Do Private Beneﬁts of Control Affect Corporate Social Responsibility? Evidence from China

: In this study, we examined whether private beneﬁts of control can inﬂuence corporate social responsibility performance. We used both separations between cash ﬂow and control rights and the length of the longest control chain to measure private beneﬁts of control. Consistent with the private beneﬁts motive, we found that ﬁrms with greater divergence between cash-ﬂow rights and control rights, with longer control chains, are associated with lower corporate social responsibility performance. Further, we found that earnings management and capital occupation by the controlling shareholder are the two effective channels through which private beneﬁts of control affect corporate social responsibility. Additionally, this negative association is more pronounced for ﬁrms located in regions with low degree of law environment and with CEOs appointed by the largest shareholder. Additional robustness tests using alternative CSR measurements, and two-stage least squares (2SLS) regression support the main ﬁndings. This study highlights a new determination channel of private beneﬁts of control and practically guides the introduction of corporate social responsibility activities in emerging markets.


Introduction
What drives a firm's corporate social responsibility (CSR) is a fundamental question in corporate governance and sustainable development. The units of enterprises have gradually become the primary carrier of social responsibility. Corporate social responsibility, involving firms' obligation to shareholders, environment and society, is also a crucial part of sustainable development. As a growing interest to market participants, it is becoming increasingly important to understand the forces that drive managers to implement CSR practices for sustainability. Previous studies on the determinations in influencing firms' CSR have focused on the effects of corporate culture and managerial characteristic [1][2][3][4]. At the same time, some scholars also conducted research on CSR performance from the perspectives of ownership structure and governance quality [5][6][7]. However, prior studies tend to ignore the possible impact of type II agency problem caused by the private motives of the ultimate controlling shareholders on CSR performance. Thus, we undertook such a study to investigate whether and how the private motives of controlling shareholders affect corporate social responsibility activities and performance in emerging market.
The rest of the paper proceeds as follows. Section 2 provides literature review on CSR and private benefits of control, and then develops our hypotheses. Section 3 discusses the model specification and sample selection. Section 4 presents empirical results and discusses the findings. Section 5 discusses and concludes the study.

Private Benefits of Control
According to studies done by Silanes et al. [13], ownership has been relatively concentrated in most countries of the world. Once ownership is concentrated in the hands of certain shareholders, to maximize their own interests, the ultimate shareholder may seek private benefits at the expense of the interests of the minority shareholders, which creates the problem of private benefits of control [14]. Thus, much research related to private benefits of control has been done. Earlier research primarily focused on the measurement of private benefits of control [15][16][17] and factors influencing the level of private benefits of control [18,19]. Recent studies focus on the impact of private benefits of control on firm decision and performance.
Empirical research shows that there is a strong link between private benefits of control and firm decision. Kang and Kim [20] showed that, when controlling shareholders have higher voting rights than their cash flow rights, they have more incentives to reduce firms' investments that are essential for business success in long-term strategic perspectives, to exploit minority shareholders' rights. Baldenius [21] studied the relationship between private benefits of control and investment decisions made by shareholder agents, noting that the controlling shareholders have higher expectation for returns on retained earnings than on conventional capital budget. Filatotchev et al. [22] found that private benefits of control is an important factor in controlling shareholders' external financing, especially when their shareholding ratio is close to absolute control ratio (50%). Therefore, it is very likely that the controlling shareholder will abandon the external financing opportunity to maintain its control position, resulting in the rejection of some investment projects with net cash flow. Beyond financial decision, Gopalan and Jayaraman [23] found that there is a strong positive link between private benefits of control and earnings management. Doidge et al. [24] noted that, considering the constraints on the consumption of private benefits, there is lower possibility for controlling shareholders with high private benefits to choose to cross-list in the United States.
Besides, there is also major research exploring the relationship between private benefits of control and firm performance. Laeven and Levine [25] found that the separation between controlling shareholder's cash-flow rights and control rights leads to a lower valuation of the corporate. Lins [26] also confirmed this relationship and further research shows that this link will be enhanced if shareholder protection is low. Using a total of 246 M&A events from Chinese state-controlled listed companies as the sample, Li Feng and Cao [27] concluded that private benefits of control correlated with acquirer announcement returns in a positive way if government shareholding is low, while no correlation was find if the government shareholding is high. Filatotchev and Mickiewicz [28] showed that high level of private benefits of control may result in lower efficiency in the use of financial resources, which is measured by the ratio of debt to investment.
To the best of our knowledge, although the impact of private benefits of control is an important research subject in the literature, its impact on firms' corporate social responsibility decision and performance remains a relatively under-researched area.

Corporate Social Responsibility
Thus far, most studies on CSR mainly concentrate on two questions: (1) In what ways and to what extent does CSR influences corporate performance? (2) By what factors and how is CSR performance influenced? To answer the first question, plenty of studies have been done [29][30][31][32][33][34]. Although some scholars [35,36] argue that the link between CSR ratings and firm performance is negative or at least neutral, according to Beurden and Gössling [31], most of the empirical findings [29,30,32,34] support that there is a positive association between CSR ratings and firm performance.
At the same time, there are also a large amount of studies trying to answer the second question [37]. Based on the analysis of these studies, we found that all these factors can be generally divided into external factors and internal factors. Among the external factors, culture environment, regulation system and industry characteristics are the three most crucial factors. For instance, Thanetsunthorn [38] showed that national culture crucially affects CSR performance both positively and negatively, depending on the dimension of CSR and the specific culture. Calveras et al. [39] noted that market regulation also plays an important in firms' CSR rating. Using 645 European companies as the sample, Gainet [40] conclude that differences of legal systems, part of the regulation system, among different countries have a significant impact on CSR performance. As for industry characteristics, Longoni et al. [41] noted that industry characteristics such as industry competition, industry uncertainty, and industry munificence are significantly associated with CSR performance. Yang, Ye and Zhu [42] empirically demonstrated that the CSR gap between firms and peer firms motivates firms perform better in CSR activities.
Financial performance, manager characteristics and firm culture are three important internal factors that should be mentioned. Financial performance has long been considered correlated with CSR performance. Based on the theory that slack resource availability and CSR are positively related, Waddock and Graves [43] found that firms with good financial performance perform better in CSR activity. From the perspective of legitimacy, Cho and Patten [44] pointed out that better-performing firms tend to pay more efforts in CSR activities to avoid regulations. In the perspective of manager characteristics, Manner [45] explored the impact of CEO characteristics on CSR performance comprehensively, finding that CEO's education background, career experience, gender and compensation have significant effects on CSR performance. Specifically, Myung, Choi and Kim [46] argued that there is a negative correlation between CEOs' negative traits and corporate social responsibility. Firm culture is also an important factor influencing CSR performance [47]. Galbreath [48] found empirical evidence that a humanistic culture can positively impact CSR performance. Beyond these studies, Chun and Shin [49] pointed out that the existence of labor unions has a negative relationship with firms' CSR activity. Zhang et al. [50] noted that higher presence of outside directors and women directors is linked to better CSR performance. Oh et al. [51] conducted a comprehensive research on the effect of ownership structure on CSR, which draws a conclusion that CSR ratings are positively related to ownership by institutions and foreign investors and negatively related to ownership by top managers.
As mentioned above, many studies explore factors that have links with CSR performance. However, as far as we know, there has been no research working on the relationship between the private motives of controlling shareholders, which is an important part of governance structure, and CSR performance. Therefore, it is necessary to conduct a study on the correlation between private benefit of control and CSR performance.

Private Benefits of Control and CSR
In modern enterprises, with the separation of ownership and management rights, shareholders and managers have a huge impact on the formulation and implementation of corporate strategies, jointly determining the decision-making of the enterprise [52]. In that situation, agency problems will appear among shareholders and management [53,54]. Recently, La Porta et al. [55] found that, with the high concentration of equity becoming a prominent feature of modern companies, the framework of theoretical analysis for studies on corporate governance has gone through major changes. The focus of studies has been gradually changing from the conflict between shareholders and managers under the situation of separate equity structure to the conflict between the controlling shareholder and the minority shareholders under the situation of concentrated equity structure. According to early studies on corporate governance, some categories of firm structures or ownership structures, where controlling shareholders can somehow obtain benefits more than just currency income, will possibly lead to conflicts between controlling shareholder and other shareholders [56,57].
Compared with listed companies from mature markets, Chinese listed firms generally have the characteristics that ownership structure is concentrated and the liquidity of shares is poor, leading to a phenomenon that most firms have controlling shareholders, whose impact on firm decision-making may exceed that of managers [9]. Bebchuk, Kraakman, and Triantis [58] found that agency costs of controlling shareholders are higher when there is a separation of control rights and cash flow rights. That is, firms with strong private benefits of control have higher agency cost between majority shareholder and minority shareholders. Besides, most research argues that controlling shareholders have a strong motivation to pursue private benefits of control by transferring, or tunneling, resources out of the company [59]. The issue of private benefits has gradually become the focus of studies on corporate governance.
Controlling shareholders are convinced to meddle in firm decisions frequently, including decisions on CSR activities, with a motivation of pursing private benefits of control. Starting from considering the moral motives of the controlling shareholders, private benefits of control is obtained by the ultimate shareholders' constantly snatching benefits from companies, minority shareholders and other stakeholders to maximize their own [5]. According to early research, it is controlling shareholders' desire of occupation that determines the degree of controlling shareholders' private benefits of control [59]. Therefore, a higher level of the private benefits of control demonstrates that the controlling shareholder concentrates less on protecting other stakeholders' interests, contrary to the morality which motivates the CSR activities. Besides, decisions on strategic investments, such as investments on CSR programs, is subject to the conflicts between controlling shareholders and minority shareholders. To purse more private benefits of control, the controlling shareholders tend to show less support for the managers to invest in projects with unclear returns, and hope that the investment can pay back in a short term [60]. However, the future value of CSR projects is highly uncertain and less likely to be apparent in the short term [61], which is obviously contrary to the controlling shareholder's motive for seeking private benefit of control.
Additionally, as discussed in the prior section, higher level of private benefits of control is likely to result in lower efficiency in the using of financial resources [28] and induce a lack of resources to invest in long-term soft assets [19]. Thus, it is possible to expect that investment decisions on firm's soft assets can be abandoned or easily postponed, even though it is essential for firm's long-term business success. In this kind of situation, CSR programs, which are absolutely not part of the core firm decisions, are very likely to be ignored.
However, the extent and the exact power of controlling shareholder's motivation to purse private benefits of control depends on the complexity of the structure of corporate governance and the availability of private benefits. First, the separation between cash flow and control rights may strengthen the incentives of controlling shareholders' motivation of pursing private benefits of control. Gopalan and Jayaraman [62] examined the private control benefits from the perspective of insider control, finding that the insider-controlled enterprises have higher earnings management levels in countries having weaker investor protection, and the greater the separation of cash flow and control rights, the higher the level of earnings management. Doidge et al. [24] found that, when private benefits are high, controlling shareholders are less likely to choose to cross-list in the United States to avoid constraints on private benefits resulting from such listings. All other conditions remaining equal, the separation of cash flow and control rights have positive link with the benefits that the controlling shareholder invades from other minority shareholders. Therefore, when the controlling shareholder's cash flow and control rights are quite separated, controlling shareholders' motivation of pursing private benefits of control will be strengthened. Second, as one of the basic characteristics of the control chain, the length of the ultimate controller's control chain of the company will induce serious agency problems [62] and thus affect the controlling shareholder's decision-making. The longer the control chain is, the more complex the corporate governance structure is. The opaque organizational hierarchy will greatly enhance the external supervision difficulty, strengthening the power of controlling shareholders' ability to purse private benefits of control [63]. In firms with longer chains, information needs to go through more intermediate-level companies to get to the top of the pyramid structure. However, managers of intermediate-level companies may not pass information objectively in time [12]. Meanwhile, longer control chains make it difficult for the ultimate controller to obtain firm information sufficiently and immediately, which makes it difficult for ultimate controller to supervise and control firm's operation [64].
As such, our first hypothesis is as follows: Firms with greater divergence between cash-flow rights and control rights, with longer control chains, perform worse in corporate social responsibility.

Two Possible Channels
Controlling shareholders' capital occupation is one of the most common tunneling actions they will do to purse their private benefits of control [11], which will obviously lead to a poor financial performance and resource constraints of firms. As one of the activities that do not belong to the core programs of firm operation, investments on CSR programs have been proven to depend largely on the slack resource of firms [43]. It is very possibly that supply constraints caused by controlling shareholders' capital occupation will restrain firms' resources putting into CSR programs. Besides, as discussed above, for the reason that controlling shareholders' have motivations to refuse investments on CSR programs, this kind of resource constraints may also give controlling shareholders' a good reason to prevent managers from supporting CSR programs.
Meanwhile, the controlling shareholders will also engage in earnings management to conceal or remedy poor financial performance caused by their tunneling behaviors [65]. Further, empirical evidence shows that earnings management will cause larger scale of controlling shareholders' tunneling behaviors [66] and higher level of earnings management is possibly correlated with less available resource to promote firm performance [67], meaning that earnings management may enhance tunneling behaviors and resource constraints, which may strengthen the impact of private benefits of control on CSR performance. Therefore, we assume that the interest grabbing behavior of the controlling shareholders, such as capital occupation and earnings management, will lead to the resource constraints against CSR activities, and then affect the CSR performance.
As such, our second hypothesis is as follows: Under private motives, firms with greater capital occupation or earnings management perform worse in corporate social responsibility.

Model Specification and Variable Definition
To investigate effects of private benefit of control on corporate CSR performance, we estimated the following ordinary least squares (OLS) baseline regressions: The dependent variable is CSR, including two sub-proxies: CSR_Score and CSR_Level. CSR_Score is a total weighted CSR score of five primary category measures of CSR scores which ranges from 0 to 100. (These CSR index systems of Hexun Net consist of 5 first-class indexes, 13 second-class indexes and 37 third-class indexes. The five first-class indexes involve the aspects of: (1) shareholders; (2) employees; (3) customers, community and suppliers; (4) environment; and (5) social responsibility. CSR_Score = 30% × CSR_1 + 15% × CSR_2 + 15% × CSR_3 + 20% × CSR_4 + 20% × CSR_5.) To keep readability, we used the score deflated by 100. CSR_Level is an indicator variable which ranges from 5 to 1. It takes the value of 5 if the CSR_Score ranks in the top fifth (score from 80 to 100) and 1 if in the bottom fifth (score from 0 to 20), and so forth. Greater CSR_Score or CSR_Level indicates better corporate social responsibility performance. We used both the separation between cash flow rights and control rights (PB1) and length of the longest control chains (PB2) to measure private benefits of control (detailed information in Appendix B). A larger PB1 or PB2 suggests a greater private benefit of control. ControlVaribles is a vector of firm-level controls. Following previous studies, we controlled the following variables in our regressions: firm size (SIZE), leverage (LEV), returns on asset (ROA), listing age (AGE) and sales growth (GROWTH), among which i indexes firms, and t indexes years. All the regressions include two-digit standard industrial classification (SIC) industry fixed effects as well as year fixed effects to control for differences across industries as well as time trends in the outcome variables. We clustered the standard errors at the firm level. The detailed definitions can be seen in Appendix A (Table A1). The coefficient of interest in our regressions is β 1 , which captures the effects of private benefits of control on CSR performance. According to the inferences above, we predicted β 1 is negative.

Sample Selection and Data Sources
The initial sample included all A-share listed firms on the Shanghai and Shenzhen Stock Exchanges over the 2010-2016 periods. We obtained financial and market information data from the China Stock Market and Accounting Research database (CSMAR provided by GTA which is a leading giant in financial information services and professional data services in China). These data have been used in several prior studies of Chinese securities markets [68]. We obtained the CSR data from HeXun Net (Hexun net is the leader of Chinese professional financial network, and thus these CSR data have excellent reliability and practicability (Website: http://www.hexun.com/)) social responsibility score of listed firms which is an annual professional evaluation report. We manually collected and identified the control chains structured diagrams of Chinese listed firms to calculate the measures of private benefits of control. To ensure the validity of the data collected and minimize the effect of other factors on the research results, we excluded from our initial sample those companies which have extreme outliers, belong to financial industry, are obviously misreported, and those industries whose observations is seriously inadequate (less than 30). The final sample consisted of 14,064 firm-year observations. Table 1 provides summary statistics of the main variables used in following analyses. The mean (median) value of CSR_Score is 0.267 (0.220) and that of CSR_Level is 2.262 (2), which manifests that the overall level of corporate social responsibility remains low in China. The mean value of PB1 is 0.054. The mean value of PB2 is 0.775. The descriptive statistics of other control variables are similar to prior paper (Hao et al., 2018). The firm characteristics variables are Winsorized at the 1st and 99th percentiles throughout the analysis.  Table 2 shows the correlation analysis results. The CSR indicators are significantly negatively related to PB1 and PB2. This result preliminarily indicates that firms with higher degree of separation between cash flow right and control rights, i.e., longer length of control chains, have lower CSR performance. From the Pearson correlation coefficients, the explanation of the variables in the model and the correlation coefficient between the control variables are less than 0.5 which proves that there is no multicollinearity between the main variables and the model selection of variables is reasonable. The following regression results have reliability and validity.

Empirical Results
First, we ran our baseline regression to test the relationship between private benefits of control and CSR performance, which is also the main context of Hypothesis 1. Table 3 generally shows the results of our baseline regression. In Columns 1 and 2, where CSR performance is measured in the format of CSR-Score, the coefficient on PB1 is negative and significant at the 5% level (t-value = −2.36) and the coefficient on PB2 is significantly positive (t-value = −4.04). In the other two columns, where CSR performance is measured in the format of CSR-Level, the empirical results are nearly the same. Coefficients on PB1 and PB2 are both negative and significant at the 1% level (t-value = −2.58 and −4.48, respectively). These results suggest that private benefits of control have a statistically and significantly negative correlation with CSR performance. Specifically, firms with greater separation of cash flow rights and control rights tend to perform worse in CSR activities, as do firms with longer control chains. In sum, the empirical results of our baseline regression strongly support Hypothesis 1.

Channel Analyses
To further analyze the possible channels in which private benefits of control could influence CSR performance, we conducted further empirical tests based on our theoretical analysis about the two possible channels above.
In these models, functioning as channel analysis, samples are divided into two groups. In detail, we first computed the industry-year mean of CSR-Score based on the data we collected. Then, controlling year and industry, we divided scores of CSR performance into two levels by the industry-year mean of CSR-Score. Firms with scores above the mean value were defined as high CSR firms, otherwise, defined as low CSR firms. After that, we separately ran regressions of PB1 and PB2 on the two possible channels, i.e., earnings management and capital occupation, in different CSR groups. Table 4 shows the regressions results for earnings management. For the measurement of earnings management, it is worth mentioning that we used a modified Jones model to calculate discretionary accruals (EM) as our proxy for earnings management following the research done by Bartov et al. [69]. In Columns 1 and 2, where the samples are firms with high CSR, t-value of coefficients on PB1 and PB2 equals 0.31 and −1.01, respectively, meaning there is no significant correlation between private benefits of control and earnings management. However, in Columns 3 and 4, where samples are firms with low CSR, coefficient on PB1 is 0.0114 with a t-value of 1.99 and coefficient on PB2 is 0.0032 with a t-value of 2.45. Results in these two columns illustrate that private benefits of control are positively related to earnings management when firms have low corporate social responsibility. Comparing regression results in high CSR samples and low CSR samples, we can draw the conclusion that earnings management indeed plays an important mediating role in the relationship between private benefits of control and CSR performance. Notes: The values of the t statistics are in parentheses. **, and *** indicate significance levels of 5%, and 1%, respectively.
As discussed above, capital occupation may be another functional channel of private benefits of control to influence CSR performance. To test the channel effect of capital occupation, we regressed the measurement of capital occupation on private benefits of control. In this model, following methods used by Wang and Xiao [10], Liang and Chen [70], we used the proportion of other receivables to the total assets of the listed companies as the measurement of capital occupation (Tunnel). Table 5 reports the regression results. Consistent with the regression results in Table 4, private benefits of control show no significant correlation with capital occupation in samples with high corporate social responsibility. A significant positive correlation is found in samples with low corporate social responsibility: coefficient on PB1 has a t-value of 2.40 and the t-value of PB2 is 2.04. These results illustrate the notion that capital occupation is one of the effective channels through which private benefits of control influence on CSR performance. Notes: The values of the t statistics are in parentheses. **, and *** indicate significance levels of 5%, and 1%, respectively.
Combing the two results above, we confirm that the interest grabbing behavior of the controlling shareholders, such as capital occupation and earnings management, will affect the CSR performance in a negative way.

Cross-Sectional Analyses
Empirical results above illustrate the fact that private benefits of control have a significant negative impact on CSR performance. Besides, the results in Tables 4 and 5 also show that earnings management and capital occupation are two important channels. In this section, we discuss the influence of legal environment and manager appointed by controlling shareholder on the relationship between private benefits of control and CSR performance. Interactions are introduced into the regressions to test the moderating effects of law environment and manager appointed by controlling shareholder on the relationship between private benefits of control and CSR performance. The analysis and results are shown as follows.
Many cross-country studies note that external environment has significant influence on corporate behavior [71]. Legal environment is also long believed to be one of the most important environment which have impact on both CSR performance [40] and private benefits of control [13]. From the view of Silanes et al. [13], the legal system is considered to be the most important mechanism to reduce controlling shareholders' private benefits of control. A well-developed legal system can obviously lead to a lower efficiency of tunneling behaviors of controlling shareholders [5]. Besides, early research also shows that firms in regions with weak legal system tend to invest less on CSR programs [63]. Therefore, we argue that the legal environment in different regions can also have a critical effect on the relationship between private benefits of control and CSR.
To measure the legal environment of different regions, we constructed an indicator variable, Law, for regions with lower level of legal environment. Specifically, we first calculated the legal index for each province-year based on the data provided by Fan, Wang and Yu [72], and then constructed an indicator variable for the province-years with law index. Law equals 1 if the legal index of a province-year is lower than the median of all provinces for the year, and 0 otherwise. In Table 6, coefficients of PB1 and PB2 are all significantly negative, which is consistent with our results before, showing that firms with greater separation of cash flow rights and control rights and longer control chains have worse performance in CSR activities. Besides, coefficients on Law itself show no significance in all four regressions, which means that the environment of law itself has no significant correlation with CSR performance. However, more attention should be paid on the coefficients on these interactions of PB1 (PB2) and Law, which are all significantly negative, indicating that firms with greater separation of cash flow rights and control rights and longer control chains have even worse performance of CSR in regions with weak legal systems compared to regions with good law system. In other words, a weak legal system will enhance the negative impact of private benefits of control.
Meanwhile, manager is also an unavoidable factor when we talk about firm decisions and performance [73]. Focusing on the relationship between private benefits of control and CSR performance, we assumed that the manager being appointed by the largest shareholder makes a difference. Prior studies show that there is a positive relationship between the existence of managers appointed by controlling shareholders and private benefits of control [74]. Through managers appointed by the controlling shareholders, controlling shareholders can exert bigger influence on firm decisions [75]. Thus, it is very likely that the impact of controlling shareholders' private benefits of control on CSR will be enhanced by managers appointed by the largest shareholder. Regressions in Table 7 aimed at testing whether manager appointed by largest shareholder (simplified as UM) will enhance the impact of private benefits of control on CSR performance. The results in Table 7 are similar to results in Table 6. Coefficients on PB1 × UM in Regressions 1 and 3 are both negative and significant at the 5% level, showing that firms with greater separation of cash flow and control rights perform worse have lower social responsibility when managers are appointed by the largest shareholder. Coefficients on PB2 × UM are also significantly negative, meaning that firms with long control chains also have worse performance in CSR activities when managers are appointed by the largest shareholder.

Sensitive Tests
Serving as a sensitive test, we controlled for the possible endogeneity of the relationship between private benefits of control and CSR by using 2SLS regressions. The OLS regression might have potential endogeneity problems by reverse causality, which might affect the effects of private benefits of control on CSR. According to the prior literature, we used the mean PB1 and PB2 of industry firms as the instrument variable in the following 2SLS analysis. Table 8 reports the results of 2SLS regression. Columns 1 and 3 report the first-stage regression results relating to PB1 as well as PB2 with MeanPB1 and MeanPB2. As expected, the coefficient on MeanPB1(2) is significantly positive associated with PB1(2) at the 1% level. Columns 2 and 4 report the coefficients from the second-stage regression. For brevity, we used the CSR_Score as dependent variables. We still found significant coefficients on the fitted value of private benefits of control. This indicates controlling shareholders do harm to CSR activity performance when private benefits of control are stronger even after controlling for the endogeneity. To reduce the bias caused by data collecting, we further used the score system of Rankins CSR Ratings as the data source for the computing of CSR_Score and CSR_Level. Ranking global (RKS) is another authoritative third-party rating agency for corporate social responsibility in China. It is committed to providing objective and scientific corporate responsibility rating information to SRI investors, responsible consumers and the public. The responsibility rating of RKS includes ESG rating (environment, social responsibility and governance), CSR report rating, social responsibility investor service, etc. Based on this, the baseline regressions of our paper are operated again. As reported in Table 9, the empirical results are the same as found above, showing that our regression analyses are not biased by the data source used. Notes: The values of the t statistics are in parentheses. **, and *** indicate significance levels of 5%, and 1%, respectively.

Discussion and Conclusions
This paper examines whether private benefits of control can influence corporate social responsibility performance. We hypothesized that private of control can negatively affect the social responsibility performance of a firm. Using the sample of Chinese listed firms from 2010 to 2016, we first proved that firms with greater divergence between cash-flow rights and control rights, i.e., with longer control chains, are associated with lower corporate social responsibility performance. Second, we found that earnings management and capital occupation induced by the controlling shareholder are the two effective channels through which private benefit of control affect corporate social responsibility. Further, we examined the effect of private benefits of control in relation to different region and manager characteristic and the results suggest that the negative association is more pronounced for firms located in regions with low degree of law environment and with CEOs appointed by the largest shareholder. Additional robustness tests using alternative CSR measurements, and two-stage least squares (2SLS) regression support the main findings.
This study also has policy and managerial implications. First, our results may compel minority shareholders and other investors to monitor the tunneling behavior of controlling shareholders against sustainability. Second, we urge managers to make decisions on CSR activities more independently, avoiding the impact from controlling shareholders' motive of private benefits of control. Third, the cross-sectional test of our research shows that legal environment has a significant impact on firms' CSR performance. Therefore, deepening the reform of legal environment in China might be an efficient way to improve firms' CSR performance. Meanwhile, regulation should pay more attention to firms with complicated ownership structure in the term of CSR. The high level of CSR performance of listed firms is crucial for sustainable development of capital market.
We acknowledge that this study has some limitations. Although adopting the method of 2SLS, the main results are not completely free of endogeneity concerns for other measurement errors or sample selection bias. In addition, we could only investigate the effects among listed firms due to data availability. Future studies could expand to unlisted firms or unique CSR database. Notwithstanding, we believe that our findings highlight a new determination channel of private benefits of control and practically guide the introduction of corporate social responsibility activities in emerging markets.
Author Contributions: Q.L. contributed to study design, data collection, and drafting and revising the manuscript. Y.S. conceptualized the research design. G.G. contributed to writing and interpretation of the findings. C.N. and X.T. reviewed the manuscript, and collected and sorted some of the data. All authors read, revised and approved the final manuscript.

CSR_Score =
The comprehensive score of five first-class CSR indexes calculated by the weighted average method, range from 0 to 100; To keep readability, we use the score deflated by 100. Year fixed effects = Indicator variables for different years; i,t = Firm i, year t subscripts.