• No results found

The impact of the audit partner’s gender on audit fees

N/A
N/A
Protected

Academic year: 2021

Share "The impact of the audit partner’s gender on audit fees"

Copied!
40
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

The impact of the audit partner’s gender on

audit fees

Final version

Paul Koomen (10003455) 23th of June 2014

First supervisor: R.W.J. van Loon Second supervisor:

MSc Accountancy & Control, variant Accountancy, 2013/2014 Amsterdam Business School

(2)

Abstract

This paper investigates the relationship between the gender of the audit partner and the audit fees that are paid by the client firm (auditee) in the United Kingdom. The gender of the audit partners is obtained from the annual reports of the companies, since the Companies Act requires audit partners to disclose their names in the audit report from April 2009 and later. Prior research document behavioural differences between males and females, and these differences could possibly influence the price setting of an audit engagement. This study focuses on the gender differences in audit planning, risk assessment, leadership and other differences. Prior studies in three Nordic countries and Belgium found that female audit partners charge higher audit fees to their auditees than male audit partners. However, when using a sample of companies listed on the London Stock Exchange in 2012 this study finds that in the United Kingdom female audit partners charge lower audit fees than male audit partners do, though the relation is not significant enough to draw conclusions upon. This study contributes to the auditing literature, because it focuses on the effect of audit partner characteristics, gender in this study on audit fees. Prior research on audit fees has focused on the audit firm or client firm characteristics, but the effect of audit partner characteristics on audit fees isn’t examined extensively so far. However, this study is subject to the limitation that the audit fee data used consists of fees for the statutory audit and fees for consultancy together, which could have influenced the results.

(3)

Table of contents

1. Introduction ... 4

1.1. Background ... 4

1.2. Research question ... 6

1.3. Motivation ... 6

1.4. Structure of the paper ... 7

2. Literature review ... 8

2.1. Determinants of audit fees ... 8

2.1.1. Auditee attributes ... 8

2.1.2. Auditor attributes ... 11

2.1.3. Engagement and other attributes ... 12

2.2. Engagement partners ... 13

2.3. Possible gender differences affecting audit fees ... 17

2.3.1. Engagement planning ... 17

2.3.2. Risk assessment ... 19

2.3.3. Conducting the audit ... 21

2.3.4. Other gender differences ... 22

3. Hypotheses ... 24

4. Sample selection and empirical models ... 25

5. Results ... 28 5.1. Descriptive statistics ... 28 5.2. Regression analysis... 32 5.2. Sensitivity analysis ... 34 6. Conclusion ... 36 7. References ... 38 3

(4)

1. Introduction

In this section, the topic of this study is introduced briefly. First, some background related to the research is provided. Second, the research question of this paper is explained. Third, the motivation for this study is presented and finally the structure of the paper is given.

1.1. Background

According to prior research there are several factors that determine audit fees (Simunic, 1980; Chan et al., 1993; Pong & Whittington, 1994; Hay et al., 2006; Palmrose, 1986). Researchers have found that there is a big eight auditor premium (Palmrose, 1986) for both small and big auditees (Chan et al, 1993). Pong & Whittington (1994) additionally found that auditee size and auditee complexity also drive the audit fees that are charged by auditors.

Additionally, Hay et al. (2006) examined prior research on determinants of audit fees from the last 25 years. Their main objective was to evaluate the different independent drivers of audit fees across different studies and countries. Hay et al. (2006) find several determinants of audit fees that are often used in the papers that they examined. Hay et al. (2006) summarize determinants of audit fees in their results section that are mostly used in prior research: Client attributes like size, complexity, risk, leverage and profitability. Auditor attributes like quality, tenure and location. And also engagement attributes like non-audit services and seasonal problems (Hay et al., 2006). Hay et al. (2006) often refer to Simunic (1980) and state that much prior research on audit fees followed the work of Simunic (1980). Simunic (1980) doesn’t state determinants of audit fees specifically, but does state some relevant information on potential benefits and costs when hiring an auditor.

However, the Public Company Accounting Oversight Board recently, on October the 11th 2011, proposed to also disclose the name of the engagement partner in the audit report in the United States (U.S.). The proposal is that the engagement partner not only signs the audit report in addition to the sign of the audit company itself, but also discloses his or her name in the audit report (King et al., 2012; PCAOB, 2011). King et al. (2012) come up with several consequences related to the disclosure of the engagement partner’s name of the audit report, both for audit quality in appearance and audit quality in fact. Audit quality in appearance is the public perception of audit quality and audit quality in fact is the actual audit quality (King et al., 2012). The authors state that the public is likely to accept this mandatory disclosure of the engagement partners name and therefor the audit quality in appearance will increase.

(5)

However, due to greater accountability of the engagement partner’s signature, the costs of the audit are likely to increase. It will lead to more conservative risk-related judgments, more audit procedures and larger sample sizes (King et al., 2012). This might have an effect on audit fees too. The authors also doubt whether the audit quality in fact will also increase when the name of the engagement partner has to be mandatory disclosed. The effect of disclosing the identity of the partner (signature) on audit fees is examined by Carcello & Li (2013). The authors find a significant increase in audit fees after the requirement of disclosing the

auditor’s signature was mandatory.

In addition to King et al. (2012) and Carcello & Li (2013), Ittonen & Peni (2012) examined the effect of the disclosure of the engagement partner’s name on the audit fee. Specifically, Ittonen & Peni (2012) assessed whether the gender of the engagement partner impacts the audit fee. In three Nordic countries (Finland, Denmark & Sweden) that Ittonen & Peni (2012) examined the disclosure of the engagement partners name is already mandatory. Additionally, in the United Kingdom disclosure of the partner’s signature is mandatory and very often the name of the audit partner is also disclosed (Carcello & Li, 2013). Ittonen & Peni (2012) found that when the engagement partner is a female, the audit fee is significantly higher than when the engagement partner is a male. However, Ittonen & Peni (2012)

conducted their research in the three Nordic countries and this research will focus on the United Kingdom.

The differences in audit fees between male and female engagement partners can be explained by several differences between males and females like planning, risk aversion or assessment and team performance/leadership. Ittonen & Peni (2012) namely state that the audit process is conducted in 4 phases: (1) planning, (2) risk assessment, (3) conducting the audit and (4) evaluate the result. This paper focuses on the differences between males and females in the first three phases, because prior research shows differences between genders in these specific concepts. Planning is a concept that girls do better than boys (Naglieri & Rojahn, 2001; Warrick & Naglieri, 1993; Bardos et al., 1992). In addition, the planning of an audit may have an effect on the audit fee (Davidson & Gist, 1996). Prior research also shows that males and females have different perceptions of risk (Gustafson, 1999; Byrnes et al., 1999) and different assessments of risk influences the audit fee (Houston et al., 1999). And an audit is conducted in a team, where the engagement partner is the leader of the team (Ittonen & Peni, 2012). Burke & Collins (2001) show that female and male accountants have different leadership styles. Additionally, Eagly & Carli (2003) show that females have advantages and disadvantages in leadership. Furthermore, there might be differences in the overconfidence of

(6)

auditors (Hardies et al., 2009) and differences in the ethical values between male and female auditors (Ameen et al., 1996; Neidermeyer, 2003).

1.2. Research question

In the previous section, some determinants of audit fees were briefly introduced and also some differences between males and females that might affect this audit fee were introduced. This study is especially interested in the differences in audit fee between male and female engagement partners that Ittonen & Peni (2012) and Hardies et al. (2009) found in their studies. Therefore this study tries to investigate whether there is a difference in audit fees between male and female engagement partners, but then in a different country than Ittonen & Peni (2012). It is mandatory to disclose the signature of the engagement partner in the United Kingdom in the audit report and Carcello & Li (2013) state that in most annual reports that they studied in the United Kingdom the name of the partner was disclosed. Therefore, this study tries to answer the following research question:

Research question: How does the gender of the engagement partner impact the height of

the audit fee?

1.3. Motivation

As mentioned in the paper of Ittonen & Peni (2012) prior studies about audit fees focused on the effects of client characteristics, audit firm characteristics, and the engagement attributes on audit fees. The authors mention that their study goes further by examining the effect of the individual engagement partner on the audit fee. The authors also mention that more research is needed in respect with the relation between the gender of the engagement partners and audit fees.

However, they did only investigate the effects in three Nordic countries. They did find significant differences between female and male engagement partners within the audit fees. Therefore, this study wants to examine whether their results also hold in a different

setting/country, the United Kingdom. The United Kingdom has been chosen because

disclosure of the partner’s signature is mandatory for financial years ending in April 2009 or later. In a lot of annual reports, the name of the partner is also disclosed, next to the signature (Carcello & Li, 2013).

(7)

This paper will then contribute to the literature, because it basically tests the results of Ittonen & Peni (2011) and it adds to the literature because this hasn’t been investigated

extensively so far. Additionally, Ittonen & Peni (2012) call for more research within this topic what makes this study contribute to existing literature. Besides that, Carcello & Li (2013) found that audit fees increased significantly after the engagement partner signature requirement in the United Kingdom. This research therefore also builds on the study of Carcello & Li (2013) by examining the effect of engagement partner’s gender on audit fees in the United Kingdom. Additionally, I will examine the effect of risky firms on the effect between audit fees and engagement partner’s gender.

The study is also interesting from a societal point of view. If a difference in audit fees is found between male and female engagement partners, it will be of importance for audit companies. It might affect the part of females that will be allowed to reach to top in audit firms, also known as partner.

1.4. Structure of the paper

The following parts of the paper consist of five more chapters. Related prior research and literature is reviewed in chapter 2. The hypotheses are developed in chapter 3. In chapter 4 the sample selection procedure is explained and the empirical models for testing the hypotheses in this paper are presented. Further, in chapter 5 the descriptive statistics, regression analysis and sensitivity analysis are described. Finally, chapter 6 shows the conclusions and limitations of this paper and some suggestions for future research.

(8)

2. Literature review

In this section of the study information is provided about the main theoretical constructs that will be used when conducting the research, based on prior literature. It will not only be based on auditing literature, but on finance, psychology and management literature as well,

especially section 2.3. This is caused by the fact that there is hardly any audit literature on differences between engagement partner’s gender and audit fees (Ittonen & Peni, 2012). The first paragraph will describe determinants of audit fees that have been used in prior research on audit fees. The second paragraph will provide information on the responsibilities of the engagement partners from a legal perspective. The third paragraph will assess differences between males and females that might have an effect on audit fees. The second and third paragraph are included because following Ittonen & Peni (2012) some aspects of audit partners, gender in this case, may influence the audit hours spent and the risk component of the audit fees.

2.1. Determinants of audit fees

There is much prior literature available regarding the determinants of audit fees (see for example: Simunic, 1980; Chan et al., 1993; Pong & Whittington, 1994; Palmrose, 1986). Audit fees reflect the economic costs of an audit that can vary with several factors (Simunic, 1980). As Hay et al. (2006) conclude most of the prior literature was about developing models by regressing audit fees against a wide variety of measures that the researchers hypothesized to be related with audit fees. Very often control variables were included that were significant in prior studies and also their experimental variables were included (Hay et al., 2006). Hay et al. (2006) summarized their results in four categories: (1) client attributes, (2) auditor

attributes, (3) engagement attributes and (4) other attributes. This research is building further on the research of Ittonen & Peni (2012) and the attributes discussed below are used in their model. The attributes that will be of value for this research are discussed briefly below.

2.1.1. Auditee attributes

Auditee size

According to Pong & Whittington (1994) the variable auditee size can be used as a proxy for the required quantity of audit services, because large auditees require more work of the

(9)

auditor. However, an audit is a statutory requirement, so the demand of an audit is inelastic to audit fees and the audit fees are therefore mostly dependent on the required amount of work, which is measured as auditee size (Pong & Whittington, 1994).

While most of the prior research used total assets as a proxy for auditee size Chan et al. (1993) used a turnover measure to capture the auditee size. The variable total assets is often transformed to a logarithm to improve the linear relationship with audit fees (Hay et al., 2006) An audit has two aspects, which are the verifications of assets and audits of transactions (Pong & Whittington, 1993). According to Pong & Whittington (1994), the first aspect can be proxied by total assets and the second can be proxied by using the turnover measure.

However, following Hay et al. (2006) the mostly used measure in prior research is the logarithm of total assets and it is expected to have a positive relationship with audit fees (Simunic, 1980; Pong & Whittington, 1994). A sales measure as a proxy for auditee size is often used too, but in my model I use the logarithm of total assets to measure auditee size.

Auditee complexity

The second attribute that is often used in models that predict audit fees is auditee complexity (Hay et al., 2006; Pong & Whittington, 1994; Thornton & Moore, 1993). The more complex a company is the harder and more time-consuming it will be to audit that specific company (Simunic, 1980). Together with the total assets variable, complexity might also relate to the amount of work required for the audit (Pong & Whittington, 1994). Different proxies for auditee complexity are used in prior literature, where the number of subsidiaries is the one most used (Hay et al., 2006). This measure is used because the more subsidiaries an auditee has, the more work is required for eliminating intra-group transactions and consolidation in conducting the audit (Pong & Whittington, 1994). The relationship between auditee

complexity and audit fees is most of the times positive and strongly significant (Hay et al., 2006).

One other factor that is related with auditee complexity is the industry of the auditee (Hay et al., 2006). Some industries are namely more difficult to audit than others (Simunic, 1980). Industry variables (SIC-codes) are often used in prior research to identify or to exclude certain industries from the model (Palmrose, 1986; Ittonen & Peni, 2012). Hay et al. (2006) state that financial institutions and utilities are often excluded from the model, because they are supposed to be audited easier. That is because these companies have relatively large assets and no extensive receivables or inventory like manufacturing companies.

(10)

Therefore I exclude the financial institutions and banks (SIC-Codes 6000-6999) from my analysis, because of the points mentioned above. I wasn’t able to get the number of subsidiaries data and therefore I use foreign assets divided by total assets (percentage) as a proxy for auditee complexity as already used in prior literature (Simunic, 1980; Ittonen & Peni, 2012). Especially Simunic (1980) mentions the foreign assets to total assets variable as a measure of diversification of the auditee related to complexity of the auditee. Therefore I suggest that foreign assets to total assets is a solid control variable in the model, because it reflects the control problem related to large decentralized companies (Simunic, 1980). Decentralized companies increase the number of decision centres and these decisions and their outcomes have to be monitored by auditors (Simunic, 1980).

Auditee risk

There are several risk attributes related to the auditee that prior research has used. Those are inherent risk, profitability, leverage (Hay et al., 2006) and business risk (Bell et al., 2001). Although business risk is a residual risk that cannot be reduced below a certain level, it still drives the audit fee according to Bell et al (2001). Inherent risk is “the risk of an error occurring in the accounting system before the recording of transactions and the exercise of internal accounting control” (Dirsmith & Haskins, 1991, p. 72). Inherent risk is also suggested to be positively related with audit fees, because certain parts of an audit have higher risks than other parts of the audit. Some parts of the audit even need specialized audit procedures

(Simunic, 1980; Hay et al, 2006). The proxies used for measuring inherent risk in prior research are receivables divided by assets, inventory divided by assets or those two combined divided by assets (Hay et al., 2006). That is, because inventories and receivables are most frequently cited as difficult to audit (Simunic, 1980). The combined measure has the strongest positive and significant result in most of the studies.

Auditee profitability is also often seen as a risk that determines the audit fee. The auditor might be exposed to loss when the client is not financially viable (Simunic, 1980). Thus the worse the performance of the auditee the more risk for the auditor (Hay et al., 2006). There are two variables in prior research that are often used to measure profitability. Those are return on assets (ROA) and a dummy variable when there is a loss.

The third risk factor often used in prior research is leverage and it also measures the risk of a failing auditee (Simunic, 1980). The proxy to measure leverage most often used is the ratio of debt to total assets. The association between leverage and audit fees is expected to be positive and most prior research confirms this association (Hay et al., 2006).

(11)

The measures that I used to capture risk are inventory and receivables divided by total assets, a dummy variable for a loss and leverage. The first is used because it captures the inherent risk best according to Simunic (1980). The second is used because Hay et al. (2006, p. 171) conclude that the most recent results of papers they examined show that a loss variable “has become an increasingly important measure of audit fees”. The third is used because it also measures the risk of a failing auditee according to Simunic (1980).

2.1.2. Auditor attributes

Auditor quality

The quality of the auditor also has a positive association with audit fees. The most important proxies for audit quality used in prior literature are dummies for Big 4 audit firms, because they are supposed to perform higher quality audits than non-big 4 audit firms (Hay et al., 2006). Other researches did indeed find that Big 4 audit firms require a fee premium (Chan et al., 1993; Francis, 2004). Chan et al. (1993) find that both small and big firms pay a fee premium when they are audited by a big 4 auditor. Additionally, Palmrose (1986) also finds a positive relation between auditor size (Big 8) and audit fees. However, Palmrose (1986) didn’t find significant results for the industry specialist relation with audit fees. Hay et al. (2006) show that three out of nine papers that examined the relation between industry specialization and audit fees found significant result. This can be explained by the fact that there is lot of debate about how to measure industry specialization (Hay et al., 2006). Following Hay et al. (2006) it is often measured by the market share that an auditor has in a specific industry. However, Palmrose (1986) also used market shares to measure industry specialization and she was not able to find significant results.

Taken the results of prior research (Hay et al., 2006; Palmrose, 1986; Chan et al., 1993) together I only use a Big-4 variable in my empirical model to proxy for audit quality as it is often found to be positively related with audit fees.

Auditor tenure

Companies that change their auditor do this most of the time to cut off the price of the audit fee (Simon & Francis, 1988; Hay et al., 2006). Most of prior studies find that after the company changed auditor the audit fees are reduced (Simon & Francis, 1988; Hay et al., 2006). However, this is only the case in the first three years after the change. In the fourth year after the auditor change the audit fees are back on normal level (Simon & Francis, 1988).

(12)

This might also be an effect of low-balling of auditors, where the auditors set the initial audit fee below its current total costs (DeAngelo, 1981). The incumbent auditors can earn future quasi-rents for one specific client and are therefore able to low-ball for the initial audit engagement. Those future quasi-rents are earned because the incumbent auditor has a cost advantage over competitive auditors, because it has already taken the start-up and contract cost in the first year of the audit, which are sunk in future years. The incumbent auditor therefor has a cost advantage and can raise the audit fees above their current total cost (DeAngelo, 1981). DeAngelo (1981) also mentions the possible decrease in auditor independence that will be caused by low-balling. However the authors finds that the quasi-rents impair the auditor independence and the quasi quasi-rents cause low-balling (DeAngelo, 1981).

For the empirical model in this paper I use a dummy variable to capture the auditor change as suggested by prior literature (Hay et al., 2006; Simon & Francis, 1988; DeAngelo, 1981). Also the auditor tenure of the audit may affect the audit fee, but following Hay et al. (2006) a dummy variable which is indicating an auditor change is a better measure and is more often used in prior literature.

2.1.3. Engagement and other attributes

Non-audit services

The provision of non-audit services to a firm has an effect on the audit fee that the firm has to pay (Firth, 1997). Non-audit services are services like tax consultancy, management advice and financial consultancies (Firth, 1997). Firth (1997) states that the revenues from non-audit services are increasing for audit firms over the last years.

Prior research states that non-audit services are related with audit fees (Hay et al., 2006; Whisenant et al., 2003). However, according to Hay et al. (2006) there are suggestions for negative and positive effects of non-audit services on audit fees. Lower audit fees are expected because synergies or knowledge spill overs as Simunic (1980) calls them, between the non-audit services and audit services might occur. Higher audit fees are expected because non-audit services may lead to changes in the auditee and this will lead to higher audit effort (Hay et al., 2006). Additionally, Firth (1997) states that auditors can pass the cost savings to their clients by charging lower audit fees, but audit firms might also choose to not pass the cost savings to their clients.

(13)

Whisenant et al. (2003) show that non-audit fees and audit fees are determined simultaneously. This indicates that non-audit fees are associated with audit fees as prior research has shown in the analysis of Hay et al. (2006). However, Firth (1997) finds a positive relation between audit fees and non-audit services and has no plausible explanation for this positive sign. Hay et al. (2006) also show that most audit fee studies include a non-audit variable in their model. When I started this research I was also willing to include a non-audit measure in my model, but due to restrictions in the audit fee data I was not able to use the non-audit fee as a control variable.

2.2. Engagement partners

While much prior research, like in the sections above, focus on auditee, auditor and engagement attributes that affect the audit fee, little research focuses on the effect that the audit partner might have on the audit fee. Evidence by Ittonen & Peni (2012) shows that the audit partner’s gender influences the height of audit fees. They support this evidence with possible differences that can occur between male and female audit partners. Those possible differences will be discussed in the third paragraph. Additionally, Taylor (2011) shows that audit partners from the same audit firm can charge different audit fees, based on their own quality. This section describes the responsibilities and duties of audit partner both from international and United Kingdom’s legislation. Also the consequences of those

responsibilities are given attention.

Following the International Standards of Auditing, audit partners have several requirements when conducting an audit. Those are stated in ISA 220 and ISA 300. The engagement partner is responsible for the whole process of the engagement that he or she is assigned to (IAASB, 2009). Several aspects of the audit process are stated in ISA 300. The planning of the audit or the strategy for the audit is important for the audit and will enhance the audit of financial statements (IAASB, 2009, ISA 300). According to ISA 300 the audit partner is also responsible for the composition of the engagement team and is responsible for the guidance and supervision of team members so that their work is reviewed. So, according to ISA 220 the audit partner is responsible for the fact that the engagement team has the capabilities and expertise to conduct an audit (IAASB, 2009, ISA 220). When needed, the audit partner can ask for expertise of consultants who are not part of the engagement team. Additionally, Ittonen & Peni (2012) state that existing literature suggests that audit fees are affected by audit team labour hours, audit team labour costs per hour and a risk component.

(14)

This statement, in combination with the requirements for audit partners from the IAASB suggests that audit partners have a reasonable effect on the audit fee.

That is supported by a study of Taylor (2011). He mentions the assumption of prior literature that firms are willing to pay a premium for audit firms, who deliver a higher audit quality than others, i.e. big 4 audit firms. But, this assumes that external users are the only users of audits are external users, who base the audit quality on a broad measure, like big 4 audit firms (Taylor, 2011). This is supported by Hardies et al. (2009) who states that users of financial statements have no other option than to look at the brand name or firm size for audit quality. However, managers are also a key group who use audits and they have more contact with the auditors, thus they can have different perception of audit quality than external users have. Following Taylor (2011) managers have to interact with audit team personnel and they receive information from the audit partner about internal controls and misstatements. The closer contact of management with the auditor will affect their perception of an audit. As Taylor (2011, p.252) states it: “from the perspective of managers, the audit is not an output generated by an audit firm, but rather a process that is carried out by people.” Additionally, the most important person of those people is the audit partner, because he or she is the one that has the most contact with management of the auditee (Taylor, 2011). This is likely to influence the perception about the auditor quality and therefore audit quality, from

management’s view, is not only based on big 4 auditors, but on the perceived quality of the audit partner (Taylor, 2011; Hardies et al., 2009). An audit partner has to have multiple skills that can differ between individual audit partners (Taylor, 2011). Above that, management chooses and pay the auditor, thus their perception of auditor quality is likely to influence the audit fees (Taylor, 2011).

However, there are additional requirements for audit partners in the United Kingdom, the country that has been studied in this research. According to the Companies Act from 2006, section 504, the auditor’s report must state the name of the auditor and be signed and dated. Secondly, if the auditor is an individual, the report must be signed by him and thirdly, when the auditor is a firm, the report must be signed by the senior statutory auditor1 in his own name, for and behalf of the auditor. Additionally, “every copy of the auditor’s report that is published by or on behalf of the company must state the name of the auditor and the name of the person who signed it as senior statutory auditor” (Companies Act, section 504 & 505). According to Carcello & Li (2013) this act requires engagement partners to sign to audit

1 The term senior statutory auditor has the same meaning as the term engagement partner (APB, 2008). Both

terms are used interchangeably with audit partner.

14

(15)

report from financial years ending in April 2009 or later. The disclosure of the signature of audit partners in the United Kingdom is somewhat different than the proposed rule in the United States, as suggested by the PCAOB. Namely, the PCAOB in 2011 suggests requiring engagement partners to disclose their identity in the auditor’s report (King et al., 2012; Carcello & Li, 2013). The PCAOB bases this proposal on four principles as cited by King et al. (2012) and the consequences as stated by King et al. (2012) and they will be elaborated with the arguments mentioned in the paper of Carcello & Li (2013).

First, the disclosure of the audit partner’s identity will increase the transparency of the total audit process. Transparency gives investors more information about the key persons who performed the audit, including the audit partner, which is the most important person of the audit (PCAOB, 2011; Taylor, 2011). Second, the disclosure of the audit partner’s identity will lead to an increased accountability of the responsible audit partner. This is because the audit partner will feel more personal accountability for the work that he or she performed which can affect his behaviour in a positive way, following the proposal of the PCAOB (PCAOB, 2011). Additionally, DeZoort et al. (2006) state that when the identity of the audit partner is available for a greater public, this will provide the audit partner with more motivation to avoid negative consequences that can follow up after an audit failure. Third, the disclosure will lead to more responsibility of the audit partner for the overall audit quality, which can differ between audit partners (PCAOB, 2011; Taylor, 2011). Fourth, the increased transparency is an opportunity for the public to assess the experience the engagement partner. According to the PCAOB (2011) this might be a trigger for audit firms, to improve the quality of all their audit partners and create competition at the partner level and to stimulate audit committees to pay a

premium for more experiences audit partners (King et al., 2012; PCAOB, 2011). This assumes that audit partners have different audit quality and partners will earn different audit fees, which depends more on their individual quality as Taylor (2011) also described. After all, the PCAOB expects an improvement in the overall audit quality (King et al., 2012).

Additionally, Carcello & Li (2013) bring forward arguments against and for the audit partner signature in their paper, of which I will discuss the effects that are different from the effects described above. Carcello & Li (2013) agree that the increased partner accountability will increase the audit quality, because the partner will change his behaviour. This will have three main effects. First, the increased accountability may lead the audit partner, together with his audit team, to perform more work through extending the audit procedures. Second, the increased accountability of the audit partner will have the effect that they will not only perform more work by extending the audit procedures, but they will probably also gather

(16)

more, better and more substantive audit evidence (Carcello & Li, 2013). Third, the increased partner accountability may lead to more conservative audit partners and audit reporting (Carcello & Li, 2013). DeZoort et al. (2006) found that when auditors were given more accountability they were less likely to pass on proposed audit adjustments, which are key decisions for audit partners and it will affect the amounts reported in the financial statements according to Carcello & Li (2013).

However, the effects described above are considered to be positive effects of the audit partner signature requirement, but the signature requirement also has negative effects or opponents who are against the signature requirement. The effect that audit partner/team will gather more and substantive audit evidence, i.e. more work, together with the more

conservative audit reporting might have the effect that audit fees will also increase (Carcello & Li, 2013). One other negative effect might be that the signature requirement focuses too much on one specific member of the audit team, the audit partner. This might lead to a decreased responsibility and accountability of the other team members who perform audit work (Carcello & Li, 2013). Others argument that audit partners are already subject to enough accountability mechanisms like PCAOB or SEC inspections, which is specific for the United States of America (Carcello & Li, 2013).

The effects of requiring an engagement partner to sign the audit report on audit quality and audit fees was examined in the United Kingdom by Carcello & Li (2013). They state that audit quality, measured by four different proxies, increased the year that the signature of the engagement partners in the audit reports became mandatory. However, Carcello & Li (2013) also examined the effects of requiring the engagement partner to sign the audit report on audit fees. The partner signature requirement in the United Kingdom leads to a significant increase in audit fees the year it became mandated (Carcello & Li, 2013). Thus, the increase in audit quality (benefits) has the consequence that audit fees (costs) also increase.

Therefore, I suggest that when there are different perceptions between males and females about the consequences described by King et al. (2012) & Carcello & Li (2013) it could lead to differences between the audit fees that male and female engagement partners will charge to their auditees. Additionally, Taylor (2011) shows that the quality of the auditor also depends on the perception of auditees’ management about the audit partners quality. Off course, the quality of audit partners might also differ between male and female audit partners which might affect the audit fee that is paid for them.

(17)

2.3. Possible gender differences affecting audit fees

This paragraph will focus on the possible differences between genders that might affect the audit fee. According to Ittonen & Peni (2012) an audit is conducted in four phases: (1) planning, (2) risk assessment, (3) conducting the audit and (4) evaluate the result. The differences between males and females in the first three sections are examined in this paragraph with support of prior literature. In the fourth section I will describe possible

differences between males and females that are different from the three mentioned before and might also affect the charged audit fees. Ittonen & Peni (2012) state that audit partner is responsible for setting the audit fee at a sufficient level so that investments for the audit can be made and that the decisions and assessments in the first two phases of the audit are the most important. When the behaviour or skills of males and females in these phases are different it might have an effect on the audit fee that is charged for the engagement.

2.3.1. Engagement planning

The first phase of the audit engagement is planning (Ittonen & Peni, 2012) and during this phase the auditor assesses control risk, detection risk and inherent of the auditee (Davidson & Gist, 1996; Houston et al., 1999). According to Ittonen & Peni (2012, p. 4) “the risk

assessments are used for audit planning decisions concerning the nature, timing and extent of audit evidence testing”. Furthermore, following Houston et al. (1999) planning an audit is about assessing the level of business risk and investing in auditing, i.e. the amount of work has to be assessed. The planning of an audit is intended to lead to an effective and efficient audit process, but not all audit planning hours are uniformly efficient (Davidson & Gist, 1996). Davidson & Gist (1996) assume that the first audit planning hours are the most efficient in reducing total audit hours. The authors give the following reasoning for that: “An audit that has not received any planning effort is expected to be quite inefficient, whereas a small number of audit planning hours should result in a fairly large savings of effort” (Davidson & Gist, 1996, p. 113). The hours spent on audit planning thereafter become somewhat less efficient until a certain point where additional audit planning hours increase the total audit hours. It might than be justified that some auditees require more audit planning hours, because they are more risky or more complex than other auditees. Especially, auditees without an internal audit committee were associated with more audit planning (Davidson & Gist, 1996). Thus, until a certain point, more hours spent on audit planning result in less total

(18)

hours spent and the more risky or complex an auditee is, the more audit planning is needed, which will increase the total number of hours spent on the audit.

But, the skill of planning or planning processes might differ between males and females and thus differences might occur in the audit planning hours and total audit hours spent on a specific audit. That might lead to different audit fees charged by male and female auditors. According to Naglieri & Rojahn (2001, p.435) “planning processing is involved in making decisions about how to do things, selection of the best method to complete a problem, monitoring the accuracy of the solution (e.g. remember to check one’s work) and

determination of when the task is accurately completed”. Good planning processes are important for people for their academic performance, but also for their everyday activities (Bardos et al., 1992; Naglieri & Rojahn, 2001). The statements of both Naglieri & Rojahn (2001) and Bardos et al. (1992) suggests that planning processes might also influence auditors in their daily work, which can affect the total planning hours that auditors need to plan an audit.

Bardos et al. (1992) show in their experimental research that girls outperform boys on the aspect of planning. In the experiment of Bardos et al. (1992) the subjects had to organize tasks, prevent perseveration and plan to shift between certain rules and together these tasks are considered as a measure of planning. All three tasks had to do with matching numbers with numbers or matching numbers with letters or pictures (Bardos et al., 1992). The measure that was used to assess which gender performed better on planning was the total time used for the task (Bardos et al., 1992). The girls needed significant less time to perform the tasks than the boys and therefor the girls outperformed the boys on.

Additionally, in the experiment of Naglieri & Rojahn (2001) the subjects also had to complete a similar task as in the experiment of Bardos et al. (1992). However, where the sample of Bardos et al. (1992) consisted of 434 subjects, the sample of Naglieri & Rojahn (2001) consisted of 2200 subjects and therefor Naglieri & Rojahn (2001) state that their results are more generalizable to a wider public than the results of previous studies. Naglieri & Rojahn (2001) also find that girls outperform boys on the planning of the assignment. Furthermore, they found that the better planning of girls led them to perform the tasks better. Warrick & Naglieri (1993) also find that girls are better planners than boys, but their results aren’t significant in contrast to the results of Naglieri & Rojahn (2001) and Bardos et al. (1992) where the girls significantly outperformed boys on planning.

Taken the results of the papers described above together suggests that female auditors are better in planning an audit than their male counterparts and are therefore faster in

(19)

completing an audit. Though, the subjects of Naglieri & Rojahn (2001) and Bardos et al. (1992) were school children with an age of 5 to 17 years and these results might not be generalizable to auditors, who are considered to be older. But still, I suggest that female auditors are better planners than their male counterparts and thus need less audit planning hours and thus less total audit hours.

2.3.2. Risk assessment

The second phase in the audit engagement is risk assessment (Ittonen & Peni, 2012). Before conducting an audit, audit partners have to assess the audit risk model as explained by Houston et al. (1999, p. 284):

The definitions used in this model are given in Houston et al. (1999, p.284): “Acceptable audit risk is the probability that auditors are willing to accept that they will render unqualified opinions on materially misstated financial statements. Inherent risk is the probability that an account balance or class of transactions contains a material misstatement before considering the effectiveness of the internal control system. Control risk is the probability that a material misstatement is not prevented or detected on a timely basis by the internal control system. Detection risk is the tolerable level of risk that auditing procedures will not detect material misstatements.” According to Houston et al. (1999) those four risks are all assessed by the responsible auditor, which is considered to be the audit partner. Based on the level of risks the auditor makes decisions and can plan the investment in the audit or charge a specific amount for the audit fee. The higher the assessed audit risk, the bigger the audit investment or charged audit fee will be for the specific auditee (Houston et al., 1999).

Additionally, auditors also have to bear in mind the business risk and assess that risk. Business risk is an additional risk that cannot be reduced below a certain level (Bell et al., 2001). Business risk has two components that are stated by the AICPA: “(1) the client’s business risk, which is associated with the client’s survival and well-being and (2) the auditor’s business risk of being associated with a particular client irrespective of whether an audit failure is asserted“(AICPA, 1992). Although the business risk cannot be reduced below a certain level it does drive the height of the audit fee, because auditors will be likely to increase the total audit hours for risky firms (Bell et al., 2001). Bell et al. (2001) didn’t find any evidence that auditors charged a risk premium for firms with a high perceived business

(20)

risk through higher audit hours rates, but the increase in audit fee solely was caused by the increase in audit hours.

However, assessments of risk(s) might differ between male and female audit partners. Gustafson (1999) states that many prior studies have found different perceptions of risk between males and females. Where most studies only describe the different perception of risks found, Gustafson (1999) tried to explain the differences between males and females. First, the one and the same risk does not always mean the same to males and females. For example, females see nuclear power as an environmental problem while males see nuclear power more as a technical problem. Second, males worry less about risks than females because risks are often created and handled by males, thus they see the risks as more acceptable than females (Gustafson, 1999).

In addition to the different perceptions of risk there might also be differences in risk-aversion or risk propensity. Johnson & Powell (1994) didn’t find any significant differences in decision making quality or risk attitudes between males and females. Although the authors mention some limitations to their study, they point out that no individual male or female can be assumed a more risk averse decision maker (Johnson & Powell, 1994). Powell & Ansic (1997) studied gender differences in financial decision making and risk taking. The authors also find that no significant differences between males and females in financial decision making, but they do find differences in risk taking or risk seeking. That females are less risk seeking than males is something that is persistently found in both the general and business specific literature (Powell & Ansic, 1997). Johnson & Powell (1994) found a lower preference for risk amongst women, when the authors examined the betting behaviour for investment decision making amongst management students. That is supported by Francis et al. (2009), who examined the effects of gender on financial reporting decision-making related to accounting conservatism. They found that when a male CFO is replaced for a female CFO there is a significant increase in the degree of accounting conservatism. The authors therefore state that female CFO’s react more risk-averse than their male counterpart, which leads for example to less equity-based compensation and reducing firm risk. Thus females are less risk seeking or more risk averse than males when they have to make financial decisions. (Francis et al., 2009).

Additionally, Byrnes et al. (1999) conducted a meta-analyses in which they included 150 studies about risk taking or risk behaviour and gender differences. Their study therefore also included risks associated with for example smoking, driving and sexual activities, so they differ from the financial decision making studies. The findings support the statement that in

(21)

prior research it was persistently found that males were less risk averse than females as was stated by Powel & Ansic (1997). Taking the results of the decision making and risk taking studies together it can be said that male auditors are less risk averse than their female counterparts. As was mentioned above, females and males have different perceptions about risk (Gustafson, 1999), but prior literature about financial decision making confirms that females take less risks than males (Powel & Ansic, 1997; Johnson & Powell, 1994; Byrnes et al., 1999). Female auditors are therefore more risk averse than their male counterparts and the higher the assessed audit risk, the higher the audit investment or audit fee will be (Houston, 1999). Thus, female auditors are more likely to address higher risks to their auditees than male auditors will do and will therefore invest more in the audit or charge higher audit fees.

2.3.3. Conducting the audit

The third phase of the audit, following Ittonen & Peni (2012), is conducting the audit. As already stated, following the IAASB, audit partners are responsible for the composition of the engagement team (IAASB, 2009). The audit partner is also responsible for the overall

performance of the audit. As the audit partner is responsible for the overall quality of the audit, he or she can be seen as the leader of the engagement team. Differences in leadership between males and females might lead to different audit fees.

Research on leadership and the differences between males and females in leadership doesn’t find hard evidence that one of the two genders is a better leader. Burke & Collins (2001) do find that females have better leadership styles than males have and these results are significant. The findings of Burke & Collins (2001) show that females make more use of the most effective leadership styles than males do. Those are transformational and contingent reward leadership styles. The first is important for the development of relationships with team members, serving as a role model and pursuing the group members to look for the best

interest for the whole team. The second is important to reward team members for good performance (Burke & Collins, 2001). However, the authors state that although their results are favourable for females it is inappropriate to say that females are better leaders than males. That is supported by the findings of Eagly & Carli (2003). Eagly & Carli (2003) analysed the differences in leadership styles between males and females from prior research and do find advantages for both genders. Furthermore, Eagly & Johnson (1990) analysed prior literature a couple of years earlier and find some small differences between males and females in

leadership. First, female leaders are more likely to emphasize on personal relations as is 21

(22)

similar with the development of relationships with team members that is named above (Eagly & Johnson, 1990; Burke & Collins, 2001). Where male leaders are more focused on the completion of the task (Eagly & Johnson, 1999). Second, female leaders are more likely to adopt a democratic leadership style where males are more likely to adopt an autocratic leadership style (Eagly & Johnson, 1999).

Altogether, females tend to be better leaders then males, though both genders have some advantages over each other so I cannot assume that males are better leaders than

females, or vice versa (Eagly & Carli, 2003; Burke & Collins, 2003; Eagly & Johnson, 1999) Therefore, I don’t suggest that one gender is a better leader than the other and leadership will thus not have a significant impact on the audit fees. This was expected, because the

differences in the first two phases of the audit are the most important in defining the audit fee (Ittonen & Peni, 2012).

2.3.4. Other gender differences

In the previous three sections I have focussed more on the gender differences in the three phases of an audit that Ittonen & Peni (2012) mention in their paper. However, during the literature review I encountered some other possible gender differences that might influence the audit fee. Those are overconfidence, and ethical values and the possible differences will now be discussed in the same sequence.

First, as described by Hardies et al. (2012) overconfidence is “the tendency of people to believe that their judgment is more accurate than it really is. As a result, overconfidence can create a mismatch between one’s confidence in one’s own judgment and the real accuracy of these judgements” (Hardies et al., 2012, p. 105). Ittonen & Peni (2012), building on prior literature, suggest that males are more overconfident than females. This is also stated in the paper of Hardies et al. (2012). However, Hardies et al. (2012) examined whether gender differences in overconfidence occur between auditors. Their findings show that there are no differences in overconfidence between male and female auditors. Thus, the assumption from other literature that males are more overconfident than females does not hold for auditors (Hardies et al., 2012). Therefore, in contrast to Ittonen & Peni (2012) I don’t suggest that the factor of overconfidence has an influence on different audit fees between male and female audit partners.

Second, the social role theory expects that society expects females to act more ethical than males do (Eagly, 1987). Further evidence also indicates that females are believed to act

(23)

more ethical than males do and that females are more likely to reject unethical behaviour in issues related to auditing (Hardies et al., 2009). This is supported by a study of Ameen et al. (1996) where the authors studied the ethical behaviour of 285 accounting majors. Their results were that female accounting students were “more sensitive to and less tolerant of unethical behaviours, less cynical, and less likely to engage in unethical academic activities” than the male accounting students (Ameen et al., 1996, p. 596). Therefore, Ameen et al. (1996) predict that female accountants that will be hired be firms will have more ethical sensitivity than their male counterparts. This has the possible effect that female accountants are less likely to allow managers to misrepresent financial information or to depart for the local accounting principles (Ameen et al., 1996). However, these results are contradictory to the results of Radtke (2000) who explicitly mentions that his results are not the same as that of Ameen et al. (1996). Radtke (2000) finds no differences in ethical sensitivity between male and female

accountants, who participated in his experiment. Therefore the ethical decision making in business will not increase when more females are hired, according to Radtke (2000).

Additionally, the ethical values of accountants/auditors might have an influence on the issue of low-balling (see section 2.1.2.). Neidermeyer et al. (2003) examined the gender differences in attitudes towards low-balling. The authors were especially interested in the perception of the auditors about low-balling and whether it breaks with the Code of

Professional Conduct of the AICPA (Neidermeyer et al., 2003). They find that both males and females think that low-balling isn’t an acceptable business practice, but females found it less acceptable than males. However, female auditors also found that low-balling breaks with the independence as stated in the Code of Professional Conduct of the AICPA, while male

auditors didn’t think so. This can be caused by the fact that the female auditors are more often of staff-level, while male auditors are more often of management level (Neidermeyer, 2003). Although, low-balling doesn’t have an effect on the independence of auditors (DeAngelo, 1981) the results of Neidermeyer (2003) show that female auditors act more ethically than male auditors do. Moreover, better ethical behaviour is associated with more independent behaviour and thus with higher quality of audits. Therefore it is assumed that female auditors are perceived to be more independent than their male counterparts (Hardies et al., 2009). This might influence the perception of management about the auditors individual quality, which might increase the audit fee that the auditees are willing to pay, which drive the audit fee that is charged (Taylor, 2011; Hardies et al., 2009). Altogether, I assume that female auditors have higher ethical values than male auditors, which suggests that female audit partners charge higher fees than male audit partners.

(24)

3. Hypotheses

Taken all findings of prior literature together two hypotheses were formed that will be tested in this research. Prior research showed that females are better in planning audits than males, which suggests that female audit partners charge lower fees than their male colleagues. Females are more risk averse than their male counterparts, which suggests that female audit partners charge higher fees than male audit partners. Females tend to be better leaders than males, but females also have disadvantages in leadership, so the effect on audit fees remains unclear. Also the suggestion that males are more overconfident than females doesn’t hold for audit partners. However, when female auditors have higher ethical values than male auditors it can influence the perceived audit quality of the individual audit partner and thus the audit fee that is charged by female audit partners. With these suggestions in mind, the following hypothesis was formed.

H1: There is a relation between the engagement partner’s gender and audit fees.

The direction is neither positive nor negative, because based on the suggestions the direction could be either. However, to get some more insight in the differences between male and female auditors one more hypothesis was formed. Prior research, as described in the previous paragraphs, suggests that female audit partners are more risk averse than male audit partners. It is suggested that female audit partners charge a higher fee for risky clients than the male audit partners will do. Therefore the following hypothesis was formed:

H2: Female engagement partners charge higher audit fees than their male counterpart for risky firms.

(25)

4. Sample selection and empirical models

I collect data from companies listed on the London Stock Exchange (LSE) in the United Kingdom. Financial information data and audit fee data is obtained from DATASTREAM. The gender of the engagement partners is hand-collected from the annual reports of the companies that were often available on their own websites. Data was collected only for the year 2012, because of the feasibility of hand-collecting the gender of the engagement partners from the annual reports. I collected data from 5205 companies in the United Kingdom for the year 2012. I deleted 3536 companies, because they lacked audit fee data. Next, 892

companies were deleted, because they lacked control variable data and also 142 companies were deleted because they were financial institutions or banks (SIC-codes 6000-6900). After I hand-collected the audit partner gender data, I further deleted 43 companies, because I wasn’t able to get the gender of the audit partner out of the annual report for 2012. These procedures result in a final sample of 585 companies for the year 2012. For the second hypothesis 49 companies were deleted, because they lacked Beta data and then 351companies were deleted, because I didn’t consider them as risky firms. Table 1 summarizes the sample selection process.

TABLE 1

Sample selection

Panel A: Sample for testing H1.

U.K. Firms listed on London Stock Exchange in 2012 5202 Deleted: Firms without necessary audit fee data 3536 Firms without necessary financial data to compute control variables 896 Firms that are financial institutions or banks (SIC-Codes 6000-6900) 142 Firms without necessary audit partner gender data 43

Final sample for testing hypothesis 1 585

Panel B: Sample for testing H2.

U.K. Firms listed on London Stock Exchange in 2012 5202 Deleted: Firms without necessary audit fee data 3536 Firms without necessary financial data to compute control variables 892 Firms that are financial institutions or banks (SIC-Codes 6000-6900) 142 Firms without necessary audit partner gender data 43

Firms without necessary Beta data 49

Firms not considered a risky firm 351

Final sample for testing hypothesis 2 185

(26)

The research methodology that has been used in this research is empirical archival study. Because prior studies about audit fees (see for example Hay et al, 2006) mostly use empirical archival research, it is likely that it is the most suitable way to investigate the possible differences between male and female engagement partners on audit fees.

Engagement partners’ gender is not the only determinant of audit fees. Prior research has shown that there are many determinants of audit fees (Hay et al., 2006; Chan et al., 1993; Pong & Whittington, 1994). Those determinants have to be included in the model. Also control variables have to be included. The control variables that I use are derived from prior research that I already discussed in chapter 2 (Hay et al., 2006; Ittonen & Peni, 2012). Therefore I suggest the following model is appropriate for testing both hypotheses:

𝐿𝐴𝐹 = 𝛼 + 𝛽1𝐺𝐸𝑁𝐷𝐸𝑅 + 𝛽2𝑆𝐼𝑍𝐸 + 𝛽3𝐼𝑁𝑅𝐸𝐴𝑆 + 𝛽4𝐹𝑂𝑅𝐸𝐼𝐺𝑁 + 𝛽5𝐿𝑂𝑆𝑆 + 𝛽6𝐿𝐸𝑉

+ 𝛽7𝐵𝐼𝐺4 + 𝛽8𝐶𝐻𝐴𝑁𝐺𝐸 + 𝜀

LAF = The natural logarithm of audit fees

GENDER = Dummy variable; 1 for female audit partners, 0 otherwise SIZE = The natural logarithm of total assets

INREAS = Inventory and receivables divided by total assets FOREIGN = Foreign assets divided by total assets (ratio) LOSS = Dummy variable; 1 for a loss, 0 otherwise LEV = Total debt divided by total assets

BIG4 = Dummy variable; 1 for Big-4 auditor, 0 otherwise

CHANGE = Dummy variable; 1 if auditee changed auditor, 0 otherwise

The way I tested the second hypothesis differs slightly from the way I tested the first hypothesis. First I have to define what risky firms are. This is not done by using the measures that Hay et al. (2006) found to be risk measures for auditees, namely (1) inherent risk, (2) profitability and (3) leverage. I have used the Beta’s (β) of the companies to measure the riskiness of the companies. “Beta measures the sensitivity of a security to market-wide risk factors. For a stock, this value is related to how sensitive its underlying revenues and cash flows are to general economic conditions” (Berk & DeMarzo, 2007, p.319). Berk & DeMarzo (2007) also mention that stocks in cyclical industries, where the revenues and profits alter greatly in time, are more likely to have beta’s that exceed 1. The average beta of stock in the market is about 1, which means that the average stock price moves 1% for every 1% change

(27)

in economic conditions (Berk & DeMarzo, 2007). Where my sample consists of listed companies in the United Kingdom, I assume that the beta is a good measure to measure the sensitivity of the stock of those companies to general economic conditions and thus their riskiness. I assumed that a firm is risky when the Beta is larger than 1 or smaller than -1.

(28)

5. Results

5.1. Descriptive statistics

The descriptive statistics for all variables used for testing both hypotheses are reported in Table 2 and Table 3. Panels A show the full sample, Panel B the sample for firms with a female auditor and Panel C show the sample for firms with a male auditor. The CHANGE variable is not reported in both tables, because, surprisingly, when I was analyzing the data I found that no auditee changed his auditor from year 2011 to 2012. Therefore I excluded this variable when I ran the regression, because it would have no added value to the model. Additionally, the summary statistics of audit fees and total assets are in nominal values and not transformed into the natural logarithms yet to show a clearer view of the firms that were examined.

Table 2, Panel A shows the summary statistics for the full sample (n=585) for hypothesis 1. The smallest audit fees paid are £2,000 and the largest audit fees paid are £552,000,000. In terms of total assets the smallest company has a value of £433,000 and the largest company has assets with a value of £219,112,200,000. The means for audit fees and total assets are £2,550,793 and £2,712,292,750 respectively. I also compared my full sample with the full sample of Ittonen & Peni (2012), who examined three Nordic countries. U.K. firms tend to be less likely to be audited by a Big-4 audit firm (mean=0.653) than firms in the Nordic countries (mean=0.918). The mean leverage of U.K. firms (16.592) is smaller than the mean leverage of Nordic firms (21.374) and further U.K. firms are more likely to have a loss (0.340 versus 0.213) and U.K. firms tend to have more foreign assets than Nordic firms (32.788 versus 18.890).

The statistics in Table 2, panel B show that for firms with a female audit partner the means of the audit fees and total assets are smaller than in the full sample, while the medians of these variables don’t differ greatly with the full sample. Furthermore, it suggests that firms with female auditors tend to be more likely to have a loss and are more often audited by a Big-4 audit firm in comparison with the full sample. Table 2, Panel C shows that overall the means of total audit fees and assets of firms audited by a male audit partner are larger than those of the full sample. Those results are similar to the results of Ittonen & Peni (2012). Further, there are no big differences between the full sample and the sample of firms with a male audit partner for the other variables.

(29)

Table 2

Descriptive statistics H1

Panel A: Summary statistics for all firms (n=585)

Variable mean median max. min. std. Dev. Audit fees (x1000) 2,550.793 196.000 552,000.000 2.000 24,563.603 Assets (x1000) 2,712,292.750 100,033.000 219,112,200.000 433.000 14,858,202.283 INREAS 0.248 0.198 0.982 0.000 0.199 FOREIGN 32.788 20.910 135.700 0.000 34.696 LOSS(dummy) 0.340 0.000 1.000 0.000 0.474 Leverage 16.592 11.624 166.198 0.000 19.491 BIG4(dummy) 0.653 1.000 1.000 0.000 0.476

Panel B: Summary statistics for firms with female auditor (n=54)

Variable mean median max. min. std. Dev. Audit fees (x1000) 572.519 162.500 4,536.000 2.000 969.243 Assets (x1000) 549,349.907 97,174.500 5,358,089.000 2,084.000 1,004,129.236 INREAS 0.235 0.193 0.758 0.000 0.179 FOREIGN 33.252 28.115 100.000 0.000 33.853 LOSS(dummy) 0.407 0.000 1.000 0.000 0.496 Leverage 14.852 9.624 68.733 0.000 15.880 BIG4(dummy) 0.704 1.000 1.000 0.000 0.461

Panel C: Summary statistics for firms with male auditor (n=531)

Variable mean median max. min. std. Dev. Audit fees (x1000) 2,751.974 196.000 552,000.000 3.000 25,774.267 Assets (x1000) 2,932,253.040 101,110.000 219,112,200.000 433.000 15,576,693.230 INREAS 0.249 0.199 0.982 0.000 0.201 FOREIGN 32.740 19.190 135.700 0.000 34.812 LOSS(dummy) 0.333 0.000 1.000 0.000 0.472 Leverage 16.769 12.031 166.198 0.000 19.826 BIG4(dummy) 0.648 1.000 1.000 0.000 0.478

Table 3, Panel A shows the summary statistics of the sample for testing the second hypothesis. For the full sample (n=185), the means of the audit fees and total assets are

£5,821,914 and £4,091,297,378 respectively. The firms for the second hypothesis tend to have more foreign assets than the firms for the first hypothesis. Further, the firms of the second hypothesis tend to have more losses, a higher leverage and are more likely to be audited by a Big-4 auditor than the firms for the first hypothesis. There are no big differences between

(30)

Table 3

Descriptive statistics H2

Panel A: Summary statistics for all firms (n=185)

Variable mean median max. min. std. Dev. Audit fees (x1000) 5,821.914 314.000 552,000.000 2.000 42,639.509 Assets (x1000) 4,091,297.378 231,683.000 184,081,185.000 1,091.000 17,101,044.669 INREAS 0.260 0.197 0.950 0.000 0.220 FOREIGN 39.312 33.090 135.700 0.000 36.444 LOSS(dummy) 0.373 0.000 1.000 0.000 0.485 Leverage 18.088 14.320 130.530 0.000 19.660 BIG4(dummy) 0.741 1.000 1.000 0.000 0.440 Beta 1.420 1.370 4.300 -2.640 0.819

Panel B: Summary statistics for firms with female auditor (n=17)

Variable mean median max. min. std. Dev. Audit fees (x1000) 570.177 210.000 3,566.000 2.000 908.441 Assets (x1000) 769,913.235 257,266.000 5,358,089.000 2,084.000 1,410,134.895 INREAS 0.244 0.197 0.760 0.000 0.194 FOREIGN 38.229 29.200 100.000 0.000 38.222 LOSS(dummy) 0.412 0.000 1.000 0.000 0.507 Leverage 13.418 9.351 44.930 0.000 12.823 BIG4(dummy) 0.647 1.000 1.000 0.000 0.493 Beta 1.737 1.730 3.420 1.060 0.630

Panel C: Summary statistics for firms with male auditor (n=168)

Variable mean median max. min. std. Dev. Audit fees (x1000) 6,353.339 324.500 552,000.000 15.000 44,721.760 Assets (x1000) 4,427,389.821 215,753.000 184,081,185.000 1,091.000 17,910,570.319 INREAS 0.261 0.197 0.950 0.000 0.223 FOREIGN 39.421 33.335 135.700 0.000 36.377 LOSS(dummy) 0.369 0.000 1.000 0.000 0.484 Leverage 18.561 15.167 130.530 0.000 20.191 BIG4(dummy) 0.750 1.000 1.000 0.000 0.434 Beta 1.388 1.345 4.300 -2.640 0.830

Table 2, Panel B and Table 3, Panel B, besides that the risky firms have a higher mean for foreign assets to total assets than the full sample of hypothesis 1. Additionally, there are some differences between the firms with male auditors in the first and in the second sample. The mean of the audit fees and assets is larger in the second sample than in the first sample. Again

(31)

the second sample has a higher foreign assets to total assets ratio than in the first sample and further no large differences are worth mentioning.

Table 4, Panel A and Panel B show the pairwise correlations for the variables that I have used in the regressions for the first and for the second hypothesis, respectively. SIZE is highly and significantly correlated with the LAF variable, which I already expected. The bigger the company, the more audit work the auditor has to perform. Further, FOREIGN, LEV and BIG4 are also positively and significantly correlated with LAF. Additionally, LOSS is negatively and significantly correlated with LAF and my variable of interest, GENDER, is negatively correlated with audit fees, though not significant. Those correlations, except the GENDER correlation with LAF are similar to the correlations that Ittonen & Peni (2012) report in their paper. That is caused by the fact that Ittonen & Peni (2012) did their

examination in Nordic countries where companies are often audited by two different auditors together. That may have the result that a Nordic company is audited by audit partners with different genders. Therefore they were able to come up with different GENDER dummy’s and not with one dummy like in my examination of U.K. companies.. For their different

GENDER dummy’s they did find negative and positive correlations with LAF. FOREIGN is also positively and significantly correlated with SIZE as both variables are total assets or are calculated with total assets in Panel A, but however not in Panel B. Furthermore, the LOSS variable is significantly and negatively correlated with SIZE, INREAS and BIG4 in both Panel A and Panel B. Overall, the correlations for the first and the second hypothesis look familiar, but only some levels of significance differ. That is probably caused by the decrease in sample size for the sample which is reported in Panel B of Table 4.

Table 4

Correlations

Panel A: Pairwise correlations first hypothesis

Variable LAF SIZE INREAS FOREIGN LOSS LEV BIG4 GENDER LAF 1.000 .842** 0.036 .127** -.329** .246** .581** -0.043 SIZE 1.000 -0.070 .158** -.375** .292** .630** -0.023 INREAS 1.000 -.269** -.243** -0.015 0.025 -0.021 FOREIGN 1.000 .136** -0.008 0.020 0.004 LOSS 1.000 0.009 -.280** 0.045 LEV 1.000 .180** -0.028 BIG4 1.000 0.034 GENDER 1.000 31

Referenties

GERELATEERDE DOCUMENTEN

Since no distinct seasonal cycles could be identified for the measured aromatic hydrocarbons (Fig. 2), the possible influ- ence of air masses passing over different source regions

Deze Big Data Revolutie wordt ook uitmuntend beschreven in het boek ‘De Big Data Revolutie’, waarin big data wordt beschreven als bron van economische waarde en

Peer-review under responsibility of organizing committee of Fourth International Conference on Selected Topics in Mobile & Wireless Networking (MoWNet’2014)..

Het verschil in effect tussen live en uitgesteld kijken op merkbekendheid en merkreputatie van de sponsor kan mogelijk worden gemedieerd door de mate van betrokkenheid bij

This chapter described the running-in of rolling-sliding contacts on macroscopic and microscopic level. 1) On macro-scale, the geometrical change of the contacting

Tiago Filipe Montes de Jesus University of

A wide range of optical modulators are used in very different application areas, such as in optical fiber communication, displays, for active Q switching or mode locking of lasers,

In conclusion, our study demonstrates that the overall compliance with reporting guidelines of studies addressing diagnostic accuracy of pelvic floor 3D