1
The impact of the CEO gender on the level of the audit fees
Name: Judith Bosma Student number: 11147040 Thesis supervisor: Dr. A. Sikalidis Date: 14 June 2017
Word count: 12.046
MSc Accountancy & Control, specialization: Accountancy Faculty of Economics and Business, University of Amsterdam
2 Statement of Originality
This document is written by student Judith Bosma who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is 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 completion of the work, not for the contents.
3 Abstract
This paper investigates if the gender of the CEO has an influence on the level of the audit fees which are paid by the company to the external auditor. Prior literature suggests that there are differences between males and females. These gender differences could influence the level of the audit fees. There have been a lot of studies about the kinds of determinants which could influence the audit fees (Hay et al., 2006), but the gender of the CEO has never been used as a determinant which could influence the level of the audit fees therefore this study contributes to the existing literature. This study compares the Netherlands to Norway, because in Norway a compulsory women quota was introduced in 2003. Because of this women quota I expect there are more female executives in Norway than in the Netherlands. To test the two hypotheses, I used two different regression models, the sample had a time period from 2009 to 2015. The results of the study suggest that the gender of the CEO has an influence on the height of the audit fees in Norway but not in the Netherlands. In the second model I added the interest coverage ratio as a variable and the interaction between the gender of the CEO and the interest coverage ratio. The results show that the interest coverage ratio has no significance influence on the height of the audit fees, either in the Netherlands or in Norway. The limitations of this study are the small sample, the sample is restricted to two countries and almost all the companies in the sample are audited by a Big Four auditor. Suggestions for future research could suggest a comparison of other countries with Norway or other countries with each other.
4
Contents
1 INTRODUCTION ... 5
2 LITERATURE REVIEW ... 8
2.1 DETERMINANTS OF THE AUDIT FEES ... 8
2.1.1 Audit size ... 8
2.1.2 Audit Risk ... 9
2.1.3 Audit complexity ... 10
2.2 CEO’S INFLUENCE ON THE AUDIT FEES ... 11
2.3 POSSIBLE GENDER DIFFERENCES ... 12
3 HYPOTHESES ... 15 4 METHODOLOGY ... 17 4.1 METHODOLOGY ... 17 4.2 SAMPLE SELECTION ... 17 4.3 MODEL ... 19 5 RESULTS ... 22 5.1 DESCRIPTIVE STATISTICS ... 22 5.2 REGRESSION ANALYSIS ... 28 5.3 SENSITIVITY ANALYSIS ... 32 6 CONCLUSION ... 37 7 REFERENCES ... 39 APPENDIX A ... 42 APPENDIX B ... 43
5
1 Introduction
This chapter will introduce the subject of the research. Firstly, some background information about the subject. Secondly, I will explain the research question. Thirdly, the motivation will be presented. Finally, I will give you the structure of the paper.
In this study, I investigate whether the gender of the CEO of a company has an influence on the audit fees which are paid to the external auditor. According to many prior researches there are a lot of factors that can influence the level of the audit fees (Hay et al., 2006; Palmrose 1986; Simunic, 1980). Simunic (1980) concludes that the economic costs of an audit are reflected in the audit fees. The economic costs of an audit can vary through different kind of factors, such as the size of the company and the amount of risk within the company (Simunic, 1980). Hay et al. (2006) did a research on the determinants that can affect the audit fees, across different studies and countries from the last 25 years. This study summarizes all the determinants which can affect the audit fees. The most common factors that are used in prior studies are size, complexity, leverage, risk and profitability. These factors are categorized as client attributes (Hay et al., 2006). The research of Pong & Whittington (1994) also shows us that the size of an audit can be a proxy for the audit fees, because when the company is big, the auditor has to do more work to complete the audit than when the company is smaller. The study of Simunic (1980) shows us that if a company is more complex it will take the auditor more time to complete the audit, so the price of the audit will get higher than for a company with a low risk. Another factor which influences the audit’s complexity is the industry in which the company operates (Hay et al., 2006). Some general ledgers, like inventory and receivables, take more time and are more complicated to audit than other general ledgers. For example, banks are easier to control compared to a retail company because banks for example have no inventory, while a retail company has a lot of inventory. The study of Simunic (1980) talks about audit risk, and about how the audit risk can affect the height of the audit fees. Some parts of an audit are more risky and complex than other parts and therefore need to be reviewed by an expert (Simunic, 1980). Experts normally cost more per hout than a normal auditor, so the price of the audit fees will increase when there is an expert involved during the audit. The audit process which is performed by external auditors contains out of four phases: planning, risk assessment, conducting the audit and evaluate the result of the audit (Ittonen & Peni, 2012).
In the past decade the number of women working in high positions in the business community increased substantially. Because of this increase researchers have begun to investigate
6 the impact of females in the business community (Bardos et al., 1992; Naglieri & Rojahn, 2001; Houston et al., 1999; Gustafson, 1999). According to the researches of Bardos et al. (1992) and Naglieri & Rojahn (2001) females are better in planning than males are. Planning can have an significant influence on the height of the audit fees, because when the audit is well planned by the CEO, there will be less hours spend on the audit by the external auditors. Another difference between male and female is the perception of risk (Gustafson, 1999) and the assessments of risk (Houston et al., 1999). As said before, risk is a factor which can influence the height of the audit fees, so there might be a difference if there is male CEO or female CEO. The difference in leadership between male and female is difficult to measure according to Eagly & Carli (2003). They have showed that there are a couple of little differences between the leadership styles of males and females, there are pros and cons for both of the genders (Eagly & Carli, 2003). Altogether, I want to investigate if these gender differences might have an effect on the height of the audit fees.
In all the recent studies about determinants which can influence the level of the audit fees and about the gender differences in the business community there has never been a study on the effect of the gender of the CEO on the height of the audit fees. Because of the significant increase in the number of female executives in the business community, I would like to determine if this increase will have an effect on the level of the audit fees. This leads to the following research question:
What is the impact of the gender of CEO on the level of the audit fees?
In this study I will compare the Netherlands with Norway. The reason why I analyze the Netherlands and Norway is because Norway is the first country in Europe which introduced a quota for women on company boards (Storvik & Teigen, 2010). The quota has been introduced in 2003 and since then the number of female board members has reached up until 40% (Storvik & Teigen, 2010). Since the introduction in Norway other European countries have been debating about females is the business community and about maybe applying women quotas. The minister of Education, Culture and Science of the Netherlands, Jet Bussemaker, is also discussing about introducing a women’s quota in the Netherlands. The minister wants that at least 30% of all the top functions in the Netherlands are performed by females. The quota in Norway is compulsory, while in the Netherlands the quota is on a voluntary basis. I want to compare the Netherlands with Norway because in Norway this law has already been applied for 14 years now and in the Netherlands the quota is still voluntary. Because of the compulsory quota in Norway I expect
7 that there are a lot more female CEO’s than in the Netherlands where the quota is still voluntary. So if you compare the Netherlands and Norway with each other we can see the effect of the gender of the CEO on the height of the audit fees.
There are some studies about the effect of women in top executives jobs. The study of Smith el al. (2006) examines if there is a relationship between women in top executive jobs and on the performance of the company. The result of the study is that females have a positive effect on the company’s performance (Smith et al., 2006). According to the study of Ryan and Haslam (2005) females have a positive effect on the share price of a company, even when there is a general economic downturn. These studies suggest that there can be a relation between the gender of the CEO and the height of the audit fees, because when a company is performancing well it will become a going concern company (Smith et al, 2006). When a company is seen as a going concern it is assumed that the company is not risky, which could have an effect on the audit risk (Chan et al, 1993).
I can contribute to the existing literature because in the prior literature no one has ever determined if female executives have an effect on the audit fees companies have to pay to the external auditor. So with this study I can fill that gap in the existing literature. This study is also interesting from a social point of view because there is a big growth in the number of female executives (Francis et al., 2009). This study is also interesting for audit companies because if the gender of the CEO has an influence on the height of the audit fees then audit companies can anticipate that audit companies can for example adjust their strategy during the client acceptance phase. This study is also interesting for the companies themselves, because if male CEO’s are charged with higher audit fees than female CEO’s, companies may choose for a female CEO instead of a male CEO.
The study is structured as following. In the second chapter I will describe the main topics of the study with the use of prior literature, to provide some background information. The third chapter will be used for the formulation of the two hypotheses. In chapter four the methodology will be explained including the sample selection and the variables for the regression model. The results of the regression model and the sensitivity analysis will be discussed in chapter five. Chapter six contains the conclusion of this study, the limitations of this study and some suggestion for future research.
8
2 Literature review
This chapter will provide information about the theoretical subjects of the study, the information will be based on prior research. The information is based not only on auditing and financial literature but also psychology literature because there is not much literature about genders in the auditing literature. The first paragraph will describe the determinants of the audit fees which have been used before in prior literature (Hay et al., 2006). The second paragraph will describe the influence which a CEO can have on the audit fees. The last paragraph will list the differences between males and females in general.
2.1 Determinants of the audit fees
There have been a lot of studies about the determinants of the audit fees (Hay et al., 2006; Palmrose, 1986; Simunic, 1980). According to the studies of Hay et al. (2006) and Simunic (1980) there are a lot of determinants that can influence the level of the audit fees. Hay et al. (2006) summarizes all the results of previous studies and classifies the results in four categories: client attributes, auditor attributes, engagement attributes and other attributes. The client attributes that will be discussed are: audit size, audit risk and audit complexity. I choose to discuss the category client attributes because I think this is the category in which the CEO can exercise the most influence. According to Pong and Whittington (1994) the audit complexity and the audit risk are the most important factors and most commonly tested.
2.1.1 Audit size
According to Pong and Whittington (1994) large companies require more work than small companies, so audit size can be used as a proxy for the audit fees. If a company has a lot of transactions during one fiscal year, it will take the external auditor more time to examine all the transactions (Pong & Whittington, 1994). The audit is a statutory requirement for firms and therefore the demand of audits is inelastic, for this reason the fees of the audits are largely dependent on amount of work that is needed (Pong and Whittington, 1994).
Simunic (1980) concludes that the most dominant determinant of audit fees in all studies is audit size. According to Simunic (1980) audit size is expected to have a positive and nonlinear relation with the audit fees. Simunic (1980) measured the audit size by the variable total assets by the end of the year. Following Hay et al. (2006) the variable total assets at the end of the year is the mostly used variable to measure audit fees, because it is expected to have a positive relation
9 with audit fees (Simunic, 1980; Jubb et al., 1996).
As mentioned before the most used variable to measure audit size is the total assets at year’s end. Chan et al. (1993) did not use total assets at year’s end as a variable, they use turnover as a variable. Chan et al. (1993) did not use total assets at year’s end as a variable because they say that the total assets can vary significantly between similar companies, for example, because of the choice in accounting policy. They say that turnover is a better variable because it is transaction based (Chan et al., 1993).
2.1.2 Audit Risk
Prior research on company risk focused mostly on inherent risks, profitability and leverage (Hay et al., 2006). Nowadays audit firms base their audit approach on the perceived risk of an audit failure, the audit risk model helps the auditor to plan the scope of audit testing (Chan et al., 1993). The audit risk model which is used by auditors looks like this (Houston et al., 1999, p.284.):
Detection risk =;<=/>/<1 879: ? @A<1>A4 879:-../01234/ -5671 879:
Houston et al. (1999, p.284) gives a definition of the audit risk model: “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.” The charged audit fees will increase if the audit risk will increase, because the scope of the audit will be bigger (Houston et al., 1999). A stronger internal control will lead to lower audit fees because the control risk decreases and the detection risk will increase (Simunic, 1980).
According to Turley and Cooper (1991) high risk audits will result in higher audit fees, because when there is a high audit risk there should be more audit testing to ensure that there are no material misstatement in the financial statements. Davidson and Gist (1996) also conclude that companies with high assessed risk require more planning by the auditor, which results in higher audit fees. The audit risk is reflected in the normal business of the company and the control environment (Chan et al., 1993).
10 Consistent with the study of Pong and Whittington (1994) risk is very important for auditors because clients can charge the auditor for negligence. Before the audit the auditor should determine how many inherent risks there are in the company. An inherent risk is: “the susceptibility of an account balance or class of account balances to error that could be material assuming that there are no related internal accounting controls” (Peters et al., 1989, p.360). If there are a lot of inherent risks in a company then the audit has a higher risk and the auditor has to do more work, that is the reason why audit risk is positively related with the audit fees (Pong and Whittington, 1994). The reason why the external auditors spend more work on companies with a high risk is because they want to prevent damage to their reputation.
According to Simunic (1980) some balance sheet accounts are more ‘risky’ than other balance sheet accounts. Receivables and inventories are for example ‘risky’ balance sheet accounts and cash is seen as a ‘simple’ balance sheet account. These balance sheet accounts sometimes need specific audit procedures like confirmation and observation (Simunic, 1980). The audit procedures require more time from the external auditor than other audit procedures. Another reason why these balance sheet accounts are more ‘risky’ than other accounts is that these accounts require a forecast of future events, which is a complex task (Simunic, 1980). For example, the measurement of the fixed assets sometimes needs the work of specialist in real estate, because the measurement of fixed assets depends on future events.
2.1.3 Audit complexity
The third attribute that is often used in models to predict audit fees is audit complexity (Hay et al., 2006). The audit complexity costs are a reflection of the nature of the business, the location of the company, the number of unusual transactions like transactions outside the normal business and the quality of the internal control system (Chan et al., 1993). Simunic (1980) shows that if the company is complex it is more difficult and more time-consuming for auditors to audit the company, which will result in higher audit fees. The number of subsidiaries can measure the complexity of a company, this is the most used proxy to measure complexity (Hay et al., 2006). The reason why this measure is used often is because there is a strong relation between the number of subsidiaries and the complexity of a company (Pong & Whittington, 1994). When there are subsidiaries in the company there need to be eliminating intra-group transactions and all the subsidiaries have to be consolidated, which requires many work (Pong &
11 Whittington, 1994). Other proxies to measure complexity are: foreign subsidiaries, number of business segments and standard industrial classification (SIC) (Hay et al., 2006). According to Hay et al. (2006) the relationship between the audit fees and auditee complexity is positive and strongly significant almost every time.
According to Simunic (1980) some industries are more time-consuming and harder to audit. Hay et al. (2006) show that financial institutions are supposed to be easier to audit and are because of that reason often not included in the model. Financial institutions are supposed to be easier to audit because they have large assets comparable with other companies and they don’t have extensive inventories or receivables (Simunic, 1980). Manufacturing companies are harder to audit because they have many inventories and more receivables than financial institutions (Simunic, 1980).
Pong and Whittington (1994) look with at audit complexity from a different perspective. In particular, they look at how two variables, which are turnover and total assets, are involved in the amount of work involved in the audit. The audit fees may also interact with the audit size, because it involves more time and/or more specific work per unit (Pong and Whittington, 1994). All the determinants described in the previous paragraph are all determinants that are used repeatedly in prior research as control variables (Pong & Whittington, 1994; Simunic, 1980; Jubb et al., 1996; Hey et al., 2006; Palmrose, 1986; Chan et al., 1993; Houston et al., 1999; Turley & cooper, 1991). The determinants which are mentioned in this section will be used during my research.
2.2 CEO’s influence on the audit fees
CEO stands for Chief Executive Officer. The CEO is the head of the management of a company and reports to the board of directors (Harrison, 1991). The CEO has the responsibility to design the corporate strategy, the board has to oversee if the strategy is well implemented (Harrison, 1991). CEOs are appointed with the expectation that they will make reasonable management decisions to increase the shareholders’ value (Habib and Hossain, 2012).
The study of Graham et al. (2013) suggest that the CEO has an integral impact on financial decision-making and the risk that are linked to those decisions. The result of the study is that the physiological characteristics like risk aversion and optimism are linked to financial decision-making (Graham et al., 2013).
12 As long as the client’s financial risk is directly connected to the audit risk of the auditor, the client’s business risk is less directly connected to the auditors’ risk assessment (Billings et al., 2013). To reduce the audit risk to a certain level, auditors must undergo sufficient audit procedures. So if there is a high financial risk or business risk at a company, the auditor will request a higher audit fee.
According to the study of Graham, Harvey and Puri (2013) risk aversion and optimism, which are managerial attitudes, are linked to corporate financial policies. This suggests that managers can influence financial and organizational practice. The interest coverage ratio is for example part of the corporate financial policies (Graham, Harvey and Puri, 2013). The interest coverage ratio is a ratio which shows if a company can pay its interest costs on the outstanding debts (Dotan, 2006).
2.3 Possible gender differences
This paragraph will focus on the gender-based differences that exist between males and females. Gender differences in behavior have long been studied in the psychological literature. In the accounting research there is an increasing interest in the gender studies. The reason of this rise is because there is a big growth in the number of female executives in the business community (Francis et al., 2009).
According to the study of Ittonen and Peni (2012) there are several behavioral differences between males and females. There are for example differences in planning, group decision-making, risk tolerance and overconfidence. These differences may affect the audit fees (Ittonen and Peni, 2012).
Women in high positions, like corporate board or CEO function, prepare themselves better for meetings than men (Huse and Solberg, 2006). This is because women have higher expectations of their job responsibilities and therefore females want to demonstrate extra competence to reach their own expectations and the expectations of others (Eagly and Carli, 2003). Men are very driven to reach certain goals and to gain financial benefits, and are therefore more able to break rules for their own success (Habib and Hossain, 2013).
According to Habib and Hossain (2013) women are less likely to break rules and to behave unethical than men. Women are more leaning towards harmonious relations and helping other people than men (Habib and Hossain, 2013). Women have an advantage towards men
13 concerning communication since they have better communicative skills (Fondas, 1997). Because of the communicative advantage women perform better in tasks where communication is required, group problem-solving and group decision-making tasks (Fondas, 1997; Dallas, 2002). Eckel & Grossman (2001) examined gender differences with the use of the ultimatum game, the ultimatum game is an economic game. Female players in the ultimatum game are more likely to reject an offer than the male players. The findings of the study show that women have different social norms regarding fairness and equality than men (Eckel & Grossmann, 2001).
Many studies find that women are more risk averse than men (Lenvin et al., 1988; Habib and Hossain, 2013). Gender differences in risk aversion may lead to lower audit fees if there is a female CEO because there is a lower audit risk (Itonnen and Peni, 2012). Schubert (2006) finds that women tend to make less extreme risks than men because women are more willing to avoid losses than men. Research also suggests that women are more conservative than men (Bonner, 2008). Gustafson (1999) found that that are differences between males and females within the perception of risk as well. The first finding of the study is that the definition of risk has a different meaning between males and females. Secondly, males worry less about risks because males create and handle risks more often than females (Gustafson, 1999).
According to Bardos et al. (1992) females are better in planning than males. Females significantly needed less time to perform the tasks which considered as a measure of planning, these tasks are: organize tasks, prevent perseveration and plan to shift between certain rules, than males, which shows that females outperformed males (Bardos et al., 1992). The study of Naglieri and Rojahn (2003) also suggest that females are better in planning than males.
There is less hard evidence on which gender is better when leadership in concerned. Burke and Collins (2001) do find that females have different kinds of leadership styles, these leadership styles seem to be better than the leadership styles that males have. Research shows that the female leadership styles are more effective than the leaderships styles man have, like the contingent reward leadership styles (Burke & Collins, 2001). Male leaders are more likely to reward employers who deliver a good performance while female leaders are more likely to develop a good relationship with the employers and serving as a role model for the rest of the company (Burke & Collins, 2001). Eagly and Carli (2003) analyzed the differences in leadership styles between males and females as well, they found positive and negative points for both genders. The study of Eagly and Johnson (1990) found the same result as the study of Burke and Collins (2001). They also found some small differences in leadership between male leaders and
14 female leaders. Female leaders focus more on personal relationships with employers, which complement the results of Burke and Collins (2001). Male leaders are more likely to create an autocratic leadership style whereas females are more likely to adopt a democratic leadership style (Eagly & Johnson, 1990).
15
3 Hypotheses
Taken all findings of the prior literature together, this chapter will explain the hypotheses that will be tested during this study. Prior research, as described in the previous chapter, showed that females are more risk averse than males (Lenvin et al., 1988; Habin & Houssain, 2013). This suggests that while using the audit risk model, the control risk will decrease and the detection risk will increase. A higher detection risk will result in less work for the auditor because there will be less evidence needed to provide reasonable assurance about the financial statements. I expect that a male CEO, who is less risk averse than a female CEO, will lead to higher audit fees for the company to pay to the external auditor. Because I expect that a male CEO will have a lower detection risk than a female CEO. According to many researches (Bardos et al., 1992; Levin et al., 1988; Eagly and Carli, 2003) females are better at planning than males. Naglieri and Rojahn (2001, p.435) describe the planning process as: “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”. The audit fees which firms have to pay to the external auditor is, among other factors, depends on the hours spend on the audit by the external auditor. If the firm is well prepared for the audit, for example gather all the information which is needed by the auditor on time and quickly bring in the information the external auditor requested, it may reduce the working hours from the external auditor. I expect that when there is a female CEO the audit will be better prepared than when there is a male CEO working at the firm and for this reason the audit fees will be lower with a female CEO.
Therefore the following hypothesis was formed:
Hypothesis 1: Companies with a male CEO will experience higher audit fees than companies with a female CEO.
Prior research has shown that females are more risk averse than males (Lenvin et al., 1988; Habib & Houssain, 2013). To measure risk aversion, I count in the interest coverage ratio. The interest coverage ratio is a ratio which is used by companies and shareholders to measure if the operations of the company can generate enough funds to pay the interest expenses on outstanding debt (Dotan, 2006). If companies have a high interest coverage ratio, it means that their operations generate enough funds to pay the interest on the outstanding debts. Therefore, I expect that firms with a high interest coverage ratio are qualified as going concern firms and
16 therefore will be seen as non-risk firms. If a company is seen as a risky firm by the external auditors, then the external auditors have to do more work to give assurance on the financial statements.
This leads to my second hypothesis:
Hypothesis 2: Companies with a male CEO will experience a lower level interest coverage ratio then companies with a female CEO, and therefore experience higher audit fees.
17
4 Methodology
This chapter will explain the methodology of the study, the sample selection and the research models. In the first part the methodology of the study will be discussed. The second paragraph will explain the sample selection procedures. The third paragraph will give an explanation about the research model for the relation between the CEO gender and the audit fees. In this paragraph will also give information about the variables that are used in the two different research models.
4.1 Methodology
The research methodology to answer the research question will be a quantitative archival study. The reason why I have choosen a quantitative archival study is because most of the prior studies about audit fees used this methodology (Simunic, 1980; Hay et al., 2006).
4.2 Sample selection
For this quantitative study I will use data from the Netherlands and from Norway. The sample will be based on observations from fiscal year 2009 till fiscal year 2015 to examine the relation between the gender of the CEO and the audit fees. I choose this time period because I wanted to use the most recent available data. The reason why I excluded the fiscal year 2016 is because not all financial data is available in the databanks yet.
I have collected my data set from Compustat, Datastream and through hand-collecting public data. First, I substracted a global dataset from Compustat with a 7 year period, running from fiscal year 2009 till fiscal year 2015. In the global dataset I filtered out the data for the Netherlands and Norway. I deleted 360 observations because not all necessary financial data was available. After that I collected the audit fees data in Datastream and connected the audit fees data with the dataset from Compustat. There were some gaps in the audit fees data from Datastream, to keep the data set as big as possible I filled up those gaps with hand collecting data. The variables Gender and Big Four were not available in any databank, so I collected that data through hand collecting. The final sample contains 507 observations for the Netherlands and 756 observations for Norway. At last I checked the normal distributions of the variables. Table 1 summarizes the sample selection process.
18
Table 1
Sample selection
Panel A: Sample for testing hypothesis 1 and 2
Observations
Dutch firms extraced from Compustat 1217
Deleted Firms withour necessary financial data 350
Firms without necessary audit fees 365
Total sample for the Netherlands 502
Norway firms extraced from Compustat 1607
Deleted Firms withour necessary financial data 454
Firms without necessary audit fees 394
Total sample for Norway 759
The Norwegian currency is the Norwegian Krone and the Dutch currency is the Euro. To make a good comparison between Norway and the Netherlands, the currencies of the two countries had to be adjusted to one similar currency. I chose to adjust to Norwegian Krone to the Euro, because the Euro is the most common currency in Europe. I looked up the currency rates from NOK to EURO on oanda.com. I used the currency rates from the end of each fiscal year. I converted the fiscal year end numbers from the financial statements of the Norway observations to the Euro. In table 2 you can see that rates which are used during the recalculations of the Norwegian observations.
Table 2
Currency rate
Date NOK EUR
31-dec-09 1 0,12008 31-dec-10 1 0,12787 31-dec-11 1 0,12872 31-dec-12 1 0,13543 31-dec-13 1 0,11863 31-dec-14 1 0,1104 31-dec-15 1 0,10452
19
4.3 Model
To answer the research question I designed two different regression models, which are derived from the model of Ittonen & Peni (2012). I adjusted the model of Ittonen & Peni (2012) with some variables that are used in prior research (Hay et al., 2006) and explained in the literature chapter, which is chapther 2. The models for the two hypotheses are two relatively similar models, the two different models are showed in the text below.
I conducted a regression model to test if my hypothesis of whether the gender of the CEO has an influence of the height of the audit fees. I designed my model with variables used from the models of Ittonen & Peni (2012) and Hay et al. (2006). The model is stated below and in table 3 you can see the variables, the definition of the variables and the sources of the data. Regression model for hypothesis 1:
BCD = E + β1 IJKLJM + β2 OPQJ + β3 SKMJET + β4 VPW DXYM + β5 [PT\ + β6 Lev + β7 Net income + c
Table 3
Variable Definition Source
LAF The fees which auditors
charge to the companies thay audit. Measured by the natural logarithm of audit fees.
Datastream and
complemented with hand collected data
Gender The gender of the CEO of a
company. Dummy variable; 1 for male and 0 for female
Hand collected
Size The total assets of the book value of the auditees total assets. Measured by the natural logarithm of the total assets.
20
Inreas Total of inventory and
receivable divided by the total assets of the companies.
Compustat
Big Four The external auditor a
company. Dummy variable; 1 for Big Four auditor and 0 for non-Big Four auditor
Hand collected
Risk Dummy variable; 1 for risky firms and 0 for non-risky firms. Companies with a SIC code beginning with number 6 are companies within the financial services. Companies in the financial services sector are less risky to audit because they have less receivables and inventories.
Compustat
Lev Total debt divided by the
total assets of the company.
Compustat
Net income The net income or loss reported in the financial statements of the company.
Compustat
IntCov Total interest expenses
divided by the earnings before interest and taxes. A good interest coverage ratio is higher than 2,5 (Dothan, 2006). Dummy variable; 1 for good ratio and 0 for bad
21 ratio.
Gender * IntCov Interaction between the variable Gender and the variable IntCov. Dummy variable; 1 for a male CEO with a good interest coverage ratio and 0 other.
Hand collected and Compustat
The second regression model is the same as the regression model which is used for hypothesis 1, there are only 2 variables added. The variables that are added to the model are the interest coverage ratio and the interaction between the interest coverage ratio and the gender of the CEO. The regression model for hypothesis 2 is stated below and in table 3 you can observe the variable overview, the definition of the variables and the sources of the data.
Regression model for hypothesis 2:
BCD = E + β1 IJKLJM + β2 OPQJ + β3 SKMJET + β4 VPW DXYM + β5 [PT\ + β6 Lev + β7 Net income + β8 IntCOv + β9 Gender ∗ IntCov + c
22
5 Results
This chapter will discuss the results of the regression models for this study. Firstly, the evaluation of the descriptive statistics and the Pearson correlation matrix for the Netherlands and Norway will be explained. Following, the results of the regression models will be analyzed. At last, the sensitivity analysis will be discussed.
5.1 Descriptive statistics
The descriptive statistics for the variables used for testing hypothesis 1 and hypothesis 2 are shown in table 4. Panel A shows the sample for the Netherlands and Panel B shows the sample for Norway. For the descriptive statistics, the values of the audit fees and the total assets are transformed into the natural logarithms.
In table 4, Panel A describes the summary statistics for the sample of the Netherlands (N=507) for hypothesis 1. The smallest audit fees that are paid in the Netherlands has a natural logarithm of 9,392 and the highest amount of audit fees paid has a natural logarithm of 17,766. The mean of the variable gender is 0,98, which means that 98% of all the CEO’s in the sample are male, and therefore 2% of the sample is a female. The mean of the variable Big Four is 0,92, which means that 92% of the companies in the sample have a Big Four audit company as their external auditor.
Table 4, Panel B describes the summary statistics for the sample of Norway (N=756). The lowest amount of audit fees paid in Norway has a natural logarithm of 4,908 and the highest amount of audit fees paid has a natural logarithm of 17,766. The mean of the variable gender is 0,92, which means that 92% of all the CEO’s in the sample are male, and therefore 8% of the sample is a female. The mean of the variable Big Four is 0,96, which means that 96% of the companies in the sample have a Big Four audit company as their external auditor.
The main difference between the samples of the Netherlands and Norway are the audit fees. The difference between the means of the natural logarithm of the audit fees is 0,989. The difference within the variable gender is small, in Norway 92% of the CEO’s are male and in the Netherlands 98% of the CEO’s are male. Another difference between the Netherlands and Norway is that in Norway 96% of the companies in the sample have a Big Four auditor, in the Netherlands 92% of the companies have a Big Four auditor. There a further no big differences
23 between the sample of the Netherlands and the sample of Norway.
Table 4 also shows the summary statistics of the variables of the sample for the Netherlands (N=507) for testing hypothesis 2. The statistics are mostly the same as the statistics from hypothesis 1, but there are 2 extra variables in this sample. The variables that are added to the sample are the interest coverage ratio and the interaction between the gender and the interest coverage ratio. The mean of the interest coverage ratio is 0,66, which means that 66% of the companies in the sample have a good interest coverage ratio. The interaction variable of the gender and the Interest coverage ratio has a mean of 0,64, which means that 64% of the companies from the sample have a good Interest coverage ratio and a male CEO. The descriptive statistics of the other variables in the model did not change in comparison to descriptive statistics of hypothesis 1.
Table 4, Panel B describes the summary statistics of the sample of Norway (N=756) for testing hypothesis 2. The mean of the Interest Coverage ratio is 0,420, which means that 42% of the companies in the sample have a good Interest Coverage ratio. The interaction variable of the gender and the Interest Coverage ratio has a mean of 0,38, which means that 38% of the companies from the sample have a good Interest Coverage ratio and a male CEO. The descriptive statistics of the other variables in the model did not change in comparison to descriptive statistics of hypothesis 1.
The biggest differences between the samples from the Netherlands and Norway are within the interest coverage ratio and in the interaction variable, gender and the interest coverage ratio. In the Netherlands 66% of the companies have a good interest coverage ratio while in Norway only 42% of the companies have a good interest coverage ratio. There are other difference in the interaction variable, in the Netherlands 64% of the companies have a good interest coverage ratio and a male CEO while in Norway only 38% of the companies have a good interest coverage ratio and a male CEO. There are further no extra differences between the samples of the Netherlands and Norway. The table with the descriptive statistics from the sample for hypothesis 2 is stated below.
24
Table 4
Descriptive statistics hypothesis 1 and 2 2
Panel A: Summary statistics for the Netherlands
Variable mean median max. min. std. Dev
Audit fees 13,776 13,839 17,766 9,392 1,781 Size 14,202 14,333 20,965 4,394 2,531 Inreas 0,263 0,262 0,869 0,000 0,210 Big Four 0,920 1,000 1,000 0,000 0,279 Gender 0,980 1,000 1,000 0,000 0,133 Risk 0,760 1,000 1,000 0,000 0,424 Lev 0,159 0,114 2,000 0,000 0,180 Net income 226 21 10.387 -2.240 865 IntCov 0,660 1,000 1,000 0,000 0,474 Gender*IntCov 0,640 1,000 1,000 0,000 0,479
Panel B: Summary statistics for Norway
Variable mean median max. min. std. Dev
Audit fees 12,777 12,751 17,824 4,908 1,372 Size 12,928 12,968 19,541 4,962 2,371 Inreas 0,172 0,098 0,776 0,000 0,181 Big Four 0,960 1,000 1,000 0,000 0,189 Gender 0,920 1,000 1,000 0,000 0,266 Risk 0,770 1,000 1,000 0,000 0,418 Lev 0,250 0,229 0,831 0,000 0,207 Net income 6.610 58 940.061 -37.500 55.701 IntCov 0,420 0,000 1,000 0,000 0,494 Gender*IntCov 0,380 0,000 1,000 0,000 0,487
25 Table 5 shows the pairwise correlation matrix for the variables from the Dutch sample which are used in the regression for both hypotheses. The variable size is as expected highly and significantly correlated with the variable audit fees. If a company is bigger, it will take more time for the external auditor to audit the whole company. This result is the same as the result of the studies from Pong & Whittington (1994), Simunic (1980) and Jubb et al., (1996). Other variables like gender, risk net income, interest coverage ratio and the interaction between the gender and the interest coverage ratio are as well positive and significant correlated with the audit fees. The variables inreas and Big Four are negatively and significantly correlated with the audit fees.
Table 6 shows the pairwise correlation matrix for the variables from the Norwegian sample which are used in the regression for both hypotheses. All the variables in the sample are significantly correlated with the variable audit fees. The variable with the highest significant correlation is the variable size, this is in line with my expectation and in line with prior studies (Pong & Whittington, 1994; Simunic, 1980). The only variable with a negative correlation with the variable audit fees is net income. The variables inreas and risk are highly and significantly correlated with each other, this is what is expected. Inventories and receivables are more difficult to audit than other general ledgers, therefore that general ledgers are more risky (Palmrose, 1986).
The most obvious difference between the samples of the Netherlands and Norway is that the variables inreas and gender are negatively correlated in the sample for the Netherlands and positively correlated in the Norwegian sample. The variable gender is more strongly correlated with the audit fees in the Dutch sample than in the Norwegian sample. The correlation in the Dutch sample is 0,359 and in the Norwegian sample is the correlation 0,219. The variable size is also less strongly correlated in the Norwegian sample than in the Dutch sample.
26
Table 5
Pairwise correlation matrix for the Netherlands
Variable Audit fees Size Inreas Big Four Gender Risk Lev Net income IntCov Gender*IntCov
Audit fees 1,000 0,743** -0,183** -0,107* 0,359** 0,101* 0,014 0,186** 0,135** 0,104* Size 1,000 -0,390** -0,070 ,453** -0,308** 0,036 0,130** 0,083 0,063 Inreas 1,000 0,059 0,016 0,523** -0,111* -0,017 0,141** 0,156** Big Four 1,000 -0,042 -0,075 -0,057 0,026 -0,097* 0,181** Gender 1,000 -0,075 0,060 0,083 0,188** 0,174** Risk 1,000 -0,163** 0,108** 0,249** 0,225** Lev 1,000 0,178** 0,087 -0,069 Net income 1,000 0,087 0,093* IntCov 1,000 0,961** Gender*IntCov 1,000
**= Correlation is significant at the 0,01 level, 2-tailed *= Correlation is significant at the 0,05 level, 2-tailed
27
Table 6
Pairwise correlation matrix for Norway
Variable Audit fees Size Inreas Big Four Gender Risk Lev Net income IntCov Gender*IntCov
Audit fees 1,000 0,598** 0,143** 0,131** 0,219** 0,089* 0,121** -0,081* 0,304** 0,341** Size 1,000 -0,188** 0,077* 0,245** 0,396** 0,315** 0,051 0,131** 0,129** Inreas 1,000 0,127** 0,071* 0,437** -0,372** -0,093* 0,380** 0,404** Big Four 1,000 0,122** 0,103** -0,070 0,031 -0,040 0,230** Gender 1,000 -0,091* 0,192** 0,022 0,018 0,087* Risk 1,000 -0,193** -0,179** 0,194** 0,213** Lev 1,000 -0,050 -0,284** -0,265** Net income 1,000 -0,038 -0,031 IntCov 1,000 0,924** Gender*IntCov 1,000
**= Correlation is significant at the 0,01 level, 2-tailed *= Correlation is significant at the 0,05 level, 2-tailed
28
5.2 Regression analysis
The result of the multivariate regression model of the analysis of the effect the gender of the CEO on the audit fees are reported in tables 7 and 8. To test my hypothesis I run two different regressions, in both of the regression the variable audit fees is the dependent variable. Table 7 represents the regression analysis for the first hypothesis and table 8 represents the regression result for the second hypothesis. Within the tables the coefficients, the standard error, the t-statistic and the significant levels are presented.
Table 7, panel A represents the regression result for the Dutch sample. The adjusted R-square of the model is 0,872. This means that this model has a high explanatory power. The regression analysis shows a significant positive association with the variables size and risk in relation to the audit fees, p-values are lower than 0,05. I expected a positive and significance based upon the result of prior studies (Pong & Whitington, 1994; Simunic, 1980; Jubb et al., 1996). This result tells us that the size of a company and the amount of risk within a company have an effect on the height of the audit fees. The variable of interest is the gender of the CEO, the result of the gender of the CEO has a positive coefficient of 0,151 with a significance level of 0,027. Which means that the gender of the CEO has an influence on the height of the audit fees, but the evidence is not very strong because if a level of 0,01 is used instead of the 0,05 this variable is not significant anymore.
Table 7, panel B represent the regression results for the Norwegian sample. The adjusted R-squared of this model is 0,522, this means that the model explains about 50% of the variability of the response data around the mean. This model has less explanatory power than the model of the Netherlands, because the Dutch sample has adjusted R-squared of 0,872. In the Norwegian sample there are a lot more variables who are significant compared to the Dutch sample. The variables size, inreas, gender and risk are significant. The variable gender has a positive coefficient of 0,691 with a significance level of 0,000. The coefficient is in line with my expectations, because I expect that if there is a male CEO the audit fees will increase. The significant level is 0,000 which means that the gender of the CEO is significant. Based on this regression result we can assume that that the gender of the CEO does have an effect on the height of the audit fees which are paid to the external auditor.
29 Based on this regression model there is an indication that in the Netherlands and in Norway the height of the audit fees are affected by the gender of the. This means that there is support for hypothesis 1 in the Dutch sample and in the Norwegian sample, although the evidence for the Netherlands is not very strong.
Table 7
Regression results for hypothesis 1
Panel A: Regression result for hypothesis 1 for the Netherlands Variables Coefficients Std. Err. t Sig
Size 0,693 0,019 37,417 0,000 Inreas -0,092 0,212 -0,433 0,666 Gender 0,151 0,197 -0,765 0,027 Big Four 0,347 0,157 -2.219 0,445 Risk 1,241 0,112 11,117 0,000 Lev 0,539 0,236 -2,290 0,023 Net income 0,0004 0,000 1,396 0,164 N 507 Adj. R-squared 0,872 F-statistic 308,97 P (f) 0,000
Panel B: Regression results for hypothesis 1 for Norway Variables Coefficients Std. Err. t Sig
Size 0,432 0,017 25,728 0,000 Inreas 0,943 0,225 4,189 0,000 Gender 0,691 0,132 5,248 0,000 Big Four 0,432 0,189 2,283 0,023 Risk 1,014 0,099 10,200 0,000 Lev -0,089 0,189 -0,473 0,637 Net income 0,000 0,000 -2,262 0,024 N 756 Adj. R-squared 0,522 F-statistic 119,107 P (f) 0,000
30 Table 8, panel A represent the results of the regression for the Dutch sample. The adjusted R-squared is 0,674 in this model. The adjusted R-R-squared decreased from 0,872 to 0,674, this means that the model has lost 0,198 of his explanatory power. In this model there are only two variables significant, the variables that are significant are size and risk. The coefficient of the variable gender is -0,176 with a significance level of 0,620, which means that the gender of the CEO has no influence on the height of the audit fees. The added variable gender*IntCov has a negative coefficient of -0,110 with a significance level of 0,284. This coefficient is in line with my expectation, but this variable has no influence on the height of the audit fees.
Table 8, panel B represent the results of the regression for the Norwegian sample. The adjusted R-squared is in this model 0,533, which means that roughly 50% of this model is explained by the variables that are used in the regression. The adjusted R-squared is increased with 0,011 by adding the variables interest coverage ratio and the interaction between the gender and the interest coverage ratio. The added variables give the model 0,011 more explanatory power. Within this model there are six variables that have a significance level lower than 0,05, the variables are: size, inreas, gender, big four, risk and net income. The variable gender*IntCov has a negative coefficient of -0,107 and a significance level of 0,680. The coefficient is in line with my expectation because I was expecting that the audit fees would decrease when there was a good interest coverage ratio. But this variable is not significant, so the audit fees are not influenced by the variable gender*IntCov.
Based on this regression there is an indication that in the Netherlands the height of the audit fees are not affected by the gender of the CEO but in Norway the gender of the CEO has an effect on the height of the audit fees. So there is support for hypothesis 1 in the Norwegian sample and not in the Dutch sample. The interest coverage ratio does not have an influence on the height of the audit fees in both of the countries. This means that the second hypothesis is not supported with the result of this study.
31
Table 8
Regression results for hypothesis 2
Panel A: Regression results for hypothesis 2 for the Netherlands Variables Coefficients Std. Err. t Sig
Size 0,585 0,023 25,569 0,000 Inreas -0,479 0,273 -1,753 0,080 Gender -0,176 0,356 -0,496 0,620 Big Four 0,112 0,186 0,602 0,547 Risk 1,664 0,134 0,397 0,000 Lev 0,364 0,262 1,386 0,166 Net income 0,00006 0,000 1,115 0,265 IntCov 0,390 0,228 1,945 0,069 Gender*IntCov -0,110 0,102 -1,072 0,284 N 507 Adj. R-squared 0,674 F-statistic 130,498 P (f) 0,000
Panel B: Regression result for hypothesis 2 for Norway Variables Coefficients Std. Err. t Sig
Size 0,409 0,017 23,459 0,000 Inreas 0,665 0,232 2,868 0,004 Gender 0,802 0,179 4,474 0,000 Big Four 0,406 0,188 2,159 0,031 Risk 0,950 0,0099 9,575 0,000 Lev 0,139 0,194 0,719 0,472 Net income -0,001 0,000 -2.231 0,026 IntCov 0,460 0,248 1,854 0,064 Gender*IntCov -0,107 0,259 -0,413 0,680 N 756 Adj. R-squared 0,533 F-statistic 97,268 P (f) 0,000
32
5.3 Sensitivity analysis
The regression results in table 10 indicate that the gender of the CEO has an effect on the height of the audit fees. I performed a sensitivity analysis to test whether the results were driven by the effect of industries and by the effect of years. I added industry dummies and year dummies to the regression model, and performed the same regression for both of the samples. Table 9 represent the variables which are used in the regression, the definition of the variables and the source of the data. The results of the sensitivity analysis are shown in table 10, panel A shows the results for the Dutch sample and panel B shows the Norwegian sample.
Regression model for the sensitivity analysis
!"# = & + β1 *+,-+. + β2 012+ + β3 4,.+&5 + β4 718 #9:. + β5 <15= + β6 Lev + β7 Net income + β8 IntCOv + β9 Gender ∗ IntCov
+ β. −. Industry dummies + β. −. Year dummies + Z
Table 9
Variable Definition Source
LAF The fees which auditors
charge to the companies they audit. Measured by the natural logarithm of audit fees in Euro’s.
Datastream and
complemented with hand collected data
Gender The gender of the CEO of a
company. Dummy variable; 1 for male and 0 for female
Hand collected
Size The total assets of the book value of the auditees total assets. Measured by the natural logarithm of the total assets.
Compustat
Inreas Inventory and receivable
divided by the total assets of
33 the companies.
Big Four The external auditor of a company. Dummy variable; 1 for Big Four auditor and 0 for non-Big Four auditor
Hand collected
Risk Dummy variable; 1 for risky firms and 0 for non-risky firms. Companies with a SIC code beginning with number 6 are companies within the financial services. Companies in the financial services sector are less risky to audit because they have less receivables and inventories (Simunic, 1980).
Compustat
Lev Total debt divided by the
total assets of the company.
Compustat
Net income The net income or loss reported in the financial statements of the company.
Compustat
IntCov Total interest expenses
divided by the earnings before interest and taxes. A good interest coverage ratio is higher than 2,5 (Dothan, 2006). Dummy variable; 1 for good ratio and 0 for bad ratio.
34 Gender * IntCov Interaction between the
variable Gender and the variable IntCov. Dummy variable; 1 for a male CEO with a good interest coverage ratio and 0 other.
Hand collected and Compustat
Industry dummy The SIC code represents a certain industry. Dummy variable; 1 for one specific industry and 0 for the other industries.
Compustat
Year dummy Dummy variable; 1 for one specific year and 0 for the other years.
Compustat
Table 10, panel A displays the adjusted R-squared of 0,874, this means that this model has large explanatory power. The coefficient of the variable gender has decreased to 0,012 with a significance level of 0,956. The coefficient of 0,012 is not in line with my expectations because I expected a higher coefficient. This suggests that by adding the industry and year dummies it had a negative effect on gender variable, which is the variable of interest, and became insignificant. Two variables that are still significant in this regression model are size, with a coefficient of 0,691 and risk with a coefficient of 1,332. The variable gender interacted with the interest coverage ratio has a coefficient of -0,191, which is in line with my expectation because a good interest coverage ratio should decrease the height of the audit fees. This variable is also significant and therefore this variable has an influence on the height of the audit fees.
Table 10, panel B shows us an adjusted R-squared of 0,561, which means that about 50% of the model is explained with these variables. The gender variable has a coefficient of 0,767 with a significance level of 0,000, this coefficient is in line with my expectations. Compared to the Dutch sample almost all the variables in the Norwegian sample are significant, only the variables leverage and interaction between the gender and the interest coverage ratio are insignificant. This
35 means that the height of the audit fees paid to the external auditor are influenced by the variables size, inreas, gender, Big Four, risk, net income and the interest coverage ratio. The coefficient of the variable gender interacted with the interest coverage ratio is -0,122, this is in line with my expectation. However the significance level of this variable is 0,632, which means that this variable is insignificant.
Based on this sensitivity regression there is an indication that in the Netherlands the height of the audit fees are not affected by the gender of the CEO but in Norway the gender of the CEO has an effect on the height of the audit fees. So there is support for hypothesis 1 in the Norwegian sample and not in the Dutch sample. The interest coverage ratio does have an influence on the height of the audit fees in the Netherlands but not in Norway. This means that the second hypothesis is not supported by the result of the study.
Table 10
Sensitivity analysis for hypothesis 2
Panel A: Regression results for the Netherlands
Variables Coefficients Std. Err. t Sig
Size 0,691 0,019 37,203 0,000 Inreas -0,174 0,216 -0,801 0,423 Gender 0,012 0,221 0,055 0,956 Big Four -0,204 0,150 -1,355 0,176 Risk 1,332 0,118 11,250 0,000 Lev -0,426 0,257 -1,659 0,098 Net income 0,000 0,000 1,125 0,261 IntCov 0,480 0,232 2,004 0,073 Gender*IntCov -0,191 0,078 -2,454 0,015 Industry dummy Yes
Year dummy Yes
N 507
Adj. R-squared 0,874 F-statistic 106,27
36 Panel B: Regression results for Norway
Variables Coefficients Std. Err. t Sig
Size 0,420 0,018 23,363 0,000 Inreas 0,601 0,256 2,352 0,019 Gender 0,767 0,175 4,375 0,000 Big Four 0,870 0,212 4,102 0,000 Risk 1,476 0,124 10,069 0,000 Lev 0,211 0,197 1,071 0,284 Net income 0,000 0,000 -2,304 0,021 IntCov 0,508 0,242 2,101 0,036 Gender*IntCov -0,122 0,245 -0,480 0,632 Industry dummy Yes
Year dummy Yes
N 756
Adj. R-squared 0,561 F-statistic 44,987
P (f) 0,000
The overall result in this study does not provide support for hypothesis 1: Companies with a male CEO will experience higher audit fees than companies with a female CEO. In Norway the gender of the CEO does have an influence on the height of the audit fees, but in the Netherlands the gender of the CEO has no influence on the height of the audit fees. The overall results of the study do not support the second hypothesis: Companies with a male CEO will experience a lower level interest coverage ratio then companies with a female CEO, and therefore experience higher audit fees. In the Netherlands and in Norway this hypothesis is not supported.
37
6 Conclusion
This study has investigated if the gender of the CEO will have an influence on the height of the audit fees which is paid to the external auditor. Because of the rising number of female executives and the women quota’s in Europe, I was interested to see if this would affect the height of the audit fees that companies pay to their external auditor. This study will contribute to the existing academic literature because there has never been a study that looks at the gender of the CEO as a determinant of the audit fees (Hay et al., 2006). During the study different kind of fields of literature have been used, I have used management, auditing and psychology literature.
The reason why I am comparing the Netherlands with Norway is because Norway is the first European country who has introduced a compulsory women quota in 2003. The quota claims that at least 40% of the board positions have to be carried out by females. Because of this quota I expected that there would be more female executives in the business community in Norway than in the Netherlands.
The literature suggests that there are certain differences between males and females. For example, females are more risk averse than males (Lenvin et al., 1988; Habib and Houssain, 2013), which could lead to a higher audit risk. Secondly, women are better at planning (Naglieri & Rojahn, 2001), this might lead to a better prepared audit. At last, females in high positions prepare themselves better for their job and want to demonstrate more competence (Huse & Solberg, 2006; Eagly & Carli, 2003), what can lead to a more organized audit. All these differences between female and male CEO’s could influence the height of the audit fees.
The sample of the study contains a seven year period, namely the period from 2009 to 2015. I gathered the data for the variables size, inreas, risk, leverage, net income, interest coverage ratio from Compustat and for the variable audit fees from Datastream. The variables gender, Big Four and a part of the audit fees were gathered by hand collecting data. After I deleted the incomplete data, I checked the variables for normal distributions. The total sample for the Netherlands contains 507 observations and the sample for Norway contains 756 observations. To test my two hypotheses I made two different regression models.
The result of the first regression model suggests that the gender of the CEO does have an effect on the height of the audit fees. In both countries, respectively the Netherlands and Norway, the gender variable is significant with a 0,05 level. In the second regression model the variable gender is not significant in the Dutch sample, but in the Norwegian sample this variable
38 is significant. After running the sensitivity analysis the variable gender is still insignificant in the Dutch sample and still significant in the Norwegian sample. The result of the first regression model suggests that hypothesis 1 is supported.
The results of the second regression model suggest that male CEO’s with a good interest coverage ratio do not influence the height of the audit fees. In both countries, the Netherlands and Norway, this variable is insignificant. After running the sensitivity analysis the variable gender interaction with the interest coverage ratio is still insignificant. These results cannot support the second hypothesis.
Therefore, the overall results of this study provide no support for hypothesis 1: Companies with a male CEO will experience higher audit fees than companies with a female CEO. Neither the second hypothesis is supported: Companies with a male CEO will experience a lower level interest coverage ratio then companies with a female CEO, and therefore experience higher audit fees. The overall result of the study is that the gender of the CEO has an influence on the height of the audit fees in Norway. The results for Norway suggest that the women quota has a positive influence on the height of the audit fees. The Netherlands might have the same results as Norway if the women quota is not voluntary anymore but changed into a compulsory quota.
This study contains some limitations. The first limitation is that the sample size is very limited, not all the data was available in the databanks. The second limitation is that almost all of the companies in the Dutch sample and in the Norwegian sample are audited by a Big Four auditor, so the result of this study are not relevant to non-Big Four auditors. Another limitation is that the gender variable almost only contains males, respectively 98% in the Netherlands and 92% in Norway. Maybe if there were some more females in the sample the results of the study will be different. The last limitation is that the research only used data from the Netherlands and Norway, so the result can be different when there are different countries included in the sample.
There are some suggestions for future research. The first suggestion is to make the sample a lot bigger by adding more non-Big Four auditors to the sample. The downside of adding non-Big Four data that it should be done by hand-collecting data which takes a lot of time. Another suggestion is to compare other countries with Norway or maybe other countries with each other. The last suggestion for future research is to investigate if the gender of the CEO has an influence of the different kind of stages of the audit process. This would give the auditors even more insight information about how to adjust their pricing strategy in the client acceptance phase.
39
7 References
Bardos, A.N., Naglieri, J.K. & Prewett, P.N. (1992). Gender differences on planning, attention, simultaneous, and successive cognitive processing tasks. Journal of School
Psychology, Vol. 30, No. 3, pp. 293-305.
Billings, B. A., Gao, X., & Jia, Y. (2013). CEO and CFO equity incentives and the pricing of audit services. Auditing: A Journal of Practice & Theory, 33(2), 1-25.
Bonner, S. (2008), Judgment and Decision Making in Accounting, Upper Saddle River, NJ: Prentice-Hall.
Burke, S. & Collins, M. (2001). Gender differences in leadership styles and management skills.
Women in Management Review, Vol. 16, No. 5, pp. 244-256.
Chan, P., Ezzamel, M. & Gwilliam, D. (1993). Determinants of audit fees for quoted UK companies. Journal of Business Finance & Accounting, 20(6), pp. 765-786.
Davidson, R.A. & Gist, W.E. (1996). Empirical evidence on the functional relation between audit planning and total audit effort. Journal of Accounting Research, Vol. 34, No. 1,
pp. 111-124.
Dallas, L. (2002). The new managerialism and diversity on corporate boards of directors. Tulance
Law Review, Vol 76, pp. 1363-1405.
Dothan, M. (2006). Costs of financial distress and interest coverage ratios. Journal of Financial
Research, 29(2), 147-162.
Eagly, A.H. & Carli, L.L. (2003), ‘The female leadership advantage: An evaluation of the evidence’, The Leadership Quarterly, Vol. 14, pp. 807–34
Eagly, A.H. & Johnson, B.T. (1990). Gender and leadership style: A meta-analysis. Psychogocial
Bulletin, Vol. 108, No. 2, pp. 233-256.
Eckel, C. C., & Grossman, P. J. (2001). Chivalry and solidarity in ultimatum games. Economic
inquiry, Vol. 39, No.2, pp. 171-188
Francis, B.B., Hasan, I., Park, J.C. & Wu, Q. (2009). Gender differences in financial reporting decision-making: Evidence from accounting conservatism. Rensselaer Polytechnic
