1
The effects of corporate governance and internal controls on audit
fees and the quality of audit fee disclosures: evidence from the
Netherlands
Abstract
This study attempts to find out whether corporate governance effectiveness, the quality of internal control and the stock market listing have influence on the audit fee and the quality of audit fee disclosure. The sample contains listed firms (AEX, AMX and AScX) and non-listed firms in the Netherlands over the years 2012 to 2017. To measure the quality of the audit fee disclosure, we use the audit fee disclosure index. All data is collected from annual reports. By running regressions, we discover that the size of the auditee, the number of meetings of the audit committee, and the stock market listing are determinants of the audit fee at listed firms. Moreover, the size of the auditee, the quality of internal control, and the size of the audit committee are related with the quality of audit fee disclosures at listed firms. Also, the number of meetings of the audit committee is related with the quality of audit fee disclosures at non-listed firms.
Thomas de Ruiter s2561808
MSc. Accountancy & Controlling (Accountancy) Word count: 12714
January, 2019
Supervisor: Dr. W. Kevelam Co-assessor: prof. Dr. R.L. ter Hoeven
2
Table of content
1. Introduction ... 4
1.1 Introduction ... 4
1.2 Scientific contribution & practical relevance ... 5
1.3 Research question ... 6
1.4 Structure of the paper ... 7
2. Theoretical background ... 8
2.1 Agency theory ... 8
2.2 Voluntary disclosure theory ... 9
2.3 Legitimacy theory ... 10
3. Literature review & hypotheses development ... 12
3.1 Determinants of audit fees ... 12
3.2 Determinants of the quality of audit fee disclosures ... 15
4. Research methodology ... 19 4.1 Introduction ... 19 4.2 Sample ... 19 4.3 Dependent variables ... 21 4.4 Independent variables ... 22 4.5 Control variable ... 23 4.6 Regression model ... 23 5. Statistical analysis ... 25 5.1 Descriptive statistics ... 25
5.2 Results on audit fees ... 30
5.2.1 Results on audit fees at listed firms ... 30
5.2.2 Results on audit fees at non-listed firms ... 31
5.3 Results on the quality of audit fee disclosures ... 33
3
5.3.2 Results on the quality of audit fee disclosures at non-listed firms ... 34
5.4 Robustness tests ... 37
5.5 Additional test ... 38
6. Conclusion ... 39
6.1 Summary & discussion ... 39
6.2 Limitations & further research ... 41
7. References ... 43
Appendix A - Table regression results on audit fees at all firms ... 49
Appendix B - Table regression results on quality of audit fee disclosure at all firms ... 50
Appendix C - Table regression results on audit fees at all firms without years 2015/2016 . 51 Appendix D - Table regression results on quality of audit fee disclosure at all firms without years 2015 and 2016 ... 52
Appendix E – Table regression internal control years 16-17 ... 53
Appendix F – Audit fee disclosure index ... 54
Appendix G – Example of an audit fee disclosure ... 55
4
1. Introduction
1.1 Introduction
In 2009, Hans Hoogervorst, at that time chairman of the Dutch supervisor of audit firms, Autoriteit Financiële Markten (AFM), claimed that auditors were partly responsible for the financial crisis (de Volkskrant, 2009). He accused auditors of being unprecise and their lack of having an overview during the audit, which occasionally resulted in bankrupt banks and companies. This statement can be considered as one of the first public allegations towards auditors during the financial recession.
The last couple of years, auditors suffered more and more from negative publicity. In 2010 and 2014, the AFM published a report about the audit quality performed by the Dutch Big Four companies (AFM, 2010) (AFM, 2014). The AFM concluded that the audit quality was insufficient and the profession should show improvements to regain trust from society. More recently, in 2017, the AFM stated that despite the implemented improvements, results were still not satisfying (AFM, 2017). These reports have given rise to questions by society what the added value of the auditor actually is and how they, despite high audit fees, constantly perform insufficiently.
As a response to the bad performing accountancy profession, the Dutch regulator has implemented many new rules. A body of the Dutch profession of auditors (NBA) implemented 53 new rules concerning the future of the profession (NBA, 2014). For example, the board of directors should have a more material function instead of a formal function (NBA, 2014). Specifically, the audit committee should take the lead and serve as a first contact of the accountant (NBA, 2014). Moreover, also a new corporate governance code is installed (MCCG, 2016). Since this most recent corporate governance code applies, large companies are obligated to employ an internal auditor, according to MCCG, 2016. However, it turns out that only 70 percent of Dutch listed companies employ an internal auditor (Bogtstra & Renes, 2017). Besides more Dutch regulation, also the European Union (EU) introduced new legislation concerning the audit activities. In 2014, the EU decided that organizations of public interest have to appoint a new auditor after ten years of cooperation (EU Regulation 537/2014). This regulation is because firms and auditors tend to become familiar with each other and therefore auditor independence could be threatened. After this, the Dutch regulator adopted this in its legislation, after they already came up with a similar regulation in 2012 (Wijzigingswet
5
financiële markten, 2016). However, did these measures improve the audit quality? In this study, we explore whether all taken measures were effective by focusing on audit fees and the quality of audit fee disclosures and linking these subjects to audit quality and auditor independence. In the next subsections, we introduce the contribution and relevance of this study, we provide the research question and conclude with the structure of this study.
1.2 Scientific contribution & practical relevance
This study about audit fees and audit fee disclosures is relevant for both academic literature and business for several reasons. First, researching audit fees is important to understand because audit fees can be interpreted as a proxy for audit quality, according to Hoitash, Markelevich & Barragato (2007). The higher the amount of audit fee, the higher the performed audit quality should be, ceteris paribus. In this study, we extend the literature of audit fees because it is important to understand for both academics and people in business, how to influence audit fees and thus audit quality. Moreover, according to Hribar, Kravet & Wilson (2014), audit fees correlate positively to measures of accounting quality. Audit fees contain information about how well accruals map into cash flows and are informative about predicting financial restatements or fraud (Hribar et al. 2014). Second, the quality of audit fees disclosures is also important to analyze. Since, according to DeFond, Raghunandan & Subramanyam (2002), the amount of fee that is paid for both auditing services and non-audit services, the audit fee disclosures clarify the degree of auditors’ independence. By taking more non-audit services and making the auditor more economically dependent on the client, the client could unintentionally reduce auditors’ objectivity. Since auditor independence is positively associated with audit quality (DeAngelo, 1981), it could result in lower quality when auditor independence is low. So, if an entity discloses information about audit fees, we can link it to auditor independence what tells us more about audit quality. In the end, this helps us understand how to execute the audit more independently.
Also, this study fills a gap and contributes to the academic literature in several ways. First, as far as our knowledge, there has been no research conducted about audit fees in combination with corporate governance controls, in the Dutch context. Most studies concerning audit fees have been researched in Anglo-Saxon countries and do less appear in Europe, especially the Netherlands. Second, the unique combination of both audit fees and its disclosures in one study, which has been studied simultaneously only once before (Kevelam, ter Hoeven & Brouwer,
6
2017), gives an opportunity to understand audit quality better. By combining these subjects, we first discover how audit fees are affected and thus how to improve audit quality. After that, we can analyze to what extent companies disclose information about audit fees and see whether the auditor has been working independently, what might have affected the audit quality. Third, the determinants used in this study have not been studied before in combination with the moderating variable of the stock market listing and where this distinction is specifically investigated. This gives the opportunity to investigate whether there are differences between listed and non-listed firms concerning audit fees and its quality of disclosures and how they are affected. Lastly, during the years of 2012 until 2017, several regulations have been implemented, and therefore it gives a great opportunity to see whether these changes have improved the audit quality. These changes have not been researched before and can help us to understand how audit quality can be improved. Thus, besides the fact that this fills a gap in the scientific literature about whether more legislation is helpful, it has a practical relevance as well. Because by clarifying whether legislation has been effective, other foreign regulators could also implement new rules if they have seen that more legislation worked out in the Netherlands.
1.3 Research question
As mentioned above, a lot of new rules have been implemented for the past five years in the Netherlands (NBA, 2014; MCGC, 2016; EU Regulation 537/2014 & Wijzigingswet financiële markten, 2016). These rules influence the corporate governance characteristics of firms and their interaction with the auditor. Parts of the new regulation had to do with corporate governance controls such as a different role of the audit committee and the board of directors and a whole new corporate governance code which makes the internal auditor function mandatory (NBA, 2014 & MCCG, 2016). Therefore, we explore whether these changes did have an impact on the audit fee, and thus audit quality. Next, we focus on the quality of audit fee disclosures. This disclosure is mandatory for listed firms, as it is optional for non-listed firms that do not meet specific requirements of being a large firm, according to the Dutch civil code, article 382a BW 2 (Langendijk, 2011). Although, the length and quality of the mandatory and voluntary audit fee disclosures fluctuate between the companies. Audit fee disclosures show us how independent an auditor has been working on a specific job. Since auditor independence is positively associated with audit quality (DeAngelo, 1981), better disclosures contain more information about audit quality. These disclosures have been subject for study
7
before, but not in combination with corporate governance controls. So, we attempt to find out whether corporate governance controls affect the quality of audit fee disclosures. The research question of this study is formulated as follows:
To what extent do internal and corporate governance controls influence the audit fees and the quality of audit fee disclosures at Dutch listed and non-listed firms?
1.4 Structure of the paper
In the following sections of this paper, we first elaborate more on the underlying theory of this study combined with earlier empirical research. This together leads to logical expectations and will be concluded with concrete hypotheses. After this, section 4 begins with the methodology where the descriptive analytics are shown and after that the statistics where the hypotheses are tested. From this point, the statistics are interpreted and analyzed, and a conclusion of the statistics is drawn from the results. In the last section, the conclusion and the limitations of the study are presented. This part includes some useful ideas for further academic research.
8
2. Theoretical background
2.1 Agency theory
The first used theory in this study is the agency theory, developed and described by Jensen & Meckling (1976). According to Jensen & Meckling (1976), an agency relationship is a contract under which one or more persons (the principal) engage another person (the agent) to perform some service on their behalf which involves delegating some decision-making authority to the agent. This is the result of separating ownership and management. Following this, conflicts can arise when both the principal and the agent try to maximize its utility. For example, an agent could try to maximize profit in a certain year and avoid expenditures, due to his limited employment contract and bonus structure, while it is better to invest regarding the long-term. In addition, in some situations, the principal pays the agent to expend resources to guarantee that he will not take certain actions that could harm the principal (Jensen & Meckling, 1976). In general, the reduction of welfare experienced by the principal due to negative actions of the agents is the residual loss (Jensen & Meckling, 1976). The sum costs of monitoring, bonding to eliminate undesirable actions of the agent and the residual loss are together the agency costs (Jensen & Meckling, 1976).
According to Eisenhardt (1989), the agency problem consists of two problems: conflicting desires of both parties and it is difficult for the principal to see what the agent is doing. To partly resolve these agency problems and to see what the agent is doing, the principal hires an auditor. The principal pays an audit fee to the auditor in exchange for receiving an opinion on how the firm performed and whether the agent fulfilled his responsibilities. By receiving this report, the principal reduces the risk of undesirable behavior performed by the agent. Eventually, the principal receives more certainty as to whether the shown numbers and statements in the report are actually in line with the current financial position of the firm. In the annual report, the audit fee is included in the audit fee disclosure, which is available for all its shareholders. In theory, the audit fee disclosure and the audited financial statements decrease the information asymmetry between agents and principals. The principals determine the extent of the acquired services. If the principal decides to acquire an integral audit from the auditor, the principal receives more certainty than in the case he decides to perform just a normal audit. However, the result is that the audit fee will be excessively high. Therefore, a small amount of uncertainty remains and this is part of the information asymmetry.
9
2.2 Voluntary disclosure theory
Healy & Palepu (2001) extensively studied voluntary disclosure, and they defined the theory as follows: given the risk of job loss accompanying poor stock and earnings performance, managers use corporate disclosures to reduce the likelihood of undervaluation and to explain away poor earnings performance. However, this theory leads also to a contradiction. According to Dontoh (1989) companies are trying to maximize their profits and disclose only information that has a positive effect on the market price. Paradoxically, companies disclose also negative information about their perspectives, and as a result, their market price declines significantly (Dontoh, 1989). But, Healy & Palepu (2001) summarized the literature until then and came up with six relevant motives for companies why they decide to voluntarily disclose information. These reasons are capital market transactions; for example, when a firm would like to issue new capital, corporate control contests, stock compensation, litigation, proprietary costs, and management talent signaling. In the end, the choice of whether to disclose specific information about the firm depends mostly on profit maximization, and apparently, the (potential) benefits exceed the costs.
In the prior subsection, we discussed the information asymmetry or so-called lemon problem in combination with the agency theory. This agency problem also plays a role in the voluntary disclosure theory. In fact, voluntary disclosure can partly resolve the information asymmetry. If managers decide to disclose extra private information, it certainly decreases the information asymmetry. Besides this, there are several solutions to the lemons problem such as optimal contracts, regulation or financial analysts and auditors (Healy & Palepu, 2001).
In this study, audit fee disclosures will be subject of research. But, audit fee disclosures can be primarily classified as a mandatory disclosure for large firms, regarding Dutch law. Specifically, there is more extensive legislation concerning disclosure for listed firms than there is for non-listed firms. Since our sample consists of ‘large’ firms, most firms will comply with standard legislation. However, the extent and quality of the disclosures still vary per company. That is why it is more interesting to see why companies decide to disclose information voluntary next to all their mandatory disclosure duties. Obviously, a more extended disclosure strategy leads to a higher audit fee. More (financial) information has to be audited, and if a firm decides to disclose specific information, it requires updating this information in the sequent year. Otherwise, people could get suspicious of what happened, and all benefits, as mentioned above, could disappear.
10
2.3 Legitimacy theory
The legitimacy theory relates to the organizational legitimacy theory, described by Dowling & Pfeffer (1975). They explained how firms operate with regard to their social system. Organizations seek to establish congruence between the social values associated with or implied by their activities and the norms of acceptable behavior in the larger social system of which they are a part (Dowling & Pfeffer, 1975). If this congruence lacks, the organizations’ legitimacy will be in danger. As a result, firms could suffer from legal, economic or social sanctions (Dowling & Pfeffer, 1975). This theory goes beyond the stakeholder theory, which states that firms should take into account all individuals or groups which depend on the company for the realization of their personal goals and on whom the company is dependent (Freeman, 1984).
So, this legitimacy theory states that firms have to earn their right to exist and this can be achieved in three different ways. First, it can adapt its goals and methods to conform to prevailing definitions of legitimacy (Dowling & Pfeffer, 1975). Second, the firm can attempt to alter the definition of social legitimacy, and finally, by using communication, the firm can attempt to become identified with the values of its social legitimacy (Dowling & Pfeffer, 1975). In the end, it means that firms have to adapt to their environment by making changes in their culture, values, mission, working methods, and other practices. As a result, they can act within the bounds and norms of their social system, by complying to their social contract. By doing this, the firm can benefit the most from its entire environment.
According to this theory, voluntary disclosures can also be justified. Deegan (2002) states that companies act within the boundaries and norms of society. But, community expectations can change across time and are not static, and therefore companies have to adapt to new circumstances (Deegan, 2002) continuously. A way to achieve this is to disclose extra information to show a willingness to change and to adapt to other boundaries and norms. So, managers tend to disclose information in their annual reports, to allay community concerns and to improve their position in society (Deegan, 2002). Based on this and pursuant to the legitimacy theory, audit fee disclosures should be disclosed as extensive as possible. By disclosing this information, the firm can show a willingness to society, but, primarily, can take away any possible concerns of the community. But, a more extensive reporting could also have implications. First, information that is more sensitive is shared with competitors. To survive and to keep a step ahead of your rivals, a certain degree of confidentiality is required. Second,
11
a more extended reporting can be more expensive. Moreover, the audit fee can increase as well, due to more the information that has to be audited.
In short, according to the legitimacy theory, firms can benefit economically from their social system if they try to congruence with it. By disclosing (audit fee) disclosures and spending more on auditors, firms show their willingness to congruence with the society. On the other side, to reach this congruence, firms have more expenditures. These costs should not be excessively high. Firms seek a certain balance in the costs and benefits of its disclosures.
12
3. Literature review & hypotheses development
3.1 Determinants of audit fees
There has been conducted a lot of research concerning audit fees and its determinants. Simunic (1980) was the first researcher to do so, and he hypothesized factors that determined audit fees, such as litigation risk and size of the auditee. More recently, research on factors related to audit fees has become more widespread (Hay, 2013). To identify what determinants influence the audit fee paid by the auditee, the most common determinants that are successfully researched are size, complexity, and risk (Hay, 2013). But, the literature on audit fees in the Netherlands is relatively underdeveloped. Nevertheless, Langendijk & Groenen (2004) found that size, complexity and listing positively influenced the audit costs in the Netherlands.
The quality of internal control as a determinant of audit fees has also been subject of research. Internal control is designed to provide reasonable assurance of the reliability of the financial reports (Lee, 2016) and therefore can complement and improve audit quality. Internal controls enable auditors to perform less substantive testing because they can rely on their internal control system (Collier & Gregory, 1996). But, if tests indicate that internal controls are not operating properly, the auditor should do more testing (Collier & Gregory, 1996). The results for internal control and the relation with audit fees are quite mixed. On the one hand, a better quality of internal control could lead to higher audit fees. A higher quality of internal control enables auditors to rely more on a firms’ internal control system. This can result in higher demand from the company to the auditor, to test more or better elements. Studies from Wallace (1984) and Kreutzfeldt and Wallace (1986) stated that more expenditures on internal controls lead to lower audit fees and weak systems demand more testing. Thus, weaker internal controls systems should lead to higher audit fees. This means, vice versa, that strong internal control could also lead to less testing and therefore a lower audit fee. The overall meta-result in the research of Hay, Knechel & Wong (2006) is not significant. In contrast, there is a positive relationship between internal control and audit fees in studies from Goodwin-Stewart & Kent (2006), Hogan & Wilkins (2008), Hoitash, Hoitash & Bedard (2008) and Wei & Son (2015). In addition, there was a positive relationship found between the disclosures of internal control weaknesses and audit fees (Raghunandan & Rama, 2006). In this study, we investigate whether internal control affects audit fees in the Netherlands. Since recent literature is relatively clear, we expect a positive association.
13
H1: The quality of internal control is positively associated with audit fees.
Corporate governance effectiveness as a determinant of audit fees has also been subject of research in the past. Corporate governance effectiveness is likely to influence audit fees because improved corporate governance implies that the control environment is more effective (Hay et al., 2006). As a result, a stronger corporate governance mechanism means that the audit fee could rise because the auditee demands higher quality. This implies that audit quality is affected as well and thus the corporate governance mechanism can be considered as an effective tool to have more control over the firm. The outcomes of multiple studies concerning audit fees are quite mixed, and it mostly depends on which measure is used (Hay, 2013). Wallace (1984) found a positive effect of internal expenditures on the audit fee. Moreover, according to Kreutzfeldt & Wallace (1986) weak control environments involve more audit work, so in turn, strong internal control environments should lead to lower audit fees (Collier & Gregory, 1996). On the other hand, there is a significant positive relation between board characteristics as independence, diligence or expertise and audit fees (Carcello et al., 2002). Furthermore, there is a positive relationship between the existence of an audit committee and audit fees (Goodwin-Stewart & Kent, 2006). In this research, we study corporate governance effectiveness in two ways. We measure this concept as the size of the audit committee and as the number of meetings of the audit committee had during a year. The audit committee assures to shareholders that auditors are in the position to safeguard their interest (Collier & Gregory, 1996). Thus, the audit committee is in line with the agency theory, because it is a control designed for shareholders, to reduce the information asymmetry between owners and managers. By measuring the number of meetings, it enables us to study whether diligence of the audit committee affects the audit fee. On the other hand, when measuring the size of the audit committee, we can explore whether more members within the audit committee lead to higher audit fees. This is important because, in general, more members contain more experience and knowledge, and therefore could demand higher audit quality. Despite contradicted results in the literature, we expect a positive relationship between the size or number of meetings of the audit committee and audit fees. We expect this because a higher audit quality is demanded nowadays. Also, we prospect that the audit committee and the auditor challenge each other and this results in a higher audit fee.
H2a: The number of meetings of the audit committee is positively associated with audit fees. H2b: The size of the audit committee is positively associated with audit fees.
14
A determinant that is underdeveloped in the current literature is whether a firm is Dutch listed, cross-listed or non-listed. Langendijk (1997, 2004) found that listed firms pay higher fees than non-listed firms in the Dutch context, but cross-listed firms were not included in this research. Seetharaman, Gul & Lynn (2002) found that the United Kingdom auditors charge higher fees for their services when their clients are also active in the United States, due to more litigation risk. Based on this study, a significant portion of the fee premium charged by auditors of cross-listed companies is related to United States regulatory and disclosure requirements (Bronson, Ghosh & Hogan, 2017). So, cross-listed firms could face higher (lower) fees, due to stronger (weaker) legal environments, according to Bronson et al. (2017). To conclude, the results in the literature are quite similar. Listed companies face higher audit fees, and cross-listed firms face even higher audit fees. Besides this, if Dutch cross-listed firms face higher audit fees than regularly listed firms, then it could mean that other legal environments calculate higher fees, which can result in higher audit quality. In addition, we study the moderating effect of listing on the relation between the other determinants and audit fees and the quality of audit fee disclosures. This might give insights whether listing leads to a higher audit fee in combination with stronger corporate governance controls. As mentioned before, we expect a positive relationship between the quality of internal control or corporate governance effectiveness and audit fees. On top of that, we expect a reinforcing effect of the stock market listing on these relations, because listed firms pay higher fees due to more litigation risk (Seetharaman et al. 2002). The other way around, if a firm is not listed, it could have an attenuated effect or no effect, because non-listed firms have less litigation risk, should comply with less regulation, and are less complex. Therefore, we expect a reinforced positive relation with regard to this moderating variable.
H3: A higher degree of listing is positively related to audit fees.
H4: A higher degree of listing is positively associated with the relation between the quality of internal control and audit fees
H5a: A higher degree of listing is positively associated with the relation between the number of meetings of the audit committee and audit fees.
H5b: A higher degree of listing is positively associated with the relation between the size of the audit committee and audit fees.
15 Figure 1
3.2 Determinants of the quality of audit fee disclosures
The literature about specific audit fee disclosures is scarce. One of the reasons for this is that these disclosures are voluntary for most companies , whilst they are compulsory for only the largest companies in the Netherlands. The Dutch civil code describes that a company can be classified as large if its assets exceed EUR 20 million, its sales exceed EUR 40 million, and the company employs more than 250 employees (BW 2:9). As a result, these companies have to comply with stronger legal requirements and are obligated to disclose their audit fee. Next,, companies tend to disclose more information, and therefore the quality of disclosures vary. Firms do this, according to Healy & Palepu (2001), because increased disclosure can decrease agency costs by reducing the information asymmetry. To examine whether the determinants affect the quality of audit fee disclosures, one could understand better how independent auditors work and whether this relates to the audit quality. In this part, we primarily focus on general disclosures instead of specific audit fee disclosures, because general disclosures are more widespread in the academic literature. However, some studies did focus on these audit fee disclosures. Studies from Dye (1991) and DeAngelo (1981) have led to a contradiction in the literature about the audit fee disclosures. Dye (1991) claims that audit fee disclosures lead to more auditor independence, while DeAngelo (1981) stated that audit fee disclosures do not affect this (Lai, 2009). According to Lai (2009), there is a controversy in the literature about the essence of audit fees disclosures and whether this impairs or improves the audit quality. According to DeAngelo (1981), disclosures do not affect transaction costs which give rise to quasi-rents (Lai, 2009). On the other hand, according to Dye (1991), if audit fee disclosures are public, the absence of quasi-rents minimizes the incentives for the auditor to distort the report
16
(Lai, 2009). This was the reason to find out whether audit fee disclosures improve auditor independence. He studied if auditors were more likely to issue modified opinions after the mandatory audit fee disclosure was implemented. It turned out that auditors indeed modified their opinion more frequently. So, he agreed with Dye (1991) and therefore, according to Lai (2009), the disclosure of audit fees will improve auditor independence and thus audit quality.
According to Garas & ElMassah, (2016) corporate governance is considered as a system concerned with ensuring that firms are running in a way that societal resources are used efficiently and effectively. To address corporate governance, corporations install both external and internal mechanisms (Garas & ElMassah, 2016). The internal corporate governance mechanism accentuates honesty and transparency in disclosing information in a timely manner, while external corporate governance mechanism is more concerned with the legal system that provides protection to shareholders (Garas & ElMassah, 2016). This internal corporate governance mechanism is in line with the legitimacy theory, because the society wherein the firm acts, demands such transparency. Subsequently, one could expect that a stronger corporate governance mechanism or internal control, could lead to more extended disclosures or better disclosure quality. Studies of Eng & Mak, (2003), Forker (1992), Ho & Wong (2001) and Markarian, Parbonetti & Previts (2007) found a positive relations between corporate governance and voluntary disclosure. Internal control or an audit department can be seen as a measure of corporate governance. However, the link between specific internal control and the quality of disclosures is missing in the academic research so far. In addition, disclosures concerning audit fees are absent as well. Therefore, the proposed hypothesis will be as follows:
H6: The quality of internal control is positively associated with the quality of audit fee disclosures.
In contrast to scarce literature about internal control as a determinant of any disclosure, more is known about the relation between corporate governance and disclosures. In line with the agency theory, more effective corporate governance could lead to more and better disclosures, because they want to reduce the information asymmetry. As mentioned above there is a positive relationship found between corporate governance and voluntary disclosures, according to Eng & Mak, (2003), Forker (1992), Ho & Wong (2001) and Markarian et al. (2007). To illustrate, Eng & Mak (2003) found a positive relationship between corporate governance, measured as ownership structure and board composition, and voluntary disclosure. In the study of Ho &
17
Wong (2001) and Forker (1992), the audit committee is used as a proxy for corporate governance and is found to be positively associated with voluntary disclosure of information. Furthermore, Li, Mangena & Pike (2010) found that several characteristics of the audit committee, such as size and the number of meetings are positively associated with intellectual capital disclosures. Also, we measure corporate governance mechanism in two ways, namely the size of the audit committee and their number of meetings during the year. On the one hand, a larger audit committee is more likely to uncover and resolve potential problems in the reporting process (Li, Mangena & Pike, 2010). On the other hand, meetings that are more frequent could mean that the audit committee has more time to monitor the reporting process (Li, Mangena & Pike, 2010). In general, studies find a positive relationship between corporate governance effectiveness and disclosures. However, the quality of audit fee disclosures has not been studied in combination with corporate governance. Since this is important for understanding how independent an auditor has been working, what we extensively described in the introduction, we extend literature. So, based on previous research, we propose the following hypothesis:
H7a: The number of meetings of the audit committee is positively associated with the quality of audit fee disclosures.
H7b: The size of the audit committee is positively associated with the quality of the audit fee disclosures.
The third determinant of the quality of audit fee disclosure to research is whether a firm has a stock market listing. Companies in our sample can be categorized in three ways; non-listed, Dutch listed or multiple listed or cross-listed. In the current academic research, studies have been primarily focusing on the relation between stock market listing and multiple kinds of disclosure, except for audit fee disclosures. One of the first studies which addressed this relation is the study of Cooke (1992). Here, a significant difference is found in the extent of disclosure between unlisted, listed on the Stockholm Stock Exchange, and multiple listed. The lower the degree of listing, the less information is disclosed in their annual report. Subsequently, if a firm is listed or multiple listed, the firm discloses more information. According to Cormier & Gordon (2000) publicly owned firms disclose more social and environmental information than privately owned companies. This is in line with the legitimacy theory, due to reasons of accountability and visibility regarding their contract with the society. Moreover, according to Branco & Rodrigues (2006), the legitimacy theory may be the reason why banks disclose their
18
social responsibilities. However, there are more reasons why a stock market listing could lead to more or better disclosures. First, listed firms are primarily big firms; therefore they should comply with several requirements from regulatory authorities (Kilic, Kuzey & Uyar, 2014). Second, a publicly firm increases ownership diffusion, therefore expectations of shareholders increase and that leads to more pressure to disclose (Kilic et al. 2014). Based on these findings, we prospect a positive relation between listing and the quality of audit fee disclosures. Also, we expect a positive moderating effect on the positive connection between corporate governance effectiveness or internal control quality and the quality of audit fee disclosures. This is because listed firms have to deal with more stakeholders, and in line with the legitimacy theory, they have a higher responsibility towards society. Therefore, we propose the following hypotheses:
H8: A higher degree of listing is positively associated with the quality of audit fee disclosures. H9: A higher degree of listing is positively associated with the relation between the quality of internal control and the quality of audit fee disclosures.
H10a: A higher degree of listing is positively associated with the relation between the number of meetings of the audit committee and the quality of audit fee disclosures.
H10b: A higher degree of listing is positively associated with the relation between the size of the audit committee and the quality of audit fee disclosures.
19
4. Research methodology
4.1 Introduction
In this section, we explain how we perform the statistical analysis. First, we describe how we selected the firms in our sample. Thereafter, we show how the dependent, independent and control variables are measured and why we do this. At last, we formulate the regression model. The dependent variables in this study are audit fees and the quality of audit fee disclosures. The independent variables are corporate governance, internal control, and the stock market listing. All independent variables serve as a determinant of both dependent variables. Furthermore, the variable stock market listing has multiple functions. One as a regular determinant on both dependent variables and besides this as a moderator in the remaining relations of the conceptual model. The control variable is the size of the auditee.
4.2 Sample
We analyze 75 listed companies (AEX, AMX, AScX) and 72 non-listed companies in the Netherlands. The timespan of this study consists of the financial years 2012 – 2017. The complete sample is shown in appendix H. We take year ending 2016 to identify which companies were listed at that specific moment in the Netherlands. Hereby, we assume these companies to be listed during the years 2012-2017. If a company is added or deleted due to an IPO or a delisting to one of these stock markets later than 2012, then those non-listed years are left out the sample. For the non-listed firms, we first selected all Dutch companies with more than 500 employees, according to company.info, and which annual reports were available in 2018. So, we exclude some companies which had not published their annual report of 2017 yet or were not obligated to release their annual report due to law exemptions. In the end, 1007 companies met our requirements and thereafter we pick the 75 companies with the most employees. We choose to select most employees as a requirement because these companies contain the most relevance with regard to society. Thus, the number of employees can be considered as a proxy for societal relevance. We collect data from annual reports, online financial statements, and other financial information resources. Specific characteristics of companies will be collected from company.info. As a result, we build a database where we can find all the information, and thereafter the research is conducted in a quantitative way.
20
Overview of observations – Table 1
Listed companies
Total obtainable number of observations
(N) 450
Delisted, IPOs, merged or liquidated 27 -
Total number of observations for AFD 423
Unavailable audit fee 7 -
Total number of observations for AUF 416
Non-listed companies
Total obtainable number of observations
(N) 432
Unavailable data 57 -
Total number of observations for AFD 375
Unavailable audit fee 25 -
Total number of observations for AUF 350
In the statistical analysis, we use different samples to test the hypotheses. We do this because not all variables apply thoroughly to non-listed firms. For example, not all non-listed firms have an audit committee. To study relations as specific as possible, we divide the sample in the listed firms' sample and non-listed firms sample. Furthermore, one of the variables is the stock market listing, and it makes no sense to investigate this factor on non-listed firms. As a result, the moderating effect of the stock market listing will not be studied for non-listed firms. To illustrate what sample we use to test every hypothesis, see figure 3 below.
Hypotheses Listed firms sample
Non-listed firms sample 1 Yes Yes 2a Yes Yes 2b Yes Yes 3 Yes No 4 Yes No 5a Yes No 5b Yes No 6 Yes Yes 7a Yes Yes 7b Yes Yes 8 Yes No 9 Yes No 10a Yes No 10b Yes No Figure 3
21
4.3 Dependent variables
In this study, two dependent variables are researched, namely audit fees and the quality of audit fee disclosures. Both subjects are included in the annual reports of the companies of our sample since it is mandatory for large firms according to the Dutch civil code 2:382a BW. Audit fees can be easily obtained because it is an objective number that indicates how much is paid for an audit only. Nevertheless, we use the natural logarithm of audit fees, since recent literature claims that the relation between audit fees and assets, and probably other determinants, is non-linear. This is mainly caused by the fact that an auditor’s evaluation of the internal control processes would likely involve a comparable level of effort for different-sized clients (Cullinan, Du & Zheng, 2016). To illustrate, whether specific assets are 10 million or 15 million worth, the audit work will not increase linearly. Therefore, we use the natural logarithm of the total audit fee.
However, the quality of audit fee disclosures is somewhat more complicated because it is more subjective. As said, it is a mandatory disclosure in the European Union (Directive 2006/43/EG), but the quality and quantity of the disclosure differ significantly among companies. According to the Dutch and European law, the audit fee disclosure should contain the following elements: the total audit fee, the total other audit services fee, the full tax services fee, and the total other non-audit related fees (Directive 2006/43/EG). Besides this, firms can opt to disclose more information about their audit fees. But, we need to measure the quality of audit fee disclosures. First of all, the quality of accounting disclosures can be defined in several ways. Studies define disclosure quality as the degree of self-interested bias in the disclosure or as the ease with which investors can read and interpret the information (Beattie, McInnes & Fearnley, 2004). One of the limitations of disclosures is the difficulty in measuring the extent of disclosure (Healy & Palepu, 2001). Besides this, there is inherent subjectivity involved when measuring disclosures. To tackle these problems, we use a disclosure index. Beattie et al. (2004) analyzed disclosure indices to measure disclosures semi-objective. This enables us to give a more objective number to these disclosures and, subsequently, we can examine this in a quantitative way. We measure this by providing points to some aspects in the audit fee disclosure. The disclosure can vary between at least 0 and max. 12 points, where 0 points mean that a firm does not publish any information about its audit fee and 12 points indicate that a particular firm discloses extensively over its audit fee. Requirements in the disclosure index are based on legislation according to Directive 2006/43/EG and general best practices. The disclosure index is shown in appendix F, and an example of an audit fee disclosure is presented in appendix G.
22
4.4 Independent variables
In previous studies, internal control has been measured in several ways. Hogan & Wilkins (2008) and Hoag & Hollingsworth (2011) found that audit fees increase when a firm has disclosed internal control deficiencies by measuring the weaknesses in the SOX section 404 in the annual report. Next to this, Wei & Son (2015) found that auditors increase the risk premium upward when internal control quality decreases, based on SOX section 404 as well. In the paper of Goodwin-Stewart & Kent (2006), internal control was measured as the number of employees in the internal audit function. The internal audit function monitors the internal controls and is, therefore, the final piece in the chain of internal control (Wallage & van Leeuwen, 2017). This implies that if a company has an internal audit function, they have internal controls that they monitor. Therefore, the internal audit function is an appropriate proxy for internal control. In short, measures of several studies are mixed, and internal control is measured as both the presence or absence of the internal auditor and internal control weakness disclosure. In the Netherlands, internal control deficiency disclosures are not available. Therefore, we choose to use the presence of an internal auditor as a proxy for the quality of internal control. Due to limited available information about how many people work within these departments, we choose to perform a dummy variable. Internal control is measured with a dummy variable and this is described as follows: 0 means absence of an internal auditor or department, and a 1 means presence of internal auditor or department.
In the study of Goodwin-Stewart & Kent (2006) the corporate governance mechanism is measured as the number of audit committee meetings. In addition, Carcello et al. (2002) use the number of board meetings as a proxy for board diligence to measure corporate governance mechanism. So the number of board meetings shows the willingness to monitor the board. More recently, the study of Zaman, Hudaib & Haniffa (2011) measured the audit committee effectiveness as the sum of independence, expertise, and size. Sometimes corporate governance mechanism is measured by taking the size of the audit committee. However, the study of Collier & Gregory (1996) made a dummy variable to measure the relation between audit fees and the audit committee. In this study, the corporate governance mechanism will be measured as the size of the audit committee and the number of meetings of the audit committee. On the one hand, we measure the diligence by counting the number of meetings of the committee. On the
23
other side, we measure the size of the audit committee and try to find out whether more members do affect the audit fee and audit fee disclosure.
To measure the stock market listing, we use publicly available information. The companies in the sample vary between whether they are listed, cross-listed or non-listed. In this study, every firm will be labeled with a zero if the firm is not listed at any stock, a one when the firm is listed at the Dutch stock market only and a two if the firm is also listed outside the Netherlands, called cross-listed.
4.5 Control variable
Besides the independent variables, also a control variable is added to this study. The control variable serves as a constant factor in order to explore the relationship between the independent and dependent variables. Usually, the control variable strongly influences the dependent variable. In this study, we use the size of the auditee as the control variable. Size has been researched extensively in combination with audit fees, but we do not specific research the size of the auditee in our study. Simunic (1980) proved that size was positively related to audit fees and thereafter, many other studies found the same positive relation, according to the meta-analysis from Hay (2006). Besides this, according to the legitimacy theory, size and disclosures should also be positively related. Since larger firms have to deal with more shareholders and other stakeholders, firms tend to disclose more information. This theory, earlier described in this study, forms the basis and therefore we expect the size of auditee to be a logic determinant of audit fee disclosures. In line with the study of Simunic (1980) and Cullinan et al. (2016), we use the natural logarithm of assets to measure the size of the auditee.
4.6 Regression model
We use panel data to test whether our hypothesis can be adopted or rejected. Panel data typically provides time observations for a number of different individual variables (Smith, 2015). In this study, we use several time observations over the years 2012-2017, and we also make use of several different variables. By analyzing panel data, two techniques can eliminate heterogeneity bias, which is a fixed effects model and a random effects model (Smith, 2015). The fixed effects model is a form of regression that can control for variables that cannot be measured. The random effects approach is that time constant explanatory variables need not be eliminated and is appropriate to make inferences about the characteristics of the population when drawing
24
individuals randomly from a population (Smith, 2015). To decide whether we use either of one model, we run the Hausman test. We divide our study into four different models. In the first two models, we test whether internal controls and corporate governance characteristics affect the audit fee and quality of audit fee disclosure at non-listed firms. According to the Hausman test, we should use a random effects model for audit fees and the quality of audit fee disclosures. Thereafter, we test the influence of internal controls, corporate governance characteristics and the degree of listing on audit fee and the quality of audit fee disclosures at listed firms.
LNAUFi,t = β1 + β2 ICi,t + β3 ACMEETINGSi,t + β4 ACSIZEi,t + β5 LISTINGi,t + β6
LNASSETSi,t
AUFDi,t = β1 + β2 ICi,t + β3 ACMEETINGSi,t + β4 ACSIZEi,t + β5 LISTING,t + β6
25
5. Statistical analysis
5.1 Descriptive statistics
The sample of this study consists of all Dutch listed companies and 72 non-listed Dutch companies over the years 2012-2017. In the best case scenario, 900 observations could have been analyzed. However, due to IPO’s, delistings or exemptions, the observed number of audit fees are lower, because unavailable audit fees are left out of the sample. Furthermore, the audit fee disclosures observations are somewhat higher. This is because firms sometimes opt to disclose no specific information about their audit fee, due to exemptions according to Dutch law. However, this disclosure can be seen as an audit fee disclosure, because they had still the choice to disclose the information. Nevertheless, firms mostly opted to reveal only the reason why they decide to disclose nothing, and therefore they received a null score in our disclosure index. Other differences in the number of observations of our dependent variables are due to missing information in the annual reports of those companies.
Listed companies – Table 2
Variable Observations Mean Median Std. Dev. Min Max Independent variables AUF 416 4071,319 1300 7583,668 45 51000 LNAUF 416 7,211 7,17 1,50 3,807 10,84 AUFD 423 0,68 0,67 0,207 0 1 Dependent variables ACSIZE 421 3,159 3 1,351 0 7 ACMEETINGS 422 4,203 4 2,035 0 10 IC 428 0,741 1 0,438 0 1 LISTING 428 1,182 1 0,386 0 2 Control variable
ASSETS 423 4,16E+07 2889027 1,37E+08 6281 1,17E+09
26
Non-listed companies – Table 3
Variable Observations Mean Median Std. Dev. Min Max Independent variables AUF 350 1433,559 706,058 2060,281 13 15611 LNAUF 350 6,487 6,56 1,352 2,565 9,656 AUFD 375 0,527 0,55 0,237 0 1 Dependent variables ACSIZE 384 1,055 0 1,681 0 8 ACMEETINGS 384 1,0412 0 1,829 0 8 IC 384 0,331 0 0,471 0 1 Control variable
ASSETS 384 8,12E+06 1927959 2,64E+07 102,397 3,35E+08
LNASSETS 384 14,271 14,472 2,013 4,629 19,629
AUF The audit fee paid by the auditee to the auditor and its network combined LNAUF Natural logarithm of the total audit fee
AUFD Score on the audit fee disclosure index in percentages ACSIZE Size of the audit committee
ACMEETINGS Number of meetings of the audit committee in a particular year
IC Dummy variable whether a company has an internal audit function (=1)
LISTING Dummy variable whether a company is not listed (=0), listed in the Netherlands (=1) or cross-listed (=2)
ASSETS Total assets of a company in millions of euros LNASSETS Natural logarithm of total assets
Since we use panel data and have multiple relations to analyze, we address the multicollinearity issue in this study. Multicollinearity appears when two or more independent variables are correlated with each other. We perform the Pearson correlation matrix, to find out whether the dependent variables affect each other. The rule of thumb says that variables can be classified as severe collinear if its coefficient is either higher than 0,8 or lower than -0,8. As displayed below in table 4, no strong collinearity is discovered. However, the correlation between the size of the audit committee and the number of meetings is correlated with a coefficient of 0,6875. But, this is quite logical, since large committees have more different opinions and therefore, they need more meetings to discuss ongoing issues. Consequently, we decide to measure both variables separately during the regression in the next subsection. Furthermore, there might be collinearity issues between LNAUF and LNASSETS, since its correlation is 0,7865. However, these are the dependent and control variable and is evidence that there is a correlation between them.
27
Pearson correlation matrix – Table 4
In figures 4 and 5 below, we show the development of the audit fee and the audit fee disclosure. In figure 4, the audit fee disclosure score is presented. The scores of the disclosure index are significantly higher at listed firms than they are at non-listed firms. The audit fee disclosure at non-listed firms (orange line) starts at 51 percent in 2012 and ends with 62 percent in 2017. This shows that the quality of the audit fee disclosure is increasing over the years at non-listed firms. Nevertheless, also the quality of the audit fee disclosure at listed firms is growing over the last six years. Where it started with an average disclosure score of 64 percent, it has been increased to 70 percent in 2017. Apparently, firms opt to disclose more about their audit fees in 2012 than they did in 2017. Although, no specific regulation is introduced during these years. Moreover, the legislation on the audit fee disclosure is already set up in 2011 (Langendijk, 2011).
Next to this, in figure 5 we show the development of the audit fees in our sample. On the left y-axis the audit fee of the listed firms are calculated, and for the non-listed firms, the fees are calculated on the right y-axis. Apparently, the audit fee for listed firms is higher than the audit fees for non-listed firms. Since the listed firms are, on average, larger and are bound to more legislation, this result is as expected. Moreover, it is in line with literature that states that size, complexity and litigation risk are the most critical determinants of audit fees (Hay, 2013). Furthermore, the audit fees for both groups undergo a similar development over the years. There is a decrease in audit fee during the years 2014 till 2016. Subsequently, the fee has grown for
LNAUF AUFD IC ACMEETINGS ACSIZE LISTING LNASSETS
LNAUF 1,0000 AUFD 0,2921*** 1,0000 IC 0,3883*** 0,2081*** 1,0000 ACMEETINGS 0,3596*** 0,3132*** 0,6186*** 1,0000 ACSIZE 0,4048*** 0,2526*** 0,5738*** 0,6875*** 1,0000 LISTING 0,3629*** 0,3140*** 0,4404*** 0,6495*** 0,5908*** 1,0000 LNASSETS 0,7865*** 0,1140*** 0,4161*** 0,3088*** 0,3765*** 0,3050*** 1,0000
28
both groups in 2017. This is possibly the result of the mandatory audit firm rotation, which started in 2014 (EU Regulation 537/2014). Due to this rotation, audit firms ask lower fees in the initial year in order to beat competitors and acquire new clients (Hay et al., 2013).
Figure 4 Figure 5
In figure 6 below, we show the different fees paid by listed firms in the sample. A notable development is that the fees for non-assurance services and tax services declined over the years 2012 to 2017. Audit firms become more and more an ‘audit only’ firm, as also remarked in the study of Kevelam, ter Hoeven & Brouwer (2017). This is partly the result of Dutch legislation, which prohibits audit firms to provide both audit and advisory services to one specific company with public interests.
Figure 6 64 66 67 69 70 70 0 50 100 2012 2013 2014 2015 2016 2017 Years
Audit fee disclosure index
listed non-listed 1200 1300 1400 1500 1600 3600 3800 4000 4200 4400 2012 2013 2014 2015 2016 2017
Audit fees
listed non-listed 0 1000 2000 3000 4000 5000 6000 2012 2013 2014 2015 2016 2017Overview different service fees
29
In figure 7 below, the development of the internal auditor is graphically presented. What we see is that internal auditor is more and more present within companies at listed and non-listed firms. In 2012 only 61 percent of the listed firms had an internal auditor, 5 years later, in 2017 this percentage increased to 84 percent. At non-listed firms, the amount doubled from 18 to 36 percent of the companies over the same period. This is probably the result of the new corporate governance code, as mentioned in the introduction (MCGC, 2016). This code prescribes to install an internal audit department and apparently, companies anticipated on this code.
Figure 7 0 20 40 60 80 100 2012 2013 2014 2015 2016 2017
Internal auditor
30
5.2 Results on audit fees
In this section, we show the results of the performed regression. First, we present the findings regarding the regression on audit fees, thereafter, we present the regression on the quality of audit fee disclosures. As said, we ran a GLS regression random effects model, as a result of the Hausman test. Initially, we test all variables individually with audit fees, and afterward, we take all the variables together. All variables and hypotheses are allocated to different models: model A till model H. The results are shown in table 5 and 6, presented on the next pages. The first table contains the results of the regression on audit fees at non-listed firms, and table 6 shows the results of the listed firms.
5.2.1 Results on audit fees at listed firms
The first hypothesis is shown in Model A, and only the size of the assets has a significant influence on the audit fees in this regression. However, the R² is quite high. The R² shows the variance of the dependent variable that explains the independent variable. This is a number between 0 and 1, and a high number means that this model fits well for our observations. Nevertheless, since internal control is not significant, we reject the first hypothesis.
In Model B and C, we test hypothesis 2, regarding the relationship between corporate governance effectiveness and audit fees. We measured this separately since the size of the audit committee and the number of meetings were quite highly correlated in the Pearson matrix. The number of meetings is not associated with audit fees. However, the size of the audit committee is related to audit fees. Therefore, we accept hypothesis 2b, but we reject hypothesis 2a for listed firms.
Model D concerns the relationship between the degree of the stock market listing and the audit fees. Obviously, we only test this hypothesis at listed firms. The results show a significant positive relation, at a significance level of 0.05. This means that cross-listed firms are related to higher audit fees than firms that are only listed in the Netherlands. Therefore, we accept hypothesis 3. In model E, F and G, we test whether the degree of listing has a reinforcing effect on the other relations. In Model E, we see that internal control in combination with the listing of a firm leads to a significant negative relationship. This significant negative relation is also discovered in model F. Despite that there is no significance found in model G between the size of the audit committee and audit fees. We can conclude that the input of listing has some effect
31
on the remaining relations. However, all hypotheses are rejected, due to the negative signs of the coefficients, while we expected positive relations. In Model H, all variables are tested simultaneously in one regression. We see similar results for most of the tested relationships. Even though, also the number of meetings of the audit committee has a positive effect on the audit fees. On the other side, the reinforcing effect of the degree of listing on the relationship between internal control and audit fees does not hold anymore. Overall, the R² is relatively high in all models at listed firms. Thus, the variables give a decent representation of the variance in the audit fee variable.
5.2.2 Results on audit fees at non-listed firms
Model A for non-listed firms also indicates that there is no statistically significant relationship between the presence of an internal auditor and audit fees. Therefore, we reject hypothesis 1 for non-listed firms. In Model B and C, we test hypothesis 2a and 2b, regarding the relationship between corporate governance effectiveness and audit fees. In the end, we reject both hypotheses. In Model D and E, we test hypotheses 1 and 2 simultaneously. Also, no evidence was found in those models, except for the size of the assets. Apparently, the size of assets is a truly determinant of audit fees and fits as a control variable, since the size of assets is significant in all models.
Regression result on audit fees at non-listed firms – Table 5
Model A Model B Model C Model D Model E
Related to hypothesis 1 2 2 1+2a 1+2b
Variable Expected sign Coefficient Coefficient Coefficient Coefficient Coefficient
Intercept 2,4823*** 2,4950*** 2,4456*** 2,4985*** 2,4839*** IC + 0,1660 0,1361 0,1556 ACMEETINGS + 0,0387 0,0226 ACSIZE + 0,0235 0,01 LNASSETS + 0,2695*** 0,2695*** 0,2740*** 0,2674*** 0,2687*** R² 0,4991 0,5180 0,5174 0,5018 0,4998 F-value 57,63 56,93 55,83 58,58 57,87 N 350 350 350 350 350 Model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model
32
Regression results on the audit fees at listed firms – Table 6
Model A Model B Model C Model D Model E Model F Model G Model H
Related to hypothesis 1 2a 2b 3 4 5a 5b All
Variable Expected sign Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
Intercept -0,4972 -0,1084 0,1926 -0,3772 0,1708 0,40310 0,1502 0,3795 IC + 0,0622 0,0835 0,1399 ACMEETINGS + 0,0258 0,0483*** 0,0298** ACSIZE + 0,1101*** 0,1275*** 0,1095*** LISTING + 0,5268** 1,1043** 0,7627** 0,9428*** 1,3369*** IC x LISTING + -0,5951** -0,3321 ACMEETINGS x LISTING + -0,0829*** -0,0567** ACSIZE x LISTING + -0,06520 -0,05 LNASSETS + 0,4773*** 0,4771*** 0,4411*** 0,4611*** 0,4555*** 0,4177*** 0,4479*** 0,4142*** R² 0,7150 0,7164 0,7147 0,7122 0,7139 0,7094 0,7148 0,7102 F-value 277,28 282,51 307,04 281,99 289,52 313,83 299,88 324,18 N 416 416 416 416 416 416 416 416 Model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model
33
5.3 Results on the quality of audit fee disclosures
In this section, we present the results of the regression on the quality of the audit fee disclosure. Again, we divide our sample into listed and non-listed firms, and the same variables are researched as before for audit fees. We ran a GLS regression random effects model, since the result of the Hausman test indicated the random effects model was most appropriate. The results are shown in table 5 and 6, on the next pages.
5.3.1 Results on the quality of audit fee disclosures at listed firms
In Model A, we test whether having an internal auditor influenced the quality of the audit fee disclosure. The result shows that there is a significant relationship between them. Furthermore, also the size of assets of the auditee affects this quality. However, the combined R² of both variables is low. This means that the variance in the quality of audit fee disclosure variable is poorly represented by the variables size of assets and internal control. Finally, we accept hypothesis 6 for listed firms.
In Model B and C, we test the corporate governance effectiveness variables separately on the quality of audit fee disclosures. In Model B, we see that the number of meetings of the audit committee has a significant positive influence on the quality of audit fee disclosures. Although, the R square is low. However, the size of the audit committee does not have a significant relationship with the dependent variable. To conclude, we accept hypothesis 7a for listed firms and reject hypothesis 7b for listed firms.
In Model D, we test hypothesis 8 concerning the relation between the degree of the listing and the dependent variable. Results do not show a significant relationship besides the significant effect of the size of assets of the auditee. Also, the R² is low. Therefore, we reject hypothesis 8. In the sequent models E, F and G, we test the moderating effect of the degree of listing on the quality of audit fee disclosures. Also, no significant relations are found in the moderating relationships. However, the variables; the presence of an internal auditor and the number of meetings of the audit committee remain positively significant. Also here, the size of the assets of the auditee stay positively significant. We conclude that we reject hypotheses 9, 10a and 10b.
34
In Model H, we test all variables simultaneously. Only the variables the number of meetings of the audit committee and the size of assets remain positively significant. The R² is a bit higher, compared to other relations in this table. However, it is still quite low.
5.3.2 Results on the quality of audit fee disclosures at non-listed firms
At non-listed firms, results show a significant negative relationship between the size of assets and the quality of audit fee disclosures. But, there is no evidence found for a significant relationship between internal control and the quality of audit fee disclosures. We reject hypothesis 6 for non-listed firms. There is a significant positive relationship between the size of the audit committee and the quality of audit fee disclosure. Although, there is no
statistically significant relationship between the number of meetings of the audit committee and the quality of audit fee disclosures. Eventually, we accept hypothesis 7b for non-listed firms and reject 7a for non-listed firms.
In table 5 for non-listed firms in model D and E, no evidence is found for the combinations of the proposed hypotheses. Except for our control variable, there is located a reversed negatively significant relation between the size of assets and the quality of audit fee disclosures.
Table regression results on the quality of audit fee disclosures at non-listed firms – Table 7
Model A Model B Model C Model D Model E
Related to hypothesis 6 7a 7b 6+7a 6+7b
Variable Expected sign Coefficient Coefficient Coefficient Coefficient Coefficient
Intercept 0,7169*** 0,7159*** 0,7296*** 0,7123*** 0,7312*** IC + 0,0062 0,0127 -0,0075 ACMEETINGS + -0,0026 -0,0043 ACSIZE + 0,0143* 0,0149* LNASSETS + (0,0129*) (0,0125*) (0,0147**) (0,0125*) (0,0147**) R² 0,0001 0,0025 0,0074 0,0037 0,0153 F-value 3,51 3,55 6,77 3,73 6,83 N 375 375 375 375 375 Model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model GLS regression random effects model
