To encourage economic development in specific regions and industries, the Chinese Central and

Transcription

1 Domestic Income Shifting by Chinese Listed Firms Terry Shevlin University of Washington Tanya Tang The University of British Columbia, Okanagan Ryan Wilson University of Iowa Abstract To encourage economic development in specific regions and industries, the Chinese Central and local governments offer a series of corporate income tax incentives (tax exemptions, reduced tax rates, tax holidays and tax refunds). In China, parent and subsidiary companies are consolidated for financial reporting, but not tax, purposes. We take advantage of a unique disclosure in the tax footnotes of Chinese listed firms to examine income shifting among consolidated group members in response to these incentives. We find that intangible intensive groups ( firms ) and firms concerned with meeting minimum earnings thresholds to issue equity shift greater amounts of income. We find no evidence that high concentrations of either Central or local government ownership affect the level of income shifting. Keywords: Income shifting; tax incentives; book-tax differences. 0

2 1. Introduction We use a sample of 320 firm-year observations of Chinese listed firms from to examine how firms respond to tax incentives provided by the Chinese governments. Enterprise zones and tax concessions are common tools for countries and states to attract investment. The Chinese tax system is no different. In an effort to attract investment to specific regions and industries the Chinese Central and local governments offer favorable tax rates for domestic Chinese corporations operating in five special economic zones, 32 economic and technology development zones, 13 free trade zones, and 52 high-tech development zones (Wu et al. 2007). The differences in tax rates among the members of a consolidated group stemming from industry-specific, enterprise-specific, and region-specific tax incentives provide strong incentives for Chinese domestic firms to shift income within the consolidated group. In many respects, to the extent that legal and non-tax considerations do not prevail, incomeshifting among entities within a consolidated group is a desirable (even ideal) strategy. Many tax-minimizing strategies result in lowering reported book income, which can lead to increased non-tax costs arising from capital market pressure or agency problems (Mills and Newberry 2001; Cloyd et al. 1996). 1 Chinese firms prepare their financial statements on a consolidated basis while taxes are levied on each firm (parent and subsidiaries) as a separate legal entity. That is, unlike U.S. firms, Chinese firms are not taxed on a consolidated basis. Therefore, shifting income from entities subject to high-tax rates to those subject to low-tax rates within a domestic 1 Scholes et al. (2009) argue that effective tax planning involves firms trading off tax benefits (i.e., reduced tax payments) and non-tax costs such as reporting lower book income resulting in possible lower market value, lower manager compensation and a greater possibility of debt covenant violations. Thus, a desirable tax planning strategy is one that reduces taxes but does not diminish income reported to investors (e.g. Shevlin 2002, Weisbach 2002, Plesko 2004, McGill and Outslay 2004). 1

3 Chinese consolidated group can substantially reduce firms tax burdens and increase reported after-tax book income. Intra-group income shifting can generate significant tax savings but it is often difficult for investors, tax authorities, and researchers to quantify the magnitude of such activity. Prior studies such as Harris (1993) and Klassen et al. (1993) examine whether patterns in reported income and taxes are consistent with the incentives to shift income provided by the Tax Reform Act of Our study avoids the need to employ an obscure proxy for income shifting. A unique disclosure in the financial statement footnotes of Chinese firms allows us to quantify the extent of domestic income shifting within the consolidated group. We measure income-shifting using a line item(s) in the tax footnote that reconciles the consolidated firm s effective tax rate to the applicable statutory tax rate applied to the parent company. This line item (or items) represents the total difference between (1) the taxes that would have been paid if the consolidated entities income had been taxed at the parent company s applicable tax rate, and (2) the taxes computed using the tax rate applied to each member of the group. We refer to this line item in the rate reconciliation as the tax rate differential adjustment (TRDA); Examples of the disclosure are provided in Appendix 1 and discussed in Section 3. 2 However, like all proxies our measure is not without its own problems which we discuss in Section 3 (for example, because TRDA is an outcome measure we do not observe pre-shifted income and thus we must assume TRDA is increasing in the amount of income actually shifted). We develop cross-sectional predictions about which groups are more likely to shift income among members. Bartelsman and Beetsma (2003) argue that, relative to tangible goods, 2 We discuss below how this Chinese tax rate disclosure compares to and differs from the rate reconciliation provided by U.S. multinationals with foreign operations. 2

4 managers have more discretion in setting transfer prices for intellectual property or knowledge intensive goods. Consequently, because of the lower costs of shifting income, we expect intangible intensive firms (e.g. pharmaceutical and software firms) can more easily shift income via accounting-based methods than heavy manufacturing and retail firms. We also expect firms that want to issue equity to shift more income to lower their taxes increasing reported after-tax earnings. Chinese listing rules require firms to maintain a minimum return-on-equity (ROE) before permission is granted to issue new equity (and to report positive profits to remain listed) (see Chen and Yuan 2004). Thus firms wishing to raise equity (or to remain listed) have strong incentives to manage their reported earnings to meet these ROE benchmarks. We also examine whether intangible intensive firms (firms facing lower costs) that wish to raise equity (firms having greater benefits) shift more income than other firms. That is, we test for an interaction effect between intangible firms and firms with equity issue incentives. Finally, we also examine whether firms in which the government is the major shareholder shift more income. Before a 2002 tax reform, income taxes collected from locally owned enterprises were exclusively assigned to local governments and income taxes collected from central-owned enterprises, national and foreign banks, and non-bank financial and insurance institutions were assigned exclusively to the Central government (SAT 1995). At the same time, the government receives less than 100% of the listed firm s dividends because it owns less than 100% of the shares. Thus prior to 2002, we expect less income shifting into lower-tax subsidiaries by government-controlled (SOE) firms relative to our other sample firms because they get to keep 100% of the tax revenue to fund the governments other activities. Beginning in 2002, tax revenues collected from centrally (locally) controlled firms were no longer exclusively allocated to the Central (local) government but are approximately equally-shared by the central 3

5 and local governments (SAT 2001). As a result, the incentive of government-controlled firms to shift income to reduce income taxes paid increased following the tax sharing agreement. If the government-controlled firms shift income into lower-taxed subsidiaries, the cash flows and earnings retained by the listed firms are increased. The government-controlled firms could then tunnel resources back to the state owned controlling shareholder that launched it at the expense of minority shareholders (see Lo et al for evidence supporting the view that listed firms shifted profits to government shareholders via favorable transfer prices). Thus we predict more income shifting by SOE firms beginning in However, complicating these predictions, even in SOE firms there is a separation between the managers and shareholders giving rise to an agency conflict. SOE firm managers might want to reduce their tax burden to increase after-tax cash flows under their control. In addition, Adhikari, Derashid, and Zhang (2006) maintain that in relationship-based economies, government privileges are given to selected firms for both personal and policy-related reasons. These favors could be special tax treatments and/or subjecting the tax positions of favored or connected firms to less scrutiny, the result being that subjective tax enforcement provides a mechanism for the government to promote some favored firms over others. In this scenario, SOE firms are predicted to shift more income to lower their tax burden. On the other hand, managers of SOE firms are often political appointees relying on the government for their continued appointment. As discussed below, in the Chinese economy, the government (both Central and local) pursue political and social goals requiring tax revenue and thus managers sharing these goals are more likely to be appointed and/or continue to hold their positions. This scenario predicts less taxmotivated income shifting by state owned firms. Thus it is an empirical question as to whether state owned firms shift more or less income. 4

6 Consistent with expectations, we find that the amount of income shifting is positively associated with the level of firms intangible assets. Managers of intangible intensive firms appear to take advantage of greater discretion in setting transfer prices to shift income into low tax jurisdictions. The fact that we observe higher levels of income shifting among intangible intensive firms is also consistent with these firms using accounting-based methods of income shifting (transfer pricing) rather than shifting real operations into low tax jurisdictions. Also consistent with expectations, we find firms that successfully obtain a rights offering in any of the next three years engage in greater amounts of income shifting. Firms concerned about maintaining the statutory minimum return on equity (ROE) that is required in order to issue new equity, appear to use income shifting as one means of keeping their ROE above the threshold. We find only mixed evidence of an interaction effect between the level of intangible assets and the desire to issue equity. Inconsistent with our expectations, we find no evidence that income shifting varies as a function of whether the largest shareholder is either the Central or local government, either before or after Our paper makes several contributions. First, we contribute to the income shifting literature by exploiting a unique disclosure in Chinese firms financial statements that allows a more precise estimate of the magnitude of income shifted (or taxes saved). Second, by studying income shifting within a country, rather than across countries as much of the prior literature has done, we avoid correlated omitted variables or inference problems because of unknown and difficult to control differences across countries that could arise from legal, capital market, and accounting differences. Third, in addition to adding to the income-shifting literature, our study contributes to research examining the incentive effects specific to Chinese capital markets. Specifically, we 5

7 find evidence of an incentive effect from the requirement to maintain after-tax ROE above certain minimum thresholds in order to issue additional equity. However, despite high levels of ownership by the Central and local government of many of our sample firms, we find no evidence that government ownership is related to income shifting. The remainder of the paper is structured as follows. In the next section we discuss prior literature, Chinese institutional details and develop our hypotheses. Section 3 describes the research design and results. Section 4 concludes. 2. Prior Literature, Institutional Details and Hypothesis Development Shackelford and Shevlin (2001) note that when studying the effects of different tax regimes across countries other forms of cross-country differences (legal systems, financial markets, etc.) potentially represent correlated omitted variables that could affect the inferences drawn from these studies. Our study largely avoids these limitations by examining the effects of different tax rates across regions within a single country. We assume that the jurisdictions within China do not vary to the same extent as cross-country comparisons along these dimensions. In this respect, our study is similar to multistate tax research on income shifting. For example, Klassen and Shackelford (1998) examine aggregated state and Canadian provincial data from and find evidence consistent with firms shifting income to favorably-taxed jurisdictions. Gupta and Mills (2002) find that corporations doing business in multiple states (regions) with different tax treatments have more incentives and opportunities for income-shifting. Figure 1 illustrates the different related parties that provide potential avenues for income shifting for a listed firm. Extant literature uses related party transactions (RPT) between the listed firms and affiliates and the state owned enterprise (SOE) outside the box as a measure of 6

8 manipulated transfer pricing. These studies document that RPT with affiliates are associated with earnings management (Jian and Wong 2010), tunneling (i.e., the transfer of assets and profits out of firms for the benefit of those who control them) (Liu and Lu 2007, Aharony et al. 2010, Lo et al. 2010) and tax-induced income shifting (Jacob 1996). Our study, focusing on income shifting within a book-consolidated group, and complements the work of Lo et al. (2010) who investigate income shifting among related parties (also known as affiliated firms) that are not part of a consolidated group. Lo et al. report evidence that firms use related party transactions to shift profits to the listed company if that firm enjoys a reduced corporate tax rate. In a similar vein, Gramlich et al. (2004) document income shifting among the affiliates of Japanese keiretsu firms. However, unlike income shifting between subsidiaries within a consolidated firm, the income shifting documented in these two studies implies a tradeoff. If the two related parties are not part of a consolidated group then any strategy to shift income into one entity will reduce profits for the other party. In contrast, shifting income among consolidated parties does not affect a listed firm s aggregate reported pretax income but can substantially reduce the firm s tax burden (see Appendix 1 for two examples, which are discussed further below). Income shifting activities can be difficult to detect and measure due to data constraints. Prior literature on tax planning demonstrates that book-tax differences (BTDs) are associated with tax aggressiveness (Mills 1998; Wilson 2009; Frank et al. 2009). For example, Mills (1998) finds that proposed U.S. audit adjustments increase as firms BTDs widen, suggesting larger positive BTDs imply aggressive tax reporting. Wilson (2009) and Frank et al. (2009) find evidence of a positive association between BTDs and identified U.S. tax shelter firms. However, book-tax differences can also be driven by normal differences between the tax and financial reporting rules (for example, depreciation rule differences, allowance for uncollectibles vs direct 7

9 write-off for tax purposes) and book-tax differences have also been shown to be reflective of or associated with earnings management (see for example, Phillips, Pincus and Rego 2003 and Hanlon 2005; and Tang and Firth 2010 using Chinese data). 3 This paper extends the above research by utilizing a setting with a unique book-tax difference disclosure the tax rate differential adjustment which provides a direct estimate of the tax savings from shifting income into lower-taxed entities within the consolidated group. 4 The tax rate differential adjustment disclosed by our sample firms has two distinct advantages as a measure of income shifting over the foreign tax rate reconciliation item disclosed under U.S. GAAP. First, the tax rate differential adjustment is a measure of cross-entity (jurisdictional) income shifting within the same country. As a result, currency and litigation risk is less likely to vary between entities/jurisdictions as compared to measures of income shifting across countries. Second, managerial discretion over the decision to designate foreign earnings as permanently reinvested confounds attempts to use the foreign tax rate adjustment disclosed under U.S. GAAP as a measure of income shifting. Only if a firm designates foreign earnings from a 3 Our paper differs from Tang and Firth (2010) in the following respects: (1) Tang and Firth (2010) investigate whether a measure of abnormal book-tax differences is indicative of earnings management, tax planning and their interaction. Abnormal book-tax differences can be used as a broad measure of earnings and tax management whereas our study focuses on a specific type of tax planning (income shifting) represented by TRDA. (2) Tang and Firth (2010) find evidence that firms expecting to issue rights have a high level of abnormal book-tax differences because they have a strong incentive to boost earnings in order to meet the ROE threshold. In this setting they are using book-tax differences as a measure of pre-tax earnings management (consistent with Phillips et al. 2003). In contrast, we are examining whether firms engage in income shifting to lower their tax expense in order to raise after-tax income when they want to issue equity. (3) We are focused on the determinants of income shifting and examine whether the level of intangible assets is associated with this particular type of tax avoidance. 4 The tax rate differential adjustment is not a traditional book-tax difference in that it does not represent a particular expense or revenue item that is either not recognized or recognized in a different period for tax versus financial reporting purposes. However, the tax rate differential adjustment does affect reported tax expense and therefore would increase total book tax differences as they are traditionally measured in academic studies. 8

10 low-tax foreign jurisdiction as permanently reinvested will there be a foreign tax rate differential adjustment in the effective tax rate reconciliation. 5 Chinese listed firms have strong tax incentives to engage in intra-group income shifting. Table 1 lists some of the tax incentives that have been available to Chinese domestic firms in recent years. When a non-parent member captures tax savings from one of these incentives, the result is a differential tax rate adjustment. The Chinese setting makes income shifting in a consolidated group easier and cheaper than cross-border income shifting for multinationals. 6 Further, Chinese tax laws require separate tax reporting: each consolidated subsidiary must calculate its current taxes on an independent legal entity basis. In other words, the profits of one subsidiary cannot be used to offset the losses of another subsidiary. However, under Chinese accounting standards in effect during our sample period, all controlled subsidiaries should consolidate their income for book purposes. The combination of substantial tax rate differences across jurisdictions within China and the lack of book-tax tradeoffs provide a powerful incentive to engage in income shifting. Despite the strong incentives for Chinese firms to shift income within the consolidated group, there are potentially significant costs associated with engaging in income shifting. According to the Ministry of Finance (1994a), business transactions between an enterprise and a 5 For financial reporting purposes, U.S. multinational firms can designate foreign subsidiary earnings as permanently reinvested under APB Opinion No. 23 (Accounting Principles Board 1972). Permanently reinvested earnings (PRE) are earnings from foreign subsidiaries that managers intend to reinvest indefinitely or that managers intend to remit in a tax-free liquidation. Firms with an average foreign tax rate below the statutory U.S. tax rate must recognize the residual U.S. tax on their foreign earnings for financial reporting purposes unless they designate them as permanently reinvested. Recognizing the residual U.S. tax eliminates the need for a foreign tax rate reconciliation adjustment. See Krull (2004) and Graham, Hanlon and Shevlin (2011) for a discussion of U.S. firms designating foreign earnings as PRE. 6 International income shifting is costly and more difficult to undertake due to numerous considerations, such as repatriation restrictions, import and export duties, foreign exchange controls and risks, inflation and political factors (Chan and Chow 1997). 9

11 related party (both inside and outside the consolidated entity) must comply with the arm s length principle. If the Chinese tax bureau determines the arm s length principle has been violated, it has the right to make adjustments and assess penalties and interest related to the adjusted tax position. 7 In addition to concerns about detection by tax authorities, it becomes difficult to use transfer prices to measure the performance of consolidated group members if distorted transfer prices are being used for the purpose of maximizing the profits allocated to entities in the lowest tax jurisdictions. The inability to accurately measure the performance of subsidiaries could lead to inefficient allocation of resources within the firm. 2.1 Hypothesis development: Which firms are more likely to shift income? Firms facing lower costs and greater benefits (incremental to the dollar tax savings) are predicted to shift more income. Thus because intangible intensive firms likely face lower costs to shifting income, our first hypothesis examines whether intangible intensive firms shift more income. For example, a firm could set up a wholly-owned separate legal entity with minimal real operations in the lower-tax jurisdictions, transfer an intangible asset such as a patent, copyright or trademark to the entity, and then use a royalty or similar arrangement to shift income from the high tax to low tax subsidiary. The firms most likely to shift income via transfer pricing without major shifts in investments (or real operating activities) are intangible intensive firms such as pharmaceutical and software firms. We assume that (1) intangible assets are more difficult to value than tangible assets, (2) intangible assets are more easily shifted than tangible assets, and (3) the increased uncertainty associated with valuing intangibles provides managers with greater opportunity to set transfer prices on 7 Article 41 of the Enterprise Income Tax Law of the People s Republic of China (NPC 2007) also indicates firms must apply the arm s length principle. See 10

12 intangibles for income shifting purposes. In contrast, we predict retail and manufacturing firms will be more constrained in their ability to alter transfer prices related to tangible assets and therefore will engage in less income shifting. Thus our first hypothesis: H1: The magnitude of income shifting is positively associated with the level of intangible assets. In our second hypothesis we examine the income shifting incentives of firms requiring additional capital through seasoned equity or equity rights issues. Chinese listed firms are required to meet minimum return on equity (ROE) thresholds if they wish to issue new equity and to avoid suspension or delisting from the stock exchange. 8 The Chinese Securities Regulatory Commission issued these guidelines in response to the strong demand for rights offerings in the early 1990s in an attempt to guide capital to better performing companies (Chen and Yuan 2004). The minimum ROE thresholds provide firms with a strong motivation to report earnings above the threshold. This analysis is similar to prior studies examining earnings management around IPOs and SEOs. These studies argue a possible incentive to manage earnings around these events is to induce investors to pay a higher offer price for the firm s shares (e.g. Teoh et al. 1998). The desire to increase the offer price could provide Chinese firms with an additional incentive to attempt to raise earnings prior to an offering. Chen and Yuan (2004) examine firms incentives to manage earnings in response to these thresholds and the Central regulator s response. We investigate an alternative mechanism to earnings management for firms to achieve the minimum ROE thresholds. We expect firms desiring to raise capital to shift income into lower-taxed 8 Specifically, the rules require the rights issuing firm to have a minimum after-tax ROE of 6% in each of the prior three years, and to maintain an average of 10% over the prior three years. After 2001, firms were required to have a minimum average after-tax ROE of 6% over the prior three years. Finally, if the listed firm reports losses for 3 consecutive years they face suspension and delisting (Chen and Yuan 2004). 11

13 subsidiaries so as to increase their reported after-tax earnings because the benefits of shifting income are higher for firms concerned about minimum ROE thresholds. As a result, the benefits to income shifting for managers of these firms are more likely to outweigh the costs associated with potential detection of altered transfer prices by tax authorities and the cost of reduced performance information caused by altered transfer prices. 9 We test for the effect of these requirements using a dummy variable for whether the firm actually issued equity in any of the three years succeeding the income shifting year. Thus our second hypothesis: H2: The magnitude of income shifting is positively associated with a dummy variable for whether the firm has a rights offering. Our third hypothesis examines whether there is an interaction effect between intangible intensive firms and rights offerings. As discussed above, we predict high levels of intangible assets make it less costly for firms to shift income. We expect that when concerns about ROE thresholds provide firms a strong incentive to shift income that intangible intensive firms will have the greatest ability to respond to this incentive by shifting large amounts of income. In contrast, retail and manufacturing firms with lower levels of intangibles will likely use alternative means of boosting income such as pre-tax earnings management. Thus our third hypothesis: H3: The magnitude of income shifting is positively associated with an interaction term between intangible intensive firms and a dummy variable for whether the firm has a rights offering. 9 There are a number of existing studies examining the use of the tax accrual for earnings management (e.g., Dhaliwal, Gleason, and Mills 2004; Phillips, Pincus, Rego and Wan; and Krull 2004). These studies focus on managerial discretion over the tax accrual via the tax reserves, valuation allowance, and the designation of permanently reinvested earnings. In contrast, we focus on income shifting (an actual tax avoidance strategy) as a means of lowering the firms tax expense to increase after-tax income. 12

14 One institutional feature in China is the number of listed firms that are spin-offs from the profitable units of state owned enterprises (SOEs) (Lo et al. 2010). The SOEs are 100 percent owned by either the Central or local governments. As a part of the transition to a market-based economy, the government decided to raise money by selling partial equity stakes in profitable components of SOEs. In response to this initiative, SOEs transferred high quality assets into profitable units and allowed these units to apply for an initial public offering. The SOEs typically retained unprofitable business units and social programs such as schools and hospitals and remain the largest shareholder in the listed firm (Lo et al. 2010). As a result, the parent SOE requires resources from the listed firm in order to fund the unprofitable programs retained by the SOE. The majority of the listed firms in our sample (90%) have either the local (77%) or Central government (13%) as their largest shareholder. While non-soe owned firms have obvious incentives to shift income into lower taxed entities, thereby maximizing the firms after-tax cash flows, the incentives for SOE firms are less clear. Before a 2002 tax reform, income taxes collected from locally owned enterprises, collective enterprises, joint ventures and individual enterprises were exclusively assigned to local governments and income taxes collected from central-owned enterprises, national and foreign banks, and non-bank financial and insurance institutions were assigned exclusively to the Central government (SAT 1995). At the same time, the government receives less than 100% of the listed firm s dividends because it owns less than 100% of the shares. Thus prior to 2002, we expect less income shifting into lower-tax subsidiaries by government-controlled firms relative to our other sample firms. Given the above we predict that state owned firms have incentives to shift less income prior to 2002: H4: The magnitude of income shifting is lower for state owned firms prior to

15 Beginning in 2002, all corporate tax revenues collected from locally owned enterprises are shared by the central and local governments on a 50:50 basis. In 2003, the central government s share increased to 60 percent (SAT 2001). Corporate tax revenues collected from centrally owned enterprises are also shared by the Central government with the local governments. As a result, the incentive of government-controlled firms to shift income to reduce income taxes paid increased following the tax sharing agreement. Tax revenues collected from centrally (locally) controlled firms were no longer exclusively allocated to the Central (local) government. If the government-controlled firms shift income into lower-taxed subsidiaries, the cash flows and earnings retained by the listed firms are increased. The government-controlled firms could then tunnel resources back to the SOE that launched it at the expense of minority shareholders (see Lo et al for evidence supporting the view that listed firms shifted profits to government shareholders via favorable transfer prices). In addition, Adhikari et al. (2006) argue that in relationship-based economies, government privileges are given to selected firms for both personal and policy-related reasons. One means for providing connected firms with an explicit subsidy would be for the Central or local government to provide the firm with tax incentives or reduced tax rates. Another means is to subject the tax positions of those firms to less scrutiny. This reduced scrutiny would give connected firms an incentive to increase income shifting because detection risk would be significantly reduced. We apply this theory to China and examine whether politically connected firms in our sample exhibit higher levels of income shifting. Following Adhikari et al., we use government ownership as our proxy for political connectedness. Specifically, if political connections reduce detection risk, we would expect greater levels of income shifting for politically connected (state owned) firms beginning in

16 In China tax audit targets are usually selected by the local tax authority, but if the Central government anticipates the incentives of local state owned firms to shift income to lower taxed entities to increase after-tax cash flows, then the Central government can subject those firms to more state administration taxation (SAT) audits thus mitigating the income shifting activities of those firms. Thus both the Central and local governments exercise influence over the audit process and strong connections with either could reduce audit scrutiny of state owned firms. Non-state owned firms are subject to scrutiny by both audit authorities which could mitigate or inhibit their ability to shift income. This leads to our final hypothesis: H5: The magnitude of income shifting is greater for state owned firms beginning in However, complicating the above SOE predictions is recognition of the argument that even in SOE firms, there is a separation between the shareholders and management which could lead to an agency problem. Managers likely prefer to lower their tax burdens so as to increase the after-tax cash flows under their control. 10 Whether this incentive is greater in SOE or non-soe firms is an empirical matter. Another alternative story relates to managerial career concerns. Kato and Long (2006) document that managers of Chinese SOE firms are often political-based appointees. Presumably these appointees share government political and social objectives and these objectives require resources including tax revenues. To enhance their political reputations and careers these managers might shift less income resulting in higher tax payments to their government superiors. Alternatively, these firms might actually shift more income lowering the 10 Cui (2010) discusses why the state might impose income taxation on 100% state owned firms as opposed to forced dividend distributions. He argues that agency problems exist even in 100% SOEs with managers likely wishing to lower their firms tax burdens so as to increase the after-tax cash flows under their control. He provides anecdotal evidence of 100% SOE firm managers exhibiting tax averse behavior such as lobbying for and obtaining special tax breaks. His arguments also apply to our sample of listed firms where there is mixed or partial ownership by the state and private shareholders. 15

17 tax burden and allowing more resources to be tunneled to their government superiors. Thus while we have signed predictions in both hypothesis 4 and 5 we recognize that it is an empirical matter as to whether SOE firms shift more or less income given the complex setting. 3. Sample research design and results 3.1 Sample selection We require information on the tax rate differential adjustment (TRDA) and these data are only available in the footnotes of B-share firms English-version financial statements. Thus we draw our sample from Chinese B-share firms listed on either the Shanghai or Shenzhen Stock Exchanges during English-version annual financial reports of Chinese B-share listed firms are obtained from various sources, including the Chinese Security Regulatory Committee (CSRC) designated official website ( and company websites, and electronic or hard copies from some listed firms upon request if their English reports were unavailable on their website. The initial sample contains 664 firm-year observations. We exclude firms with incomplete data, and firms in industries that are subject to 11 In China, there are multiple categories of shares, including state-shares, legal-person-shares, employee-shares, A- shares (ordinary domestic individual shares), and foreign individual shares, such as B-shares, H-shares, and N- shares. Only A-shares and B-shares are publicly tradable shares on the domestic Stock Exchanges. A-shares are traded in Renminbi (RMB) by domestic investors. B-shares are traded in either U.S. dollars on the Shanghai Exchange or Hong Kong dollars on the Shenzhen Exchange, and held by foreign entities and foreign individuals, including overseas Chinese residents in Hong Kong, Macau, or Taiwan. In early 2001, B-shares were opened to domestic investors who had access to foreign currencies. Based on the financial information disclosure regulations issued by the Ministry of Finance (MOF) and the Chinese Securities Regulatory Commission (CSRC), B-share firms must prepare financial reports in two languages and use two sets of accounting standards: the Chinese-version under Chinese GAAP and the English-version under International Accounting Standards (IAS). A unique feature of the IAS-based report of a B-share company is the disclosure reconciling the firm s effective tax rate to the applicable tax rate applied to the parent company, which allows us to obtain our measure of income shifting among subsidiaries. We use 1999 as a starting point because most B-share firms did not disclose English reports prior to that time period. In order to make data collection more manageable we restricted our sample period to 6 years covering During our sample period China used its own version of International Accounting Standards (IAS). In 2007 the government issued new accounting standards that substantially converged Chinese GAAP with International Financial Reporting Standards. 16

18 special tax policies or accounting rules (e.g., agriculture and mining). 12 Given that tax losses and past tax losses utilized in the current period make it difficult to interpret our measures of income shifting (TRDA and TRD*) we exclude observations with negative consolidated pre-tax book income and/or tax losses utilized in the current period. We exclude one extreme outlier firm-year observation with return-on-equity of greater than 22. We also eliminate 8 firms that appear to have errors in their reported TRDA from the tax rate reconciliation. 13 Finally, we eliminate 74 firms with no variation in the tax rates applied to the firms in their consolidated group. In other words, the tax rate range for these firms is 0 and they do not have the incentive to shift income for tax purposes. As shown in Table 2, the final number of firm-year observations is reduced to 320. Panels B and C show the sample by year and by industry affiliation. Panel B shows that the number of observations varies slightly across years but no one year dominates the sample reducing concerns about an unusual year affecting our results. Similarly, Panel C shows that while there is some variation across the number of observations in each industry, no single or small group of industries dominate the sample. 3.2 Income shifting and the tax rate differential adjustment (TRDA) As discussed in Section 2, tax rate differentials and separate tax reporting offer the incentive and possibility for undertaking income shifting. The consolidated reporting in accounting standards requires a listed firm to consolidate all controlled subsidiaries income into its financial statements. However, according to the separate tax reporting requirement under Chinese tax law, 12 For example, agricultural enterprises with total assets of 50 million Yuan (eastern), 30 million Yuan (middle region) and 20 million Yuan (western) are corporate income tax free (MOF 1997) and have special tax credit policies for their R&D (MOF 1996). Agriculture and mining also represent industries that are physically tied to their geographic location to conduct their basic operations. 13 Inferences based on our tests are unchanged if we include these 8 observations. 17

19 entities file their tax return on an independent entity basis, meaning that an individual entity s current taxes are the result of its own taxable income multiplied by its applicable tax rate. Therefore, for tax purposes, a consolidated group s financial statement reported current tax expense is the sum of each of the consolidated group member s current income tax expense unlike the case of a domestic U.S. firm which multiplies the consolidated entity s total taxable income by its applicable tax rate. This book and tax treatment can be represented mathematically as: Under consolidated tax reporting, as in the U.S., the consolidated group s current tax expense N = (TI 0 +TI 1 +TI 2 +TI 3+.. )*t 0 = ( TIi)* t (1) 0 Under separate tax reporting approach used in China, the consolidated group s current tax expense i 0 = TI 0 t 0 +TI 1 t 1 +TI 2 t 2 +TI 3 t 3+ = N ( TIi* ti) (2) i 0 where TI denotes taxable income, t denotes the tax rate and subscript i=0 denotes the parent in the entity, and i=1,n denotes the controlled subsidiaries within the consolidated group. If the entity transfers taxable income from high-taxed subsidiaries to low-taxed subsidiaries, (2) will be lowered. As a result, the gap between (2) and (1) represents the results of different requirements in consolidated book and independent tax reporting if the parent tax rate is the highest (discussed below). The amount of the gap between (1) and (2) above is separately stated as a reconciling item in the footnotes of Chinese B-share listed firms and is referred to as the tax rate differential adjustment (TRDA). In other words, TRDA is the group s total pretax income multiplied by the parent company s applicable tax rate (t p ) minus the total tax expense that results when each individual subsidiary s applicable tax rate is applied to the subsidiary s share of pretax income. 18

20 Two examples illustrating the disclosure and calculation are presented in Appendix 1. In the first example, the Income Tax Expense footnote of Chengde Dixian Textile Co. Ltd is reproduced. For Chengde, the 2004 profit before tax was RMB 112,629 and the tax calculated at the parent rate of 33% was 37,234 (this is an application of equation (1)). This amount is reduced by a tax holiday for some subsidiaries (of 28,604) and a reduced tax rate arising from some export sales (of 4,297). Thus the tax rate differential adjustment (TRDA) is RMB 32,901 due to these lower tax rates. We define TRDA as a positive number the larger the number the greater the amount of income shifted (or tax saved) from shifting from the higher-taxed parent company tax rate to lower-taxed subsidiaries. 14 The second example in Appendix 1 presents the tax footnote for Shanghai JinJiang International Hotels Development Company. Unlike our first example, the parent company tax rate for this firm is equal to the consolidated group s lowest tax rate. This case is true for 108 of our 320 firm-year observations where the parent company s tax rate is the lowest in the group. As a result, these firms either have no TRDA adjustment or a negative TRDA adjustment (94 observations). 15 In the case of firms with a negative TRDA adjustment, this configuration of tax rates results in a higher tax than would be the case if all pretax income was subject to the parent company s lower applicable rate. Shifting income to the lower tax parent company is still income 14 As described above, we define TRDA as the difference between equation (2) and (1) resulting from different tax rates being applied to the separately filing members of the consolidated group. Consequently, TRDA only includes the effect of different tax rates as indicated in the effective tax rate reconciliation in the footnotes. TRDA excludes any adjustments arising from different expense and revenue recognition for tax versus financial reporting purposes, such as non-deductable expense, non-taxable income, unrecognized tax losses and their utilization, and other permanent differences. 15 Note it is also possible for firms where the parent company s tax rate is between the consolidated group s lowest and highest tax rate (t h t p t l ) to have a TRDA of 0. This outcome occurs either because all the income is reported in the parent company or the allocation of income between subsidiaries does not result in a group tax rate that differs from the parent company tax rate. 19

21 shifting but leads to difficulty in interpreting a negative TRDA. TRDA in this case actually measures the increase in the consolidated entities taxes due to having some of the income left/located in the higher taxed subsidiary(ies). For this reason, we use three alternative approaches to dealing with these firms. First we retain observations with a negative TRDA, but reset TRDA to zero for these observations. Second we exclude these observations. Third, we retain these observations but calculate an alternative measure of income shifting. We calculate a modified measure (TRD*) that uses the consolidated groups highest applicable tax rate to construct a benchmark for purposes of quantifying the amount of income shifting. Specifically, TRD* is calculated as follows: TRD* = PTBI * (t h t p ) + TRDA; (3) TRD* = PTBI * t h PTBI * t p + (PTBI * t p - current tax expense); TRD* = PTBI * t h current tax expense where PTBI is pretax book income; t h is the highest applicable tax rate of any entity in the consolidated group; t p is the statutory tax rate applicable to the parent company; and current tax expense is the total of the current tax expense separately calculated for each subsidiary (basically equation 2). If the parent faces the highest tax rate then t h = t p and equation (3) collapses to the reported TRDA the case for 95 firm-year observations. Below are four examples illustrating how both TRDA and TRD* are calculated for firms with different tax rates and income levels between the parent company and a subsidiary. In each case the consolidated income is the same at $11,000 (the sum of $10,000 and $1,000) and the two tax rates are the same (at 33% and 15%) but the income allocated to the parent and subsidiary and tax rate faced by each is systematically varied across the 4 cases. 20

22 Case A Parent Sub Consolidated PTBI 1,000 10,000 11,000 TRD* Statutory tax rate Actual tax (eqn 2) 330 1,500 1,830 Tax if all income was taxed at t p (eqn 1) 3,630 TRDA 1,800 1,800 In Case A above, the parent company s tax rate (t p = 33%) is greater than the tax rate of the subsidiary (t s = 15%). The majority of the firm s income ($10,000 of $11,000) is located in the low tax subsidiary. TRDA then reflects this benefit of having $10,000 taxed at 15% instead of 33% (i.e. $10,000 18%). Because the parent company in Case A faces the highest tax rate, TRDA is equal to TRD*. Case B Parent Sub Consolidated PTBI 10,000 1,000 11,000 TRD* Statutory tax rate Actual tax 3, ,450 Tax if all income was taxed at t p 3,630 TRDA Case B has the same parent and subsidiary tax rates as Case A, but the majority of the income ($10,000) is now in the parent company. TRDA is $180 reflecting the $1,000 of income taxed in the subsidiary at 15% instead of 33%. Note that TRDA is substantially lower in Case B than in Case A reflecting the fact that most of the income is taxed at the parent s higher tax rate. 21

23 Alternatively stated, TRDA is low in Case B because the company could have shifted substantially more income into the lower tax subsidiary. Case C Parent Sub Consolidated PTBI 1,000 10,000 11,000 TRD* Statutory tax rate Actual tax 150 3,300 3,450 Tax if all income was taxed at t p 1,650 TRDA -1, In Case C the majority of the income is allocated to the subsidiary ($10,000), but the parent company now has the lower tax rate (15%) relative to the subsidiary s rate (33%). In this case the firm can save taxes by shifting income into the parent company. TRDA is not a good measure of income shifting in this case because the -$1,800 actually estimates the amount of higher taxes due to leaving income in the higher taxed subsidiary. However, TRD* of $180 does reflect the savings of having $1,000 taxed at 15% instead of 33%. (TRD* is calculated using equation (3) as ($11, $3,450 = $3,630 $3,450 = $180)). Importantly, we view the level of income shifting in Cases B and C to be equivalent and this equivalence is reflected in TRD* of $180 in both cases. Case D Parent Sub Consolidated PTBI 10,000 1,000 11,000 TRD* Statutory tax rate Actual tax 1, ,830 Tax if all income was taxed at t p 1,650 TRDA ,800 22

24 In Case D the parent company continues to have the lower tax rate relative to the subsidiary, but now the majority of the income ($10,000) is allocated to the parent. Because the parent company s tax rate is below that of the subsidiary, TRDA continues to be a poor measure of income shifting (estimating the foregone tax savings of having some income left in the higher taxed subsidiary). However, TRD* reflects the significant tax savings from having the majority of the firm s income allocated to the low tax parent company ($11, $1,830 = $3,630 - $1,830 = $1,800 or $10,000 18%). Note we view the level of income shifting to be equivalent in Cases A and D and this equivalence is reflected in TRD* which exhibits the same values in both scenarios. 16 These examples illustrate that TRDA is not a measure of abnormal income shifting (because we cannot observe pre-shifted income) and the existence of positive TRDA thus does not necessarily imply managers have altered transfer prices to shift income. However, increased manipulation of transfer prices for the purpose of shifting income into low tax jurisdictions will result in greater TRDA and we expect firms engaged in income shifting to exhibit higher levels of TRDA in the cross-section of firms. In untabulated analyses we compare descriptive statistics across four partitions of the sample: the 95 observations when the parent tax rate (t p ) is the highest rate in the consolidated entity (and TRDA = TRD*), the 108 cases when t p is the lowest rate (and TRDA < 0), and the 16 While the use of t h in the calculation of TRD* might be questioned on the grounds that its use leads to high estimates of the income shifted, the use of t h gives more meaningful estimates than use of an average tax rate. First if an average tax rate is used, TRD* will not equal TRDA in the 95 cases where t p = t h and TRDA is a meaningful measure. Second, we believe that the amount of income shifted and tax savings in Case B and C are the same and TRD* based on t h provides the same estimate of $180 compared to an estimate of a negative TRD* of $810 ($11,000x.24 - $3,450) using the average tax rate of 24% [(33%+15%)/2)] in Case C. Third, we believe the amount of income shifted and tax saved is the same in Cases A and D and TRD* based on t h provides the same estimate of $1800 compared to an estimate of $810 ($11,000x.24 - $1,830) using the average tax rate of 24% in Case D. Thus we use t h in our estimate of TRD*. 23