Female representation in top management and (aggressive) corporate tax avoidance

(1)

Female representation in top

management and (aggressive)

corporate tax avoidance

Manon Boots 10194436

Date: June 22, 2015 Word count: 14,136

MSc Accountancy & Control, variant Accountancy Amsterdam Business School

Faculty of Economics and Business, University of Amsterdam Supervisor: dhr. dr. W.H.P. Janssen

(2)

2

Statement of Originality

This document is written by student Manon Boots 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.’

(3)

3

Abstract

This study examines the effect of female representation in the top management of an

organization on (aggressive) corporate tax avoidance, and, supplementary, whether this effect (if any) is stronger for organizations operating in the retail industry than for organizations operating in industries other than the retail industry. The sample consists of S&P 1500 companies in the period 2000 to 2011. The empirical findings indicate that for the broad spectrum of corporate tax avoidance there is a negative association between firms that have one or more female executives in their top management team (however, the findings seem to be driven by those firms having one female executive in their top management) and corporate tax avoidance. There is no evidence that such a negative association exists for firms with a top management team that consists of two or more female executives, and that the found

relationship is more negative for firms operating in the retail industry as compared to the firms operating in other industries. Further, the empirical findings indicate that for the aggressive form(s) of corporate tax avoidance there is no association between female representation in the top management of an organization and aggressive corporate tax avoidance. However, for retail firms, as compared to the firms operating in industries other than the retail industry, that have a top management team that consists of one or more female executives, there is a negative association with aggressive corporate tax avoidance. In general, the results are contradictory to prior studies that investigated the effect of gender diversity on corporate tax ‘reduction’, and suggest that more risk-tolerant women self-select into the managerial professions. In situations where female executives perceive the risks and benefits of engaging in (aggressive) corporate tax avoidance activities to be close together (risk(cost)-benefit trade-off), it could be the case that only when they have the support of another or other female executive(s) in the top management team they actually dare to take the risks, implying that even though more risk-tolerant women self-select into the managerial professions, they are not at the complete risk-taking level as their male counterparts.

(4)

4

1 Introduction

Corporate tax avoidance is a common phenomenon, especially in the Netherlands where companies seem to jeopardize the Dutch favorable corporate tax system (because of the existence of regulations that prevent double taxation), that received an increased attention of policymakers throughout the years. As such, policymakers around the world have either enacted or are considering regulations relating to (multinational) firm’s tax avoidance behavior. An example of an enacted regulation is Schedule M-3, which came in effect in the United States for tax years ending on or after December 2004. The overall goal of Schedule M-3 is to increase transparency in reporting data by requiring companies to provide more detailed information into their tax reporting statements (Boynton, DeFilippes and Legel, 2008). Greater transparency in reporting data will, in turn, increase the likelihood to reveal tax avoidance behavior.

More recent, compared to Schedule M-3, is the OECD/G20 BEPS (Base Erosion and Profit Shifting) Project, which brought the topic corporate tax avoidance again vividly under society. Just as Schedule M-3 is the OECD/G20 BEPS Project generated to reduce the likelihood of tax avoidance behavior: in order to ensure that profits are taxed in the

country/place they should legally be taxed, that is, where the economic activities generating the profits are performed and where value is created, fifteen actions, containing the tools countries need to fulfil this task, are (currently) developed (completed in December 2015) (OECD, 2014).

The increased attention on the topic corporate tax avoidance not only applies to policymakers. It also applies to economic researchers. The corporate tax avoidance literature is an upcoming, although young, stream of research (Hanlon and Heitzman, 2010). Here, most of the existing studies have focused on the determinants and consequences of corporate tax avoidance. To mention a few, by looking at the determinants of corporate tax avoidance, Hope, Ma and Wayne (2013) have found a negative association between corporate tax avoidance and geographic earnings disclosure: in the U.S., after the adoption of SFAS 131, tax avoidance behavior increased for companies that, from that moment, chose to no longer disclose geographic earnings in their financial statements. Such a negative relation can also be found between the existence of board members in an organization with at least one tax-related affiliation and corporate tax avoidance, resulting from the study of Taylor and Richardson (2014), investigating the incentives of corporate tax planning and reporting in an Australian context.

(6)

6 Next to the growing corporate tax avoidance literature does the inclusion of gender diversity into economic research grow, mainly because of the increase (even it is a slow increase) of women in ‘top positions’ of organizations that took place the last 15 years (Catalyst, 2014). In the psychology literature, much is written about the difference in risk-taking behavior/attitudes of men and women, and recently, studies have begun to

examine the effect of this difference on ‘economic phenomena’. For example, Hili and Affes (2012) investigate, in the case of French listed companies, the impact of gender diversity of the board of directors on earnings persistence, a proxy for earnings quality. They found no significant differences between firms with male and female directors sitting on the board. Peni and Vähämaa (2010) examine, in an American setting, the relation between gender diversity of the CFO of an organization and its effect on earnings management. They conclude from their research that female CFOs can be associated with higher income-decreasing

discretionary accruals, thereby implying that female CFOs are more conservative than male CFOs. Also the study of Thiruvadi and Huang (2011) is worth mentioning here. They investigate whether the presence of a gender diverse audit committee is associated with an increase or decrease in a firm’s earnings management. Thiruvadi and Huang (2011) find evidence that, just as with the previous mentioned study, there is a positive relation between the presence of a female director on an audit committee and income-decreasing accruals: a reduction takes place in the level of a firm’s engagement in earnings management activities.

Going further, combining the previous two topics (corporate tax avoidance and gender diversity), the relation between gender diversity and corporate tax ‘reduction’ (see the literature review section, especially paragraph 2.1, for the meaning of ‘reduction’) has been examined. In this setting, there are 2 studies. Their results are mixed. Khaoula and Ali (2012) investigate, in an American context, whether gender diversity of the board of directors of an organization affect the level of corporate tax planning; they found no effect. In contrast, Francis, Hasan, Wu and Yan (2014) investigate whether female CFOs are less tax aggressive than male CFOs. From their research, Francis et al. (2014) conclude that female CFOs are indeed less tax aggressive compared to their male counterparts. Moreover, by applying a similar research methodology as in the case of aggressive corporate tax avoidance, Francis et al. (2014) examine the relation between the difference in gender of the CFO of an

organization and corporate tax avoidance. In this case, they found no distinction between female and male CFOs.

This paper builds on and extends the previous 2 mentioned studies that examine the effect of gender diversity on corporate tax ‘reduction’, by taking a different point of view as

(7)

7 compared to those studies : a ‘top management team’ point of view. Besides, I will add a new dimension by differentiating between industries: there will, supplementary, be a distinction (and comparison) made between retail firms and non-retail firms (see paragraph 2.1 and paragraph 2.5 for a more detailed explanation). The research question I compiled is the following: What is the effect of female representation in the top management of an

organization on (aggressive) corporate tax avoidance? – An investigation of listed companies in the US from 2000 to 2011, with a supplemental focus on the retail industry. In the research

question, the term aggressive is placed between brackets because I will look to the broad concept of corporate tax avoidance, as well as aggressive corporate tax avoidance (see paragraph 2.5and methodology section).

By examining the research question What is the effect of female representation in the top

management of an organization on (aggressive) corporate tax avoidance? – An investigation of listed companies in the US from 2000 to 2011, with a supplemental focus on the retail industry I will contribute to existing literature in three ways. First, as said before, the

corporate tax avoidance literature is very young (Hanlon and Heitzman, 2010), but is of growing interest (especially under policymakers). This interest is about the magnitude, determinants and consequences of corporate tax reporting, and calls for more detailed examination on, among others, the impacts of (individual) executives on corporate tax avoidance. This study responds to these calls. Second, not only corporate tax avoidance is of growing interest, also the inclusion of gender diversity into economic research grows. The relation between gender diversity and corporate tax ‘reduction’ has been examined. However, research in this setting is scarce. In giving a short summarization: Khaoula and Ali (2012) investigate the impact of gender diversity of the board of directors of an organization on the level of corporate tax planning, while Francis et al. (2014) examine whether there is a

difference in the tax aggressiveness of female CFOs and their male counterparts. In contrast to the previous, my study will be the first study that takes a ‘top management team’ point of view, in this setting. Does female representation in the top management of an organization in general, who set the tone at the top of an organization instead of having a direct link to corporate tax activities as in the case of the CFO of an organization, affect the level of (aggressive) corporate tax avoidance? The outcome may have implications for, for example, policymakers: in the Netherlands there is an ongoing debate about the introduction of a quota for women in ‘top positions’ (30%-40%, already in effect in Germany) (Waaijers, 2014). Research that examines the effect of executive gender diversity on ‘economic phenomena’ can contribute to the debate. If it proofs that female executives have positive impacts, it can

(8)

8 result in the common acceptance of such a quota, which will facilitate the introduction of it (societal contribution). Third, according to Hanlon and Heitzman (2010), corporate tax avoidance is hard to define and measure. Several studies are not clear in which construct of corporate tax ‘reduction’ (tax avoidance, tax evasion and tax planning) they investigate, and across studies the terms are used interchangeably. By researching and explaining the concept, and moreover, through a consistent use of one term and a clear explanation of the reasons behind the kind of measurement choice of (aggressive) corporate tax avoidance, this research

could contribute to a better understanding of the definition and measurement of corporate tax

avoidance.

The results indicate that for the broad spectrum of corporate tax avoidance there is a negative association between firms that have one or more female executives in their top management team and corporate tax avoidance. There is no evidence that the relation is more negative for firms operating in the retail industry as compared to those firms operating in other industries, and that such a negative relationship also exists between firms with a top management team that consists of two or more female executives and corporate tax

avoidance. Further, the empirical findings indicate that for the aggressive form(s) of corporate tax avoidance there is no association between female representation in the top management team of an organization and aggressive corporate tax avoidance. However, for retail firms, as compared to those firms operating in industries other than the retail industry, that have a top management team that consists of one or more female executives, there is a negative

association with aggressive corporate tax avoidance.The results suggest that, in general, more risk-tolerant women self-select into the managerial professions.

The structure of the paper is as follows. The next paragraph describes the different constructs of corporate tax reduction, reviews prior studies on the impact of top management team members on corporate tax avoidance and on the relation between gender diversity and (aggressive) corporate tax avoidance, and develops the hypotheses. Paragraph 3 describes the sample selection procedures and the research method. The results of the empirical tests are expressed in paragraph 4. Finally, paragraph 5 concludes the results, complemented with some suggestions for future research.

(9)

9

2 Literature review

The literature review consists of 5 subparagraphs. In the first subparagraph, paragraph 2.1, a description, derived from various definitions, of corporate tax avoidance, is given. This is followed by an explanation of the different constructs of corporate tax reduction and a discussion of the risks (aggressive) corporate tax avoidance entails. In paragraph 2.2 some studies that investigated the impact of top management team members on corporate tax avoidance are outlined. A more detailed description of the studies of Khaoula and Ali (2012) and Francis et al. (2014) takes place in paragraph 2.3. Paragraph 2.4 is marked by the theory of gender diversity. Finally, in paragraph 2.5 the hypotheses are developed.

2.1 Corporate tax avoidance: definition and risks

In the corporate tax literature, different definitions of corporate tax avoidance come forward (the most broadly: ‘the reduction of explicit taxes’ (Hanlon and Heitzman, 2010, p. 137)): corporate tax avoidance is hard to define. Next to this, the terms tax avoidance, tax evasion and tax planning (constructs of corporate tax reduction) are often used interchangeably in the corporate tax literature. Even though their objective is the same (reduction of payable taxes), in technical sense they differ from each other. Making a distinction between them, done hereafter, is important for measuring the level of tax avoidance, tax evasion and tax planning: corporate tax reduction can be measured in many ways, however not all measures/proxies are appropriate in capturing all of the different constructs.

To begin with corporate tax avoidance (focused on in this research), derived from different definitions, it basically means the following: in a legal way reducing the payment of taxes, tax liability, by making use of methods that are not prevented nor approved by the tax law, i.e. loopholes. In this context, someone can think of the distribution of profits from countries marked by high effective tax rates to countries with lower effective tax rates (Cai and Liu, 2009). Another example is the grant of stock options to the executives of an organization. Until the ‘Cut Unjustified Loopholes Act’ of February 2013, an organization could provide their executives with stock options without reducing their profit reported in the financial statements at the one hand, while the organization at the other hand could take a tax deduction at an amount equal to the difference between the price of the stock option and the

(10)

10 value of the option upon exercise, thereby causing a book-tax difference (Citizens for Tax Justice, 2014). To refer to the most aggressive (although legal) forms of these corporate tax avoidance activities (activities at the end of the corporate tax avoidance spectrum), sometimes the term aggressive corporate tax avoidance (also tax aggressiveness) is used (Hanlon and Heitzman, 2010).

Going further to tax evasion, the distinction between tax avoidance and tax evasion is characterized by illegality. This comes forward in the following citation: ‘Evasion is

concerned with concealing or misrepresenting the nature of a transaction; when avoidance takes place the facts of the transaction are admitted but they have been arranged in such a way that the resulting tax treatment differs from that intended by the relevant legislation.’

(Kay, 1980, p. 136). In other words, tax evasion implies the reduction of tax payments (tax liability) by circumventing the law; a company escapes the payment of taxes. This can be done in many ways. One example is to intentionally underreport the profits presented in the income statement, thereby falsifying the income statement towards tax authorities.

The distinction between tax avoidance and tax planning is less clear than the

distinction between tax avoidance and tax evasion. As mentioned, tax avoidance implies the reduction of the tax liability by making use of loopholes in the tax law. In the case of tax planning, tax liability is reduced by using provisions (e.g. tax deductions; payments to

pension schemes) given by the law (Merks, 2006). Both are legal. However, in the case of tax avoidance, a tax reduction is obtained that originally is/was not the intention of the

government.

Looking to (aggressive) corporate tax avoidance (and tax evasion), it brings costs and benefits. A research of the Citizens for Tax Justice (2014) to the use of loopholes in and circumvention of the tax law by 288 profitable Fortune 500 companies between 2008 and 2012 reveals that over the time period 2008-2012 the whole group paid a 19.4% effective federal income tax rate, which is slightly more than half of the statutory 35% tax rate. During the period 2008-2012 there were five companies (AT&T, General Electric, IBM, Merizon and Wells Fargo) who relished a ‘gain’ of 77 billion by circumventing the tax law. (Aggressive) corporate tax avoidance comes at the cost of society and at the benefit of tax avoiders (by reducing the tax liability, after-tax income will be higher). However, these benefits don’t come without risks. Risks (indirect costs for tax avoiders) are associated with corporate tax avoidance, especially when it comes to aggressive corporate tax avoidance (tax

(11)

11 aggressiveness). Two risks one can think of, are: the incurrence of political costs and the risk of image loss (reputational costs).

The political cost hypothesis states that if high profits causes political costs, managers are more likely to shift income from the current period to future periods (Watts and

Zimmerman, 1978). In this context, Mills, Nutter and Schwab (2013) investigate whether there is an association between politically-sensitive contractors and the payment of higher taxes. They find evidence that this is the case: higher effective tax rates were observed for politically-sensitive contractors; contractors that are highly sensitive to political costs. The view that (aggressive) corporate tax avoidance imposes political costs on firms is

substantiated by this.

The second risk factor, image loss, is probably the biggest concern companies have nowadays. Both the firm as a whole as the managers of a firm can be exposed to image loss. It may result in reputational costs, which can cause negative stock price reactions, demonstrated by Hanlon and Slemrod (2009). Hanlon and Slemrod (2009) study the relationship between aggressive corporate tax avoidance and stock price reactions (resulting after a news

announcement that associates a company with aggressive corporate tax avoidance: having a tax shelter). In their hypothesis section they state that firms with a higher Cash ETR have a more positive, or less negative, stock price reaction upon the news announcement, because having a high Cash ETR leads to the market assuming that such a firm is sufficiently tax aggressive. In this case, the news announcement will be seen as a positive signal of optimal aggressive tax avoidance. Besides, Hanlon and Slemrod (2009) hypothesize that the stock price reaction is more negative for firms in the retail industry (because of their direct link with consumers) compared to firms in other industries. On average, Hanlon and Slemrod (2009) find evidence that news about firms engaging in aggressive corporate tax avoidance results in a stock price decline. This decline is more negative for firms operating in the retail sector.

In contrast to Hanlon and Slemrod (2009), Austin and Wilson (2013) investigate whether firms with a valuable consumer reputation can be associated with less (aggressive) tax avoidance behavior. So, Austin and Wilson (2013) study the same relationship (aggressive tax avoidance and reputational costs) as Hanlon and Slemrod (2009) did, but then the other way around: ex-ante instead of ex-post. By using Harris Interactive’s EquiTrend survey, Austin and Wilson (2013) identified a set of firms designated by consumers as having valuable brands. They found evidence for their hypothesis that those firms have higher effective tax rates, implying that firms with valuable brands (valuable consumer reputation) engage in lower levels of tax avoidance.

(12)

12 2.2 Top management team member impacts on corporate tax avoidance

At first sight, someone might think that individual top executives can’t affect the level of corporate tax avoidance, because they are not directly involved (with the exception of the CFO) in the tax activities of an organization. However, abstracted from prior research it can be stated that this thought is not correct (e.g. Dyreng, Hanlon and Maydew, 2010; Gaertner, 2014)

The top management team of an organization consists of the CEO of the organization and the senior executives that have the responsibility over one or more functional areas in that organization (Menz, 2012). Senior executives that can be imagined, are: the chief financial officer (CFO), the chief operating officer (COO), the chief marketing officer (CMO), the chief information officer (CIO) and the chief strategy officer (CSO). The top management team members can individually affect corporate tax avoidance, demonstrated by Dyreng et al. (2010). Dyreng et al. (2010) asked themselves the question whether there are other factors (individual top executive effects, due to their tone at the top) than just firm characteristics, on which prior research have mainly focused on, that determine the level of corporate tax

avoidance. To investigate this, they examine whether and how a firms’ tax avoidance level changed after a top executive left the company, and whether this also happened at the time the executive was hired by the company. In this way, firm effects are excluded. In doing this, Dyreng et al. (2010) find evidence that individual top executives have a significant effect on corporate tax avoidance, with the CEO having the biggest impact. On average, they find a change in GAAP effective tax rates of 11%. Moreover, Dyreng et al. (2010) examine whether the mentioned executive effects are due to their biographical characteristics (e.g. background, age) or ‘styles’ (i.e. strategies). From the empirical analyses it appear that there isn’t a

significant connection.

After Dyreng et al. (2010) have shown the significant role individual top executives have in the determination of corporate tax avoidance, but failed to find a relation with the characteristics of the executives in this context, Gaertner (2014) investigates what may drive the effect: he examines the association between managerial incentives, after-tax

compensation, and corporate tax avoidance. In comparing pretax and after-tax compensation with each other, after-tax compensation entails more risk, because managers can’t control taxes (rate). In reducing risks, Gaertner (2014) assumes that the CEO attempts to lower the effective tax rate. Indeed, a negative association between the CEO’s after-tax incentives and

(13)

13 effective tax rate is found, implying that the after-tax compensation of a CEO is a determinant of corporate tax avoidance.

2.3 Prior studies on the effect of gender diversity on corporate tax ‘reduction’

As mentioned and shortly explained, there are 2 studies that investigate the effect of gender diversity on corporate tax reduction: Khaoula and Ali (2012) and Francis et al. (2014). Khaoula and Ali (2012) examine the relationship between gender diversity of the board of directors of American companies and corporate tax planning. They asked themselves the question whether tax planning improves in the case an organization has a diverse board of directors. They consider a board of directors as diverse if 20% (or more) of the directors sitting on the board are female. Based on prior research, Khaoula and Ali (2012) hypothesize that the presence of women in the board of directors, gender diversity, is positively associated with tax planning activities. Next to this (their main research), they look to two other

characteristics of the board of directors, degree of independence and board size, and their effect on corporate tax planning. For the degree of independence, Khaoula and Ali (2012) hypothesize that there is a positive relation between the independence of the board of

directors and corporate tax planning. In contrast, a negative association is expected between board size and tax planning behavior. To test the hypotheses, Khaoula and Ali (2012) use a sample of 300 American companies, abstracted from the S&P 500, for the period 1996-2009. An explanation is given for the choice of making use of the S&P 500 companies: larger firms tend to have higher levels of corporate tax planning. ETR (effective tax rate) is used as a proxy for corporate tax planning, and 2 control variables were added: ROA (return on assets) and firm size. From comparing diversified (heterogeneous) and homogeneous boards of directors with each other, Khaoula and Ali (2012) conclude that gender diversity of the board of directors of an organization doesn’t have an effect on corporate tax planning, which may be caused by the degree of men sitting on the board of directors and their strategies, dominating the female strategies. For the second hypothesis, the results indicate that there is indeed a positive relation between corporate tax planning and the independence of the board of

directors. In contrast, no evidence is found for the third hypothesis: board size isn’t negatively associated with corporate tax planning. This also applies to the firm size of an organization. Firm size seems to doesn’t have any effect on corporate tax planning.

(14)

14 Francis et al. (2014) attempted to examine whether female CEOs are less tax

aggressive than their male counterparts. However, because of a limited sample size, they replaced their focus from the CEO to the CFO of an organization. In their main research, Francis et al. (2014) hypothesize that female CFOs engage in lower levels of aggressive corporate tax avoidance compared to male CFOs. This is due to the attitudes women have towards risk-taking behavior. Women tend to be more risk averse than men. They, however, take into account the probability of the into managerial professions self-selection of more risk-tolerant women (see the next paragraph). If this is the case, no gender effect will be found. Francis et al. (2014) use a sample of S&P 1500 companies, a sample period of 1988-2007, and apply 3 different approaches to test their hypothesis. In the measurement of

aggressive corporate tax avoidance, 3 proxies are used: probability of tax sheltering, predicted unrecognized tax benefits and permanent DTAX (discretionary book-tax differences). After comparing the level of aggressive corporate tax avoidance between the pre- and post-periods of male-to-female CFO transitions (difference-in-differences approach, to rule out the self-selection ‘problem’), between firms with a female CFO and matched firms with a male CFO (propensity score matching), and between the pre- and post-periods of female-to-male CFO transitions, Francis et al. (2014) conclude that female CFOs are indeed less tax aggressive than their male counterparts. They state that this effect can only be found in the case of aggressive corporate tax avoidance, and not in the case of (broad) corporate tax avoidance. More uncertainty (i.e. risk) is involved with aggressive corporate tax avoidance: the risk aversion of women comes forward. Corporate tax avoidance captures not only the uncertain tax avoidance activities but also the less uncertain (less risky) tax avoidance activities, whereby the risk attitudes of men and women are supposed to be indifferent. By applying the same approaches as in the case of aggressive corporate tax avoidance, and the use of ETR, Cash ETR and total book-tax difference (BT) as proxies for corporate tax avoidance, Francis et al. (2014) find evidence for their statement.

2.4 Gender diversity: risk aversion and ethical behavior

In psychology, and later in economics, much is researched and written about the differences between men and women, and especially between their preferences in risk-taking behavior, ethical behavior, and under professionals their leadership style. In the remainder of this

(15)

15 subparagraph, I will explain the first two of the previous three mentioned differences between men and women: preferences in risk-taking behavior and ethical behavior. First, relating to risk-taking behavior, women tend to be more risk averse than men (e.g. Bruce and Johnson, 1994; Powell and Ansic, 1997; Weber, Blais and Betz, 2002). By drawing a sample from the general population, Bruce and Johnson (1994) explore the impact of gender differences on (off-course) betting decisions. To be more detailed, they investigate how men and women differentiate in their gambling behavior: putting money on the outcome of horseraces in UK off-course betting offices. The results provide evidence that women are more reluctant in taking risks than men. Also Powell and Ansic (1997) investigate the impact of gender differences in risk-taking behavior on decision-making. However, in contrast to Bruce and Johnson (1994), Powell and Ansic (1997) examine decision-making in the context of financial decision-making, instead of betting decisions. Hereby, Powell and Ansic (1997) make use of a series of realistic financial decisions, based on real financial data. The results are consistent with the conclusion Bruce and Johnson (1994) drew from their research. Going further to the study of Weber et al. (2002), evidence is provided that women are more risk averse, compared to men, if it comes to financial decisions, health/safety, recreational and ethical decisions. Only in the case of social risk, Weber et al. (2002) conclude from their research that the risk preferences between men and women are the same.

The three studies descripted above are only a few of the many that come to the same result: women tend to be more risk averse than men. However, most of these studies draw inferences on the general population, and therefore the outcome can’t just be generalized to all real-life settings. Looking to the managerial professions, not all researchers are convinced that among professionals gender differences towards risk attitudes exist: risk preferences between men and women are indifferent because more risk-tolerant women (women that are in the possession of masculine characteristics) self-select into the managerial professions (e.g. Kumar, 2010).

Second, relating to ethical behavior, women tend to have a stronger attitude towards compliance regarding rules and regulations than men. This comes forward in, for example, the study of Simon and Corbett (1996). They investigate the driving behavior of men and women. By using interviews and questionnaires Simon and Corbett (1996) conclude, inter alia, that women in overall offended the traffic laws less often than men. Further, and of strong importance here, when it comes to tax compliance, there is evidence that also in this context women have a stronger attitude towards compliance than men (Hasseldine, 1999; Kastlunger, Dressler, Kirchler, Mittone and Voracek, 2010).

(16)

16 2.5 Hypothesis development

To summarize, as seen in paragraph 2.2, not only the CFO of an organization, who has a direct link with the corporate tax activities (financial policies) of a firm, can affect the level of (aggressive) corporate tax avoidance. It also applies to other executives in the top

management of an organization. They are all responsible for formulating the risk-related policies of the firm (Baixauli-Soler, Belda-Ruiz and Sanchez-Marin, 2015). Besides, by setting the tone at the top with regard to, inter alia, tax activities, the executives create, or should create, a culture leading to the common acceptance and the implementation of those policies by the members of the organization. So, this study investigates the combined effect of top management executives on the level of (aggressive) corporate tax avoidance. Taken into account the in paragraph 2.4 mentioned tendency of women to be more risk averse than men/the tendency to have a stronger attitude towards tax compliance, the accompanying of (aggressive) tax avoidance activities with risks, and the statement of Francis et al. (2014) that the effect of gender diversity only can be found in the case of aggressive corporate tax

avoidance (more uncertainty (risk) is contained with the engagement in aggressive corporate tax avoidance activities), the following hypothesis can be made:

H1: Female representation in the top management of an organization is negatively associated with (aggressive) corporate tax avoidance, whereby the effect is bigger for aggressive

corporate tax avoidance than for corporate tax avoidance.

Next to the previous, from the research of Hanlon and Slemrod (2009) (paragraph 2.1) it stems that the risk of image loss (stock price decline), one of the biggest concerns faced by organizations, is on average higher for retail firms as compared to firms operating in

industries other than the retail industry. Taken into consideration that women tend to be more risk averse than men, this may emerge more clearly in the retail industry (because of the higher risk that is associated with (aggressive) corporate tax avoidance, on this point), leading to the second hypothesis:

H2: The effect of female representation in the top management of an organization on

(aggressive) corporate tax avoidance is more negative for firms in the retail industry than for firms operating in other industries.

(17)

17

3 Research methodology

This paragraph is divided into 2 subparagraphs. Paragraph 3.1 describes the sample selection. This is followed by a description of the proxies that will be used for the measurement of (aggressive) corporate tax avoidance, complemented with the research design, set out in paragraph 3.2.

3.1 Sample selection

For examining the research question, archival research is used as empirical approach (database). The sample consists of S&P 1500 companies in the period 2000 to 2011. I have chosen the S&P 1500 because of data availability and a sample period-starting year of 2000, because from then on female representation in the top management team of organizations rose (see introduction). With a sample period of 12 years, I arrive at a sample that is as large as possible, increasing the reliability of the results. For gender information and financial

accounting information I use ExecuComp database and Compustat respectively. In cases were ExecuComp provides gender information for more than 5 executives (forming a top

management team), I select the in rank 5 highest paid (based on salary and bonuses)

executives, thereby assuming that a top management team consists of 5 executives, which are the 5 highest in rank of payment (Baixauli-Soler et al., 2015).

After merging data from ExecuComp with Compustat data, the sample consists of 20,068 firm-year observations. From these firm-years observations, firm-year observations related to financial institutions (SIC codes 6000-6999) and utility firms (SIC codes 4900-4999) are subtracted, due to specific regulatory requirements faced by those organizations. Further, firm-year observations with missing data to estimate the proxies for (aggressive) corporate tax avoidance, Long-run Cash ETR and DTAX, and to calculate the control variables are excluded. The previous results in a sample of 6,213 firm-year observations for the Long-run Cash ETR and a sample of 7,879 firm-year observations for the DTAX (here, also the firms with less than 20 observations in 1-digit SIC industry-year groups are

excluded).1

1

(18)

18 3.2 Measurement (aggressive) corporate tax avoidance and research design

In order to measure the level of (aggressive) corporate tax avoidance, the distinction made in the literature review between different constructs of corporate tax reduction is of importance here. Corporate tax reduction can be measured in many ways, for example:

GAAP/Current/Cash effective tax rate (ETR), book-tax differences, unrecognized tax benefits and tax shelter activity. Not all measures are appropriate in capturing all of the different constructs of corporate tax reduction, and there is no common agreement among researchers on which measure fits best; all of the measures have their own implications, e.g. in measuring DTAX a lot of assumptions have to be made to deal with missing data values (see Appendix B) (Hanlon and Heitzman, 2010). For selecting the proxies for (aggressive) corporate tax avoidance, I reviewed prior studies. To capture corporate tax avoidance, researchers mainly apply the measures GAAP ETR, Current ETR and Cash ETR (e.g. Hope et al., 2013; Dyreng et al., 2010), while the proxies discretionary permanent book-tax differences (DTAX) and tax shelter activity are more common in measuring the level of aggressive corporate tax

avoidance (e.g. Kim, Li and Zhang, 2011; Francis et al., 2014; Rego and Wilson, 2012). As such, as a proxy for corporate tax avoidance activities, I use long-run Cash ETR (more accurate in measuring long-run corporate tax avoidance than annual effective tax rates (Dyreng, Hanlon and Maydew, 2008)) and for the measurement of aggressive corporate tax avoidance, I make use of DTAX (Frank, Lynch and Rego, 2009: actual tax sheltering (aggressive corporate tax avoidance) cases are related to DTAX).

Long-run cash ETR2 = ∑ Worldwide cash taxes paid

∑ Worldwide total pre-tax accounting income

DTAX = PERMDIFFi,t = β0 +β1 INTANGi,t + β2 UNCONi,t + β3 MIi,t + β4 CSTEi,t +

β5 ΔNOLi,t + β6 LAGPERMi,t + εi,t

Derived from Stickney and McGee (1982), Hope et al. (2013) and Francis et al. (2014), I include the following control variables (determinants of (aggressive) corporate tax avoidance,

2

See Appendix B for the calculation of the proxies for (aggressive) corporate tax avoidance. Note that for the calculation of the long-run Cash ETR, the sum of the 3-year lead, instead of the lagged (as most studies using this proxy do), cash taxes paid and pretax income is taken. My motivation to deviate from existing studies and to use lead cash taxes paid and pretax income, is that is more reasonable that the occurrence of female

representation in the top management team of an organization has an effect on future (aggressive) corporate tax avoidance activities and not on activities already occurred.

(19)

19 i.e. previously shown factors having an association with (aggressive) corporate tax avoidance (even though researchers are not always consistent in whether a positive or negative

association exists)): return on assets (ROAi,t: incentives to engage in tax avoidance activities

may be higher for high-generating profit firms; political cost theory), leverage (LEVi,t: using

debt as means of financing can bring tax benefits, i.e. interest expense is deductible), growth opportunities (MTBi,t:firms with high growth opportunities have more funds and incentives to

engage in tax avoidance activities), net operating loss carry forward (NOLi,t: utilization of

prior (year(s)) net operating loss carry forwards reduces the tax liability of the current period), the change in the net operating loss carry forward balance (∆NOLi,t), property, plant and

equipment (PPEi,t: there is a difference in book-tax treatment of depreciation expenses, which

has a bigger impact on more capital-intensive firms), research and development (RDi,t: for tax

and financial reporting purposes, research and development expenses are treated differently), intangible assets (INTANGi,t: equity method is accompanied with a difference in book-tax

treatment of intangible assets), foreign income (FORINCi,t: for tax and financial reporting

purposes, foreign income is treated differently; firms with foreign operations may have more possibilities to engage in (aggressive) corporate tax avoidance activities (difference in statutory tax rates)) and equity income (EQINCi,t: equity method is accompanied with a

difference in book-tax treatment of consolidated earnings (Cheng, Huang, Li and Stanfield, 2012)). Besides, researchers are inconclusive whether firm size is a determinant of

(aggressive) corporate tax avoidance. However, because there are some studies that find evidence for this (e.g. Mills et al., 2013), also firm size (SIZEi,t: political cost theory) is

included as control variable.3 Expected is that all of the control variables, with the exception of the control variable LEV (leverage), are negatively (positively) associated with long-run Cash ETR (DTAX).4

The previous results in the following multiple regression model:

TAXi,t = β0 + β1 GENDERi,t + β2 RETAILi,t + β3 ROAi,t + β4 LEVi,t + β5 MTBi,t + β6 NOLi,t +

β7 ∆NOLi,t + Β8 PPEi,t + β9 RDi,t + β10 INTANGi,t + β11 FORINCi,t + β12 EQINCi,t + β13 SIZEi,t

+ β14 GENDERi,t * RETAILi,t + εi,t

where TAXi,t represents the measures long-run Cash ETR and DTAX for firm i in year t.

3

See Appendix B for the calculation of the control variables.

4

Note that lower levels of the long-run Cash ETR are consistent with higher corporate tax avoidance-levels, and that higher levels of DTAX are consistent with higher corporate tax avoidance-levels.

(20)

20 GENDERi,t, the main variable of interest,captures the number of women the top management

team of an organization consists of. For GENDERi,t I make use of 2 dummy variables:

GENDER1 and GENDER2, where GENDER1 equals 1 if the top management team of an organization consists of one or more female executives and 0 otherwise, and GENDER2 equals 1 in the case a firm has two or more female executives in their top management team and 0 otherwise. By using 2 dummy variables, the ‘risk’ that masculine strategies dominate the effect(s) of female representation in a top management team is flatten (GENDER2): if a team consist of only one women (i.e. underrepresentation), there is a high change she takes over masculine characteristics, and vice versa (e.g. Godwin, Stevens and Brenner, 2006; Khaoula and Ali, 2012). If firms with female top executives engage in lower levels of (aggressive) tax avoidance, as expected, then the coefficient GENDER should be significant and positive (negative) in regressions where long-run Cash ETR (DTAX) is the dependent variable. For the second hypothesis, RETAILi,t and the interaction term

GENDERi,t * RETAILi,t are included in the regression model. RETAIL is a dummy variable

that equals 1 in the case a firm operates in the retail industry (NAICS Codes 44 and 45) and 0 for firms operating in industries other than the retail industry. If the effect, if any, of female representation in the top management of an organization on (aggressive) corporate tax avoidance is larger for firms operating in the retail industry than for firms operating in other industries, then the coefficient on the interaction term GENDER * RETAIL should be significant and positive (negative) in regressions where long-run Cash ETR (DTAX) is the dependent variable. Further, the relationship between the variable RETAIL and (aggressive) corporate tax avoidance is expected to be positive. Retail firms have close customer-supplier relationships. According to Cen, Maydew, Zhang and Zuo (2014), the existence of close customer-supplier relationships leads to (more) opportunities to engage in (aggressive) corporate tax avoidance activities (to engage in such activities, mostly a third party-involvement is required).

4 Multivariate analysis – Empirical results

In this paragraph the results of the empirical tests are expressed. The paragraph is divided into 2 subparagraphs. The first subparagraph, paragraph 4.1, is dedicated to corporate tax

(21)

21 Both subparagraphs begin with an overview and explanation of the descriptive statistics and the Pearson correlation matrix. This is followed by the results of the long-run Cash ETR and DTAX tests respectively. An answer to and an explanation on the hypotheses is provided, with a subdivision of the 2 hypotheses within each subparagraph. Moreover, paragraph 4.2 is complemented with a robustness test for the results of the DTAX test.

4.1 Corporate tax avoidance 4.1.1 Descriptive statistics

As mentioned in paragraph 3.2, as a proxy for corporate tax avoidance, I use long-run Cash ETR. The descriptive statistics on the long-run Cash ETR (the dependent variable), GENDER (the independent variable/variable of interest), RETAIL (the independent variable), and the control variables are provided in Table 2.

Table 2 – Descriptive statistics for the long-run Cash ETR tests

Variable # Obs Mean Std. Deviation Min. Max.

Long-run Cash ETR 6213 0.252 0.138 0 0.989

GENDER1 1713 0.276 0.447 0 1 GENDER2 365 0.059 0.235 0 1 RETAIL 635 0.102 0.303 0 1 ROA 6213 0.123 0.108 -0.152 0.524 LEV 6213 0.193 0.196 0 1 MTB 6213 3.675 4.115 -10.191 25.49 NOL 6213 0.603 0.489 0 1 ChangeNOL 6213 0.004 0.053 -0.189 0.284 PPE 6213 0.293 0.241 0.018 1.102 RD 6213 0.030 0.050 0 0.236 INTANG 6213 0.216 0.227 0 1.088 FORINC 6213 0.025 0.042 -0.030 0.197 EQINC 6213 0.001 0.005 -0.007 0.029 SIZE 6213 7.249 1.437 4.461 10.77

For the long-run Cash ETR sample, there are 6,213 firm-year observations. Of these 6,213 firm-year observations, 1,713 firm-year observations (27.6%) are related to GENDER1 and 365 firm-year observations (5.9%) to GENDER2: the number of firm-year observations for which the top management team of organizations consists of at least one female executive is 1,713 and the number of firm-year observations for which the top management team of

(22)

22 organizations consist of two or more female executives is 365, considerably less than the 1,713 GENDER1 firm-year observations. Further, of the 6,213 firm-year observations, 635 firm-year observations (10.2%) are related to firms operating in the retail industry.

The mean value of the long-run Cash ETR is 0.252. This mean value is comparable with the reported mean value of the long-run Cash ETR in Dyreng et al. (2008) and Hoopes, Mescall and Pittman (2012). Based on the descriptive statistics, the following comes, inter alia, forward for the control variables: a mean value of 0.123 for the control variable ROA (return on assets), a mean value of 0.193 for the control variable LEV (leverage), a mean value of 0.603 for the control variable NOL (net operating loss carry forwards), indicating that roughly 60% of the firm-year observations have a net operating loss carry forward, and a mean value of 0.025 and of 0.001 for the control variables FORINC (foreign income) and EQINC (equity income) respectively. The reported mean values for the control variables FORINC and EQINC imply that for the 6,213 firm-year observations, as a source of revenue for organizations, the amount of foreign income and equity income is relatively small.

4.1.2 Pearson correlation matrix

The Pearson correlation matrix, which measures the degree of correlation (strength of relationship) between variables, for the long-run Cash ETR, GENDER, RETAIL and the control variables is presented in table 3 (next page).

As can be seen in table 3, more than half of the correlations are significant at the 1% level. To mention, the most important variables in the Pearson correlation matrix are the independent variable GENDER and the independent variable RETAIL. As expected, both definitions of the variable GENDER, GENDER1 and GENDER2, are positively (0.053; 0.039) and significantly correlated with the dependent variable, long-run Cash ETR. This also applies to the variable RETAIL: RETAIL is positively (0.145) and significantly correlated with the dependent variable, although this is not as expected. Furthermore, the correlations between the independent/control variables are one average not very high, reducing the risk of multicollinearity. To be sure there is no multicollinearity problem, I tested for the variance inflation factor (VIF). The rules of thumb states that if the VIF is above 10, there is

(23)

23

Table 3 – Pearson correlation matrix for the long-run Cash ETR tests

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (1) Long-run 1 Cash ETR (2) GENDER1 0.053 1 (3) GENDER2 0.039 0.405 1 (4) RETAIL 0.145 0.150 0.178 1 (5) ROA 0.108 0.028 0.065 0.084 1 (6) LEV - 0.118 - 0.033 - 0.050 - 0.108 - 0.176 1 (7) MTB - 0.031 - 0.006 - 0.004 - 0.015 - 0.429 - 0.049 1 (8) NOL - 0.197 0.001 - 0.025 - 0.178 - 0.254 0.143 - 0.065 1 (9) Change- 0.008 0.003 - 0.001 - 0.028 - 0.120 0.064 0.010 0.009 1 NOL (10) PPE - 0.061 - 0.030 - 0.013 0.097 0.054 0.233 0.001 - 0.157 - 0.019 1 (11) RD - 0.162 - 0.068 - 0.030 - 0.201 0.055 - 0.173 0.179 0.160 0.074 - 0.294 1 (12) INTANG - 0.022 0.019 - 0.034 - 0.181 - 0.103 0.328 - 0.023 0.177 0.069 - 0.314 0.005 1 (13) FORINC - 0.092 - 0.017 0.010 - 0.167 0.221 - 0.054 0.167 0.197 0.013 - 0.097 0.269 - 0.008 1 (14) EQINC 0.014 - 0.046 - 0.053 - 0.086 0.004 0.060 0.006 - 0.004 - 0.006 0.065 - 0.063 - 0.032 0.033 1 (15) SIZE - 0.064 - 0.005 - 0.037 0.041 - 0.188 0.213 - 0.036 0.166 0.009 0.092 - 0.151 0.097 0.173 0.148 1

Significance at the 1% level is indicated by the bold correlations. All variables are defined as in Appendix B.

4.1.3 Long-run Cash ETR test – Hypothesis 1

Table 4 (next page) reports the results of the long-run Cash ETR test for hypothesis 1: Female

representation in the top management of an organization is negatively associated with (aggressive) corporate tax avoidance, whereby the effect is bigger for aggressive corporate tax avoidance than for corporate tax avoidance. At this point, only for the first part of the

hypothesis (female representation in the top management of an organization is negatively associated with corporate tax avoidance), a statement can and will be made.

As shown in table 4, in the regressions where the long-run Cash ETR is the dependent variable and GENDER is defined as GENDER1, the coefficient on GENDER is positive (0.010) and significant at the 1% level. In the regressions where the long-run Cash ETR is the dependent variable and GENDER is defined as GENDER2, the coefficient on GENDER is positive (0.011), but insignificant. The results suggest that when the top management team of an organization consists of one female executive (the effect of GENDER1 seems, for a big part, to be driven by those firms having one female executive in their top management), the

(24)

24 Table 4 – Results for the long-run Cash ETR test on hypothesis 1

GENDER1 GENDER2

Variable Exp. Sign Coef. T-stat. Prob. Coef. T.stat. Prob.

GENDER + 0.010*** 2.58 0.010 0.011 1.48 0.139 ROA - 0.137*** 7.26 0.000 0.137*** 7.25 0.000 LEV + - 0.060*** - 5.95 0.000 - 0.060*** - 6.02 0.000 MTB - - 0.002*** - 3.80 0.000 - 0.002*** - 3.77 0.000 NOL - - 0.040*** -10.52 0.000 - 0.039*** -10.46 0.000 ChangeNOL - 0.114*** 3.56 0.000 0.115*** 3.58 0.000 PPE - - 0.072*** - 8.69 0.000 - 0.072*** - 8.74 0.000 RD - - 0.484*** -12.89 0.000 - 0.489*** -13.06 0.000 INTANG - 0.000 0.02 0.982 0.001 0.10 0.920 FORINC - - 0.153*** - 3.43 0.001 - 0.155*** - 3.47 0.001 EQINC - 0.602* 1.66 0.097 0.587 1.62 0.106 SIZE - - 0.001 - 1.13 0.259 - 0.001 - 1.10 0.272 Intercept 0.323*** 31.43 0.000 0.325*** 31.72 0.000 Adjusted R2 0.093 0.092 F-statistic 54.12 53.71

Significance at the 1%, 5%, and 10% level is indicated by ***, **, and *, respectively. All variables are defined as in Appendix B.

Long-run Cash ETR is winsorized to the range [0,1].The continuous variables are winsorized at the 1st and 99th percentiles.

level of a firm’s corporate tax avoidance (broad spectrum) declines. This supports hypothesis 1, but is contradicting with the outcomes documented in the studies of Khaoula and Ali (2012) and Francis et al. (2014). Further, the second set of results (originating from the test where the long-run Cash ETR is the dependent variable and GENDER is defined as

GENDER2) shows that there is no evidence that firms that have two or more female

executives in their top management team engage in lower levels of corporate tax avoidance than those firms that have homogeneous top management teams/only one female executive representative in their top management team. The previous is not in line with hypothesis 1, for which several possible explanations can be given, thereby not taking into consideration the results of the DTAX test (at this stage). First, for the outcome of the test where GENDER is defined as GENDER2, sample size could be a limitation: the number of firm-year

observations for which the top management team of organizations consist of two or more female executives is 365, considerably less than the 1,713 GENDER1 firm-year observations. Second, the results can be supported by/be in the support of the studies that argue that more risk-tolerant women self-select into the managerial professions, mentioned in paragraph 2.4. In this case, the reduction in the level of corporate tax avoidance of firms that have a top

(25)

25 management team with one female executive can be explained as follows: long-run Cash ETR measures the broad spectrum of corporate tax avoidance, including the less aggressive forms of corporate tax avoidance. It could be the case that the more risk-tolerant female executive (even though she is not completely at the full risk-taking level as their male counterparts) doesn’t believe, after making a risk(cost)-benefit trade-off, it is worth taking the risks of the less aggressive (also less profitable compared to the aggressive) forms of corporate tax avoidance (and has the power of persuasion to refrain their male counterparts to ‘create’ a culture leading to this type of corporate tax avoidance) if the risks and benefits of engaging in corporate tax avoidance activities are perceived to be close together, but can come to the conclusion that she actually dares to take those risks if she has the support of another or other female executive(s) in the top management team (TSM, 2012).

In addition to the previous, also shown in table 4 are the results of the control variables for the regressions where long-run Cash ETR is the dependent variable. As can be seen, for both definitions of GENDER, GENDER1 and GENDER2, most coefficients on the control variables are significant at the 1% level and are in line with the expected relationship they have with the dependent variable. For example, the coefficient on the control variable MTB (growth opportunities) is negative (-0.002) and significant: firms with high growth

opportunities have more funds and incentives to engage in tax avoidance activities. Next to this, a significant control variable that, in contrast to the preceding, does not meet the expectations, is the control variable LEV (leverage). However, the outcome, a negative

instead of a positive coefficient, is consistent with Kim et al. (2011) and Francis et al. (2014).

4.1.4 Long-run Cash ETR test – Hypothesis 2

Table 5 (next page) reports the results of the long-run Cash ETR test for hypothesis 2: The

effect of female representation in the top management of an organization on (aggressive) corporate tax avoidance is more negative for firms in the retail industry than for firms operating in other industries. As shown in this table, in the regressions where the long-run

Cash ETR is the dependent variable, for both definitions of GENDER, GENDER1 and GENDER2, the coefficient on the variable of interest (GENDER*RETAIL) is positive, but insignificant. The results do not support hypothesis 2, for which several possible explanations can be given. First of all, the limited sample size (number of firm-year observations) for the interaction variable, as reported in table 5, could be a limitation. Second, as with hypothesis1,

(26)

26 Table 5 – Results for the long-run Cash ETR test on hypothesis 2

GENDER + 0.006 1.51 0.131 0.001 0.17 0.864 RETAIL - 0.036*** 4.64 0.000 0.037*** 5.83 0.000 ROA - 0.126*** 6.67 0.000 0.127*** 6.69 0.000 LEV + - 0.055*** - 5.52 0.000 - 0.056*** - 5.55 0.000 MTB - - 0.002*** - 3.72 0.000 - 0.002*** - 3.72 0.000 NOL - - 0.038*** -10.00 0.000 - 0.038*** - 9.95 0.000 ChangeNOL - 0.112*** 3.50 0.000 0.112*** 3.51 0.000 PPE - - 0.071*** - 8.66 0.000 - 0.072*** - 8.70 0.000 RD - - 0.452*** -11.97 0.000 - 0.456*** -12.08 0.000 INTANG - 0.008 0.89 0.372 0.008 0.95 0.340 FORINC - - 0.112** - 2.49 0.013 - 0.112** - 2.49 0.013 EQINC - 0.849** 2.33 0.020 0.833** 2.29 0.022 SIZE - - 0.002* - 1.88 0.060 - 0.002* - 1.89 0.058 GENDER* + 0.002 0.20 0.838 0.005 0.32 0.751 RETAIL Intercept 0.322*** 31.45 0.000 0.324*** 31.72 0.000 N (# of interaction 301 116 observations) Adjusted R2 0.098 0.098 F-statistic 49.42 49.21

Long-run Cash ETR is winsorized to the range [0,1].The continuous variables are winsorized at the 1st and 99th percentiles.

I can continue with the argument made in paragraph 4.1.3 that more risk-tolerant women self-select into the managerial professions. The risk of image loss (which may result in stock price declines) is on average higher in the retail industry. However, after reviewing the pretax income of sample-firms operating in this industry, I can make the assumption that also the profits generated in the retail industry are generally above average (increasing the incentives to engage in corporate tax avoidance activities). Taking the previous into account, this could explain that, as opposed to hypothesis 1, for both definitions of GENDER there is no evidence that there is a (more) negative relationship between female representation in the top

management team of a retail firm, compared to firms operating in other industries, and corporate tax avoidance.

(27)

27 4.2 Aggressive corporate tax avoidance

4.2.1 Descriptive statistics

As mentioned in paragraph 3.2, as a proxy for aggressive corporate tax avoidance, I use DTAX. The descriptive statistics on DTAX (the dependent variable), GENDER (the

independent variable/variable of interest), RETAIL (the independent variable), and the control variables are provided in Table 6.

Table 6 – Descriptive statistics for the DTAX tests

Variable5 # Obs Mean Std. Deviation Min. Max.

DTAX 7879 - 0.000 0.062 -0.504 0.378 GENDER1 2207 0.280 0.449 0 1 GENDER2 472 0.060 0.237 0 1 RETAIL 758 0.096 0.295 0 1 ROA 7879 0.086 0.141 -0.460 0.489 LEV 7879 0.198 0.208 0 1.071 MTB 7879 3.311 4.166 -11.48 25.46 NOL 7879 0.636 0.481 0 1 PPE 7879 0.289 0.240 0.017 1.089 RD 7879 0.036 0.059 0 0.288 FORINC 7879 0.021 0.041 -0.068 0.191 EQINC 7879 0.001 0.005 -0.008 0.028 SIZE 7879 7.157 1.461 4.215 10.78

For the DTAX sample, there are 7,879 firm-year observations. Of these 7,879 firm-year observations, 2,207 firm-year observations (28.0%) are related to firm-year observations for which the top management of an organization consists of one or more female executives (GENDER1), and 472 firm-year observations (6.0%) are related to GENDER2: the number of firm-year observations for which the top management team of organizations consist of two or more female executives is 472. As with the long-run Cash ETR tests, the number of firm-year observations associated with GENDER2 is considerably less than the number of firm-year observations associated with GENDER1. Next to the previous, also shown in table 5 is that 758 firm-year observations (9.6%) are related to firms operating in the retail industry.

The mean value of the DTAX is -0.000. This mean value ispractically, although not completely, similar as the reported mean value of the DTAX in prior studies, where the mean value is around 0.020 (e.g. Badertscher, Katz and Rego, 2013; Francis et al., 2014). The

5

Note that the control variables changeNOL (change in the net operating loss carry forward balance) and INTANG (intangible assets) are excluded, because they are inherently part of the DTAX-model.

(28)

28 reported mean values of the control variables are generally in line with those reported for the long-run Cash ETR tests: the descriptive statistics for the DTAX tests (table 6) show, inter alia, a mean value of 0.086 for the control variable ROA (return on assets), a mean value of 0.636 for the control variable NOL (net operating loss carry forward), indicating that almost 64% of the firm-year observations can be associated with a net operating loss carry forward, and a mean value of 0.021 for the control variable FORINC (foreign income); as a source of revenue, foreign income is relatively small.

4.2.2 Pearson correlation matrix

The Pearson correlation matrix, which measures the degree of correlation (strength of

relationship) between variables, for the DTAX, GENDER, RETAIL and the control variables is presented in table 7.

Table 7 – Pearson correlation matrix for the long-run Cash ETR tests

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (1) DTAX 1 (2) GENDER1 0.000 1 (3) GENDER2 0.013 0.405 1 (4) RETAIL - 0.007 0.155 0.181 1 (5) ROA 0.383 0.015 0.056 0.081 1 (6) LEV - 0.021 - 0.027 - 0.041 - 0.089 - 0.166 1 (7) MTB 0.113 - 0.024 - 0.005 - 0.010 0.354 - 0.075 1 (8) NOL - 0.037 - 0.001 - 0.028 - 0.178 - 0.269 0.132 - 0.078 1 (9) PPE 0.038 - 0.033 0.001 0.096 0.091 0.258 - 0.004 - 0.158 1 (10) RD - 0.010 - 0.076 - 0.043 - 0.195 - 0.143 - 0.114 0.140 0.184 - 0.316 1 (11) FORINC 0.178 - 0.023 0.005 - 0.135 0.308 - 0.066 0.172 0.123 - 0.069 0.149 1 (12) EQINC - 0.000 - 0.040 - 0.035 - 0.075 0.068 0.037 0.009 - 0.026 0.073 - 0.081 0.070 1 (13) SIZE - 0.032 - 0.012 - 0.031 0.049 - 0.022 0.205 - 0.024 0.119 0.144 - 0.237 0.167 0.165 1

Significance at the 1% level is indicated by the bold correlations. All variables are defined as in Appendix B.

As can be seen in table 7, two-thirds of the correlations are significant at the 1% level. To mention, the most important variables in the Pearson correlation matrix are the independent variable GENDER and the independent variable RETAIL. Not in line with the expectations is that there is no statistically significant correlation between GENDER1/GENDER2 and the dependent variable, DTAX. This also applies to the variable RETAIL. Furthermore, the

(29)

29 correlations between the independent/control variables are one average not very high,

reducing the risk of multicollinearity. To be sure there is no multicollinearity problem, I tested for the variance inflation factor (VIF). The rules of thumb states that if the VIF is above 10, there is multicollinearity. However, with a mean VIF of 1.21 this is not the case.

4.2.3 DTAX test – Hypothesis 1

Table 8 reports the results of the DTAX test for hypothesis 1: Female representation in the

top management of an organization is negatively associated with (aggressive) corporate tax avoidance, whereby the effect is bigger for aggressive corporate tax avoidance than for corporate tax avoidance.

Table 8 – Results for the DTAX test on hypothesis 1

GENDER - - 0.000 - 0.20 0.845 - 0.002 - 0.70 0.483 ROA + 0.177*** 32.31 0.000 0.177*** 32.31 0.000 LEV - 0.014*** 4.34 0.000 0.014*** 4.32 0.000 MTB + - 0.001*** - 3.38 0.001 - 0.001*** - 3.39 0.001 NOL + 0.007*** 4.97 0.000 0.007*** 4.97 0.000 PPE + 0.005* 1.71 0.088 0.005* 1.72 0.086 RD + 0.035*** 2.80 0.005 0.034*** 2.80 0.005 FORINC + 0.092*** 5.25 0.000 0.092*** 5.25 0.000 EQINC + - 0.322** - 2.27 0.023 - 0.324** - 2.28 0.022 SIZE + - 0.002*** - 3.69 0.000 - 0.002*** - 3.71 0.000 Intercept - 0.012*** - 3.32 0.001 - 0.012*** - 3.32 0.001 Adjusted R2 0.159 0.159 F-statistic 150.32 150.38

DTAX is winsorized to the range [-1,1].The continuous variables are winsorized at the 1st and 99th percentiles.

As shown in table 8, in the regressions where the DTAX is the dependent variable, for both definitions of GENDER, GENDER1 and GENDER2, the coefficient on GENDER is negative (-0.000 and -0.002 respectively; as expected), but insignificant. There is no evidence that female representation in the top management of an organization is negatively associated with aggressive corporate tax avoidance: the results do not support hypothesis 1 and especially the

Female representation in top management and (aggressive) corporate tax avoidance