Sustainability reporting and corporate tax avoidance

(1)

Amsterdam Business School

(Sustainability Reporting and Corporate Tax Avoidance)

Name: Daan de Bree

Student number: 11112492

Thesis supervisor: prof. dr. B.G.D O’Dwyer Date: 08-06-2016

Word count: 14760, 0

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

Statement of Originality

This document has been written by student Daan de Bree, who declares that he takes 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)

Abstract

This thesis examines the relationship between sustainability reporting and corporate tax avoidance. In addition, the effect of sustainability reports that have been prepared in accordance with Global Reporting Initiative (GRI) guidelines on corporate tax avoidance is tested. Furthermore, this study investigates whether providing assurance on sustainability reports will have an effect on the usage of corporate tax avoidance measures. The study sample consists of organizations from the Standard & Poor’s 500 in the period 2013 to 2015. The dependent variable is measured as a continuous variable with a value from zero to one. The independent variables are measured as dummy variables, where the variable is valued as one if it complies with the requirements in at least two of the three sample years. For example, the variable sustainability reporting is valued as one if the organization has issued a sustainability report in at least two of the three sample years. The empirical findings indicate that there is no relationship between sustainability reporting and corporate tax avoidance. Additionally, the results designate that the GRI guidelines do not have an impact on the use of corporate tax avoidance. Moreover, the results show that providing assurance on sustainability reports does not influence the amount of corporate tax avoidance as well. These findings are contradictory with the expectations outlined in the literature study. This thesis sketches several possible reasons for this contradiction. First, organizations do sometimes have incentives to develop corporate social responsibility policies and thereby sustainability reports, for example, for marketing purposes. These organizations might issue sustainability reports to be congruent with the values of their customers in order to make more profits. In some cases these firms will not particularly be interested in society or their social responsibilities. Furthermore, being compliant with the GRI guidelines or providing assurance on sustainability reports could be seen as a method to legitimize behaviour in some cases, but not particularly to ensure that no negative impacts on environment and society by organizational actions will occur. This might explain why none of the hypotheses outlined in this thesis hold. This paper is subject to a few limitations. One of the limitations might be the way in which the dummy variables were computed. These variables are valued as one if they are compliant in two out of three sample years. A stronger connection may be found if these organizations must be compliant with the requirements in all of the sample years. Further, measuring corporate tax avoidance by only one proxy could give a narrow view of the situation. Using more proxies could provide a more complete picture of the relationship between sustainability reporting and tax avoidance.

(4)

(5)

4.2.2 Correlation Matrix ... 32 4.2.3 Results ... 33 4.3 Hypothesis 3 ... 35 4.3.1 Descriptive statistics ... 35 4.3.2 Correlation Matrix ... 36 4.3.3 Results ... 37 5 Conclusion ... 39 6 References ... 42 7 Appendices ... 46 7.1 Appendix A ... 46 7.2 Appendix B. ... 46 7.2.1 Sample Selection ... 46 7.2.2 Variable measurement ... 47

(7)

1 Introduction

1.1 Relevance and Research Question

Over the last few years big companies such as Amazon, Google and Starbucks have been in the news because they avoid taxes on a large scale1_{. These companies use several strategies for this} purpose, for instance the so-called “Double Irish Dutch Sandwich2_{” strategy and income} shifting. Despite the fact that most of these strategies are not illegal, many people think using these strategies is unethical. Tax incomes are of crucial importance for governments (Christensen & Murphy, 2004). Corporate tax income accounts for 11% of the total income of the Dutch government3_{. Tax avoidance behaviour will allow organizations to pay less tax. This results in a} lower amount of corporate taxation. However, the government does still require the same amount of tax revenues. Therefore, tax avoidance by organizations will cause a higher tax burden for citizens. In other words, corporate tax avoidance is very costly to governments and citizens and is therefore often seen as unethical behaviour (Christensen & Murphy, 2004).

On the other hand, corporate social responsibility (CSR) is gaining more importance in society. The European Commission defines CSR as: “the responsibility of enterprises for their impacts on society” (European Commission, 2011, p. 6). CSR activities implores companies to act responsibly towards society and the environment beyond what they are legally obligated to do (Vallaster, Lindgreen, & Maon, 2012). In this light, tax avoidance is interesting for companies because it is often seen as irresponsible behaviour (Christensen & Murphy, 2004). Because of the increasing importance of CSR, sustainability reporting is becoming more and more important for organizations these days; see Figure 1 (Coppola, 2015).

Sustainability Reporting is an extension of the financial report. The sustainability report is an organizational report on economic, social and environmental performance, whereas financial reporting involves solely reporting the economic performance of the organization (Gray, 2006). Corporations develop sustainability policies for various reasons. First, numerous corporations decide to enhance their sustainability practices for economic motivations. They want to protect or enhance the shareholder value by displaying their sustainability performance. Showing this performance would enhance shareholder value because these organizations expect that sustainability performance will be of great importance to shareholders. Economic indicators

1_{http://www.bbc.com/news/magazine-20560359}

2_{As Zucman (2014) defines in his research, the Double Irish Dutch Sandwich strategy as a strategy that reduces}

taxes by routing profits through Irish subsidiaries and the Netherlands and then to the Caribbean.

(8)

focus on the manner in which the organization affects the stakeholders with whom it has direct and indirect interactions (Hayes, Dassen, Schilder & Wallage, 1999). Second, a number of organizations make an ethically-motivated decision to enhance their sustainability performance. These organizations actually attempt to ensure no negative impacts on environment and society by organizational actions (Hayes et al., 1999). Such organizations would make decisions with the purpose of enhancing their sustainability performance, even if this goes against the will of stakeholders. Especially the second argument is of great importance to the research question in this thesis. In case organizations deploy a sustainability report because they attempt to ensure no negative impacts on society and the environment, this study argues that such organizations are less likely to be involved in tax avoidance activities. This follows from the argument that corporate tax avoidance is seen as irresponsible and unethical behaviour within society.

This research is relevant in practice because many governments are troubled by lost taxes as a result of corporate tax avoidance. By searching for an effect of sustainability reporting on corporate tax avoidance, governments could obtain more insight into this problem. Besides, governments could develop solutions for reducing corporate tax avoidance, for instance by moving towards mandatory sustainability reporting. Research on the subject of corporate tax avoidance is thin but increasingly investigated in several studies over the last few years (Desai & Dharmapala, 2006; Dyreng, Hanlon & Maydew, 2008; Dyreng, Hanlon & Maydew, 2010). Further, the link between CSR and corporate tax avoidance has been examined several times (Hoi, Wu & Zhang, 2013; Christensen & Murphy; 2004; Sikka, 2010). However, the question of whether companies who produce sustainability reports are less likely to develop corporate tax avoidance activities has not previously been tested. This research contributes to the existing literature because of three reasons. First, the amount of existing literature on corporate tax avoidance is small but is of growing interest. In particular, the factors that contribute to corporate tax avoidance and the impact that it produces are desired for further research. By examining the effects of sustainability reporting on corporate tax avoidance, this thesis responds to these calls. Second, sustainability reporting literature is also just emerging. Specifically, studies on the effects of sustainability reporting are required. This thesis attempts to fulfil these demands by examining the effect of sustainability reports on tax avoidance. Third, the consequences of specific guidelines used in the preparation of sustainability reports and of assurance provided on sustainability reports have not been extensively studied. Therefore, the second hypothesis in this study will examine the effect of Global Reporting Initiative (GRI) guidelines on corporate tax avoidance, and the third hypothesis will test the impact of assurance on corporate tax avoidance. The research question of this study is therefore as follows:

(9)

What effect do different types of sustainability reports have on the level of corporate tax avoidance used by an organization?

To answer this research question, a database research is conducted. A distinction is made between organizations with any form of sustainability report and organizations without sustainability reports. The sample to be used is the Standard and Poor’s 500. Thereafter, the amount of corporate tax avoidance is measured by the Long-Run Cash Effective Tax Rate4_. The results indicate that organizations that have issued two or three sustainability reports over the last three years do not have higher tax rates. In other words, issuing sustainability reports does not have an effect on the use of corporate tax avoidance. Subsequently, the empirical findings exhibit an insignificant relationship between sustainability reports prepared in accordance with the GRI guidelines and corporate tax avoidance. Furthermore, organizations with assured sustainability reports are not associated with higher tax rates than organizations with non-assured sustainability reports. Put differently, organizations with assured sustainability reports do not make less use of corporate tax avoidance. The results suggest that these different types of sustainability reports do not have an affect on the amount of corporate tax avoidance utilized by organizations.

1.2 Structure

The structure of this thesis is as follows: In the next section, the underlying literature is explained. Firstly, the topic of corporate social responsibility is discussed, whereby stakeholder theory and legitimacy theory is illuminated. Additionally, at the end of this section the hypotheses are presented. Secondly, sustainability reporting is addressed, in which the Global Reporting Initiative receives special attention. Lastly, the subject of corporate tax avoidance is covered. Section 3 focuses on the sample and the research design. Further, results are explained in Section 4 and the conclusion in the fifth section is used for final remarks.

(10)

2 Literature

In this chapter, the literature that is used to develop the hypotheses is discussed. First, corporate social responsibility, stakeholder theory and legitimacy theory are addressed. Second, sustainability reporting is discussed; here, the Global Reporting Initiative and assurance are very important topics. Thereafter, the topic of corporate tax avoidance is presented. Finally, the hypotheses are stated and linked to the previous literature.

2.1 Corporate Social Responsibility

In this section the concept of Corporate Social Responsibility (CSR) is discussed. First, three different theoretical approaches to CSR will be identified, which will be converted into two basic approaches. Thereafter, stakeholder theory and legitimacy theory are addressed.

2.1.1 Different approaches to CSR

In past research, various approaches to CSR have been discussed, and these basically constitute three categories: the shareholder approach, the stakeholder approach and the societal approach (van Marrewijk, 2003).

The shareholder approach is based on the classical idea that “the social responsibility of business is to increase its profits” (Friedman, 1970). A company should focus on the needs of the shareholders, as they are expected to be in favour of profit maximization. This implies that social responsibility activities are not relevant for companies but are the responsibility of governmental organizations. Quazi and O’Brien (2002) cite the shareholder approach as the classical approach in which the social responsibility of an organization is a one-dimension activity. Here, organizations are only responsible for supplying goods and services to society at a profit. Finally, van Marrewijk (2003) states that the organizations are only concerned with CSR activities to the extent that they contribute to the creation of long-term value for the shareholders.

The stakeholder approach is based on the idea that organizations are not only accountable to their shareholders but they should also take various other stakeholders into account (van Marrewijk, 2003). The preferences of the stakeholders could affect or be affected by the accomplishment of an organization’s targets (Freeman, 1984). Stakeholders are defined as “People or small groups with the power to respond to, negotiate with, and change the strategic future of the organization” (Eden & Ackermann, 2013).

The society approach can be seen as comprehensive. Here, the organization is responsible to society as a whole, of which they are an integral part (van Marrewijk, 2003). This view is broader

(11)

than the stakeholder approach because here society as a whole is seen as an entity that should be taken into account by organizations whilst conducting business.

Within this research, the first and the second approaches are considered the economic direction and the third approach is viewed as the ethical direction. This distinction is made because one could argue that developing CSR policies for shareholders or stakeholders could be seen as a method to maximize the profits of an organization. If the shareholders and stakeholders demand CSR policies from an organization, it would be in the organization’s economic self-interest to develop these policies. When an organization develops CSR policies with the sole purpose of maximizing profits, it is likely that it would not refrain from corporate tax avoidance activities, because these tax avoidance activities are also associated with maximizing profit (Bartelsman & Beetsma, 2003). On the other hand, the society approach is seen as the ethical direction within this study. Here, the organization is responsible to society as a whole. These organizations would develop CSR policies for ethically motivated reasons. With its CSR policies, such organizations actually attempt to ensure no negative impacts on the environment and society as a result of organizational actions (Hayes et al., 1999). These organizations would deploy CSR policies even if they would run counter to the favour of shareholders and stakeholders. Therefore, it is likely that such organizations would not engage in tax avoidance activities because these activities are seen as harmful for society. Conclusively, an organization would implement CSR policies within the economic direction because the shareholders and stakeholders require them. Using the ethical direction, an organization would deploy CSR policies because they want no negative impacts on the environment and society. Such organizations would implement these policies even if they were at the expense of the stakeholders.

2.1.2 Stakeholder Theory and Legitimacy Theory

In this sub-section, the two subjects of stakeholder theory and legitimacy theory are addressed. First the definitions of these theories will be discussed and then the connections between these definitions and the two CSR directions discussed in the previous paragraph will be made.

Clarkson (1995) divides stakeholders in two categories, primary and secondary. Primary stakeholders are those who are vital for an organization. Without continuing participation of these stakeholders the organization cannot survive. Primary stakeholder groups usually consist of shareholders, employees, customers, suppliers and the so-called public stakeholder group. This public group is comprised of the government and those bodies whose policies and laws must be followed. Usually, there is much cohesion between these stakeholders and the organization. Secondary stakeholders are defined as stakeholders who influence, or are influenced by, an

(12)

organization. However, they are not vital for the organization and they are not engaged in transactions with the organization. Examples of secondary stakeholder groups are the media and certain special interest groups. These groups can sway public opinion in favour of or in opposition to the organization’s performance. An example which connects these secondary stakeholder groups with corporate tax avoidance could be the growing attention for corporate tax issues in the media. This has the result that public opinion could be turned against organizations that deploy tax avoidance activities (Dyreng et al., 2015).

Freeman (1984) states that it is important for organizations to acknowledge the needs of the stakeholders in order to achieve the firm’s objectives. According to the stakeholder theory, an organization should disclose information in order to communicate to the stakeholders that their needs have been taken into account. In addition, the stakeholder theory assumes that there are various social contracts with multiple stakeholder groups. However, some of these stakeholders have opposing interests, which has the consequence that not all the needs of the stakeholders can be met (Deegan, 2002). Organizations should measure the relative importance of the stakeholders with respect to power, urgency and legitimacy, in order to recognize where the organization should focus their CSR actions (Mitchell, Agle & Wood, 1997). The stakeholder theory could be linked with the stakeholder approach discussed by van Marrewijk (2003). The stakeholder approach suggests that the main reason for organizations to deploy CSR activities is to meet the preferences of their stakeholders. Therefore, the stakeholder theory is in line with the economic approach.

According to Lindblom (1994), legitimacy is a status that exists when an organization’s value system is congruent with the value system of the larger social system of which the organization is a part. Legitimacy theory is derived from the idea that the organization is part of the social system upon which it has an influence or by which it is influenced. This social system has certain values and ideas of the activities of the organization. When there is a difference between these values and ideas of the society and of the organization, a legitimacy gap arises. There are multiple reasons why such a legitimacy gap could develop. First, a legitimacy gap could be the result of changed values and ideas of society. For example, the concept of tax avoidance is becoming more and more important to societies these days. In order to eliminate or reduce the legitimacy gap, organizations should address these changed values of society. Second, the gap could be the result of some change in the activities of the organization. Such a change could be a result of the organization deciding to deploy a new type of tax avoidance strategy, possibly with the effect that an increasing legitimacy gap develops. Third, a legitimacy gap could arise when information of an organization’s activities becomes publicly available. In other words, information which was

(13)

previously unknown becomes known. For example, an organization deploys tax avoidance activities that suddenly become publicly known because the media are writing about it. All these reasons will have the effect that the value system of society is no longer congruent with the value system of the organization (Palazzo & Richter, 2005). However, it could be argued that the legitimacy theory is connected with the society approach of van Marrewijk (2003). Organizations might also try to legitimize their actions for economic reasons. For example, some organizations may try to eliminate the legitimacy gap for marketing purposes. Such a firm would like to improve its reputation. In this perspective, the legitimacy theory can also be seen as the economic approach.

According to the stakeholder theory, an organization should acknowledge the needs of its stakeholders in order to achieve the firm’s objectives. In most organizations the government is an important stakeholder (Clarkson, 1995). An organization must follow the rules set by the government and must pay taxes to the government. Christensen and Murphy (2004) argue that receiving taxes can be seen as an important necessity of the government. Further, if it is important to stakeholders that the organization pays a fair amount of taxes, firms with an interest in CSR practices are less likely to use corporate tax avoidance (Hoi et al., 2003). In addition, the legitimacy theory argues that a company has one social contract with society as a whole. To obtain legitimacy, an organization should align its value system with the value system of society. If society believes that an organization should pay a fair amount of taxes, the organization will have to comply with them according to the legitimacy theory (Sikka, 2010). In other words, both theories could also be seen as a variant of the ethical direction. It is important, however, to acknowledge whether an organization is willing to improve its CSR activities at all costs. As long as an organization is engaging in CSR activities only because they are in the self-interest of the organization, the behaviour belongs under the economic direction.

To demonstrate their corporate social responsibility activities to their stakeholders and society in general, companies often issue a sustainability report.

(14)

2.2 Sustainability Reporting

In this sub-section, sustainability reporting in general is discussed. Then the Global Reporting Initiative guidelines are presented. finally, the importance of assurance will be addressed.

2.2.1 Definition

According to the Global Reporting Initiative (2016) a sustainability report is “a report published

by a company or organization about the economic, environmental and social impacts caused by its everyday activities. A sustainability report also presents the organization's values and governance model, and demonstrates the link between its strategy and its commitment to a sustainable global economy”. In other words, a sustainability report addresses organizational performance towards the goal of sustainability development, while being accountable to the primary and secondary stakeholders. Sustainability reporting could be considered a synonym for corporate social responsibility reporting, where the results and improvements of CSR activities are published. In this manner, all the stakeholders are able to verify whether the CSR activities are at the level they require.

2.2.2 Global Reporting Initiative

In 1997 the Global Reporting Initiative (GRI) was founded by a number of organizations that were part of the Coalition for Environmentally Responsible Economies (CERES)5_{. The GRI is a} non-profit organization that provides standards on sustainability reporting that are widely used around the globe. The organization’s goal is to create a future where sustainability is integral to every decision-making process. By the establishment of sustainability standards, the GRI strives for a more sustainable economy and world. The standards of the GRI will improve the quality of sustainability reports for a number of reasons. First, by establishing specific guidelines a certain form of standardization will be fostered (Steurer, 2010). This standardization will improve quality because standardized reports will be easier to compare and organizations will not have the opportunity to selectively choose which topics they want to address and which they want to omit in the report. In other words, the emphasis on standardization will have the effect that a more comprehensive sustainability report will be issued. Second, with the use of the GRI guidelines a more transparent sustainability report will be developed. Transparency will improve reporting quality because stakeholders who are interested in the sustainability report will be better able to observe the quality of CSR activities (Kolk, 2008). An organization may issue its sustainability report according to the GRI framework for two reasons. First, an organization could use the

(15)

GRI framework because it actually highly values society and the environment, which in this research is referred to as the ethical direction. Such an organization is likely to be less inclined to engage in tax avoidance activities. Second, an organization may use the GRI framework for economic reasons. Such an organization tries to legitimize its CSR policies by using the GRI framework to satisfy the shareholders and stakeholders but it is not actually interested in society or the environment. It is likely that such organizations are not particularly concerned with paying a fair amount of taxes. In other words, these organizations are likely to engage in tax avoidance activities.

There has been some criticism in the past about the quality of the sustainability reports. Some organizations issue a sustainability report without using a proper guideline. This could have the effect that only positive performance is shown, with any negative performance being eliminated from the report. In this way, stakeholders would not get a comprehensive view of the sustainability performance. Besides, several organizations use the GRI guidelines but these reports contain some problems, too. First, there was some critique that several organizations do report CSR performance but do not truly take action. They act as if they are concerned with CSR activities but they report only to legitimize their behaviour to the stakeholders and society. These organizations are not concerned with society but are simply trying to improve their economic performance. Second, a number of organizations do not apply the GRI guidelines in the correct manner. These organizations only use some of the guidelines or they apply the guidelines in their own manner. This reduces the report quality because the reports then become less trustworthy and are less suitable for comparison (Brammer & Pavelin, 2006; Morhardt, Baird & Freeman, 2002). A possible solution for reducing these problems with the GRI guidelines is assurance. Hodge, Subramaniam and Stewart (2009) concluded that assurance increased the perceived reliability of the sustainability report by users. Further, Ioannou and Serafeim (2014) argued that organizations employ more ethical practices in countries where assurance of sustainability reports is more widespread. There is a dilemma here as well. Some organizations demand assurance of their sustainability reports for ethical motivations, other organizations want their sustainability reports to be assured for legitimizing purposes because it is economically the best choice to do so. The organizations with ethical motivations are likely to be less inclined to engage in tax avoidance activities. The organizations with economical motivations are likely to be interested in tax avoidance behaviour because it makes sense economically.

(16)

2.3 Corporate Tax Avoidance

This section addresses corporate tax avoidance, discussing its positive effects and negative effects.

Previously, there has been some inconsistency about what exactly falls under corporate tax avoidance. In addition, terms such as tax sheltering, tax evasion, tax aggressiveness and tax non-compliance are often used as synonyms for tax avoidance (Hanlon & Heitzman, 2008). According to Dyreng et al. (2008), corporate tax avoidance can be seen as “the ability to pay a low amount of tax per dollar of reported pre-tax financial accounting income” (p. 3). This definition is relatively broad because it basically entails every measure that has a decreasing effect on the amount of taxes per dollar of income. Subsequent definitions elaborated on this explanation. Chen, Chen, Cheng and Shevlin (2010) argued that tax avoidance “encompasses tax planning activities that are legal, or that may fall into the grey area, as well as activities that are illegal” (p. 42). This definition is different from that given by Dyreng et al. (2008) in that it takes the illegal component into account. Furthermore, Hanlon and Heitzman (2010) defined tax avoidance as “the reduction of explicit taxes” (p. 137). This definition is different from that used by Dyreng et al. (2008) because it includes illegal activities and it distinguishes different types of tax avoidance. Tax evasion is somewhat easy to understand, as it represents the illegal spectrum of tax avoidance. Tax aggressiveness could be seen as a wide range of transactions with the primary goal of lowering the tax liability without involving a real response by the firm. Further, tax aggressiveness is thought of as stretching the legal concept of tax avoidance. Further, tax shelters are defined as methods such as investments with the sole purpose of lowering the tax rate (Hanlon & Heitzman, 2010). Finally, tax non-compliance can be defined with the following statement. “When taxpayers try to find loopholes with the intention to pay less tax, even if technically legal, their actions may be against the spirit of the law and in this sense considered non-compliant” (Wenzel, 2002; p, 2). This research follows the definition of Dyreng et al. (2008) because the dependent variable used will be the long-run cash effective tax rate, which has also been used in the research of Dyreng et al. (2008).

According to Fischer (2014), tax avoidance activities are deployed for one of three reasons. First, an organization can pay less tax than is actually required as a result of a certain interpretation of a country’s law. For example, in the Netherlands, organizations are able to make tax arrangements with the government6_{; this could have the effect that certain organizations pay less tax than} actually required. Second, several organizations declare their profits in a country other than

(17)

where the profits were actually earned. A way of doing this is with profit shifting through transfer prices. Profit shifting can be referred to as the allocation of income and expenses among different entities of the same organization, for example with the use of transfer pricing, in order to reduce the tax liability of the organization (Harris, Morck & Slemrod, 1993). Third, some organizations make their tax payment somewhat later than when the profits were earned. This has the effect that the net income for the current period is higher because the tax liability is reduced.

Corporate tax avoidance can be profitable but it also has some potential risks. In research from the Citizens for Tax Justice (2014) it was demonstrated that the use of tax avoidance measures has led to a total income tax rate of 19.4% between 2008 and 2012 for the profitable firms of the Fortune 500. The statutory tax rate of the United States is 35%. In other words, the use of corporate tax avoidance has led to a diminishing of almost half of the tax rate. This enormous economic benefit does not come without risks. The two main risks addressed in this thesis are the political costs hypothesis and reputational costs. First, according to Watts and Zimmerman (1978), political costs are potential costs that can be imposed on organizations by external parties as a result of political actions. The political sector is incentivized to reallocate wealth. Organizations do probably shift income from one period to another. This will lower the profits and thereby avert attention from politicians. Corporate tax avoidance can be used to influence potential political costs. When more corporate tax avoidance is applied, net income rises, which will increase the scrutiny of politicians and could increase political costs. Second, in all likelihood the biggest concern organizations currently have is the fear of loss of image. It takes a very long time and a tremendous amount of money to build a brand. However, one negative news item and the organizational brand value is diminished. Therefore, organizations are intensely concerned about the risks of brand loss and the reputational costs that come with it (Rangaswamy, Burke & Oliva, 1993). According to Austin and Wilson (2013), organizations with higher brand values pay a higher rate of taxes. This could be due to the reputational risk of tax avoidance. As argued by Hanlon and Slemrod (2009), reputational costs can cause negative stock price reactions. Hanlon and Slemrod (2009) also found that news about organizations with more corporate tax avoidance resulted in a decline of the stock price. Besides these risks, tax avoidance behaviour is often seen as unethical and irresponsible. According to Christensen and Murphy (2004, p37), “tax revenues are the lifeblood of democratic government and the social contract, vital to the development and maintenance of physical infrastructure and to the sustenance of the infrastructure of justice that underpins liberty and the market economy”. In other words, corporate tax avoidance has a negative influence on society and its citizens because

(18)

the money available to governments will be impaired. This will have the effect that governments will have fewer resources to enrich the habitats of its citizens. Further, many stakeholders see corporate tax avoidance as irresponsible and unethical. Several stakeholders require information about tax avoidance in the sustainability reports. Therefore, an organization must take this into consideration when devising its CSR strategy (Dyreng et al., 2008). As argued in the sections above, this research distinguishes two directions of CSR, the economic direction and the ethical direction. According to the literature used in this section it is clear that despite the risks it entails, corporate tax avoidance is economically profitable. For this reason, it is predicted that organizations with CSR policies based on the economic direction are likely to be engaged in corporate tax avoidance activities. Further, it is mentioned that corporate tax avoidance is seen as unethical and irresponsible. Therefore, organizations with CSR policies based on the ethical direction are expected to be less involved with corporate tax avoidance activities.

(19)

2.4 Hypotheses

This research aims to provide a broad explanation of the research question and related topics . In order to provide this explanation, three hypotheses are required.

Sustainability reporting can be seen as describing the performance of CSR activities to the stakeholders and society. In addition, organizations with a sustainability report are considered to be more aware of their corporate social responsibilities. Further, many stakeholders see corporate tax avoidance as unethical and require organizations to take this into account when developing their CSR policies. Therefore, this thesis argues that organizations using a sustainability report are less likely to develop corporate tax avoidance activities (Wieriks, 2013).

1. Organizations with a sustainability report make less use of corporate tax avoidance than organizations without a sustainability report.

Hedberg and Von Malmborg (2003) argued that the GRI guidelines are primarily used to increase credibility of CSR and not actually to improve CSR guidelines. Furthermore, Isaksson and Steimle (2009) asserted that the GRI guidelines are not sufficient to make sustainability reporting relevant and clear. On the other hand, GRI guidelines have the effect that reports are more standardized and more transparent (Willis, 2003). Therefore, it is argued that sustainability reports prepared according to GRI guidelines are generally of higher quality. Because these reports are more transparent, organizations do have fewer opportunities to credibly engage in corporate tax avoidance activities. For this reason, it is argued that organizations with a sustainability report issued in accordance with the GRI probably engage less in corporate tax avoidance activities (GRI, 2006).

2. Organizations with sustainability reports prepared according to the Global Reporting Initiative make less use of corporate tax avoidance than organizations with sustainability reports prepared without using the Global Reporting Initiative.

Sustainability reports on which assurance has been done are usually of greater perceived reliability. This perception derives from the thought that an assured sustainability report does not give an organization the opportunity to misrepresent information, therefore users usually value such reports as being more reliable (Hodge et al., 2009). If the organization chooses to have

(20)

assurance done on its sustainability reports, it usually attaches more value to its CSR activities because it spends more money on them. Ioannou and Serafeim (2014) stated that organizations employ more ethical practices in countries where assurance of sustainability reports is more widespread. If organizations require assured sustainability reports for ethical motivations, then these organizations are less likely to be inclined to engage in corporate tax avoidance.

3. Organizations with assurance of their sustainability reports make less use of corporate tax avoidance in comparison with organizations using sustainability reporting without assurance.

(21)

3 Methodology

This chapter outlines the methodology that is used in this thesis. First the sample will be addressed and then the variables used are explained. A discussion of the research design concludes this chapter.

3.1 Sample

For examining the research question, database research has been conducted. The sample consists of Standard & Poor’s 500 companies for the period of 2013 until 2015. The S&P 500 was chosen because of the ready availability of the data. The period starts from 2013 as from this year sustainability reporting was quite heavily used among S&P 500 companies. The database of the Global Reporting Initiative has been used for information about sustainability reports. In addition, the websites of certain companies have been consulted for sustainability reports. Information about corporate tax avoidance was found in the Compustat database. In order to measure the effect of sustainability reports on corporate tax avoidance, this thesis first identified which organizations of the S&P 500 have issued a sustainability report in the period of 2013 until 2015. If the organization has issued a sustainability report, it was checked for preparation in accordance with the GRI guidelines and whether some sort of assurance is provided of the report. In order to create a reliable connection with the dependent variable, organizations must have issued the same sort of sustainability report for at least two-thirds of the sample period. Further, organizations with missing data were excluded from the sample.

After merging the GRI database with the Compustat information, the S&P 500 data from 2013 through 2015 consisted of 1,862 observations. Due to special regulations with respect to financial institutions, companies with SIC codes between 6000 and 7000 were deleted from the sample. This resulted in the sample of 1,310 observations. Further, the long-run cash effective tax rate has been measured over three years. For this reason, it was determined to delete companies from the sample that didn’t have three years of data because of, for instance, bankruptcy or acquisitions. This had a decreasing effect on the sample volume; after this action the sample consisted of 1,209 observations. Moreover, empty values of research and development, intangible assets and property, and plant and equipment are adjusted to zero. Finally, with respect to the long-run cash effective tax rate, the organizational information was measured as the mean of the entire sample period. In order to measure the control variables and the sustainability data, the means of these variables must be computed as well. The sustainability reports are measured as a dummy variable, in which the organization will be given a 1 if they

(22)

have issued a sustainability report in two of the three sample years. Whether the report was issued in accordance with the Global Reporting Initiative and whether assurance is provided is calculated in the same manner. However, due to this calculation the sample has decreased to one observation for each company, which resulted in a final sample of 403 observations for Hypothesis 1.

For Hypothesis 2, the dependent variable will be the long-run cash effective tax rate as well. Furthermore, the fact of whether the sustainability report has been issued in accordance with the Global Reporting Initiative will serve as the main independent variable. However, for a sustainability report to be in accordance with the Global Reporting Initiative, it must of course exist. Therefore, the organizations that have not issued a sustainability report were deleted from the sample. This resulted in a final sample of 202 observations for Hypothesis 2.

Hypothesis 3 likewise uses the long-run cash effective tax rate as the dependent variable. Furthermore, the fact of whether assurance is provided for the sustainability report will be the main independent variable in this hypothesis. This variable is measured as a dummy variable, in which organizations receive the value of one if assurance is provided of their sustainability report in at least two of the three sample years. However, assurance is only provided of sustainability reports that have been prepared in accordance with the Global Reporting Initiative. Therefore, the final sample of Hypothesis 3 consisted of 142 observations.

3.2 Research design

The variable corporate tax avoidance is impossible to measure directly. Therefore, it is necessary to use a proxy for this variable. In the literature a distinction is made among various proxies of corporate tax avoidance. However, some of these proxies do also measure tax sheltering, tax evasion and tax planning. To measure corporate tax avoidance, researchers commonly use Cash ETR, GAAP ETR and Current ETR (Dyreng et al., 2010; Hope, Ma & Thomas, 2013). In this thesis the long-run Cash ETR of Dyreng et al. (2008) is used. Using this proxy has many advantages compared to other proxies such as Cash ETR, GAAP ETR or Current ETR. First, the LRETR proxy is most accurate in measuring corporate tax avoidance in the long term (Dyreng et al., 2008). This is important because all the variables in this thesis are measured as a single observation for each organization. All the firm-year observations have been merged into one observation. Therefore, the LRETR proxy has the main advantage that it is the best proxy to measure long-term corporate tax avoidance within one observation for each organization. Second, it is argued that organizations issue sustainability reports because of long-term reasoning. In other words, a company hypothetically desires to add value to society over a long

(23)

term and issues a sustainability report to reflect that. For this reason, tax avoidance should also be measured over a longer period. Third, because the LRETR method measures tax avoidance over multiple years, the tax expenses are more solid. Organizations are able to use accounting mechanisms to reduce or increase tax expenses in the current year relative to next year. For instance, firms can shift income from this year to next year, which reduces the net income of the firm and reduces the tax expense. Therefore, the LRETR proxy measures tax expenses more accurately than other proxies. Conclusively, this thesis argues that the LRETR proxy is the proxy most suitable for the research design used here.

The effective tax rate is measured over a three-year period. This is done by dividing the organization’s total cash taxes paid by the sum of its total pre-tax income reduced by the special items. Because this measure takes three years into account it is able to reflect the effective tax rate more closely. Hopefully, the use of this measure will have the effect that payments that relate to long-ago tax disputes are included. The formula that will be used for the long-run cash effective tax rate is the following:

𝐿𝑅𝐸𝑇𝑅7

𝑖𝑡 =

∑𝑡 𝐶𝑎𝑠ℎ 𝑡𝑎𝑥 𝑝𝑎𝑖𝑑_𝑖𝑡

𝑡=𝑡−3

∑𝑡𝑡=𝑡−3(𝑃𝑟𝑒𝑡𝑎𝑥 𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑡− 𝑆𝑝𝑒𝑐𝑖𝑎𝑙 𝑖𝑡𝑒𝑚𝑠𝑖𝑡)

In order to provide reliable research, certain control variables8_{are included. First, according to} Zimmerman (1983), larger firms face greater political costs in the form of higher tax payments. It is argued that larger firms are more inclined to engage in tax avoidance activities (Mills, Nutter & Schwab, 2012). Therefore, Size (SIZEi,t) is included in the control variables. Second,

Leverage (LEVi,t) is added to the control variables because organizations which use debt

financing may not need to engage in tax avoidance activities due to the deductible interest expense (Desai & Dharmapala, 2009). Third, organizations with relatively high profits could have more incentives to engage in tax avoidance activities (Chen et al., 2010). For this reason, the return on assets (ROAi,t) will be controlled. Fourth, growth opportunities (MTBi,t) are

controlled because growing firms make more significant investments in tax-favoured assets (Manzon & Plesko, 2001). Fifth, the presence of a net operating loss carry forward (NOLi,t) and

the direction of the change in the net operating loss balance (∆NOLi,t) are being controlled in

this research because loss carry forwards will give some tax benefits to the organization (Hope et

7_{In Appendix B the calculation of the proxy for corporate tax avoidance is explained in more detail.} 8_{Appendix B provides a more extensive explanation for the calculation of the control variables.}

(24)

al., 2013). Sixth, capital-intensive firms are more subject to the different depreciation expense methods for tax and financial reporting purposes. For this reason, property, plant and equipment (PPEi,t) is also included in the control variables. Furthermore, research and development

(RDi,t), intangible assets (INTANGi,t) and equity in earnings (EQINC_i,t) are included in the

control variables because of the different book and tax treatments of intangible assets and consolidated earnings according to the equity method (Rego & Wilson, 2012).

To test whether having a sustainability report is associated with tax avoidance, the following regression model is used for Hypothesis 1:

𝐿𝑅𝐸𝑇𝑅_𝑖,𝑡 = 𝛽₀+ 𝛽₁𝑆𝑈𝑆𝑅𝐸𝑃𝑂𝑅𝑇_𝑖,𝑡+ 𝛽₂𝑆𝐼𝑍𝐸_𝑖,𝑡+ 𝛽₃𝐿𝐸𝑉_𝑖,𝑡+ 𝛽₄𝑅𝑂𝐴_𝑖,𝑡+ 𝛽₅𝑀𝑇𝐵_𝑖,𝑡 + 𝛽₆𝑁𝑂𝐿_𝑖,𝑡+ 𝛽₇∆𝑁𝑂𝐿_𝑖,𝑡+ 𝛽₈𝑃𝑃𝐸_𝑖,𝑡 + 𝛽₉𝑅𝐷_𝑖,𝑡+ 𝛽₁₀𝐼𝑁𝑇𝐴𝑁𝐺_𝑖,𝑡 + 𝛽₁₁𝐸𝑄𝐼𝑁𝐶_𝑖,𝑡 + 𝜀_𝑖,𝑡

Here, “i” stands for the organization and “t” stands for the year. The dependent variable tax avoidance (LRETRi,t) is measured by the long-run cash effective tax rate. A positive relationship

between the independent variables and the LRETR proxy means that the independent variables have a positive influence on the tax rate. In other words, a positive relation between the independent variables and the LRETR proxy means a negative relation between the independent variables and corporate tax avoidance. The main independent variable (SUSREPORTi,t) is

measured as a dummy variable in which it is listed with a value of one if the organization has issued a sustainability report in at least two of three sample years. The variable is listed as a zero if the organization has not issued a sustainability report in one of the three years or only published such a report in one of the sample years. This distinction is made to ensure that the organization has not incidentally issued a sustainability report.

To test Hypothesis 2 the variable (GRIi,t) is used as the main independent variable:

𝐿𝑅𝐸𝑇𝑅𝑖,𝑡 = 𝛽0+ 𝛽1 𝐺𝑅𝐼𝑖,𝑡+ 𝛽2 𝑆𝐼𝑍𝐸𝑖,𝑡+ 𝛽3 𝐿𝐸𝑉𝑖,𝑡+ 𝛽4 𝑅𝑂𝐴𝑖,𝑡+ 𝛽5 𝑀𝑇𝐵𝑖,𝑡+ 𝛽6 𝑁𝑂𝐿𝑖,𝑡

+ 𝛽7 ∆𝑁𝑂𝐿𝑖,𝑡+ 𝛽8 𝑃𝑃𝐸𝑖,𝑡+ 𝛽9 𝑅𝐷𝑖,𝑡+ 𝛽10 𝐼𝑁𝑇𝐴𝑁𝐺𝑖,𝑡 + 𝛽11 𝐸𝑄𝐼𝑁𝐶𝑖,𝑡+ 𝜀𝑖,𝑡

The sample of the second hypothesis is somewhat smaller than the first. This derives from the fact that (GRIi,t) can only be measured if the organization has issued a sustainability report in

two of the sample years. Therefore, observations that are listed with a zero for the variable (SUSREPORTi,t) have been deleted. The measure of Global Reporting Initiative guidelines

adherence, (GRIi,t), is measured as a dummy in which it is assigned a value of one if the

(25)

guidelines in at least two of the three sample years. An organization measures a zero here if the sustainability report was not prepared in accordance with the GRI guidelines in at least two of three sample years.

To test Hypothesis 3, the variable (ASSURANCEi,t) is used as the main independent variable:

𝐿𝑅𝐸𝑇𝑅_𝑖,𝑡 = 𝛽₀+ 𝛽₁𝐴𝑆𝑆𝑈𝑅𝐴𝑁𝐶𝐸_𝑖,𝑡 + 𝛽₂𝑆𝐼𝑍𝐸_𝑖,𝑡+ 𝛽₃𝐿𝐸𝑉_𝑖,𝑡+ 𝛽₄𝑅𝑂𝐴_𝑖,𝑡+ 𝛽₅𝑀𝑇𝐵_𝑖,𝑡 + 𝛽₆𝑁𝑂𝐿_𝑖,𝑡+ 𝛽₇∆𝑁𝑂𝐿_𝑖,𝑡+ 𝛽₈𝑃𝑃𝐸_𝑖,𝑡+ 𝛽₉𝑅𝐷_𝑖,𝑡+ 𝛽₁₀𝐼𝑁𝑇𝐴𝑁𝐺_𝑖,𝑡 + 𝛽₁₁𝐸𝑄𝐼𝑁𝐶_𝑖,𝑡 + 𝜀_𝑖,𝑡

The sample for this third hypothesis is smaller than that used for the second hypothesis. This is because within the GRI database each organization that has commissioned assurance of their sustainability reports had issued their report in accordance with the GRI guidelines. In other words, organizations which are listed as a zero at the variable (GRIi,t) within the second

hypothesis are deleted from the sample for Hypothesis 3. The main independent variable (ASSURANCEi,t) is measured as a dummy variable, as well as (SUSREPORTi,t) and (GRIi,t).

In this dummy, an organization is valued at one if it has assured its sustainability report in at least two of three sample years. If not, the observation in that case will be stated as zero.

(26)

4 Empirical results

This chapter outlines the descriptive and correlating statistics measured and results for each of the three hypotheses stated.

4.1 Hypothesis 1

4.1.1 Descriptive statistics

The long-run cash effective tax rate is used as a proxy for corporate tax avoidance as mentioned in the third chapter. In the table below the descriptive statistics for the first hypothesis are presented. The long-run cash effective tax rate (LRETRi,t), the dummy for sustainability reports

(SUSREPORTi,t) and all the control variables are included in this table. Table 1: Descriptive statistics for hypothesis 1

Variable Observations Mean Standard Deviation Minimum Maximum

LRETR 403 .213 .160 0 1 SUSREPORT 403 .501 .501 0 1 SIZE 403 4.756 .399 3.968 5.843 LEV 403 .247 .174 0 .781 ROA 403 .075 .075 -.148 .310 MTB 403 13.677 18.364 -35.724 108.422 NOL 403 .581 .494 0 1 ∆NOL 403 .045 .103 0 .593 PPE 403 .306 .312 .007 1.107 RD 403 .019 .034 0 .157 INTANG 403 .230 .237 0 .960 EQINC 403 .002 .005 -.002 .031

All the variables contain 403 observations. In other words, data on all the variables has been collected for 403 organizations of the S&P 500. These organizations have an average long-run cash effective tax rate of 21.3%. This long-run cash effective tax rate is comparable with those used in the research of Dyreng et al., (2008). The mean of (SUSREPORTi,t) is 0.501; this

implies that 50.1 per cent of the organizations in the sample have issued a sustainability report in at least two of the three sample years. Further, (SIZEi,t) has a mean value of 4.756, (LEVi,t) is

(27)

and (NOLi,t) has a mean of 0.581. This implies that 58.1% of the companies within the sample

do have a net operating loss. The (∆NOLi,t) is 0.045 on average, the (PPEi,t) has a mean value

of 0.306, and (RDi,t) has an average value of 0.019. Lastly, (INTANGi,t) has a mean value of

0.230 and (EQINC_i,t) is 0.002 on average, indicating that the equity income is almost negligible.

4.1.2 Correlation Matrix

The Pairwise correlation matrix for Hypothesis 1 is presented in Table 2. This matrix measures the strength of the relationship between the dependent variable (LRETRi,t) , the main

independent variable (SUSREPORTi,t) and the control variables. The underlined numbers are

the p-values of the correlation. The bold p-values are significant at the 5% level.

Table 2: Correlation Matrix for hypothesis 1

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) LRETR (1) 1.000 - SUSREPORT (2) 0.001 1.000 0.991 - SIZE (3) -0.007 0.261 1.000 0.895 0.000 - LEV (4) -0.191 0.069 0.008 1.000 0.000 0.169 0.873 - ROA (5) 0.219 0.065 -0.033 0.094 1.000 0.000 0.192 0.509 0.058 - MTB (6) 0.054 0.035 -0.113 -0.019 0.158 1.000 0.279 0.486 0.023 0.702 0.002 - NOL (7) 0.027 -0.043 -0.206 -0.086 -0.004 0.050 1.000 0.596 0.388 0.000 0.085 0.940 0.317 - ∆NOL (8) -0.182 -0.013 -0.110 0.165 -0.092 0.031 0.371 1.000 0.000 0.791 0.028 0.001 0.065 0.535 0.000 - PPE (9) -0.212 0.170 0.104 0.475 -0.127 -0.143 -0.269 0.039 1.000 0.000 0.001 0.037 0.000 0.011 0.004 0.000 0.437 - RD (10) -0.102 0.028 -0.045 -0.134 0.191 0.185 0.088 0.120 -0.267 1.000 0.041 0.581 0.371 0.007 0.000 0.000 0.076 0.016 0.000 - INTANG (11) 0.091 -0.019 0.040 0.291 0.216 0.000 0.150 0.075 -0.331 0.099 1.000 0.069 0.711 0.427 0.000 0.000 0.998 0.003 0.131 0.000 0.047 - EQINC (12) 0.134 0.099 0.139 0.076 0.204 0.022 -0.033 -0.006 0.138 -0.066 -0.004 1.000 0.007 0.046 0.005 0.130 0.000 0.658 0.508 0.904 0.006 0.186 0.936 -

(28)

As stated in Table 2, almost half of the p-values are significant. For this hypothesis the most important independent variable would be (SUSREPORTi,t) . Surprisingly, the correlation

between (SUSREPORTi,t) and (LRETRi,t) is extremely low (0.001) and highly insignificant

(0.991). This means that the strength of the relationship between the main independent variable and the dependent variable is exceptionally weak. Furthermore, the control variables are not highly correlated with each other. The highest correlation is 0.475 between (PPEi,t) and

(LEVi,t), which means that a higher amount of property, plant and equipment leads to a higher

amount of leverage. This occurs when a large part of the property, plant and equipment is bought with borrowed money. Because most of the control variables are not highly correlated with each other, the risk of multicollinearity is reduced. To ensure that there is no multicollinearity threat, the variance inflation factor (VIF) is tested. The mean VIF is 1.39, with 2.18 as the highest factor. According to some STATA web books9_{a VIF of higher than 10} indicates multicollinearity. For this reason, the multicollinearity risk is ruled out for Hypothesis 1.

4.1.3 Results

The first hypothesis is formulated as: Organizations with a sustainability report make less use of corporate

tax avoidance than organizations without a sustainability report. This is tested by (SUSREPORTi,t) as the

independent variable and (LRETRi,t) as the dependent variable. Here, the long-run cash

effective tax rate serves as a proxy for corporate tax avoidance. The long-run cash effective tax rate is winsorized at the range of [0,1], and the other variables are winsorized at the 1st_{and 99}th percentiles. Significants of the coefficients are indicated with *** for 1%, ** for 5% and * for 10%.

(29)

The coefficient of the main independent variable (SUSREPORTi,t) on (LRETRi,t) is positive

(0.007) but insignificant (0.665). The results are not in line with the expectation stated in the first hypothesis, in which a significant positive relation was expected. Several possible explanations exist for this result. First, the outcome could be due to the small sample size, with only 403 observations. Second, the measurement approach for the dummy variable (SUSREPORTi,t) could provide an explanation. In order to be assigned a value of one for this dummy variable, organizations must have issued a sustainability report in two of the three sample years. Requiring a sustainability report in all the sample years could possibly lead to significant results. These are merely limitations of this thesis. However, several other possible factors may also have contributed to this insignificant result. First, according to Bridges and Wilhelm (2008), organizations have incentives to develop sustainability reports for marketing purposes. As Lindblom (1994) argued, organizations could try to legitimize their behaviour to be congruent to the values of society. However, this could have the effect in some cases that sustainability reports do not actually change the behaviour of organizations but are issued to increase the sales and reputation of the organization. Second, organizations implement sustainability reports for different reasons. Many organizations enact corporate social responsibility activities to benefit society even if it is at the expense of their own stakeholders. However, some organizations will not engage in CSR activities if it is not lucrative for the stakeholders (Freeman, 1984). These organizations would also use tax avoidance to increase their profits and thereby increase the

Table 3: Results hypothesis 1 – LRETR

Variable Coefficient T - statistic P - Value Number of Observations 403

SUSREPORT 0.007 0.43 0.665 F - value 7.440

SIZE -0.010 -0.48 0.628 Prob > F 0.000

LEV -0.179*** -3.05 0.002 R-squared 0.173

ROA 0.421*** 3.91 0.000 Adjusted R-squared 0.150

MTB 0.000 0.73 0.466 Root MSE 0.148 NOL 0.009 0.50 0.616 ∆NOL -0.192** -2.38 0.018 PPE -0.059* -1.69 0.092 RD -0.925*** -3.89 0.000 INTANG 0.063 1.57 0.117 EQINC 3.598** 2.33 0.021 Constant 0.283*** 2.90 0.004

(30)

stock price, which is beneficial to the stakeholders (Fischer, 2014). These factors could explain the insignificant connection between sustainability reports and corporate tax avoidance.

Table 3 also exhibits information about the control variables and their relation with the (LRETRi,t) as the dependent variable. Some of the control variables are significant and are in

line with the relationship, which was expected. However, several control variables are not in line with the predicted relationship. First, (SIZE_i,t) is expected to have a negative relationship with (LRETRi,t) because a bigger organization would have higher political costs in the form of tax

payments and is therefore more inclined to engage in tax avoidance activities (Mills et al., 2012) and therefore have a lower long-run cash effective tax rate. Although the relationship between (SIZEi,t)and (LRETRi,t) is indeed negative, it is insignificant and therefore does not meet the

expectations. Second, the control variable (LEV_i,t) is significantly negative with (LRETRi,t) and

therefore not in line with the predicted outcome. However, this result is in line with some literature such as Francis, Hasan, Wu and Yan (2014). Third, an unexpected and strange result is the outcome of (ROA_i,t) . Organizations with high profits were expected to have more incentives to engage in tax avoidance activities. However, the relation between (ROAi,t) and

(LRETRi,t) is significantly positive and therefore contrary to the expectation. Fourth,

(∆NOL_i,t)is expected to have a significant negative relation with (LRETRi,t). The actual result is

in line with this expectation. Fifth, more PPE is associated with more use of tax avoidance. Therefore a significant negative relationship between (PPEi,t) and (LRETRi,t) is expected. The

outcome conforms with the expectation. Sixth, (RDi,t) is expected to have a negative

relationship with (LRETRi,t) . The results are in accordance with the expectations. Last,

(EQINC_i,t) is expected to be negative and significant. The results are in contrast with these predictions, as they are positive and significant. It could be the case that this result is due to the small sample size of this research design.

(31)

4.2 Hypothesis 2

4.2.1 Descriptive statistics

The long-run cash effective tax rate is used as a proxy for corporate tax avoidance as mentioned in the third chapter. In the table below, the descriptive statistics for the second hypothesis are presented. The long-run cash effective tax rate (LRETRi,t), the dummy for sustainability reports

prepared in accordance with the Global Reporting Initiative (GRI_i,t) and all the control variables are included in this table.

Table 4: Descriptive statistics for hypothesis 2

Variable Observations Mean Standard Deviation Minimum Maximum

LRETR 202 .216 .158 0 1 GRI 202 .703 .458 0 1 SIZE 202 4.859 .402 3.978 5.843 LEV 202 .259 .163 0 .742 ROA 202 .080 .078 -.148 .310 MTB 202 14.313 17.075 -35.724 108.422 NOL 202 .559 .498 0 1 ∆NOL 202 .044 .101 0 .593 PPE 202 .359 .334 .007 1.107 RD 202 .0196 .032 0 .157 INTANG 202 .226 .230 0 .960 EQINC 202 .002 .006 -.002 .031

All the variables contain 202 observations. The sample in Hypothesis 2 is almost half of the sample used in Hypothesis 1. This is a result of excluding the organizations that have not issued a sustainability report in at least two of the three sample years. These observations are deleted because the main independent variable in this hypothesis is a dummy variable which tests whether the sustainability report is prepared in accordance with the GRI guidelines (GRIi,t).

Obviously, a sustainability report can only be prepared in accordance with GRI guidelines if it is actually issued. The organizations in the sample used for Hypothesis 2 have an average long-run cash effective tax rate of 21.6%. This long-run cash effective tax rate is comparable with those used in the research of Dyreng et al., (2008) and is also comparable to the tax rate of the first hypothesis. In addition, the mean of (GRIi,t) is 0.703, which implies that 70.3% of the

organizations in the sample have issued a sustainability report in accordance with the GRI guidelines in at least two of the three sample years. Put differently, only 29.7% of the sample has issued two or more sustainability reports which were not in accordance with the GRI framework.

(32)

Further, the mean values of the control variables are highly comparable with those of the first hypothesis.

4.2.2 Correlation Matrix

The Pairwise correlation matrix for Hypothesis 2 is presented in Table 5. This matrix measures the strength of the relationship between the dependent variable (LRETRi,t) , the main

independent variable (GRI_i,t) and the control variables. The underlined numbers are the p-values of the correlation. The bold p-p-values are significant at the 5% level.

Table 5: Correlation Matrix for hypothesis 2

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) LRETR (1) 1.000 - GRI (2) -0.014 1.000 0.846 - SIZE (3) -0.005 0.058 1.000 0.949 0.412 - LEV (4) -0.263 0.013 0.021 1.000 0.000 0.851 0.766 - ROA (5) 0.272 -0.060 0.011 -0.014 1.000 0.000 0.395 0.876 0.847 - MTB (6) 0.075 0.022 -0.096 0.061 0.297 1.000 0.289 0.755 0.173 0.388 0.000 - NOL (7) 0.060 0.034 -0.273 -0.118 0.003 0.138 1.000 0.398 0.630 0.000 0.094 0.962 0.051 - ∆NOL (8) -0.179 0.052 -0.123 0.173 -0.098 0.018 0.385 1.000 0.011 0.459 0.081 0.014 0.165 0.803 0.000 - PPE (9) -0.162 0.166 0.113 0.501 -0.177 -0.172 -0.307 -0.044 1.000 0.021 0.018 0.111 0.000 0.012 0.014 0.000 0.532 - RD (10) -0.046 0.041 0.047 -0.143 0.284 0.105 0.122 0.123 -0.391 1.000 0.513 0.559 0.505 0.043 0.000 0.137 0.084 0.083 0.000 - INTANG (11) 0.088 -0.133 -0.010 0.239 0.239 0.041 0.168 0.183 -0.380 0.193 1.000 0.212 0.059 0.885 0.001 0.001 0.564 0.017 0.009 0.000 0.006 - EQINC (12) 0.182 0.005 0.149 0.002 0.206 0.067 -0.028 -0.038 0.138 -0.083 -0.014 1.000 0.010 0.943 0.034 0.978 0.003 0.341 0.691 0.590 0.050 0.239 0.845 -

As stated in Table 5, almost half of the p-values are significant. For this hypothesis the most important independent variable would be (GRIi,t). Surprisingly, the correlation between (GRIi,t)

Sustainability reporting and corporate tax avoidance