Corporate Governance Quality, Ownership Structure, Agency Costs and Firm Performance. Evidence from an Emerging Economy

The purpose of this paper is to investigate the effect of corporate governance quality and ownership structure on the relationship between the agency cost and firm performance. Both the fixed-effects model and a more robust dynamic panel generalized method of moment estimation are applied to Chinese A-listed firms for the years 2008 to 2016. The results show that the agency–performance relationship is positively moderated by (1) corporate governance quality, (2) ownership concentration, and (3) non-state ownership. State ownership has a negative effect on the agency–performance relationship. Various robust tests of an alternative measure of agency cost confirm our main conclusions. The analysis adds to the empirical literature on agency theory by providing useful insights into how corporate governance and ownership concentration can help mitigate agency–performance relationship. It also highlights the impact of ownership type on the relationship between agency cost and firm performance. Our study supports the literature that agency cost and firm performance are negatively related to the Chinese listed firms. The investors should keep in mind the proxies of agency cost while choosing a specific stock. Secondly; the abuse of managerial appropriation is higher in state-held firms as compared to non-state firms. Policymakers can use these results to devise the investor protection rules so that managerial appropriation can be minimized.


Introduction
Opportunistic managers, rather than maximizing the shareholder's wealth, tend to misuse the organizational resources for their own benefit. A good set of governance practices and ownership structures can mitigate the conflict of interest between the principles and the agents, hence enhancing the firm value. The purpose of this paper is to investigate the effect of corporate governance quality and ownership structure on the relationship between agency cost and firm performance. This is one of the few studies to explore the relationship between agency costs and firm performance in a dynamic modeling approach. Specifically, this study aims to address the following research questions: (1) Does corporate governance quality mitigates the relationship between agency cost and firm performance? (2) How do ownership concentrations affect the relationship between agency cost and firm performance? (3) How do state and non-state companies moderate the relationship between agency cost and firm performance?
Data is extracted of 2248 Chinese A-listed companies for the period 2008-2016. Using both fixed effects and dynamic panel generalized methods of moment estimation, the results show that agency cost is negatively related to firm performance. At the same time, corporate governance and ownership concentration enhances firm performance. When corporate governance and ownership concentration are taken as a moderating variable, we find a positive impact on the agency-performance relationship. We also studied the effect of ownership type on the association between agency cost and firm performance. Non-state ownership positively moderates the relationship between agency cost and firm performance. In contrast, the agency cost keeps its negative sign when the state ownership is taken as a moderating variable. Our conclusions are supported by taking alternative measures of independent variables for robustness.

Background of the Study
The motivations for this study can be broken down into the following aspects: (1) Why China should be used as the test case to conduct this research. (2) Why corporate governance quality matters in an emerging economy like China. (3) Why dynamic panel modeling approach was used in this study. The following Sections 2.1-2.3 briefly discuss these questions.

Why China?
Emerging markets are prone to managerial discretion to a greater extent compared to in Anglo-American countries. Managers in these economies tend to manage funds inefficiently, which directly affects firm performance. Compared to developed economies, the extent of agency cost is different in emerging economies, specifically in China. Many researchers have used the proxy of agency cost based on either the managerial discretion or ineffective use of shareholder's funds.
We chose Chinese listed firms for this study because China's market for corporate control and the stock market mechanism is unique. Established in the early 1990s, the Chinese stock market was used as a vehicle to transform the "planned economy" to a "market economy." The Chinese Securities Regulatory Commission (CSRC hereafter) under the umbrella of the Chinese government made these reforms possible. Chinese listed firms rely heavily on internal financing, such as retained earnings, rather than external funding. The reliance on internal financing gives managers discretion to manipulate funds for self-empire building or investing inefficiently.
During the wave of recent privatization, about 60% of the Chinese market is still under the direct or indirect control of the state. The word corporatization is used instead of privatization by Lin (2001). He suggested that although the market for corporatization falls in the hands of the state, the firm's governance will be improved. The corporatization process has made it difficult for stakeholders to distinguish between state-held and non-state-held companies (Milhaupt and Zheng 2015). State control leads to more market power and easy access to external financing, which ultimately leads to a considerable amount of funds at the disposal of managers. Firms with concentrated state ownership behave differently when compared to non-state firms. For this purpose, the agency cost in state-owned institutions is much higher than in privately held firms. Clarke (2003) suggested that state-owned enterprises lack any clearly defined principles, which incites managers to fulfill their interest rather than the interest of the shareholders. Additionally, these managers have strong political backing, which makes them unaccountable for their actions. Therefore, as described by Ding et al. (2007), the performance of state-held firms falls well short of that of privately owned firms. Based on the argument above, we try to answer the research question, whether the ownership type affects the agency-performance relationship or not?

Why Governance Quality Matters in China
In 2001, China became a member of the World Trade Organization (WTO) and adopted the Organization for Economic Co-operation and Development (OECD) principles of corporate governance and started improving the corporate governance of its listed companies. CSRC, in cooperation with National Trade and Economic Commission, issued the code of corporate governance in 2002. The laws were based on the principles of investor protection and the code of conduct of the directors and managers. A continuous improvement took place that led the listed companies to shift the reporting from Chinese Accounting Standards to the International Accounting Standards in 2006. A two-tier board system is a unique aspect of the Chinese corporate governance mechanism. A one-tiered system such as that of USA has all the directors (executives and non-executives) being part of one Board, known as the Board of directors. In the two-tier board system, there is an executive board (consisting of all executive directors) and a supervisory board (consisting of all non-executive directors). The CSRC has taken many steps for the good of corporate governance in listed companies and to protect minority shareholders. Now companies are being encouraged to have at least one third independent directors on Board, the information disclosure act (2007) was explicitly introduced during IPO, and rules relating to shareholder meetings (2006) related to the convening of shareholder meetings and their resolution were introduced. Most of the state and legal person companies were either fully or partially privatized during this transition period.
The steps taken by CSRC to improve the corporate governance mechanism in China are remarkable. They can be used as one of the tools to curtail the opportunistic behavior of the managers. As an emerging economy, the Chinese market is still in the earliest steps of adopting good corporate governance practices.

Why Dynamic Panel Data?
The traditional agency framework was reexamined outside the jurisdiction of the Anglo-Saxon market, especially in the emerging markets. The development in the corporate governance literature has suggested that the governance variables plays an important part (endogenously) in the value maximization process of the shareholders (Nguyen et al. 2015). The agency cost can affect firm performance. Still, due to the link between the agency cost and firm governance variables, we cannot be sure about the relationship, e.g., studies have shown that state ownership is positively related to the presence of agency cost (Wei et al. 2005). Therefore, a dynamic model is required to cater to the endogeneity problem, specifically in the case of China, where investor protection is weak. Additionally, the motivation for using a dynamic panel model in the corporate governance literature is derived from the recent calls by Zhou et al. (2014) and Nguyen et al. (2015).

Theoretical Literature Review
Two sets of approaches in literature have been distinguished by the researchers to mitigate the agency problem. The first approach is the refraining approach, which proposes that the manager's interest could be aligned with that of the shareholders only if they are forced to refrain from opportunistic behavior. The refraining approach consists of leverage (Emanuel et al. 2003;Malmquist 1990;Siregar and Utama 2008), dividend payment (Easterbrook 1984;Lang and Litzenberger 1989), the risk of corporate takeover (Bethel and Liebeskind 1993;Shleifer and Vishny 1991), a strong board structure (Bathala et al. 2017;Jackling and Johl 2009;Miller 2002), independent audit committees (Collier and Gregory 1999;Islam 2010), well-reputed external auditors (Eshleman and Guo 2014;Hope et al. 2012), and oversight by institutional shareholders (Singh and Davidson 2003).
The second approach is known as the encouraging approach, and motivates the managers to do desirable actions. This approach includes performance-based remuneration (Abowd 1990) and employee stock ownership programs (Fox and Marcus 1992;Nikoskelainen and Wright 2007;Singh and Davidson 2003). Agrawal and Knoeber (1996) suggested a different agency-mitigating mechanism and concluded that a single measure could be misleading. Shan (2015) prepared a corporate governance index for eight different corporate governance measures and explored their effect on earnings management and value relevance. Our study also focusses on the design of the corporate governance index, consisting of agency-mitigating variables, and investigates the impact of corporate governance in moderating the relationship between agency cost and firm performance. The result shows that an effective monitoring mechanism (corporate governance quality) can align the interests of shareholders and managers.
Studies on ownership concentration and firm performance follow two contradicting theories. The monitoring argument implies that in the presence of weak governance mechanism, the majority shareholders helps in reducing the agency cost and increasing the overall firm value (Porta et al. 1999;Shleifer and Vishny 1986;Bhagat et al. 2017;Su et al. 2008;Li et al. 2008). The principal-principal theory states that the minority shareholders are exploited when the control of the ownership lies with the majority shareholders. They are the key decision-makers and appoint the management based on personal preferences. The management works to maximize value for the majority shareholders, while the minority shareholders are continuously overlooked (Denis and McConnell 2003;Hu and Izumida 2009). China is regarded as an emerging economy with corporate governance procedures still at an evolving stage. Shareholders are subject to managerial expropriations, and hence concentrated ownership may help mitigate the agency problem.

Agency Cost and Firm Performance
Emerging markets are prone to managerial discretion to a greater extent compared to Anglo-American countries. The managers in these economies tend to manage the funds inefficiently that directly effects the firm performance. Compared to developed economies, the extent of agency cost is different in emerging economies, specifically in China. Many researchers have used a proxy of agency cost based either on managerial discretion or effective use of shareholder funds. The proxies for agency cost mostly used in China are discretionary accruals as a mean to earnings management Wang and Campbell 2012;Guo and Ma 2015), free cash flow coupled with low growth opportunities (Chung et al. 2005b;Chen et al. 2016;Chiou et al. 2010), research and development expenditure (Shust 2015;Dinh et al. 2016;Ruiqi et al. 2017), and administrative expense ratio, which usually includes executives' salaries, travelling allowances, conference levies, etc. Huang et al. 2011;Zhang et al. 2016).
Most of the literature cited on the relationship between the agency cost and firm performance has established a negative relationship. For example, management earnings lead to the negative market performance of firms listed in Hong Kong (Ching et al. 2006). Higher levels of leverage in terms of short-term debt and long-term debt also have a negative effect on firm performance (Yazdanfar and Öhman 2015). A study conducted by Lang et al. (1995) showed that managers' discretion in selling assets led to lower firm performance. Khidmat and Rehman (2014) empirically tested the relationship between agency cost and firm performance in the emerging economy of Pakistan. They found a negative association between the selling and administrative expense ratio and firm performance. The Chinese market is prone to agency cost, and we expect a negative effect of agency cost on Chinese listed firm performance.

H1.
Agency costs have a negative effect on the firm performance.

Corporate Governance, Agency Cost, and Firm Performance
Managers' opportunistic behavior increases their wealth, which leads to a decrease in firm performance. This opportunistic behavior of managers can be curtailed through a good set of internal and external corporate governance principles. Leverage is considered to be an agency mitigating mechanism, as outsiders monitor the actions of managers with respect to efficient contracting (Jensen 1986;Lang et al. 1995;Malmquist 1990). Debt covenants are considered an essential part of efficient contracting that, in addition to active monitoring, prevents specific risk-taking activities by management. All these efforts reduce the agency cost on one hand, while the value of the firm is increased on the other.
Dividend policy is also considered to be an agency mitigating variable, as elaborated by Ghosh et al. (2000). Two ways in which shareholder wealth can be maximized, either by increasing the share price by investing in value-enhancing projects or by distributing the excess cash in the form of dividends if managers fail to identify value-maximizing projects. Managers try to gain hold of the firm's resources when they have no positive NPV projects; hence, the agency cost of free cash flow occurs. In this case, dividends play an essential role in alleviating the agency problem (Jensen 1986). Porta et al. (2000) claimed that with the country having low investor protection, dividend policy becomes a robust governance mechanism for alleviating agency cost.
Studies have shown the effects of different board characteristics on firm performance in emerging economies (Borlea et al. 2017). A strong and independent board structure aligns the interest of managers and shareholders. Jackling and Johl (2009), in their study, concluded that from the perspective of agency theory, having independent directors on the Board enhances firm performance. Independent directors not only the have experience and the knowledge required to scrutinize the opportunistic behavior of the managers, but can also dissent from the other board members if they find any irregularities (Marchetti et al. 2017). The literature has shown mixed results regarding the size of the board, e.g., Mappadang et al. (2018) found that larger board size led to tax avoidance. Some studies have suggested that the optimal board size is either very small or very large (U-shaped), when assessed with respect to performance (Coles et al. 2008). However, a study conducted by Beiner et al. (2004) shows that the choice of the board size is dependent on environmental factors. Pearce and Zahra (1992) suggested that large boards were characterized by efficient monitoring and had a larger impact on corporate performance than small boards. Similarly, in advanced economies, studies have shown a positive association between board size and firm performance (Guest 2009). In the context of the Chinese market, board size has a negative impact on risk taking (Huang and Wang 2015;Haider and Fang 2016). Since board size leads to less risk taking in Chinese listed firms, we can propose that managerial appropriations can be curtailed through having a larger board size, thus boosting firm performance. CEO duality is taken as an important component of board characteristics, and generally, empirical research has shown that the separation of the CEO and the chairman results in an alleviation of the agency cost (Goyal and Park 2002;Kula 2005;Lei et al. 2013). Board diversity is also considered an important element of corporate governance. Research has shown that the representation of females on the board reduces the agency cost and enhance the firm value. For example, the relationship between gender diversity and firm performance in Chinese listed firms was investigated by Sial et al. (2018a). They concluded that board diversity influences firm performance, and that corporate social responsibility mediates the relationship. Similarly, research conducted by Sial et al. (2019) highlighted the importance of gender diversity in moderating the relationship between corporate social responsibility and earnings management. Board activity has a significant negative effect on agency costs. Frequent meetings of the Board of directors mean that they are actively involved in the matters of the company, and managers refrain from self-empire building (Ma and Tian 2009;Sahu and Manna 2013).
Some other corporate governance variables, in addition to board structure, also help in mitigating agency cost. The presence of an audit committee with independent members can proactively identify misappropriations in the financial records and can play a significant role in mitigating the agency problem. The agency cost is reduced when an independent committee is devised voluntarily (Collier and Gregory 1999). Additionally, it enhances firm value (Chan and Li 2008). A well-reputed and experienced external auditor can carefully scrutinize the financial statements. They have the required expertise, as well as having market knowledge of harmful financial practices. A Big Four auditor can mitigate the agency cost, as well as help in enhancing the firm value (Eshleman and Guo 2014;Hope et al. 2012). Some researchers have shown a profound effect of corporate social responsibility (CSR) on firm performance. The study conducted by Sial et al. (2018b) suggested CSR to be an important determinant of corporate governance in enhancing the firm performance. However, the earnings management had the negative impact on the relationship between CSR and firm performance.
The encouraging approach to agency cost states that managers can be motivated to carry out specific desirable behavior. Jensen and Meckling (1976) proposed the convergence of interest hypothesis, in which managers who were given stock ownership have a better effect on firm performance. Brander and Poitevin (1992) explained that the terms offered in the compensation contract reduced the agency cost. In China, the same results were established by Zhang et al. (2016), who concluded that the perks of senior executives were negatively related to the agency cost. Managerial ownership is one of the ways to align the interest of the shareholders and managers. By having an ownership stake in the company, the managers would now take ownership of the company and would do their best to increase its value. However, in the literature, this relationship has not been not found to be linear. Still, there is a monotonic relationship, which suggests that at a lower level of managerial ownership, the agency cost is reduced, but the agency cost increases when a certain level of managerial ownership is reached (Jensen and Meckling 1976;McConnell and Servaes 1990).
Many recent studies have examined a combination of agency mitigating governance variables instead of focusing on the effect of an individual variable. Agrawal and Knoeber (1996) suggested a different agency mitigating mechanism and concluded that a single measure could be misleading. Shan (2015) prepared a corporate governance index for eight various corporate governance measures and explored their effect on earnings management and value relevance. Achim et al. (2016) investigated the effect of overall corporate governance quality in the performance of Romanian firms. They found a positive association between the governance quality and business performance in the emerging economy. This study focuses on the design of the corporate governance index comprising agency mitigating variables, and examines the impact of corporate governance in moderating the relationship between agency cost and firm performance. Dey (2008) used seven different proxies of agency conflict and generated seven principal factors from 22 individual governance variables. She concluded that the existence and role of governance mechanism is a function of the level of agency conflict in the firm. The link between corporate governance, agency cost and the firm performance is elaborated in Table A1, Appendix A.
From the literature above, we establish the link between corporate governance variables, agency cost and firm performance. The corporate governance quality mitigates the corporate expropriation through efficient monitoring. The corporate governance is also linked with better performance specifically in emerging markets (Klapper and Love 2004). With effective corporate governance mechanisms, the agency cost can be curtailed, while higher firm performance can simultaneously be achieved. Based on the link between corporate governance attributes and agency cost reduction, as well as the positive association between the corporate governance dimensions and firm performance, we can devise our hypothesis: H2. Corporate governance quality moderates the agency-performance relationship.

Ownership Concentration, Agency Cost, and Firm Performance
The literature on ownership provides two contradicting theories with respect to ownership concentration and the agency problem. The first theory is based on efficient monitoring. The theory postulates that the majority shareholders have more stake in the firm. They are more vigilant than the minority shareholders. Their monitoring skills make them distinct from the rest of the minority shareholders. Due to efficient monitoring, they are able to reduce managerial expropriation. The second theory, known as principle-principle agency theory, postulates that the majority shareholders exert undue power on management to obtain their own benefits. The minority shareholders are hence exploited by the managers as well as the majority shareholders (Denis and McConnell 2003;Hu and Izumida 2009). In countries where the corporate governance mechanism is weak, ownership concentration works as a substitute for corporate governance (Porta et al. 1999). An increase in ownership concentration leads to shareholder activism. Therefore, agency costs can be reduced (Kroll et al. 1993;Li et al. 2008;Su et al. 2008). As the percentage of individual ownership increases in the firm, more individuals are inclined to incur monitoring costs (Porta et al. 1997). Ma et al. (2010) studied the effect of ownership concentration and firm performance in Chinese listed companies. They found a positive impact of ownership concentration and firm performance, irrespective of who the majority shareholder was. Heugens et al. (2009) performed a meta-analysis of the relationship between ownership concentration and firm performance in Asian countries. They concluded a positive association between the two variables. They further elaborated that ownership concentration was an active corporate governance mechanism for protecting the minority shareholders from managerial appropriation.
In summary, concentrated ownership is linked with better firm performance in emerging economies (Heugens et al. 2009). The concentrated ownership structure helps in the protection of the minority shareholders from the managerial expropriation. Based on the alignment of interest argument, the concentrated ownership mitigates the agency cost in the emerging economies, resulting in improved performance (Chen 2001). The link between ownership concentration, agency cost, and firm performance is stated in the following hypothesis: H3. Ownership concentration moderates the agency-performance relationship.

Ownership Type, Agency Cost, and Firm Performance
Although ownership concentration has a significant effect on alleviating the agency problem, different studies have shown that it depends upon the type of majority shareholder. The objective of the government is to provide employment and social solidity in the economy. In this way, a conflict of interest arises between state-owned enterprises and shareholders (Chong-En et al. 2002). As illustrated by Gunasekarage et al. (2007), the performance of firms decreases if the ultimate shareholder is the state. State-owned companies have substantial market power, easy access to finance, and less monitoring, which makes them difficult to default . Keeping in view the agency perspective, many researchers have found a negative relationship between state ownership and firm performance (Chen 2001;Jia et al. 2005;Wei and Varela 2003;Xu and Wang 1999). Clarke (2003), in his article "Corporate Governance in China: An Overview", demarcated the objectives of state-owned firms as the generation of employment, direct control over strategic industries, and politically motivating employment. This results in state-owned firms having a suboptimal level of performance and higher agency costs. According to Xu and Wang (1999), the ownership concentration in Chinese listed companies is positively related to firm performance. Additionally, state-owned firms have an adverse effect on firm performance and labor productivity. Similarly, Chen et al. (2016), in their study, investigated the impact of free cash flows and corporate governance characteristics on a firm's investment decisions, using data from 865 Chinese listed firms. The results showed that state ownership concentration boosted over-/underinvestment, while firms with good governance attributes mitigated the over-/underinvestment. On a similar note, Huang et al. (2011) examined the effect of agency cost on the relationship between top executives' overconfidence and investment-cash flow sensitivity. Their results showed that investment-cashflow sensitivity was higher in state-owned companies.
Furthermore, they constructed an agency cost proxy and concluded that the agency cost was higher in state-owned companies. A comparative study between state and non-state firms and their effect on earnings management was carried out by Ding et al. (2007). They analyzed 273 privately owned and state-owned companies. They concluded that the privately owned companies tended to maximize their earnings more than the state-owned companies, despite the reported discretionary accruals reported in non-state companies exceeding those of the state-owned companies.
In summary, state-owned firms in China are characterized by having higher agency costs. Hiring in Chinese firms is based on political connections (Jonge 2014). The performance of non-state firms is better than that of state-owned firms (Hess et al. 2010). The effect of state ownership on agency cost and firm performance will be different from that of non-state ownership.
Accordingly, our next hypothesis would be as follows: H4. State (non-state) ownership negatively (positively) moderates the agency-performance relationship.

Data Collection
The data collection started by taking all the Chinese A-listed firms over the period 2008 to 2016. However, we dropped firms based on the following criteria: (i) Firms belonging to financial sector. These firms have different accounting mechanisms and are not subject to current data collection process.
(ii) Firms with ST (special treatment) and PT (particular treatment) status. These firms are financially distressed and may give spurious results. (iii) Firms with missing values for dependent, independent or control variables. (iv) We winsorized 1% of the data from upper and lower values to control for outliers. After all the adjustments, the final sample comprised 2248 firms. The data was extracted from the Chinese Stock Market and Research (CSMAR) database.

Model Specification and Estimation Techniques
The fixed-effects model and system GMM proposed by Blundell and Bond (1998) were used to test our hypothesis and cater to the un-observed endogeneity problem (Nakano and Nguyen 2013;Nguyen et al. 2014).
In the first equation, we want to explore the effect of agency cost and corporate governance on firm performance.
where FP represents firm performance and has more than one measure, AC is the measure of agency cost, while CGQ is the corporate governance quality index. CTR represents the control variables used in the equations. We added the interaction term in Equation (1) to capture the impact of corporate governance quality and ownership structure on the relationship between agency cost and firm performance.
where the moderators are corporate governance quality, ownership concentration, and ownership type, respectively. The financial performance of the firm is time-dependent, i.e., the current performance of the firm is affected by the past performance and testing the effect of two-year lagged performance on current performance does not give us a significant impact. This leads us to conclude that the AR (1) dynamic panel model is sufficient. The literature on corporate governance suggests that corporate governance, as well as ownership structure, are endogenously determined (Nguyen et al. 2015). Therefore, this study first uses the fixed effect model to control the governance variables when estimating the agency-performance relationship. However, the use of a fixed-effect estimator does not eliminate the endogeneity lag performance measures. Following Nguyen et al. (2015), this study uses the system GMM recommended by Blundell and Bond (1998). The major advantage of constructing the system GMM estimator is that it enhances the power of estimation (Hoechle et al. 2012).

Dependent Variables
Following Le and Buck (2011), we incorporated two measures of performance return on assets (ROA) and earnings per share (EPS) as dependent variables. We decided to include only the book measures of performance as the market measure of performance lacks accuracy and precision of valuation (Le and Buck 2011;Park and Luo 2001;Wan and Yuce 2007).

Agency Cost
Our first independent variable is the proxy for different agency cost measures. We measured the agency cost as the ratio of administrative expense to sales. The administrative expense includes the majority of the costs that have been incurred by the management, such as salaries, executive travel allowances, entertainment expenses, conferences and tour expenses, welfare payments, utilities, and other expenses that fall under this category. According to Li et al. (2008), Chinese managers misuse administrative expenses in automobiles, lavish office designs, recreational activities, and traveling. Therefore, administrative overhead can be regarded as a close proxy of agency costs.
Free cash flow is also considered to be a determinant of agency costs (Jensen and Meckling 1976). Still, the financial flexibility view suggests that managers intentionally keep a higher proportion of cash to meet future needs (Arslan-Ayaydin et al. 2014). An alternative definition of free cash flow as a measure of agency cost was used by Rahman and Mohd-Saleh (2008). First, we calculated free cash flow and growth opportunities following Chung et al. (2005a). Then, we created a dummy variable for agency cost that took the value of 1 if the firm in a particular industry in a specific year had free cash flow in excess of the industry median and a price-to-book ratio less than the industry median.
Three additional variables for measuring the extent of agency cost were provided as a robust check of our results. Earnings management, measured by absolute discretionary accruals following Dechow et al. (1995), is considered to be a proxy of agency cost. Research has determined that firms having high agency costs tend to manage their accruals (Christie and Zimmerman 1994;Teoh et al. 1998;Cormier and Martinez 2006). Next, we measured research and development expenditure (R&D-AC). R&D-AC is a dummy variable that takes the value of 1 if the firm 'i' in the year t has a price-to-book ratio less than the industry median and research and development expenditures higher than the industry median. Finally, we constructed an agency cost index through principal component analysis. We took the first principle component as the measure of agency cost.

Corporate Governance Quality
The second independent variable used in this study was the corporate governance quality index. Due to the adoption since 2007 of the new accounting standards and the availability of data under the compliance and explanation statement, we constructed the index with different agency mitigating governance variables. The variables used in the construction of the index were dividend payment, board size, board independence, board diversity, board meeting, CEO duality, Big Four auditor, managerial ownership, managerial compensation, institutional investors, number of commissions established, and separation of control rights and cashflow rights. The collection of corporate governance variables data was based on the study conducted by Shan (2015), who developed a governance index for Chinese listed firms. The detailed measurement of these variables is given in Table 1 below. Following Achim et al. (2016), we used the compliance and explanation statement for the measurement of corporate governance index.

CEO Duality
When CEO is also the board chairman.
Assigned value 1 to firm i in year t if CEO Duality does not exist, 0 for otherwise (Dey 2008;Gaio 2010;Lei et al. 2013).

Independent Directors
The number of independent directors on the Board of directors  Award 1 mark if Board Meeting held in the firm i in year t is greater than the median value of the sample in fiscal year t, 0 mark otherwise (Dey 2008;Vafeas 1999).

Female Director
If the Board has female representation or not.
Award 1 mark if firm i in fiscal year t has a female director on the Board, 0 mark otherwise (Ararat et al. 2010;Carter et al. 2003;Erhardt et al. 2003).

Managerial Compensation
Total compensation awarded to the top three highest-paid managers in the same industry.
Award 1 mark if managerial compensation paid in the firm i in year t is greater than the median value of the sample in fiscal year t, 0 mark otherwise (Lei et al. 2013).

Managerial Ownership
It is the shareholding percentage of top three officials of the firm.
Award 1 mark if managerial ownership held in the firm i in year t is greater than the median value of the sample in a specific industry, 0 marks otherwise (Chong-En et al. 2002;Agrawal and Knoeber 1996).
Dividend Payment Measured by the dividend per share.
Award 1 mark if Dividend paid in the firm i in year t is greater than the median value of the sample in a specific industry, 0 marks otherwise (Easterbrook 1984;Porta et al. 2000).

Institutional Ownership Measured as the ownership held by institutions in the firm
Award 1 mark if institutional ownership held in the firm i in year t is greater than the median value of the sample in a specific industry, 0 marks otherwise (Tang and Chang 2015).
Big Four auditor Hiring a Big Four auditor Award 1 mark if firm i in fiscal year t hires a Big Four auditor, 0 marks otherwise (Gao and Kling 2008;Lei et al. 2013).

Separation of control rights and cash flow rights
Degree of separation between the control right and cash flow right Award 1 mark if separation between the control right and cash flow right in the firm i in year t is less than the median value of the sample in a specific industry, 0 marks otherwise (Lei et al. 2013).

Number of committees established
Total number of committees a firm has.
Award 1 mark if the Committee established in the firm i in year t is greater than the median value of the sample in a specific industry, 0 marks otherwise.
Note: The corporate governance index is made under the compliance and explanation statement rule.

Ownership Concentration
Large shareholders have a lot of stake in the company and can actively monitor the activities of the managers. The greater the degree of ownership, the more active the shareholders are (Kroll et al. 1993;Su et al. 2008;Li et al. 2008). Therefore, we measured ownership concentration as the percentage of shares held by the largest shareholder.

Ownership Type
We divided ownership type into two categories depending upon whether it was held by the state or by a non-state entity. State ownership was measured as the percentage share held by the state. Non-state ownership was defined as the percentage of shares held by non-state entities (Ding et al. 2007).

Control Variables
In this study, we controlled the firm size by taking the natural logarithm of total assets (Nguyen et al. 2014;Richardson et al. 2003). The second control variable was leverage, which was measured as the ratio of total debt to total assets (Arthurs et al. 2008). Growth opportunities also affect the firm value, so their effect was also controlled. This was calculated as the difference between current year sales and previous sales divided by the previous sales (Gill and Biger 2013). We also controlled the firm age, which was calculated as the number of years for which the firms had been listed in the stock exchange (Shan 2015). The summary of all the variables is given below in Table 2. Earnings per share EPS EPS is calculated as the net income scaled by several shares outstanding (Zhang et al. 2014).

Independent Variables
Administrative Expense Ratio AC1 AC1 is measured as the ratio of administrative expenses to sales (Lei et al. 2013).

Free Cash Flows FCFE_AC
FCFE_AC is a dummy variable that takes the value of 1 if the firm 'i' in year t has a price-to-book ratio less than the industry median and free cash flows greater than the industry median (Rahman and Mohd-Saleh 2008).
The absolute value of discretionary accruals EM EM is the measure of earnings management. The discretionary accruals are calculated through a modified Jones model (Dechow et al. 1995).

Research and Development expenditures R&D-AC
R&D-AC is a dummy variable that takes the value of 1 if the firm 'i' in year t has a price-to-book ratio less than the industry median and R&D greater than the industry median.

Industry Dummy INDUSTRY
Industry dummy shows the effect of each company listed in a particular the industry according to CSRC coding (Arora and Dharwadkar 2011;Filatotchev et al. 2007).

Year Dummies YEAR
The year dummies represent the year effect from the year 2008 to 2016. Table 3, Panel A reports the descriptive statistics of all the variables used in our methodology. The overall descriptive statistics show the mean value of performance measures used in our analysis at 4.3 percent, 6.49 percent, and 0.39 per share for ROA, ROE, and EPS, respectively. The average value of the administrative expense ratio is 9.8 percent, while the free cash flows to total assets have a negative average value of −18.7 percent of total assets. The absolute value of discretionary accruals denoted by EM has a mean value of 5.43 percent of total assets. The mean value of the corporate governance quality index represented by CGQ is 5.79. The maximum amount of the shareholding percentage of the largest shareholder (top 1) is 89.99 percent.  Table 3 reports the summary statistics of the variables used in Equations (1) and (2). Additionally, the We divided the descriptive statistics into two more panels based on state ownership and non-state ownership. The mean values of ROA, ROE, and EPS in Non-SOE (4.9 percent, 8.6 percent, and 4.01 respectively) are higher than those of the SOE (3.5 percent, 3.27 percent, 4.02 per share respectively). State-owned enterprises have a higher level of absolute discretionary accruals (0.07 > 0.06), as well as a higher administrative expense ratio (0.102 > 0.081) compared to non-SOE. The non-state companies have a better corporate governance score of 6.07 compared to state companies, which have a score of 5.35. The state-owned companies are larger (22.46 > 21.49), as well as having a higher average leverage (0.505 > 0.353), than non-state-owned companies.

Descriptive Statistics and Multicollinearity Diagnostic
Panel B presents the descriptive statistics of the corporate governance variables used in this study divided into three groups: Overall, State, and Non-State. The A-listed companies from the period 2008 to 2016 have an average independent director of 3.29, reported CEO duality of 26.27 percent of the total sample, average board size of 8.8, while 16.7 percent on average present female representation on the board. On average, non-state firms pay a higher level of dividends (0.13 > 0.11), have a higher female ratio on the board (0.18 > 0.13), have higher managerial ownership (16.63 percent > 11.57 percent), and a higher degree of separation between control rights and cash flow rights (6.26 percent > 4.38 percent) as compared to state-owned companies. All these governance attributes lead to a better governance environment, resulting in less managerial expropriation. State-owned companies have a greater board size (9.46 > 8.38), more institutional ownership (8.8 > 5.8), more access to getting audited by Big Four auditors (0.087 > 0.02) and more independent directors on the board (3.44 > 3.09). Table 4 shows the results for pairwise correlation analysis. Looking at all of the independent variables, we find no sign of multicollinearity, as the values of the coefficients are less than 0.8. Additionally, we performed individual VIF analysis and found all the values to be less than the critical level of 10 in every case (Shan 2015). When performing the regression analysis for the moderating effect, the interaction term usually gives a value for VIF greater than 10. Following Allison (2010), we mean-centered the interaction terms (agency cost variables and their interaction with CGQ, Top1, SOE, and NSOE). The use of mean centering does not affect the probability values, and at the same time, the multicollinearity is reduced, as seen from Table 5. All the dependent variables are positively correlated with corporate governance quality. On the other hand, the agency cost proxies are negatively related to the performance measures. Ownership concentrations also have a positive association with the firm performance. Table 6 shows the effect of corporate governance and agency cost on firm performance by using the fixed-effect model and system dynamic panel data estimation. ROA and EPS were the dependent variables, while agency cost and corporate governance were the independent variables. Four variables, namely, Size, Leverage, Growth, and Firm age, were used as control variables. The asset size used has a positive effect on firm performance while, on the contrary, a higher level of leverage hampers performance. These results are similar to the study conducted by Vithessonthi and Tongurai (2015).  Note: Table 4 reports the correlation coefficients of the variables used in Equations (1) and (2); the detailed calculation is presented in Table 1. * signify p-values of 1 percent.     Table 6 reports the regression results from estimating Equations (1) and (2) respectively. Variable definitions are provided in Table 2. *, **, *** signify p-values at 1 percent, 5 percent, and 10 percent, respectively.

Moderating Effects of Corporate Governance Quality
Looking at the primary results, agency cost (administrative expense ratio and free cash flow) has a significant adverse effect on firm performance measured by ROA and EPS. Corporate governance has a positive impact on both returns on ROA (FE: β = 0.003, p < 0.01; GMM: β = 0.002, p < 0.01) as well as EPS (FE: β = 0.003, p < 0.01; GMM: β = 0.004, p < 0.01). The objective of the study was to investigate the moderating effect of corporate governance quality on the relationship between agency cost and firm performance. To do so, we incorporated the interactions terms of corporate governance quality with agency cost proxies. The results are depicted in Table 6 with ROA and EPS as dependent variables. Corporate governance quality significantly positively moderates the relationship between agency cost and firm performance. When interaction terms are introduced, the coefficient of agency cost has changed the sign from negative to positive for both AC1 × CGQ (ROA-FE: β = 0.004, p < 0.01; GMM: β = 0.012, p < 0.01 and EPS-FE: β = 0.002, p < 0.05; GMM: β = 0.03, p < 0.1) and FCF-AC × CG (ROA-FE: β = 0.009, p < 0.01ta; GMM: β = 0.007, p < 0.1 and EPS-FE: β = 0.002, p < 0.05; GMM: β = 0.007, p < 0.1). These results indicate that the objectives of the principles and agents are aligned if the firms have the adopted good corporate governance practices reported by Jensen and Meckling (1976). Looking at the control variables, the size of the firm has a positive effect on the firm performance in all cases, which is in agreement with what we found in the literature. Leverage is negatively related to the performance, as can be seen by the signs of the negative coefficients in the table. Based on these results, we can accept our alternative hypothesis that corporate governance positively moderates the agency-performance relationship. Table 7 indicates the effect of ownership concentration measured by the shareholding percentage of the top shareholder on the relationship between agency cost and firm performance. The results show a positive impact of ownership concentration on the ROA and EPS for both the fixed-effect model and GMM. These results are in line with those of Shleifer and Vishny (1986) and Li et al. (2008). We introduced two interaction terms to measure the moderating effect of ownership concentration on the relationship between firm performance and agency cost. As shown in the results, the interactions terms possess a positive coefficient value, with ROA (TOP_1 × AC1 β = 0.002, p < 0.01 and TOP_1 × FCF-AC β = 0.0014, p < 0.1) and EPS (TOP_1 × AC1 β = 0.002 and TOP_1 × AC2 β = 0.006) in the fixed effect model. Similarly, the GMM model also exhibits a positive moderating effect of ownership concentration on the relationship between agency cost and firm performance. At higher levels of ownership concentration (for the largest shareholder), the shareholders are highly vigilant, and this helps facilitate the alignment of interests among the agents and the principles. The results show that as the ownership concentration is increased, the firm performance increases on the one hand, while the agency cost is decreased on the other side. These results support the second hypothesis that the ownership concentration positively moderates the relationship between agency cost and firm performance.

Moderating Effects of Ownership Types
The effect of ownership type on the relationship between agency cost and firm performance is reported in Table 8 State ownership is taken as a primary independent variable. A firm is described as state-owned if more than 50 percent of shares are held by the government and its affiliates. The results of the table report the adverse effect of state ownership on firm performance for both the fixed-effect model and GMM. These two sets of results are in line with the studies conducted in the Chinese context by Wei and Varela (2003) and Wei et al. (2005). According to these authors, the agency cost in state enterprises is higher, and this negatively affects the firm value. To test our third hypothesis, we introduced two interaction terms for agency costs with the state ownership variable to measure the effect on firm performance. All the interaction terms (SO × AC1, SO × FCF-AC) report negative coefficients, showing that state enterprises are significantly motivated by political motives; therefore, they focus less on performance. The effect of non-state ownership on the relationship between agency cost and firm performance is shown in Table 9. Non-state ownership has a positive impact on both performance measures, as depicted in the fixed-effect model and GMM. We introduced two interaction terms to specifically explore the effect of non-state ownership on agency-performance relationships. The interaction terms show a positive and significant moderating effect on firm performance in the fixed effect model as well as GMM. Looking at the ROA model, NSO × AC1 and NSO × AC2 have coefficients of 0.0381 and 0.0027 in the fixed-effect model, and of 0.068 and 0.0026 in GMM. The EPS model also shows positive coefficients with both fixed-effect models (FE: β = 0.019, p < 0.01; GMM: β = 0.005, p < 0.01) and GMM (FE: β = 0.597, p < 0.05; GMM: β = 0.028, p < 0.1). These results are in alignment with the study conducted by Ding et al. (2007), who found better performance by non-state enterprises as compared to state-owned ones. Based on the above analysis, we support our third hypothesis, that state ownership has an adverse effect on the agency-performance relationship. In contrast, non-state ownership positively moderates the agency-performance relationship.

Alternative Measures of Agency Costs
Tables 10 and 11 display the alternative measures of agency cost incorporated into our analysis. We have taken three variables, namely, earnings management using the absolute value of discretionary accruals as a proxy denoted by EM; research and development expenditures, denoted by R&D-AC; and the first principle component, obtained through principal component analysis using four agency cost proxies used in this paper, denoted by PC-AC. The results show the positive moderating effect of alternative measures of agency cost and corporate governance (R&D-AC × CGQ, EM × CGQ, and PCA-AC × CGQ) on ROA and EPS for both the fixed-effect model and the system GMM approach. These results again strengthen our hypothesis that corporate governance positively moderates the agency-performance relationship.
Tables 10 and 11 also explore the moderating effect of ownership concentration on the agencyperformance relationship. Again, from the results, we can accept our alternative hypothesis that the ownership concentration has a positive influence on the agency-performance relationship. Ownership structure has a negative moderating effect on the agency-performance association when the firms are state-owned. On the contrary, we witness a positive moderating effect when the ownership rests in the hand of non-state entities. These results support our third hypothesis that SOE has a negative impact on the agency-performance relationship, while NSOE has a positive moderating effect.

Summary and Conclusions
Emerging markets with weak investor protection have much-execrated agency problems as compared to developed markets. The sources and types of agency cost differ in emerging economies. The purpose of this study was to investigate the effect of corporate governance quality and ownership concentration on the relationship between agency cost and firm performance. A-share listed firms in China were taken as a sample. Both the fixed-effect model and dynamic panel generalized method of moment estimation were employed in order to cater for the unobserved endogeneity problem. The results show that agency cost is negatively related to firm performance, while corporate governance and ownership concentration enhance firm performance. When corporate governance and ownership concentration were taken as moderating variables, we found a positive impact on the agency-performance relationship.
We also studied the effect of ownership type on the association between agency cost and firm performance. Non-state ownership positively moderated the relationship between agency cost and firm performance, while the agency cost kept its negative sign when the state ownership was taken as a moderating variable. Alternative measures of independent variables were also considered for the robustness of our results, such as the absolute value of discretionary accruals, denoted as (EM), research and development expenditures (R&D-AC), and first principle component (PCA-AC) generated through the principal component analysis of agency cost.
This study adds to the literature on corporate governance, specifically with respect to emerging economies. China is one of the largest emerging economies, and possesses a unique corporate governance system. Most of the companies are state owned, and are characterized by political influence and corporate expropriation. By incorporating effective corporate governance mechanisms, Chinese listed firms can enhance their financial performance. The theoretical evidence on ownership structures postulates that concentrated ownership can help firms to reduce agency problems. The results of this study show that concentrated ownership aligns the interests of managers and shareholders, hence increasing the overall performance of firms. Our study divides the ownership structure of Chinese listed firms into state and non-state. Studies on emerging economies have shown that non-state firms can effectively curtail agency costs (Ding et al. 2007). The results of this study show that the performance of Chinese listed firms is enhanced when they are owned by non-state entities. State ownership has a negative impact on the agency-performance relationship.
This study supports the literature in concluding that agency cost and firm performance are negatively related in Chinese listed firms. Investors should keep in mind the proxies of agency cost when choosing a specific stock. Secondly, the abuse of managerial appropriation is higher in state-held firms compared to non-state firms. Policymakers can use these results to devise the investor protection rules so that managerial appropriation can be minimized.
In summary, our results support all of our hypotheses, indicating a positive moderating effect of corporate governance quality (H1) and ownership concentration (H2) on the relationship between agency cost and firm performance. Additionally, non-state (state) ownership of companies positively (negatively) moderates the agency-performance relationship (H3).
Our study does have certain limitations. First, in constructing the corporate governance index, we tried to take account of all of the agency-mitigating governance variables. However, due to data unavailability, we dropped several governance variables, such as CEO compensation (instead, we used top three compensation), stock options, independent director dissent report (although this data is present on CSMAR, it lacks the data of financial statements), and audit committee (CSMAR contains the total committee data, but does not further elaborate), to name several.