Does board independence affect firm performance? A
perspective from China
Amsterdam Business School
Name Danni SunStudent number 10707565
Program Economics & Business Specialization Finance
Number of ECTS 15
Supervisor Ilko Naarborg Completion July 6th, 2015
Abstract
The board of directors system is introduced from corporate America to corporate China, whereas empirical studies and the practitioners’ experience find mixed evidence on the efficacy of independent directors. This paper conducts a large-sample, long-horizon and comprehensive examination on the relationship between board independence and firm performance in China, with the effort to minimize the endogeneity problem inherent in corporate governance mechanism. I find: generally there is no correlation between board independence and firm performance; whether a firm is owned by the state does not affect the board-performance relationship.
Statement of Originality
This document is written by Student Danni Sun 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.
Table of Contents
1. Introduction ... 1
2. Literature review ... 3
2.1 International evidence ... 3
2.2 Institutional background in China ... 7
3. Hypotheses and methodology ... 10
4. Data ... 13
4.1 Performance variables ... 13
4.2 Board independence variables ... 14
4.3 Control variables ... 15
4.4 Instrumental variables ... 16
5. Analysis ... 18
5.1 Fixed-effects estimation ... 21
5.2 Simultaneous equations ... 22
5.3 State-owned enterprise, board independence and firm performance ... 26
6. Robustness check ... 26
7. Conclusion and discussion ... 31
1. Introduction
Board independence has long been seen as a crucial composition of corporate internal control mechanism since the last century. Literally, independent directors are independent of material interests in the company. Thus ideally they are there to ensure the boards to carry out fiduciary duty and monitor the top management.
However, it is still an unsolved puzzle whether board independence can indeed promote firm profitability, since empirical studies regarding this area show controversial outcomes, and are highly curved by the endogeneity problem. Bhagat and Black (2002) document the non-correlation between board independence and long-term firm performance in the United States. Nevertheless, under non-American context, studies present a reversed outcome that there is a consistent causal relationship between board independence and firm performance. For example, Dahya and McConnell (2007) and Black and Kim (2012) document empirical evidence from the UK and Korea respectively. As the second largest economy in the world, China has its unique institutional circumstance and corporate characteristics. For instance, according to Wang (2014), a large amount of Chinese public firms still have the state as their major shareholder, similar to numbers of other emerging economies but not typical in western countries. Thus, whether this positive relationship outside the United States is also the case in China leaves open.
While prior research has examined this question in China, researchers mostly examine the year-by-year correlations with the attempt to minimize the effects of endogeneity problem, which could arise from unobserved heterogeneity, simultaneity and reverse causality (Adams et al., 2008). Therefore, my study focuses on the dynamic process of the influence of changing independent directors on future firm performance. The rationale is that once board structure changes, the present firm profitability does not necessarily reflect this immediately, but the future performance will, i.e. a time lag is required for the effect to become visible (Bhagat and Black, 2002). In addition, with the development of corporate governance and maturity of corporates in China, it is highly likely that the outcome will turn out to be
different from the past. Although Liu et al. (2015) and other articles report positive correlation between board independence and firm performance through their long horizon investigation, my study find a non-correlation result in the end.
Considering the specific institutional background in China, I also study the impact of state-owned enterprises (SOEs) on the relationship between board independence and firm performance. One fact in China is that most firms are controlled by the government, and related studies show that because of social or political goals, the government may lack the ability to monitor management (Chen et al., 2011), leading to agency problems and under-functioned directors. What’s more, prior research mostly simply focus on the board composition and firm performance, while I also pay attention to the time specialty. As is all known the 2008 global crisis has great effect on global financial market, it influenced Chinese stock market and corporate operation as well. I test the impact of this special period in robustness section and eventually find out that the non-correlation results is regardless of the crisis.
An important and inevitable question in board-performance relationship in our condition is the endogeneity problem. While board composition could affect firm performance under our assumption, once a firm is underperforming, it is highly likely to increase its independent directors due to common wisdom. This is the so-called reverse causality. Endogeneity may also come from simultaneity and omitted variables. In order to robustly test the impact of independent directors on firm performance, I will use fixed effects (FE) and two-stage least squares (2SLS) estimation methods, with clustered standard errors to adjust the potential heteroskedasticity.
My paper is organized as follows. Section 2 reviews the empirical and theoretical literature on corporate boards and performance, along with the Chinese institutional context. Section 3 gives a description of my hypothesis and methodology. Section 4 gives a data description and section 5 presents the results of empirical tests as well as the corresponding analysis. Then section 6 provides additional robustness checks. In the end section 7 provides concluding contents and policy implications.
2. Literature review
2.1 International evidence
A large number of studies have investigated the relationship between corporate governance mechanism and firm finance performance. Among the extensive corporate governance system, scholars further analyze different aspects of corporate boards, including the board composition (the ratio of independent directors on inside directors or the number of independent directors), the background of directors, the characteristics of boards and so on (e.g. Baysinger and Butler, 1985; Jensen, 1993; Yermack, 1996). However, it is still a controversy issue to make a unified conclusion regarding this topic, no matter in China or in an international background. Generally, three kinds of empirical results are found, namely a positive, a negative or no correlation between independent directors and firm performance. As Fama and Jensen (1983) point out, corporate boards have the incentive to “carry out their tasks and do not collude with managers to expropriate residual claimants”. From this understanding, the existence of directors is able to lower the possibility of collusion from inside directors and top managers, thereby impede the stockholders’ wealth or firm value from being divested as much as possible. Other studies like Baysinger & Butler (1985) and Rosenstein & Wyatt (1990) also hold this point of view and provide empirical supports. Specifically, the 266 U.S. firms that Baysinger and Butler uses provide factual evidence that having more independent directors on board increases firm performance that is measured by the relative financial performance (RFP). While Rosenstein and Wyatt find a positive correlation as well, they manage to test it in a cleaner way to address the intractable endogeneity problem by examining the stock price reaction on the day of the announcement that new independent directors will be added on board. They find a significant 0.2% stock price growth, which indicates that independent directors could increase shareholder value indirectly.
In addition, related studies vary from using single-country data as well as multi-country data. For example, Dahya and McConnell (2007) demonstrate a positive effect in examining the connection between board independence and firm performance in the UK over the period 1989–1996, a period that includes a exogenous shock of the publication of the Cadbury Report calling for at least three outside directors for publicly-traded corporations. Black and Kim (2012) report the 1999 Korean law of mandating 50% outside directors to produce economically large share price increases for large firms; their share prices boost in 1999 when the reforms are announced. Black and Khanna (2007) have the similar findings in India. As for multinational studies, Dahya et al. (2008) investigate data from 22 countries and report a significant positive relation between corporate value and the proportion of independent directors. Furthermore, this relation is specifically pronounced in countries that have weak legal protection on shareholders.
Contrary to the above empirical findings, there are a number of research from other scholars that find a negative relationship between board independence and corporate performance. For example, in his main aim of exploring the association of board size and firm performance, Yermack (1996) also presents a consistently negative effect of the proportion of independent directors on firm performance in a panel of main US companies.
There are also abundant researches reporting non-correlation results as well as controversial evidence. For example, Hermalin and Weisbach (1991) report in their study of 142 NYSE firms that, except for a negative relation between the different proportion of independent directors and firm performance measured by Tobin’s Q, they also document that “there appears to be no relation” between board composition measured by the percentage of independent directors and firm performance. They give a possible explanation that perhaps the effect of independent directors simply does not exist. That is to say, inside and outside directors are “equally bad (or, possibly, good)” due to top managements’ control during the board member selection process. Actually, studies end up with
weak-correlation outcomes are not unusual at all.
One influential empirical research to this field of research is from Bhagat and Black (2001). They study the main firm performance measures from 1985-1995 with the board directors’ data in early 1991 and seek to find the answer to the relationship between board independence, board size and firm performance. Using Tobin’s q, return on assets, ratio of sales to assets and market adjusted stock price returns, it turns out that there is a strong inverse correlation between board independence and firm performance. Hence they conclude that having more independent directors in the board does not help to improve the firm’s profit, and perhaps these firms even perform worse than the others. In discussing the potential rationale behind the result deviating from conventional wisdom, three reasons are given: (1) Insiders are probably better at strategic planning decisions; (2) the information asymmetry: independent directors sometimes are not well-informed about the company while inside directors relatively know more about corporate issues; (3) tradeoff between independence and incentives. Their finding is also in line with a stream of empirical studies (Klein, 1998; Mishra and Nielsen, 2000; Bhagat and Bolton, 2008). Furthermore, they document two approaches through which board behavior can affect a firm: discrete tasks and overall performance. Discrete tasks involve replacing CEO, making take over bids and other top decisions, while overall performance is rather difficult to measure and prone to be noisy due to long observation span. Despite the aforementioned international empirical evidences, much has been learned about independent directors in Chinese listed companies. According to Wang (2014), this field of research started late in China with the earliest one from Li & Li (2001), because the independent directors is first introduced from U.S. in 2001. But in the mean time, related study has been highly centralized and rich in the past decade till now, which demonstrates that corporate governance remains a hot topic in China. Similar to those peer studies, most investigations in China also use measurements including ROA, ROE, and Tobin’s q to proxy firm profitability and ratio of independent board members or the absolute numbers of independent directors to proxy board independence. Other indication of firm performance variable include
the profit stability measured by E/P (initial stock price per share, Zhao et al., 2008), cumulative abnormal return (Hu and Shen, 2002), EPS (earnings per share, e.g. Ma, 2004; Shen et al., 2007), SOA (sales on assets, e.g. Lü and Lü, 2003; Shen et al., 2007) etc. As a summary of conclusion, Wang (2014) reports a ratio of approximately 2 to 1 that studies documenting a positive to negative and no relationship between independent directors and firm performance from his 30 selected empirical studies. He further evaluates the empirical evidence as a certain extent of confusing because these articles fail to identify which kind of role are independent boards playing in Chinese companies. In China it is widely conceived that independent directors are “vase directors” that do not play an effective monitoring role but only advisors; some extreme case is that they merely present to meet the regulatory requirements towards corporates. However, in a more recent study proposed by Liu et al. (2015), the authors provide robust and positive relationship between board independence and firm performance using fixed effects, IV-2SLS and diff-in-diff estimations that well tackled the endogeneity problem. Not included in Wang’s meta-empirical survey, this literature is one of my key references.
However, in formal theory, there is a vacuum regarding this corporate finance field. Only a few studies manage to discuss and make a conclusion in a theoretical way. One of these studies is the survey of the economic literature on boards of directors from Hermalin and Weisbach (2001). They find that most related economic works mainly focus on three pairs of relationships: (1) board characteristics, such as board composition, and firm profitability 2) board characteristics and board actions 3) the factors affecting the makeup of boards and their evolution over time? This survey provides integrated economic supports and background knowledge including several critical conceptual issues. For example, regarding the necessity of board, they point out that as en economic institution, boards of directors help to solve the agency conflicts inherent in managing an organization. Here they use an “organization” instead of “firm” because they believe that the existence of governing boards “predates” the regulatory requirements. Regarding the first question in the survey, which is closely related to my study, Hermalin and Weisbach report two approaches of empirical tests: one is to examine the relationship between accounting
performance measure and ratio of independent directors; the other one is to use Tobin’s q. Tobin’s q is first suggested by Morck et al., 1988. The rationale behind the latter measure is that this ratio reflects the added value of invisible factors such as governance. In other peer studies (see, e.g. Yang & Zhao, 2014; Liu et al., 2015), Tobin’s q is deemed as a market-based measure that captures all firm operation aspects. Overall this wide-scope study provides us another method to detect firm performance and gives a deep understanding of corporate governance system. Further, Hermalin and Weisbach evaluate these studies by considering whether they not only test contemporaneous correlations, but also use simultaneous equation methods to correct the reverse causality effect. Finally, except for finding a negative relationship between board size and profitability, they conclude from the empirical literature a general poor result in testing the correlation of board composition and firm performance.
A recent and also the first theoretical claim on the efficacy of independent directors within China’s terrain is from Wang (2014). This integrated paper collects 30 related articles in Chinese public companies in order to generalize different empirical evidence and in the end find that the evidence from the Chinese context is consistent with most of the international studies: mixed with either positive or negative or no correlation.
2.2 Institutional background in China
Institutional background is extraordinarily important for corporate governance studies. It is especially paramount in China since the whole system of independent directors is introduced from America to China.
Generally, there are two fundamental theories in western corporate governance research concerning this conventional wisdom. Agency theory, put forward by Fama (1980), Jenson and Meckling (1976), assume that individuals usually act at their own interests in nature and it is necessary to resort to outside directors to monitor managerial behavior. Therefore agency problems can be alleviated and interests of
shareholders are protected. An alternative prescription is the stewardship theory. Davis et al. (1997) assumed human behavior to be positive, and managers can fulfill their own duty to maximize the benefit of a firm. These two theories are contradictory with each other to some extent, and modern corporate reform in China is more partial to the Agency Theory, implying that to some extent top managers are very much likely to make strategic decisions over the best choice for the interests of the firm owners.
While pyramidal schemes are common in western corporates, one-share-one-vote rule and one dominant shareholder is more frequently observed in Chinese firms. This phenomenon inherits from the late start of China’s modern corporatization process and stock trading system (Xu and Wang, 1999). The stock companies did not go through a heated growth until time comes to the 1990s. Chinese stock exchange does not appear before 1990, when the Shanghai Stock Exchange (SHSE) was established and before long the Shenzhen Stock Exhcange (SZSE) opened in 1991. Since then, the Chinese corporates has been growing rapidly in its immature state, and under the significant impact of strict regulations from the government. A noticeable policy change is in August 2001, when the China Securities Regulatory Commission (CSRC) issued the Guidance Opinion on the Establishment of an Independent Director System in Listed Companies (hereafter the 2001 Guidance). All companies listed on Chinese stock exchanges must adjust their institution of boards of directors in accordance with this unprecedented comprehensive guidance, namely, to regulate independent directors at a least proportion of one-third. However, till present the mainstream viewpoint from scholars and social groups that independent directors are “vase directors” in China remains unchanged. Hence it is time to mention the ownership structure, which is also a distinct characteristic in China. Currently, most corporate are transformed from originally state-owned enterprises, leading to a massive fraction of state controlling. As a matter of fact, in listed companies, as much as 44.8% of the total shares are hold by a single major shareholder on average (Bai et al., 2004), including the state as the major owner. Having said that though, the proportion of state shares has declined slightly over time, but experienced a momentous change after the 2007 Split-share Structure
Reform.
One of the most crucial background information in China’s context is the composition of shares in the two stock markets. Shares are classified into six different categories: domestic (A-shares), foreign shares (B-, H-, N-, S-, T-shares). Specifically, A-shares are denominated in RMB and for domestic investors. B-shares are also denominated in RMB but designed for foreigners, so can only be bought by foreign currencies. Companies issuing H-shares are listed in Hong Kong Stock Exchange (HKSE) and these shares are priced by Hong Kong dollars or US dollars. N-, S-, T-shares are Chinese companies but publicly traded in US, Singaporean and Japanese market respectively. Among A-shares, there are further four classifications according to the nature of equity holders: (1) state shares, (2) legal person shares, (3) individual shares and (4) foreign shares. The state shares are held by central and local governments, or state-owned enterprise (Xu & Wang, 1999). Besides, not all A-shares are tradable. Before 2005, the outset of Split-share Structure Reform, two-thirds of A-shares in China’s stock market were still non-tradable shares, while by the end of 2008 most were mandated to transfer into tradable-shares under the regulation proposed by the CSRC (Liu et al., 2015). As its name manifests, non-tradable shares cannot be traded on secondary market, giving rise to severe information asymmetry. Therefore, the large-scale, radical reform is highly likely to cause heterogeneity in the long investigation period of related study. It is also why my study period starts from 2005 in order to mitigate its influence as much as possible but still manage to keep certain amount of samples. Because of the disagreement among academia on how to measure the market value of non-tradable shares on Chinese stock markets (Yeh et al., 2009), this paper mainly focuses on tradable A-shares.
The contribution of this thesis lies in three aspects. First of all, it extends the study of Bhagat and Black (2002) by expanding the time horizon of study in a Chinese context. Although Bhagat and Black state that their study is over a 10-years time span, they actually only use data of board of directors for 1 single year, while I intend to panel a more comprehensive series of data including 9-years’ corporate performance and board details of Chinese listed companies. Secondly, as Liu et al. (2015) point out,
most existing literature under China’s context fails to robustly test the impact of independent directors on firm operating performance; they either ignores the econometric challenge of endogeneity or do not manage to address it. Thus this work proposes to reach statistically and economically meaningful consequences after controlling for firm size, ownership, industry membership, growth opportunities and other alternative corporate governance structures. Third, bringing in the state-owned enterprise as supplementary research allows us to examine the links between performance and the presence of independent board members with Chinese characteristics. Overall, my thesis follows the common practice of empirical test in order to explore the relationship between board composition and firm performance.
3. Hypotheses and methodology
Although various literature report ambiguous correlation between board independence and overall firm performance, an interesting phenomenon is that more and more official institutions and policymakers are treating a majority-outside board as “good” corporate governance and mandating a certain amount of independent directors on the boards of public companies. During the 1990s, at least 26 countries call for the representation of outside directors on public companies’ boards. For example, in 1992, in an official document known as the Cadbury Report, the UK established a minimum of three outside directors in publicly traded firms. The Stock Exchange of Hong Kong (HKEX) required at least 2 independent non-executive directors for each listed companies in its 1993 Growth Enterprise Market Listing Rules. When it comes to China, the CSRC issued its 2001 Guidance, requiring at least one-thirds of the board representations to be outside directors for Chinese publicly traded firms. Not only these official regulations but informal surveys reveal people’s preference on independent board as well. McKinsey & Company (1999-2002) offers evidence from their surveys that 80 percent of institutional investors are willing to pay a premium to firms with good governance. The logic here is that an independent board consists a great part of effective internal governance mechanism. Hence we can conclude that most practitioners, namely regulatory authorities, corporations and investors believe that an independent board helps enhance the firm value,
although scholars hold controversial opinions.
In essence, the paramount assumption of my first hypothesis is that independent di- rectors lead to a stronger board so that this majority-independent board has monitoring functions restricting the behavior of management, keep insider self-dealing from happening, mitigating agency problem and protecting minority shareholders. Decades ago, Berle and Means (1932) documents that managers usually pursue their own interests rather than those of shareholders. Under this context, board appears and acts as a monitoring role to counterbalance the power of managers. When the board is strong enough to prevent the major shareholders from taking advantage of corporate resources to earn their own benefits, the board is said to be effective and the corporate value will not be sacrificed due to managers’ personal exploitation. As cited in Dhaya et al., 2008, “Dominant shareholders have an incentive and, in the absence of a countervailing force, the ability to divert corporate resources from other shareholders to themselves for personal consumption. Such diversion reduces the observed market value of the firm.” Thus independent directors have an incentive to conduct their own tasks and protect the interest of firm owners.
However, even if more independent directors means a more independent board and suppose they could function well, whether this will increase firm performance in the end seems unclear. This question happens to have the same view with Dhaya et al., 2008: What power does an outside director have to control the dominant shareholder even if he chooses to be an effective monitor? Thus the second fundamental premise underlying my first hypothesis is that with the supervision of independent directors, the decision that the management makes increases firm performance. These discussions give rise to the following hypothesis:
Hypothesis 1: An increase of independent directors in board will promote firm performance in the short run in China.
Regarding the effect of state-owned enterprise on firm value, numerous studies has demonstrated that the state being the major shareholder of a company threatens
the functioning of board (Liu et al., 2012) and has significant negative effect on firm performance (Qi et al., 2000; Bai et al., 2004; Liu et al., 2015). The reasons are as following. First of all, most of the time the government pursues other objectives such as political aims and creating employment instead of maximizing profits through SOEs. So the board either has no strong incentive or is not able to monitor management behavior. Secondly, firms owned by the government represent the “ultimate separation between ownership and control” (Liu et al., 2012). In China, an SOE is said to be owned by the nation’s citizens but actually operated by bureaucrats or managers who usually have not enough expertise. Shareholder rights are represented by finance bureaus of the local governments. The well being of managers are not directly tied to corporate profitability. Hence in this context, agency conflict is most severe and I assume that being owned by the state influences the original relationship between board independence and corporate profitability. These reasons lead to my second hypothesis:
Hypothesis 2: The monetary benefits arising from greater board independence are more pronounced in state-owned enterprises (SOEs) in China.
Regarding the research design, this paper follows Bhagat & Black, 2002 and Liu et al., 2012. As mentioned in Bhagat & Black, 2002, the authors expect board composition to affect performance only gradually. As I understand it, a firm’s explicit performance is not increased or decreased as soon as the board composition goes through some structural change. That is to say, we allow a time lag for performance data to reveal its fluctuation. It might be imprecise if someone directly regress firm performance proxies on explaining variables within the same year. However, my investigation reveals that most other researches in this field use the simple method, which is estimating the impact of either the independent directors’ proportion or the absolute number of their position on board on firm performance simultaneously, for example, they use the data from the same year. Therefore, following Bhagat & Black’s approach, I use cumulative performance data over two years after the change of independent directors occurs instead of measuring firm performance and board composition within the same year. In practice, for instance, if I intend to estimate the impact of board change in year 2005, independent directors and other control
variables data will be collected on the year 2005 while the performance data will be the average value of a “prospective” period: from 2005 to 2006. By doing so, it might capture the influence that firm performance comes under more completely and objectively.
To test the second hypothesis, samples are divided into SOEs and non-SOEs based on whether the major owner of a firm is the state. Dummy SOE equals to 1 if a firm is state-owned and 0 otherwise. Then regressions similar to the aforementioned equations are explicitly conducted in order to check the effect of government on the relationships between board independence and firm profitability.
4. Data
The data is obtained through the Chinese Securities Market and Accounting Research (CSMAR) database, which provides comprehensive Chinese public companies' financial statements data and corporate governance indicators. Samples covering all industries except financial sector in Shanghai Stock Exchange (SHSE) and Shenzhen Stock Exchange (SZSE) from 2005 to 2013 are collected. To avoid the influence of abnormal finance status from some companies, I exclude stocks carrying “ST” (special treatment) tags from all domestic shares (A-shares); this is the first step by Chinese stock exchanges in delisting a stock. Additionally, missing samples are discarded. Thereby the final sample consists of 957 public firms and 8613 firm-years.
4.1 Performance variables
To measure firm performance, this thesis relies on two commonly used indicators: one is an accounting measure, return on assets (ROA); the other is a market-based measure, Tobin's q. ROA is calculated by operating income divided by a firm’s total assets. The canonical Tobin’s q is defined as the ratio of market value of a firm’s assets over the book value of its assets. A relatively high Tobin’s q value reflects a high market expectation on that firm; it also means satisfactory firm management to
its investors, great growth opportunity and promising profitability. However, following Bhagat and Black (2002), both of these two measures are not exploited directly, but generate an average of two consecutive years, which is the key feature of my analysis. For example, the dependent variable for fiscal year 2005 is the mean of Tobin’s Q or ROA on fiscal year 2005 and 2006, denoted as avg_Q and avg_ROA respectively.
As panel A of Table 1 shows, from 2005 to 2013 the average 2-year cumulate Tobin’s q and ROA are 1.467 and 5.2% respectively, after winsorizing the variables by 1% at both tails in order to reduce the influence of extremums. Compared to Liu et al. (2015) and Jiang et al. (2010), my results show no extraordinary abnormal values on cumulative ROA since their single year mean ROA are 3.6% and 2.8% respectively. Regarding Tobin’s q, Zheng & Lü (2009) report a mean yearly Q of 1.16, while my 2-years value is 1.467, which may arise from the instability of market expectation on listed firms, but still comparable with the international norm. Back to the Q value itself, we can conclude that overall the market values of these firms exceed their replacement costs and that our domestic investors on stock market hold an optimism view towards these companies.
4.2 Board independence variables
An equivocal issue in this research is the so-called ''independent director''. Due to various purposes, different studies seek different norms of independent directors despite their different meanings, extant manifestations including "non-interested," "independent," "outside," "non-executive," "non-employee," and "disinterested" (Clarke, 2006). According to the definition from Chinese Securities Regulatory Commission (CSRC), the term in my thesis represents the directors that are independent of management and the relatives of the management, the controlling shareholder, and the persons providing financial, legal or consulting services to the company.
board members divided by total number of directors. As a supplementary measurement of board independence, I follow Ye et al. (2007) and also use the dummy of whether the number of a firm’s independent directors exceeds the median of the entire sample, which is denoted as Ddirct.
From Panel B of Table 1, we can see that the mean fraction of independent directors in China is 36.2% over the period 2005-2013. Taking the average absolute number of directors on board into consideration, which is 9, this percentage means approximately 3 out of 9 directors are independent of the top managers in the firm, which is in accordance with the regulatory push that the CSRC conducted. What’s more, the standard deviation is 0.091, consisting only 14% of the mean, which reflects that the composition of the boards in Chinese public firms has relatively small variation.
4.3 Control variables
Following the categories that classified in Liu et al., 2015, control variables that potentially affect performance in all regressions can be divided into three categories: 1. firm and board characteristics; 2. ownership variables; 3. monitoring costs proxies. For the first classification, I include firm size (ln_assets), board size (ln_board), firm age (ln_age), firm leverage (leverage) and CEO-chair separation (Ddual). Firm size is denoted by natural log of total assets in an accounting year; the natural log of number of directors on board represents board size. Firm age is the year that I measure board composition minus the year that a firm is listed on stock market; natural log is also taken on that. CEO-chair separation is a dummy that indicates whether the CEO of a firm and the chairman on board are the same person. Secondly, ownership variable includes four items: the fraction of foreign ownership (bhshare), a dummy that whether a firm is state-owned (SOE), the fraction of shares that the biggest shareholder possesses (top1shr) and finally the shares that the state owns (stshare). Bhshares is computed by the percent of aggregated B-shares and H-shares over total shares outstanding. As for monitoring costs, this study controls sales growth (salesgr). As literature shows, board independence and cost of
monitoring are significantly and negatively related (Linck et al., 2008), thus this criterion is included in the control group. Sales growth is computed by one year’s total sales over its value of last year’s and reveals a firm’s growth opportunity. Table 2, panel C presents summary of all the control variables. On average, the state holds a proportion of 17.2% shares, with its standard deviation 0.225, which is quite surprising to me since in related literature (Bai et al., 2004, Liu et al., 2015) there are more than half of the companies controlled by the government. Companies issuing B- or H-shares are in the minority, at about 3% level, which implies that there are few multi-listing corporates. The CEO-Chair duality is not common in China as well, since the mean value is only 12.9%. The mean logarithm of board size is about 2.2, implying the mean of total directors at about 9 people. The average and standard deviation of the logarithm of assets is 21.862 and1.269 respectively. Sales growth has a distinguished average number: 22.3% and varies a lot since its standard deviation is 0.481. Besides, approximately one fourth of the Chinese listed companies are state-owned; this data is collected on the basis of the actual controller statistic in CSMAR dataset. Ln_age reflects that the average firm age in China up to 2013 is about 11.5 years over the 9-years period. The major shareholder in each sample firm holds a large stake of shares since the mean of the top 1 shareholder’s stake is 36.3%, and the highest one reaches 85.2%. Last but not least, control variables include the three-digit industry code in terms of The 2012 Revised Industry
Classification Guidelines for Listed Companies as well as years’ dummy.
4.4 Instrumental variables
Endogeneity is one crucial problem that exists in almost every corporate governance analysis but still needs to be addressed with caution. Within the framework of this paper, the number of independent directors and firm performance are expected to be endogenous. One effective and widely used way to tackle this problem is to use instrumental variables. On the one hand, instruments are correlated with explanatory variables; on the other hand, IV should not be correlated with the error
term. In our circumstance, firm performance and independent directors to some extent have reverse causality. Not only independent directors might enhance firm profitability by curbing agency problems, but the other way round. When an intelligent, capable and powerful CEO operates a firm, it is very likely that the firm has overwhelming performance compared to times without him, while in the meantime he might has the incentive to reduce independent board members in order to hide from monitoring. Another example is when a firm does not perform well, the board might vote to increase the number of independent directors intentionally. In one word, instrumental variables can help obtaining consistent estimates by capturing the sole part that the dependent variable affects the independent variable.
Following prior literature (Yermack, 1996; Knyazeva et al., 2013; Liu et al., 2015; Zhang & Yang, 2014), the corresponding instrumental variables for board independence measure are proindep and indusindep. Specifically, the former instrument is measured by the mean value of the fraction of independent directors of other firms that are located in the same province in the same year. As Knyazeva et al. (2013) document, local director labor markets have significant effects on board independence. Because more often than not it is a relatively stable social and working circle, perhaps they know each other before due to overlapped acquaintances, while on the other hand a firm’s location is always predetermined before the selection of independent directors and rarely changes. Thus ideally the two crucial criteria for instrumental variables: (1) strong relevance to the endogenous variable and (2) orthogonality with the error term, are met. In addition, in line with Liu et al. (2015), this paper also uses indusindep, which is calculated by the mean of the proportion of independent directors in other firms that are within the same industry in the same year. The rationale is that within the same industry, firms tend to have similar business mix and investment opportunities, such that a firm’s characteristics, specifically in our condition its board composition, is likely to correlate with that of their peers’, but the industry average is unlikely to influence firm performance directly (Zhang & Yang, 2014).
Table 2 presents the correlation matrix of all the key variables used in the regressions. The correlation coefficients among the 11 variables are all blow 0.5, hence the potential multicollinearity problem is ideally free of concern.
5. Analysis
In this section, I investigate empirically the effects of board members as well as other firm characteristics on firm performance. The dependent variable is computed from two ratios, namely Tobin’s Q and return on assets (ROA), where Tobin’s Q is calculated by the market value of a firm’s assets over the book value of a firm’s assets and ROA is computed by a firm’s operating income over its total assets. They further create the final dependent variables: Cum_Q and Cum_ROA. The independent variables include the percent of independent directors, which is the key variable of interest, a dummy of whether a firm’s independent directors exceed the sample median, a dummy variable indicating whether a firm is state-owned (SOE), the fraction of shares that the largest shareholder holds (top1shr), the weight of foreign shares (bhshare), the sales growth compared to last fiscal year (salesgr), firm size (ln_sassets), board size (ln_board), leverage ratio (leverage), firm age (ln_age), duality of CEO and chair (Ddual) and industry dummies. For firm size, I take the natural logarithm of operating income and denote it as ln_assets, and natural logarithm is also taken on the number of board members in order to proxy the board size. In addition, the leverage is defined by the ratio of a firm’s book value of debt over its book value of total assets. Ln_age is calculated by each year of the paneldata minus the year that the firm is listed on the Chinese stock market, using a natural logarithm form as well. Ddual starts with a capital D because it is a dummy variable indicating whether the CEO of a firm is also the chairman of board concurrently. Finally, the industry dummies are classified into 16 three-digit categories according Table 1
Summary statistics
This table reports the summary statistics of key variables used in the regression models. The sample consists of 957 unique firms and 8613 firm-years during 2005-2013. Panel A presents
statistics for the dependent variables, measured by two-years average of Tobin’s q and ROA, both winsorized at a 1% level. Panel B is the independent variable, measured as the proportion of independent directors on board (indep) and a dummy equals 1 if a firm’s independent directors are greater than the sample median (Ddirct). Panel C reports all the control variables except industry and year controls. State shares (stshare) and B- H-shares (bhshare) are the percentage of a firm’s state shares and foreign shares respectively. Ddual is a dummy that equals 1 if the Chief Executive Officer also chairs the board of a firm. Board size is represented by the natural log of the sum of board members (ln_board). Leverage is calculated by a firm’s book value of debt over its book value of total assets. Ln_assets is the logarithm of total assets. Sales growth (salesgr) is the arithmetic average of three years sales growth. SOE is a dummy equals 1 if a firm is owned by the state or state-owned enterprise. Ln_age is the firm age counting from the year that it is listed on the Chinese exchange, in a natural logarithm form. Top1shr is the percentage of the major shareholders’ stock holding amount.
Variable Obs Mean Std. Dev. Min 25th Pctl Median 75th Pctl Max Panel A: Dependent variables
avg_Q 7656 2.244 1.136 0.191 2.345 2.345 2.345 38.360 avg_ROA 7656 0.053 0.039 -0.449 0.059 0.059 0.059 1.619 Panel B: Independent variable
indep 8613 0.362 0.051 0.091 0.333 0.333 0.375 0.714
Ddirct 8613 0.303 0.459 0 0 0 1 1
Panel C:Control variables
stshare 8613 0.172 0.225 0 0 0 0.351 0.862 bhshare 8613 0.031 0.098 0 0 0 0 0.543 Ddual 8613 0.129 0.335 0 0 0 0 1 ln_board 8613 2.205 0.207 1.386 2.197 2.197 2.303 2.944 leverage 8613 2.241 1.074 1.212 1.514 1.866 2.529 5.381 ln_assets 8613 21.862 1.269 16.520 21.010 21.762 22.615 27.955 salesgr 8613 0.223 0.481 -0.281 -0.038 0.113 0.274 1.800 SOE 8613 0.609 0.488 0 0 1 1 1 ln_age 8613 2.362 0.422 0 2.197 2.398 2.639 3.135 top1shr 8613 0.363 0.156 0.022 0.236 0.341 0.481 0.852
Table 2
Correlation matrix
This table presents the correlation matrix among the variables used in the econometric models in this paper. The significance of correlation coefficients are signed with stars (*** p<0.01, ** p<0.05, * p<0.1). Variable descriptions are in Table 1.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 avg_Q 1 2 avg_ROA 0.041*** 1 3 indep 0.028** -0.013 1 4 Ddirct 0.004 -0.003 0.298*** 1 5 stshare -0.014 0.023** -0.105*** 0.095*** 1 6 bhshare 0.026** -0.019* 0.012 0.068*** 0.006 1 7 Ddual 0.010 -0.006 0.028** -0.086*** -0.065*** 0.025** 1 8 ln_board -0.017 0.018 -0.307*** 0.656*** 0.140*** 0.064*** -0.109*** 1 9 leverage 0.063*** 0.048*** 0.001 -0.075*** 0.016 -0.018* 0.024** -0.072*** 1 10 ln_assets -0.114*** 0.059*** 0.044*** 0.247*** 0.047*** 0.131*** -0.102*** 0.253*** -0.237*** 1 11 salesgr 0.004 0.052*** 0.032*** 0.029*** 0.116*** 0.001 -0.044*** 0.017 -0.090*** 0.232*** 1 12 SOE -0.019* 0.038*** -0.033*** 0.145*** 0.192*** 0.032*** -0.117*** 0.165*** -0.044*** 0.229*** 0.033*** 1 13 ln_age 0.025** -0.010 0.106*** -0.052*** -0.351*** 0.174*** 0.008 -0.097*** -0.039*** 0.142*** 0.040*** -0.002 1 14 top1shr -0.025** 0.027** -0.026** 0.033*** 0.405*** 0.000 -0.117*** 0.039*** -0.022** 0.299*** 0.145*** 0.261*** -0.155* ** 1
to the Chinese Security Regulatory Commission (CSRC). One thing to notice is that although the data is collected over the interval 2005-2013, in the estimation, however, the dependent variables ends at 2012 since they are calculated by a 2-years sum. Thus the final panel actually consists of 8 years time span when I am testing the relationship between board independence and firm performance using the average performance measure as the dependent variable.
In order to robustly test the impact of independent directors on firm performance, this paper estimates fixed effects as well as two-stage least squares model using a panel data set. The econometric challenges that this empirical study faces mainly come from three aspects: simultaneity, reverse causality and omitted variables, leading to the most common problem in corporate finance research: endogeneity. Industry- and year-fixed effects model control heterogeneity that is time-invariant and time-variant respectively, hence fixed effects approach mitigates endogeneity problem, but does not necessarily eliminate it (Bai at al., 2004; Liu et al., 2015). To further address simultaneity and reverse causality, instrument variables and 2SLS model are used as well.
5.1 Fixed-effects estimation
Column 1, 2, 5 and 6 of Table 3 presents the results of ordinary least squares (OLS) regression with fixed effects. Column 1 and 2 are the estimation of average Tobin’s q with indep and Ddirct as independent variables respectively, while Colume 5 and 6 are the estimation results of average ROA likewise. Since I use a 2-years’ average value to proxy firm profitability (year 2013 is ommited), the total panel consists of 7093 firm-years, instead of 8613. Model 1(a) for Tobin’s q is like following:
avg_Qi,t= 𝛽𝛽0+ 𝛽𝛽1indepi,t+ 𝛾𝛾∗controli,t+ 𝜆𝜆𝑖𝑖 + 𝜏𝜏𝑡𝑡+ 𝑢𝑢𝑖𝑖,𝑡𝑡
Model 1(b) uses the dummy of whether a firm’s independent directors exceeds the median of sample: 5 people as independent variable:
avg_Qi,t = 𝛽𝛽2+ 𝛽𝛽3Ddircti,t+ 𝛾𝛾∗controli,t+ 𝜆𝜆𝑖𝑖 + 𝜏𝜏𝑡𝑡+ 𝜀𝜀𝑖𝑖,𝑡𝑡
independent variables and other control variables; the models for avg_ROA are alike. Since the general performance (as well as indep and some control variables such as
leverage) of Chinese listed companies varies along with time significantly, this paper
adopts year-fixed effects (𝜆𝜆𝑖𝑖) and industry-fixed effects (𝜏𝜏𝑡𝑡) in the meantime. 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖,𝑡𝑡 includes the aforementioned 10 control variables. In order to address the potential heteroskedasticity and serial correlation problems, this paper always uses robust standard errors clustered at firm levels. As is shown in the regression results, most variables are insignificant statistically especially our key explanatory variables
indep and Ddirct. One reason that may lead to such an outcome is that board
independence and firm profitability do have no correlation, as illustrated in literature (e.g. Hermalin and Weisbach, 1991; Bhagat and Black, 2001). One thing to mention is that generally the year dummies are statistically significant (included in the regression but not present in the table). This result is consistent with my expectation and other references. However, what is inconsistent with my expectation is that the variable ln_assets is significantly and negatively correlated when regressing Tobin’s q. Moving to the continued table, we can see the firm size proxy is positively correlated with the average two-years’ ROA, although not statistically significant. The coefficient of SOE is positive, which is also beyond my expectation because state-owned enterprise usually undergo more severe obstruct in the functioning of independent directors. On the contrary, the regression shows that a firm being state-owned will increase its Tobin’s Q by about 0.0981.
When we focus on the regression of average ROA and board independence, sales growth and firm leverage are quite notable, because an increase of 1% on sales will increase the return on assets by 0.31% on average, and leverage ratio is also associated with better firm performance.
5.2 Simultaneous equations
In line with Dahya et al. (2008), Bhagat & Black (2002) and Liu et al. (2015), I use simultaneous equations methodology along with two instrumental variables to address the potential reverse causality and simultaneity. As aforementioned in part
3.1.4, I adopt the mean of the fraction of independent directors of other firms that are located in the same province in the same year (proindep) and the same computing method but for different industry groups (indusindep) to deal with the endogenous board composition variable, namely, the independent directors. The model 2 for average Tobin’s q is like following:
First stage equation 1: 𝐼𝐼𝐶𝐶𝐼𝐼𝐼𝐼𝐼𝐼� = 𝛽𝛽𝚤𝚤,𝑡𝑡 0+ 𝛽𝛽1𝐼𝐼𝐶𝐶𝐶𝐶𝑝𝑝𝐶𝐶𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡+ 𝛽𝛽1𝑝𝑝𝐶𝐶𝐼𝐼𝑢𝑢𝑖𝑖𝑝𝑝𝐶𝐶𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡+ 𝜂𝜂∗control
i,t+ 𝑣𝑣1𝑖𝑖,𝑡𝑡
Second stage equation 2: 𝑎𝑎𝑣𝑣𝑎𝑎_𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛾𝛾0+ 𝛾𝛾1𝐼𝐼𝐶𝐶𝐼𝐼𝐼𝐼𝐼𝐼� +𝚤𝚤,𝑡𝑡 𝜛𝜛∗controli,t+ 𝑣𝑣𝑖𝑖,𝑡𝑡
In the first stage of regressions, I regress the endogenous independent variable Indep against the instruments proindep and indusindep along with the ten control variables that are classified into firm characteristics, ownership features and monitoring costs. In the next step, the predicted value of indep is used as a replacement of the actual value of indep in the OLS model to regress 2-years’ average Tobin’s q or ROA. Control variables are the same as in the linear model. By doing so, we can use 𝐼𝐼𝐶𝐶𝐼𝐼𝐼𝐼𝐼𝐼� as an 𝚤𝚤,𝑡𝑡 OLS estimator to predict consistent and unbiased coefficients in the second stage. Column 3, 4, 7 and 8 are the 2SLS results for Tobin’s q and ROA respectively. The first stage F test statistics are around 98 and 39 respectively when using indep and Ddirct as key variable, indicating that the instruments are significantly exogenous. Besides, in unreported first stage regression results, the instruments also demonstrate to have strong correlation with the problematic variable indep and Ddirct, which means that the norm of correlation of an instrumental variable is also met. The Stata weak identification test (Cragg-Donald Wald F statistic and Kleibergen-Paap Wald rk F statistic) indicates that our instruments are strong and helpful.
As we can see from Panel A of table 3, the independent variable indep is significantly correlated with both of the firm performance measures at a 90% confidence level when regressing average Q. Given the average board size of 9.3, we can interpret the coefficients of key interest variable as: an increase of 10% indep, equivalent to increasing about 1 independent director on board, is going to make contribution of
This table reports the regression results for ordinary-least-squares (OLS) regression with firm- and year-fixed effects and two-stage-least-squares with instrumental variables. All regressions are estimated with robust standard errors clustered at the firm level. Variable descriptions are in Table 1. Figures in the brackets below the coefficients are the t-statistics of the estimators. The statistical significance is denoted as: *** p<0.01, ** p<0.05, * p<0.1 respectively.
Panel A: Dependent variable: avg_Q
(1) (2) (3) (4) FE 2SLS indep -0.3084 0.2268* (-0.48) (1.71) Ddirct 0.0185 0.5503* (0.25) (1.72) SOE 0.0981*** 0.0985*** 0.0885** 0.0828** (2.77) (2.79) (2.53) (2.35) bhshare -0.3369 -0.3572 -0.7883 -0.6099 (-0.38) (-0.40) (-0.74) (-0.61) top1shr -0.0201 -0.0121 0.0004 -0.0051 (-0.09) (-0.05) (0.00) (-0.02) stshare 0.1275 0.1286 0.0451 0.0404 (1.21) (1.22) (0.56) (0.50) salesgr 0.0756 0.0757 0.0725 0.0788* (1.60) (1.59) (1.63) (1.68) ln_assets -0.4094** -0.4109*** -0.3897*** -0.3894*** (-2.57) (-2.60) (-2.72) (-2.72) ln_board 0.0153 0.0260 0.2657 -0.6413 (0.06) (0.10) (1.01) (-1.27) leverage -0.0129 -0.0129 -0.0050 -0.0005 (-0.38) (-0.38) (-0.17) (-0.02) ln_age 0.0987 0.1030 0.4812** 0.4933*** (0.85) (0.87) (2.57) (2.59) Ddual -0.0296 -0.0289 -0.0231 -0.0198 (-0.29) (-0.28) (-0.23) (-0.20) Constant 10.7402*** 10.6235*** (3.47) (2.86)
1st stage F test statistics 98.730 39.570
J-test p-value 0.949 0.397
Obs 7,093 7,093 7,093 7093
Table 3 (continued)
Panel B: Dependent variable: avg_ROA
(5) (6) (7) (8) FE 2SLS indep -0.0041 0.0647 (-0.25) (1.16) Ddirct -0.0035 0.0157 (-1.36) (1.15) SOE 0.0036 0.0036 0.0037 0.0035 (1.52) (1.53) (1.58) (1.49) bhshare -0.0163 -0.0168 -0.0241 -0.0190 (-0.83) (-0.84) (-1.13) (-0.95) top1shr 0.0028 0.0027 0.0020 0.0018 (0.32) (0.30) (0.22) (0.20) stshare -0.0000 -0.0001 -0.0002 -0.0003 (-0.02) (-0.03) (-0.06) (-0.10) salesgr 0.0031*** 0.0031*** 0.0031*** 0.0033*** (3.40) (3.36) (3.58) (3.57) ln_assets 0.0006 0.0007 0.0007 0.0007 (0.22) (0.23) (0.26) (0.27) ln_board -0.0068 -0.0022 0.0001 -0.0257 (-1.10) (-0.48) (0.02) (-1.30) leverage 0.0017** 0.0017** 0.0018** 0.0019** (2.01) (1.97) (2.15) (2.33) ln_age -0.0042 -0.0044 -0.0002 0.0002 (-0.95) (-0.99) (-0.04) (0.05) Ddual -0.0034 -0.0034 -0.0034 -0.0033 (-1.33) (-1.35) (-1.31) (-1.26) Constant 0.0551 0.0441 (0.79) (0.69) 1st stage F test statistics 98.730 39.910 J-test p-value 0.240 0.561 Obs 7,093 7,093 7,093 7093 R-squared 0.006 0.006 0.005 0.007
2.27 to the 2-years’ Tobin’s q on average. As for Ddirct, a firm having more than 5 independent directors is highly likely to enhance the average Tobin’s q by 0.5533. This also shows that the simultaneous equation helps to conduct the endogeneity problem. Again, SOE shows significant positive correlation with firm performance, which is consistent with the test results in the fixed-effects model. Firm age turns out to be positively associated with firm value. In panel B, the significance level remains unchanged with prior test in columns 5 and 6.
5.3 State-owned enterprise, board independence and firm performance
As is shown in Table 1, more than 60 percent of Chinese companies are owned by the state, by saying that, I mean the actual controllers of these companies are either the state, administrative institutions or state-owned institutions. Thus a notable characteristic on the Chinese Stock Markets is that the state has strict controlling power over our public corporates. In order to test the particular effect of SOE on the relation between board independence and firm performance, I separate all the samples into two subsamples: SOEs and non-SOEs, and then conduct the fixed-effects regression.
Liu et al. (2015) reports that the positive correlation between board independence and firm performance in their study is largely driven by state-owned enterprises. While I do not report strong significant relationship in table 3, after dividing all the firms into SOEs versus non-SOEs, table 4 shows no noticeable difference in the regression results, which is consistent with the prior signs. However, several interesting aspects are: firms that are not under the control of the government show a strong positive correlation between foreign shares and firm performance, perhaps this is because issuing foreign stocks makes a firm go through stricter restrictions and supervision system such as in Hong Kong Exchange. What’s more, having a major shareholder benefits firm profitability for a non-government-controlled firm. Overall there is no consistent result between the measurements of Tobin’s q and ROA for firm performance.
6. Robustness check
Since the investigation period includes the global financial crisis times, according to the data from National Bureau of Statistics of China, the GNP growth dropped from 14.16% in 2007 to 9.63% in 2008, and in 2009 it was 9.21%, implying a negative economic growth. It can be concluded that the crisis affected not only mature western capital markets, but China’s as well, and consequently could skew firm performance and other variables that we observe in this research. Hence a
robustne-Table 4. State-owned enterprise and its effect on board-performance relationship
This table presents regression results for board-performance relationship using fixed-effects model and two subsamples: one consists of state-owned firms (SOE=1) and the other one is non-state-owned firms (SOE=0). All regressions are estimated with robust standard errors clustered at the firm level. Independent variable is the percentage of independent directors on board (indep). State shares (stshare) and B- H-shares (bhshare) are the percentage of a firm’s state shares and foreign shares respectively. Ddual is a dummy that equals 1 if the Chief Executive Officer also chairs the board of a firm. Board size is represented by the natural log of total board members (ln_board). Leverage measures a firm’s book value of debt over its book value of total assets. Ln_assets is the logarithm of total assets. Sales growth (salesgr) is the arithmetic average of three years sales growth. Firm age is adopted by its natural log. Top1shr is the percentage of the major shareholders’ stock holding amount. The statistical significance is denoted as: *** p<0.01, ** p<0.05, * p<0.1 respectively.
(1) (2) (3) (4)
SOE=0 SOE=1
avg_Q avg_ROA avg_Q avg_ROA
indep -0.3240 -0.0423 -0.6027 0.0093 (-0.27) (-1.34) (-0.81) (0.63) bhshare 0.4132** 0.0459 0.8977 0.0046 (-2.17) (-0.95) (1.19) (0.36) top1shr 1.3218* 0.0016 -0.4043 0.0075 (1.80) (0.09) (-1.56) (0.78) stshare 0.0244 -0.0005 0.0826 -0.0011 (0.09) (-0.05) (0.66) (-0.32) salesgr 0.1335 0.0031* 0.0452* 0.0016 (1.04) (1.66) (1.85) (1.57) ln_assets -0.8602** 0.0011 -0.0487 0.0027 (-2.44) (0.19) (-1.44) (1.63) ln_board 0.3395 -0.0243 -0.1928 0.0021 (0.58) (-1.56) (-0.75) (0.41) leverage -0.0118 0.0010 0.0098 0.0013* (-0.19) (0.67) (0.44) (1.73) ln_age 0.3157 -0.0060 0.0552 -0.0060 (1.32) (-0.71) (0.45) (-1.43) Ddual -0.1497 -0.0087** -0.0349 -0.0009 (-0.72) (-2.01) (-0.67) (-0.32) Constant 18.6619** 0.1021 3.8712*** -0.0061 (2.53) (0.70) (3.26) (-0.15) Obs 2,785 2,785 4,308 4,308 R-squared 0.061 0.009 0.009 0.006
Table 5. Same-year correlation of board independence on firm performance
This table reports the regression results in testing the correlation between board independence and firm performance pre- and post financial crisis, using fixed effects model. All regressions are estimated with robust standard errors clustered at the firm level. Variable descriptions are in Table 1. The dependent variable is the average Tobin’s Q and ROA in year 2005-2006 and 2011-2012 respectively. T-statistics of the estimators are in parentheses. The statistical significance is denoted as: *** p<0.01, ** p<0.05, * p<0.1 respectively.
(1) (2) (3) (4)
2005-2006 2011-2012
avg_Q avg_ROA avg_Q avg_ROA
indep -1.2085 -0.1518* 1.4297 -0.0334 (-1.56) (-1.94) (0.95) (-0.91) SOE 0.1152* 0.0013 -0.1039 -0.0074 (1.90) (0.17) (-0.37) (-1.36) bhshare -0.3536 0.2285 -7.9197* 0.0957 (-0.24) (0.54) (-1.81) (0.62) top1shr -0.2457 0.0422 -1.5656 0.0536* (-1.35) (1.63) (-0.98) (1.65) stshare -0.0716 -0.0256 -0.1900 -0.0062 (-0.61) (-1.45) (-0.61) (-0.66) salesgr 0.0459 0.0077 0.1153 0.0127** (0.87) (0.90) (0.76) (2.54) ln_assets 0.0080 -0.0375*** -1.2027** 0.0189** (0.11) (-2.96) (-2.30) (2.27) ln_board -0.2844 -0.0314 0.2663 -0.0372 (-1.50) (-1.53) (0.52) (-0.92) leverage 0.0760* 0.0003 0.0850 0.0106*** (1.70) (0.09) (1.14) (3.04) ln_age -0.1169 0.0019 0.6091 -0.0029 (-0.63) (0.10) (0.57) (-0.08) Ddual -0.0093 0.0066 0.3164 0.0142* (-0.09) (0.90) (1.23) (1.85) Constant 2.114 0.928*** 26.14** -0.325 (1.695) (0.279) (11.86) (0.229) Observations 1,541 1,539 2,832 2,832 R-squared 0.024 0.043 0.068 0.040
ss check from pre-crisis period (2005-2006) and post-crisis period (2011-2012) is examined and the results are reported in table 5. In the first test, I explicitly run two regressions over these two separate periods. If the results are consistent with before, the financial crisis performs no great influence on the public firms’ relationship between board and firm performance. In table 5 column 1, we can conclude that before the crisis, board independence proxy is negatively related to firm
performance measured both by ROA. But after the crisis the correlation is again insignificant. This implies that our results in part 4 are robust and consistent regardless of the exogenous shock of financial crisis.
In the second test, I re-run the equation of model 1(a) and model 2 using all the data in the same year, which is a common practice in many related literature to estimate the relationship between board independence and firm performance. As can be seen in table 6, column 1 reflects that an increase of 10% in indep (equivalent to adding 1 independent director onto the board) is going to improve the same-year Tobin’s q by almost 2. Taking the average Tobin’s q in sample into consideration, we can conclude that one more independent director will boost approximately twice of the market value of a firm measured by Tobin’s q. This relevance is significant. Such estimation is close to my references, for example, Liu et al. (2015) report a coefficient of 0.178 before the variable of independent directors’ ratio.
To summarize, the robustness tests in Section 6 performs as an additional evidence for the regression results that I conduct in Section 5. No matter before or after the 2008 global financial crisis, the association between board independence and firm performance turns out to be inexistence according to my estimation. However, when relying on the year-by-year firm performance measures, the existence of independent directors will increase the firm value on the whole.
Table 6. Same-year correlation of board independence on firm performance
This table reports the same-year correlation between board independence and firm performance, using fixed effects model and two-stage-least-squares model with instrumental variables. All regressions are estimated with robust standard errors clustered at the firm level. Variable descriptions are in Table 1, except the dependent variable is the Tobin’s Q and ROA with the same year of the independent variables. Figures in the brackets below the coefficients are the t-statistics of the estimators. The statistical significance is denoted as: *** p<0.01, ** p<0.05, * p<0.1 respectively.
Tobin’s q ROA (1) (2) (3) (4) FE 2SLS FE 2SLS indep 0.1977** 0.1204*** 0.0280 0.2548*** (2.11) (3.00) (0.78) (2.75) SOE -0.0178 -0.1081 -0.0079 -0.0084* (-0.17) (-1.02) (-1.62) (-1.76) bhshare -2.3178** -4.4256*** -0.0291 -0.0531 (-2.18) (-2.92) (-0.45) (-0.79) top1shr 1.3273*** 0.5578 0.0444** 0.0302 (3.23) (1.28) (2.11) (1.55) stshare 0.6669*** 0.3651** 0.0010 0.0012 (4.05) (2.29) (0.14) (0.20) salesgr 0.2414*** 0.2933*** 0.0228*** 0.0239*** (3.27) (4.02) (7.14) (7.57) ln_assets -1.8791*** -1.8371*** 0.0095* 0.0091* (-5.93) (-6.31) (1.75) (1.79) ln_board 0.3078 1.1730** -0.0270 -0.0049 (1.37) (2.54) (-1.54) (-0.26) leverage 0.0113 0.0331 0.0232*** 0.0231*** (0.16) (0.49) (11.36) (1.69) ln_age 0.5933** 2.3177*** -0.0197** -0.0068 (2.57) (6.17) (-2.33) (-0.99) Ddual 0.0715 0.0682 -0.0014 -0.0017 (0.33) (0.30) (-0.25) (-0.31) Constant 37.7531*** -0.1662 (5.63) (-1.22)
1st stage F test statistics 98.23 98.33
J-test p-value 0.521 0.598
Obs 8037 8037 8037 8037
7. Conclusion and discussion
In this paper, I use a panel data consisting of 8613 firm-years and 9 years’ horizon to test the impact of independent directors on firm performance in Chinese listed firms, and further discover the additional effect of governance on this board-performance relationship. In order to address the inherent endogeneity problem, several econometric models are used, including ordinary least squares model with fixed effects and two-stage least squares model with instrumental variables. To sum up, I find no strong correlation between board independence and firm performance in China. This is in line with the common cognition and existing literature that outside directors are “ vase” directors that only on board because of regulation mandate. They do not have real function in dealing with the self-dealing problem or other inefficiencies in corporates. Thus it provides some evidence that the 2001 Guideline does not achieve its expected effect.
In addition, in testing the correlation between board independence and firm performance, I also consider the unique context in China. Briefly, regarding the ownership structure, the government owns more than 60 percent of public firms in China. This is a potential threaten to the functioning of independent directors in firm. The reason is that the firm’s top management and directors (who represent shareholders) usually undergo severe agency conflict in such background. Overall, according to the empirical test results, a firm being state-owned or not affects firm performance measured by Tobin’s q. However when the samples are divided into SOEs and non-SOEs, our key variable shows no necessary correlation to firm performance.
In conclusion, my investigation manages to address the endogeneity problem in corporate finance study; provides evidence that by far the tunneling in Chinese firms still exists after years emphasize on the independent directors system; and Chinese public firms as well as regulation institutions have a long way to go in the construction of firm internal control systems.
Reference
Adams, R., Hermalin, B. E., & Weisbach, M. S. (2008). The role of boards of directors in corporate governance: A conceptual framework and survey (No. w14486). National Bureau of Economic Research.
Bai, C. E., Liu, Q., Lu, J., Song, F. M., & Zhang, J. (2004). Corporate governance and market valuation in China. Journal of Comparative Economics, 32(4), 599-616.
Barnhart, S. W., & Rosenstein, S. (1998). Board composition, managerial ownership, and firm performance: An empirical analysis. Financial Review, 33(4), 1-16.
Baysinger, B. D., & Butler, H. N. (1985). Corporate governance and the board of directors: Performance effects of changes in board composition. Journal of Law, Economics, & Organization, 101-124.
Bhagat, S., & Black, B. S. (2002). The non-correlation between board independence and long-term firm performance. Journal of Corporation Law, 27, 231-273.
Bhagat, S., & Bolton, B. (2008). Corporate governance and firm performance. Journal of Corporate Finance, 14(3), 257-273.
Black, B. S., & Khanna, V. S. (2007). Can corporate governance reforms increase firm market values? Event study evidence from India. Journal of Empirical Legal Studies, 4(4), 749-796.
Black, B., & Kim, W. (2012). The effect of board structure on firm value: A multiple identification strategies approach using Korean data. Journal of Financial Economics, 104(1), 203-226.
Campa, J. M., & Kedia, S. (2002). Explaining the diversification discount. The Journal of Finance, 57(4), 1731-1762.
Clarke, D. C. (2006). The independent director in Chinese corporate governance. Delaware Journal of Corporate Law, 31(1), 125-228.
Dahya, J., Dimitrov, O., & McConnell, J. J. (2008). Dominant shareholders, corporate boards, and corporate value: A cross-country analysis. Journal of Financial Economics, 87(1), 73-100.
Desender, K. A. (2009). The Relationship Between the Ownership Structure and Board Effectiveness. University of Illinois at Urbana-Champaign, College of Business Working Papers, 09-0105.
Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. Journal of law and economics, 301-325.
