Corporate governance and firm performance
in dual-class firms
Name: Jingdan Tan
Student number: 11377321
Thesis supervisor: dr. V. O’Connell Date: 26 June 2017
Word count: 14,659
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
2 Statement of Originality
This document is written by student Jingdan Tan 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.
3 Abstract
This study aims to expand the understanding of corporate governance mechanism in dual-class firms. Based on the sample of S&P 1500 firms from 2008 to 2015, I investigate the association between corporate governance quality and firm performance in the dual-class setting. I also examine whether share structure and firm size have moderating effect on the corporate governance quality-firm performance relationship in dual-class firms. Consistent with the expectation, I find evidence that the positive relationship between corporate governance quality and firm performance which is widely reported for single-class firms is also applicable in dual-class context, and the corporate governance-firm performance relationship is significantly more positive in larger dual-class firms. However, empirical result fails to support the hypothesis that firms with dual-class share structure have a significantly more positive relationship between corporate governance quality and firm performance. The result addresses the potential two-sided effect of dual-class structure on corporate governance and firm performance, namely the negative effect caused by severe agency problems, and the positive effect brought by the anti-takeover protection.
4 Contents 1 Introduction ... 5 2 Literature review ... 7 2.1. Dual-class firm ... 7 2.2. Corporate governance ... 14 2.3. Firm performance ... 17 2.4. Hypothesis development ... 19
2.4.1. Corporate governance quality and firm performance in dual-class firms... 19
2.4.2. The moderating effect of share structure on corporate governance quality-firm performance relationship ... 20
2.4.3. The moderating influence of firm size on corporate governance quality-firm performance relationship ... 22
3 Methodology ... 25
3.1. Sample and data ... 25
3.2. Measurement of corporate governance quality ... 25
3.3. Measurement of firm performance ... 27
3.4. Control variable ... 28 3.5. Empirical models ... 29 4 Results ... 30 4.1. Descriptive statistics ... 31 4.2. Empirical findings ... 34 4.3. Test of endogeneity... 37 5 Discussion ... 41 6 Conclusion ... 44 Reference ... 47
5
1 Introduction
Dual-class share structures have long been criticized for their severe agency problems (Grossman and Hart, 1988; Hossain 2014). The separation of voting rights and cash-flow rights enables insiders of dual-class firms to hold the majority of voting power, leaving outside shareholders in a relatively weak position. The excess control rights provide opportunities for insiders to pursue private benefits (Masulis et al., 2009).
Many empirical studies have addressed various issues brought by the disproportional cash-flow rights and voting rights in dual-class firms. For example, Lim (2016) find that firms with dual-class share structure suffer from higher level of information asymmetry than firms with single-class share structure, indicating that the entrenchment effects (arising from voting rights) dominate the incentive effects (arising from cash-flow rights); Masulis et al. (2009) report that as the divergence between cash-flow rights and voting rights widens, dual-class firms encounter more agency problems; Gompers et al. (2010) conclude that firm value in dual-class firms is positively associated with insider’s cash-flow rights and negatively associated with insider’s voting rights.
Following Gompers et al. (2010), it is interesting to investigate whether firms with dual-class share structure adopt other additional governance mechanisms, such as increasing outside representation of directors on board and performance-based compensation, to compensate for the potential agency costs. This study aims to shed light on governance mechanisms and firm performance in dual-class firms.
The tests in this study are based on a sample of S&P 1500 firms from 2008 to 2015. Among the total 1,949 firms, 173 are classified as dual-class firms, accounting for 8.88% of the total sample. Following the methodology of Baber and Liang (2008), I measure the internal corporate governance quality using B-Index, the sum of six governance variables. Referring to O’Connell and Cramer (2010), I use Tobin’s Q, return on assets (ROA) and stock return as firm performance metrics.
6 dual-class firms. Lacking effective corporate governance mechanism, managers have more opportunities to pursue private benefits at the expense of outside shareholders, and firms tend to underperform. I expect that firm performance is positively related to governance quality in dual-class firms. Empirical results support the hypothesis when firm performance is measured by ROA and stock return, indicating that in dual-class firms, there is a significant positive association between corporate governance quality and firm performance.
Secondly, I analyze whether share structure has a moderating impact on the corporate governance-firm performance relationship. Given the fact that the potential effects of dual-class share structure on firm performance may be positive or negative, I assume that the corporate governance-firm performance relationship is significantly more positive in dual-class firms. However, the empirical results fail to support the hypothesis, since none of the results are significant. When taking the potential endogeneity effect into consideration, result based on the performance metric of stock return becomes significant. Taken as a whole, I cannot safely draw the conclusion that dual-class share structure has a moderating effect on the corporate governance-firm performance relationship, indicating that dual-class firms cannot significantly improve their firm performance simply through enhancing corporate governance mechanism.
Thirdly, I investigate whether firm size has a moderating effect on the corporate governance quality-firm performance relationship in dual-class firms. Prior literature reveals that corporate governance does not work identically in larger and smaller firms, and firm size is one of the essential factors that worth considering in the adoption of dual-class share structure. I predict that the corporate governance quality-firm performance relationship is significantly more positive in larger dual-class firms. Consistent with my expectation, I find that larger dual-class firms have a significantly more positive association between corporate governance quality and firm performance, when ROA and stock return are used as performance metrics. This study makes two main contributions to the existing literature. Firstly, this study expands the understanding of corporate governance in a dual-class setting. Most of the prior studies on corporate governance are done in single-class context, without taking the share structure into
7 consideration. In addition, most of the studies on corporate governance in dual-class context focus on individual governance attributes, while few studies adopt a holistic view, using aggregated measurements of corporate governance. This study addresses these gaps in the literature. Secondly, this study sheds light on the size effect on the corporate governance-firm performance relationship in a dual-class setting, which has been largely neglected in prior studies.
The remainder of the paper is structured as follows. Section 2 reviews the prior studies on dual-class share structure, corporate governance and firm performance, and explains how the three hypotheses are developed. Section 3 describes the methodology and research design, introducing the sample, measurement of variables and the empirical models I used in this study. Section 4 outlines the core findings of the empirical results and the test of endogeneity. Section 5 provides the discussion of the empirical results. Summary and conclusions are presented in Section 6.
2 Literature review
In this section, prior studies are reviewed. I start with the existing literature on dual-class firms, comparing dual-class share structure with single-class share structure and summarizing characteristics of dual-class firms from prior studies. Then, studies on corporate governance and firm performance are introduced. Lastly, hypotheses development is presented.
2.1. Dual-class firm
As dual-class share structure is adopted by more and more firms across different countries, it received great attention from the public and academic researchers. Dual-class share structure is substantially different from traditional single-class share structure, while both kinds of share structures have their benefits and costs.
Single-class share structure is also known as one share-one vote structure, with each shareholder owning equally cash-flow rights and voting rights. Under this share structure,
8 firms are less likely to be controlled by insiders or wealthy individual outside shareholders. Being regarded as a way of doing fair business (Hossain 2014), single-class share structure is adopted as the rule in many exchanges of different countries. The one-share-one-vote rule in NASDAQ exchange is defined as all shareholders having equal voting rights in public firms and each shareholder having one vote.
As the firm is growing larger, the funding needs inevitably lead to the dilution of control rights over the single-class firm. As each share has equal cash-flow rights and voting rights, management has to sacrifice their control rights when they decide to get funds from the market. There are a great many examples showing that the founders of firms gradually lose their corporate control rights as they adopt stock financing under the traditional one share-one vote structure.
Control right over the firms is viewed as the most important thing to the founders of the firms. As a result, founders of the companies attempt to find effective ways to protect their corporate control rights. Many of founders of firms adopt dual-class share structure, reducing the voting power of shares held by outside shareholders. Dual-class share structure is becoming prevalent across different countries.
Typical dual-class share structure divides common stock into two different categories, namely Class A and Class B shares. Class A shares are with inferior voting rights and offered to general public, while Class B shares are with superior voting rights and often held by insiders of the firms. Such a share structure enables insiders and outside shareholders to have disproportional voting rights and cash-flow rights.
Under most of the dual-class share structures, outside shareholders own only shares with inferior or even no voting rights, while the majority of voting power is held by insiders of firms. Such a share structure offers great protection to the controller of dual-class firms, and effectively reduces the possibility of hostile takeovers. Gompers et al. (2010) report that US dual-class firms from 1995 to 2002 are significantly older than one share-one vote firms, indicating that firms with dual-class share structure are less likely to be acquired, compared to firms with traditional single-class share structure. Therefore, dual-class share structure is
9 considered to be an effective anti-takeover defense.
Dual-class firms are quite common in Asia, where family corporations are prevalent. Although dual-class share structure has long been criticized for its potential agency problem (Grossman and Hart, 1988; Hossain 2014), nowadays many large international firms, such as Google, Dell, Facebook and Ford Motor, adopt this share structure. These dual-class firms are commonly dominated by founders or founders’ families (Hossain 2014). DeAngelo and DeAngelo (1985: 33) find significant family involvement in dual-class firms, and top managements and their families hold a median 56.9% of voting rights and 24.0% of cash-flow rights.
Gompers et al. (2010) investigate the determinants of dual-class adoption, and find that the following five factors are significantly related to the adoption of dual-class share structure: the company is named for a founder, firm is in a media industry, the sales of other firms in the same area are higher, there are fewer firms in the same area, and the firm has lower sales ranking compared to other firms during the IPO year.
Dual-class firms construct their share structure in various ways, such as classifying different classes of shares, setting the unequal voting structure for the superior and inferior shares, and determining the proportion of publicly trading superior and inferior shares. Gompers et al. (2010) illustrate that 85% of US dual-class firms from 1995 to 2002 have at least one untraded class of common stock, which always has superior voting power. The most common dual-class structure is a 1-to-10 voting structure, in which inferior voting share has one vote for each share, while superior voting share has ten votes for each share, and approximately 60% of voting rights and 40% of cash-flow rights are held by insiders of dual-class firms (Gompers et al., 2010: 1056).
The disproportional cash-flow rights and voting rights of shares, on the one hand, provides anti-takeover protection to insiders of the firms, on the other hand, potentially results in more agency problems. Compared to the one share-one vote structure, dual-class share structure is considered to have weaker alignment of interest between management (controller) and outside shareholders, which leads to higher agency cost (Smith et al., 2009). The excess
10 control rights provide opportunities for insiders to pursue private benefits (Masulis et al., 2009).
Prior literature discusses the characteristics of dual-class firms in the following aspects: trading prices of superior and inferior voting shares (e.g., DeAngelo and DeAngelo., 1985; Amoako-Adu and Smith, 2001; Smart and Zutter, 2003; Smart et al., 2008; Smith et al., 2009), agency cost (e.g., Bebchuk et al., 1999; Masulis et al., 2009; Hoi and Robin, 2010; Lim 2016), firm valuation (e.g., Gompers et al., 2004; Gompers et al., 2010), and earning informativeness (e.g., Fan and Wong, 2002; Francis et al., 2005).
Discounted trading price of dual-class shares
Publicly traded dual-class shares, often with inferior voting rights, are traded at a discount relative to single-class shares, while shares with superior voting rights often require a premium. One possible explanation for this phenomenon is that outside shareholders are only willing to buy inferior shares at a lower price, as they perceive that managers of dual-class firms may extract for personal benefits at the expense of outside shareholders.
Many empirical studies demonstrate that the trading share prices of dual-class firms are relatively lower than that of single-class firms (Smart et al., 2008). Smart and Zutter (2003) report that firms with dual-class shares structure suffer from less underpricing than firms with single-class structure, but they have lower trading prices relative to fundamentals than single-class IPOs. This situation can be explained by the fact that dual-class managers do not have the incentive to disperse equity ownership through underpricing. Smart et al. (2008) conclude that both at IPO and the subsequent five years, trading prices of dual-class firms are significantly lower than those of single-class firms. The discount trading price of dual-class shares is mainly due to the fact that insiders hold the majority of voting power, making it difficult for outside shareholders to replace the incumbents (Smart et al., 2008). However, they also point out that the lower trading prices of dual-class share do not indicate lower abnormal stock or accounting returns. Smith et al. (2009) investigate dual-class firms in the Canadian context and reach similar results, stating that Canadian dual-class firms, when compared to closely-held single-class firms, sell at a significant discount. Such a
11 phenomenon is because of the higher level of agency problem brought by the dual-class structure (Smith et al., 2009).
Agency problem
The disproportional voting rights and cash-flow rights in dual-class firms leave room for insiders to exercise managerial discretion and engage in shareholder value-destroying activities, reflecting severe agency conflicts between insiders and outside shareholders. Agency problem, arising from the separation of ownership and control (Jensen and Meckling, 1976), is expected to be more significant in dual-class firms. As the majority of voting rights are held by insiders in dual-class firms, it is relatively hard for outside shareholders to intervene in the management activities, even though they hold a large proportion of shares. Outside shareholders are unlikely to successfully replace the managements through exercising their inferior voting rights, even if the management performance is relatively poor. Empirical results from Smart et al. (2008) support this argument, illustrating that dual-class firms have less frequent CEO turnover, and CEO turnover is only responsive to firm performance of single-class firms.
Lim (2016) find that firms with dual-class share structure suffer from higher levels of information asymmetry than traditional one share-one vote firms. Agency costs are higher in dual-class firms, as inferior voting shareholders are less powerful to control insiders' acts, even though they have incentives to monitor the firms (Lim 2016). In order words, the excessive control rights of insider make it more costly for outside shareholders to monitor dual-class firms.
As the divergence between cash-flow rights and voting rights widens, agency conflicts in dual-class firms aggravate (Masulis et al., 2009). Bebchuk et al. (1999) define dual-class share structure as a controlling minority structure, since it allows shareholders to control the firm with only a small fraction of its equity. They find that agency costs imposed by controllers who only hold a minority of cash-flow rights are larger than those imposed by controllers who obtain control over the firm through owning the majority of cash-flow rights, under the controlling minority structure. The disproportional agency costs can be explained
12 by the non-linear relationship between the size of agency costs and size of cash-flow rights, in which size of agency costs increases at a sharply rate when the holding size of cash-flow rights decreases (Bebchuk et al., 1999). Masulis et al. (2009) document several phenomena reflecting agency problems in dual-class firms, and find that as the cash-flow rights-voting rights divergence widens, an extra dollar of cash holdings means less to outside shareholders, CEOs of firms receive higher compensation, and managers engage more in shareholder value-destroying acquisition. The results imply that the disproportional cash-flow rights and voting rights enable and facilitate managerial extraction of personal benefits of control. Earning informativeness
Agency conflicts in dual-class firms can also be reflected from their earning informativeness. Fan and Wong (2002) investigate the earning informativeness in East Asia context, and find that higher ownership concentration and larger separation of ownership and control lead to lower earning informativeness. They provide two possible explanations for this phenomenon: controlling owner’s entrenchment weakens the perceived earning credibility of outside shareholders, and the concentrated ownership enables controllers to conceal the proprietary information about their rent-seeking activities. Francis et al. (2005) reach similar results in the US setting, and report that earnings are less informative in dual-class firms, while dividends are more informative. Francis et al. (2005) also indicates that the net effect of dual-class share structure is reducing the credibility of earnings, and improving perceived performance through substantial dividends.
Firm valuation
Prior literature suggests that the excessive control rights held by insiders of dual-class firms have adverse impacts on firm valuation (Gompers et al., 2004). Because of the perceived significant agency problems, firms with dual-class share structure are more likely to underperform and suffer from lower firm valuation.
Hoi and Robin (2010) examine whether proximity of controller of dual-class firms has an impact on firm valuation. Proximity is defined to be high, medium and low when the controller of dual-class firm is a top executive, a board of director, and an outsider. They find
13 that higher proximity of controller in dual-class firms, which provides greater opportunities for insiders to engage in expropriation such as empire building and excessive managerial compensation, leads to lower firm value.
Gompers et al. (2004) investigate the relationship between the cash-flow rights-voting rights divergence and firm valuation in dual-class firms, and come to the conclusion that the firm value is decreasing in the level of voting rights and increasing in the square of cash-flow rights, and is increasing in the level of cash-flow rights and decreasing in the square of cash-flow rights. The results imply that management entrenchment (arising from voting rights ownership) impairs firm values of dual-class firms, while the alignment of incentive (arising from cash-flow rights ownership) enhances firm values. Building on the previous study, Gompers et al. (2010) report that the firm value of dual-class firms is negatively related to insider’s ownership of voting rights and positively related to insider’s ownership of cash-flow rights.
However, dual-class share structure does not necessarily do harm to shareholder wealth. In order words, the separation of voting rights and cash-flow rights is not always suboptimal to outside shareholders. The benefits from hostile takeover protection may exceed the agency costs caused by the share structure (Böhmer et al., 1999). Empirical studies reveal that dual-class share structure can be beneficial to outside shareholders. DeAngelo and DeAngelo (1985) illustrate that dual-class share structure encourages managers to invest in corporate-specific human capital. Dimitrov and Jain (2006) state that dual-class share structure enhances firm value, and firms with dual-class recapitalization experience a higher growth rate of sales, assets and operating earnings compared to their single-class competitors. The results support the value enhancing argument that firms adopt dual-class recapitalization to achieve growth without losing control, and provide no evidence to the value destroying argument that managers choose dual-class recapitalization to consume personal advantages at the cost of outside shareholders (Dimitrov and Jain, 2006).
14
2.2. Corporate governance
Agency problems between managers and shareholders at public corporations arise as a result of the separation of ownership and control, the conflicting interests and the information asymmetry between owners and managers (Jensen and Meckling, 1976; Fama and Jensen, 1983). As interests of managers are not in line with those of shareholders, managers may have incentives to pursue private benefits at the price of shareholders (Shleifer and Vishny, 1997; Dey 2008). CEO receiving excessive compensations, empire building, and managers’ approving shareholder value-destroying investments are examples of agency problems in firms. When managers and owners have conflicting interests, agency costs arise (Jensen and Meckling, 1976; Ang et al., 2000).
Agency cost defined by Jensen and Meckling (1976) refers to the principals’ monitoring cost, agents’ bonding cost and residual loss. Ang et al. (2000) provide an explanation for monitoring costs and free-rider problems in public firms. Since monitoring effort is a quasi-public good, the free-rider problem easily arises. Monitoring shareholders incur full monitoring costs in monitoring activities but enjoy limited monitoring benefit, which is limited to the percentage of shares they hold. On the contrary, non-monitoring shareholders enjoy full benefits brought by monitoring activities without incurring any monitoring costs. Dalton et al. (2007) conclude three main approaches to mitigate agency problems, namely independence approach, equity approach and market for corporate control. Independence approach suggests that independent directors on board can effectively monitor managers’ activities and reduce the possibilities of managers’ misconducts. Equity approach indicates that the increase in managers’ equity ownership can help to ease the conflicts of interests between managers and owners. Market for corporate control implies that the threat of being acquired by other firms has disciplinary effect on managers. In fact, all of the approaches refer to firms-level corporate governance mechanisms.
Corporate governance is a set of mechanisms that help to ensure firms are directed and managed to create value for owners while fulfilling responsibilities to other shareholders (Merchant and Stede, 2007). The aim of corporate governance is to mitigate agency problems
15 and reduce agency conflicts between owners and managers.
Some prior studies reveal that corporate governance choice is an endogenous determination (e.g., Larcker et al., 2007; Dey 2008; Black et al., 2006; Drobetz et al., 2004). Dey (2008) report that firms with severe agency conflicts also have better corporate governance mechanism in place, indicating that corporate governance mechanism is an endogenous response to the economic and business environment of the firms.
As corporate governance is a mechanism to mitigate agency problems, firms with weaker governance tend to have more agency problems (Core et al., 1999). Core et al. (1999) examine the relationship of corporate governance, CEO compensation and firm performance, and suggest that CEOs in firms with weaker corporate governance receive greater compensations, while firms with more agency problems are more likely to underperform. Armstrong et al. (2010) reach the similar conclusion, suggesting that CEOs in firms with weaker corporate governance get higher compensation.
Corporate governance mechanisms can be further classified as either external governance mechanisms (market for corporate control) or internal corporate governance mechanisms (shareholder activism), and the two kinds of mechanisms interact with each other (Cremer and Nair, 2005; Brown and Caylor, 2005; Baber and Liang, 2008). Baber and Liang (2008) find evidence that two kinds of corporate governance mechanisms act as substitutes.
Internal corporate governance
Internal corporate governance refers to the interaction between firm’s managements, board of directors and employees (Baber and Liang, 2008). Prior studies on internal governance focus on the characteristics of board of directors (e.g., Rosenstein and Wyatt, 1990; Yermack 1996; Core et al., 1999; O’Connell and Cramer, 2010; Dulewicz and Herbert, 2004), such as the composition of board, board size, specialized board committees and the outside representation of board of directors.
Firms with better corporate governance provisions are expected to have better firm performance as well as higher firm valuation. Many studies confirm this expectation and come to the similar conclusion that internal corporate governance is closely related to firm
16 performance and firm value. For example, Dahya et al. (2007) report that there is a positive association between percentage of independent directors on board and firm value; O’Connell and Cramer (2010) conclude that the board size is negatively related to firm performance. However, better corporate governance mechanism does not necessarily contribute to better performance. Both Dulewicz and Herbert (2004) and Andres et al. (2005) find no evidence of the association between the composition of board and firm value. Agrawal and Knoeber (1996) even report that there is a negative relationship between outside representation on board of directors and firm performance. The result may be explained by the political purpose-driven expansion of boards, meaning that firms include outside directors like politicians and environmental activists to achieve certain targets.
External corporate governance
External corporate governance refers to the mechanisms that discourage stakeholder participation in the management process (Baber and Liang, 2008). Anti-takeover provisions are typical representations of external corporate governance mechanisms. The most common anti-takeover provisions include staggered board, golden parachutes, poison pills, unequal voting rights, the supermajority requirement for merger and the supermajority requirement for amendment of the bylaws. Adopting more anti-takeover provisions, on the one hand, enhances the management power, on the other hand, impairs shareholder rights (Gompers et al., 2003).
Bebchuk et al. (2009) refer to the anti-takeover provision as entrenchment, which has a negative impact on management behaviors and incentives. As managers are protected from removals, they might have incentives to involve in self-interest serving activities, such as shirking, empire-building and tunneling, to exploit private benefits (Bebchuk et al., 2009; Hoi and Robin, 2010). However, anti-takeover provisions are not necessarily linked to adverse effects. For example, managers can prevent firms from price-distorted takeovers by implementing anti-takeover provisions (Bebchuk et al., 2009).
Firms with stronger shareholder rights are more likely to outperform and have higher firm valuation (Gompers et al., 2003; Bebchuk et al., 2009). Gompers et al. (2003) find evidence
17 that weak shareholder rights lead to additional agency costs. Core et al. (2006) report that firms with weak shareholder rights have poorer operating performance, but they find no evidence that the poor governance leads to poor stock returns.
2.3. Firm performance
Firm performance is the process of quantifying firms’ efficiency and effectiveness (Neely et al., 1995). The measurement of firm performance varies across different studies. Firm performance can be measured through accounting performance measures and market performance measures.
Accounting performance measures, based on historical data, are considered to be backward-looking. Although accounting performance measures are less susceptible to noise compared to market performance measures, they are prone to manipulation. For example, when managers’ compensations depend on firm performance, managers may have incentives to manipulate earnings to boost their compensations. Typical accounting performance measures include Return on Assets (ROA), Return on Equity (ROE), the growth rate of sales and net operating income.
As share price reflects the expected future cash-flow of firms, market performance measures are regarded as forward-looking measures. However, since stock prices of firms are volatile to external events, market performance measures are relatively noisy when compared to accounting measures. Stock return is a typical example of market performance measures. In the studies on the association between corporate governance quality and firm performance, Tobin’s Q (e.g., Agrawal and Knoeber, 1996; Gompers et al., 2003; Gompers et al., 2004; Klapper and Love, 2004; Brown and Caylor, 2006; Villalonga and Amit, 2006; Dahya et al., 2007; Bebchuk et al., 2009; Ammann et al., 2010; Gompers et al., 2010; Hoi and Robin, 2010), return on assets (ROA) (e.g., Böhmer et al., 1999; Klapper and Love, 2004; Larcker et al., 2007) and stock return (Core et al., 2006; Larcker et al., 2007) are the most popular performance measures.
18 metrics as proxies for firm performance. For example, O’Connell and Cramer (2010) verify the board size-firm performance relationship, and they use return on assets (ROA), stock return and Tobin’s Q as performance metrics; Dulewicz and Herbert (2004) use two performance measures, cash flow return on total assets and sales turnover, to examine whether compositions and practices of board bear any relationship to firm performance. In the dual-class setting, abnormal stock returns are widely used as market performance measures, while ROE and ROA are the most commonly used accounting performance measures in IPOs and SEOs (e.g., Jarrell and Poulsen, 1988; Böhmer et al., 1999; Dimitrov and Jain, 2006; Smart et al., 2008; Hossain 2014).
Böhmer et al. (1999) investigate the difference of performances between dual-class and single-class IPOs. Their analysis includes both market and accounting performance measures. In addition to the commonly used market-adjusted abnormal stock market return, they include market-to-book ratio of equity, Tobin's Q and ratio of market value of equity to assets as alternative market performance measures, while they use the growth rate of assets and sales, ROE, net operating income and ROA as accounting operating performance measures. They find similar results from both market and accounting performance measures, indicating that firms with dual-class share structure significantly outperform matched single-class firms in post IPOs. Smart et al. (2008) adopt the similar metrics for post-IPOs-performance of dual-class firms. 3-year-horizon and 5-year-horizon stock return are used as market performance measures, while ROE and ROA are used as proxies for operating performances. When examining agency problems and corporate governance in dual-class firms, Tobin’s Q (e.g., Morck et al., 1988; Gompers et al., 2004; Gompers et al., 2010; Hoi and Robin, 2010) is widely used as the key performance measure.
Both Gompers et al. (2004) and Gompers et al. (2010) use Tobin’s Q as metric for firm valuation in the dual-class setting, and they find that insider ownership of voting rights is negatively associated with firm value. Hoi and Robin (2010) examine the relationship between proximity of controller to the locus of management and firm valuation, measured by the market-to-book ratio of equity, in dual-class firms, and suggest that higher proximity
19 leads to lower firm value.
2.4. Hypothesis development
2.4.1. Corporate governance quality and firm performance in dual-class firms
Although different studies discuss various aspects of corporate governance (e.g., internal governance mechanisms and external governance mechanisms) in different context and use diverse aggregated metrics for corporate governance quality, they come to a similar conclusion that there is a positive relationship between corporate governance quality and firm value (e.g., Gompers et al., 2003; Drobetz et al., 2004; Klapper and Love, 2004; Brown and Caylor, 2006; Bhagat and Bolton, 2008; Bebchuk et al., 2009; Ammann et al., 2010; Cremers and Nair, 2015).
Governance Index (G-Index) introduced by Gompers et al. (2003) and Entrenchment Index (E-Index) introduced by Bebchuk et al. (2009) are examples of external governance indices (Chi and Su, 2016), which are also known as anti-takeover protection indices. Gompers et al. (2003) introduce a Governance Index, turning 24 governance provisions collected from the Investor Responsibility Research Center (IRRC) database into an index, as a proxy for the level of shareholder rights. They report that firms with stronger shareholder rights (lower score of G-Index) tend to have the higher valuation, higher profits, higher growth rate of sales, lower capital expenditure and fewer corporate acquisition activities. Based on the work of Gompers et al. (2003), Bebchuk et al. (2009) introduce an Entrenchment Index with six provisions taken out of the 24 IRRC governance provisions and find that firms with higher Entrenchment Index have a lower firm valuation. Bhagat and Bolton (2008) support the claims that the better governance quality as measured by G-Index and E-Index leads to better current and subsequent operating performance, but they find no evidence to the association between governance quality and future stock performance.
While the above studies concentrate on external corporate governance, there is no reason to ignore internal governance mechanism in the examination of corporate governance-firm
20 performance relationship. Cremers and Nair (2005) report that both external and internal governance mechanisms are linked to firm value, and the two mechanisms act as complements in the association with firm performance.
Some studies attempt to construct a comprehensive measure of governance quality taking both internal and external governance mechanisms into consideration. For example, Ammann et al. (2010) introduce a governance index using 64 individual governance attributes obtained from Governance Metrics International (GMI) dataset of 22 developed countries, and find a significant positive relationship between firm-level governance quality and firm value. Brown and Caylor (2006) create a Gov-Score, based on 51 firm-specific provisions derived from International Shareholder Services (ISS) database, and report that Gov-Score is significantly positively associated with firm value, which is measured by Tobin’s Q, and seven provisions out of 51 fully drive the association between Gov-Score and firm value. While most of the prior studies on the governance-performance relationship are done in single-class context, few studies look into this relationship in a dual-class setting. Since the dual-class share structure is considered to be an effective external governance mechanism, it is interesting to investigate whether this positive relation widely exists in single-class context is also applicable to firms with dual-class share structure.
Consistent with the findings of Hoi and Robin (2010) and Masulis et al. (2009), which states that dual-class firms with higher level of agency conflicts have lower firm value, I expect that there is a strong positive association between corporate governance quality and firm performance in dual-class firms.
Hypothesis 1: Firm performance is positively associated with corporate governance quality in dual-class firms.
2.4.2. The moderating effect of share structure on corporate governance quality-firm performance relationship
Prior studies reveal that ownership structures have mixed effects on corporate governance and firm value (e.g., McConnell and Servaes, 1990; Lemmon and Lins, 2003). Although there
21 is no consensus on whether dual-class share structure impairs corporate governance as well as firm value or not, it is reasonable to anticipate that share structure will have impacts on the corporate governance-firm performance relationship.
The potential effects of dual-class share structure on corporate governance and firm performance may be two-sided (Gompers et al., 2004; Hoi and Robin, 2010; Hossain 2014). On the one hand, dual-class share structure results in more agency problems and has an adverse impact on firm performance. Due to the disproportional voting rights and cash-flow rights, the majority of voting rights are controlled by insiders of dual-class firms, while outside shareholders are at a disadvantageous position. As a result, insiders might have more opportunities and incentives to consume personal benefits, instead of acting in line with the interests of outside shareholders. The conflict of interest between insiders and outside shareholders reflects severe agency problems in dual-class firms. As corporate governance is a set of mechanisms to mitigate agency problems, firms with weaker corporate governance are expected to have more agency problem (Core et al., 1999). Therefore, dual-class firms tend to have higher levels of agency problem as well as weaker corporate governance, compared to one share-one vote firms (Masulis et al., 2009; Hoi and Robin, 2010).
As insiders of dual-class firms hold the majority of voting power, it is rather hard for outside shareholders to vote for a replacement of managements, even if management performance is poor. As managers of dual-class firms face the little stress of replacement, they may have fewer incentives to achieve superior performance, but higher incentives to engage in shareholder value-destroying activities. As a consequence, the firms with dual-class share structure are more likely to underperform, when compared to one share-one vote firms. On the other hand, dual-class share structure allows managers to concentrate more on the long-term growth of firms, without worrying about losing control rights. Dual-class share structure is believed to offer powerful anti-takeover protection to the firms, and it effectively reduces the possibility of hostile takeovers. When managers of dual-class firms focus on value creation for shareholders, internal corporate governance acts as a complement to the share structure. Gompers et al. (2004) compare the external corporate governance between
22 dual-class and single-class firms, and report that dual-class firms have a significantly lower level of G-Index (also known as anti-takeover index), indicating that dual-class share structure offers such powerful anti-takeover protection that it is unnecessary for dual-class firms to take any other protection provisions.
Dual-class share structure does not necessarily result in high level of agency conflict. As most of the superior voting shares are illiquid and held by insiders, it is not that easy for holders of superior shares to exit. Therefore, insiders are forced to act in line with the best interest of outside shareholders (Böhmer et al., 1996; Hossain 2014).
Hoi and Robin (2010) report that the proximity effect, namely the negative association between controller proximity and firm value, is only significant in firms with dual-class share structure, meaning that the lower proximity of controller is more valuable in dual-class firms. In other words, management entrenchment effect may provide potential benefits to dual-class firms, and they are more capable of improving firm performance through enhancing their internal corporate governance. The result indicates that shareholders of dual-class firms that with high proximity must seek additional governance mechanisms to restrain agency costs (Hoi and Robin, 2010).
Taken as a whole, the inherent agency conflicts of dual-class firms arising from the disproportional voting rights and cash-flow rights are expected to result in corporate governance working differently between dual-class firms and one share-one vote firms. The above discussion leads to the second hypothesis:
Hypothesis 2: The corporate governance-firm performance relationship is significantly more positive in dual-class firms.
2.4.3. The moderating influence of firm size on corporate governance quality-firm performance relationship
Klapper and Love (2004) point out that the effect of firm size on corporate governance is unclear. On the one hand, larger firms may suffer from more agency problems, because it is more difficult for shareholders to monitor them. Therefore, larger firms are willing to
23 implement better corporate governance mechanisms to avoid the potential agency costs. On the other hand, smaller firms have incentives to adopt better corporate governance mechanisms to obtain external funds.
Some prior studies confirm that firm size plays a role in the corporate governance-firm performance relationship in single-class context (e.g., Andres et al., 2005; Ariff et al., 2007; O’Connell and Cramer, 2010).
Andres et al. (2005) report that board size is negatively associated with the firm value, indicating that the disadvantages brought by large boards, such as worse coordination and inefficient communication, outweigh the benefits of better management control. They further report that the board size-firm value relationship is influenced by firm size, as firm size influences board size as well as firm value. Results from Ariff et al. (2007) confirm this argument in Malaysia context, suggesting that firm size has a substantial influence on corporate governance ratings. O’Connell and Cramer (2010) verify the moderating effect of firm size on the association between board size and firm performance in the Irish setting, and they show that the board size-performance relationship is less negative for smaller firms. In contrast to the positive association between firm size and corporate governance quality, Gompers et al. (2003) find that the G-Index, also known as the anti-takeover protection index, is positively related to firm size, indicating that larger firms implement more anti-takeover provisions and have weaker shareholder rights.
When moving to a dual-class setting, the situation may be similar but more complex. There are two reasons to suspect that corporate governance works differently in larger and smaller dual-class firms.
Firstly, the size effect discussed above in the single-class setting may also exist in dual-class context. The excessive control rights held by insiders put outside shareholders of dual-class firms in relatively disadvantageous position. For larger dual-class firms, the agency problems may be aggravated, and it is more costly for inferior shareholders to monitor larger dual-class firms.
24 compare the firm size of dual-class and single-class firms and come to the similar conclusion that firms with dual-class share structure are significantly larger than firms with single-class shares structure. For example, Amoako-Adu and Smith (2001) find that firms with dual-class share structure going public on Toronto Stock Exchange (TSE) between 1983 and 1998 are significantly larger than those with one share-one vote structure, when firm size is measured by total assets. Gompers et al. (2010) summarize the characteristic of US dual-class firms in 2000 and find a significant difference in firm size between single-class and dual-class firms. The median total assets and equity market value of dual-class firms ($482 million and $138 million respectively) are significantly larger than those of single-class firms ($138 million and $100 million respectively) (Gompers et al., 2010: 1057). Based on the sample of US firms from 1996 to 2010, Lim (2016) conclude that firm size is positively related to the adoption of dual-class share structure.
Larger firms are more capable of deterring hostile takeover and reducing the likelihood of takeover (Amoako-Adu and Smith, 2001). Comment and Schwert (1995) conclude from prior studies that firm size is the only consistent variable which is significantly associated with the likelihood of hostile takeover. For example, Palepu (1986) obtain negative coefficient for firm size to the likelihood of takeovers; Shleifer and Vishny (1988) find that larger sizes of firms tend to deter hostile takeovers. In order words, firm size deters hostile takeovers.
The protection of hostile takeover is believed to be the primary reason for firms to adopt dual-class share structure. As larger firm size discourages takeovers, it is reasonable to suspect that anti-takeover protection is not the sole reason for larger firms to take dual-class share structure. The underlying incentives for larger and smaller firms to adopt dual-class share structure may be different, which result in distinct designs of dual-class systems, such as the equity ownership structure and the divergence between superior and inferior voting rights. The above facts may lead to different corporate governance quality and firm performance across different size of dual-class firms.
The above discussion leads to the third hypothesis:
25 significantly more positive in larger dual-class firms.
3 Methodology
This empirical study aims to investigate corporate governance as well as firm performance in a dual-class setting. The research sample is based on S&P 1500 firms from 2008 to 2015. Corporate governance quality is the independent variable, while firm performance is the dependent variable. In this section, the sample, data collection, measurement of variables and empirical models are described.
3.1. Sample and data
The sample of the analysis consists of public firms in S&P 1500 index from 2008 to 2015. S&P 1500 firms are chosen as the sample because US stock market is the biggest market which allows firms with dual-class share structure to go public.
Data about corporate governance are collected from Institutional Shareholder Services (ISS). Firm performance and related company information are obtained from Compustat.
The initial sample consists of 2,089 firms (13,687 observations) with data about directors available from ISS-Directors. CUSIP code is used as the identity code to match data about corporate governance with firm performance. After merging the data from ISS with that from Compustat, the sample is reduced to 1,949 firms (11,505 total observations), among which 137 firms are with dual-class share structure, accounting for 8.88% of the total sample. Dual-class firms are identified through the variable “Dual Class Stock” in ISS-Governance database.
3.2. Measurement of corporate governance quality
Corporate governance quality is a key independent variable in all the tests of three hypotheses of this research. Larcker et al. (2007) point out that measuring corporate governance is difficult due to the multi-dimensional nature of corporate governance and the lack of
26 conceptual basis for the selection of relevant governance attributes. Therefore, the measurement of corporate governance does not have any formalized method or consensus, and the metrics for corporate governance used in different studies differ.
In this study, I only discuss internal corporate governance quality. The reason why I do not take external corporate governance into consideration is that dual-class share structure is believed to be a powerful defense for hostile takeover as well as an external corporate governance mechanism. Gompers et al. (2004) report that the average median of G-Index in firms with dual-class share structure is lower than those with single-class share structure, indicating that dual-class share structure is powerful enough for anti-takeover protection, and dual-class firms are unlikely to further implement other anti-takeover provisions. In other words, the share structure is a key factor to consider when firms decide which anti-takeover provisions to take. The above fact leads to weak comparability of the external corporate governance mechanisms across firms with different share structures. Therefore, external corporate governance is not relevant in the discussion of corporate governance quality in this study.
Internal corporate governance quality is measured using B-Index introduced by Baber and Liang (2008). The B-Index comprises of six governance attributes. The B-Index is increased by 1 when the following criteria are met: (1) more than two-thirds of directors on board are independent; (2) all the directors on audit committee are independent; (3) all the directors on compensation committee are independent; (4) firm has a separate independent nominating committee; (5) board size is smaller than the median of distribution for all firms; (6) CEO is not the board chair. The higher values of B-Index indicate the firm has a higher quality of corporate governance.
Table 1 Computation of B-Index
Variable description Measurement
BOD_INDEP Percentage of independent directors on board
BOD_INDEP takes the value 1 if two-thirds of the directors on board are independent directors, and 0 otherwise
27
directors on audit committee
directors in audit committee are independent directors, and 0 otherwise
COM_INDEP
Percentage of independent directors on compensation committee
COM_INDEP takes the value 1 if all of the directors in compensation committee are independent directors, and 0 otherwise
NOM_INDEP
Percentage of independent directors on nominating committee
NOM_INDEP takes the value 1 if there is a separate independent nominating committee, and 0 otherwise
BSIZE Number of board members
BSIZE takes the value 1 if the board size is smaller than the median of distribution for all firms, and 0 otherwise
SEP_CHAIR Whether CEO is the board chair
SEP_CHAIR takes the value 1 if CEO is not the board chair, and 0 otherwise
The B-Index is the sum of the above 6 variables.
3.3. Measurement of firm performance
Firm performance is the dependent variable in this study. To capture an unbiased view of the firm performance, three firm performance metrics are used as dependent variables. Referring to O’Connell and Cramer (2010), the measurement of firm performance consists of three different indicators: Tobin’s Q (TOBINSQ), Return on Asset (ROA) and Stock Return (STOCKRETURN).
Tobin’s Q is widely used for firm valuation. Many prior studies on the corporate governance-firm value relationship in the single-class setting use Tobin’s Q as a measurement of firm value (Brown and Caylor, 2006). In dual-class context, Tobin’s Q is also commonly used as a metric for firm valuation (Gompers at al., 2004; Smith et al., 2009; Gompers et al., 2010). Assuming that the market value and the book value of liabilities are equivalent, Tobin’s Q is also known as the ratio of equity market value to equity book value. In this study, I compute Tobin’s Q as the market value of equity divided by equity book value of equity. ROA is an important indicator measuring the profitability of a company, and it is calculated using net income divided by average total assets. ROA is a typical accounting performance measure, and it is commonly used in prior literature verifying the corporate governance-firm performance association (e.g., Core et al., 1999; Bhagat and Bolton, 2008).
28 While the above two indicators are accounting performance measures, which are backward-looking and prone to manipulations, the stock return is a market performance measure providing a forward-looking and unbiased perspective of the firm (O’Connell and Cramer, 2010). However, stock market-based performance measures are sensitive to investor anticipations (Bhagat and Bolton, 2008). Although stock return might be relatively noisy, it is an effective proxy for firm performance used by prior studies (Core et al., 1999).
Overall, Tobin’s Q, ROA and stock return can be viewed as complementary instead of competing metrics, capturing different features of firm performance (O’Connell and Cramer, 2010).
3.4. Control variable
Firm performance is closely related to firms' characteristic. Therefore, it is necessary to introduce control variables to filter the impacts brought by elements other than corporate governance characteristics.
Here I include firm size, the two-year growth of sales, research intensity, capital intensity and leverage as control variables. The selection of control variables is guided by Hoi and Robin (2010) and Ammann et al. (2010).
Firm size is an essential factor which has a great impact on firm performance. Firm size (LOGSIZE) is computed as the natural logarithm of total assets.
Following Amman et al. (2010), the past two-year growth of sales (GROWTH) is used to control the impact of sales growth among different firms.
Following Hoi and Robin (2010), research intensity and capital intensity are proxies for potential investment opportunities, which may also lead to different firm performances. Research intensity (RDINT) is the ratio of Research and Development (R&D) expenditure to net sales, while Capital intensity (CAPINT) is the ratio of capital expenditure to total assets (Hoi and Robin, 2010).
McConnell and Servaes (1995) report that the effect of leverage on firm value is double-sided. Therefore, it is necessary to include the leverage as a control variable. Leverage
29 (LEVERAGE) is calculated as the ratio of book value of total liabilities to total assets.
3.5. Empirical models
I define a panel dataset as the data comprise of 1,949 firms (cross-section) from 2008 to 2015 (time series). The variation across different firms is assumed to be random and is uncorrelated with independent variables in the model. Thus, tests of all the three hypotheses are based on Random-effects Generalized Least Squares (GLS) regression model.
In each test of the hypotheses, regressions are run using three different firm performance metrics mentioned above. The empirical models of the three hypotheses are defined as follow. Hypothesis 1 expects that firm performance is positively associated with internal corporate governance quality in dual-class firms. Regressions are run based on the equation (1) only in the sample of dual-class firms. If the result supports the hypothesis, the coefficient of CGQ will have a positive value.
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑖𝑡 = 𝛽0+ 𝛽1𝐶𝐺𝑄𝑖𝑡+ 𝛽2𝐿𝑂𝐺𝑆𝐼𝑍𝐸it+ 𝛽3𝐺𝑅𝑂𝑊𝑇𝐻𝑖𝑡+ 𝛽4𝑅𝐷𝐼𝑁𝑇𝑖𝑡+ 𝛽5𝐶𝐴𝑃𝐼𝑁𝑇𝑖𝑡+ 𝛽6𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡 + 𝜀𝑖𝑡 (1)
Hypothesis 2 predicts that the corporate governance quality-firm performance relationship is significantly more positive in dual-class firms. Regressions are run based on the equation (2) in the full sample.
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑖𝑡 = 𝛽0+ 𝛽1𝐶𝐺𝑄𝑖𝑡+ 𝛽2𝐷1it+ 𝛽3(𝐷1it× 𝐶𝐺𝑄it) + 𝛽4𝐿𝑂𝐺𝑆𝐼𝑍𝐸it+ 𝛽5𝐺𝑅𝑂𝑊𝑇𝐻𝑖𝑡+ 𝛽6𝑅𝐷𝐼𝑁𝑇𝑖𝑡+ 𝛽7𝐶𝐴𝑃𝐼𝑁𝑇𝑖𝑡+ 𝛽8𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡+ 𝜀𝑖𝑡 (2)
Dummy variable D1 is introduced to classify whether firms are single-class firms or dual-class firms. When firms have dual-class share structure, D1 takes the value 1, and otherwise 0.
To test if there is any share structure-related effect influencing the corporate governance-firm performance relationship, interaction item D1*CGQ is introduced in the model. Hypothesis 2 predicts that the coefficient of D1*CGQ will have a positive value.
Hypothesis 3 anticipates that larger dual-class firms have a significantly more positive corporate governance quality-firm performance relationship. Regressions are run based on
30 the equation (3) only in the dual-class sample.
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑖𝑡 = 𝛽0+ 𝛽1𝐶𝐺𝑄𝑖𝑡+ 𝛽2𝐷2it+ 𝛽3(𝐷2it× 𝐶𝐺𝑄it) + 𝛽4𝐺𝑅𝑂𝑊𝑇𝐻𝑖𝑡+ 𝛽5𝑅𝐷𝐼𝑁𝑇𝑖𝑡+ 𝛽6𝐶𝐴𝑃𝐼𝑁𝑇𝑖𝑡+ 𝛽7𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡+ 𝜀𝑖𝑡 (3)
Dummy variable D2 is introduced to capture the size effect. Firm size is classified based on the market capitalization at the end of each year. If the firm's market capitalization is higher than the median market capitalization, then the firm is classified as larger firm, otherwise smaller firm. When firms are classified as larger firms, D2 takes the value 1, and otherwise 0. To test whether firm size is a moderator on the association between governance quality and performance, interaction item D2*CGQ is introduced in the model. If the results align with the expectation, the coefficient of D2*CGQ will have a positive value.
Table 2 Variable definitions
Variable Definition
CGQ The value of B-Index
D1 D1=1 if firm is dual-class firms, and 0 otherwise
D2 D2=1 if firm’s market capitalization is larger than median of distribution for all firms, and 0 otherwise
LOGSIZE Natural logarithm of total assets GROWTH Two-year growth of sales
RDINT Ratio of R&D expenditure to net sales CAPINT Ratio of capital expenditure to total assets LEVERAGE Ratio of total debt to total assets
TOBINSQ Ratio of market value to book value of equity
ROA Ratio of earnings before interest and tax to total assets STOCKRETURN Ending stock price less initial stock price plus dividends
4 Results
Results of the empirical models mentioned above are shown in this section. Firstly the descriptive summary of the variables is presented, followed by the empirical findings. Lastly, a test of endogeneity is carried out to identify whether potential impact of endogeneity exists in the empirical findings.
31
4.1. Descriptive statistics
Table 3 displays the descriptive summary of the sample. The sample is further described under classifications of share structure and firm size.
Table 3 Summary statistics of variables
Variable ALL D1=1 D1=0 P-value D2=1 D2=0 P-value
Observation 11,505 680 10,825 342 338 CGQ Mean 4.796 4.363 4.823 0.000 4.418 4.308 0.189 Std dev. 0.853 1.094 0.828 1.106 1.081 Median 5 5 5 0.000 5 4 0.155 LOGSIZE Mean 3.577 3.550 3.578 0.338 3.849 3.248 0.000 Std dev. 0.742 0.722 0.743 0.650 0.663 Median 3.529 3.491 3.535 0.226 3.754 3.168 0.000 GROWTH Mean 5.174 6.002 5.122 0.582 12.822 0.355 0.045 Std dev. 40.406 72.570 37.480 107.141 2.760 Median 0.022 0.022 0.023 0.813 0.106 -0.010 0.000 RDINT Mean 0.033 0.022 0.033 0.271 0.023 0.031 0.098 Std dev. 0.259 0.058 0.266 0.053 0.073 Median 0.000 0.000 0.000 0.007 0.000 0.000 0.548 CAPINT Mean 0.035 0.035 0.035 0.933 0.044 0.026 0.000 Std dev. 0.049 0.051 0.049 0.051 0.049 Median 0.020 0.018 0.020 0.476 0.028 0.008 0.000 LEVERAGE Mean 0.574 0.560 0.575 0.103 0.592 0.527 0.002 Std dev. 0.238 0.270 0.236 0.289 0.246 Median 0.571 0.543 0.573 0.010 0.571 0.508 0.000 TOBINSQ Mean 2.757 2.599 2.767 0.850 3.621 1.564 0.004 Std dev. 22.425 9.387 22.999 12.807 3.070 Median 1.706 1.632 1.709 0.501 2.401 1.221 0.000 ROA Mean 0.076 0.081 0.075 0.187 0.106 0.055 0.000 Std dev. 0.100 0.084 0.101 0.085 0.074 Median 0.065 0.072 0.065 0.218 0.095 0.041 0.000 STOCKRE TURN Mean 209.384 150.281 213.097 0.040 284.606 46.699 0.000 Std dev. 774.903 410.024 792.097 553.904 173.244 Median 26.589 27.436 26.457 0.962 99.424 11.616 0.000 The p-value for difference represents Two-sample t-test with equal variance in means and Wilcoxon rank-sum test of the variance of median across dual-class and single-class firms, larger and smaller dual-class firms, with 95% confidence level.
32 Among the 11,505 observations, 680 are from dual-class firms, accounting for 5.91% of the total sample.
From the description of the variables, although the median of CGQ are identical (both are 5) among dual-class and single-class firms, the mean of CGQ in dual-class firms is lower than that of one share-one vote firms (4.363 versus 4.823). Both the difference of mean and median between the two categories of firms is significant at 1% level.
When turning to control variables, firms with dual-class share structure have slightly lower LOGSIZE mean (3.550) and median (3.491) than those with single-class share structure (3.578 and 3.491 respectively), while the mean and median of GROWTH present mixed results. Both the mean (56.0%) and median (54.3%) of LEVERAGE of dual-class firms are lower than those of single-class firms (57.5% and 57.3% respectively). All of the control variables, LOGSIZE, GROWTH, RDINT, CAPINT, and LEVERAGE, are similar across dual-class firms and single-class firms, and there is no significant difference between the two categories of firms, except for median of RDINT and LEVERAGE (with the p-value of 0.007 and 0.010). The results imply that the sample dual-class firms and single-class firms have similar firm size, sales growth and capitalization intensity. The results of firm size and leverage are in contrast to the results from Gompers et al. (2004), which report that US dual-class firms in 2001 have significantly higher values of total assets as well as market value, and higher levels of leverage compared to single-class firms.
When looking into firm performance, both the mean (2.599) and median (1.632) of TOBINSQ in dual-class firms are lower than those of single-class firms (2.767 and 1.709 respectively), although the differences are not statistically significant. The mean of STOCKRETURN in dual-class firms (150.281) is substantially lower than that of single-class firms (213.097), but the median of STOCKRETURN across the two kinds of firms shows a small difference (27.436 versus 26.457). The difference of mean between STOCKRETURN among two categories of firms is significant at 5% level (with p-value=0.04), while the difference of median is not significant. However, when measuring firm performance by ROA, dual-class firms have average ROA of 8.1% while one share-one vote firms have 7.5%,
33 indicating that dual-class firms have a higher rate of return on assets than single-class firms. Such a phenomenon might be due to the effect of firm size and the disproportional sample size across dual-class and single-class firms. The differences of mean and median of ROA between the two categories firms are insignificant.
Overall, the firm performance of sample single-class firms is better than that of sample dual-class firms. There is a significant difference when using STOCKRETURN as the performance metric, while there is no significant difference in TOBINSQ and ROA between the two categories firms. We can conclude from the sample description that dual-class firms have poorer corporate governance quality and lower firm performance, compared to one share-one vote firms.
When classifying firms based on market capitalization, 342 observations out of 680 are from larger dual-class firms, while the rest 338 observations are from smaller dual-class firms. Larger dual-class firms have higher mean (4.418) as well as higher median (5) of CGQ than smaller dual-class firms (4.308 and 4 respectively). However, none of the differences of mean and median between larger and smaller dual-class firms is significant. The result indicates that larger dual-class firms tend to have better internal corporate governance quality than smaller firms, but the difference is not significant.
When looking into control variables, larger dual-class firms have higher mean and median of GROWTH (1,282.2% and 10.6% respectively) than smaller dual-class firms (35.5% and -1% respectively). Size effect also exists in CAPINT and LEVERAGE. Both the mean (59.2%) and median (57.1%) of LEVERAGE in larger dual-class firms are significantly higher than smaller dual-class firms (52.7% and 50.8% respectively), revealing that larger dual-class firms use more debt financing and rely more on leverage effects than smaller firms. Both the mean and median of GROWTH, CAPINT and LEVERAGE reveal significant variance between larger firms and smaller dual-class firms. The result implies that the sales growth and the capital intensity differ among firms with different size, and it is necessary to control for these three factors when investigating the corporate governance quality-firm performance relationship.
