Corporate tax avoidance and firm value in the light of the 2008 financial crisis

(1)

Corporate tax avoidance and firm value in the light

of the 2008 financial crisis

By

Reinier M.M. Hovius

University of Groningen

Faculty of Economics and Business

MSc International Economics & Business

June 2015

Supervisor: Prof. dr. H. (Hans) van Ees

Co-assessor: Dr. D.J. (Dirk) Bezemer

Abstract: The primary goal of this paper is to analyze the effect of corporate tax avoidance activities on the profitability of U.S. firms listed in the S&P 500. The paper predicts that the effect of tax avoidance activities on firm value should vary systematically with the quality of corporate governance. The simple presumption that corporate tax avoidance activities result in a transfer of value from the state to shareholders has not appeared to be validated in the data. The pattern found in this paper is consistent with the shareholder-agency perspective and claims that the effect of corporate tax avoidance activities on firm value is larger for well-governed firms. The 2008 financial crisis has not led to a larger effect for well-well-governed firms to engage in corporate tax avoidance activities. But it did led to a positive significant effect for poorly-governed firms to engage in corporate tax avoidance activities to generate off-the-book revenue.

(2)

LIST OF TABLES

TABLE 1: Explanation Tests in Stata………...20 TABLE 2: Summary Statistics………...22 TABLE 3: Corporate tax avoidance, Firm value, and Governance institutions………...25 TABLE 4: Tax Avoidance, Firm Value and Governance Institutions: pre-Crisis and Crisis….27 TABLE 5:Alternative measures of Earnings Management………..30 TABLE 6: Tax Avoidance Robustness Checks………32

LIST OF ABBREVIATIONS

AMEX American Stock Exchange

CFTE Current Federal Tax Expense EBIT Earnings Before Interest Taxes

EU European Union

GAAP General Accepted Accounting Principles GAO Government Accountability Office GICS Global Industry Classification Standard

IMF International Monetary Fund

IRS Internal Revenue Service

IT Information Technology

NASDAQ National Association of Securities Dealers Automated Quotations

NOL Net Operating Loss

NYSE New York Stock Exchange

OLS Ordinary Least Squares

ROA Return On Assets

SEC Security and Exchange Commission S&P 500 Standard & Poor's 500

U.S. United States of America

(4)

1. | INTRODUCTION

Starbucks, Google and Amazon have been under discussion for avoiding paying their taxes (BBC News Magazine, 2013). Apple is accused by the U.S. government for avoiding billions of dollars in taxes by setting up dummy headquarters in Ireland (Forbes, 2013). Where,

McDonald’s is tackled by trade unions for super-sized tax avoidance activities in the EU (Bloomberg, 2015).

Tax consequences are a motivating factor in many corporate decisions. Managerial actions designed solely to minimize corporate tax obligations are thought to be an increasingly important feature of U.S. corporate activity (Bankman, 2004; Slemrod, 2004; Desai & Dharmapala, 2009). Previous research on corporate tax avoidance activities performed by Yin (2003) reports effective tax rate reductions in S&P 500 firms from an average of 28.9% in 1995 to 24.2% in 2000. Statistics from the U.S. Government Accountability Office (GAO, 2008a) indicate that over the 1998–2005 period, 24% of U.S controlled corporations and 34% of foreign-controlled corporations reported zero tax liabilities. The reasons behind the zero reported tax liabilities in the U.S. include the (ab)use of transfer pricing rules and utilization of tax havens. The U.S. GAO (2008b) also state that in 2007, 83 out of the 100 largest publicly listed U.S. firms had subsidiaries in jurisdictions listed as tax havens. Graham and Tucker (2005) indicate a decline in U.S. corporate tax payments. For example, in 1994 S&P 500 firms paid federal taxes of 29 cents per dollar of reported profits, but this fell to just 18 cents per dollar one decade later1. Leonhardt (2011) reports that 115 firms from the S&P 500 paid a total corporate tax rate of 20% or less over the past 5 years as opposed to the U.S. federal statutory rate of 35%. Arguably, the United States has one of the highest corporate tax rates in the world. This has motivated companies to devote enormous time and effort in finding loopholes which resulted in a decrease of U.S. government taxable income (Leonhardt, 2011).

Corporate tax avoidance, unlike tax evasion, is legal (The Economist, 2012). However, that does not imply that engaging in tax avoidance activities is not costly. Besides the direct costs of engaging in such activities, intra-firm friction between managers and shareholders could result in managers pursuing self-serving objectives instead of directly transferring money from the state to its shareholders. Thus, this raises the question how do investors actually value such corporate tax avoidance actions?

(5)

The extant literature on the effects of taxes in financing and investment decisions (Auerbach, 2002; Graham, 2003) does not typically incorporate agency problems in the valuation of, and response to, tax avoidance opportunities. Research on corporate tax avoidance activities emphasizes the legal distinction and accounting consequences of corporate tax avoidance, but does not consider the question how institutional investors value such activities.

This paper analyzes the degree to which corporate tax avoidance activity are valued by investors in a large sample of U.S. firms. The direct effect of tax avoidance activities is to increase the after-tax value of the firm. However, the effect of tax avoidance activity on firm value is also a function of firm governance, as predicted by the agency perspective on corporate tax avoidance activities. This is because in well-governed firms managers are more limited to pursue self-serving objectives and therefore the benefits of tax avoidance activities are better valued by investors. In poorly-governed firms, managers gain more freedom to engage in self-serving objectives which could result managerial opportunism and rent diversion (Desai & Dharmapala, 2009). Therefore the net effect of corporate tax avoidance activities on firm value should be greater for firms with a better quality of corporate governance. This first hypothesis is tested using a sample of S&P 500 firms, comprising a dataset of 1271 observations at the firm-year level for 221 participating firms for the period 2005-2010.

The constituted dataset also creates an incentive to analyze the role of corporate tax avoidance activities in relation to the 2008 financial crisis, which had severely hit the world economy (Hemmelgarn & Nicodeme, 2010). The commencement of the 2008 financial crisis splits the sample period into a 3-year pre-crisis and a 3-year crisis period.

(6)

Overall, this paper analyzes the effect of corporate tax avoidance activities on the profitability of U.S. firms listed in the S&P 500, with distinctive emphasis for the 2008 financial crisis.

The rest of the paper proceeds as follows. The following section reviews the related literature on corporate tax avoidance activities, firm value, the quality of corporate governance institutions and the commencement of the 2008 financial crisis. Section 3 describes the data, methodology and empirical specification. Section 4 discusses the empirical results, a series of extensions and robustness checks. Where Section 5 presents the conclusion, limitations and inquiries for further research.

2. | LITERATURE REVIEW

Hanlon and Heitzman (2010) define tax avoidance as the reduction of explicit tax payments. The literature on corporate tax avoidance activities hold the view that positive book-tax differences and low effective tax rates reflect tax avoidance behavior (Kim et al., 2011). The book-tax gap value is known as the difference between income reported to capital market and tax authorities (Manzon & Plesko, 2002; Desai, 2003, 2005). The effective tax rate for corporations is the average rate at which its pre-tax profits are taxed (Dyreng et al., 2008). Accordingly, growing book-tax differences and declining effective tax rates for U.S. public corporations since the mid-1990s have stimulated researchers to investigate determinants and consequences of corporate tax avoidance activities (Schackelford & Shevlin, 2001; Graham, 2003; Desai & Dharmapala, 2009). Purported growth on corporate tax avoidance activities has resulted in two alternative views on the motivation and effects of this activity.

(7)

regarding the engagement in corporate tax avoidance activities, like a firm’s reputation or any potential penalty or threat that could harm the firm. This view purely focuses on the potential gain for investors while engaging in corporate tax avoidance activities. In other words, corporate tax shelters are merely tax-saving devices without any agency dimension.

The second view on corporate tax avoidance activities focuses more on dimensions of the principal-agent relationship between managers and investors. This agency view of tax avoidance is attracting increased attention in the literature (Desai, 2005; Desai & Dharmapala, 2006, 2007, 2009; Hanlon and Heitzman, 2009). The agency dilemma explains that managers do not always pursue the same goal and interest as the investors. Investors want to increase the after-tax value of the firm, where the managers might seek personal advantages and therefore act in their own best interest. This problem often arises due to asymmetric information (i.e. the manager having more information), such that investors cannot directly ensure that managers are always acting in their best interest.

(8)

benefits to shareholders, given the opportunities for diversion that these vehicles provide. In their most recent article, Desai and Dharmapala (2009) indicate that according to the agency problem, corporate tax avoidance activities can create a shield for managerial opportunism and the diversion of rents. Hanlon and Heitzman (2009) further contribute to this claim. In their research they found a negative market reaction to tax shelter disclosure, suggesting that investors are concerned about the possibility that tax shelters are intertwined with managerial opportunism and performance manipulation.

Both views imply that the engagement in corporate tax avoidance activities have a different result on firm value. The first, and traditional, view of corporate tax avoidance suggests that shareholder value should increase with tax avoidance activity. Thus, corporate tax avoidance activities should have a positive effect on firm value. The second view, the agency perspective on corporate tax avoidance activities provides a more nuanced prediction (Desai & Dharmapala, 2009).

(9)

differences in corporate ownership implies different incentives to engage in corporate tax avoidance activities.

Following the second view, this paper focuses on the consequences of corporate tax avoidance activity under the agency perspective. The role of corporate governance will be taken into consideration. Corporate governance is defined as the ‘determination of the broad uses to which organizational resources are deployed and the resolution of conflicts among the myriad participants in organizations’ (Daily et al., 2003). The agency perspective implies a conflict of interest between managers and corporate shareholders. This conflict is central in the analysis of modern corporations in which managers have less than full ownership of the cash flow rights of the firm (Jensen & Meckling, 1976). Based on their analysis, Lemmon and Lins (2003) suggest that firm`s ownership structure is a primary determinant in the extent of agency problems between controlling managers and external investors. This has an important implication for the valuation of the firm. Therefore ownership structure is a relevant indicator to further analyze the quality of the corporate governance.

The quality of corporate governance will moderate the relationship of corporate tax avoidance activities on firm value. Also it will distinguish well-governed and poorly-governed firms. Nevertheless, it is difficult to determine a priori what the relative effects on firms with poorly-governed and well-poorly-governed institutions are upon the relation of corporate tax avoidance activities and firm value. In well-governed firms (i.e. a high quality of corporate governance) activities are indicated with well-defined mechanisms, processes and relations by which corporations are controlled and directed. A clear distribution of the rights and responsibilities among different participants in the corporation is set. There is less room for negotiation between principals and agents, rules and procedures are determined, which makes the engagement in managerial opportunism and the diversion of rents less likely.

(10)

increase in tax sheltering at this firm would lead directly to the same increase in the after-tax value of the firm, but would also lead to a higher level of diversion (i.e. lower income reported to the shareholders). Therefore, the net increase in the after-tax value of the firm would be smaller than in the case of the well-governed firm.

Desai and Dharmapala (2009) contribute by explaining that the goal of corporate tax avoidance is to increase the after-tax value of the firm. The quality of corporate governance has a moderating effect on this relationship and the effect will be larger for well-governed firms then for poorly-governed firms, due to the increased opportunities for managerial opportunism and rent diversion. Therefore the following hypothesis is formulated:

H1: The effect of corporate tax avoidance activities on firm value will be larger for

well-governed firms relative to the effect for poorly-well-governed firms.

Here, a contribution will be made to the empirical research of Desai and Dharmapala (2009) who found in their sample period between 1993-2001 that the view of corporate tax avoidance as a transfer of resources from the state to shareholders is incomplete given the agency problems characterizing shareholder-manager relations. The hypothesis is not about the effects of corporate tax avoidance activities on firm value per se but rather about the effects of the interaction between tax avoidance activities with the quality of corporate governance.

The first hypothesis will be further examined for the time period in which the sample is selected. A sample of large U.S. firms is taken in the period 2005-2010 to indicate the role of corporate tax avoidance activities in relation to the global recession. The 2008 financial crisis has severely hit the world economy (Hemmelgarn & Nicodeme, 2010). The U.S. suffered tremendously from this crisis, where several large financial institutions have threatened to collapse2, the stock market has fallen dramatically, and the spread of a variety of different types of loans over comparable U.S. Treasury securities have widened dramatically (Chari et al., 2008). The crisis played a significant role in the failure of key businesses, declines in consumer wealth estimated in trillions of U.S. dollars, and a downturn in economic activity leading to the 2008–2012 global recession. The effects of the global recession are evident and it has changed the operating behavior of corporations. The roots of the 2008 financial crisis

2

(11)

are found in bad regulation, lack of transparency, and market complacency brought about by several years of positive returns (Zingales, 2008). This has affected the value of a firm. Evidently, the cause of every crisis derives from a different nature. The 2008 financial crisis was caused by an expectation of defaults of payments on subprime mortgages in the U.S.. Where the East Asian crisis (1997) was marked by a devaluation of stock markets, currencies, and other asset prices. Nevertheless, the East Asian crisis provides a well-defined example to observe how the quality of corporate governance has affected firms behavior during a period of crisis.

Lemmon and Lins (2003) investigated the effect of ownership structure on firm value during the East Asian crisis. Under the agency perspective, they highlight that the crisis negatively impacted firms investment opportunities and that stock returns in poorly-governed firms where 10-20 percentage points lower than those of well-governed firms. Lemmon and Lins (2003) found that the ownership structure played an important role upon the value of a firm during the East Asian financial crisis. Mitton (2002) implied that corporate governance practices in East Asia could have made countries more vulnerable to the crisis and therefore exacerbated the crisis once it started. Mitton (2002) found the quality of corporate governance has an significant influence on firm-level performance during a crisis. Mitton (2002) also suggests that the negative impact is greater on firms in which controlling shareholders have stronger incentives and means to expropriate resources. His article explains that corporate governance activities could become more critical in a financial crisis for two reasons. First, expropriation of minority shareholders could become more severe during a crisis. Also Johnsen et al. (2000) contributes that a crisis can lead to greater expropriation because the expected return on investment falls. Secondly, a crisis could force investors to recognize and take into account weaknesses in corporate governance that already existed. An example based on the East Asian crisis is explained by Rajan and Zingales (1998), who argue that investors ignored weaknesses of East Asian firms while the region was doing economically well, but quickly pulled out once the crisis began because they believed the region lacked adequate institutional protection for their investments. The literature found evidence that during the East Asian financial crisis the quality of corporate governance influenced firm behavior. It states that poorly-governed firms have lost relatively more value during the East Asian crisis in comparison to well-governed firms (Mitton, 2002).

(12)

policies and regulations, faced a drop in property and asset prices, experienced an increase in volatility of the stock markets and observed a withdrawal of funds available in the market. Additionally, many countries faced a period of significant economic growth before the crisis started (Allen & Snyder, 2009). Therefore this paper contributes on the existing literature of the East Asian crisis and assumes that the quality of corporate governance also played an important role on firm value during the 2008 financial crisis.

(13)

Literature indicate that the quality of corporate governance institutions are important indicators for firm value during a period of crisis (Mitton, 2002; Lemmon & Lins, 2003). Overall, firm value tends to decrease during a period of crisis, profits diminish and investors might try alternative methods to sustain or create revenue. Poorly-governed firms are likely to lose more value during a period of crisis then well-governed firms (Mitton, 2002). This could be explained trough the ease for well-governed firms to derive revenue from alternative sources, for example corporate tax avoidance activities. Since well-governed firms are able to mitigate the effect for managerial opportunism that may complicate the engagement in corporate tax avoidance activities (Desai & Dharmapala, 2009).

The commencement of the 2008 financial crisis was, by all accounts, an unexpected event, and therefore presents an interesting opportunity to study the proximate effect of the quality of corporate governance institutions on the relation of corporate tax avoidance activities and firm value. A second hypothesis will be tested to infer the interaction of tax avoidance activities for governed firms during a period of pre-crisis and crisis. Implying that well-governed firms engage more in corporate tax avoidance activities during the financial crisis to generate `additional` revenue. Therefore the following hypothesis is constituted;

H2: Corporate tax avoidance by well-governed firms have a larger effect on firm value during

crisis years then pre-crisis years.

Taking a sample of 2005-2010, this paper analyzes the effect of the commencement of the 2008 financial crisis upon the effect of corporate tax avoidance on firm value moderated for corporate governance institutions.

3. | DATA AND EMPIRICAL SPECIFICATION

In order to analyze the phenomenon of corporate tax avoidance activities and to empirically test the hypotheses described in section 2, data is retrieved from the Thomson Reuters Datastream. A large sample of S&P 500 listed firms is taken into account. Publicly available data will be collected. This section describes each of these steps in detail, discusses the data and the empirical specifications.

(14)

The 500 leading companies capture approximately 80% of the available market capitalization on the American capital market (McGraw Hill Financial, 2015). This has made the S&P 500 one of the most commonly used benchmarks for the overall U.S. stock market. All major companies are categorized according to the Global Industry Classification Standard (GICS). The GICS has divided the S&P 500 into 10 different sectors, 24 industry groups, 67 industries and 156 sub-industries. Firms in the financial sector are omitted from the sample because of unrepresentative data and potential biases in the available data. Furthermore, firms with missing data and potential outliers are restricted from the model, eventually comprising a dataset of 1271 observations at the firm-year level from 221 participating firms for the period 2005-20103.

3.1 Measuring firm value, governance, and corporate tax avoidance

In the existing literature, the standard since Demsetz and Lehn (1985) is to use the Tobin`s Q as a measure of firm value. The Tobin`s Q is the ratio between a physical asset's market value and its replacement value. It is computed as the market value of equity plus the book value of debt (computed as the book value of assets minus the book value of equity) divided by the book value of assets. The definition of Tobin’s Q has been widely used in the existing literature. (La Porta et al. 2002; Doidge et al. 2004; Gozzi et al. 2006). Desai and Dharmapala (2009) modify the Tobin`s Q by excluding deferred tax expenses in the definition, because current tax avoidance activity may result in changes to future tax liabilities and thus create a mechanical correlation between the dependent variable and the measure of tax avoidance. Tobin`s Q will be the dependent variable in the analysis, where the effects of corporate tax avoidance activities on firm value will be measured. The modification regarding the deferred tax expenses will be taken into consideration and, therefore, firm value is defined according to the following equation;

TobinsQ =

Market value of equity + (Total Assets−Accumulated depreciation)− Total shareholder`s equity − Deferred tax expense

(Total Assets − Accumulated depreciation) (1)

As an indirect measure of the quality of corporate governance, the institutional ownership of a firm will be used as a proxy variable. Institutional ownership refers to the ownership stake in

3

(15)

a company that is held by large financial organizations, pension funds or endowments. The primary measure of institutional governance is the fraction of the firm`s shares owned by institutional investors, this relationship to firm value has been thoroughly examined in previous literature (Morck et al., 1988; McConnell & Serveas, 1990). For measuring the fraction of the firm`s shares owned by institutional investors data is retrieved from the free float datatypes available in Thomson Reuters Datastream. The percentage of total strategic holdings is measured as the percentage of total shares in issue of 5 percent or more held strategically by institutional investors (based on Schedule 13F filings with the U.S. Security and Exchange Commission (SEC) by large institutional investors).4 In other words, institutional holdings are evaluated as strategic if they obtain 5 percent or more of the total share of capital in a company. The total percentage of strategic holdings represent the sum of strategic holdings held by, government institutions, cross holdings by one company in another, pension funds, endowment funds, investment banks, employees and foreign investors. The fraction [0, 1] of the percentage of total strategic holdings is taken into consideration (Desai & Dharmapala, 2005; 2009) and will therefore be defined as;

Institutional Ownership =Percentage of total strategic holdings

100

(2)

The basic motivation underlying this proxy is that institutional investors have greater incentives and capacity to monitor managerial performance (Desai & Dharmapala, 2009). Thus, the higher the fraction of firm shares owned by institutional investors, the greater the degree of scrutiny to which managerial actions are subjected, the less important are agency problems between managers and shareholders. This is stated under the assumption that institutional investors will not expropriate minority shareholders.

In order to test the first hypothesis, a distinction between high- and low-institutional ownership will be constructed. The mean of the fraction of institutional ownership is used to distinguish between high- and low-institutional ownership, thereby creating two sub-samples of well-governed and poorly-governed firms.

(16)

Corporate tax avoidance is more complicated to measure. The problem arises from the fact that tax returns are often confidential and therefore income reported to tax authorities cannot be observed directly and must be inferred using financial accounting data, which is described in Manzon and Plesko (2002) and implemented in Desai and Dharmapala (2006). In the literature, the book-tax gap value is used as a proxy to determine corporate tax avoidance activities. The book-tax gap value is the difference between income reported to capital markets, using General Accepted Accounting Principles (GAAP), and to tax authorities. This measure of corporate tax avoidance has attracted considerable interest in recent years (Manzon & Plesko, 2002; Desai, 2003, 2005). The approach uses firms current federal tax expense (CFTE) and divide this tax liability by the U.S. federal corporate tax rate5. In other words, a firms inferred taxable income can simply be subtracted from the firms reported U.S. domestic pretax financial statement income;

The book tax gap value = U. S. pretax financial income − U. S. inferred taxable income (3)

Devoted as the difference between the book income reported by a firm to its shareholders and the tax income reported to the IRS. Manzon and Plesko (2002) use the current federal tax expense to estimate the inferred U.S. taxable income for firms. CFTE is measured as the portion of the total income tax provision that is based on domestic taxable income.

CFTE = U. S. inferred taxable income × Statutory tax rate (4)

Assuming that each firm faces a statutory tax rate that is given by the top statutory federal corporate income tax rate, which is 35% during our sample period, the firm’s inferred taxable income is simply;

U. S. inferred taxable income = CFTE

Statutory tax rate (5)

Given this inferred value of the firms taxable income, the book-tax gap value can be calculated. The firms pretax financial income refers to the income that is accumulated before

(17)

income taxes are paid, and is calculated as the total pretax income on the balance sheet. In the rest of the paper the book-tax gap value is denoted by BTi,t for firm i in year t.

Increases in tax avoidance are likely, at the margin, to be less valuable for loss-making firms with no current tax liability (Desai & Dharmapala, 2005). Therefore the analysis restricts the sample to firm-years in which firms are estimated to have a positive taxable income. Another reason for excluding those observations is that for a negative taxable income it is not feasible to calculate the total tax expense due to the fact that the marginal tax rate is unclear (i.e. not 35%).

The book-tax gap does not automatically represents a growth in corporate tax avoidance. Therefore the measure of corporate tax avoidance must control for other factors. Existing literature found that the over-reporting of financial income, also known as earnings management, contributes to the measured book-tax gap value (Lev & Nissim, 2004; Hanlon, 2005). Lev and Nissim (2004) and Hanlon (2005) investigated how the book-tax gap predicted the quality of future earnings, and interpreted the entire book-tax gap as being due to earnings management activity. They analyzed the consequences of managed earnings for subsequent accounting and market outcomes, but did not focus on the contemporaneous valuation of the tax avoidance component of the book-tax gap value (Desai & Dharmapala, 2009). Desai and Dharmapala (2006) state that the aim of earning management is the smoothing of reported income over time in order to reach bonus targets and avoid reporting losses. Healy (1985) argued that earnings management most likely occurs through the exercise of managerial discretion in determining accounting accruals, which are the adjustments to realized cash flows that are used in calculating the firms net income. Earnings management is by itself an indicator of inefficient corporate governance and therefore the quality of corporate governance is also dependent on the engagement of a firm in earnings management practices6.

Following the work of Desai & Dharmapala (2009) this paper states that book-tax gaps differences are attributable to either earnings management or tax avoidance activity. To determine the degree to which earnings management is responsible for the book-tax gap value, data on accruals will be used to isolate the component of the book-tax gap value that is attributable to earnings management. The firms total accruals will be used as an independent variable for earnings management activity and implemented in the model as a proxy variable

6

(18)

for determining corporate tax avoidance activity. Existing literature found many different calculations for accrual earnings measurement (Healy, 1985; DeAngelo, 1986; Jones, 1991; Dechow et al., 1995; Bartov et al., 2001). In this paper total accruals will be denoted as TAi,t, for firm i in year t and defined as follows;

Total accruals = Accrual earnings − Cash earnings (6)

Here, accrual earnings are determined as the firms net income on the balance sheet and cash earnings are denoted as the net cash flow from operating activities. This cash-flow approach is better suited to describing accruals in all situations7.

To infer that corporate tax avoidance activities affect firm value and should be greater for firms with stronger governance institutions, a selection of control variables will be taken into the model to control for firm value. Six control variables will be implemented in the regression analysis. First, Return On Assets (ROA) is used to control for the firms past performance. Fama and French (2000) explain that ROA can either control for mean reversion or persistence in earnings. Secondly, the financial leverage ratio will be used as a control variable that may affect firm value. Haushalter (2000) finds evidence that firms with higher leverage have greater risk-reducing activities which affects firm value. Thirdly, capital expenditures, scaled by the book value of assets, are used to control for changes in firms investment levels. It emphasizes the unobserved shocks to the investment opportunities available to the firm. This may cause variations in firm value that are unrelated to tax avoidance activity, as firm value is generally thought to be closely associated with firms investment decisions (Desai & Dharmapala, 2005). A fourth variable where the model will be controlled for is the firm’s annual growth rate. The percentage of one-year sales change is used to measure a firm’s sales growth. A growth in sales is generally found to be positively correlated with a firm’s value (Schmalensee, 1989; Hirsch, 1991). A fifth variable will control for the changes in firm size over time by using the total market value of the firm scaled by the book value of assets. Recent research showed evidence that managers within larger firms are more likely to destroy shareholder wealth, i.e. through bad acquisitions, and therefore decrease the overall value of the firm (Offenberg, 2009). Lastly, this model measures corporate tax avoidance activity as restricted to U.S. tax expense and U.S. federal taxes.

7

(19)

However, the exclusion of foreign taxes and income from this calculation avoids problems associated with inferring the applicable foreign tax rates. But, foreign activities can affect U.S. tax liability under the worldwide system of taxation that applies to U.S. corporations. Therefore, a control variable that proxies for foreign activity, the absolute value of foreign income or loss scaled by the book value of assets, will be taken into the regression analysis.

3.2 The empirical specification

The central claim of the paper concerns the interaction of corporate governance institutions and tax avoidance activity on firm value. First, the question whether corporate tax avoidance activities affects the value of the firm is of considerable interest. This should be addressed using the following specification;

TobinsQit = β1BTit + β2TAit + β3CONTROLit + μi + ɛt + υit` (7)

Where the variables BTit and TAit are as defined above, μi and ɛt are firm and year fixed effects, respectively, and υit is the error term. The control variables are explained in the following equation, denoted for firm i and year t;

CONTROLit = β1ReturnOnAssetsit + β2FinancialLeverageit + β3CapitalExpendituresit

+ β4GrowthRateit + β5FirmSizeit + β6ForeignIncomeit (8)

In order to test the first hypothesis, a specification has to be made for testing the moderating effect of the quality of corporate governance on the relationship between corporate tax avoidance activities and firm value. Therefore the measure of institutional ownership (Iit) is added into the model. Also, an interaction term between institutional ownership and corporate tax avoidance activities will be specified;

TobinsQit = β1BTit + β2TAit + β3Iit + β4(Iit *BTi,t) + β5CONTROL + μi + ɛt + υit` (9)

(20)

avoidance activities on firm value is larger for firm-years in which institutional ownership is higher (and corporate governance is stronger).

The second hypothesis will be tested for a 3-year pre-crisis (2005, 2006, 2007) and a 3-year crisis (2008, 2009, 2010) period. This will be done by simply splitting the sample in half. By splitting the sample it is possible to estimate the effect of corporate tax avoidance activities on firm value during both periods and to investigate whether the effects of tax avoidance activities have increased or decreased. Also a distinction between poorly-governed and well-governed is inferred to test whether corporate tax avoidance by well-well-governed firms have a larger effect on firm value during crisis years then pre-crisis years.

3.3 Methods and assumption tests

Before proceeding with the regression analysis, several assumptions need to be analyzed. The variables as described in section 3.1 will be examined on the normal distribution of the dependent variable, multicollinearity, heteroskedasticity and serial autocorrelation. Furthermore, a test will be performed for the appropriateness of a pooled Ordinary Least Squares (OLS) regression. If OLS cannot be performed, a second test is investigated whether the regression will be performed using a random effect model or a fixed effect model. Table 1 summarizes the tests that will be performed in order to verify the assumptions corresponding to the regression analyses.

Table 1: Explanation Tests in Stata

Breusch-Pagan Lagrange Multiplier Test Test for panel regression (xttest0)

Hausman Test Test for fixed or random effect (hausman)

Wooldridge Test for Autocorrelation Test for serial autocorrelation (xtserial) Modified Wald Test Test for heteroskedasticity (xttest3) Jarque-Bera Normality Test Goodness-of-fit test (JB) Variance Inflation Factor Test Test for multicollinearity (VIF)

(21)

hypothesis is rejected8, there is a significant difference in the variance between firms meaning that it is not possible to perform pooled OLS estimations. If the null hypothesis is rejected under the Breusch-Pagan Lagrange Multiplier test, the Hausman test has to be performed in order to detect whether a random or a fixed effect model is most appropriate. The underlying idea in the Hausman test is that both the random and fixed effects estimators are consistent if there is no correlation between the error component (ui) and the explanatory variables (xit). Accepting the null hypothesis explains that the fixed effects estimator and random effects estimator are consistent and therefore the random effects model is preferred due to a higher efficiency, while under the alternative hypothesis only fixed effects are consistent with the model and therefore preferred.

To test for serial autocorrelation and heteroskedasticity, the Wooldridge test for autocorrelation and a modified Wald test for group wise heteroskedasticity are performed. The Wooldridge test for autocorrelation in panel data tests the null hypothesis of no first-order autocorrelation against the alternative hypothesis of first-order autocorrelation (Drukker, 2003). The modified Wald test tests the null hypothesis that data is homoscedastic against the alternative hypothesis that the data is heteroscedastic. To test for normality, a Jarque-Bera test is performed. This test is based on two measurements, skewness and kurtosis. Skewness refers to how symmetric the residuals are around zero, where kurtosis refers to the degree of peakedness of the graph of a statistical distribution. The null hypothesis under the Jarque-Bera test implies that a variable is normally distributed, the alternative hypothesis implies it is not normally distributed.

Furthermore, multicollinearity will be tested by investigating the ‘variance inflation factors’ which indicates the existence of multicollinearity among variables. This provides an index that measures how much the variance of an estimated regression coefficient is increased because of multicollinearity. As a rule of thumb, a variable whose VIF values are greater than 10 may merit further investigation9. The results for these assumption tests can be found in the next section.

8

Var(u) = 0

(22)

4. | DATA ANALYSIS AND RESULTS

4.1 Summary statistics

The descriptive variables in section 3 are summarized and reported in table 2. The sample constitutes of 1271 observations over the period of 2005-2010. This time-period is chosen in order to make a distinction between a three-year pre-crisis period (2005, 2006, 2007), and a three-year period of crisis (2008, 2009, 2010).

Table 2: Summary Statistics

Variable Mean

Standard

Deviation Number of Observations TobinsQ (including deferred tax

expense) .3537494 .4859874 1271

TobinsQ (excluding deferred tax

expense) .4024159 .4328625 1271

Book-tax gap (scaled) .0068241 .0730403 1222

Institutional ownership (fraction) .1423918 .1262766 1271

Total accruals (scaled) -.0787292 .168716 1271

Return On Assets (%) 8.718345 7.112872 1269

Financial Leverage (ratio) .5705504 .1928242 1271 Capital Expenditures (scaled) .0517992 .0435247 1271

Growth Rate (%) 10.8448 19.77667 1271

Firm Size (scaled) .0019848 .0015665 1271

Foreign income (scaled) .0608004 .0948288 1118

Net Operating Loss (scaled) .4572881 .367804 1271 Discretionary accruals (scaled) :

Jones Model -.7054057 .4455776 1271

Discretionary accruals (scaled):

Modified Jones Model -.6976223 .4482512 1269

Note: These variables are defined as explained in section 3. Scaled variables are deflated by the book value of assets.

4.2 Regression assumption tests

(23)

hypothesis is rejected, implying that pooled OLS regression is not appropriate. Therefore, the Hausman test is performed in order to test for the appropriateness of fixed or random effects regression analysis. The results of this test indicate that a fixed effect model is most appropriate, therefore rejecting the null hypothesis that the fixed effects estimator and the random effects estimator are consistent.

Furthermore a Wooldridge test is performed to test for serial autocorrelation. This is tested because serial autocorrelation in linear panel-data models biases the standard errors and causes the results to be less efficient (Drukker, 2003). In other words, the obtained firm-level data might not be independent, and if so, the data should be clustered. Performing the Wooldridge test indeed indicates that serial autocorrelation is present in the model. Thus, the model will be controlled for serial autocorrelation using the cluster option. This option clusters the time-series observations on individuals. Additionally, the existence of heteroskedasticity is tested by performing a Wald test. The model is subjected to heteroskedasticity, which will be controlled for by using a robust option when running the regression. This results that robust standard errors are valid for both heteroscedastic and homoscedastic errors. The data distribution of the dependent variable is also tested on normality by a Jarque-Bera test. Based on this test it can be concluded that the dependent variable is normally distributed, and therefore no further adaptations in the model has to be performed. Finally, variance inflation factors are investigated to test for multicollinearity. None of the variance inflation factors, however, are above the rule of thumb value 10 (or tolerance value <0.01).

4.3 Results

The results using a fixed effect model for testing the first hypothesis (i.e. equations 7 and 9) are reported in table 3, using the controls and sample as described above. A fixed effect model is used to allow for differences in the intercept parameter for each individual firm. It is a statistical model that represents the observed quantities in terms of explanatory variables that are treated as if the quantities were non-random. The intercept therefore includes a subscript i, as noticed in the equations of the empirical specification, indicating that it is firm-specific. This is the case since every listed company in the sample has firm-specific, time-invariant characteristics. Furthermore, robust standard errors are clustered at the firm level.

(24)

using TAi.t, this indicates that corporate tax avoidance activity has no effect on firm value. However, the aim of the first hypothesis is not to test the effect of corporate tax avoidance activities on firm value per se but rather about the effect of the interaction between corporate tax avoidance activities and the quality of corporate governance. The quality of corporate governance is resembled by the level of institutional ownership and, as explained in section 2, a key determinant for measuring the engagement in corporate tax avoidance activities (Desai et al., 2007; Chen et al., 2010). Therefore the first hypothesis is tested by inserting equation (9), which is reported in table 3, column 2. Here, the coefficient on the interaction between the book-tax gap value and institutional ownership is positive, β4 > 0, and highly significant. This implies that, the overall effect of corporate tax avoidance activities on firm value is indistinguishable from zero, and that the effect of corporate tax avoidance activities on firm value is larger for firm-years in which institutional ownership is higher (and corporate governance is stronger). This is consistent with the first hypothesis that well-governed firms (i.e. high institutional ownership) have a larger positive effect on corporate tax avoidance activities to increase the value of the firm. Thus, the agency problems which arise in poorly-governed firms mitigate the benefits to shareholders when engaging in corporate tax avoidance activities.

Additionally, this finding is checked by running equation (7) separately on the well-governed and poorly-governed subsamples (table 3, column 3 and 4). The distinction between well-governed and poorly-well-governed firms is defined as the mean fraction of the institutional ownership variable. Where, well-governed firms are those firms for which `high` institutional ownership exceeds the sample mean of I>0.14. Table 3, column 3 indicate the results for poorly-governed firms. The book-tax gap value has a small positive effect on firm value, however, statistically insignificant. Colum 4 shows that the effect of corporate tax avoidance activities on firm value is positive, of greater magnitude then in column 3, and of borderline significance for well-governed firms. This also contributes in explaining the first hypothesis, which states that the effect for well-governed firms would be larger relative to the effect for poorly-governed firms.

(25)

the first hypothesis explaining that the effect of corporate tax avoidance on firm value will be larger for well-governed firms relative to the effect for less well-governed firms.

Table 3: Corporate tax avoidance, Firm value, and Governance institutions Dependent Variable:

TobinsQ TobinsQ TobinsQ TobinsQ

(1) (2) (3) (4)

All firms All firms Poorly-Governed Firms

Well-Governed Firms

Book-Tax Gap (scaled) 0.0859 -0.0343 0.0386 0.256*

(0.0749) (0.0757) (0.0676) (0.132) Total Accruals (scaled) 1.137*** 1.135*** 0.634*** 1.298***

(0.134) (0.133) (0.128) (0.0930)

Institutional Ownership (fraction) -0.0333 (0.0280) Book-Tax Gap interacted with

Institutional Ownership 1.014*** (0.361) Return On Assets (%) -0.0118*** (0.00222) -0.0118*** (0.00220) -0.00655*** (0.00248) -0.0126*** (0.00330) Financial Leverage (ratio) 1.363*** 1.369*** 1.249*** 1.353*** (0.0553) (0.0546) (0.0757) (0.0580)

Capital Expenditures (scaled) 0.672* 0.672* 1.049 0.0111

(0.363) (0.364) (0.696) (0.787)

Growth Rate (%) 0.000453 0.000439 0.000298 4.22e-05

(0.000330) (0.000330) (0.000290) (0.000576)

Firm Size (scaled) -26.26*** -27.50*** -26.88*** -13.34

(8.731) (8.609) (9.128) (12.53)

(26)

In order to test the second hypothesis, the sample is divided in half and indicating a 3-year pre-crisis period (2005, 2006, 2007) followed by a 3-year period of crisis (2008, 2009, 2010). The literature, as explained in section 2, showed that a crisis affected the quality of corporate governance institutions and had an impact on the value of a firm (Mitton, 2002; Lemmon & Lins, 2003). Firm value tends to decrease during a period of crisis, profits diminishes and managers might try alternative methods to create revenue. This could imply a further engagement in corporate tax avoidance activities. Therefore the second hypothesis states that corporate tax avoidance activities for well-governed firms should have a larger effect on firm value during crisis years then pre-crisis years. To further examine this hypothesis, a distinction between poorly-governed and well-governed firms is inferred. The results are given in table 4.

Column 1 and 2 identify the interaction effect of the book-tax gap value with institutional ownership for pre-crisis and crisis years. Column 1 provides consistent results with the first hypothesis stating that under normal conditions (i.e. no crisis), the interaction variable is positive and highly significant indicating that the effect of corporate tax avoidance activities on firm value is larger for firm-years in which institutional ownership is higher (and corporate governance is stronger). Column 2, which only observes the period of crisis, does not find a significant effect on the interaction variable.

In column 3 and 4 the distinction is made between poorly-governed and well-governed firms for pre-crisis years. The results show that for well-governed firms (high institutional ownership) the effect of engaging in corporate tax avoidance activities is of borderline significance, positive, and of greater magnitude then in poorly-governed firms. Which is in line with the first hypothesis.

(27)

Table 4: Tax Avoidance, Firm Value and Governance Institutions: pre- and within- Crisis

Note: The dependent variable is Tobin’s q, as defined in section 3. The sample over the period 2005-2010 is drawn from the Thomson Reuters Database and is restricted for firm-years in which data on S&P500 listed firms was available. Companies with missing data and potential outliers are restricted from the model, comprising 221 participating firms. Pre-crisis is defined as the firm-years 2005, 2006, 2007 where crisis-years are denoted for the firm-years 2008, 2009, 2010. Column 1 and 2 includes the interaction terms for the Book-tax gap value with the dummy for institutional ownership. In column 3 and 5, the sample is restricted to firm-years with institutional ownership < 0.14 (the mean fraction of Institutional Ownership). In column 4 and 6, the sample is restricted to firm-years with institutional ownership > 0.14.Robust standard errors that are clustered at the firm level are presented in parentheses; *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively.

Dependent Variable

TobinsQ TobinsQ TobinsQ TobinsQ TobinsQ TobinsQ

(1) Pre-Crisis (2) Crisis (3) Pre-Crisis Poorly-Governed Firms (4) Pre-Crisis Well-Governed Firms (5) Crisis Poorly-Governed Firms (6) Crisis Well-Governed Firm

Book-tax gap (scaled) -0.208** 0.0190 -0.102 0.182* 0.203** -0.194

(0.103) (0.0989) (0.0796) (0.104) (0.0957) (0.192)

Total accruals (scaled) 0.552*** 0.717*** 0.128 0.313*** 0.377** 0.932***

(0.201) (0.0939) (0.140) (0.110) (0.151) (0.0463) Institutional Ownership (fraction) -0.0376 (0.0443) 0.0329 (0.0554) Book-Tax Gap interacted with Institutional Ownership 1.125*** (0.356) 0.794 (0.617) Return On Assets (%) -0.00371 -0.00865*** 0.000137 -0.00127 -0.00750* -0.00631*** (0.00244) (0.00170) (0.00135) (0.00177) (0.00391) (0.000902) Financial Leverage (ratio) 1.413*** (0.0675) 1.317*** (0.130) 1.295*** (0.0820) 1.446*** (0.0564) 1.428*** (0.178) 1.268*** (0.107) Capital Expenditures (scaled) -0.110 (0.278) 1.683*** (0.440) -0.264 (0.288) 0.178 (0.400) 1.330* (0.692) 1.353*** (0.311) Growth Rate (%) 1.97e-05 0.00109** 7.11e-08 0.000498*** 0.00116** 4.68e-06

(0.000263) (0.000430) (0.000286) (0.000170) (0.000478) (0.000172) Firm Size (scaled) -29.15** -50.87*** -19.48** -27.97** -48.92** -12.81

(28)

An explanation for this phenomenon is found in the reasoning behind the second hypothesis. Mitton (2002) indicated that poorly-governed firms have lost relatively more value in comparison to well-governed firms during a period of crisis. This was explained through the ease for well-governed firms to derive revenue from alternative sources, for example corporate tax avoidance activities. Since well-governed firms are better able to mitigate the effect of agency problems that complicate the engagement in corporate tax avoidance activities (Desai & Dharmapala, 2009). However, not well-governed but poorly-governed firms are better able to derive revenues from engaging in corporate tax avoidance activities during crisis-years10. A threat of losing firm value, potential bankruptcy, firing employees, or a loss in the salaries of managers might resulted that managers (in poorly-governed firms) became more willingness to engage in corporate tax avoidance activities to enhance firm value. So, agency problems that complicate the engagement in corporate tax avoidance activities (i.e. hypothesis one) do not seem to arise in poorly-governed firms during crisis years.

There has not been any research conducted on the effect of corporate tax avoidance activities during a period of crisis. The finding is remarkable considering the existing literature about corporate tax avoidance activities, which explains that the effect of corporate tax avoidance activities on firm value is larger for well-governed firms (i.e. Mitton 2002; Desai & Dharmapala 2005; 2009; Chen et al. 2010). Therefore the observation that poorly-governed firms significantly benefit from corporate tax avoidance activities during a period of crisis provides a new finding in the emerging paradigm of the existing literature (Desai & Dharmapala, 2006; Graham & Tucker, 2006; Desai et al., 2007, Desai & Dharmapala, 2009) and opens up an inquiry for future research11.

4.4 Alternative measure of earnings management

The results of Table 3 and 4 are partly consistent with both hypotheses, including the economic reasoning beyond these findings. However, there are a number of possible measurement errors with the variables that has to be taken into consideration to test similar results. Following the research of Lev & Nissim (2004) and Hanlon (2005) that earnings management contributes to the book-tax gap value. Total accruals (TAi,t) is used as a proxy for earnings management in the equation. However, the measurement of total accruals could

10_{At least in this paper`s constituted sample period of 221 participating firms for the crisis-years 2008, 2009,} and 2010.

(29)

be incomplete, where the remaining component of the book-tax gap value may be mischaracterized as tax avoidance while it actually represents earnings management. Therefore alternative measures of total accruals will be taken into consideration. Two alternative measures, named `abnormal` or `discretionary` accruals, have been developed by Jones (1991) and Dechow et al. (1995) to better isolate the components of accruals that are truly under managerial control. Both measurements follow the work of DeAngelo (1986) who used total accruals from a prior period as a measure of total accruals. She defined `abnormal` total accruals as the difference between current total accruals and normal total accruals. Which can be separated into discretionary, the non-obligatory expenses such as bonuses for management, and nondiscretionary accruals, such as any upcoming bills or next month's salary. Jones (1991) assumes that the difference between current- and prior-year accruals is due solely to changes in discretionary accruals because nondiscretionary accruals are assumed to be constant from period to period. Implying that managers have little opportunity to manipulate the nondiscretionary accruals. Computing discretionary accruals (DAi,t) is based on the assumption that the change in revenue from sales is nondiscretionary; it involves regressing TAi,t on the change in sales and the value of plant, property and equipment, with each variable being scaled by the book value of assets (Jones, 1991)12. The residual (ɛt) represents the firm-specific discretionary portion of total accruals. Dechow et al. (1995) modifies this original Jones model (1991) by assuming that managers may exercise some discretion in determining the value of credit sales. Therefore refining the model by excluding the change in credit sales, by deducting the change in accounts receivable, from the nondiscretionary component of accruals. The discretionary part of the total analysis is again the error component13.

Table 5, column 1 and 2, indicate that both alternative measurements of earnings management are also highly significant and that the interaction between the book-tax gap value with institutional ownership is also significant and positively indistinguishable from zero (β4 > 0). As explained in section 2, many different measures for earnings management are described in the existing literature, all providing different calculations and indicators of accruals accounting (Healy, 1985; DeAngelo, 1986; Jones, 1991; Dechow et al., 1995; Bartov et al.,

12_{Denoted as follows: Discretionary Accruals}

= ɛt = TAit/Ait - αi [l / Ait-1] + β1i [ ∆REV / Ait-1] + β2i [ PPEit / Ait-1] Where TAit = total accruals in year t for firm i; ∆REVit = revenues in year t less revenues in year t - 1 for firm i; PPEit = gross property, plant, and equipment in year t for firm i; Ait-1 = total assets in year t -1 for firm i; ɛt = error term in year t for firm i.

13_{Denoted as follows: Discretionary Accruals}

(30)

2001). Nevertheless, inferring two different measurements of accruals provide consistent results with the outcomes in table 3 and 4. Therefore using total accruals, which also includes the nondiscretionary part of accruals, is a reliable observation to observe the institutional ownership structure of the firm for the full sample period.

Table 5: Tax avoidance, Firm value, and Governance institutions: Alternative measures of Earnings Management

Dependent Variable

TobinsQ TobinsQ

(1) (2)

Book-Tax Gap (scaled) 0.133

(0.118)

0.133 (0.119) Discretionary Accruals (scaled) 0.444***

Jones Model (0.133)

Discretionary Accruals (scaled) 0.424***

Modified Jones Model (0.132)

Institutional Ownership (fraction) -0.0378 -0.0416

(0.0352) (0.0351)

Book-Tax Gap interacted with Institutional Ownership 1.206* (0.613) 1.249** (0.623) Return On Assets (%) -0.00862* -0.00889*

Financial Leverage (ratio)

(0.00490) 1.363***

(0.00501) 1.351***

(0.0890) (0.0958)

Capital Expenditures (scaled) 3.075** 3.165**

(1.233) (1.270)

Growth Rate (%) 0.00271*** 0.00244***

(0.000533) (0.000477)

Firm Size (scaled) -67.16** -69.54**

(28.93) (29.55)

Foreign Income (scaled) 0.597 0.608

(0.410) (0.414)

Observations 1,072 1,072

R-squared 0.585 0.581

Number of Firms 220 220

(31)

4.5 Additional Robustness checks

The results in table 3 and 4 are based on the construction of the book-tax gap outlined in section 3. However, several factors may lead to an incorrect measurement of tax avoidance activities. A particular important feature here is the role of future tax liabilities (Desai & Dharmapala, 2005). This paper follows the book-tax gap measurement as it was constructed by Manzon and Plesko (2005) and implemented by Desai and Dharmapala (2009). Therefore the measure of firm value is excluded for deferred tax expenses. Table 6, column 1, tests for any differences in the book-tax gap using the standard calculation of TobinsQ (La Porta et al. 2002; Doidge et al. 2004; Gozzi et al. 2006) and provides consistent results with table 3. Furthermore, the specifications implemented in obtaining the results from table 3 use a number of control variables that may affect firm value. Desai and Dharmapala (2009) argue that a measurement for potential Net Operating Loss (NOL) carryforwards has to be taken into consideration. Since NOL carryforwards could affect the incentives to engage in corporate tax avoidance activities (Desai & Dharmapala, 2009). NOL carryforwards is an accounting technique that applies the current year’s net operating losses to future years profits in order to reduce tax liability14. U.S. GAAP specifies that loss carryforwards can be used in any of the seven years following the occurred loss. NOL will be measured by net sales minus all the costs of goods sold and other capital expenditures and scaled by the book value of assets. Table 6, column 2, shows that adding NOL carryforwards into the equation does not generate a significant impact on firm value and that the interaction term for corporate tax avoidance activities with the institutional ownership measurement is not affected and remains positive and significant.

Desai and Dharmapala (2009) specify an interaction variable between institutional ownership and corporate tax avoidance. A positive coefficient (β4 > 0) indicate that the effect of corporate tax avoidance activities on firm value is larger for firm-years in which institutional ownership is higher (and corporate governance is stronger). In an earlier paper, Desai and Dharmapala (2005) use a different interaction variable to infer the same effect of corporate tax avoidance activities on firm value. They defined the dummy variable WELLGOVi to partition the sample of firms into those that are relatively well-governed (Iit > 0.14) and those that are poorly-governed (Iit < 0.14). This is similar to the article of Desai and Dharmapala (2009), however, they interacted the book-tax gap value with the dummy variable for

(32)

governed firms instead of institutional ownership. This paper followed the work of Desai and Dharmapala (2009), however, a positive and of borderline significant effect is also found in table 6, column 3, when implementing the interaction variable as constituted by the earlier work of Desai and Dharmapala (2005).

Table 6: Tax avoidance, Firm Value and Governance Institutions: Tax Avoidance Robustness Checks

Note: The dependent variable is Tobin’s q, as defined in section III. The sample over the period 2005-2010 is drawn from the Thomson Reuters Database and is restricted for firm-years in which data on S&P500 listed firms was available. Companies with missing data and potential outliers are restricted from the model, comprising 221 participating firms. Robust standard errors that are clustered at the firm level are presented in parentheses; *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively

Dependent Variable

TobinsQ TobinsQ TobinsQ

(1)

Excluding deferred taxes

(2) (3)

Book-Tax Gap (scaled) -0.0242 -0.0316 0.0351

(0.0667) (0.0731) (0.0770)

Total Accruals (scaled) 0.929*** 1.135*** 1.135***

(0.0889) (0.140) (0.134)

Institutional Ownership (fraction) -0.0277 -0.0334 0.0335

(0.0231) (0.0277) (0.0358)

Book-Tax Gap interacted with Institutional Ownership

0.474* (0.242)

1.006*** (0.349) Book-Tax Gap interacted with a

dummy for Well-Governed Firms

0.220* (0.123)

Net Operating Loss (scaled) -0.0602

(0.0512)

Return On Assets (%) -0.00873*** -0.0116*** -0.0116***

(0.00166) (0.00233) (0.00221)

Financial Leverage (ratio) 1.324*** 1.382*** 1.364***

(0.0556) (0.0568) (0.0551)

Capital Expenditures (scaled) 0.714* 0.583* 0.691*

(0.379) (0.332) (0.363)

Growth Rate (%) 0.000455* 0.000427 0.000443

(0.000274) (0.000321) (0.000327)

Firm Size (scaled) -28.65*** -23.92*** -26.91***

(7.934) (7.511) (8.497)

Foreign Income (scaled) -0.0507 0.109 0.0567

(0.147) (0.181) (0.179)

Observations 1,072 1,072 1,072

R-squared 0.835 0.821 0.820

(33)

Overall, inferring for the standard calculation of TobinsQ, adding an additional control variable into the equation, or constituting a different interaction effect with the book-tax gap value leads to results that are generally consistent with those in table 3. Therefore, the constituted calculations, as explained in section 3, provided a reliable and solid measurement for investigating the effect of corporate tax avoidance activities on firm value with the moderating effect for the quality of corporate governance.

5. | CONCLUDING REMARKS

5.1 Conclusions

The aim of this paper is to analyze the effect of corporate tax avoidance activities on the profitability of U.S. firms listed in the S&P 500, with distinctive emphasis for the 2008 financial crisis.

The simple presumption that corporate tax avoidance activities result in a transfer of value from the state to shareholders has not appeared to be validated in the data. In the full sample, tax avoidance activities by firms has not lead to an automatic increase in firm value (table 3, column 1). The pattern found in the data is more consistent with the agency perspective on corporate tax avoidance activity with emphasis on the mediating role of the quality of corporate governance. The main finding of the first hypothesis is that for well-governed firms the effect of corporate tax avoidance activity on firm value is larger than for poorly-governed firms. The interaction of the quality of corporate governance with corporate tax avoidance is positively indistinguishable from zero and highly significant (table 3, column 2). Where also the estimated effects of corporate tax avoidance on firm value has a stronger effect for well-governed firms then for poorly-well-governed firms (table 3, column 4). These findings contribute to the work of Slemrod (2004) that incorporating agency issues into the analysis of corporate tax avoidance leads to theoretical and empirical conclusions that are substantially different from those that would be predicted by a model where managers are perfect agents. Furthermore, it contributes to the emerging literature on the effects of corporate tax avoidance activities on firm value (Desai & Dharmapala, 2005; 2009).

(34)

engage in corporate tax avoidance activities. Therefore the 2008 financial crisis has not further stimulated well-governed firms to engage in corporate tax avoidance activities.

However, it did find a positive significant effect for poorly-governed firms during crisis-years. Agency problems, that arise in poorly-governed, firms do not have complicated the engagement in corporate tax avoidance activities during crisis-years. This finding shed new light on the effect of tax avoidance activities during a period of crisis. It contributed to the work of Weisbach (2002) who asked why firms do not engage in tax sheltering more extensively, given the widespread availability of shelters and the low risk of penalties. Apparently, the 2008 financial crisis stimulated poorly-governed firms to engage more in corporate tax avoidance activities to generate off-the-book revenue.

5.2 Limitations

The analysis discussed in the paper suggests that moderating for the quality of corporate governance has an important effect on the relation of corporate tax avoidance activities and firm value. However, there could be some alternative explanations for the link between the valuation of tax avoidance and firm governance. Three alternative explanations seem particularly relevant.

First, it is possible that the differences in valuation of tax avoidance between well-governed and poorly-governed firms relate to differences in the types of tax shelters used by these firms (Desai & Dharmapala, 2005). It could be possible that the smaller effect for poorly-governed firms can be explained due to those firms investing in riskier shelters that are discounted at higher rates. Nevertheless this seems unlikely, given the wide availability of tax shelters with very high returns (Slemrod, 2004).

Corporate tax avoidance and firm value in the light of the 2008 financial crisis