System. American Political Science Review, 48: Shleifer, A. and Vishny, R. (1986) Large Shareholders and Corporate Control. Journal of Political Economy, 94: Short, H., Zhang, H. and Keasey, K. (2002) The Link Between Dividend Policy and Institutional Ownership. Journal of Corporate Finance, 8: Zeckhauser, R.J. and Pound, J. Are Large Shareholders Effective Monitors? An Investigation of Share Ownership and Corporate Performance. In: R. G. Hubbard, editor. Asymmetric Information, Corporate Finance, and Investment. University of Chicago Press: Chicago,

18 Table 1: Sample Characteristics Variable All Firms Positive EBIT Firms Mean Median Std Mean Median Std Amount spent on dividends to Common and Pref. Stocks Earnings Book value of the firm Leverage Tobin's Q Firm-years Note: All numbers are expressed in $ millions. Dividends are equal to the sum of dividends to common shares and preferred shares. Earnings are defined as EBIT in a particular year. The market value of a firm is computed as the sum of the market value of equity and the book value of total debt at the end of a given year. Firm size is defined as the book value of the total assets. Leverage is defined as the ratio of total debt to the book value of the total assets. Tobin's Q is defined as the ratio of the market value of the firm to the book value of the total assets. Table 2: Average Earnings Payout Ratios All Firms Positive EBIT Firms Year N Dividends N Dividends % % % % % % % % % % % % Total % % 17

19 Table 3: Distribution of Stock Ownership and Voting Power across Large Shareholders Panel A. Distribution of stock ownership across different types of block holders Variable Pool Affiliated entities Outside Block holders ESOPs Directors Officers Total Total firms in sample Panel B. Distribution of voting power across the largest three shareholders Largest block Affiliated entities ESOPs Directors Officers Outside Block holders % held Voting power nd largest block % held Voting power rd largest block % held Voting power Note: Voting power is measured by the normalized Banzhaf index. 18

20 Table 4: Effect of Large Shareholders on Dividend Dynamics Effect measured by Equity ownership Voting power PAM FAM ETM PAM FAM ETM Intercept Firm size **** **** **** **** Leverage Tobin's Q Earnings **** Lagged dividend **** **** **** **** Lagged earnings **** Earnings(t)-Earnings(t-1) Earnings * ownership [Earnings(t)-Earnings(t-1)]*ownership Earnings(t-1)*ownership Earnings*voting power **** [Earnings(t)-Earnings(t-1)]*voting power Earnings(t-1)*voting power **** Industry effects significant? No No No No No No Year effects significant? No No No No No No Adjusted R square N Note: *,**,***,**** indicate statistical significance at the 0.1,0.05,0.01 and level. Equity ownership is the stock ownership of the largest shareholder. Voting power is the normalized Banzhaf index of the largest shareholder. The three dividend models are defined as follows: PAM is the partial adjusted model; FAM is the full adjusted model; ETM is the earnings trend model.