MSc Accountancy & Control
Track: Control
Tax Avoidance
Does CEO equity incentives lead to more tax avoiding activities? And
what is the role of corporate governance?
Name: Joris ’t Hart
Student number: 10892982
Date: 20-6-2016
Word count: 12.008
Thesis supervisor: Prof. Dr. V.R. O’Connell
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
I
Statement of originality
This document is written by student Joris ‘t Hart, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economic and Business is responsible solely for the supervision of completion of the work, not for the contents.
II
Abstract
This research first examines whether tax avoiding activities are related with the extent of equity incentives of the Chief Executive Officer (CEO). The motivation behind this research is mixed results in prior literature. Second, this research examines whether the quality of corporate governance is related to extent of tax avoiding activities and whether the corporate governance quality moderates the effect between equity incentives and tax avoidance. The motivation behind this is mixed results in prior literature and statements that equity incentives itself is a corporate governance mechanism.
I find no significant relationship between the extent of equity incentives of the CEO and tax avoidance measured with different proxies for tax avoidance. The results also indicates that there is no significant relationship between the quality of internal governance and tax avoiding activities. Furthermore, there is no significant evidence that the quality of internal governance moderates the effect between the extend of equity incentives for the CEO and tax avoidance.
Prior literature finds evidence that CEO’s view the tax departments as profit centres and therefore use their influence to pressure the tax department to maximize profits. However, since the results indicates no significant relationship between CEO equity incentives and tax avoidance it might be interesting to examine whether the equity incentives of the tax director is associated with tax avoidance.
III
Table of content
1. Introduction ...1
2. Literature review and Hypothesis development ...2
2.1 Tax avoidance ...2 2.2 Equity incentives ...4 2.3 Corporate Governance ...5 2.4 Hypothesis development ...6 3. Methodology ...8 3.1 Sample ...8 3.2 Research design ...9 3.3 Measures ... 11 3.3.1 Equity incentives ... 11 3.3.2 Tax avoidance ... 11 3.3.3 Corporate governance ... 14 4. Results ... 14 4.1 Descriptive statics ... 14 4.2 Results ... 18 5. Conclusion ... 24 6. References ... 27
1
1. Introduction
This paper examines whether the extend of tax avoidance is associated with the extent of granted equity incentives in the CEO’s compensation plan. It also examines whether the quality of corporate governance is associated with tax avoidance and whether or not the quality of corporate governance mitigates the expected effect of equity incentives on tax avoiding activities. This question derives from recent news articles in the Dutch media and contradictory results in prior literature on this subject. Large company’s like Starbucks and Microsoft are known for their tax sheltering. Recently the European Commission declared the tax arrangements between the Dutch government and Starbucks illegal and decided that Starbucks needs to pay between 20 and 30 million euro’s to the tax authority (Kleinnijenhuis, 2015). The European Commission now also want to know about other large multi nationals and their tax arrangements, like Microsoft for instance. Tax avoidance is a relative subject, and according to Hanlon and Heitzman (2010) thereare still a lot of interesting questions unanswered or unclear. There are mixed results in prior literature on this subject. Desai and Dharmapala (2006) find evidence that high powered incentivized executives are associated with less tax avoiding firms. Rego et al (2011) find evidence that equity incentives have a positive relationship with tax avoiding activities of firms.
The question whether corporate governance quality affects the tax avoiding activities is derived from statements in prior literature. Larcker and Tayan (2011) define the corporate governance framework as control mechanisms that prevent or dissuade self-interested managers from making decisions that are not in the interest of the investors and stakeholders. According to Hanlon and Heitzman (2010) risk-neutral shareholders expect that the executives of the firm will act in the best interest of the shareholders, and therefore try to maximize profit. Maximizing profit includes efforts to reduce tax liabilities as long as the incremental benefits exceeds the incremental costs. Therefore, it seems interesting to examine whether the quality of corporate governance is associated with tax avoiding activities. The traditional view of tax avoidance is that avoiding taxes will increase the firms wealth, like Hanlon and Heitzman (2010) and Desai and Dharmapala (2006) describe in their papers. The alternative view on tax avoidance is that it reduces transparency and may decreases the firms wealth on the long term. According to Khurana and Moser (2013), long term investors view tax avoiding activities as unbeneficial on the long term. Given the fact that most equity incentives like exercisable options commonly are mid-term orientated, we can conclude that according to the alternative view the interests of the executives concerning the tax avoiding activities are misaligned. According to the traditional view tax avoiding is a shared interest of equity incentivized executives and the shareholders of the firm. Assuming that shareholders want a transparent and long term wealth increase, it is interesting to examine whether corporate governance moderates the expected effect between equity incentives and tax avoidance.
In order to examine whether tax avoidance have a relationship with equity incentives and corporate governance and to examine whether corporate governance moderates the relationship between equity incentives and tax avoidance I collected data to determine the level of tax avoidance that is present in S&P 500 firms from fiscal year 2007-2014. I have chosen for US listed firms because it is mandatory to provide compensation information regarding equity incentives. Furthermore, I have
2 chosen S&P 500 firms as they are perceived to be large firms and therefore their tax avoiding activities are perceived to have a bigger impact on society than smaller firms. Prior literature states that there is no single best measure for tax avoidance (Hanlon & Heitzman, 2010). Different measures for tax avoidance capture different strategies of avoiding taxes. To capture different forms of tax avoidance I used three different proxies to measure the tax avoiding activities, namely: SHELTER, DTAX and CASHETR. SHELTER is used to capture the unexplained book-tax difference. DTAX is used to capture the abnormal book-tax differences. CASHETR is used to capture tax deferral strategies. The extent of equity incentives is measured according to the method Cheng & Warfield (2005) use in their paper. It consists of five elements, namely: option grants in the current period, exercisable options (option in current period excluded), exercisable option, restricted stock grants, and stock ownership. These measurements are divided by the total outstanding shares of the company. The potential benefits caused by an increase in short-term stock prices, is shared by all shareholders. The benefit enjoyed by the CEO is thus the proportion to the ratio of equity incentives elements held by the CEO divided by the total outstanding shares.
The quality of corporate governance is measured with the B-index score taken from the Baber et al (2012) paper. This index measures the internal governance quality with six elements: 1)CEO is not chairman of the board, 2) two/third or more is independent, 3) all audit committee members are independent, 4) all compensation committee members are independent, 5) separate independent nominating committee and 6) the board size is less than the median of distribution for all firms. The internal governance is relevant for this study as it measures the quality of internal independence.
The results of this paper show no significant relation between tax avoiding activities and CEO equity incentives. The results also indicates there is no significant relationship between the quality of internal governance and tax avoiding activities. Furthermore, the results indicates that there is no significant evidence that corporate governance moderates the effect between CEO equity incentives and tax avoidance. Overall, all three hypothesis are rejected.
2. Literature review and Hypothesis development
2.1 Tax avoidance
In order to understand this research as intended, I will first explain what the conceptual definition is of tax avoidance. Hanlon and Heitzman (2010) define tax avoidance as the reduction of explicit taxes. It reflects all activities that reduces the explicit tax liabilities. Other terms used are tax aggressiveness and tax sheltering. In this research the term tax avoidance is used for corporate tax avoidance and thus not include personal or individual tax avoidance. Besides the constantly changing law for taxes which creates opportunities for companies to engage in tax avoidance by exploiting the loopholes, Stiglitz, J. E. (1986) states there are three basic principles for avoiding taxes in income. First the postponement of taxes, which basically means that present value of taxes which are payed in the future are much less worth than paying the same amount today. Second the arbitrage of different tax brackets. This is very effective for a “family” of companies where for example the family member located in a low tax country
3 gets most of the profit allocated. Third the arbitrage across different tax treatments. Stiglitz, J. E. (1986) argues that many tax avoidance devices contains a combination of the three principles.
According to Hanlon and Heitzman (2010) risk-neutral shareholders expect that the executives of the firm will act in the best interest of the shareholders, and therefore try to maximize profit. Maximizing profit includes efforts to reduce tax liabilities as long as the incremental benefits exceeds the incremental costs. If tax liabilities reduces, assuming pre-tax income remains constant, the profit after taxes increases. Owners of firm tend to structure the incentives for the firm in such manner that the tax decisions are efficient, and by this I mean that the tax decisions executives make increase the after tax wealth of the firm (Hanlon and Heitzman, 2010). This approach predicts that incentive structures have effect on tax efficient decisions by executives. However, Khurana and Moser (2013) find results that shareholders with a long term horizon discourage tax avoiding activities as such activities can encourage managerial opportunism and reduce transparency. operations, tax planning and other factors have been examined as determinant of tax avoidance, where tax avoidance itself is measured in various ways.
Hanlon and Heitzman (2010) provide an overview of measures used in prior literature to measure tax avoidance. Tax effective measures like GAAP ETR and CASH ETR capture the average rate of paid taxes per dollar of income and are computed by dividing the an estimate of the tax liability by a measure that indicates the profit before taxes such as pre-tax income. CASH ETR is computed by dividing cash taxes paid with pre-tax income and is, according to Hanlon and Heitzman, affected by tax deferral strategies but is not affected by changes in the tax accounting accruals. GAAP ETR, on the other hand, is computed by dividing total tax expenses with pre-tax income and affects accounting earnings. Long run CASH ETR, computed by dividing the sum of cash taxes paid over n years with the sum of pre-tax accounting income over n years, has the advantage that it avoids the year to year volatility in annual effective tax rates. Using this tax effective rates as measures for tax avoidance are beneficial because it avoids tax accruals effects present in the current tax expense. On the other hand, it also reflects transactions that have upward effect on the tax liability, does not capture the beneficial tax benefits caused by lobbying activities, does not directly capture implicit taxes and does not capture conforming tax avoidance as the measure uses the book pre-tax income as the denominator (Hanlon and Heitzman, 2010). Book-tax differences (BTD), which captures the difference between book and taxable income, seems to be a valid measure for tax avoiding activities. However, according to Hanlon and Heitzman (2010) argue that the information in book-tax differences about tax avoidance is harder to document due to the difficulty to obtain valid tax outcomes. In the paper of Desai and Dharmapala (2006) the authors use a measure of abnormal book-tax differences. They regress the BTD on total accruals which control for the possible earnings management and the residual is taken as the proxy to measure the extend of tax avoiding behaviour. Frank et al (2009) and Wilson (2009) use DTAX as their proxy for tax avoidance. DTAX is the estimated residual from their PERMDIFF measure, where PERMDIFF is the difference between the effective and statutory tax rates multiplied by the pre-tax accounting income. This measure, DTAX, is used to capture items that reduce the firms GAAP ETR’s. Because PERMDIFF is basically a function of GAAP ETR, it does not (as mentioned above) capture conforming tax avoidance and tax deferral strategies and therefore can’t be seen as an overall measure for tax avoidance.
4
2.2 Equity incentives
As mentioned in the previous paragraph risk-neutral shareholders expect executives to act in their best interest. To align the interest of the executives and shareholders to mitigate for agency problems, the compensation of these executives can consist of equity awards. According to Jensen and Meckling (1976) ownership is an important and effective way to align executives interest with the those of the shareholders. These equity incentives have an upward potential and therefore, according to the convergence of interest theory, align the interest of the executives with the shareholders, as they are shareholders when they receive this awards and benefit from increasing firm wealth (Pergola and Joseph, 2011). Jensen and Meckling (1976) even refer to it as one of the most effective mechanisms to align the CEO’s interest with the interest of the owners. According to this theory the interests of the executives have a linear relationship with equity incentives. The second theory for equity incentives is the entrenchment theory. According to this theory the executives in the middle range of the linear regression, described in the convergence of interest theory, have enough power to overcome the control mechanisms and use their position in ownership and consume perks which reduces the value of the firm. They can act in their own self-interest and have no fear for getting dismissed or other sanctions (Pergola and Joaseph, 2011).
According to Cheng and Warfield (2005) there are five elements of equity incentives: stock ownership, restricted stock grants, current period option grants, unexercisable options and exercisable options. The granted options are often unexcercisable for three or four years and can be excersized when this time period expires(Cheng & Warfield, 2005). When CEO’s receive these equity incentives, they have in any point in time various forms of equity-based holdings (Cheng & Warfield, 2005) and thus is their wealth sensitive to the firms stock prices. According to Cheng and Warfield (2005), executives view these sensitivity in to ways. The incentive alignment theory states that executive equity incentives motivates them to align their interests with the shareholders and thus reduce agency costs. The risk diversification theory states that risk-adverse executives want to reduce their exposure to the firms risk and therefore might want to sell their shares to reduce their exposure to this risk. Cheng and Warfield (2005) also state that executives with higher equity incentives are more likely to sell their shares and their wealth is therefore sensitive to short-term stock prices. This can lead to incentives for the CEO to boost the firms performance on short-term.
According to Wilson (2009) firms that engage more in tax avoiding activities have larger BTD’s, more foreign assets and less leverage. The survey based research from Philips (2003) tests whether executives with compensation based on after tax profits have lower GAAP ETR’s, and finds evidence supporting this hypothesis. This findings are not consistent with the findings of Desai and Dharmapala (2006), who find a negative relationship between equity incentives and tax avoiding activities. Desai et al (2007b) explain this findings by the theory that managers who engage in tax avoiding activities to create managerial diversion, the equity incentives increases in order to further align the interest of the executives with the shareholder, which then decreases the tax avoiding activities of the executives which are used to accomplish the diversion.
5
2.3 Corporate Governance
Corporate governance is a framework of principles and best practices to ensure accountability, fairness and transparency of the board of directors to the shareholders of the company. Larcker and Tayan (2011) define the framework as control mechanisms that prevent or dissuade self-interested managers from making decisions that are not in the interest of the investors and stakeholders. They also state that corporate governance is a nowadays a topic we see and discuss frequently discuss about due to accounting scandals, inside trading, compensation of the executives and frauds. Larcker and Tayan (2011) argue this can be blamed on the quality of corporate governance within these firms as this mechanism should prevent this behaviour of the executives.
Lacker and Tayan (2011) define corporate governance as the collection of control mechanism than an organisation adopts to prevent or dissuade potentially self-interested managers from engaging in activities detrimental to the welfare of shareholders and stakeholders. They also acknowledge that not every executives is self-interested when corporate governance mechanisms is not in place, in fact executives can also act to the interest of the shareholders because of their personal moral value.
In the paper of Desai and Dharmapala (2006) they argue that tax avoidance also facilitates managerial rent extraction because firm become less transparent due to tax avoiding structures. In this view the quality of corporate governance moderates the expected positive relationship between equity incentives and tax avoidance. According to Desai and Dharmapala (2006) and Wilson (2009) the strength of corporate governance will moderate the corporate tax avoiding activities.
Baber et al (2012) separate two kinds of corporate governance mechanism, external corporate governance and internal corporate governance. According to these authors external governance systems facilitates or discourages active stakeholder participation in the governance process. If the costs of the stakeholder participation in order to establish restrictions and provisions is relatively high, the quality of corporate governance is low. The internal governance reflects the interactions among firm management, directors and employee’s and focuses mainly on the independence of the board of directors. Based on Baber et al (2012) I assume that firms with higher internal governance quality prevents executives to act in their self-interests. The authors refer to multiple papers who find a positive relationship between the quality of internal governance and firm performance (Baysinger & Butler, 1985; Hermalin & Weisbach, 1991; Yermack, 1996; Agrawal & Knoeber 1996; Klein, 1998; Bhagat & Black, 1999). Moreover, public policy debates advocates independence of the oversight committees (Baber et al, 2012). SOX even requires the audit committee to be solely and independent.
The Council of Institutional Investors (2003) advise that a substantial part of the board should be independent and that its oversight committees consist of only independent members.
Corporate governance codes are originally called into life to mitigate for the so called agency problems. Agency problem arise when self-interested managers act and take decisions to increase their personally wealth. These decisions can adversely decrease the firms wealth and thereby the shareholders wealth. The costs arriving from this problem are so called agency costs. In order to control for this agency problems, firms can adopt control mechanisms to control for this problems which are called corporate governance (Larcker & Tayan, 2011). According to Larcker and Tayan (2011) each
6 corporate governance system should include a board of directors who oversee the management of the firm. The Organization for Economic Corporation and Development (2004) state that the board of directors should ensure strategic guidance, effective monitoring of the management and accountability to the company and shareholders.
The audit committee is responsible for monitoring the external audit activities and is responsible for appointing the external auditor. The label audit committee could be misleading, since the committee does not audit the firm. The audit committee is called into live to prevent the management to influence the auditor selecting process and the auditor itself (Walker, 2004). Walker (2004) also states that audit committees strengthen the quality of corporate governance and the quality of financial reporting. I find it important to note that audit committees itself do not improve the quality of the audit. Only auditors can improve this quality according to Flint (1980). Results of prior literature find evidence that independent and active audit committees are related to less internal control problems (Krishnan, 2005). Furthermore, Abbot et al (2004) finds evidence that independent and active audit committees are related with significant lower restatements.
The nominating committee is responsible for nominating, evaluating and identify the directors. They are also responsible for the CEO’s succession planning (Lacker & Tayan, 2011). Vafeas (1998) finds evidence that the nominating committee can influence the independence of the outside directors and thereby strengthen the quality of corporate governance. Furthermore, Brown and Caylor (2004) find evidence that independent nominating committees are associated with better firm performance.
The responsibility of the compensation committee is to set the CEO’s compensation plan and give advice to the CEO on the compensation plans for other executives (Lacker & Tayan, 2011). However the results from Gregory-Smith (2012) indicates there is no significant relationship between the independence of the compensation committee and compensation plan for the CEO.
2.4 Hypothesis development
As discussed in the previous paragraphs tax avoidance is a way to decrease the tax liabilities of a firm and therefore increase the firms after tax profit. If firms engage in such tax avoiding activities the tax percentage per dollar is transferred to the shareholders. According to Hanlon and Heitzman (2010) there are multiple ways to measure tax avoiding activities, and different tax avoiding activities are reflected in different measures. The proxies this research uses to measure the extent in which firms avoid taxes is further explained in paragraph 3.3.2. Prior literature of Pergola and Joseph (2011); Desai and Dharmapala (2006); Desai et al (2007b) and Jensen and Meckling (1976) shows that equity incentives are often used to align the interests of the executives with the shareholders. According to Pergola and Joseph (2011) there are two theories on the equity incentives. According to the convergence of interest theory, equity incentives align the interest of the executives with the shareholders of the firm in a linear line and thus granting more equity incentives results in more alignment of these interests. Based on this theory, we can expect that executives with more equity incentives are incentivized to avoid taxes to increase the after tax profit and thereby the firms overall wealth. Pergola and Joseph (2011) also introduce a second theory on equity incentives. According to the entrenchment theory the middle range of the linear relation between equity incentives and the interests alignment with the shareholders
7 overcome the control mechanism and are incentivized to act in their own self-interest without fairing dismissal or consequences if caught. Based on this theory, we can expect that managers will not focus on tax avoiding activities that increase the firms wealth but on activities that benefits only their own wealth like consuming perks. Because this kind of behaviour is difficult if not impossible to obtain from archival database data, I will base my expectations on the convergence of interest theory and expect that executive with more equity incentives will engage more in tax avoiding activities in order to increase the firms wealth and therefore the shareholders wealth. Because the executives with equity incentives are part of the shareholders they also increase their own wealth. In this study, I focus only on the CEO’s equity compensation. Even though he or she has no direct responsibility over the tax function, CEO’s are considered to see tax departments as profit centres (Crocker & Slemrod, 2005) and use their influence to maximize the profit within this profit centre. This is consistent with the findings of Dyreng et al (2010) who find evidence that CEO’s influence the level tax avoiding activities. Therefore the first hypothesis is as follows:
Hypothesis 1: The level of equity incentives a CEO receives is positively related with the extend firms engage in tax avoiding activities.
It is important to note that Khurana and Moser (2013) find evidence that long term orientated shareholders will discourage tax avoidance as it could decrease the firms wealth on the long term. This could be an explanation if the results suggests that equity incentives are negatively related with tax avoiding activities. Given the statement of Cheng and Warfield (2005) that CEO’s with higher equity incentive are more likely to sell their shares and therefore their wealth is sensitive for short-term stock prices it could be that equity incentives do not align the interests of the shareholders. In fact this could deviate the interests of the CEO’s and shareholders. Given the fact that unexercisable option often are exercisable after three or four years, I assume that equity incentives align the interests of the CEO with the short-term investors and misalign the interest of the CEO with the long-term investors. This assumption is consistent with the expectation that CEO’s with larger extents of equity incentives are related with more tax avoiding activities.
As discussed in paragraph 2.3 Larcker and Tayan (2011) define the framework as control mechanisms that prevent or dissuade self-interested managers from making decisions that are not in the interest of the investors and stakeholders. It is even considered as the most important mechanism to prevent executives act in their own self-interest. Armstrong et al (2015) argues that equity incentives themselves are a governance mechanism which align the interests of the executives with the shareholders. Based on the argument of Desai and Dharmapala (2006) and Wilson (2009) that tax avoidance will reduce the corporate transparency and that corporate governance mechanisms are put in place to make firms more transparent, I expect that firms with higher levels of corporate governance will engage less in tax avoiding activities, as such activities tend to decrease the transparency of the firm. Therefore the second hypothesis is as follows:
8
Hypothesis 2: The level of corporate governance have a negative relationship with tax avoidance.
As the first hypothesis predicts that equity incentives will lead to more tax avoiding activities and the second hypothesis predict that the level of corporate governance will lead to less tax avoiding activities, the third and final hypothesis will test whether the level of corporate governance will mitigate the relationship between equity incentives and tax avoiding activities. Therefore, the third hypothesis is as follows:
Hypothesis 3: The level of corporate governance will mitigate the relationship between equity incentives and tax avoidance.
3. Methodology
To test the hypothesis, I used quantitative archival databases. This method is commonly used in this research area and therefore assumed to be the most appropriate method to conduct this research.
3.1 Sample
For this research study the empirical approach will be archival/databased research. The sample consists of the 500 largest companies of the U.S., also called the S&P 500 companies. The S&P 500 is used because of the availability of data needed to determine the equity incentives. Furthermore, large companies are more likely to affect the society when tax is avoided relatively to smaller companies. This is due to the fact that tax authorities collect a certain percentage of the pre-tax book income and larger firms are expected to have a greater amount as pre-tax income. If these large firms avoid taxes by, for example, decrease their pre-tax taxable income with 1%, this will have more impact on the amount of taxable income then when a smaller firm with expected smaller pre-tax income decrease this with 1%. Because of availability of data to compute the B-index the fiscal year range will be from 2007 until 2014. Following Armstrong et al (2015), all foreign registrant and real estate firms are eliminated because they are subjected to different tax laws.
The first step was to obtain all financial data from COMPUSTAT. The information needed for equity incentives is obtained from the EXUCOMP database within COMPUSTAT and the North America database within COMPUSTAT. The data needed for measuring tax avoidance is obtained from the North America database within COMPUSTAT. The data needed for the corporate governance is obtained from the ISS database, formerly known as RiskMetrics.
The second step was to merge the three different datasets by common company identifiers (GVKEY) and fiscal year. After the merge of these datasets the observations with missing data were deleted from the sample.
Starting the data gathering process for the B-index the sample size consists of 2,278 firm year observations. Due to unclear data regarding the CEO flag, which indicates yes when the executive is a CEO and no if not, in the ISS database a substantial portion of the sample has to be dropped in order to be sure the observation contains the actual CEO of that firm. After some manually checks whether
9 the data is conforming the annual reports, I found that this flag also indicates yes if the executive is a CEO at another firm. After deleting all missing observation within the ISS database and COMPUSTAT, 490 observation remained.
3.2 Research design
This paper first examines the relation between CEO equity incentives and corporate tax avoidance and second examines if the corporate governance quality has an effect on this relationship. Equity incentives for CEO’s are determined through four out of the five elements of Cheng and Warfield’s (2005) paper. Tax avoidance is determined utilizing three measures of the Rego et al (2012) paper. These measure are DTAX, CASHETR and SHELTER. In order to determine corporate governance quality I used the B-Index taken from the Baber et al (2012) paper. According to Baber et al (2012) this measure captures the internal governance quality. It focusses primarily on the role of director independence. B-index total score is computed as the sum of dummy variables created for each of the six elements of this index. This dummy variables is 1 if the firm in year t is compliant to the conditions and 0 if not.
Since other factors can potentially influence the relationship between CEO equity incentives, tax avoidance and corporate governance this research includes several control variables. According to Desai and Dharmapala (2006) (p.164) the control variable SIZE capture the changes in the scale of or size of the firm. To capture the SIZE proxy this research uses the natural logarithm of the firm’s market capitalization (LOG_MARKET) following Armstrong et al (2012). In order to control for the underlying economic activity of the firm the proxy return on assets (ROA) is included. Also, following Armstrong et al (2012). Leverage (LEV), change in goodwill(∆GDWL), and foreign assets (FORAS) are included as control variables. Because of data limitations the foreign assets need to be estimated by the method of Older et al (2007). Due to data limitations this research made a variation on the method. The return on sales can be composed by dividing sales with the total assets (AT). In COMPUSTAT data for domestic and foreign sales is available wherefrom the proportion of foreign sales to total sales can be determined. Assuming that the foreign sales and domestic sales have the same ratio to asset total, it is possible to estimate the foreign assets. COMPUSTAT historical segments provide information on foreign and domestic sales and income. By multiplying the asset total to foreign sales divided by the total sales, I estimated the foreign asset total.
To test the first hypothesis, which state that equity incentives are expected to lead to more tax avoiding activities, this research uses the following model:
= , + , + , + , + ∆ , + ,
Where:
TAXAV = Tax avoidance
EQINCEN = Equity incentives as the sum of OWNERSHIP, EXERCISABLE and UNEXERCISABLE SIZE = The natural logarithm of the firms’s market capitalization
10
LEV = leverage: Long term debt divided by asset total
∆GDWL = change in goodwill: goodwill in year t-1 minus goodwill in year t
FORAS = estimation of foreign assets: foreign sales divided by total sales times total assets
To test the second hypothesis, which state that the quality of corporate governance has a negative relationship with tax avoiding activities, this research uses the following model:
= , + , + , + , + ∆ , + ,
Where:
TAXAV = Tax avoidance
BINDEX = Total of dummy variable of each of the six elements. Each dummy variable is one if firm is compliant
and zero if not: 1) CEO is not chairman of the board, 2) two third or more members of the board is independent, 3) all audit committee members are independent, 4) all compensation committee members are independent, 5) separate and independent nominating committee, and 6) the board size is less than the median of distribution for all firms (adjusted for firm size)
SIZE = The natural logarithm of the firms market capitalization ROA = The return on assets: net income divided by total assets LEV = leverage: Long term debt divided by asset total
∆GDWL = change in goodwill: goodwill in year t-1 minus goodwill in year t
FORAS = estimation of foreign assets: foreign sales divided by total sales times total assets
To test the third and las hypothesis, which state that the quality of corporate governance moderates the relationship between equity incentives and tax avoiding activities, this research uses the following model:
= , + ( ∗ ), + , + , + , + ∆ ,
+ ,
Where:
TAXAV = Tax avoidance
EQINCEN = Equity incentives as the sum of OWNERSHIP, EXERCISABLE and UNEXERCISABLE BINDEX = Total of dummy variable of each of the six elements. Each dummy variable is one if firm is compliant
and zero if not: 1) CEO is not chairman of the board, 2) two third or more members of the board is independent, 3) all audit committee members are independent, 4) all compensation committee members are independent, 5) separate and independent nominating committee, and 6) the board size is less than the median of distribution for all firms (adjusted for firm size)
11
SIZE = The natural logarithm of the firms market capitalization ROA = The return on assets: net income divided by total assets LEV = leverage: Long term debt divided by asset total
∆GDWL = change in goodwill: goodwill in year t-1 minus goodwill in year t
FORAS = estimation of foreign assets: foreign sales divided by total sales times total assets
3.3 Measures
3.3.1 Equity incentives
To determine equity incentives for CEO’s I followed Cheng & Warfield (2005). They used five elements to calculate the total equity incentives. These elements are: option grants in the current period, exercisable options (option in current period excluded), exercisable option, restricted stock grants, and stock ownership. Each of these measurements are divided by the total outstanding shares of the company. The potential benefits caused by an increase in short-term stock prices, is shared by all shareholders. The benefit enjoyed by the CEO is thus the proportion of the ratio of equity incentives elements held by the CEO divided by the total outstanding shares.
Cheng and Warfield (2005) report that restricted stock appear to have a small role in stock based compensation and therefore have a limited stake in CEO equity based holding. The restricted stock represents only 0.008% of the total outstanding shares according to Cheng and Warfield (2005) (p.450). Based on this information and due to data limitations, this research excludes restricted stock awards.
3.3.2 Tax avoidance
According to Hanlon and Heitzman 2010 (p. 139) there is no best single measure for tax avoidance because each measure captures different aspects of tax avoiding activities. The effective tax rate measures, GAAP ETR and CASH ETR capture the average rate of tax per dollar of income. Whereby the GAAP ETR is affected by the accounting earnings but do not reflect tax deferral strategies. CASH ETR on the other hand do reflect these tax deferral strategies but have no impact on accounting earnings. The book-tax difference measures, like BTD and SHELTER, capture the difference between the taxable income reported to the tax authorities and the book income reported in the annual reports. Wilson (2009) find evidence that firms with large book-tax differences are more accused of tax avoidance than similar firms with lesser extents of book-tax difference. Discretionary or abnormal measures of tax avoidance, like DTAX, capture the portion driven by intentional tax avoidance. The discretionary portion attempts to remove underlying determinants that are not driven by intentional tax avoidance, leaving the portion driven by intentional tax avoidance in the residual (Hanlon & Heitzman, 2010). As mentioned earlier in this paragraph Hanlon and Heitzman (2010) state that there is no single measure that captures all different aspects of tax avoidance. And thus, there is no single best measure to measure tax avoiding activities. Because my research is focused on all tax avoiding activities that increases the firms wealth, regardless if this increase is short, mid or long term, I will use a measure to capture tax avoiding activities that affect accounting earnings, a measures that capture tax deferral
12 strategies and a measures that capture discretionary or abnormal book-tax differences. Therefore I combine three measures for calculating TAXAV, namely: CASH_ETR, SHELTER and DTAX. Using this three measures I am following Rego et al (2012). According to Hanlon & Heitzman (2010) CASH ETR is affected by tax deferral strategies but not by changes in the tax accounting accruals. The CASH ETR is computed by cash taxes paid (TXPD) in year t divided by pre-tax income (PI) less special items (SPI) in year t. The outcome of this formula is then scaled between 0 and 1, whereby 0 represents the smallest observation and 1 the largest observation. Therefore, I used the following formula:
= − min( ) max( ) − min( )
In order to capture the unexplained book-tax difference, I use the SHELTER proxy from the paper of Wilson (2009). SHELTER, which is a tax shelter prediction score which explains the unexplained book-tax difference. Prior research from Wilson (2009) shows that SHELTER are significantly related with real case tax avoidance. SHELTER is a dummy variable which equals to 1 for firms in the top quintile of the predicted probability of engaging in tax sheltering and 0 if not. Like Rego et al (2012) I follow Wilson (2009) using this measure. The findings in the paper of Wilson (2009) provided the following weights which I use in calculating the measure:
Intercept = -4.30
Weight for book to tax difference = 6.63 Weight for leverage = 1.72
Weight for size = 0.66
Weight for return on assets = 2.26 Weight for foreign income = 1.62
Weight for research and development = 1.56
SHELTER = −4.30 + 6.63 × BTD − 1.72 × LEV + 0.66 × SIZE + 2.26 × ROA + 1.62 × FOR INCOME + 1.56 × R&D
Following McGuire et al (2012) I calculate book-tax difference (BTD) as pre-tax book income less estimated taxable income scaled by total assets, where pre-tax book income is calculated by subtracting minority income interest from pre-tax income. Taxable income is calculated by summing up the current federal tax expense and current foreign tax expense and dividing this sum by the top U.S. statutory tax rate less the change in net operating carry forwards losses (McGuire et al, 2012). Leverage (LEV) is calculated by dividing the long term debt (DLTT) by the total assets (AT) at the end of year t (McGuire et al, 2012). Size is the natural log of market value of equity which is calculated by annual closing price (PRCC_F) times common shares outstanding (CSHO) in year t (McGuire et al, 2012). ROA is calculated by dividing income before extraordinary items (IB) with the average asset total (AT) for year t (McGuire et al, 2012). Foreign income is calculated by scaling pre-tax foreign income(PIFO) by total assets (AT) at the beginning of year t for year t (McGuire et al, 2012). Research and development (R&D) is
13 calculated by scaling the research and development expenses (XRD) to the total assets (AT) at the beginning of the year t (McGuire et al, 2012).
To capture the abnormal book-tax differences, I used a measure which is similar to the measure of Desai and Dharmapala (2006, 2009). Frank et al (2009) developed a proxy for tax avoidance (DTAX). This measure is the predicted residual of the PERMDIFF regression according to Frank et al (2009), where PERMDIFF is basically the difference between the effective and statutory tax rate. DTAX explains the unexplained portion of the ETR differential. According to Hanlon and Heitzman (2010) the intent of the measure is to capture items that reduce the GAAP ETR rate of the firms. When GAAP ETR rates are reduced the bottom line earnings or after-tax earnings will increase, and thereby the firms wealth. Hanlon and Heitzman (2010) also state this measure is appropriate to capture tax avoiding activities that lower the firms GAAP ETR and boost accounting accruals. This is based on the fact that is a function of GAAP ETR which captures the changes in accounting accruals and it reflect the non-conforming tax avoidance.
I calculate the DTAX by the residual of the following regression:
= + + + + + ∆ +
+
Where:
PERMDIFF = BI – [(CFTE + CFOR)/STR] – (DTE/STR) according to Hoi et al (2013) BI = pre-tax book income (PI) for firm i in year t.
CFTE = current federal tax expense (TXFED) for firm i in year t. CFOR = current foreign tax expense (TXFO) for firm i in year t. DTE = deffered tax expense (TXDI) for firm i in year t.
STR = Statutory tax rate in year t.
INTANG = goodwill and other intangibles (INTAN) for firm i in year t.
UNCON = income (loss) reported under the equity method (ESUB) for firm i in year t. MI = income (loss) attributable to minority interest (MII) for firm i in year t.
CSTE = current state income tax expense (TXS) for firm i in year t.
∆NOL = change in net operating loss carry forwards (TLCF) for firm i in year t. LAGPERM = the one year lagged PERMDIFF for firm i in year t.
Like Hoi et al (2013) I follow Frank et al (2009) regarding possible missing data that is needed to calculate the variables for this regression. This entails that if minority interest (MI), current foreign tax expenses (TXFO), income from unconsolidated entities (ESUB) or current state tax expense (TXS) is missing in the dataset, I set the variables to zero. If current federal taxes expense (TXFED) is missing in the dataset, I set the value to total tax expense (TXT) less current foreign tax expense (TXFO) less current state tax expense (TXS) less deferred tax expense (TXDI). If the data for goodwill and other intangibles is missing from the dataset, I set the value to zero. If COMPUSTAT marks the INTAN with “c” I will set INTAN equally to the value of goodwill (GDWL)
14
3.3.3 Corporate governance
Baber et al (2012) separate two kinds of corporate governance mechanism, external corporate governance and internal corporate governance. According to these authors external governance systems facilitates or discourages active stakeholder participation in the governance process. If the costs of the stakeholder participation in order to establish restrictions and provisions is relatively high, the quality of corporate governance is low. The internal governance reflects the interactions among firm management, directors and employee’s and focuses mainly on the independence of the board of directors.
In order to obtain the level of corporate governance, I used the B-index. This measure reflects the internal governance and focuses on the role of director independence. To calculate the B-index the index increases by 1 if 1)CEO is not chair, 2) two/third or more is independent, 3) all audit committee members are independent, 4) all compensation committee members are independent, 5) separate independent nominating committee and 6) the board size is less than the median of distribution for all firms (adjusted for firm size)
4. Results
4.1 Descriptive statics
The descriptive statics in table 1 provides the following insights. The first dependent variable for tax avoidance, SHELTER, has a mean of .2510 and a standard deviation of .4340. The second dependent variable for tax avoidance, CASHETR, has a mean of .8877 and a standard deviation of .1154, so there can be concluded that most firms within the sample are in the right rail of distribution. The third and last dependent variable, DTAX, has a mean of 6.2148 and a standard deviation of 1.3509. The first independent variable EQINCEN has a mean of .8979 and a standard deviation of .1154. The mean for the dummy variable B-INDEX is 4.0918 which indicates that this sample consists of firms that on average score 4 point out of 6. For elements 1) independence of executives, 2) independence of audit committee member, and 3) independence compensation, all means are above 0.9, which indicates that this sample consists of more companies that score 1 on this elements. This sample also consists of more firms that scored 1 on the element of board size, according to the mean of 0.7. The mean of element CEO = CHAIR and separate nominating committee is 0.4224, which indicates that slightly more than half of the sample firms do not comply to this element. The mean of element separate nominating committee members is 0.0469, which indicates that most firms within the sample do not comply to this elements. All variables except for the elements of B-index, B-INDEX itself and SHELTER are checked for normality. The elements of B-index, B-INDEX itself and SHELTER are not checked for normality because they are dummy variables, with either the value “zero” or “one”, and therefore not need to be normally distributed. Bases on this, variable DTAX is transformed into the natural logarithm of the variable. EQINCEN, FORAS and LEV are transformed into the square root of variable. SIZE is transformed into one dived by the square root of the variable.
15
Variables N mean sd min max p.25 p.75 Var skewness kurtosis
TAXAV: SHELTER 490 .25102 .43404 0.00000 1.00000 .00000 1.00000 .18839 1.14843 2.31889 CASHETR 490 .88774 .11541 0.00000 1.00000 .89576 .93707 .01332 -4.45304 27.63187 DTAX 490 6.21479 1.35094 1.36690 9.94159 5.228263 7.00838 1.82504 .23021 3.30635 EQINCEN (%) 490 .89793 .70801 .00000 5.12705 .48530 1.06553 .50128 3.02598 15.51849 B-INDEX: 490 4.09184 .73874 2.00000 6.00000 4.00000 5.00000 .54574 -.26862 2.89586 CEO CHAIR 490 .42245 .49445 .00000 1.00000 .00000 1.00000 .24448 .314004 1.09859 INDEPEN 490 .94286 .23235 .00000 1.00000 1.00000 1.00000 .05399 -3.81583 15.56061 AUDITCOM 490 .98163 .13441 .00000 1.00000 1.00000 1.00000 .01807 -7.173783 52.46316 COMPENCOM 490 .99796 .04518 .00000 1.00000 1.00000 1.00000 .00204 -22.06812 488.00200 NOMICOM 490 .04694 .21172 .00000 1.00000 .00000 .00000 .04483 4.28411 19.35360 BOARDSIZE 490 .70000 .45873 .00000 1.00000 .00000 1.00000 .21043 -.87287 1.76190 SIZE 490 .32172 .02045 .27371 .40930 .30718 .33550 .00042 .0166118 2.95228 ROA 490 .07883 .05859 -.31074 .36052 .04569 .10962 .00343 -.58589 11.65226 LEV 490 .42990 .17393 .00000 .91362 .35301 .53670 .03025 -.73778 3.75401 490 240.19430 987.41600 -3789 9858 -7.25000 187.32980 974990 3.77198 29.02042 FORAS 490 89.74685 87.51874 .00000 740.9106 37.34856 104.94400 7659 2.58871 13.53448
16 Table 2 provides the Pearson correlation matrix. Pearson correlation matrix calculates the correlation between each separate variable which are used in the regression. All significant correlations are stated in bold. When the correlation coefficient exceeds 0.8 it could indicate that there is a multi-collinearity between the variables. Multi-multi-collinearity could lead to unreliable inferences bases on the data set.
The correlation matrix indicates that SHELTER and DTAX are positively correlated with each other, but negatively correlated with CASH ETR. Despite the relatively high correlation between each other, there can be noted that the significance is low. The correlation between equity incentives (EQINCEN) and SHELTER and DTAX are negatively correlated at a low significance. This is contradicting with the expectation that more equity incentives is positively related with more tax avoidance. Equity incentives are, according to this table positively correlated with CASHETR where I note a small correlation of 0.1399. B-INDEX is not correlated with the tax avoidance variables which is not in line with the expectations. SIZE is negatively correlated with tax avoidance measures SHELTER and DTAX with low significance, which indicates that larger firms may engage less in tax avoidance practices. ROA is positively associated with SHELTER and DTAX with low significance, which is in line with the expectation that more tax avoidance results in higher ROA ratio’s. LEV has a low correlation with DTAX with low significance. The change in goodwill (∆GDWL) has low correlation with SHELTER and DTAX with low significance. Foreign assets has relatively high correlation with SHELTER and DTAX with low significance, where from there can be concluded that firms with more foreign assets engage in more tax avoiding practices.
17 n = 490 1 2 3 4 5 6 7 8 9 10 1. SHELTER - 2. DTAX 0.6392 * - 3. CASHETR -0.5844 * -0.6270 * - 4. EQINCEN -0.1222 * -0.1517 * 0.1399 * - 5. B-INDEX 0.0746 0.0791 -0.0340 -0.0841 - 6. SIZE -0.6545 * -0.8351 * 0.6398 * 0.2068 * -0.0460 - 7. ROA 0.3555 * 0.1337 * -0.2178 * 0.0488 -0.0407 -0.2016 * - 8. LEV -0.0875 0.0925 * 0.0714 -0.0397 0.0871 -0.0762 -0.2247 * - 9. ∆GDWL 0.1878 * 0.2383 * -0.1833 * 0.1203 * 0.0657 -0.2209 * 0.1299 * 0.0411 - 10. FORAS 0.5316 * 0.6314 * -0.6825 * -0.1310 * 0.0676 -0.6452 * 0.0228 0.0104 0.2059 * - Note: Significant effects are shown in bold, *significant at p<0.10, **significant at p<0.05, ***significant at
p<0.01.
18
4.2 Results
In order to effectively perform the regression some assumptions must be made. Firstly there must be no multi-collinearity between the independent variables. This assurance is tested with the variance inflation factor (VIF) test. The VIF test for all three regressions shows that most of the variables in all three regressions have a value lower than 10. Only the variable that is created for H3, (EQINC*B-INDEX) and EQINC shows a value higher than 10 (11). This is explained by the fact that EQINCEN * B-INDEX is variable EQINCEN multiplied by the score between 1 and 6 of the B-INDEX. To test whether the residuals are normally distributed I performed the Shapiro-Wilk test. The alpha level is set at 0.05. Residuals are normally distributed when p-value exceeds this alpha level. The p-value of the performed Shapiro-Wilk test is <0.05 and therefore we can be concluded that the residuals are not normally distributed. Because the fact that the variables are already checked, and transformed when needed, I accept the distribution of the residuals as a limitation and proceed with the multi variate regression.
Table 3 provide the regression results for model 1. This model tests the relationship between tax avoidance and equity incentives, main effect, and supported by the control variables. The dependent variable TAXAV is divided in three measures, namely: SHELTER, DTAX and CASHETR. Table 3 provides the results of the regression for each of these measures. The R-sqaure’s are repectively 0.5170, 0.7157 and 0.5596. The R-square for the model measuring the relationship between SHELTER and equity incentives indicates that 51.70% of variation is explained by the model. The model for testing the relationship between DTAX and equity incentives shows a R-sqaure of 0.7157, indicating that 71.57% of the variation is explained by the model. The R-square for the model that tests the relationship between CASHETR and equity incentives indicates that 55.96% of the variation is explained by the model. This percentage is in general higher if more control variables are applied. The table shows that all dependent variables, SHELTER, DTAX and CASHETR, have no significant relationship with the independent value for equity incentives. The p-values are respectively 0.678, 0.520 and 0.439. Providing this information we can conclude that providing more equity incentives causes more tax avoiding behaviour of the firm, and thereby we can reject H1. This results are not in line with results of prior literature of Desai and Dharmapala (2006) and Rego et al (2012). It is notable that control variables ROA, SIZE, LEV, and FORASS have a significant relation with all three measures for tax avoidance. It is Interesting to note that SIZE is negatively related with tax avoidance, implying that smaller firm engage more in tax avoiding activities. Control variable FORASS is significant positive related with both SHELTER and DTAX and significant negatively related with CASHSETR, implying that firms with more foreign assets engage in more tax avoiding activities. Overall, this regression table indicates there is no significant relationship between tax avoidance and CEO equity incentives and thus hypothesis 1 is rejected. Despite the fact that the main effect tested is not significant, the coefficients SHELTER and CASHETR are not as predicted. The coefficient for DTAX is as predicted.
19
Table 3: Multivariate regression annalysis, effect on equity incentives
Dependent variable → SHELTER DTAX CASHETR
Independen variable ↓ Coeff t-Stat Coeff t-Stat Coeff t-Stat
Main effect: EQINC -.08388 -0.42 .31007 0.64 .03966 0.77 Control variables ROA 1.7805 7.05 *** -.39091 -0.65 *** -.25652 -4.00 *** SIZE -9.7717 -10.42 *** -48.29911 -21.58 *** 1.70697 7.17 *** LEV -.17960 -2.19 ** .23508 1.20 ** .0473798 2.27 ** ∆GDWL .00001 0.39 .00006 1.74 -0.00001 -0.20 FORASS .00112 5.34 *** .00236 4.71 *** -.00063 -11.92 *** Observations 490 490 490 Prob > F 0.000 0.000 0.000 R-squared 0.5170 0.7157 0.5596
20 Table 4 provides the regression results for model 2, testing whether tax avoidance is related with the corporate governance quality of the firm. Similar to the regression used to test H1 the table provides the results for the main effect whether firms with higher corporate governance quality, measured with the B-INDEX, engage in more tax avoiding activities. The R-sqaure’s are repectively 0.5197, 0.7163 and 0.5591. The R-square for the model measuring the relationship between SHELTER and B-IDNEX indicates that 51.97% of variation is explained by the model. The model for testing the relationship between DTAX and B-INDEX shows a R-sqaure of 0.7163, indicating that 71.63% of the variation is explained by the model. The R-square for the model that tests the relationship between CASHETR and B-INDEX indicates that 55.91% of the variation is explained by the model. The second model indicates that there is no significant relationship between the quality of corporate governance, measured by the B-INDEX, and tax avoiding behaviour of firms. Given the lack of significance H2 can be rejected. The values and significance of the control variables are similar to the first model presented in table 3. I note the difference in significance for DTAX and ROA , which is highly significant in the first model and not significant in the second. However, the significance for the change in goodwill is also different compared with the first model. Overall, this regression table indicates there is no significant relationship between tax avoidance and quality of corporate governance and thus the second hypothesis is rejected. Even though the relationship between tax avoidance and internal governance is not significant it is notable that the results in table 4 shows coefficients that are contradicting to the expectations except for CASHETR.
21
Table 4: Multivariate regression annalysis, effect on B-INDEX
Dependent variable → SHELTER DTAX CASHETR
Independen variable ↓ Coeff t-Stat Coeff t-Stat Coeff t-Stat
Main effect: B-INDEX .03185 1.71 .05392 1.21 .00023 0.05 Control variables ROA 1.78517 7.11 *** -.34096 -0.57 -.25274 -3.95 *** SIZE -9.85071 -10.74 *** -48.02882 -21.89 *** 1.74239 7.46 *** LEV -.18975 -2.31 ** .21611 1.10 .04715 2.25 ** ∆GDWL .00001 0.24 .00006 1.80 * .00001 -0.08 FORASS .00110 5.27 *** .00233 4.66 *** -.00063 -11.91 *** Observations 490 490 490 Prob > F 0.000 0.000 0.00 R-squared 0.5197 0.7163 0.5591
22 Table 5 provides the results for the third and last model, measuring whether the quality of corporate governance has an effect on the relationship between tax avoidance and equity incentives. the R-squares are respectively 0.5181, 0.7171 and 0.5597. These R-squares indicates that 51.81% of the variation of the dependent variable SHELTER is explained by the model, 71.71% of the variation of the dependent variable DTAX is explained by the model and 55.97% of the variation of the dependent variable CASHETR is explained by the model. In table 5 the result indicate that corporate governance quality has nog significant effect on the relationship between equity incentives and tax avoiding activities. The p-values for EQINCEN are 0.257, 0.206 and 0.580. The p-values for EQICEN * B-INDEX are 0.290, 0.125 and 0.738. Overall, table 5 provides evidence that there is no significant relationship between equity incentives and tax avoidance and there is no significant evidence that corporate governance moderates this relationship and thus we can reject hypothesis 3. Given this information there can be concluded that the quality of corporate governance measured by the B-index does not have an significant effect on the relationship between equity incentives and tax avoiding activities. Despite the insignificance it is interesting to note that the coefficients for all three proxies change their direction. The results for the control variables are similar to the results of the previous regressions.
23
Table 5: Multivariate regression annalysis, effect on B-INDEX
Dependent variable → SHELTER DTAX CASHETR
Independen variable ↓ Coeff t-Stat Coeff t-Stat Coeff t-Stat
Main effect: EQINCEN -.75282 -1.14 -2.00120 -1.27 .00023 0.05 EQINCEN * B-INDEX .16091 1.06 .55598 1.53 -.01293 -0.34 Control variables ROA 1.78585 7.08 *** -.37238 -0.62 -.25695 -4.01 *** SIZE -9.75035 -10.40 *** -48.22541 -21.57 *** 1.70525 7.16 *** LEV -.18391 -2.24 ** .22018 1.12 .04772 2.28 ** ∆GDWL .00001 0.36 .00006 1.68 * .00001 -0.19 *** FORASS .00111 5.27 *** .00232 4.64 *** -.00063 -11.88 *** Observations 490 490 490 Prob > F 0.000 0.000 0.00 R-squared 0.5181 0.7171 0.5597
24
5. Conclusion
This research first examines whether equity incentives have a relationship with tax avoiding activities. Second, it examines whether the quality of corporate governance is related with tax avoidance. Third, it examines whether the quality of corporate governance moderates the predicted positive relationship between equity incentives and tax avoidance. All data used to perform this research is obtained from archival databases. The extend of tax avoidance is measured with three different proxies. According to Hanlon and Heitzman, there is no single measure that reflects all tax avoiding activities and therefore a combination of several measures should be made. Following Rego et al (2011) and Wilson (2009) the first measure used to capture tax avoiding activities is SHELTER, which is a tax shelter prediction score which explains the unexplained book-tax difference. Prior research from Wilson (2009) shows that SHELTER are significantly related with real case tax avoidance. The second proxy for tax avoidance used in this research is DTAX. DTAX measures the abnormal book-tax differences and is the predicted residual of the PERMDIFF regression according to Frank et al (2009) and Wilson (2009). DTAX explains the unexplained portion of the ETR differential. Because PERMDIFF is a function of GAAP ETR it captures the changes in accounting accruals and it reflect the non-conforming tax avoidance. The third and final measure used to capture the tax avoiding activities is CASH ETR. This measure is used to determine the tax deferral strategies which have impact on the ETR’s . To determine the quality of corporate governance I used the B-index score from the Baber et al (2012) paper. This measure determines the quality of internal governance and consists of 6 elements: 1)CEO is not chairman of the board, 2) two/third or more is independent, 3) all audit committee members are independent, 4) all compensation committee members are independent, 5) separate independent nominating committee and 6) the board size is less than the median of distribution for all firms.
Prior literature found that investors expect that the executives of the firm act in such way that the firms wealth increases. Tax avoidance is a way to increase the firms wealth as it increases the after-tax income. Prior literature also found that equity incentives are used to align the interest of the executives with the shareholders as equity award have an upwards potential. It is therefore expected that executives with larger amount of equity incentives engage more in tax avoiding activities. Prior research from Rego et al (2011) find evidence that equity incentives are significantly positively related with SHELTER and significantly negatively related with CASH ETR, which confirms the expected relationship between equity incentives and tax avoidance. On the other hand, Desai and Dharmapala (2006) find that high powered incentivized executives are associated with lower level of tax avoidance. Desai and Dharmapala (2006) findings suggest that the negative relationship between equity incentives and tax avoidance loses its significance when firms are well governed. Rego et al (2011) find similar results, their findings of the positive relationship between equity incentives and tax avoidance also becomes insignificant except for CASH ETR where they find a weak significance on a positive relationship between CASH ETR and tax avoidance, which suggests that corporate governance quality mitigates for the positive relationship between equity incentives and tax avoidance. Based on theoretical logic this research expects that equity incentives have a positive relationship with tax avoiding activities. This research also expects that corporate governance quality have a mitigating effect on this relationship.
25 In order to capture multiple aspects of tax avoidance, this research used multiple proxies to capture tax avoiding activities. The first measure is SHELTER, the prediction score of tax sheltering, captures the unexplained boot-tax difference. The second measure used is DTAX, the residual of the PERMDIFF regression, is used to capture the abnormal book-tax difference. The third and final measure is CASH ETR, cash taxes paid per dollar of pre-tax book income, captures the tax deferral strategies. Since other factor possibly also influence the tax sheltering activities, the model includes a couple of control variables. SIZE, ROA, LEV, ∆GDWL and FORASS are added to the model as control variable. Where SIZE captures the size of the firm, ROA captures the return on assets, LEV captures the leverage of the firm, ∆GDWL captures the change in goodwill of the firm and FORASS captures the foreign assets the firm has. To capture the quality of corporate governance, this research used the B-index score from Baber et al (2012) paper. This measure consists of six elements, which in total captures the internal governance quality of a firm: 1)CEO is not chairman of the board, 2) two/third or more is independent, 3) all audit committee members are independent, 4) all compensation committee members are independent, 5) separate independent nominating committee and 6) the board size is less than the median of distribution for all firms. All elements are 1 when the firm is compliant and 0 if not. To test the first hypothesis, whether equity incentives have a positive relationship with tax avoidance, I run a regression for each of the tax avoiding measures. To test the second hypothesis, whether the quality of corporate governance have a relationship with tax avoiding activities, I run a multiple regression for each of the tax avoiding proxies. To test the third and final hypothesis, which tests whether the quality of corporate governance moderates the effect of equity incentives on tax avoidance, I run a regression for each of the tax avoiding proxies where the equity incentives are multiplied with the corporate governance score.
The results of the first hypothesis regression test, displayed in table 3, provide evidence that there is no significant relationship between equity incentives and tax avoiding activities. However, there is evidence that size, leverage, return on assets and foreign assets are related with all of the tax avoiding proxies. The results of the second hypothesis regression test, displayed in table 4, provide evidence that there is no significant relationship between the quality of internal governance and tax avoidance. This test find similar results for the control variables size, return on assets, change in goodwill and foreign assets. The results of the third hypothesis regression test, displayed in table 5, find evidence that corporate governance have no moderating effect on the relationship between equity incentives and tax avoidance. In fact all of the three proxies of tax avoidance have no significant relationship with equity incentives times the b-index score. Here I find also some similar results on the control variables. In the third regression, the change in goodwill also has a significant relationship with the proxies DTAX and CASHETR. In sum, all the hypothesis are rejected. In contrast with prior research, my results suggest there is no significant relationship between equity incentives and tax avoiding activities. The results also suggest there is no evidence for a significant relationship between corporate governance quality and tax avoiding activities. According to the results of this research, the quality of corporate governance does not moderate the effect between equity incentives and tax avoidance.
A possible explanation and limitation of this study is the fact that I excluded tax directors and their equity incentives from the sample. Armstrong et al (2011) find evidence that CEO’s and CFO’s
