Corporate risk profiles, female board members and their effect on audit fees and the quality of audit fee disclosure: Evidence from Dutch listed and unlisted companies

(1)

Corporate risk profiles, female board

members and their effect on audit fees and

the quality of audit fee disclosure: Evidence

from Dutch listed and unlisted companies

A.H.J.F. (Bram) Maarleveld

Master Thesis Accountancy & Controlling

Supervisors:

drs. W. Kevelam & prof. dr. R.L. ter Hoeven

Student number: S2291347

Date: 2019/01/21

(2)

2

Abstract

The main purpose of this paper is to determine the effect of gender diversity of the board and a corporate’s risk profile on the level of audit fees and the quality of audit fee disclosure. The sample of this research consists of the 75 Dutch companies with a listing on the Amsterdam Stock Exchange at the end of 2017 and the 75 largest unlisted companies in terms of the number of employees. All the data are hand selected from the annual reports of the years 2012-2017. The level of audit fees is positively related to gender diversity in the board (after controlling for tokenism) and listing status. The relation between a risk profile and audit fees is ambiguous. The model for audit fee disclosure show ambiguous results and have low explanatory power. The main finding in this model is the positive relation between listing status and the quality of audit fee disclosure.

Key words: Audit fees; Audit fee disclosure; Dutch audit market; Gender diversity; Risk profile; Listing status

(3)

3

1. Introduction

1.1 Introduction

In August 2015, Fortune published an article about the different audit fees paid by two telecom companies in the United States: Apple and Microsoft. Apple, with total assets of $231.8 billion and $181.8 billion in revenues paid Ernst & Young $10.6 million for the annual audit. Microsoft on the other hand, with total assets of $172.4 billion and revenues of $82.7 billion paid $46.2 million to its auditor Deloitte (Fortune, 2015). Another example is the different level of audit fees paid by AT&T and Verizon. Both companies have respectively $292.8 and $232.7 billion in total assets and $132.5 and $127.1 billion in revenues. AT&T paid $27.7 billion to its auditor, while Verizon’s annual audit cost $36.3 million. Remarkable in this situation is that both companies were once part of the Bell System (Verizon, 2016), are in the same business and share Ernst & Young as its auditor. What is the basis for diverging audit fees for similar companies in related businesses? In most cases, this question remains unanswered since companies disclose little about the drivers and the composition of audit fees and the perceived value of the annual audit decreases for investors and analysts (Gray, Turner, Corman & Mock, 2011).

The fees charged by auditors are discussed for a long time, both internationally and within the Netherlands. Frequently, auditors are accused of charging high fees while the perceived value of the auditor’s report decreases and audit quality is insufficient. In 2014, the Dutch supervisor of the Financial Markets, the Autoriteit Financiële Markten (AFM), published a report about the structural deficiencies in audit quality and required concrete improvements from the audit firms (AFM, 2014). In order to regain trust in the audit profession, a working group of The Dutch Institute of Chartered Accountants (Nederlandse Bedrijfsorganisatie van Accountants, or ‘NBA’) presented a plan with 53 points of improvement. However, in June 2017, after an evaluation of the implementation and embedding of the improvement program, the AFM concluded that the quality of statutory audits by Big 4 audit firms still is insufficient (AFM, 2017). This culminated in the announcement of an advisory committee for the auditing industry by the Dutch minister of finance, in what he called: “The moment approaches that regulation replaces self-regulation.”

The absence of quality improvements and the decrease in its perceived value questions the relevance of the annual audit. In a paper on auditing practices during the financial crisis, Sikka (2009) states that characteristics of the financial crisis were ineffective regulatory systems and corporate risk-taking through complex financial instruments. This resulted in the

(5)

5 financial distress of Dutch financial institutions like Fortis- and ING Bank. In addition, Sikka (2009) concludes that during the crisis, firms with risky balance sheets received unqualified audit opinions just before the public declaration of financial difficulties. This situation occurred while these organizations paid high audit- and non-audit fees to the external auditor.

In the wake of the this crisis, the interest in the role that corporate governance plays in effective monitoring and oversight increased. One of the possible explanations for the risk-taking behavior during the financial crisis is the limited proportion of women in Dutch corporate business, since women empirically evidenced are more risk-averse than their male counterparts (Byrnes, Miller & Schafer, 1999). A stream of literature determined this relation by associating corporate governance variables to audit fees (Gul, Srinidhi & Tsui, 2008). The under-representation of women in Dutch business is an ongoing issue, since at the end of 2017 only 7% of the Dutch listed companies met the quota of 30% female representation in both the executive and non-executive board (Lückerath-Rovers, 2017). If an organization does not comply with this quota, it has to disclose the reason for this non-compliance in the annual report and needs to explain which appropriate measures are taken to achieve the 30% target in the future. Since this regulation did not have the desired effect, different parties argue for a legal quota in which companies are obliged to have a determined percentage of female representation in both the executive- and non-executive board.

1.2 Research question

The main purpose of this study is to find empirical evidence on the determinants of audit fees and the quality of audit fee disclosure. The implementation of EU-regulation regarding the mandatory audit firm rotation increased the public interest in auditor remuneration. Regulation No 537/2014 requires organizations of public interest to rotate audit firms after an audit engagement of at most ten consecutive years. After this rotation, a cooling-off period of four years is required. By the adoption of this regulation in 2016, the European Parliament and the European Council intend to improve the audit quality of organizations of public interest. The Dutch regulator already introduced legislation in 2012, which requires audit firm rotation for Dutch public-interest entities every eight years. Eventually, the Dutch regulator adopted the EU-regulation. As a result of the Dutch and European regulation, a large amount of Dutch PIEs changed auditors between 2014 and 2016.

The large number of audit firm rotations between 2014 and 2016 add new perspectives to the current stream of scientific research in the field of auditor remuneration. Starting point

(6)

6 in most audit fee research is the original seminal paper by Simunic (1980), the first study in the field of audit fee research that examines different determinants of audit fees. In this paper regarding the then ‘Big 8’ accounting firms, Simunic (1980) found that client size, risk and -complexity are positively related to audit fees. Much research followed the work of Simunic (1980), investigating both client- and auditor-related variables. In a meta-analysis, Hay, Knechel & Wong (2006) evaluated the different studies that examine the determinants of audit fees and came with five topics for future research. In the updated meta-analysis, Hay (2013) concluded that the three determinants estimated by Simunic (1980) still are the most used variables in audit fee research.

Due to its partly voluntary nature, the amount of studies related to the quality of audit fee disclosure is however limited. Most Dutch companies disclose the necessary numerical information, but the annual reports often lack a detailed explanation of the composition and development of audit fees (Kevelam, Ter Hoeven & Brouwer, 2017). In 2002, the Dutch audit profession adopted a recommendation that audit fees should be made public, in order to guarantee auditor independence. However, due to the lack of obligation, this recommendation did not have a real impact in practice. EU-directive 2006/43/EG states that providing additional non-assurance services by the statutory auditor can pose a threat to its independence. According to this directive, the disclosure of audit fees can mitigate or remove this threat. It requires organizations to disclose the charges paid for the annual audit, other audit engagements, tax advisory and non-audit services. As a result from this directive, the Dutch legislator introduced regulation on audit fee disclosure. Since 2008, with the implementation of article 382a, BW 2 of the Dutch Civil Code, Dutch PIEs and large entities that meet certain size related requirements are obliged to incorporate disclosure on audit fees in the annual report. This article requires disclosure of the total amount of audit fees charged by the audit firm and the audit firm network, regarding the audit- and advisory engagements in the relevant fiscal year.

In this study, we determine the effect of female representation in corporate boards on audit fee and the quality of audit fee disclosure. Since this representation empirically evidenced has its effect on the firm’s riskiness, we also attempt to investigate the relation between firms with a higher risk profile and the audit fee. Besides that, we attempt to find out what the influence is of gender diversity of the board and the firm’s riskiness on the quality of audit fee disclosure. Therefore, the research question of this study is:

What is the effect of female board representation and the corporate’s risk profile on the audit fee and the quality of audit fee disclosure of Dutch listed and unlisted auditees?

(7)

7

1.3 Academic contribution and practical relevance

As mentioned before, the literature on the quality of audit fee disclosure is relatively limited. Besides the academic contribution, the relevance of determining the effect of gender diversity and a corporate risk profile on the quality of audit fee disclosure serves a practical use as well. The purpose of regulation on audit fee disclosure is to inform users of financial statements about the independence of the external auditor (Langendijk, 2011). The ratio between the level of fees paid for audit and non-audit services is a good proxy for the perceived independence of an auditor (DeFond, Raghunandan & Subramanyam, 2002). Francis & Ke (2006) found that auditor independence increased after the introduction of the SEC’s mandatory fee disclosure regulation, meaning that an increase in disclosure leads to an increase in auditor independence. In addition, Francis (2011) found that audit quality increases when auditors are more independent. Additional to the scientific contribution, these findings demonstrate the practical relevance of estimating the determinants that affect the quality of audit fee disclosure, since higher disclosure quality could lead to an increase in auditor independence and consequently to an increase in audit quality. The literature on audit fees is comprehensive. In most audit fee-related studies, the audit fee is used as a proxy for audit- or accounting quality. Hribar, Kravet & Wilson (2014) found empirical evidence that audit fees are positively related to other measures of accounting quality. The main implication of this research is that the level of audit fees affects audit quality. This result confirms the academic and practical relevance of estimating the determinants of audit fees.

In addition to the general relevance of research on audit fees and audit fee disclosure, this study contributes to the current state of literature in different ways. First, we estimate the moderating effect of gender diversity of the board on the relation between a corporate’s risk profile and the level of audit fees and the quality of the audit fee disclosure. If this relation is significant, it determines that female presence in the boardroom increases the demand for higher auditing effort and consequently, higher audit quality. In addition, this is a potential argument in favor of a statutory female quota in Dutch corporate business, since the level of audit fees and the quality of the audit fee disclosure serve as proxies for respectively audit quality and auditor independence (Abbott, Parker, Peters & Raghunandan, 2003; Carcello, Hermanson, Neal & Riley, 2002). Second, this study focuses on the Dutch market and extends current research in the field of auditor remuneration, which is predominantly Anglo-Saxon based. The results demonstrate whether the same indications apply to the Dutch audit market and whether the Dutch audit market is comparable to the Anglo-Saxon ones. Third, in this paper we add the 75 largest unlisted Dutch firms. This extends the current stream of literature, which is mainly

(8)

8 focused on listed firms only. The results demonstrate whether the same conclusions can be made on listed and unlisted companies, since they face different legislation regarding the annual audit and audit fee disclosure. Last, the fiscal year 2017 is added to the sample as a new firm-year in order to establish whether prior research is affected by the wave of audit firm rotations between 2014 and 2016.

1.4 Methods & results

Our sample consists of both listed and unlisted companies. The 75 listed observations consist of the companies that are listed at The Amsterdam Stock Exchange at the end of 2017. The unlisted observations consist of the 75 largest non-listed companies in terms of the number of employees. The main result on audit fees is that the size of an auditee is the single most important determinant in explaining the level of audit fees. Using pooled OLS models and the likelihood ratio test we find that listing status, inherent risk and the ratio of receivables and inventories to total assets are positively associated with the level of audit fees. The leverage ratio shows a significant negative relation with audit fees. In addition, after controlling for tokenism, we find positive and statistically significant results on the relation between gender diversity in the board and audit fees. For the quality of audit fee disclosure, the results are ambiguous and have a low adjusted R-square. The complex nature of the determinants that explain the quality of audit fee disclosure forms the main opportunity for future research.

1.5 Structure of the paper

The remainder of this paper structures as follows. The following section describes the theoretical framework regarding audit fee and the quality of audit fee disclosure. In section 3, we develop hypotheses that are based on the theoretical framework and empirical findings. Section 4 describes the sample size, followed by the methodology of this research. In section 5, we present the results of our research. The concluding section provides a discussion, implications and limitations of this study and ends with opportunities for future research.

(9)

9

2. Theoretical framework

In this section, we present the theoretical foundations on which the hypotheses in this study are based. The first section contains the agency theory, which is predominantly used in audit fee-related research. The second section presents the legitimacy theory, concerning the relation between organizations and society. In the last section, we describe the voluntary disclosure theory.

2.1 Agency theory

Agency theory primarily deals with the principal-agent relationship, existing because of the separation of ownership and control. The principal-agent relation is defined by Jensen & Meckling (1976) as a contract under which one or more persons (principals) engage another person (the agent) to perform some service on their behalf which involves delegating some decision-making authority to the agent. The shareholder-manager relationship is a well-known example within agency theory, in which the shareholder is the principal and the manager the agent. The theory is based on two assumptions. First, both the principal and the agent are utility maximizers who tend to maximize their returns by all means. Second, the interests of the principal and the agent are not aligned, which results in opportunistic behavior driven by self-interest (Yi, Howard & Eggleton, 2011). The resulting conflict is called the agency problem.

A characteristic of most business settings is that one party has an information advantage over another party, the so-called information asymmetry (Subramaniam, 2006). This information advantage arises because managers (agents) are more executively involved in the daily business than shareholders (principals) are. The annual audit, particularly through the assurance that is given by an external auditor, reduces information asymmetry that the agent has over the principal on financial statements and mitigates managerial opportunism in financial reporting (Piot, 2001). In this respect, audit fees can be seen as the agent’s monitoring costs (Verbruggen, Christiaens, Reheul & Van Caneghem, 2015). Besides that, the disclosure of the audit fee reduces information asymmetry and therefore reduces agency costs (Jensen & Meckling, 1976).

2.2 Legitimacy theory

Legitimacy theory is a systems-oriented theory in which an organization is part of a broader social system and is assumed to be influenced by and have influence upon the society in which it operates (Gray, Owen & Adams, 1996). The theory concerns the relationship

(10)

10 between a firm and society. Therefore it extends stakeholder theory which principally focuses on the relation with stakeholders, in contrast to legitimacy theory which focuses on society as a whole. In the social system, firms have no inherent right to resources or even have no inherent right to exist; they exist to the extent that society considers an organization as legitimate (Deegan, 2002). Legitimacy is determined by the requirements of a ‘social contract’ and organizations should comply with the terms of the social contract while conducting their business. Survival depends on the extent that a company is capable of meeting those terms. (Yi, Howard & Eggleton, 2011). According to Deegan (2006), legitimacy is not a static concept; organizations continually have to adapt their activities to the changing requirements of the social contract.

Since the social contract is a theoretical construct, its terms cannot be determined with precision. Gray et al. (1996) state that legal requirements provide explicit terms of the contract, while societal and political expectations form the implicit terms. The annual audit serves both the implicit and explicit terms. The explicit terms because the Dutch Civil Code (article 396, BW 2) requires an external audit and disclosure audit fees of PIEs and organizations that meet specific requirements. The implicit terms since society expects an unqualified audit report issued by an external auditor, to obtain assurance on the annual accounts. When legitimacy is into question, managers can adopt different strategies which are relying on the use of external reporting (Dowling & Pfeffer, 1975). Through corporate disclosure, an organization can educate society that its actual behavior corresponds with the requirements of the social contract or it can change the perception of legitimacy while not changing its actual behavior. Audit fee disclosure can be used to inform society that organizations are behaving within the requirements of the social contract in which an independent auditor provides assurance on the financial statements.

2.3 Voluntary disclosure theory

In reporting, there is a reporting threshold for which information is disclosed and which is not. Voluntary disclosure theory states that ‘good’ news is disclosed while ‘bad’ news is concealed (Chu, Chatterjee & Brown, 2012). The theory initially referred to the voluntary disclosure of financial information, but in current research the theory is used to explain non-financial disclosure as well. This stream of research argues that organizations disclose superior non-financial performance voluntarily in order to reveal the nature of its actual performance and to (potentially) increase market value (Hummel & Schlick, 2016).

(11)

11 If the market for accounting regulations and auditing would work efficient, managers’ accounting decisions and disclosures communicate changes in the firm’s performance to outside principals. Since this is not the case, managers make tradeoffs betweenreporting private information on the firm’s performance to principals and conceal other information for different business reasons (Healy & Palepu, 2001). Managers have private information regarding the independence of the external auditor. Since regulation on disclosure of audit fees only applies to certain Dutch entities (Langendijk, 2011), organizations have incentives to voluntarily disclose additional information on the composition and development of the fees paid to its external auditor. The credibility of corporate disclosures is enhanced by regulators, standard setters, auditors and other capital market intermediaries. Healy & Palepu (2001) state that third-party assurance on the quality of management disclosure is a mechanism for increasing the credibility of voluntary disclosure.

(12)

12

3. Empirical findings & Hypotheses development

In this section, we present the empirical findings regarding audit fees and audit fee disclosure. We hypothesize the risk of an auditee, the number of listings and the gender diversity of the board on audit fees and the quality of the audit fee disclosure. Based on the empirical findings and the theoretical framework in the previous section, we develop the hypotheses.

3.1 Hypotheses development on audit fees

The relation between an entity’s risk profile and the level of audit fees is frequently determined in previous literature. In a factor analysis on the nature of risk regarding audit fees, Jubb, Houghton & Butterworth (1996) explain the two types of risk in audit engagements that are relevant to an auditor. Business risk is defined as the probability that an auditor suffers loss or injury in his professional practice. The audit risk, on the other hand, is the likelihood that an auditor renders an inappropriate opinion on an auditee’s financial statements. Jubb et al. (1996) state that in most studies, empirical evidence is found for a positive relation between some concept of risk and the level of audit fees. According to legitimacy theory, society expects a broader audit effort for risky auditees. In addition, in line with agency theory, an auditor compensates for both types of risk in order to reduce inherent risk to an acceptable level. This risk compensation results in additional audit effort by the external auditor and consequently results in higher audit fees. Bell, Doogar & Solomon (2008) found empirical evidence for the adoption of a business risk audit (BRA) approach in which an auditor responds to the increased complexity of financial statements of risky auditees. Bell, Landsman & Shackelford (2001) found that an increase in business risk results in higher audit fees due to the additional audit hours spent in order to compensate for the business risk. Last, in a study regarding Dutch audit engagements, Blokdijk et al. (2003) found lower levels of planning materiality for auditees with an increased risk of earnings management. Planning materiality is an important audit judgment, since its value is inversely related to audit effort, and consequently with audit fees. Based on theory and previous empirical findings, we expect a positive relation between the auditee’s risk and the audit fees. Therefore the hypothesis is:

H1: The risk of an auditee is positively associated with higher audit fees.

Going public is an important strategic decision for organizations to raise capital. It is typically followed by increased financial reporting and disclosure regulation and increased

(13)

13 supervision by the AFM. This increased supervision is based on the Financial Supervision Act (Wet op het financieel toezicht or Wft) and the Financial Reporting Supervision Act (Wet toezicht financiële verslaggeving or Wtfv). For an auditor, it is also relevant whether an organization has a listing or not, since public audit engagements are associated with greater exposure to risk than non-public ones (Palmrose, 1986). The increase in audit risk causes a greater auditing effort by the auditor. In the Dutch context, prior research found an audit fee premium for listed companies (Langendijk, 1997; Langendijk & Groenen, 2004). This is in line with agency theory in which an auditor could necessitate additional auditing effort as a risk premium for expected litigation costs. According to legitimacy theory, additional audit procedures are required to comply with the explicit terms of the social contract, since being listed entails additional legislation. Multiple studies found an audit fee premium for organizations that are cross-listed. These studies predominantly found higher audit fees for cross-listings in countries with stronger legal regimes than the regime of their home country (Choi, Kim, Liu & Simunic, 2009; Bronson, Ghosh & Hogan, 2017; Carson & Fargher, 2007; Seetharaman, Gul & Lynn, 2002). Wingate (1997) developed the Wingate Litigation Index, an index that determines the strength of legal regimes and the risk of doing business as an auditor in a particular country. Common law countries like the US and the UK receive high scores in this index. Since most multiple listed organizations in our sample are listed in the Netherlands and a common law country, we expect a positive relation between the number of listings and audit fees. Therefore, we formulate the hypothesis for listed and cross-listed companies regarding audit fees as follows:

H2: The number of listings of an auditee is positively associated with higher audit fees.

The stream of literature regarding the structure and effectiveness of corporate boards and its subcommittees is comprehensively. Essential factors for board effectiveness are characteristics like independence, expertise and experience. However, female representation in the board and its committees currently obtains more attention. Compared to boards consisting of men only, women bring different perspectives and improve informed decision making (Rose, 2007). In this way, in line with agency theory, transparency increases which consequently reduces information asymmetry at the board level (Srinidhi, Gul & Tsui, 2011). Two commonly used, contradicting approaches relating board effectiveness to audit fees are based on the demand and supply side. According to the demand side approach, audit fees are seen as a proxy for audit quality; it hypothesizes that effective governance requires more audit effort from the external auditor, therefore increasing audit fees. The supply side, on the other hand, links audit fees to

(14)

14 auditor risk; it argues that organizations with a strong control environment decrease the auditor’s assessed level of control risk, and consequently reduces audit effort (Simunic, 1980). Studies regarding gender-related governance characteristics argue that female directors are more risk-averse, are more diligent and require a higher level of monitoring compared to their male counterparts (Harjoto, Laksama & Lee, 2015). Most empirical findings are in line with the demand side approach; these findings emphasize that gender diversity increases the demand for higher quality audits in terms of higher audit effort. Higher audit fees are found for organizations with a female CEO (Huang, Huang & Lee, 2014; Harjoto et al. 2015) and with higher female representation in the audit committee (Aldamen, Hollindale & Ziegelmayer, 2018; Ittonen, Miettinen & Vähämaa, 2010). Lastly, audit fee premiums are found for organizations with a higher proportion of female directors (Lai, Srinidhi, Gul & Tsui, 2017; Gul et al., 2008). Concerning corporate risk and gender diversity of the board in relation to audit fees, Aldamen et al. (2018) found higher audit fees for high-risk organizations with gender diverse audit committees. In line with the demand side approach and the findings of Aldamen et al. (2018), we expect higher monitoring and auditing requirements for high-risk auditees with gender diverse boards. Therefore, we hypothesize that gender diverse boards strengthen the association between higher-risk auditees and audit fees. Based on theory and previous findings, we formulate the hypotheses as follows:

H3: The proportion of women in the board of an auditee is positively associated with higher audit fees.

H4: The proportion of women in the board of an auditee strengthens the positive association between the risk of an auditee and audit fees.

H5: The proportion of women in the board of an auditee strengthens the positive association between the number of listings of an auditee and audit fees.

3.2 Hypotheses development on the quality of audit fee disclosure

As mentioned in the section of the research question, the research regarding the quality of audit fee disclosure is relatively limited. This gives reason to fill this gap in the current state of literature, in order to improve our understanding of the determinants of the quality of audit fee disclosure. To overcome the problem of the lack of available literature, we also use the literature on other types of disclosure in the development of the hypotheses. We predominantly use the literature on voluntary disclosure, since the disclosure of audit fees is only mandatory for PIEs and large entities. In addition, the same motives apply for both voluntary disclosure

(15)

15 and audit fee disclosure according to agency theory and voluntary disclosure theory. Both theories use disclosure to overcome the problem of information asymmetry. The nature of disclosure and the theoretical motives confirm that the comparison of these types of disclosure is a valid one. We research the quality of audit fee disclosure for the risk of an auditee, whether an auditee has a (cross)listing and for the gender diversity of the board.

In contrast to the relation between auditee’s risk and audit fees, the relation between an organization’s risk profile and voluntary disclosure is less determined in previous literature. The relation between the level of risk and disclosure levels can be hypothesized in a positive and in a negative direction. The positive direction is in line with agency theory, legitimacy theory and voluntary disclosure theory. These theories posit that higher-risk organizations disclose information in order to overcome the increased information asymmetry and to provide information to explain the extent to which an organization is capable of managing the risks. In addition, Jensen & Meckling (1976) argue that higher-risk organizations in terms of leverage have higher monitoring costs. These organizations disclose more information in their annual report in order to reduce these costs. The negative direction hypothesizes decreased disclosure since higher-risk organizations may not want to attract market attention (Elshandidy, Fraser & Hussainey, 2013). In line with the theoretical foundation, the empirical results on the relation between organizational risk and voluntary disclosure are varying as well. Different studies found higher disclosure levels for organizations with higher leverage ratios (Michelon & Parbonetti, 2012; Elshandidy et al. 2013), while other studies did not find a relation (Chow & Wong-Boren, 1987; Raffournier, 1995) or even found a negative relation (Meek, Roberts & Gray, 1995). Abraham & Cox (2007) and Marshall & Weetman (2007) found a positive relation between risk disclosures and organizational risk, respectively using the variance of stock returns and liquidity risk as proxies for organizational risk. In a meta-analysis on 2,473 corporate annual reports, Ahmed & Courtis (1999) found an overall significant positive relation between leverage and disclosure levels. These results give rise to formulate our hypothesis as follows: H6: The risk of an auditee is positively associated with the quality of audit fee disclosure.

As mentioned in the hypothesis development of listing status in relation to the level of audit fees, organizations whose shares are listed face additional disclosure regulation, under supervision of the AFM. Besides that, stakeholders of listed companies have diverse interests and consequently exert different pressures on companies to disclose information they consider necessary. Meek et al. (1995) argue that organizations that are cross-listed face additional

(16)

16 capital market pressure to disclose information, compared to domestic listed firms. Companies that give in to these pressures act in line with legitimacy theory, demonstrating that their operations are in line with the expectations of such pressure groups (Haniffa & Cooke, 2005). Voluntary disclosure theory suggests that in disclosure choices, it can be assumed that the expected proceeds are perceived to exceed the costs. In relation with listing status, organizations use disclosure to raise capital at the lowest possible costs. The empirical results on disclosure levels for (cross)listed companies are unambiguous. The findings state that disclosure levels increase when a company is listed and even increase more when a company is cross-listed. Cooke (1989, 1992) found significant relations between disclosure levels and listing status, for companies that are respectively listed on the Stockholm and Tokyo Stock Exchange. Meek et al. (1995) found higher strategic information disclosure for companies that are listed internationally, especially for European multinational companies. Lastly, in a meta-analysis, Ahmed & Courtis (1999) concluded that listing status has a significant positive relation with voluntary disclosure. Based on the empirical findings and the theoretical foundation, we formulate the hypothesis as follows:

H7: The number of listings of an auditee is positively associated with the quality of audit fee disclosure.

Empirical literature is restricted to analyzing the publicly available determinants that reflect the outcomes of corporate operations. The role of corporate governance on social reporting and voluntary disclosure is well recognized. Empirical findings indicate that female representation in the board is positively associated with corporate social reporting (Ibrahim & Angelidis, 1994) and higher levels of social performance (Siciliano, 1996). Effective corporate governance and -disclosure decisions can be seen as complementary mechanisms to obtain or maintain organizational legitimacy (Michelon & Parbonetti, 2012). Disclosure decisions form a strategy that organizations can use to meet the changing requirements of society (Dowling & Pfeffer, 1975). According to agency theory, the appointment of female directors in the board is an effective mechanism in restricting opportunistic behavior of board members and encourages organizations to disclose a wider range of information which increases transparency and reduces information asymmetry (Tejedo-Romero, Rodrigues & Craig, 2017). Two major factors explain the increased voluntary disclosure tendency for gender diverse boards. First, women have a more “trust-building leadership style” than man (Trinidad & Normore, 2005). Trust-building requires more information exchange and therefore, mitigates information asymmetry. Second, women more than men attach value to a monitoring environment which

(17)

17 enhances transparency and a richer information environment (Srinidhi et al. 2011). Multiple studies confirm the relation between gender diversity and disclosure levels. A higher proportion of women in the board is positively associated with the extent of voluntary environmental disclosure (Amar, Chang & McIlkenny 2017; Liao, Luo & Tang, 2014), corporate social disclosure (Barako & Brown, 2008) and corporate governance disclosure (Elmagrhi, Ntim & Wang, 2016). In some studies, no evidence is found for an association between gender diversity and disclosure levels. This often is explained by a process called occupational socialization which hypothesizes that gender differences tend to disappear in the work environment. This has its influence on both male and female directors and therefore differences between them dilute. Smith & Rogers (2000) further researched this process and concluded that no gender differences appeared when specific quotable rules are tested. However, in ambiguous situations with limited specific regulation, Smith & Rogers (2000) found that women tend to act more ethical than their male counterparts. Since audit fee disclosure in the annual report is not entirely regulated, we hypothesize a positive association between gender diversity in the board and the quality of audit fee disclosure. In addition, since women are more risk-averse and attach more value to an information environment and monitoring effectiveness, we expect an increase in the quality of audit fee disclosure for higher-risk auditees. Therefore, we formulate the hypotheses as follows:

H8: The proportion of women in the board of an auditee is positively associated with the quality of audit fee disclosure.

H9: The proportion of women in the board of an auditee strengthens the positive association between the risk of an auditee and the quality of audit fee disclosure.

H10: The proportion of women in the board of an auditee strengthens the positive association between the number of listings of an auditee and the quality of audit fee disclosure.

(18)

18

4. Research method

In this section, we explain the research method for this study. First, we explain our sample selection and how the data for this research are gathered, followed by a further look into our dependent variables, namely the audit fee and the quality of audit fee disclosure. Subsequently, we take a deeper look into the independent variables and control variables.

4.1 Sample selection and data collection

This study focuses on Dutch entities, both listed and unlisted. The listed companies are selected from the stock market index of Euronext Amsterdam, the Amsterdam Stock Exchange. It concerns the data of large (AeX), Mid cap (AmX) and Small cap (AScX) companies that are listed at the end of 2016. For certain fiscal years, not all the required data are available due to the annual composition change of the index, initial public offerings or the delisting of organizations. The remaining companies are the 75 largest unlisted companies based on the number of employees. First, we selected all Dutch companies with more than 500 employees from company.info; this results in 2,152 entities. We removed 102 companies because of their listing status and 1,043 companies because they did not lodge their financial statements of 2017 already. Out of the 1,007 unlisted entities that remained, the 75 largest companies in terms of the number of employees are selected. We replaced the unlisted companies that apply the exemption to disclose audit fees (article 382a part 3, BW 2 of the Dutch Civil Code). A list of the selected companies is included in appendix A. We hand collected all data used in this study from annual reports of the fiscal years 2012-2017. For some fiscal years, the required information is not available; these observations are not included in this paper. Therefore our sample consists of 418 listed and 341 unlisted observations across 150 entities. The following sections set out the extracted data from the annual reports that are relevant in this study.

4.2 Dependent variables: audit fees and quality of audit fee disclosure

This research consists of two dependent variables, namely audit fees and the quality of the audit fee disclosure. Like most audit fee-related research, we measure the audit fee by the level of the audit fees (AUF) charged by a company’s auditor (Collier & Gregory 1996; Ittonen & Peni 2012; Hardies, Breesch & Branson 2015). Since we use audit fee as a proxy for audit quality, we use the sum of the audit fees and other audit engagements, since auditors use the outcomes of those engagements for the annual audit as well. The quality of the audit fee disclosure is complicated to measure objectively. An audit fee disclosure index with

(19)

19 unweighted disclosure items is developed to measure this quality. The advantage of using an unweighted disclosure index is the reduction of subjectivity in assigning equal weight to the disclosure items (Barako & Brown, 2008). We drew the list of disclosure items from article 390.3 from the Dutch Guidelines of Annual Reporting (Richtlijnen Jaarverslaggeving). To estimate the audit fee disclosure score, we use a binary coding technique. This technique is consistent with the approach of prior studies that use a disclosure index (Haniffa & Cooke, 2005; Cooke, 1992). The index consists of twelve equally weighted disclosure items. The number of points obtained is divided by the highest possible score; this results in an audit fee disclosure quality (AFD) between 0% - 100%. The list of disclosure items that determines the quality of the audit fee disclosure is provided in appendix B.

4.3 Independent variables

Both the models with the level of audit fees and the quality of audit fee disclosure have three independent variables, namely, board diversity, corporate risk and listing. In this section, we describe the different proxies that are used to measure the independent and control variables. A summary of the variables and its measurements is tabulated in table 1:

Table 1 Summary of variables

Variables Name Measurement

Audit fee AUF The sum of the audit fee for the annual audit

and other audit engagements

Quality of audit fee disclosure AFD Audit fee disclosure index between 0% -

100%

Gender diversity in Board of Directors FBD Proportion of women in the board of directors

Gender diversity in combined boards FBT Proportion of women in both boards combined

Inherent risk INH

Dummy variable equal to 1 if company is in the financial institutions or utilities sector,

0 otherwise

Leverage LEV Total debt divided by total assets

Inventories and receivables to total assets InvRec Sum of the inventories and trade receivables,

divided by total assets

Listing LIST

Dummy variable equal to 0 if company is unlisted, 1 if the company is listed on the Amsterdam Stock Exchange, 2 if the company

is multiple listed

(20)

20 4.3.1 Board diversity

We use two proxies to measure the gender diversity of the board. Since a potential statutory quota in Dutch business concerns both the board of directors and the supervisory board, both these determinants are incorporated in our model. The proxies we use are the female proportion in the board of directors (FBD) and the female proportion in the board of directors and supervisory board combined (FBT) (Amar et al. 2017; Hummel & Schlick, 2016; Barako & Brown, 2008).

4.3.2 Risk

Many proxies for risk are used in studies on auditor remuneration. Jubb et al. (1996) explained the different perceptions of risk in audit fee-related research, since they argued that the audit risk concept was ill-defined in the current state of literature. For robust results, we use three different proxies to define risk in this study. Inherent risk (INH) is a dummy variable that equals 1 if a company is in the financial institutions or utilities sector, since Hay, Knechel & Wong (2006) stated that those industries are most frequently assigned as risky in audit fee research. Leverage (LEV) is measured by a firm’s total debt divided by its total assets (Aldamen, 2018; Bell, Landsman & Shackelford, 2001). The third proxy (InvRec) is the sum of the inventories and receivables, divided by total assets. The study of Simunic (1980) was the first that identified inventories and receivables as risky balance sheet items that require additional audit procedures, since the valuation of both assets requires a forecast of future assumptions.

4.3.3 Listing

We hypothesize that the listing status of a company has its effect on both the amount of audit fees and the quality of audit fee disclosure. The listing variable measures whether an organization is unlisted, listed on the Amsterdam Stock exchange or has a cross-listing. The listing status is measured by a dummy variable (LIST) which equals 0 if it is not listed, equals 1 if it has a listing on the Amsterdam Stock Exchange and equals 2 if the company is listed on multiple stock exchanges.

4.4 Control variables

Control variables are added in order to reduce the influence of variables that are not part of the primary research. This paper controls for the relation between the size of an auditee and the dependent variables, which is determined frequently in prior audit fee research. In most studies related to audit fees, the size of the company is added as a control variable, because of its high explanatory value on audit fees. The size of an auditee is positively associated with

(21)

21 audit fees since larger auditees require more audit effort. This relation is determined in both the international (Simunic, 1980; Carson & Fargher, 2007) and the Dutch context (Langendijk, 1995). Besides that, large companies have better resources and face more pressure from different stakeholder groups to disclose information (Michelon & Parbonetti, 2012). The size of an auditee is measured (SIA) by a firm’s total assets (Abraham & Cox, 2007; Linsley & Shrives, 2006). In order to ensure that the results are not affected by potential outliers, we first winsorize the total assets at 1% levels.

Since societal pressure for transparency on audit- and non-audit fees increased over the years, organizations tend to disclose more comprehensively in their annual report. In addition, the introduction of new regulation on audit firm rotation affects the level of audit fees between 2014 and 2016. To ensure that the year of observation does not affect the regression results, control variables are included for the year of observation. The dummy for 2012 is not included in the regression, as it serves as the reference year.

4.5 Research model

To test the hypotheses presented in the previous section, we use univariate, bivariate and multivariate analyses. In the univariate analysis, we obtain general information of the variables like means, standard deviations and medians. Besides that, we analyze the development of audit fees and the quality of audit fee disclosure over the years, in order to capture an accurate view of the evolving audit market. For the bivariate analysis, we use a Pearson correlation matrix in order to examine whether potential multicollinearity problems are affecting the results.

We use pooled OLS models in our regression on audit fees and the quality of audit fee disclosure. To estimate the moderating effect of gender diversity in the boardroom, we conduct the likelihood ratio test. The regression models that determine the effect of the independent variables on the level of audit fees and the quality of audit fee disclosure are as follows:

𝐴𝑈𝐹_𝑖,𝑡 = 𝛽₀+ 𝛽₁𝐹𝐵𝐷_𝑖,𝑡+ 𝛽₂𝐹𝐵𝑇_𝑖,𝑡+ 𝛽₃𝐼𝑁𝐻_𝑖,𝑡 + 𝛽₄𝐿𝐸𝑉_𝑖,𝑡+ 𝛽₅𝐼𝑛𝑣𝑅𝑒𝑐_𝑖,𝑡 + 𝛽₆𝐿𝐼𝑆𝑇_𝑖,𝑡+ 𝛽₇ (𝑆𝐼𝐴)_𝑖,𝑡+ 𝜀_𝑖,𝑡

𝐴𝐹𝐷_𝑖,𝑡 = 𝛽₀+ 𝛽₁𝐹𝐵𝐷_𝑖,𝑡+ 𝛽₂𝐹𝐵𝑇_𝑖,𝑡+ 𝛽₃𝐼𝑁𝐻_𝑖,𝑡 + 𝛽₄𝐿𝐸𝑉_𝑖,𝑡 + 𝛽₅𝐼𝑛𝑣𝑅𝑒𝑐_𝑖,𝑡 +

(22)

22

5. Results

In this section, we present the results of the univariate, bivariate and multivariate analyses. First, we analyze the descriptive statistics for all the observations. Second, we present the Pearson correlation matrix in order to determine whether potential multicollinearity problems affect the results. In the multivariate analysis, we present the regression results of the audit fee and audit fee disclosure models. Last, we conduct multiple robustness tests in order to determine whether the results are robust and consistent.

5.1 Descriptive statistics

In table 2 the descriptive statistics are presented separately for listed and unlisted observations. The data are hand collected from annual reports of the years 2012-2017 and less unlisted observations are included due to the unavailability of some required data. This results in 418 listed and 341 unlisted company observations. In figure 1, we present the development of the audit fees and the quality of audit fee disclosure throughout the years. Except for the year 2015, the development of audit fees over the years is similar for listed and unlisted companies. Due to the regulation on auditor rotation, the table shows a decrease in audit fee in 2016 and for listed companies in 2015 as well. The large difference in average audit fees between listed and unlisted companies confirms the audit fee premium for listed companies. The average quality of the audit fee disclosure is relatively low with 68.8% and 56.9% for listed and unlisted companies respectively. These results suggest higher quality of audit fee disclosure for listed companies. The low disclosure quality is mainly due to the fact that certain disclosure items are frequently missing. These are the method of audit fee allocation, whether subsidiaries are included, the fees for consolidated subsidiaries if these are a material part of the total audit fees and last, a qualitative explanation. Figure 1 shows a stable increase in audit fee disclosure quality for listed companies. Unlisted companies increasingly understand the relevance of disclosure on audit fees, since a substantial increase can be seen during 2015 and 2016.

Figure 2 presents the development of the average ratio of tax advisory fees and non-audit service fees to the total amount of external non-auditor fees. From 1 January 2013 (with a transition period of two years), article 24b of the Audit Firms Supervision (Wet toezicht accountantsorganisaties or Wta) introduces a prohibition for audit firms to provide both audit- and advisory services to Dutch PIEs. Since this regulation is applicable for entities of public interest, figure 2 presents lower ratios for listed companies. The regulation on the separation of audit and advisory services did on average not affect the ratio for unlisted companies. For these

(23)

23

Figure 1: Development of audit fees and the quality of audit fee disclosure. Audit fees for listed companies are presented on the left y-axis, for unlisted companies on the right y-axis.

companies, the ratio develops up and down between 16% and 21% between 2012 and 2017. For listed companies, the effects of Article 24b Wta and the expiration of the transition period are visible in figure 2. Since the introduction in 2013, the ratio of advisory- to audit fees consistently decreased, from 12.10% in 2013 to 3.82% in 2017. After minor decreases in 2013 and 2014, figure 2 shows the highest decrease after the transition period at the end of 2014. This suggests that PIEs and audit firms postponed the separation of advisory- and audit services to the end of the transition period. At the end of 2017, there are still some audit firms that provide both audit and non-audit services to public-interest entities.

In appendix C, snapshots of audit fee disclosures of a listed and an unlisted company are included. We drew these audit fee disclosures from the annual reports of Nutreco N.V. 2016 and Stern Groep N.V. 2015 (AScX). The audit fee disclosure quality of Nutreco is well above average, with a score of 91%. The missing disclosure element is the qualitative explanation. Overall, Nutreco’s audit fee disclosure gives a thorough representation of the composition and development of its audit fees. The quality of the audit fee disclosure of Stern Groep N.V. is, with a disclosure score of 36%, well below the average of listed companies. The only elements that are disclosed by Stern Groep are the fees for the audit of the financial statements and the method of audit fee allocation. These examples and its varying disclosure scores demonstrate that in Dutch corporate business there is still no common practice in the disclosure of audit fees

(24)

24 Table 2 Descriptive Statistics Listed Unlisted

Variable n Mean Median Minimum Maximum Std. Dev. n Mean Median Minimum Maximum Std. Dev. AUF ('000) 418 4,600.8 1527.0 49.000 53,000 8,211.9 341 1,601.6 833,00 13,000 17279 2,271.9 AFD 418 0.6881 0.6667 0.2500 1.0000 0.1943 341 0.5688 0,5455 0,0909 1.0000 0.1941 FBD 418 0.0857 0.0000 0.0000 1.0000 0.1696 341 0.0981 0,0000 0,0000 1.0000 0.1647 FBT 418 0.1744 0.1716 0.0000 0.5556 0.1218 341 0.1280 0,0833 0,0000 1.0000 0.1610 INH 418 0.3397 0.0000 0.0000 1.0000 0.4742 341 0.1994 0,0000 0,0000 1.0000 0.4001 LEV 418 0.6117 0.5656 0.0040 1.2727 0.2159 341 0.6789 0,6815 0,0701 1.6034 0.1984 InvRec 418 0.2070 0.1773 0.0000 0.8779 0.1787 341 0.3510 0,3522 0,0000 0.8695 0.2181 LIST 418 1.1818 1.0000 1.0000 2.0000 0.3862 341 - - - - -

SIA ('000) 418 4.21e+07 21,914,790 10.762 1.17e+09 1.38e+08 341 5,316,654 2,087,909 0.687831 9.32e+07 1.14e+07

Table 3 Pearson correlation matrix

AUF AFD FBD FBT INH LEV InvRec LIST SIA

AUF 1.0000 AFD 0.2662*** 1.0000 FBD 0.0968*** 0.0996*** 1.0000 FBT 0.2620*** 0.1333*** 0.6175*** 1.0000 INH 0.2843*** 0.0152 -0.0349 0.1123*** 1.0000 LEV 0.1111*** -0.0732** 0.1125*** 0.0522 0.3203*** 1.0000 InvRec -0.1455*** -0.0838** -0.0230 -0.2012*** -0.3329*** 0.0239 1.0000 LIST 0.4032*** 0.2851*** -0.0142 0.2412*** 0.1816*** -0.1048*** -0.3528 1.0000 SIA 0.6337*** 0.0862** 0.0347 0.1234*** 0.3113*** 0.2113*** -0.2063 0.3226 1.0000 ***, ** and * significant at 1, 5 and 10 percent respectively. Refer to table 1 for definitions of variables.

(25)

25

Figure 2: Development of the average ratio of tax advisory- and non-audit services to the total amount of external auditor fees.

Table 2 shows that the proportion of women on the board of directors and supervisory board combined is higher for listed companies, while the female representation in the board of directors is slightly higher for unlisted organizations. In general, both listed and unlisted companies do not nearly fulfill the target of 30% female representation in the board of directors and the supervisory board. Table 2 shows that 34% of the listed observations is a financial institution or operates in the utilities sector. For unlisted companies, this is only 20%. On average, unlisted companies are slightly more leveraged than listed companies. In addition, the ratio of inventories and receivables to total assets is higher for unlisted companies. Out of the 418 listed company observations, 76 are multiple listed.

Table 3 presents the correlations between the variables in our regression models. The Pearson correlation matrix identifies potential multicollinearity problems. Obviously, the proportion of women in the board of directors is highly correlated to the proportion of women in both boards combined. Inherent risk is moderately correlated to the size and leverage of an auditee, suggesting that financial- and public utility institutions are relatively larger and more leveraged. The listing status of a company is slightly correlated to the level of audit fees. As expected, there is a positive, statistically significant correlation between the size of an auditee and the amount of audit fees. This correlation is the highest in the correlation matrix, but the value (0,6337) is below the critical value of 0,80 (Greene, 1999). This suggests that multicollinearity is not problematic in this research.

0% 5% 10% 15% 20% 25% 2012 2013 2014 2015 2016 2017

Average ratio of tax advisory- and non-audit service fees to total amount of external auditor fees

(26)

26

5.2 Regression results on audit fee

In table 4, we present the regression results for the audit fee model. We use the specified research model of section 4.5 in this analysis. The relation between the independent variables and the level of audit fees is determined separately, in order to determine the significance of hypotheses H1, H2 and H3. The moderating effect of female diversity in the board on the relation between risk and audit fees and the relation between listing status and audit fees is presented in the models D and E respectively. Model E tests the results for all the determinants in our research model.

Model A presents the results for hypothesis 1, which estimates the relation between an organization’s risk profile and the amount of audit fees. The adjusted R² is high, which means that the independent variables decently declare the variations in the level of audit fees. The inherent risk of a company has a significant relation with the level of audit fees, albeit at a p-value of 6.3%. Contrary to our expectations, leverage shows a significant negative relation with audit fees. The ratio of receivables and inventories to totals assets is positively and significantly related to audit fees. Since the different coefficients for the risk variable show contradictory results, we reject hypothesis 1.

In model B the regression results for the hypothesis on listing status are presented. The adjusted R² of model B is high and all the determinants have highly significant coefficients. These results confirm an audit fee premium for organizations that are listed on the Amsterdam Stock Exchange and multiple listed on foreign exchanges. The regression results of model B mean that we cannot reject hypothesis 2. Therefore, the amount of listings is positively associated with audit fees.

Model C presents the results of hypothesis 3 that estimates the relation between female board representation and the level of audit fees. The model shows an insignificant negative result for the relation between the female proportion in the board of directors and the amount of audit fees. As expected, the relation between the female proportion in the board of directors and the supervisory board combined is significantly positive. Since the two determinants that measure the female proportion in the board show contradictory results, we reject hypothesis 3.

In model D and E, the effect of female presence in the board is added as a moderator to the models A and B. The results of model D are similar to the results of model A. To calculate whether the difference between model D with and without the moderating effect of female presence in the board is significant, we use the likelihood ratio test. This test compares whether the difference in log likelihoods of the two models are statistically significant. If this difference is significant, adding the moderating effect significantly improves the fit of the model,

(27)

27 compared to the model that only contains the individual variables. The likelihood ratio test shows significant chi-squared values. Therefore, the relation between the risk profile of an organization and the level of audit fees is stronger when gender diversity in the board is added as a moderator. Besides that, adding gender diversity in the board as a moderator results in a stronger relation between the number of listings and the level of audit fees. Due to the significance of these results, we cannot reject the fourth and fifth hypotheses.

Model F presents the regression results of the determinants that are separately presented in model A to C. The model shows similar significance results and the coefficients have the same directions as the individual models. The high adjusted R² states that the variation in the level of audit fees is explained well by the determinants in this model. As prior research confirmed, the size of an auditee is the most important determinant in explaining the level of audit fees, since removing this variable out of our model decreases the adjusted R² to 0.2436.

5.3 Regression results on the quality of audit fee disclosure

In this section, we present the regression results for the quality of audit fee disclosure. Similar to the regressions on the level of audit fees, we first present the regression results of the individual determinants to test the significance of the hypotheses H6, H7 and H8. The models J and K show the results when we add gender diversity of the board as a moderator. The last model of the result section presents the results when all the determinants of the quality of audit fee disclosure are included.

Model G presents the results of hypothesis 6 on the relation between an auditee’s risk and the quality of audit fee disclosure. The low adjusted R² indicates that the independent variables do not explain a large part of the variance in the quality of audit fee disclosure. For inherent risk, the coefficient points in the expected direction, but is not significant. Similar to the audit fee model, the leverage ratio shows a significant result with a negative coefficient. The ratio of inventories and receivables to total assets shows a significant negative association with the quality of audit fee disclosure as well. Since the determinants of an auditee’s risk show contradictory results, we reject hypothesis 6.

(28)

28

Table 4 Regression results for the level of audit fees.

***, ** and * statistically significant at 1%, 5% and 10% levels.

Model A Model B Model C Model D Model E Model F

Hypothesis H1 H2 H3 H4 H5 Variable Expected relation Intercept 2825.28*** 1056.58*** 1201.76*** 1697.74** 556.32 132.22 INH + 752.62* 682.10* 645.25* LEV + -2311.84*** -2349.58*** -1585.75** InvRec + 1619.10* 2539.75*** 3588.65*** LIST + 1580.29*** 1303.31*** 1497.06*** FBD + -1361.71 -1122.49 -168.60 -100.80 FBT + 7890.83*** 8282.08*** 5824.77*** 6414.37***

SIA + 6.91e-05*** _{6.30e-05*** 6.58e-05***} 6.74e05*** 6.21-e05*** 6.34e-05***

N 759 759 759 759 759 759 Adjusted R² 0.5627 0.5800 0.5799 0.5875 0.5935 0.604 Model Pooled OLS-regression Pooled OLS-regression Pooled OLS-regression Pooled OLS-regression/Likelihood ratio test Pooled OLS-regression/Likelihood ratio test Pooled OLS-regression

(29)

29 In model H, we present the relation between the number of listings and the quality of audit fee disclosure. The adjusted R² still is very low, but much higher compared to model G, declaring that listing status has a higher explanatory value than the risk determinant. As expected, the coefficient is positive and highly significant. Therefore, we cannot reject hypothesis 7, which means that the amount of an auditee’s listings is positively associated with the quality of audit fee disclosure.

Model I shows the results for the eighth hypothesis regarding the relation between gender diversity in the board and the quality of audit fee disclosure. The adjusted R² is very low, which means that gender diversity in the boardroom does not explain a large part of the variation in audit fee disclosure. Both determinants show a positive relation, but only the gender diversity in the board of the directors and supervisory board combined is statistically significant. Therefore, we reject hypothesis 8.

In the models J and K, the regression results are presented after gender adding diversity in the board as a moderator. The adjusted R²’s are slightly higher compared to the individual models, but the models still have low explanatory power in declaring variances in the dependent variable. Model J shows similar results in comparison with model G, besides the FBT coefficient which changed direction and is not significant in this model. The coefficients are further removed from zero and therefore more pronounced. We use the likelihood ratio test to determine whether a significant difference between the models is ascertained when gender diversity in the board is added as a moderator. The test shows a chi-squared value of 0.0016 which means that gender diversity in the board strengthens the relation between an auditee’s risk and the quality of audit fee disclosure. Therefore, we cannot reject hypothesis 9. The same test is conducted for the interaction effect of gender diversity in the board on the relation between listing status and audit fee disclosure. This results in a significant chi-squared as well, and hence we cannot reject hypothesis 10.

In the last model, we add all the determinants of our regression. The low R² values in all the models indicate that the determinants in our regression have low explanatory power in explaining variances in the quality of audit fee disclosure. The models I and K show an increase in adjusted R² when listing status is added as a determinant. Due to its effect on the adjusted R² and its high significance, the main finding is that the listing status of an auditee explains the most of the variance in the quality of audit fee disclosure in our model.

(30)

30

Table 5 Regression results for the quality of audit fee disclosure

***, ** and * statistically significant at 1%, 5% and 10% levels.

_{Model G}

Model H Model I Model J Model K Model L

Hypothesis H6 H7 H8 H9 H10 Variable Expected relation Intercept 0.6746*** 0.5442*** 0.5846*** 0.6586*** 0.5144*** 0.5582*** INH + 0.001 0.0539* 0.0008 LEV + -0.0825** -0.1113** -0.0427 InvRec + -0.0650* -0.1441*** 0.0172 LIST + 0.0955*** 0.1419*** 0.0960*** FBD + 0.0360 0.1330 0.1940** 0.1295** FBT + 0.1359** -0.1720 0.0364 -0.0149

SIA + 1.62e-10 -1.54e-10 1.08e-10 1.92e- 10 -1.39e-10 -1.24e-10

N 759 759 759 759 759 759 Adjusted R² 0.0172 0,0915 0,0195 0.0464 0.1079 0.0966 Model Pooled OLS-regression Pooled OLS-regression Pooled OLS-regression Pooled OLS-regression/Likelihood ratio test Pooled OLS-regression/Likelihood ratio test Pooled OLS-regression

Corporate risk profiles, female board members and their effect on audit fees and the quality of audit fee disclosure: Evidence from Dutch listed and unlisted companies