Corporation is considered as an entity, so it must develop a series of policies, strategies, and actions to grow to position itself well in a complex business environment or maximize shareholder wealth. According to agency theory, management considers the cost of corporate social responsibility as the cost of reputation or political costs when maximizing profits [2] . As a legal entity, corporate needs to take social responsibility even though it needs cost. Corporations are considered socially unacceptable if they tend to be tax-aggressive [4] .

Risk management strategy theory believes that CSR activities can enhance its reputation for social responsibility to avoid risks of anti-political, regulatory and social sanctions [6] . Godfrey (2005) [7] argues that positive social reputation is important when there is a negative event in a business. This reputation helps to improve sanctions on the company’s business, thus providing a degree of insurance coverage. Tax aggressiveness can lead to negative sanctions such as loss of corporate reputation, increased political pressure, penalty imposed by tax bureaus and even consumer boycotts [8] . Corporation will hedge the consequences of the negative sanctions by implementing a social responsibility strategy if reputation of social responsibility can reduce the radical tax-aggressive. In the meantime, Lanis and Richardson (2012) [4] find that the higher the degree of disclosure of corporate social responsibility, the lower the actual tax rate of enterprises by studying the degree of disclosure and the tax-aggressiveness of social responsibility in 408 listed companies in Australia in 2008-2009. Corporations disclosing social responsibility mislead people believe that they are responsible and will not be tax-aggressive. However, the financial fraudulent cases of Enron and WorldCom make us believe that many enterprises are carrying out taxation activism while carrying out social responsibility activities.

Corporate culture theory assumes that corporate social responsibility is the common philosophy of enterprises, and corporate tax payments have a positive impact on communities and society [9] . Companies that are tax-aggressive are widely viewed as unethical by the general public and mass media, and similarly as irresponsible corporate social responsibility [10] . Hoi et al. (2013) [5] argue that corporate social responsibility activities are closely linked to the interests of many stakeholders, while taxation has a radical impact on government demands on corporations and social welfare. They test the tax incentives in different ways and the results show that firms with four or more irresponsible CSR activities are more likely to be tax-aggressive and have higher accounting-tax differences. In order to promote a good corporate culture, management will convey the belief of social responsibility from the top to the bottom, objectively and truthfully record economic transactions from the bottom up, and will focus on avoiding downward tax revenue when conducting tax planning adjustment, reflecting the concept of corporate culture.

Disclosure of social responsibility information does not equate with social responsibility performance, but it can explain social responsibility performance. Disclosure of social responsibility information refers to the annual social responsibility report, environmental report, sustainability report and corporate citizen report disclosed by the enterprise each year, which contains other information such as social responsibility performance. Radical tax exposure makes the disclosure of social responsibility information is considered as “whitewashing” and reduces the credibility of the disclosure of social responsibility. Moreover, on the one way, the current tax system in our country is not perfect, and enterprises are vulnerable to the tax revenue radical tax loopholes. On the other way, corporate social responsibility information disclosure is a voluntary action; therefore, one of our focuses is on the link between tax aggressiveness and disclosure of social responsibility information. As such, the paper formulates the corresponding prediction from the risk-management perspective as follows:

H1: The more full disclosure of corporate social responsibility information, the higher level of the tax aggressiveness.

Many foreign studies show that having a political background in the management of a company brings policy preferences such as financing facilities and tax incentives to the enterprise, thereby increasing the enterprise value [11] . Adhikari et al. (2006) [11] study the relationship between the actual tax rate and political links based on a sample of 257 enterprises in Malaysia from 1990 to 1999 and find that there is a significant negative correlation between them which means that enterprises with political connections has significantly lower actual tax rates.

In China, seeking political ties may be beneficial to the economic development of itself and society. However, it may also hinder economic development because of rent-seeking [12] . 28.8% of private entrepreneurs think “the most important thing is to strive for being deputies and CPPCC members,” and they show strong interest in political participation [13] . Weng Jianying (2015) [14] finds that the accounting earnings of private listed companies with political connections are significantly higher than that of private listed companies without political connections, which shows that political connections can increase the content of earnings information. Political identities can create tax avoidance effects, and taxpayers with political identities are more likely to become tax-aggressive [15] .

Enterprises pay the cost of political connections as to look for political connections, but also receive the support of preferential tax policies. Corporates disclosing social responsibility information also need to pay a certain cost, but reap the rewards of social responsibility reputation. If companies get tax breaks when getting political connections, are they willing to pay social responsibility costs to take tax activism? As two voluntary mechanisms, will companies choose to pay a double cost to achieve the goal? This paper argues that enterprises with no political connection to get the government resources will tend to use the disclosure of social responsibility information to obscure tax aggressiveness. Therefore, this paper proposes the second research hypothesis:

H2: Relative to enterprises with political connection, enterprises with no political connection are more likely to be tax-aggressive by disclosing social responsibility information.

If management has served in local government departments or has local government resources, it is likely to receive more financial subsidies and more political support [16] . From the county, city, provincial and to even higher levels, enterprises increasingly pay more and more costs for their gradual search for political ties. This paper argues that companies with weaker political connections pay relatively lower costs, receive less government subsidies and receive less tax incentives will show red-eye to those who are stronger political connections. Therefore, they can only imply social responsibility activities to carry out radical taxation, with the expectation of obtaining the same tax consequences as those with strong political ties. For enterprises with strong political connections, the policy support and tax concessions they have obtained are in a favorable position to compete in the market so there is no need to spend a large amount of costs to conduct additional disclosure of social responsibility information and exacerbate the expenditure pressure on enterprises Therefore, this paper proposes the third research hypothesis:

H3: Relative to enterprises with strong political connections, enterprises with weaker political connections are more likely to be tax-aggressive by disclosing social responsibility information.

Changes in political relations will lead to greater uncertainty and increased management risks, so management will become more cautious. From no political connection to political connection, management will positively demonstrate “positive energy” and is likely to consciously reduce tax incentives, so as to establish long-term and stable political ties with the government. From political connections to political gaps, management may want to “wash the water” through disclosing of social responsibility information. Therefore, this paper proposes the forth research hypothesis:

H4: Political changes have weakened the relationship between the disclosure of social responsibility and tax aggressiveness.

Due to 2008 corporate tax reform and data availability, the paper chooses the A-share listed companies from 2008 to 2014 as the research sample, and rejects the corresponding data according to the following criteria: ① Financial, insurance industry; ② Financial data, corporate governance data missing; ③ Newly listed companies; ④ The taxation level (the actual tax rate) is greater than 1 or less than 0 companies, and then get the final sample of 11,314 companies in the year. Financial data, governance structure data are from CSMAR, Wind database. Data of political connection is collected by hand which is from the resumes of corporate executives.

Empirical studies show that the real tax rate reflects the tax-aggressive level of the firm [17] , and academics often use the actual tax rate as an alternative tax revenue radical [18] . Therefore, The real tax rate (ETR) is the tax revenue level and is calculated using the effective tax rate = income tax expense/EBIT [3] [18] [19] [20] . The lower the tax rate, the higher the level of tax aggressiveness is.

The index method is widely used in the research of social responsibility information. In this paper, we refer to the classification of corporate social responsibility information disclosure index by Zhu Song et al. (2011) [21] , Shen Hongtao (2007) [22] , Lanis and Richardson (2012) [4] and the annual disclosure of listed companies by CSMAR database. The content of sample corporate social responsibility disclosure is divided into five categories: employees, suppliers, consumers, communities and the environment. At the same time for each type of information disclosure assignment, the disclosure is assigned a value of 1, not assigned 0, resulting in social responsibility information disclosure index, so the value ranges between 0 and 5.

This paper draws on Hu Xuyang et al. (2008) [23] and Li Weian (2013) [15] , defining political contact (PC) as chairman, general manager and chairman of the board of supervisors as former deputies or CPPCC members (county, city, province and nationwide). In H2, the Chairman, General Manager and Chairman of the Board of Supervisors served as the current NPC deputies or CPPCC members (county, city, province and nationwide) with a value of 1, or 0 otherwise. In H3, the Political Link Strength (PCI) is measured using the exponential approach. The three methods for measuring the political connection index are as follows: the chairman, the general manager and the chairman of the board of supervisors are currently serving as officials at the rank of level above, NPC deputies and CPPCC members with an assignment of 3; Under the provincial level people’s congress deputies, CPPCC members or served as officials at the level of above, NPC deputies or CPPCC members or provincial people’s congress, CPPCC Standing Committee, the assignment is 2; served at the level of the following officials, provincial people’s congress below CPPCC members Assignment is 1; no political background is assigned 0. This gives the political link index, which ranges between 0 and 3. In H4, the change of political connection (△PCI) adopts the assignment method. The change of the political ties between chairman, general manager and chairman of the supervisory board relative to the previous year has the value of 1, and the change of 0.

Other control variables are in Table 1.

In order to test H1 and H2, this paper constructs model (1).

L O G ( E T R i t ) = β 0 + β 1 × S D I i t + β 2 × S D I i t × P C i t + β 3 × S I Z E i t + β 4 × L E V i t + β 5 × C A P I N T i t + β 6 × I N V I N T i t + β 7 × R O A i t + β 8 × T O P 1 i t + β 9 × M B i t + β 10 × T A X R A T i t + β 11 × A G E P U B i t + β 12 × O P I N I O N i t + β 13 × S T A T E i t + ∑ Y E A R + ∑ I N D U R + ε i t (1)

If the corporate social responsibility information is disclosed more fully, the actual tax rate is lower and then the higher level of tax aggressiveness then β 1 should be negative. Enterprises with no political connection are more likely to be tax-aggressive by disclosing social responsibility information, and then β 2 should be positive.

In order to test H3, this paper constructs model (2).

L O G ( E T R i t ) = β 0 + β 1 × S D I i t + β 2 × S D I i t × P C I i t + β 3 × S I Z E i t + β 4 × L E V i t + β 5 × C A P I N T i t + β 6 × I N V I N T i t + β 7 × R O A i t + β 8 × T O P 1 i t + β 9 × M B i t + β 10 × T A X R A T i t + β 11 × A G E P U B i t + β 12 × O P I N I O N i t + β 13 × S T A T E i t + ∑ Y E A R + ∑ I N D U R + ε i t (2)

If β 2 is positive, it indicates that enterprises with weaker political connections are more likely to be tax-aggressive by disclosing social responsibility information.

In order to test H4, we build a model (3).

L O G ( E T R i t ) = β 0 + β 1 × S D I i t + β 2 × S D I i t × Δ P C I i t + β 3 × S I Z E i t + β 4 × L E V i t                                           + β 5 × C A P I N T i t + β 6 × I N V I N T i t + β 7 × R O A i t + β 8 × T O P 1 i t                                           + β 9 × M B i t + β 10 × T A X R A T i t + β 11 × A G E P U B i t                                           + β 12 × O P I N I O N i t + β 13 × S T A T E i t + ∑ Y E A R + ∑ I N D U R + ε i t (3)

If β 2 is positive, it indicates that political changes have weakened the relationship between the disclosure of social responsibility and tax aggressiveness.

Table 2 reports descriptive statistics of all variables except year and industry control variables. The mean value of the effective tax (ETR) is 0.16, with maximum and minimum of 0.99 and 0.000 respectively. This is because samples with the actual tax rate of less than 0 and greater than 1 are deleted. The average of the socially responsible disclosure index (SDI) is only 1.11 with the median 0, indicating that most of the sample companies do not disclose social responsibility information, because China has not yet formulated related laws and regulations about the disclosure of social responsibility information and the capital market demand for such information is not strong. The average political connection (PC) was 0.23 with a median of 0, indicating that most sample companies had no background for political connections. The change in political linkages (△PCI) averages 0.15, suggesting that more than half of businesses with a background in political ties have undergone political transitions. The average SIZE is 21.90; the minimum is 20.09 and the maximum is 24.27, indicating that most enterprises are close in size. LEV mean and median value are 0.45 and 0.46; the maximum and minimum values are 0.11 and 0.82, indicating that the level of debt between enterprises is different, and there is a certain gap, but the overall condition of the financial structure of China’s listed companies is good. The average capital intensity (CAPINT) is 0.23, with the maximum and minimum values of 0.58 and 0.01 showing a large gap, indicating that the long-term investment gap between enterprises is large. INVINT has a mean of 0.17, a median of 0.13, a maximum and a minimum of 0.55 and 0.01, a small mean and median, indicating that sample companies have less investment in inventory. In addition, the gap of the return on assets (ROA) is large; the minimum is negative; the maximum value of 0.14. Equity concentration (TOP1) averages 36.33%, with a median of 34.86%, indicating that sample companies are more concentrated in equity. The mean of investment opportunities (MB) is 2.24, with a median of 1.86, indicating that 50% of sampled firms have lower expected value. The average age of listed companies (AGEPUB) is 9.27, with a median of 9, indicating that the general business listing is less than 10 years and the business operation may not be stable enough. The average of STATE is 0.48, indicating that 48.2% of sample companies are stated-own companies.

The Regression results are shown in Table 3 as belows. Model (1) regression shows that β 1 is negative and significantly correlated, confirming H1, that is, the more full disclosure of corporate social responsibility information, the higher level of the tax aggressiveness. It shows that the purpose of corporate social responsibility activities is not to serve the society. Maybe it is for the economic motivation and the disclosure of social responsibility information may be the “facade decoration”, hoping to enhance the corporate reputation which can provide certainty to the enterprise of the security and to buffer the impact of its negative events. β 2 is positive and it is significantly related to the actual tax rate, indicating that relative to enterprises with political connection, enterprises with no political connection are more likely to be tax-aggressive by disclosing social responsibility information, in line with the expectation of the second assumption. Companies with weaker political connections pay relatively lower costs, receive less government subsidies and receive less tax incentives will show red-eye to those who are stronger political connections. For the politically connected enterprises, the policy support and tax concessions they have obtained are in a favorable position to compete in the market. Therefore, there is no need to spend a large amount of additional information on social responsibility disclosure to exacerbate the expenditure pressure on enterprises.

β 2 is positive and significant correlations in model (2), indicating that relative to enterprises with strong political connections, enterprises with weaker

Note: ***indicates significant at 1% level (bilateral); **indicates significant at 5% level (bilateral); *indicates significant at 10% level (bilateral).

political connections are more likely to be tax-aggressive by disclosing social responsibility information, in line with H3 expectation. Enterprises with higher political connection have paid a high cost to obtain the corresponding policy support and taxation resources. In contrast, the enterprises with weak political connections have relatively lower cost but gain less policy support and tax preference will have “red-eye” towards companies’ additional resources.

β 2 is positive and significant correlations in model (3), which confirms that the changes in political connections in H3 cause the enterprises to face greater uncertainty. From the political gap to the political connection, management will actively demonstrate “positive energy”. From political connection to political gap, management may wish to “cleanse” by disclosing social responsibility information while companies with stable political ties tend to mask tax incentives by disclosing social responsibility information.

In order to explain the relationship between social responsibility information disclosure and the level of tax aggressiveness, this paper uses Book-Tax Difference (BTD) and Corporate Social Responsibility (CSR) as the alternative variables of explanatory variables and explanatory variables respectively.

Chen et al. (2010) [2] find that BTD is applicable in China as a measure of tax avoidance. Using Desai’s (2006) [24] approach, BTD = (total profit − current income tax expense/nominal tax rate)/total assets, the greater the BTD difference, the higher level of tax aggressiveness. The results are shown in Table 4 below. It shows that the more full disclosure of corporate social responsibility information, the higher level of the tax aggressiveness. Relative to enterprises with political connection, enterprises with no political connection are more likely to be tax-aggressive by disclosing social responsibility information. Relative to enterprises with strong political connections, enterprises with weaker political connections are more likely to be tax-aggressive by disclosing social responsibility information. Political changes have weakened the relationship between the disclosure of social responsibility and tax aggressiveness.

In addition, the index of social responsibility information disclosure is reassigned, that is, if the enterprise has disclosed social responsibility information, the value of 1 is assigned; otherwise, the value is 0. The results in Table 5 below indicate that the more full disclosure of corporate social responsibility information, the higher level of the tax aggressiveness. Relative to enterprises with political connection, enterprises with no political connection are more likely to be tax-aggressive by disclosing social responsibility information. Relative to enterprises with strong political connections, enterprises with weaker political connections are more likely to be tax-aggressive by disclosing social responsibility information. Political changes have weakened the relationship between the disclosure of social responsibility and tax aggressiveness.

Note: ***indicates significant at 1% level (bilateral); **indicates significant at 5% level (bilateral); *indicates significant at 10% level (bilateral).

Note: ***indicates significant at 1% level (bilateral); **indicates significant at 5% level (bilateral); *indicates significant at 10% level (bilateral).