The effects of agency conflicts on the demand of non-audit services:
Evidence from Germany
Name: Diem Mi Nguyen Student number: 11108819
Thesis supervisor: dhr. dr. Alexandros Sikalidis Date: June 26, 2017
Word count: 13,683
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
2
Statement of Originality
This document is written by student Diem Mi Nguyen who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no
sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of
3
Abstract
The joint provision of audit and non-audit services has been an ongoing and highly-debated topic. It has been argued that the economic bond between auditors and its clients may compromise the auditor’s independence. Regulators have therefore taken actions to curtail auditor-provided audit services, even though studies have shown that providing non-audit services to non-audit clients could result in positive knowledge spillovers. This paper examines whether firms with higher agency costs purchase less non-audit services from their incumbent auditor. The empirical results show that performance-based compensation determines the audit client’s demand for non-audit services, while there is no significant relation found with leverage and ownership structure. Contrary to prior extant literature which focused on Anglo-Saxon countries, this study examines a continental European setting. The main contributions of this paper are in terms of its timeliness with respect to regulatory changes in Germany, in which it may shed light on the NAS mechanisms of the audit profession in other continental European countries as well.
Keywords: Auditor independence, Non-audit services, Ownership composition, Management compensation, Leverage
4
Table of Contents
1 INTRODUCTION... 5
2 BACKGROUND AND LITERATURE REVIEW ... 8
2.1 THE AUDITING ENVIRONMENT IN GERMANY ... 8
2.2 AGENCY THEORY ... 9
2.3 AUDITOR INDEPENDENCE ... 10
2.4 JOINT PROVISION OF AUDIT SERVICES AND NON-AUDIT SERVICES ... 13
2.5 HYPOTHESES DEVELOPMENT ... 16 2.5.1 Ownership composition ... 16 2.5.2 Management compensation ... 17 2.5.3 Leverage ... 18 3 METHODOLOGY ... 20 3.1 RESEARCH DESIGN ... 20 3.2 SAMPLE COLLECTION ... 20 3.3 VARIABLES ... 21 3.3.1 Dependent variable ... 21 3.3.2 Independent variables ... 24 4 RESULTS ... 27 4.1 DESCRIPTIVE STATISTICS ... 28 4.2 CORRELATION MATRIX ... 28 4.3 REGRESSION ANALYSIS ... 30
4.3.1 Main regression model ... 30
4.3.2 Sensitivity analysis... 31 5 DISCUSSION ... 34 6 CONCLUSION ... 37 REFERENCES ... 39 APPENDICES ... 42 APPENDIX 1 ... 42 APPENDIX 2 ... 43
5
1
Introduction
The role of the auditor is to provide an independent opinion so that users of the financial statements can rely on the information presented in the reports. Ideally, auditors should therefore not have any business interest in their audit clients due to the fact that the auditor should serve as an independent link between management and stakeholders who rely on the financial statements for decision-making. Hence, the audit practice has played an influential role in enhancing the effective functioning of the business environment (Salehi, 2009). Auditors are deemed to be independent, yet they are hired and compensated by their client, which makes it difficult for the auditor to stay completely integer. Due to a series of accounting scandals that marked the beginning of a new millennium, great concern has clearly been on the rise over the last decades regarding the independence of auditors and the audit profession has been undermined. Consequently, at the center of the debate among the accounting profession, investors and regulators regarding auditor independence is the provision of non-audit services by audit firms to their clients. Non-audit services (NAS) are defined as any other professional service provided by a registered public accounting firm, other than an audit or a review of the financial statements of the client. From the perspective of the agency theory, management (agents) may have different motivations than shareholders (principals). So while it is important for capitals markets that the auditor provides an objective independent opinion; the audit client hires the auditor and pays for its compensation. It has therefore been advocated that companies with more agency conflicts are predicted to purchase less NAS from their incumbent auditor (Beattie & Fearnley, 2002; Firth, 1997 as cited in Svanström & Sundgren, 2012; Parkash & Venable, 1993).
While prior literature is not conclusive, there is evidence that when auditors provide NAS to their audit clients, the auditor independence is impaired by favoring the interests of clients over the interest of capital markets (Beattie & Fearnley, 2002; Beck et al. 1988 as cited in Quick, Sattler & Wiemann, 2013; Zerni, 2012). On the other hand, it has been posited that providing NAS enhances an auditor’s knowledge of the client’s business by ensuing a more effective audit (Dedman, Kausar & Lennox 2014; Knechel, Sharma & Sharma, 2012; Svanström & Sundgren, 2012). Similarly, investors are not disturbed by the impairment of auditor independence because providing NAS is considered as a means to generate profits which in return compensates for the loss in auditor independence (Campa & Donnelly, 2015). In contrast to the aforementioned studies, other studies indicated that there is no significant
6 relationship between agency costs and non-audit services (Alshawish, Abed & Hamdallah, 2015; Simunic, 1984 and Beck et al., 1988 as both cited in Quick et al., 2013; Quick et al., 2013). The International Federation of Accountants (IFAC) Code of Ethics and the Eighth Directive of the European Union (EU) both included restrictions and safeguards to strengthen the auditor independence. The main objective to reform the regulatory framework was to mitigate the increasing risk of conflicts of interests for audit firms when auditing public interest entities (PIEs) (European Commission, 2014). The European Commission (EC) introduced a so-called ‘black list’ to avoid any circumstances that might compromise the independence of auditors. This topic remains to this day an important subject for two reasons. First, the framework can be amended once more if it is still found to be ineffective in its initial function. Second, it only applies to PIEs and is therefore not binding on all firms such as private or non-listed firms, so the effects of NAS on the auditor’s independence remain relevant in those firms. Thus, the following research question in this research stands central:
“Do agency conflicts impact the demand of non-audit services?”
In this study the following three factors will be examined – which are served as proxies for agency costs – are (1) ownership structure, (2) leverage and (3) performance-based compensation. To the best of my knowledge, very few researches have focused on continental European countries. I expect different results in a continental European setting as opposed to an Anglo-Saxon setting due to the different legal and financial reporting environments. The former has Civil law where there is generally a weaker corporate governance and low audit quality, whereas the latter has Common law where they typically have a more effective corporate governance and higher audit quality (Barniv, Myring & Thomas, 2005). Audit committees are rare in continental Europe because it is voluntary to institute an audit committee. An audit committee can help to offset the wrong incentives that management might have and in its place provide incentives to act in the shareholder’s best interests. Moreover, the corporate ownership in Anglo-Saxon companies is diffused through public shares offering, while the ownership in continental European countries is more concentrated. Within these companies, firms are predominantly controlled by private investors, as opposed to German companies where institutional ownership is more relevant. This is because agency costs impact the demand more for NAS in countries where private shareholding is dominant (Parkash & Venable, 1993 as cited by Quick et al., 2013).
7 This study contributes to the existing literature by a number of considerations. First, the primary contribution and aim of this paper is to determine whether agency conflicts in a firm increases the demand for non-audit services in Germany. Köhler and Ratzinger-Sakel (2012) showed that the audit information of the Germany industry is internationally comparable thus, implying that the results of this research could shed some light to other continental European countries. As mentioned before, prior research was largely conducted in Anglo-Saxon countries. Second, the banking, insurance and financial services industries are also examined. Auditing activities play an essential role in mitigating information asymmetries in an opaque environment, but the intermediation risk and the intricacies of loans and financial assets makes it more difficult to monitor for outsiders. Moreover, they deal with more transaction accounts with complex activities which increase the difficulties in monitoring (Cameran & Perotti, 2014). This means that the need for an independent auditor increases as well. Since these three industries are mostly excluded in prior literature it can enlarge our knowledge of the interdependent economy related to auditing. Third, the paper of Quick et al. (2013) only covers a short timeframe of three fiscal years; therefore this study increases the validity of the research by examining a longer time-period from 2005 to 2014. This longer time-period shows the possible consequences of the after-math of prohibiting several NAS in 2004 and it includes the period where Germany in 2009 reformed its audit regulation.
Overall, the results of this study show that agency conflicts do not determine the demand of non-audit services. However, the proportion of performance-based compensation, size of a company, BIG-4 auditor and an audit committee play a significant role in the client’s willingness to purchase non-audit services from their incumbent auditor. In contrast with prior literature, this study finds a significant relation between performance-based compensation and NAS, suggesting that managers purchase more NAS from their incumbent auditor as it will generate more profits for the firm and lead to a higher bonus. Campa and Donnelly (2015) posited that this action compensates for the loss in auditor independence since it will likewise lead to higher dividend payouts. The remainder of this thesis is organized as follows. The next chapter states the literature review and the hypotheses development. The third chapter describes the research design and the data and sample selection. The results are presented in the fourth chapter, followed by the discussion in the fifth chapter. The final chapter contains the conclusions and limitations of this study and provides suggestions for further research.
8
2
Background and literature review
2.1 The auditing environment in Germany
Due to a series of corporate scandals in the United States (U.S.) the Sarbanes-Oxley Act (SOX) was introduced in 2002 and this act restricted auditors from providing NAS to their audit clients. SOX also forced auditors to put more effort in their audit tasks and a better monitoring of potential fraudulent activities (Amir, Guan & Livne, 2010). The extant research pertaining to auditor independence is therefore largely centered on the U.S. as a result of earlier audit failures. This has ultimately led to radical changes in the EU where the EC reformed the Eighth Company Law Directive on Statutory Audit to enhance the role of internal auditors, audit committees and corporate governance (Jankovič, Ivankovič & Jerman, 2010). Despite the harmonization of IFAC’s International Standards on Auditing (ISA) and national auditing standards, Germany maintained several characteristics that could impact the results. For example, the corporate governance in German limited corporation (Aktiengeselleschaften) has certain particularities. They (a) are oriented around stakeholders, (b) have an insider control system and (c) have a clear separation between management and supervision, where there is a two-tier system as opposed to the Anglo-Saxon one-tier system. The German Stock Corporation Act obliges all listed firms within Germany to have a dual board structure that includes a supervisory board (Aufsichtsrat) and a management board (Vorstand) (Cromme, 2005 as cited in Woods, Kajüter & Linsley, 2007). As a result, shareholders and investors may be less apprehensive about the independence of the auditors, since they rely on the monitoring of the supervisory board (Quick & Warming-Rasmussen, 2009). Another typical characteristic of the audit market in Germany is the highly-concentrated ownership composition (Prowse, 1995 as cited by Gassen & Skaife, 2009). It has been argued that a highly-concentrated ownership composition is coherent with firms that hold more liabilities than equity (La Porta et al., 1997 as cited Gassen & Skaife, 2009). Moreover, Gassan and Skaife (2009) claimed that German private debt holders place less reliance on audits as a resource of information since private debt holders generally have access to firm information already.
Germany effected Article 49 of Directive 2006/43/EG through the Accounting Law Reform Act (Bilanzrechtsreformgesetz (BilReg)) in 2005, as a means to require listed companies to disclose information regarding audit and non-audit fees. On top of that, the Accounting Law Modernization Act (Bilanzrechtsmodernisierungsgesetz (BilMoG)) became effective in 2009
9 and this extended the requirements concerning fees disclosures. Although Germany has certain distinct particularities as mentioned before, Köhler and Ratzinger-Sakel (2012) showed that the results of Germany are internationally comparable, which means that Germany provides a suitable setting to conduct this research, since it could shed some light to other continental European countries as well.
2.2 Agency theory
The agency theory seeks to explain information asymmetries that are caused by the misalignment of interests between agents (management) and principals (shareholders). In general, shareholders appoint a team of managers who are authorized to make decisions on the shareholder’s behalf to maximize the firm’s value. This means that shareholders are delegating their responsibilities to management, but such delegation also requires that shareholders should trust managers that they act in the shareholder’s best interests. Because, managers might be influenced by various factors such as personal and financial gains and will therefore pursue personal interest of increasing incentives that will not fully benefit the shareholders. Such “hidden action” behavior cannot be observed by investors or creditors (Arrow, 1985 as cited in Quick & Warming-Rasmussen, 2009). The misalignment of interests arises due to the separation of ownership and control (Jensen & Meckling, 1979) and, the agency problem is likely to increase where one party has an information advantage over another party, which is defined as information asymmetry. Hence, management might have a tendency to report more optimistically or have an inducement to bias information flows. Different motivations lead to uncertainties about the reliability of the financial statement which will have an impact on the level of trusts that shareholders have in their selected managers. For this reason, shareholders are skeptical of the decisions made by management, because any decision made by management which leads to a loss in forgone wealth to the shareholders represents an agency cost. To deal with the agency problem, audit committees perform an oversight role and are charged with supervising and monitoring roles on behalf of the investors. In addition, different mechanisms are put in place to align the interests of agents with principals such as performance-based compensations. These bonuses are expected to restrain potential opportunistic behavior of managers. Jensen and Meckling (1976) define agency costs as the sum of (a) the monitoring expenditures by the principles, (b) the bonding expenditures by the agent and (c) the residual loss. Capital markets consider the financial statements as doubtful of companies that are done by auditors who are not
10 perceived as independent, and thus social costs are incurred. Because these firms are considered as risky investments therefore, investors and shareholders demand a higher risk premium, and thus, the cost of capital increases (Quick & Warming-Rasmussen, 2009). Under these circumstances, investors are concerned by the close relationship between the audit firm and its clients, and fear that the auditor independence may be compromised, leading to a reduction of audit quality (Quick & Warming-Rasmussen, 2015). Therefore, audit committees have an important role in the audit process because of their responsibilities to appoint, terminate and determine the compensation of the auditor. Jensen and Meckling (1976) hypothesized that company size plays a vital role on agency costs because within a larger firm it is more likely that monitoring is more complicated and expensive. With this in mind, investors in firms with relatively high agency costs are peculiarly troubled by the economic bond between the auditor and its clients. Due to this, they are faced with a tradeoff of costs. It is important to note that external audits can only serve its purpose in mitigating agency conflicts if auditor independence is achieved (Hallak & Silva, 2012). Otherwise, it would only mean that the information provided in the audited financial statements is more likely to be in favor of the client firm. Therefore, the demand for high audit quality and independent audits increases with high agency problems (Chow, 1982 as cited by Svanström & Sundgren, 2012).
2.3 Auditor independence
Auditor independence has traditionally been regarded as a keystone of the audit profession since various stakeholders use the audited financial statements to make decisions, based upon the information that is provided within these statements. To assure that the financial statements are free from material misstatements, the role of the auditor is to provide users with reasonable assurance that the information is reliable by expressing an independent opinion on the financial statements (Jankovič et al., 2010). Although it is very important for auditors to remain independent; in practice it has often showcased that it could be quite challenging. Especially when the fees of the client signify a significant portion of the auditor’s income. Because on the same note, auditors are operating in a highly competitive market, whereby clients have an extensive choice of auditors. In a service provider-client relationship, it is particularly important that the auditor meets the needs of its clients (Patel & Prasad, 2013). Clients are therefore able to abuse their position to pressurize auditors into adopting incorrect accounting criteria or fraudulent activities.
11 The definition of auditor independence has been divided by the IFAC into ‘independence of mind’ and ‘independence in appearance’. The former is when the auditor maintains an unbiased attitude and his professional judgment has not been compromised throughout the audit whereas, independence in appearance means that the auditor should avoid all circumstances that a reasonable informed third party – bearing in mind all relevant facts and circumstances – would conclude that the integrity, objectivity or professional skepticism of the professional auditor has been comprised (Hayes, Gortemaker & Wallage, 2014). The term independence in appearance is regularly interchanged with ‘perceived independence’. The joint provision of audit and non-audit services negatively affects both attributes of auditor independence. The independence of mind is impaired when auditors are economically depended on the compensation of their client and therefore report more favorably because they are afraid of losing the client.
The International Ethics Standards Board for Accountants (IESBA) sets standards of conducts and states five fundamental principles that all auditors should comply with. These are integrity, objectivity, professional competence and due care, confidentiality and professional behavior (Hayes et al., 2014). Nonetheless, the regular treatment to auditor independence remains for the most part the same after the amendment of the Statutory Audit Directive (Knechel et al., 2012). The Statutory Audit Directive still embodies the vital independence aspects, which are the five potential threats to the fundamental principles (IFAC, 2015). The first threat is a self-interest threat and occurs when an auditor benefits from a financial or other self-interest conflict with an audit client. In the case of non-audit services, this could be caused by the added fees derived from NAS. Second, a self-review threat, which is the threat where (1) a professional auditor relies on the judgment of a previous engagement or (2) when the auditor previously was an employee of the client, where he was in the position to exert significant influence over the subject matter. Third, advocacy threats can occur when an auditor seems to promote the positions of their clients, because the risk that the auditor identifies with the interest of the client is higher when the provision of NAS is more significant. Fourth, familiarity threats may occur from having close relationships, since the provision of NAS such as advisory service entails a great deal of mutual trust between the audit firm and the client. Familiarity threats may also arise when the auditor of the engagement team has an immediate family member who is a director. Consequently, this can result in excessive trust in the client and an inadequate objective
12 testing of accounting data (Quick et al. 2013). Lastly, an intimidation threat; auditors can be dissuaded to act objectively because clients can threaten auditors with cessation of their contract (IFAC, 2015). In order to reduce or eliminate the aforementioned threats, safeguards depending on the situation are identified and applied, for example, an additional auditor who is not on the assurance team will review the work done or the senior assurance team personnel will be rotated. Auditors should decline or terminate the engagement when safeguards are unable to eliminate the aforementioned threats (Hayes et al., 2014), considering the fact that they should be deemed as ‘independent of mind’ and ‘independent in appearance’ (IFAC, 2015).
The accusation of inappropriate accounting treatments due to the compromised independence has resulted in several restrictions on non-audit services to PIEs by statutory auditors, audit firms or members within their networks (EC, 2014). The objective to reform the EU regulatory framework once more – which became applicable on 17 June 2016 – was to increase the quality of statutory audit. When auditors provide non-audit services to their audit clients, regulators fear two consequences that NAS has on auditor independence. The first concern is that NAS fees make auditors financially dependent on their clients, because economic dependence could influence the auditor to report favorably for fear of losing the client and its compensation. The other concern is that the nature of some non-audit services puts auditors in managerial roles, which could potentially threat their objectivity (DeFond, Raghunandan & Subramanyam, 2002). The new measure includes a list of prohibitions of provisions of certain non-audit services such as (a) tax and tax compliance services, (b) services that play a role in management or decision-making, (c) services linked to the financing, capital structure and allocation as well as investment strategy of the audit client, and finally (d) either promoting, dealing in or underwriting shares (PwC, 2016). It also includes the requirement that when auditors wish to provide NAS to their clients, that are not specifically prohibited, it should be permitted first by the audit committee and also disclosed to their shareholders (Hayes et al., 2014). Besides restricting certain NAS, auditors are also required to disclose the amounts of fees received. However to this day, there is no clear evidence whether current regulations are working and whether the changes of the EC are actually necessary (Campa & Donnelly, 2015). Ratzinger and Schönberger (2015) also argued that it is debatable whether restricting NAS will solve the perception problems of the profession.
13 Before the amendment of the Eight Directive, SOX was introduced in 2002 to enhance auditor independence to reduce the cost of capital (Amir et al., 2010). Anecdotal evidence pertaining to the United States also supports the regulators’ argument that substantial amounts of NAS fees paid to the auditor played a factor in some of the audit failures such as Enron (US Senate, 2002 as cited in Knechel et al., 2012). The authors argued that SOX led to a better alignment of the conflicting interests between shareholders and managers, but this came at the expense of lenders (DeFond et al., 2008 as cited by Amir et al., 2010). The alleged benefit of an enhanced independence may therefore only be applicable to shareholders and not for lenders as they will be unable to moderate their cost of capital (Amir et al., 2010). In addition, stakeholders can be skeptical about the independence of auditors when they are aware that the same auditor receives a lucrative amount of NAS fees from the same clients as well (Quick &Warming-Rasmussen, 2009). Campa and Donnelly (2015) revealed that independence of mind is compromised when auditors provide NAS to their clients, especially if audit expenditures are less than anticipated. But they further suggest that investors are not concerned by this impairment, because the provision of NAS is deemed as a means to generate more profits for the firm which recompense for the loss in auditor independence. Furthermore, Ratzinger-Sakel and Schönberger (2015) showed that even in the case of enhanced independence in appearance, restricting NAS might be a trade-off between the loss of audit quality-enhancing aspects of NAS via knowledge spillovers and production efficiencies. Knowledge spillover is the general term that is use to define knowledge transfers (Patel & Prasad, 2013). The presumption that the regulators have, is that it will strengthen the overall positioning of external auditors and its independence when auditors do not provide NAS to their clients and disclose more information (Amir et al., 2010). However, other authors scrutinized these assumptions and found no impairment of auditor independence (Köhler & Ratzinger-Sakel, 2012). In her recent research, Ratzinger-Sakel and Schönberger (2015) argued that if auditor independence is properly secured, the joint provision of audit services and NAS may possibly strengthen the role with its client.
2.4 Joint provision of audit services and non-audit services
The provision of NAS provided by the incumbent auditor has been a much-debated topic pertaining to auditor independence, and evidence of prior literature shows contradictory effects. The primary belief is that over time auditors rely on the enlarged fees of these clients and may likely be coerced to concede dubious accounting practices, which threatens the
14 integrity of the financial statements (Mitra & Hossain, 2007). Consequently, this damages the independence in appearance and reputation of auditors because it reduces the level of trust that investors and stakeholders have. For firms who face acute agency conflicts this is particularly risky. It has therefore been argued that these firms are less likely to purchase NAS from their incumbent auditor (Beck et al., 1988 as cited in Quick et al., 2013; Knechel et al., 2012; Salehi, 2009; Zerni, 2012). The objection of the profession suggests that prohibiting NAS from auditors leads to a decrease of audit effectiveness and efficiency. Prior research has generally supported the argument that providing NAS by the incumbent auditor does not diminish the quality of the audit (Knechel et al., 2012). However, the academic literature studying this subject matter does not provide distinct findings nor does it constantly reach consistent conclusions as those adopted by the legislators. From the perspective of the shareholders, Salehi (2009) showed that shareholders strongly believe that providing NAS to audit clients significantly compromises the auditor’s independence. Moreover, the results indicated that separating the audit and non-audit departments are not perceived as beneficial, suggesting that shareholders prefer to separate the audit and non-audit services by having different audit firms to perform one of the tasks. In the early research of Beck et al. (1988, as cited in Quick et al., 2013) it was found that firms with higher agency costs want to reduce their costs of capital by improving their perceived independence. In order to do so, they found that these firms reduce the cut of NAS that they acquire from the audit firms. Zerni (2012) examined whether agency conflicts affect a client firm’s decision to purchase NAS from their incumbent auditor too. His findings reveal that the willingness of client firms with embedded agency problems are negatively correlated with the demand of NAS from their incumbent auditor. As an explanation for his findings, he claims that client firms with high agency costs demand higher quality audits to safeguard the auditor’s independence in appearance, and therefore take proactive measures to protect their perceived integrity of the audit process.
On the other hand, it has been posited that when auditors provide non-audit services to their audit clients, they already possess the necessary knowledge, capabilities, and technical competence resulting in an economically efficient service (Patel & Prasad, 2013). Because the incumbent auditor already has better intimate knowledge of the client’s needs and business and is therefore more capable to offer better services at a lower cost resulting in knowledge spillovers. Knechel et al. (2012) tested their hypotheses in New Zealand which provided a neutral setting since the government does not prohibit the joint provision of audit
15 and non-audit services, thus there is no regulation concerning audit firms. The authors found that the provision of NAS to their audit clients resulted in knowledge spillovers, which in due course leads to audit efficiency and reductions in costs. Likewise, Dedman et al. (2014) found that companies with higher agency costs who purchased NAS were also inclined to continue purchasing an audit.
Contrary to the previous studies mentioned, some studies did not find any relationship between the level of agency costs and the demand for non-audit services. Svanström and Sundgren (2012) examined the choices of purchasing NAS from the incumbent auditor of 322 Swedish SMEs. Their results indicated that the purchase of non-audit services from the incumbent auditor significantly increases with the duration of the auditor-client relationship. They also found that there is a positive association between the joint provision of audit and NAS and the perceived audit quality. Even though their result is consistent with the view that knowledge spillover arises and the incumbent audit firm becomes more cost efficient, they were unable to find any support that agency costs are negatively associated with the joint provision in SMEs. Yet, ample evidence from prior literature (Alshawish et al., 2015; Amir et al., 2010; Campa & Donnelly, 2015; Quick et al., 2013; Zerni, 2012) showed that company size plays a role in purchasing NAS. It is alleged that larger firms require more consulting services and therefore purchase higher amounts of NAS. Quick et al. (2013) focused on the 160 largest German stock-listed companies. They find that larger firms were positively correlated with the demand of consulting services. But, the results showed that overall, agency conflicts do not determine the demand of NAS, where they failed to find evidence that performance-based compensation and leverage is significantly correlated with NAS. These insignificances might be caused because the joint provision is not perceived as harmful by investors and creditors due to several reasons. First, large audit firms usually separate their departments; therefore different teams will provide the different services. Second, management and investors might even benefit from the knowledge spillovers since the auditor becomes more cost efficient. Third, prohibiting certain NAS could have established a sense of security, thereby protecting investors in spite of prominent agency conflicts (Quick et al., 2013). The results of Alshawish et al. (2015) also showed that there is no significant impact of agency cost on the demand of NAS. But contrary to the study of Quick et al. (2013) the authors found significant correlation between leverage and NAS purchases.
16
2.5 Hypotheses development
This study focuses on three important factors that may affect the demand of non-audit services: (1) ownership structure, (2) leverage and (3) performance-based compensation. The effects of these determinants as hypothesized in this research study are presented in the conceptual model in Figure 1. Related notions are posited in the subsequent subparagraphs.
Figure 1 Conceptual model
2.5.1 Ownership composition
Agency conflicts arise due to the separation of ownership and control (Jensen & Meckling, 1976). This separation leads to two parties with different goals in a firm and to information asymmetry, where management typically has an information advantage since they are more embedded in the firm’s operations. While the abundance of prior literature on the information role looked at common law countries, where information asymmetry produces a homogenous demand for audits, this study focuses on the German market where the demand for audits is more heterogeneous due to the disparities in the ownership concentration and capital structure of German companies (Ashbaugh & Warfield, 2003 as cited by Gassen & Skaife, 2009). The composition of ownership is vital to the levels of monitoring by investors, since management often have different incentives than its investors. The incentive that investors have to monitor
Agency cost Management compensation Ownership composition Leverage H2 (+) Non-Audit Services
17 depends on the sizable stake of blockholders (Mitra & Hossain, 2007). For the reason that dispersed shareholders have smaller stakes than blockholders which makes it more unattractive for them to monitor, given the fixed costs of monitoring. In contrast, blockholders enjoy a larger fraction of the benefits that will outweigh the agency costs. Thus, larger economic stakes supply ample incentives to monitor the actions of managers and to get involved with managerial decision-making with regards to the audit and financial reporting processes. For instance, they are able to induce management to reduce purchasing NAS of their incumbent auditors if they perceive that the provision of significant fees potentially impairs the objectivity of auditors (Mitra & Hossain, 2007). Moreover, managers of firms with substantial institutional ownership may be provoked to voluntary disclose information to acquire for the sake of gaining confidence of shareholders (Quick et al., 2013).
Mitra and Hossain (2007) demonstrated that the institutional stock ownership is negatively related to the NAS fees ratio. Investors tolerate more non-audit services if the concentration of ownership is higher (Quick et al, 2013). Ownership concentration refers to the amount of stock owned by individual investors and blockholders. This can be measured by free float, which is the percentage of shares that investors hold with less than five per cent of total equity. More authors have included the composition of ownerships in their research. Zerni (2012) used the separation of voting rights from cash-flow rights as a proxy for the degree of entrenchment of the controlling shareholder and found that NAS are more likely to be purchased from a different auditor, rather from their incumbent audit firm. Thus, I hypothesize that:
H1. Free float is negatively associated with non-audit services
2.5.2 Management compensation
Auditors have intimate knowledge of their client’s business process and best industry practices therefore, auditors can add significant value for their audit clients through NAS. NAS are assumed to increase firm performance and value as long as the benefits, in the form of cost reductions resulting from the joint provision, exceeds the NAS fees paid to auditors (Chen, Du, Krishnan & Su, 2009). The decision on purchasing NAS from their external auditor has often been explained in terms of agency-related variables such as management compensation (Beattie & Fearnley, 2002). Compensation for performances such as bonuses is
18 given to management to incentivize them to make decisions in the best interest of the firm, and to align the conflicting interests between management and investors within an organization. This leads to a reduction in the necessary active monitoring and as a result the relevance of information in financial reporting and need for an independent auditor decreases (Quick et al. 2013). Additionally, the combined fee of audit and non-audit is more likely to be lower when the auditor performs both services. This leads to a reduction of total costs, since some of the engagement phases already have been performed, and it leads to an enhanced firm profitability (Chen et al., 2009). Evidence has consistently demonstrated that firms with higher agency costs and a lower fraction of performance-based management compensation have lower NAS fee ratios (Beattie & Fearnley, 2002). Therefore, compensation that is given to managers serves as a proxy for the level of agency conflicts, and it is assumed that it would increase the demand of non-audit services. I therefore hypothesize the following:
H2. Performance-based compensation for management is positively associated with non-audit services.
2.5.3 Leverage
The expected sign of the relation between distressed firms and the ratio of non-audit fees is ambiguous. On the one hand, firms with high leverage decrease their discretionary spending to preserve cash but on the other hand they wish to improve their financial condition by spending more on consultancy services (DeFond et al., 2002, 1268). Black (1976, as cited in Quick et al., 2013) argued that management are more likely to misinterpret information when there is either a high degree of leverage or transferring wealth from creditors to management and shareholders through the payment of dividends. Furthermore, firms with high leverage tend to breach covenants to manipulate the financial statements to make them appear more favorable (DeFond & Jiambalvo, 1994 as cited in Quick et al., 2013). Thus, in the event of high liabilities, auditor independence is considered more important than usual. Capital markets find that these firms are risky to invest in, and are uncertain about the credibility of the financial statements. For this reason, they demand a risk premium. Consequently, this risk premium adds to the cost of debt, hence an external independent auditor is therefore more crucial. Earlier empirical research (Parkash & Venable, 1993; and Firth, 1997 as both cited by Chen et al., 2009) found that firms with higher agency costs caused by high leverage be
19 inclined to purchase lower amounts of NAS to decrease their agency costs. This finding is consistent with the view that firms wish to manage their independence in appearance.
Major accounting scandals around the world have led German regulators to prohibit several types of NAS in 2004. These restrictions to provide NAS by the incumbent auditor might have availed a sense of security by assuring investors that the independence of auditors has not been compromised in any way. However, new evidence supports the conclusion that highly-leveraged firms purchase more non-audit services (Alshawish 2015; Ghosh and Pawlewicz, 2009 as cited in Alshawish et al. 2015; Hallak & Silva, 2012; Patel & Prasad, 2013; Zaman et al. 2011 as cited in Kent, 2011). Several possible explanations have been claimed for the change of direction. First, Kent (2011) implies that poor performances could lead firms to obtain help from externals to improve the profitability of the company. Second, Patel and Prasad (2013) imply that highly-leveraged firms require more consulting to manage their position. Third, Zaman et al. (2011 as cited in Hallak & Silva, 2012) argued that monitoring is inevitable for companies that are highly leveraged in order to protect the firm from financial and market risks. Lastly, Alshawish et al. (2015) posit that when any firm needs more funding, it owes an endorsement from various certified sides to get the approval for lending. Therefore, the decision to purchase NAS does not entirely depend on the management. Contrary, other studies claimed that purchasing NAS from their incumbent auditor actually results in a reduction of agency costs, which is vital for firms with high debt levels (Dedman et al., 2014; Knechel et al., 2012; Svanström & Sundgren, 2012). DeFond et al. (2002) showed that market-based institutional incentives such as litigations costs outweigh the economic dependency. Therefore, I hypothesize that highly-leveraged firms purchase more non-audit services to aid the company’s situation.
20
3
Methodology
3.1 Research design
A quantitative research is used to examine whether agency conflicts explain the demand of non-audit services. This study focuses on publicly listed companies in Germany, because listed firms are the firms that are most likely to be impacted by the reforms of the European Commission. The data collection process starts by using Datastream to retrieve an overview of all firms that are listed in Germany. The initial sample of this study originated from an intersection of three databases: Datastream (Thomson Reuters), Bureau van Dijk and Capital IQ's People Intelligence. Information regarding non-audit fees, total fees and management compensation were not available for every year or for every firm and were therefore added manually by hand-collecting the missing data. Data regarding total debt and presence of an audit committee were also hand-collected as these data were also missing. The sample period is from 2005 to 2014. This time period is chosen because following the major accounting scandals which have drawn a lot of attention towards the audit profession; German regulators banned a few types of NAS in 2004. As from 2005 it was required for listed firms to disclose audit fees, where all German entities who are subjected to statutory audits had to comply with the “Accounting Law Modernization Act (Bilanzrechtsmodernisierungsgesetz (BilMoG)) (Köhler & Ratzinger-Sakel, 2012). The disclosures of fees will give more information about the services that are provided by the audit firm. Moreover, the year 2014 was chosen to avoid the results being influenced by the revision of the statutory audit directive that became effective in June 2016, where the EU restricted several non-audit services. Although this reformed directive is significantly important for the audit profession, it would impact the results and would therefore be more problematic to assess whether agency conflicts impact the demand of purchases of NAS.
3.2 Sample collection
The consolidated financial statements of listed German companies for the years 2005 to 2014 are selected. The initial sample generated by Orbis resulted in 773 listed-firms. I further collected information from the three above mentioned databases and hand-collected the missing data. From my initial sample the databases only generated data for 253 listed-firms. Therefore, the initial number of listed-firms was reduced by 520 listed firms. From the 253 listed-firms the data regarding non-audit fees were only available for approximately 34 firms and for management compensation approximately 41 firms, both depending on the year. By
21 hand-collecting the missing data this number was increased to 193 listed-firms and 27,020 firm-year observations.
3.3 Variables
Various authors have used different factors to determine the effects on the demand of non-audit services. Based upon their findings, their results are incorporated into the model that will be used to test the three hypotheses as theorized in chapter 2 paragraph 5. I examine whether the demand for non-audit services are associated with free float, leverage and performance-based compensation (variables of interest), where non-audit fees is the dependent variable and the variables of interest are the independent variables. The model is specified below:
Fee ratio = β0 + β1 Free float + β2 Leverage + β3 Performance Based compensation
+ β4 Size + β5 Growth + β6 ROA + β7 CFO + β8 Market-to-book ratio
+ β9 Stock-price volatility + β10 Audit committee + β11 Big 4 + β12 Loss
+ β13 Audit opinion + ∑ β 14-22 Industry + ε.
3.3.1 Dependent variable
An important factor in research that investigates the effect of non-audit service fees on auditor independence is the valuation of the former variable. The dependent variable is the non-audit services which is measured by the NAS fees ratio. The NAS fee ratio is defined as the ratio of non-audit fees to total fees.
Table 1
Fee ratios per year (2005-2014)
Fiscal year N Mean Median SD Min. Max.
2005 1629 .294 .258 .225 .000 .908 2006 1629 .277 .260 .201 .000 .822 2007 1629 .273 .230 .227 .000 .877 2008 1629 .255 .219 .199 .000 .801 2009 1629 .261 .221 .206 .000 .909 2010 1629 .267 .233 .200 .000 .840 2011 1629 .243 .226 .185 .000 .836 2012 1629 .240 .205 .189 .000 .778 2013 1629 .233 .190 .198 .000 .876 2014 1629 .239 .200 .184 .000 .801
23
Table 2
Fee ratios per industry (2005-2014)
Fiscal year Construction
Finance, Insurance and
Real Estate Manufacturing Mining
Retail trade Services Transportation and Logistics Wholesale trade Total 2005 .152 .335 .288 .371 .257 .325 .299 .111 .294 2006 .023 .323 .300 .360 .211 .275 .232 .204 .277 2007 .420 .276 .255 .591 .174 .286 .309 .265 .273 2008 .139 .253 .269 .000 .318 .228 .268 .220 .255 2009 .377 .325 .273 .049 .152 .225 .284 .263 .261 2010 .135 .341 .264 .512 .218 .264 .249 .370 .267 2011 .298 .111 .248 .470 .165 .250 .254 .310 .243 2012 .030 .202 .234 .084 .241 .250 .271 .290 .240 2013 .691 .240 .236 .227 .116 .232 .218 .205 .233 2014 .227 .146 .236 .486 .284 .223 .278 .305 .239 Total .249 .255 .260 .294 .213 .256 .266 .254 .258 Note: N=1629
24 Table 1 presents the descriptive statistics and illustrates the differences between the NAS fee ratios per year, which differ significantly. The highest portion of NAS entails 90,9% of total audit fees whereas the smallest portion is 0% which means that some firms have not purchased any non-audit services. In later years the fees ratio have decreased, implying that clients have purchased less non-audit fees after 2010. A plausible reason could be the consequences of the reformed changes in the audit profession in Germany that occur in 2009, after the regulators already prohibited some non-audit services in 2004. This could suggest that investors might have found that providing non-audit services compromises the independence of the auditor. As a result, audit clients try to maintain their perceived independence by purchasing less non-audit services from their incumbent auditor.
As exhibited in Table 2, the use of NAS varies somewhat according to the industry where the firm is operating in as well as within which years. It is also evident that the NAS fee ratio has decreased from 2010 onwards. This occurrence is mainly due to the NAS fees ratio in the industry Finance, Insurance and Real Estate as well as Mining and Retail trade where there is a significant drop in the purchase of non-audit services from the audit firm. Overall, the fee ratio per industry does not differ greatly.
3.3.2 Independent variables
The independent variable is the agency costs, which is defined by three variables of interests. With the variables of interest, I test the hypotheses as stated in chapter 2 paragraph 5, and the control variables are included to increase the quality of the multiple regressions.
3.3.2.1 Variables of interest
The following variables of interest are presented in Table 3 and represent agency costs. It includes the ownership composition (free float percentage), management compensation (relation between bonuses and total compensation given to the management board) and leverage (total debt divided by total assets).
25 3.3.2.2 Control variables
Several control variables are included in the regression model. These variables might potentially influence the demand of non-audit services as documented by previous literature, which is shown in Table 4. I posit that larger firms have a higher need of consultancy services, thus I presume that there is a positive correlation between firm Size and NAS. Prior literature indicated that the size of the client accounts for 70 percent in variation in audit fees (Patel & Prasad, 2013). The logarithm of total assets is used as a proxy for company size (Alshawish et al., 2015; Campa & Donnelly, 2015; Quick et al., 2013; Zerni, 2012). Moreover, Campa and Donnelly (2015), Quick et al. (2013) and Zerni (2012) controlled for Growth and ROA. I expect that Growth has a positive correlation with NAS and I measure this as the change in percentages in gross sales in comparison with the preceding year. The growth of a company might either imply that the firm is expanding or rapid changes in its economic environment. Thus, drastic changes within the company are likely to necessitate consultancy services for possible restructuring of the organization. Additionally, Dedman et al. (2014) also controlled for ROA. ROA and Loss are both used as proxies for firm performance, because bad firm performance indicates that distressed firms are more likely to
Table 3
Agency costs: variable of interests.
Variable Definition Expectation Prior research
Free float The proportion of shares available to trade
– Alshawish et al. (2015); Mitra & Hossain (2007); Quick et al. (2013); Zerni (2012) Management compensation Performance-based compensation divided by total compensation
+ Beattie & Fearnley (2002); Chen et al. (2009); Quick et al. (2013)
Leverage Total debt divided by total assets
+ Alshawish et al. (2015); Amir et al. (2010); Campa & Donnelly (2015); Dedman et al. (2014); Knechel et al. (2012); Mitra & Hossain (2007); Quick et al. (2013)
Notes: “+”, Positive relationship between NAS fees ratio and independent variable predicted; “–”, negative relationship between NAS fees ratio and independent variable predicted.
26 obtain help from externals to improve the company’s profitability (Kent, 2011). DeFond et al. (2002) also argues that distressed firms possibly demand more audit work which results in higher audit fees. I further integrate the variables Market-to-book ratio and Stock price volatility into the model to serve as proxies for expectations of probability by investors.
Several different dummy variables are included in the model such as: the existence of an audit committee, BIG 4/Non-BIG 4 , reported loss, type of audit opinion and type of industry. BIG-4 firms are in general more capable of having the resources to provide higher standards of NAS. A BIG-4 auditor has access to more resources and is able to attract more capable personnel (Patel & Prasad, 2013). The audit committee plays a noteworthy part in the audit process because of their responsibilities to appoint, terminate and determine the compensation of the auditor. Given that the economic bonding between auditors and their clients may jeopardize the independence in appearance, it may also impact the will to purchase NAS, especially for those in charge of corporate governance (Zerni, 2012). For this reason, some firms have established an audit committee even though this is not mandatory in Germany. Therefore, an audit committee represents good corporate governance that could reduce information asymmetries because they handle the required independence of the auditor. In addition, a qualified audit opinion signals that the financial statements are not free from material misstatements (Woods et al., 2007). Audit reports are only deemed as beneficial if the reports contain reliable information. Moreover, when auditors give a qualified audit opinion, it leads to a decrease in the demand of non-audit services because the willingness to engage the auditor as a consultant is also expected to be reduced (Quick et al., 2013).
27
Table 4 Control variables
Variable Definition Expectation Proxy for Prior research
Control variables
Size Logarithm of total assets in million euros
+ Company size
Amir et al. (2010); Alshawish et al. (2015); Campa & Donnelly (2015); Quick et al. (2013); Zerni (2012)
Growth Growth in gross sales compared to prior year
+ Company growth
Campa & Donnelly (2015); Quick et al. (2013); Zerni (2012)
ROA EBIT / total assets +/– Economic situation
Amir et al. (2010); Campa & Donnelly (2015); Dedman et al. (2014); Quick et al. (2013); Zerni (2012)
CFO Cash flow from operations/total assets
+ Performance indicator
Quick et al. (2013)
Market-to-book-ratio Market capitalization / book value of equity + Expectation of probability by investors
Amir et al. (2010); Quick et al. (2013)
Stock price volatility
Volatility over the 250 days before the balance sheet date – Management vs. investors Quick et al. (2013) Dummy variables Audit committee 1 if an audit committee exists and 0 otherwise
– Corporate Governance
Dedman et al. (2014); Knechel et al. (2012); Quick et al. (2013)
BIG 4 1 if firm is a BIG 4 and 0 otherwise
+ Audit quality Campa & Donnelly (2015); Dedman et al. (2014); Knechel et al. (2012); Quick et al. (2013); Zerni (2012) Loss 1 if firm reports
negative income and 0 otherwise
– Performance indicator
Amir et al. (2010); Campa & Donnelly (2015); Mitra & Hossain (2007); Quick et al. (2013); Knechel et al. (2012)
Audit opinion
1 if firm received a qualified audit opinion and 0 otherwise
– Audit quality Beattie & Fearnley (2002); Quick et al. (2013)
Industry 1 from a particular industry and 0 otherwise
Notes: “+”, Positive relationship between NAS fees ratio and independent variable predicted; “–”, negative relationship between NAS fees ratio and independent variable predicted.
28
4
Results
4.1 Descriptive statistics
Before the regressions are performed, two outliers were removed from the data sample since extreme observations can have a high influence on the descriptive statistics and regression coefficients. Table 5 presents the descriptive statistics of all the variables. From the variables of interest, Free float (mean=0.617; median=0.640), Leverage (mean=0.510; median=0.534) and Performance-based compensation (mean=0.371; median=0.375) are slightly lower in comparison with prior literature. The descriptive statistics show that on average 25,8% from the total audit fees are related to non-audit services. The proportion of shares available to trade which is measured by free float is around 61,7% within firms. Moreover, the average leverage ratio is 51% so there are roughly half as many liabilities as there is equity. Plus, the compensation given to management consists out of 37,1% performance-based compensation.
Table 5
Descriptive statistics of the variables of interest (N=1629).
Variable Mean Median SD Variance Q25 Q75
NAS .258 .224 .202 .041 .089 .390 Free float .617 .640 .287 .082 .390 .890 Leverage .510 .534 .200 .040 .380 .663 Performance-based compensation .371 .375 .204 .042 .232 .519 Size 13.072 12.654 2.364 5.588 11.306 14.478 Growth .086 .054 .377 .142 -.022 .145 ROA 4.285 5.150 13.133 172.487 2.230 8.490 CFO .063 .074 .227 .051 0.034 .114 MTBV 2.029 1.520 2.661 7.082 1.000 2.380
Stock price volatility 31.576 29.730 11.659 135.928 23.220 38.865
Audit committee .530 1.000 .499 .249 .000 1.000
BIG 4 .685 1.000 .465 .216 .000 1.000
Audit opinion .003 .000 .056 .003 .000 .000
Loss .020 .000 .141 .020 .000 .000
4.2 Correlation matrix
For the correlation matrix, both the Spearman-Rho and Pearson correlation between the dependent variable NAS and the independent and control variables have been conducted, and
29 the results are shown in Table 6 (see Appendix 1, p. 42). The Pearson’s correlation shows that the dependent variable NAS is significantly correlated with the variables of interest Free float (p=0.086; r=-0.039) at a significance level of 10% and Performance-based compensation (p=0.002; r=-0.072) at a 1% level as expected. However, Leverage (p=0.430; r=-0.018) is negatively correlated with NAS fee ratio. This relation is in contrast with the third hypothesis which states the assumption that audit clients purchase more non-audit fees from their auditor when firms are highly-leveraged. The control variables Size (p=0.061; r=0.007), Audit Committee (p=0.000; r=-0.121) and BIG-4 (p=0.000; r=0.142) are significantly correlated at a significance level of 1% as well. In addition MTBV (p=0.091; r=0.039) is correlated with NAS at a 10% level. These relations were expected and confirm prior research. Apart from leverage and stock price volatility, all variables have the expected sign.
Moreover, Free float is correlated with the following variables at a 1% significance level: Performance-based compensation (p=0.000; r=0.097), Size (p=0.000; r=0.097), Audit committee (p=0.000; r=0.194), BIG 4 (p=0.000; r=0.085) and Loss (p=0.004; r=-0.065). In addition, Stock price volatility (p=0.012; r=0.057) is correlated at a 5% significance level and ROA (p=0.092; r=-0.038) at a 10% level. Quick et al. (2013) argued that performance-based compensation leads to a reduction in the necessary active monitoring of audit committees, however, the results of the Pearson’s correlation matrix show that there is a positive relation between performance-based compensation and audit committee with a p-value of 0.000 with a related beta of 0.179, which is significant at a 1% significance level. This signifies that the need for an audit committee increases when there is a higher proportion of performance-based compensation. What is also evident from the table is that the variable Performance-based compensation is significantly correlated at a 1% level with most variables, whereas leverage is only correlated with performance-based compensation and audit committee at a 5% significance level. These correlations and its associated P-values provide more or less a descriptive explanation about the coherence between the variables. Additionally, I performed the Spearman-Rho correlation, which is presented in Table 7 (see Appendix 2, p. 43). What is notable, is that under the Pearson correlation – which measures only linear relationships – is that under the Spearman-Rho method the correlations between some variables are less significantly correlated or could not be statistically proven.
30
4.3 Regression analysis
The regression is first performed without the control variables, the results are shown in Table 8. Contrary to the third hypothesis (H3), leverage does not show the expected sign and is negatively correlated with the demand of NAS. The other variables of interest show the expected sign and are significantly correlated with NAS fee ratio. Performance-based compensation (p=0.001; r=0.075) is significantly associated at a 1% significance level and free float is (p=0.043; r=-0.033) is correlated with NAS at a 5% level. However, when the control variables are added, free float is no longer significant with NAS fee ratio and performance-based compensation is only correlated at a 5% significance level.
Table 8
Regression analysis between NAS fee ratio and the variables of interest
Expectation Coefficient Std. Error t Sig.
(Constant) β0 .258 *** .018 14.497 .000
Free float β1 – -.033 ** .016 -2.025 .043
Leverage β2 + -.014 .023 -.591 .555
Performance-based compensation β3 + .075 *** .023 3.300 .001
Notes: Correlation is significant at: *10, **5 and ***1 per cent levels (two–tailed). N=1629.
4.3.1 Main regression model
Table 9 reports the results of the correlation between the dependent variable non-audit fees ratio and the variables of interest and control variables of the first model (Model 1). This model includes all three variables of interest: free float, leverage and performance-based compensation and all control variables. Despite the Pearson correlation which shows that free float is significantly correlated at a 10% level with the NAS fee ratio, the regression concludes that only the proportion of performance-based compensation affects the demand for NAS. As mentioned before, when the control variables are included, free float (p=0.118; r=-0.026) is no longer significant correlated. The previous regression showed a significant correlated at a 5% level. As can been seen in the table, performance-based compensation is now significantly positive at a 5% level (p=0.019) with a related beta of 0.057. The effects of the control variables reduced the significance level of performance-based compensation from 1% to a 5% significance level. This means that the demand for NAS increases with 5,7% when the proportion of performance-based compensation in relation with total compensation
31 is high. Therefore the third hypothesis is confirmed. The influence of performance-based compensation is significant, however, as shown in Table 9 it is not the only factor that significantly influences the demand of NAS. The following variables: Size (p=0.009; r=0.007), Audit Committee (p=0.000; r=-0.087) and BIG-4 (p=0.000; r=0.075) with a significance level of 1% also play a noteworthy role in the demand of NAS. What is remarkable is the high standardized beta of the control variables Audit committee and BIG-4. This suggests that when an audit committee exists, the demand of NAS decreases with 8,7% whereas if the client is audited by a BIG-4, the demand of NAS increases with 7,5%. Of the variables of interest, only free float (p=0.118; r=-0.026) and performance-based compensation (p=0.019; r=0.057) are of the expected signs and are therefore in line with our hypotheses, however, only the variable performance–based compensation is significant at a 5% significance level. In contradiction with the hypothesis, leverage is not positively correlated with NAS, implying that clients actually purchase less non-audit fees from their auditor when there are more liabilities than equity present in the company. With the exception of leverage and stock price volatility, all variables are consistent with the expected sign. In summary, the relative values of the non-audit fees imply that firms purchase more non-audit fees when the proportion of performance–based compensation is higher. Whereas, free float and leverage decrease the demand of non-audit services, but the relationships of the latter were insignificant. Moreover, the result on leverage is inconsistent with the assumption that agency conflicts affect the demand of non-audit services.
4.3.2 Sensitivity analysis
An additional sensitivity analysis is performed where each of the three variables of interest are tested separately to see whether any differences between the variables would occur. The results are presented in Table 10 and can be compared to the results of Table 9. When we test the variables separately, Model 2A and Model 2B display a slightly stronger significant direct relationship regarding Size (p=0.002 and p=0.004) as opposed to Model 1 (p=0.009), where both variables remain significant at a 1% level. However, as exhibited by Model 2C, the size of the company plays a less significant role with performance-based compensation with a p-value of 0.017 with a related beta of 0.006, which is significant at a 5% level, instead of the previous 1% significance level. In addition, performance-based compensation (p=0.022; r=0.056) shows a slightly weaker relationship when tested separately without free float and leverage. Moreover, the p-values of free float and leverage both decreased, but not to the
32 point where it can be statistically confirmed that the relation between these variables and NAS is significant.
Taken together, the insignificant findings of the multiple regressions can be explained by the insignificant relation of free float, leverage and NAS. However, when the three variables of interest are all included in the model there is a higher adjusted R2, as opposed to testing the variables separately. Under the main regression model the adjusted R2 is 6,1% whereas under Model 2A, 2B and 2C are 5.8%, 5.7% and 5.9% respectively. This means that when the three variables are combined the model only explains 6,1%. Overall, the sensitivity analysis has not resulted in any big fluctuations in comparison with the results of Table 9.
Table 9
Main regression model: Model 1
Variable Expected Coefficient Std. Error Model 1 t Sig.
(Constant) β0 .150 ** .041 3.704 .000 Free float β1 – -.026 .016 -1.566 .118 Leverage β2 + -.020 .022 -.899 .369 Performance–based compensation β3 + .057 ** .024 2.346 .019 Size β4 + .007 *** .003 2.614 .009 Growth β5 + .005 .013 .360 .719 ROA β6 +/– .000 .000 -1.113 .266 CFO β7 + .008 .025 .304 .761 MTBV β8 + .001 .002 .676 .499
Stock price volatility β9 – .001 .000 1.356 .175
Audit Committee β10 – -.087 *** .010 -8.381 .000
BIG 4 β11 + .075 *** .011 6.672 .000
Audit opinion β12 – -.072 .081 -.891 .373
Loss β13 – -.014 .035 -.399 .690
∑Industry β14–22 -4.640 .002 -0.019 .985
33
Table 10
Sensitivity analysis: Model 2A–2C
Model 2A Model 2B Model 2C
Ownership composition Leverage Performance-based compensation
Variable Coefficient Std. Error t Sig. Coefficient Std. Error t Sig. Coefficient Std. Error t Sig.
(Constant) .144 *** .039 3.712 .000 .153 *** .040 3.772 .000 .136 *** .039 3.501 .000 Free float -.024 .016 -1.455 .146 Leverage -.024 .022 -1.051 .293 Compensation .056 ** .024 2.296 .022 Size .008 *** .003 3.088 .002 .007 *** .003 2.872 .004 .006 ** .003 2.397 .017 Growth .006 .013 .459 .646 .006 .013 .475 .635 .004 .013 .342 .732 ROA .000 .000 -.679 .497 .000 .000 -.563 .574 .000 .000 -1.000 .317 CFO .005 .025 .206 .836 .005 .025 .201 .841 .007 .025 .289 .773 MTBV .001 .002 .874 .382 .001 .002 .879 .380 .001 .002 .668 .504
Stock price volatility .001 .000 1.179 .239 .000 .000 1.007 .314 .000 .000 1.147 .251
Audit Committee -.086 *** .010 -8.251 .000 -.088 *** .010 -8.464 .000 -.089 *** .010 -8.531 .000
BIG 4 .077 *** .011 6.855 .000 .077 *** .011 6.894 .000 .075 *** .011 6.717 .000
Audit opinion -.072 .081 -.894 .371 -.071 .081 -.878 .380 -.075 .081 -.931 .352
Loss -.010 .035 -.107 .786 -.006 .037 -.175 .861 -.009 .035 -.258 .796
∑Industry -.000 .002 -.046 .000 -2.746 .002 .011 .991 -1.794 .002 -.007 .994
Notes: Correlation is significant at: *10, **5 and ***1 per cent levels (two–tailed).
