Auditing in cities
the effect on audit fees and audit fees disclosure
Master thesis, Msc, Accountancy at the University of Groningen Faculty of Economics and Business
January 21, 2019 PAUL SPRUITENBURG Studentnumber: 2405105 Hertspieghelweg 44-2 1055KP Amsterdam tel.: +316 13 93 55 47 e-mail: p.m.spruitenburg.1@student.rug.nl Supervisor: drs. W. (Wilfred) Kevelam Co-assessor: prof. dr. R.L. (Ralph) ter Hoeven
Table of content
1 Introduction ... 3 2 Scientific contribution ... 5 3 Theoretical framework ... 7 3.1 Agency theory ... 7 3.2 Legitimacy theory ... 83.3 Voluntary disclosure theory ... 8
3.4 Urban theory ... 9
4.1 Audit fees and hypotheses development ... 10
4.1.1 Auditee complexity ... 10
4.1.2 Audit firm rotation ... 11
4.1.3 Audit location ... 11
4.2 Audit fees disclosure and hypotheses development ... 12
4.2.1 Auditee complexity ... 13
4.2.2 Audit firm rotation ... 13
4.2.3 Audit location ... 14
5 Research method ... 16
5.1 Data collection ... 16
5.2 Audit fees and audit fees disclosure ... 16
5.3 Independent variables ... 17
5.3.1 Auditee complexity ... 17
5.3.2 Audit firm rotation ... 18
5.3.3 Audit location ... 18
5.3.4 Control variables ... 18
5.4 Model development ... 18
6 Results ... 20
6.1 Descriptive statistics ... 20
6.2 Pearson Correlation Matrix ... 23
6.3 Results for the audit fees ... 23
6.4 Results for audit fees disclosure ... 27
6.5 Robustness tests ... 29
7 Conclusion ... 31
8 Limitations and future research ... 34
9 References ... 36
3
1
Introduction
Large companies are important to society. This can be considered an understatement if we take a look at recent Dutch political developments. Prime Minister Mark Rutte insisted that the dividend tax should be eliminated in order to keep the large companies headquartered in the country. Furthermore, he was convinced that the Netherlands would thrive after the elimination because it would invite other large companies to move their headquarters to the country. Eliminating dividend tax would mean diminishing tax gains, so the question remains how society would profit from this elimination. The answer is partly given by the Prime Minister, pointing at the potential loss of jobs in case of keeping the dividend tax.1 Although this might have become a problem, it was questioned by the public if this would outweigh the tax loss.
Another reason for wanting large companies to settle in the country is the positive impact they have on the surrounding society. At the beginning of 2018, Amsterdam ranked 19th when it comes to the number of headquarters of large global companies (67 company headquarters) situated in the city.2 Tokyo holds the biggest number with 613 headquarters, New York (217)
and London (193) complete the top three. When we compare these cities with the top three wealthiest cities on earth, the same names appear in New York ranking the wealthiest ($3000 billion), followed by London ($2700 billion) and Tokyo ($2500 billion). This number consists of the total of private wealth (assets minus any liabilities) held by all the citizens of the city.3
Large companies with their headquarters in the Netherlands could improve society’s wellbeing, but could also be a risk to that wellbeing. This was strikingly displayed by the large-scale money laundering at the biggest bank of the Netherlands, ING.4 In order to protect society, the audit firms should have a high standard of audit quality to prevent fraud schemes and money laundering.5 In previous literature, the audit fees are mentioned as an important indicator of audit quality because it implies more audit activities performed at companies. But why do audit
1 Mark Rutte over de dividendbelasting, VVD https://www.vvd.nl/nieuws/mark-rutte-over-de-dividendbelasting/ 2 Cities With The Most Fortune 500 Companies, Worldatlas https://www.worldatlas.com/articles/cities-with-the-most-company-headquarters.html
3 The Smartest Cities In The World In 2018, Forbes https://www.forbes.com/sites/iese/2018/07/13/the-smartest-cities-in-the-world-in-2018/#96ab2dc2efc0
4 Hoogste boete ooit voor ING: winst bank ging boven controle op witwassen, Volkskrant.
https://www.volkskrant.nl/nieuws-achtergrond/hoogste-boete-ooit-voor-ing-winst-bank-ging-boven-controle-op-witwassen~ba845cd1/
5 ING zaak werpt vraag over rol accountant bij fraude weer op, Accountancy van Morgen & FD
https://www.accountancyvanmorgen.nl/2018/09/17/ing-zaak-werpt-vraag-over-rol-accountant-bij-fraude-weer-op/
4 fees, and the disclosure of it in annual reports, differ between companies? To answer this question, we should consider several factors. An important influencer of the height of the audit fees is the complexity of the audited company, since it increases the risks and audit activities.6
But there is more to the audit fees. A few months ago, two researchers from Baruch College in New York, Ghosh and Siriviyakul (2018), published their research about the impact of audit firm tenure on audit fees. They found that audit fees decrease enormously in the first year after an audit firm rotation. They also found that, subsequently, audit fees increase on average by 13 percent between the first year and the second, by about 22 percent between the first year and the third and then creep up to 28 percent above the original fee by year 12.
Another potential influencer of the height of the audit fees is the location of the company. Large companies are moving their headquarters more and more to the cities because of the presence of knowledge and skills.7 “I don’t think you can find a company that’s built more than $10 billion in shareholder value in a three-year period that isn’t a bike ride away from a world-class engineering university.”, claims Scott Galloway, author of The Four (book about Amazon, Apple, Google, Facebook). It seems no coincidence that the cities with the most headquarters of big companies, Tokyo, New York and London, are also considered to be the smartest cities, with New York ranking 1st, London 2nd and Tokyo 4th.8 This presence of knowledge and skills
potentially influence the audit fees.
In this study, we examined the influence of being located in a big city on the audit fees paid by companies and the quality of audit fees disclosure. Furthermore, we looked into the influence of the auditee complexity and the audit firm rotation. We performed this research by collecting data from annual reports of both the biggest listed as well as the biggest non-listed companies in the Netherlands.
The main research question in this study is:
“Does the audit location have impact on audit fees and audit fees disclosure?”
6 Analysis of Audit Fees by Industry Sector, Audit Analytics https://www.auditanalytics.com/blog/analysis-of-audit-fees-by-industry-sector/
7 Headquarters checklist companies pick location, Wharton
http://knowledge.wharton.upenn.edu/article/headquarters-checklist-companies-pick-location/
8The Smartest Cities In The World In 2018, Forbes https://www.forbes.com/sites/iese/2018/07/13/the-smartest-cities-in-the-world-in-2018/#96ab2dc2efc0
5
2
Scientific contribution
This paper contributes to existing literature in several ways. We will look into four of them. To begin with, this study takes into account a scarcely examined subject: the moderating effect of the audit location. According to Hay (2013), some studies, especially in the UK, include a measure for “city effect”: companies audited in expensive cities have higher audit costs. This can be explained by the existing competition (Buzby, 1975; Firth, 1979), the available facilities (Boyle, 1990; Klier and Testa, 2002), the learning opportunities (Duranton and Puga, 2004) and the matching possibilities present in cities (Duranton and Puga, 2004). As mentioned before, it is no coincidence that the cities with the most headquarters of big companies, Tokyo, New York and London, are also considered to be among the top four smartest cities.9 Despite these few studies, little research has focussed on the effect of the audit location on audit fees.
In addition, this study will not only investigate the audit fees but also the company’s disclosure regarding the paid audit fees. Whereas audit fees are a common research topic, little research has been done regarding the quality of audit fees disclosure. Since 2008, companies in the Netherlands are obliged to disclose the amount paid to the audit firms. The information is relevant for several stakeholders because it is an indication for the auditor’s independence (DeAngelo, 1981; Lai, 2009). Although disclosure regarding the paid audit fees is to a certain degree mandatory, the company has the opportunity to disclose other relevant information about the audit fees. This voluntary aspect results in different qualities of audit fees disclosure. But why should a company disclose more than necessary? The answer is found in the research of Dieleman (2008) and Langendijk (2012). They state that the height of the audit fees in comparison to the height of the non-audit fees gives an insight into the perceived independence of the auditor. Defond et al. (2002) agree and explain that non-audit services, as an addition to the audit, increase an economic dependency between the audit firm and the client. Companies should also disclose voluntarily because voluntary disclosure expresses the transparency and accountability of management in conducting business (Akhtaruddin and Haron, 2010; Li et al., 2012). To sum it all up, the relevance of the audit fees disclosure seems clear and this study tries to fill the existing gap in literature regarding the determinants of the audit fees disclosure.
9The Smartest Cities In The World In 2018, Forbes https://www.forbes.com/sites/iese/2018/07/13/the-smartest-cities-in-the-world-in-2018/#96ab2dc2efc0
6 Also, examining companies in the Netherlands is interesting because of the early legislation regarding mandatory audit firm rotation. Audit firm rotation has been found to influence the height of the audit fees paid by a company (Ghosh and Siriviyakul, 2018). Hay (2013) also suggests that auditors with a short tenure charge lower fees. The audit firms compete with the height of the audit remuneration when it comes to attracting new clients. Consequentially, audit firms are repeatedly accused of “low-balling”: discounting the initial audit engagement to earn the right to future quasi-rents of audit fees (DeAngelo, 1981; Beck et al., 1988). DeAngelo states that low-balling has a negative impact on auditor’s independence, resulting in a lower audit quality. In order to mitigate familiarity between the auditor and the auditee, various governments introduced legislation regarding auditor’s independence. In Europe, listed companies need to comply with new legislation, resulting in mandatory audit firm rotations every 10 years. The Netherlands was an early adapter with regard to this legislation. As a result,
all listed companies had to switch from their audit firm before January 2017 (NBA, 2016). Using a sample period between the years 2012 until 2017, this study is able to examine the impact of audit firm rotation by looking at the years before and after the rotations, including the potential quasi-rents obtained in the consecutive years after the rotation.
Finally, this study contributes to science because it looks at both listed and non-listed companies. Prior research regarding audit fees often used US-listed companies (Hay et al., 2006). The distinction between listed and non-listed is especially interesting when focusing on the Netherlands, since the Dutch mandatory audit firm rotation only applies to the listed companies and, therefore, audit firm rotations of non-listed companies are voluntary in nature. Both mandatory and voluntary audit firm rotations were subject to research in the past, but, to the best of my knowledge, little research measured both mandatory and voluntary audit firm rotation in one and the same study. The outcomes can differ since a voluntary change of audit firm, in contrast to a mandatory change, can be motivated by the client.
7
3
Theoretical framework
The theoretical framework used in this study is based on theories concerning audit fees, audit fees disclosure and their various determinants.
3.1 Agency theory
This study used the Agency theory as a theoretical basis. The agency problem was already highlighted by Adam Smith in 1776 (Bendickson et al., 2016) and is studied over time by various researchers (Berle and Means, 1932; Jensen and Meckling, 1976; Fama and Jensen, 1983). Jensen and Meckling (1976) define the agency relationship as a contract under which a person (the principal) engages another person (the agent) to perform some service on his behalf, which involves delegating some decision-making authority to the agent. This separation causes an information gap between principal and agent, since the principal is not able to gather the same knowledge from outside the company as the manager from inside the company. Jensen and Meckling (1976) state that possible problems occur with this so-called information asymmetry between the agent and the principal, because the agent tries to maximize his own interests. The principal can limit this divergence of interest by monitoring the agent, for example with the audit of the financial statements of the company. Prior research shows that auditors provide stakeholders with two important opportunities: knowledge and assurance. Thus, it helps reduce information asymmetry between the principal and the agent (Muller and Riedl, 2002; Hakim and Omri, 2010). According to Varici (2013), the audit is an effort to demonstrate that financial statements are reliable, and that managers behave properly and honestly. In order to substantiate these promises, the audit firm performs audit activities. These activities come with a price since it requires knowledge, expertise and personnel and therefore audit firms charge audit fees to the company. The audit fees are examples of agency costs (Williams, 1988) because auditors are likely to spend much more time inspecting managers’ activities in cases where agency problems are suspected (Dopuch and Simunic, 1982; Nikkinen and Sahlström, 2004; Leventis et al., 2011). According to Defond (1992), the audit fees should be at a sufficient level to prevent information asymmetry. The level of prevention depends on the audit quality. This is defined by DeAngelo (1981) as the joint probability of detecting and reporting financial statement errors. The audit quality depends on the auditor's independence and a higher amount of non-audit fees threatens this independence (Mautz and Sharaf, 1961; DeAngelo, 1981; Parkash and Venable, 1993; Firth, 1997; SEC, 2000). It is therefore plausible
8 to assume that higher audit fees and higher quality of the audit fees disclosure indicate a higher auditor independence and, subsequently, a higher audit quality.
3.2 Legitimacy theory
Legitimacy theory broadens the Agency theory by stating that the agent is not only obliged to disclose information to the principal, but also to other stakeholders since there is a social contract between the company and society (Magness, 2006). The social contract is an understanding between a company and society: a company receives permission from society to operate and is ultimately accountable to that society for how it operates, because society provides corporations the authority to own and use natural resources and to hire employees (Deegan, 2004). In other words, society allows the company to exist, but only if the company complies with societal expectations. In this manner, the company gains society’s trust. Audit firms play a crucial role in fostering trust in financial reporting (Power, 1994, 2003; Financial Reporting Council (FRC), 2008; Al-Ajmi 2009; Lin and Liu, 2010; Solomon, 2010). Thus, the paid audit fees originate from the need of the company to gain legitimacy from society. Also, since the Legitimacy theory is based on society’s perception, the agent is compelled to disclose information that would positively influence the external users’ opinion about his company (Cormier and Gordon, 2001). As mentioned earlier, a higher amount of non-audit fees threatens auditor independence. In order to stay legitimate and keep society’s trust, the company should disclose the audit fees and non-audit fees in the annual report.
3.3 Voluntary disclosure theory
For medium to large companies in the Netherlands it is mandatory to disclose information regarding the paid audit fees. However, criteria regarding the disclosed information are vague, giving opportunity to companies to vary in the quality of disclosure. Since the companies are able to determine the amount of audit fees disclosure, they find themselves in the field of the Voluntary disclosure theory. The Voluntary disclosure theory may be explained by the Agency theory (Firth, 1980; Chow and Wong-Boren, 1987; Cooke 1989, 1991, 1992; Hossain et al., 1994). The manager (agent), in the knowledge that the shareholder (principal) will try to reduce the information asymmetry through monitoring, will have an incentive to try to convince the shareholder that he is acting optimally, and voluntary disclosure may be a means of achieving this. Singhvi and Desai (1971) agree with these hypotheses by stating that the quality of disclosure will affect the quality of investment decisions made by investors. The fact that firms
9 provide more information than required by regulation is, in fact, a quintessential accounting problem because it raises questions about the role of accounting regulation (Verrecchia, 1990).
3.4 Urban theory
The Urban theory holds the assumption that an urban industry gains a great advantage from the fact that it offers a constant market for skill (Marshall, 1890). According to Klier and Testa (2002), the central function of corporate headquarters is the acquiring and distributing of information. They explain that stakeholders require corporate headquarters to stay ahead of developments in their industries. Meanwhile, competition requires that companies must adopt new production technologies and management strategies. Both sides will often require a flow of information and administration to a wide-ranging geography of operations. Cities provide various opportunities to distribute information in an easy manner, for example by airports and major highways (Dow Jones, Inc., 1977). Furthermore, airports allow their managerial personnel to travel and direct their own operations, as well as to interact with others in their industry at conventions and trade shows (Boyle, 1990). Also, major airports bring suppliers and customers into the city.
Positive effects of urban locations also include processes of matching and learning (Duranton and Puga, 2004). Better matching is a result of the local availability of many alternative choices to purchasers and sellers of services or labor. If the workforce grows and the number of firms increases, the average worker is able to find an employer that is a better match for his or her skills and, therefore, the salaries of the average worker increase. Another advantage of cities is based on the learning process. A fundamental feature of learning is that it often involves interactions with others and many of these interactions have a face-to-face nature (Duranton and Puga, 2004). Glaeser (1999) states that the number of meetings between skilled and unskilled workers increases with city size. Therefore, by bringing together a large number of people, cities stimulate the learning process. DeAngelo (1981) states that larger audit firms deliver higher audit quality because they are more independent from their clients. Furthermore, Kim (1989) shows that, as the size of the local market increases, the investment in specific skills increases relatively more than that in general skills. This presence of both expertise and more specialized auditors in cities results in higher audit quality. This can be interpreted as increased specialization due to an increase in city size. Specialization not only leads to a higher audit quality, but the audit fees also increase. Ferguson et al. (2003) find higher audit fees for audit firms that are the industry specialists in the city.
10
4
Hypotheses development
In this chapter, we developed the hypotheses used in this study, concerning audit fees and audit fees disclosure.
4.1 Audit fees and hypotheses development
The determinants used in this study are based on prior research regarding the audit fees (Simunic, 1980; Hay et al., 2006; Francis and Yu, 2009; Hay, 2013; Bills et al., 2014; Corbella et al., 2015). Simunic (1980) looked at various factors, for example firm size, complexity, industry and audit risk. He proposed a model that is still broadly accepted and used by recent research (Hay et al., 2006; Francis and Yu, 2009; Hay, 2013). Hay et al. (2006) extended the research of Simunic by looking at various client attributes, auditor attributes and engagement attributes. Francis and Yu (2006) studied the Big Four accounting firms and the impact of firm size on the office level. Hay (2013) updated his prior research and concluded that future research should focus more on a few determinants that seem to influence the audit fees.
4.1.1 Auditee complexity
One of the determinants regarding the audit fees researched in this paper is auditee complexity. Findings in earlier research have consistently shown the audit fees to be significantly associated with complexity. Bills et al. (2014) state that audit costs are likely to be higher in complex industries because complexity increases the risk of material misstatement, which requires additional audit effort, investments in technology, and more experienced personnel to respond to these additional risks. In addition, complexity requires more time and effort to be spent by the auditor in planning, coordinating and executing the audit function (Gerrard et al, 1994). Rubin (1988) adds that more complex organizations require additional audit time because of the increased variety of transactions and internal control systems that need to be audited. Concluding, the increasing complexity of companies can make it more difficult for external auditors to audit and detect fraud (Firth, 1985). Building further on this research, we expect that there is a positive relation between audit complexity and the audit fees. The hypothesis is as follows:
H1: There is a positive relation between auditee complexity in the Netherlands and the audit fees.
11 4.1.2 Audit firm rotation
The research regarding audit firm rotation is broad, especially with respect to voluntary audit firm rotation. According to DeAngelo (1981), the Security and Exchange Commission (SEC) is concerned that situations where the audit firm agrees to a fee significantly less than is normal in order to obtain the client, may constitute a lessening of independence. In line with the Agency theory and the Legitimacy theory, the potential loss of auditor’s independence and expertise could negatively influence the decision-making process of the shareholders and stakeholders. Nashwa (2004) validated the theory by concluding that voluntary change is often associated with audit failures like financial distress and fraud. Also, previous research showed that mandatory rotation can lead to a loss of previously gathered client-specific knowledge (Johnson and Lys, 1990; Shu, 2000). Research on this topic in relation to the audit fees is done by Simon and Francis (1988). They found that the audit fees decreased after a new audit firm was appointed. When a client switches between audit firms, the client should initially enjoy lower audit fees because potentially new audit firms low-ball or discount the initial audit engagement (DeAngelo, 1981; Beck et al., 1988). To my knowledge, there is little research done with regard to the distinction between voluntary and mandatory audit firm rotation, especially in Europe. The hypotheses are as follows:
H2: There is a negative relation between voluntary audit firm rotation and the audit fees. H3: There is a negative relation between mandatory audit firm rotation and the audit fees.
4.1.3 Audit location
According to the Urban theory, geography plays a crucial role in the acquisition of skills. Jovanovic and Rob (1989) state that unskilled workers can only become skilled after face-to-face interactions with skilled workers. Hence living in a city enhances the chance to acquire skills. More recently, Choi et al. (2007) show that the geographical location of the auditor’s office is an important engagement-specific determinant of audit quality. The existence of expertise in cities is not limited to audit firms, other established companies also attract better employees in urban areas. As a result, these companies become more complex and have to be audited by an audit firm that has enough expertise to understand the company. Therefore, we expect that the size of the audit location positively influences the relation between auditee complexity and the audit fees. In addition, audit firm rotations took place at both listed and non-listed companies in the Netherlands. As explained, the nature of the rotation differs, but there is another distinction to be made. The majority of the companies is located in cities, but a considerable number resides in smaller towns and even in outlying villages. These differences
12 potentially influence the impact of audit firm rotation on the audit fees, because of the higher number of audit firms in cities, enabling competition between firms. This increase in competition results in a decrease in the audit fees (Kinney, 1986; Maher et al, 1992). Furthermore, Casterella et al. (2004) state that the audit fees are lower if clients have more bargaining power due to more competition. Therefore, the impact of audit firm rotation on the audit fees is influenced by the audit location. The following hypotheses are formulated:
H4: There is a positive relation between the size of the audit location and the audit fees. H5: The size of the audit location has a strengthening effect on the relation between auditee complexity and the audit fees.
H6: The size of the audit location has a strengthening effect on the relation between mandatory audit firm rotation and the audit fees.
H7: The size of the audit location has a strengthening effect on the relation between voluntary audit firm rotation and the audit fees.
4.2 Audit fees disclosure and hypotheses development
Previous literature barely examined the quality of the disclosure of audit fees, probably because it is a recent phenomenon. Research into this phenomenon might be important since more disclosure of audit fees is associated with enhanced audit quality (Chen, 2016). Due to this existence of an apparent gap in literature concerning audit fees disclosure, we rely on more general theories regarding disclosure. In line with the Agency theory, there is a greater possibility of implicit “auditor-manager side-contracting” when fees are undisclosed, because of the risk that managers may provide auditors with financial incentives to give favorable reports (Acemoglu and Gietzmann, 1997). The Dutch government tends to mitigate this risk with the mandatory disclosure legislation. Furthermore, in order to earn the shareholders’ trust, managers have the opportunity to enhance the quality of the audit fees disclosure. This is in line with the Voluntary disclosure theory. The need to improve the quality of audit fees disclosure depends on several factors. Prior research focused on, among others, firm size, listing status, earnings margin and country of incorporation (Singhvi and Desai, 1971; Wong-Boren, 1987; Meek et al., 1995). In this study, we extended this research by looking at auditee complexity, audit firm rotation and audit location.
13 4.2.1 Auditee complexity
Increased disclosure by a firm could be the result of a more complex business (Cooke, 1981). Operating in a number of geographical areas, including other countries, increases the amount of information to be controlled by a firm. Also, according to Cooke (1989), the amount of information necessary for the managers’ bonding activities towards the shareholders increases if the firms’ operations become more international. Wallace et al. (1994) suggested that firms in a specific industry might face particular circumstances that may influence their disclosure practice. Furthermore, companies are forced to comply with the usual disclosure practices of countries in which they operate. Based on previous literature, the hypothesis is:
H8: There is a positive relation between auditee complexity and the quality of audit fees disclosure.
4.2.2 Audit firm rotation
The influence of audit firm rotation on audit fees disclosure has been scarcely examined. One of the few studies on this topic comes from Myers, Myers and Omer (2003). They show that an audit firm rotation might be a good idea because they find that the quality of the disclosures decreases with an increase of audit firm tenure. Additionally, the possibility of auditor-manager side-contracting, mentioned by Acemoglu and Gietzmann (1997), is probably higher when the auditor is less independent. Therefore, audit firm rotation may enhance the quality of audit fees disclosure since rotation results in more independence (Firth et al., 2012). Subsequent studies found a significant relation between disclosure and listing status of a company (Firth, 1979; Cooke, 1989a, 1989b, 1991, 1992, 1993; Malone et al., 1993; Hossain et al., 1994; Wallace et al., 1994; Hossain et al., 1995). Since mandatory audit firm rotation occurs at the listed companies, the influence of mandatory audit firm rotation on audit fees disclosure is expected to be positive. Additionally, voluntary audit firm rotation is also expected to have a positive influence on audit fees disclosure. The hypotheses are:
H9: There is a positive relation between voluntary audit firm rotation and the quality of audit fees disclosure.
H10: There is a positive relation between mandatory audit firm rotation and the quality of audit fees disclosure.
14 4.2.3 Audit location
Previous research argues that most innovations develop in cities (Jacobs 1969; Bairoch 1988). The packing of individuals and industries into close quarters provides an environment in which ideas flow quickly due to face-to-face interaction (Glaeser et al., 1992). Jacobs (1969, 1984) argues that interpersonal communication in cities helps to innovate. Without an opportunity to learn from others, there would be little reason for people to pay high rents just to work in a city. As a consequence of this preference for cities by potential employees, the companies grow and competition increases. Porter (1990) favors local competition because it speeds up the adoption of technology and stimulates innovation. Furthermore, other research examined issues such as how competition affects disclosure (e.g., Verrecchia, 1983; Darrough and Stoughton, 1990). Most large companies are located in big cities. In 2000, the fifty most crowded city areas in the US were home to 87% of all large company headquarters (Klier and Testa, 2002). It is argued in literature that firm size is a variable which can proxy for competitive advantage (Buzby, 1975; Firth, 1979; Leftwich et al., 1981; Ball and Foster, 1982). Firth (1979) suggests that gathering information is an expensive endeavour and perhaps larger firms can afford such expenses better. Furthermore, smaller firms may feel that more disclosure of their activities will put them in a competitive disadvantage with other, larger companies in their industry. On the other hand, Lang and Lundholm (1993) and McKinnon and Dalimunthe (1993) state that large firms have incentives to voluntarily disclose more information than smaller firms in order to enhance company value, because non-disclosure may be viewed by (potential) investors as bad news. Also, the larger the company, the more likely it will be able to attract a wide variety of highly skilled individuals necessary to introduce more sophisticated reporting systems that can disclose more and better information (Cooke, 1989). There may also be greater demands on large companies to provide information for customers, suppliers and analysts as well as public in general. As mentioned earlier, the expertise and skills of audit firms are also developed better in cities in comparison to rural areas. These skills possessed by high quality audit firms increase the precision of the financial information reported by the auditee. Verrecchia (1990) shows that the quality of the information encourages voluntary disclosure. With more expertise in the city, it is plausible to assume that the complexity of the company has more influence on the quality of the audit fees disclosure if the company is located in a big city. Another reason for the audit location to have a strengthening effect on the relation between audit firm rotation and audit fees disclosure is the fact that the audit firm has more expertise and therefore is expected to pay more attention to publicly relevant disclosure. The hypotheses are as follows:
15 H11: There is a positive relation between the size of the audit location and the quality of audit fees disclosure.
H12: The size of the audit location has a strengthening effect on the relation between auditee complexity and the quality of audit fees disclosure.
H13: The size of the audit location has a strengthening effect on the relation between mandatory audit firm rotation and the quality of audit fees disclosure.
H14: The size of the audit location has a strengthening effect on the relation between voluntary audit firm rotation and the quality of audit fees disclosure.
16
5
Research method
This section describes the methods used to perform the research. We gathered quantitative data from annual reports and used Company.info to obtain the SIC codes and the number of (foreign) subsidiaries.
5.1 Data collection
Regarding the listed companies located in the Netherlands, we expanded the existing database of Kevelam et al. (2017), which contained data from 75 listed companies over the period 2012-2016, by adding data from the annual reports with respect to the financial year 2017. These companies are listed at the AEX, AMX or AScX respectively, according to their market cap. The 75 biggest non-listed companies in the Netherlands were selected out of all companies with more than 500 employees, resulting in 2152 companies. After that, 102 companies were removed because they were listed companies and 1043 companies were removed because they did not yet publish their financial statements of 2017. Out of the remaining 1007 unlisted companies, the 75 largest companies, in terms of the number of employees, were selected. We gathered information from the annual reports of these companies from the period between 2012 and 2017. We added the data to the above-mentioned database. An overview of the listed as well as the non-listed companies can be found in Appendix A and Appendix B respectively. The total of company years in the sample amounts to 900. After excluding the company years without information about the audit fees or audit fees disclosure, the remaining sample consisted of 753 company years. An overview is given in Appendix C. Other relevant data was gathered by using Company.info.
5.2 Audit fees and audit fees disclosure
The amount of the audit fees was part of the data derived from the annual reports. There is a distinction made between the audit fees, audit-related fees, tax fees and other fees (Kevelam et al., 2017). We added up the amounts of the audit fees and the audit-related fees since the sum equals the amount payed to the audit firm regarding the audit of the financial statements and, therefore, represents the audit remuneration in the best manner. In contrast to the audit fees, the quality of the disclosure of the audit fees is not easily measured. To be able to measure and compare the quality of the audit fees disclosure between the companies, we introduced a disclosure index. According to Marston and Shrives (1991) it is clear that the construction of an index is difficult and generally involves subjective judgement on the part of the researchers.
17 To reduce this subjectivity, we used an unweighted disclosure index. The list of disclosure items is based on the items mentioned in article 390.3 from the Dutch Guidelines of Annual Reporting (Richtlijnen Jaarverslaggeving). The used disclosure index consisted of twelve equally weighted questions that form the criteria for the disclosure score of the company. The index can be found in Appendix D.
5.3 Independent variables
Both the audit fees and audit fees disclosure have three determinants: auditee complexity, audit firm rotation and audit location. These determinants have been measured in multiple ways in order to make the results more robust. We added the size of the auditee (SIZ) as a control variable in order to make the results more reliable. An overview of the variables is given in table 1.
Table 1: description of all the research variables.
5.3.1 Auditee complexity
In literature the measurement of auditee complexity has proved to be difficult, with numerous of measurement possibilities available. For instance, it is possible to assess the complexity by examining organizational charts, product diversity, diversity of the salesforce, type and diversity of financing arrangements, the geographic spread of the auditee, and the diversity of the legal structures (Simunic, 1980; Palmrose 1986; Gerrard et al., 1994; Hay, 2013). In most cases, these measures are unobservable. In this study, the measure is based on the number of subsidiaries (SUB), the number of foreign subsidiaries (FSU) and the number of SIC codes (SIC). This information is available in The Netherlands using Company.info. We used the sum of the original SIC codes and the SIC codes added by Company.info since this number approaches a complete view of the company’s complexity the most.
Description of variables
AF Audit fees Interval
AFD Audit fees disclosure Interval
AMS Location Amsterdam Dummy
BCI Location four biggest cities Dummy
CPO City population Interval
AFR Audit firm rotation Dummy
SIC Number of sic codes Interval
SUB Number of subsidiaries Interval
FSU Number of foreign subsidiaries Interval
18 5.3.2 Audit firm rotation
The influence of audit firm rotation is measured in two separate ways: rotations of listed companies and rotations of non-listed companies. The first rotation is to be labelled as a mandatory rotation due to the previously mentioned legislation regarding audit firm rotations. This legislation is only applicable to listed companies and, therefore, switches between audit firms by non-listed companies are voluntary in nature. We used a dummy variable for both voluntary audit firm rotation and mandatory audit firm rotations. The sample consisted of 59 mandatory rotations and 29 voluntary rotations.
5.3.3 Audit location
The listed and non-listed companies are headquartered across the country. The influence of the location of the audit was measured in multiple ways. First of all, we measured the “city effect” mentioned by Hay (2013) by using a dummy variable for the four biggest cities (BCI), since these cities together form the beating heart of “Randstad”; the biggest business region in the Netherlands. A similar dummy variable has been used to measure the city effect of Amsterdam (AMS), with value being 1 if the company is headquartered in the capital and 0 if this is not the case. In previous literature, the same has been done for companies headquartered in London (Kharuddin and Basioudis, 2014). Finally, we looked at the population of the city. The city population (CPO) was recorded by using the 2017 report of CBS (Centraal Bureau Statistiek).
5.3.4 Control variables
Simunic (1980) and Hay (2013) mentioned the size of an audit as the single biggest influencer regarding the audit fees. Therefore, the size of the audit was added as a control variable (SIZ), in order to measure the influence of the dependent variables.
5.4 Model development
Since the dataset consisted of both time series and cross-section data, we worked with panel data. Due to the exclusion of company years without data about the audit fees and the audit fees disclosure, the dataset is unbalanced regarding the years per company. Since our dataset had a remaining 753 company years, the impact on the outcomes of the analyses was limited. Panel data can be analyzed with either a fixed effects model or a random effects model. We performed a Durbin-Wu-Hausmann test to determine the best fitting model and concluded that the random
19 effects model was preferable to the fixed effects model. To be sure a Pooled OLS regression would not be better in comparison to a random effects GLS regression, we performed a Breusch-Pagan Lagrangian multiplier test. The following models were tested with the random effects model:
𝐴𝐹𝑖,𝑡 = 𝜇 + 𝐵1𝑆𝑈𝐵𝑖,𝑡 + 𝐵2𝑆𝐼𝐶𝑖,𝑡+ 𝐵3𝐴𝐹𝑅𝑖,𝑡 + 𝐵4𝐵𝐶𝐼𝑖,𝑡 + 𝐵5𝑆𝐼𝑍𝑖,𝑡 + 𝑈𝑖+ 𝑊𝑖,𝑗
20
6
Results
The results section is built up as follows: first, the descriptive statistics are shown. In addition, in order to estimate the collinearity between variables, a Pearson Correlation Matrix is included. After that, the results for the audit fees and audit fees disclosure, respectively, are discussed. Finally, we showed the results of the robustness tests.
6.1 Descriptive statistics
An overview of the statistics is shown in table 2. A distinction is made between “overall”, “between” and “within”. The between statistics refer to the cross-sectional differences, with a sample of 141 companies. The within statistics refer to the time series differences, the differences within a company over the years, with an average sample of 5,3 years per company. The within deviations equal to 0 when it comes to SIC codes, Amsterdam and Four biggest cities. This is due to the fact that these variables were time invariant. The audit firm rotation had a mean of 0,117, which means that, on average, 1 out of the 10 companies switched from audit firm.
Figure 1 shows the development of the average audit fees in the period between 2012 and 2017. We separated the development of the audit fees for listed as well as non-listed companies. The primary axis shows the amounts of the audit fees that correspond to the listed companies, while the secondary axis corresponds to the trend line of the non-listed companies. Both trends show a decrease in the fees until 2016. In 2017, the audit fees rose significantly for listed companies as well as for non-listed companies. An explanation for listed companies could be found in the number of audit firm rotations. In the years 2014, 2015 and 2016 the audit fees reached the lowest amounts while the highest number of mandatory audit firm rotations (12, 15 and 24, resp.) took place. After that, in 2017, the audit fees increased rapidly while the number of audit firm rotations decreased (5). In 2012 and 2013, when the audit fees were relatively high, only a few mandatory audit firm rotations took place, 2 and 4 respectively.
The development of the audit fees disclosure is also shown in figure 1. Both trends show an increase in audit fees disclosure. Although this is reassuring from society’s point of view, the audit fees disclosure could still be at a higher level, especially with respect to the non-listed companies. Dutch legislation, Article 382a (Burgerlijk Wetboek), requires companies to disclose information about the first four questions and the 10th question of the disclosure index
21 (Appendix D). That means, if a company obliges to the law, the disclosure must be at least 42% already. When we looked at the listed companies in 2017, we concluded that at least two companies failed to meet the required disclosure standard of 42%. In comparison, the non-listed companies scored lower with at least 9 companies which failed to meet the 42% benchmark. These numbers might be even higher, because companies are able to score 42% on the voluntary components of the disclosure index. Table 1 shows that listed companies voluntarily disclosed, on average, at least 29%, while non-listed companies disclosed at least 20%. The difference might be explained by Singhvi and Desai (1971). They suggest that smaller companies do not usually earn money in the securities market and, therefore, cannot benefit from a better quality of disclosure.
Table 2: description statistics of the research variables
* amounts x € 1000
Variable Specifics n Mean Standard
deviation Minimum Maximum
Audit fee*
overall 753 3258,15 6461,41 24,90 53.000 between 141 6069,79 46,68 50.666,67 within 5,3 1306,26 -4908,52 22.091,48
Audit fees disclosure
overall 753 0,636 0,203 0,090 1 between 141 0,181 0,144 0,985 within 5,3 0,090 0,220 1,123 Subsidiaries overall 753 96,236 206,127 0 1954 between 141 200,785 0 1954 within 5,3 0,827 76,736 102,7364 Foreign subsidiaries overall 753 75,754 171,950 0 1560 between 141 167,760 0 1560 within 5,3 0,796 57,004 82,004 SIC codes overall 753 2,883 1,302 1 6 between 141 1,340 1 6 within 5,3 0 2,883 2,883
Audit firm rotation
overall 753 0,117 0,321 0 1 between 141 0,129 0 0,667 within 5,3 0,301 -0,550 0,950 Amsterdam overall 753 0,412 0,492 0 1 between 141 0,496 0 1 within 5,3 0 0,412 0,412
Four biggest cities
overall 753 0,595 0,491 0 1 between 141 0,492 0 1 within 5,3 0 0,595 0,595 City population overall 753 483.197 348.472 15.156 844.947 between 141 350.049 15.156 844.947 within 5,3 0 483.197 483.197 Auditee size* overall 753 7.525.550 10.900.000 268.257 34.800.000 between 141 10.500.000 268.257 34.800.000 within 5,3 2.024.255 -14.300.000 29.800.000
22 Figure 1: development of the audit fees and audit fees disclosure
Figure 2 shows the development of the audit fees/non-audit fees ratio. The audit fees consist of the audit fees and the other audit-related fees paid by the company to the same audit firm. The non-audit fees are calculated by adding the tax fees and the other non-audit fees paid by that same company to the same audit firm. Regarding the listed companies, the government introduced legislation in January 2013 regarding the separation of audit activities and non-audit activities at the same company, in order to enhance the auditor’s independence. This resulted in an increase to almost 100% in 2017. The decrease in 2013 and 2014 is explained by the Dutch Authority for Financial Markets (AFM). The AFM published a paper in 2013 in which they concluded that the audit firms approached listed companies to sign non-audit fees contracts for the years 2013 and 2014. The audit firms used a loophole in legislation, because the non-audit were tolerated in the first two years after the introduction of this specific legislation. The non-listed companies remained stable over the years.
Figure 2: the development of the relation between audit fees/non-audit fees
71% 62% 50% 55% 60% 65% 70% 75% 2012 2013 2014 2015 2016 2017
Development of audit fees disclosure
Listed companies Non-listed companies €1.400 €1.450 €1.500 €1.550 €1.600 €1.650 €1.700 €1.750 € 4.100 € 4.200 € 4.300 € 4.400 € 4.500 € 4.600 € 4.700 € 4.800 € 4.900 2012 2013 2014 2015 2016 2017 T h o u san d s T h o u san d s
Development of the audit fees
Listed companies Non-listed companies 98% 88% 75% 80% 85% 90% 95% 100% 2012 2013 2014 2015 2016 2017
Audit fees/non-audit fees development
Listed-companies Non-listed companies
23 6.2 Pearson Correlation Matrix
In order to obtain reliable statistical evidence, we measured the auditee complexity and the audit location in three different ways. As mentioned, the complexity was measured with the number of subsidiaries (SUB), the number of foreign subsidiaries (FSU) and the number of SIC codes (SIC). The different proxies have been checked for correlation via Pearson’s correlation coefficient. Table 3 shows a matrix with the different correlation coefficients between variables. Obviously, the SUB variable had a high correlation with FSU (0,99). SUB also had a high correlation with the audit fees (0,783), therefore SUB was expected to be a significant determinant of the audit fees. With respect to the regressions, variables with a correlation coefficient higher than 0,8 must be exempted. Therefore, the influence of variable FSU was estimated separately and used to test for robustness. The same is applicable to the audit location, where we measured the influence of the four biggest cities (BCI), Amsterdam (AMS) and the city population (CPO). Due to the high correlation with BCI, AMS (0,838) and CPO (0,937) were measured separately and were used to make the results more robust, regarding the influence of the audit location.
6.3 Results for the audit fees
The results regarding the audit fees are shown in table 4. As described in the methodology section, we performed a Durbin-Wu-Hausman test and a Breusch-Pagan Lagrangian multiplier test and concluded that the random effects model was applicable. Therefore, we performed a Random-effects GLS regression for each variable separately, and together with BCI.
The first model, model A, explains the influence of the number of subsidiaries (SUB) and SIC codes (SIC) on the audit fees. We controlled for auditee size (SIZ). As expected, the audit fees were positively influenced by SUB and SIC. All variables, including SIZ, were statistically significant. The first hypothesis, H1, was therefore accepted. The model had a high adjusted R- squared of 0,7549. This is consistent with prior research (Hay et al., 2006; Francis and Yu, 2009; Hay, 2013), stating that auditee complexity and auditee size have high explanatory power for the audit fees.
24 Table 3: Pearson Correlation Matrix (* = p<0,1, ** = p<0,05 and *** = p<0,01)
AF AFD SUB FSU SIC AFR AMS BCI CPO SIZ
AF 1,000 AFD 0,224*** 1,000 SUB 0,783*** 0,157*** 1,000 FSU 0,811*** 0,145*** 0,990*** 1,000 SIC 0,198*** 0,134*** 0,168*** 0,138*** 1,000 AFR -0,001 0,073** 0,037 0,036 0,020 1,000 AMS 0,056 -0,027 -0,068 -0,040 -0,182 -0,002 1,000 BCI 0,209*** -0,024 0,111*** 0,142*** -0,134*** -0,011 0,838 1,000 CPO 0,151*** -0,035 0,039 0,073** -0,164*** -0,012 0,869*** 0,937*** 1,000 SIZ 0,689*** 0,0913** 0,502*** 0,525*** 0,103*** 0,004*** 0,150 0,227*** 0,200*** 1,000
25 Model B gives insight in the influence of audit firm rotation on the audit fees. As expected, AFR significantly influenced the audit fees negatively. Despite the significant influence, we did not accept hypotheses 2 and 3 at first, since Model B measured the overall influence of audit firm rotations, mandatory as well as voluntary. In order to be able to accept or reject the hypotheses, we performed the same model with only the listed companies and, subsequently, the non-listed companies. The results are shown in Model C (mandatory audit firm rotation) and Model D (voluntary audit firm rotation) and were both statistically significant. Both hypothesis 2 and hypothesis 3 were accepted. By looking at the coefficient and the adjusted R-squared of both models, the mandatory audit firm rotation had a bigger impact and explanatory power than voluntary audit firm rotation.
The audit location, shown in Model E, had positive and significant influence on the audit fees. We accepted hypothesis 4. Due to collinearity, audit location was tested with BCI, but in the robustness section we describe the results for Amsterdam (AMS) and city population (CPO). Models F and G also showed significant results and with higher coefficients than Model A and B, implicating that BCI had a strengthening effect on the relation between the audit fees and auditee complexity, and between the audit fees and audit firm rotation. After performing a Wald test, the implications were as expected; hypothesis 5 was therefore accepted. Hypotheses 6 and 7 were tested with samples of respectively listed companies and non-listed companies (Model H and I). After performing a Wald test, hypothesis 6 was accepted and hypothesis 7 was rejected. We concluded that the influence of mandatory audit firm rotation on audit fees was strengthened by the size of the audit location.
26 Table 4: statistical results with respect to the audit fees (* = p<0,1, ** = p<0,05 and *** = p<0,01)
Variable Expected
influence
Model A Model B Model C Model D Model E Model F Model G Model H Model I
Hypothesis 1 Hypothesis 2 Hypothesis 3 Hypothesis 4 Hypothesis 5 Hypothesis 6 Hypothesis 7
SUB + 20,33*** 20,96** SIC + 308,10* 368,55** AFR - -407,22** -541,69* -213,83* -407,63** -544,09* -213,83* BCI + 1615,70** 1029,987** 1622,14** 2958,46** -64,10 SIZ + 0,0001572*** 0,0002059*** 0,0001189*** 0,0001598*** 0,0002022*** 0,0001527*** 0,000205*** 0,0001117*** 0,0001595*** Intercept -781,52 1611,55*** 2960,75*** 776,84*** 625,17*** -1508,10** 684,32 1151,52 813,14*** N 753 753 411 342 753 753 753 411 342 Adjusted R2 0,7549 0,4730 0,5443 0,4432 0,4512 0,7607 0,4492 0,4697 0,4442 f 520,72*** 98,90*** 50,93*** 68,41*** 99,27*** 532.84*** 103,82*** 55,59*** 67,58*** Model Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Rho 0,8823 0,8990 0,8827 0,8844 0,8976 0,7800 0,8990 0,8840 0,8859
27 6.4 Results for audit fees disclosure
The results regarding audit fees disclosure are displayed by table 5. As described in the methodology section, we performed a Durbin-Wu-Hausman test and Breusch and Pagan Lagrangian multiplier test and concluded that the random effects model was applicable. Therefore, we performed a Random-effects GLS regression for each variable separately and together with BCI.
Model J explains the influence of the number of subsidiaries (SUB) and SIC codes (SIC) on the audit fees. We controlled for auditee size (SIZ). As can be seen in the model, all variables, including SIZ, were statistically insignificant. The first hypothesis of AFD, hypothesis 8, was therefore rejected. The model had a low adjusted R-squared, which means that even if the results were significant, the model holds little explanatory power with respect to the quality of audit fees disclosure.
The same can be said about Model K when it comes to the low adjusted R-squared. We concluded that, despite significant results for AFR, audit firm rotations had little influence on the quality of audit fees disclosure. Model L and Model M confirmed this conclusion, even though Model L showed that mandatory audit firm rotation was significantly related to audit fees disclosure. Voluntary audit firm rotation showed insignificant results. Hypotheses 9 was accepted and hypothesis 10 was rejected. The size of the auditee, the control variable, was insignificant in all models.
The audit location, shown in Model N, had no significant impact on the disclosure of the audit fees. Hypothesis 11 was rejected. The moderating effect of audit location, shown in Model O and P, was denied after performing a Wald-test. Hypothesis 12 was therefore rejected. It made no difference making a distinction between listed companies (Model Q) and non-listed companies (Model R). Hypothesis 13 and hypothesis 14 were both rejected. All models showed a low adjusted R-squared value, which means that even if the models were significant, the model was of little help in explaining the difference in the quality of audit fees disclosure.
28 Table 5: statistical results with respect to the audit fees disclosure (* = p<0,1, ** = p<0,05 and *** = p<0,01)
Variable Expected influence
Model J Model K Model L Model M Model N Model O Model P Model Q Model R
Hypothesis 8 Hypothesis 9 Hypothesis 10 Hypothesis 11 Hypothesis 12 Hypothesis 13 Hypothesis 14
SUB + 0,0000947 0,0000961
SIC + 0,111497 0,0107055
AFR + 0,02841** 0,02321* 0,03746 0,02843** 0,02325* 0,03741
BCI + 0,00097905 0,0074272 0,0098197 0,0265089 0,0009516
SIZ + 1,11e-09 1m,82e-09 -8,39e-11 1,78e-09 1,85e-09 1,16e-09 1,88e-09 -4,93e-13 1,79e-09
Intercept 0,5895*** 0,6202*** 0,6899*** 0,5641*** 0,6293*** 0,5947*** 0,6256*** 0,7059*** 0,5646*** N 753 753 411 342 753 753 753 411 342 Adjusted R2 0,0338 0,0128 0,0019 0,0048 0,0099 0,0347 0,0145 0,0080 0,0049 f 5,12 8,23** 3,60 2,90 2,61 5,14 8,32 3,94 2,90 Model Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Random-effects GLS regression Rho 0,7554 0,7593 0,8143 0,6546 0,7590 0,7568 0,7606 0,8162 0,6585
29 6.5 Robustness tests
In order to test the validity of the above-mentioned results, we performed several robustness tests. First, we measured the auditee complexity with a different proxy, after that we measured the audit location on the level of Amsterdam (AMS) and city population (CPO). Finally, we performed a GLS regression and a Pooled OLS regression with all variables.
After testing the auditee complexity with SUB and SIC, the complexity was tested with the number of foreign subsidiaries (FSU), performing a random-effects GLS Regression. The results were equally significant and positive. This was expected, since FSU had a high correlation with SUB (0,99). We also tested the effect of SUB and SIC on the audit fees without controlling for SIZ. The results were significant and positive with an adjusted R-squared of 0,6434. We concluded that auditee complexity is an important factor when it comes to predicting the height of the audit fees. The results of FSU were the same as SUB and SIC regarding the quality of audit fees disclosure. The complexity of the auditee did not have influence on audit fees disclosure.
We looked at the influence of audit location in three different ways. First, a dummy variable of the four biggest cities of the Netherlands (BCI) was used in the models in the previous section. To test the outcomes, we performed the same models with the other audit location proxies, AMS and CPO. By testing AMS, the results were insignificant, but with CPO the results were similar to BCI, significant and positive. We also tested the impact of audit location on the audit fees without controlling for SIZ. The results were positive and significant, with an adjusted R-squared of 0,0435. The explanatory value of BCI was limited, but it was a significant predictor. All audit location variables were of insignificant value when it comes to audit fees disclosure, making the explanatory power negligible.
We performed a random-effects GLS regression and a Pooled OLS regression with all variables for both the audit fees and the audit fees disclosure, in order to make the individual models more robust. The outcomes are shown in table 6 for the audit fees (Model S and T) and table 7 for the audit fees disclosure (Model U and V). Model S and Model T showed significant results in the expected direction, with an exception of AFR in the OLS regression. As mentioned before, the time variant aspect of the panel data was ignored in the Pooled OLS regression, making the Model less valuable. Model U showed the same results as the individual models regarding the variables. Model V showed unexpected results with three significant variables.
30 Table 6: statistical results with respect to the audit fees Table 7: statistical results with respect to the audit fees disclosure
Variable Expected influence Model S Model T SUB + 20.15*** 19,08*** SIC + 367,76** 334,9819*** AFR - -426,35** -502.63 BCI + 1030,414** 894,61*** SIZ + 0,0001516*** 0,0002145*** Intercept -1450,76** -1630,98*** N 753 753 Adjusted R2 0,7611 0,7663 f 536,97*** 494,21*** Model Random-effects GLS regression Pooled OLS regression Rho 0,7823 - Variable Expected influence Model U Model V SUB + 0,0000933 0,0001271*** SIC + 0,010761 0,0164199*** AFR + 0,02824** 0,0414597* BCI + 0,0073845 0,0119225
SIZ + 1,20e-09 4,14e-10
Intercept 0,5904*** 0,5760*** N 753 753 Adjusted R2 0,0386 0,0418 f 10,77* 6,51*** Model Random-effects GLS regression Pooled OLS regression Rho 0,7587 -
31
7
Conclusion
As mentioned in the introduction section, large companies with headquarters located in the Netherlands have great impact on society’s wellbeing. This impact can either be positive (i.e. increase in available jobs, increase in wealth and/or increase in accessible knowledge) or negative (i.e. fraud schemes, money laundering or tax evasion). In order to prevent large companies to negatively impact society, the audit of those companies should be of high quality and therefore the auditor has to be independent. In this study, the height of the audit fees and the quality of audit fees disclosure were used as proxies for auditor independence. Furthermore, we looked at the impact of the size of the audit location on the audit fees and the quality of the audit fees disclosure in annual reports. Also, we looked into the impact of the auditee complexity and audit firm rotations on both audit fees and audit fees disclosure.
The sample used in this research consisted of the biggest listed and non-listed companies headquartered in the Netherlands. The data was collected from the annual reports of the companies from the period between 2012 and 2017 and from Company.info. There are many large companies in the Netherlands, which is why our results can be generalized to other countries, even though the Netherlands is a small country. Furthermore, companies headquartered in the Netherlands are interesting to research because of the present legislation regarding mandatory audit firm rotation for listed companies. Also, because most listed companies needed to switch between audit firms before the 1st of January 2017, the effect of low-balling could be measured since the sample included the financial year 2017. Another argument to perform this study in the Netherlands is the fact that audit fees disclosure is, to a certain degree, mandatory. Because of recent developments regarding the Brexit10, more large companies may decide to move their headquarters to the Netherlands, making this study possibly more relevant.
With respect to the determinants, we found that the size and the complexity of companies have the biggest influence on the height of the audit fees. Furthermore, we found that the audit fees were higher in the biggest cities, ceteris paribus. Also, when using city population as a proxy for audit location, the height of the audit fees was positively influenced by the size of the population. In addition, the complexity of the company had more impact on the audit fees if the
10 Discovery en Japanse bank naar Nederland om Brexit, AD https://www.ad.nl/economie/discovery-en-japanse-bank-naar-nederland-om-brexit~adcf5a3d/
32 company was audited in one of the four biggest cities. This might be explained by the Urban theory, stating that more knowledge and expertise results in higher costs in general. We concluded that the size of the audit location is an important determinant of the audit fees and should be taken into account in future research.
The results regarding the impact of audit firm rotations on the audit fees were significant for both mandatory audit firm rotations and voluntary audit firm rotations. We concluded that the audit fees decreased after a mandatory as well as a voluntary audit firm rotation. We stated that the quasi rents were in the years after the mandatory audit firm rotations took place, indicating the presence of low-balling. This could not be concluded if it comes to voluntary audit firm rotation. The audit fees did rise in 2017, but the number of audit firm rotations also increased in the same year. Furthermore, the mandatory audit firm rotations had more explanatory power when the company was located in a bigger audit location. This might be explained by the increased competition between audit firms in the cities. We concluded that future research should take into account audit firm rotation with the consideration that the mandatory type has more explanatory power than the voluntary variant.
We tested the same determinants regarding the quality of the audit fees disclosure and concluded that the most important determinants of the audit fees, auditee size and auditee complexity, have no influence on the quality of the audit fees disclosure. The same conclusion was applicable to the audit location and the voluntary audit firm rotations. On the contrary, the mandatory audit firm rotation had a significant impact on the audit fees disclosure but had low explanatory power. In this study, we found that the quality of audit fees disclosure was not explained by the determinants that we examined in this study. Furthermore, the average quality of audit fees disclosure was higher at listed companies when compared to the non-listed companies. One explanation can be found in the fact that listed companies, in general, have more shareholders than non-listed companies. Chau and Gray (2002) conclude that a positive association exists between the number of shareholders and the extent of voluntary disclosure.
Overall, we summarized that the determinants examined in this study had little to zero effect on the quality of the audit fees disclosure. While previous literature suggested that the auditee size has an increasing effect on the quality of the audit fees disclosure, this study found no such relation. In general, we doubt if the quality of the audit fees disclosure can be explained by firm
33 characteristics since the audit fees disclosure might be subject to the judgment of individual controllers of the companies.
The most important findings in this study regarding audit fees are as follows. First, we found that, in line with previous literature (Francis and Yu, 2009; Hay, 2013), the auditee complexity and size are the most important predictors for the height of the audit fees. In addition, this study found that the size of the audit location has a positive influence of the height of the audit fees. These findings are in line with the Urban theory and expand the literature mentioned by Hay (2013) by looking at the four biggest cities, instead of looking at only the biggest city of the country. Furthermore, when we tested the audit location with the city population, the results showed the same influence. Finally, this study contributed to existing literature by being the first study that examined the influence of the size of the audit location in a Dutch context.
