The effect of dual-class ownership on
market valuation
Name: Martijn Vreeker
Student number: 10649980
BSc. in: Finance & Organization Supervisor Rafael P. Ribas
Date: 19-6-2016
Abstract
This study examines the effect of dual-class ownership on market valuation. Panel data are tested in consideration of two main effects. The null-hypothesis suggests that dual-class ownership structures have a value enhancing effect. On the other side, the alternative hypothesis states that a dual-class structure decreases market valuation due to managerial entrenchment. Using Tobin’s Q as a proxy for market valuation, I find evidence for a negative relationship. Therefore the null-hypothesis is rejected, which suggests that the entrench-ment theory holds. Theoretically this implies that one should not invest in dual-class companies. However, market valuation is influenced by many other factors, and a dual-class structure might also have positive effects. Therefore, no hard recommendations can be made.
1 Statement of Originality
This document is written by student Martijn Vreeker who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2 Table of contents
1. Introduction………... 3
2. Effects of a dual class structure……… 5
2.1 Entrenchment Theory……… 5
2.2 Value Enhancement Theory………. 6
2.3 Other Theories..………... 7
3. Empirical Strategy……….……….. 8
3.1 Data……… 8
3.2 Model and Hypotheses………... 11
3.3 Used variables……….. 12 3.4 Reverse Causality……… 13 3.5 Summary Statistics………. 14 4. Results……… 17 4.1 Empirical results………. 17 4.2 Discussion………. 20 5. Conclusion……….. 21 6. References……….. 22
3 1. Introduction
On the 28th of April 2016, Eric Lam posted an article on Bloomberg regarding the
ownership structure of Bombardier Inc. and Alimentation Couche-Tard Inc. While having the same ownership structure, Bombardier underperforms heavily in comparison to Alimentation. The 10-year return on their shares are minus 5% and plus 20% respectively. Their peers have shown an average 12% return over de last ten years, while their counterparts only generated 7,1%. Lam mentioned these companies to argue that their ‘dual-class structure’ can work out in either way, magnifying the effect on the managers’ performance and therefore firm valuation. Evidence to this argument is ambiguous. On one hand, literature finds that a dual-class structure decreases firm value due to entrenchment. On the other hand, a value enhancing effect is observed as well.
This thesis examines the impact of a dual-class ownership structure on firm value. The main objective is to solve the puzzle in existing literature.
Empirical evidence should clarify whether the entrenchment theory or the value enhancement theory is observed in the long-term. As a secondary objective, legislators and shareholders might obtain guidance on how to deal with such ownership structures. The tested hypothesis states that dual class structures have a value enhancing effect on market valuation and therefore increase shareholder value. The alternative hypothesis suggests that the entrenchment theory is applicable, which implies that market valuation decreases with a switch to a dual-class structure.These hypotheses are tested using panel data of North American firms. In a linear model, Tobin’s Q is regressed on the dual class status to estimate its effect.
Within a dual-class structure, a company has two classes of stock with unequal voting rights. In general, the class A stock has 10 votes per share and the Class B stock has only 1 (Gompers et al., 2010). The class A stocks are in the hands of the management, the founder or his family (hereafter referred to as the manager), granting these ‘insiders’ the majority of the voting power (DeAngelo and DeAngelo, 1985). The B stocks are the ones that are publicly traded, and therefore mainly in the hands of investors (hereafter referred to as shareholders). Hence, the voting rights are with the management and the majority of the cash
4
flow rights are with the shareholders. From the viewpoint of the shareholders, Villalonga and Amit (2006) point out two agency problems that arise. The
shareholders lose the power to incentivize the manager. They no longer have the ability to control the manager in order to maximize firm value. Also, the
management may abuse their controlling position to extract resources from the company for private goals. This would be at the expense of the shareholders. From these agency problems, the entrenchment theory follows. On the other hand there’s the value enhancing theory, which states that managers make better decisions when they are not controlled by outsiders and do not worry about voting power dilution. Both theories will be further discussed in the following section.
Literature on the topic is inconclusive. In an estimation of Tobin’s Q using Canadian dual class firms, King and Santor (2008) find a 17% discount on
market valuation. In contrast, Dimitrov and Jain (2006) find a positive effect of 23% in their survey to long term cumulative abnormal returns. Instead of
focusing on returns after announcement, this study examines the effect of a dual class structure in the long run. Thus, not the first periods after recapitalization are assessed, but a sample containing data of multiple years per dual-class company.The main finding is a negative coefficient for dual class structure, implicating that it has a long-term negative effect on firm value. This finding can be explained by the entrenchment theory, which is described below. Strictly speaking, a dual class structure is not beneficial for shareholders according to the results. This conclusion is avoided however, because other positive effects are possible and yet to be examined.
The remainder of the thesis is organized as follows: in section (2) related literature is discussed. In section (3) the empirical strategy is described, after which section (4) presents and discusses the results. Finally, section (5)
5 2. Effects of a dual-class structure
This section discusses the existing theories about the impact of a dual-class structure and focuses on managerial behavior. The entrenchment theory is discussed first, after which it is compared to the value enhancement theory. Finally, some other important theories are explained.
2.1 Entrenchment theory
When firms adopt a dual-class structure, the majority of the voting rights are in hands of the management. Claessens et al. (2002) find that in situations like these, firm value decreases. This would form evidence for the entrenchment theory. Entrenched managers do not feel pressure of all control factors in their acting. These factors include controlling by outside shareholders, monitoring by the board of directors or the treat of a hostile takeover (Berger et al., 1997). Hence, entrenchment occurs when the management’s controlling power is large enough for other parties to not be able to incentivize them.1
In case of dual-class shares where the voting rights exceed the cash flow rights, the incentive effect of cash flow rights on managers is overshadowed by the entrenchment effect. McGuire et al. (2014) state that entrenchment provides the opportunity for managers to not perform optimally. This leads to managers pursuing objectives that don’t result in value maximization. Bebchuk et al. (2000) support this vision and add that next to performing badly, managers can take advantage of their job security and extract resources from the firm on the wrong purpose. Examples for this are perquisite consumption and excessive
remuneration. These are costs for the agency problems pointed out by Villalonga and Amit (2006).
Evidence is found by Masulis et al. (2009). They find that as the difference between voting rights and cash flow rights (the wedge) widens, managers are more likely to make value destroying decisions, and capital expenditures lead in a lesser extent to value maximization. Gompers et al. (2010) find that firm value
6
decreases as the wedge widens. Thus, combining these papers it can be concluded that dual-class ownership leads to a decrease in firm value due to entrenchment.
Entrenchment can be seen as an anti-takeover mechanism as well. Grossman and Hart (1988) show that such a mechanism blocks the increase of firm value. By their controlling power, management is protected against hostile takeovers. These could be value enhancing for the shareholders by generating synergy, and would not be turned down in single class structure. Therefore, having a dual-class structure would always be suboptimal. In a resembling analysis of this subject, Jarrell and Poulsen (1988) set dual-class equal to other defensive structures. They say it can have a similar effect on shareholder value as greenmail payments or poison pills; a drop in shareholder wealth of between 1% and 3%. Bauguess et al. (2012) do not find any evidence for a dual-class structure acting as an anti-takeover device.
2.2 Value enhancement theory
In contrast to the negative effect of entrenchment, dual-class is said to have a positive effect on firm value. Dimitrov and Jain (2006) find no evidence for the entrenchment effect, but a value enhancing effect instead. In their research on long-term abnormal returnsthey find a positive effect of dual-class structures.
Several reasons can be found in existing literature. Lehn et al. (1990) find that industry-adjusted operating income increases in dual-class companies, which points to value enhancement. They also find that firms that adopt a dual-class structure have significantly more growth opportunities. Attracting capital to exploit these opportunities should increase firm value. The usual way to realize this, is a normal issuance of shares. However, a normal issue leads to voting power dilution. This translates into costs for the current shareholders. As a result, underinvestment will occur since profitable investment opportunities may be declined due to the costs of voting power dilution. A dual-class structure eliminates this effect, and therefore enhances firm value (Bauguess et al., 2012). Dimitrov and Jain (2006) validate this theory by showing that growth is indeed beneficial for shareholder value. Their tests cover a time frame of four years, in which the effect of a dual-class recapitalization on abnormal returns are found to
7
be significantly positive. The dual-class firms would outperform their counterparts by over 23% in these four years.
Next to exploiting growth opportunities, literature offers another explanation for the value enhancement effect. Chemmanur and Jiao (2012) present the managers’ claim that the dual-class structure allows them to pursue investment opportunities without being bothered about the temporarily
fluctuating share price. They assume there are two options: a more profitable investment in the long run, or a less profitable investment in the short run. The manager wants to invest in the long run, but it creates more uncertainty to
outsiders than investing in the short run. When the manager decides to follow his plans, the company’s share price fluctuates. This comes to displeasure of the shareholders and forms a threat for the job certainty of the manager. Thus, it provides an incentive to invest in the less profitable short run. With a dual-class structure, managers don’t have to worry about this threat and invest in the long run. It increases their ability to maximize firm value.
2.3 Other theories
Besides the entrenchment effect and the value enhancement effect, there are other effects that might influence the relation between dual-class ownership and firm value. Firstly, Chemmanur and Jiao (2012) distinguish between talented managers and untalented managers ‘mimicking’ the actions of the talented. In case of a talented manager, the dual-class structure will likely result in value enhancement. When the manager in untalented, entrenchment will lead to value destruction. Since it is nearly impossible to retrieve data on the abilities of the management as a person, this variable is omitted in the regression equation.
Secondly, there is the situation in which the generation coming after the management inherits the A class shares. This generation might not have been concerned with the company until the inheritance, and suddenly gain all control. In a normal situation this manager would be monitored and controlled by the shareholders, but this is prevented by the voting power. The same situation will occur as in the case of the untalented manager. Thus, the effect probably depends on the ability and reputation of the incumbent manager (Anderson et al., 2009).
8
A third factor of influence is whether the dual-class structure is exploited by family ownership. Anderson and Reeb (2003) find that family ownership has a significantly positive effect on performance. Their sample, consisting of 403 firms in total, is relatively small. Yet, the positive effect can be explained by the
monitoring effect of family shareholders. When the voting power is divided over family members who are involved with the profitability and growth of the
company, this creates an incentive to monitor the incumbent manager.
Monitoring reduces the effects of potential entrenchment and incentivizes the manager to maximize value.
Empirical research should give us a better understanding of the net effect of the dual-class structure on a firm’s market value. The observed relation should clarify the literature and can be interpreted as evidence for one of the theories. It should help investors in dual-class firms to make adequate decisions.
3. Empirical strategy
In this section, the construction of the dataset is described, followed by a list of variables. The usage of these variables is explained after the empirical model and the corresponding hypotheses are presented. Finally, the summary statistics are presented and briefly discussed.
3.1 Data
The control group of this study consists of every company that has a ‘single-class’ share structure. Together with the dual-class firms, this means data is required on both single- and dual-class companies. The annual fundamentals are retrieved from the Compustat database for all companies in North-America: Total Assets, Total Liabilities, Capital Expenditures, and Operating Cash-Flows. The SIC codes, used to indicate the industry division are retrieved from Compustat as well. For the variables of total assets and total liabilities, many observations are missing. Since these variables make up for an important part of the control variables, these observations are eliminated from the dataset. After calculating the ratio’s as described in Table 1 below, large outliers such as extremely high leverage are excluded, as well as errors in terms of duplicates.
9
Data on shares outstanding, share price at the end of the fiscal year and dividends are retrieved from the CRSP database. In contrast to the annual data of Compustat, the CRSP database presents monthly data. Shares Outstanding and dividends paid are merged with the Compustat dataset using the date on which the fiscal year ends. Outliers with respect to market value are excluded. They occur when the share price is below 1 USD. These ‘penny stocks’ result in a low market value, passing outliers to the dependent variable Tobin’s Q. The market value for these companies is about 2000 times smaller than the average, what makes it hard to find a significant coefficient.
Finally, companies in the sample with a dual-class structure have to be identified. ISS database (formerly RiskMetrics) contains data on this subject. Every company in the database with a dual-class structure is merged into the master file. A dummy variable for dual-class structure is created to be able to run regressions.
After the all the data is merged into one master file, the sample is checked for outliers and strange observations again for each variable. Only the most extreme observations are deleted immediately. The doubtful observations left are checked with financial statements to validate them. To balance the panel data, companies with only one or two years in the panel data are excluded. After cleaning, 33762 year observations are left, including 616 dual-class firm-years. Table 1 on page 10 presents a list of the used variables along with their definitions and sources.
10 TABLE 1: List of
variables
Variable Description Source Database Items
Dependent variable
Tobin’s Q Market value of equity/book value of equity CCM* PRCC_F
CRSP SHROUT
CCM* AT
ln(Q) Natural log of Tobin's Q - -
Independent variable
Dual Class Dummy variable for Dual Class firms ISS dualclass Control
variables
ln(Firm age)
Log of the difference between the year of observation and the first observation in the CRSP database.
CRSP PRC
Leverage Total debt/Total assets CCM* AT/LT
Investment
ratio Capital expenditures/Lagged total assets** CCM* CAPX Cash-flow ratio Operating Cash flow/Lagged total assets** CCM* OANCF
Dividend
Dummy variable for whether a company pays dividend. CRSP DIVAMT ln(Assets) Natural log of Total assets. CCM* AT
*Database in which Compustat and CRSP are merged **Total assets lagged by one fiscal year.
11 3.2 Model and hypotheses
To assess the extent to which value is maximized by the management in a dual-class structure, a series of empirical tests is conducted. Initially, market
capitalization was used as the dependent variable. However, market
capitalization varies heavily across the sample. This makes it hard to find a significant coefficient, as the residuals are big and fluctuate much. To
compensate for this, Tobin´s Q is used as a measure for relative market value. Tobin’s Q measures a firm’s market value in relation to the book value of assets. This way the size of a company does not influence the coefficient for dual-class status. The equation for Tobin’s Q (hereafter Q) is as follows:
= ℎ (1)
Usage of Q is in line with the papers of Gompers et al. (2010) and King and Santor (2008). Q is estimated with dual-class as the main explanatory variable, along with control variables suggested by related literature. The full model to be estimated looks as follows:
= + ∗ + ∗ ln( )
+ ∗ + ∗ +
∗ ℎ + ∗ +
∗ ln( ) + +
(2)
To assess elasticity, the natural logarithm of Q is also predicted using this equation. The model is tested using a linear regression model for panel data with year effects, firm fixed effects and industry fixed effects. Due to collinearity, firm fixed effects are only included when industry effects are not, and vice versa. Industry effects in the model are based on the SIC-codes of the companies in the sample. The first two digits in a SIC-code indicate the industry division a
company operates in. For each division, a dummy is generated to control for the effect a certain industry has. The non-classifiable division is omitted to prevent overidentification. Bauguess et al. (2012) match their sample to a control group
12
using 2-digit SIC codes, which suggests the importance of the industry division in the effect of dual-class structures. The model is estimated using a two sided test.
As stated in the previous section, two main scenarios are considered. The coefficient of interest, , is negative if a dual-class structure negatively affects market value. In that case, evidence is found for the entrenchment theory. is positive if a dual-class structure positively affects market value. In this opposite case, evidence if found for the value enhancement theory.
Literature finds evidence for both theories. Lehn et al. (1990) find that firms adopting a dual-class structure have significantly higher growth
opportunities. From this, it can be derived that the current incumbent is a
talented manager. Chemmanur and Jiao (2012) prove that talented managers are indeed value enhancing. This evidence is followed by the tested hypothesis:
Tested hypothesis: A dual-class ownership structure increases the market
valuation of a firm
However, arguably the most comprehensive dataset of dual-class companies was constructed by Gompers et al. (2010). Having the most
comprehensive dataset, the results should be the most reliable. Their research is somewhat similar to the one in this thesis, which implies that a corresponding negative impact of dual-class structures should be found. Therefore, the
alternative hypothesis is as follows:
Alternative hypothesis: A dual-class ownership structure decreases the market
valuation of a firm
3.3 Used variables
Table 1 on page 10 presents the exact definition for all used variables, along with their source. The reasons to include each variable are discussed here.
Dual-class firms tend to have more growth opportunities (Lehn et al., 1990). Growth causes the share price to increase, since future dividend payments and capital gains are expected to grow with it. A change in the share price causes
13
growth opportunities with capital expenditures over assets as a proxy. This variable is listed as Investment ratio. Profitability is also reflected in the share price of the company, as it influences current dividends and future growth.2 To
control for profitability, the proxy operating cash flows over total assets is added to the model. In both variables, the numerator influences the assets of the
corresponding year. For that reason, lagged assets are used.
From prior research it is known that smaller, younger firms are more profitable, and grow at a faster pace. This increases market valuation and therefore Q. Bigger, more mature firms tend to have a lower Q. The dataset contains firms of different sizes, which results in an increase in the error term of the regression. Size is usually controlled for by total assets. But since the spread of total assets in the dataset is large causing the error term to increase, total assets are transformed into a logarithm.
Leverage is added to control for solvency. When a company is highly levered, the interest costs and potential financial distress costs increase. It influences profitability of the firm and therefore the share price and firm value. Paying dividends influences the share price and firm value as well. A dummy variable for whether a company pays dividend is included. Finally, following Bauguess et al. (2012) firm age is the last control variable. Firm age reflects the number of years the firm is traded in the CRSP database prior to the date of the observation. Leverage, capital expenditures, total assets, age and dividends are in line with the paper of Gompers et al. (2010) and King and Santor (2008).
3.4 Reverse causality
Gompers et al. (2010) point out an error in the relation between Q and the dual-class status. Managers can base the decision to adopt a dual-class structure on whether the firm is over- or undervalued. If the firm is undervalued, its shares are sold at a relatively low price. It becomes a takeover target for potential rivals. As can be read in the literature review, the dual-class structure takes care of this
2 When profits are not paid out as dividend, they are most likely invested in growth opportunities causing future growth. Expected dividends and capital gains determine the share price
14
problem. Hence, the decision to go dual-class depends to some extent on the level of Q. For the model, this implicates that the dependent variable might influence the main explanatory value, also known as reverse causality. Reverse causality would cause a bias towards a negative coefficient. Accordingly, this paper reports a negative coefficient. Gompers et al. (2010) also show that the decision to adopt a dual-class structure depend on many other factors, so Q might not have a considerable influence on dual-class after all.
Usually reverse causality can be solved using an instrumental variable. Such a variable should correlate with the dual-class dummy but cannot be correlated to the error term, and is not easily obtainable. Gompers et al. use two variables as instruments for their regression: the % of firms in Compustat that are in the same metropolitan area as the observation, and the % of total sales of these firms. The used dataset doesn’t contain information on variables like these, so it is assumed that the influence is considerably small.
3.5 Summary Statistics
Table 2presents some summary statistics of the final dataset. The dataset is split into single class companies (control group) and dual-class companies (group of interest).
Worth noticing is that the dual-class companies have a lower average Q than the control group. The lower Q is caused by the fact that the difference in asset value, in comparison to the control group, is higher than the difference in market value. Both of these values are higher for the dual-class group, what implicates that the firms in this group are bigger on average. Furthermore, their capital expenditure ratio is lower while the cash flow ratio is almost twice as high, what suggests that the firms that form the this are more mature companies.3 The maturity is
also reflected in their age. Average ln(Firm Age) represents a difference of 7 years on average. The lower investment ratio contradicts the finding of Lehn et al. (1990) that dual-class companies would have more growth opportunities. Furthermore, the almost double cash flow ratio suggests that the dual-class firms
3More mature firms typically invest less R&D and new assets in general. They are grown firms generating cash flows that
15
are more profitable than their single class counterparts. This is in line with the findings of Bauguess et al. (2014). The average of the dual-class variable in the full dataset is equal to 0,018, which means that 1,8% of my total observations are cases with two share classes. The sample of Gompers et al. (2010) consisted for approximately 6% of dual-class observations. Hence, this dataset is not as comprehensive as theirs.
Finally, leverage is somewhat lower in the dual-class sample, which is in line with the entrenchment theory. Berger et al. (1997) mention that entrenched managers may prefer a lower level of leverage than optimal. The preference would arise from the desire to preserve their undiversified human capital by lowering firm risk. King and Santor (2008) add that a higher level of leverage restricts the control effect of dual-class shares, since creditors can impose
constraints via covenants to monitor the management. They don’t find empirical significance. Theoretically, by taking on more debt the management loses part of the controlling power.
TABLE 2: Summary
Statistics
Full dataset Single-Class companies Dual-class companies
Mean Dev. Std. Mean. Dev. Std. Mean Dev. Std.
Variable Q 1.120 1.308 1.122 1.312 1.028 1.082 ln(Q) -0.512 1.248 -0.512 1.251 -0.471 1.113 Dual class 0.018 0.134 ln(Age) 2.545 0.943 2.538 0.943 2.927 0.834 Leverage 0.549 0.282 0.550 0.283 0.505 0.228 Investment 0.052 0.109 0.052 0.110 0.043 0.043 Cash-flow 0.064 0.192 0.063 0.193 0.111 0.088 Dividend 0.142 0.350 0.142 0.350 0.141 0.349 ln(Assets) 20.681 2.066 20.662 2.069 21.748 1.630 N 33,762 33,146 616
16
To give a better understanding of the evolution of Q in both subsamples, Figure 1 presents Q over the years of the subsample. Over the full sample period
Q increased more for the dual-class companies. This suggests that the value
enhancement theory is applicable. Regarding the financial crisis, which
commenced in late 2007, it can be observed that Q decreased at a faster paste for the single class firms. Theory behind this might be that the management of dual-class companies were more able to ignore temporary price fluctuations in their decision making, mitigating the overall drop in share price. Between 2011 and 2013, Q increases for both groups. In 2012, the lines cross. The crossing implies that in a period of economic growth, dual-class companies outperform their counterparts in term of market valuation. This again would confirm the value enhancement theory.
FIGURE 1: Average Q per year for both ownership structures
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 Single-Class Dual-Class
17 4. Results
4.1 Empirical results
Equation 2 is estimated in Table 3 controlling for several fixed effects. In all columns, the dual-class dummy is the main explanatory variable. As mentioned, firm fixed effect and industry fixed effects are never used in the same regression due to collinearity. When fixed effects are not included, heteroscedasticity is assumed. Hence, there is controlled for robust standard errors when fixed effects are excluded.
Starting off, column 1 is a control estimation with only year effects. The control for size is left out, to check whether the -0,3 correlation with Q is
troublesome. It shows that the effect of the dual-class status on Q is a drop of 14 points, indicating that the entrenchment theory is applicable. This negative effect is observed for all significant coefficients of dual-class throughout the regression panel. In column 2, control for size is included in combination with industry fixed effects. The industry fixed effects cause all control variables to be significant, but the main explanatory variables is not. Hence, it is not possible to derive a
conclusion from this. In column 3, the same regression is controlled for firm fixed effects. The coefficient for dual-class is significant, along with the control for size. For this reason, it is assumed that the correlation causes no problems. To be able to interpret the elasticity, in columns 4 and 5 the dependent variable is the natural log of Q. Controlling for industry fixed effects results again in no
significance for the dual-class variable. In contrast, adding firm fixed effect shows a mildly significant result. Columns 3 and 5 have a significant coefficient for dual-class, and show the best statistical precision in terms of standard errors. Therefore, results of these regressions will be interpreted as the main results of the paper.
According to column 3, the main explanatory variable dual-class has a coefficient of -0,158. Thus, the ratio of market value of assets to book value of assets declines if the firm switches to a dual-class structure. From this can be concluded that the stock market valuation of the firm in comparison to the assets, decreases. A drop of 15,8 points in Q causes the market value of assets to drop by this amount times the book value of assets. Similarly, column 5 shows a
18
coefficient for dual class of -0,087. Since the dependent variable is a logarithm, we can use it to interpret the elasticity. A drop of 8,7 points represents a
percentage drop in Q of approximately 8,7%. To put this result into perspective, consider the following example. For the sake of simplicity, the book value of assets is held constant. This makes sense in general, since the share price is much more volatile than the book value of assets. Let’s take a company from the dataset: Kellogg Company has a Q value of approximately 1,5, a market
capitalization of $18,2 billion and book value of assets of $11,9 billion. From the coefficient in column 5, this represents a drop in market value of 18,2*0,087 = $1,58 billion. According to column 3, switching to a dual-class structure causes a drop in market value of 0,158*11,9 = $1,88 billion. In reality, the drop can be alleviated by an increase in the book value of assets. Hence, the exact effect can’t be measured from this dataset, but the negative relation between market value of assets and dual-class status is assumed to be established.
The negative coefficients are in line with the literature, King and Santor (2008) find that family owned companies with a dual class structure are valuated 17% lower in comparison to widely held firms. The dual class status on its own also has a negative coefficient, slightly smaller than the effect mentioned above. Gompers et al. (2010) find a negative coefficient for dual class on Q of 5,4 points. This is approximately 3 times smaller than the coefficient in my regression. Their use of an instrumental variable can explain this difference, as it’s included to solve the potentially negative bias.
19
TABLE 3: Estimated effect of a dual class structure on Tobin’s Q.
(1) (2) (3) (4) (5) Dependent variable: Q Q Q Ln(Q) Ln(Q) Dual Class -0,144 (0,072)** (0,075) -0,070 (0,075)** -0,158 (0,059) -0,042 (0,048)* -0,087 Ln(Age) -0,105 (0,016)*** (0,016)*** -0,056 (0,018) -0,025 (0,012)** -0,024 (0,011)**** -0,052 Leverage -0,854 (0,086)*** (0,082)*** -0,602 (0,034)*** -0,622 (0,058)*** -0,868 (0,022)*** -0,844 Investments 0,250 (0,078)*** (0,078)*** 0,278 (0,053)*** 0,314 (0,067)*** 0,339 (0,034)*** 0,366 Cash-Flow to Assets (0,076) 0,092 (0,080)** 0,195 (0,030)*** 0,293 (0,056)*** 0,294 (0,020)*** 0,306 Dividend -0,021 (0,140) (0,014)*** 0,040 (0,017)** 0,036 (0,012)*** 0,094 (0,011)*** 0,083 Ln(Total Assets) -0,197 (0,011)*** (0,012)*** -0,382 (0,009)*** -0,212 (0,008)*** -0,342 Industry Fixed
Effects YES YES
Firm Fixed
Effects YES YES
Year Effects YES YES YES YES YES
Number of
Observations 31851 31851 31851 31851 31851
Number of Firms 5232 5232 5232 5232 5232
Overall R2 0,113 0,193 0,133 0,372 0,216
Notes to table 3. Data are retrieved from CRSP, Compustat and ISS, and include companies observed from 2007-2014. A panel regression is used. The Dual Class dummy equals 1 if a company has a dual class structure, zero otherwise. The Dividend dummy equals 1 if a company pays dividend, zero otherwise. In regressions where fixed effects isn’t controlled for,
heteroscedasticity is assumed. Standard errors are in parentheses and the asterisks *,** and *** indicate significance of 10%, 5% and 1% respectively.
20 4.2 Discussion
The main objective, solving the puzzle in existing literature, has been fulfilled. A negative relation between dual-class structures and market valuation has been established, though it’s difficult to examine the reliability of this result. The reverse causality problem has not been dealt with, which implies that the outcome may be biased negatively. However, Gompers et al. (2010) do control for it, and they find a negative relation as well. This indicates that the result is only slightly magnified by the reverse causality problem, but may still be valid.
The secondary objective of this paper is to possibly offer guidance to
legislates and investors about how to deal with dual class structures. They might want to reject a motion to recapitalize or restrict firms in their ability to adopt such a structure. The negative effect on market valuation does point in this direction, as it can be interpreted as evidence for the entrenchment theory. Strictly speaking, this should lead to the government regulating the firms in their ownership structure and shareholders selling their shares. However, it is impossible to give hard recommendations when evaluating this result. Dual-class structures may for example have a positive influence on environmental decision-making or the working conditions of a firm. Therefore, additional research is required.
Some important limitations of this thesis must be mentioned. Firstly, this study does not account for a concrete difference in voting rights and cash flow rights. The extent to which these differ might influence the observed effect. Secondly, one should keep in mind that the result cannot be interpreted as it is. The endogeneity bias has to be solved first using an instrumental variable. Thirdly, the ability of the manager should be distinguished for. A
recommendation for future studies is to indicate the talent of the manager and divide them into subgroup to test them separately. This might lead to new
insights. If Eric Lam (2016) is right about the effect going both sides, this should be the factor it is depending on.
21 5. Conclusion
A dual class ownership structure is a contentious topic. Literature is inconsistent on whether its economic impact is negative of positive. The scope of this thesis is to assess impact of a dual-class structure on market valuation in the long run. A negative relation is found, which suggests that the entrenchment theory applies. To protect existing shareholders against this effect, dual class voting structures might have to be regulated in the future. Also, shareholders might want to use their vote to reject a dual-class recapitalization when the idea is introduced. It remains difficult to give hard recommendations based on this result, due to limitations and other influences.The model is limited by excluding the wedge between voting rights and cash flows rights, and not assessing the ability of the management. Including this in further research can lead to more significance and increase the usefulness of my findings.
22 6. References
Adams, R., and Ferreira, D. (2008). One share-one vote: The empirical evidence.
Review of Finance, 12(1), 51-91.
Anderson, R. C., Duru, A., and Reeb, D. M. (2009). Founders, heirs, and corporate opacity in the United States. Journal of Financial Economics, 92(2), 205-222.
Anderson, R. C., and Reeb, D. M. (2003). Founding-family ownership and firm performance: evidence from the S&P 500. The Journal of Finance, 58(3), 1301-1328.
Bauguess, S. W., Slovin, M. B., and Sushka, M. E. (2012). Large shareholder diversification, corporate risk taking, and the benefits of changing to differential voting rights. Journal of Banking & Finance, 36(4), 1244-1253.
Bebchuk, L. A., Kraakman, R., and Triantis, G. (2000). Stock pyramids, cross-ownership, and dual class equity: the mechanisms and agency costs of separating control from cash-flow rights. Concentrated Corporate Ownership, 295-318.
Berger, P. G., Ofek, E., and Yermack, D. L. (1997). Managerial entrenchment and capital structure decisions. The Journal of Finance, 52(4), 1411-1438.
Chemmanur, T. J., and Jiao, Y. (2012). Dual class IPOs: A theoretical analysis.
Journal of Banking & Finance, 36(1), 305-319.
Claessens, S., Djankov, S., Fan, J. P., and Lang, L. H. (2002). Disentangling the incentive and entrenchment effects of large shareholdings. The Journal of
Finance, 57(6), 2741-2771.
DeAngelo, H., and DeAngelo, L. (1985). Managerial ownership of voting rights: A study of public corporations with dual classes of common stock. Journal of Financial
Economics, 14(1), 33-69.
Dimitrov, V., and Jain, P. C. (2006). Recapitalization of one class of common stock into dual-class: Growth and long-run stock returns. Journal of Corporate Finance, 12(2), 342-366.
Gompers, P. A., Ishii, J., and Metrick, A. (2010). Extreme governance: An analysis of dual-class firms in the United States. Review of Financial Studies, 23(3), 1051-1088.
Grossman, S. J., and Hart, O. D. (1988). One share-one vote and the market for corporate control. Journal of Financial Economics, 20, 175-202.
Jarrell, G. A., and Poulsen, A. B. (1988). Dual-class recapitalizations as
antitakeover mechanisms: The recent evidence. Journal of Financial Economics, 20, 129-152.
23
King, M. R., and Santor, E. (2008). Family values: Ownership structure, performance and capital structure of Canadian firms. Journal of Banking &
Finance, 32(11), 2423-2432.
Lam, E. (2016). Bombardier Aside, Canada Dual-Class Shares Are Top Gainers. Retrieved May 21, 2016, from
http://www.bloomberg.com/news/articles/2016-04-28/bombardier-aside-canada-dual-class-shares-are-top-gainers
Lehn, K., Netter, J., and Poulsen, A. (1990). Consolidating corporate control: Dual-class recapitalizations versus leveraged buyouts. Journal of Financial
Economics, 27(2), 557-580.
Masulis, R. W., Wang, C., and Xie, F. (2009). Agency problems at dual-class companies. The Journal of Finance, 64(4), 1697-1727.
McGuire, S. T., Wang, D., and Wilson, R. J. (2014). Dual class ownership and tax avoidance. The Accounting Review, 89(4), 1487-1516.
Villalonga, B., and Amit, R. (2006). How do family ownership, control and management affect firm value? Journal of Financial Economics, 80(2), 385-417.
