Ownership and ownership concentration: which is important in determining the performance of China’s listed firms?
This research is supported by the New Research Grant from the Faculty of Commerce, University of Wollongong. We gratefully acknowledge the critical comments and helpful suggestions provided by two anonymous referees. We also thank the editor for his valuable inputs and advice.
Abstract
This article investigates the impact of ownership and ownership concentration on the performance of China’s listed firms. By recognizing the differences between ownership and ownership concentration and between total ownership concentration and tradable ownership concentration, we find that ownership concentration is more powerful than any category of ownership in determining firm performance; tradable ownership concentration has a more significant and positive influence on firm performance than total ownership concentration; the highest level of firm performance is approached when a firm is characterized by both total ownership concentration and tradable ownership concentration. Thus, we propose a conclusion that ownership concentration enhances firm performance regardless of who the concentrated owners are.
1. Introduction
In December 1978, China initiated an economic reform in an attempt to replace its centrally planned economy with a market-oriented one. Prior to the reform, private ownership in business and production was only allowed at the family level. Enterprises were fully owned by either the state or collective organizations. Ownership restriction was gradually released with the permission of private enterprises, foreign investors and joint-stock companies.
China’s securities market was established with the opening of the Shanghai and Shenzhen stock exchanges in December 1990 and April 1991, respectively. The first several blocks of listed firms were formerly state-owned enterprises (SOEs) that had been reformed into joint-stock companies. These firms issued common shares – mostly to Chinese residents and some to foreign investors. Later on, and after satisfying the listing rules, collective enterprises and private companies were allowed to be listed on the market.
State shares are shares that have been converted from state-owned assets or investments. Representatives of state share ownership are the state government and local government (Bureau of State Asset Administration). Legal person shares are shares that have been transformed from the assets or investments of legal entities, such as institutions or enterprises, when the legal entities became the owners of joint-stock companies before public listing. An enterprise may restructure to carve out an independent firm for listing. The assets counted as investments from the parent companies into the carved firms are viewed as legal person shares as well. Both state and legal person shares are non-tradable on the public market of stock exchanges.
Common shares sold publicly to Chinese residents are entitled A shares and those sold initially to foreign investors are entitled B shares. These B shares have been accessible to Chinese residents since the restriction was removed in April 2001. Both A and B shares are exclusively tradable shares on China’s stock market. Shares that are issued by local Chinese companies on the Hong Kong Exchange are entitled H shares. These H shares are not accessible by the Chinese people residing on the mainland.
Both state and legal persons own, on average, about one-third each of the ownership of the listed firms in the domestic market, with the remaining ownership held by the massive numbers of individuals and some financial institutions. Thus, China’s stock market is characterized by a high level of ownership concentration and a low level of marketability. This unique feature of China’s stock market has attracted increasing research. Scholars are trying to answer several critical research questions, such as: what are the influences of the different categories of ownerships on firm performance? (Qi et al., 2000; Sun and Tong, 2003; Delios and Wu, 2005; Gunasekarage et al., 2007); does the concentrated ownership have a significant impact on firm performance? (Xu and Wang, 1999; Wang et al., 2004; Gunasekarage et al., 2007; Ng et al., 2009) and can the theories developed from other markets explain the phenomena of China’s emerging market?
Prior research, however, might not obviously recognize several factors. First, there is the distinction between ownership and ownership concentration. Ownership mainly refers to the attributes of the owners, while ownership concentration refers to the control power that top owners have in making a firm’s decisions. A high proportion of a category of ownership does not necessarily mean high ownership concentration. For example, high state ownership may refer to high ownership concentration, because there is only one state. However, state shares are directly owned by different state agents that represent the interests of various organizations (Chen et al., 2009). High legal person ownership may also not always imply high ownership concentration, because there may be a number of legal persons holding shares of a firm. High tradable ownership can indicate low ownership concentration, because massive numbers of shareholders own a firm and each individual typically holds a minuscule proportion of shares. Second, total ownership concentration differs from tradable ownership concentration. For instance, about nine out of ten top shareholders are non-tradable share owners. As they are not able to sell the shares, they are not pure shareholders with complete shareholders’ rights. The top non-tradable shareholders can control the firm and top tradable shareholders can influence the market price. Thus, they may have different incentives in monitoring the firm. Finally, in most firms, there are state, legal persons, large non-tradable and large tradable shareholders. The different categories of ownerships and ownership concentrations ultimately have either interactive or joint impact on firm performance.
Therefore, this article’s innovation is that the research design distinguishes ownership and ownership concentration and also total ownership concentration and tradable ownership concentration. Through referring to state ownership, legal person ownership and tradable ownership, this research focuses on the total ownership concentrations and tradable ownership concentrations. The research is developed from investigating the individual impact of a category of ownership or ownership concentration and tests the interactive and joint impact of ownerships and ownership concentrations on firm performance.
Our new findings are first, that ownership concentration has a more powerful and positive effect on the performance of China’s listed firms than any category of ownership, such as state and/or legal person ownership; second, that tradable ownership concentration has a more significant and positive influence on firm performance than total ownership concentration that mainly represents non-tradable ownership concentration; third, that the interactive effect of state and legal person ownerships enhances the firm management and is positively associated with firm performance and finally, that the interactive effect of total ownership and tradable ownership concentrations is more powerful than individual effects of either total ownership concentration or tradable ownership concentration, so that the highest level of firm performance is approached when a firm is characterized by both total ownership concentration and tradable ownership concentration.
Research on the impact of ownership and ownership concentration on the performance of China’s listed firms has either theoretical or empirical significance. China’s economy is the largest transition economy, and China’s securities market is the largest emerging market worldwide. It is broadly argued that China’s economy is not a complete market economy because the government is the main manipulator of the economy, rather than the market mechanism. The securities market is an immature market as the state and legal persons retain a large proportion of non-tradable shares. The recent financial crisis has shaken developed economies such as the United States, the United Kingdom and Japan. China’s economy seems to be barely affected by the worldwide slowdown. Therefore, China’s economy – which is characterized by state ownership and control – has been put forward as a successful model (Yao, 2009). Does the state ownership and control really help China’s economic success? Does the state and legal persons retaining a large proportion of non-tradable shares improve firm performance? This research provides us with critical evidence that ownership concentration, rather than a high level of state or legal person ownership, has a more positive impact on firm performance.
Section 2 of this article reviews relevant literature on China’s listed firms and develops hypotheses. Section 3 explains the data, variables and basic statistics. Section 4 analyses the individual impact of each category of ownership and ownership concentration. Section 5 analyses the interactive effect between two categories of ownerships and between two categories of ownership concentrations. Section 6 analyses the joint effect with different types of ownerships and ownership concentrations. Section 7 shows a robustness test with a market measure of Tobin’s Q. Section 8 summarizes this research and presents our conclusions.
2. Relevant literature on China’s listed firms and hypothesis development
2.1. Relevant literature
An important feature of a publicly listed company is its establishment and operation with significant amounts of capital gathered from diverse investors. Large shareholders, however, may end up in control, forcing such a firm to take actions that benefit themselves at the expense of minority shareholders. It is also possible that large shareholders with sufficiently large ownership stakes will be aligned with the interests of the firm and will fully engage in monitoring the firm (Shleifer and Vishny, 1986, 1997). In the situation without controlling shareholders, dispersed investors may lack the incentive to monitor firms, giving the managers a ‘free ride’. The different classes of owners may have either common or conflicting interests and hence may influence the firm in various ways.
Publicly listed firms in China’s new emerging market have many special characteristics. As many firms are reformed SOEs, the state is the largest shareholder in most of them. The state possesses about one-third ownership of total market capitalization. Legal person ownership was mostly converted from existing assets of legal entities when the firms were restructured for listing. Legal person ownership comprises about one-third of total market capitalization. State and legal person shares are non-tradable on stock exchanges.1 The remaining shares are publicly tradable and owned by either individual or institutional investors. The unique ownership structure of China’s listed firms has drawn much attention from academics, with an increasing number of papers being published in this area.
Some relatively early research on the ownership and ownership concentration of China’s listed firms is conducted by Qi et al. (2000). They find that firm performance is positively related to the proportion of legal person shares but negatively related to the proportion of state-owned shares. Further analyses show that firm performance increases with the degree of relative dominance of legal person shares over state-owned shares. On the other hand, there is no evidence supporting a positive correlation between corporate performance and the proportion of tradable shares owned by either domestic or foreign investors.
Sun and Tong (2003) evaluate the performance of 634 firms listed on China’s two stock exchanges from 1994 to 1998. In their univariate analysis, they find that firms with more than 50 per cent state ownership perform better – as measured by real net profit and earnings before interest and tax – than firms with less than 50 per cent state ownership before they are publicly listed, however, they perform worse after they are publicly listed. Gunasekarage et al. (2007) investigate the influence of state ownership and ownership concentration on the performance of China’s listed firms from 2000 to 2004. They find that, on average, firm performance is negatively influenced by state ownership. Recent research by Ng et al. (2009) examines 4315 firm year observations of Chinese firms from 1996 to 2003. Results support the hypothesis of a convex relationship between state ownership and firm performance, showing benefits from strong privatization and state control.
Xu and Wang (1999) document that legal persons, like institutional investors, have greater initiative and expertise in supervising management and improving firm values. Sun and Tong (2003) find evidence from a regression analysis that the proportion of legal person ownership is positively correlated with firm performance as measured by Tobin’s Q. Delios and Wu (2005) focus on the impact of legal person ownership on firm performance using a sample of listed firms in China’s markets from 1993 to 2001. They argue that the change in legal person ownership drives the firm value according to a U-shaped function. They suggest that a low level of legal person ownership may lead to over-diversification and reduce firm value, whereas a high level of legal person ownership has a monitoring effect and increases firm value.
Xu and Wang (1999) undertake early research on the relationship between ownership concentration and firm performance of China’s listed firms for the years 1993–1995. They use the top 10 ownership ratio and the Herfindahl index as proxies for ownership concentration and find significant coefficients for the two ownership concentration variables in the regressions on market returns and return on equity. They demonstrate a linear relationship between ownership concentration and firm performance.
Wang et al. (2004) examine how ownership concentration by top shareholders and the balance of power among them affect the performance of China’s listed firms from 1994 to 2000. They find that publicly listing the firm lowers the state ownership. They also generate evidence that the ownership concentration by a few large shareholders is positively correlated with firm performance and that a more balanced ownership structure among these top shareholders is also good for firm performance.
Gunasekarage et al. (2007) conduct a series of regressions on Tobin’s Q with substitutes of ownerships of largest shareholder, top five shareholders, top ten shareholders and top two to ten shareholders. They find that they all have significantly positive coefficients except for the largest shareholder. Thus, they conclude that a balanced ownership structure enhances firm performance and that block ownership is detrimental to firm value in China’s market. Ng et al. (2009) also find a significant and positive association between ownership concentration represented by the top five ownership ratio and firm performance represented by Tobin’s Q.
2.2. Hypothesis development
One feature of a modern corporation is the general separation of ownership and management. Productivity is improved by promoting the comparative strengths of the owners’ funds and the managers’ skills. However, a conflict of interest arises between the owners and managers when the owners do not manage the firm by themselves (Jensen and Meckling, 1976). The managers of firms might pursue their own interests rather than the interests of the owners, which is against the principle of maximization of shareholders’ wealth.
Two governance strategies are broadly used by investors to eliminate the agency problem. One is to rely on the disciplinary force of external governance systems – such as capital markets and the legal system – to protect investment against managerial opportunism (Gillan, 2006; Heugens et al., 2009). The other is to concentrate investors’ ownership, so that they can exert direct influence on top managers to run the firm in their interests. Concentrated ownership provides the large investors with both sufficient incentive and power to discipline management, and thus improve firm performance by decreasing monitoring costs (Shleifer and Vishny, 1986, 1997).
La Porta et al. (1999) demonstrate that ownership is more concentrated in countries with poor legal protection to investors. Heugens et al. (2009) also state that ownership concentration is an efficient corporate governance strategy in markets with poor legal protection of minority shareholders. Strong legal protection of shareholders makes ownership concentration inconsequential and therefore redundant. China’s new emerging market is well known for its immature legal system and criticism of fraud and corruption (Chen et al., 2006). The high level of ownership concentration of China’s stock market is not merely a result of economic reform from existing SOEs to publicly listed firms, but it is also a necessary governance strategy that protects investors and maintains profitability.
In China’s stock market, a high level of legal person ownership may not always imply a high level of ownership concentration, because there may be a number of legal persons in a firm. However, a high level of tradable ownership only implies a low level of ownership concentration, because massive numbers of shareholders own a firm and each individual typically holds a minuscule proportion of shares. A high proportion of state ownership seems to represent a high level of ownership concentration as there is only one state. However, a recent study by Chen et al. (2009) shows that state-owned shares can be owned by various state agencies, such as State Asset Management Bureaus, SOEs affiliated to the central government, and SOEs affiliated to local government. Each state agency has its own goals. This is probably one reason that previous research on the effects of legal person and state ownership on firm performance generates mixed results. Research on the effects of ownership concentration on firm performance provides consistent evidence. Therefore, our first hypothesis is
H1: Ownership concentration, rather than state or legal person ownership, has a more powerful and positive effect on the performance of China’s listed firms.
Gedajlovic et al. (2005) classify the shareholders of the Japanese market into three categories: ‘stable investors’, ‘market investors’ and ‘inside investors’. Each one of these investors has a relatively distinct investment objective. The stable investors are affiliated firms, banks and insurance companies. Stable investors are willing to keep the ownership primarily to cement and grow stable business relationships rather than to earn returns on their equity investment. Market investors are ‘pure investors’ in that they are typically tied to the firm by just their equity stakes and have maximizing equity returns as their primary investment objective.
We use the two concepts of ‘stable investors’ and ‘market investors’ to explain the ownership of China’s listed firms. The holders of state shares and legal person shares are definitely stable investors as their shares are not tradable on stock exchanges. However the share price changes, the stable investors have no other choice but to hold onto the shares. It is true that one of the objectives of state and legal person shareholders is to maintain control and involvement in the firms. By contrast, the holders of tradable shares are market investors who have the exclusive objective of maximizing returns.
Heugens et al. (2009) state that firms may be sensitive to the pressures of market investors for two reasons. First, market investors will be more willing to sell their ownership stake when they are dissatisfied with the firm’s performance. Selling off large blocks of shares is a powerful disciplining force, because it is likely to decrease the firm’s share price and thus increase the cost of equity capital. Second, market investors will require a risk premium when they fear problems like tunnelling (expropriating firm assets) or managerial entrenchment, which increases the firm’s cost of equity capital. These disciplinary forces make it more likely that the firm’s managers will engage in strategies consistent with the investment objectives of market investors and mitigate large stable investors’ ambition of tunnelling the firm. Based on the prior statement made by Heugens et al., we further argue that not only do the managers take care of the disciplinary forces of the market investors, but the stable investors are sensitive to the activities of market investors as well. The selling off of large blocks of shares by market investors will reduce the wealth of stable investors as it decreases share price. Therefore, our second hypothesis is
H2: Tradable ownership concentration has a more significant and positive effect on firm performance than total ownership concentration.
It is a general fact that there are large proportions of non-tradable state ownership and legal person ownership in most of China’s listed firms. Either the state or a legal person is the controlling shareholder, and both the state and legal persons rank at the top of shareholders in those firms. The state and legal person shareholders monitor each other while they compete for corporate control. For example, when a legal person takes action in corporate governance, the legal person has to consider the reaction of the state owner, and vice versa. The competition between state owners and legal person owners will eventually mitigate the misuse of voting rights and improve firm value. Therefore, our third hypothesis is
H3: The interactive effect of state and legal person ownerships enhances firm management and is positively associated with firm performance.
In a mature market, almost all of the shares are tradable. Concentrated ownership gives investors a powerful incentive to be involved in governance, as well as the means to monitor managers by direct checking of corporate management and the threat of using their determined voting rights. In a situation that the firm is confronted with financial problems, the concentrated owners are able to transfer private resources to support the firm. Furthermore, concentrated shareholders are able to sell shares when they are dissatisfied with the firms’ management, which is an effective discipline that forces management decisions to align with the goals of shareholders.
However, in China’s new emerging market, two-thirds of shares are non-tradable. About nine out of ten top shareholders are non-tradable share owners. The large non-tradable shareholders have the incentive to monitor management but lack threat power on the managers because they cannot sell their shares. By contrast, although the large tradable shareholders have the threat power on the managers by selling shares, their incentive in monitoring the firm may not be as strong as their non-tradable counterparts because of their relatively lower level of ownership. Nevertheless, the large non-tradable shareholders and tradable shareholders can be complementary. The function of concentrated ownership in a mature market can be approached by the interactive function of concentrated non-tradable ownership and tradable ownership in China’s market. Therefore, our fourth hypothesis is
H4: The interactive effect of total ownership concentration and tradable ownership concentration is positively related to firm performance, and is more powerful than the individual effect of either total ownership concentration or tradable ownership concentration.
3. Data, variables and statistics
3.1. Data
Our research focuses on firms listed on either the Shanghai or Shenzhen stock exchanges in 2003 and 2004 to avoid possible temporary shakes from the new policy of floating non-tradable state shares starting from 2005.2 We exclude some different types of firms in the sample. First are firms that have been classified by the China Securities Regulatory Commission (CSRC) as ‘special treatment’ (ST) or ‘particular transfer’ (PT) firms. These ST and PT firms are under special monitoring owing to their poor operation. Restrictions on the trading of their shares have also been imposed.3 Second are financial firms. These firms are specially regulated and usually have much higher leverage ratios than other firms. Third are firms with foreign ownership, such as those issued B shares on domestic markets and H shares on the Hong Kong Stock Exchange. Firms with foreign ownership are subject to different requirements for listing, reporting and accounting standards. Lastly are firms with missing data or incomplete information for our modelling. Therefore, our pooled sample retains 1975 sets of observations of firms, primarily obtained from the China Stock Market and Accounting Research Database created by the GTA Information Technology Company and the University of Hong Kong. We have made a number of corrections on the data with references from several other data sources.4
3.2. Variables and statistics
We apply the accounting indicator of return on assets (Return_assets) as the main measure of firm performance and a market indicator of Tobin’s Q for the robustness test. Return on assets was chosen as the main measure because we consider return (represented by the net profit of the firm) maximization to be a universal objective of firms. Furthermore, return on assets is more reliable than other measures so we are able to achieve a relatively accurate analysis over a short sample period. For example, return on equity seems an appropriate measure of investment profitability, but it is useless for firms with negative equity or both negative profit and equity, which is not unusual for China’s firms. The accounting measured performance of a firm is eventually reflected in the stock price. Tobin’s Q is argued to have the advantage of reflecting the firm’s current value and future profit potential. However, in China’s new emerging market, share prices are relatively volatile. Therefore, we only use Tobin’s Q for the robustness test.
The return on assets is the annual net profit divided by the average book value of assets at the beginning and end of the year. The statistics of designated variables are reported in Table 1. The return on assets is 0.0247 on average. The return on assets varies considerably, with a minimum of −0.6121 and a maximum of 0.3138. The mode is 0.0011, far to the left of the mean and median, which implies that some firms experience a lower accounting performance than the average. Tobin’s Q is calculated using market prices of shares multiplied by the number of outstanding shares plus the book value of liabilities then divided by the total value of assets. The average of Tobin’s Q is 1.7334 with a standard deviation of 0.7392. The maximum of 12.3443 is 16 times the minimum of 0.7964. The mode is smaller than both the mean and the median, which implies that more firms have a lower market performance than the market average.
| Variable | Mean | Std | Median | Mode | Minimum | Maximum |
|---|---|---|---|---|---|---|
| Firm value/performance | ||||||
| Return_asset | 0.0247 | 0.0626 | 0.0261 | 0.0011 | −0.6121 | 0.3138 |
| Tobin’s Q | 1.7334 | 0.7392 | 1.5271 | 1.1387 | 0.7964 | 12.3443 |
| Ownership | ||||||
| Ratio_state | 0.3221(0.3885) | 0.2701 | 0.3475 | 0 | 0 | 0.8500 |
| Ratio_legal | 0.3159(0.3003) | 0.2623 | 0.2838 | 0 | 0 | 0.8638 |
| Ratio_trade | 0.3620(0.3112) | 0.1160 | 0.3750 | 0.3333 | 0.0868 | 1.0000 |
| Ownership concentration | ||||||
| Top10_total | 0.6155(0.6489) | 0.1224 | 0.6331 | 0.6083 | 0.0786 | 0.8948 |
| Top10_trade | 0.0848(0.0998) | 0.0936 | 0.0468 | – | 0.0077 | 0.8757 |
| Firm attributes (control variables) | ||||||
| Total_asset (billions) | 2.5762 | 5.6832 | 1.5337 | 0 | 0.1259 | 150.0546 |
| Ratio_debt | 0.4783(0.5021) | 0.1771 | 0.4903 | 0 | 0.0081 | 1.1913 |
| Num_dir | 9.79 | 2.23 | 9 | 9 | 5 | 21 |
| Ratio_indir | 0.3350 | 0.0545 | 0.3333 | 0.3333 | 0.0714 | 0.6667 |
| Dir_mting | 7.42 | 3.01 | 7 | 6 | 2 | 26 |
| Util_indry | 0.1073 | 0.3096 | 0 | 0 | 0 | 1 |
| Prop_indry | 0.0506 | 0.2192 | 0 | 0 | 0 | 1 |
| Cong_indry | 0.1240 | 0.3297 | 0 | 0 | 0 | 1 |
| Manu_indry | 0.6518 | 0.4765 | 1 | 1 | 0 | 1 |
| Comm_indry | 0.0663 | 0.2489 | 0 | 0 | 0 | 1 |
- Summary of statistics of 1975 sets of observations of firms listed on either the Shanghai or the Shenzhen stock exchange in 2003 and 2004. Return_asset is the return on assets. Tobin’s Q is a market measure of firm performance. Ratio_state is the ratio of state shares. Ratio_legal is the ratio of legal person shares. Ratio_trade is the ratio of tradable shares. Top10_total is the ratio of shares held by the top 10 shareholders to total outstanding shares. Top10_trade is the ratio of tradable shares held by the top 10 tradable shareholders to the total tradable shares. Total_asset is total assets. Ratio_debt is the ratio of debt to total assets. Num_dir is the number of directors. Ratio_indir is the ratio of independent directors. Dir_mting is the frequency of director meetings. Util_indry is the industry dummy for utilities. Prop_indry is the industry dummy for property. Cong_indry is the industry dummy for conglomerations. Manu_indry is the industry dummy for manufacturing. Comm_indry is the industry dummy for commerce. Figures in parentheses are weighted averages.
The ownership variables are the ratio of state ownership (Ratio_state), the ratio of legal person ownership (Ratio_legal) and the ratio of tradable share ownership (Ratio_trade). These variables are calculated by the number of shares in the relevant category divided by the total number of outstanding shares. The means of the state, legal person and tradable share ownerships are 0.3221, 0.3159 and 0.3620, respectively. Although the tradable share ownership is the largest, the ratios of the three ownerships are relatively close together. However, if the means are calculated using market capitalization-weighted averages, then state ownership is the largest, with a value of 0.3885, and tradable share ownership is the second largest, with a value of 0.3112. From Table 1, we also see that the maximum state and legal person ownerships are 0.8500 and 0.8638, respectively, and that the minimum is zero in each case, which implies that some firms have no state or legal person ownership at all. The mean, median and mode are similar only for tradable share ownership. However, the maximum of 1 indicates the existence of firms with 100 per cent tradable shares, while the minimum of 0.0868 shows that some firms have only 8.68 per cent tradable shares.
As the ownership variables do not represent ownership concentration, we define two ownership concentration variables. The first is the top 10 ownership ratio (Top10_total), which equals the number of shares held by the top 10 shareholders divided by the total number of outstanding shares. Because tradable shareholders are rarely ranked in the top 10, the top 10 share ratio almost represents non-tradable ownership concentration. The second ownership concentration variable is the top 10 tradable ownership ratio (Top10_trade), which is the ratio of the number of tradable shares held by the top 10 tradable shareholders to the total number of outstanding tradable shares.
The mean of the top 10 ownership ratio is 0.6155, differing only slightly from the median of 0.6331 and the mode of 0.6083, which indicates a high ownership concentration in China’s listed firms. The maximum top 10 share ratio is extreme, at 0.8948, meaning that almost 90 per cent of shares are held by the top 10 shareholders for some firms. This enables the top 10 shareholders to control the firms and dominate other shareholders. The top 10 tradable shareholders own 8.48 per cent tradable shares on average, with extremes as low as 0.0077 and as large as 0.8757. The top 10 tradable ownership ratio is much smaller than the top 10 ownership ratio. Under the circumstance of a large proportion of non-tradable shares and a relatively highly dispersed tradable share ownership, however, the top 10 tradable shareholders can influence market prices.
With the control variables necessarily employed in the regression analyses, we include total assets (Total_asset) and the debt-to-asset ratio (Ratio_debt) to control for firm size and capital structure effects. We also select corporate governance variables to control for management effects: the number of board directors (Num_dir), the ratio of independent directors (Ratio_indir) and the frequency of director meetings (Dir_mting). Industry control variables follow the comprehensive classifications most popularly used in China: utilities, manufacturing, commerce, conglomerations, finance and property. In the regression, we adopt four dummy variables: utilities (Util_indry); manufacturing (Manu_indry); conglomerations (Cong_indry) and commerce (Comm_indry). Property (Prop_indry) will be carried in the intercept to avoid the dummy variable trap that leads to perfect multicollinearity.
4. Simplex analyses
Ownership and ownership concentration have an impact on firm performance. Empirical research on China’s market provides us with mixed results. In this section, we apply the simplex method to test the impact that different categories of ownerships and ownership concentrations have on firm performance. The objective of this section is not for agreement with previous findings in the literature, instead, it is to compare the impact of ownership with the impact of ownership concentration in determining firm performance.
We sort the firms into five classes in terms of specific ownership or ownership concentration. For state ownership, the first class is firms with zero state shares. The subsequent four classes are defined by the quartile numbers of firms ranked by proportion of state ownership, from minimum to maximum. Similarly, for legal person ownership, the first class is firms with zero legal person shares. The next four classes are defined by the quartile numbers of firms ranked by proportion of legal person ownership, from minimum to maximum. For tradable ownership, top 10 ownership concentration, and top 10 tradable ownership concentrations, three sets of five classes are defined, respectively, in terms of the quintile numbers of firms, ranked by proportion of ownership or ownership concentration, from minimum to maximum. Accordingly, we employ analyses of variance (anova). The results are reported in Table 2.
| 1 | 2 | 3 | 4 | 5 | F-stat | P > F | |
|---|---|---|---|---|---|---|---|
| Panel 1: Firm performance (return on assets) with state ownership | |||||||
| Ratio_state | |||||||
| Range | ≤0 | 0–0.2994 | 0.2994–0.4864 | 0.4864–0.6260 | >0.6260 | ||
| Means | 0.0243 | 0.0170 | 0.0130 | 0.0283 | 0.0411 | 11.11 | 0.001 |
| Obs | 561 | 353 | 353 | 353 | 355 | ||
| Panel 2: Firm performance (return on assets) with legal person ownership | |||||||
| Ratio_legal | |||||||
| Range | ≤0 | 0–0.1033 | 0.1033–0.3467 | 0.3467–0.5972 | >0.5972 | ||
| Means | 0.0374 | 0.0261 | 0.0152 | 0.0194 | 0.0320 | 7.18 | 0.001 |
| Obs | 210 | 441 | 441 | 442 | 441 | ||
| Panel 3: Firm performance (return on assets) with tradable ownership | |||||||
| Ratio_trade | |||||||
| Range | ≤0.2973 | 0.2973–0.3529 | 0.3529–0.4004 | 0.4004–0.4675 | >0.4675 | ||
| Means | 0.0443 | 0.0228 | 0.0244 | 0.0210 | 0.0110 | 15.40 | 0.001 |
| Obs | 395 | 395 | 395 | 395 | 395 | ||
| Panel 4: Firm performance (return on assets) with top 10 ownership concentration | |||||||
| Top10_total | |||||||
| Range | ≤0.5216 | 0.5216–0.6045 | 0.6045–0.6575 | 0.6575–0.7174 | >0.7174 | ||
| Means | 0.0136 | 0.0113 | 0.0212 | 0.0221 | 0.0553 | 33.93 | 0.001 |
| Obs | 395 | 395 | 395 | 395 | 395 | ||
| Panel 5: Firm performance (return on assets) with top 10 tradable ownership concentration | |||||||
| Top10_trade | |||||||
| Range | ≤0.0253 | 0.0253–0.0377 | 0.0377–0.0605 | 0.0605–0.1327 | >0.1327 | ||
| Means | 0.0013 | 0.0102 | 0.0192 | 0.0342 | 0.0586 | 58.24 | 0.001 |
| Obs | 395 | 395 | 395 | 395 | 395 | ||
- Firm performance is represented by return on assets. Ratio_state is the ratio of state shares. Ratio_legal is the ratio of legal person shares. Ratio_trade is the ratio of tradable shares. Top10_total is the ratio of shares held by the top 10 shareholders to total outstanding shares. Top10_trade is the ratio of tradable shares held by the top 10 tradable shareholders to total tradable shares.
In panel 1 of Table 2, we know that firm performance as a function of change of state ownership displays a U(V) shape. The return on assets is down towards the bottom in the middle class and approaches the top in the latest class, where state ownership is at the highest level. The result is consistent with Ng et al.’s (2009) argument of a convex relationship. In panel 2, a similar U(V) shape of firm performance can be observed with changes in legal person ownership. This function U(V) is relatively flat, because discrepancies in return on assets between the classes are smaller than those in state ownership. This result supports Delios and Wu’s (2005) findings. In panel 3, firm performance displays a U(V) shape in the first three classes, with a low tail in the last two classes, and an obvious downward trend as tradable ownership increases (Xu and Wang, 1999).
In panel 4, the return on assets associated with top 10 ownership concentration displays an asymmetric U(V) shape. The return on assets in the second class is the smallest. However, the top 10 ownership concentration obviously drives firm performance in an upward trend. This U(V) shape has a smaller curvature and a more obvious upward trend than the U(V) shapes in panels 1 and 2. Finally in panel 5, the top 10 tradable ownership concentration drives firm performance in a persistent upward trend. It looks like a linear relationship even though it is not strictly a straight line. Apparently, tradable ownership concentration has a unique effect on firm performance that is distinct from any other categories of ownership or ownership concentration.
To further verify our analyses, we regress firm performance against each category of ownership or ownership concentration as well as against a set of control variables. Table 3 shows that before we add the quadratic terms, only the coefficient of tradable ownership (model 5) is negative and at 1 per cent significance, representing a high level of tradable share ratio combined with a low firm performance because of dispersed ownership. The coefficients of the level terms of state and legal person ownerships (models 1 and 3) are positively insignificant, indicating that neither state nor legal person ownership has determined influence on firm performance. The top 10 and top 10 tradable ownership concentrations (models 7 and 9) have the coefficients of 0.074 and 0.1534 with the t-values of 7.03 and 11.43, respectively, larger and more significant than any other coefficients of level terms, which implies that high levels of ownership concentration have a more positive impact on firm performance than high levels of state or legal person ownership.
| Model | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
|---|---|---|---|---|---|---|---|---|---|---|
| Intercept | −0.3524** (−9.71) | −0.3290** (−8.92) | −0.3655** (−10.04) | −0.3490** (−9.43) | −0.3357** (−9.31) | −0.3030** (−7.72) | −0.3814** (−10.66) | −0.3149** (−7.08) | −0.2832** (−7.98) | −0.2743** (−7.81) |
| Ownership and ownership concentration | ||||||||||
| Ratio_state | 0.0051 (1.71) | −0.049** (−2.87) | ||||||||
| Ratio_state 2 | 0.0834** (3.29) | |||||||||
| Ratio_legal | 0.0083 (1.08) | −0.034 (−1.84) | ||||||||
| Ratio_legal 2 | 0.0612* (2.37) | |||||||||
| Ratio_trade | −0.0551** (−5.03) | −0.1492** (−3.25) | ||||||||
| Ratio_trade 2 | 0.1047* (2.11) | |||||||||
| Top10_total | 0.074** (7.03) | −0.088 (−1.64) | ||||||||
| Top10_total 2 | 0.1416** (2.51) | |||||||||
| Top10_trade | 0.1534** (11.43) | 0.3025** (11.85) | ||||||||
| Top10_trade 2 | −0.3350** (−6.84) | |||||||||
| Control variables | ||||||||||
| Ratio_indir | 0.1111** (4.64) | 0.1129** (4.73) | 0.1105** (4.62) | 0.1097** (4.59) | 0.1104** (4.64) | 0.1107** (4.66) | 0.1093** (4.62) | 0.1083** (4.58) | 0.1022** (4.41) | 0.1024** (4.46) |
| Num_dir | 0.0096 (1.61) | 0.0106 (1.77) | 0.0101 (1.69) | 0.0100 (1.68) | 0.0082 (1.38) | 0.0078 (1.30) | 0.0071 (1.19) | 0.0065 (1.10) | 0.0084 (1.45) | 0.0092(1.60) |
| Dir_mting | −0.0040 (−1.11) | −0.0044 (−1.23) | −0.0046 (−1.26) | −0.0044 (−1.22) | −0.0032 (−0.88) | −0.0034 (−0.94) | −0.0035 (−0.97) | −0.0037 (−1.03) | −0.0052 (−1.48) | −0.0057 (−1.65) |
| Ratio_debt | −0.1478** (−19.31) | −0.1447** (−18.81) | −0.1480** (−19.36) | −0.1462** (−19.05) | −0.1469** (−19.31) | −0.1455** (−19.06) | −0.1436** (−18.94) | −0.1413** (−18.51) | −0.1412** (−19.00) | −0.1409** (−19.17) |
| Total_asset | 0.019** (11.98) | 0.0179** (11.05) | 0.0196** (12.45) | 0.0190** (11.90) | 0.0194** (12.46) | 0.0188** (11.90) | 0.0185** (11.94) | 0.0175** (10.94) | 0.0156** (10.10) | 0.0149** (9.69) |
| Industry dummy | Included | Included | Included | Included | Included | Included | Included | Included | Included | Included |
| Inflection point and proportion of observations on downside and upside of U or ∩ shape | ||||||||||
| Inflection point | 0.2960 | 0.2805 | 0.7124 | 0.3091 | 0.4515 | |||||
| Obs% on downside | 45.97 | 49.77 | 99.20 | 1.57 | 0.56 | |||||
| Obs% on upside | 54.03 | 50.23 | 0.80 | 98.43 | 99.44 | |||||
| Adj R2 | 0.2028 | 0.2119 | 0.2087 | 0.2106 | 0.2176 | 0.219 | 0.2270 | 0.2291 | 0.2569 | 0.2738 |
| F | 52.82 | 49.26 | 53.06 | 48.87 | 55.91 | 51.32 | 58.97 | 54.33 | 69.25 | 68.67 |
| P>F | 0.0001 | 0.0001 | 0.0001 | 0.0001 | 0.0001 | 0.0001 | 0.0001 | 0.0001 | 0.0001 | 0.0001 |
- The dependent variable is the return on assets. Ratio_state is the ratio of state shares and Ratio_state2 is its quadratic term. Ratio_legal is the ratio of legal person shares and Ratio_legal2 is its quadratic term. Ratio_trade is the ratio of tradable shares and Ratio_trade2 is its quadratic term. Top10_total is the ratio of shares held by the top 10 shareholders to total shares outstanding and Top10_total2 is its quadratic term. Top10_trade is the ratio of tradable shares held by the top 10 tradable shareholders to the total tradable shares and Top10_trade2 is its quadratic term. Ratio_indir is the ratio of independent directors. Num_dir is the logarithm of the number of directors. Dir_mting is the logarithm of director meeting frequency. Ratio_debt is the ratio of debt to total assets. Total_asset is the logarithm of total assets. Industry dummy variables include utilities, property, conglomerations, manufacturing and commercial. Here * and ** represent significance at 5% and 1%, respectively. Figures in parentheses are t-statistics.
After we add the quadratic terms, the coefficients of the level terms become negative and the coefficients of the quadratic terms are positive for state, legal person, tradable ownership and top 10 ownership concentration (models 2, 4, 6 and 8). The coefficient of the level term remains positive, and the coefficient of the quadratic term is negative only for the top 10 tradable ownership concentration (model 10). Importantly, all coefficients of quadratic terms are significant at the 5 per cent and 1 per cent conventional levels. Thus, this seems to indicate that firm performance is a U-shaped function of state ownership, legal person ownership, tradable ownership and top 10 ownership concentration. Firm performance looks like a ∩ shape with top 10 tradable ownership concentration.
To gain insight into firm performance, we differentiate the returns on assets with respect to ownership or ownership concentration within the regression equations that have quadratic terms (McConnell and Servaes, 1990). Inflection points of ownership or ownership concentration ratios occur where firm performance is minimum for a U-shaped function or firm performance is maximum for a ∩-shaped function. We also calculate the percentages of observations on each side of the inflection points, which are reported at the bottom of Table 3.
By observing the proportions of observations on each side of the inflection points, we find that, interestingly, only firm performance with respect to state and legal person ownership are true U-shaped functions, as the proportions of observations on the downside and upside are almost equivalent. Firm performance with respect to tradable ownership actually represents a downward linear trend, as 99.20 per cent of firm observations are on the downside and 0.8 per cent of firm observations are on the upside of the U shape. In contrast, firm performance with respect to top 10 ownership concentration is almost a positively linear trend, as 1.57 per cent of firm observations are on the downside and 98.43 per cent of firm observations are on the upside of the U shape. Firm performance with respect to top 10 tradable ownership concentration is closer to a positively linear trend with 0.56 per cent of firm observations on the downside and 99.44 per cent of firm observations on the upside of the ∩ shape. The evidence in the anova and regression analyses is consistent. Therefore, we are aware that some specious U or ∩ shapes detected from the signs of regression coefficients may just be the result of several extreme observations. When we consider the majority of observations, we have to acknowledge the upward or downward trends behind the shapes.
The results of the preceding anova and regression analyses provide supporting evidence for our hypotheses. First, a high ratio of a category of ownership does not necessarily mean a high ownership concentration. The high level of tradable ownership may in fact illustrates low ownership concentration, because massive numbers of shareholders may own a firm and each individual holds only a tiny proportion of shares. Low ownership concentration incurs the ‘free ride’ problem and hence leads to low firm performance. A high level of legal person ownership can represent either high or low ownership concentration, because there can be a number of legal person shareholders in a firm. A high level of state ownership seems to represent high level of ownership concentration as there is only one state. However, because different state agencies are the actual controllers of the state shares, the possible concentrated state ownership is diluted. Therefore, the above analyses validate our first hypothesis (H1): ownership concentration, rather than state or legal person ownership, has a more powerful and positive effect on the performance of China’s listed firms.
Second, in China’s unique ownership setting, ownership concentration can be represented as either total ownership concentration or tradable ownership concentration. Even though tradable ownership concentration is much lower than that of total ownership concentration, it significantly reinforces discipline on firm management. Larger tradable shareholders monitor firms through voting by either hands or ‘feet’– that is, selling shares and leaving the firms – while non-tradable shareholders monitor firms through voting by hands only. Selling shares is a threat to both the management team and non-tradable controlling shareholders and thus disciplines that shareholders’ interests are not neglected. In addition, the top tradable shareholders do not have enough power from their ownership to expropriate minority shareholders. Therefore, the above analyses also validate our second hypothesis (H2): tradable ownership concentration has a more significant and positive effect on firm performance than total ownership concentration.
5. Interactive effect analyses
We have investigated firm performance with respect to categories of simplex ownership or ownership concentration. In the real world, owners interactively determine firm performance. They compete with each other for their own interests and objectively monitor each other for common interests. It is a fact that the state and legal persons are always the major owners of most of China’s firms and so we now examine their interactive effect on firm performance.
The results of an interactive anova with state and legal person ownerships are arranged in panel 1 of Table 4. The rows represent firms according to their level of state ownership, and the columns represent firms in terms of their level of legal person ownership. Ownership ranges are the same as designated in Table 2. Thus, we find that firm performance in three of the five classes in the rows (1, 2 and 4) and in four of the five classes in the columns (1, 3, 4 and 5) do not display a U shape. Instead, firm performance reveals more increasing patterns of intervals in interlaced levels of state and legal person ownership.
| Panel 1: Firm performance by interactive determination of state and legal person ownerships | ||||||||
|---|---|---|---|---|---|---|---|---|
| Ratio_state | Ratio_legal | |||||||
| Quintuple | 1 | 2 | 3 | 4 | 5 | Total | ||
| Class | Range | =0 | 0–0.1033 | 0.1033–0.3467 | 0.3467–0.5972 | >0.5972 | ||
| 1 | =0 | Means | 0.0350 | 0.0781 | 0.0234 | 0.0164 | 0.0278 | 0.0243 |
| Obs | 13 | 5 | 31 | 192 | 320 | 561 | ||
| 2 | 0–0.2994 | Means | 0.0392 | 0.0419 | 0.0056 | 0.0185 | 0.0280 | 0.0170 |
| Obs | 4 | 18 | 113 | 172 | 46 | 353 | ||
| 3 | 0.2994–0.4864 | Means | 0.0242 | 0.0080 | 0.0115 | 0.0257 | – | 0.0130 |
| Obs | 20 | 78 | 217 | 38 | – | 353 | ||
| 4 | 0.4864–0.6260 | Means | 0.0293 | 0.0228 | 0.0392 | 0.0333 | 0.0456 | 0.0283 |
| Obs | 55 | 194 | 67 | 30 | 7 | 353 | ||
| 5 | >0.6260 | Means | 0.0436 | 0.0365 | 0.0189 | 0.0251 | 0.0531 | 0.0411 |
| Obs | 118 | 146 | 13 | 10 | 68 | 355 | ||
| Total | Means | 0.0374 | 0.0261 | 0.0152 | 0.0194 | 0.0320 | ||
| Obs | 210 | 441 | 441 | 442 | 441 | |||
| F = 6.79, p < 0.001 | ||||||||
| Panel 2: Firm performance by interactive determination of top 10 and top 10 tradable share ownership concentrations | ||||||||
|---|---|---|---|---|---|---|---|---|
| Top10_total | Top10_trade | |||||||
| Quintuple | 1 | 2 | 3 | 4 | 5 | Total | ||
| Quintuple | Range | ≤0.0253 | 0.0253–0.0377 | 0.0377–0.0605 | 0.0605–0.1327 | >0.1327 | ||
| 1 | ≤0.5216 | Means | −0.0037 | 0.0002 | 0.0196 | 0.0318 | 0.0425 | 0.0136 |
| Obs | 122 | 71 | 85 | 77 | 40 | 395 | ||
| 2 | 0.5216–0.6045 | Means | 0.0020 | 0.0016 | 0.0115 | 0.0245 | 0.0262 | 0.0113 |
| Obs | 85 | 94 | 93 | 82 | 41 | 395 | ||
| 3 | 0.6045–0.6575 | Means | 0.0029 | 0.0153 | 0.0159 | 0.0340 | 0.0454 | 0.0212 |
| Obs | 81 | 87 | 85 | 85 | 57 | 395 | ||
| 4 | 0.6575–0.7174 | Means | −0.0025 | 0.0127 | 0.0171 | 0.0307 | 0.0481 | 0.0221 |
| Obs | 70 | 86 | 76 | 77 | 86 | 395 | ||
| 5 | >0.7174 | Means | 0.0196 | 0.0251 | 0.0380 | 0.0526 | 0.0798 | 0.0553 |
| Obs | 37 | 57 | 56 | 74 | 171 | 395 | ||
| Total | Means | 0.0013 | 0.0102 | 0.0192 | 0.0342 | 0.0586 | ||
| Obs | 395 | 395 | 395 | 395 | 395 | |||
| F = 13.17, p < 0.001 | ||||||||
- Firm performance is represented by return on assets. Ratio_state is the ratio of state shares. Ratio_legal is the ratio of legal person shares. Top10_total is the ratio of shares held by the top 10 shareholders to total outstanding shares. Top10_trade is the ratio of tradable shares held by the top 10 tradable shareholders to the total tradable shares.
To have an overall picture of the interactive effect of state and legal person ownerships on firm performance, we design an interactive variable of (1 + Ratio_state) × (1 + Ratio_legal) – 1, which is represented by Ratio_state * Ratio_legal, and its quadratic term is represented by Ratio_state * Ratio_legal2. We regress the firm performance measure of return on assets with the interactive variable and its quadratic term, respectively. In model 11 of Table 5, the coefficient of the level term is positive at 1 per cent significance, while in model 12, neither the coefficient of the level term nor the coefficient of the quadratic term is significant at the conventional level. The results demonstrate that the interactive effect of state and legal person ownerships associates with firm performance in a positively sloped linear shape.
| Model | 11 | 12 | 13 | 14 | 15 | 16 |
|---|---|---|---|---|---|---|
| Intercept | −0.3583** (−9.96) | −0.3676** (−9.78) | −0.3233** (−9.24) | −0.3557** (−9.94) | −0.3041** (−8.39) | −0.3237** (−9.24) |
| Ownership and ownership concentration | ||||||
| Ratio_state | 0.0104 (1.37) | |||||
| Ratio_legal | 0.0119 (1.58) | |||||
| Top10_total | 0.0335** (2.55) | |||||
| Top10_trade | 0.1417** (9.97) | |||||
| Ratio_state * Ratio_legal | 0.0129** (3.06) | 0.0251 (1.7) | 0.0013 (0.32) | |||
| Ratio_state * Ratio_legal2 | −0.006 (−0.86) | |||||
| Top10_total * Top10_trade | 0.0626** (11.68) | 0.1207** (7.82) | 0.0622** (11.24) | |||
| Top10_total * Top10_trade2 | −0.03** (−4.01) | |||||
| Control variables | ||||||
| Ratio_indir | 0.1086** (4.54) | 0.1095** (4.58) | 0.1033** (4.46) | 0.1037** (4.49) | 0.1008** (4.36) | 0.1031** (4.45) |
| Num_dir | 0.0098 (1.63) | 0.0096 (1.61) | 0.0063 (1.09) | 0.0062 (1.08) | 0.0073 (1.26) | 0.0063 (1.1) |
| Dir_mting | −0.0042 (−1.16) | −0.0040 (−1.11) | −0.0041 (−1.19) | −0.0040 (−1.15) | −0.0047 (−1.36) | −0.0041 (−1.19) |
| Ratio_debt | −0.1469** (−19.22) | −0.1475** (−19.22) | −0.1391** (−18.70) | −0.1395** (−18.82) | −0.1391** (−18.75) | −0.1391** (−18.68) |
| Total_asset (billion) | 0.019** (12.12) | 0.0192** (12.12) | 0.0159** (10.35) | 0.0163** (10.59) | 0.0154** (9.85) | 0.0159** (10.34) |
| Industry dummy | Included | Included | Included | Included | Included | Included |
| Inflection point and proportion of observations on downside and upside of U or ∩ shape | ||||||
| Inflection point | 2.0402 | |||||
| Obs% on downside | 0.20 | |||||
| Obs% on upside | 99.8 | |||||
| Adj R2 | 0.2113 | 0.1212 | 0.2590 | 0.2664 | 0.2635 | 0.2585 |
| F | 53.88 | 49.04 | 65.99 | 65.58 | 55.32 | 63.60 |
- The dependent variable is return on assets. Ratio_state is the ratio of state shares. Ratio_legal is the ratio of legal person shares. Top10_total is the ratio of shares held by the top 10 shareholders to total outstanding shares. Top10_trade is the ratio of tradable shares held by the top 10 tradable shareholders to total tradable shares. Ratio_state * Ratio_legal is an interactive variable that is equal to (1 + Ratio_state) × (1 + Ratio_legal) – 1 and Ratio_state * Ratio_legal2 is its quadratic term. Top10_total * Top10_trade is an interactive variable that is equal to (1 + Top10_total) × (1 + Top10_trade) – 1 and Top10_total * Top10_trade2 is its quadratic term. Ratio_indir is the ratio of independent directors. Num_dir is the logarithm of the number of directors. Dir_mting is the logarithm of director meeting frequency. Ratio_debt is the ratio of debt to total assets. Total_asset is the logarithm of total assets. Industry dummy variables include utilities, property, conglomerations, manufacturing and commercial. Here * and ** represent significance at 5% and 1%, respectively. Figures in parentheses are t-statistics.
The explanation for the U(V) shape of firm performance determined by either legal person or state ownership in quite a number of publications (such as Wei et al., 2005; Delios and Wu, 2005; Ng et al., 2009) is that as the concentrated legal person or state ownership increases, the large shareholders – whether a legal person or a state agency – may expropriate the minority shareholders by transferring assets or funds to their parent companies. When the ownership increases above a certain level, the interests of the legal person or state agency become fully aligned with the firm. Thus, a further increase in legal person or state ownership will improve firm value. If this statement is correct, the competition between legal person and state agency mitigates the possible expropriations and retains the monitoring incentive. Thus, our third hypothesis (H3) is verified: The interactive effect of state and legal person ownerships enhances firm management and is positively associated with firm performance.
The concentrated non-tradable ownership and tradable ownership can complement each other. The interactive effect of total ownership concentration and tradable ownership concentration is more powerful than the individual effect of either total ownership or tradable ownership concentration in China’s market. To test our fourth hypothesis, we conduct an interactive anova with top 10 ownership and top 10 tradable ownership concentrations. In panel 2 of Table 4, the rows are the firms classified in quintuples of the top 10 ownership concentration and the columns are the firms classified in quintuples of the top 10 tradable ownership concentration. The rows show that (except for a minor difference in the second quintuple) firm performance increases as tradable ownership concentration goes up at almost every level of top 10 ownership concentration. The change in total ownership concentration does not alter the linear relationship between tradable ownership concentration and firm performance. The columns show that firm performance in three of five quintuples (3, 4 and 5) displays an asymmetric U shape and in one quintuple (2) represents a straight line. Nevertheless, performance in every column reveals an upward trend.
Interestingly, by observing panel 2 of Table 4, we find that firm performance can be represented by an inclined and increasing plane from the top-left corner to the bottom-right corner. A minimum return on assets of −0.0037 appears at the intersection of the lowest levels of the two categories of ownership concentrations; meanwhile, a maximum return on assets of 0.0798 occurs at the intersection of the highest levels of the two categories of ownership concentrations. More attractively, firm performance with the lowest levels of both top 10 and top 10 tradable ownership concentrations has a smaller value (−0.0037) than that of the single lowest level of either the top 10 (0.0136) or top 10 tradable (0.0013) ownership concentrations. In contrast, firm performance with the highest levels of both total and tradable ownership concentrations has a greater value (0.0798) than that of the single highest level of either the top 10 (0.0553) or top 10 tradable (0.0586) ownership concentration.
We apply the second interactive variable in our regression analyses with models 13 and 14 in Table 5. Here, an interactive variable of (1 + Top10_total) × (1 + Top10_trade) – 1 is represented by Top10_total * Top10_trade, and its quadratic term is represented by Top10_total * Top10_trade2. Before and after adding the quadratic term, all the coefficients of the level and quadratic terms are at 1 per cent significance. In particular, the level term in model 13 is significant, with an extremely large t-value of 11.68, i.e., larger than t-values of the level terms in any other category of ownership or ownership concentration.
As the quadratic term is positively significant in model 14, firm performance seems to represent a ∩ shape with the interactive effect of two categories of ownership concentrations. However, we differentiate the return on assets with respect to the interactive variable within the regression equation that has the quadratic term. Thus, we find the inflection point of (1 + Top10_total) × (1 + Top10_trade) – 1 to be 2.042. In the sample, only four of 1975 observations (0.2 per cent) are on the downside of the ∩ shape. Interactive effect of ownership concentrations drives firm performance upward higher than that of any single ownership concentration, which has been pre-detected in the anova analysis. Overall, the results strongly support our fourth hypothesis (H4): the interactive effect of total ownership concentration and tradable ownership concentration is positively related to firm performance and is more powerful than the individual effect of either total ownership concentration or tradable ownership concentration.
6. Joint effect analyses
We have previously argued that a high proportion of a category of ownership does not necessarily represent a high ownership concentration. For convenience, we treat ownership and ownership concentration differently. This section considers the joint effect of different ownerships and ownership concentrations on firm performance. We ascertain whether ownership or ownership concentration has more power in determining firm performance. Accordingly, we find the factors which are most relevant to performance. To do so, we undertake a new regression analysis with models 15 and 16 in Table 5.
In model 15, the explanatory variables of state ownership, legal person ownership, top 10 ownership concentration and top 10 tradable ownership concentration are parallel. Tradable ownership is omitted to avoid the problem of multicollinearity. We find that both the coefficients and the t-statistics of either state or legal person ownership are a lot smaller than those of either top 10 ownership concentration or top 10 tradable ownership concentration. The former are statistically insignificant, while the latter are significant at the 5 per cent level or higher.
We add the quadratic terms for each ownership or concentration variables in model 15.5 Owing to the joint effect, the U shape of firm performances detected in Section 4 completely disappears for state ownership and legal person ownership and is very weak for top 10 ownership concentration. The ∩ shape of firm performance with respect to top 10 tradable ownership concentration is still significant. As only 12 of 1975 observations are on the downside, firm performance displays an almost linear relation with top 10 tradable ownership concentrations.
In model 16, we substitute the individual variables of ownership and ownership concentration with two interactive variables, Ratio_state * Ratio_legal and Top10_total * Top10_trade. Also owing to the joint effect, the interactive variable of state and legal person ownerships has a small and insignificant coefficient. In contrast, the interactive variable of the top 10 and top 10 tradable ownership concentrations has a large and significant coefficient at the 1 per cent significance with a t-value of 11.24.
Likewise, we add the quadratic terms for each interactive variable in model 2.6 The coefficient of the quadratic term of the interactive variable of state and legal person ownerships is small and insignificant. Meanwhile, the coefficient of the quadratic term of the interactive variable of the top 10 and top 10 tradable ownership concentrations is large and negatively significant at 1 per cent. As only 7 of 1975 observations are located on the downside of the ∩ shape, the interactive effect of ownership concentrations is associated with high firm performance.
Thus, the preceding results for considering a joint effect strongly indicate that ownership concentration is more powerful than the categories of ownership in influencing firm performance. The evidence here and in Section 4 is consistent and therefore verifies again our first hypothesis that ownership concentration has a more powerful effect on firm performance than state or legal person ownership, regardless of whether it is joint or individual.
7. Robustness test
In this section, we replace the accounting measure of return on assets with the market measure of Tobin’s Q to conduct a robustness test on the firm performance determined by ownership or ownership concentration.7 We first apply the simplex analysis by repeating models 1 to 10 of Table 3, i.e. we regress Tobin’s Q with the different categories of ownership and ownership concentration variables, respectively. We find that, first, the signs of the coefficients of any ownership or ownership concentration variables and their quadratic terms have no meaningful change. Second, the significance of coefficients is either mostly maintained in the original level or improved to some extent. Third, the allocation of observations on the downside or upside around the inflection point of a category of ownership or ownership concentration varies a margin. Although the Adjusted R-squares and F-tests are a bit smaller than their counterparts in Table 3, the evidence generated from the regression on the market measure of Tobin’s Q is consistent with those from the regression on the accounting measure of return on assets. Ownership concentration increases firm performance more than any other category of ownership. Tradable ownership concentration has a more powerful effect in increasing firm performance.
We then apply the interactive analysis by repeating models 11–14 of Table 5, in which we regress Tobin’s Q with either the interactive variable of state ownership and legal person ownerships or the interactive variable of total ownership concentration and tradable ownership concentration, respectively. We find that the coefficients of the level term and quadratic term of model 12 become statistically significant. As the number of observations on the downside is only 3.80 per cent of total observations, the interactive effect of state and legal person ownerships seems to have a possible linear relationship with the market measure of firm performance. However, the coefficients of the level term and quadratic term of model 14 become more significant. In particular, the number of observations on the downside is merely 0.1 per cent of total observations, and so the interactive effect of total ownership concentration and tradable ownership concentration is a dominating factor that has positive association with firm performance.
Finally, we apply the joint analysis by repeating models 15 and 16 of Table 5, in which we regress Tobin’s Q with ownership variables and ownership concentration variables or the interactive ownership variable and the interactive ownership concentration variable, respectively. We find that the signs and the levels of significance of the coefficients are not meaningfully different between the two regressions. The coefficients related to ownership concentration are positively significant at the 1 per cent level while the coefficients related to state and legal person ownership are insignificant at conventional level. Ownership concentration has a more powerful positive effect on firm performance than any other category of ownership.
8. Conclusions
The impact of ownership and ownership concentration on firm performance has been investigated worldwide. Research on ownership and ownership concentration regarding China’s market has generated a number of publications. However, prior research on China’s listed firms may not have considered important differences between ownership and ownership concentration and between total ownership concentration and tradable ownership concentration. This article fills this void to some extent. By recognizing those differences, this article focuses on the effects of total ownership concentration and tradable ownership concentration on firm performance, in comparison with the effects of state ownership and legal person ownership.
We find, first, that a high level of a category of ownership does not necessarily represent a high level of ownership concentration. For example, there may be a number of legal person owners in a firm. State shares of a firm may be held by different state agencies. A high level of individual shareholders only represents massive numbers of smaller shareholders of a firm. Our results show that ownership concentration has a more powerful and positive effect on the performance of China’s listed firms than the levels of state or/and legal person ownerships.
Second, ownership concentration can be classified as total ownership concentration (that mainly represents non-tradable ownership concentration) and tradable ownership concentration. The large tradable shareholders may sell their shares when they are dissatisfied with management, which decreases share prices and increases capital costs. This is a threat that disciplines managers’ and controlling shareholders’ behaviour. Thus, tradable ownership concentration has a more significant and positive influence on firm performance than total ownership concentration.
Third, it is a general phenomenon that state and legal persons are large shareholders in most of the firms. The state owner represented by different state agencies and legal person owners compete with and monitor each other, which mitigates expropriation by controlling shareholders and increase efficiency. The interactive effect of state and legal person ownerships enhances firm management and draws the performance to be an approximate upward linear trend.
Fourth, the large non-tradable and large tradable shareholders can be complementary to each other in monitoring the firms and discipline the managers’ behaviour. Our evidence shows that a firm with a high level of both total ownership and tradable ownership concentration has an extremely high firm performance. In particular, the interactive effect of total ownership and tradable ownership concentrations has a more powerful determination on firm performance than either any individual effect of ownership concentration or interactive effect of state and legal person ownerships.
Our findings propose a final conclusion that ownership concentration has a positive impact on firm performance, regardless of whether the block shareholders are the state or legal persons. The concentrated pure shareholders – such as tradable shareholders who have complete rights of ownership – have a greater power in determining firm performance than impure shareholders, such as state and legal person shareholders who have incomplete rights of ownership for selling shares on the public market. We believe our proposition needs further evidence.
