2015 DETERMINANTS OF TAX TRANSPARENCY IN EUROPEAN LISTED FIRMS

(1)

DETERMINANTS OF TAX TRANSPARENCY IN

EUROPEAN LISTED FIRMS

by

BOB HOLLAND

Prof. dr. I.J.J. Burgers, Supervisor

Thesis in Accountancy & Controlling, for the

requirements of the Degree of Master of Science

UNIVERSITY OF GRONINGEN

Faculty of Economics & Business

(2)

2 | M a s t e r ’ s t h e s i s A c c o u n t a n c y & C o n t r o l l i n g , 2 0 1 5

DETERMINANTS OF TAX

TRANSPARENCY IN EUROPEAN LISTED

FIRMS

Personal information

Name Bob Holland

Date of birth 24/10/1992

Student number 2076888

E-mail b.holland@student.rug.nl

E-mail (alternative) bob_holland@live.nl

Address Spirealaan 64, Groningen

Postal code 9741 PD

Phone +31 (0)6 549 120 56

Study program Accountancy & Controlling

Supervisor Prof. dr. I.J.J. Burgers

(3)

Abstract

Governments have been struggling with the collection of taxes in their jurisdictions for decades. This

problem arises partly because some firms tend to be secretive about where they pay their taxes. In

order to ensure that taxes are paid in the jurisdiction where the operations that generate them take

place, governments are increasingly interested in tax transparency. In this thesis, tax transparency is

defined as the way an organization communicates its approach to tax and how it discloses the amount

of taxes paid. Shareholders are also interested in this information, as it may aid them in making

efficient investment decisions. This thesis presents several theories underlying tax transparency. In

addition to four proposed determinants of tax transparency (use of Big 4 auditor, firm size, leverage,

and effective tax rate), differences across industries are investigated. The thesis uses a method of

manual assessment of annual reports in order to determine tax transparency scores. For the sample of

140 large European listed firms, multiple regression analysis shows a positive and significant

relationship between leverage and tax transparency. No statements can be made about the effects of

the use of a Big 4 auditor, and no significant evidence is found for firm size or effective tax rate.

Although there is also no significant evidence for the existence of industry differences, this may be the

result of small industry subpopulations in the sample. Further research is necessary in order to confirm

the current conjecture of the existence of industry differences. This information may be beneficial for

several parties including policymakers. The results of this research can be incorporated in the

development of instruments aimed at enhancing tax transparency. It may be necessary to develop

(4)

Foreword

About a year ago, I handed in my bachelor’s thesis for the Bachelor of Accountancy & Controlling. In the foreword, I wrote “a new year, a new thesis” – aiming at the other Bachelor thesis that I wrote a semester earlier. This statement seems to be applicable again, but should be slightly amended: “a new year, new theses.” The past year has been a busy one, in which I have started the journey to conclude two MSc programs (Economics of Taxation and Accountancy). It took the necessary hours and sacrifices so far, but I expect to finish both degrees before summer.

Every year I learn more, and this is especially the case whilst doing research. I have gained a lot of insights in writing a thesis, proposing hypotheses, and especially managing my time efficiently. Balancing two theses with a part-time job that includes four hours of travelling each day called for a creative use of the preciously little time available. Luckily, I had the tools and mindset available to make use of these travel hours, and I have completed a product that I am satisfied with.

I would like to thank a number of people that have made this thesis possible. Foremost my supervisor Prof. dr. Burgers, who has provided me with valuable feedback, as well as patience with my busy schedule. I also would like to thank the other students from the theme group. The sessions were really helpful, and were of benefit for everyone by providing useful tips and feedback. Furthermore I would like to thank family, friends and colleagues for support and revision activities.

(5)

ABSTRACT ... 3

FOREWORD ... 4

1. INTRODUCTION... 6

2. THEORETICAL FRAMEWORK ... 9

3. RESEARCH METHODS...22

4. RESULTS ...29

5. DISCUSSION...32

6. CONCLUSION ...36

BIBLIOGRAPHY...37

APPENDIX ...44

(6)

1. Introduction

1.1 Attention for tax transparency

Taxes have become a major topic in the public debate in the past few years. The media report more and more stories concerning large multinationals (MNEs) that pay little to no taxes on their profits. Attention has increased for the amount of taxes paid by companies, as well as where they pay them. The world of taxes has lacked transparency for years, allowing for aggressive tax strategies and company structures that reduce the taxable base tremendously. This has led to a call to stop tax avoidance and to communicate tax behavior clearly to the public.

In the past few years several initiatives were introduced aimed at increasing tax transparency. Tax transparency is a relatively new term and can be interpreted in different ways. For this research, tax transparency is viewed as the way an organization communicates its approach to tax and how much tax it pays:

“We see tax transparency as being how an organization chooses to communicate its approach to tax and how much tax it pays. It is how it seeks to provide clarity on the complex area of tax.” (EY, 2013)

Transparency in modern financial reporting is considered crucial in helping users to understand, and reach their own conclusions about business (Billings & Capie, 2009). Increasing tax transparency serves several purposes. Its prime purpose is to reduce tax avoidance (OECD, 2013a). As the Task Force on Financial Integrity and Economic Development states:

“Tax avoidance on such a large scale is facilitated by a lack of transparency in the way MNCs report and publish their accounts. Making MNC accounts more transparent would help tackle tax avoidance at very low cost.” (Murphy, 2009)

Tax transparency may help to reveal punishable illegal acts, and may also reveal the need for greater spending, better training or a stronger revenue authority (Devereux et al., 2011). For companies an incentive to be transparent in their approach to tax can be to regain public trust (P wC, 2013).

Although several advantages are expected to arise from enhanced transparency, some questions remain to its usefulness. For example, Devereux et al. (2011) say:

“There nevertheless remains a question as to whether more transparent reporting of tax payments would provide information which would aid public debate about the appropriate structure of international tax principles.” (p.9)

(7)

1.2 Country-by-country reporting initiatives

Several initiatives are aimed at increasing transparency in taxes by reporting on a country-by-country basis, i.e. country-by-country reporting (CbCR). For example, the extractive industries are guided by the Extractive Industries Transparency Initiative (EITI). This industry is also targeted by the Dodd-Franks Act and the Publish What You Pay (PWYP) initiative. Banks are specifically targeted by the Capital Requirements Directive (CRD) IV. The OECD develops an instrument applicable across industries, being the OECD CbCR instrument resulting from Base Erosion and Profit Shifting (BEPS) Action Point 13 (OECD, 2013a; OECD, 2015). In the literature, some questions are raised about the usefulness of such instruments (International Tax Review, 2012; Evers, Meier & Spengel, 2014; Grau Ruiz, 2014).

Some of the available instruments are principle based. Those instruments are not so strict in their requirements regarding disclosure. This leads to variations in which information is published and in which form. These variations in implementation of instruments and regulations result in different effects on the quality of tax transparency.

1.3 Relevance of research

Understanding which factors may be of influence on tax transparency in annual reports can help regulators and investors in several ways. First, this understanding identifies the characteristics of firms that constantly outperform others in tax transparency. This information can be used as input for the development of new regulations or instruments that are especially focused on certain types of firms; this development can include more guidance on reporting (i.e. use of templates); a mandatory approach taking the place of a voluntary approach; and industry-specific elements (such as the EITI instrument, focused on the extractive industries). Devereux et al. (2011) indicate that, in line with the objectives of EITI and PWYP, adaptations to these CbCR instruments may be necessary for multinational companies from other sectors. Also, this information can be used to identify those firms that need increased attention from regulators that aim at enhanced tax transparency. Those firms can then be targeted more efficiently for increased monitoring or penalties.

Specifically, this research tries to find evidence for variables that influence tax transparency. The central question to be discussed is, “What are determinants of tax transparency in large European listed firms?” In this research, four determinants are investigated, being Big 4 Audit, Firm size, Leverage and the Effective Tax Rate. In addition, differences across industries are considered. Appendix B captures the expected relationships in a conceptual model. This model leads to the following research questions (RQ):

(8)

8 | M a s t e r ’ s t h e s i s A c c o u n t a n c y & C o n t r o l l i n g , 2 0 1 5 RQ1: Is tax transparency influenced by a Big 4 audit?

RQ2: Is tax transparency influenced by the size of the firm? RQ3: Is tax transparency influenced by leverage?

RQ4: Is tax transparency influenced by the ETR?

RQ5: Can differences across industry sectors be identified for the four determinants of tax transparency?

1.4 Research contribution

This research focuses on a sample of 140 large listed firms in Europe. Large firms are the object of research, since their owners and directors are more likely to benefit from tax transparency, or feel relatively more pressure from the public to be transparent about taxes. Additionally, Europe is progressive in the development and implementation of transparency-enhancing instruments. Results from this research can be particularly useful as input for these instrument developments.

Answering the research questions stated above as well as the central question contributes to literature in several ways. First, it provides insights in possible determinants of tax transparency in a developed market. Further research can add to this list of possible determinants, in order to come up with a comprehensive overview of factors that are of influence on tax transparency. Second, this research provides insights in the methods and instruments to measure tax transparency. Since tax transparency is a relatively new topic in accounting research about transparency, some pioneering approaches are taken by researchers in order to adequately capture tax transparency. Third, by investigating whether (strength of) relationships differ across industry sectors, problematic sectors can be pinpointed and subsequently targeted specifically for the development of transparency enhancing instruments.

1.5 Thesis structure

Chapter 2 of this thesis discusses the theoretical framework and develops the relevant hypotheses; it addresses relevant theories, disclosure requirements and insights from previous research. Next, chapter 3 discusses the research methods to be used. It includes a description of the sample, measurement methods for the included variables and statistical analyses to be performed. Chapter 4 then lists the results. Chapter 5 discusses those results, as well as research limitations and opportunities for further research. Chapter 6 concludes this thesis.

(9)

2. Theoretical framework

2.1 Defining tax transparency

Tax transparency is a broad concept that is interpreted differently by different groups of stakeholders. The appropriate definition depends on the context, and as a result, tax transparency needs to be defined in this thesis in order to frame the research question. In general, three perspectives dominate, which can be called (1) the fiscal perspective; (2) the accountancy perspective; and (3) the controlling perspective. The first and second perspectives are focused on external tax transparency from the viewpoint of the firm. The third perspective is focused on internal tax transparency.

The first perspective on tax transparency, i.e. the fiscal perspective, is the perspective that tax authorities are especially interested in. Tax authorities are increasingly searching for ways to effectively collect taxes on profits that are generated in that jurisdiction. As a result, authorities are interested in information about the firm’s operations, number of employees, and profits per jurisdiction. Several organizations are developing tools that are aimed at providing this information, an example of which is country-by-country reporting (OECD, 2013a). In addition, exchange of information between the competent authorities is of great assistance in tackling tax avoidance (Global Forum on Transparency and Exchange of Information for Tax Purposes, 2014). In the fiscal perspective, the primary purpose of tax transparency is to ensure that reliable and relevant information reaches the tax authorities timely. This information is not necessarily published in an annual report.

The second perspective on tax transparency, i.e. the accountancy perspective, focuses on information needs of investors and shareholders. These stakeholders may want to include tax-related information in their investment decisions. The annual report is the most obvious tool for a firm to communicate this information to investors and shareholders. In this perspective, tax transparency is about the choice that the firm makes in how it communicates its approach to tax and how much tax it pays (EY, 2013). Although this communication is principally directed at shareholders and investors, the annual report can also be used by other stakeholders. The communication can also be directed at tax authorities or governments. The accountancy perspective encompasses a broad audience as a result.

The third perspective, i.e. the controlling perspective, is focused on the internal environment of the firm. This perspective is about mapping tax risks (likelihood and impact) and formulating risk responses in order to be in control of tax. It is also about creating a transparent internal environment in the area of tax, and as such it is closely related to tax risk management (e.g. PwC, 2014). A Tax Control Framework is an example of a tool that can be used in this perspective. For example, the PwC Tax Control Framework includes “a clearly defined and transparent communications strategy which communicates the approach to managing tax both internally and externally”1_.

(10)

The accountancy perspective will be the focus of this thesis. Providing insights in externally focused tax transparency is of importance for both the firm and its stakeholders. Moreover, because the accountancy perspective primarily focuses on the use of the annual report as a means of communication on tax matters, several stakeholders are addressed at the same time. Results of this thesis may be beneficial for a broad audience as a consequence (tax authorities, shareholders, investors, and creditors). The annual report is also an object that can be assessed according to quantified measures, as for example in (voluntary) disclosure research. This allows performing multiple regression analyses and may result in a model that predicts tax transparency of annual reports based on several firm characteristics.

2.2 Information asymmetry and agency theory

In the introduction, several benefits of increased tax transparency have been mentioned (reducing tax avoidance, revealing punishable illegal acts, and revealing the need for a stronger revenue authority). The underlying problem is incongruent information availability and needs between MNEs, shareholders, governments, regulators, the public and other stakeholders. This is also known as information asymmetry (Akerlof, 1970; Jensen & Meckling, 1976). Different parties have different incentives, which can result in a need to obscure information. Obscured information disrupts effective functioning of the market (Healy & Palepu, 2001). The goal of increasing transparency is often to reduce information asymmetry; in the same line, increased tax transparency is expected to reduce the specific tax related information asymmetries.

What directors do or do not disclose is often linked with their personal interests and incentives. This decision making process is part of the research area called the Agency Theory (Fama, 1980; Fama & Jensen, 1983). Research in this area investigates multiple factors that have an influence on transparency, for example the influence of director stock ownership (Rose et al., 2013; Bhagat et al., 2008), culture characteristics (Hooghiemstra et al., 2015), adoption of IFRS (Tendeloo & Vanstraelen, 2005; De La Bruslerie & Gabteni, 2014) and amount of outside directors (Armstrong et al., 2010). Traditional research considers general transparency in financial reports. Transparency, however, can be subdivided in multiple areas. An annual report can be considered transparent in the amount of bonuses paid to directors or the amount of pollution, but less transparent with regards to taxes or the firm’s carbon footprint. Results from previous research that have focused on general AR transparency may not be applicable with regards to tax transparency. It is therefore necessary that tax transparency is investigated separately, so that it can be verified which determinants are of influence on this specific topic.

(11)

2.3 Mandatory versus voluntary disclosure

Agency conflicts arise when parties, usually referred to as the principal and the agent, have different interest or unaligned goals. In a firm-setting, the directors (the agent) may be motivated to make decisions that are in their own interest, instead of the interest of the company and its investors (the principal). Reaching these self-interested goals can lead to a strategy of hiding information. Ultimately, unsolved conflicts can result in high agency costs for both the principle and the agent, so that the parties are seeking for solutions to the agency conflict problem. The tax strategy of the firm is a possible area of agency conflicts.

Voluntary disclosure is mentioned as a possible solution to resolve agency conflicts (Healy & Palepu, 2001). Voluntarily disclosing information reduces information asymmetries between the firm and its investors (Diamond & Verrecchia, 1991; Kim & Verrecchia, 1994) and helps to resolve adverse selection problems between investors (Brown et al., 2001; Hooghiemstra et al., 2015). Cheung et al. (2010) conclude that this is the result from increased transparency. Directors may want to disclose certain information voluntarily because it protects them from reputational consequences (Skinner, 1994); however, they may refrain from disclosing information that can result in a competitive disadvantage for the firm, or when costs of collecting, processing and publishing the information are too high (Verrecchia, 2001). Directors make a trade-off between providing good news to shareholders and bad news to competitors (Dontoh, 1989).

Skinner (1994) finds that bad news disclosures tend to be qualitative statements, as opposed to quantitative good news disclosures. He argues that voluntary disclosure is often motivated by the fear of legal consequences. Voluntary disclosure reduces expected legal costs because it will be more difficult for the plaintiff to argue that information is withheld, and because disclosing earlier rather than later reduces damages that the plaintiff can claim.

Cheung et al. (2010) specifically investigate the difference between mandatory and voluntary disclosure. In their statistical analyses, they determine separate Voluntary Disclosure Index and Mandatory Disclosure Index scores. They find that investors are more responsive to higher voluntary disclosure than mandatory disclosure; this result is stronger among more profitable companies. Ball et al. (2012) find that mandatory and voluntary disclosure are complements, and Einhorn (2005) develops a model for the role of a firm’s mandatory disclosure in determining their voluntary disclosure strategy. De la Bruslerie & Gabteni (2014) state that the problem of strategic disclosure processes is complex, since voluntary and mandatory information have to be combined as requested by the institutional environment of the stock market. This implicates that the level of mandatory information is not set forever and may change.

(12)

2.4 Regulatory framework

Financial reporting is being guided by different sets of rules. In the US, firms are required to disclose under guidance of US GAAP. From 2005 onwards, the EU adopted International Accounting Standards (IAS) and International Financial Reporting Standards (IFRS). In the setting of this research (Europe), especially the disclosure requirements from IAS 12 are of interest. IAS 12 Income Taxes implements the comprehensive balance sheet method of accounting for income taxes2_{. The Standard}

states that “an entity should account for the tax consequences of transactions and other events in the same way it accounts for the transactions or other events themselves”. As a result, the entity should recognize such items as current tax, deferred tax assets and liabilities (DTA and DTL) and tax amount for the period. IAS 12 requires disclosure on major components of tax expense and other items that may be of interest to other parties. Appendix C lists IAS 12 disclosure requirements.

As said, the purpose of IAS 12 is to increase transparency for income (profit) taxes. However, discontent exists as to the usefulness of IAS 12. Van den Ende et al. (2012) take note of the criticism on IAS 12; the Standard is found to be complex, outdated and it is not in line with economic reality. The accounting regulations deviate from fiscal regulations and fail to show economic reality. Due to criticisms under preparers and users of the Standard, it is possible that its requirements will be amended in the near future. In this thesis, the research is framed by IAS 12 requirements that were in effect per the latest available annual reports (2013/2014 in most cases). IAS 12, being not-ideal and susceptible of change, can therefore be a confounding factor in this research.

2.5 Determinants for increased (tax) transparency

Previous research has found all sorts of determinants for transparency. For a sample of Chinese firms, Cheung et al. (2010) found that profitable, overseas-listed, and companies with nomination, audit and compensation subcommittees tend to be more transparent. Behn et al. (2010) find that larger firm size, higher leverage, a higher contribution ratio, a better NTEE classification of Higher Education, or a high compensation expense ratio are factors that result in higher transparency in the nonprofit sector. Hope, Kang, & Wim (2013) find that foreign firms cross-listed in the US are less transparent than comparable US firms. For US-listed Asian companies, Kumar (2013) finds that larger firm size, greater ownership dispersion and lower leverage provide more voluntary disclosure of intangibles information. That firm size is associated with higher transparency is also shown by Botosan (1997), Lang & Lundholm (1993) and Oliveira et al. (2006).

Lan et al. (2013) propose eleven hypotheses based on the voluntary disclosure theory to measure voluntary disclosure in the Chinese stock market. These hypotheses are subdivided in six areas:

agency theory (leverage and assets in place); signaling theory (liquidity, ROE and auditor type);

(13)

corporate governance (ownership diffusion, proportion of non-executive directors on the board and

state ownership); litigation risk (intermediary and legal environment); proprietary theory (firm size); and cost of equity. They find that the Chinese stock market has some distinct features in connection with voluntary disclosure. It is found that firm size, leverage, assets in place, ROE, and ownership diffusion are significantly associated with voluntary disclosure. Auditor type and intermediary and legal environments have larger influences; especially the auditor type result is different from results in other samples. Another research performed in China finds that the probability of voluntary disclosure on internal control weaknesses is associated with profitability, age, business complexity, corporate governance attributes and ownership structure (Ji et al., 2015).

Evidence from a one-tier board structure country (Singapore) suggests that firms with a higher proportion of independent directors on the board are associated with higher levels of voluntary disclosure (Cheng & Courtenay, 2006). No evidence was found for the influence of board size and CEO duality. In contrast, Ho & Wong (2001) find no evidence of a relation between the level of voluntary disclosure and board independence, also for a sample with one-tier board structures (Hong Kong). In these researches, independent directors are directors that are not involved in the direct operations of the firm and have no family or business relationship with management. Ghazali & Weetman (2006) find that director ownership is significantly associated with voluntary disclosure, while government ownership, new governance initiatives and industry competitiveness are not. Chau & Gray (2002) find that the extent of outside ownership is positively associated with voluntary disclosures.

Auditor type is a frequently researched topic in accountancy literature. Often, the question is whether large auditing firms enhance audit quality. Although some regulators and small auditors have criticized large auditors in the past and claimed that auditor size does not affect audit quality, several researchers subsequently provided arguments and evidence to the contrary. DeAngelo (1981), Simunic & Stein (1987) and Francis & Wilson (1988) argue that large international accounting firms have incentives to protect their reputation by providing high-quality audits, but Francis (2004) stresses that this does not mean that Big 4 audits are always superior; individual failures can and do occur. Francis & Krishnan (1999) show that Big 4 auditors are more conservative and Lennox (1999) finds that Big 4 auditors are more accurate in the UK.

Due to their size and relatively large influence on society, Big 4 auditing firms receive increased media attention. The public debate, that includes discussions about corporate social responsibility and responsible tax behavior, ensures that the largest auditors - all of which also provide tax advisory services - are being scrutinized at all times. When things go wrong, consequences can be sizeable: it can be seen as case of ‘The bigger they hit, the harder they fall’. A famous example is the downfall of Arthur Andersen in 2002 following the Enron scandal (then part of the then Big 5 auditors). Although Enron represented only a small part of Andersen’s global revenues, the scandal was enough to

(14)

effectuate its surrendering of licenses to operate. After being cleared of criminal charges, Andersen effectively had no choice but to cease operations due to irreparable reputation damages.

The example above indicates that Big 4 auditors are indeed fallible and do not reach 100% audit quality. However, the example is also evidence for the argument that DeAngelo (1981) made: the risk of reputation loss and the subsequent risk of firm failure effectuate incentives to provide high quality audits. Not a single client is large enough to warrant taking opportunistic risk in their audit. Greenwood et al. (2005) argue that professional service firms need to generate a superior reputation to succeed. In line with this reasoning, the argument is maintained that the size of the audit firm, i.e. a Big 4-auditor in the context of this research, is related to the quality of the services provided.

Finally, a study by Hope, Ma, & Thomas (2013) investigates the level of geographic earnings disclosure for United States MNEs. They find that, after the adoption of Statement of Financial Accounting Standards No. 131 in 1998, geographic earnings disclosure in financial reports decreased especially for firms with a low effective tax rate. This result indicates that firms with low effective tax rates choose not to voluntarily report earnings on a geographic basis and are less transparent than their peers.

2.6 Hypothesis development

Previous articles have focused almost exclusively on general transparency levels in annual reports. Deriving from this research, four possible determinants of tax transparency are identified in this paper. Using these determinants for the specific case of tax transparency enhances contemporary literature because these previous results cannot be indiscriminately generalized to a tax setting. New evidence can be provided that either extends previous results to the area of taxes, or marks determinants as having distinct effects for tax transparency. For this thesis, four determinants are selected that are assessed to be most likely to influence tax transparency levels. Other variables, such as those related to board characteristics, corporate governance, or litigation risk, are not included, although they are possibly interesting starting points for further research. Instead, this thesis is framed by the determinants referred to as Big 4 Audit, Firm size, Leverage, and Effective Tax Rate.

Big 4 Audit

First, it is expected that tax transparency is influenced by the type of auditor hired by the firm, similar to the results of Lan et al. (2013). Specifically, it is expected that when a firm is audited by a Big 4 auditor, tax transparency levels will be higher. As has been stressed in paragraph 2.5, Big 4 auditors are widely considered to be an effective proxy for higher quality audits, since large accounting firms have more incentives to protect their reputation, and because they do not need to opportunistically reduce audit quality in order to retain a single client. In addition, Big 4 auditors make use of standardized audit technology (Jeong & Rho). In accountancy research, Big 4 auditors are by far the

(15)

most used proxy for audit quality (e.g. Kanagaretnam et al., 2009; Chi & Weng, 2014; Lai, 2013). Second-tier audit firms can also be expected to consider their reputation in performing audits, but they differ greatly from Big 4 auditors in terms of income and market capitalization3_{. As such, researches}

argue that the theorized proxy for audit quality is strongest for Big 4 auditors. For this reason, the proxy is also used in this research. It should be noted, however, that there is evidence that suggests that there is little difference in actual audit quality between Big 4 and second-tier firms (e.g. BDO and Grant Thornton), but a more pronounced difference in perceived audit quality (Boone et al., 2010).

In researches by Armstrong et al. (2014) and Dhaliwal et al. (2014), Big 4 Audit was a control variable but showed no significant influence. Results from Francis & Krishnan (1999) and Lennox (1999), conducted in the US and the UK respectively, indicate that Big 4 auditors are more conservative and accurate. The results from these researches are likely contingent on the specific country situation, which makes generalization to all of Europe difficult. Countries can be different on several dimensions, for example political, cultural, differences in law systems, and differences in reporting standards and disclosure requirements. Investigation of specific country elements falls outside the scope of this research. Instead, the European market is considered as a whole. This is a limitation to this research, as results may only be significant for some countries within the sample. This limitation provides opportunity for further research.

The arguments above result in the expectation that an overall positive relationship exists for the European market, ergo hypothesis H1:

H1: Ceteris paribus, firms that were audited by a Big 4 auditor will show higher tax

transparency.

Firm size

Second, firm size is expected to be positively related to tax transparency. The relationship between firm size and disclosure has previously been investigated from a transaction cost perspective. King et al. (1990) hypothesize that the level of disclosure increases with firm size, as the incentives for private information acquisition are greater for large firms. Hence, a positive relationship is expected between firm size and tax transparency. Firms can have multiple reasons to be transparent about their taxes. The directors can be intrinsically motivated to be transparent, because they are aware of their social roles and responsibilities. In contrast, they can also see transparency as a necessary boundary condition to operate. Concealing information can be perceived as bad behavior by stakeholders and can have reputational consequences.

In addition to the transaction cost theory, the positive relationship can also be explained from a legal cost perspective. Skinner (1994) suggests that the value of securities litigation damages is positively

(16)

related to firm size. As a result, larger firms are more motivated to be transparent. In the case of taxes , large firms may be sensitive to the threat of imposed fines and legal procedures. It must be noted, however, that Skinner theorizes from a USA perspective, where litigation levels are generally higher than in other markets (Lam & Mensah, 2006) and litigation costs are 2,6 times the average level of litigation costs in the European market (Insitute for Legal Reform, 2013). As a result, the relationship may be less strong for the sample of European listed firms used in this thesis.

Several researches, conducted in diverse settings, provide evidence for the expectation that larger firms are motivated to be transparent (Behn et al., 2010; Kumar, 2013; Botosan, 1997; Lang & Lundholm, 1993; Lan et al., 2013; Ettredge et al., 2011; and Oliveira et al., 2006). Expecting that this relationship is not different for large listed European firms in the specific case of tax transparency, this yields hypothesis H2:

H2: Ceteris paribus, larger firms will show higher tax transparency. Leverage

Third, the way a firm is financed can influence various decisions made by directors. Generally, a firm makes a trade-off between debt and equity financing. Debt can be attractive from a tax perspective, since interest payments are deductible from taxable profit in most cases. A firm is called highly leveraged when debt financing is relatively high as compared to equity financing. A disadvantage of high leverage can be the higher risk of insolvability. Banks and other finance institutes therefore demand that certain criteria are met before additional debt is provided. As a consequence, creditors also become increasingly focused on the financial health of an organization when debt levels are high. They may demand more monitoring on the financial health of the firm, in order to ensure repayment of provided loans.

Jensen & Meckling (1976) argue that leverage creates a demand for monitoring. In order to prevent costs associated with the risk of defaulting on debts, shareholders want to be ensured that opportunistic behavior is held in check. Behn et al. (2010) extend the argument by theorizing that firms with debt must also bend to monitoring demands by creditors of the loans. In an empirical study on the relationship between tax avoidance and the cost of bank loans, Hasan et al. (2014) find evidence that firms with greater tax avoidance incur higher spreads when obtaining bank loans. Whilst the potential benefits of tax avoidance accrue to shareholders, creditors do not necessarily benefit from this behavior and face a downside risk. In order to minimalize risk of defaulting at the debtor’s side, creditors demand more insight in a company’s tax matters. As a result, leverage can be expected to be positively related to tax transparency, because firms try to avoid extra financing costs through higher spreads by providing sufficient insight in their tax behavior. Tax transparency in the annual report then serves as a risk-reducing strategy that benefits both the debtor and the creditor. Although tax behavior can be communicated to creditors in other ways, the annual report is an audited instrument that grants

(17)

assurance to multiple parties at the same time. When debt needs to be attracted, (audited) accounting information is used in the contracting process (Zimmer, 1986; Christie and Zimmerman, 1994). As a result, the annual report is a preferable form of communication for MNEs from an efficiency perspective.

On the other hand, a negative relationship between leverage and tax transparency is also possible. Tarca (2004) argues that leverage leads to less dependence on equity capital; therefore, these firms are less likely to be subject to shareholders’ demand for information. This reasoning suggests that firms with higher leverage are subjected to a lower demand for information and have fewer incentives to be transparent about taxes. In some researches, no significant relationship was found between leverage and disclosure levels although the underlying theories suggested otherwise (Aly et al., 2010; Bauwhede & Willekens, 2008; Lopes & Rodrigues, 2007; Reverte, 2009; Wang et al., 2008). Although Behn et al. (2010) and Lan et al. (2013) find that leverage is positively associated with transparency, Kumar (2013) finds conflicting results. These conflicting arguments yield hypothesis H3:

H3: Ceteris paribus, leverage in firms has an influence on the level of tax transparency. Effective Tax Rate

Fourth, a tax-specific variable is considered. Gallemore & Labro (2014) have connected the firm’s effective tax rate (ETR) with internal information quality in firms. The ETR is the amount of tax expense borne on pre-tax income. Effective tax rates are found to be high in general, about 22,4% world-wide (OECD, 2013b). Gallemore & Labro (2014) find that high internal information quality lowers ETR, especially for high coordination needs firms and high uncertainty firms. They associate a lower ETR with tax avoidance, and also argue that firms with high external transparency “might be more comfortable adopting complex tax avoidance strategies”. This last statement seems to indicate that a negative relationship is expected between transparency and ETR, i.e. high transparency mea ns more tax avoidance and hence a lower ETR.

On the other hand, Balakrishnan et al. (2012) find evidence that suggests that aggressive tax planning decreases corporate transparency; it is suggested that firms face a trade-off between financial transparency and aggressive tax planning. This suggests that firms with high tax aggressiveness are less transparent. Resulting from this, a positive relationship can be expected between ETR and (tax) transparency. It should be noted that although (adjusted) ETR is a generally used proxy for tax avoidance or tax aggressiveness, it is not guaranteed that tax avoidance or tax aggressiveness are actually captured by the measure. Lacking sophisticated tools to measure this tax behavior, ETR is often chosen because the calculations are simple and data is more easily obtainable. Unfortunately, ETR measures have difficulties with eliminating tax facilities that cannot be characterized as

(18)

aggressive but that do lower ETR at the same time. ETR is therefore a weak measure, so that results may not be robust.

Summarizing, the relationship between ETR and tax transparency is ambiguous, resulting in hypothesis H4:

H4: Ceteris paribus, the ETR of a firm has an influence on the level of tax transparency. Industry differences

Lastly, it is expected that differences exist across industries. Some previous research in disclosure quality found mixed results. Barako et al. (2006) found that firm characteristics have an influence on the extent of voluntary disclosure in Kenya. In contrast, Beretta & Bozzolan (2004) and Van Oorschot (2010) find that the amount of risk disclosure is not significantly influenced by industry. No predictions are made about which specific industries can be expected to outperform others in terms of transparency.

Although these last cited articles investigated risk disclosure, it can be assumed that the underlying processes can also be extended to tax transparency. It is worthwhile to take possible industry differences into account, especially since some industries are targeted by specific initiatives (EITI for extractive industries and CRD for banks, for example). This thesis does not predict that these industries will strictly outperform other industries, but instead takes a neutral stance (hypothesis H5):

H5: The effects of the four identified determinants of tax transparency differ across industries.

2.7 Measuring Tax transparency

The main object and dependent variable in this research is tax transparency. Tax transparency encompasses the way in which an organization chooses to communicate its approach to tax and how much tax it pays. Measuring tax transparency can be difficult, because definitions differ and several sources of information can be taken into account. The annual report and its included financial statements are a prime instrument of communication between a firm and its shareholders, but large companies often also maintain a website and publish statements about specific topics.

In a research on transparency of corporate governance practices, Cheung et al. (2010) use a Transparency Index that is based on five OECD principles of corporate governance (OECD, 2004); these five principles are the rights of shareholders, equitable treatment of shareholders, the role of stakeholders, disclosure and transparency, and board responsibilities and composition. Cheung et al. (2010) only incorporate information from annual reports, resulting in a Voluntary Disclosure Index score and a Mandatory Disclosure Index score. Hodgdon et al. (2008) incorporate IAS and IFRS to determine a disclosure index, used to measure compliance with IFRS.

(19)

In this research, tax transparency will also be measured with a disclosure index score. In addition to information from the AR, certain selected published information is also included into the assessment, if it meets the criteria of an instrument specifically developed to increase tax transparency. This includes for example a country-by-country for financial institutions in order to comply with CRD IV. Because some firms include them in their AR and others publish the document separately, only the inclusion of the separately published document ensures comparability of tax transparency scores.

The AR is considered the most reliable source of information, because the document is audited. Accountants and auditors have assessed the report and provided their audit opinion. In contrast, information from a corporate website is more problematic to use, since its form of presentation is not guided by universally applicable guidance, and as a result it is not as trustworthy as audited statements. Marston & Shrives (1991) implicate that the AR is the major source of information to various user-groups, although other information sources are available (medial release, interim reporting, letters to shareholders, and employee reports).

Factors of tax transparency

A number of factors will constitute the variable of tax transparency. These factors are the same ones as used in tax transparency research by Deloitte4_{. In this research, tax transparency was compared}

between FTSE 100 companies from the UK and 162 listed companies from seven other European countries (The Netherlands (AEX Amsterdam), Germany (DAX Frankfurt), Belgium (BEL Brussels), Denmark (OMX Copenhagen), Sweden (OMX Stockholm), Norway (OBX Oslo) and Austria (ATX Vienna)).

For all entities within the sample, at least seven factors are identified. These factors are (1) Use of havens; (2) Tax governance; (3) Tax policy / strategy; (4) Taxes paid / “contribution”; (5) Country-by-country reporting; (6) Uncertain tax positions and (7) Narrative on ETR drivers.

Use of havens refers to the use of tax havens. Tax havens are considered to be countries where the STR is close to 0% or where effectively little taxes are paid on income. Tax governance refers to tax related governance, including the appointment of a committee dedicated to the subject or the Audit committee discussing tax matters. Tax policy / strategy refers to the communication of the firm’s strategy or policy on tax matters. Taxes paid / “contribution” refers to the disclosure of the amount of taxes paid. Some firms call this “tax contribution”. Country-by-country reporting refers to a disclosure of taxes paid per country or jurisdiction; in other words, a geographical split of tax payments. Uncertain tax positions include comments aimed at clarifying uncertainties in tax positions. Narrative on ETR drivers refers to disclosures on the difference between ETR and STR.

4_{Due to my employment at Deloitte Belastingadviseurs BV I have access to source material and (prelimin ary)}

results of conducted transparency research that may or may not have been published yet. For confidentiality reasons, this source material cannot be disclosed by my discretion.

(20)

20 | M a s t e r ’ s t h e s i s A c c o u n t a n c y & C o n t r o l l i n g , 2 0 1 5 Extra factor for Dutch companies

A recent trend in tax governance is the establishment of collaborative and trust based relationships between tax authorities and large corporate taxpayers and tax advisors, which is referred to as co-operative compliance (OECD, 2013c). Early adopters of co-co-operative compliance models are Ireland, the USA and the Netherlands, each starting in 2005 with pilot projects and continuously developing the models and integrating them in compliance risk management strategies. These voluntary relationships between MNEs and tax authorities are part of the firm’s tax governance. Discussing these relationships openly in the AR contributes to the tax transparency of the AR.

In this sample, co-operative compliance is added for Dutch companies as an eighth factor. The Netherlands is the sole representative of the three leading countries in co-operative compliance in the sample of 140 firms in this research (for a total of seven firms). In this case, a Dutch company refers to a company that is incorporated and listed in the Netherlands. Dutch companies within the medium to very large businesses segment have the possibility to engage in an agreement with the tax authorities, known as ‘Horizontal Monitoring’ (Horizontaal Toezicht) (see text box).

Because the introduction of Horizontal Monitoring has had a positive influence on the tax business climate in the Netherlands (The Netherlands Tax and Customs Administration, 2010), it is an important component of tax transparency in the Netherlands. Especially firms that engage in an agreement with the tax authorities are expected to mention this in their AR when full transparency is pursued. Firms that do not participate in Horizontal Monitoring may still want to motivate this decision towards shareholders.

Other countries also have possibilities to engage in co-operative compliance arrangements, but MNEs that are listed there have had a shorter timeframe to gain experience in this area than early adopters such as Ireland, the USA and the Netherlands. Therefore, Dutch companies are chosen in this thesis to explore the possible influence of including this factor in the Tax Transparency Score. Although the extra factor is unlikely to have a material impact on transparency scores, its inclusion may yield useful insights in the possible extension of the measurement tool. It is theorized that mentioning the approach

Dutch State Secretary on Horizontal Monitoring

Horizontal monitoring refers to mutual trust between the taxpayer and the Tax and Customs Administration, as well as to the more precise specification of each other’s responsibilities and the options available to enforce the law and comply with the mutual agreements. As a result, the underlying relationships and communications between citizens and the government shift to a more equal situation.

(21)

towards Horizontal Monitoring in the AR enhances tax transparency. Since the addition of an eighth factor for a part of the sample can result in disparities between subsets, statistical tests will be performed both with inclusion and exclusion of the extra factor.

Composing a score for tax transparency by consulting annual reports allows performing quantitative analyses. However, this method has its limitations, primarily because assessment is based on the researcher’s selection and judgment. Cheung et al. (2010) discuss the use of disclosure indices compiled by analysts, such as the AIMR database. The AIMR database is based on an annual survey that produces firm rankings of voluntary disclosure. Cheung et al. (2010) point out a possible limitation:

“(…) it provides a more general measure of voluntary disclosure. However, the AIMR

rankings rely heavily on analysts’ selection and judgment, which may bring bias to the data. Critiques of self-constructed measures argue that these metrics typically rely on annual reports and may ignore other sources of information such as press briefings, analyst meetings, etc.

In this research, part of this limitation is reduced by checking for consistency after all observations are assessed on tax transparency. Due to progressive insight, judgment may evolve over time. The procedure for assessing the identified factors in order to come up with a score on tax transparency, and the procedure for ensuring consistency will be discussed in chapter 3.

(22)

3. Research methods

3.1 Sample

In this research, the sample consists of 140 large firms that are listed in Europe (see appendix D). This sample is the same that was used in my Bachelor’s thesis5_{. The list of firms in the sample is composed}

by ranking European MNEs in terms of market capitalization, and selecting the top 140 entities.

The European market is especially suited for this research, since several reporting initiatives are effective in Europe. These initiatives include some adoption of EITI, the CbCR instrument by the OECD and, for some European firms, a country-by-country initiative from the European Union. Some of these instruments are not mandatory, or exist only in a conceptual form. Due to this, a firm that voluntarily discloses certain information can be considered to be more transparent than a firm that does not. This sample includes firms from different countries and from different industrie s. This allows testing for industry differences (hypothesis H5).

Listed firms have to publish financial information at least annually by law. Therefore, enough information should be available for these firms. Large firms from different industries are used to allow comparison between the firms. In this research, the latest available financial information is used to assess the quality of tax transparency. This ensures that the same disclosure requirements and initiatives apply to all firms and that comparison is possible.

Using information from one single year can be limiting to the research. However, it falls outside the scope to perform a multi-year analysis. Future research could compare different time periods. For example, when CbCR from BEPS Action Point 13 (transfer pricing documentation and country-by-country reporting) becomes mandatory, it would be interesting to see its effects on the quality of tax transparency. See also paragraph 5.2 on Research limitations & future research.

3.2 Measurement

Tax transparency

As has been discussed in paragraph 2.7, seven factors (eight in the case of Dutch firms) constitute a score on tax transparency. These factors, Use of havens, Tax governance, Tax policy / strategy, Taxes paid / “contribution”, Country-by-country reporting, Uncertain tax positions, Narrative on ETR drivers and Horizontal Monitoring will be assessed manually by the researcher for all 140 firms. This is done by consulting the annual report and determining whether the annual report (AR) mentions the factor. If the factor is not mentioned, ‘NO’ is recorded; if the factor is mentioned but only with very general

5_{Holland, B. (2014). Disclosure quality of goodwill impairments: The influence of firm performance, relative}

size of goodwill and experience with impairments. Bachelor’s thesis Accountancy & Controlling, University of Groningen, the Netherlands

(23)

descriptions, ‘Partial’ is recorded; and if the factor is mentioned with enough details, ‘YES’ is recorded. Results will be tracked in an Excel-file.

Assessment of these factors is done by the subjectivity of the researcher. In order to be as consistent as possible, 25 entities from the sample will be randomly selected and reassessed. Due to progressive insight, assessments and ratings may develop over time. If this results in large or material restatements of tax transparency scores, the whole sample will need to be reassessed. If not, the assessment is expected to be performed consistently.

After this procedure, a final score on tax transparency can be computed. The most common practice is that all items are weighted equally (Glaum et al., 2013) and this practice will be followed in this research. Because the measurement method is not a widely applied one, it is difficult to determine which factors are relatively more important and deserve different weighting. In any case, it is assumed that slightly different weightings would not result in materially different tax transparency scores. Factors that are answered with ‘NO’ receive a score of ‘0’; factors answered with ‘Partial’ receive a score of ‘0.5’ and factors answered with ‘YES’ receive a score of ‘1’.

(1) 𝑇𝑎𝑥 𝑇𝑟𝑎𝑛𝑠𝑝𝑎𝑟𝑒𝑛𝑐𝑦 𝑆𝑐𝑜𝑟𝑒 = 𝐶ℎ𝑒𝑐𝑘 𝑙𝑖𝑠𝑡 𝑠𝑐𝑜𝑟𝑒 𝐴𝑚𝑜𝑢𝑛𝑡 𝑜𝑓 𝑑𝑖𝑠𝑐𝑙𝑜𝑠𝑢𝑟𝑒 𝑖𝑡𝑒𝑚𝑠

For the first test, this means that for all firms in the sample the Tax Transparency Score will be computed according to formula (2):

(2) 𝑇𝑅𝐴𝑁𝑆1 =𝐶ℎ𝑒𝑐𝑘 𝑙𝑖𝑠𝑡 𝑠𝑐𝑜𝑟𝑒 7

For the second test, the additional factor ‘Horizontal Monitoring’ will be included for Dutch firms. This means that for non-Dutch firms, the Tax Transparency Score will be computed according to formula (2), and for Dutch-firms according to formula (3):

(3) 𝐹𝑜𝑟 𝐷𝑢𝑡𝑐ℎ 𝑓𝑖𝑟𝑚𝑠: 𝑇𝑅𝐴𝑁𝑆2 =𝐶ℎ𝑒𝑐𝑘 𝑙𝑖𝑠𝑡 𝑠𝑐𝑜𝑟𝑒 8

Big 4 Audit

The variable Big 4 Audit will be measured as “1” for firms that have been audited by PwC, EY, KPMG or Deloitte, and “0” for firms that have been audited by any other auditor (Chi & Weng, 2014). In France, MNEs are required to be subjected to a joint audit6_{. During a joint audit, two or more}

statutory auditors perform the audit. When at least one of the participating auditors is a Big 4 auditor, the measurement will be “1”.

6_{Article L823-2 French Commercial Code: “All legal entities which are obliged to p ublish consolidated}

(24)

24 | M a s t e r ’ s t h e s i s A c c o u n t a n c y & C o n t r o l l i n g , 2 0 1 5 Firm Size

Firm size can be measured on several items that vary with firm size. In many studies, firm size emerges as a robust variable. Some studies imply that it is unnecessary to pay much attention to the choice between alternative measures of firm size, but Shalit & Sankar (1977) state that there is also evidence that different conclusions can be reached depending on the particular size measure used.

According to Shalit & Sankar (1977), the choice for measures of firm size can depend on a number of factors, including practical considerations of data availability. Shalit & Sankar identify the five most commonly used measures as being annual sales, total assets, total number of employees, stockholders’ equity, and market value of the firm at year-end. Results based on these five measures for firm size are generally found to be robust (Ben-Zion & Shalit, 1975; Shalit & Ben-Zion, 1974), but Shalit & Sankar find that employment is a relatively poor substitute for the other four measures. Research conclusions can be different if employment is used as a proxy for firm size, although strong correlation between the measures indicates that the likelihood is small. Firm assets and stockholders’ equity have the lowest inequality coefficient, indicating that these two measures are interchangeable.

Because the dependent variable in this research (tax transparency) is not directly associated with any of these five measures, three of the five most commonly used measures are chosen. Using three different methods allows for comparing these variables’ relative impacts. Firm size will be measured as being the natural log of respectively total assets (equation 4a), sales (equation 4b), and number of employees (equation 4c); see also Lan et al. (2013) and Hope et al. (2012). Based on the findings of Shalit & Sankar (1977), the assets measure is interchangeable with stockholders’ equity. Stockholders’ equity is dropped in this thesis out of data availability considerations. Market valuation is dropped for the same reason, resulting in three formulas for firm size:

(4𝑎)𝑆𝐼𝑍𝐸1 = ln(𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑦𝑒𝑎𝑟−𝑒𝑛𝑑)

(4𝑏)𝑆𝐼𝑍𝐸2 = ln(𝑆𝑎𝑙𝑒𝑠𝑦𝑒𝑎𝑟−𝑒𝑛𝑑)

(25)

25 | M a s t e r ’ s t h e s i s A c c o u n t a n c y & C o n t r o l l i n g , 2 0 1 5 Leverage

Financial leverage can be measured in several ways. Three common approaches can be distinguished, where one measure focuses on the balance sheet, another measure on the income statement and yet another on a comparison of rates of return (Ghandhi, 1966). Ghandhi further examines whether the choice of measure really matters. It is found that the different measures yield highly correlated results in stable industries, less correlated results in relatively unstable industries, and moderately correlated results overall. Ghandhi concludes that “in practice in circumstances other than the most extreme it would appear to be a matter of relative indifference which measure were [to be] adopted”. Ghandhi warns that the use of a proxy has no constant or assured reliability, so that it is worthwhile to select a suitable proxy in most circumstances.

For this thesis, the most important analysis is performed at the total sample level. The choice between measures for leverage is therefore made out of practical considerations. The balance sheet approach is selected because the necessary data is easily obtainable from the ORBIS database. The balance sheet approach identifies two standard measures, being the ratio between debt and equity, or the ratio between debt and total assets. As Ghandhi (1966) points out, the choice between these two is a matter of complete difference, since “any ranking of firms by one of the measures must yield the same ordering as that derived from the other measure”.

In this research, the choice is made for the ratio between total debt and total assets at year-end. The values of total debt and total assets are based on the balance sheet valuation instead of market valuation. Market valuations can result in displacement when securities are valued differently for firms with a similar capital structure. Therefore, the most widely applied measure of leverage by economists is used:

(5) 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 = 𝑇𝑜𝑡𝑎𝑙 𝑑𝑒𝑏𝑡𝑦𝑒𝑎𝑟−𝑒𝑛𝑑 𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑦𝑒𝑎𝑟−𝑒𝑛𝑑

It must be noted, however, that the indiscriminate use of this measure for all industries is not perfect for the industry analysis. Ghandhi (1966) finds evidence that there is an increased chance of displacement. Although he finds generally high coefficients between measures and concludes that the choice of measure will not lead to severely conflicting results, the choice of measure is ideally contingent on industry characteristics. Because the likelihood of conflicting results is small, this thesis opts for using one measure only.

(26)

26 | M a s t e r ’ s t h e s i s A c c o u n t a n c y & C o n t r o l l i n g , 2 0 1 5 Effective Tax Rate

The Effective Tax Rate (ETR) is the amount of tax expense borne on pre-tax income and is equal to the ratio between tax expense and pre-tax income. This method of measurement is rather simplistic and variations could be used (Dyreng et al., 2008), but these variations are more sensible when timing issues of certain specific tax items are considered. Some researchers adapt ETR variables in order to use it as a proxy for aggressive tax behavior. Using only the ratio between tax expense and pre -tax income would not be an adequate proxy for tax aggressiveness, because a low tax rate does not necessarily vary with tax aggressiveness. Instead, low tax rates can result from statutory tax credits and tax reliefs, such as investment credits, loss carry forwards and the participation exemption. Suitable proxies for tax aggressiveness should remove these elements because although the use of these facilities lowers the firm’s effective tax rate, it does not fall under the concept of aggressive tax behavior.

In this research, capturing tax aggressiveness is not a goal. This pleads for using the simplest measure of ETR which is assumed to be the same as the effective tax rate disclosed by the firm. Thus, ETR is measured according to formula (6) with information from the latest fiscal year:

(6) 𝐸𝑇𝑅 = 𝑇𝑎𝑥 𝑒𝑥𝑝𝑒𝑛𝑠𝑒 𝑃𝑟𝑒 − 𝑡𝑎𝑥 𝐼𝑛𝑐𝑜𝑚𝑒

3.3 Data collection

For the independent and control variables, online databases are used. ORBIS is used to collect data on Auditor, Total assets, Total debt, Tax expense and Pre-tax income. Where data is unavailable, other databases are used, such as Compustat and Asset4. Some data that is still missing will then be collected manually whenever appropriate. Special care is taken for currency exchange effects, and the method of calculating certain variables. For example, year-end data is converted to Euro at the closing-date exchange rate. In the end, no firm needed to be dropped from the sample. Information on the industry (Standard Industry Classification (SIC) codes) is only retrievable as 3-digit codes; these codes are manually converted to 2-digit codes.

The tax transparency variable is determined manually. At first, the annual reports of the 140 MNEs from the sample are collected. The AR used for the assessment of tax transparency will be matched to the AR that is the source for the other collected data (latest available year). Subsequently, an Excel model will be built that facilitates the tracking of the scores for the tax transparency components. A formula calculates two different scores for each firm: a tax transparency score that includes only the seven general components (TRANS1), and a tax transparency score that also takes into account the additional factor for Dutch companies (TRANS2).

The assessment of the annual reports is done by searching for relevant search terms for each component. These search terms are assumed to be the most reliable in providing the right place to find