Government ownership, corporate governance and tax aggressiveness: evidence from China
Abstract
This study investigates how government ownership and corporate governance influence a firm's tax aggressiveness. Using Chinese listed companies during 2003–2009, we find that compared with government-controlled firms, non-government-controlled firms pursue a more aggressive tax strategy. In particular, non-government-controlled firms with a higher percentage of the board shareholdings and with a CEO who also serves as the board chairman are more aggressive. For government-controlled firms, we find that board shareholding has an impact on tax aggressiveness and it does not differ between local and central government-controlled firms. However, local government-controlled firms in less developed regions where the implementation of corporate governance measures is generally less effective are more tax aggressive than those in other regions.
1. Introduction
This study examines how government ownership and corporate governance affect a firm's tax aggressiveness in China.1 We give particular attention to the influence of government ownership on tax aggressiveness because government control of corporations is pervasive in most transitional economies (Chen et al., 2009). In recent years, many governments have relinquished control of a segment of their economies through privatization programmes. However, these governments or their agencies still maintain control of a significant number of companies especially those in key industries such as telecommunication and defence (Krivogorsky, 2000; Chen et al., 2009). There have been recent calls for evidence on the role that government ownership plays in corporate tax aggressiveness (e.g. Hanlon and Heitzman, 2010). As government-controlled and non-government-controlled companies have different stakeholders and governance practices, it will be interesting to examine how they may differ in their tax strategies.
Recent studies have investigated how state ownership affects firms' tax reporting practices in China. Zeng (2011), for example, finds that government-controlled firms are less tax aggressive compared with non-government-controlled firms. Wu et al. (2013) also provide evidence that local state-owned enterprises (SOEs) pay a higher level of tax than private firms. However, these studies do not examine the effect of corporate governance on firms' tax strategies. On the other hand, there are studies that provided evidence on the linkages between corporate governance and a company's tax avoidance in the West. For example, Minnick and Noga (2010) find that corporate governance plays a role in the long-term tax management of US firms. In particular, compensation incentive for directors is an important driver for tax management. In addition, Lanis and Richardson (2011) find that having a higher proportion of independent directors on the board reduces the likelihood of tax aggressiveness in Australia. However, these studies do not deal with the effect of ownership structure on tax reporting. In this study, we examine the impact of government ownership, corporate governance and their interactive effect on a firm's tax reporting. Specifically, we focus on several specific board characteristics which are indicative of corporate governance and compare their impacts on the tax aggressiveness for government versus non-government-controlled firms.
We take advantage of the institutional setting in China to develop our hypotheses. First, for many Chinese listed companies, the ultimate controlling shareholder is the Chinese government, which influences company management through its shareholding and political power (Liu and Lu, 2007; Lo et al., 2010). As such, managers of government-controlled firms may have different tax incentives compared with their counterparts in the private sector. On the one hand, as they are appointed and evaluated by the government owners, they may be eager to protect government revenues by avoiding aggressive tax planning because such tax planning will raise the after-tax profits that other shareholders keep at the expense of the government (e.g. Crocker and Slemrod, 2005; Zhang, 2006). On the other hand, managers of government-controlled firms may want to maximize corporate resources under their control through aggressive tax planning and to divert such resources for individual gains as their salary and compensation level is quite low compared with the private sector (e.g. Bushman et al., 2004; Wang et al., 2008; Guedhami et al., 2009). In this study, we provide empirical evidence to shed light on the dominance of these two competing incentives in China's setting.
Second, Chinese stock market provides a high-power context for our research as under-developed institutional infrastructure and corporate governance leave minority shareholders vulnerable to tunnelling (Jiang et al., 2010). Third, prior studies provide mixed and inconclusive results on the impact of board characteristics on firm performance and earnings management in China (Chen et al., 2006; Liu and Lu, 2007; Lo et al., 2010). While Minnick and Noga (2010) find that, other than compensation contracts, board characteristics do not influence a company's tax management in the United States, how corporate governance in China affects tax aggressiveness is yet to be explored.
Using data from all non-financial A-share companies listed in China during 2003–2009, we examine how ownership type and board characteristics affect tax aggressiveness. Overall, our results show that government ownership and corporate governance play an influential role in tax management in China. Compared with government-controlled firms, non-government-controlled firms are more tax aggressive. The finding suggests that managers of government-controlled firms are more eager to achieve the political objective of protecting government revenue in their tax strategy. For board characteristics, we find that firms with a higher percentage of board shareholdings and having a CEO who also serves as the board chairman are more tax aggressive. The effect of duality role of board chairman is primarily driven by non-government-controlled firms, and there is an absence of such a relationship for government-controlled firms. Furthermore, the proportion of independent directors on the board does not have any effect on tax aggressiveness for all firms. Our additional analysis of government-controlled firms shows that local government-controlled firms located in less developed regions with generally less effective implementation of corporate governance measures are more tax aggressive than those in more developed regions. We do not find regional differences for other government or non-government-controlled firms.
This study contributes to the extant literature by providing empirical evidence on the influence of government ownership, corporate governance and their interactions on tax aggressiveness. Extant studies on China document that bureaucrats in SOEs are detrimental to shareholders in terms of firm performance (e.g. Fan et al., 2007). Our study provides evidence that managers of government-controlled firms are more eager to fulfil their political objectives of protecting government revenue than to exploit complex tax planning. In addition, prior literature has not compared the impact of board effectiveness on tax aggressiveness for firms with different ownership type. This study fills this gap in the literature by providing a direct examination of the impact of board effectiveness on tax aggressiveness of government versus non-government-controlled firms. Government ownership of corporations is traditionally a typical phenomenon in most transitional economies (Krivogorsky, 2000; Chen et al., 2009), and as the trend to privatize continues in these economies, our research results should serve as a useful reference for them to prepare for ways to improve tax compliance.
The rest of the study proceeds as follows. Section 2 discusses the relevant institutional background in China. Section 3 reviews the related literature and develops the hypotheses. Section 4 describes the data, research methodology and regression model. Section 5 presents the results and sensitivity tests, and Section 6 concludes.
2. Institutional background
2.1. Government ownership in China
The Chinese economy had been dominated by SOEs. The ownership structure of SOEs in China is in the form of pyramid holdings held primarily by the government (Liu and Sun, 2005). This highly concentrated government ownership often leads to inefficient operations in the SOEs as their main objective was to carry out the government's political agenda rather than to maximize shareholders' wealth. To improve the efficiency and hence the economic performance of the firms, the Chinese government has undertaken measures to reduce its holdings in listed companies through the sale of state shares to other groups of investors such as institutional and individual investors (Chan et al., 2007). This ownership structure reform aims at transforming the SOEs to give more autonomy to managers to run the firms with reduced government influence and to make them more responsive to the market and investors. However, for the majority of listed firms in China, the government still maintains a controlling ownership, particularly in important industries such as banking, energy, transportation, natural resources and telecommunication (Liu and Lu, 2007).
Compared with the diffuse ownership structure in most Western countries, the ownership of many Chinese listed companies is concentrated in the hands of the government through state-owned shares or legal-person shares (indirectly owned by the government). For these firms, power is concentrated and power distance is large (Chan et al., 2003). Very often, it is the government that appoints, motivates and disciplines managers, and investors in government-controlled firms are largely outsiders as the government seldom appoints directors representing minority shareholders (Zhang, 2006; Fan et al., 2007). The government pursues its own objectives and often limits the firm's ability to maximize wealth at the expense of other stakeholders. Thus, the agency problem for government-controlled firms can be quite different from non-government-controlled firms.
2.2. Corporate governance in China
An important milestone for the development of corporate governance in China was the issuance of the Code of Corporate Governance for Listed Companies in China (the ‘Code’) by China Securities Regulatory Commission (CSRC) in 2001. The Code sets up the principles on investor protection and the code of conduct for managers, directors as well as supervisors. Other regulations that relate to corporate governance include the Company Law and the Establishment of Independent Directors Systems by Publicly Listed Companies Guiding Opinion (the ‘Guiding Opinion’). However, taken together, the corporate governance in Chinese listed companies is still under-developed. For example, several characteristics of the board of directors of Chinese companies indicate potential problems. First, many independent directors have close relationships with company owners. Although this facilitates information sharing, it raises questions on the board's independence (Su, 2010). Second, the duality role of the board chairman is not prohibited in China. A firm's CEO can also serve as the chairman of the board. This may weaken the supervisory function of the board. Third, the board of directors of government-controlled firms often consists of representatives or officials from the government and other state entities, whose interests and motivations are not the same as those of outside investors (Tenev and Zhang, 2002; Su, 2005).
3. Literature review and hypothesis development
3.1 Government ownership and tax aggressiveness
Prior studies provide evidence on the association between ownership structure and tax aggressiveness. For example, Desai and Dharmapala (2008) find that firms with concentrated ownership have greater incentives to avoid taxes as they have lower non-tax costs. However, Khurana and Moser (2009) find that firms with higher levels of long-term institutional ownership are less tax aggressive because institutional owners are more concerned with the long-term consequences of aggressive tax strategy. Similarly, Chen et al. (2010) find that family-owned firms are less tax aggressive. In this study, we focus on the effect of government ownership on tax aggressiveness.
In China, as discussed earlier, the economy had been dominated by SOEs. Although the Chinese government has recently relinquished some controls in SOEs via privatization programmes, the governments (central and local) still maintain as a controlling shareholder in a large number of corporations (Liu and Lu, 2007). Similar to institutional investors, government ownership is a long-term institutional ownership. In this sense, according to Khurana and Moser (2009), we expect government-controlled firms are less tax aggressive. In addition, managers of government-controlled firms, who are appointed by the government, are required to achieve political and social objectives, and some of them may actually be eager to be a ‘leader’ in paying taxes even at the expense of firm value (Shleifer and Vishny, 1994; Chang and Wong, 2004). Furthermore, a serious violation of tax laws not only involves penalty (which averages to 25 per cent of the additional tax for transfer pricing violations) (Ernst & Young, 2012), but can also jeopardize their political career (Cao and Dou, 2007). Besides, the government owners often have an incentive to ‘cash out’ the firm's earnings in the form of tax collection. In summary, managers of government-controlled firms should be eager to protect government revenues and thereby less aggressive in tax reporting.
On the other hand, according to recent evidence, some of these managers may have competing incentives to collude with insiders to divert corporate resources for individual gains (Bushman et al., 2004; Bushman and Piotroski, 2006; Wang et al., 2008; Guedhami et al., 2009). Prior research suggests that diversion activities are rather common in China (e.g. Jiang et al., 2010). To facilitate their diversionary practices, they can exploit complex tax planning to keep more resources in the firms or for tunnelling purposes (e.g. Desai and Dharmapala, 2006; Lo et al., 2010). In addition, the Chinese tax regulations stipulate that tax revenue collected from all local government-controlled firms and some central government-controlled firms must be shared among different layers of government. For example, for the corporate income tax paid by local government-controlled firms, the local government can take only 40 per cent of the tax revenue (State Council, 2001). Because the respective local government does not get 100 per cent of the tax dollar collected from its own enterprises, this may motivate the respective government to direct their firms to minimize tax payment to keep more resources in its controlled firms.
- H1 Ceteris paribus, government-controlled firms are less tax aggressive than non-government-controlled firms.
3.2 Corporate governance and tax aggressiveness
The board of directors plays an important role in a firm's corporate governance. Some recent studies provide evidence that board characteristics affect a firm's tax strategies (e.g. Minnick and Noga, 2010; Lanis and Richardson, 2011). In this study, we specifically examine the impact of board composition, duality role of the board chairman and board shareholdings on the tax aggressiveness of government versus non-government-controlled firms.
3.2.1 Board composition
The composition of the board of directors is an important determinant of its effectiveness, and there should be a proper mixture of insiders and outsiders on the board (Fama and Jensen, 1983). On the one hand, the board with more insiders may potentially face ‘independence’ problem because their monitoring role is diminished by engaging in self-reviewing activities. On the other hand, a large number of outsiders can create a wider gap between the board and the management as they have less insight on firm operations.
Prior studies generally find that independent directors have a positive impact on deterring earnings management. For example, Park and Shin (2004) find that independent directors with financial expertise are able to reduce earnings management. Similarly, Peasnell et al. (2005) find that firms with a high percentage of independent directors are less likely to engage in opportunistic earnings management. Kato and Long (2006) also assert that independent directors who are truly independent of the controlling shareholders have potential to improve the quality of corporate governance (which we argue should reduce tax aggressiveness). Uzun et al. (2004) find that firms with a high percentage of independent directors have less financial fraud because independent directors have fewer incentives for the firms to commit fraud. Based on a small sample of matched tax-aggressive and non-tax-aggressive Australian firms, Lanis and Richardson (2011) find that a higher proportion of independent directors reduces the likelihood of tax aggressiveness.
In the Chinese context, based on 169 fraud cases investigated by the CSRC, Chen et al. (2006) find that firms with a high proportion of non-executive directors on board are less likely to engage in security-related frauds. Similarly, Lo et al. (2010) provide evidence that firms with a board that has a high percentage of independent directors have a smaller magnitude of transfer pricing manipulations. These results appear to suggest that outside directors are able to monitor the actions of managers and deter fraudulent acts. Moreover, compared with executive directors, independent directors, given their social status, should be more concerned about the long-term reputational effect of management decisions. Tax avoidance activities are short-term opportunistic managerial behaviours, which could incur significant long-term costs. In particular, government appointed outside board members in government-controlled firms should be even less interested in tax aggressiveness as paying tax is an important political ‘duty’ for them in China.
On the other hand, many independent directors in China maintain close relationships with management and are often political allies or friends and relatives of the senior managers/owners. They are also in a weak position within Chinese boards, and they are more decorative than functional (Su, 2005). They do not want to antagonize senior management or major shareholders as they have the incentive to maintain their directorship and their director fees are in part dependent on the company's after-tax profit. Furthermore, corporate tax avoidance behaviours in China are generally conceived by shareholders as value-increasing, especially those of non-government-controlled firms. Many independent directors who act on behalf of the shareholders may not curb firms' aggressive tax strategy.
- H2a Ceteris paribus, there is a difference in tax aggressiveness between companies with different percentages of independent directors on board.
3.2.2 CEO as chairman of the board
The duties of the board chairman include the handling of board meetings and overseeing the hiring, termination, and compensation of the CEO and other senior executives (Jensen, 1993). The CEO, as the executive leader of a firm, is the final decision-maker in terms of entity operations. It is not uncommon that the chairman and CEO are the same person in the United States, while in most European countries, these two roles are most often separated (Lin and Liu, 2009). The different practices across countries reveal the costs and benefits of the chairman's duality role (Braun and Sharma, 2007). On the one hand, the combined leadership structure creates efficiency in decision-making and effective leadership in ensuring that firm strategy formulation and implementation by the CEO will be better coordinated (Chen et al., 2006). Moreover, the duality role of chairman can avoid potential rivalry between the CEO and chairperson and eliminate the confusion as a result of the existence of two spokespersons. On the other hand, agency theory suggests that more effective control over managers to align their interests to the shareholders will be better achieved by the separation of the CEO from the chairman. The co-services performed by the board chairman may impair effective monitoring and also hinder honest evaluation of firm performance by the board which in turn leads to long-term adverse consequences (Dalton and Kesner, 1997; McWilliams and Sen, 1997).
- H2b Ceteris paribus, companies with the same person serving as CEO and the board chairman are more tax aggressive, particularly for non-government-controlled companies.
3.2.3 Board shareholdings
Equity ownership by board members creates incentive for directors to protect their financial stake in the firm. Johnson et al. (1993) find that the more equity holdings by board members, the more their involvement in strategic restructuring. Morck et al. (1988) suggest a positive relation between board shareholdings and firm performance. Kren and Kerr (1997) find that a firm with a board of directors that has significant shareholdings exhibits a stronger linkage between firm performance and compensation. In this study, we analyse board shareholding instead of CEO shareholding because we want to address the influence of the board as a whole on tax reporting.
With board shareholdings, the interests of the board of directors are directly aligned with firm performance. Tax aggressiveness, which serves as a way to increase after-tax firm value, can help the board members enjoy the benefits of increased share values. Moreover, McWilliams and Sen (1997) show that when directors increase their ownership of the firm, the potential managerial entrenchment incentives increase. With more shareholdings, the directors have greater incentives to work to increase the value of the company. One possible method to increase firm value is through aggressive tax management.
- H2c Ceteris paribus, companies with a higher percentage of board shareholdings are more tax aggressive.
4. Research method
4.1 Data collection
Our sample consists of all A-share non-financial companies listed in Shanghai and Shenzhen Stock Exchanges for the period 2003–2009. These A-share companies are all domestic Chinese companies (i.e. they are not foreign investment enterprises). We chose the sample period because the Corporate Governance Code was effective from 2002, and the related controlling shareholder data are available from 2003 onwards. Consistent with prior studies (e.g. Gupta and Newberry, 1997; Frank et al., 2009; Chan et al., 2010), we excluded finance and insurance firms because of their special financial reporting requirements. We deleted 1454 observations with missing data and 1337 observations with insufficient data to calculate measures of tax aggressiveness (i.e. ETRs and alternative measures). Furthermore, following prior studies (Gupta and Newberry, 1997; Chen et al., 2010; Wu et al., 2013) and to eliminate outliers, we restricted observations with ETRs within the range of [0, 1]. The final sample consists of 6032 firm-year observations. We collect all financial data and board characteristics from China Stock Market and Accounting Research (CSMAR) database, which has been used by several recent studies (Chen et al., 2006; Lo et al., 2010). Panel A of Table 1 shows the summary of the sample selection process. Panel B reports the distribution of sample firms by year. The number of firms is almost evenly distributed over the 7-year sample period.
| Panel A: Sample selection | |
|---|---|
| Firm-year observations of all listed A-share non-financial companies between 2003 and 2009 | 10122 |
| Less observations with missing data | 1454 |
| Less observations with insufficient data to calculate ETR | 715 |
| Less observations with insufficient data to calculate RETR | 622 |
| Less observations with ETR < 0 or ETR > 1 | 1299 |
| Total number of observations for regression | 6032 |
| Panel B: Yearly distribution | |
|---|---|
| Years | No. of firm-years (%) |
| 2003 | 773 (13) |
| 2004 | 852 (14) |
| 2005 | 780 (13) |
| 2006 | 842 (14) |
| 2007 | 946 (16) |
| 2008 | 931 (15) |
| 2009 | 908 (15) |
| Total | 6032 (100) |
4.2 Model
(1)The effective tax rate, ETR, is a common measure of corporate tax aggressiveness in prior literature (e.g. Gupta and Newberry, 1997; Hanlon and Slemrod, 2009; Wilson, 2009; Chen et al., 2010). Following Wu et al. (2013), we define ETR as the ratio of the current portion of tax expense to adjusted taxable income.2 Given the variation in tax rates within China due to various tax preferential policies, a low ETR may be due to a preferential tax rate and therefore has little to do with tax reporting aggressiveness. To resolve this problem, we adjust the ETR by the applicable tax rate (ATR) of a company in a given year.3 We use the ratio of ETR to ATR (i.e. RETR) as the dependent variable in our main analysis. As the RETRs are bound to fall between 0 and 1, we estimate the model using a double-censored Tobit regression. We also estimate a separate regression for each ownership type.
The explanatory variable, Govt, is a dummy variable used to test the effect of government ownership on tax aggressiveness. Govt equals to 1 if the controlling shareholder is the Chinese government (central or local), and 0 otherwise. Controlling shareholder is defined according to Article 41 of the Guidelines for the Articles of Association of Listed Companies.4 As hypothesized in H1, a government-controlled firm has less incentive to engage in aggressive tax strategy. Therefore, we expect β1 to be positive. The Indep_director, Co_services and BOD_shares are variables for board characteristics. Indep_director denotes the percentage of independent directors served on the board. According to H2(a), we do not predict a sign for β2. Co_services is a dummy variable which equals to 1 if the board chairman also serves as a firm's CEO, and 0 otherwise. Consistent with H2(b), we predict that the duality role of the board chairman will lead to more tax aggressiveness and therefore β3 should be negative. BOD_shares measures the board shareholdings, which is the percentage of a firm's shares held by all directors of the board. As hypothesized in H2(c), we expect β4 to be negative.
Following prior literature, we include six control variables that are found to have influence on tax aggressiveness. LEV denotes a firm's capital structure, which is measured as total liabilities divided by total assets. On the one hand, a firm with high financial leverage would have lower ETRs (and thus a lower RETR, other things being equal) because of the deductibility of interest payments for tax purpose. On the other hand, a firm with a low tax burden has less incentive to use debt financing. Graham and Tucker (2006) find that tax shelter participants use less debt because firms use tax shelter deductions as a substitute for the interest deduction associated with debt. Similarly, Gupta and Newberry (1997) also find a positive relation between ETR and debt financing. Size is measured as the natural logarithm of the total assets. While larger firms have more resources for tax planning and are better able to reduce their tax burdens (Shevlin and Porter, 1992; Dyreng et al., 2008), they are also subject to a greater level of public scrutiny that results in less tax aggressiveness. MB is the market to book ratio and is used to measure the firm's investment opportunities. Spooner (1986) argues that ETR may be higher for firms with greater investment opportunities. However, Derashid and Zhang (2003) and Chen et al. (2010) find inconsistent results with different measures of ETRs. ROA denotes a firm's profitability and is calculated as the pre-tax income divided by the total assets. Prior studies find that more profitable companies would have higher ETRs (Gupta and Newberry, 1997; Wilson, 2009). CAPINT and INVINT are used to control for assets mix (Gupta and Newberry, 1997). CAPINT equals property, plant and equipment divided by total assets, and INVINT is the year-end total inventory divided by total assets. Capital-intensive (CAPINT) firms are more affected by the differences in accounting and tax treatments of depreciation. In China, after the adoption of IFRS, the significant increase in book-tax differences would lead to more scrutiny by tax authorities (Chan et al., 2010). To reduce the risk of being selected for tax audit, CAPINT firms may want to have a higher ETR. On the other hand, with the tax benefits associated with capital investments, CAPINT firms may have lower ETRs. Finally, the book-tax differences in the valuation of inventory may motivate inventory intensive (INVINT) firms to maintain a higher level of ETR.
To control for the effect of earnings management on a firm's tax aggressiveness, we include a dummy variable, RIGHTS, which equals to 1 if the firm applies for a rights offering in one of the next 3 years (Chan et al., 2010). Firms with a rights offering have more incentive to manage earnings to satisfy Chinese regulatory requirements. Besides, in China, different geographical regions have different institutional developments. Implementation of regulations is often less effective in less developed regions. To control for this effect, we include a dummy variable, MKTIND (=1 if a firm is located in a well-developed province).5 Finally, we also include industry dummies and year dummies to control for potential industry and year fixed effects.
5. Empirical results
5.1 Descriptive statistics and univariate tests
Table 2 shows the descriptive statistics of the regression variables. The mean of the dependent variable, RETR, is 67.7 per cent (the mean ETR is 16.2 per cent, and the mean ATR is 25.1 per cent). About 58 per cent of the firms are government-controlled which indicates the vital role played by the Chinese government in controlling listed companies. For the board characteristics, on average, the sample firms have 35 per cent independent board members and about 14 per cent of firms having the same person served as the board chairman and CEO. On average, board members hold only about 3 per cent of shares as government appointed board members normally do not hold any shares. For the control variables, the average leverage ratio (LEV) is 47.7 per cent, which is considerable higher than US firms in the Chen et al. (2010) study.
| Variable | Whole sample (N = 6032) | Government-controlled firms (N = 3512) | Non-government-controlled firms (N = 2520) | Diff. in Mean | t-stat | |||
|---|---|---|---|---|---|---|---|---|
| Mean | SD | Mean | SD | Mean | SD | |||
| RETR | 0.677 | 0.478 | 0.689 | 0.469 | 0.659 | 0.491 | 0.030 | 2.39** |
| Govt | 0.582 | 0.493 | — | — | — | — | — | — |
| Indep_director | 0.351 | 0.048 | 0.346 | 0.047 | 0.358 | 0.049 | −0.011 | −8.92*** |
| Co_services | 0.137 | 0.344 | 0.093 | 0.290 | 0.199 | 0.400 | −0.106 | −11.98*** |
| BOD_shares | 0.030 | 0.107 | 0.002 | 0.012 | 0.069 | 0.157 | −0.067 | −25.36*** |
| LEV | 0.477 | 0.179 | 0.483 | 0.177 | 0.470 | 0.183 | 0.013 | 2.86*** |
| SIZE | 21.453 | 0.999 | 21.631 | 0.994 | 21.204 | 0.952 | 0.427 | 16.73*** |
| MB | 3.943 | 8.703 | 3.357 | 4.523 | 4.760 | 12.317 | −1.404 | −6.20*** |
| ROA | 0.079 | 0.062 | 0.074 | 0.053 | 0.086 | 0.073 | −0.012 | −7.11*** |
| CAPINT | 0.292 | 0.185 | 0.322 | 0.193 | 0.251 | 0.164 | 0.071 | 14.99*** |
| INVINT | 0.174 | 0.152 | 0.166 | 0.147 | 0.187 | 0.158 | −0.021 | −5.32*** |
| RIGHTS | 0.022 | 0.146 | 0.025 | 0.156 | 0.018 | 0.133 | 0.007 | 1.81** |
| MKTIND | 0.805 | 0.396 | 0.777 | 0.416 | 0.844 | 0.362 | −0.067 | −6.47*** |
- *** and ** indicate statistical significance at the 1 and 5 per cent levels (two-tailed tests), respectively. Variable definitions: RETR, ratio of effective tax rate to applicable tax rate; Govt, 1 if controlling shareholder is the government, 0 otherwise; Indep_director, percentage of independent directors on the board; Co_services, 1 if the chairman of the board is also the CEO of the firm, 0 otherwise; BOD_shares, percentage of shares held by the board members; LEV, ratio of year-end total liabilities to total assets; SIZE, nature logarithm of year-end total assets; MB, ratio of year-end market value per share to net assets per share; ROA, ratio of year-end pre-tax income to total assets; CAPINT, ratio of year-end property, plant, and equipment (PPE) to total assets; INVINT, ratio of year-end inventory to total assets; RIGHTS, 1 if the firm has rights offering in the next 3 years, 0 otherwise; MKTIND, 1 if the firm is located in a well-developed province, 0 otherwise.
Table 2 also presents the descriptive statistics of all variables partitioned by ownership type. The univariate tests indicate that there are significant differences in board and other firm characteristics between government-controlled and non-government-controlled firms. Government-controlled firms have higher RETRs (68.9 per cent) than non-government-controlled firms (65.9 per cent). This is consistent with our hypothesis that government intervention results in less tax aggressiveness. For the board characteristics, government-controlled firms have a lower ratio of independent directors, less likely to have a CEO serving as the board chairman and much smaller board shareholdings than non-government-controlled firms. For the firm characteristics, government-controlled firms are more leveraged (LEV), larger in size (SIZE), exhibit lower return (ROA) and lower market to book value (MB). In addition, these firms are more CAPINT but less INVINT. Finally, more government-controlled firms issue rights during the sample period but fewer government-controlled firms are located in well-developed regions compared with non-government-controlled firms.
Table 3 presents the correlations among the explanatory variables. Most of the correlations among the test and control variables are small, and all of them are <0.51. Thus, multicollinearity should not be an issue.
| RETR | Govt | Indep_director | Co_services | BOD_shares | LEV | SIZE | MB | ROA | CAPINT | INVINT | RIGHTS | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RETR | 1.000 | |||||||||||
| Govt | 0.031** | 1.000 | ||||||||||
| Indep_director | −0.015 | −0.114*** | 1.000 | |||||||||
| Co_services | −0.027** | −0.152*** | 0.056*** | 1.000 | ||||||||
| BOD_shares | −0.060*** | −0.310*** | 0.106*** | 0.139*** | 1.000 | |||||||
| LEV | −0.042*** | 0.037*** | 0.013 | −0.074*** | −0.166*** | 1.000 | ||||||
| SIZE | −0.023 | 0.211*** | 0.006 | −0.131*** | −0.196*** | 0.332*** | 1.000 | |||||
| MB | −0.011 | −0.080*** | 0.038*** | 0.074*** | 0.028** | 0.093*** | −0.075*** | 1.000 | ||||
| ROA | 0.006 | −0.091*** | 0.029** | 0.050*** | 0.141*** | −0.250*** | 0.042*** | 0.176*** | 1.000 | |||
| CAPINT | −0.151*** | 0.190*** | −0.085*** | −0.054*** | −0.132*** | −0.063*** | 0.109*** | −0.046*** | 0.024** | 1.000 | ||
| INVINT | 0.102*** | −0.068*** | 0.057*** | −0.008 | −0.004 | 0.309*** | 0.087*** | 0.015 | −0.092*** | −0.504*** | 1.000 | |
| RIGHTS | −0.006 | 0.023 | 0.021 | 0.003 | −0.029** | 0.057*** | 0.065*** | −0.003 | 0.021 | 0.010 | 0.005 | 1.000 |
| MKTIND | 0.020 | −0.083*** | 0.012 | 0.039*** | 0.099*** | −0.018 | 0.044*** | −0.024 | 0.017 | −0.107*** | 0.050*** | 0.135*** |
- ** and *** indicate statistical significance at the 5 and 1 per cent level (two-tailed tests), respectively. See Table 2 for variable definitions.
5.2 Multivariate analysis
Table 4 presents the Tobit regression results for the model.6 The industry and year dummies are included but not tabulated. The coefficient on Govt is positively significant at the 5 per cent level indicating that government control can reduce companies' incentives to manage taxes aggressively because managers of those firms are under pressure to pursue political objectives of protecting government revenue. Whereas for non-government-controlled firms, managers may exploit complex tax planning to reduce tax expenses for their individual and shareholders benefits. The results are consistent with the prediction in H1 that government-controlled firms are less tax aggressive.
| Predicted sign | Full sample | Government-controlled firms | Non-government-controlled firms | |
|---|---|---|---|---|
| Coefficient (t-statistic) | Coefficient (t-statistic) | Coefficient (t-statistic) | ||
| Intercept | 0.7713 (7.46)*** | 0.9741 (7.22)*** | 0.5133 (3.07)*** | |
| Govt | + | 0.0195 (2.12)** | — | — |
| Indep_director | ? | −0.1474 (−1.72) | −0.1059 (−0.93) | −0.1688 (−1.28) |
| Co_services | — | −0.0332 (−2.79)*** | 0.0075 (0.42) | −0.0662 (−4.19)*** |
| BOD_shares | — | −0.1736 (−4.21)*** | −1.3976 (−3.28)*** | −0.1136 (−2.63)*** |
| LEV | ? | −0.2639 (−9.66)*** | −0.2747 (−7.57)*** | −0.2338 (−5.47)*** |
| SIZE | ? | 0.0058 (1.23) | −0.0086 (−1.42) | 0.0223 (2.97) |
| MB | ? | −0.0009 (−1.83) | 0.0002 (0.12) | −0.0009 (−1.73) |
| ROA | + | 0.3762 (5.28)*** | 0.6444 (5.50)*** | 0.2343 (2.59)*** |
| CAPINT | — | −0.2429 (−8.62)*** | −0.1856 (−5.16)*** | −0.3341 (−7.33)*** |
| INVINT | + | 0.1602 (4.24)*** | 0.3159 (6.01)*** | −0.0362 (−0.66) |
| RIGHTS | ? | −0.0034 (−0.13) | −0.0530 (−1.58) | 0.0753 (1.61) |
| MKTIND | ? | 0.0126 (1.22) | 0.0262 (2.04)** | −0.0034 (−0.20) |
| Year Dummies | Included | Included | Included | |
| Industry Dummies | Included | Included | Included | |
| Chi-square | 710.57*** | 468.27*** | 321.63*** |
- ** and *** indicate statistical significance at the 5 per cent and 1 level (two-tailed tests), respectively. See Table 2 for variable definitions.
For the board characteristics, the coefficients on Indep_director are negative but not significant. On the other hand, the coefficients on Co_services and BOD_shares are significant in the expected direction. The significant effect of Co_services is mainly driven by the non-government-controlled firms. Therefore, consistent with H2(b), non-government-controlled firms with the same person serving as both the board chairman and CEO are more tax aggressive. The dual appointment results in ineffective monitoring of the board and obstructs its oversight and governance role. The significant negative coefficient of board shareholdings for both government- and non-government-controlled firms indicates that when more shares are owned by directors, there will be greater incentives to engage in aggressive tax strategy. The result is in line with H2(c).
Regarding the control variables, for both government- and non-government-controlled firms, higher leverage (LEV), lower profitability (ROA), higher capital (CAPINT) and lower inventory (INVINT) intensiveness are associated with more tax aggressiveness. Furthermore, government-controlled firms located in less developed (MKTIND) regions are more tax aggressive.
To further investigate what are the incentives that drive some managers of government-controlled firms to pursue aggressive tax strategies, we perform additional analysis for the government-controlled subsample by comparing the results of the local government-controlled firms with those of the central government-controlled firms. As mentioned earlier, starting from 2002, for corporate income tax paid by local government-controlled firms, the local government can take only 40 per cent of the tax. In other words, the respective local government does not get 100 per cent of the tax dollar collected from its own enterprises. This can provide an incentive for these firms to evade tax to keep 100 per cent of the profit that otherwise will be taxed away in the local government-controlled firms. The results in Table 5 show that there is no significant difference in tax aggressiveness between local and central government-controlled firms, and the impact of board-related corporate governance on tax aggressiveness is also the same for local and central government-controlled firms. However, the MKTIND variable is significant only for local government-controlled firms, indicating that local government-controlled firms in less developed regions are more tax aggressive. This is probably due to the less effective implementation of corporate governance measures in those regions.
| Predicted sign | Government-controlled firms | Local government-controlled firms | Central government-controlled firms | |
|---|---|---|---|---|
| Coefficient (t-statistic) | Coefficient (t-statistic) | Coefficient (t-statistic) | ||
| Intercept | 0.9723 (7.20)*** | 0.9457 (4.92)*** | 1.0226 (5.19)*** | |
| Local_govt | — | 0.0061 (0.56) | ||
| Indep_ director | ? | −0.1056 (−0.93) | −0.1689 (−1.14) | 0.0359 (0.20) |
| Co_services | — | 0.0080 (0.44) | −0.0004 (−0.02) | 0.0219 (0.75) |
| BOD_shares | — | −1.3933 (−3.27)*** | −1.2004 (−2.35)** | −2.0463 (−2.64)*** |
| LEV | ? | −0.2741 (−7.55)*** | −0.2856 (−5.54)*** | −0.2463 (−4.65)*** |
| SIZE | ? | −0.0087 (−1.43) | −0.0054 (−0.62) | −0.0152 (−1.72) |
| MB | ? | 0.0001 (0.09) | −0.0019 (−0.63) | 0.0006 (0.45) |
| ROA | + | 0.6455 (5.51)*** | 0.8280 (4.91)*** | 0.4620 (2.69)*** |
| CAPINT | — | −0.1836 (−5.08)*** | −0.1754 (−3.70)*** | −0.1801 (−3.19)*** |
| INVINT | + | 0.3163 (6.02)*** | 0.2419 (3.61)*** | 0.4359 (5.10)*** |
| RIGHTS | ? | −0.0529 (−1.58) | −0.0776 (−1.77) | −0.0185 (−0.36) |
| MKTIND | ? | 0.0259 (2.01)** | 0.0332 (2.00)** | 0.0148 (0.73) |
| Year dummies | Included | Included | Included | |
| Industry dummies | Included | Included | Included | |
| Chi-square | 468.58*** | 272.54*** | 226.60*** |
- ** and *** indicate statistical significance at the 5 and 1 per cent level (two-tailed tests), respectively. See Table 2 for variable definitions.
5.3 Sensitivity tests
To examine the robustness of our results, we performed several sensitivity tests. First, we use an alternative measure of tax aggressiveness, DETR, which is the difference between ETR and ATR (i.e. ETR – ATR), synonymous to a book-tax difference. The larger the difference between ETR and ATR, the more aggressive the company is. The overall regression results shown in Table 6 are similar to our main results in Table 4 except that the Indep_director becomes significantly negative. This possibly suggests a collusion effect between the independent directors and corporate management as discussed earlier. In addition, the MKTIND variable shows that non-government-controlled firms are more tax aggressive in well-developed regions, probably because there is more tax planning expertise for private firms in those regions. Second, some prior studies often define controlling shareholder as the one who holds more than 20 per cent of a firm's shares rather than the 30 per cent as stated in Guidelines for the Articles of Association of Listed Companies in China. This 20 per cent requirement is also the threshold for preparing consolidated financial statements. We therefore use the 20 per cent threshold for controlling shareholding and rerun the regression. The results are reported in Table 7. The results for ownership type and board characteristics are consistent with our hypotheses and those in Table 4.
| Predicted sign | Full sample | Government-controlled firms | Non-government-controlled firms | |
|---|---|---|---|---|
| Coefficient (t-statistic) | Coefficient (t-statistic) | Coefficient (t-statistic) | ||
| Intercept | −0.0506 (−1.65) | 0.0009 (0.02) | −0.1219 (−2.47)** | |
| Govt | + | 0.0069 (2.51)** | ||
| Indep_ director | ? | −0.0533 (−2.09)** | −0.0474 (−1.39) | −0.0493 (−1.29) |
| Co_services | — | −0.0109 (−3.07)*** | −0.0058 (−1.06) | −0.0161 (−3.45)*** |
| BOD_shares | — | −0.0480 (−3.92)*** | −0.2345 (−1.84)* | −0.0342 (−2.68)*** |
| LEV | ? | −0.0848 (−10.43)*** | −0.0893 (−8.21*** | −0.0729 (−5.79)**** |
| SIZE | ? | 0.0030 (2.13)** | −0.0005 (−0.27) | 0.0071 (3.21)*** |
| MB | ? | −0.0001 (−0.98) | 0.0003 (0.73) | −0.0002 (−1.19) |
| ROA | + | 0.0078 (3.71)*** | 0.1419 (4.04)*** | 0.0428 (1.61) |
| CAPINT | — | −0.0817 (−9.76)*** | −0.0600 (−6.12)*** | −0.1022 (−7.60)*** |
| INVINT | + | 0.0275 (2.44)** | 0.0782 (4.96)*** | −0.0322 (−2.00)** |
| RIGHTS | ? | 0.0023 (0.28) | −0.0108 (−1.08) | 0.0214 (1.55) |
| MKTIND | ? | −0.0077 (−2.51)** | −0.0056 (−1.45) | −0.0104 (−2.03)** |
| Year dummies | Included | Included | Included | |
| Industry dummies | Included | Included | Included | |
| Chi-square | 607.34*** | 395.00*** | 274.79*** |
- *, ** and *** indicate statistical significance at the 5 and 1 per cent level (two-tailed tests), respectively. DETR is the difference between ETR and applicable tax rate. See Table 2 for other variable definitions.
| Predicted sign | Full sample | Government-controlled firms | Non-government-controlled firms | |
|---|---|---|---|---|
| Coefficient(t-statistic) | Coefficient(t-statistic) | Coefficient(t-statistic) | ||
| Intercept | 0.8251 (7.80) | 1.0093 (7.38)*** | 0.5508 (3.18)*** | |
| Govt | + | 0.0196 (2.06)** | ||
| Indep_director | ? | −0.1610 (−1.82) | −0.1135 (−0.99) | −0.1860 (−1.35) |
| Co_services | — | −0.0266 (−2.17)** | 0.0060 (0.32) | −0.0560 (−3.39)*** |
| BOD_shares | — | −0.1978 (−4.60)*** | −1.3772 (−3.23)*** | −0.1425 (−3.16)*** |
| LEV | ? | −0.2756 (−9.81)*** | −0.2805 (−7.58)*** | −0.2501 (−5.64)*** |
| SIZE | ? | 0.0041 (0.86) | −0.0098 (−1.58) | 0.0216 (2.78)*** |
| MB | ? | −0.0007 (−1.39) | 0.0003 (0.24) | −0.0007 (−1.36) |
| ROA | + | 0.3553 (4.70)*** | 0.6192 (5.20)*** | 0.1837 (1.85) |
| CAPINT | — | −0.2382 (−8.27)*** | −0.1824 (−4.99)*** | −0.3259 (−6.90)*** |
| INVINT | + | 0.1624 (4.20)*** | 0.3292 (6.16) | −0.0456 (−0.81) |
| RIGHTS | ? | −0.0082 (−0.30) | −0.0603 (−1.80) | 0.0802 (1.65) |
| MKTIND | ? | 0.0131 (1.23) | 0.0252 (1.93) | 0.0029 (0.16) |
| Year dummies | Included | Included | Included | |
| Industry dummies | Included | Included | Included | |
| Chi-square | 668.07*** | 445.81*** | 296.74*** |
- ** and *** indicate statistical significance at the 5 and 1 per cent level (two-tailed tests), respectively. See Table 2 for variable definitions.
Third, China implemented a tax reform of reducing the tax rate for domestic companies beginning in 2008. During the transition period of 2008–2009, firms may have different behaviour towards tax management. To control for the effect of tax reform in our analysis, we add a dummy variable, Tax_reform, which equals to 1 if the sample period is after the tax reform (2008–2009), and 0 otherwise. The results (not tabulated) are the same as Table 4. We also further restrict our sample period to only the pre-tax reform period (2003–2007) and rerun the regression. Again, the main results (not tabulated) remain unchanged.
6. Conclusions
This study examines the role of government ownership and corporate governance in tax aggressiveness, providing new insights on how these variables together affect tax avoidance activities. Using data for all A-share non-financial companies listed in Shanghai and Shenzhen Stock Exchanges between 2003 and 2009, we find that government-controlled firms pursue less aggressive tax strategies as compared to non-government-controlled firms. Non-government-controlled firms with higher board equity holdings and duality duties performed by the board chairman are more tax aggressive. These results provide evidence that managers of government-controlled firms have the political objectives of protecting government revenues, and they push their firms to avoid pursuing aggressive tax planning. However, management of non-government-controlled firms, particularly those with dominant CEOs, tend to exploit aggressive tax planning. These results should alert public policy-makers in transitional economies with privatization programmes to design appropriate tax compliance measures.
Future extensions of this research include an investigation on whether the executives of government-controlled firms do actually benefit from making generous tax payments in terms of political appointments, promotions or career advancements. On the other hand, for those aggressive tax planners, do they actually divert corporate resources for individual gains such as investing in their own pet projects? If so, tax aggressiveness may not enhance the value of these firms.
