Amsterdam Business School
CEOs and CFOs equity incentives and Tax avoidance
Thesis
Name: Arjanto Dekkers Student number: 10823875 Date: 20 june 2016
Faculty of Economics and Business, University of Amsterdam Supervisor: dhr.dr. A. Sikalidis
Statement of originality
This document is written by student Arjanto Dekkers 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 Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
This study examines the relationship between executives equity incentives compensation and corporate tax avoidance. The empirical findings are based on a sample of companies listed on the S&P 1500 from the years 2010 to 2014. For the corporate tax avoidance measure, GAAP ETR, there is evidence that executives equity incentives compensation is negatively associated with that particular ETR. These results suggest that equity incentives compensation induce management to lower the effective tax rate and thus avoid taxes. Moreover, the association is stronger for CFOs than for CEOs meaning that CFO equity incentives have more influence on corporate tax avoidance strategies than CEO equity incentives. Furthermore, I found evidence for a negative association between CFOs equity compensation and CURRENT ETR, Long-Run CASH ETR and CASH ETR. The evidence suggests that shareholders incentivize not the CEO but the CFO to avoid taxes since he is responsible for the tax management of the firm. In general, the results are consistent with previous literature that studied executives compensation packages and tax avoidance strategies and suggest that equity incentives induces lower ETRs that reflect corporate tax strategies.
1 Introduction
Tax avoidance is a popular topic in research as well as in the news lately. Especially after the Organisation of Economic Development (OECD) had announced that Governments increasingly are dependent on taxes of individuals. Corporate taxpayers developed ways to pay less, while individuals have to fill in the gap by paying more income taxes, social contributions and value added taxes (2015). Furthermore, the OECD reported that between 4 and 10 percent of global revenue is lost due to avoiding taxes. This is costing countries between $100bn- $240bn a year. Major multinational companies, Google, Facebook, Starbuck, J.P. Morgan, Pepsi, Proctor & Gamble and Netflix avoid taxes (Guardian, 2014; Guardian, 2015). At a time when governments spending’s are being cut and having an impact on the lives of everyday people, multinationals are willing to avoid their social responsibilities to the public (Guardian, 2013).
Because of this governmental and international institutions are increasing their focus on companies that avoid taxes and have set up several projects, such as the 15 step approach by the OECD, to decrease the likelihood of companies reducing the tax burden raised upon them. The institutions argue that profits earned should be taxed in the countries where profits are generated and thereby implying that profit shifting should be punished. To accomplish this they suggest that governments across the world should work together (OECD, 2015; HMRC, 2015; OECD, 2015).
However, not only governmental institutions have increased their focus on tax avoidance. Researchers also targeted tax avoidance and what causes management to engage in these strategies. However, prior studies show contradicting results. For example, prior research has already established that tax director compensation is associated with lower GAAP effective tax rates (ETR), however they do not find association between CEO or CFO compensation and tax avoidance (Armstrong, Blouin and Larcker, 2012). On the contrary, Rego and Wilson (2012) found that greater equity risk incentives are associated with higher tax risk. Furthermore, they found some evidence of the positive relation between equity risk incentives and tax risk for CFOs is less than that for CEOs. Moreover, Dyreng, Hanlon and Maydew (2010) results suggests that both the CEO and CFO have a significant impact in lowering the ETR. While Armstrong et al. (2015) were not able to find association between management incentives and lower ETRs. A survey study from Graham et al. (2014) examined the incentives for executives to engage in tax avoidance strategies based on evidence from the field. He found that financial accounting incentives play a role for executives to engage in such strategies. However, he also found that the most prominent reason not to engage in tax avoidance is reputational damage.
This study builds on and extends the tax avoidance literature that examines the relationship between equity incentives and corporate tax avoidance by using a more recent sample from 2010 to 2014, this might impact the results due to the increased focus of governmental institution. Moreover, I will use the tax avoidance measurements GAAP ETR, CURRENT ETR, Long-Run CASH ETR and CASH ETR in searching for an association between equity incentives and tax avoidance. Furthermore, the measurement of equity incentives is based on Bergstresser & Philippon’s model (2006) and is according to my knowledge not been used in tax avoidance studies. The research question that I developed is as follows: To what extent is the influence of CEO equity incentives and CFO equity incentives associated with corporate tax avoidance?
Based on the research question I have developed two hypothesis to test and the results indicate that for the extensive range of corporate tax avoidance there is evidence for a negative association between GAAP ETR and CEO and CFO equity incentives, whereby the association is stronger for the CFO than for the CEO. Suggesting that CFOs equity incentives have more influence on corporate tax avoidance than CEOs equity incentives. Furthermore, I found evidence for an negative association between CURRENT ETR, Long-Run CASH ETR and CASH ETR and CFO equity incentives. The evidence suggests that shareholders incentivize not the CEO but the CFO to avoid taxes since he is responsible for the tax management of the firm.
The structure of the paper is as follows. Chapter 2 elaborates on corporate tax avoidance and related economic literature. In particular, it reviews reports of governmental and international institutions, literature concerning the costs and benefits of tax avoidance strategies, performance measures of tax departments and the relation between equity incentives and riskier projects. Chapter 3 describes the data selection process and the models which are used in determining tax avoidance. Moreover, the tax avoidance measures are described and their weaknesses and strengths are highlighted. Chapter 4 focuses solely on the results of the empirical tests. And finally, chapter 5 concludes the results and the limitations of this study are mentioned.
2 Background & Literature
In this chapter, I will explain the phenomenon of corporate tax avoidance. After that, I will elaborate on the possible benefits and costs of tax avoidance. Thereafter, I will provide literature that acknowledges shifts in the performance evaluation of tax departments. And finally, I will explain equity incentives and the link with management taking on riskier projects. After this extensive literature I will provide the hypotheses.
2.1 Corporate Tax Avoidance
Tax avoidance is since recently a popular topic for society, but for researchers it is a popular topic for over decades. Tax avoidance is considered to be the reduction of explicit taxes, which is accomplished by engaging in tax strategies. This definition is also consistent with Dyreng et al. (2008) which reflects all transactions that have any effect on the firm’s explicit tax liability. However, tax avoidance does not have to imply any illegal activities. There are areas in the tax code that allow and encourage companies to reduce their tax liability. Moreover, the law is in many areas unclear, especially in complex transactions, and firms may take a certain position in which the outcome is unclear (Dyreng et al., 2008, p. 62). In this study, I consider tax avoidance as encompass anything that reduces the firm’s taxes relative to pre-tax accounting income. However, corporate tax avoidance is considered to be unethical and several governmental institutions increased their focus on this behaviour.
The Organisation for Economic Co-operation and Development (OECD) has done an extensive amount of research on tax avoidance and has introduced a 15 step plan to counteract tax avoidance. Measuring the scope of the amount of avoiding taxes is difficult. However, the OECD estimated that between 4 and 10 percent of global tax income is lost which is 100 to 240 billion U.S. dollars (2015). Moreover, aggressive tax planning has increased among multinational enterprises. Consequently, the affiliates of those multinational enterprises showed almost twice the profit rate of their global groups suggesting that these enterprises shift profits to countries where the tax burden is lower (p.4). Furthermore, the OECD also acknowledges the fact that the globalization of the economy provides opportunities for multinational enterprises to avoid taxes. Therefore, the report emphasizes that governments need to cooperate and refrain from harmful tax practices.
The Congressional Research Service (CRS) has done research about tax avoidance in the U.S. and has estimated that the potential revenue shifted by multinationals vary noticeably in
excess of $100 billion annually. The Joint Committee on Taxation estimated the potential magnitude of revenue lost from deferral of $83.5 billion in 2014 for U.S. multinationals (Gravelle, 2015). While the U.S. Administration estimated $63.4 billion. Moreover, they estimated for 2002 that corporate tax could be cut to 28% from 35% if deferral were ended. The study also emphasizes the fact that there is an increase in avoiding taxes from 2004 to 2010. For example, the Netherlands has an ETR below 12% and the profits of U.S. controlled corporations as a percentage of the GDP (Gross Domestic Product) was in 2004 in the Netherlands 4.6% which increased to 17.1% in 2010. This means that extensive amounts of money is avoided by U.S. multinational corporations (pp. 17-18).
The Her Majesty’s Revenue and Customs (HMRC), the department of taxes of Great Britain, measures the tax gap annually. In their report of 2015 they give an insight of the tax gap of 2013-2014 whereby the tax gap the difference is between the amount of tax that should be collected by the HMRC against what is actually collected. They estimated the total tax gap for corporations at 3 billion pounds. However, the overall corporation tax gap has declined from 14 cent in 2005-2006 to 7 cents in 2013-2014. This has to do with an increase of litigation costs and investigations of the HMRC. For example, the HMRC introduced in 2009 the code, this encourages banks and organisations that provide banking services in the UK to be transparent and open in their own tax affairs, and not to promote or facilitate tax avoidance by others. Meaning that there is an increased focus by HMRC for transparent tax transactions and to cut off any tax avoidance strategies.
To summarize, international tax issues has never been as high on the political agenda as they are today. The reports show that multinational enterprises try to reduce their tax burden and that there is an increased focus by national tax departments on avoiding taxes by increasing tax regulations.
2.2 Tax avoidance Benefits vs Costs
There are a few studies that investigate whether tax avoidance benefits exceeds the costs of this type of activity. Engaging in tax avoidance strategies can be profitable for companies if the benefits exceed the costs. Benefits of avoiding taxes is straightforward, they reduce tax liabilities which has an positive impact on the cash flow and can also increase net income. First, lowering the ETR will increase reported after-tax earnings. As a result, this will increase the stock price and market value of the firm since earnings and stock price share content (Ball and Brown,
For example, debt contracts and debt covenants use earnings to determine the likelihood of default. Graham and Tucker (2006) argue that by reducing the tax rate the earnings per share (EPS) increases which also increased market capitalization of the firm. Moreover, it reduces expected bankruptcy costs, enhances credit quality, reduces the risk of covenant violation, and reduces the cost of debt. In addition, they also found that credit ratings improve one notch relative to matched firms, most likely because of falling debt ratios (p.566). Third, earnings and after-tax earnings are also used to evaluate management and are used as a performance measure in management bonus plans. Therefore, performance measures motivate managers to make use of tax avoidance strategies and consequently the ETR is an important benchmark that is compared across firms according to Graham et al. (2011). Fourth, executives can increase performance with little effort by hiring third parties to lower ETRs, as long as the benefits from avoiding taxes exceeds the costs. Bankman (2007) acknowledges this and he argues that the dollar benefits far outweigh the costs due to the low probability of detection by the tax authority. Finally, Slemrod (2004) argues that there are three parties involved in avoiding taxes; the shareholders, executive management and the tax authority. Avoiding taxes benefits both the shareholders and executive management at the cost of the tax authority (p.894). This fact incentivises CEOs and CFOs to conduct in tax avoidance.
Although engaging in tax avoidance has its advantages, it can also have disadvantages for the company. Tax aggressiveness can impose high costs on a firm. First, the firm has to invest in resources to develop and maintain the tax strategy, this includes fees paid to tax experts, accountants and attorneys and hours spend of internal employees to plan and manage the tax planning. On the other hand, Wilson (2008) found that the average tax avoidance transactions generates savings of $375.5 million. Second, cost can increase if the tax authorities successfully confront the tax position of the company. For example, once the company is targeted by the tax authority for avoiding taxes, its policy allows it to expand the scope of information that it requests from the firm, which can lead to the discovery of other aggressive tax avoidance (Gallemore et al., 2014, p.1105). However, Bankman (2007) argues that the probability of detection by the tax authority is low. Moreover, when the company avoids taxes within the legal boundaries than the tax authority is not able to fine the company. Third, tax avoidance can have a serious impact on the reputation of the company and distrust of the public according to Graham et al. (2014) and Hardeck and Hertl (2014). For instance, expected penalties decrease the expectations of future cash flows and bad reputation might affect earnings. Moreover, Hanlon and Slemrod (2009) studied the relationship between tax aggressiveness and stock price. They found a stock decline when investors are confronted with corporate tax aggressiveness news. On the other hand, Gallemore et al. (2014) used a different research design and found no
evidence that firms or their top executives face significant reputational costs from tax shelter involvement. They do find a decline in stock price around the tax shelter revelations but this decline fully reverses within 30 days (p.1127). However, tax avoidance is considered to be a risky project and the outcome maybe uncertain (Badertscher et al. 2013).
To summarize, prior research shows that tax avoidance benefits are able to exceed the costs. Therefore, it can be a strategy to increase or to maximize company value.
2.3 Performance measurement of corporate tax departments
Corporate administrative departments were before the 1990’s evaluated based on the cost model, including the tax departments. However, after the 1990’s the evaluation of tax departments gradually started to change from a cost model towards a profit-centre. Meaning that the department is evaluated based on how much profit it generates and how much it contributes to the firm’s income.
As a result, more and more performance measures are set by management to evaluate tax departments. Consequently, tax departments use their knowledge and skills to lower the ETRs as much as possible. Before the 1990’s they were evaluated based on minimizing costs of tax compliance. But after that, the evaluation of tax departments shifted to profitability (Robinson et al., 2010, p.1035). Moreover, they found that the choice to manage and measure the tax department as a profit centre is positively associated with the degree of importance a firm places in taxes as a mechanism to meet earnings expectations and with the percentage of the total tax and the percentage of the total tax budget that the firm allocates to tax planning as opposed to compliance (p.1055). In addition, firms that use profit centres to evaluate tax departments are associated with lower ETRs. According to Robinson et al’s model they observe a reduction in ETRs of 1.5 percent for every 10 percent increase in the likelihood that a firm would adopt a profit centre measurement model (p.1058). This shift means that tax departments are incentivized to produce profit which is consistent with the survey of Hollingsworth (as cited in Slemrod, 2004). Results show that the value added by tax departments was most cited and went up from 75 percent to 86 percent. Moreover, the effective tax rate was cited as a measure to evaluate performance by 58 percent of respondents, up from 48 percent (p.885). Thus companies more and more try to reduce the tax burden upon the company.
2.4 Equity compensation and tax avoidance
Overall equity incentives motivate managers to increase the risk profile of the company and consequently take on riskier net present value projects since this increases stock price and stock price volatility. Prior research has established an association between equity incentives and increase in managerial risk taking. Coles et al. (2006) used Execucomp to extract data about salary, bonus and compensation for the period 1992 – 2002. They calculated the dependability of the CEOs compensation to the stock price, whereby they defined delta as the change in dollar value of the executive’s wealth for a one percent change in stock price (pp.434-435). They found that stock price volatility is significant and positively associated with R&D expenditures, firm focus and leverage which are considered to be risky projects. Therefore, equity incentives suggests that managers will make use of tax avoidance strategies.
Furthermore, Guay (1999) examines the relationship between CEO wealth and equity risk. He used compensation and CEO stock wealth data. He found that stock-return volatility is positively associated with CEOs wealth (p.65). The payoff in management equity compensation is large enough to influence the investment decisions of the CEO and the overall risk profile of the company. Guay (1999) his equity risk incentives theory indicates that managers who receive an proper amount of equity incentives take on more riskier projects and therefore might be more willing to invest in tax avoidance practices. Tax avoidance projects should increase stock return volatility, since this increases the uncertainty of the future tax outcomes. Investors will notice the increase of uncertainty and will lead to higher stock prices, higher valuations of stock options and greater managerial wealth.
Moreover, Ju et al. (2013) examines the link between corporate risk taking and managerial option compensation. They explain that option compensation can provide the right incentives to determine corporate risk. However, they also describe that this type of compensation can create too much or too little corporate risk taking, depending on the risk aversion of the manager. When provided with too much compensation risk averse executives might want to lessen corporate risk to lessen their portfolio risk. When provided with too little compensation executives do not have enough incentives to engage in riskier project. On the other hand, according to the findings of their study, if provided with the right amount of option compensation the executive will encourage risk policies that will increase the shareholders wealth and his own personal wealth (pp.13 -15).
When linking managerial compensation to after tax performance measures this could lead to lower ETRs. This is acknowledged by Philips (2003). His study investigates the role of compensation based incentives of CEOs and Business unit (BU) managers using after-tax based
performance measures and whether this leads to corporate tax-planning strategies. Survey data is used to determine whether BU managers and CEOs are compensated based on after-tax performance measures. Philips (2003) found that when the BU manager is compensated on an after-tax basis the CEO is compensated on a pre-tax measure. Moreover, when the CEO is compensated on an after-tax measure the BU manager is more likely to also be evaluated on an after-tax basis (p.860). Suggesting that when the shareholders include after-tax performance measures in the CEOs compensation package this measure will also be included in the BU manager compensation package. Furthermore, Philips (2003) argues that CEOs have more incentives to focus on after-tax than pre-tax measures due to the shareholders who expect a certain profitability level (p.862). However, he did not find any association between CEO compensation and lower ETR’s indicating that CEOs are willing to implicitly avoid taxes through their BU managers.
On the contrary, Gaertner (2014) examined the relationship between CEO after-tax incentives and corporate tax avoidance. He used a larger dataset of 309 samples instead of the 209 samples Philips (2003) used. Gaertner (2014) found that when CEOs have after-tax incentives it will decrease the ETR with 5.7 percent (p.1091). Moreover, he found an association between CEO compensation and after-tax incentives due to the additional risk that is taken to lower the ETR. Given these results it might be that the equity incentives of the CEO and his dependability on the stock price, is associated with lower ETRs.
To sum, equity incentives induces management to engage in more risky projects. This might suggest that equity incentives in executive compensation packages might induce executives to lower the ETRs of the company.
2.5 Hypothesis
Based on the elaboration in the current chapter, two hypotheses are developed to test. Most literature and theory predict that equity incentives induces riskier strategies, more tax departments are evaluated on the effective tax rate and tax avoidance strategies are profitable. Therefore, it is possible to say that equity incentives encourage executive management to avoid taxes. Consequently, the two hypothesis are;
H1: CEO equity incentives are negatively associated with a firm’s ETR H2: CFO equity incentives are negatively associated with a firm’s ETR
3 Research Method
In this paragraph the research method is explained and is divided into 4 sections. Section 3.1 describes the process of the sample selection. Section 3.2 illustrates the measurement of corporate tax avoidance. Section 3.3 clarifies the measurement of equity incentives and the accompanying calculations. And finally, section 3.4 points out the model that will be used for the empirical tests.
3.1 Sample selection
To investigate the research question, archival (database) research is used as empirical approach. I obtain CEO and CFO compensation data from Compustat’s Execucomp Annual compensation database. For the financial statement data I used Compustat’s Fundamental Annual database. The sample consists out of companies from the Standard & Poor’s 1500 index from the period 2010 to 2014. I chose the S&P 1500 index since this covers most of the market capitalization in the U.S. With a sample period of 4 years, the total sample size consists of 45.613 firm-year observations.
For the data to be included in my sample, I eliminated all the observations related to financial institutions (SIC codes 6000-6999) and utility firms (SIC codes 4900-4999) since these institutions have specific regulatory requirements to comply with. Moreover, I delete all the firms who have a negative pre-tax income because I expect that the association between equity incentives and tax avoidance is weak for firms that are not profitable (Rego & Wilson, 2012). Furthermore, all the firm-year observations with missing data to estimate corporate tax avoidance, GAAP ETR, CURRENT ETR, Long-Run CASH ETR and CASH ETR, and data to calculate the control variables are excluded. After that, I delete the ETRs which are negative, are equal to 1 or higher and I further winsorize the control variables at 1st and 99th percentile
following prior studies (Dyreng et al. 2010; Hope et al. 2013). When merging the data for CEO and CFO compensation to the annual financial statement data, I arrive at a dataset of 2.156 firm-year observations for GAAP ETR, for CURRENT ETR the dataset consists of 1.970 firm-firm-year observations and for CASH ETR and Long-Run CASH ETR the dataset consists of 2.120 and 388 firm-year observations.1
3.2 Measurement of corporate tax avoidance
Reviews of prior literature established that corporate tax avoidance can be measured in several ways and using one proxy does not give a good indication of tax avoidance planning. Moreover, not all measures are appropriate in capturing tax avoidance, however, researchers have already accept that every single measure has its own limitations (Hanlon & Heitzman, 2010). The measurement of tax avoidance is divided into whether the firm avoid taxes and the level of tax aggressiveness and tax planning. I selected the GAAP ETR as one of my proxies to capture tax avoidance because in this report tax avoidance is defined as encompassing anything that reduces the firm’s taxes relative to pre-tax accounting income (Dyreng et al., 2008). First, the GAAP ETR is the tax rate which firms are required to disclose in the notes of their financial statements and is the rate that affects accounting earnings. Second, compensation contracts could be affected by the GAAP ETR because increasing after-tax earnings can increase stock price and thus affect compensation. Moreover, GAAP ETR is an important benchmark to compare firms with since this reduces expected bankruptcy costs, enhances credit quality, reduces the risk of covenant violation and reduces the cost of debt (Graham & Tucker, 2006). GAAP ETR is the ratio of total income tax expense divided by the total pre-tax accounting income minus special items, thus the GAAP ETR for a given firm i for year t is given by:
GAAP ETR𝑖 𝑡 = 𝑊𝑜𝑟𝑙𝑑𝑤𝑖𝑑𝑒 𝑡𝑜𝑡𝑎𝑙 𝑖𝑛𝑐𝑜𝑚𝑒 𝑡𝑎𝑥 𝑒𝑥𝑝𝑒𝑛𝑠𝑒𝑖 𝑡 Worldwide total pre − tax accounting income 𝑖 𝑡
This measure reflects non-conforming tax avoidance and is computable by jurisdiction. However, several items that are not tax planning strategies could have an effect on the GAAP_ETR. For example, changes in valuation or changes in the tax contingency reserve might have an impact on the GAAP ETR and tax deferral strategies are not reflected, which is according to Dyreng et al. (2008) a frequently used activity by companies.
To overcome the limitations of GAAP ETR I will include CURRENT ETR as a second tax avoidance measure. Despite the fact that this measure is not often used by researchers it has some advantages over GAAP ETR and is defined as the current tax expense per dollar of pre-tax book income and reflects tax deferral strategies. CURRENT ETR reflects non-conforming tax avoidance strategies and tax deferral strategies(Hanlon & Heitzman, 2010). However, it does not have an effect on accounting earnings nor does it reflect conforming avoidance. This ETR is
calculated by dividing current income tax expense minus deferred taxes with pre-tax accounting income minus special items, thus the CURRENT ETR for a given i for year t is given by:
CURRENT ETR = 𝑊𝑜𝑟𝑙𝑑𝑤𝑖𝑑𝑒 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑡𝑜𝑡𝑎𝑙 𝑖𝑛𝑐𝑜𝑚𝑒 𝑡𝑎𝑥 𝑒𝑥𝑝𝑒𝑛𝑠𝑒𝑖 𝑡 Worldwide total pre − tax accounting income𝑖 𝑡
To overcome the limitations of both the GAAP ETR and CURRENT ETR I will use CASH ETR as a third tax avoidance measure. This measure is extensively used in the tax avoidance literature (Dyreng et al., 2010; Rego & Wilson, 2008; Armstrong, 2015; Hope et al., 2013). The CASH ETR has several advantages. It reflects all transactions that have any effect on the firm’s explicit tax liability, this includes tax positions with certain and uncertain outcomes. CASH ETR captures tax avoidance that generates temporary book–tax differences, it is not based on a set of broad firm characteristics and reflects deferral tax strategies and non-conforming tax strategies (Hanlon and Heitzman, 2010). Moreover, firms that do aggressively accelerate their expenses or defer income for tax purposes will be reflected in a lower CASH ETR and is therefore more volatile than GAAP ETR. Finally, this measure takes into account the tax benefits of employee stock options (Dyreng et al., 2005, p.12). This tax avoidance proxy is calculated by dividing cash taxes paid with total pre-tax accounting income minus special items. CASH ETR for a firm i for year t is given by:
CASH ETR𝑖 𝑡 = Worldwide cash taxes paid𝑖 𝑡
Worldwide total pre − tax accounting income𝑖 𝑡
The final tax avoidance measure that I will use is the Long-Run CASH ETR, this measure will be used to measure the ETR over a five-year period and better reflects the ETRs over a long period of time because it is less volatile. Moreover, CASH ETR measured over a short time period includes payments and refunds from the IRS and other tax institutions of settling tax agreements that arose years ago. The Long-Run CASH ETR overcomes these inconsistencies and therefore better reflects tax avoidance strategies (Dyreng et al., 2008). This tax avoidance proxy is based on a five-year period and is calculated by dividing the sum of cash taxes paid with the sum of total pre-tax accounting income minus the sum of special items. This measure for a given firm i for year t is given by:
𝐿𝑜𝑛𝑔 − 𝑅𝑢𝑛 CASH ETR𝑖 𝑡 = ∑ 𝑊𝑜𝑟𝑙𝑑𝑤𝑖𝑑𝑒 𝑐𝑎𝑠ℎ 𝑡𝑎𝑥𝑒𝑠 𝑝𝑎𝑖𝑑𝑖 𝑡
∑ 𝑊𝑜𝑟𝑙𝑑𝑤𝑖𝑑𝑒 𝑡𝑜𝑡𝑎𝑙 𝑝𝑟𝑒 − 𝑡𝑎𝑥 𝑎𝑐𝑐𝑜𝑢𝑛𝑡𝑖𝑛𝑔 𝑖𝑛𝑐𝑜𝑚𝑒𝑖 𝑡
Even though, CASH ETR and Long-Run CASH ETR captures more tax avoidance strategies than GAAP ETR and CURRENT ETR, it still has disadvantages since tax avoidance cannot be measured perfectly. For example, CASH ETR and Long-Run CASH ETR reflects all transactions that have an effect on the firm’s explicit tax liability. Moreover, the transactions do not distinguish real tax activities that are tax favoured and real tax activities that are specifically taken to lower taxes. Next, it might be that some may interpret low ETRs as the result of upward earnings management if taxes remain constant but if taxes remain constant as the firm manages earnings upwards then the firm is still avoiding taxes (Hanlon & Heitzman, 2010, p. 141). Moreover, these measures do not impact accounting earnings. However, in this article tax avoidance is considered to encompass anything that reduces the firm’s taxes relative to pre-tax accounting income.
3.3 Measurement of equity incentives
EQ_INC are the equity incentives of the CFO (CFO EQ_INC) and CEO (CEO EQ_INC) and consist out of option grants in the current period, un-exercisable options, exercisable options, restricted stock and stock ownership. The proxy for equity incentives is calculated according to the Bergstresser and Phillipon’s (2006) model. The reason for choosing this model is because it is well used and referenced by researchers in calculating the sensitivity of equity incentives (Kim et al. 2011; Jiang et al. 2010; Kim et al. 2015; Coles et al. 2006). The model focuses on the sensitivity of the equity incentives. By sensitivity we mean the sensitivity of the CEOs and CFOs stock and stock option holdings to changes in stock price. More specifically, it measures the dollar change in the value of a CEO’s stock and stock option holdings arising from a one percentage point increase in the firm’s stock price (Bergstresser & Phillipon, 2006, p.519). My expectation is that when sensitivity is high, executives are more likely to avoid taxes. The calculation of the equity incentives model is as follows:
𝑂𝑁𝐸𝑃𝐶𝑇 𝑖 𝑡 = 0.01 ∗ 𝑃𝑅𝐼𝐶𝐸𝑖 𝑡 ∗ (𝑆𝐻𝐴𝑅𝐸𝑆𝑖 𝑡 + 𝑂𝑃𝑇𝐼𝑂𝑁𝑆𝑖 𝑡)
owns), PRICE is the company’s share price, SHARES is the number of shares held by the CEO and OPTIONS is the number of options held by the CEO. Whereby, the option value is calculated by summing up the estimated value of in-the-money unexercised exercisable options (OPT_UNEX_EXER_EST_VAL) and the estimated value of in-the-money un-exercisable options (OPT_UNEX_UNEXER_EST_VAL). After that, ONEPCT is used to calculate the EQ_INC which reflects the portion of the CEOs total annual compensation from the firm stemming from a one percentage point increase in the firm’s stock price (Bergstresser & Philippon, 2006, p.520).
𝐸𝑄_𝐼𝑁𝐶𝑖 𝑡 =
ONEPCT𝑖 𝑡
ONEPCT𝑖 𝑡+ 𝑆𝐴𝐿𝐴𝑅𝑌𝑖 𝑡+ 𝐵𝑂𝑁𝑈𝑆𝑖 𝑡
The measure is based on the assumption that the delta of the option in the CEOs and CFOs portfolio is one where a dollar increase in the price of the firm’s share translates one for one to the value of an option (Bergstresser & Philippon, 2006, p. 520).
3.4 Modelling tax avoidance and equity incentives
The hypothesis predicts that larger CEO and CFO equity incentives and therefore sensitivity is associated with corporate tax avoidance. The models that I will be using are derived from Hope et al (2013) and are stated below;
𝑇𝐴𝑋𝑖 𝑡 = 𝛽0+ 𝛽1𝐶𝐸𝑂 𝐸𝑄_𝐼𝑁𝐶𝑖 𝑡+ 𝛽2 𝑆𝐼𝑍𝐸𝑖 𝑡+ 𝛽3 𝐿𝐸𝑉𝑖 𝑡+ 𝛽4𝑅𝑂𝐴𝑖 𝑡 + 𝛽5𝑁𝑂𝐿𝑖 𝑡+
𝛽6𝑃𝑃𝐸𝑖 𝑡+ 𝛽7𝑅&𝐷𝑖 𝑡+ 𝛽8𝐼𝑁𝑇𝐴𝑁𝐺𝑖 𝑡 + 𝛽9𝐹𝑂𝑅𝐼𝑁𝐶𝑖 𝑡 (1)
𝑇𝐴𝑋𝑖 𝑡 = 𝛽0+ 𝛽1𝐶𝐹𝑂 𝐸𝑄_𝐼𝑁𝐶𝑖 𝑡+ 𝛽2 𝑆𝐼𝑍𝐸𝑖 𝑡+ 𝛽3 𝐿𝐸𝑉𝑖 𝑡+ 𝛽4𝑅𝑂𝐴𝑖 𝑡 + 𝛽5𝑁𝑂𝐿𝑖 𝑡+ 𝛽6𝑃𝑃𝐸𝑖 𝑡+ 𝛽7𝑅&𝐷𝑖 𝑡+ 𝛽8𝐼𝑁𝑇𝐴𝑁𝐺𝑖 𝑡 + 𝛽9𝐹𝑂𝑅𝐼𝑁𝐶𝑖 𝑡 (2)
Equation (1) models tax avoidance (GAAP ETR, CURRENT ETR, Long-Run CASH ETR and CASH ETR) as a function of CEO EQ_INC as the portion of the CEOs total annual compensation from the firm stemming from a one percentage point increase in the firm’s stock price. This model will be used to test hypothesis 1 and would support my hypothesis that CEO
EQ_INC would reduce the GAAP ETR, CURRENT ETR, Long-Run CASH ETR and CASH ETR. Equation (2) is used to test hypothesis 2 and substitutes CEO EQ_INC for CFO EQ_INC compared to model (1) and is the portion of the CFOs total annual compensation from the firm stemming from a one percentage point increase in the firm’s stock price. I expect that CFO EQ_INC is associated with tax avoidance and therefore reduce the ETR.
However, to make the model trustworthy, I have to include control variables. Previous literature showed that some factors have an association with corporate tax avoidance (although researchers do not always agree whether there is a positive or negative association). Therefore, I include the following control variables. First, according to Zimmerman (1983) larger firms face greater political costs in the form of higher tax payments. Mills et al. 2012 argues that larger firms are more willing to engage in tax avoidance activities. For that reason, size (𝑆𝐼𝑍𝐸𝑖 𝑡) is included
in the model. Second, Leverage( 𝐿𝐸𝑉𝑖 𝑡 ) is added to the control variables because organizations
which use debt financing may not need to engage in tax avoidance strategies due to the deductible interest expense (Desai & Dharmapala, 2009). Third, Chen et al. 2012 argues that organizations with relatively high profits could have more incentives to engage in tax avoidance activities and foreign income is treated differently for reporting for tax purposes. As a result, the return on assets (𝑅𝑂𝐴𝑖 𝑡) and foreign income (𝐹𝑂𝑅𝐼𝑁𝐶𝑖 𝑡) are added to the model. Fourth, the
presence of a net operating loss carry forward (𝑁𝑂𝐿𝑖 𝑡) because loss carry forwards will give
some tax benefits to the organization (Hope et al., 2013). NOL is a dummy variable which is equal to 1 if the firm has a positive amount of net operating loss carry forward and when the firm has no amount it is equal to 0. Fifth, different depreciation expense methods for tax financial reporting purposes are more likely to be used by capital intensive firms. Therefore, property, plant and equipment (𝑃𝑃𝐸𝑖 𝑡) is also included in the model. Furthermore, research and
development expenses (𝑅&𝐷𝑖 𝑡) and intangible assets (𝐼𝑁𝑇𝐴𝑁𝐺𝑖 𝑡) are included in the control
variables because of the different book and tax treatments of intangible assets according to Rego & Wilson (2012).
I expect that all of the control variables within this model are negatively associated aside from the control variable LEV (leverage) which is expected to be positively associated with Long-Run CASH ETR, CASH ETR, CURRENT ETR and GAAP ETR.2
4 Empirical results
In this paragraph the results of the empirical tests are shown. The paragraph is divided into 3 sections. Section 4.1 is dedicated to GAAP ETR and CEO and CFO equity incentives. Section 4.2 is devoted to CURRENT ETR and CEO and CFO equity incentives. Section 4,3 is committed to the CASH ETR and Long-Run CASH ETR and CEO and CFO equity incentives. The sections start with descriptive statistics and the Pearson correlation matrix. After that, the results of the regression of corporate tax avoidance are shown and explained.
4.1 Corporate Tax Avoidance GAAP ETR 4.1.1 Descriptive statistics
One of the proxies that is used, as mentioned in paragraph 3.2, is GAAP ETR. The descriptive statistics on GAAP ETR (dependent variable), CEO EQ_INC and CFO EQ_INC (independent variables/variables of interest) and the control variables are provided in table 1.
Table 1 - Descriptive statistics for GAAP ETR tests
Variables # Obs Mean
Std.
Deviation Min. Max.
GAAP ETR 2156 0,333 0,109 0,002 0,997 CEO EQ_INC 2156 0,309 0,232 0 1,000 CFO EQ_INC 2156 0,141 0,134 0 1,000 SIZE 2156 3,461 0,685 1,747 5,267 LEV 2156 0,161 0,138 0 0,565 ROA 2156 0,122 0,067 0,003 0,356 NOL 2156 0,526 0,499 0 1 PPE 2156 0,229 0,185 0 0,766 R&D 2156 0,017 0,025 0 0,107 INTAN 2156 0,235 0,192 0 0,739 FORINC 2156 0,031 0,035 0 0,147
The variable for tax avoidance, GAAP ETR, has 2,156 firm-year observations from which all are related to CEO EQ_INC and CFO EQ_INC which represents the portion of the annual compensation stemming from a one percent increase in share price.
The GAAP ETR has a mean value of 0,333 and is slightly higher than the reported mean from Dyreng et al. (2010). Meaning that on average a firm’s tax expense is 33,3% (0,333 dollars ) per 1 dollar of pre-tax book income. CEO EQ_INC and CFO EQ_INC have respectively mean
values of 0,309 and 0,141 suggesting that when the share price of the firm increases with 1% the value of their annual compensation rises on average by 30,9% for the CEO and 14,1% for the CFO. This gap between CEO and CFO EQ_INC is consistent with Jiang et al. (2012) who argues that CEO equity incentives are much larger than that of the CFO. Further, the descriptive statistics for the control variables are: a mean value for SIZE of 3,461, a mean value of 0,161 for LEV meaning that each firm is on average financed for 16,1% with debt, a mean value of 0,122 for ROA meaning that on average the return on assets is 12,2%, a mean value 0,526 for NOL which means that approximately 52,6% of the firms have a net operating loss carried forward, a mean value of 0,229 for PPE suggesting that on average the firm’s assets consist for 22,9% out of the value of their property, plant and equipment and for R&D and INTAN mean values of 0,017 and 0,235. Therefore, on average firms spend 1,7% of their value of assets on R&D and the total asset value of each firm covers 23,5% out of intangibles. Finally, the control variable, FORINC has a mean value of 0,031 meaning that on average 3,1% of the firm’s total income comes from outside the U.S.
4.1.2 Pearson Correlation Matrix
The Pearson Correlation Matrix measures the degree of correlation between variables for the GAAP ETR, CEO EQ_INC and CFO EQ_INC and the control variables are shown in table 2.
In table 2 half of all the variables are correlated at the 1% level. The variables of interest in the Pearson Correlation Matrix are the independent variable GAAP ETR and the dependent variables, CEO EQ_INC and CFO EQ_INC. As expected, CEO EQ_INC is significant and negatively correlated (-0.062) at the 1% level with the independent variable GAAP ETR and supports H1. Moreover, CFO EQ_INC (-0.090) is also significant and negatively correlated with GAAP ETR and thus supports H2. When looking at the control variables and their correlation with the variable GAAP ETR, I find significant correlations for; SIZE (-0,159) suggesting that larger firms have more political costs and therefore more incentives to lower their tax burden, R&D (-0,221) meaning that firms with more R&D expenses have different book-tax treatments and therefore lower the GAAP ETR and FORINC (-0,324), implying that firms with foreign income have more opportunities to avoid taxes. On the contrary, PPE (0,093) is positive and significant which is also against my expectations.
The correlations between the independent and control variables are all in the range of 0,450 and -0,324. To make sure there is no multicollinearity I tested for the variance inflation
Table 2 Pearson correlation matrix for GAAP ETR test (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (1) GAAP ETR 1 (2) CEO EQ_INC -0,062 1 (3) CFO EQ_INC -0,090 0,361 1 (4) SIZE -0,159 0,099 0,208 1 (5) LEV -0,025 -0,086 0,058 0,377 1 (6) ROA -0,170 0,204 0,145 -0,193 -0,269 1 (7) NOL -0,046 -0,072 -0,009 0,042 0,101 -0,164 1 (8) PPE 0,093 0,035 -0,077 0,012 0,139 0,020 -0,159 1 (9) R&D -0,221 0,080 0,093 0,041 -0,119 0,048 0,165 -0,272 1 (10) INTANG -0,033 -0,035 0,070 0,185 0,299 -0,162 0,192 0,450 0,064 1 (11) FORINC -0,324 0,033 0,072 0,270 0,010 0,082 0,120 -0,145 0,382 -0,005 1
Significance at the 1% level is indicated by the bold correlations All variables are defined as in appendix E
factor (VIF). If VIF is above 10 then multicollinearity is present. However, in this case the maximum number of VIF is 1,521.
4.1.3 GAAP ETR test – Hypothesis 1 and 2
The regression analysis which is reflected in table 3 shows the results of the GAAP ETR test for hypothesis 1 and 2. Which are the following; (H1) CEO equity incentives are negatively associated with a firm’s ETR and (H2) CFO equity incentives are negatively associated with a firm’s ETR. At this stage, only for the first hypothesis, second hypothesis and GAAP ETR an explanation will be given.
The regression analysis , as shown in table 3, where GAAP ETR is the dependent variable and EQ_INC is defined as the CEO EQ_INC, shows that the coefficient is negative (0,018) and significant (t-statistic= -1,853) and thus supports H1. For the regression where EQ_INC is defined as the CFO EQ_INC, the coefficient is negative and significant (-0,035; t-statistic= -2,052) and therefore supports H2. Both results can be interpreted to mean that when the total equity incentives compensation for CEO (CFO) increases with 1% the GAAP ETR decreases with 1,8% (3,5%). The results suggest that equity incentives induce the CEO and CFO to engage in riskier projects and thus tax avoidance strategies to lower the firm’s GAAP ETR to increase after-tax earnings and future tax outcomes of the company which will lead to higher stock – and option prices to increase his own personal wealth. Moreover, the results suggests that CFOs might have more influence on the firm’s tax avoidance behaviour than the CEOs. This is consistent with the findings of Dyreng et al. (2008) who found that the effects of executives have a significant impact on the GAAP ETR.
In extension to the previous, also shown in table 3 are the results of the control variables for the regression where GAAP ETR is the dependent variable. Derived from the table can be seen that both from the CEO EQ_INC and CFO EQ_INC regressions, three of the coefficients are significant at the 1% level, FORINC, SIZE and R&D, which are in line with the expected relationship with the dependent variable. For example, the coefficient of FORINC (under CEO EQ_INC as the independent variable) is -0,806 and significant (t-statistic= -11,021) meaning that companies who have more foreign income (FORINC) have more opportunities to engage in tax avoidance strategies. On the other hand, the non-significant coefficient of INTAN, ROA and NOL are positive (while negative coefficients were expected) and therefore do not meet my expectations. However, the outcome, a positive number, is consistent with Rego & Wilson (2012) and Desai & Dharmapala (2009).
4.2 Corporate Tax avoidance CURRENT ETR
The second proxy that is used to measure corporate tax avoidance is CURRENT ETR. The descriptive statistics on CURRENT ETR (dependent variable), CEO EQ_INC and CFO EQ_INC (independent variables/variables of interest) are stated in table 4.
Table 3 -Results for GAAP ETR
CEO EQ_INC CFO EQ_INC
Variable Exp. Sign Coef. T-stat. Prob. Coef. T.stat. Prob.
EQ_INC - - 0,018* -1,853 0,064 - 0,035** -2,052 0,040 SIZE - - 0,012*** -3,266 0,001 - 0,012*** -3,087 0,002 LEV + - 0,014 -0,722 0,471 - 0,009 -0,491 0,623 ROA - 0,000 -0,010 0,992 0,001 0,023 0,982 NOL - 0,002 0,475 0,635 0,002 0,527 0,598 PPE - 0,020 1,384 0,166 0,017 1,154 0,249 R&D - - 0,480*** -4,883 0,000 - 0,481*** -4,900 0,000 INTANG - 0,003 0,190 0,849 - 0,002 -0,170 0,865 FORINC - - 0,806*** -11,021 0,000 - 0,806*** -11,022 0,000 Adjusted R² 0,122 0,122 F-statistic 34,168 34,231
Significance at the 1%, 5%, and 10% level is indicated by ***, **, and *, respectively. All variables are defined in Appendix E
GAAP ETR is winsorized to the range [0,1]. The control variables are winsorized at the 1st and 99th percentiles.
Table 4 - Descriptive statistics for Current ETR tests
Variables # Obs Mean
Std.
Deviation Min. Max.
CURRENT ETR 1970 0,372 0,173 0,001 0,992 CEO EQ_INC 1970 0,322 0,235 0 1,000 CFO EQ_INC 1970 0,189 0,193 0 1,000 SIZE 1970 3,457 0,677 1,747 5,236 LEV 1970 0,157 0,138 0 0,565 ROA 1970 0,128 0,065 0,003 0,356 NOL 1970 0,506 0,500 0 1 PPE 1970 0,235 0,182 0,009 0,772 RD 1970 0,016 0,024 0 0,099 INTAN 1970 0,231 0,189 0 0,720 FORINC 1970 0,028 0,034 0 0,144
The variable for tax avoidance, CURRENT ETR, has 1,970 firm-year observations from which all are related to CEO EQ_INC and CFO EQ_INC which represents the portion of the annual compensation stemming from a one percent increase in share price.
As can be seen from table 4 is that CURRENT ETR has a mean value of 0,372 which is slightly higher than the mean value of Hope et al. (2013) meaning that entities on average have current tax expenses of 37,2% (0,372 dollars) per 1 dollar of pre-tax book income. The variables CEO EQ_INC and CFO EQ_INC have a mean value of 0,322 and respectively 0,189 which suggests that when the share price increases with 1% the value of their annual compensation rises on average with 32,2% for the CEO and 18,9% for the CFO. This difference in CEO and CFO equity incentives is consistent with the study of Aggarwal and Samwick (2003). They argue due to their different responsibilities within the firm, the CEO gets the bigger part of total equity compensation.
For the control variables, the following can be noticed; SIZE has a mean value of 3,457, LEV has a mean value of 0,157 meaning that on average each entity is financed with debt for 15,7% of their total assets, ROA has a mean value of 0,128 saying that on average each firm had a return on their assets of 12,8%, NOL has a mean value of 0,506 suggesting that 50,6% of all the entities have a net carry loss forward, PPE with a mean value of 0,235 meaning that on average 23,5% of their total assets consist out of property, plant and equipment, R&D has a mean value of 0,016 saying that on average each firm spends 1,6% of their value of total assets on R&D, and INTAN and FORINC with a mean value of 0,231 and respectively 0,028 suggesting that on average the total assets consist for 23,1% of intangible assets and that on average 2,8% of each firm’s income comes from outside the U.S.
4.3 Pearson Correlation Matrix
The Pearson Correlation Matrix measures the degree of correlation between variables for the CURRENT ETR, CEO EQ_INC and CFO EQ_INC and the control variables and are shown in table 5.
Table 5 Pearson correlation matrix for CURRENT ETR test (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (1) CURRENT ETR 1 (2) CEO EQ_INC -0,085 1 (3) CFO EQ_INC -0,082 0,042 1 (4) SIZE -0,120 0,107 0,084 1 (5) LEV 0,040 -0,105 0,066 0,406 1 (6) ROA -0,147 0,227 -0,022 -0,206 -0,302 1 (7) NOL 0,023 -0,066 0,051 0,089 0,127 -0,172 1 (8) PPE 0,174 0,036 -0,123 0,033 0,139 0,007 -0,164 1 (9) R&D -0,222 0,092 0,043 0,034 -0,133 0,044 0,125 -0,271 1 (10) INTANG -0,032 -0,057 0,211 0,187 0,296 -0,192 0,165 -0,471 0,028 1 (11) FORINC -0,337 0,035 0,029 0,258 0,005 0,071 0,115 -0,135 0,363 -0,310 1
Significance at the 1% level is indicated by the bold correlations All variables are defined as in appendix E
In table 5, more than half of all the correlations are significant at the 1% level. The most important independent variables are CEO EQ_INC and CFO EQ_INC. As hypothesized, both definitions of the variable CEO EQ_INC and CFO EQ_INC, are negatively (-0,085; -0,082) and significantly correlated with the dependent variable CURRENT ETR. When looking at the control variables, 5 of them are significantly correlated with the dependent variable. For example, ROA is negative and significantly correlated with CURRENT ETR suggesting that firms with higher return on assets have more incentives to lower the ETR. However, PPE is significant and positively correlated (0,174) and is not in line with my expectation.
Furthermore, the correlation of the control variables are on average not very high and lies between -0,471 and 0,406 which reduces the risk of multicollinearity. To be certain I tested for variance inflation factor (VIF). The variable with the highest VIF score is INTANG with 1,653 meaning that there is no multicollinearity since this is only the case when the score is above 10.
4.4 CURRENT ETR tests – hypothesis 1 and 2
The regression analysis is reflected in table 6 and shows the results of the CURRENT ETR test for hypothesis 1 and 2. Which are the following; (H1) CEO equity incentives are negatively associated with a firm’s ETR and (H2) CFO equity incentives are negatively associated with a firm’s ETR.
The regression analysis , as shown in table 6, where CURRENT ETR is the dependent variable and EQ_INC is defined as the CEO EQ_INC, shows that the coefficient is negative (-0,019) however insignificant (t-statistic= -1,441) and does not support H1. With respect to the regression where EQ_INC is defined as CFO EQ_INC the coefficient is negative and significant (-0,043, t-statistic= -2,713). This result can be interpreted to mean that when total equity incentive compensation increases with 1% the CURRENT ETR decreases with 4,3%. This supports H2 and is consistent with the idea that equity incentives induce the CFO to engage in riskier projects such as tax avoidance to increase the uncertainty of the tax outcomes which can increase stock prices and consequently increase his or her personal wealth. Moreover, the result suggests that CFOs have more influence on the CURRENT ETR than the CEO. On the other hand, how could it be that CFO EQ_INC is negatively associated to CURRENT ETR and CEO EQ_INC not? And why is this different to the GAAP ETR? (where CEO EQ_INC and CFO EQ_INC are associated) An possible explanation could be that CEOs are not being compensated by shareholders to generate a favourable impact on the entities cash holdings by lowering the CURRENT ETR since CURRENT ETR does not have an effect on after-tax
Table 6 -Results for CURRENT ETR
CEO EQ_INC CFO EQ_INC
Variable Exp. Sign Coef. T-stat. Prob. Coef. T.stat. Prob.
EQ_INC - - 0,019 -1,441 0,150 - 0,043*** -2,713 0,007 SIZE - - 0,017*** -3,285 0,001 - 0,018*** -3,512 0,000 LEV + - 0,009 -0,331 0,740 - 0,004 -0,167 0,867 ROA - - 0,271*** -5,442 0,000 - 0,284*** -5,844 0,000 NOL - - 0,020*** -3,267 0,001 - 0,021*** -3,344 0,001 PPE - 0,105*** 5,140 0,000 0,102*** 5,024 0,000 R&D - - 0,540*** -3,842 0,000 - 0,553*** -3,943 0,000 INTANG - 0,004 0,226 0,821 0,013 0,627 0,530 FI - - 1,098*** 11,258 0,000 - 1,087*** 11,162 0,000 Adjusted R² 0,160 0,162 F-statistic 42,630 43,314
Significance at the 1%, 5%, and 10% level is indicated by ***, **, and *, respectively. All variables are defined in Appendix E
CURRENT ETR is winsorized to the range [0,1]. The control variables are winsorized at the 1st and 99th percentiles.
earnings. Another explanation could be that shareholders only incentivize the CFO since he is fully responsible of the financial system of the firm which includes tax management. Moreover, the CFO has more knowledge about the possibilities of lowering the ETRs burdened upon the firm (Dyreng et al., 2010). Therefore, shareholders do not find it necessary to incentivize the CEO. This suggests that shareholders expect that the CFO is able to influence the CEO regarding lowering the ETRs by engaging in tax avoidance strategies (pp.1163-1165). A final explanation could be linked to the agency theory framework and is that the contracting weight on a performance measure is related to its signal-to-noise ratio (Banker & Datar, 1989; Lambert, 2001). Meaning that an increase in the sensitivity of the performance measure to the agent’s actions, or a decrease in the conditional variance will increase the weight placed on the performance measure in the ideal compensation contract. In case of the current study, GAAP ETR (“signal”) is more likely to be sensitive towards the CEOs decisions than the CURRENT ETR. Since the CURRENT ETR incorporates deferral tax strategies. From a “noise” viewpoint the CURRENT ETR of my sample has more variation than GAAP ETR. Focusing on the values of the standard deviations of GAAP ETR and CURRENT ETR which are respectively 10,9% and 17,3%. This is most likely due to the distinct character of the tax audit outflows and deferral components. Therefore, GAAP ETR is more informative to investors and the CEOs actions. This leads to more weight allocated to the GAAP ETR in the compensation package and also equity incentives for CEOs (Armstrong et al., 2012).
Also shown in table 6 are the results of the control variables. Most of the variables are significant at the 1% level for both definitions of EQ_INC, CEO EQ_INC and CFO EQ_INC and are in line with the expected relation (ROA, R&D , NOL, FORINC and SIZE). On the contrary, PPE is significant (t-statistic=5,024; 3,344) and positively associated with CURRENT ETR, while a negative association was expected. However, a positive PPE is also found by Hope et al. (2013).
4.5 Corporate Tax Avoidance CASH ETR and Long-Run CASH ETR
The third and fourth proxies that are used for corporate tax avoidance are CASH ETR and Long-Run CASH ETR. The descriptive statistics on CASH ETR (dependent variable), CEO EQ_INC and CFO EQ_INC (independent variables/variables of interest) are stated in table 7 and the descriptive statistics on Long-Run CASH ETR are stated in table 8. The variable to measure tax avoidance strategies, CASH ETR, has 2,120 firm-year observations from which all are related to CEO EQ_INC and CFO EQ_INC which represents the portion of the annual
CASH ETR, revulsive of the descriptive statistics in table 7, has a mean value of 0,287 meaning that on average a firm’s taxes paid is 28,7% (equivalent of 287 dollar cents) per 1 dollar pre-tax book income. This value is similar to the amount that is reported by Hope et al. (2013) and Dyreng et al. (2010). The independent variables of interest, CEO EQ_INC and CFO EQ_INC have a mean value of 0,337 and 0,250 suggesting that when the share price increases with 1% the value of their annual compensation on average rises with 33,7% for the CEO and 25% for the CFO. This is consistent with Jiang et al. (2012) they argue that CEO equity incentives are larger than that of the CFO. Focusing on the control variable values which are also reflected in the descriptive statistics and include; SIZE with a mean value of 3,463, LEV has a mean value of 0,159 saying that each firm is on average for 15,9% financed with debt, ROA with a mean value of 0,127 meaning that on average the return on assets of the firms is 12,7% , NOL
Table 7 - Descriptive statistics for CASH ETR tests
Variables # Obs Mean Std. Deviation Min. Max.
CASH ETR 2120 0,287 0,211 0 0,994 CEO EQ_INC 2120 0,337 0,244 0 1 CFO EQ_INC 2120 0,250 0,226 0 1 SIZE 2120 3,463 0,693 1,825 5,286 LEV 2120 0,159 0,138 0 0,568 ROA 2120 0,127 0,065 0,007 0,344 NOL 2120 0,556 0,497 0 1 PPE 2120 0,164 0,164 0,009 0,719 R&D 2120 0,027 0,027 0 0,114 INTAN 2120 0,188 0,188 0 0,743 FORINC 2120 0,330 0,327 0 1,239
Table 8 - Descriptive statistics for Long-run CASH ETR tests
Variables # Obs Mean Std. Deviation Min. Max.
Long-Run CASH ETR 388 0,287 0,159 0,013 0,684
CEO EQ_INC 388 0,339 0,210 0,005 0,999 CFO EQ_INC 388 0,251 0,118 0,002 0,632 SIZE 388 3,490 0,699 1,903 5,294 LEV 388 0,156 0,128 0 0,513 ROA 388 0,128 0,059 0,025 0,306 NOL 388 0,642 0,480 0 1 PPE 388 0,210 0,160 0,013 0,700 R&D 388 0,020 0,027 0 0,108 INTAN 388 0,236 0,183 0 0,716 FORINC 388 0,331 0,317 0 1,180
has a mean value of 0,556 which means that 55,6% of the firms have a net operating loss carried forward, PPE has a mean value of 0,164 which means that on average the firm’s assets consist for 16,4% out of property, plant and equipment, R&D with a mean value of 0,027 and intangibles has a mean value 0,188 which means that on average each firm’s expenses 2,7% of their total assets on R&D and that on average each firm’s value of total assets consist for 18,8% out of intangible assets. And the final control variable, FORINC, has a mean of 0,330 which suggest that on average 33% of the firm’s income comes from outside the U.S.
The descriptive statistics on the tax avoidance measure, Long-Run CASH ETR (five-year period) is stated in table 8 and has 388 observations from which all are related to CEO EQ_INC and CFO EQ_INC. As can be seen from table 8 is that Long-Run CASH ETR has a mean value of 0,287 meaning that on average a firm’s taxes paid is 28,7% (equivalent of 287 dollar cents) per 1 dollar pre-tax book income. This value is similar to the amount that is reported by Hope et al. (2013) and Dyreng et al. (2010). The independent variables of interest, CEO EQ_INC and CFO EQ_INC have a mean value of 0,339 and 0,251 suggesting that when the share price increases with 1% the value of their annual compensation on average rises with 33,9% for the CEO and 25,1% for the CFO. Based on the control variables in table 8, the following mean values comes forward; SIZE with a mean value of 3,490, LEV with a mean value of 0,156, ROA has mean value of 0,128, a mean value of 0,642 for NOL, PPE has a mean value of 0,210, R&D with a mean value of 0,020 and INTAN and FORINC with mean values of 0,236 and 0,331.
4.5.1 Pearson Correlation Matrix
The Pearson Correlation Matrix measures the correlation between variables for the CASH ETR, CEO EQ_INC, CFO EQ_INC and the control variables. The results are shown in table 9 for the CASH ETR and table 10 for the Long-Run CASH ETR.
As can be seen from the Pearson Correlation Matrix, table 9, less than half of all the correlations are significant at the 1% level. To point out, the most important variable is the dependent variable CASH ETR and the independent variables CEO EQ_INC and CFO EQ_INC. As hypothesized, the dependent variables CFO EQ_INC (-0,200) is negatively correlated with CASH ETR. However, the correlation of CEO EQ_INC is not significant with the variable CASH ETR and is not consistent with hypothesis 1. Furthermore, when focusing on the control variables and their correlation with CASH ETR you are able to see that NOL (-0.123) is significant and negatively correlated meaning that entities with a net carry loss forward
correlated and significant suggesting that companies with more R&D expenses induce tax avoidance planning due to the different book and tax treatment and FORINC (-0.133) is also significant and negatively correlated suggesting that firms with more foreign income have more opportunities to engage in tax avoidance.
Table 9 Pearson correlation matrix for CASH ETR test
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (1) CASH ETR 1 (2) CEO_EQ INC -0,046 1 (3) CFO_EQ INC -0,200 0,118 1 (4) SIZE -0,045 0,105 0,151 1 (5) LEV 0,017 -0,075 0,124 0,354 1 (6) ROA -0,022 0,089 -0,046 -0,158 -0,264 1 (7) NOL -0,123 -0,092 0,103 0,040 0,085 -0,198 1 (8) PPE 0,043 0,028 -0,169 0,029 0,151 0,040 -0,176 1 (9) R&D -0,212 0,072 0,092 0,027 -0,133 0,037 0,143 -0,280 1 (10) INTANG 0,000 -0,043 0,366 0,187 0,310 -0,225 0,161 -0,409 0,011 1 (11) FORINC -0,133 -0,023 0,065 0,280 0,078 -0,187 0,165 -0,175 0,336 0,004 1
Table 10 Pearson correlation matrix for Long-run CASH ETR test
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (1) L-R CASH ETR 1 (2) CEO_EQ INC -0,074 1 (3) CFO_EQ INC -0,326 0,221 1 (4) SIZE -0,077 0,130 0,258 1 (5) LEV 0,020 -0,096 0,236 0,385 1 (6) ROA 0,055 0,233 -0,110 -0,205 -0,279 1 (7) NOL -0,139 -0,062 0,179 0,019 0,075 -0,221 1 (8) PPE 0,036 0,030 -0,312 0,046 0,182 0,029 -0,154 1 (9) R&D -0,308 0,099 0,178 0,018 -0,154 0,086 0,125 -0,269 1 (10) INTANG 0,014 -0,048 0,691 0,182 0,302 -0,223 0,146 -0,406 -0,029 1 (11) FORINC -0,194 -0,025 0,125 0,297 0,077 -0,173 0,132 -0,171 0,376 -0,023 1
Significance at the 1% level is indicated by the bold correlations All variables are defined as in appendix E
The Pearson Correlation Matrix for the tax avoidance measure Long-Run CASH ETR is reflected in table 10. As can be seen from this table less than half of the correlations are not significant at the 1% level. To mention, the most important variable is the dependent variable Long-Run CASH ETR and the independent variables CEO EQ_INC and CFO EQ_INC. As hypothesized, the independent variable CFO EQ_INC is negatively correlated and significant. On the contrary, the independent variable CEO EQ_INC is negatively correlated however not
significant and is not consistent with hypothesis 1. Furthermore, the significant correlations between Long-Run CASH ETR and the control variables are; NOL with a value of -0,139, R&D with a value of -0,308 and FORINC with a value of -0,194. All the significant correlation are in line with my expectations.
The correlations between the independent and the control variables for table 9 are all in the range of 0,366 and -0,409. To make sure there is no multicollinearity I tested for the variance inflation factor (VIF). If VIF is above 10 then multicollinearity is present. However, in this case the maximum number of VIF is 1,690. With respect to the results from table 10, all correlations are in the range of -0,406 and 0,691 and the maximum number of VIF in this case is 2,687.
4.5.2 CASH ETR and Long-Run CASH ETR test – hypothesis 1 and 2
The regression analysis, table 11, shows the results of the CASH ETR. The results of the regression analysis of the Long-Run CASH ETR is reflected in table 12. Both regression are used to test hypothesis 1 and 2. Which are the following; (H1) CEO equity incentives are negatively associated with a firm’s ETR and (H2) CFO equity incentives are negatively associated with a firm’s ETR.
The regression analysis, as presented in table 11, where CASH ETR is the dependent variable and EQ_INC is defined as the independent variable CEO EQ_INC, shows that the coefficient is insignificant (-0,018; t-statistic= -1,46). Therefore, H1 is not supported. With respect to H2, in the regression where the independent variable EQ_INC is defined as CFO EQ_INC, shows that the coefficient is negative and significant (-0,125; t-statistic = -9,025) and therefore supports H2. This means that equity incentives compensation induce the CFO to engage in riskier projects and thus tax avoidance strategies to lower the firm’s CASH ETR to increase future tax outcomes of the company which can lead to higher stock – and option prices to increase his own personal wealth. These results suggest that CFOs have more influence on tax avoidance strategies to lower the CASH ETR than CEOs.
