The Nonlinear Impact of Corporate Social Responsibility on Stock Returns

“Can companies benefit from corporate social responsibility?” is still un-solved. This study investigates how corporate social responsibility impacts stock returns with Hexun’s CSR rating for first time. The sample is based on Chinese A-share listed companies for the period 2010-2017. We adopt threshold regression and multiple regression and find that: first, there is a U-shaped relationship between CSR and stock returns. This finding suggests that the relationship between CSR and stock return is affected by cut-off point, which is negative before cut-off point, and positive after cut-off point. In addition, this relationship has nothing to do with the nature of enterprise property rights. Second, fulfilling CSR affects corporate reputation and re-lieves information asymmetry, corporate reputation and institutional investors are the channels through which CSR influences stock returns. Fourth, we demonstrate the moderating effect of external legal environment between CSR and stock return. This paper provides new evidence for understanding the impact of CSR on enterprise value.

Theoretical Economics Letters cial responsibilities (CSR). CSR is a practical activity that enterprises voluntarily and rationally allocate their resources in order to improve social welfare and enhance the relationship between enterprises and core stakeholders. CSR is the cornerstone of firms' sustainable development, and it is needed to adapt to social development and economic globalization. It is a kind of intangible assets of firms. Among the 250 largest enterprises in the world, the number of enterprises that publish CSR reports has increased from 44% in 2011 to 78% in 2017. The survey of 4900 firms also shows that the number of firms that publish CSR reports has increased from 30% to 81% from 2015 to 2017 (KPMG) [1]. CSR is so important. Why do some enterprises ignore it? There has always been a dispute about it. Specifically, can CSR make firms benefit from it? So the purpose of this paper is to study how CSR performance impacts stock returns. We aim to address the following questions. Does CSR increase stock return? Is the effect of CSR on stock return different across firms? How CSR affects stock returns?
While a lot of studies investigate the CSR and firm performance relationship and offer valuable insights (Luo and Bhattacharya; Peloza and Shang; Servaes and Tamayo; Mishra and Modi; Lenz et al.) [2] [3] [4] [5] [6]. While most of them have been conducted in developed economies, they have not reached a consistent conclusion at present. One view is that CSR is conducive to optimize corporate image, reduce financing costs, improve financial performance and competitiveness, and contribute to the sustainable development of enterprises, for example, Lenz et al. [6] find CSR activities enhance firm value. Another view is that the resources of an enterprise are limited; the only responsibility of an enterprise is to maximize the interests of shareholders, to make profits for shareholders by using limited resources, and to fulfill CSR will consume the resources, reduce profits and damage financial performance, for example, Margolis and Walsh [7] find that CSR activities may damage firm value. There is still gap about CSR research in developed and developing countries. There are two reasons for choosing China Securities Market as an example to study CSR. First, more and more Chinese enterprises have begun to realize the importance of CSR and have launched an upsurge of active involvement in CSR. The total number of CSR reports was 24 from 2001 to 2005. After the publication of the Guidelines on Social Responsibility of Listed Companies in 2006, the number of social responsibility reports was 33 in 2006. By the end of 2018, the number of social responsibility reports increased to 2097, and 1779 of 2097 are by Chinese listed companies, which accounted for 84.8%. Second, with the increasing importance and influence of China in the global economy, the attention of international investors for Chinese stocks is also increasing, and the influence of Chinese stock market in the world is gradually increasing.
Compared with using financial index to measure firm performance or enterprise value, using accounting indicators to measure enterprise value has certain shortcomings, on the one hand, accounting indicators are difficult to reflect the long-term performance of firms, on the other hand, it is easy to be manipulated. Therefore, this paper chooses stock returns in capital market to measure firms' We find that there is a nonlinear relation between CSR and stock returns, specifically, there is a U shape between them. In a nutshell, stock returns tend to decline with the increase of CSR at first, when CSR exceeds a certain threshold, stock returns increase with the increase of CSR. At first, with the increase of CSR investment, firms are unable to recover the actual cash flow generated by CSR performance temporarily, not only stock returns not be stimulated by CSR but also may decline at first. However, CSR is a type of long-term investment, fulfilling CSR can provide competitiveness and profitability for firms, this positive effect of CSR will be highlighted after some time, investors also discover the growth potential of these companies and thus increase their stock purchases, which is conducive to improving stock returns, so we can find the positive effect of CSR on financial performance. Besides, there is no difference between CSR and stock return for state-owned enterprises and non-state-owned enterprises.
We take our findings one step further, we study the impact of different parts of CSR (based on first-level) on stock returns, we find that there is a U-shaped relationship between CSR of different groups and stock returns except employee responsibility, there is an inverted U-shaped relationship between employee responsibility and stock return. Why cannot the concern of firms for employees produce lasting promotion for the development of enterprises? This is a very interesting question. We will study it in future.
Furthermore, we explore intermediary mechanisms to explain the relationship between CSR and corporate performance. We find that corporate reputation and institutional investors' investment strategies are two potential channels for CSR performance influence stock returns. Specifically, CSR is conducive to improve corporate reputation, which will affect investors' attention to enterprise, thus increasing the purchase of stocks, which lead the rise of stock returns. Firms with high CSR scores send positive signals, institutional investors has abundant professionals and abundant funds, therefore, institutional investors respond to CSR information issued by firms in timely, institutional investors will adjust their investment strategies in timely according to information, so we find that institutional investors play a mediating role in the impact of CSR on stock returns. We also find the moderating role of external legal environment in the impact of CSR the understanding of CSR system interpretation and implementation is also greatly different from the expectations of relevant policymakers. The uncertainty of the external legal environment will lead to the uncertainty of CSR. We consider the relation between CSR and stock return in conjunction with external legal environment. This allows us to evaluate the role of a local's legal environment in relation to CSR, and additionally yields important implications for regulators.
The remainder of the paper is structured as follows. Section 2 presents a literature review and hypotheses development. Section 3 discusses our data and methodology. In Section 4, we analyze how CSR scores impact stock returns and the influence mechanism. Section 5 presents additional analyses and robustness test, and Section 6 concludes the paper.

Corporate Social Responsibility and Stock Return
There are two main lines in the current areas about CSR and stock returns. One is that CSR is positively correlated with stock returns. Mcguire and Schneeweis [8] prove that firms with low CSR have lower returns on stock than those with high CSR. Luo and Bhattacharya [2] measure corporate performance by stock returns and use the CSR index provided by FAMA database, and they find that CSR has a positive correlation with stock return. Hill et al. [9] point out that CSR can filter out high-quality enterprises, so the better CSR of company, the more favored by value investors, therefore, holding a portfolio of stocks with high CSR will benefit in the long run. Deng et al. [10] find that compared to low CSR acquirers, high CSR acquirers realize higher merger announcement returns. Lins et al. [11] find that firms with high CSR ratings outperform firms with low CSR ratings during the crisis. Another viewpoint is that CSR has nothing to do with stock returns or that CSR is not conducive to improving stock returns. Brammer et al. [12] find that CSR was significantly negatively correlated with stock returns, which can be rationalized by multi-factor models for explaining the cross-sectional variation in returns. Mackey et al. [13] argue that stock prices have nothing to do with CSR in a balanced market, because the effect of CSR on stock returns will be eliminated through arbitrage.
These studies show mixed results of CSR and stock return: positive and negative. Such questionable conclusions create a research gap, allowing for more research in an effort to acquire satisfactory results. In fact, the impact of CSR on stock returns is a gradual process, with a certain lag. CSR investment as a hidden investment will lead to higher costs in the short term but not necessarily get a return in the current period. That is, within a certain period with the increase of CSR investment, firms are unable to recover the actual cash flow generated by However, in the long term, firms' active performance of social responsibility is conducive to their sustainable development, thus helping improve the value of firms, which is beneficial to improve stock returns. Enterprises win the trust and support from stakeholders by undertaking CSR, which help to boost share price.
As Balabanis et al. [14] argue CSR is not conducive to corporate financial performance in the short term, but after a while, we can see the positive effect of CSR on financial performance. Servaes & Tamayo [4] argue that CSR promotes corporate performance only under certain conditions. Based on the above analysis, we raise hypothesis 1: H1: There is a nonlinear relation between CSR and stock returns.

The Mediating Role of Corporate Reputation
According to the theory of enterprise resources, corporate reputation is an important intangible asset, which is scarce, difficult to imitate and replace, and can create value for enterprises (Hall; James et al.) [15] [16]. Research has proved that CSR helps enhance corporate reputation. Specifically, Firms with better CSR tend to disclose more positive image to investors and other stakeholders, the disclosure of CSR-related information sends a signal of incomplete self-interest to stakeholders, which helps enterprises to form a good reputation for investors.
When such signals are continuously released and effectively received by external investors, firms can accumulate and form reputation capital. As time goes by, for companies that perform CSR better, investors are more likely to judge that these companies have a good social reputation and low operating risk, and make a higher expectation of the company's future earnings and stock value, and then have a strong willingness to invest in the company's stock, with the increase of purchase, it will promote the stock price, such as, Luo and Bhattacharya [2] argue that CSR increases customer loyalty, leading to firms having more pricing power, Ailawadi et al. [17] and Hilger et al. [18] find that firms to sell more or at higher prices those products that have CSR features.
In reality, a large number of investors cannot fully understand the social responsibility information disclosed by firms due to the limitation of their abilities, media supervision provides strong support for investors to understand firms (Miller [19]). On the one hand, the media actively publicize CSR activities to build a good reputation for the enterprise, thereby increasing customer loyalty and satisfaction to the firm. On the other hand, the exposure and tracing of irresponsible acts of firms by the media result in the loss of responsible reputation and deterred irresponsible acts of firms, such as tax evasion, manipulation of financial information, food safety risks (Luo et al.) [20], forming a "reputation responsibility." Therefore, investors can understand the corporate reputation

The Mediating Role of Institutional Investors
It has been proved that firms with better CSR will disclose more information, reducing the information asymmetry between firms and capital markets (Mohr  [25]. From the perspective of investors, the disclosure of CSR information can be used as an effective non-financial information communication channel to alleviate information asymmetry, especially when the quality of financial information is insufficient, investors can get a better understanding of the financial health of enterprises (Lizzeri) [26]. Investor's investment decision-making is more reasonable and wise, such as, Simms [27] surveys of more than 400 fund managers and analysts in nine European countries and finds that about 50% of respondents expressed that they would use the CSR information disclosed by management in investment decision-making. Flammer [28] analyzes the stock market and also finds that investors dislike companies with bad social responsibility problems, it means that corporate social responsibility (CSR) affects investors' evaluation and trust in the firms, which changes investors' portfolios and then may affect those firms' stock returns, Disclosure of CSR information will change the decision-making of information users, however, there are differences in the market for various types of investors to receive information, including time and cost for information processing. Compared with retail investors, institutional investors have abundant professionals and funds, which can take advantage of information gathering as professional investment institutions. Institutional investors respond to CSR information issued by firms soon and adjust their investment strategies in timely, retail investors will imitate the investment strategies of institutional investors to improve stock returns, such as, Wang and Qian [29] find that institutional investors voted with their feet on the behavior of dairy firms without social responsibility based on China's melamine incident in 2008, furthermore, individual investors follow the pace of institutional investors. According to the above discussion, we raise the following hypothesis: H3: Institutional investors play a mediating role in the impact of CSR on stock returns.

The Moderating Role of External Legal Environment
Marquis et al. [30] find that the uncertainty of regulatory systems will lead to the uncertainty of CSR performance. In the social environment with higher marketization process, stakeholders have enhanced their ability to protect their rights by using laws, regulations and market regulations, and have more information channels (media and intermediary organizations, etc.) to obtain and screen the authenticity and reliability of CSR information (Cambell; Wang et al.) [31] [32].
In other words, a higher institutional environment in the process of marketization will help improve the authenticity, reliability, and transparency of CSR information.
On the contrary, in the social environment with low marketization process, due to the lack of effective external supervision and restraint mechanism, the imperfection of laws and regulations and market mechanism, managers are inclined to adopt symbolic disclosure of CSR information, such as "dishonest" to improve corporate image. However, due to the imperfection of market information transmission mechanism and the lack of effective external supervision and restraint mechanism. The underdeveloped market intermediary organizations make it difficult for stakeholders to review and verify the authenticity and reliability of CSR information. Therefore, we have reason to believe that good external legal environment will help firms to better fulfill their social responsibilities and propose the Hypothesis 4.
H4: External legal environment plays a moderating role in the impact of CSR on stock returns.
This paper intends to explore the CSR role in stock return, with the mediating effects of corporate reputation and institutional investors, and the moderating effects of external legal environment, as shown in Figure 1.

Sample
The  excluded. Second, the companies whose CSR score is missing are also excluded.
Third, all variables are winsorized at the 1st and 99th percentiles of their distributions except CSR score. score to evaluate the firm performance of CSR, the higher the score, the more active the firm is fulfilling its social responsibility. The more detailed descriptions of the CSR data are in Table A1 in Appendix.

Other Variables
This paper uses the number of times that news media and newspapers report on firms to measure the reputation of firms. In order to measure corporate reputation more realistic, we choose the number of positive reports from news media and newspapers to further measure the reputation of firm. Previous studies have revealed risk factors and firms' characteristics that affect stock return, and these factors or characteristics should be controlled. In multivariate analysis motivated by prior research, we include the following control variables: debt asset ratio, market value, price-to-book ratio, firm risk, firm growth. Table 1 describes and defines all variables used in this study.   between enterprises, it is of great significance to study the impact of CSR on stock returns. The third to seventh rows in Table 2 also provide descriptive statistics for firm characteristics that we use as control variables, for example, the average of growth value is −0.1492, the average beta value is 1.1049.  (3) to model (4).

Model Setting
where , i t median represents the number of news media and newspapers report on firm i at year t reported by Internet and Newspapers, where , i t lawindex represents the legal environment index of the region where the firm is located at year t.

Results and Discussion
This section describes the main empirical results. Specifically, the influence of

Corporate Social Responsibility and Stock Return
To examine how CSR performance affects stock return, we regress stock return on CSR proxies and control variables by panel threshold model and OLS regression (model 1 and model 2). Table 3 reports our main results. We find that score variable has a threshold from Panel A of Table 3, based on threshold estimator in Panel B of Table 3 we find that, when score is less than 59.99 the coefficient on score is negatively and statistically significant at the 1% level, when score is more than 59.99 the coefficient on score is positively and statistically significant at the 5% level, this shows that the relationship between CSR and stock return is not monotonically increasing.  Furthermore, we add the square term to the regression to study the relationship between CSR performance and stock return. Additionally, we examine whether the relationship between CSR and stock return changes with ownership.
Column (1) shows that the coefficient on CSR*CSR is positively and statistically significant at the 1% level. Column (2) and column (3)  whether the products produced by firms are safe and reliable. Becchetti et al. [34] find that various CSR activities might affect firms' cost of equity in different ways. So we further study the impact of different parts of CSR (based on first-level) on stock returns. In Table 4, we find that there is a U-shaped relationship between CSR of different groups and stock returns except employee responsibility, there is an inverted U-shaped relationship between employee responsibility and stock return, and this is a puzzle. Why cannot the concern of firms for employees produce lasting promotion for the development of enterprises? This is a very interesting question; we will study it in future.

Mediating Effects of Corporate Reputation
In this section, we follow a three-step procedure to test hypothesis 2. In the first step, we model stock return as a function of CSR. In the second step, we model corporate reputation as a function of CSR. In the third step, we include both corporate reputation and CSR as determinants of stock return. We assess the mediating role of corporate reputation in the relationship between CSR and stock return by examining the significance and magnitude of the coefficients of CSR in all three steps. From Table 3, we find that CSR could significantly explain stock return (step 1); the conclusions of steps 2 and 3 are detailed in Table   5 and Table 6.
In Table 5, we use positive reported data of firm as a substitute variable for corporate reputation, according to the source of news media, the positive reported data of firms can be divided into three categories: positive reported data from internet and newspapers (column1), positive reported data from newspapers (column3), and positive reported data from internet (column5). In step 2,  (1) Return (2) Positive Reports by Newspapers (3) Return (4) Positive Reports by Internet (5) Return (  Note. Positive reports by internet and newspapers stands for the total number of all positive news about the company in financial newspapers and network news media in a day, which includes news title and news content, we calculate the daily data to get the monthly data, and then calculate the monthly data to get the annual data. Return stands for yearly stock return. Positive Reports by Newspapers stands for the total number of all positive news, which comes from financial newspapers. Positive Reports by Internet stands for the total number of all positive news, which comes from network news media, we also convert daily data into annual data.***, ** and * show the significance levels at 1%, 5% and 10% respectively. Theoretical Economics Letters   Note. Return stands for yearly stock returns. Reports by internet and newspapers stand for the total number of all news about the company in financial newspapers and network news media in a day, which includes news title and news content, we calculate the daily data to get the monthly data and then calculate the monthly data to get the annual data. Reports by Newspapers stand for the total number of all news, which comes from financial newspapers. Reports by Internet stands for the total number of all news, which comes from network news media, we also convert daily data into annual data. ***, ** and * show the significance levels at 1%, 5% and 10% respectively. column (1) of Table 5 shows that the coefficient on score is positively and statistically significant at the 1% level, column (3) of Table 5 shows that the coefficient on score is positively and statistically significant at the 10% level, column (5) of Table 5 shows that the coefficient on score is positively and statistically significant at the 1% level, this shows that there is a positive correlation between CSR and corporate reputation.
In step 3, the regression equation includes CSR and corporate reputation, the coefficients on score*score are positively and statistically significant at the 1% level in column (2), column (4) and column (6), this means that the corporate reputation to serve as a moderator on CSR and stock return.
In Table 5, we have analyzed in detail the mediation role of good reputation of firms, but there is also bad reputation of firms. Under the influence of bad reputation, the stock price of firms will inevitably decline, so what is the role of corporate reputation when there is good reputation and bad reputation? In Table 6, we perform similar analysis as in Table 5 using all the reported data of firm as a substitute variable for corporate reputation. We found the same results  Table 5. In summary, Table 5 and Table 6 provide the results of the regression analysis testing the mediating role of corporate reputation in the relationship between CSR and stock return from different perspectives.

Mediating Effects of Institutional Investor and Moderating Effect of External Legal Environment
We use the same method proposed in hypothesis 2 to test the mediating effect of institutional investors. The results of our empirical analyses were in Table 7.
Column (1) of Table 7 shows a significant coefficient on score (p < 0.05), indicating that CSR has a significant positive effect on institutional investors. Furthermore, we find that the coefficient on shareholding is significantly positive in columns (2). These results suggest that institutional investors play a mediating role in the impact of CSR on stock returns. So far we have proved the hypothesis 3. Table 7. CSR and stock return with the mediating effect of corporate reputation and the moderating effect of external legal environment. Note. Dependent variable is yearly stock return. Shareholding stands for the shareholding ratio of institutional investors. External legal environment stands for the legal situation of the region where the firm is located. ***, ** and * show the significance levels at 1%, 5% and 10% respectively. Theoretical Economics Letters To test the moderating effect of the external legal environment, regression analysis was carried out according to model 6. In column 3 of Table 7 presents the estimations of coefficients of variables to test the relationship between CSR and stock return with the moderating effect of external legal environment, lawindex is positive and statistically significant at the 1% level with a coefficient value of 0.4322, the positive coefficient value of interaction term lawindex*score 2 is 0.0129, and is statistically significant at the 10% level. This means that external legal environment is conducive to enhancing the positive impact of CSR on stock returns and reducing the hindrance of CSR on stock returns. Hypothesis 4 was proved.

Additional Analyses
In the analysis of Section 4, we did not consider the differences in CSR of enterprises itself. In this section, we will further analyze the impact of CSR on stock returns from the perspective of the differences in CSR of enterprises. In reality, not all enterprises publish CSR reports, and not all enterprises voluntarily publish CSR reports. So we test whether voluntary disclosure of CSR and mandate disclosure of CSR have different effects on stock returns. In column 1 of Table 8   Table 8. CSR and stock return with different sample. Note. Dependent variable is yearly stock return. According to whether the enterprise independently discloses the corporate social responsibility information, we divide them into two groups: the disclosure group and the non-disclosure group, we further divided them into different sub samples in the disclosure group. ***, ** and * show the significance levels at 1%, 5% and 10% respectively.
reports the regression results of the group that does not disclose corporate social responsibility information, score*score is positive and statistically significant at the 5% level with a coefficient value of 0.0001. In column 2 of Table 8 presents the estimations of coefficients of variables to test the relationship between CSR and stock return in the sample of voluntary disclosure corporate social responsibility, score*score is positive and statistically significant at the 5% level with a coefficient value of 0.0001. In column 3 of Table 8 reports the results of companies that required disclosure CSR, score*score is positive and statistically significant at the 1% level with a coefficient value of 0.0002. This means that there is a U shape between CSR and stock return has nothing to do with whether the corporate social information is disclosed.
Some of firms' CSR reports are audited by a third party, while most of firms' CSR reports have not been audited by a third party. So we further divide the disclosure group based on this. In column 4 of Table 8 reports the results of the sample group whose CSR report is audited by the third party, score*score is positive and statistically significant at the 5% level with a coefficient value of 0.003.
In column 5 of Table 8 reports the results of the sample group whose CSR report is not audited by the third party, score*score is positive and statistically significant at the 1% level with a coefficient value of 0.0002. This means that there is a U shape between CSR and stock return has nothing to do with whether CSR report is audited by the third party.

Robustness Tests
We take one step forward by using different CSR measures to check the robustness of the relationship. We learn from Stig and Jesper [35], we orthogonalize score with respect to mean score of the same property right in the same industry in the same year by running the regression: where , i t CSR is the third party's evaluation of the enterprise's corporate social responsibility in firm i at year t, t meanscore is then mean score of the same industry in the same year or the same property right in the same industry in the same year, CSR ⊥ is the residual, to evaluate how much independent explanatory power is left by mean score after controlling for the information in industry. As the next step, we then use CSR ⊥ (lagged) as the explanatory variable in our predictive regressions for one-year-ahead stock return ( , 1 i t ret + ), run the regression according to model 2. The results are in Table 9. The results show that there is significant explanatory power left in CSR when adjusted for industry information, there is a stable U-shaped relationship between corporate social responsibility and stock return.

Conclusions
In this paper, we not only study the relationship between CSR performance and stock returns but also discuss the underlying mechanism, which affects their  relationship. We find that there is a U-shaped relationship between CSR and stock returns, which is not affected by the nature of firm property rights. Furthermore, we try to find out that the channels of CSR performance affect stock returns. We find that corporate reputation and institutional investors are two channels for CSR performance to influence stock returns. Specifically, CSR can improve corporate reputation, which helps to enhance the attention of investors to the company and boost the rise of stock returns. CSR can reduce information asymmetry, and institutional investors can timely grasp favorable information, so as to change their investment strategies. Furthermore, retail investors will follow institutional investors, thus promoting the rise of stock returns. We also examine the moderator role of external legal environment on CSR and stock return. External legal environment promotes CSR to enhance its positive role and slow down its negative role. Our research contributes to the literature in several ways. First, we broaden the range of empirical evidence on the relation between CSR and firm performance in stock market with new data. This paper first uses score on CSR of Hexun to study the impact of CSR on stock market. Second, while prior studies produce conflicting results on the relation between CSR and firm performance, we provide a new way to understand the role of CSR on firm performance. Third, this paper uncovers corporate reputation and investment strategy of institutional investors are the channels through which CSR affects stock market, contributes to a growing body of CSR literature by providing new evidence on the moderating effect of external legal environment between CSR and stock return.
Based on the findings of this paper, we have several suggestions to further improve CSR activities in China. For investors, they should pay more attention to the information about CSR, which is conducive to improving the accuracy of  Income tax divided by total profit (weight 10% or 5% or 15%) Amount of public welfare donation (weight 10% or 5% or 15%) Note. Description of distribution of weight proportion in different industries: by default, the weight of shareholders' responsibility accounts for 30%, the weight of employees' responsibility accounts for 15%, the weight of rights and interests of suppliers, customers and consumers accounts for 15%, the weight of environmental responsibility accounts for 20%, and the weight of social responsibility accounts for 20%. Among them, the responsibility weight of employees in the consumer industry accounts for 10%, the rights and interests of suppliers, customers and consumers account for 20%, and the weight of other indicators remains unchanged; the environmental responsibility weight of manufacturing industry accounts for 30%, the social responsibility weight accounts for 10%, and the weight of other indicators remains unchanged; the environmental responsibility weight of service industry accounts for 10%, the social responsibility weight accounts for 30%, and the weight of other indicators remains unchanged. Table A2. CSR performance and stock return based on the nature of industry. Note. Dependent variable is yearly stock return. ***, ** and * show the significance levels at 1%, 5% and 10% respectively.