The effect of dual class share structure on dividend payout policy

(1)

The effect of dual class share structure

on dividend payout policy

by Roy Luten

A thesis submitted in fulfilment of the requirements for the degree of Master of

Science in International Financial Management

Student number: s2012650

Supervisor: dr. Lammertjan Dam

Date of submission: 09-01-2015

(2)

1

Abstract

This thesis investigates the effect of disproportionate voting power held by insiders of dual class firms. While the expropriation hypothesis predicts that (controlling) shareholder-managers have both the opportunity and incentive to expropriate value from shareholders, the mitigation hypothesis states that shareholder-managers will actively attempt to reduce the agency costs brought about by a dual class share structure. A sample consisting of 545 US dual class firms in the years 1995-2002 is compared to a control sample of 11,979 single class firms. The results indicate that, while the dual class structure can increase the propensity to award more dividends, there are no significant differences between the dividend payout policy of dual class and single class firms.

JEL classification: G30, G34, G35

(3)

2

1. Introduction

This thesis investigates the effect of dual class share structures on the dividend payout behaviour of publicly traded firms in the US.* The issue of ownership and control of a company has been a widely researched topic since the first half of the last century, when Berle and Means (1932) released their seminal work The Modern Corporation and Private

Property. Their research advocated the importance of greater shareholder rights to ensure that

corporations were not run in the interests of just management. Since then, investor protection laws have been developed and implemented to protect shareholders from their inability to effectively control the actions of management. This principal-agent relationship has been at the centre of corporate governance research, and has been debated extensively since Jensen (1986) formalised his well-known agency theory. In his paper, Jensen discusses the conflicts of interest that can arise between shareholders and managers when free cash flow is substantial, which can lead to sub-optimal investments and contribute to organizational inefficiencies. In adopting a combination of both monitoring and incentives, shareholders can encourage managers to align their interests with those of shareholders, even when managers themselves are also shareholders. Whenever a manager-shareholder possesses a controlling stake in the firm, a substantial amount of capital invested in the firm should provide enough incentive to encourage a focus on shareholders’ wealth maximization and long-term value creation (Gomes, 2000).

However, dual-class share structures present a challenge to corporate governance practices. A dual-class structure means that a firm can issue more than one class of stock, with varying cash flow and voting rights for different classes of shares. In theory, this would make it less difficult for management to consolidate control and grant more room for discretionary actions, which might ultimately allow the firm to be run in a more successful manner. In practice, however, the consolidation of voting rights in the hands of management can promote entrenchment, and insulate managers from the discipline of the market for corporate control (DeAngelo and DeAngelo, 1985). In effect, managers also become insulated from shareholder disagreement because they no longer face the pro rata consequences of their actions, which opens up the possibility for management to exploit private benefits of control at the expense of the company and its investors.

*_{I would like to thank my thesis supervisor Lammertjan Dam for his time and effort, and Andrew Metrick}

(4)

3

How are investors enticed to invest in such a firm then, knowing they will not be able to control management’s actions?I argue that it is possible to interpret giving up voting rights as the ‘sale’ of voting rights to management, because shareholders that buy an inferior share class in effect give up a part of their control rights. If this interpretation is valid, it implies investors should receive higher cash flows in the form of dividends or share repurchases in return. Classic literature has typically posited that firms have no reason to pay cash dividends, or that a firm’s dividend policy comes down to indifference (Stiglitz, 1973; Miller and Modigliani, 1961). Following pecking order theory, firms would prefer to include more debt in their capital structure rather than more equity, because debt issuance does not dilute management control and interest payments are tax deductible (Myers and Majluf, 1984). Due to the increasing risk of bankruptcy issuing equity to raise capital becomes preferable, which in turn increases the firm’s propensity to pay dividends to please shareholders. However, in the case of firms with a dual class share structure, there are more factors which influence the propensity to pay dividends.

(5)

4

Using a dataset comprised of 545 US dual class firms and 11,979 US single class firms that spans from 1995 to 2002, the results indicate that the dividend payout ratio for dual class firms actually tends to be higher than for single class firms. Consistent with the literature, firm size as measured by total assets is found to be an important predictor of the propensity of firms to pay dividends (Fama and French, 2001); in a similar vein, firms displaying high asset growth over the past three years displayed a tendency to reduce the amount of dividends they pay. Net executive compensation has also been included in the regressions, but has no significant effect on the dividend payout ratio. This contrasts studies where CEO compensation is used rather than total executive compensation, and indicates that at a broader level, executive compensation does not negatively influence cash flows for shareholders as much as previously thought. As such, this study shows that dual class share structures do not necessarily lead to expropriation. In addition, when using a sample of only dual class firms, the effect of dividends received by insiders is economically and statistically stronger than the effect of executive compensation, showing that shareholders can even benefit from a dual class structure, while shareholder-managers do create personal gains in the process.

(6)

5

2. Literature review and hypotheses

The starting point for research on the effects of a different type of control structure is agency theory (Jensen, 1986), which posits that agency costs arise because owners (or the ‘principals’) seek to ensure that managers (or ‘agents’) act in their interests, even though managers may feel more inclined to look after their own interests first. The problem is further compounded by the presence of asymmetric information, because managers have more complete information about the firm and its investment opportunities than outside shareholders do. For example, while managers have a sufficient amount of information to determine which investments would create maximum net present value for the firm, they may instead decide to invest free cash flows in a project that is sub-optimal for the firm, but is better for their career. Because managers can have both the opportunity and the incentive to expropriate value in such a way, shareholders seek ways to ensure that managers maximize their wealth without engaging in opportunistic or self-serving behaviour. Seeing as complete contracts are impossible to write due to informational asymmetries and bounded rationality, interest alignment should be reached through either incentives or monitoring, and agents that do not fall in line can be disciplined by the market for corporate control. In some cases, managers also possess a controlling stake in the company, which can pose a challenge for non-controlling shareholders. According to Gomes (2000), conflicts between controlling and non-controlling shareholders can result in additional agency costs, because minority shareholders may not be sufficiently protected by the legal system, or because the controlling shareholder is protected by the corporate governance structure. However, Gomes shows that managers with a controlling stake may not have any incentive to expropriate wealth from the firm, even if there is a poor corporate governance structure. The reason for this is that managers with a controlling stake are likely to have a substantial amount of capital invested in the firm, providing a strong incentive to maximize firm value and shareholder wealth in the long run.

(7)

6

rights and their cash flow rights. According to Gompers et al. (2010), around 6 percent of all publicly traded firms in Compustat feature stock with both an inferior and superior share class, representing 8 percent of the total stock market capitalization. Dual class firms are characterized by a high degree of insider ownership, as well as a strong divergence between voting rights and cash flow rights (Francis et al., 2005; Jordan et al., 2014). Rather than a one-to-one ratio of both voting rights and cash flow rights, insiders in dual class companies typically possess less than 40 percent of the cash flow rights, while possessing 60 percent of the voting rights (Gompers et al., 2010; Amoako-Adu et al., 2014). Indicative of how important control is, Smart et al. (2008) document that managers are willing to receive a price discount during the initial public offering of the dual class firm, in order to retain control of the company.

(8)

7

cash flow rights and voting rights grows, firm value decreases. Similarly, as the divergence grows, chief executives of dual class firms receive significantly higher remuneration, acquisitions are more likely to destroy value, and capital expenditures ultimately provide a lower return for shareholders (Masulis et al., 2009). This is further supported by Amoaku-Adu et al. (2011), who show that controlling shareholder-managers of dual-class firms receive more compensation than controlling shareholder-managers in single-class firms, and this excess compensation is usually awarded in the form of bonuses and stock options at the cost of distributed dividends. The lower the equity stake of controlling shareholders, the higher their share of excess compensation tends to become compared to actually distributed dividends (Mancinelli & Ozkan, 2006; Gugler & Yurtoglu, 2003). In short, dual class share structures have been shown to promote expropriating behaviour at the cost of outside shareholders, especially when managers are also controlling shareholders. As such, I test the following set of hypotheses:

H0: insider voting power does not significantly reduce the dividend payout ratio.

H1: insider voting power significantly reduces the dividend payout ratio.

(9)

8

encourages higher bidding prices, benefitting the existing shareholders (Burkart et al., 1998). Finally, DeAngelo and DeAngelo (1985) assert that dual class share structures can encourage managers to invest in firm-specific human capital, leading to more long-term growth. To summarise, even though dual class share structure can create opportunities for expropriation and increase agency costs, outside shareholders could also benefit from higher dividend payments and higher firm value. Therefore, the second set of hypotheses to be tested is the following:

H2: insider voting power does not significantly increase the dividend payout ratio.

H3: insider voting power significantly increases the dividend payout ratio.

(10)

9

3. Data and methodology

Sample construction

I acquire a dataset of dual class companies from Andrew Metrick, co-author of Gompers et al. (2010), which contains company-specific information such as the number of share classes, the dividends and voting rights assigned to shares, and how many dividends and voting rights insiders receive. Full documentation with regard to the creation of this dataset is included in their article. Initially, there are 741 unique firms with a total of 3,730 firm-year observations. Initial trimming takes place by removing companies from the list that do not have GVKEY codes or Compustat ticker symbols, which are required in order to be able to access their data in Compustat.

The years 1990 up to and including 1994 do not have enough observations to warrant retaining these years in the data set, as they would produce rather incomparable results compared to the other years. For the purposes of this research, I focus solely on companies with exactly two classes of stock, i.e., both a superior and inferior class of stock. Companies with more or less than two classes of stock are removed from the sample. Companies which do not have reported voting rights per share in the dataset are also removed, because it implies that one of the main variables of interest is not available. A very small number of firm years (around 1%) in the sample show a number of outstanding shares of either Class A or Class B shares which is lower than the number of shares that insiders possessed. Clearly such numbers are not realistic, which may be due to faulty input during the original dataset creation. Because these firms constitute such a small part of the total dataset, they have been omitted to avoid introducing statistical errors.

(11)

10

SIC codes starting with 99 are also present in the sample at this point, meaning they are difficult to classify for the governing body; more often than not, they are financial holdings. In line with past research (e.g. Masulis et al., 2009; Gompers et al., 2010; Jordan et al. 2014) and to prevent introducing statistical inconsistencies, these firms are removed from the sample.

In addition to the dual class sample, a control sample comprised of publicly traded single-class firms is created by searching the entire Compustat database for the years 1995 up to and including 2002. Firms with GVKEYs that appear in the initial list of dual class companies are removed from the single class firm list, because such firms have more than one class of shares. This leaves the single class sample with an initial count of 17,234 unique firms, and 92,836 firm-year observations. As with the dual class sample, firms with SIC codes 60-67, 99, and 49 are removed from the sample to prevent introducing biases into the dataset.

Using Compustat, I download data for the following variables: total assets, total liabilities, the book value of common stock, annualized closing stock prices, dividends paid per share, and number of shares outstanding. I use these variables to calculate factors such as Tobin’s Q, dividend payout ratios, return on assets, leverage, and the 3-year asset growth. Consistent with prior research, a ‘zero voting premium’ is realistic for non-traded classes of stock in dual class firms. Using a robustness check, Gompers et al. (2010) demonstrate that even a 10% increase in the share price of the non-traded class does not affect their results significantly, because the non-traded shares only make up a small part of the total capital structure of a firm. Consistent with Ang and Megginson (1989), I add a 5.4% voting premium to the price of the publicly traded superior voting shares for which I could not obtain the data from CRSP.

Empirical design

In order to research the relationship between a firm’s dividend payout ratio and the voting power held by insiders, I estimate a number of equations. By employing a stepwise approach and gradually adding more variables, it becomes less difficult to discern what effects certain variables have on the results. To start with, I first estimate a baseline regression that makes no distinction between dual class and single class firms to assess the uncontrolled effect of insider voting power on the dividend payout ratio:

(12)

11

where Dividend payout ratio is the ratio of total dividends divided by the total net income paid in a firm-year, Insider voting power is the percentage of total shares held by the insiders of the firm, α is the intercept, ε is an error term, subscript i represents a firm, and subscript t represents a year. By choosing the ratio of total dividend payments to net income as the dependent variable, I can approximate the effect of insider voting power on a firm’s general dividend payout policy. The reason for using a ratio rather than an absolute number as the dependent variable, is because the total amount of dividends also acts as a measure of size. Together with the total amount of assets included later as a control variable for firm size, this can erroneously inflate the predictive power of the regressions. The relationship makes sense intuitively, because the amount of dividends paid can be influenced by the managers of the firm, who may or may not decide to utilize their private benefits of control. By using net income as the denominator, cash flows from interest are automatically excluded. I found that the data for single class firms suffers from heteroskedasticity and that there is a certain amount of variance between observations of different firms, but not between observations of the same firm. In other words, the residuals of the cross-sections appear to be correlated. Intuitively this makes sense, as total dividend payments by the same firm tend to stay the same each year, for example. As such, I decide to employ White period standard errors to account for the issues heteroskedasticity may cause (White, 1980).

Next, I introduce a distinction between dual class and single class stock into the equation. To do so, I create a dummy variable for both classes of stock, which allows me to capture the effect of insider voting power on the dividend payout policy of dual class and single class firms, respectively. The regression equation looks as follows:

(2) Dividend payout ratio i,t = α + β1 * Insider voting power i,t * Class i,t + β2 *

Insider voting power i,t * (Class i,t - 1) + β3 * Class i,t + ε i,t,

(13)

12

improve the explanatory power of the regression. Including the control variables leads to following regression equation:

(3) Dividend payout ratio i,t = α + β1 * Insider voting power i,t * Class i,t + β2 *

Insider voting power i,t * (Class i,t - 1) + β3 * Class i,t + Χ i,t + ε i,t,

where Χ is a vector of control variables that captures firm characteristics, some of which are theorized to affect a firm’s dividend payout ratio. According to Fama and French (2001) and DeAngelo et al. (2006), firms are more likely to award dividends when they are larger, and have higher book-to-market ratios. As a measure of firm size, the natural logarithm of the amount of total assets is used, rather than the natural log of sales as has been done before in past research (e.g., Amoako-Adu et al. 2014). The reason for this is that it is not possible to include the natural log of total sales for firm-years where sales are negative, which would bias the sample by only considering more profitable companies. However, it is still important to consider that some firms are more profitable than other firms, which is why return on assets (defined as net income divided by total assets) is also included as a control variable.

(14)

13

Growth opportunities become more abundant for firms as shareholders value them more highly, providing the company with more capital which it can use to invest in its business. Intuitively then, it makes sense to measure growth opportunities using Tobin’s Q, which can reliably be measured using a model similar to that of Kaplan and Zingales (1997) and Gompers et al. (2010):

(ln) Tobin′_{s Q = (ln)}BV of total assets + (MV of equity − BV of equity) BV of total assets .

Most importantly, the book value of assets is used as the denominator here rather than the replacement cost of assets, because that would be too difficult to estimate reliably. The more growth companies experience and the more growth opportunities they are presented with, the higher the opportunity cost becomes if they decide to award more dividends (Jordan et al., 2014).

(15)

14

A final control variable is added to control for industry selection bias. More specifically, media companies tend to adopt dual class share structures much more often than companies in other industries (DeAngelo and DeAngelo, 1985; Smart and Zutter, 2003), because control over a media company can provide substantially more opportunities to consume private benefits. Firms with SIC codes 2710, 2711, 2720, 2721, 2730, 2731, 4830, 4832, 4833, 4840, 4841, 7810, 7812, and 7820 are defined as media companies.

(16)

15 Table 1: descriptive statistics for dual class and single class firms

Dual class (N = 2,562) Single class (N = 61,381)

Obs. Mean Median Max. Min. Std. Dev. Obs. Mean Median Max. Min. Std. Dev. Dividend payout ratio 2,550 0.36 0.00 194.65 -69.43 5.66 57,847 0.06 0.00 1,787.98 -3,309.48 15.87 Insider voting power 2,562 0.60 0.64 1.00 0.00 0.27 56,609 0.01 0.00 0.89 0.00 0.05 (ln) Total assets 2,555 19.67 19.79 26.39 10.00 1.83 57,953 18.19 18.18 26.64 6.91 2.57 Return on assets 2,555 -0.06 0.03 2.45 -13.69 0.59 57,953 -1.15 0.01 1,827.00 -6,309.00 47.04 Leverage 2,543 0.56 0.55 4.45 0.02 0.33 57,953 2.04 0.52 8,233.00 0.00 65.74 (ln) Tobin's Q 2,551 0.64 0.53 5.42 -1.57 0.52 56,549 0.70 0.54 12.31 -8.87 0.82 Total asset growth 2,373 0.57 0.11 201.82 -0.54 5.61 40,804 3.70 0.12 32,527.66 -0.91 290.47 Media 2,562 0.11 0.00 1.00 0.00 0.32 61,380 0.02 0.00 1.00 0.00 0.13 (ln) Executive compensation 858 15.48 15.39 19.03 13.12 0.92 11,181 15.38 15.26 20.86 9.14 1.06 ∆ Executive compensation 824 101.89 30.50 13,605.59 -92.46 543.44 10,514 141.83 32.34 298,292.22 -94.76 2,959.52 Insider dividends 966 0.36 0.33 0.91 0.01 0.21 9,958 0.02 0.00 0.87 0.00 0.06

(17)

16

4. Results

Table 2 displays the results for baseline regression (1), which contains no control variables or interactions and is thus only a simple check to see whether the regression intuitively makes sense. At a confidence level of 5%, insider voting power significantly affects the dividend payout ratio, which signifies that the relationship between dividend payout ratio and insider voting power does make sense statistically. With a positive sign, Insider voting power indicates that it increases the amount of dividends paid, which supports the mitigation hypothesis. Of course, without any other variables, the explanatory power of this regression is low.

Table 2: the uncontrolled effect of insider voting power on the dividend payout ratio.

Variable Coefficient p-Value

Intercept 0.061 0.374

Insider voting power 0.506 0.034**

R² 0%

The sample is comprised of 63,943 firm-years, with a total of 58,819 panel observations. The contents of the columns are self-explanatory. Please refer to Appendix A for variable definitions. The p-values are based on standard errors, adjusted for heteroskedasticity (White, 1980), and no year or industry fixed effects have been used. p-Values with *, **, and *** are significantly different at 10%, 5%, and 1% confidence intervals, respectively. Sources: data on voting power is provided by Andrew Metrick, co-author of Gompers et al. (2010). Data on executive compensation and insider holdings for single class firms is obtained from Execucomp. All other data is obtained from Compustat.

Table 3 contains the results for regression (2), where a distinction is made between the effect

Insider voting power has on the dividend payout ratio for both dual class and single class

(18)

17

Table 3: the effect of insider voting power on the dividend payout ratio, controlling for share classes.

Insider voting power * Class 0.725 0.075*

Insider voting power * (Class-1) -0.273 0.407

Class -0.141 0.395

R² 0%

Adding control variables to the specification leads to the results in Table 4. Both the interaction variable for dual class firms and the dummy variable for the firm category are significant, which indicates that the distinction between dual class and single class firms becomes important when control variables are added. Still, the interaction variable Insider

voting power * (Class-1) remains statistically insignificant, and reinforces the belief that

voting power held by insiders is a much less important predictor of dividend payments in single class firms. The measure of size, (ln) Total assets, is positive and statistically significant, following the prediction that larger firms are more likely to award dividends (Fama and French, 2001). In addition, Total asset growth appears to be positive and statistically significant, which indicates that an increase in asset growth would increase the dividend payout ratio. However, that would not really make sense, because the theory predicts the opposite. Because the coefficient is not economically significant at all, this variable cannot be considered an important predictor of dividend payout policies. The variables Return on

assets and Leverage are both statistically and economically insignificant, which is especially

(19)

18

variable), this selection bias does not affect the results significantly. In general, the results from this specification suffer from a lack of economic significance, which tells that even though the variables might be important predictors, it is not possible yet to make any definitive statements concerning their predictive power. As noted earlier, the data on executive compensation and dividends received by insiders as measured by Compustat is limited, and is therefore not added at first to the specification of regression (3). However, doing so changes the results dramatically.

Table 4: the controlled effect of insider voting power on the dividend payout ratio, excluding executive compensation.

Intercept -0.852 0.000***

Insider voting power * Class 0.825 0.077*

Insider voting power * (Class-1) 0.500 0.124

Class -0.325 0.066*

(ln) Total assets 0.055 0.000***

Return on assets 0.000 0.595

Leverage 0.000 0.132

(ln) Tobin’s Q 0.005 0.741

Total asset growth 0.000 0.032**

Media -0.117 0.219

R² 0%

(20)

19

power, Class, (ln) Total assets and Total asset growth have completely disappeared. As a

whole, these results seem to suggest two things: first off, there are variables which I suspect contain a substantial amount of noise, which would not be surprising. For example, what exactly determines the amount of compensation an executive receives is different for every firm, and is also influenced by industry differences, among other things. Secondly, attempting to capture the effect of corporate governance data is very difficult, not only because it is still unclear exactly how governance practices influence firm performance, but also because such data is difficult to measure. As they are, the results from Table 5 imply that I cannot reject the null hypotheses for either the expropriation theory or the mitigation theory, because there are simply no significant positive or negative effects.

Table 5: the controlled effect of insider voting power on the dividend payout ratio, including executive compensation.

Insider voting power * Class -3.520 0.259

Insider voting power * (Class-1) 5.215 0.211

Class 0.592 0.580

(ln) Total assets -1.381 0.328

Return on assets 2.824 0.381

Leverage 21.140 0.286

(ln) Tobin’s Q 2.547 0.307

Total asset growth -0.003 0.204

Media -1.796 0.250

(ln) Executive compensation 0.414 0.373

∆ Executive compensation 0.000 0.568

Insider dividends 5.245 0.174

R² 1.5%

(21)

20

though the mean dividend payout ratio of dual class firms would appear to be substantially higher at first glance.

Table 6: t-test of equality for the mean of the dividend payout ratio.

Variable Mean Std. Dev.

Single class dummy 0.064 15.874

Dual class dummy 0.356 5.658

t-Value -0.926

p-Value 0.354

The sample is comprised of 60,397 observations. The contents of the columns are self-explanatory. Sources: data on voting power is provided by Andrew Metrick, co-author of Gompers et al. (2010). Data on executive compensation and insider holdings for single class firms is obtained from Execucomp. All other data is obtained from Compustat.

Because earlier results in Table 4 show that the effect of insider voting power appeared to be stronger for dual class firms, and because the literature indicates that agency theory is less applicable when voting rights and cash flow rights diverge, I decide to conduct an additional robustness test by running the regression for only the dual class firms in the sample. The results, seen in Table 7, indicate that they suffer from the same problems mentioned earlier (e.g. noise, measurement errors), although they suffer less severely compared to the results in Table 5. Nonetheless, because Insider voting power is still statistically insignificant, I definitively cannot reject the null hypotheses for both the expropriation (H0) or mitigation

(H2) theories, or accept the alternative hypotheses H1 or H3. Considering the prior research

that found predominantly negative effects (e.g. Masulis et al., 2009; Gompers et al., 2010), this result seems out of line. However, what it indicates is that agency theory is still useful in reaching interest alignment in dual class firms, even though there may be a strong divergence between voting power and cash flow rights. Consistent with the literature that predicts the expropriation hypothesis, executive compensation has a significantly negative effect on dividend payouts, because free cash flow within the firm is redirected to the benefit of managers. In addition, Insider dividends has an even stronger effect in this regression than (ln)

Executive compensation, both economically and statistically. This is in line with the

(22)

21

Table 7: the controlled effect of insider voting power on the dividend payout ratio of dual class firms.

Intercept 7.008 0.003***

Insider voting power -0.924 0.112

(ln) Total assets -0.095 0.429

Return on assets -1.107 0.549

Leverage 0.916 0.218

(ln) Tobin’s Q 0.346 0.341

Total asset growth -0.459 0.380

Media -0.255 0.394

(ln) Executive compensation -0.353 0.091*

∆ Executive compensation 0.002 0.149

Insider dividends 2.620 0.004***

R² 2.35%

The sample is comprised of 2,562 firm-years, with a total of 478 panel observations. The contents of the columns are self-explanatory. Please refer to Appendix A for variable definitions. The p-values are based on standard errors that have not been adjusted for heteroskedasticity, and no year or industry fixed effects have been used. p-Values with *, **, and *** are significantly different at 10%, 5%, and 1% confidence intervals, respectively. Sources: data on voting power is provided by Andrew Metrick, co-author of Gompers et al. (2010). Data on executive compensation and insider holdings for single class firms is obtained from Execucomp. All other data is obtained from Compustat.

(23)

22

5. Conclusion

This thesis examines the effects of insider voting power on the dividend payout ratio of both dual class and single class firms, and aims to determine whether the divergence between the voting rights and cash flow rights of insiders makes agency theory less applicable. The initial regressions seem to be in line with the mitigation hypothesis, which predicts that dual class firms will display a higher propensity to pay more dividends. The inclusion of executive compensation data reveals that there are no significant differences between the effect of insider voting power on dividend payout ratios of either dual class or single class firms, which implies that the difference in control structures has no notable effects on the dividend policy of firms and that incentives are equally strong among both dual class and single class firms. It is only when considering a sample consisting of exclusively dual class firms that the effects predicted by both the expropriation and mitigation hypotheses become noticeable; while shareholder-managers have an incentive to expropriate value from shareholders through excessive compensation, their investment in the firm creates a stronger incentive to instead award more dividends, which is in the interest of all shareholders and increases the insiders’ personal gains at the same time.

In summary, while the effects of a dual class share structure do influence the dividend payout ratio, it is not significantly affected by the amount of voting power held by insiders, and does not differ strongly from single class firms in the control sample. This means that agency theory, which recommends a combination of both incentives and monitoring to achieve interest alignment, is effective for both dual class firms and single class firms. In dual class firms where shareholder-managers have a substantial amount of cash flow rights, outside shareholders may receive higher dividend payments than they would as shareholders of single class firms, although it would not be significantly different. However, these relatively short-term gains come at the cost of long-short-term value creation, because cash is still being redirected from potential investments to dividend payments.

(24)

23

simply would not have been possible within the time frame of this thesis. Secondly, dividends themselves do not completely capture the cash flows received by shareholders. Share repurchases are also a possible source, but these fall outside the scope of this thesis because they are difficult to identify using Compustat, and can be subject to measurement or estimation errors when calculated manually. Thirdly, the data on insider voting power for single class firms is obtained from Execucomp, which is the only source where this kind of data is readily stored. The problem is that this means the insider voting power is only measured based on executives, which means other insiders with voting power are not included in this definition. Again, the alternative would have been to gather such data by hand, for which I do not have the time or human resources. Finally, Execucomp is quite incomplete compared to the normal Compustat database, which results in very different sample sizes for the firm-level data and the executive-level data. As discussed in section 4, these differences in the number of observations do impact the results of the regressions.

(25)

24

References

Amoako-Adu, B., Baulkaran, V & Smith, B., 2011. Executive compensation in firms with concentrated control: the impact of dual class structure and family management, Journal of

Corporate Finance, vol. 17, no. 5, pp. 1580-94.

Amoako-Adu, B., Baulkaran, V & Smith, B., 2014. Analysis of dividend policy of dual and single class U.S. corporations, Journal of Economics and Business, vol. 72, no. C, pp. 1-29.

Ang, J. & Megginson, W., 1989. Restricted voting shares, ownership structure, and the market value of dual-class firms, The Journal of Financial Research, vol. 12, no. 4, pp. 301-18.

Bebchuk, L. & Cohen, A., 2005. The costs of entrenched boards, Journal of Financial

Economics, vol. 78, no. 2, pp. 409-553.

Bebchuk, L. Kraakman, R. & Triantis, G., 2000. Stock pyramids, cross-ownership and dual class equity: The Mechanisms and agency costs of separating control from cash-flow rights, in Morck, R. (ed.), Concentrated Corporate Ownership, The University of Chicago Press, Chicago, IL.

Berle, A. & Means, G., 1932. The Modern Corporation and Private Property, Macmillan, New York.

Burkart, M., Gromb, D. & Panunzi, F., 1998. Why higher takeover premia protect minority shareholders, Journal of Political Economy, vol. 106, no. 1, pp. 172-204.

Chemmanur, T. and Jiao, Y., 2012. Dual class IPOs: A theoretical analysis, Journal of

Banking and Finance, vol. 36, no. 1, pp. 305-19.

(26)

25

DeAngelo, H., DeAngelo, L. & Stulz, R., 2006. Dividend policy and the earned/contributed capital mix: a test of the life-cycle theory, Journal of Financial Economics, vol. 81, no. 2, pp. 227-54.

Easterbrook, F., 1984. Two agency cost explanations of dividends, American Economic

Review, vol. 74, no. 4, pp. 650-9.

Faccio, M., Lang, L. & Young, L., 2001. Dividends and expropriation, American Economic

Review, vol. 91, no. 1, pp. 54-78.

Fama, E. & French, K., 2001. Disappearing dividends: changing firm characteristics or propensity to pay?, Journal of Financial Economics, vol. 60, no. 1, pp. 3-43.

Fama, E. & Jensen, M., 1983. Separation of ownership and control, Journal of Law and

Economics, vol. 26, no. 2, pp. 301-25.

Farinha, J., 2003. Dividend policy, corporate governance and the managerial entrenchment hypothesis: An empirical analysis, Journal of Business Finance and Accounting, vol. 30, no. 9-10, pp. 1173-1209.

Francis, J., Schipper, K. & Vincent, L., 2005. Earnings and dividend informativeness when cash flow rights are separated from voting rights, Journal of Accounting and Economics¸ vol. 39, no. 2, pp. 329-60.

Gomes, A., 2000. Going public without governance: Managerial reputation effects, The

Journal of Finance, vol. 55, no. 2, pp. 615-46.

Gompers, P., Ishii, J. & Metrick, A., 2010. Extreme governance: an analysis of dual class firms in the United States, Review of Financial Studies, vol. 23, no. 3, pp. 1051-88.

Grossman, S. & Hart, O., 1988. One share/one vote and the market for corporate control,

(27)

26

Gugler, K. & Yurtoglu, B., 2003. Corporate governance and dividend pay-out policy in Germany, European Economic Review, vol. 47, no. 4, pp. 731-58.

Jensen, M., 1986. Agency cost of free cash flow, corporate finance and takeovers, American

Economic Review, vol. 76, no. 2, pp. 323-9.

Jordan, B., Liu, M. & Wu, Q., 2014. Corporate payout policy in dual-class firms, Journal of

Corporate Finance, vol. 26, no. C, pp. 1-19.

Kaplan, S. & Zingales, L., 1997. Do investment-cash flow sensitivities provide useful measures of financing constraints?, Quarterly Journal of Economics, vol. 112, no. 1, pp. 169-216.

La Porta, R., Lopez-de-Silanes, F. and Shleifer, A., 1999. Corporate ownership around the world, The Journal of Finance, vol. 54, no. 2, pp. 471-516.

Lehn, K., Netter, J. & Poulsen, A., 1990. Consolidating corporate control: Dual class recapitalization versus leveraged buyouts, Journal of Financial Economics, vol. 27, no. 2, pp. 557-80.

Mancinelli, L. & Ozkan, A., 2006. Ownership structure and dividend policy: evidence from Italian firms, The European Journal of Finance, vol. 12, no. 3, pp. 265-82.

Masulis, R., Wang, C. & Xie, F., 2009. Agency problems at dual class companies, The

Journal of Finance, vol. 58, no. 4, pp. 1445-68.

Miller, M. and Modigliani, F., 1963. Dividend policy growth and the valuation of shares,

Journal of Business, vol. 34, no. 4, pp. 411-33.

(28)

27

Rozeff, M., 1982. Growth, beta and agency costs as determinants of dividend payout ratios,

Journal of Financial Research, vol. 5, no. 3, pp. 249-59.

Ruback, R., 1988. Coercive dual-class exchange offers, Journal of Financial Economics, vol. 20, no. 1-2, pp. 153-73.

Smart, S., Thirumalai, R. & Zutter, C., 2008. What’s in a vote? The short- and long-run impact of dual class equity on IPO firm values, Journal of Accounting and Economics, vol. 45, no. 1, pp. 94-115.

Smart, S. & Zutter, C., 2003. Control as a motivation for underpricing: a comparison of dual and single-class IPOs, Journal of Financial Economics, vol. 69, no. 1, pp. 85-110.

Smart, S., Zutter, C. & Thirumalai, R., 2007. Why dual-class shares don’t add up, NACD

Directorship, vol. 33, no. 4, pp. 14-15.

Stiglitz, J., 1973. Taxation, corporate financial policy and the cost of capital, Journal of

Public Economics, vol. 12, no. 3, pp. 1-34.

(29)

28

Appendix A: variable definitions

Variable Definition

Dividend payout ratio The total amount of dividends awarded during a year, divided by the total net income from the same year.

Insider voting power The percentage amount of votes insiders possess.

Class A dummy variable that indicates whether a firm is a dual class or single class firm.

(ln) Total assets The natural log of total assets.

Return on assets Profitability ratio, calculated as total net income divided by total assets.

Leverage Leverage ratio, calculated as total liabilities divided by total assets.

(ln) Tobin’s Q The natural log of the market value of assets, divided by the book value of assets.

Total asset growth The three-year average growth in total assets.

Media A dummy variable that indicates whether a firm belongs to an industry associated with the media, i.e., SIC codes 2710, 2711, 2720, 2721, 2730, 2731, 4830, 4832, 4833, 4840, 4841, 7810, 7812, and 7820.

(ln) Executive compensation The natural log of total executive compensation. ∆ Executive compensation The year-to-year percentual change in total executive

compensation.

(30)

29

Appendix B: correlations between all variables

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)

∆ Executive compensation (1) 1

Observations ₁₁₃₃₈

(ln) Executive compensation (2) 0.057 1

Observations ₁₁₃₃₈ ₁₂₀₃₉

Dividend payout ratio (3) -0.001 0.007 1

Observations ₁₁₃₂₆ ₁₂₀₁₆ ₆₀₃₉₇

Insider dividends (4) -0.026 -0.167 0.005 1

Observations ₅₃₀₀ ₅₃₇₉ ₁₀₈₈₈ ₁₀₉₂₄

Insider voting power (5) -0.001 -0.066 0.004 0.926 1

Observations ₁₁₃₀₁ ₁₁₉₅₈ ₅₈₈₁₉ ₁₀₉₂₄ ₅₉₁₇₁ Leverage (6) _-0.009 _0.099 _0.000 _-0.058 _-0.006 ₁ Observations ₁₁₃₂₄ ₁₂₀₁₇ ₆₀₁₅₈ ₁₀₉₁₆ ₅₈₉₄₉ ₆₀₄₉₆ Return on assets (7) 0.000 0.050 0.000 -0.008 0.006 -0.402 1 Observations ₁₁₃₃₂ ₁₂₀₂₆ ₆₀₁₇₀ ₁₀₉₁₈ ₅₈₉₆₁ ₆₀₄₉₆ ₆₀₅₀₈ Media (8) -0.001 0.048 0.001 0.152 0.139 -0.001 0.002 1 Observations ₁₁₃₃₈ ₁₂₀₃₉ ₆₀₃₉₇ ₁₀₉₂₄ ₅₉₁₇₁ ₆₀₄₉₆ ₆₀₅₀₈ ₆₃₉₄₂

Total asset growth (9) 0.002 0.002 0.000 0.001 -0.003 0.000 0.000 -0.001 1

Observations ₁₀₈₄₄ ₁₁₂₄₄ ₄₃₀₈₆ ₁₀₀₁₆ ₄₂₆₆₉ ₄₃₁₃₄ ₄₃₁₄₅ ₄₃₁₇₇ ₄₃₁₇₇

(ln) Tobin’s Q (10) 0.029 0.179 -0.002 0.015 -0.001 0.102 -0.127 -0.012 0.000 1

Observations ₁₁₃₁₄ ₁₂₀₀₈ ₅₈₇₈₁ ₁₀₉₁₆ ₅₇₅₆₁ ₅₉₀₈₈ ₅₉₁₀₀ ₅₉₁₀₀ ₄₂₃₅₁ ₅₉₁₀₀

(ln) Total assets (11) 0.00 0.66 0.01 -0.09 0.15 -0.09 0.08 0.06 0.00 -0.23 1

Observations ₁₁₃₃₂ ₁₂₀₂₆ ₆₀₁₇₀ ₁₀₉₁₈ ₅₈₉₆₁ ₆₀₄₉₆ ₆₀₅₀₈ ₆₀₅₀₈ ₄₃₁₄₅ ₅₉₁₀₀ ₆₀₅₀₈

(31)

30

Appendix C: development of the amount of firms with a dual class share structure

0 50 100 150 200 250 300 350 400 1995 1996 1997 1998 1999 2000 2001 2002