The Nature of Controlling Shareholders, Political Background and Corporate Anti-Corruption Practice Disclosure

Abstract

The purpose of this paper is to examine the relationship between the nature of controlling shareholders and corporate anti-corruption practice disclosure (ACPD) as well as the mediating role of political background of the chairman or CEO of the firm on the relationship between the two. The content analysis was conducted to extract ACPD from standalone corporate social responsibility reports (CSRR) of 703 China's A-share listed companies. A dummy variable was constructed according to whether a firm disclosed ACPD or not. Logistic regression analysis was used then. Results show that the nature of controlling shareholders has a significant impact on corporate ACPD, with central enterprises disclosing the most frequently, local state-owned enterprises the second and private enterprises the least. Political background of the chairman or CEO has a negative impact on corporate ACPD of state-owned enterprises. These findings have some useful insights in understanding the rent-seeking behavior and information disclosure behavior of corporates in emerging markets. In order to curb the serious corruption problem which is commonplace in developing countries like China, the government should exert certain pressure to strengthen the supervision of information disclosure of listed firms and improve information transparency.

1. Introduction

In recent decades, corporate social performance and information disclosure have been hot topics in practical and academic areas. A lot of research findings on corporate social responsibility disclosure (CSRD) have emerged. However, those findings mainly focus on the level of environmental disclosure or the overall level of social disclosure (Newson & Deegan, 2002; Islam & Deegan,2010; Orij, 2010; Fernandez-Feijoo, Romero, & Ruiz, 2014; Yekini, Adelopo, Andrikopoulos, & Yekini, 2015). Fewer findings have been reported on the disclosure level and influencing factors of other social items such as anti-corruption practice. Only two studies have been found on corporate ACPD until now, i.e. Dissanayake, Islam, and Dellaportas (2011), Joseph, Gunawan, Sawani, Rahmat, Noyem, and Darus (2016). Dissanayake et al. (2011) employs a case study to investigate the pressures of government, NGOs and the media on corporate anti-bribery disclosures. While Joseph et al. (2016) conducts only a descriptive statistical analysis of the status quo of anti-corruption disclosure. No literature has been found to take a large sample data to test the motivation and influencing factors of corporate ACPD. In addition, researches on the motives of corporate social responsibility disclosure in the past mostly conduct research on the basis of agency theory, signal transmission theory, stakeholder theory and resource dependence theory. Rare research is conducted on the basis of rent-seeking theory. Based on the above research opportunities, this paper attempts to study the influence of the nature of controlling shareholders on corporate anti-corruption disclosure and the mediating role of political connections between the two on the basis of rent-seeking theory, making up for the shortcomings of existing research.

Corporate social responsibility covers a wide range of topics, including economic responsibility, environmental responsibility, product responsibility, community responsibility, etc. And it involves different interest groups (Freeman & Reed, 1983). Corruption means the use of public power for private gains (Heidenheimer, 2008), or the exchange of political power for wealth (Huntington, 1968). The concept of corruption has its broad sense and narrow sense. In a narrow sense, corruption refers to the act of government officials abusing power for personal gains. In a broad sense, corruption refers to the abuse of power by all behavior bodies in their own interests. Corruption in this paper is used in a broad sense, whether it is a public sector or a private one. Corruption and bribery not only endanger the interests of shareholders, but also damage a firm’s reputation and image, even undermine the market competition order in the long run. Thus, corporate corruption and bribery governance should be a key issue disclosed in corporate social report.

Traditionally, China is a country where interpersonal relationship and social networks play an important part, leading to serious corruption problems (Jin & Zhao, 2010). According to Transparency International, China scored only 40 points, 43 points below the average, ranking 79th out of 178 countries or regions surveyed, far below neighboring countries such as Singapore (No.7), Japan (No.20) and South Korea (No.52). It is a long way to go to for China to curb corruption and improve information transparency of government and enterprises. Whereas, when enterprises disclose social responsibility information, they often ignore the disclosure of corruption and bribery governance information, intentionally or unintentionally. Research findings on corporate ACPD are also rare. In order to make up for this gap, this paper attempts to study the motives of anti-corruption disclosure in China on the basis of rentseeking theory, complementing the extant research of social disclosure in emerging countries.

Taking China’s A-share listed firms which published standalone corporate social report in 2015 as a sample, this paper employs the motives of corporate anti-corruption disclosure. The findings show that the possibility of stateowned firms disclosing anti-corruption information is significantly greater than that of non-state-owned holdings. A big difference exists in the probability of anti-corruption disclosure by state-owned enterprises (SOE) at different government levels. Central enterprises disclose significantly more ACPD than local state-owned enterprises. Furthermore, the study divides the samples into two groups according to political background of the chairman or CEO of the firm, and tests the mediating role of political background of chairman or CEO on the relationship of the nature of controlling shareholders and corporate ACPD. The empirical results show that political background has a negative impact on state-owned corporates’ ACPD, which indicates that political connection may allure corporate corruption and bribery, which is harmful to corporate governance.

This paper may contribute in the following aspects: First, the existing literature mostly examines the impact of the nature of controlling shareholders on the overall level of corporate social disclosure, and less studies the current situation and motives of corporate ACPD (Dissanayake et al., 2011; Joseph et al., 2016). This study complements the literature by empirically testing the relationship between the nature of controlling shareholders and corporate ACPD, exploring the motives of ACPD of Chinese firms. Second, the previous literature on corporate social disclosure is mostly based on the agency theory, legitimacy theory, new institutionalism theory or stakeholder theory. However, considering the history and reality of China, this paper examines the impact of the nature of controlling shareholders and political background on corporate ACPD on the basis of rent-seeking theory of new political economy, providing a new perspective for studying firms’ behavior in a transitional economy.

2. Theoretical Background and Hypothesis Development

As a transitional economy, China’s construction of various systems is relatively backward. Government intervenes too much in the operation, financing and investment of enterprises. The relationship between government and enterprise is distorted. Many resources are controlled in the hands of the government, with key areas or industries monopolized by the government. China’s economic structure presents the characteristics of "dual ownership structure". Because of the natural connection between state-owned enterprises and the government, it is easier for state-owned enterprises to obtain economic resources than private ones. Private enterprises are at a competitive disadvantage. In order to obtain the benefits of financial subsidies, land use rights and tax incentives, private enterprises are more inclined to engage in rent-seeking activities (or bribery instead). These rent-seeking activities are essentially corruption in general. The serious corruption problems in private enterprises make them unwilling to disclose anti-corruption information.

Furthermore, the primary goal of private firms is to maximize economic benefits. So when making decisions upon whether and how to disclose, private firms tend to consider the cost and benefits of disclosure. Only when disclosure brings more benefits than cost will they choose to disclose. In China's capital market, information users lack attention to social responsibility information (Chen & Ma, 2005). Private firms lack motivation and pressure to disclose social information, including anti-corruption practice information. The objectives of state-owned enterprises, however, are diversified. In addition to maintaining and increasing the value of state-owned assets, state-owned enterprises also bear a variety of policy burdens, such as solving employment problems and realizing the political goals of the government (Liao & Shen, 2014). The government, the media and the public all have higher expectations for state-owned enterprises to undertake social responsibility, so state-owned enterprises are facing greater impetus and pressure to disclose social information (Yin, Wang, & Liu, 2014). In addition, the management of stateowned enterprises holds fewer shares than that of private ones. Thus, the cost of information disclosure is mainly borne by the enterprise not by the management. When making decisions on disclosure, the management of SOE seldom considers the cost of disclosure. Compared with private enterprises, the level of voluntary information disclosure of state-owned enterprises is often higher.

Since the 18th National Congress of the Communist Party of China, the Party Central Committee has promoted the party's work style and clean government in an unprecedented way, insisting on strictly administering the party. From implementing the "eight regulations" of the central government to the central patrol for full coverage, anti-corruption has been advancing into the new normal. In this context, anti-corruption has become a hot issue in political life. The executives of SOE are mostly appointed by the government and belong to state cadres. Their personal goals are not only to obtain economic benefits, but also to achieve promotion on the official stage (Li, 2012). In the current political environment, anti-corruption effectiveness has become a key indicator for assessing local officials. In order to help superiors to achieve anti-corruption performance and gain personal promotion, executives of SOE have greater incentives to promote anti-corruption construction and disclose anti-corruption information within the enterprise. In other words, anti-corruption practice and disclosure are essentially manifestations of SOE executives' political rent-seeking. Central enterprises are directly under the jurisdiction of the central government, and the government's intervention is stronger. In the anti-corruption campaign launched by the central government, central enterprises cooperate more actively and have stronger motivation for political rent-seeking. Therefore, the probability for SOE to disclose anti-corruption information may be greater.

Taking into account all the above, we put forwards the following hypothesis:

H1a: The probability of anti-corruption information disclosed by state-owned enterprises tends to be higher than that disclosed by non-state-owned enterprises.

H1b: The probability of anti-corruption information disclosed by central enterprises tends to be higher than that disclosed by local state-owned enterprises.

The political connection of state-owned enterprises is not established actively for the purpose of maximizing profits. It is usually gained through appointment by higher-level government departments for the consideration of social or political goals. Top executives of SOE with political background tend to care more about their political performance than economic performance (Zhang & Fang, 2013). When anti-corruption becomes a performance indicator for government officials, politically motivated executives of SOE have more incentives to disclose ACPD in order to win the promotion tournament. On the other hand, top executives of SOE with political background can use their power to seek rent in the officialdom, because their social network is made up of "their own people" and it is easy to "coordinate" and "communicate". The social network has virtually become the "umbrella" of rent-seekers (Wang, 2011), facilitating for them to embezzle state-owned assets for their own use. Furthermore, once the information is disclosed to the public, it will be subject to the supervision of the public. Once the false information is found, it will directly affect the reputation and interests of the information discloser. Given the risks and hidden costs of information disclosure, executives with political backgrounds may choose not to disclose. Previous studies have also shown that the probability of CSR disclosure in state-owned holding companies with political background is low (Wang, Lin, & Yu, 2013) and the quality of disclosure is low (Si, 2013). And they tend to disclose less voluntary internal control information (Zhang & Fang, 2013). Considering the above, whether the state-owned holding companies with political background will disclose more ACPD becomes a question to be tested by empirical evidence. Based on this, the following hypothesis is put forward:

H2: The probability of anti-corruption information disclosed by SOE with political background tends to be higher than that disclosed by SOE without political background.

3. Research Method

Taking China’s A-share listed firms which issued standalone corporate social responsibility report in 2015 as a sample, this paper empirically examines the relationship between the nature of controlling shareholders and corporate ACPD as well as the mediating effect of political background of the chairman or CEO of the firm on the relationship between the two. The specific research design is as follows:

3.1. Sample Selection and Data Source

We use a content analysis method to study anti-corruption information disclosed in corporate social responsibility reports. Content analysis is a method often used in the research of corporate voluntary disclosure. Our data comes from Juchao website, a service utility which aims to provide timely, accurate and complete information disclosure services of Chinese firms for capital market participants, especially small and medium investors. We use "2015 Social Responsibility Report", "2015 Corporate Citizenship Report" and "2015 Sustainability Report" as keywords to search for the 2015 CSRR released in 2016. Our search period ranges from January 1, 2016 to December 31, 2016. After deleting the reports issued by B-share firms (because A-share firms and B-share firms face different institutional context), we get 724 sample firms. Then, the following screenings are made: (1) deleting firms with incomplete financial data; (2) deleting firms with less than five firms in the same industry; and (3) deleting firms that are insolvent. Finally, 703 observations remain. The anti-corruption disclosure data used in this paper and social responsibility committee data are manually collected by reading the reports. The nature of controlling shareholders, political background and control variables data are collected from CSMAR Economic and Financial Research database. The CSMAR Economic and Financial Research Database is a professional database in China, drawing on the professional standards of the University of Chicago CRSP, Standard & Poor's Compustat, NYSE TAQ, I/B/E/S, Thomson and other internationally renowned databases, combined with the actual situation in China.

3.2. Models and Variables

To statistically test the relationship between the nature of controlling shareholders and corporate ACPD as depicted in H1a, this paper designs the following model:

\(\begin{aligned} A C P D_{i}=& \beta_{0}+\beta_{1} \text { State }_{i}+\beta_{2} \text { Size }_{i}+\beta_{3} \text { Board }_{i}+\\ & \beta_{4} \text { Dual }_{i}+\beta_{5} \text { List }_{i}+\beta_{6} G R I_{i}+\\ & \beta_{7} \text { CSR Committee }_{i}+\sum \text { Industry }+\varepsilon \end{aligned}\)       (1)

ACPD_i represents corporate anti-corruption practice disclosure, which is the dependent variable of this paper; ܵState_i represents the nature of controlling shareholders, which is the independent variable of this paper; other variables are control variables that may affect the disclosure decision of a firm.

In order to test whether there is a difference in ACPD between state-owned enterprises of different levels (H1b), model (2) is designed as follows:

\(\begin{aligned} A C P D_{i}=& \beta_{0}+\beta_{1} \text { Central }_{i}+\beta_{2} \text { Local}_{i}+\beta_{3} \text { Size }_{i}+\\ & \beta_{4} \text { Board }_{i}+\beta_{5} \text { Dual}_{l}+\beta_{6} \text { List }_{i}+\beta_{7} G R I_{i}+\\ & \beta_{8} \text { CSR Committee }_{i}+\sum \text { Industry }+\varepsilon \end{aligned}\)       (2)

On the basis of model (1), in model (2) state-owned enterprises are further divided into central enterprises and local state-owned enterprises according to controlling shareholdings. If the coefficients of Central_i and Local_i are greater than 0, it means that central enterprises and local state-owned enterprises are more inclined to disclose ACPD than private enterprises. If the coefficient of Central_i is greater than that of Local_i ,it means that central enterprises are more inclined to disclose ACPD than local state-owned enterprises.

In order to test H2 and further verify the mediating effect of political background on the relationship between the nature of controlling shareholders and corporate ACPD, this paper sets up models (3) and models (4):

\(\begin{aligned} A C P D_{i}=& \beta_{0}+\beta_{1} \text { State }_{i}+\beta_{2} \text { State }_{i} * \text { Connection}_{i}+\\ & \beta_{3} \text { Size }_{i}+\beta_{4} \text { Board }_{i}+\beta_{5} \text { Dual }_{i}+\beta_{6} \text { List }_{i}+\\ & \beta_{7} G R I_{i}+\beta_{8} \text { CSR Committee }_{i}+\sum \text { Industry }+\varepsilon \end{aligned}\)       (3)
 

\(\begin{aligned} A C P D_{i}=& \beta_{0}+\beta_{1} \text { Central}_{i}+\beta_{2} \text { Local}_{i}+\beta_{3} \text { Central}_{i} * \\ & \text { Connection }_{i}+\beta_{4} \text { Local }_{i} \text { * Connection }_{i}+\beta_{5} \text { Size }_{i}+ \\ & \beta_{6} \text { Board }_{i}+\beta_{7} \text { Dual}_{i}+\beta_{8} \text { List }_{i}+\beta_{9} G R I_{i}+\\ & \beta_{10} \text { CSR Committee }_{i}+\sum \text { Industry }+\varepsilon \end{aligned}\)       (4)

On the basis of model (1), in Model (3) we introduce the interaction terms of controlling shareholders nature and political background. On the basis of model (2), in Model (4) we introduce the interaction terms of different government background levels and political background (\(\text { Central }_{i} * \text { Connection}_{i}\) and \(\text { Local }_{i} * \text { Connection }_{i}\)). In model (3), if the coefficient of \(\text { State }_{i} * \text { Connection }_{i}\) is greater than 0, state-owned enterprises with political background may have a greater probability of disclosing anti-corruption information, and vice versa. In model (4), if the coefficient of \(\text { Central }_{i} * \text { Connection}_{i}\) is greater than 0, it shows that the probability of anti-corruption disclosure of central enterprises with political background is greater, and vice versa. If the coefficient of \(\text { Local }_{i} * \text { Connection }_{i}\) is greater than 0, it shows that the probability of anti-corruption disclosure of local state-owned enterprises with political background is greater, and vice versa.

Furthermore, this paper divides the sample firms into two groups according to the nature of controlling shareholder (state-owned group and non-state-owned group) to run a regression on political background and ACPD.

The purpose of this research is to examine the influence of the nature of controlling shareholders and the political background of the chairman or CEO on corporate ACPD decision. Therefore, the explained variable is corporate ACPD. Draw on the research of Dissanayake et al. (2011) and Joseph et al. (2016), we classified corporate ACPD into the following categories: anti-bribery policies and procedures, commitment of the board of directors and executives to anti-corruption and anti-bribery, anti-bribery human resources construction, responsible commercial operations (i.e. anti-corruption in the supply chain), external assurance of anti-bribery policies and procedures, establishment and implementation of anti-corruption codes of conduct, and reporting of violations. As long as the sample firm discloses one of these items in the social report, ACPD of the firm will take 1, otherwise it is 0. The explanatory variables in the model include:

(1) Nature of controlling shareholders (State). If the sample firm is state-owned, this variable takes 1, otherwise it is 0. In addition, the sample firms are further divided into central ones (Central) and local state-owned ones (Local) according to the level of state-owned controlling shareholders.

(2) Political background (Connection). Political background is measured by whether the chairman or CEO of the sample firm has a political background. If the chairman or CEO is a government official, a deputy to the National People's Congress, a member of the Chinese People's Political Consultative Conference, or a party representative, it will be defined as a firm with a political background and the variable Connection will take 1, otherwise it is 0.

In terms of control variables, we select size of the firm, board of directors, CEO duality, listings places, reference to GRI or not (GRI), and whether to set up a social responsibility committee. We also control the impact of industry in the model(Cowen, Ferreri, & Parker, 1987; Said, Zainuddin, & Haron, 2009; Giannarakis, 2014; FernandezFeijoo et al., 2014). All variables are shown in Table 1. The financial data in the paper is drawn from China Stock Market & Accounting Research Database (CSMAR database). Data processing is completed by Stata12.0 software.

Table 1: Definition of Major Variables

4. Empirical Results

4.1. Descriptive Statistics

Table 2 reports descriptive statistics for the main variables. It can be seen from the table that there are 391 out of 703 firms disclosed ACPD, accounting for 55.62% of the total sample size. 59.03% of the sample firms are state-owned, of which about one-third are central enterprises. The proportion of firms with political background is low, accounting for only 20.48% of the total. Most firms implement separation between the chairman and CEO. In the preparation of social responsibility reports, 20.63% refer to G3 standard or G4 standard issued by the Global Reporting Initiative (GRI). Only 64 firms have set up social responsibility committees or social responsibility leading groups, less than 10% of the total sample size.

Table 2: Descriptive Statistics of Main Variables

4.2. Pearson Correlation

Table 3 provides the results of correlation analysis between main variables. As can be seen from the table, State and ACPD are significantly positively correlated at the 5% level, indicating that state-owned firms are more likely to disclose anti-corruption information. Central and ACPD are significantly positively correlated at the 5% level, while the coefficients of Local and ACPD is not significant. This indicates that central enterprises are more likely to disclose ACPD, while local state-owned enterprises and private enterprises exhibit little difference in ACPD, supporting both H1a and H1b. The coefficient of Connection and ACPD is not significant, indicating whether a firm has political background or not does not impact its anti-corruption disclosure decision. This may be due to the correlation between Connection and other variables, so it needs to be further verified by multiple regression analysis. In terms of control variables, Size is significantly positively correlated with ACPD at the 5% level, while Dual is negatively correlated with ACPD. List, GRI, and CSR Committee are significantly positively correlated with ACPD at the 1% level. This means that large firms, firms with non-duality, firms listed at Shenzhen Stock Exchanges, firms which produce social responsibility reports with reference to GRI standards, and firms with social responsibility committees are more likely to disclose anti-corruption information. The correlation coefficients between the explanatory variables and control variables are relatively small, which means that our models do not have serious multicollinearity.

Table 3: Pearson Correlation Coefficient

Note: ***, **, and * indicate significant levels at 1%, 5%, and 10%, respectively.

4.3. Regression Analysis

Table 4 reports the results of Logitistic regression analysis of the nature of controlling shareholders, political background on corporate ACPD. The regression results for Model (1) show that State and ACPD are significantly positively correlated at the 5% level, which indicates that state-owned enterprises are more inclined to disclose anticorruption and anti-bribery information than private enterprises. As for the regression results of Model (2), both Central and Local are significantly positively correlated with ACPD, and the coefficient of Central is greater than that of Local, indicating that central enterprises are most likely to disclose ACPD, followed by local state-owned enterprises and private firms have the lowest probability of disclosure. Model (3) adds the interaction term between State and Connection on the basis of Model (1), exploring the mediating effect of political background on the relationship between the nature of controlling shareholders and corporate ACPD. The regression results show that State is positively correlated with ACPD at the 1% level. Compared with Model (1), the level of significance is improved. State*Connection and ACPD are significantly negatively correlated, indicating that state-owned enterprises with political background are less likely to disclose ACPD. This may be due to the fact that political connections are prone to corruption and corruption decreased corporate transparency. Model (4) adds the interaction term between Central and Connection and the interaction term between Local and Connection on the basis of Model (2). The regression results show that the coefficients of Central and Local are both positive at the 5% level but the coefficients of Central*Connection is not significant statistically. The coefficient of Local*Connection is significantly negative, which indicates that political background mainly impacts the disclosure decision of local state-owned firms rather than central firms.

Table 4: Logit Regression Result

Note: ***, **, and * indicate significant levels at 1%, 5%, and 10%, respectively.​​​​​​​

To further verify the influence of the nature of controlling shareholders and political background of the chairman or CEO on corporate ACPD, we grouped the sample firms into two groups according to the nature of controlling shareholders and then the political background. Logit regression analysis is performed afterwards. The results are shown in Table 5 and Table 6. Table 5 is a group test based on the nature of controlling shareholders, and Table 6 is based on the political background. As can be seen from Table 5, as for the political-background group, State and ACPD are significantly positively correlated at the level of 1%, while for non-political-background group, the coefficient of State is not significant. This indicates that political background does impact the disclosure decision of state-owned enterprises.

Table 5: Group Test by Political Background

Note: ***, **, and * indicate significant levels at 1%, 5%, and 10%, respectively.

Table 6 shows that political background is negatively correlated with ACPD to the state-owned group, while to the non-state-owned group, the coefficient of Connection is not significant. That is to say, the negative impact of political background on corporate ACPD is only found in stateowned firms, supporting H2.

Table 6: Group Test by the Nature of Controlling Shareholders

Note: ***, **, and * indicate significant levels at 1%, 5%, and 10%, respectively.​​​​​​​

4.4. Robust Test

In order to verify the robustness of our research, the following sensitivity tests are conducted: First, the measurement of the dependent variable is changed. We count the number of sentences of anti-corruption information disclosed in CSRR by each firm. If the number of sentences of anti-corruption information disclosed by the firm exceeds the median of the industry, it is considered to be a highdisclosure firm; otherwise it is considered to be a lowdisclosure firm. We assign 1 to high-disclosure firms and 0 to low-disclosure firms. Substituting this dummy variable into the model (1)-(4), we have re-run the Logit regression. The results are shown in Table 7, which are consistent with the previous results and the significance level of the coefficient of Local*Connection even improved.

Table 7: Robust Test (1)

Note: ***, **, and * indicate significant levels at 1%, 5%, and 10%, respectively. Considering that some industries have serious multicollinearity when industry variables are controlled, the sample size decreases after the firms in these industries are deleted.​​​​​​​

Second, considering that the disclosure decision on corporate social responsibility information by firms listed at Hong Kong Stock Exchange is subject to different reporting regulations. To eliminate this impact, we deleted those firms listed at Hong Kong Stock Exchange from the sample. With the remaining 622 firms, we re-ran the regression model. The results are shown in Table 8. It can be seen from Table 8 that in Model (1) and Model (2), the coefficients of State are significantly positive at the level of 5% and 1%, respectively, and the coefficients of State*Connection are significantly negative at the level of 1%, supporting hypotheses H1a and H2b.

Table 8: Robust Test (2)

Note: ***, **, and * indicate significant levels at 1%, 5%, and 10%, respectively.​​​​​​​

5. Conclusions

The frequent exposure of official corruption scandals in recent years has caused global concern. Regional subcultures in East Asia are particularly tolerant of corruption, leading to frequent reports of corruption scandals in East Asia. What are the causes of the frequent occurrence of corruption and how to resolve them have become hot topics for discussion by the government, academia, media and public. In view of the causes of corruption of Chinese officials, scholars hold different ideas, but most believe that improper and distorted governmententerprise relation is one of the root causes. Differences between Chinese mode and western social development mode are obvious. China adopts a mixed ownership system dominated by the public economy. In terms of resource allocation, government regulation and intervention are much larger than in the West. Therefore, under such a special political and economic context, it is of practical significance to study the impact of government intervention on the microbehavior of enterprises.

Taking China’s A-share listed firms that issued standalone CSRR in 2015 as a sample, this paper empirically tests the relationship between the nature of controlling shareholders and the decision-making of corporate ACPD and the mediating effect of political background on the relationship between them. The study finds that currently China is vigorously promoting the anti-corruption construction, from the central government to the local government, with anticorruption effectiveness included in the key performance indicators for assessing local officials. Under such a context, in order to help superiors achieve anti-corruption goals and finally obtain personal promotion, the executives of stateowned enterprises in China show great enthusiasm in carrying out anti-corruption initiatives and disclosing anticorruption practice information. Compared with that of private enterprises, the possibility of anti-corruption practice disclosure of state-owned enterprises is significantly higher. Anti-corruption performance and disclosure have become means of political rent seeking for SOE executives. Central enterprises are under the direct jurisdiction of the central government, and government intervention is stronger. In the anti-corruption campaign launched by the central government, central enterprises are more willing to cooperate, and the probability of anti-corruption practice disclosure is probably higher.

In addition, the study shows that political background inhibits the anti-corruption disclosure of SOE, but has no significant impact on the decision-making of ACPD of private firms. The reason may lie in the fact that the maintenance of political connections of SOE executives often implies corruption and bribery, such as hospitality and gifts. The existence of corruption and bribery prevents the disclosure of ACPD. To guarantee the reliability of this research, we did the following robust tests: changing the measurement of the dependent variable (measured by the sentences of ACPD) and excluding firms listed at Hong Kong Stock Exchange. The results of this study remain unchanged, which shows that the conclusions of this research are robust.

Our study provides a useful addition to the existing literature on corporate voluntary disclosure. As a link between the company and the capital market as well as information users, the quality of information disclosure has a direct impact on the interests of information users, which is a key issue affecting the healthy development of the capital market. Information disclosure is cost-effective. When deciding whether and how to disclose, management should take into account the needs of stakeholders. In the anticorruption campaign launched by the central government of China, state-owned enterprises are facing unprecedented pressure on legitimacy. Anti-corruption practice disclosure has become a strategic tool for SOE management to cope with public pressure. This paper provides evidence for the legitimacy theory and stakeholder theory in the field of corporate voluntary disclosure by studying the impact of macro-environment on micro-governance of enterprises.

This research is supplementary to Dissanayake et al. (2011) and Joseph et al. (2016). Joseph et al. (2016) conducted a statistical analysis of corporate ACPD in Malaysia and Indonesia. Dissanayake et al. (2011) employed a case study to investigate the motivations of corporate ACPD on the basis of media agenda-setting theory. In this research, we use a large sample to empirically test the relationship between the nature of controlling shareholders and corporate ACPD and the mediating role of political connections in the relationship between the two data on the basis of rent-seeking theory, enriching the literature in the field of corporate social responsibility disclosure. As governments, NGOs, the media, and the public are paying increasingly attention to corporate social responsibility, more and more companies choose to issue social reports and even upgrade their social responsibility policies to strategic level. Corporate social responsibility includes responsibility for shareholders, creditors, employees, suppliers, communities, etc. Corruption and bribery will decrease the values of shareholders and damage the sustainable development of enterprises, so it should be incorporated into corporate social responsibility. The globally accepted social responsibility report guidelines GRI also incorporate anticorruption and anti-bribery information into key performance indicators. However, practically, many companies still miss anti-corruption and anti-bribery disclosure in their social reports, intentionally or accidentally. Once the report is issued, it must be supervised by the public. Therefore, improving anti-corruption and anti-bribery disclosure will surely help prevent and curb corruption and bribery.

The findings of this research have some useful insights in corruption governance. Government intervention can, to some extent, solve the market failure caused by information asymmetry by influencing the allocation of resources. In order to curb corruption and bribery, the government should exert some legitimacy pressure on enterprises, strengthen the supervision on corporate disclosure, and improve the information transparency. But too much government intervention, especially government officials holding a post in corporates, is likely to breed corruption. Government officials are executives of enterprises. Political connection, i.e. government officials holding a post in corporates, is an implicit contract formed between corporates and the government. The establishment of "relationship" can win important resources for corporates which are controlled by the government. But it also implies corruption which is the so-called "power and money transaction". So to solve the problem of corruption, it is necessary to separate the government from enterprises, to introduce competition mechanism, and to promote information transparency.

This research has some limits. First, we take only crosssectional data as our sample, and do not consider the time trend of corporate ACPD. Future research can examine the trend of corporate ACPD through a long-term longitudinal study. Second, we conduct this research in the context of China, a special and transitional economy. Corporates in other countries may operate in a different political, economic and cultural context, thus exhibiting different disclosure preferences. Future research can be carried out in other countries to study the impact of legal system, culture and investor protection on corporate ACPD. Third, we employ a content analysis to study whether corporate disclose ACPD or not and the amount of disclosure. The quality of disclosure is not considered. Even if a firm chooses to disclose, there is a big difference in the content of disclosure and the quality of disclosure. Generally speaking, the quality of quantitative disclosure is higher than that of narrative description. As for disclosure content, there are also significant differences between those who only disclose visions and goals, and those who disclose visions and goals, specific measures and performance indicators at the same time. Future research can study the factors and motivations that affect the quality of anti-corruption disclosure by assigning different weights on different quality ACPD and constructing ACPD index.