The influence of ownership structure on audit fee and audit fee disclosure in the Dutch market

(1)

Supervisor: W. Kevelam Co-assessor: R.L. ter Hoeven

Master Thesis Accountancy MSc Accountancy & Controlling

S2742381 Faculty of Economics and Business

January 21, 2019 University of Groningen

The influence of ownership structure on audit

fee and audit fee disclosure in the Dutch market

Stefan Blokzijl

University of Groningen

ABSTRACT: High quality corporate governance mechanisms are assumed to have the power to influence the demand for audit services. Ownership structure is considered to be part of the corporate governance of corporations. Conclusive evidence on the influence of ownership characteristics and audit fees, however, remains unknown. This paper aims to provide convincing evidence on the relationship between ownership characteristics and audit fee (disclosure), by examining the association in a market that is characterized as concentrated. By using a generalized least squares regression analysis for a sample of 377 Dutch listed firms, we find evidence that ownership concentration and institutional ownership are positively associated with audit fees. Based on the same sample, we also find evidence that ownership concentration and institutional ownership are negatively associated with the quality of audit fee disclosure. Finally, evidence of a positive association between social media usage and audit fee disclosure is found.

Keywords: Audit fees; audit fee disclosure; corporate governance; ownership structure; ownership concentration; inside ownership; outside ownership; social media usage.

(2)

Table of contents | 2

ecently, the external audit received a vast amount of attention due to the collapse of construction company Carillion in the UK and insufficient prevention of money laundering at ING in the Netherlands (Brooks, 2018; Piersma, 2018). Inefficient and malfunctioning corporate governance mechanisms are believed to be among the major reasons that cause such corporate scandals (Mustapha & Ahmad, 2011). In response to this, corporate governance and the external audit have received a significant amount of attention from academics, regulators and the media (e.g. Gotti, Han, Higgs & Kang, 2012; Griffin, Lont & Sun, 2009). Evidence shows that good corporate governance mechanisms increase the demand for high quality audits (Fan & Wong, 2005), which can mitigate the occurrence of future scandals. Corporate governance mechanisms are primarily focused on solving agency problems that emerge from the difficulty for shareholders to monitor the firm’s management (Aguilera & Jackson, 2003). Ownership structure can influence this monitoring role, as ownership concentration, for example, can either mitigate or exacerbate agency problems (Setia-Atmaja, 2009). Therefore, ownership structure is argued to be an important governance mechanism. Effective monitoring by shareholders, in turn, can influence the demand for audit services (Teoh & Wong, 1993). Corporate monitoring could on the one hand lead to a demand for higher audit quality (Kane & Velury, 2004), but could on the other hand reduce the control risk which decreases audit fees (Tsui, Jaggi & Gul, 2001). As the audit fee is considered to be a good proxy for audit quality (Hoitash, Markelevich & Barragato, 2007), the purpose of this study is to explore whether effective governance mechanisms influence the audit quality. This relationship is examined by investigating to what extent audit fees and audit fee disclosures are influenced by different ownership structures.

The traditional audit fee model of Simunic (1980) states that audit fees are determined based on the function of (i) the effort of the auditor during the engagement and (ii) the client-specific risk incurred by the auditor to execute its audit. From this model it can be argued that higher audit risks, occurring from client-specific risks, imply more audit activities which will result in a higher audit fee. With the introduction of article 2:382a1 in the Dutch Civil Code, it has become mandatory to disclose the audit fee starting from the financial year of 2008 (Kevelam, ter Hoeven & Brouwer, 2017). This makes it relevant to conduct research on the determinants

1_{Art. 2:382a Burgerlijk Wetboek: Bijzondere voorschriften omtrent de toelichting.}

http://wetten.overheid.nl/jci1.3:c:BWBR0003045&boek=2&titeldeel=9&afdeling=5&artikel=382a&z=2018-09-19&g=2018-09-19

(5)

Introduction | 5 of the audit fee. Also, article 2:382a explains what (at the minimum) should be included in the audit fee disclosure. Those items are mandatory for large Dutch legal entities (medium and small corporations are exempted for this rule), however the extensiveness and readability of the disclosure can vary between companies. Prior research of Dye (1991) found that audit fee disclosure is positively associated with auditor independence. Therefore, this research will also focus on the effect of ownership characteristics on the quality of audit fee disclosure.

1.2 Academic contribution

Recently, linking ownership concentration to audit fees has become more popular. This could be explained by the study of Hay, Knechel & Wong (2006) that indicated several gaps in the existing literature on audit fees. They suggested that future research on audit fees can be particularly useful in the research area of different forms of ownership. In the years following, different studies investigated the link between certain ownership characteristics and audit fees in other (national) settings. However, these studies have not come to a unanimous conclusion on the effect of ownership on the level of audit fees. This study aims to provide convincing evidence to this research gap, by investigating the issue in the Dutch context, which can be characterized by concentrated ownership (Kabir, Cantrijn & Jeunink, 1997). This is relevant, since La Porta, Lopez-de-Silanes & Shleifer (1999) examined the 20 largest firms in 27 countries and found that concentrated ownership is the norm, not the exception. This research, therefore, aims to provide conclusions that are generalizable to other markets that are characterized as concentrated. In the context of concentrated markets, ownership concentration functions as a corporate governance mechanism itself (Edwards & Nibler, 2000) and the audit can be seen as the outcome of the effectiveness of this mechanism. This makes it interesting to focus on a setting that is characterized by ownership concentration, to find out to what extent different ownership types demand a different extent of audit quality.

When significant effects of ownership structure on audit fees are found, this can be valuable for investors as the ownership structure can determine the demand for audit quality. This will also be useful for regulators, the accounting profession and other users of financial statements.

Apart from focusing on audit fees, this research will also look into the quality of the audit fee disclosure. In contrast to the amount of literature on audit fees, there is limited existing research on audit fee disclosure and determinants that may result in a higher audit fee disclosure quality. This study aims to fill this gap by investigating the effect of ownership structure on the quality of audit fee disclosure. The mandatory disclosure of the audit fee arises from the European

(6)

Introduction | 6 Regulation 2006/43/EG, which resulted in article 2:382a of the Dutch Civil Code. This article forces Dutch public interest entities (PIEs), starting from the financial year of 2008, to disclose the audit fee in the annual report. This law article, however, raised some questions about ambiguities in the formulation of the article. Subsequently, the NBA (at that time called the NIVRA), the Dutch professional organization for auditors, presented a standardized format2 in 2009 in which they prefer organizations to disclose details about the audit fee. Later on, this format is adopted in RJ 390 (Raad van Jaarverslaggeving, 2017), a Dutch guideline on annual reporting. However, there still exist differences in the quality of the audit fee disclosure, as the extensiveness and readability of the disclosure is up to the auditee. These differences cause the foundation to conduct research on determinants that potentially influence the audit fee disclosure.

Finally, investigating audit fees in the period of 2012-2017 is extra relevant because of the implementation of mandatory audit rotation by the Dutch government at the end of 2012. This regulation is implemented in 2012 and forces Dutch PIEs to switch audit firm every 8 years. Two years later, the EU implemented this rule in the European Regulation (537/2014)3, which demands specific requirements regarding statutory audits of PIEs. Since most audit firms used to have longstanding relationship with their clients, this has forced most of the PIEs to change auditors before the 1st_{of January 2016 (Kevelam et al., 2017). Kevelam et al. (2017) found a}

decrease in audit fees for AEX corporations in the period of 2014-2016. It will therefore be extra relevant to indicate the role of ownership structure on the level of audit fees in this time period.

1.3 Research question

Ownership concentration will be the starting point of this research. After that, the study will focus more specific on the identity of the shareholders. To do this a distinction is made between: inside ownership and outside ownership. Inside ownership consists of (a) managerial ownership, which refers to the shares held by the management, and (b) family ownership, which refers to the shares held by a (founding) family. Outside ownership is split up into institutional ownership, which refers to the shares hold by institutions (i.e. banks, insurance companies, pension funds), and non-institutional ownership, which refers to cross-holdings by other companies and holdings of other non-institutional shareholders.

2_{NIVRA (2009): Vermelding accountantshonorarium in de jaarrekening. NIVRA-Wijzer 1.}

3_{EU Regulation No 537/2014: Specific requirements regarding statutory audit of public-interest entities.}

(7)

Introduction | 7 Large institutional shareholders have bigger incentives to monitor management more closely due to the large stakes of the company that they own (Chung & Zhang, 2011). The same accounts for non-institutional ownership. Due to their outside monitoring function, outsiders are expected to demand higher audit quality. Hence, a higher audit fee and better audit fee disclosure is expected. Inside owners (i.e. managerial- and family owners) are assumed to have access to internal information sources (Chen, Chen, Cheng, 2008) and this is expected to have a decreasing effect on the level of audit fee and the quality of audit fee disclosure. As prior research found a decreasing effect of the audit fee after the rotation of the audit firm, this is expected to have a negative moderating effect. Finally, the effect of social media usage will be investigated on the quality of audit fee disclosure. It is expected that firms that use social media more intensively will disclose a higher quality audit fee disclosure.

This study therefore examines the relationship between different types of ownership characteristics and the level of audit fees and the quality of audit fee disclosure. It is expected that different types of ownership will affect the level and disclosure of audit fees. Hence, the research question guiding this research is:

To what extent are audit fees and audit fee disclosure influenced by different types of ownership structure in the Netherlands?

In the remainder of this research paper, the theoretical framework will be elaborated as well as the hypothesis development. Thereafter, the research method will be provided along with the results. Finally, the conclusion and discussion of this study will be presented.

(8)

Theoretical framework | 8 2. THEORETICAL FRAMEWORK

In this theoretical framework, the theory behind the concepts of audit fees and audit fee disclosure will be provided as well as the expected effect of each type of ownership structure on the amount of audit fees and the quality of audit fee disclosure. First, a general explanation and application of the theories will be provided. Subsequently, different types of ownership characteristics are explained and the hypotheses will be constructed.

2.1 Underlying theories

In this section, the underlying theories that are relevant for this topic will be elaborated. First, the agency theory is presented, followed by the signaling- and substitution theory.

2.1.1. Agency theory

Agency theory is the main theory that is used for research on corporate governance mechanisms. The agency theory arises from the separation between ownership and control and was introduced by Jensen & Meckling (1976). Their study came up with a theory that assumes a principal-agent relation, where shareholders (principals) are the ones that own the firm, whereas managers (agents) are the ones who are in charge of the control of the firm. This separation can create agency problems, as managers possess more relevant information about the firm than the shareholders, which is referred to as information asymmetry (Berle & Means, 1932). The existence of information asymmetry provides the agents the possibility to act in their own interests instead of the interests of the owners, which can be referred to as the moral hazard problem (Berle & Means, 1932). Another agency problem arises from the difficulty for the principals to monitor the actions of the managers (Eisenhardt, 1989). In order to overcome these problems the interests of principals and agents should be aligned.

One mechanism to reduce the information asymmetry is the audit function, as the audit function provides assurance to the outside stakeholders that the disclosed information is correct (Varici, 2013). More specifically, auditing reduces agency problems by ensuring the quality of the financial statements, which reduces the moral hazard problem that the principals face towards the managers (Francis & Wang, 2008). The execution of high quality audits is an important check on opportunistic managerial behaviour, which can be substantiated by the research of Teoh & Wong (1993). This research finds a positive relationship between earnings quality and audit quality, which indicates the importance of high quality audits. As audit fees are a good proxy for audit quality (Hoitash et al., 2007), a certain level of audit fee can be assumed to be sufficient for the principals to prevent the moral hazard problem. With the implementation of

(9)

Theoretical framework | 9 the European guideline it is mandatory for Dutch listed firms to disclose on the audit fee starting from the financial year of 2008 (Kevelam et al., 2017). This gives principals the opportunity to compare the audit fee with the audit fee of other (rival) companies.

According to the agency theory, two common types of agency problems are prevalent, i.e. Type I and Type II. In studies investigating ownership structure, both types are of much importance. Type I agency problems are related to differences in managements’ interests and shareholders’ interests, whereas Type II agency problems refer to the differences in interests between large and small shareholders (Shleifer & Vishny, 1986). Type II agency problems, for example, are present when large shareholders try to extract personal benefits over the smaller shareholders. These different types of agency problems play an important role in the area of research on ownership structure, as the different types of agency problems determine the amount of monitoring by the shareholders.

2.1.2. Signaling Theory

Next to the agency theory, the signaling theory is applicable to this research. Signaling theory is founded by Hughes (1986) and focuses on reducing information asymmetry. The main idea is that in a market with different quality level firms, “the party possessing superior quality information can send signals to other interest-related parties” (An, Davey & Eggleton, 2011). A signal in the audit market, for example, can be the purchase of more audit services (Carcello, Hermanson, Neal & Riley, 2002) or to choose a reputable audit firm to increase the credibility of the financial figures for outside stakeholders (Fan & Wong, 2005). In order to be effective, “a signal should be difficult to be imitated by the low quality firms” (An et al., 2011).

Due to the information asymmetry that exists between the firm and its investors, a high quality firm has the incentive to highlight its superior quality over other firms in order to attract more investors (An et al., 2011). Therefore, a firm which possesses more favourable private information will purchase a high quality audit, as this signal is hard to copy for a firm with less favourable private information (Titman & Trueman, 1986). Of course, the auditee will not purchase a high quality audit primarily to send a signal, but rather as a motive to obtain financing, boost firm value, reduce financing costs or attract investors (Wu, 2012).

Therefore, following the signaling theory, companies with better corporate governance will send such signals to the investors. This will affect the audit fee and the audit fee disclosure in a positive way. Audit fee disclosure in particular could be a very effective way to signal organizational superiority and transparency to investors.

(10)

Theoretical framework | 10

2.1.3. Substitution Theory

Substitution theory takes an opposing view compared to the signaling theory. This theory suggests that corporate governance mechanisms “effectively substitute for one another and/or operate in concert” (Dalton, Daily, Certo & Roengpitya, 2003). Therefore, for a company with superior corporate governance it is less needful to purchase more audit services (Wu, 2012). Following the traditional audit fee model, lower agency costs will reduce the risks the auditor faces, which will lead to a lower audit fee. This is confirmed by Knechel & Willekens (2006) who noticed that “the demand for external audit is influenced by risks faced by corporate shareholders and mechanisms employed to minimize these risks”. Therefore, part of the external audit substitutes for effective internal corporate governance (Wu, 2012). As a result, following the substitution theory, effective corporate governance will negatively affect audit fees and will lead to lower audit fee disclosure quality. So, when different types of ownership structure affect the internal governance of the firm, this may also potentially affect the workload of the auditor. Yet, there is no concluding evidence which points to one particular direction.

2.2 Hypotheses development – Audit fees

In this section, the hypotheses that relate to audit fees will be developed based on the underlying theories. First the hypotheses regarding outside ownership will be elaborated, followed by the hypotheses regarding inside ownership.

2.2.1. Ownership concentration and audit fees

Ownership concentration will be taken as a starting point to investigate the influence of ownership structure on the level of audit fees. Hamadi & Heinen (2015) state that ownership is assumed to be concentrated when the largest shareholder owns more than 10% of the outstanding shares. According to Shleifer and Vishny (1997) “ownership concentration itself is an important determinant of effective corporate governance", which illustrates the monitoring function that is present under concentrated ownership. In other words, when there is complete diffused ownership, the small shareholders are less inclined to monitor the management themselves. As their stakes are low, they are not willing to put in a lot effort to monitor the management, they rather rely on others. This is generally considered as the ‘free rider problem’ (Morck, 2000).

By purchasing more shares, a shareholder can become a substantial shareholder. In 2013, the threshold to report a substantial shareholding decreased from 5% to 3%4. Therefore, in the

4_{Wet op het financieel toezicht.}

(11)

Theoretical framework | 11 Netherlands a shareholder is perceived as substantial when at least 3% of the outstanding shares is owned. Shareholders owning 3% or more of the issued capital of a listed company are obliged to report this to the Autoriteit Financiële Markten5_{(AFM), the Dutch Authority for the Financial}

Markets. When shareholdings increase and ownership becomes more concentrated, the shareholders are able to overcome the ‘free rider problem’ (Kabir et al., 1997). In this sense, shareholders become more inclined to monitor the actions of the management, which will constrain managers to take actions that are in their own interest.

Watts & Zimmerman (1983) argue that greater conflicts of interest increase the demand for audit quality. This would assume that, in contrast to Morck (2000), under diffused ownership there is a higher demand for more audit services. However, others argue that with major shareholdings there is the opportunity of an agency type II problem. In order to signal non-expropriation of minority shareholders, large shareholders will demand a high quality audit (Fan & Wong, 2005). Hay, Knechel & Ling (2008) also conclude that more concentration in ownership will increase the investments in the external audit. Therefore, we expect a positive relationship between ownership concentration and audit fees, which is reflected in the following hypothesis.

H1: Ownership concentration is positively associated with the level of audit fees. 2.2.2. Outside ownership and audit fees

Outside ownership focuses on the shareholding parties that do not have any inside relationship with the firm. Outside ownership will be divided into institutional ownership and non-institutional ownership.

Institutional ownership

Institutional investors are investors that exert a high ownership engagement. Traditionally, banks, investment funds, insurance companies and pension funds are characterized as institutional investors. A unique characteristic of institutional investors is that they invest money of the public that they have at their disposal, due to their business activities6. Monitoring is a key aspect when discussing the relationship between institutional shareholdings and audit fees. The long-term horizon that institutional investors have, makes them more likely to monitor the firm more closely (Maug, 1998). Further, as institutional ownership is related to significant

5_See:

https://ww.afm.nl/en/professionals/doelgroepen/effectenuitgevende-ondernemingen/meldingen/substantieel

6_See:

(12)

Theoretical framework | 12 stockholdings in a company (Chung & Zhang, 2011), these parties have more incentives to monitor managers compared to the ones that invested only little or no wealth in the company (Shleifer & Vishny, 1986). When relatively few shares are hold, they can easily liquidate their shares which reduces the incentive to monitor the firm (Maug, 1998), which is in accordance with hypothesis 1.

Institutional investors also have more expertise to monitor management. For example, Balsam, Bartov & Marquardt (2002) found that institutional investors recognize earnings management more quickly than individual investors. Kane and Velury (2004) argue that institutional investors demand high quality information, as earnings information is important for valuation purposes and they need to rely on the financial numbers set up by the company. Combined with the low turnover of shares, institutional investors encourage managers to improve performance and to increase the quality of financial reporting (Velury, Jenkins, 2006). This monitoring role will demand more audit quality and hence higher audit fees.

However, too much ownership by institutional blockholders could also ensure them to extract private benefits from minority shareholders, the agency type II problem (Barroso, Ali, Lesage, 2018). When the shareholdings reach a certain threshold, the institutional blockholders could focus more on direct monitoring of the management and the audit function becomes an unwanted cost (Bohinc & Bainbridge, 2001). Also, their monitoring function can result in the implementation of better internal governance mechanisms, which has the potential to reduce the audit workload (Prawitt, Sharp & Wood, 2011). This could cause a tendency to put pressure on the company to reduce the audit fees.

We find the arguments that assume a positive relationship, however, more compelling. Mainly because of the argument of Kane & Velury (2004), that reliable financial information is important for institutions to make their decisions. This results in the following hypothesis.

H2: Institutional ownership is positively associated with the level of audit fees. Non-institutional ownership

Following Donker, Santen & Zahir (2009), non-institutional owners are defined as “cross-holdings by other companies and “cross-holdings of other institutional shareholders”, i.e. non-institutional ownership refers to the shares held by corporations that are not institutions. The distinctive difference with institutional investors is that they invest their own money. Following the agency theory, non-institutional ownership and auditing both function as a corporate governance mechanism (Jensen & Meckling, 1976), as both mechanisms have the ability to

(13)

Theoretical framework | 13 reduce information asymmetry. Claessens, Djankov & Lang (2000) confirm this by arguing that investors with large shareholdings will have strong incentives to collect useful information in order to reduce information asymmetry and to overcome the classic agency problem.

According to Mitra et al. (2007), ‘”when individual shareholdings increase, they become more actively engaged in the company’s affairs, including the financial reporting process”. This active monitoring could result in a lower probability of material misstatements and therefore a lower inherent risk for the auditee. Hence, it could be expected that this will reduce the audit fee that is charged by the audit firm.

Alternatively, the monitoring function of non-institutional shareholders can also increase the demand for high quality audits in order to safeguard the reliability of financial information (Mitra et al., 2007). Also, as monitoring is not the primary business objective of non-institutional shareholders, it can be expected that they will delegate the task of monitoring to the auditor (Han, Kang & Rees, 2013). Another argument could be that the client firm purchases a high quality audit in order to attract future blockholders to invest in the firm. The above mentioned arguments lead to the following hypothesis.

H3: Non-institutional ownership is positively associated with the level of audit fees. 2.2.3. Inside ownership and audit fees

In this study, inside ownership refers to the shares owned by the (previous) management and by family members of the company. As the individual ownership share of managers and family owners is relatively small in the Dutch context, they will be taken together as one group in the hypothesis development.

Managerial ownership

Managerial ownership is regarded as being an essential solution to mitigate agency conflicts (Jensen & Meckling, 1976). An increase in managerial stock ownership will ensure that managers’ wealth is more tied to the long-term value of the company and therefore the interests of shareholders and managers are more aligned.

Mustapha & Ahmad (2011) find that total monitoring costs related to auditors will be lower if large portions of company shares are owned by the directors or managers. Gul, Chen & Tsui (2003) explain this by suggesting that high managerial stock ownership diminishes opportunistic behaviour. Therefore, managers are less likely to opportunistically misreport financial information, increasing the reliability of accounting data. In such a low-risk environment, auditors reduce the amount of risk premium and/or the level of planned audit

(14)

Theoretical framework | 14 investment, both of which have a declining effect on audit fees (Mitra et al., 2007). Warfield, Wild & Kenneth (1995) find a similar result and state that “managers are willing to show the company’s realistic economic value when their interests are aligned with the interests of shareholders”. Therefore, it is expected that increasing managerial shareholdings will have a negative effect on audit fees.

Family ownership

Family businesses are distinctive in terms of ownership as “families represent a unique type of shareholders that hold poorly diversified portfolios, are long-term investors and in control of senior management positions” (Anderson & Reeb, 2003). According to the agency theory, this enables them to influence and monitor the firm, resulting in less Type I agency problems (Ho & Khang, 2013). The distinctive characteristics of these firms are supposed to be instrumental in the auditor choice and the level of audit fees (Lin & Liu, 2009). This difference is argued to be caused by the different agency conflicts which lead to varied demands for audit quality. On the one hand, family owned firms face the risk that expropriation of own benefits by the family members is relatively high. Moreover, the influence they have over the financial reporting can cause higher audit risks and accordingly a higher audit fee (Lei & Lam, 2018). On the other hand there are researchers that argue that family ownership can lead to the opportunity that families pursue their own benefits over other shareholders, the agency type II problem. This may reduce the incentive to hire high quality audits, resulting in a lower audit fee (Fan & Wong, 2005). Also, according to Ali & Lesage (2014), “families have less diversified portfolios and large blocks of shares, which gives them strong incentives to monitor management, which in turn reduces potential manager opportunism”. Therefore, the demand for more audit services and hence a higher fee is expected to be weak in family firms.

Taken the previous arguments in consideration, this study predicts a negative association between inside ownership and the level of audit fees. This leads to the following hypothesis.

H4: Inside ownership is negatively associated with the level of audit fees. 2.2.4. Audit firm rotation and audit fees

As indicated in the introduction section, the EU Regulation regarding specific requirements of statutory audits of PIEs has made it mandatory for PIEs to regularly change their auditor. More specifically, it forces Dutch PIEs to switch their audit firm in the period of 2012-2016, which makes it highly relevant to include this in this research study.

(15)

Theoretical framework | 15 - + + + - - - -

Audit firm rotation can be expected to influence the audit fee. It would be plausible to expect an increase in the audit fee, as the new audit staff needs to gather knowledge of the auditee, which will take time and hence more costs (Raiborn, Schorg & Massoud, 2006). However, Corbella, Florio, Gotti & Mastrolia (2015) indicate a reduction in the audit fee of Italian listed firms in the first year after the audit firm change. This can be explained by the low-balling effect, which refers to the practice that audit firms use to charge lower audit fees for an initial audit engagement, in order to earn the right to charge future quasi-rents of audit fees (DeAngelo, 1981). This is confirmed by Kevelam et al. (2017) that report a reduction in the audit fee for Dutch listed firms after they switched to a new audit firm. As the association between auditor rotation and audit fee is expected to be negative, it is expected that audit firm rotation will have a negative moderating effect on the previous variables. This study will therefore extend the existing research by investigating the moderating effect of audit firm rotation on the relationship between ownership characteristics and the audit fee.

H5: Audit firm rotation has a negative moderating effect on the relationship between ownership concentration and audit fees in the year after the audit firm rotation. H6: Audit firm rotation has a negative moderating effect on the relationship between institutional ownership and audit fees in the year after the audit firm rotation. H7: Audit firm rotation has a negative moderating effect on the relationship between non-institutional ownership and audit fees in the year after the audit firm rotation. H8: Audit firm rotation has a negative moderating effect on the relationship between inside ownership and audit fees in the year after the audit firm rotation.

The different hypotheses are schematically showed in Figure I.

Ownership concentration Institutional ownership Non-institutional ownership Inside ownership

Level of audit fee Audit firm rotation

(16)

Theoretical framework | 16 2.3 Hypotheses development – Audit fee disclosure

In this section, the hypotheses regarding audit fee disclosure will be elaborated. As the literature on audit fee disclosure is in its emerging phase, there is limited research available on the topic of audit fee disclosure. However, as the extensiveness of the audit fee disclosure can be decided in collaboration with the auditor and the auditee, the disclosure shares common ground with literature regarding voluntary disclosure. Therefore, literature on voluntary disclosure will also be used to build up the hypotheses.

2.3.1. Ownership concentration and audit fee disclosure

Fama & Jensen (1983) propose that when share ownership is widely held, there are more potential conflicts of interest between manager and shareholder. As a result, information disclosure is likely to be more extensive compared to closely held firms, as principals will signal that they act in the interest of the shareholders, to attract potential investors. Firms that are closely held, are often assumed to be irresponsive to demand full disclosure, as the shareholders are expected to have access to the information that they require (Cormier, Magnan & van Velthoven, 2005). The following arguments lead to the development of the following hypothesis.

H9: Ownership concentration is negatively associated with the quality of audit fee disclosure.

2.3.2. Outside ownership and audit fee disclosure

The hypotheses development focusing on audit fee disclosure will follow the same structure as the hypotheses that focus on the level of the audit fee. Therefore, outside ownership will be again divided into institutional ownership and non-institutional ownership.

Institutional ownership

As mentioned before, the unique characteristic of institutional investors is that they invest third parties’ money that they have at their disposal. Due to their large long-term shareholdings, institutional owners play an active role in the monitoring of the firm (Chung & Zhang, 2011). According to Ajinkya, Bhojraj & Sengupta (2005) “institutional investors desire and demand more disclosures as they are not able to directly oversee the activities of the managers”. Managers are assumed to have a tendency to withhold information, whereas institutional investors, given their incentives to monitor the managers, are assumed to play an important role in the information function (Shleifer & Vishny, 1986). Empirical evidence from Bushee & Noe

(17)

Theoretical framework | 17 (2000) provides evidence that documents a positive relationship between institutional ownership and the extent of corporate disclosure. This results in the following hypothesis.

H10: Institutional ownership is positively associated with the quality of audit fee disclosure.

Non-institutional ownership

Large shareholders are particularly assumed to have fewer agency conflicts, as they are able to overcome the ‘free rider problem’ (Morck, 2000). As mentioned before, the ‘free rider problem’ is an economic concept that deals with the lack of interest by diffused shareholder ownership to take action against the management of the firm, as diffused owners will both lack the means and motives to address agency problems. Therefore, when stockholder ownership is concentrated and substantial in nature, investors become more interested in monitoring management actions to increase firm value (Mitra & Hossain, 2007). As corporate- or non-institutional shareholdings are held by firms of which their primary activity is not monitoring their investments, but running their own business, it can be expected that they demand more disclosure.

H11: Non-institutional ownership is positively associated with the quality of audit fee disclosure.

2.3.3. Inside ownership and audit fee disclosure

The hypothesis development of inside ownership will follow the same structure as the hypothesis development of inside ownership on audit fees. Inside ownership will therefore refer to managerial ownership and family ownership.

Managerial ownership

Managerial ownership (i.e. shares held by executive directors and executive officers) is also expected to affect the audit fee disclosure. In order to align the interests of the principals and agents, Jensen & Meckling (1976) suggest to pay managers in stocks. This will align the interests of the managers with the interests of the shareholders, as both strive for the highest stock price. However, high levels of managerial ownership will encourage the retention of information because managers rely on internal information sources rather than external information sources (Sartawi, Riyad & Ala’eddin, 2014). Therefore, according to Khliff, Ahmed & Souissi (2017) “when the proportion of insider ownership becomes greater, the monitoring capacity of outsiders is reduced, which results in less voluntary disclosure by firms”.

(18)

Theoretical framework | 18 Empirical evidence presented by Eng & Mak (2003) supports this view. They report that lower levels of managerial ownership is associated with increased levels of voluntary disclosure. This leads to the proposition that managerial ownership negatively affects the audit fee disclosure.

Family ownership

Family firms possess some unique characteristics that make it interesting to focus on. Compared to other shareholders, family owners normally have a much longer investment horizon. Furthermore, they are assumed to have better access to private information (Chen et al., 2008). Since family members are assumed to have direct access to these information sources, this will decrease the problem of information asymmetry. Thus, they can be expected to demand less detailed disclosure of the audit fee. In the end this will match the interests of both family owners and managers, as they both possess the relevant information. This will reduce the agency problem between family owners and managers, which in turn leads to the proposition that family owned firms potentially demand less disclosure.

The unique characteristics of family firms lead to the proposition that family owners have less preferences for audit fee disclosure. The above stated arguments lead to the following hypothesis.

H12: Inside ownership is negatively associated with the quality of audit fee disclosure.

2.3.4. Social media usage and audit fee disclosure

Finally, the effect of social media usage will be investigated on the quality of audit fee disclosure. In the past years, social media has come up as an important tool for organizations to interact with their stakeholders (Zhou et al., 2015). It can, for example, be used to increase equity and business values and to disclose corporate information (Yu, Duan & Cao, 2013). A key study in this research area is the study of Blankespoor, Miller & White (2014). They found that dissemination of corporate information through Twitter is associated with a decrease in information asymmetry. This study proposes that when organizations are more active on social media, they find it more important to communicate information to stakeholders. Hence, they will be more likely to disseminate more corporate disclosures in the annual report. The rationale behind this association is that organizations that intensively use social media are more willing to close the information gap with their stakeholders (Blankespoor et al., 2014).

Culnan, McHugh & Zubillaga (2010) found that Twitter is the most frequently used social media platform that U.S. organizations use. In the Netherlands, organizations are also relatively

(19)

Theoretical framework | 19 - - + + +

active on Twitter7_{. Therefore, the focus of this study will be on the Twitter activities of Dutch}

firms to find out if social media intensive firms disclose better on the audit fee.

H13: Social media usage is positively associated with the quality of audit fee disclosure.

The different hypotheses are schematically showed in Figure II.

7_{CBS (2014): Nederlandse bedrijven zeer actief op Twitter.}

https://www.cbs.nl/nl-nl/nieuws/2014/27/nederlandse-bedrijven-zeer-actief-op-twitter Ownership concentration Institutional ownership Non-institutional ownership Inside ownership

Quality of audit fee disclosure Social media usage

(20)

Research method | 20 3. RESEARCH METHOD

In this section the research method is explained. First the data and sample will be explained, followed by the operationalisation of the dependent-, independent- and control variables. After that, the research model is explained, followed by the relevant data adjustments that are executed.

3.1 Data and sample

This study focuses on the financial years of the period 2012 - 2017. As mentioned before, this period is extra relevant due to the implementation of mandatory audit rotation regulation, which forces Dutch PIEs to change their audit firm before the 1st of January 2016. The initial sample comprises 75 Dutch listed firms and 75 Dutch non-listed firms for the specified years. The listed firms will be included in all the analyses to test the hypotheses. The non-listed firms are only included in the analysis to test the effect of social media usage on audit fee disclosure.

The listed firms comprise the 75 firms that are listed on the Amsterdam Exchange index (AEX), Amsterdam Mid Cap index (AMX) or the Amsterdam Small Cap index (AScX). The index composition of year-end 2016 of these indices will be used for this research. As the composition of the indices is based on market capitalisation, the composition changes every year. Due to this, in combination with delistings and IPOs, some years will have less observations. The final sample of listed firms comprises 377 observations of 72 corporations. Table I gives an overview of the total sample of companies and the explanations for deleting certain items.

The unlisted firms comprise 75 firms for the period 2012-2017, which are selected based on the amount of employees at the end of 2017. The company.info database was used to select all the companies with an employee amount of 500 or higher. This resulted in 2.152 companies. Subsequently, 102 companies were deleted as they were stock listed companies and 1.043 companies were deleted due to the absence of an annual report for the financial year 2017. This resulted in a population of 1.007 organizations. From these organizations, the 75 companies with the highest amount of employees were selected, as their public relevance is assumed to be the highest. Due to the unavailability of data, some observations were deleted. The explanations that justify the deletion of observations are showed in Table I. The final sample of unlisted firms comprises 218 observations of 45 corporations. A complete overview of all included firms can be found in Table B.I in Appendix B.

(21)

Research method | 21

TABLE I

Sample Overview Listed companies

Total firm-year observations of listed companies Delisted companies and IPOs

Incomplete data

Total observations listed companies

450 (44) (29)

377

Non-listed companies

Total firm-year observations of unlisted companies Missing company data

Missing social media data Missing audit fee disclosure data

Total observations unlisted companies

450 (49) (153) (30) 218 3.2 Dependent variables

The dependent variables in this research are both the audit fee and the quality of audit fee disclosure by the Dutch firms. Audit fee and audit fee disclosure are mandatory topics of the annual report of Dutch PIEs. The audit fee can be found in the audit fee disclosure. The audit fee data will be hand-collected from annual reports. As this research mainly focuses on the demand for audit services in general, the fiscal and non-audit related costs are included in the audit fee.

The quality of audit fee disclosure is subjective and therefore a proxy will be used to measure the quality. There are multiple ways to analyse narratives in annual reports, such as content analyses, linguistic analyses, analyses of ratings of experts and disclosure indices (Beattie, McInnes & Fearnley, 2004). This research uses a disclosure index to measure the quality of audit fee disclosure, as a disclosure index is an approach that is widely used to analyse narratives in annual reports. In this sense, disclosure quality refers to the extent of compliance to rules and regulations (D’Arcy & Grabensberger, 2003) and the extent of readability of the disclosure (Richards & van Staden, 2015). The index consists of 12 questions and is mainly based on the requirements of article 2:382a of the Dutch Civil Code, and the preferred presentation of disclosure8 issued by the Dutch professional organization for auditors (NBA). Additionally, some subjective items that aim to measure the readability of the disclosure were added. The

8_{NBA (2010): Handleiding Regelgeving Accountancy (p. 176/177)}

(22)

Research method | 22 index can be found in Table A.I in the Appendix. The disclosure index is an unweighted index (Beretta & Bozzolan, 2008), hence each component of the index results in 1 point if it is present in the disclosure and 0 points if the component is not present. As the total amount that can be achieved equals to 12 points, the total component score will be divided by 12 in order to end up with a range of 0% – 100%, whereby a higher score reflects a higher disclosure quality. Consistent with other disclosure index studies (e.g. Allegrini & Greco, 2013), the internal consistency of the disclosure index is tested by calculating the coefficient alpha, also known as the Cronbach’s alpha. According to Cronbach (1951), “when the coefficient alpha value is high, this suggests that all items are reliable and the entire test is internally consistent”. Hair, Ringle & Sarstedt (2013) state that a coefficient alpha should exceed 0.7 for an acceptable internal consistency of the disclosure index. The results show a coefficient alpha of 0.7494 for the audit fee disclosure index. This provides support that the items in the audit fee disclosure index are a valid capture of the same underlying construct.

3.3 Independent variables

The ownership data in this research will be hand-collected from the annual reports of Dutch listed firms. This differs from most of the research papers in the area of ownership characteristics, as most of the studies rely on a database to collect ownership data (e.g. Han et al., 2013; Mitra et al., 2007). As ownership data in databases is often incomplete and in order to achieve consistency and data reliability throughout the research, all ownership data is hand-collected from the annual reports. This is, furthermore, in accordance with Fogarty (2006) who states that at least one variable should be hand collected in order to safeguard data reliability. The ORBIS database will be used in addition, because it provides more insight in the specific types of ownership data and subdivides these types into different categories. For this research, these types are subdivided into four categories, which are summarized in Table II. To be clear, the ORBIS database is only used to identify the identity of the specific shareholder.

As it is mandatory for a substantial shareholder to report to the AFM a substantial shareholding that reaches, exceeds or goes below a pre-defined threshold9_{, this data can be used to perform}

cross-checks. These cross-checks with the AFM database are regularly executed to safeguard the data accuracy.

9_See:

(23)

TABLE II

Labels ORBIS categories

Category ORBIS categories

Institutional A (Insurance companies), B (Banks), E (Mutual & Pension fund, nominee, trust, trustee) F (Financial company)

Non-institutional C (Corporate ownership)

Managerial I (One or more named individuals or families) Family I (One or more named individuals or families)

3.3.1. Ownership concentration

Ownership concentration measures the extent of concentration of ownership and will be measured by the sum of all substantial shareholders who hold at least 5% of the outstanding shares. This is similar to study of AlQadasi & Abidin (2018). The data will be gathered by using the annual reports of the companies over the sample years.

3.3.2. Institutional and non-institutional ownership

Institutional and non-institutional ownership is measured by the percentage of common stocks held by (non)-institutional investors that own more than 3% of the outstanding shares. The data is obtained by using both the firm’s annual reports in combination with the ORBIS database.

3.3.3. Inside ownership

Comparable to the research design of Nelson & Mohamed-Rusdi (2015), managerial stock ownership will be measured by the percentage of outstanding shares held by managerial personnel. The data will be extracted from the annual reports in combination with the ORBIS database. Following prior research (e.g. Ali, Chen & Radhakrishnan, 2007; Ho & Khang, 2013), family firms are defined as those firms in which members of the founding family are taking positions in top management or the board, or own the largest block of shares of the company. Inside ownership will be measured by taking the percentage of shares held by family members (Donker et al., 2009) plus managerial personnel. Moreover, following the research of Ho & Kang (2013), a dummy variable will be created that equals 1 if the largest shareholder is an insider. The data is obtained using annual reports in combination with the ORBIS database.

3.3.4. Social media usage

Social media usage focuses on the activity of Dutch firms on the Twitter platform. Twitter does not provide statistics to retrieve the amount of tweets placed per year by an account. However, it does provide the total amount of tweets placed by a company and the date the account was

(24)

Research method | 24 registered on the Twitter site. Therefore, the social media usage will be measured by taking the total amount of tweets placed by the corporate account and divide this by the amount of years the account is active on Twitter. This will result in an average amount of tweets per year placed by the organization. This data is hand-collected for both listed and non-listed firms in the Netherlands.

3.3.5. Audit firm rotation

Audit firm rotation takes place when an auditee decides to hire a different auditor to audit the financial statements compared to the preceding year. Audit firm rotation is measured by taking a dummy variable that equals 1 if the auditee decided to switch its audit firm during the year. A value of 0 indicates that the auditee did not decide to switch to a different audit firm during the year, which is consistent with the research of Corbella et al. (2015).

3.4 Control variables

Consistent with prior studies (Hay et al., 2008; Carcello et al., 2002) there will be controlled for client size, -complexity and -leverage, as it is argued that these variables will have an increasing effect on the level of audit fee (disclosure).

Client size. It is generally accepted that on the basis of the audit fee model of Simunic (1980), client size influences the level of audit fees. According to Hackston & Milne (1996) there also exists a positive relationship between company size and disclosure. Therefore, in this research there will be controlled for client size. Following the research of Mitra et al. (2007), client size will be measured by the natural log of total assets.

Client complexity. According to Nelson & Mohamed-Rusdi (2015), client complexity refers to “the nature of the company that will lead to the need of extensive audit work”. It is, therefore, expected that client complexity will be positively related to audit fees. Following Niemi (2005), client complexity will be measured by the ratio of inventories and receivables to total assets.

Client leverage. Following the findings of Chaney, Jeter & Shivakumar (2004) it is expected that the leverage of the client has a negative effect on the audit fee, hence a high leverage reduces the audit fee. It is expected that this relation will also hold for the quality of the audit fee disclosure. Therefore, following Francis, Richard and Vanstraelen (2009), client leverage will be measured by dividing total debt by total assets.

(25)

TABLE III

Description of variables

Variable Description Measurement

Dependent variables

AUF Audit fee paid to the auditor Natural log of audit fees AFD Total score on items in the audit fee

disclosure index

Disclosure score / total score Score between 0 and 100

Independent variables

OC Ownership concentration of shareholders that own > 5%

Total % of shares hold by

shareholders that individually own > 5%

INSTO Ownership of institutional investors > 3%

Total % of shares hold by banks, insurance companies and pension funds

NINSTO Ownership of non-institutional investors > 3%

Total % of shares hold by non-institutional shareholders INO Ownership of managers and family

members > 3%

Total % of shares hold by insiders in the company

DINO A dummy variable representing a dominant insider

Dummy variable that equals 1 if the largest shareholder is an insider

SMU Social media usage Average number of tweets per year

AFR A dummy variable representing audit firm rotation

Dummy variable that equals 1 if the auditee changed its audit firm

Control variables

CSIZE Client size Natural log of total assets

CCOM Client complexity Inventories and receivables / Total

assets

CLEV Client leverage Total debt / Total assets

3.5 Research design

A panel data regression analysis is used to investigate the relationship between the different variables and audit fee (disclosure). Panel data entails multidimensional data that is gathered over multiple years. This is applicable to this research, as multidimensional data on ownership characteristics, social media usage, audit firm rotation, audit fee and audit fee disclosure is gathered over a time period of six years. The statistical models that are used are presented below:

𝐴𝑈𝐹_𝑖,𝑡 = 𝛽_1,𝑖+ 𝛽₂𝑂𝐶_𝑖,𝑡+ 𝛽₃(𝑂𝐶_𝑖,𝑡∗ 𝐴𝐹𝑅_𝑖,𝑡) + 𝛽₄𝐼𝑁𝑆𝑇𝑂_𝑖,𝑡+ 𝛽₅(𝐼𝑁𝑆𝑇𝑂_𝑖,𝑡∗ 𝐴𝐹𝑅_𝑖,𝑡) + 𝛽₆𝑁𝐼𝑁𝑆𝑇𝑂_𝑖,𝑡+ 𝛽₇(𝑁𝐼𝑁𝑆𝑇𝑂_𝑖,𝑡∗ 𝐴𝐹𝑅_𝑖,𝑡) + 𝛽₈𝐼𝑁𝑂_𝑖,𝑡

+ 𝛽₉(𝐼𝑁𝑂_𝑖,𝑡∗ 𝐴𝐹𝑅_𝑖,𝑡) + 𝛽₁₀𝐶𝑆𝐼𝑍𝐸_𝑖,𝑡+ 𝛽₁₁𝐶𝐶𝑂𝑀_𝑖,𝑡+ 𝛽₁₂𝐶𝐿𝐸𝑉_𝑖,𝑡 + 𝑟𝑎𝑛𝑑𝑜𝑚 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 + 𝜀

(26)

Research method | 26 𝐴𝐹𝐷_𝑖,𝑡 = 𝛽_1,𝑖+ 𝛽₂𝑂𝐶_𝑖,𝑡+ 𝛽₃𝐼𝑁𝑆𝑇𝑂_𝑖,𝑡+ 𝛽₄𝑁𝐼𝑁𝑆𝑇𝑂_𝑖,𝑡+ 𝛽₅𝐼𝑁𝑂_𝑖,𝑡+ 𝛽₆𝑆𝑀𝑈_𝑖,𝑡

+ 𝛽₇𝐶𝑆𝐼𝑍𝐸_𝑖,𝑡+ 𝛽₈𝐶𝐿𝐸𝑉_𝑖,𝑡+ 𝑟𝑎𝑛𝑑𝑜𝑚 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 + 𝜀

3.5.1. Fixed effects vs. random effects

When dealing with panel data, it is important to distinguish between random and fixed specific effects. The rationale for this comes from the fact that every individual company has its own specific characteristics that have the potential effect to influence the dependent variables. The Hausman test will be used to determine whether random or fixed effect model should be used. The outcome of this test shows that the random effect model is the most appropriate model. This model assumes that the unobserved effects are independent from the explanatory variables. Hence, a generalized least squares (GLS) regression is the most appropriate technique to test the hypotheses.

3.5.2. Data adjustments

In order to safeguard the quality and reliability of the data, some adjustments have been made to the data. First, to reduce the effect of outliers, the process of winsorizing is used for all the accounting data and the social media usage data. To minimalize the potential effect of outliers, winsorizing has been applied to the top and bottom 1% of the data. Numbers that are higher or lower than the upper or lower bound value will be replaced by the calculated value.

Secondly, to reduce the skewness of the data, the natural logarithm of client size is calculated to proxy for client size.

(27)

Results | 27 4. RESULTS

In this section the results and findings of this study will be elaborated. First, an overview of the descriptive statistics is presented. Subsequently, the results of the regression analyses will be presented, followed by the additional- and robustness tests.

4.1 Descriptive statistics

An overview of the descriptive statistics can be found in Table IV. A total of 377 firm observations for listed firms and 595 observations for listed- and non-listed firms are included in the data. The data shows that for the main model the mean of audit fees is EUR 4.913.780 with a standard deviation of EUR 8.494.060. The average score on the audit fee disclosure index is 70% for listed firms and 65% for the total sample. The average percentage of shareholdings hold by owners that individually own more than 5% of the outstanding shares is 40%, which indicates that ownership concentration has decreased with approximately 7% compared to the research of Kabir, de Jong, Marra & Röell (2001). Subsequently, the identity of the major shareholders is identified. The average percentage of shares hold by institutions is higher compared to non-institutions, 27% against 11% respectively. The average percentage of shares hold by insiders equals 8%.

TABLE IV

Descriptive Statisticsa

Variableb,c _N _Mean _SD _Min. _Max.

AUF 377 4.914 8.494 0 55.000 AFD 377|595 0,7009 |0,6528 0,1895 |0,1932 0,25 |0 1 |1 OC 377 0,4017 0,2160 0 1 INSTO 377 0,2675 0,1958 0 1 NINSTO 377 0,1132 0,1597 0 0,9761 INO 377 0,0802 0,1714 0 0,8250 DINO 377 0,1883 0,3915 0 1 SMU 377|595 2.314 |3.477 12.732 |19.863 0 100.874|172.050 AFR 377 0,1300 0,3367 0 1 CSIZE (log) 377|595 15,033 |14,686 2,189 |2,328 10,28 |7,24 20,55 |19,87 CCOM 377 0,2224 0,1788 0 0,8779 CLEV 377|595 0,5964 |0,6239 0,2085 |0,2163 0 |0 1 |1

a_{As a different population is used to test the hypothesis regarding social media usage, the descriptive}

statistics of those variables are listed after the vertical line ( | ).

b_{All accounting numbers are taken in thousands of Euro’s.}

c_{All accounting numbers are winsorized at the 1}st_{and 99}th_percentiles.

All variables are defined in Table III.

The Pearson correlation matrix can be found in Table V. The correlation statistics show that audit fee disclosure, non-institutional ownership, audit firm rotation and the three control variables are correlated with the level of audit fee. Institutional ownership and social media

(28)

Results | 28

TABLE V

Pearson Correlation matrix for independent variables

Variable AUF AFD OC INSTO NINSTO INO DINO SMU AFR CSIZE CCOM CLEV

AUF 1,000 AFD 0,2027*** 1,000 OC -0,0546 -0,1169 1,000 INSTO -0,026 -0,1287** 0,3689*** 1,000 NINSTO -0,0971* 0,0122 0,3942*** -0,2141*** 1,000 INO 0,0182 0,0175 0,3486*** -0,3951*** -0,1702*** 1,000 DINO 0,0111 -0,0007 0,2705*** -0,3794*** -0,1844*** 0,8974*** 1,000 SMU -0,0103 -0,0603 -0,0487 0,0592 -0,0597 -0,0570 -0,0613 1,000 AFR -0,0853* -0,0046 -0,0509 0,0370 -0,0212 -0,0581 -0,0450 -0,0459 1,000 CSIZE 0,6930*** 0,097* -0,1426*** -0,1123** -0,1120** 0,0199 0,0250 0,0443 -0,0148 1,000 CCOM -0,1204** -0,0027 0,1282** -0,0231 0,0975* 0,0854* 0,0964* -0,0610 0,0554 -0,4239*** 1,000 CLEV 0,1829*** -0,1086** -0,0577 0,0721 -0,0153 -0,1748*** -0,2134*** -0,1945*** 0,0327 0,5005*** -0.2336*** 1,000

*, ** and *** denote significance at the p < 0.10, p < 0.05 and p < 0.01 levels respectively. All variables are defined in Table III.

(29)

Results | 29 Figure III: Development of dependent variables

usage are correlated with audit fee disclosure. Noteworthy is the correlation between inside ownership and the dummy variable that indicates a dominant insider, as Hair, Black, Babin & Anderson (2010) argue that correlation coefficients between independent variables that exceed the value of 0.7 may indicate the presence of multicollinearity. Due to the high correlation of 0.8974, it is decided to drop the dummy variable that indicates a dominant insider, as this proxy attempts to measure the same construct as the percentage of inside ownership. Another way to test for multicollinearity is to calculate the corresponding VIF values. Following Hair et al. (2010), if the VIF values are less than 10, this indicates the absence of multicollinearity. The results show that all the VIF values are less than 10, with the highest VIF value of 7.64. Therefore, there are no signals that point to the presence of multicollinearity.

Figure III shows the mean value of the two dependent variables over the six years of the sample

period regarding listed companies. The development of the audit fee for listed firms shows an interesting development. In the period of 2012 – 2014, it shows a relatively steady trend. This is abruptly stopped in 2015, where the average audit fee shows a decrease of more than half a million of euros. In the financial year 2016 the audit fee remains stable, whereas in 2017 the average audit fee returns to the level of 2014. For the unlisted firms, the audit fee shows a relatively steady increase over the sample years. An overview of the proportion of different audit fees (audit-, non-audit- and tax fees) is presented in Figure IV. What stands out in this figure is the decrease to almost nil tax- and non-audit fees. This is in accordance with article 24b of the Wet toezicht accountsorganisaties (Wta) introduced in 2013 (with a transition period of two years), that prohibits audit firms of PIEs to both perform audit- and consulting activities.

(30)

Results | 30 Figure IV: Overview development specific audit fees of listed firms

Figure IV: Overview development of specific audit fees

The development of the average score on the audit fee disclosure index shows a more stable increase over the sample years. It becomes clear from Figure III that listed firms report a higher quality audit fee disclosure compared to unlisted firms. Apart from this, both listed and unlisted firms show a slight increase over the years. The increase indicates that the corporations that are captured in the sample produce a higher quality audit fee disclosure over time. To illustrate the difference between a low and high audit fee disclosure, three snapshots of audit fee disclosures are included in Table A.II in Appendix A. This table shows an example of a high-, moderate- and low quality audit fee disclosure. The first snapshot is subtracted from Koninklijke BAM Groep N.V.’s annual report of 2017. This disclosure scores high as it gives a good graphical overview of the different audit fees and also provides a qualitative explanation of the development of the audit fee. The second disclosure of Koninklijke DSM N.V. scores moderate, for example because the method of disclosure and a qualitative explanation are missing. The last disclosure of Stern Groep scores low as they provide very limited information regarding the audit fee. Moreover, the different audit fees are not provided, which results in a low disclosure score. These three examples are an indication of how the disclosure scores are determined. A full overview of the average disclosure score per sample firm is provided in

Table B.I in Appendix B.

4.2 Regression results – Audit fee hypotheses

In this section, the results of the regression models that are used to test the hypotheses are discussed. Table VI shows the outcomes of the results for the hypotheses related to audit fees. In the following section, these outcomes will be elaborated. The first models are aimed to test

The influence of ownership structure on audit fee and audit fee disclosure in the Dutch market

The influence of ownership structure on audit

fee and audit fee disclosure in the Dutch market

Stefan Blokzijl

University of Groningen

TABLE OF CONTENTS