Tax Avoidance, Corporate Governance. and Firm Value

Transcription

1 Tax Avoidance, Corporate Governance and Firm Value Jingjing Chen 1 This paper examines whether corporate governance regulates the influence of tax avoidance on firm value and how this influence affects the valuation of shareholders. Taking the data of FTSE 350 firms on London Stock Exchange from 2008 to 2015 as the sample, the empirical results show evidence that the effect of tax avoidance is positive on firm value for firms with strong corporate governance and insignificant for firms with weak corporate governance. Among the corporate governance techniques, compensation incentives and board structure appear to have impacts on corporate tax avoidance and other proxies of corporate governance show no statistical significance on tax avoidance measures. As shown in the findings, the relationship between corporate tax avoidance and firm value is found to be conditional to corporate governance at disaggregate level and the overall effect of corporate governance on the relationship is insignificant. Keywords: Tax Avoidance; Corporate Governance; Firm Value; Agency theory; 1 Jingjing Chen (jingjing.chen@wsu.edu) is a PhD student of Finance and Management Science Department, Carson College of Business, Washington State University in Pullman, WA.

2 1. Introduction This study aims to examine the relationship between tax avoidance, corporate governance and firm value in a sample of UK companies. The research puts forward two fundamental questions, how shareholders consider the value of corporate tax avoidance activities, and to what extent corporate governance affects the shareholder s valuation on tax avoidance. From a traditional point of view, in other words, in the absent of agency costs, corporate tax avoidance activities impose both costs and benefits. With a nature of minimizing the tax liability of a company, tax avoidance activities arguably transfer the state wealth to shareholders value. In terms of costs of tax avoidance activities, there are direct costs regarding resources used to engage in such activities, and potential costs of market scrutiny and reputation damages (Desai and Dharmapala 2009). The trade-off between potential costs and benefits of tax avoidance results in an uncertain effect of corporate tax avoidance on firm value. Furthermore, agency theory considers managerial tax avoidance actions as possible opportunities for managers to pursue self-interests (Desai and Dharmapala 2006). Would the effect of agency costs be significant enough to reveal a loss of firm value? Under the agency framework, corporate governance is relevant to mitigate such effect of managerial diversion. Desai and Dharmapala (2009) further propose that the benefits of tax avoidance activities in saving tax charges are possibly offset by the potential managerial rent extraction for firms with weak corporate governance. Therefore, such benefits and the net effect of tax avoidance are likely to be greater for firms with stronger corporate governance. Tax management generally refers to the activities that reduce firm s tax liability. This broad definition presents a continuum of which one end is perfectly legal activities as strategic tax planning and the other extreme is illegal tax evasion (Dyreng et al. 2008; Hanlon and Heitzman2010). To avoid distinguishing the legality of tax-reduction activities, tax avoidance activities are those strategies between the two extremes of tax management (Hanlon and Heitzman 2010). Definition of tax avoidance from tax authority is relatively narrower, for instance, HMRC (2015) refers to tax avoidance as gaining tax advantages in an unintended way of the law. The interest of tax authority is more likely about estimating and tackling the tax gap rather than maximizing shareholder wealth or reconciling principle-agent problems. The question, which part of the continuum does tax avoidance activities fall in, depends on the aggressiveness of tax management activities. The interest of prior research is intentional tax reduction strategies at the aggressive extreme of the continuum. This research also focuses on the aggressive end of tax management and defines tax avoidance following Hanlon and Heitzman (2010) and Slemrod (2004) as a subset of tax management activities which is to pay lower taxes by engaging in a wide range of transactions without an actual response by the company. The practices of corporate tax-related behaviours closely link to changes in tax regime and the attitude of managers, shareholders and the general public with regards to tax responsibility. According to the survey of KPMG (2016), 75% of companies respond that a country s tax system is the main reason that drives the choice of where to locate the corporation. Compared to international peers, the UK has recently been considered as a considerably attractive destination of business activities due to its competitiveness of tax system (KPMG 2016). The competitiveness arises from the stability of the regime,

3 forewarning of big changes and simplicity resulting from previous tax reform. The UK corporate tax rate has been reduced steadily since 1982 with a peak of 52% to 2016 with 20% as the lowest rate among G8 2 (Brooks et al. 2016). The environment in the UK has recently shown a changing attitude towards corporate tax avoidance activities. Some practices of tax avoidance become unacceptable by the general public(brooks et al. 2016). After a few years of calling for increased tax transparency and responsibility, companies are responding along with improved transparent reports of tax affairs (KPMG 2016). In terms of corporate tax-related activities, the study of UK firms reflects the respond to a stable and well-developed tax regime. The Recent publication of corporate tax gap 3 from HM revenue and Customs (HMRC) is estimated to be 3.0 billion for , which compromises seven percent of the estimated corporate tax liability. According to the report, the tax gap of Large Business Service Corporation (LBS), which is 1.0 billion for , 1.1 billion for and , and 1.8 billion for , could be attributed to avoidance risks with 80%, 82% and 83% respectively. Although the tax gap of corporate income tax has shown a reduction over the period of to , the interest and behaviour of reducing tax payment remain active. Prior literature has long-standing interests on the tax effect of real corporate financial decisions, such as tax effect on capital structure, dividend policy, compensation policy, risk management, and organizational forms (Graham 2003). Instead of taking taxes as one of the determinants of those decisions, could tax itself possibly be a decision of a company? The answer to the question attracts attentions to corporate tax avoidance activities. How tax avoidance decisions, which relates to reducing corporate tax liability, affect firm value and cost of debts 4, and could corporate governance mitigate or encourage tax avoidance? US research presents empirical results more in line with the notion of Desai and Dharmapala (2006) agency framework. Cloyd et al. (2003), Hanlon and Slemrod (2009) and Kim et al. (2011) find strong evidence suggesting that the market reaction to tax avoidance activities is negative. Kim et al. (2011) also document that wellgovernance firms tend to have greater net effect of tax avoidance resulting in an increased firm value, consistent with the findings of Huseynov and Klamm (2012) using Cash ETR. However, the results of empirical research seem to be sensitive to tax avoidance measures and corporate governance quality. Desai and Dharmapala (2009) fail to find a significant relationship between book-tax difference (BTD) and firm value at an aggregate level which implies that the average effect of tax avoidance on firm value could also be offset. Recent UK study by Brooks et al. (2016) shows evidence that no negative or significant relationship has been found between corporate tax payment and financial performance. The study of Brooks et al. (2016) investigates the price reaction to tax rate changes by 2 Noted that Russia, which is currently suspended as a G8 member, also has a 20% corporate tax rate. 3 The tax gap is defined in the HMRC (2015) report as the difference between theoretical tax liability and the amount of tax which is actually collected. 4 Despite firm value, another influences of tax avoidance is on the cost of debts. Beck et al. (2014) argue that corporate tax avoidance, particularly illegal tax evasion activities, have been accused of being an important cause of sovereign debt crisis since they cause fiscal instability. The examination of tax avoidance could contribute to resolving the asymmetric information problem between shareholders and managers in the process of financial crisis (Beck et al. 2014).

4 using long-term ETRs during 1988 to In contrast, Abdul Wahab and Holland (2012) find UK evidence consistent with the notion of agency framework on tax avoidance that tax planning is negatively associated with the market value of equity, but no evidence is found that corporate governance plays a role in mitigating the relationship. They show that permanent book and tax difference is the component that shapes the negative relationship. Other studies further investigate which of the corporate governance techniques influence corporate tax avoidance behaviour. Dyreng et al. (2010) show that individual executives have a significant impact on the aggressiveness of tax avoidance level. Lanis and Richardson (2011) find that the proportion of independent directors on the board lowers the possibility of engaging in aggressive tax avoidance activities. Kubick and Lockhart (2016) show strong evidence that incentives of the external labour market, which motivates directors to be competitive for their employment position, is important in determining the corporate aggressive tax strategies. By using various proxies to examine the role of corporate governance mechanisms in tax avoidance activities, Minnick and Noga (2010) find that compared to incentive compensation, especially pay-performance-sensitivity, other corporate governance measures have less impact on tax avoidance. They also document that tax avoidance with better corporate governance results in higher returns for shareholders. Instead of using various proxies to measure the corporate tax avoidance levels, Armstrong et al. (2015) use quantile regressions associated with the extreme levels of tax avoidance distribution to examine the potential shifts in the relation of tax avoidance in condition to different corporate governance techniques. They find that board independence and financial sophistication could mitigate the agency problem but high equity incentives increase the risk of moral hazard by managers, consistent with the findings of Rego and Wilson (2012). This paper contributes in the following. First, it provides UK evidence with regard to shareholders valuation on tax avoidance of companies under UK tax regime. Second, this research advances the knowledge of the effect of corporate governance on the relationship between tax avoidance behaviours and firm value under agency framework. Third, this paper employs various proxies of tax avoidance and corporate governance and jointly examines the validity and reliability of those measures. Fourth, the discussion of the interplay between corporate governance and tax avoidance may shed light on the fundamental question whether better corporate governance techniques result in increasing firm value (Minnick and Noga 2010). Fifth, the analysis is based on a period of eight years and reflects the changing attitudes of tax avoidance over time. In summary, the regression results show evidence that there is no significant relationship between tax avoidance and firm value. The relationship between corporate tax avoidance and firm value is found to be conditional to the level of corporate governance at disaggregate level. The effect of tax avoidance is positive on firm value for firms with strong corporate governance and insignificant for firms with weak corporate governance. The results are consistent with the hypothesis that corporate governance moderates the principle-agent problem. Among those corporate governance techniques that have been examined, compensation incentives and board structure appear to have impacts on corporate tax avoidance and other proxies of corporate governance show no statistical significance on tax avoidance measures.

5 The remainder of the paper is structured as follows. Section 2 explores the underlying theories and empirical evidence as well as develops the hypotheses for the relation between tax avoidance, corporate governance and firm value. Section 3 discusses the models used to test the hypotheses and the choices and measurement of key variables. Section 4 describes the process of data collection and summary statistics of the sample. Section 5 analyses the empirical results. Section 6 concludes and acknowledges the limitations of the research. 2. Literature review and hypotheses development There is a young and growing literature studying the effect of corporate governance on tax avoidance and firm value. Hanlon and Heitzman (2010) argue that it would be hard to determine the legality in advance of the fact of tax avoidance activities. The legal ambiguity of tax avoidance strategies provides a ground for agency framework to perceive tax avoidance as a risky investment which contains both positive potentials and negative damages on firms. The following section gives a general review of the recent stream of the related literature and discusses how corporate governance mechanism influences the relationship between tax avoidance and firm value. Section 2.1 explains the fundamental framework of corporate governance, tax avoidance, and firm value. Section 2.2 reviews the previous literature on tax avoidance and firm value. Section 2.3 discusses empirical evidence regarding different corporate governance characteristics relating to tax avoidance activities. Section 2.4 develops the control variables included in regression analysis. Section 2.5 emphasizes on the measurement problem engendered in empirical studies. Section 2.6 develops the hypotheses followed by section 2.7 the conclusion. 2.1 Agency theory: a theoretical framework Slemrod (2004) and Crocker and Slemrod (2005) construct a foundation of the principle-agent framework in explaining the variation in corporate tax avoidance. The separation of ownership and control indicates that shareholders cannot directly make the tax decision and managers may have private information about the possible reduction strategies in income taxes. Two thoughts under agency theory link corporate governance to tax avoidance. One traditional notion is that higher level of incentive compensation could encourage managers to behave more aggressively in tax avoidance planning in order to increase the wealth of shareholders (Armstrong et al. 2015). In contrast, another situation when engaging in tax avoidance activities allows managers for rent extraction could alter this positive relationship (Desai and Dharmapala 2006; Desai et al. 2007). Corporate governance in turns plays an important role in resolving the principal and agent conflicts and thus reduces agency costs (McKnight and Weir 2009). Existing literature does not make a consensus on whether there is an optimal value-maximizing corporate governance structure (Coles et al. 2008; McKnight and Weir 2009). The research question regarding the effect of corporate governance on tax avoidance activities provides insight on understanding the cross-sectional variation of tax avoidance activities and the relationship between corporate governance, firm value and tax avoidance.

6 2.1.1 Traditional view Traditional view considers tax avoidance as value-added activities which managers are motivated to take for maximizing the value of organization (Desai et al. 2007). Since tax avoidance decision made by managers affects firm value in a positive way, early studies focus on the efficiency of the decision which is to assess the benefits and the cost of being scrutinized (Phillips 2003). The implication, therefore, is that the contractual relationship may or may not lead to more aggressive tax planning due to market scrutiny. However, Hanlon and Heitzman (2010) pinpoint that earlier research on companies engaging in tax avoidance activities makes no assumption of agency costs when considering managers making corporate decisions on the aggressiveness of tax management. Under the traditional thought, shareholders could simply address the agency problem by motivating managers to increase after-tax performance (Kim et al. 2011), e.g. to align compensation with after-tax measurement (Phillips 2003). The Traditional view suffered from the ignorance of agency costs. Tax avoidance activities may allow for managerial rent extraction which could offset the reductions effect of tax expense on maximizing shareholders wealth and alter the proposed positive relationship (Desai and Dharmapala 2006; Desai et al. 2007). Therefore, shareholders may no longer value corporate tax avoidance activities (Desai and Dharmapala 2009). Desai and Dharmapala (2006) propose that managers may extract rents via tax-related activities. For instance, costs of engaging in tax sheltering activities could more or less attribute to the degree of tax avoidance which depends on the own perspective of the manager. Tax sheltering and managerial diversion are decisions made at the same time, and more aggressiveness in tax avoidance could imply more managerial extraction (Desai and Dharmapala 2006) Tax avoidance and managerial rent extraction The model of Desai and Dharmapala (2006) takes account of the problem of managerial diversion or opportunism by incorporating agency costs into the model of tax avoidance and firm value. Corporate governance under the agency model mitigates the negative relationship between tax avoidance and firm value. Good quality of corporate governance aligns the interests of managers and shareholders by enforcing better governance techniques in preventing managerial diversion. Desai et al. (2007) argue that good practices of corporate governance limit the extent to private benefits extracted by managers and the extent to the impact of agency costs on firm value. Desai and Dharmapala (2006) maintain that the effect of tax avoidance is conditional to corporate governance. They predict that firms with well corporate governance are more effective in preventing managerial diversion and thus the corporate governance incentives could lead to higher levels of tax avoidance but positive influences on firm value. However, firms with poor corporate governance are more interested in reducing tax sheltering for lowering the opportunity of rent extraction (Desai et al. 2007). Following the framework of Desai and Dharmapala (2006), empirical research shows evidence of the interdependence of managerial opportunism and tax avoidance. Desai and Dharmapala (2006) find that stronger equity incentives result in lower level of tax avoidance for weak governance companies. Chen et al. (2010) show that non-family firms are more tax avoidance than family firms. They argue that since other shareholders may commonly presume that family has more opportunities to mask

7 benefits by various activities including tax avoidance, family owners respond to forgo the benefit to avoid discounting share value by other shareholders who concern family may extract rents Limitation on the agency model of tax avoidance Some research questions the underlying assumption of the agency model that tax avoidance facilitate managers to seek their own interests. The decision of managers for firms engaging in tax avoidance depends on factors such as the relevance of information (Gallemore and Labro 2015), organizational forms (Robinson et al. 2010), business strategy (Higgins et al. 2015). Evidence from other studies also challenges that corporate governance is not necessarily the predominant reason for explaining the negative relationship between tax avoidance and firm value. For example, managers may be concerned about market scrutiny, or shareholders themselves may not consider tax avoidance activities as a value maximizing activities regarding the risk of rigorous scrutiny and severe penalty (Graham and Tucker 2006; Brooks et al. 2016; Bebchuk and Fried 2003; Hanlon et al. 2015). Agency theory is useful for understanding tax avoidance activities related to the compensation, especially in the case of tax avoidance activities enabling the managerial diversion. However, when this assumption (that tax avoidance provides a channel for managers to seek their own interest) is not sufficiently important, agency framework could not provide a whole explanation of the direct effect between corporate governance and tax avoidance (Lanis and Richardson 2011). Also, when managers have a limited discretion grant to make tax decision, corporate governance mechanism could no longer explain the effect of tax avoidance on firm value. The decision of managers could be significantly influenced by other variables, such as concerns of reputation cost (Graham Tucker 2006), information quality and availability (Gallemore and Labro 2015), controllability (Robinson et al. 2010), tax knowledge (Cook et al. 2008; McGuire et al. 2012), corporate social responsibility (Huseynov and Klamm 2012a). 2.2 Tax avoidance and firm value Empirical evidence shows mixed results on the relationship between tax avoidance and firm value. The inference contains a debate on whether the share price is negatively, positively or not associated with bad news with regard to corporate tax liability. In the world without agency costs, shareholders with different levels of risk aversion could value tax avoidance on opposite directions and the overall effect of tax avoidance news might therefore be offset. From the traditional point, aggressive tax planning activities save costs and increase firm value (Kim et al. 2011). However, the traditional view is unlikely to fully explain mixed responses. As previously discussed tax avoidance may facilitate managerial diversion and imposes agency costs (Desai and Dharmapala 2006). Cloyd et al. (2003) investigate the reactions of stock price to the news announcement that firms will expatriate to tax haven countries and conclude that the market does not believe that the benefits of fewer tax expenses outweigh or offset the costs. In terms of news about the involvement in tax sheltering activities, Hanlon and Slemrod (2009) adopt a method of event study and examine the response of share prices by employing data of 108 US firms during the year 1990 to They find that share prices drop

8 0.5% as a reaction to tax shelters participation disclosure but the price decreases are muted for well-governance firms. Kim et al. (2011) extend the explanatory power of the agency model by employing a sample of US listed companies. They documented a long run effect of more aggressive tax avoidance on firm value. All measures they used for tax avoidance, cash ETR, large book-tax differences (BTD) and tax sheltering incentive are found having a significant relationship with stock price drops. They also find evidence that higher tax avoidance could predict crashes in future stock price and external governance mechanisms diminish the risk of dropping in firm value. In contrast, Desai and Dharmapala (2009) show evidence that there is no significant relationship between book-tax difference (BTD) 5, a proxy for measuring aggressive tax avoidance activities (discussed more in detail in section 3.3.2) and firm value, measured by Tobin s q in their study. A significant effect is found in a disaggregate level consistent with the agency framework. They show that firms with strong corporate governance appear to have a significantly positive association between tax avoidance and firm value, and firms with weak corporate governance shown a negative relationship. Recent UK study by Brooks et al. (2016) uses data of FTSE All share companies and they show no relationship between corporate tax payment and financial performance. Brooks et al. (2016) find that despite short term falls for some small firms, there is no long-term drop in share price relating to current ETRs, long term ETRs, BTDs, which contradicts to the findings of Cloyd et al. (2003). They argue that tax aggressiveness of managers or any decision on changes in tax rates represents an evaluated and rational decision process accounting for the costs and benefits of corporate tax avoidance. They conclude that by considering responsible investors who keep an eye on corporate tax avoidance, the aggressiveness of senior managers in tax avoidance will not likely to affect stock prices. However, unlike the study of Cloyd et al. (2003) which tests the direct relationship between stock price and corporate expatriation news in a two-year window, the study of Brooks et al. (2016) investigates the price reaction to tax rate changes by using long-term ETRs. The information included in ETRs and BTDs is subject to measurement error which could be caused by the limitation of accounting-based measures. 2.3 Corporate governance and tax avoidance As previously discussed, tax avoidance activities may be conditional to the effectiveness of corporate governance. Prior literature suggests that not only the overall quality of corporate governance structure but also the individual corporate governance mechanism relates to tax avoidance activities and have an effect on firm value. Minnick and Noga (2010) make a comprehensive investigation of the role of corporate governance plays in tax aggressive management and estimate whether tax planning with better corporate governance increases the shareholder wealth in the long run. They find the important effect of corporate governance, of which the most important driver of tax avoidance decisions is incentive compensation, especially pay-for-performance sensitivity. However, their study has yet to explore the world outside the US. In order to investigate the UK practices in this research, the following sections adopt the classification by McKnight and Weir (2009) of corporate governance characteristics: board structure, ownership structure, incentives compensation. 5 Noted that book-tax difference could also be caused by earning management, Desai and Dharmapala (2009) used total accruals as control variable to separate the effect of tax avoidance activities.

9 2.3.1 Board structure and tax avoidance A substantial literature has investigated the link between performance and board structure, referring to the board size and the components. However, mixed results are shown towards the optimal board structure of a corporation. Early literature suggests that smaller and more independence board tends to have higher firm value(jensen 1993; Eisenberg et al. 1998; Yermack 1996). However, recent studies argue that there is no optimal board structure. Boone et al. (2007) and Coles et al. (2008) show evidence that board sizes and outside directors vary by firm-level characteristics. In addition, some argue that larger board size tends to compromise more to make a consensus. Cheng (2008) shows empirical evidence that large board is significantly related to less volatility in firm value. Prior studies suggest that more outside directors on board reduce the agency costs and the new of appointment to outside directors leads to an increase in equity value (Rosenstein and Wyatt 1990). In particular of all the research on board structure, the study of Richardson et al. (2014) links the board effect with corporate tax avoidance and the decision to capital structure in terms of debt policy 6. Their study show intervention from the broad on tax avoidance activities and find evidence that firms with a higher fraction of outside directors increase their debt-substitution effect where the debt-substitution effect is defined as debt being a substitute to tax aggressive activities. The implication would be consistent with other empirical results that companies with a higher level of board independence result in less influence of managers (Coles et al. 2008). Other board characteristics have also been examined in prior research but have shown weak impact on lowering the agency costs. In contrast to the argument that gender diversity could result in higher risk of the firm, Sila et al. (2016) find no evidence that more women on the board affect the firm performance. McKnight and Weir (2009) found little evidence in supportive of an important effect of the duality of CEO/Chair on firm value. However, Bhagat and Bolton (2008) document a significantly positive relationship between the separation of CEO/Chair and corporate operational performance. They also conclude that an overall quality of corporate governance could be measured by board independence or ownership structure Ownership structure and tax avoidance Ownership structure with a composition of institutional ownership, managerial ownership, and family ownership contributes to another part of corporate governance mechanisms. Institutional investors are argued to be powerful shareholders with more active engagement of voting. They are considered as having superior knowledge and resources to monitor managers (McKnight and Weir 2009; Desai and Dharmapala 2009). Thus, a higher level of institutional ownership could better monitor managers in order to reduce agency costs. Lim (2011) extends Desai and Dharmapala (2006) model by examining and documenting a negative relationship between tax avoidance and cost of debt. It is found that the negative relationship is magnified when institutional ownership is higher. Their study adopts institutional ownership as the proxy of 6 Graham and Tucker (2006) document a debt-substitution effect. They find evidence that firms engaging in tax sheltering activities appear to have an 8% less debt ratios compared to matched pre-sheltering firms.

10 corporate governance 7 and suggests the important effect of institutional ownership on monitoring managerial extraction. A similar study by Kim et al. (2011) also shows that institutional ownership as one of the external governance sources has an important effect on mitigating the negative impact of tax avoidance on firm value. Desai and Dharmapala (2009) also adopt institutional ownership as the proxy for governance quality and find evidence consistent with the proposal that benefits from corporate tax avoidance depend on the quality of corporate governance. Overall, prior evidence suggests that higher institutional ownership leads to lower chance of rent extraction. Under the agency model, increased managerial ownership provides more incentives for managers to act on behalf of the interest of the shareholders. Kim and Lu (2011) document that under weak external governance (EG) measured by industry concentration ratio and institutional ownership concentration, CEO ownership has more room to mitigate agency problems via motivation effect. When EG is strong, the relationship between CEO ownership and Tobin s q is shown being insignificant. They also find evidence that the level of R&D investment, which is used for measuring the risk-taking decisions of CEOs, is affected by the quality of external governance. Their findings also provide insights on the notion of tax avoidance research under agency model. In terms of family ownership, family firms appear to manage tax payments less aggressively than that of non-family firms (Chen et al. 2010). However, in this case, family owners avoid tax avoidance strategy for the reason of reputation costs and possible penalties but not interest alignment Incentives compensation and tax avoidance Most literature on executive incentives and CEO tax aggressive preferences focuses mainly on compensation incentives. Empirical evidence shows that managers decision on corporate tax avoidance strategies is most sensitive to managerial compensation compared to board structure and age of the CEO (Minnick and Noga 2010). Specifically, managerial compensations such as equity-based incentives (Rego and Wilson 2012) and Pay-for-performance sensitivity (Minnick and Noga 2010) have been found driving executives to behave in the way that increases shareholders wealth. Tournament incentive which refers to the difference between total compensation of the CEO and those of the near-highest paid CEO in the industry, has been found playing an important role in affecting tax decisions. A Recent study of Kubick and Lockhart (2016) show evidence that the external labour market has been an effective motivation for CEOs to undertake aggressive tax strategies Other corporate governance characteristics and tax avoidance Other corporate governance characteristics may also influence corporate tax avoidance. In terms of characteristics of the CEO, Francis et al. (2016) show that whether the CEOs are Democratic or Republican affect their decisions to engage in tax sheltering activities. Other factors such as overconfidence (Huang et al. 2016) and national culture (DeBacker et al. 2012) has shown to affect the risk adverse preference of the managers. From the view of stakeholders, Chyz et al. (2013) examine the influence of labour unions on corporate aggressive tax management and they document a negative 7 Noted that Desai and Dharmapala (2009) also measure institutional ownership which as the proxy of good corporate governance has been widely used in empirical literature under Desai and Dharmapala (2006) model and find consistence results.

11 association between tax avoidance and union power, measured by unionization rate and bargaining power. Their findings further suggest that stakeholders are likely to have an important effect on corporate tax aggressive strategies. Overall, prior literature provides extensive evidence on the effect of corporate governance and develops various measures of governance techniques. Nevertheless, it is important to note that corporate governance varies significantly by countries. An empirical study on UK practices by McKnight and Weir (2009) shows no significant effect of board structure on the agency costs. Abdul Wahab and Holland (2012) also find UK evidence contradicts to US results that corporate governance shown no mitigating effect on agency conflicts and tax avoidance activities. They conduct a factor analysis to determine the governance mechanism which could reflect the quality of corporate governance. The following sections of this paper also investigate the effectiveness of different corporate governance techniques with the purpose of testing the relations to tax avoidance activities. 2.4 Control variables: firm characteristics The prior literature identifies several firm characteristics for controlling the crosssectional variation on corporate tax avoidance activities. Earlier research has extensive focuses on the rationale of firm size and different effective tax rates. Mixed results have been found due to different measures of firm sizes 8, sample period and the specific model (Rego and Wilson 2012). Zimmerman (1983) show evidence that firm size is positively related to effective tax rates (ETR) 9. In terms of UK setting, Holland (1998) documents a size effect is related to average ETR during and finds a negative relation from the late 1970s to early 1980s. In contrast, Stickney and McGee (1982) and Gupta and Newberry (1997) find size factor has no significant relationship with ETRs but other firm characteristics such as capital intensity, leverage, capital structure and asset mix are shown significantly associated with ETRs. Shiing-Wu (1991) present the result that net operating loss (NOL) has statistically importance on the variation of ETRs. Empirical studies also document that the effect of determinants is not widely spread across industries. Samples from oil and gas companies were found to have highest ETRs but wholesale and retail industries resulted in lowest ETRs (Zimmerman 1983). Klassen and Laplante (2012) identify another firm characteristic that the high level of foreign reinvestment signals companies with subsidiaries located in lower foreign tax rates places having more incentives as well as opportunities to shift income. Companies with high R&D percentage are shown have more chance to shift profits overseas (Klassen and Laplante 2012). In term of leverage, Graham and Tucker (2006) show that compared to a control sample, tax shelter companies have lower leverage ratios. Brooks et al. (2016) argue that capital intensive companies could have more opportunities to avoid taxes through depreciation of assets. 8 Moore (2012) summarise that firm size can be calculated by numerous measure, e.g. net income, gross receipts, business value and number of employees. The result of this study also shows patterns in long term ETR changes. 9 The notion of earlier studies investigating variation of ETR is that under the political risk hypothesis, large firms are subject to greater possibility of market scrutiny and ETRs are expected to partly measure the political risks(zimmerman 1983).

12 2.5 Endogeneity problem and sample bias Endogeneity problem of corporate governance research could cause serious measurement problems. Endogeneity occurs in a multiple regression when the independent variable is associated with the error term. Endogeneity problem causes the OLS estimators to be biased and inconsistent which will result in either overestimate or underestimate the parameters. There are few cases which lead to endogeneity. First, the regression model may exclude an important variable and the omitted variables could affect both the independent and dependent variables. For example, there might be an uncontrolled or unrecognized variable that affects both the CEO characteristic and corporate tax sheltering activities (Francis et al. 2016). Second, the measurement error of independent variable could also cause endogeneity problem. Separating tax avoidance effect from earnings management or aggressive reporting with regard to accounting measures of tax avoidance could be complicated. Desai and Dharmapala (2009) state that the implication of the regression results is limited because of the possibility of having measurement errors of tax avoidance proxies. Third, endogeneity arises from simultaneity 10 when independent variables are jointly determined with the dependent variable. For instance, tax sheltering activities could increase company s market capitalization as well as motivate executives to exercise their share option(desai and Dharmapala 2006). Neglecting the above kinds of measurement problem could lead to bias results or spurious relationship (Wintoki et al. 2012). Prior empirical research makes a lot of effort in controlling for endogeneity problem by conducting robustness test. Substantial studies include control variables and fixed effect into the model to avoid the effect of omitted variables. Robustness test is also widely tested in additional to the regression analysis. With regard to measurement error in independent variables, the approach of instrumental variables (IV), which is to include an instrument unrelated to the error term into the regression model, can solve the inconsistency problem caused by endogeneity. However, the estimates are still bias under IV method and the efficiency of the method is limited to the quality of the instrument. The robustness could also be tested by including lagged variables of the independent variables (Wintoki et al. 2012). For example, Desai and Dharmapala (2009) check the robustness by incorporating lagged variables of the tax avoidance proxy. Minnick and Noga (2010) include lagged differences of ETR into the model. Moreover, a majority of studies also conduct robustness tests by using alternative measures of the independent variables, i.e. other proxies of tax avoidance and corporate governance techniques (Lim 2011; Abdul Wahab and Holland 2012; Desai et al. 2007). A great number of empirical studies of corporate governance and firm value focus on different samples from a single country to the worldwide practices. However, there is a lack of research in investigating the links between corporate governance, tax avoidance, and firm value apart from the view from US sample. A few studies have extended the perspective towards evidence from Russia (Desai et al. 2007; Mironov 2013; DeBacker et al. 2012), Korea (Lim 2011), China (Zhang et al. 2016) and the UK (Brooks et al. 2016; Abdul Wahab and Holland 2012). Little is known about the practices outside the US. Furthermore, prior studies tend to exclude the financial industry from their sample and the results would fail to account for the unknown characteristics of financial institutions regarding corporate governance and tax avoidance. 10 Simultaneity is defined when X is not random but a function of Y (Wintoki et al. 2012).

13 For studies adopting the measurement of tax sheltering, the sample of tax sheltering firms would be limited to those which have been publicly accused of engaging in tax sheltering activities. The result from tax sheltering companies can be biased since it ignores other silent firms. For instance, research using tax sheltering proxies may ignore companies which have hidden tax avoidance activities since the sample only includes firms being exposed of engaging in tax sheltering activities. The result of tax sheltering model would fail to capture the firm characteristics of those silent companies(graham and Tucker 2006; Desai and Dharmapala 2006). Moreover, the current tax shelter sample is based on studies of US firms, and thus the tax sheltering regression results may be biased to non-us companies. Also, including firms which are accused of involving in tax sheltering activities may cause endogeneity problem since the included firms are likely to be those with the least care of social response such as reputation cost (Graham et al. 2014). 2.6 Hypotheses development To examine the relationship between corporate governance, tax avoidance, and firm value, the first step would be to test the hypotheses under agency framework. The agency framework proposes that tax avoidance activities provide greater opportunities for managers to extract rents from the companies. Therefore, tax avoidance would imply a decrease in firm value. A negative relationship between tax avoidance and firm value indicate that the effect of managerial diversion outweighs the potential benefit of tax avoidance strategies. Prior research found result consistent with the notion of agency framework. Mironov (2013) shows that diversion activities negatively affect the firm value. Kim et al. (2011) find a significantly positive relationship between tax avoidance and firm idiosyncrasy risk in stock price. Thus, it could be expected that companies with a higher level of tax avoidance have lower firm values. Accordingly, this research test the following hypothesis (H1). H1: Tax avoidance is negatively related to firm value. According to Desai and Dharmapala (2006) model, the quality of corporate governance influences the ability of managers to extract rents since well-governance companies are more likely to have stronger control techniques that prevent managerial diversion. According to the level of corporate governance, firms could be categorized into two extremes. Strong corporate governance is expected to mitigate the problem of managerial diversion, and weak governance may worsen the problem. H1 considers the average effect of shareholder s valuation on tax avoidance. H2 is expected to investigate whether the quality of corporate governance makes a difference to the relationship between firm value and tax avoidance activities. H2: All else being equal, the negative relationship of tax avoidance and firm value is mitigated by strong corporate governance. There is extensive evidence that corporate governance techniques improves firm value. In another aspect of agency framework, tax avoidance can be seen as a risky investment available to managers similar to other investments which are influenced by corporate governance. Without the assumption of managerial diversion, several studies examine how corporate governance would improve firm value by testing the direct relationship

14 between corporate governance and tax avoidance. Armstrong et al. (2015) argue that given the unsolved Principle-agent problem, managers may choose a level of tax avoidance which shareholders may disagree. The third hypothesis (H3) examines the direct relationship between corporate governance and tax avoidance by incorporating various corporate governance mechanisms into the model. H3: Increased quality of corporate governance techniques leads to higher tax avoidance, in other words, lower tax expenses. The test of H3 does not necessarily assume that tax avoidance activities create opportunities for rent extraction. But it assumes that various corporate governance techniques mitigate the agency problem by controlling tax avoidance activities (Armstrong et al. 2015). Within the agency model, strong governance better aligns the interest of managers with corporate value-maximization. H3 further investigates the specific corporate governance techniques (board size, board independence, the duality of the CEO, ownership structure, and stock compensation) that tax avoidance activities are sensitive to. Larger boards may have more difficulties in convincing managers to allocate resources to tax avoidance activities. Independence board can be more flexible in diverting resources to tax avoidance which implies that more independence board has greater incentives to pursue a higher level of tax avoidance. In terms of compensation, since the purpose of managerial incentives is to motivate managers acting on behalf of shareholders, higher stock compensation is expected to result in lower tax expenses, in other words, higher tax avoidance is likely to be associated with higher stock incentives. Alternatively, H3 can be stated as above. Noted that the tests of H1, H2 and H3 are all single-sided test. 2.7 Summary of literature review The agency framework lays a foundation for empirical research to investigate the relationship between corporate governance tax avoidance and firm value. The model of Desai and Dharmapala (2006) extends the theoretical ground by assuming managerial opportunism with regards to tax avoidance activities. Tax-related literature also contributes to the knowledge of how corporate governance affect firm value. Research under agency framework advances the knowledge of the effect of corporate governance on the relationship between tax avoidance behaviours and firm value. Second, research on this areas jointly examines the role of corporate governance as well as tax avoidance activities on firm value. Third, the research also sheds light on the other consequences of the interaction between corporate governance and tax avoidance, such as debt policy, the cost of debt, bondholders. Further research of these areas will possibly push the boundaries of existing body of knowledge to corporate governance regarding many other disciplines. 3. Methodology 3.1 Research design 6 FV it = β 0 + β 1 TA it + β n Conrol it + θ t + δ i + ε it (1) n=2 Following the regression model from previous studies by Mironov (2013), Desai and Dharmapala (2009) and Kim and Lu (2011), equation (1) is the regression model for

15 testing H1. FV it is the firm value which has been measured by Tobin s q (most common proxy) and market capitalization (Brooks et al. 2016; Desai and Dharmapala 2009). β 0 is the constant. The coefficient of tax avoidance is β 1 which implies the sensitivity of the firm value to changes in tax avoidance levels across firms. According to H1, β 1 is expected to be negative for BTD based measures, and be positive for ETR measure. TA it is the measure of corporate tax avoidance which is measured by current ETR, BTD and permanent BTD. θ t controls for firm-fixed effect and δ i controls for year-fixed effect. 6 n=2 β n Conrol it are control variables for various firm characteristics including size measured by sales, foreign income, R&D expense, leverage and capital intensity (Zimmerman 1983; Gupta and Newberry 1997; Shiing- Wu 1991; Klassen and Laplante 2012). FV it = β 0 + β 1 TA it + β 2 CG it + β 3 TA it CG it + β n Conrol it + θ t + δ i + ε it (2) Equation (2) is the regression model for testing H2 where CG it is the proxy for corporate governance which is measured by corporate governance score. According to the H2, β 3 is expected to be positive when using BTD-based measures and to be negative when using ETR-based measures, which implies that when corporate governance has higher quality, the effect of tax avoidance on firm value is greater across time and firms(desai and Dharmapala 2009). Equation (3) is the regression model for testing H3 following Minnick and Noga (2010) and Huseynov and Klamm (2012), where β 1 6 measure the effect of each corporate governance techniques on corporate tax avoidance activities. CG it represent three types of corporate governance techniques, board structure, ownership structure and managerial incentives. According to available corporate governance data in Datastream, CG it includes corporate governance mechanisms proxies as board size, board independence, CEO/Chairman duality, ownership structure, CEO stock compensation and total senior executives compensation. 6 8 n=4 TA it = β 0 + β n CG it + β m Conrol it + θ t + δ i + ε it (3) n=1 11 m=7 It is noticed that substantial literature focuses on the most aggressive form of tax avoidance. Several studies using GAAP ETR and Current ETR find no significant relationship between firm value and tax avoidance (Brooks et al. 2016). Equation (2) captures the average effect of corporate governance on the mitigation of agency conflicts using Ordinary least square (OLS) method. Prior research has adopted two methods in examining the effect of corporate governance without narrowing the view on the central location of the sample distribution. Armstrong et al. (2015) use quantile regressions to provide a view of the distribution and to allow for analysis of special quantiles (Valta 2012). The other method is to classify the whole sample into subsamples of good corporate governance and bad corporate governance firms, and then re-run the regression for H1 and H2 on each sub-samples for comparing the coefficients, β 1 and β 3, of each subsample as well as the whole sample (Kim and Lu 2011; Desai and Dharmapala 2009). It is also noticeable that the regression model can incorporate long term measures of tax avoidance, such as long-run ETRs (Dyreng et al. 2008), matching with other variables along the same period of time (Minnick and Noga 2010). The discussion of long term tax avoidance activities is not included in this paper, however, it is a further aspect of investigation.