Corporate Social Responsibility and Earnings Management of Listed Companies in Power Industry

Due to the particularity of business operation of electric power industry enterprises, their social responsibility will have an important impact on national economic development and residents’ life. In this paper, 303 Listed Companies in the power industry from 2013 to 2018 are taken as samples to measure the degree of social responsibility of the power industry based on the contribution of the power industry enterprises to the government, employees, shareholders, creditors, suppliers and customers, and to test the correlation between the Listed Companies in the power industry and earnings management. The empirical results show that the more the power industry enterprises pay to the outside world, the lower the degree of earnings management, that is, the better the degree of social responsibility of listed companies in the power industry, the lower the degree of earnings management. Through the research results of this paper, we can think that the social responsibility behavior of power industry enterprises is “altruistic” rather than “egoistic”, which explains the importance of non-marketization of large state-owned power enterprises such as State Grid Corporation of China.

, and will also be punished by the capital market (Xiao et al., 2010). In recent years, with the outbreak of many malignant events of enterprises that do not attach importance to social responsibility, such as melamine incident, Foxconn employees" jumping off buildings, etc., Chinese enterprises have gradually attached importance to the concept of social responsibility, and gradually implemented the behavior of fulfilling social responsibility, such as donating Hope Project, donating after Wenchuan earthquake, spending on post disaster construction, etc., which Chinese enterprises have fulfilled The event of social responsibility has also been recognized by the capital market and market investors (Chen & Zhang, 2012).
It can be seen that social responsibility is an internal and external issue of enterprises. From the internal point of view, social responsibility is the "cost" of an enterprise and the performance of an enterprise will be affected to a certain extent by the performance of social responsibility. Brammer et al. (2006) and Li et al. (2006) found that the higher the level of performance of social responsibility is, the worse the performance will be. From the external point of view, although the performance of corporate social responsibility will bring better impact and reputation to the enterprise, it will make the enterprise Facing more external supervision, enterprises can reduce the degree of information asymmetry between enterprises and investors, but it may also make enterprises transfer investors" attention to financial reports through the disclosure of social responsibility information, so as to achieve the purpose of operating financial reports. So from the perspective of earnings management, it is still an unknown question whether the performance of social responsibility will make enterprises have a better sense of social responsibility and reduce the degree of earnings manipulation, or whether enterprises will use social responsibility as a means to cover up their manipulation. Deng et al. (2011) found that there was a significant negative correlation between corporate social responsibility and earnings management, and social responsibility behavior may also be a cover for listed companies to smooth earnings and reduce or increase earnings in loss years. Deng et al. (2013) also found that enterprises with better performance in social responsibility do less earnings management, indirectly indicating that enterprises selective disclosure of information, so that the disclosed information can best serve their own interests (Meng et al., 2010). Therefore, this paper holds that: Hypothesis 2: The higher the degree of corporate social responsibility performance in the power industry, the higher the degree of earnings management in the corresponding power industry.

Variable Definition and Empirical Model
Explained Variable: Earnings Management. The measurement method of earnings management was first proposed by Jones (1991), and then most of the literature revised the original model accordingly.
This paper uses the modified model proposed by Dechow et al. (1995) for measurement. Dechow et al. (1995) believed that the earnings management value of listed companies is the value of manipulated accrued profits, which is the difference between the total accrued profits and the non manipulated accrued profits, that is, DA=TA-NDA. TA is the total accrued profit value, the difference between the net profit and the net cash flow of operating activities, which is adjusted by the total assets of the previous period, while NDA is the non manipulative accrued profit value, which is not disclosed in the financial statements, that is, it needs to be calculated. The specific calculation formula is as follows: In the formula (1), REV t and REV t-1 are the operating revenue of the current period and the previous period respectively, REC t and REC t-1 are the change value of accounts receivable of the current period and the previous period respectively, PPE is the fixed assets of the current period, INTAN is the intangible assets of the current period, and a is the coefficient to be appraised, which needs to be obtained through the following regression equation: According to the research content and design variables, this paper sets the following empirical model.
In formula (3), it is the coefficient of variable to be estimated, C is the constant term. In this paper, we will test formula (3) separately according to the classification of variable CSR.

Data Description
This paper chooses the listed companies in the power industry as the empirical analysis sample, because the listed companies will regularly disclose information. Considering the amendment of industry classification made by CSRC in 2012, this paper will locate the sample period in 2012-2018.
At the same time, this paper will delete the sample of listed companies in the power industry which are specially processed in the sample year and the sample of Listed Companies in the power industry which lack more data. Finally, this paper gets 303 samples of Listed Companies in the power industry  Table 2 shows the descriptive statistical results of the main variables in this paper. From the results of Table 2, the mean value of the explained variable DA is -0.0001, almost 0, that is to say, the degree of earnings management of listed companies in the power industry is not high, while the median value is 0.0032, which indicates that more listed companies in the power industry operate positive earnings management, but from the maximum and minimum value of the variable DA, the degree of earnings management of listed companies in different power industries is quite different. From the six explanatory variables of social responsibility, the mean value of variable CSR 1 is 0.0813, which shows that the net value of tax of listed companies in the power industry accounts for about 8% of the cash inflow value of operation, but the tax returns received by some listed companies in the power industry are greater than the taxes paid, that is to say, some listed companies in the power industry have not made corresponding contributions to the government; the mean value of variable CSR 2 is 0.0973. It shows that the cash paid to employees by listed companies in the power industry accounts for nearly 10% of the operating revenue, but the gap between different companies in terms of employee benefits is very large; the average value of variable CSR 3 is 0.0238, which indicates that listed companies in the power industry will give about 2% of the operating revenue to shareholders.  In addition, from the standard error value of each variable, except that the standard error values of ln(Size) and year are greater than 1, the standard error values of other variables are less than 1, which means that most of the variables are relatively concentrated and there is no large dispersion.   and control variables are small, and the largest is not more than 0.5, which indicates that there is no collinearity problem between variables, which shows that the empirical results in the following paper will not be distorted due to the problem of multicollinearity.

Table 4. Regression Results
(1) (2) (3) (4) (5)  Table 4 shows the regression results in this paper. From the perspective of the six regression results as a whole, although the R 2 value of each regression result is not high, it can not be simply compared with the R 2 value, but also needs to be combined with the F statistics of each regression result, and the F statistics of each regression result can pass the significance test of the conventional confidence level, which shows that each regression result has a certain degree of credibility, and the regression result is credible.
Specific to the regression results, when the explanatory variable is CSR 1 , its coefficient value is 0.1901, and it can pass the test at a significant level of 10%, which shows that from the perspective of the contribution to the government, there is a significant negative correlation between the degree of social responsibility of the power industry and earnings management, that is, the higher the contribution rate of listed companies to the government, the lower the degree of earnings management. When the variable is CSR 2 , the coefficient value is -0.0427, which can be tested by 5% significant level. It shows that from the perspective of contribution to employees, there is a significant negative correlation between the degree of social responsibility of the power industry and earnings management, that is, the higher the contribution rate of listed companies to employees in the power industry, the less likely they are to manipulate earnings management. When the variable is CSR 3 , the coefficient value is -0.2647, which can be set by routine setting the significance test of credit level shows that from the perspective of shareholder contribution rate, the more return the listed companies in the power industry give to shareholders, the less likely they are to carry out high-level earnings management. When the variable is CSR 4 , the variable coefficient value is -0.1242, which can also be tested by the significance test of conventional confidence level of 5%, indicating the greater the contribution of the listed companies in the power industry to the social responsibility of creditors, the corresponding degree of earnings management will be lower. The coefficient value of variable CSR 5 is -0.0395, which can pass the significance level of 10% confidence level, indicating that from the contribution rate of suppliers, the larger the ratio of cash paid by the listed companies in the power industry to cash outflow value is, the corresponding degree of earnings management will be lower. The coefficient value of variable CSR 6 is -0.0463, which can pass The significance test of 1% confidence level shows that from the perspective of customer contribution rate, the higher the contribution of listed companies to customers in the power industry, the less easy it is to carry out earnings management. According to the six indicators of social responsibility set in this paper, all the coefficient values are negative, which means that the higher the degree of social responsibility performance in the power industry, the lower the corresponding degree of earnings management, that is to say, the original intention of the listed companies in the power industry to perform social responsibility is to be responsible for all stakeholders, not for their own private interests.
industry among the municipal companies, the audit governance of the four major international accounting firms is higher, but there is instability, that is to say, among the listed companies in the power industry, the audit effect of the four major international accounting firms seems to be no significant difference from that of the domestic accounting firms. The regression results between the variable YEAR and the explained variable do not pass the significance test of the conventional confidence level, and the results show that the listed companies in the power industry have no significant impact on their earnings management behavior, and the long and short listing time will not affect their earnings management behavior. There is a significant positive correlation between the variable ROE and the explained variables, indicating that the listed companies in the power industry with signs of earnings will indeed manipulate earnings management to a greater extent.
In order to test the robustness of the research conclusion, this paper conducts the corresponding can be considered that the regression results of this paper are credible.

Research Conclusion
From the research of existing literature, the social responsibility of power industry enterprises is not good, but because of the particularity of power industry, the social responsibility of power industry enterprises is very important. In this paper, 303 listed companies in the power industry from 2012 to 2018 are selected as samples to empirically test the relationship between social responsibility and earnings management of listed companies in the power industry. The results show that based on the contributions of government, employees, shareholders, creditors, suppliers and customers, we find that listed companies in the power industry contribute to the government the more employees, shareholders, creditors, suppliers and customers pay, the lower the corresponding degree of earnings management.
That is to say, the better the degree of social responsibility of listed companies in the power industry, the lower the degree of earnings management. From the empirical results of this paper, we can see that although the degree of the listed companies in the power industry to fulfill their social responsibilities is not very high, but the behavior of the listed companies in the power industry to fulfill their social responsibilities is "altruistic", not "egoistic", that is to say, the original intention of the listed companies in the power industry to fulfill their social responsibilities is relatively simple, and there are no special purposes. At the same time, the empirical evidence of this paper also shows that the lower the debt ratio and the signs of surplus manipulation, the higher the degree of earnings management. Although there is a positive correlation between the asset size and the year of listing and earnings management of the listed companies in the power industry, and there is a negative correlation between the audit behavior and earnings management of the listed companies in the power industry some relationships are unstable. For the power industry, because its business involves the national economic development and the normal life of residents, its social responsibility is very important, so the power industry enterprises not only need to pay attention to the implementation of social responsibility, but also need to fulfill the social responsibility. On the one hand, power industry enterprises should learn from the practices of other industry enterprises, ensure the issuance of social responsibility reports of power industry enterprises, so as to release the behavior of social responsibility reports, disclose the performance status of their social responsibility to the society and ensure the accuracy of relevant information; on the other hand, power industry enterprises should also increase their efforts to fulfill social responsibility, in addition to ensuring the industry In addition to the implementation of environmental protection related social responsibilities, we should also increase efforts in such aspects as hope projects and charitable donations, so as to ensure that more stakeholders' interests are met. At the same time, considering the unity of data availability and measurement standards, this paper selects the general listed companies in the power industry as the sample of empirical evidence, but some large-scale power industry enterprises, including the State Grid, and the better performance of social responsibility, are not included. Therefore, in the follow-up study, we should strengthen the non listed power such as the State Grid Attention to the social responsibility of industrial enterprises.