Auditor's Gender and Audit Fees

Abstract

Documented behavioral differences between women and men suggest that the gender of the audit engagement partner may have implications for the price setting of the audit. Thus, this paper examines the effect of the auditor's gender on audit fees in an environment where the responsible audit partners can be identified. Using a sample of public firms from the NASDAQ OMX exchanges in three Nordic countries, we find evidence indicating that firms with female audit engagement partners have significantly higher audit fees. Although this is an interesting finding, it should be interpreted with caution since there is no clear theoretical explanation to support it. Potential reasons are introduced, such as the gender differences in risk tolerance, which may affect the pricing decisions by increasing the audit investment and/or increasing the audit fee risk premium. Alternatively, female auditors' diligence, lower overconfidence, and higher level of preparation could also lead to an increase in audit investment, and thereby result in higher audit fees.

SUMMARY

This paper examines the effect of the audit engagement partner's gender on the audit fees in an environment where the audit partner(s) responsible for the audit can be identified. Prior audit fee studies have focused on the effects of client characteristics, audit firm characteristics, and the engagement attributes on audit fees. Moreover, the audit fee literature, concentrating on for example the Big-4 premiums and industry specialization, has for long assumed that the pricing of an engagement is identical across the entire audit firm. However, in more recent literature the focus has increasingly been on examining how office level attributes affect audit fees. This study goes one step further by considering the individual audit engagement partner's potential effect on the audit fees. The purpose of this paper is to examine whether the gender of the auditor has an effect on the fees paid by the client. More specifically, it is of interest to study whether documented gender differences, for example in cognitive functioning and risk preferences, impact the audit engagement partner's decisions on the audit investment and/or audit fee risk premium, and thereby affect audit fees. Using a sample of public firms from the NASDAQ OMX exchanges in three Nordic countries, namely Finland, Denmark, and Sweden, we find exploratory evidence indicating that firms with female audit engagement partners have significantly higher audit fees than firms with male auditors. Since the existing literature does not provide a clear theoretical explanation for this finding, interpretations should be made with caution. Our results may, however, suggest that gender differences in risk tolerance influence the pricing decisions by causing either an increase in the audit investment and/or an increase in the risk premium of audit fees. Alternatively, female auditors' diligence and higher level of preparation could also lead to an increase in the audit investment, and thereby result in higher audit fees. However, it should be noted that because the theoretical background of the study is drawn from management, psychology, and finance literature, the reported findings are somewhat exploratory.

1. INTRODUCTION

The vast majority of previous studies on audit fees have focused on the effects of client characteristics, audit firm characteristics, and the engagement attributes on audit fees. The literature on audit fees assumes that the pricing of an engagement may be affected, for example, by the characteristics of the audit firm or office. Given that there may exist differences in individual auditors' attributes that could have an effect on the pricing of the audit engagement, for example distinctions in engagement planning, negotiation skills, team management capabilities, and risk preferences, more research in this area is called for (e.g. DeFond & Francis, 2005). The purpose of this study is to examine whether the gender of the auditor has an effect on the fees paid by the client. More specifically, we focus on studying whether and how the gender of the audit engagement partner affects the fees paid to the external auditors. If the audit partners can be associated with gender differences, for example in cognitive functioning, decision-making, leadership style, and risk preferences (Wood et al., 1985; Johnson & Powell, 1994; Eagly & Carli, 2003; Schubert, 2006) which may affect the decisions on the audit investment, the audit fees could be affected by the gender of the audit engagement partner.1 The potential consequences of gender differences have been subject to increasing interest in the recent corporate finance literature. The findings indicate, for instance, that female directors have to demonstrate superior competence to reach top positions (Eagly & Carli, 2003), they have higher expectations regarding their responsibilities and, thus, prepare themselves for the tasks more thoroughly (Fondas & Sassalos, 2000; Huse & Solberg, 2006), women tend to be less overconfident than men (Bonner, 2008) and, finally, women are more risk averse than men (Levin et al., 1988; Jianakoplos & Bernasek, 1998; Schubert, 2006). The possible gender-based differences in all of these areas could have an impact on the fees paid by the audit clients. As a consequence, it is important to empirically investigate the relationship between auditor gender and audit fees to find evidence on whether and how the audit engagement partner characteristics affect the audit fees.

We conduct this study by focusing on the listed firms in three Nordic countries: Denmark, Finland, and Sweden. In these countries, the responsible audit engagement partners' names, as well as the names of the audit firms they are representing, are mandatory information on the audit reports. This is in contrast with many other countries, for instance the US and the UK, where only the name of the audit firm is printed on the audit report.

Using a sample of 715 firm-year observations from NASDAQ OMX listed firms, we find evidence suggesting that firms with female audit engagement partners have significantly higher audit fees. Thus, in general, our empirical findings indicate that the auditor's gender may have an impact on audit fees. The results are interesting; however, interpretations should be made keeping in mind that there is no clear theoretical background for this finding and, therefore, more research is needed. The findings are consistent with the existing studies suggesting that women are on average more risk averse than men. The previous audit fee studies indicate that the risk tolerance of the audit partner may affect the audit fees in two ways: it could affect the level of the audit investment (e.g. audit effort), or the risk premium of the engagement may be influenced (Houston et al., 1999, 2005; Johnstone & Bedard, 2001, 2003). Thus, in both cases, a higher level of risk aversion may lead to a higher level of audit fees. Due to the fact that our theoretical background is based on evidence from the psychology, management, and finance literatures, and this issue has not been addressed previously in audit research, the reported empirical findings should be considered largely exploratory in nature. The remainder of this paper is organized as follows. Section 2 reviews the related literature and presents our research hypothesis. Section 3 describes the data on the OMX listed firms and presents the methodology used in the analysis. The empirical findings on the effect of auditor gender on audit fees are reported in Section 4. Finally, Section 5 provides concluding remarks.

2.RELATED LITERATURE AND HYPOTHESIS

Audit partners, gender, and audit fees The existing literature suggests that audit fees are to a large extent a function of audit team labor hours, audit team labor costs per hour, and a risk component. Following Francis' (2004) motivation concerning the shift from firm-level analysis to office-level analysis, it may be warranted to go one level further and consider how the characteristics of individual audit partners, such as the gender of the audit partner, can affect the audit investment component (i.e. the number of audit hours) and the risk component of the audit fees. The requirements for an adequate audit and the responsibilities of the engagement partner are regulated comprehensively. The International Standards of Auditing (ISA) (IAASB, 2009: ISA 220 and ISA 300) state that the audit partner is responsible for the overall quality of each engagement s/he is assigned to. The responsible audit partner should plan the nature, timing, and extent of guidance and supervision of team members, and review their work. The audit partner also needs to be assured that the engagement team has the required capabilities, competence, and time to perform the audit according to the professional standards and regulatory requirements. According to ISA requirements, the audit partner is required to use appropriate consultation and include specialized experts on the team in the case of complicated or contentious matters.2

Additionally, the Code of Ethics for Professional Accountants (IESBA, 2009) states that the audit engagement partner is responsible for setting the audit fee at a level which allows a sufficient amount of resources to be invested in the engagement. The Code emphasizes that there are no circumstances under which the amount of the audit fee can justify the lack of resources or time to properly perform the audit. As a consequence, individual behavioral attributes could have an effect on the engagement partner's estimation of the required amount of work. Thus, gender differences in risk aversion may lead female auditors to demand higher audit fees simply because they require a higher level of assurance, i.e. more work. The audit process consists of four major phases: (1) planning, (2) risk assessment, (3) conducting the audit, and (4) evaluating the results and issuing the report. In defining the size of the audit fee, the decisions and assessments made during the first two phases are particularly important. According to the professional regulation, the audit partner is responsible for the decisions related to the whole auditing process.

The documented behavioral differences between women and men, for example in planning, group decision-making, risk tolerance, or overconfidence, may affect the formation of the audit fee. Davidson and Gist (1996) examine how audit planning can affect the labor hour component of audit fees. They find that audit planning reduces the total audit effort up to a certain level, and thereafter increases the audit hours spent on an engagement. Additionally, they report that clients with higher assessed riskiness require more planning. According to the literature on gender differences, there may be features related to gender that affect the planning of the audit engagement. First, Huse and Solberg (2006) argue that women in high positions, such as corporate boards, are better prepared for meetings than men. Second, women tend to have higher expectations regarding their responsibilities (Fondas & Sassalos, 2000), and they have to demonstrate extra competence to reach top positions (Eagly & Carli, 2003). Third, women have better communicative skills and may, as a consequence, have a comparative advantage over men in tasks where communication within and among different groups is required (see, e.g., Wood et al., 1985; Maznevski, 1994; Fondas, 1997; Schubert, 2006). Thus, women tend to perform better in group problem-solving and decision-making tasks (Wood et al., 1985, Robinson & Dechant, 1997; Dallas, 2002). Communication skills may also give female auditors a comparative advantage over men in client negotiations concerning the engagement and, thus, female auditors may be able to sell a higher level of assurance to the client. Consequently, gender differences among audit partners may have an effect on both the audit planning as well as the audit process and, therefore, the gender of the audit partner may affect the number of hours the audit team spends on an engagement. If the female auditors use significantly more time in planning the task, it may increase the audit fees, whereas a reasonable investment in planning may reduce the audit effort and, thus, also the audit fees. Many studies propose that women are more conservative and risk averse than men (see e.g. Levin et al., 1988; Johnson & Powell, 1994; Powell & Ansic, 1997; Jianakoplos & Bernasek, 1998; Byrnes et al., 1999). Schubert (2006) suggests that women try to avoid losses and therefore tend to take less extreme risks. Planning an audit engagement includes the assessment of the inherent risk, control risk, and setting the detection risk, which are done by the engagement partner. The risk assessments are used for audit planning decisions concerning the nature, timing, and extent of audit evidence testing. Houston et al. (1999, 2005) and Johnstone and Bedard (2001, 2003) show that auditors address changes in different aspects of risk by adapting the audit investment and/or the risk premium. Consequently, if there are gender differences in the risk assessment process, risk tolerance, or assessment of the persuasiveness of evidence, female audit engagement partners could have higher audit fees because of the increased audit investment and/or the risk premium. Evidence from psychology research indicates that, in general, people tend to overestimate their knowledge and abilities (Fischhoff et al., 1977; Lichtenstein et al., 1982). Kennedy and Peecher (1997), Messier et al. (2008), and Owhoso and Weickgenannt (2009) find evidence of overconfidence at all levels of the audit team, namely partners, managers, seniors, and staff. Moreover, auditors seem to be overconfident in their own knowledge and abilities, as well as the abilities of other team members. Given that overconfidence exists also among the audit partners, as reported by Kennedy and Peecher (1997), Bonner (2008), Messier et al. (2008), and Owhoso and Weickgenannt (2009), it may affect the audit engagement partner's decisions on the audit investment during the audit procedures, or the audit fee risk premium. The earlier literature indicates that men are generally more overconfident than women, particularly in masculine domains (see Bonner 2008 for a review). Thus, the gender differences in overconfidence may also cause female audit partners to charge higher fees. Despite recent developments in gender equality, the world of business remains mainly a world of men. Thus, because the literature has reported a tendency for executives to hire subordinates like themselves (see e.g. Anderson-Gough et al., 2005), it can be more difficult for female auditors to achieve high positions, such as partnerships (Fogarty, 1996; Collin et al., 2007). This may lead to the females having fewer possibilities to charge higher fees, and thus a negative relationship between female auditors and audit fees could exist. Finally, if the well-documented gender wage gap also affects the audit fees, the female auditors could be associated with lower audit fees (e.g. Blau & Kahn, 1992, 2000; Munasinghe et al., 2008). However, particularly in engagements with large audit teams, the gender wage gap may not necessarily have any significant effect in the audit pricing context, as the salary of the responsible audit partner is only a part of the total audit fees charged from the client. Finally, it should be noted that the audit market is highly competitive, and potential individual partner effects may be mitigated by firm-level competition and audit tenders. Thus, due to market competition, the individual auditors may not necessarily have any effect on audit fees. Also, the exploratory nature of the study may influence the interpretation of the results.

Hypothesis The existing literature indicates that audit fees are a function of audit staff effort, audit staff labor costs, and client risk (see e.g. Niemi, 2002; Hackenbrack & Hogan, 2005). As the audit fee is preliminarily determined before the audit engagement, it is calculated using budgeted hours of each grade of auditing personnel and budgeted unit prices. However, due to the nature of audit fee formation, planning and risk assessment can significantly impact the final audit fee. Given the components affecting the audit fees and assuming that individual audit engagement partner characteristics may have an effect on decisions regarding the audit fees, it is of interest to examine the relationship between the gender of the audit engagement partner and audit fees. Thus, the hypothesis to be examined in this paper is: H1: There is a relation between the gender of the audit engagement partner and the audit fees. If there indeed exists an association between the gender of the audit engagement partner and the audit fees, it may be either positive or negative, as suggested by the gender-based differences reported in earlier studies. A positive relation is supported by the literature suggesting that the diligence, more thorough preparation, lower level of overconfidence, and the higher risk aversion of females would imply that women require a higher level of audit investment or risk premium, thereby leading to an increase in the audit fees (Wood et al., 1985; Levin et al., 1988; Blau & Kahn, 1992, 2000; Powell & Ansic, 1997; Huse & Solberg, 2006; Schubert, 2006; Bonner, 2008; Munasinghe et al., 2008). Furthermore, the better communication skills of females (see e.g. Wood et al., 1985; Maznevski, 1994; Fondas, 1997; Schubert, 2006) may give them an advantage in the bidding phase, as they may be able to convince the client to accept a higher priced audit, or for example to reassure the client that a more thorough and thus also more costly audit has to be performed. By contrast, however, also a negative relationship between female auditors and audit fees is supported by the earlier studies. Provided that the audit fees are a function of audit team effort, audit team labor costs, and client risk, the existing literature on gender differences suggests that better communication and teamwork skills of women, together with the gender wage gap, may reduce the effort needed and, as a result, decrease the cost of the audit engagement (see e.g. Wood et al., 1985; Blau & Kahn, 1992, 2000; Maznevski, 1994; Fondas, 1997; Schubert, 2006). Thus, the proposed superior communication skills of females may cause differences in audit fees in either direction. Moreover, the tendency for homophily in the hiring process of audit partners may give women fewer opportunities for charging high audit fees and, thus, lead to a negative relationship between female auditors and audit fees (Anderson-Gough et al., 2005). Finally, the potential partner effects may be cancelled out by competition in the audit market. Thus, this empirical study is largely exploratory in nature.

3.DATA AND METHODOLOGY

The initial sample used in the empirical analysis consisted of 1,210 firm-year observations from firms listed in the NASDAQ OMX exchanges in Denmark, Finland, and Sweden as of the end of 2007. OMX Nordic Exchange is a part of the NASDAQ OMX Group Inc. that operates in the Nordic and Baltic countries. The data used in the analysis cover the fiscal years 2005–2006. Following prior research, we excluded financial institutions (SIC codes 6000–6900) from the sample due to their unique features (396 firm-year observations excluded). Next, we excluded observations with insufficient data (99 observations excluded). This left us with a sample of 715 engagements from two years, which are classified by country and industry in Table 1. We utilized data from these countries because of the regulatory feature they provide for studying how the characteristics of an audit engagement partner affect the charges of the audit services. In the Nordic countries, auditors are required to personally sign the audit reports on behalf of the audit firm and, thus, we were able to identify the audit partner(s) responsible for each engagement. This is in contrast with the regulation in many other countries, e.g. the US and the UK, where only the name of the audit firm responsible for the audit is public information. Interestingly, the issue of publishing the audit engagement partner signatures is on the agenda of the Public Company Accounting Oversight Board (PCAOB) and, as a consequence, the engagement partner signatures may become obligatory in the US as well (PCAOB, 2009).

The data on the gender of the audit engagement partner were manually gathered from the firms' audit reports. In the Nordic countries, the audit report must be signed by at least one auditor, even if an audit firm is appointed. The signing auditor(s) is (are) here defined as the engagement partner(s), and by reviewing the signatory auditors' names it was possible to identify the gender of the engagement partner(s). For some engagements, the audit report was signed by more than one auditor from the same firm and in some cases also from two separate firms. In the engagements with multiple auditors signing the audit report, the clients have voluntarily opted for more than one audit partner or for more than one audit firm for the engagement. The legislation does not necessitate joint audits or multiple audit engagement partners in any of the three countries. The Danish legislation, however, required joint auditing (two audit firms) for listed firms until the end of 2004 and, as a consequence, it is more common for Danish firms to have joint auditing compared to the other two countries. Opting for a joint audit may also be caused by financial matters, as suggested by Thinggaard and Kiertzner (2008). Their results indicate that joint audits, where both auditors are responsible for a significant part of the audit task, tend to reduce audit fees in larger firms, as compared to audits where one auditor is in charge of the audit. Additionally, it is worth noting that, consistent with most of the other markets around the world, a great majority of the firms in our sample are audited by Big-4 auditors. The legal environment, as well as the listing requirements concerning corporate governance, financial reporting, and auditing in the three sample countries (Denmark, Finland, and Sweden) are relatively homogeneous and, thus, they can be analyzed as one group. La Porta et al. (1998) consider the Nordic countries (Denmark, Finland, Norway, and Sweden) to have similar laws to each other, but distinct from other countries. In contrast, Sinani et al. (2008) find significant differences in the formal board and ownership structures between the Scandinavian countries (Denmark, Norway, and Sweden). The main difference in ownership structure of the Nordic companies is that the Swedish firms have, on average, a higher proportion of family ownership, whereas state ownership is more common in Denmark and Finland (La Porta et al., 1998; Sinani et al., 2008). Moreover, the NASDAQ OMX Nordic Exchange Statistics from year 2007 reveal that, based on the market capitalization measured in the number of firms as well as the total value of the yearly trades, the Swedish stock exchange alone is as big as the Finnish and Danish exchanges together. This is also apparent from our sample, as over half of the firms are Swedish (NASDAQ OMX, 2008).

However, despite the observed differences in formal structures, Sinani et al. (2008) suggest that the corporate governance practices have strong similarities in these countries. The reason for this, according to Sinani et al. (2008), is the similarity in key corporate governance characteristics, such as trust, quality of enforcement, absence of corruption, quality of government, and freedom of speech, which are all results of the countries' equal small size and ethnical homogeneity. Our sample countries have close similarities in the requirements for becoming a CPA and, moreover, they comply with the Eighth Directive of the European Union. In all of the three sample countries, before being certified the auditors need to have the required theoretical education, professional experience, and have passed a practical examination. The relationship between audit fees and the gender of the responsible auditor is examined by employing cross-sectional panel regressions. Consistent with the literature on audit fees (see e.g. Simunic, 1980; Abbott et al., 2003), our model includes firm-specific variables to control for the known factors affecting audit fees. Furthermore, we also control for the industry-, country-, and year-specific factors by including corresponding control variable dummies in the equation. Thus, the following model is applied to examine the possible relation between auditor gender and audit fees where AFEEj,t denotes the natural logarithm of audit fees for firm j at t, INVRECj,t is the receivable and inventory intensity, LEVj,t is the percentage of total debt to total assets, LOSSj,t is a dummy variable which equals 1 if the net income of firm j during year t is negative, SIZEj,t is the natural logarithm of total assets, ROAj,t is the percentage of return on assets, FOROPj,t is the percentage of foreign assets (i.e. the assets owned by the firm abroad) to total assets, JOINTj,t is a dummy variable which has a value of 1 if at least two audit partners from different audit firms signed the audit report and each is responsible for a minimum of 20% of the task, CHANGEj,t is a binary variable, which equals 1 if the responsible audit firm(s) changed compared to the previous year, BIG4j,t is a dummy variable, which is assigned a value of 1 if the responsible audit firm is one of the Big-4 group and, thus, the variable is considered to control for the auditor reputation, NASj,t is the natural logarithm of non-audit service fees, YEAR2005 is a binary variable that indicates fiscal years, is a dummy variable according to industry classification (SIC) codes, and is a dummy variable indicating the company's country of origin. The data on the financial control variables are obtained from Thomson Financial Worldscope, and the data on audit firm-specific factors are gathered manually from the firms' annual reports.

The test variable in our equation is the female representation variable FEMj,t, which varies in different model specifications. We define four binary variables and a ratio to measure female audit engagement partner representation, and models from 2 to 6 are organized in order of decreasing female representation: FGROUP equals 1 if only women are acting as responsible auditors, FRATIO measures the share of female auditors compared to the number of auditors signing the audit report, FDUM equals 1 if a female audit partner participates in or is responsible for signing the audit report alone or together with a male, MIXED is set to 1 if a female audit partner is engaged with a male audit partner, and MGROUP is set to 1 if only male auditors sign the audit report. Only one female representation variable is included in the equation at a time in order to avoid possible multicollinearity problems. Therefore, we estimate six different regression specifications, where female variables are as follows: (i) no female variable, (ii) FGROUP, (iii) FRATIO, (iv) FDUM, (v) MIXED, and (vi) MGROUP. The data on female representation are gathered from the annual reports of the firms. Throughout the panel regressions, a three-way fixed-effects specification is used, because it allows for a different intercept for each industry in the sample. The method also controls for the country-specific effects and for the possible change in audit fees from one year to another. Moreover, in order to account for contemporaneous correlation and different variances in the disturbances of each cross-section, we employ the White cross-section robust covariances in the regression specifications.

4.RESULTS

Descriptive statistics

The descriptive statistics for the female audit engagement partner variables, as well as for the control variables, are reported in Table 2. Panel A shows the statistics for the entire sample of 715 observations. The statistics demonstrate that the observations are relatively heterogeneous in all respects. For example, the largest audit fees paid amounts to €14.491 million, whereas the smallest fee is only €0.004 million. In terms of total assets the range is from €0.513 million to 41.528 billion. Panel B of Table 2 presents the descriptive statistics for four sub-samples of the observations, which are categorized by female audit engagement partner representation. Some interesting remarks can be made on this table. For example, in the three sub-samples with female audit partner representation, the means and medians of audit fees, total fees, and total assets are in general smaller than in the sample as a whole. The median values of the mixed group are an exception to this, as the median values are larger for the mixed sub-sample than for the sample as a whole. Furthermore, it can be noted that the firms in the sub-sample with only male auditors are larger in terms of both audit fees and total assets. Finally, one interesting remark is that in the sub-sample with female audit partners only, the mean of LOSS is significantly higher than in the sample as a whole or the other sub-samples. In Panel C, we tabulate the summary statistics by country. There are significant differences between countries in the amount of leverage and joint auditing. Furthermore, fewer of the Finnish firms have reported losses and they have a higher proportion of foreign operations than the sample in general (significance levels noted by asterisks). Table 3 reports the pairwise correlations for the variables used in the regressions. As expected, SIZE is strongly correlated with our dependent variable audit fees (and total fees). Additionally, SIZE is negatively correlated with LOSS. FOROP is positively correlated with audit fees and total fees, which is natural as firms with significant international operations require a more time-consuming and, thus, a more expensive audit, which again increases the audit fees. Not surprisingly, our female representation variables are correlated with each other. Regression results The estimation results of our alternative fixed-effects panel regressions are reported in Table 4. We report regression results with six different model specifications, where our dependent variable audit fees is first regressed on our set of control variables and, then, in Models 2–6 we include one of our five different variables for female audit engagement partners in the equation. In Model 1, audit fees (AFEE) is regressed on firm-specific control variables (INVREC, LEV, LOSS, SIZE, ROA, FOROP, JOINT, CHANGE, BIG4, and NAS) and additional controls for industry, country, and year. As can be seen from the table, all our control variables are statistically significant at least at the 5% level, and they have the expected signs. Furthermore, our control variables have a good explanatory power for audit fees, since the adjusted R2 is 87.2% and the F-statistic is significant at the 1% level. Consistent with most of the literature following Simunic (1980), the inherent risk, leverage, and firm size are all positively associated with audit fees. Interestingly, the findings of Thinggaard and Kiertzner (2008) seem to hold also in our sample, as opting for a joint audit seems to significantly reduce the audit fees. In Models 2–6, we examine the relationship between female audit engagement partners and audit fees by regressing the audit fees on a female audit engagement partner variable and the firm control variables. In Model 2 the female audit engagement partner variable is FGROUP, which equals 1 if the responsible auditor(s) is (are) female. The estimation results indicate that audit fees are higher when a female audit partner is engaged. The coefficient for FGROUP is positive and statistically significant at the 1% level. All the control variables appear as expected and are statistically significant, and the explanatory power of the model is 87.2%. In Model 3, we use a ratio for female audit engagement partner representation (FRATIO) for each firm-year observation, and in Model 4 the female representation variable is FDUM, which equals 1 if the firm has at least one female audit engagement partner. Consistent with Model 2, the results for Models 3 and 4 presented in Table 4 also show that female audit engagement partner representation increases audit fees. Again, the results are statistically significant at the 1% level. In Model 5, we introduce the female representation variable MIXED, which equals 1 if the firm has appointed both female and male audit partner(s). Interestingly, the estimation results concerning our test variable turn out to be insignificant, thereby indicating that if both male and female partner(s) are assigned to the task, the auditor gender does not have a statistically significant impact on audit fees. Finally, Model 6 further confirms the results of our previous models by suggesting that firms with male audit engagement partners (MGROUP) have statistically significantly lower audit fees. In all models, the F-statistics and adjusted R2's indicate a good fit of the specification and, moreover, the results for all the female variables, except for MIXED, are statistically significant at the 1% level.

The results reported in Table 4 generally indicate that female audit engagement partners may have a positive effect on the audit fees. The coefficient estimates for our female auditor representation variables are positive and statistically highly significant, whereas the coefficient for the male auditor representation variable is negative and statistically significant. The results for mixed group are not statistically significant. The female representation variables seem to be important in our regressions, as they have a high significance level in all the model specifications, except for MIXED. In fact, according to our results, the gender variables seem to be even more important than leverage in explaining the cross-sectional variation in audit fees. In all, our results may indicate that the female auditors charge higher audit fees than the male auditors. There may be several factors that help explain this phenomenon. After describing our additional tests, we discuss the possible alternative explanations based on the literature on audit fees and gender differences. Robustness checks The regression results presented in Table 4 indicate that the gender of the responsible auditor may have an effect on the size of the audit fees. In the following, we examine the robustness of our results by conducting several additional tests. First, we test whether our results hold if the dummies for industry and country are removed from the models. The results of these regression specifications (not tabulated) are very similar to those presented in Table 4, and again in Model 4 the female representation variable MIXED lacks significance. The adjusted R2's are still close to 80% in all models, and the F-statistics are significant at the 1% level, suggesting that the model fits the purpose well even after removing the variables controlling for country and industry of the company.

In order to test that the results are not caused by a few outlying observations, we winsorize the audit fees and the control variables at the 0.5% and 99.5% levels, and re-estimate the six alternative models presented in Table 4. Furthermore, we also winsorize the data at the 2.5% and 97.5% levels to confirm that the results are not based on the winsorizing percentage. The results (not tabulated) further confirm our previous findings on gender-based differences in audit fee pricing. We have divided our sample into small and big firms to see if the audit pricing differs between the two groups. The regression results for the two sub-groups (not tabulated) are similar to those presented in the paper. However, they lack significance mainly because the variances in female dummy variables become substantially low. For this reason, we have also run our regression specifications including an interaction term between the alternative gender variables and client size, e.g. FGROUP ? SIZE. These tests provide rather similar results to those reported in Table 4, the main difference being that the coefficient for MIXED is here positive and significant at the 1% level. Furthermore, we have also gathered data for auditor-specific factors, namely auditor experience and tenure.3 Unfortunately, these data are available only for the Swedish firms, and thus we are unable to use the variables in the full sample regressions. We have, however, run a regression including the variables experience and tenure for the Swedish firms. When considering only one country from our sample, the small number of female observations becomes a problem. Therefore, we focus on testing these two additional variables in the model specification including the female representation ratio (FRATIO). FRATIO is chosen for these analyses because it has more variance than the dummy variables and, thus, it may be more useful in a smaller sample. This regression gives similar results to those presented in Table 4, i.e. the coefficient for FRATIO is positive. The results, however, lack significance and no conclusions can be drawn based on them. In general, it seems that experience tends to decrease the audit fees, whereas tenure has no impact on the fees. The negative impact of experience on audit fees can be explained by the routine and knowledge of the more experienced partners. Furthermore, if there is a market segment where the price competition is higher or, alternatively, if there are clients that are known to focus only on the price, the more experienced engagement partners may be able to perform the audit at a lower cost due to improvements in the efficiency of the engagement and, thus, they may be able to gain or maintain clients. As explained above, we have chosen to use the White cross-section robust covariances and to hold the firm-specific effects fixed in our models. To examine whether our results depend on the panel estimation specifications, we re-estimate the models presented in Table 4 with the ordinary coefficient covariance method, and with no fixed-effects specifications. These regressions (not tabulated) further indicate that the female audit engagement partners charge higher audit fees. These results, however, lack significance in the majority of the model specifications. We used the natural logarithm of the audit fees as our dependent variable throughout the regressions. In addition to audit fees, the companies often also pay their auditing companies other fees, for consulting, for example. Next, we re-estimate the models presented in Table 4, but this time using the natural logarithm of total fees as our dependent variable. Our regression results indicate that the firms with female auditors pay higher total fees than the firms with male auditors. The results, however, are insignificant in some of our model specifications (not tabulated). In all, the regression results with total fees as a dependent variable give results similar to those regressions with audit fees as a dependent variable. To complete our robustness checks, we re-estimate the models presented in Table 4 by including an interaction between the female representation variables and different risk measures, added one at a time to the model. The risk measures used here are the return on assets (ROA) and two common bankruptcy prediction measures developed by Altman (1968) and Zmijewski (1984). All of these variables are considered to measure the financial viability of the client. These tests are fairly consistent in showing that even after controlling for the riskiness of the client, the female audit engagement partners still have significantly higher fees. Summarizing the additional tests, we conclude that these robustness analyses generally provide support to the regression results reported in Table 4, as the estimated coefficient for the female variables appear positive and significant in most of our additional regressions.

Interpretation of the results In general, our results suggest that female auditors are associated with higher audit fees and possible explanations for the findings are discussed next. First, following prior literature on audit planning (Davidson & Gist, 1996) and gender differences that may affect the amount of resources allocated to planning (Fondas & Sassalos, 2000; Huse & Solberg, 2006), it may be argued that female audit engagement partners may invest more in planning the engagements, which could increase the audit effort and, therefore, also the audit fees. The higher risk aversion of females documented in, for example, the behavioral finance literature (Johnson & Powell, 1994; Powell & Ansic, 1997; Jianakoplos & Bernasek, 1998; Schubert, 2006) suggests that female auditors may assess some of the clients' risk components higher than their male counterparts. Higher assessed level of risk may affect the composition of the audit fees by either increasing the audit investment or increasing the risk premium (Houston et al., 1999, 2005; Johnstone & Bedard, 2001, 2003). Given that the audit engagement partners' assessment of the client's business risk affects the audit fees, our results may indicate that female auditors have a lower level of risk tolerance. Moreover, the gender differences in overconfidence (see Bonner, 2008, p. 104 for a review) may have an impact on the audit fees. Men are documented to be more overconfident than women, which can cause the female auditors to invest more time and effort in a task before being prepared to give the auditor's opinion. Thus, the lower overconfidence level of females can result in higher audit fees in the audits with a female audit partner. These exploratory findings do raise many questions, which are left to be answered by future research. Most importantly, perhaps, from the demand side perspective, it is of interest why any firm would hire a female auditor in a competitive audit market, if a male auditor is cheaper. First, this could simply be because of ignorance, as this kind of evidence has not, to the best of our knowledge, been discussed in the prior audit literature. Second, as the audit fees may also contain a flexible component, which is determined by the findings during the audit process, the prices may be difficult to estimate and compare beforehand. Third, if female audit partners are indeed more risk averse, they may increase audit testing more as a response to any risk component observed during the audit procedures. Finally, a female auditor can, however, also be hired because she is considered to provide actual or perceived benefits to the client. If, for example, female auditors are able to communicate in their tender that they are charging a higher audit fee because of more exhaustive audit testing, they may provide benefits to the quality-conscious clients and, thus, justify accepting a more expensive audit engagement. In addition, for well-documented gender-based differences, females may be perceived to produce a higher quality audit or to perform a more efficient audit of equal quality. Moreover, by hiring a female auditor the company may also attempt to gain social benefits comparable to those of, for example, nominating females to executive positions, such as boards of directors. Nevertheless, it should be kept in mind when interpreting our empirical findings that the analysis here is largely exploratory in nature.

5.CONCLUSIONS AND LIMITATIONS

This paper examines whether the gender of the audit engagement partner affects audit fees. Following earlier studies related to audit fees and gender-based behavioral differences in communication, overconfidence, decision-making, and risk-taking, we empirically examine the potential relationship between the gender of the audit engagement partner and the audit fees. We use a sample of NASDAQ OMX Nordic Exchange listed firms to study the relationship between the gender of the audit engagement partner(s) and audit fees. This setting enables us to identify the audit engagement partners for each firm from their audit reports. The results provide exploratory insights into the question of how audit engagement partner attributes, rather than audit firm- or office-level attributes, may affect audit fees. Our empirical findings suggest that the gender of the audit engagement partner may have an effect on the audit fees. In particular, after controlling for the client attributes, the female audit partners are documented to have significantly higher audit fees. This is an interesting finding; however, interpretations should be made with caution due to the lack of evident and convincing theoretical explanations. Our general theoretical background is drawn from the psychology, management, and corporate finance literatures. Thus, the reported empirical findings have to be regarded as somewhat exploratory and more work on the relationship between auditor gender and audit fees is needed. Our findings may suggest that gender differences related to planning, preparation, and diligence affect the engagement partner's decisions concerning the audit investment. Alternatively, the auditor's risk assessment of the client may be an important determinant in explaining the relationship between audit partner gender and audit fees. Previous studies suggest that the auditor-assessed riskiness of the client may increase the audit investment (i.e. number of audit hours) or the risk premium included in the audit fees and, therefore, the gender differences in risk tolerance may be a significant factor affecting the relationship between the auditor gender and audit fees. In interpreting the findings of this study, it is also necessary to consider the following limitations. First, an important assumption made in this study is that individual audit engagement partners are able to influence the audit fees, at least to some extent. Although the bidding process is obviously regulated to a certain extent by the headquarters, we find it reasonable to assume that the audit partner in charge of the engagement is integrated in the process and has an influence on the audit investment or the risk premium. This view is also supported by the ISA (IAASB, 2009), the auditor's code of ethics (IESBA, 2009), and our interviews with Big-4 auditors. Second, our sample consists of NASDAQ OMX listed firms from Denmark, Finland, and Sweden and, therefore, it is uncertain to what extent our results apply in other settings. Finally, our analysis may suffer from a self-selection bias. We have controlled for a set of client attributes which, according to the previous literature, affect the audit fees. However, it is possible that we have omitted some variables, or that some firm characteristics simultaneously affect the appointment of female audit engagement partners and audit fees. In particular, we acknowledge that auditor experience, specialization, and tenure may have an impact on the formation of the audit fees. These limitations should be considered when interpreting our findings, until corroborated by future research.

ACKNOWLEDGEMENTS

We would like to thank Editor Anne Loft, two anonymous reviewers, Lynette Chou, Juha-Pekka Kallunki, Timo Salmi, Ann Vanstraelen, and Sami Vahamaa for helpful discussions and comments. We gratefully acknowledge financial support from the NASDAQ OMX Nordic Foundation. In addition, Peni further acknowledges support from the Evald and Hilda Nissi Foundation, the Foundation for Economic Education, the Oskar Oflund Foundation, the OP-Pohjola Group Research Foundation, the Finnish Foundation for Economic and Technology Sciences (KAUTE), and the Finnish Foundation for Share Promotion.