Audit fees under rules based and principle based accounting standards (U.S. GAAP as rules based and IFRS as principle based)

Amsterdam Business School

Audit fees under rules based and principle based accounting

standards (U.S. GAAP as rules based and IFRS as principle based)

Name: Onur Ata

Student number: 10428828 Date: 17 June, 2016

Word count: 12.710

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam Supervisor: Dr. A. Sikalidis

2 Statement of Originality

This document is written by student Onur Ata who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

3 Abstract: This study contributes to the ongoing debate whether rules based or principle based standards are more beneficial. In the current environment the international financial reporting standards (IFRS) are seen as principle based while the generally accepted accounting principles (GAAP) used in the United States are seen as rules based standards. Research about audit fees under the two competing accounting standards have shown that audit fees are higher under IFRS due to an increase in complexity and audit effort, whereas research about fees under U.S. GAAP shows that fees are higher due to an higher risk of litigation. This paper compares audit fees in the period 2011 to 2014 for 170 companies of the Fortune Global 500 that report under IFRS and U.S. GAAP, and finds evidence that the fees are higher for firms under U.S. GAAP. Further, the study investigates the effect of the level of investor protection and legal tradition of a country on audit fees, which helps to understand what influence a country’s institutional characteristics have. The results suggests that audit fees increase with the level of investor protection and are higher in civil law countries compared to common law countries.

4

Content

1 Introduction ... 6

2 Literature review ... 8

2.1 Audit fees and its determinants ... 8

2.1.1 Determinants of audit fees ... 8

2.1.2 Size of the auditee ... 10

2.1.3 Complexity of auditee ... 10

2.1.4 Risk ... 10

2.1.5 Audit quality ... 11

2.1.6 Litigation propensity... 11

2.2 Principle and rules based accounting standards ... 12

2.2.1 U.S. GAAP and IFRS ... 12

2.2.2 Rules and principle based standards ... 13

2.2.3 Audit fees under rules based and principle based standards ... 15

2.3 Investor protection and legal tradition ... 17

2.3.1 Investor protection ... 18

2.3.2 Civil and common law ... 18

2.4 Hypotheses ... 20 3 Research methodology ... 21 3.1 Data... 21 3.2 Empirical models ... 22 4 Results ... 24 4.1 Descriptive statistics ... 24 4.2 Correlation coefficients ... 26 4.3 Multicollinearity ... 27 4.4 Regression analysis ... 29 4.5 Sensitivity analysis ... 32

5 5 Conclusion and discussion ... 35 References ... 37

6

1 Introduction

The two most important accounting standards at this day are the International financial reporting standards (IFRS) that are issued by the International Accounting Standards Board (IASB), and the generally accepted accounting principles (GAAP) issued by the Financial Accounting Standards Board (FASB).

In the current environment IFRS is described as a principle based standard while U.S. GAAP is seen as rules based (Agoglia, Doupnik, & Tsakumis, 2011). Rules based standards provide detailed guidance with “bright-lines” and a perceived benefit of it is greater comparability of financial statements (Schipper, 2003). But it also has its down sides, for example a study of Segovia, Arnold, & Sutton (2009) suggests that auditors are more willing to allow earnings management under rules based standards. On the other hand principle based standards, which provide less guidance and more room for professional judgement by auditors, are seen as the solution to the problems that are caused by rules bases standards (Agoglia et al. 2011).

In the ongoing debate about which standard is more beneficial a lot of research has been conducted. There are studies that research the impact of rules and principle based standards on a range of different outcomes. For example a study of Jamal and Tan (2010) looks at the financial reporting quality under principle and rules based standards and finds that the financial reporting quality improves with principle based standards. A study by Sin, Moroney, & Strydom (2015) looks at material control weaknesses of companies under a principle and rules based setting. Another study of Agoglia et al. (2011) measures aggressive reporting under both approaches and suggests that CFOs are less likely to report aggressively under principle based standards than under rules based standards. Some studies make a more direct comparison between IFRS and U.S. GAAP and look at comparability, financial reporting outcomes, and earnings management under the two standards and show different results (Barth, Landsman, Lang, & Williams, 2012; Gordon, Jorgensen, & Linthicium, 2010).

From prior literature it is also obvious that a lot is investigated about audit fees, there are studies that look at the effect of auditor size on audit fees, the effect of auditor change on audit fees and the effect of board characteristics on audit fees (Palmrose, 1986; Simon & Francis, 1988; Carcello, Hermanson, Neal, & Riley, 2002). Some newer studies focus on the effects of accounting standards on auditor fees. For example a study of Seetharaman, Gul and Lynn (2002) investigates UK-firms that are cross-listed on US markets. They find that audit fees are higher under rules based standards, due to an increase in litigation risk. Choi, Kim, Lui and Simunic (2009) find that auditors charge higher fees for firms that are cross-listed in countries with stricter regimes and standards.

7 While these two studies show a significant increase in audit fees due to an increase in litigation risk under rules based standards, other studies find that audit fees increase under principle based standards because these standards require more professional judgement and complex estimates of auditors. For example a study of George, Ferguson and Spear (2013) find that audit fees in Australia increase with 8 percent after adopting IFRS due to an increase in complexity. Another study of Kim, Liu and Zhen (2012) shows that audit fees increase for European countries after switching from their local GAAP to IFRS. But the focus of these studies is mainly on the differences in audit fees of cross-listed firms or at the differences after the adoption of a certain standard.

So, it becomes clear that studies about audit fees under both accounting standards show contradictory results, and to the best of my knowledge there hasn’t been any direct study on the differences in audit fees under principle and rules based accounting standards, and thereby comparing audit fees under IFRS and U.S. GAAP. The goal of this study is to fill this gap in knowledge and thereby answering the following research question: What is the impact of principle and

rules based accounting standards on audit fees? (IFRS as principle based accounting standard and US GAAP as rules based).

Studies from Choi et al. (2009), Taylor and Simon (1999) and Jaggi and Low (2011) show that institutional factors also have a significant influence on audit fees. Therefore, besides looking at the effect of accounting regulations on audit fees this study also investigates the impact of institutional factors on audit fees, which helps to understand what influence these country specific factors have.

The research is based on data that range from 2011 till 2014 and compares 170 firms that are members of the Fortune Global 500 which report under IFRS and U.S. GAAP. The results show that audit fees are higher for firms that report under U.S. GAAP and thus under rules based standards, which can be explained by an increase in litigation risk. In addition the findings show that audit fees increase with higher levels of investor protection of a country, which is caused by the fact that higher levels of investor protection is associated with an increase in litigation risk. Finally, it shows that audit fees are in general lower for common law countries, compared to civil law countries. Suggesting that the stronger law enforcement in common law countries does not have a positive effect on audit fees.

The remainder of this paper will be structured as follows, section 2 presents the literature review, section 3 the research methodology, section 4 the results and finally section 5 will show the conclusion of the study.

8

2 Literature review

In this section existing literature about determinants of audit fees, principle and rules based standards, investor protection and legal tradition will be discussed. First, the most important literature about audit fees will be discussed and what the most important determinants of these fees are according to prior literature. Second, an explanation will be given about principle and rules based standards, followed by the existing research about audit fees under both standards. Finally, an explanation will be given about the relation between audit fees, investor protection and the legal tradition of a country.

2.1 Audit fees and its determinants

Simunic (1980) describes the audit fee as the product of unit price and the quantity of audit services demanded by the management of the auditee. Thereby he states that audit fees reflect the economic costs of an audit that can vary with several factors (Simunic, 1980). Because there are several factors that influences the pricing of an audit, it is useful to understand how these variables affect the audit fees. So, in the section below the most important determinants of audit fees, according to prior literature, will be discussed.

2.1.1 Determinants of audit fees

The paper of Simunic (1980) introduced the basic model for determining audit fees, and tried to investigate the existence of competitiveness in the audit industry. To this day the model has been used by a wide range of different studies (Lyon & Maher, 2005; Seetharaman, Gul, & Lynn, 2002; Taylor & Simon, 1999; Bell, Landsman, & Shackelford, 2001; Palmrose, 1986). A major part of these studies has introduced several experimental variables and tried to improve the basic audit fee model. According to the study of Simunic (1980) audit fees are determined by: the size of the auditee, complexity of the auditee’s operations, receivables and inventories which is stated as a risk, principal industry of the auditee and finally whether the auditee is a public or closely held company.

After the paper of Simunic (1980) audit fees have been investigated on different outcomes. A study that builds further on the study of Simunic (1980) is a paper by Palmrose (1986), which focuses on the effect of auditor size and audit fees. The paper investigates whether big audit firms charge higher fees because of monopolistic powers, to reflect the higher quality of their audit services, or that large audit firms charge relative lower fees because of scale effects. Palmrose (1986) focuses on 1200 public and non-public companies in the United States. The results show that there

9 is a significant positive association between audit firm size and audit fees which is, after controlling for different variables, explained by the fact that big eight audit firms provide higher quality audits.

A subject that has drawn a lot of attention in audit fee literature, and which is not included in the audit fee model of Simunic (1980), is the litigation risk. Before giving an explanation of this risk it can be important to understand the risk assessment in the auditing process. The statement on auditing standards (SAS) no. 47 requires auditors to use the audit risk model (ARM) as part of the audit planning process (Houston, Peters, & Pratt, 1999). The ARM distinguishes the overall audit risk into inherent risk, control risk and detection risk. Houston et al. (1999, p. 284) describe the three types of risk as follows: “Inherent risk is the probability that an account balance or class of transactions contains a material misstatement before considering the effectiveness of the internal control system. Control risk is the probability that a material misstatement is not prevented or detected on a timely basis by the internal control system. Detection risk is the tolerable level of risk that auditing procedures will not detect material misstatements”. Besides these three risks, auditors also have to bear with business risk, which is a residual risk that in principle cannot be eliminated below a certain level (Bell, Landsman, & Shackelford, 2001). The risk of litigation arising from association with the client is seen as the primary part of this business risk (Bell et al., 2001; Houston et al., 1999).

The litigation risk can be described as the risk that legal actions will be taken against an individuals or corporations actions, products or services (Seetharaman, Gul, & Lynn, 2002). A study of Bell et al. (2001) investigates whether the auditee or auditor bear the costs of this litigation risk. They use a survey to investigate the number of audit hours spent on the client and the charged fee per hour. They predict that the hourly audit fees and the number of audit hours will increase when more risk is involved. The results of this study show that auditors increase the number of audit hours, but not the fee per hour when there is a higher risk of litigation. So it can be argued that while auditors put in more effort, they don’t demand higher wages for their efforts. However the results of this study are interesting, the paper focuses only on one big audit firm. Thus, the results are not generalizable.

A study in an international context of Taylor and Simon (1999) also examines the importance of litigation risk on audit fees. In addition to the litigation risk, they investigate the effect of disclosure requirements and the regulation intensity of a country on audit fees. According to the authors political, regulatory and economic variables, what they call macro-economic variables, have a significant impact on audit fees. They predict that the disclosure requirements of a country will increase the complexity of an audit and that a stricter regulation will require more audit work, which both will ultimately lead to higher audit fees. The results of their study shows

10 that the three variables have a significant positive influence on audit fees and are important determinants of audit fees.

In a more recent study of Hay, Knechel, & Wong (2006) a meta-analysis is used to test the effect of the most commonly used variables for determining audit fees in audit literature, the results of this study are helpfull to determine the most important determininants and will also be mentioned hereafter.

2.1.2 Size of the auditee

Based on prior literature about audit fees it is clear that firm size is an important determinant of audit fees, large auditees will require more work and this will have an influence on the charged audit fees (Seetharaman, Gul, & Lynn, 2002; Taylor & Simon, 1999; Simunic, 1980; Hay, Knechel, & Wong, 2006). In the first introduced model of Simunic (1980) the size of the auditee is determined by the natural log of total firm assets, the same method is also used in more recent studies (Kim, Liu, & Zhen, 2012; George, Ferguson, & Spear, 2013; Seetharaman, Gul, & Lynn, 2002).

Hay et al. (2006) show in their study that of the 147 audit papers they have investigated 105 papers use the log of total firm assets to proxy for the size of the firm. So it can be concluded that the size of the auditee is an important determinant of audit fees. The expectation is that the size of an auditee will have an positive effect on the charged audit fee.

2.1.3 Complexity of auditee

Like auditee size complexity is seen as an important determinant of audit fees (Simunic, 1980; Hay, Knechel, & Wong, 2006). The logical expectation is that more complex a client is the harder it is to audit, and more time consuming the audit will be (Simunic, 1980; Hay, Knechel, & Wong, 2006). The concept of complexity has been measured in different ways in prior studies, but the most typical indicator is the number of subsidiaries and the number of foreign subsidiaries (Hay et al., 2006). The reason that this measure is used as a proxy for complexity is that a higher amount of subsidiaries requires more work from the auditor in terms of consolidation (Pong & Whittington, 1994). In the analysis of Hay et al. (2006) it is concluded that there is no doubt that the relationship between fees and complexity is positive and significant.

2.1.4 Risk

Risk plays an important role in determining audit fees. Two areas that are stated as being difficult to audit are inventory and receivables, these items require more evidence gathering and have valuation challenges (Simunic, 1980; Hay, Knechel, & Wong, 2006). The most common way to

11 measure this risk are receivables divided by total assets, inventory divided by total assets, and the combination of both divided by the total assets (Hay et al., 2006). The expectation is that this measure will have a positive association with audit fees.

Another measure for risk that is commonly used in prior literature is leverage, this measures the risk of a client failing which exposes the auditor to a potential loss (Simunic, 1980; Hay, Knechel, & Wong, 2006). Leverage is used in most prior studies about audit fees, and is measured by dividing the total amount of debts by the total amount of assets (Simunic, 1980; Seetharaman, Gul, & Lynn, 2002). Leverage is expected to have a positive association with audit fees (Hay et al., 2006).

The last measure of risk that is often used in prior literature is profitability, this reflects the extent to which the auditor may be exposed to a loss because a client is not financially viable (Simunic, 1980; Hay, Knechel, & Wong, 2006). The general perception is that the worse the performance of an auditee the more the risk an auditor is exposed to, and the higher the audit fee that is charged (Hay et al., 2006). The two most used variables for profitability are the net income divided by the total assets (ROA), and a dummy variable for the existence of a loss (Hay et al., 2006). It is expected that the relationship between audit fees and ROA is negative and with loss positive (Simunic, 1980).

2.1.5 Audit quality

Prior literature makes clear that audit quality also plays an important role in determining audit fees (Basioudis & Francis, 2007; Seetharaman et al., 2002; Hay et al., 2006). Higher audit fees are expected when it is recognized that an auditor provides high quality audits (Hay et al., 2006). A significant part of the available literature finds a strong support between high audit quality and big 4 (or big 5,6) audit firms, this suggests that auditors of these big firms are able to demand premiums for their audits compared to auditors of non-big 4 firms (Hay et al., 2006). The most common way to measure this is using a dummy variable for firms that are audited by big 4 auditors (Hay et al., 2006). So, the expectation is that there is a positive relation between audit fees and big 4 audit firms.

2.1.6 Litigation propensity

Audit fees are also related with auditor exposure to losses from legal liabilities (e.g., Taylor & Simon, 1999; Seetharaman et al., 2002; Venkataraman, Weber, & Willenborg, 2008). The perception is that while effort increases the cost of performing the audit, it decreases the expected litigation costs

12 (Venkataraman et al., 2008; Seetharaman et al., 2002). This is explained by the fact that more effort will increase the likelihood that auditors detect a material misstatement, and satisfy audit standards. Seetheraman et al. (2002), Taylor and Simon (1999) and Venkataraman et al., (2008) state in their study that a more “litigious environment” will increase the likelihood of lawsuits against auditors, which will lead to an increase in audit fees. So, it is expected that a more litigious environment will have a positive effect on audit fees.

2.2 Principle and rules based accounting standards

The fierce debate between principle and rules based standards have been going on for several years, and in the current environment the accounting standards of the IFRS are seen as principle based, whereas the accounting standards of U.S. GAAP is described as being rules based (Collins, Pasewark, & Riley, 2012; Benston, Bromwich, & Wagenhofer, 2006; Jamal & Tan, 2010). Rules based standards are often seen as providing very detailed guidance with “bright-lines”, while principle based accounting standards are normally characterized as containing clear statements but lacking detailed implementation guidance (Collins et al., 2012). A lot is investigated in prior literature about rules and principle based standards and why one is more beneficial than the other in a specific situation. In the following section a short introduction about IFRS and the U.S. GAAP will be given, after that the most important studies which investigate the effect of rules and principle based standards on different outcomes will be discussed, which will help to understand why there is such a fierce debate between the two competing standards.

2.2.1 U.S. GAAP and IFRS

Generally Accepted Accounting Principles (GAAP) varies from country to country in terms of bodies issuing it, level of authority and in term of its sources (Ampofo & Sellani, 2005). For example, there is a Mexican, Chinese and Canadian GAAP and generally the authorities that are responsible for setting these GAAP is the national accounting standards board in a country. Currently the Financial Accounting Standards Board (FASB) is the body which is responsible for issuing U.S. GAAP in the United States. U.S. GAAP is often described as containing extensive disclosures, rule based and comprehensive instructions (Ampofo & Sellani, 2005; Agoglia et al., 2011).

In March 2002 the European parliament passed a resolution which required all firms listed on stock exchanges of European countries to apply IFRS when preparing financial statements. The

13 use of IFRS was a substantial shift because of the many differences of IFRS and the domestic standards used in the countries in Europe (Armstrong, Barth, Jagolinzer, & Riedl, 2010). The goal of the EU was to achieve a capital market integration with a step towards global convergence of financial reporting. However, firms have to make use of IFRS, which are issued by the International Accounting Standards Board (IASB), the standards must be approved by the European Commission before they are required in the EU (Armstrong et al., 2010). A distinctive feature of the IFRS is that they are principle based instead of rules based (Carmona & Trombetta, 2008; Agoglia et al., 2011).

2.2.2 Rules and principle based standards

As mentioned earlier rules based accounting standards contain “bright-lines” and detailed implementation guidance. But according to Hail, Leuz and Wysocki (2010, p. 376) “the exact meaning of the term bright-lines is not well defined and often unclear”. According to Donelson, McInnis and Mergenthaler (2012) the characteristics of rules based standards include scope and legacy exceptions, a high level of detail and extensive implementation guidance. The intention of these characteristics is to lead to more precision and lower the amount of professional judgement required (Donelson et al., 2012).

Those favoring rules based standards often mention that principle based standards allow too much flexibility and room for interpretation when preparing financial statements, what ultimately will lead to an impairment of audit quality and resulting in more diverse financial reporting outcomes (Peytcheva, Wright, & Majoor, 2014). These proponents believe that standards based on rules will reduce diversity, and enhance consistency and comparability (Collins, Pasewark, & Riley, 2012).

There are several studies where the findings suggest a preference for rules based standards instead of principle based standards. For example, a study by Trompeter (1994) investigates the impact of more precise standards on audit judgements involving audit-client conflicts. The study involves cases and questionnaires about auditor judgements and responses in specific situations, which were answered by 54 audit partners. The results show that more rules based standards significantly limits the client’s ability to influence auditor’s judgement, which will allow the auditor’ judgements to be more credible (Trompeter, 1994).

Hackenbrack and Nelson (1995) investigate whether less precise standards give experienced auditors the opportunity to approve reporting decisions that are in line with their incentives. The authors argue that “vague” standards may provide auditors a mechanism to justify more aggressive reporting methods. From the results of the study it becomes clear that auditors approve reporting

14 decisions that are consistent with their incentives and make use of the vagueness of the imprecise standard to justify their actions (Hackenbrack & Nelson, 1995).

Another study of Ng and Tan (2003) find that when there is a weak audit committee under a principle based environment, auditors are more likely to allow aggressive revenue recognition than when there is a detailed guidance/rule specifying the appropriate reporting treatment. In the same study auditors also conclude that precise standards enhance auditor effectiveness and the financial reporting quality (Ng & Tan, 2003).

However, after a wave of different corporate accounting scandals in past years, and especially after the Enron corporation scandal, the rules based accounting standards became largely under fire (Benston et al., 2006). According to proponents of the principle based approach Enron was “complying” with the strict rules, but did not comply with the underlying principles of accounting and was thereby able to provide a misleading economic view of the company (the so called “check-the-box” mentality) (Beston et al., 2006). The proponents also suggest the increased disclosure requirements under principle based standards will lead to less diversity in judgement, and the absence of “bright-lines” may require preparers to look beyond the form of a transaction which will ultimately lead to a higher reporting quality (Collins et al., 2012). The accounting scandals and the pressure from the community who favor principle based standards led to the concern that U.S. accounting standards became too rules based, because of the “bright-lines” and the very detailed guidance many people within the financial community felt that the standards invite opportunistic interpretation (Agoglia, Doupnik, & Tsakumis, 2011). These developments led to the call for a more principle based approach in the United States and eventually to the debate of which sort of standard to use, that has been going on to this day (Agoglia et al., 2011).

In a specific financial reporting situation it can be stated that principle based standards require the auditor to exercise judgement, rather than following specific rules (Collins et al., 2012). One of the studies that examines the effect of rules based versus principle based accounting standards is by Psaros and Trotman (2004). In their study they examine how accountants’ consolidation judgements are impacted by rules based and principle based accounting standards (substance-over-form). They find that accountants make more aggressive consolidation judgements under rules based accounting standards than under principle based standards (Psaros & Trotman, 2004).

Another study that focuses on aggressive reporting under the two standards is by Jamal and Tan (2010), which investigates whether a principle or rules based oriented type auditor affects the financial managers reporting decisions, under rules and principle based accounting standards. The results of the research shows that the type of auditor has no effect on the managers reporting

15 decisions in a rules based setting, but managers are less likely to report aggressively in a principle based setting under a principle based type auditor. So, they conclude that a move toward principle based accounting standards will likely lead to an improvement of financial reporting quality, but only when the auditors’ mindset is also principle based (Jamal & Tan, 2010).

Some of the existing studies investigate the impact of rules and principle based standards on the behavior of different involved parties. For example a study of Segovia, Arnold and Sutton (2009) researches the effect of rules and principle based standards on the behavior of auditors. Their study explores the effect of the two accounting standards on the auditors’ willingness to allow leeway in reporting practices and how the auditors’ behavior is influenced by the standards (Segovia et al., 2009). They used a survey of 114 auditors from international and regional audit firms in the United States, and find out that auditors are more willing to allow clients to manage earnings under rules based standards (Segovia et al., 2009). The findings of this study suggests that principle based standards reduce earnings management in some way, while auditors under a rules based approach are more likely to allow earnings management.

Agoglia, Doupnik and Tsakumis (2011) investigate the effect of rules based and principle based standards on the reporting behavior of CFOs and other experienced financial statement preparers. Their results show that the participants of the study are less likely to report aggressively when applying a less precise reporting standard (Agoglia et al., 2011). They also find that the audit committee strength affects aggressive reporting in a rules based regime, while audit committee has no effect in a principle based regime. Further, the study shows that there is a significantly less variability amongst preparers’ financial reporting decisions under principle based standards, which suggests an increase in comparability of financial statements when using a principle based approach (Agoglia et al., 2011).

2.2.3 Audit fees under rules based and principle based standards

In the last years there have been a lot of discussion about whether the potential adoption of IFRS in the U.S. would be beneficial (Hail, Leuz, & Wysocki, 2010; George, Ferguson, & Spear, 2013). This will mean that the rules based standards (U.S. GAAP) will make room for more principle based standards (IFRS). The adoption of a different kind of standards will have consequences for firms, investors and other stakeholders (Hail et al., 2010). The use of different standards will also have an effect on audit fees, which is a subject that has drawn a lot of attention in the audit literature.

As mentioned above rules based standards have bright-lines and detailed guidance, because of the strict “rules” the litigation risk in countries who make use of this kind of standards will be

16 higher (Seetharaman et al., 2002). This theory is based on “contends that the specificity of rules-based standards provides plaintiffs with a clear path to successful litigation” (Donelson et al., 2012, p. 1248).

An important paper that focuses on the litigation risk in audit pricing is by Seetharaman et al. (2002), they state that auditors are likely to charge a risk premium to cover for possible future litigation costs. They examine whether UK firms that are cross-listed in the United States have higher audit costs than similar companies that are not cross-listed on the U.S. markets. The results show that companies that are cross-listed on US markets pay an approximately 20% higher fee than companies that are not cross-listed, which is according to the authors explained by auditors charging a risk premium to cover for possible litigation costs or due to an increase in effort in defense against the increased likelihood of future litigation (Seetharaman et al., 2002).

Another study that looks at the effects of cross-listing on audit fees is by Choi, Kim, Lui and Simunic (2009). They build further on the study of Seetharaman et al. (2002), but instead of investigating cross-listing effects of only one country (U.K.) they use cross-listing data from 14 countries. Their results also show that auditors charge higher fees for firms that are cross-listed in countries with stricter regimes and rules, this shows that audit fees increase with the increase in the strength of legal regimes and rules (Choi et al., 2009).

Venkataraman et al. (2008) examines the effect of litigation risk on auditor compensation, they compare audit fees of companies before they go public with the audit fees the companies pay once they are public. The results show that the investigated firms pay higher audit fees after they are public, the authors associate this with the fact that auditors demand higher fees because they are exposed to more litigation risk when auditing public companies (Venkataraman et al. 2008).

So Choi et al. (2009), Seetharaman et al. (2002) and Venkataraman et al. (2008) all conclude in their studies that audit fees are higher in a more litigious environment, due to an increase in effort in defense against the increased likelihood of future litigation, or to cover for possible litigation costs.

On the other hand principle based standards require a large amount of professional judgment and more complex estimates of auditors compared to rules based standards (Kim, Liu, & Zhen, 2012). Because of the “vagueness” of these standards auditors have to use more of their experience instead of “following the rules”, this makes principle based standards more complex for auditors (Kim et al., 2012). As mentioned earlier IFRS is seen as containing principle based accounting standards, and since the European parliament required listed companies in Europe to prepare financial statement according to IFRS many studies investigated the effect of this adoption (Kim et al., 2012).

17 One of those studies is from Kim et al. (2012) which focuses at the effect of IFRS on audit fees. In their study they investigate the effect of IFRS adoption on audit fees in European countries, the focus is on countries that switched from their local GAAP to IFRS in 2005. Their sample consists of 2.860 firm years from 11 European countries. Based on the results of the study it is stated that IFRS related audit fees increase because of the complexity that principle based standards brought by (Kim et al., 2012).

A research by George, Ferguson and Spear (2013) also investigate the effect of IFRS adoption on audit fees, but in an Australian context. They use a sample of publicly traded companies on the Australian stock exchange for the period 2002-2006. The findings suggests an increase in audit fees of 8 percent in the year after IFRS adoption, beyond the yearly normal increases of the fees. In addition they find that smaller firm will incur a higher increase in audit fees than larger firms.

Another study that focuses on the effect of IFRS adoption is by Vieru and Schadewitz (2010), they use a sample of 73 Finnish companies that switched to IFRS. Their results support their presumption that IFRS is more complex and the risk associated with this complexity causing an increase in audit fees.

From these studies mentioned above it becomes clear that prior literature about audit fees under a rules and principle based setting shows contradictory results.

2.3 Investor protection and legal tradition

From previous studies it becomes clear that firm specific factors are important determinants of audit fees (e.g., Simunic, 1980; Palmrose, 1986). But institutional factors and characteristics of a country also play an important role in determining audit fees (Choi et al., 2009; Taylor & Simon 1999). This research focuses on two institutional factors namely the investor protection and legal tradition of a country.

Defond and Hung (2004) describe investor protection as the terms and laws’ extent to protect investors’ rights and the strength of the legal institutions to enforce these protective laws. Strong investor protection rights can have an impact on the litigation environment of a country, which will thereby affect auditors’ remuneration (Jaggi & Low, 2011).

Scholars in general identify two broad legal traditions: civil law and common law (Porta, Lopez-de-Silanes, Shleifer, & Vishny, 1998). However there is no unanimity how to exactly define the two legal families, legal scholars often use the following six criteria to identify the two legal

18 traditions: Historical background and development of the legal system, theories and hierarchies of sources of law, the working methodology of jurists within the legal systems, the characteristics of legal concepts employed by the system, the legal institutions of the system, and finally the divisions of law employed within a system (Porta et al., 1998). A broader explanation about investor protection, the two legal families and how these are associated with audit fees will be explained hereafter.

2.3.1 Investor protection

As described earlier investor protection can be seen as the laws that protect the rights of investors. Investor protection empowers investors to take legal action against managements and auditors if information in the financial statements are considered to be questionable (Jaggi & Low, 2011).

The study of Jaggi and Low (2011) is one of the few studies that investigates the effect of investor protection on audit fees. Their findings show that there is a significant positive association between audit fees and investor protection. According to the authors this suggests that auditors charge a risk premium for potential litigation costs because higher investor protection in a country increases the risk for litigation (Jaggi & Low, 2011).

On the other hand Leuz Nanda and Wysocki (2003) investigate the association between the investor protection level of a country and earnings management. They find in their study that discretionary accruals are lower in countries with strong investor protection, which means that the quality of financial information is higher in countries with high levels of investor protection. They state that higher quality information can lead to lower audit risk and effort, which can ultimately lead to lower audit fees (Leuz et al., 2003). An important explanation for the reason of higher financial information quality in countries with high investor protection is the threat for legal suits from investors if the reported information is inaccurate or lacking quality (Jaggi & Low, 2011; Francis & Wang, 2008).

So it can be stated that in countries with higher investor protection the litigation risk for auditors is higher (Jaggi & Low, 2011; Francis & Wang, 2008). From the studies that are discussed earlier we know that a more litigious environment is associated with higher audit fees (e.g., Seetharaman et al., 2002; Choi et al., 2009). So, the expectation is that higher levels of investor protection is associated with higher audit fees.

2.3.2 Civil and common law

The civil, also called Romano-Germanic, law is the oldest, most influential and the most widely distributed legal tradition around the world (Porta et al., 1998). The civil law is generally seen as

19 highly systematized and structured and relies on broad, general principles while often ignoring the details (Tetley, 2000).

Its origin comes from the Roman law, makes use of statutes, comprehensive codes and relies on legal scholars to determine and formulate the rules (Porta et al., 1998). Typically three common families are recognized within the civil law tradition: French, German and Scandinavian (Porta et al., 1998). The French code was written under Napoleon in 1807 and was brought to Belgium, the Netherlands, Poland, Italy and some western regions of Germany (Porta et al., 1998). The French code has also been of significant influence in Luxembourg, Portugal, Spain and Italy. The German code was written in 1897 and because it was produced several years later it is not as widely adopted as the French code. The German code has had an important influence on the legal theory in Austria, Greece, Hungary, Italy, Switzerland, Yugoslavia, China, Japan and Korea (Porta et al., 1998). The Scandinavian family is also seen as part of the civil-law tradition, however its law is not been based on the Roman law as much as the German and French laws, but the Nordic countries stopped using the law. (Porta et al., 1998).

On the other hand, the common law is usually seen as much more detailed in its instructions than the civil law (Tetley, 2000). The common law family includes the law of England and the laws that are based on the English law (Porta et al., 1998). An important characteristic is that common law consists of judges who have to resolve specific conflicts. In the past common law has been spread over the British colonies, including the United States, Canada, Australia, India and many more countries (Porta et al., 1998).

One of the few studies that focuses on the differences in common and civil law countries is by Porta et al. (1998), for instance they look at the difference in strength of law enforcement in 49 countries around the world, which can measure how litigious a country is. They examine laws governing investor protection, the quality of enforcement of these laws, and ownership concentration. From their results they conclude that law enforcement and investor protection is stronger in common law countries than in civil law countries (Porta et al., 1998).

Francis, Khurana and Pereira (2001) build further on the study of Porta et al. (1998) and find from a sample of 31 countries that accounting standards are timelier and more transparent in common law countries, which is explained by the fact that these countries require more detailed public disclosure of accrual-based accounting information.

So common law countries have stronger law enforcement and investor protection and can thereby be seen as more litigious (Jaggi & Low, 2011; Francis & Wang, 2008). Based on prior literature the expectation is that the audit fee increases with the strength of a legal regime, thus it

20 is expected that audit fees are higher in common law countries (e.g., Seetharaman et al., 2002; Choi et al., 2008, 2009).

2.4 Hypotheses

In summary, audit fees are determined by the size, complexity, risk, litigation propensity and the provided audit quality. In the current environment IFRS is seen as having principle based and U.S. GAAP having rules based standards, which according to existing literature both have their pros and cons. Research about the audit fees under both principles also showed contradictory results, studies of Choi et al (2009), Seetharaman et al. (2002) and Venkataraman et al. (2008) find that fees under rules based standards are higher due to the increase in litigation risk, while research from Kim et al. (2012) and George et al. (2013) find that audit fees increase when using more principle based standards due to an increase in complexity. So the first hypothesis will be:

H1: Audit fees are higher under rules based accounting standards, and thus under U.S. GAAP

I also want to test whether high investor protection will lead to higher audit fees. From the literature review section is becomes clear that higher investor protection increases the risk for potential litigation, and a more litigious environment is associated with higher audit fees (e.g., Jaggi & Low, 2011; Seetharaman et al. 2002; Choi et al., 2009). This leads to the following hypothesis:

H2: Audit fees are higher in countries with higher levels of investor protection

In addition to the first two hypotheses I will also test the audit fees in the two different legal traditions. From what is mentioned in the literature review it is clear that law enforcement is stronger in common law countries while the accounting standards are also more transparent and timely in this legal system compared to civil law countries (Porta et al., 1998). Because of this common law countries are seen as more litigious (Porta et al., 1998). Based on prior literature the expectation is that audit fees increase with stronger legal regimes (e.g., Seetharaman et al., 2002; Choi et al., 2008, 2009). Thereby the third hypothesis will be:

21

3 Research methodology

In this chapter the research method to examine the effect of the two accounting standards, investor protection and legal tradition on audit fees will be explained. First the used databases and the chosen sample will be discussed which will be followed by the used empirical models.

3.1 Data

The research is carried out by a database research, so a quantitative archival study will be performed. Compustat Fundamentals annual database, Audit Analytics and financial statements are used to collect the audit fee and company data. The investor protection level data is collected from the Worldwide Governance Indicators via Worldbank and for legal tradition information is collected from the World Factbook via the CIA website library (Worldbank, 2015; CIA, 2015).

Consistent with a study of Collins, Pasewark, & Riley (2012) this study also compares the companies that are members of the Fortune Global 500 that report under IFRS and U.S. GAAP. The collected data spans from 2011 till 2014, this period is chosen because it is a stable period without the influence of the financial crisis or whatsoever.

The initial sample consisted of the 500 companies listed on the Fortune Global 500 list. After merging data from Compustat and Audit Analytics several companies were deleted because they lacked audit fee data. After this process the sample consisted of 193 of the 500 companies with 772 firm-year observation. Companies which lacked control variable data were also deleted from the sample (57 companies). I hand collected audit fee data for 36 companies to make sure I have companies in the sample that operate in the same industry with different accounting standards (IFRS or U.S. GAAP). This led to the final sample of 170 companies and 680 firm-year observations.

Table 1 Sample data

Data from 2011 till 2014 Number of companies

Number of firm-year observations Initial sample 500 2000 Missing fee data -278 1108 Missing control variable data -88 332 Hand-collected data +36 140

22 3.2 Empirical models

The tests are based on regressions of the natural logarithm of audit fees on several variables, including indicator variables. In the literature review section the determinants of audit fees and the most common used variables are already discussed. These are all based on prior literature and will be used in the models (e.g., Simunic, 1980; Seetharaman et al., 2002; George et al., 2013). After collecting data for the research it became clear that the majority of the Fortune Global 500 companies didn’t have foreign assets reported, because of this the control variable for foreign assets is not used in the models.

The main dependent variable is the natural logarithm of audit fees paid to auditors for audit services. In addition the natural logarithm of non-audit fees and the total amount of auditor remuneration, which is the audit fee plus the non-audit fee, will be tested as alternative dependent variables. Adding these alternative outcome variables help determine how robust the results are to inaccuracies in the allocation of audit remuneration between audit and non-audit services (Seetharaman et al., 2002).

Based on the study of Porta et al. (1998) for investor protection by law enforcement, the average score of the following three variables from the Worldwide Governance Indicators are used: estimate of control of corruption, estimate of government effectiveness and estimate of the rule of law. This score range from -1 to 2.

So for the first hypothesis the following models will be used:

LogAf = 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROA+ 𝛽6LOSSit + 𝛽7Big4it +

𝛽8IFGAit+ 𝜀I (1)

LogNonAf = 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROA+ 𝛽6LOSSit + 𝛽7Big4it +

𝛽8IFGAit+ 𝜀I (2)

LogTotAf = 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROA+ 𝛽6LOSSit + 𝛽7Big4it +

𝛽8IFGAit+ 𝜀I (3)

Variable definitions:

LogAf = The natural logarithm of total audit fees for firm i in year t LogNonAf= The natural log of non-audit fees

LogTotAf = The natural log of total auditor remuneration, including both audit fee and non-audit fee

23 Inv = Ratio of total inventories to total assets

Rec = Ratio of total receivables to total assets Lev = Long term debt divided by total assets

ROA = Return on assets, net income before extraordinary items divided by total assets Loss = An indicator variable equaling 1 if a firm reported a loss and 0 for otherwise Big 4 = An indicator variable equaling 1 for audited by big 4 and 0 for otherwise

IFGA = An indicator variable equaling 1 for firm reports under IFRS and 0 for U.S. GAAP

For the second hypothesis the following models will be used:

LogAf= 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROAit + 𝛽6LOSSit + 𝛽7Big4it +

𝛽8InPrit+𝜀it (4)

LogNonAf = 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROAit + 𝛽6LOSSit + 𝛽7Big4it +

𝛽8InPrit+𝜀it (5)

LogTotAf = 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROAit + 𝛽6LOSSit + 𝛽7Big4it +

𝛽8InPrit+𝜀it (6)

Where the variables are the same as the variables used in the previous model except InPr which is the level of investor protection that range from -1 to 2.

And for the third hypothesis the models that will be used is:

LogAf= 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROAit + 𝛽6LOSSit + 𝛽7Big4it +

𝛽8LAWit+𝜀it (7)

LogNonAf = 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROAit + 𝛽6LOSSit + 𝛽7Big4it +

𝛽8LAWit+𝜀it (8)

LogTotAf = 𝛽0 + 𝛽1Sizeit + 𝛽2Invit + 𝛽3Recit + 𝛽4Levit + 𝛽5ROAit + 𝛽6LOSSit + 𝛽7Big4it +

𝛽8LAWit+𝜀it (9)

Where the variables are the same as indicated above except LAW which is an indicator variable equaling 1 if the firm is located in a common law country and 0 in a civil law country.

24

4 Results

In this section the results of this research will be shown. First, the descriptive statistics will be discussed, secondly the correlation coefficients will be explained, thirdly the model will be tested on multicollinearity, after that the regression results will be shown and finally the sensitivity analysis will be discussed.

4.1 Descriptive statistics

The descriptive statistics for the used variables in this research are shown in two tables. Table 2 represents the descriptive statistics for companies that report under IFRS and table 3 for companies that report under U.S. GAAP.

Table 2 shows that the smallest amount of audit fees paid by the auditees in IFRS jurisdictions are $1375.93 and the highest $82751.2. The smallest company in terms of the total amount of assets are $2861 and the biggest $2807491. The mean for the audit fees that are paid by companies under IFRS are respectively $21150.31 and the mean for the total assets are $251199.63. Furthermore, the table shows that all of the companies in the sample that report under IFRS are audited by big 4 audit companies, which is expected with companies that are in the Fortune Global 500. The ratio variables shows that the inventory ratio is 0.0817 which means that most of the firms have a low portion of their assets tied up in inventory. The leverage ratio of 0.1719 shows that the firms that report under IFRS have a relatively low debt ratio. The ROA ratio tells us how profitable the companies are relative to their assets, the ratio of 0.0359 is low and shows that the firms have a low income compared to the amount of assets. The loss dummy shows that the majority of the companies did not report a loss in the sample period. The mean of the dummy variable LAW shows us that the majority of the companies that report under IFRS in the sample are located in civil law countries.

In comparison with table 2, table 3 shows that the minimum amount of audit fees paid by firms that report under U.S. GAAP are $721.5 and the highest amount $96600, which means that minimum fees are lower for U.S. GAAP companies while the maximum amount is higher for firms under this accounting standard. The mean of $17740.01 for the audit fees tells us that average fees for firms under U.S. GAAP are lower than firms that report under IFRS. Similar to the firms under IFRS, almost every company is audited by a big 4 firm. So the difference in audit fees will not likely be driven by the fact that large accounting firms charge fee premiums for their clients (Taylor & Simon, 1999). The ratio variables show us that the inventory ratio of 0.1179 is higher, also the

25 leverage ratio of 0.2144 means that the U.S. GAAP firms have a relatively higher debt ratio. As mentioned earlier we know that assets like receivables and inventory require more audit effort which can lead to higher fees (e.g., Taylor & Simon, 1999). The low differences between the receivables and inventory ratios under both standards leads to a better comparison of the charged audit fees, without the influence of a complexity factor.

The ROA and loss ratio are very low which also means that the companies that reported a loss under U.S. GAAP are very low, compared to firms that made a profit. This means that auditors under both standards are, relatively speaking, not too much exposed to extra risk due to a bad financial performance of the auditee. Furthermore, the LAW dummy variable of 1.00 shows that the companies that report under U.S. GAAP are located in a common law country.

Table 2

Descriptive statistics IFRS firms (N = 340)

Mean Std. Deviation Minimum Maximum Audit fees (in $ x1000) 21150.31 15776.55 1375.93 82751.2 LogAf 16,6002 ,81562 14,13 18,23 LogNonAf 14,7788 1,38639 8,42 17,92 LogTotAf 16,7909 ,84486 14,13 18,60 Assets ($ million) Control variables 251199,63 420643,391 2861 2807491 Size 11,5574 1,31869 7,96 14,85 Inv ,0817 ,06988 0,00 ,31 Rec ,1732 ,13993 ,02 ,67 Lev ,1719 ,11331 0,00 ,46 ROA ,0359 ,04628 -,14 ,23 Loss ,0972 ,29678 0,00 1,00 Big 4 1,0000 0,00000 1,00 1,00 Variables of interest IFGA 1,0000 0,00000 1,00 1,00 InPr 1,3092 ,73927 -,26 2,00 LAW ,2535 ,43576 0,00 1,00

26 Table 3

Descriptive statistics U.S. GAAP firms (N = 340)

Mean Std. Deviation Minimum Maximum Audit fees (in $ x1000) 17740.01 17518.96 721.5 96600 LogAf 16,2447 1,00035 13,49 18,39 LogNonAf 14,6505 1,49684 8,52 17,47 LogTotAf 16,4787 1,03320 13,70 18,60 Assets ($ million) Control variables 170421,75 407141,312 2491 2573126 Size 10,8893 1,36677 7,82 14,76 Inv ,1179 ,11937 0,00 ,47 Rec ,1569 ,14267 0,00 ,74 Lev ,2144 ,16230 0,00 ,98 ROA ,0618 ,05520 -,15 ,35 Loss ,0408 ,19812 0,00 1,00 Big4 ,9898 ,10063 0,00 1,00 Variables of interest IFGA 0,0000 0,00000 0,00 0,00 InPr 1,4700 0,00000 1,47 1,47 LAW 1,0000 0,00000 1,00 1,00 4.2 Correlation coefficients

Table 4 shows the correlation matrix for all the used variables in the models. It becomes clear that size is highly and significantly correlated with the audit, non-audit and total fees. Which is expected by the fact that bigger firms require more audit work (e.g., Seetharaman et al. 2002; Kim et al. 2012). Furthermore Inv, Lev and ROA are significantly and negatively correlated with the audit and total amount of fees. Inv, Rec and Lev are also significantly correlated with each other. The variable Loss shows a negative correlation with audit fees but not significant, however it is significantly correlated with the non-audit fees. In the sample there is no correlation between big 4 audit firms and the audit fees, which can be explained by the fact that almost all companies in the sample are audited by a big 4 audit firm. Further, there is a significant positive correlation between the dummy variable IFGA and the paid audit fees and total audit fees, but not with the

27 non-audit fees. There is also a significant positive correlation between the variable InPr and the audit, non-audit and total audit fees. Finally, the dummy variable LAW shows a significant negative correlation with the amount of audit and total fees, but not with non-audit fees.

4.3 Multicollinearity

An important concern when a model is tested with more than one predictor variable is multicollinearity, which means that there is a strong relationship between two or more predictor variables (Field, 2013). If there is a strong relationship between several predictor variables it is impossible to obtain a unique estimate of the regression coefficients, so multicollinearity makes it difficult to assess the individual importance of a predictor variable (Field, 2013).

One way of identifying multicollinearity is looking at the variance inflation factor (VIF). The VIF shows if a predictor variables has a strong relationship with other predictor variables. The general assumptions is that if the largest VIF is greater than 10 there is cause for concern of multicollinearity (Field, 2013; Taylor & Simon, 1999). From table 5 it becomes clear that there is no concern for multicollinearity issues in this sample, because none of the VIFs approach the level of 10.

Table 4

Correlation matrix

LogAf LogNonAf LogTotAf Size Inv Rec Lev ROA Loss Big4 IFGA InPr LAW LogAf 1 LogNonAf ,728** ₁ LogTotAf ,983** _,822** ₁ Size ,762** _,612** _,763** ₁ Inv -,165** _-,172** _-,172** _-,277** ₁ Rec ,271** _,208** _,271** _,263** _,108** ₁ Lev -,168** _-,088* _-,148** _-,185** _-,110** _-,170** ₁ ROA -,112** _{-0,002 -,091}* _-,207** _{-0,074 -,182}** _{0,028 1} Loss -0,05 -,122** _{-0,063 -,083}* _{0,073 -,092}* _{0,063 -,466}** ₁ Big4 0,065 ,115** _{0,074 0,028 0,057 0,057 -,147}** _{0,061 0,02 1} IFGA ,187** _{0,043 ,159}** _,238** _-,174** _{0,057 -,145}** _-,241** _,113** _{0,066 1} InPr ,186** _,279** _,224** _{0,047 ,129}** _,114** _{-0,058 0,06 -0,029 -0,011 -,163}** ₁ LAW -,172** _{0,029 -,126}** _-,192** _,103** _{-0,016 ,100}** _,275** _-,143** _{-0,052 -,793}** _,312** ₁

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

Table 5

Multicollinearity

Variables VIFa _VIFb _VIFc

Size 1,357 1,394 1,357 Inv 1,232 1,250 1,232 Rec 1,187 1,196 1,187 Lev 1,121 1,127 1,121 ROA 1,548 1,529 1,548 Loss 1,388 1,379 1,388 Big4 1,044 1,045 1,044 IFGA 2,901 2,799 2,901 InPr 1,181 1,143 1,181 LAW 3,065 2,895 3,065

a. Dependent variable LogAf b. Dependent variable LogNonAf c. Dependent variable LogTotAf

4.4 Regression analysis

For testing the hypotheses in this research three regression analysis are made, with respectively the audit fee for audit services, the fees for non-audit services and the total auditor remuneration (audit fee plus the non-audit fee) as the three different dependent variables. In table 6 the regression results are shown, the first analysis has an adjusted R2 _{of 0.617, the second 0.465 and the third 0.628}

this means that respectively 61.7%, 46.5% and 62.8% of the used variables is explained by the model, so the overall fit of the model is good.

The first column with LogAf as the dependent variable shows a significant relation at the 1% level with Size and the variable Rec. This is explained by the fact that the firm size increases the audit work and thereby the audit fees. Rec is stated as a risk for auditors which increases the audit fees, both these findings are as expected and consistent with prior literature of Simunic (1980), Hay et al. (2006) and Choi et al. (2009). The positive significant relation of ROA with the audit fee is not as expected, but this might be caused by the fact that the majority of the firms in the sample had a high positive income before extraordinary items and relatively high audit fees.

Furthermore, for the variables of interest, there is a negative significant relation at the 5% level with a coefficient of -0.154 between IFGA and LogAf which means that the paid audit fees, for audit services, by firms which use IFRS are in general lower than firms that use U.S. GAAP.

30 The expectation was that the audit fees for firms that report under U.S.GAAP were higher, so this means that the first hypothesis is supported. This finding is also in line with previous studies of Choi et al. (2009), Seetharaman et al. (2002) and Taylor and Simon (1999) who also found that fees are higher in a more litigious environment. Further, there is a significant positive relation at the 1% level with a coefficient of 0.337 between InPr and LogAf, which means that higher investor protection levels of a country is associated with higher audit fees. Thus, it can be stated that the second hypothesis is also supported and the findings are similar to the findings of Jaggi and Low (2011). Finally, there is a significant negative relation at the 1% level with a coefficient of -0.332 between LAW and LogAf which tells us that the fees paid by auditees in common law countries are lower than the fees paid by auditees in civil law countries. Therefore based on the regression results the third hypothesis is rejected.

The second column of table 6 provide the regression results with LogNonAf as the dependent variable. We can see that the size of a company also has a significant positive relation and a coefficient of 0.667 with the amount of non-audit fees paid. There is also a significant positive relation at the 5% level between Lev and LogNonAf. This can be caused by the fact that companies with high leverage ratios have potential going-concern problems and are therefore more likely to utilize such services from auditors (Seetharaman et al., 2002). Furthermore there is a non-significant negative relation between IFGA and LogNonAf. A significant relation at the 1% level with a coefficient of 0.810 is there between InPr and LogNonAf, from what is mentioned earlier we know that higher investor protection levels are associated with a more litigious environment (Jaggi & Low, 2011). So this finding is in line with the research of Seetharaman et al. (2002) where non-audit fees are also higher in a more litigious environment. The relation of LogNonAf and LAW is positive but not significant.

The third column with LogTotAf as the dependent variable shows almost similar results as with LogAf as the dependent variable, which is as expected. There is also a significant relation at the 1% level between LogTotAf, Size, Rec and ROA. The coefficient of Lev is not significant, but positive which is line with studies of Choi et al. (2009) and Simunic (1980). For the variables of interest there is a negative relationship, but not significant, between IFGA and LogTotAf. This means that it’s not certain that the total audit fees are lower for firms under IFRS, which wasn’t the case when looking at the paid audit fees for audit services. Further, InPr and LogTotAf are significantly and positively related with a coefficient of 0.401. Which means that the total audit fees are higher in countries with higher investor protection. And finally the significant negative relation at the 1% level with a coefficient of -0.237 between LAW and LogTotAf suggests that the total audit remuneration is lower in common law countries.

31 Table 6

Audit fee regression results for testing H1, H2 and H3 (N = 680)

Variables

Dependent variable

LogAf Dependent variable LogNonAf Dependent variable LogTotAf

(Constant) [9,957] (26.187) 0,000** [3.749] (5.301) 0.000** [9.561] (24.800) 0,000** Size [0,510] (27.069) 0,000** [0,667] (18.798) 0,000** [0.530] (27.750) 0,000** Inv [0,138] (0.572) 0,568 [-0,384] (-0.861) 0,389 [0.069] (0.281) 0,779 Rec [0,497] (2.882) 0,004** [0,412] (1.294) 0,196 [0.510] (2.921) 0,004** Lev [-0,010] (-0.059) 0,953 [0,675] (2.230) 0,026* [0.135] (0.816) 0,415 ROA [1,736] (3.302) 0,001** [2.953] (3.053) 0,002** [1.921] (3.604) 0,000** Loss [0,208] (1.941) 0.053 [-0,667] (-0.028) 0,977 [0.196] (1.805) 0,072 Big4 [0,372] (1.243) 0,214 [1.964] (3.603) 0,000** [0.558] (1.842) 0,066 IFGA [-0,154] (-1.994) 0,047* [-0.108] (-0.762) 0,446 [-0.122] (-1.566) 0,120 InPr [0,337] (6.749) 0,000** [0.810] (7.888) 0,000** [0.401] (7.924) 0,000** LAW Adjusted R2 [-0,332] (-3.936) 0,000** 0.617 [0.153] (0.992) 0,322 0.465 [-0.237] (-2.769) 0,006** 0.628 B coefficient in brackets. T-statistic in parentheses.

32 4.5 Sensitivity analysis

In addition to the previous regression analysis a sensitivity analysis is performed, where each variable of interest is run separately. The results in table 7, 8 and 9 are mostly similar to the results in table 6 from the previous section. However, in table 7 the negative relation between LogAf and IFGA is now significant at the 10% level instead of the previous 5% level.

In table 8 the positive relation between InPr and the audit, non-audit and total fees still exists, with no differences in the significance levels. Table 9 shows that there is also still a negative relation between LAW, LogAf and LogTotAf, but now significant at the 5% level instead of the previous 1% level.

Overall, the sensitivity analysis doesn’t show very big changes compared to the results in table 6.

Table 7

Audit fee regression results with variable of interest IFGA (N = 680)

Variables

Dependent variable

LogAf Dependent variable LogNonAf Dependent variable LogTotAf

(Constant) [9,971] (26.053) 0,000** [5.129] (7.091) 0.000** [9.745] (24.887) 0,000** Size [0,524] (27.052) 0,000** [0,671] (18.315) 0,000** [0.546] (27.562) 0,000** Inv [0,390] (1.578) 0,115 [-0,138] (-0.296) 0,767 [0.341] (1.348) 0,178 Rec [0,557] (3.173) 0,002** [0,684] (2.039) 0,042* [0.607] (3.342) 0,001** Lev [-0,040] (-0.237) 0,812 [0,400] (1.251) 0,211 [0.091] (0.528) 0,598 ROA [1,736] (3.223) 0,001** [3.028] (2.976) 0,003** [2.064] (3.747) 0,000** Loss [0,239] (2.158) 0.053 [-0,028] (-0.132) 0,895 [0.233] (2.057) 0,040*

33 Big4 [0,328] (1.061) 0,289 [1.874] (3.206) 0,001** [0.498] (1.574) 0,116 IFGA [-0,111] (-1.516) 0,083 [-0.097] (-0.662) 0,538 [-0.085] (-1.268) 0,215 Adjusted R2 _0.590 0.404 0.594 B coefficient in brackets. T-statistic in parentheses.

*,** significant at 5% and 1% respectively two tailed.

Table 8

Audit fee regression results with variable of interest InPr (N = 680) Variables Dependent variable LogAf Dependent variable LogNonAf Dependent variable LogTotAf (Constant) [9,666] (25.579) 0,000** [4.366] (6.262) 0.000** [9.354] (24.597) 0,000** Size [0,517] (27.425) 0,000** [0,635] (18.250) 0,000** [0.535] (28.161) 0,000** Inv [0,172] (0.715) 0,475 [-0,447] (-1.005) 0,315 [0.102] (0.419) 0,675 Rec [0,463] (2.657) 0,008** [0,.447] (1.389) 0,165 [0.486] (2.773) 0,006** Lev [-0,038] (-0.235) 0,815 [0,565] (1.868) 0,062 [0.120] (0.729) 0,466 ROA [1,350] (2.593) 0,010** [2.661] (2.769) 0,006** [1.670] (3.186) 0,002**

34 Loss [0,222] (2.049) 0.041* [-0.107] (-0.537) 0,592 [0.205] (1.881) 0,060 Big4 [0,410] (1.358) 0,175 [1.948] (3.496) 0,001** [0.581] (1.912) 0,059 InPr [0,265] (5.570) 0,000** [0.738] (8.405) 0,000** [0.352] (7.344) 0,000** Adjusted R2 _0.608 0.456 0.625 B coefficient in brackets. T-statistic in parentheses.

*,** significant at 5% and 1% respectively two tailed.

Table 9

Audit fee regression results with variable of interest LAW (N = 680) Variables Dependent variable LogAf Dependent variable LogNonAf Dependent variable LogTotAf (Constant) [10.078] (26.018) 0,000** [4.697] (6.447) 0.000** [9.781] (24.557) 0,000** Size [0,522] (27.052) 0,000** [0,674] (18.565) 0,000** [0.546] (27.578) 0,000** Inv [0,402] (1.643) 0,101 [-0,104] (-0.225) 0,822 [0.352] (1.403) 0,161 Rec [0,574] (3.236) 0,001** [0,615] (1.845) 0,066 [0.608] (3.343) 0,001** Lev [-0,030] (-0.180) 0,857 [0,402] (1.274) 0,203 [0.097] (0.564) 0,573 ROA [1,872] (3.548) 0,001** [2.645] (2.598) 0,010** [2.092] (3.770) 0,000**