The effect of the general anti-avoidance rule on corporate tax avoidance in China

Accepted Manuscript The Effect of the General Anti-Avoidance Rule on Corporate Tax Avoidance in China Sidney C.M. Leung, Grant Richardson, Grantley Taylor PII: DOI: Reference:

S1815-5669(18)30188-7 https://doi.org/10.1016/j.jcae.2018.12.005 JCAE 148

To appear in:

Journal of Contemporary Accounting & Economics

Received Date: Revised Date: Accepted Date:

31 March 2018 16 October 2018 7 December 2018

Please cite this article as: Leung, S.C.M., Richardson, G., Taylor, G., The Effect of the General Anti-Avoidance Rule on Corporate Tax Avoidance in China, Journal of Contemporary Accounting & Economics (2018), doi: https:// doi.org/10.1016/j.jcae.2018.12.005

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

The Effect of the General Anti-Avoidance Rule on Corporate Tax Avoidance in China

Sidney C.M. Leung* Department of Accountancy College of Business City University of Hong Kong Tel: +852-3442-7924 Fax: +852-3442-0349 E-mail: [email protected]

Grant Richardson Department of Accounting and Corporate Governance Macquarie University Eastern Road North Ryde NSW Australia 2109 Tel: +61-2-9850-7994 Fax: +61-2-9850-8497 E-mail: [email protected]

Grantley Taylor School of Accounting Curtin Business School Curtin University GPO Box U1987 Perth, Western Australia, Australia 6845 Tel: +61-8-9266-3377 Fax: +61-8-9266-7196 E-mail: [email protected]

*Corresponding author We are grateful for the helpful comments and suggestions of an anonymous reviewer. The work described in this paper was fully supported by a grant from the Research Grant Council of the Hong Kong Special Administrative Region, China (Project No. CityU 11503014). 1

The Effect of the General Anti-Avoidance Rule on Corporate Tax Avoidance in China Abstract: This study examines the effect of the general anti-avoidance rule (GAAR), introduced on January 1, 2008, to enforce corporate tax avoidance laws in China. Based on a sample of 517 Chinese firms over the 2006–2010 period (2,585 firm-years), we find a reduction in tax avoidance following the implementation of the GAAR that appears to be the result of the new and stringent tax legislation and the consolidation of Chinese tax law. We also observe that the effects of firms’ engaging a Big Four auditor and directors with tax expertise in deterring tax avoidance significantly decreased following implementation of the GAAR. To all intents and purposes, it seems that the implementation of the GAAR in China has moderated the effects of and substituted for these particular monitoring and disciplining mechanisms. Keywords: General anti-avoidance rule (GAAR); corporate tax avoidance; China. 1. Introduction This study investigates the corporate tax avoidance1 activities of Chinese multinational firms to determine whether any significant changes in those activities have taken place since the implementation of the general anti-avoidance rule (GAAR) in China on January 1, 2008.2 We also examine whether the effects of state control and the engagement of a Big Four auditor or directors with tax expertise in reducing tax avoidance activities are moderated post-GAAR implementation. The GAAR provides a unique opportunity to assess how new regulations affect tax avoidance in a relation-based economy. Currently, very little is known or understood about Chinese firms’ tax avoidance strategies used to minimize their corporate tax liabilities (Chan et al. 2013). Given that tax reforms are infrequent; few studies have been able to successfully examine their effect on corporate tax avoidance practices (Gupta & Newberry 1997; Richardson & Lanis 2007; Hanlon & Heitzman 2010). In the Chinese context, research by An and Tan (2014), Lin et We define corporate tax avoidance as any transaction or event (“passive” or “aggressive”) that leads to a reduction in the amount of corporate taxes paid by a firm (Dyreng et al. 2008). This includes tax-planning activities that are legal or that may fall into a gray area, along with illegal activities. Hence, tax avoidance can range along a continuum with many cases falling in the disputed gray zone in between the two extremes of legal and illegal (Hanlon and Heitzman 2010). 2 The National People’s Congress of China promulgated the new Enterprise Income Tax Law (EITL) on March 16, 2007 to take effect on January 1, 2008. 1

2

al. (2014), and Huang (2015) finds that firms alter their tax avoidance activities substantially in the periods leading up to and immediately following major tax reforms. Xu et al. (2011) and Huang (2015) also claim that in China, tax enforcement can act as an important corporate governance mechanism by ensuring that managers act in accordance with tax laws and by reeducating firms to avoid engaging in particularly aggressive tax evasion activities. We are motivated to undertake this study for several reasons. First, research has been carried out on corporate tax avoidance in Australia, Europe, and the U.S., but very little is known about the tax avoidance practices of Chinese firms (Chan et al. 2010; Zeng 2011; Wu et al. 2012; Chan et al. 2013). Second, tax avoidance, and especially aggressive transfer pricing activities, has gained increasing traction in China recently (Sutherland et al. 2012). The Chinese tax authorities have reported a large reduction in corporate tax revenue due to transfer pricing aggressiveness3 (SAT 2014). Our research on Chinese firms’ tax avoidance strategies provides some valuable insights into the extent of transfer pricing aggressiveness activity. Third, the introduction of the GAAR in China on January 1, 2008 was intended to overcome a major perceived deficiency in Chinese tax law enforcement. The GAAR consolidated tax laws and unified tax rules, rates, and incentives, resulting in a wholly transformed corporate tax environment in China. The law introduced several new and more stringent tax provisions, particularly geared toward preventing the more aggressive forms of tax avoidance such as transfer pricing. 4 At the same time, information on penalties for tax law violations has become more widely dispersed, leading to an increase in the use of tax authority audit teams (Lin et al. 2014). We analyze how Chinese firms

3

We define transfer pricing aggressiveness as the set of transactions and arrangements used to reduce corporate tax liabilities by shifting profits (or losses) and tax deductible expenses between group members located in variably taxed jurisdictions as an intentional manipulation of related party transfer prices (Taylor et al. 2015). 4 A 2007 survey conducted by the National Bureau of Statistics claimed that almost two-thirds of apparently lossmaking foreign firms had deliberately made false reports and used transfer pricing to avoid paying approximately RMB30 billion (USD4.39 billion) in corporate taxes (Global Times 2009).

3

altered their tax avoidance activities in response to the GAAR. Given that major changes to country-level tax systems occur infrequently, this study takes advantage of a unique opportunity to evaluate firms’ reaction to stricter tax laws, especially in a relation-based economy such as China’s. Fourth, the Chinese economy is unique in terms of the importance of state control, external audits, and the expertise of boards of directors (Chen et al. 2006; Fan et al. 2007). The Chinese government plays an important role in business activities through its majority ownership of state-owned enterprises (SOEs), as well as its links with external auditors and with government-connected boards of directors, particularly in SOEs (Chen et al. 2011a). The effectiveness of anti-avoidance tax rules, such as the GAAR, could possibly interact with state control, engagement of a Big Four auditor, and boards of directors’ tax expertise; this study therefore explores how control, monitoring, and disciplining mechanisms affect corporate tax avoidance in the post-GAAR period in China. Based on a sample of 517 Chinese firms during the period from 2006 to 2010 (2,585 firmyears), we report that the GAAR has been effective in curbing corporate tax avoidance in China. The reduction in tax avoidance following the implementation of the GAAR appears to be due to the introduction of new and more stringent tax legislation and to the consolidation of Chinese tax law. We also find that the effects of engaging a Big Four auditor and directors with tax expertise in deterring tax avoidance significantly decreased following the GAAR’s implementation. Essentially, evidence suggests that the implementation of the GAAR in China has moderated the effects of, and substituted for, these specific monitoring and disciplining mechanisms. This study makes the following contributions. First, although research has examined the effect of stricter tax enforcement regulations on corporate tax avoidance in developed (“rule-based”) economies such as the U.S. (e.g., Hoopes et al. 2012), little is known about whether anti-

4

avoidance rules are effective in deterring tax avoidance activities in developing (“relationbased”) economies, such as China. By examining how Chinese firms respond to the GAAR, this study extends prior research to consider whether stricter anti-avoidance rules constrain tax avoidance activity in a major relation-based economy. Opportunities to undertake such research are limited as corporate tax reforms are infrequent. As such, our study adds to the sparse literature about the effects of tax reform in relation-based economies. The rise of Chinese firms in terms of economic prominence and international influence suggests that a study of the GAAR and its effects on tax avoidance in China is important and timely. Second, the size of the Chinese economy and its distinct institutional background (e.g., the importance of state control) warrants a detailed study of corporate tax avoidance. The determinants of tax avoidance in the context of China are not well understood; this study adds to our knowledge of these determinants, particularly for transfer pricing aggressiveness. Third, we use both a broad-based measure of tax avoidance (i.e., book-tax gaps) and a more specific measure as stipulated as part of the GAAR (i.e., transfer pricing aggressiveness) to evaluate a broad spectrum of tax avoidance proxy measures as originally envisaged by Hanlon and Heitzman (2010). Fourth, our findings provide valuable feedback to the government and tax authorities in China about whether the GAAR achieved its intended purpose in curbing Chinese firms’ tax avoidance activities. Finally, our study’s results will also be of interest to policymakers and regulators in other developing (relation-based) economies who are considering tax reform to curb corporate tax avoidance. This paper proceeds as follows. Section 2 provides a brief review of the background to Chinese tax law and Chinese firms’ corporate tax avoidance activities. Section 3 develops our hypotheses, Section 4 discusses the research design, and Section 5 summarizes and analyzes the empirical results. Finally, Section 6 concludes the paper.

5

2. Background 2.1. Chinese tax law China’s tax laws evolved radically after the implementation of the GAAR (Deloitte 2013). The introduction of the GAAR on January 1, 2008 was a major event in corporate tax reform in China: it was the first law in Chinese history to impose income tax on all enterprises5 (Li 2007). The introduction of the GAAR by the National People’s Congress of China (NPC) was listed in three documents: (1) the Enterprise Income Tax Law (EITL), (2) the EITL implementing regulations, and (3) the State Administration of Taxation (SAT)6 notice issuing the measures for special tax adjustments for a trial implementation (Ernst & Young 2014a; SAT 2014). The GAAR replaced the Foreign Investment Enterprise Income Tax Law,7 which was applicable to firms with foreign direct investment, and the Interim Enterprise Income Tax Regulations, which were relevant to Chinese-owned firms (Li 2007). The NPC and the Ministry of Finance (MOF) were motivated to apply these tax reforms for several reasons: to promote equal taxation for all firms, to foster a more sustainable development of China’s economy, to develop tax law, policy, and practices as per international norms, and to realize efficiency and simplicity in tax administration and compliance. The GAAR denotes a major shift in Chinese tax policy and practices (Li 2007). The first significant change that occurred under the GAAR was the introduction of new and more stringent anti-avoidance tax laws. These allowed the SAT to assess and make tax The term “enterprise” is not defined in the new tax legislation. However, the State Administration of Taxation (SAT) explanatory notes on EITL indicate that the definition of enterprise includes, but is not limited to, a firm. 6 The SAT is the body charged with collecting tax and enforcing compliance and is assisted by the state and local tax bureaus at the provincial level (PWC 2012). 7 Foreign investment enterprises include Chinese–foreign joint venture firms and wholly foreign-owned firms. 5

6

adjustments in areas such as transfer pricing, tax haven utilization, thin capitalization, and the use of foreign-controlled corporations (Deloitte 2009). The new tax laws also included general anti-avoidance provisions to determine whether transactions leading to a reduction in taxable income passed a reasonable business purpose test. 8 The GAAR altered the corporate tax landscape in China by introducing these new rules, including “catch-all” general anti-avoidance provisions that would examine all business transactions and arrangements based on a reasonable business purpose test (Cheung 2012). In short, the GAAR was designed to target abusive, aggressive, or illegal tax structures that were not able to be addressed prior to its introduction. The second significant change following the GAAR was the increased power of the SAT to enforce tax rule compliance. For instance, Article 92 states that the tax authorities may initiate an audit of a firm’s tax arrangements based on the reasonable business purpose test and the principle of economic substance over legal form under Article 93. In addition, Article 94 allows the SAT to review, remodel, or re-characterize firms’ tax arrangements based on the economic substance of transactions, thereby removing any tax benefits that the firm may have received (Xu et al. 2011).9 Moreover, under the GAAR, a firm may be targeted by the SAT for a transfer pricing audit if it conducts a significant number of related party transactions, if its profit is lower than the industry norm, if it fails to prepare documentation on transfer pricing methodologies and pricing, or if it fails to adhere to the arm’s length principle of commercial transactions (Cheung 2012). The transfer pricing rules included in the GAAR ensure that related party transactions are conducted on an arm’s length basis (SAT 2014). The SAT requires that documentary evidence

8

For instance, Article 47 of the EITL allows the SAT to make tax adjustments related to arrangements or transactions in which the main business purpose is to avoid, reduce, defer, or make exempt the payment of corporate taxes. 9 The SAT is able to investigate a lack of commerciality in transfer pricing transactions if they are not able to pass the reasonable business purpose test. It can retract tax benefits and apply tax penalties if a firm does not pass that test (Li & Huang 2008; Cheung 2012).

7

be provided to show that transfer prices for intra-group transactions comply with the arm’s length principle (SAT 2014). Further, as per the MOF (which overseas audit practices in China), business transactions must comply with the arm’s length principle (Shevlin et al. 2012). To deal with audit requirements under the new rules, the SAT and tax bureaus at the provincial and municipal levels formed specialist teams to deal with transfer pricing compliance (Ernst & Young 2014b). Transfer pricing audit case selection is run by a central decision-making authority. The GAAR introduced a transfer pricing penalty regime, with serious violations of transfer pricing laws attracting about 25% in additional taxes (Ernst & Young 2014b). The third significant change under the GAAR was the consolidation of previously separate tax laws for domestic enterprises and foreign investment enterprises (FIEs), in addition to a general consolidation of tax rates, tax rules, and tax incentives (Lin et al. 2014). Prior to 2008, FIEs and domestic firms operating in certain industries (e.g., utilities) received preferential tax treatment. However, the SAT claimed that FIEs were responsible for large declines in tax revenue following China’s accession to the World Trade Organization in 2001. The SAT was responding to the transfer of profits out of China and into offshore localities (e.g., tax havens) and the manipulation of transfer pricing methods as the main methods by which FIEs evaded its tax liabilities (An & Tan 2014). The GAAR also included a reduction in the statutory tax rate for domestic firms from 33% to 25%, effective January 1, 2008. While the legislated statutory tax rate was 33%, the effective tax rate (ETR) had been 25% for domestic firms and 15% for FIEs. Under the GAAR, both domestic firms and FIEs became subject to a tax rate of 25% (i.e., no change in the ETR for domestic firms and an increase from 15% to 25% for FIEs) (Li & Huang 2008). Thus, in terms of the application of tax laws, the GAAR removed the effect of local or

8

regional taxation variations for firms, regardless of ownership, and introduced a consistent statutory tax rate. In short, the establishment of new and more stringent tax laws, the increased power of the SAT to ensure compliance with the new tax regime, the increased resourcing of the SAT at both the local and central levels, and the consolidation of tax laws after the implementation of the GAAR, are all likely to reduce a Chinese firm’s incentive or opportunity to engage in corporate tax avoidance activities in general and to participate in particularly aggressive forms of tax avoidance, such as transfer pricing, in particular.

2.2. Chinese firms’ tax avoidance activities Several studies have examined Chinese firms’ tax avoidance strategies. Chan et al. (2013) analyze the nature of tax avoidance in China from 2003 to 2009. They find that non-governmentcontrolled firms with a higher percentage of board shareholdings and with a CEO who also served as chairman of the board displayed higher levels of tax avoidance. They also observe that the board of directors’ independence and local versus central government connections did not significantly affect the extent to which a firm engaged in tax avoidance. Wu et al. (2012) find that firm size, state control, and tax status were key determinants of Chinese firms’ ETRs. Lin et al. (2014) examine how public and private firms in China responded to the statutory tax rate reduction in 2008. They document that private firms were able to shift taxable income from a high (2007) to a low (2008) tax year, achieving tax savings of around 8.58% in 2007. They also find that firm managers altered their financial reporting decisions to exploit the 2008 corporate tax cuts. Shevlin et al. (2012) explore income shifting between members of a consolidated group in China in response to a series of tax incentives (e.g., tax holidays, tax exemptions, and reduced

9

rates) that were in place in the pre-GAAR environment (i.e., pre-2008). They report that intangible-intensive firms shifted more income, with managers of those firms using their discretion to set transfer prices while moving income to low-tax jurisdictions. Finally, one of the most important targets of the GAAR regulations is transfer pricing aggressiveness (Li 2007; Li & Huang 2008; Wang 2016). The U.S. research and advocacy group Global Financial Integrity claim that China was subject to an USD3.79 trillion in illegal capital outflow between 2000 and 2011. Of the USD2.83 trillion that flowed illicitly out of China from 2005 to 2011, a total of USD595.8 billion wound-up as cash deposits or financial assets (e.g., stocks, bonds, mutual funds, and derivatives) in tax havens (Global Financial Integrity 2012). Aharony et al. (2010) analyze a sample of 185 Chinese IPO firms from 1999 to 2001 and find that related-party sales of goods and services were used opportunistically to manage earnings upwards in the pre-IPO period. They report a relation between tunneling behavior in the postIPO period and earnings management through abnormal related-party sales in the pre-IPO period, but did not explore the arm’s length nature of these related-party transactions.

3. Hypotheses development 3.1. Corporate tax avoidance and GAAR Lin et al. (2014) finds that Chinese domestic firms deferred tax payments to the years after the 2008 reduction in the corporate tax rate, which generated tax savings of 8.58% compared what would have been paid in 2007, the year immediately prior to the reduction. They report that firms altered their tax planning decisions leading up to and immediately following the tax reforms. An and Tan (2014) explore how FIEs responded to the removal of preferential tax treatment and the unification of the Chinese corporate tax regime, finding that FIEs

10

reacted to these changes by shifting income out of China. However, SOEs undertook much less income shifting than private firms, suggesting that the former may have received favorable tax treatment from the Chinese government or were subject to more rigorous scrutiny. Overall, these studies provide further evidence that corporate tax reforms alter a firm’s tax avoidance activities leading up to and immediately following implementation. As discussed above, the GAAR provision drastically strengthened the tax law enforcement landscape in terms of corporate tax avoidance in China by increasing the marginal costs of tax avoidance, such as the potential tax penalties imposed by the tax authorities and the costs associated with implementation (e.g., the time and effort and transaction costs involved in executing tax transactions) and agency costs (e.g., managerial rent extraction) used in tax avoidance activities (Scholes & Wolfson 1992; Desai & Dharmapala 2006; Chen et al. 2010). Accordingly, Chinese firms are less likely to engage in tax avoidance activities in the post-GAAR period than in the pre-GAAR period. Based on the above discussion, we develop the following (directional) hypothesis: H1: Chinese firms’ corporate tax avoidance is lower in the post-GAAR period than in the pre-GAAR period.

3.2. Control, monitoring and disciplining mechanisms, and GAAR We also develop several additional hypotheses to analyze whether the effect of control, monitoring, and disciplining mechanisms, including state control, engagement of a Big Four auditor, and the presence of directors with tax expertise in reducing corporate tax avoidance, was moderated following implementation of the GAAR.

11

3.2.1. State control, tax avoidance, and the GAAR SOEs continue to dominate China’s publicly-listed firms, accounting for 310 of the top 500 (62%) in 2012 (Wang 2013). Although China has carried out several economic reforms since the 1980s (i.e., corporatization and privatization), the government has retained substantial control over Chinese firms. This affects the shareholdings and political power of firm managers (Chan et al. 2013; Kusnadi et al. 2015). Hence, the management of government-controlled firms are likely to have tax incentives that differ from their nongovernment-controlled counterparts. Further, state control may either constrain or facilitate tax avoidance activities and thus influence the corporate response to the GAAR implementation. There is evidence that SOEs are likely to comply with tax regulations and pay more taxes, because the state is the controlling owner.10 Zeng (2011) and Chan et al. (2013) find that nongovernment-controlled Chinese firms are more likely to pursue more tax avoidance schemes than government-controlled firms. They argue that government control plays an important role in tax risk management in China; as such, government-controlled firms are less tax avoidant than their counterparts, as firm managers are aware of the role that they play in protecting and maintaining government tax revenue. This is likely to have social and political objectives that extend far beyond shareholder wealth maximization. 11 In addition, as the government is the controlling shareholder of SOEs, it can exert a great deal of influence on managerial decision-making in areas such as the appointment, promotion, compensation, and dismissal of firm managers (Lo et al. 2010). There could be pressure on SOEs to align their

10

The corporate tax paid by SOEs is similar to dividends paid on state investment (Zeng 2011; Chan et al. 2013). The government has incentives to use collected tax revenue for social welfare purposes (e.g., the provision of public goods) and thus encourages SOEs to pay a higher amount of corporate taxes (Chan et al. 2013). 11

12

tax-planning strategies with those of the government (Wu et al. 2009, 2012; Wu 2013).12 However, the effect of state control on restricting corporate tax avoidance in China is likely to be weaker in the post-GAAR period than in the pre-GAAR period: as mentioned above, the GAAR has strengthened the tax law enforcement landscape in China and has increased the level of penalties and punishments related to corporate tax avoidance. This is likely to have reduced the incentives and opportunities related to tax avoidance. In other words, state control as a tool to reduce tax avoidance may have been substituted by the GAAR implementation. Based on the above discussion, we develop the following (directional) hypothesis: H2: The effect of state control in reducing corporate tax avoidance is moderated following implementation of the GAAR.

3.2.2. Big Four auditors, tax avoidance, and the GAAR Since the 2006 tax year, Chinese firms have been required to have their internal control systems reviewed by an external auditor (Lin et al. 2014). Research finds that the provision of high-quality audits and monitoring by Big Four audit firms may reduce financial statement fraud and corporate tax avoidance (e.g., Matsumura & Tucker 1992). Big Four audit firms have reputational capital at stake, possess the financial backing to reduce client pressure, and retain audit and specialist team expertise to more effectively detect financial statement fraud relative to non-Big Four firms (Rezaee 2005). Big Four audit firms are also considered to provide higher quality audits because of their superior expertise and judgment (Francis 2004;

12

An alternative view is that SOEs may carry out more aggressive forms of tax avoidance than non-SOEs because SOEs may receive preferential tax treatment from the government (Zhang 2012). Further, managers of SOEs may also have greater self-interest incentives to use the savings from tax avoidance for personal benefit (Chan et al. 2013).

13

Xiao & Yuan 2007). Therefore, firms that employ Big Four auditors are likely to exhibit less tax avoidance than firms that use non-Big Four auditors.13 In the case of a relation-based economy such as China, linkages built on familiarity and trust between the external auditor and managers are likely to be essential for a firm to achieve its financial and tax objectives (Chua et al. 2009). Liu et al. (2011) claim that auditormanagement affiliations increase the likelihood that a firm will receive a clean audit opinion. The Chinese audit market is also effectively controlled by the government through the MOF (Liu et al. 2011), 14 so it is not unreasonable to expect that closer connections between external auditors and managers could reduce a firm’s tax avoidance activities. Finally, as alluded to above, the GAAR has enhanced tax law enforcement in China and restricted firms from engaging in aggressive tax avoidance practices. Consequently, the effect of Big Four auditors as a corporate governance monitoring mechanism in reducing a firm’s tax avoidance activities is likely to be weaker in the post-GAAR period. Based on the above discussion, we develop the following (directional) hypothesis: H3: The effect of engaging a Big Four auditor in reducing corporate tax avoidance is moderated following implementation of the GAAR.

3.2.3. Directors’ tax expertise, tax avoidance, and the GAAR Boards of directors play a critical role in monitoring managers and mitigating the risk that they carry out aggressive or fraudulent activities (Beasley 1996). Directors are responsible for establishing the compliance tone and framework at the top of the organization and are 13

However, if Big Four auditors were to assist their client firms in developing innovative and aggressive tax avoidance strategies, then this would likely lead to more corporate tax avoidance activity, not less (Sikka 2010). 14 The Chinese government controls the audit market through the Chinese Institute of Certified Public Accountants (CICPA), which is effectively controlled by the MOF. The Chinese government also controls the audit market, as the services of CICPA are used by local governments (Wang et al. 2008).

14

likely to influence a manager’s engagement in risky activities (Dyreng et al. 2010). This oversight and monitoring may affect a firm’s propensity to engage in tax avoidance (Lanis & Richardson 2011; Richardson et al. 2013a). Board directors with tax expertise are better able to monitor compliance with tax laws, keep abreast of changes to existing tax laws, and effectively add to a firm’s overall tax function and management of tax-related risk (Maydew & Shackelford 2007). Overall, firms with board members who are also tax experts are less likely to be devoted to aggressive and risky tax positions (Carter et al. 2010). In a climate of high compliance costs, directors may be inclined to use their tax expertise to manage a firm’s tax liabilities. Consistent with our previous discussion that the GAAR has generally strengthened the law enforcement landscape in China and has boosted the level of penalties and punishments related to tax avoidance, high compliance costs are likely to have reduced the incentives and opportunities to engage in tax avoidance. Thus, the effect of tax expertise of board members as a disciplining mechanism in reducing a firm’s engagement in tax avoidance activities is expected to be less effective in the post-GAAR period than in the pre-GAAR period. Based on the above discussion, we develop the following (directional) hypothesis: H4: The effect of directors with tax expertise in reducing corporate tax avoidance is moderated following implementation of the GAAR.

4. Research design 4.1. Sample selection and methodology Our sample consists of the top 900 Chinese firms (by market capitalization) listed on the Shanghai or Shenzhen stock exchanges according to the China Stock Market and Accounting

15

Research (CSMAR) database over the five-year period from 2006 to 2010 (4,500 firm-year observations). After eliminating purely domestic firms (225), and financial and insurance firms (158), we collected data for 517 firms (2,585 firm-year observations) for our empirical analysis. We restrict our sample to Chinese firms with foreign income (i.e., multinational firms) as opposed to purely domestic firms because multinational firms have more capacity and incentive to engage in tax avoidance activities that significantly reduce corporate taxes (Klassen & Laplante 2012a, 2012b). Financial institutions and insurance firms were excluded from the sample due to significant differences in the use of accounting policies and in the calculation of accounting estimates, together with the different regulatory constraints that these firms faced. Table 1 presents the sample industry distribution according to Chinese industry classification codes. The sample includes more firms in the manufacturing and petrochemical sector (50.10%), social services sector (8.70%), transportation sector (7.35%), utilities sector (6.96%), and wholesale trade and retail sector (5.80%) than in other sectors. [Insert Table 1 Here] 4.2. Dependent variable Our dependent variable is corporate tax avoidance (CTA). A large body of tax research uses the difference between accounting income and taxable income, i.e., the “book–tax gap” (BTG), as a proxy measure for tax avoidance (see Hanlon and Heitzman 2010); large differences between accounting (or book) income and taxable income are typical for firms attempting to maximize accounting income while reducing taxable income (e.g., Frank et al. 2009). Our first tax avoidance proxy measure is computed as the difference between accounting income and taxable income scaled by lagged total assets (BTG1) (Manzon & Plesko 2002).

16

Taxable income is calculated as income tax expense scaled by the statutory tax rate in China of 33% (pre-2008) or 25% (post-2008). Our second tax avoidance proxy measure (BTG2) is calculated according to the BTG measure that Desai and Dharmapala (2006) develop. They decompose BTG into both earnings management and tax avoidance components. The component related to earnings management is removed to leave a residual value that is inferred to measure tax avoidance (Chen et al. 2010).15 Our third tax avoidance proxy measure is transfer pricing aggressiveness (TPRICE), which is a specific form of tax avoidance (Taylor et al. 2015). As a disclosure requirement, Chinese firms must provide a statement on the commerciality or arm’s length nature of its related-party dealings under IAS 24 Related Party Transactions in their annual reports.16 A firm’s level of TPRICE can be estimated by the extent to which it engages in non-arm’s length transactions and from its lack of assurances about transfer pricing and documentation (Usmen 2012). Consistent with Richardson et al. (2013b), TPRICE is measured as a dummy variable, coded 1 if the parent entity cannot substantiate that its related-party transactions with group members were made on an arm’s length or commercial basis, and 0 otherwise.17

4.3. Test variables Our main test variable of interest is the introduction of the new tax rules (GAAR). We measure GAAR as a period dummy variable, coded 1 if the observations correspond to the postGAAR period (i.e., 2008–2010), and 0 otherwise (i.e., 2006–2007) (Chan et al. 2013).

15

See Appendix A for a brief description of the Desai and Dharmapala (2006) BTG measure. The Chinese government introduced IFRS equivalent accounting standards in 2006. 17 We note that firms could be either silent on the commerciality of transactions between group affiliates when they should be disclosing pursuant to IAS 24 that such transactions were conducted on an arm’s-length basis or they may disclose that such transactions were based on what management deemed to be reasonable (i.e., again with no benchmarking of the commerciality of transactions). In such cases, these firms are scored as 1 in this study. 16

17

For the control, monitoring, and disciplining variables, we measure GOV as a dummy variable, coded 1 if the firm is majority (> 50%) owned and controlled by the government, and thus deemed to be an SOE, and 0 otherwise (Zeng 2011). AUD is measured as a dummy variable, coded 1 if the firm uses a Big Four external auditor, and 0 otherwise (He et al. 2012). TDIR is measured as a dummy variable, coded 1 if at least one member of the firm’s board of directors has tax expertise (i.e., experience in tax accounting or tax-related audit work), and 0 otherwise (Taylor & Richardson 2014). Finally, the interaction terms are computed by multiplying GAAR with GOV, AUD, and TDIR, respectively (GOV*GAAR, AUD*GAAR, and TDIR*GAAR).

4.4. Control variables In our regression model, we also include several control variables that have been identified as important determinants of tax avoidance in prior research; these include firm size (SIZE), leverage (LEV), capital intensity (CINT), inventory intensity (INVINT), R&D intensity (RDINT), foreign operations (FOR), return on assets (ROA), the market-to-book ratio (MKTBK), firm age (AGE) and industry sector effects (INDSEC) (e.g., Stickney & McGee 1982; Gupta & Newberry 1997; Rego 2003; Chen et al. 2010; Taylor & Richardson 2012;). Given that large firms have the resources and expertise to significantly reduce the amount of corporate taxes payable (Taylor & Richardson 2012), SIZE is included as a control variable in our regression model. Firm size is measured as the natural logarithm of total assets (Stickney & McGee 1982). We expect SIZE to have a positive sign (Taylor & Richardson 2012). Highly leveraged firms are able to shift debt between variably taxed jurisdictions to gain greater tax deductions on interest expenses (Rego 2003). Alternatively, firms may also pursue

18

debt-based rather than aggressive tax avoidance strategies (Graham & Tucker 2006). Thus, LEV is included as a control variable in our regression model. It is measured as long-term debt scaled by total assets (Gupta & Newberry 1997). No sign prediction is made for LEV. CINT and INVINT are included as variables in our regression model to control for highly capital-intensive or inventory-intensive firms (Stickney & McGee 1982). CINT is measured as net property, including plant and equipment, scaled by lagged total assets, and INVINT is measured as total inventory scaled by lagged total assets. We expect CINT to have a positive sign due to the accelerated depreciation charges based on asset lives. To the extent that INVINT acts a substitute for CINT, we expect INVINT to have a negative sign (Stickney & McGee 1982). In keeping with prior research (e.g., Gupta & Newberry 1997), we also incorporate RDINT as a control variable in our regression model. We measure RDINT as R&D expenditure scaled by total assets. RDINT is expected to have a positive sign (Gupta & Newberry 1997). Firms with substantial foreign operations are also likely to have greater opportunities and capacity to engage in tax avoidance activities (Rego 2003), so FOR is included as a control variable in our regression model. FOR is measured as the total number of foreign incorporated subsidiaries scaled by the total number of subsidiaries, and is expected to have a positive sign (Rego 2003). ROA is incorporated in our regression model to control for firm-specific operating performance (Gupta & Newberry 1997). ROA is measured as the pre-tax income scaled by the total assets. We expect ROA to have a positive sign (Gupta & Newberry 1997). We also include MKTBK as a control variable in our regression model, as research shows that tax avoidance may vary based on firm growth (Chen et al. 2002). MKTBK is measured as the

19

market value of equity scaled by the book value of equity (Gupta & Newberry 1997). No sign prediction is made for MKTBK due to conflicting research (e.g., Cheng et al. 2012). We also include AGE as a control variable in our regression model, as research shows that the longer a firm has traded in public markets, the more likely it is that it will have tax avoidance strategies in place (Taylor & Richardson 2014). AGE is measured as the number of years since firm incorporation. It is expected to have a positive sign (Taylor & Richardson 2014). Finally, it is also possible for tax avoidance to vary across industry sectors (Rego 2003). We incorporate 13 INDSEC dummy variables as controls based on Chinese industry classification codes.18 No sign predictions are made for the INDSEC dummies.

4.5. Regression models The regression model used to test H1 is estimated as follows: CTAit =

αit + β1GAARit + β2SIZEit + β3LEVit + β4CINTit + β5INVINTit + β6RDINTit + β7FORit + β8ROAit + β9MKTBKit + β10AGEit + β11-22INDSECit + εit,

(1)

where i = firms 1–517; t = the financial years 2006–2010; CTA = corporate tax avoidance proxy measures (BTG1, BTG2 and TPRICE); GAAR = a period dummy variable, coded 1 if the observations correspond to the post-GAAR period (i.e., 2008–2010), and 0 otherwise (i.e., 2006– 2007); SIZE = the natural logarithm of total assets; LEV = long-term debt scaled by total assets; CINT = net property, plant and equipment scaled by total assets; INVINT = total inventory scaled by total assets; RDINT = R&D expenditure scaled by total assets; FOR = the total number of foreign incorporated subsidiaries scaled by the total number of subsidiaries; ROA = pre-tax income scaled by total assets; MKTBK = the market value of equity scaled by the book value of 18

The 13 INDSEC dummy variables in our regression model are as follows: agriculture, mining, manufacturing and petrochemicals, utilities, construction, transportation, telecommunications, wholesale trade and retail, real estate, social services, media and communication, conglomerates, and other (omitted in our regression model).

20

equity; AGE = firm age (in years); INDSEC = a dummy variable, coded 1 if the firm is represented in a specific industry sector, and 0 otherwise; and  = the error term. The regression model used to test H2, H3 and H4 is estimated as follows: CTAit =

αit + β1GAARit + β2GOVit + β3GOVit*GAARit + β4AUD + β5AUDit*GAARit + β6TDIRit + β7TDIRit*GAARit + β8SIZEit + β9LEVit + β10CINTit + β11INVINTit + β12RDINTit + β13FORit + β14ROAit + β15MKTBKit + β16AGEit + β17-28INDSECit + εit,

(2)

where GOV = a dummy variable, coded 1 if the firm is majority (> 50%) owned and controlled by the government and thus deemed to be an SOE, and 0 otherwise; AUD = a dummy variable, coded 1 if the firm uses a Big Four external auditor, and 0 otherwise; TDIR = a dummy variable, coded 1 if at least one member of the firm’s board of directors has tax expertise (i.e., prior experience in tax accounting or tax-related audit work), and 0 otherwise; and GOV*GAAR, AUD*GAAR, and TDIR*GAAR are interaction terms between GAAR and GOV, AUD, and TDIR.

5.

Empirical results

5.1. Descriptive statistics Table 2 reports the descriptive statistics for the dependent variable (BTG1, BTG2, and TPRICE), test variables (GAAR, GOV, AUD, and TDIR) and control variables (SIZE, LEV, CINT, INVINT, RDINT, FOR, ROA, MKTBK, and AGE). We find that BTG1, BTG2, and TPRICE have a mean (median) of 0.003 (0.003), –0.001 (0.000) and 0.494 (0.000), respectively. We also observe that GAAR has a mean (median) of 0.600 (1.000), while GOV, AUD, and TDIR have a mean (median) of 0.122 (0.000), 0.152 (0.000), and 0.087 (0.000), respectively. Finally, the descriptive statistics for the control variables are also shown in Table 2. 21

[Insert Table 2 Here] 5.3. Regression results – H1 The regression results for the association between GAAR and tax avoidance (see Eqn. (1)) are presented in Table 3. The regression coefficient for GAAR is negative and statistically significant for BTG1, BTG2 (p < 0.05), and TPRICE (p < 0.01), which is consistent with H1. Our results show that tax avoidance was reduced following the implementation of the GAAR. This suggests that the new tax rules were effective in restricting tax avoidance generally and transfer pricing aggressiveness specifically. It appears that firms facing the new tax rules in early 2008 may have perceived the risks of tax avoidance (e.g., tax penalties, implementation costs, agency costs, and perhaps guanxi factors) to be more salient than the potential gains (e.g., improving after-tax cash flows). The new tax rules may thus have provided a sufficient incentive for firms to reduce their propensity to engage in tax avoidance activities in the post-GAAR period. Finally, we also observe that several of the regression coefficients for the control variables are significantly associated with BTG1 and BTG2 (SIZE, LEV, INVINT, RDINT, FOR, ROA, and MKTBK) (p < 0.05 or better), and/or with TPRICE (SIZE, LEV, RDINT, and AGE) (p < 0.10 or better), which is consistent with prior research (e.g., Stickney & McGee 1982; Gupta & Newberry 1997; Chen et al. 2002; Rego 2003; Graham & Tucker 2006; Taylor & Richardson 2012). [Insert Table 3 Here] 5.4. Regression Results – H2, H3, and H4 Next, we include the GOV, AUD, and TDIR test variables and their respective interaction terms with GAAR in the regression model (see Eqn. (2)) to analyze the effects of state control,

22

engagement of a Big Four auditor, and directors’ tax expertise on the association between GAAR implementation and tax avoidance. Table 4 (Panel A) reports the pooled cross-sectional regression results. The interaction terms permit the shift in the slopes of each of the explanatory variables after the tax reform to be analyzed and help to determine whether these associations with tax avoidance changed after the GAAR implementation. The test variables GOV, AUD, and TDIR provide evidence of associations with tax avoidance in the pre-GAAR period. Finally, Table 4 (Panel B) reports the t-statistics for differences in the coefficients of the test variables GOV, AUD, and TDIR in the post-GAAR implementation period. Table 4 (Panel A) shows that the regression coefficient for GAAR is negative and statistically significant for BTG1, BTG2, and TPRICE (p < 0.05 or better), which is again consistent with H1. The regression coefficients for AUD and TDIR using BTG1 and BTG2 as the dependent variables are negative and statistically significant (p < 0.10 or better), as is the regression coefficient for TDIR using TPRICE as the dependent variable (p < 0.05). Taken together, the engagement of a Big Four auditor and having board directors with tax expertise appear to reduce tax avoidance generally. Further, the presence of a tax expert on the board also reduces the prospect of a firm engaging in more aggressive forms of tax avoidance, such as transfer pricing. We also report that the regression coefficient for GOV is not statistically significant in any of the regression models for BTG1, BTG2, and TPRICE. Taken together, these results suggest that the monitoring and disciplining mechanisms related to AUD and TDIR restrict tax avoidance. We are also interested in the coefficients of the interaction terms between GOV, AUD, and TDIR and GAAR, as the significance and sign of the coefficients for these interaction terms allows us to determine whether the joint effect of the GAAR and our other test variables have altered a firm’s propensity to engage in tax avoidance after the tax reforms. The regression

23

coefficients for the interaction terms AUD*GAAR and TDIR*GAAR are positive and statistically significant (p < 0.10 or better) for BTG1 and/or BTG2. For TPRICE, the AUD*GAAR regression coefficient is positive and statistically significant (p < 0.05). We also find that the regression coefficients of the interaction term GOV*GAAR are not statistically significant in any of the regression models for BTG1, BTG2, and TPRICE. Overall, these results are consistent with H3 and H4, but not H2. Our results suggest that the associations between AUD and TDIR and tax avoidance changed after the GAAR. Following the implementation of the new regulations, the effect of AUD and TDIR on BTG1 and BTG2 was largely negated, as illustrated by the change in regression coefficients for AUD and TDIR and their interaction terms with GAAR. The GAAR appears to have initiated a change in firms’ propensity to engage in tax avoidance; the new tax rules have thus reduced the effects of some mechanisms as the GAAR substituted the monitoring and disciplining functions of AUD and TDIR in place prior to the GAAR. We also find that some of the control variables (SIZE, LEV, INVINT, RDINT, FOR, ROA, MKTBK, and AGE) are significantly associated with at least one of the tax avoidance measures (p < 0.05 or better). Table 4 (Panel B) reports the t-statistics for hypothesis tests of the significance of GOV, AUD, and TDIR in the post-GAAR period based on the coefficient estimates for BTG1, BTG2, and TPRICE shown in Panel A. For the general tax avoidance proxy measures of BTG1 and BTG2, we observe that the t-statistics for AUD and TDIR (and GOV) are not statistically significant after the GAAR, which is consistent with the finding that the new tax rules reduced the effectiveness of AUD and TDIR in curbing tax avoidance in the post-GAAR period. However, we do find the t-statistic for AUD to be significant for TPRICE (p < 0.01). While the new tax rules may have been effective in reducing tax avoidance generally, this monitoring

24

mechanism remained important in the post-tax reform period for transfer pricing. [Insert Table 4 Here] 5.5. Robustness checks We perform a number of robustness checks to evaluate the reliability of the regression results presented in Tables 3 and 4 of the paper. First, a firm may not necessarily change its tax avoidance activities in the first year of the GAAR implementation. We thus repeated our analysis after excluding the 2008 tax year (i.e., the first year of GAAR implementation) from our sample and use the 2009–2010 tax years as the post-GAAR period. Our (untabulated) regression results for GAAR and the interaction terms GOV*GAAR, AUD*GAAR, and TDIR*GAAR are similar to those reported in Tables 3 and 4. Second, we added each of the test variables GAAR, GOV, AUD, and TDIR and the interaction terms GOV*GAAR, AUD*GAAR, and TDIR*GAAR consecutively into the regression model to test the stability of the regression coefficients and robustness of our empirical findings. Our (untabulated) results show that the regression coefficients for GAAR, GOV, AUD, and TDIR and interaction terms GOV*GAAR, AUD*GAAR, and TDIR*GAAR were stable and had identical signs and similar levels of statistical significance to those presented in Table 4. Finally, as our last robustness check, we also made use of the fixed-effects panel regression model technique (e.g., Hsiao 2003; Wooldridge 2010) to control for the possibility of correlated omitted variable bias in our regression models. Our (untabulated) regression results for GAAR and the interaction terms GOV*GAAR, AUD*GAAR, and TDIR*GAAR are comparable to those shown in Tables 3 and 4.

25

6. Conclusion This study analyzes the effect of the introduction of the GAAR on January 1, 2008 on corporate tax avoidance in China. We find that the GAAR has been effective in curbing tax avoidance in China. The reduction in tax avoidance following the implementation of the GAAR appears to be attributable to the introduction of new and stringent tax legislation and the overall consolidation of Chinese tax law. We also report that the effects of engaging a Big Four auditor and having a director with expertise in deterring tax avoidance were significantly reduced in the post-implementation period. Ultimately, it appears that the GAAR moderated the effects of and substituted for these monitoring and disciplining mechanisms. Our findings are important because the GAAR represents the first major corporate tax reform in China. The effect of China’s tax reform is consistent with the idea that government policies and intervention play a key role in dictating corporate behavior and in allocating resources in China (Chen et al. 2011b). Our assertions and findings are consistent with research by Xu et al. (2011) and Huang (2015), which shows that tax enforcement acts as a corporate governance mechanism in China by ensuring that firms behave according to tax laws and refrain from engaging in aggressive tax avoidance activities. The governance role of tax enforcement is likely to be quite pronounced after the introduction of a major tax reform, such as the GAAR in China. Thus, our study provides some key insights into tax avoidance in China that should be of interest to policymakers and regulators in other relation-based economies around the world. Our study is subject to at least one limitation. The sample is based on Chinese firms listed on the Shanghai or Shenzhen stock exchanges and as such, the results could potentially differ for Chinese firms listed separately on the Hong Kong stock exchange. Future research could consider this issue.

26

References Aharony, J., Wang, J., & Yuan, H. 2010. Tunneling as an incentive for earnings management during the IPO process in China. Journal of Accounting & Public Policy, 29 (1), 1–26. An, Z. & Tan, C. 2014. Taxation and income shifting: Empirical evidence from a quasi-experiment in China. Economic Systems, 38 (4), 588–596. Beasley, M.S. 1996. An empirical analysis of the relation between the board of director composition and financial statement fraud. The Accounting Review, 71 (4), 443–465. Cai, H., & Liu, Q. 2009. Competition and corporate tax avoidance: evidence from Chinese industrial firms. The Economic Journal, 119 (537), 764–795. Carter, D.A., D’Souza, F., Simkins, B.J., & Simpson, W.G. 2010. The gender and ethnic diversity of U.S. boards and board committees and firm performance. Corporate Governance: An International Review, 18 (5), 396–414. Chan, K.H., Lin, K.Z., & Mo, P.L.L. 2010. Will a departure from tax-based accounting encourage tax noncompliance? Archival evidence from a transition economy. Journal of Accounting & Economics, 50 (1), 58– 73. Chan, K.H., Mo, P.L.L., & Zhou, A.Y. 2013. Government ownership, corporate governance and tax aggressiveness: evidence from China. Accounting & Finance, 53 (4), 1029–1051. Chen, S., DeFond, M., & Park, C. 2002. Voluntary disclosure of balance sheet information in quarterly earnings announcements. Journal of Accounting & Economics, 33 (2), 229–251. Chen, G., Firth, M., Gao, D.N., & Rui, O.M. 2006. Ownership structure, corporate governance and fraud: evidence from China. Journal of Corporate Finance, 12 (3), 424–448. Chen, S., Chen, X., Cheng, Q., & Shevlin, T. 2010. Are family firms more tax aggressive than non-family firms? Journal of Financial Economics, 95 (1), 41–61. Chen, S., Sun, Z., Tang, S., & Wu, D. 2011a. Government intervention and investment efficiency: Evidence from China. Journal of Corporate Finance, 17 (2), 259–271. Chen, C.J.P., Li, Z., Su, X., & Sun, Z. 2011b. Rent-seeking incentives, corporate political connections, and the control structure of private firms: Chinese evidence. Journal of Corporate Finance, 17 (2), 229–243. Cheng, C.S.A., Huang, H.H., Li, Y., & Stanfield, J. 2012. The effect of hedge fund activism on corporate tax avoidance. The Accounting Review, 87 (5), 1493–1526. Chua, R.Y.J., Morris, M.W., & Ingram, P. 2009. Guanxi vs networking: distinctive configurations of affect- and cognition-based trust in the networks of Chinese vs American managers. Journal of International Business Studies, 40 (3), 490–509. Cull, R., & Xu, L. 2005. Institutions, ownership and finance: The determinants of profit reinvestment among Chinese firms. Journal of Financial Economics, 77 (1), 117–146. DeChow, P.M., Sloan, R.G., & Sweeney, A.P. 1995. Detecting earnings management. The Accounting Review, 70 (2), 193–225. Deloitte, 2013. Taxation and investment in China 2013: Reach, relevance and reliability. Available at: http://www.deloitte.com/assets/DcomGlobal/Local%20Assets/Documents/ Tax/Taxation%20and%20Investment%20Guides/2013/dttl_tax_guide_2013_China.pdf. Desai, M.A., & Dharmapala, D. 2006. Corporate tax avoidance and high-powered incentives. Journal of Financial Economics, 79 (1), 145–179. Dyreng, S, Hanlon, M., & Maydew, E. 2008. Long-run corporate tax avoidance. The Accounting Review, 83 (1), 61– 82. Dyreng, S.D., Hanlon, M., & Maydew, E. 2010. The effects of managers on corporate tax avoidance. The Accounting Review, 85 (4), 1163–1189. Eberhart, A.C., & Senbet, L.W. 2010. The effects of executives on corporate tax avoidance. The Accounting Review, 85 (4), 1163–1189. Ernst & Young 2014a. China released administrative measures for general anti-avoidance rules (GAAR) Available at: http://www.ey.com/Publication/vwLUAssets/EY-CTIN-2014002-ENG/$FILE/EY-CTIN-2014002-ENG.pdf. Ernst & Young 2014b. Global transfer pricing tax authority survey. Available at: http://www.ey.com/Publication/vwLUAssets/EY-global-transfer-pricing-tax-authority-survey/$FILE/ey-2014global-transfer-pricing-tax-authority-survey.pdf.

27

Fan, J., Wong, T., & Zhang, T., 2007. Politically connected CEOs, corporate governance, and Post-IPO performance of China’s newly partially private firms. Journal of Financial Economics, 84 (2), 330–357. Francis, J.R. 2004. What do we know about audit quality? The British Accounting Review, 36 (4), 345–368. Francis, B. Hasan I., & Sun X. 2009. Political Connections and the Process of Going Public: Evidence from China. Journal of International Money & Finance 28 (4), 696-719. Frank, M.M., Lynch, J.L., & Rego, S.O. 2009. Are financial and tax reporting aggressiveness reflective of broader corporate policies? The Accounting Review, 84 (2), 467–496. Global Financial Integrity. 2012. Illegal financial flows from China and the role of trade misinvoicing. Available at: http://www.gfintegrity.org/storage/gfip/documents/reports/ChinaOct2012/gfi-china-oct2012-report-web.pdf. Global Times. 2009. SAT steps up watch on transfer pricing by multinationals. July 31, 2009. Available at: http://www.globaltimes.cn/business/china-economy/2009-07/453131.html. Government Accountability Office (GAO). 2008. Tax Administration. Comparison of the Reported Tax Liabilities of Foreign- and US-Controlled Corporations, 1998–2005. Government Accountability Office, Washington, DC. Graham, J., & Tucker, A. 2006. Tax shelters and corporate debt policy. Journal of Financial Economics, 81 (3), 563–594. Gupta, S., & Newberry, K. 1997. Determinants of the variability in corporate effective tax rates: evidence from longitudinal data. Journal of Accounting & Public Policy, 16 (1), 1–34. Hanlon, M., & Heitzman, S. 2010. A review of tax research. Journal of Accounting & Economics, 50 (2-3), 127– 178. Hanlon, M., & Slemrod, J. 2009. What does tax aggressiveness signal? Evidence from stock price reactions to news about tax shelter involvement. Journal of Public Economics, 93 (1-2), 126–141. He, X., Wong, T.J., & Young, D. 2012. Challenges for implementation of fair value accounting in emerging markets: evidence from China. Contemporary Accounting Research, 29 (2), 538–562 Healy, P.M. 1985. The effect of bonus schemes on accounting decisions. Journal of Accounting & Economics, 7 (1– 3), 85–107. Hoopes, J., Mescall, D., & Pittman, J. 2012. Do IRS audits deter corporate tax aggressiveness? The Accounting Review, 87 (5), 1603–1639. Hsiao, C. 2003. Analysis of Panel Data. Second Edition. New York, NY, Cambridge University Press. Huang, W. 2015. Tunneling through related-party loan guarantees: evidence from a quasi-experiment in China. Review of Quantitative Finance & Accounting, in press. Internal Revenue Service (IRS). 2013. Pacific Association of Tax Administrators (PATA) transfer pricing documentation package. Available at: http://www.irs.gov/uac/Pacific-Association-of-Tax-AdministratorsFinalizes-Transfer-Pricing-Documentation. Jameson, M., Prevost, A., & Puthenpurackal, J. 2014. Controlling shareholders, board structure, and firm performance: Evidence from India. Journal of Corporate Finance, 27, 1–20. Klassen, K., & Laplante, S. 2012a. The effect of foreign reinvestment and financial reporting incentives on crossjurisdictional income shifting. Contemporary Accounting Research 29 (3), 928–955. Klassen, K., & Laplante, S. 2012b. Are U.S. multinational corporations becoming more aggressive income shifters? Journal of Accounting Research 50 (5), 1245–1286. Kusnadi, Y., Yang, Z. & Zhou, Y. 2015. Institutional development, state ownership, and corporate cash holdings: evidence from China. Journal of Business Research 68(2), 351–359. Lanis, R., & Richardson, G. 2011. The effect of board of director composition on corporate tax aggressiveness. Journal of Accounting & Public Policy, 30 (1), 50–70. Li, J. 2007. Fundamental enterprise income tax reform in China: motivations and major changes. Bulletin for International Taxation, 61 (12), 519–528. Li, J., & Huang, H. 2008. The transformation of Chinese enterprise income tax: Internationalization and Chinese innovations. CLPE Research Paper No. 41. Lin, K.Z., Mills, L.F., & Zhang, F. 2014. Public versus private firm responses to the tax rate reduction in China. Journal of the American Taxation Association, 36 (1), 137–163. Liu, J., Wang, Y., & Wu, L. 2011. The effect of guanxi on audit quality in China. Journal of Business Ethics, 103 (4), 621–638. Lisowsky, P. 2010. Seeking shelter: Empirically modeling tax shelters using financial statement data. The Accounting Review, 85 (5), 1693–1720. Lisowsky, P., Robinson, L., & Schmidt, A. 2013. Do publicly disclosed tax reserves tell us about privately disclosed tax shelter activity? Journal of Accounting Research, 51 (3), 583–629.

28

Lo, A.W.Y., Wong, R.M.K., & Firth, M. 2010. Tax, financial reporting, and tunneling incentives for income shifting: An empirical analysis of the transfer pricing behavior of Chinese-listed companies. Journal of the American Taxation Association, 31 (2), 1–26. Maksimovic, V., & Titman, S. 1991. Financial policy and reputation for product quality. Review of Financial Studies, 4 (1), 175–201. Manzon, G., & Plesko, G. 2002. The relation between financial and tax reporting measures of income. Tax Law Review, 55 (2), 175–214. Matsumura, E.M., & Tucker, R.R., 1992. Fraud detection: A theoretical foundation. The Accounting Review, 67 (4), 753–782. Maydew, E.L., & Shackelford, D.A. 2007. The changing role of auditors in corporate tax planning. In Auerbach, A.J., Hines, J.R., & Slemrod, J., (Eds.). Taxing corporate income in the 21st century (pp. 307–337). Cambridge University Press. Cambridge, UK. Organisation for Economic Cooperation & Development (OECD). 2006. The OECDs project on harmful tax practices: 2006 update on progress in member countries. OECD, Washington, DC. Available at: http://www.oecd.org/dataoecd/1/17/37446434.pdf. PWC, 2012. The peoples’ republic of China tax facts and figures–2012. Available at: http://www.pwchk.com/webmedia/doc/634737952050870749_cn_tax_facts_figures_2012.pdf. Rego, S.O. 2003. Tax-avoidance activities of U.S. multinational firms. Contemporary Accounting Research, 20 (4), 805–833. Rezaee, Z. 2005. Causes, consequences, and deterrence of financial statement fraud. Critical Perspectives on Accounting, 16 (3), 277–298. Richardson, G., & Lanis, R. 2007. Determinants of the variability in corporate effective tax rates and tax reform: evidence from Australia. Journal of Accounting & Public Policy, 26 (6), 689–704. Richardson, G., Taylor, G., & Lanis, R. 2013a. The impact of board of director oversight characteristics on corporate tax aggressiveness: an empirical analysis. Journal of Accounting & Public Policy, 32 (3), 68–88. Richardson, G., Taylor, G., & Lanis, R. 2013b. Determinants of transfer pricing aggressiveness: empirical evidence from Australian firms. Journal of Contemporary Accounting & Economics, 9 (2), 136–150. Scholes, M., & Wolfson, M. 1992. Taxes and business strategy: A planning approach. Prentice-Hall, Englewood Cliffs, NJ. Shevlin, T., Tang, T.Y.H., & Wilson, R. 2012. Domestic income shifting by Chinese listed firms. Journal of the American Taxation Association, 34 (1), 1–29. Sikka, P. 2010. Smoke and mirrors: corporate social responsibility and tax avoidance. Accounting Forum, 34 (3-4), 153–168. State Administration of Taxation (SAT). 2014. Replies to questions from the press regarding administrative measures on the general anti-avoidance rule (trial) by in-charge SAT officials. Available at: http://www.chinatax.gov.cn/2013/n2925/n2957/c1396975/content.html. Stickney, C., & McGee, V. 1982. Effective corporate tax rates: the effect of size, capital intensity, leverage, and other factors. Journal of Accounting & Public Policy, 1 (2), 125–152. Sutherland, D. Matthews, B., & El-Gohari, A. 2012. An exploration of how Chinese companies use tax havens and offshore financial centres: ‘round-tripping’ or ‘capital-augmenting’ OFDI TMD Working Paper Series, University of Nottingham. Available at: http://www.tmd-oxford.org/sites/www.tmdoxford.org/files/publications/ SLPTMD-WP-042.pdf. Taylor, G., & Richardson, G. 2012 International corporate tax avoidance practices: evidence from Australian firms. International Journal of Accounting, 47 (4), 469–496. Taylor, G., & Richardson, G. 2014. Incentives for corporate tax planning and reporting: empirical evidence from Australia. Journal of Contemporary Accounting & Economics, 10 (1), 1–15. Taylor, G., Richardson, G., & Lanis, R. 2015. Multinationality, tax havens, intangible assets, and transfer pricing aggressiveness: an empirical analysis. Journal of International Accounting Research, 14 (1), 25–57. Tu, G., Lin, B., & Liu, F. 2013. Political connections and privatization: evidence from China. Journal of Accounting & Public Policy, 32 (2), 114–135. Usmen, N. 2012. Transfer prices: a financial perspective. Journal of International Financial Management & Accounting, 23 (1), 1–22. Wang, Z. 2013. China’s Top 500 Slows Down. China.org.cn, August 31, 2013. Wang, J. 2016. The Chinese approach to transfer pricing: problems faced and paths to improvement. British Tax Review (1), 89–118. Wooldridge, J.M. 2010. Econometric analysis of cross section and panel data, 2nd ed., MIT Press, Cambridge, MA.

29

Wu, W. 2013. Institutional environment, ownership and firm taxation. Economics of Transition 21 (1), 17–51. Wu L., Wang, Y., Luo, W., & Gillis, P. 2012. State ownership, tax status and size effect of effective tax rate in China. Accounting & Business Research, 42 (2), 97–114. Wu S., Xu N. & Yuan Q. 2009. State control, legal investor protection, and ownership concentration: evidence from China. Corporate Governance: An International Review 17 (2), 176–192. Xiao, H. & Yuan J. 2007. Ownership structure, board composition and corporate voluntary disclosure. Managerial Auditing Journal 22 (6), 604–619. Xu W., Zheng, Y. & Zhang, J. 2011. Tax enforcement as a corporate governance mechanism: empirical evidence from China. Corporate Governance: An International Review, 19 (1), 25–40. Zeng, T. 2011. Ownership concentration, state ownership, and effective tax rates: evidence from China’s listed firms. Accounting Perspectives, 9 (4), 271–289. Zhang, H. 2012. How does state ownership affect tax avoidance? Evidence from China. Working paper #2012/1318, Singapore Management University, Singapore.

30

APPENDIX A Description of the Desai & Dharmapala (2006) Method for Computing the Book-tax Gap Residual Taxable income is calculated as TIit = accounting income tax expense scaled by the corporate statutory tax rate of 33% (pre-2008) or 25% (post-2008). The BTG is calculated by subtracting TI from pre-tax accounting income (AI): BTGit = AIit – TIit. The BTG is scaled by lagged total assets. Total accruals (TA) were calculated for each firm in each year using the measure of total accruals developed by Healy (1985). TAs is considered to measure the earnings management component of BTG and is computed as follows: TAit = EBEIit – CFOit (1) where i = firms 1–517; t = financial years 2006–2010; TA = total accruals; EBEI = earnings before extraordinary items (i.e., pre-tax income); and CFO = cash flows from operations. The following OLS regression is performed to account for the component of BTG attributable to earnings management: BTGit = β1TAit + μit + εit (2) where BTG = the book-tax difference scaled by lagged total assets; TA = total accruals scaled by lagged total assets; μ = the residual; and ε = the error term. The residual value of BTG is deemed by Desai & Dharmapala (2006) to reflect corporate tax avoidance (CTA): CTAit = μit + εit.

31

TABLE 1 Sample Industry Distribution Industry Description Agriculture Mining Manufacturing and petrochemicals Utilities Construction Transportation Telecommunications Wholesale trade and retail Real Estate Social services Media and communication Conglomerate Other Total

No. of Firms 5 22 259 36 14 38 18 30 17 45 10 18 5 517

No. of Firm-years 25 110 1,295 180 70 190 90 150 85 225 50 90 25 2,585

Relative Frequency (%) 0.97 4.26 50.10 6.96 2.71 7.35 3.48 5.80 3.29 8.70 1.93 3.48 0.97 100%

32

TABLE 2 Descriptive Statistics Variable N Mean Std. Dev. Minimum Median Maximum BTG1 2,585 0.003 0.050 –0.830 0.003 0.323 BTG2 2,585 –0.001 0.050 –0.839 0.000 0.320 TPRICE 2,585 0.494 0.500 0.000 0.000 1.000 GAAR 2,585 0.600 0.490 0.000 1.000 1.000 GOV 2,585 0.122 0.327 0.000 0.000 1.000 AUD 2,585 0.152 0.359 0.000 0.000 1.000 TDIR 2,585 0.087 0.283 0.000 0.000 1.000 SIZE 2,585 22.796 1.267 15.770 22.595 30.231 LEV 2,585 0.568 0.222 0.021 0.578 4.465 CINT 2,585 0.075 0.060 0.000 0.060 0.710 INVINT 2,585 0.172 0.165 0.000 0.130 0.730 RDINT 2,585 0.003 0.016 0.000 0.000 0.410 FOR 2,585 0.057 0.174 0.000 0.000 0.833 ROA 2,585 0.049 0.066 –0.830 0.040 0.654 MKTBK 2,585 0.294 0.160 –0.955 0.270 0.935 AGE 2,585 10.171 4.069 0.000 10.439 20.047 Variable definitions: BTG1 = pre-tax accounting income less taxable income (where taxable income is computed as income tax expense scaled by the statutory corporate tax rate of 33% (pre-2008) or 25% (post-2008)) using the method developed by Manzon & Plesko (2002) scaled by lagged total assets; BTG2 = book-tax gap residual calculated using the method developed by Desai & Dharmapala (2006); TPRICE = a dummy variable, coded 1 if the parent entity cannot substantiate that related party transactions group members were made on an arm’s length or commercial basis, and 0 otherwise; GAAR = a period dummy variable, coded 1 if the observations correspond to the post-GAAR period (i.e., 2008–2010), and 0 otherwise (i.e., 2006–2007); GOV = a dummy variable, coded 1 if the firm is majority (> 50%) owned and therefore controlled by the government and is thus considered to be an SOE, and 0 otherwise; AUD = a dummy variable, coded 1 if the firm employs a Big 4 external auditor, and 0 otherwise; TDIR = a dummy variable, coded 1 if at least one member of the firm’s board of directors has tax expertise (i.e., prior experience in tax accounting or tax-related audit work), and 0 otherwise; SIZE = the natural logarithm of total assets; LEV = long-term debt scaled by total assets; CINT = net property, plant and equipment scaled by total assets; INVINT = net inventory scaled by total assets; RDINT = research and development expenditure scaled by total assets; FOR = the total number of foreign incorporated subsidiaries scaled by the total number of subsidiaries; ROA = pre-tax income scaled by total assets; MKTBK = the market value of equity scaled by the book value of equity; and AGE = firm age (in years).

33

TABLE 3 Regression Results – H1 Variable

Predicted Sign

Intercept

+/–

GAAR

–

SIZE

+

LEV

+/–

CINT

+

INVINT

–

RDINT

+

FOR

+

ROA

+

MKTBK

+/–

AGE

+

INDSEC

?

BTG1 OLS –0.064 (–4.07)*** –0.002 (–1.70)** 0.001 (2.23)** 0.020 (2.39)** 0.006 (0.69) –0.019 (–4.30)*** 0.129 (1.80)** 0.008 (2.09)** 0.375 (12.80)*** 0.027 (3.28)*** 0.001 (0.56) Yes

BTG2 OLS –0.072 (–4.86)*** –0.003 (–1.73)** 0.002 (2.47)*** 0.021 (2.47)** 0.005 (0.54) –0.018 (–4.29)*** 0.128 (1.78)** 0.008 (2.23)** 0.380 (13.01)*** 0.027 (3.28)*** 0.001 (0.72) Yes

TPRICE LOGIT –8.682 (–9.21)*** –0.136 (–2.36)*** 0.409 (12.011)*** –0.702 (–2.71)*** –0.103 (–0.28) 0.304 (1.25) 3.797 (1.56)* –0.094 (–0.47) 0.672 (0.96) –0.250 (–1.05) 0.019 (2.64)*** Yes

Adj. R2 / Pseudo R2 (%) 26.83% 27.91% 9.66% N 2,585 2,585 2,585 Regression coefficient estimates with t-statistics reported in parentheses. The t-statistics are based on the Huber/White/Sandwich estimator of standard errors (see Wooldridge 2010). Variable definitions: INDSEC = industry sector dummy variable, coded 1 if the firm is represented in the specific industry sector, and 0 otherwise; and see Table 2 for other variable definitions. *, **, and *** indicate significance at the 0.10, 0.05, and 0.01 levels, respectively. The p-values are one-tailed for directional hypotheses and two-tailed otherwise.

34

TABLE 4 Regression Results – H2, H3 and H4 BTG1a BTG2a TPRICEa OLS OLS LOGIT Panel A: Regression coefficient estimates with t-statistics reported in parentheses. The t-statistics are based on the Huber/White/Sandwich estimator of standard errors (see Wooldridge 2010). –7.400 –0.070 –0.078 Intercept +/– (–7.37)*** (–3.86)*** (–4.59)*** –0.004 –0.004 –0.157 GAAR – (–2.26)** (–2.53)*** (–2.35)*** –0.001 –0.001 0.126 GOV – (–0.01) (–0.03) (0.94) 0.002 0.002 0.037 GOV*GAAR + (0.74) (0.70) (0.22) –0.005 –0.005 –0.160 AUD – (–1.63)* (–1.56)* (–1.25) 0.003 0.004 0.313 AUD*GAAR + (1.02) (1.31)* (1.96)** –0.011 –0.010 –0.286 TDIR – (–2.26)** (–2.22)** (–1.74)** 0.013 0.013 0.120 TDIR*GAAR + (2.47)*** (2.46)*** (0.59) 0.002 0.002 0.351 SIZE + (2.42)*** (2.59)*** (9.50)*** 0.021 0.020 –0.631 LEV +/– (2.41)** (2.37)** (–2.48)*** 0.006 0.006 –0.057 CINT + (0.60) (0.60) (–0.15) –0.019 –0.019 0.321 INVINT – (–4.46)*** (–4.29)*** (1.23) 0.149 0.127 4.315 RDINT + (2.10)** (1.75)** (1.68)** 0.009 0.009 –0.096 FOR + (2.45)*** (2.40)*** (–0.46) 0.375 0.381 0.689 ROA + (12.75)*** (13.04)*** (1.03) 0.028 0.028 –0.320 MKTBK +/– (3.37)*** (3.30)*** (–1.33) 0.001 0.001 0.019 AGE + (0.62) (0.60) (2.57)*** INDSEC ? Yes Yes Yes Variable

Predicted Sign

Adj. R2/Pseudo R2 (%) 27.54% 27.40% 10.46% N 2,585 2,585 2,585 Variable Hypothesis BTG1 BTG2 TPRICE Panel B: t-statistics for hypotheses tests of significance of the explanatory variables in the postGAAR period based on the coefficient estimates reported in Panel A. The t-statistics are calculated as follows: (βi + βj)/[Var(bi) + Var(bj) + 2cov(bi, bj)]1/2. GOV Β2 + β3 = 0 –0.95 –0.88 1.40 –0.36 –0.10 –4.19*** AUD Β4 + β5 = 0 0.86 0.91 –1.44 TDIR Β6 + β7 = 0 Variable definitions: GAAR*(GOV or AUDIT or TDIR) = an interaction term comprising the GAAR dummy variable multiplied by each of GOV, AUDIT and TDIR; and see Tables 2 and 3 for other

35

variable definitions. *, **, and *** indicate significance at the 0.10, 0.05, and 0.01 levels, respectively. The p-values are one-tailed for directional hypotheses and two-tailed otherwise.

36