DOES FEMALE REPRESENTATION ON AUDIT COMMITTEES AFFECT AUDIT FEES?

Kim Ittonena, Johanna Miettinenb, Sami Vhmaaa,*

a

University of Vaasa, Department of Accounting and Finance b Lahti University of Applied Sciences First Draft: August 2007 This Draft: June 2011 Abstract

This paper examines the association between female audit committee representation and audit fees. We hypothesize that gender diversity on audit committees has implications for external auditing, and thus affects audit fees. Using a sample of the S&P 500 firms, we find considerable evidence to suggest that firms with female audit committee chairs have significantly lower audit fees. From the audit demand perspective, our findings may indicate that female audit committee chairs reduce the need for assurance provided by external auditors. Alternatively, from the supply-side perspective, female chairs and committee members may decrease audit fees by affecting the auditors assessment of audit risk, for instance, by improving the effectiveness of internal control activities or by generally enhancing the integrity of the financial reporting process. If female representation reduces the inherent risk of misstatements, gender diversity on audit committees may be negatively associated with audit fees. Keywords: audit fees; audit committees; gender diversity; female directors

We would like to thank three anonymous referees, Liesbeth Bruynseels, Mary Jane Lenard, Alexandra Niessen, Paroma Sanyal, Andrew Yim, Mikko Zerni, and participants at the 2008 American Accounting Association Meeting, the 2008 AAA Auditing Section Meeting, the 2008 Southwest AAA Meeting, the 2008 Midwest Finance Association Meeting, the 2008 Southwestern Finance Association Meeting, the 2008 AFAR Workshop, and the 2009 EARNet Symposium for helpful discussions and comments. The authors gratefully acknowledge the support of the Emil Aaltonen Foundation, the Finnish Foundation for Advancement of Securities Markets, the Finnish Savings Banks Research Foundation, the Foundation for Economic Education, the NASDAQ OMX Nordic Foundation, the Ostrobothnia Chamber of Commerce, and the Paulo Foundation. * Corresponding author. Address: University of Vaasa, Department of Accounting and Finance, P.O. Box 700, FI-65101 Vaasa, Finland; Tel. +358 6 324 8197; Fax: +358 6 324 8344; E-mail address: sami@uwasa.fi

INTRODUCTION Psychology and management literature have long acknowledged the existence of significant gender differences, for instance, in cognitive functioning, communicative skills, decision-making, and leadership styles. Given these differences, the recent corporate finance literature has examined the effects of female representation on corporate boards and committees, and noted that gender diversity may have important implications for corporate governance (see e.g., Carter et al. 2003; Erhardt et al. 2003; Huse and Solberg 2006; Rose 2007; Adams and Ferreira 2009; Huse et al. 2009). In general, prior studies indicate that female representation contributes to board effectiveness and may enhance the boards monitoring activities within the firm. Nevertheless, the issue of gender diversity has so far been largely ignored in accounting and auditing literature. In this paper, we examine the role of gender diversity on corporate audit committees in relation to external auditing. In particular, we attempt to assess whether and how female audit committee representation affects the fees paid to the external auditors. Given that recent legal reforms impose strict requirements on the composition of the audit committee and emphasize the committees monitoring responsibilities in ensuring the integrity of the financial reporting process (Sarbanes-Oxley Act 2002; SEC 2003), it is important to consider the potential effects of gender diversity on the functioning and effectiveness of audit committees. To the best of our knowledge, this study is the first attempt to address the role of gender diversity on audit committees. Thus, by focusing on the association between female audit committee representation and audit fees, this paper provides novel insights into empirical audit pricing literature.

Over the past few years, the relationship between audit committee characteristics and audit fees has been examined in Abbott et al. (2003), Lee and Mande (2005), GoodwinStewart and Kent (2006), and Vafeas and Waegelein (2007). In brief, these studies indicate that audit committee characteristics are important determinants of audit fees. Consistent with demand-side argumentation, the findings reported in previous studies suggest that audit fees are positively related to audit committee size, expertise, independence, and meeting activity. The present paper aims to extend the existing literature by considering the potential effects of audit committees gender diversity on audit fees. Based on the recent corporate governance literature (e.g., Adams and Ferreira 2009; Huse et al. 2009), we presume that female representation may enhance the monitoring activities of the audit committee. Our empirical findings demonstrate that female representation on audit committees affects audit fees. In particular, using a sample of the S&P 500 firms, we find considerable evidence to suggest that female representation is negatively related to the fees paid to the external auditors. The results are particularly strong for firms with female audit committee chairs. Consistent with previous studies, our control variables for other audit committee characteristics (size and activity level) are positively associated with audit fees. From the demand-side perspective, our results may indicate that female chairs and audit committee members reduce the need for assurance provided by external auditors. Alternatively, from the supply-side perspective, female chairs may decrease audit fees by affecting the auditors assessment of audit risk, for instance, by improving the communication with external and internal auditors, and/or by enhancing the effectiveness of internal monitoring

and the integrity of the financial reporting process. If female representation reduces the inherent risk of misstatements, gender diversity on audit committees may be negatively associated with audit fees. Overall, our findings are broadly consistent with the view that gender diversity may lead to improvements in the effectiveness and monitoring activities of the audit committee. However, given the somewhat exploratory nature of the analysis, our empirical findings should be interpreted with some caution. The remainder of this paper is organized as follows. The next section reviews the related literature and presents our research hypothesis. The third section describes the data on the S&P 500 firms, while the fourth section presents the methodology used in the analysis. The empirical findings on the relationship between female audit committee representation and audit fees are reported in the fifth section. Finally, the last section provides concluding remarks.

BACKGROUND AND HYPOTHESIS DEVELOPMENT

Audit Committees and External Auditing

The importance of audit committees within corporate governance has received increasing emphasis from regulatory authorities in the last ten years. The current regulations emphasize the role of the audit committees monitoring activities in ensuring the integrity of the financial reporting process (BRC 1999; Sarbanes-Oxley Act 2002; SEC 2003). Among other requirements, the Sarbanes-Oxley Act and the current SEC rules require that audit committees are directly responsible for the appointment, compensation,

and oversight of external auditors.1 Following the increased regulatory interest, a considerable body of empirical auditing literature has recently focused on the relationship between audit committees and external auditing. In brief, this literature indicates that audit committees may have a critical role in enhancing audit quality. Lee et al. (2004) and Turley and Zaman (2004) argue that audit committees may affect the quality of auditing in several respects. First, the audit committee may demand that the firm appoints a more knowledgeable auditor with a better reputation (see e.g., Abbott and Parker 2000; Chen et al. 2005). Second, the audit committee may improve audit quality by demanding a greater quantity of audit effort from the incumbent auditor (e.g., GoodwinStewart and Kent 2006; Abbott et al. 2003; Carcello et al. 2002). Furthermore, the audit committee may increase the independence of external auditors by providing support for the auditor in potential disputes with the firms management (see e.g., Knapp 1987; DeZoort and Salterio 2001; DeZoort et al. 2003). Fourth, audit committees may strengthen the internal audit function of the firm, and thereby reduce the need for assurance provided by external auditors (Turley and Zaman 2004; Abbott et al. 2008). Finally, audit committees may enhance the integrity of audit reports by mitigating the potential threat of auditor replacement based on managerial motives (see e.g., Carcello and Neal 2000; Lee et al. 2004). Over the past few years, empirical studies have addressed the association between audit committee characteristics and audit fees. These studies use observable audit
1

Other important responsibilities of audit committee members include, for instance, the oversight of internal

audit functions and the evaluation of internal control systems. In this paper, we presume that female representation on the audit committee may have important implications for these monitoring activities.

committee characteristics to measure effectiveness, and assume that effectiveness increases with the committees size, financial expertise, independence, and meeting activity.2 Moreover, prior research notes that the relation between audit committees and audit fees is affected by both the client firms demand for audit services and the supply of these services by the external auditor (see e.g, Abbott et al. 2003; Turley and Zaman 2004; GoodwinStewart and Kent 2006). According to the demand-side argumentation, effective audit committees may be expected to demand more audit effort and greater assurance in order to achieve higher audit quality. These demands obviously lead to higher audit fees. The supply-side arguments, on the other hand, suggest that the involvement of an effective audit committee is bound to strengthen the internal controls of the firm. Because of the audit committees monitoring activities, the external auditor may reduce the assessed level of control risk and decrease the number of auditing hours. Thus, the supply-side story implies lower audit fees. In general, the empirical findings in previous studies provide more support for the demand-side argumentation, as audit fees seem to increase with the effectiveness of the audit committee. Carcello et al. (2002), using data from 1992-1993, find that board characteristics, rather than audit committee characteristics, are associated with audit fees. However, as noted by Abbott et al. (2003), the findings of Carcello et al. (2002) are affected by the fact that boards were the main corporate governance mechanisms in the early 1990s. Since then, the regulatory environment has emphasized the audit committees
2

It should be noted that the assumption about the positive relationship between audit committees size and

effectiveness is inconsistent with the literature on the effectiveness of the board of directors (see e.g., Yermack, 1996). Considerable empirical evidence suggests that smaller boards are more effective.

role in financial monitoring (BRC 1999; Sarbanes-Oxley Act 2002; SEC 2003). Accordingly, using more recent data, Abbott et al. (2003), Lee and Mande (2005), Goodwin-Stewart and Kent (2006), and Vafeas and Waegelein (2007) find a positive association between audit committee characteristics and audit fees. Abbott et al. (2003), Lee and Mande (2005) and Vafeas and Waegelein (2007) document that audit fees increase with audit committees independence and financial expertise. Moreover, Lee and Mande (2005) and Goodwin-Stewart and Kent (2006) find a positive relationship between audit fees and audit committees meeting activity, while Vafeas and Waegelein (2007) report that audit fees increase with the size of the committee. Finally, Goodwin-Stewart and Kent (2006) show that the mere existence of the audit committee tends to increase audit fees.

Gender Diversity and Corporate Governance

It is widely acknowledged in the psychology and management literature that gender differences exist e.g. in cognitive functioning, communicative skills, decision- making, and leadership style. Considerable evidence suggests that women have better communicative capabilities, and may thereby have a comparative advantage over men in tasks that require communication within and among different groups (see e.g., Wood et al. 1985; Maznevski 1994; Fondas 1997; Schubert 2006). Because women communicate and listen more effectively, they tend to perform better than men on group problem-solving and decisionmaking tasks requiring discussion and consensus (Wood et al. 1985, Robinson and Dechant 1997; Dallas 2002). Moreover, gender diversity in decision-making groups obviously provides greater diversity of opinions and interests, which in turn may lead to better

decisions (see e.g., Hillman et al 2007). The alternative view is that social diversity (gender, age, ethnic groups) may lead to reduced within-group communication and increase disagreements and conflicts (see e.g., Williams and OReilly, 1998; Jehn et al., 1999), thereby reducing the effectiveness of decision-making groups. Thus, the recent corporate governance literature has recognized that gender diversity may affect the functioning and efficiency of corporate boards and committees (see e.g., Fondas and Sassalos 2000; Erhardt et al. 2003; Huse and Solberg 2006; Rose 2007; Adams and Ferreira 2009; Huse et al. 2009; Nielsen and Huse 2010). In general, prior studies indicate that female representation may enhance the effectiveness and monitoring activities of the board. Besides the potential effects on group dynamics, at least two additional features of gender differences merit discussion in corporate governance context. First, as noted by Fondas and Sassalos (2000), women tend to have higher expectations regarding their responsibilities as directors, which may induce them to expend more effort on their tasks. In a similar vein, Huse and Solberg (2006) show that women in corporate boards are better prepared for board meetings than men, and female representation may thus improve board behavior and effectiveness. Further, given the glass ceiling phenomenon, women have to demonstrate extra competence in order to reach managerial positions and corporate boards (Eagly and Carli, 2003). Consequently, gender diversity may improve the efficiency of corporate boards and committees simply because the female representatives, in general, are presumably highly competent and hard-working. On a related note, recent empirical studies by Green et al. (2009) and Kumar (2010) indicate that female financial analysts are generally more competent than their male counterparts. Kumar (2010) argues that female

financial analysts possess better-than-average skills due to gender discrimination in the labor market. Second, a vast body of literature suggests that women are more conservative and risk averse than men (see e.g., Levin et al. 1988; Johnson and Powell 1994; Powell and Ansic 1997; Jianakoplos and Bernasek 1998; Byrnes et al. 1999). According to Schubert (2006) and Watson and McNaughton (2007), women try to avoid losses and are less inclined to take extreme risks. This suggests that gender diversity in corporate boards and committees may help to prevent excessive risk-taking. Although willingness to bear risk is a managerial necessity for creating shareholder value, gender differences in risk aversion may be good news from the perspective of corporate governance, as female representation may, for instance, help to better perceive potential biases in strategy formulation and risk assessments. Thus, it may be argued that gender diversity in the board of directors and audit committee should enhance monitoring and oversight. The conservatism and risk-aversion of females may also have implications for the integrity of the financial reporting process.3 Empirical support for this view is provided by Krishnan and Parsons (2008) who document that earnings quality is positively associated with gender diversity in managerial positions. Although the issue of gender diversity has so far been largely neglected in the accounting and auditing literature, female board representation has received considerable attention in the corporate finance literature over the past few years. In brief, these empirical studies indicate that gender diversity may have positive effects on corporate governance,

Anecdotal evidence from the recent high profile accounting scandals indicates that females are often acting

as the whistleblowers (Sherron Watkins at Enron and Cynthia Cooper at WorldCom).

10

financial performance and firm value.4 Carter et al. (2003) use a sample of Fortune 1000 firms to examine whether gender diversity on the board of directors affects firm value, and find a significant positive association between female board representation and firm value. Rose (2007), however, using data on Danish firms, finds no evidence of any systematic relationships between firm value and gender diversity. Erhardt et al. (2003) examine the effects of female board representation on financial performance, and document that firms with more diverse boards have higher profitability. Similar findings on a positive association between gender diversity and financial performance are also reported in Farrell and Hersch (2005). Finally, Adams and Ferreira (2009), Huse et al. (2009) and Nielsen and Huse (2010) focus on the effects of female directors on the boards monitoring and control tasks. These studies indicate that gender diversity may be positively associated with the effectiveness and monitoring activities of the board.

Hypothesis

Given that audit committee characteristics are associated with audit fees (Abbott et al. 2003; Lee and Mande 2005; Goodwin-Stewart and Kent 2006; Vafeas and Waegelein 2007), and moreover that gender diversity presumably affects the functioning of corporate boards and committees (Carter et al. 2003; Erhardt et al. 2003; Huse and Solberg 2006; Adams and Ferreira 2009; Huse et al. 2009), it is of interest to examine whether female

Watson (2002) and Rose (2007), however, document that there is no significant link between gender

diversity and firm performance.

11

representation on audit committees affects audit fees. Thus, the hypothesis to be examined in this paper is:

H1: There exists a relation between female audit committee representation and audit fees.

The association between the audit committees gender diversity and audit fees, if it exists, may be positive or negative. Based on the recent corporate governance literature, we presume that female representation improves the functioning and enhances the monitoring activities of the audit committee. However, as discussed above, the relationship between the effectiveness of the audit committee and audit fees is rather complex. The demand-side argumentation typically suggests that audit fees increase with the effectiveness of the audit committee, while the supply-side arguments imply a negative relation between the committees effectiveness and audit fees. Moreover, it should be noted that the demandside and supply-side arguments are not mutually exclusive hypotheses, and both mechanisms are likely to affect audit fees at the same time. Thus, the question of whether and how female audit committee representation affects audit fees is ultimately addressed empirically.

DATA

The initial sample used in the empirical analysis consists of the S&P 500 firms, and our sample period covers the fiscal years 2006-2008. Following prior research, we exclude

12

financial institutions (SIC codes 6000-6900) from the sample due to unique aspects of their regulation. This leaves us with a sample of 941 firm-year observations, which are classified by industry in Table 1. These firms are audited by Big 4 auditors, except for eight observations where the auditor is BDO. It should be noted that under SOX requirements, all firms in our sample are required to maintain an audit committee. Under the rules of the major U.S. stock exchanges, audit committees must be composed of at least three independent, financially literate directors of whom at least one is a financial expert as defined by the SOX.5

(insert Table 1 about here)

Our data on audit fees are obtained from Audit Analytics database, and comprise the fees paid to external auditors for performing audit services. The data on female audit committee representation and other audit committee characteristics are collected mainly from Audit Analytics. Since these data were not available from Audit Analytics for some of the S&P 500 firms, we completed our sample by gathering the missing values from proxy statements. For each firm, we determine the gender and potential financial expertise of each audit committee member, and quantify the size and the meeting frequency of the audit
5

In this paper, we use the final SOX definition of financial expertise. We acknowledge that there has been

some controversy over the financial expertise definition. DeFond et al. (2005) examine how the market values financial expertise on audit committees under alternative definitions of expertise. They find a positive market reaction to the appointment of accounting financial experts (i.e., the original SEC definition) and no reaction to the appointment non-accounting financial experts (i.e., the final SOX definition). Thus, it should be noted that our results may be affected by the broader definition of financial expertise.

13

committee. Committees financial expertise (CEXP), size (CSIZE), and meeting frequency (CMEET) are used as control variables in the analysis.6 Following previous studies on audit fees (see e.g., Simunic 1980; Francis and Wang 2005; Raghunandan and Rama 2006; Huang, Raghunandan and Rama 2009), we use several firm-specific control variables to account for the effects of client size, complexity, and audit risk.7 These firm-specific controls used in the analysis are: (i) the natural of log total assets (TA), (ii) inventory and receivable intensity (INVREC), (iii) the number of industries the firm is operating in (SQSIC), (iv) the proportion of foreign sales to total sales (FOREIGN), (v) the current ratio (LIQ), (vi) financial leverage (LEV), (vii) return on assets (ROA), (viii) whether the client firm has incurred loss during the fiscal year (LOSS), (ix) material weakness in internal controls (ICW), (x) restatements of financial statements (RESTATE), (xi) the ratio of nonaudit fees to audit fees (NAFRATIO), and (xii) an initialyear of the audit engagement (INITIAL). The data on these control variables are obtained from Thomson Financial Worldscope and Audit Analytics.

(insert Table 2 about here)

Table 2 presents descriptive statistics for the firm-specific and the audit committee related control variables. Panel A contains the statistics for the complete sample of 941

Our sample consists of the S&P 500 firms, and according to the SOX, the audit committees of these firms

must be composed of independent members. Thus, we are not using audit committees independence as a control variable in our empirical analysis.
7

Hay et al. (2006) provide a comprehensive discussion on the different attributes that affect audit fees.

14

firm-year observations. As can be noted from the table, the sample is quite heterogeneous in terms of client firm size and the level of audit fees. Panel A demonstrates that firm size, as measured by the amount of total assets, varies from about $785 million to about $798 billion, with a mean (median) of $11,053 million ($10,802 million). Similarly, the level of audit fees paid to external auditors appears to vary considerably, with a mean (median) fee of $5.341 million ($5.101 million), and a minimum and maximum of $0.493 and $94.300 million, respectively. Turning the focus to the audit committee controls, Panel A shows that an average audit committee holds about nine meetings per fiscal year, and has four members of whom two are typically financial experts. Panel B of Table 2 reports the descriptive statistics for three sub-samples of firms with female representation on the audit committee. Several interesting features can be noted from these statistics. First, Panel B indicates that firms with female audit committee chairs are quite similar to the complete sample of firms. Interestingly, however, it should be noted that firms with female audit committee chairs have no internal control problems, have a lower number of restatements, and have a larger fraction of financial experts in the audit committee. Panel B also shows that the financial expert to member ratio is higher in firms with female experts. This indicates that female experts are seldom the only financial experts on the audit committee. Finally, Panel B suggests that firms with female audit committee members are somewhat larger than other firms, and moreover, have slightly larger audit committees.

(insert Table 3 about here)

15

Table 3 provides pairwise correlations among the variables used in the analysis. As expected, our dependent variable AFEE is strongly positively correlated with TA, and also notably positively correlated with control variables FOREIGN, SQSIC, CMEET, and ICW, and negatively correlated with LIQ. Female representation variables exhibit low correlations with AFEE. Regarding our control variables, Table 3 shows that INVREC, LIQ and CSIZE are strongly correlated with TA, LEV with FOREIGN and LIQ, and LOSS with ROA. Finally, it can be noted from Table 3 that our female representation variables are highly positively correlated with each other.

METHODOLOGY

We examine the relationship between audit fees and female audit committee representation with cross-sectional regressions. Consistent with the literature on audit fees (see e.g., Simunic 1980; Francis and Wang 2005; Raghunandan and Rama 2006; Huang, Raghunandan and Rama 2009), our model includes several firm-specific variables to control for the effects of firm size, complexity, and audit risk on audit fees. Furthermore, given that recent studies by Abbott et al. (2003), Goodwin-Stewart and Kent (2006), and Vafeas and Waegelein (2007) document a relation between audit fees and audit committee characteristics, we include three audit committee related variables in our analysis as additional controls. We presume that this set of control variables will enable us to examine the potential incremental effect of female audit committee representation on audit fees. Thus, we estimate the following regression model to examine the association between audit fees and female representation on the audit committee:

16

AFEE i ,t
6

TAi ,t
7

INVREC i ,t
8

SQSIC i ,t
9

FOREIGN i ,t
10 14 CMEET i , t

LIQ i ,t
(1)

LEV i ,t CSIZE i ,t

ROAi ,t

LOSS i ,t

ICW i ,t

RESTATE i ,t

11 NAFRATIO i , t 15 16

12 INITIAL i ,t

13 CEXPi ,t i ,t

FEMALE i ,t

where AFEEi,t denotes the natural logarithm of audit fees for firm i at time t, TAi,t is the natural logarithm of total assets, INVRECi,t is inventory and receivables to total assets, SQSICi,t is the square root of two-digit SIC-codes the firm is operating in, FOREIGNi,t is the proportion of foreign sales to total sales, LIQi,t is the current ratio, LEVi,t is total debt to total assets, ROAi,t is the return on assets, LOSSi,t is a dummy variable that equals one if the firm has incurred loss during the fiscal year, ICWi,t is a dummy that equals one if there is a material weakness in internal controls, RESTATEi,t is a dummy that equals one if there is a restatement of financial statement, NAFRATIOi,t is the ratio of nonaudit fees to audit fees, INITIALi,t is a dummy for initial audit engagement years, CEXPi,t is the proportion of financial experts to committee members, CMEETi,t is the number of audit committee meetings, and CSIZEi,t is the number of audit committee members. The test variable of interest in our regression specification is FEMALEi,t, which is defined as one of the following six binary variables: FCHAIR equals one if the chair of the committee is a female, FEXPERT is one if the committee has at least one female expert, FMEMB equals one if the committee has at least one female member, FCE is set to one if the committee has a female chair and at least one female expert, FCM is one if the committee has a female chair and at least one additional female member, and FTOM is one

17

if the committee has at least two female members. These female representation variables are used in the regressions one at a time in order to avoid problems with multicollinearity. Based on prior research (see e.g., Francis and Simon 1987; Simon and Francis 1988; Carcello et al. 2002; Abbott et al. 2003; Knechel and Willekens 2006; Raghunandan and Rama 2006; Mitra et al. 2007; Huang et al. 2009), we expect audit fees to be positively associated with the natural of log total assets (TA), inventory and receivable intensity (INVREC), the number of segments the firm is operating in (SQSIC), the proportion of foreign sales to total sales (FOREIGN), financial leverage (LEV), whether the client firm has incurred loss during the fiscal year (LOSS), material weakness in internal controls (ICW), and restatements of financial statements (RESTATE), and negatively associated with the current ratio (LIQ), return on assets (ROA), the ratio of nonaudit fees to audit fees (NAFRATIO), and an initial-year of the audit engagement (INITIAL). Moreover, consistent with the recent findings of Abbott et al. (2003), Lee and Mande (2005), Goodwin-Stewart and Kent (2006) and Vafeas and Waegelein (2007), we expect audit fees to increase with the size, expertise, and meeting frequency of the audit committee. As discussed in Abbot et al. (2003), the positive relationship between audit committee characteristics and audit fees relies on demand-side arguments. It should be noted, however, that these audit committee characteristics may also enhance the effectiveness of internal monitoring, thereby reducing the perceived audit risk, which then in turn may decrease the fees paid to external auditors.8

In a recent paper, Abbott et al. (2008) argue that increased audit committees oversight may lead to

improvements in internal audit function, which in turn, may increase the external auditors reliance on the clients corporate governance. Using data on the Fortune 1000 firms, Abbott et al. (2008) document that the external auditors increased reliance on internal auditing reduces audit fees. Moreover, according to Cohen et

18

RESULTS

Regression Results

Table 4 reports the estimation results of the alternative regression models used to test the relation between audit fees and female representation in the audit committee. We begin our analysis with Model 1 by regressing audit fees on firm-specific control variables and the controls for audit committee characteristics. As can be seen from the table, the control variables have the expected signs and are mostly statistically significant. Moreover, our control variables seem to have good explanatory power for audit fees, as the adjusted R2 of the regression is 69.6 % and the F-statistic is significant at the 1 % level.9 Consistent with Abbott et al. (2003) and Vafeas and Waegelein (2007), the coefficient estimates for the control variables related to audit committee characteristics are positive, suggesting that companies with large audit committees and frequent meetings have higher audit fees.

(Insert Table 4 about here)

In Models 2-7, we examine the relationship between female representation and audit fees by regressing audit fees on a set of dummy variables for female representation and the

al. (2010), auditors consider active and diligent audit committees to improve corporate governance, and that these improvements affect the planning and performance of the audit.
9

It should be noted that high R2s are typical in this type of audit fee regressions (see e.g., Abbott et al. 2003;

Raghunandan and Rama 2006; Huang et al. 2009).

19

controls for firm specific factors and audit committee characteristics. The female representation dummy in Model 2 is FCHAIR, which equals one if the chair of the audit committee is female. The estimation results of Model 2 indicate that audit fees are lower for companies with female audit committee chairs, as the coefficient estimate for FCHAIR is negative and statistically significant at the 1 % level. The magnitude of the coefficient suggests that the logarithm of audit fees is about 0.136 units lower for firms with female audit committee chairs, corresponding to about $0.679 million for the average firm in our sample. The explanatory power of Model 2 is relatively high, with an adjusted R2 of 69.8 %. Finally, the incremental F-statistic (not reported) associated with testing the restricted Model 1 against Model 2 is 8.31 and significant at the 1 % level, thereby suggesting that female audit committee representation helps to explain the cross-sectional variation in audit fees. Models 3 and 4 in Table 4 regress audit fees on dummy variables for female experts (FEXPERT) and female members (FMEMB), respectively. As can be noted from the table, the coefficients for both FEXPERT and FMEMB are statistically insignificant. The adjusted R2s of these models are 69.6 % and 69.5 %, respectively. In Model 5, we use a dummy variable that is set to one if the audit committee has a female chair and at least one female expert (FCE). Consistent with Model 2, the estimate for FCE is negative and significant at the 5 % level, and the F-statistic and adjusted R2 indicate good fit of the specification. Moreover, the incremental F-statistic (not reported) testing Model 1 against Model 5 is statistically significant.

20

The female representation variable used in Model 6 (FCM) identifies firms that have female chairs and at least one additional female member on their audit committee. Again, the estimation results suggest that female representation is negatively related to audit fees. The coefficient for FCM is negative and significant at the 5 % level. Finally, in Model 7 we use a female dummy that equals one if the audit committee has at least two female members (FTOM). Our estimations indicate that FTOM is negatively and statistically significantly related to the level of audit fees, and thus provides further evidence to suggest that companies with female audit committee representation pay lower audit fees. The Fstatistic of Model 7 is significant at the 1 % level, and the adjusted R2 is 69.6 %.

Alternative Interpretations of the Results

Overall, our results suggest that female audit committee representation is negatively related to audit fees. The coefficient estimates for the female representation variables are negative and statistically significant in most of our regression specifications, and the coefficient for FCHAIR is always negative and statistically highly significant. This finding obviously merits further discussion. Are firms with female chairs and audit committee members associated with lower audit fees because gender diverse audit committees are more effective and better monitors and need less support from the external auditor? Or alternatively, are the fees lower because gender diverse audit committees are less effective in insisting on increased external audit coverage and enhanced audit quality? From the demand-side perspective, female representation may reduce audit fees by affecting the audit committees assessment of the required level of external audit work.

21

Because females tend to expend more time and effort to gather information and are generally better prepared for meetings and have better attendance records (see e.g., Huse and Solberg 2006; Adams and Ferreira, 2009), the women on audit committees may be less dependent than the men on the assurance provided by external auditors. In a similar vein, the diligence and effort of women may be expected to inherently reduce the need for support from the external auditors. An alternative demand-side interpretation for the lower audit fees is that gender diversity may decrease the effectiveness of the audit committee. Williams and OReilly (1998) and Jehn et al. (1999) argue that social diversity may reduce communication and increase disagreements and conflicts among group members, and thereby reduce the effectiveness of decision-making groups. Consistent with previous studies (Abbott et al. 2003; Lee and Mande 2005; Goodwin-Stewart and Kent 2006; Vafeas and Waegelein 2007), we find that other measures of audit committee characteristics (size and activity level) are positively associated with audit fees. Given that the existing literature has assumed that the committees effectiveness increases with its size, financial expertise and meeting activity, our findings would indicate that audit committees with female chairs are less effective in insisting on increased audit coverage and enhanced audit quality. Nevertheless, considering that prior literature has documented gender diversity to enhance the functioning of corporate boards, we are more inclined to assume that female representation improves the effectiveness and diligence of the audit committee. Considering the supply-side perspective, female representation on audit committees may decrease audit fees by reducing the assessment of audit risk. Female chairs and audit

22

committee members could affect the auditors assessment of audit risk, for instance, by improving the effectiveness of internal auditing and/or by generally enhancing the integrity of the financial reporting process.10 It should be noted that chairpersons of the audit committee are often responsible for the communication between the audit committee and the external as well as the internal auditors. Based on the findings of Adams and Ferreira (2009) and Huse et al. (2009), it is reasonable to expect that gender diversity enhances the monitoring activities of the audit committee. The potential improvements in the audit committees diligence and oversight may reduce audit fees by affecting the planning and performance of the audit (see e.g., Abbott et al. 2008; Cohen et al. 2010). Moreover, the psychology and management literature suggests that women tend to be more conservative and risk averse than men (see e.g., Powell and Ansic 1997; Jianakoplos and Bernasek 1998; Byrnes et al. 1999). These gender differences in risk aversion may also directly and indirectly affect the auditors judgment on audit risk. Finally, given the better communicative capabilities and meetings preparations of women, female chairs and audit committee members may reduce audit fees simply by reducing the hours spent by the auditor (Goodwin-Stewart and Kent 2006; Stewart and Munro 2007).

Endogeneity Concerns

So far, we have documented that firms with female audit committee representation are associated with lower audit fees. We acknowledge that it is implausible that the mere

10

Krishnan and Parsons (2008) document that gender diversity in the top managerial positions improves the

23

presence of females in the audit committee, per se, would affect the pricing decisions of the external auditor. Instead, our findings essentially suggest that female audit committee representation may improve the effectiveness and monitoring activities of the audit committee or some other, potentially unobservable, corporate governance attribute of the firm. Nevertheless, it is important to recognize that our empirical analysis may potentially suffer from endogeneity problems. It is possible that certain firm characteristics simultaneously affect the appointment of females to the audit committee and the level of audit fees. Moreover, our results may also be affected by a self-selection bias if female directors tend to self-select into less risky companies with lower audit fees due to gender differences in risk tolerance. In order to mitigate these endogeneity issues, we utilize a matched-firm approach and we also analyze changes in audit fees after female chairs are appointed to the audit committee. Endogeneity problems could also be addressed with twostage instrumental variable techniques. This approach would require a suitable instrumental variable that is strongly correlated with the female representation variables and uncorrelated with audit fees. Unfortunately, we are unable to use instrumental variable techniques due to a lack of valid instruments for the female representation variables.11 We first attempt to moderate endogeneity concerns by re-estimating all regression models using matched-firm samples. For this purpose, each sample firm with female audit committee representation is matched with a firm without corresponding female audit committee representation based on firm size and riskiness. The estimation results of these

quality of financial reporting.

11

The findings of Larcker and Rusticus (2010) indicate that the use of weak instrumental variables may lead

to biased estimates and wrong statistical inference.

24

matched-firm regressions are reported in Table 5. Overall, the estimates based on the matched-firm samples are broadly consistent with the results reported in Table 4. Most importantly, the coefficients for FCHAIR and FTOM are negative and statistically significant, indicating that audit fees are lower for firms that have female audit committee chairs or at least two female members on the audit committee. However, in contrast to Table 4, the negative coefficients for FCE and FCM appear statistically insignificant in the matched-firm regressions.

(Insert Table 5 about here)

If female representation on audit committees reduces audit fees, we should ultimately observe a decrease in audit fees after females members are appointed to the audit committee. Hence, we next examine the relationship between female audit committee representation and audit fees by regressing changes in audit fees on changes in the control variables and a dummy variable that equals one if the chair of the audit committee changes from a male to a female.12 The coefficient estimate (not tabulated) for the gender change dummy appears statistically insignificant in this regression specification, thereby suggesting that the appointment of a female chair does not affect audit fees, at least not during the appointment year. This may indicate that female representation on the audit committee is endogenously determined. Nevertheless, it should be noted that our three-year
12

We focus on chairperson changes because our findings are mostly pertained to firms with female audit

committee chairs. Moreover, audit committee chair is typically the person who communicates with the external as well as internal auditors.

25

sample period includes only 14 observations of male chairs being replaced by a female chair, and thus the estimates are obviously plagued by the lack of statistical power. Moreover, it is also necessary to recognize that governance structures change slowly and therefore the potential implications of female appointments may be seen with a significant lag, rather than during the year of the appointment. Overall, the regressions based on matched-firm samples and changes in audit fees provide somewhat mixed evidence on the effects of female audit committee representation on audit fees. Although the matched-firm regressions indicate that firms with female audit committee chairs have significantly lower audit fees than size- and risk-matched firms with male chairs, the results of the change regression do not rule out the possibility that female representation on the audit committee is endogenously determined. Consequently, causal interpretations of our empirical findings should be made with caution.

Robustness Checks

We examine the robustness of our findings by conducting several additional tests. First, in order to ascertain that the results are not affected by extreme observations or few outliers, we winsorize audit fees and the firm-specific control variables at the 0.5 % and 99.5 % levels and then re-estimate all regression models. In general, the estimation results with the winsorized data are consistent with those reported in Table 4. Most importantly, the coefficients for the female representation variables have the same signs and significance levels. Moreover, the explanatory power of the estimated models is equally high, with adjusted R2s ranging from 69.8 % to 70.0 %. Thus, the regressions based on winsorized

26

data further confirm that female audit committee representation is associated with lower audit fees, and suggest that this finding is not driven by outliers. To examine the sensitivity of our results to the definition of the test variables, we introduce three alternative female representation variables: (i) percentage of female members on the audit committee (PFMEMB), (ii) percentage of female experts on the audit committee (PFEXPERT), and (iii) a dummy variable for female chair with financial expertise (FCHAIREXP). Our estimations with these alternative test variables (not tabulated) once again suggest that female audit committee representation is negatively related to audit fees, as the estimated coefficients for PFMEMB and FCHAIREXP are negative and statistically significant. Furthermore, due to somewhat lower correlations of these new variables, we are able to estimate alternative regressions that simultaneously include two female representation variables. In particular, we estimate the following three specifications: (i) FCHAIR and PFMEMB, (ii) FCHAIREXP and PFMEMB, and (iii) FCHAIREXP and PFEXPERT. In these regressions (not tabulated), only the coefficient for FCHAIR appears statistically significant, thereby suggesting that our findings are mostly pertained to firms with female audit committee chairs. Given that manufacturing firms constitute a major part of our sample (see Table 1), we next examine whether our empirical findings are induced by industry-related factors. In particular, we re-estimate all regressions with industry dummies. These estimates (not tabulated) are generally similar to the results reported in Table 4, and thus indicate that our findings are not affected by industry-specific factors. Considering the variables of interest,

27

the estimated coefficients for FCHAIR, FCE, and FCM are negative and statistically highly significant, while the coefficient for FTOM appears insignificant. The results reported in Tables 4 and 5 are based on Whites heteroskedasticity consistent standard errors. To examine whether our results depend on the standard error estimates, we use three alternative estimation methods: (i) the Newey-West

heteroskedasticity and autocorrelation consistent standard errors, (ii) standard errors clustered by firm, and (iii) standard errors clustered by year. The regression results based on these alternative standard error specifications are broadly consistent with our main findings. Regardless of the estimation technique, the coefficient for FCHAIR is always negative and statistically significant at least at the 5 % level, thereby providing considerable evidence to suggest that firms with female audit committee chairs are associated with lower audit fees. The estimates based on the Newey-West standard errors are similar to the results reported in Table 4, except for the insignificant coefficient for FTOM. When the standard errors are clustered by firm, the coefficients for FCHAIR, FCE, FCM, and FTOM are negative and statistically highly significant, but inconsistent with our main findings, the coefficient for FEXPERT appears positive and significant. Finally, the coefficient estimates for FCHAIR and FCM are negative and statistically significant when the standard errors are clustered by year, while the coefficients for the other female representation dummies are insignificant. As can be noted from Table 2, the data we have obtained from Audit Analytics contains some observations with zero audit committee meetings and zero financial experts. We randomly checked some of the proxy statements of these firms, and observed that the

28

firms either do not have audit committee meetings and/or financial experts or do not report them. To control for potential data biases, we re-estimate our main models (i) by excluding all firms with zero meetings or financial experts, and (ii) by excluding the controls for financial experts and audit committee meetings. The results of these regressions are consistent with our main findings. The estimated coefficients for FCHAIR, FCM and FCE are statistically highly significant, while the coefficients for FEXPERT, FMEMB and FTOM are insignificant. In order to further explore whether female representation is associated with improved monitoring and lower audit risk, we next take into account the strict requirements set for the firms listed in the New York Stock Exchange (NYSE). Specifically, this analysis is motivated by the NYSE rules which emphasize the audit committees role in risk monitoring and internal control activities. Therefore, we divide the sample into NYSE (739 firm-year observations) and non-NYSE (202 firm-year observations) firms, and re-estimate all regressions. The estimates for the sample of the NYSE firms (not tabulated) are generally similar to the results reported in Table 4. Again, the estimated coefficients for FCHAIR, FCE, FCM, and FTOM are negative and statistically highly significant. However, in the subsample of the non-NYSE firms, only the coefficient for FCM remains statistically significant at conventional levels. Overall, these estimates indicate that our results are much stronger for the NYSE-listed firms. Given the emphasized role of the audit committee in risk monitoring and internal control activities under the NYSE rules, our findings may indicate that female audit committee representatives, and female chairs in particular, may enhance monitoring and reduce audit risk.

29

Limitations

In interpreting the findings, it is necessary to consider several limitations of our empirical analysis. First, it should be noted that this paper is the first attempt to assess the role of gender diversity on audit committees. Our general theoretical background is drawn from psychology, management, and corporate finance literature. Thus, our findings have to be regarded as somewhat exploratory. Second, our sample consists of the S&P 500 firms, and is thereby limited to very large publicly traded U.S. firms. This sample is not necessarily representative of the entire market, and thus it is uncertain to which extent our findings apply in other settings. Moreover, it should be noted that our sample is limited to three fiscal years. Finally, our analysis may potentially suffer from a self-selection bias. We have attempted to control for the firm and audit committee characteristics that previous literature has documented to affect audit fees. It is nevertheless possible that we have omitted some correlated variables, or that certain firm characteristics simultaneously affect the choice of females in an audit committee and the level of audit fees. As discussed above, we were unable to rule out the possibility that female representation on the audit committee is endogenously determined, and therefore any causal interpretations of our empirical findings should be made with caution.

CONCLUSIONS

This paper examines the relationship between audit fees and female representation on corporate audit committees. Motivated by the gender differences in human behavior, and

30

recent findings in the corporate governance literature, we hypothesize that gender diversity on audit committees affects the level of fees paid to the external auditors. Given that recent legal reforms set strict requirements for the composition of the audit committee and emphasize the committees monitoring responsibilities in ensuring the integrity of financial reporting process (Sarbanes-Oxley Act 2002; SEC 2003), it is important to consider the potential effects of gender diversity on the functioning and effectiveness of audit committees. Moreover, by focusing on the association between female audit committee representation and audit fees, this paper adds novel insights into audit pricing literature. Our empirical findings demonstrate that female representation on audit committees may affect audit fees. In particular, using a sample of S&P 500 firms, we find considerable evidence to suggest that firms with female audit committee chairs have significantly lower audit fees. From the audit demand perspective, our findings may indicate that female audit committee representatives and especially female chairs reduce the need for assurance provided by external auditors. Alternatively, from the supply-side perspective, female chairs may decrease audit fees by affecting the auditors assessment of audit risk, for instance, by improving the effectiveness of internal control activities and the communication with the external and internal auditors, or by generally enhancing the integrity of the financial reporting process. If female representation reduces the inherent risk of misstatements, gender diversity on audit committees may lead to lower audit fees. Overall, our results suggest that gender diversity may lead to improvements in the effectiveness and monitoring activities of the audit committee.

31

The analysis presented in this paper is somewhat exploratory in nature. Therefore, our empirical findings should be viewed as suggestive until confirmed by further research. Our results suggest a number of potential avenues for future research. First, an obvious extension would be to determine whether these findings hold true for smaller firms. Future research could also explore the association between audit committee gender diversity and the quality of financial reporting. Our findings could be extended, for instance, by examining whether firms with female audit committee representation have lower discretionary accruals, fewer restatements, or lower cost of capital. Future research addressing these issues could provide valuable information about the role of gender diversity on corporate audit committees.

32

REFERENCES

Abbott, L. and S. Parker (2000), Audit committee characteristics and auditor selection, Auditing: A Journal of Practice & Theory, Vol.19, pp. 4766. Abbott, L., S. Parker, G. Peters and K. Raghunandan (2003), The association between audit committee characteristics and audit fees, Auditing: A Journal of Practice & Theory, Vol.22, pp. 1732. Abbott, L., S. Parker and G. Peters (2008), The internal audit functions organizational position and its impact on the external auditors reliance decision, Proceedings of the 2008 Annual Meeting of the American Accounting Association. Adams, R. and D. Ferreira (2009), Women in the boardroom and their impact on governance and performance, Journal of Financial Economics, Vol.94, pp. 291309. Blue Ribbon Committee, BRC (1999), Blue Ribbon Committee on Improving the Effectiveness of Corporate Audit Committees, New York Stock Exchange, New York. Byrnes, J., D. Miller and W. Schafer (1999), Gender differences in risk taking: A metaanalysis, Psychological Bulletin, Vol.125, pp. 367383. Carcello, J. and T. Neal (2000), Audit committee composition and auditor reporting, The Accounting Review, Vol.75, pp. 453467. Carcello, J., D. Hermanson, T. Neal and R. Riley (2002), Board characteristics and audit fees, Contemporary Accounting Research, Vol.19, pp. 365384. Carter, D., B. Simkins and W. Simpson (2003), Corporate governance, board diversity, and firm value, The Financial Review, Vol.38, pp. 3353. Chen, M., R. Moroney and K. Houghton (2005), Audit committee composition and the use of an industry specialist audit firm, Accounting and Finance, Vol.45, pp. 217239. Cohen, J., G. Krishnamoorthy and A. Wright (2010), Corporate governance in the post Sarbanes-Oxley era: auditors experiences, Contemporary Accounting Research, Vol.27, 751786. Dallas, L. (2002), The new managerialism and diversity on corporate boards of directors, Tulane Law Review, Vol.76, pp. 13631405.

33

DeFond, M., R. Hann and X. Hu (2005), Does the market value financial expertise on audit committees of boards of directors?, Journal of Accounting Research,Vol.43, pp. 153193. DeZoort, T. and S. Salterio (2001), The effects of corporate governance experience and financial reporting and audit knowledge on audit committee members judgments, Auditing: A Journal of Practice and Theory, Vol.20, pp. 3147. DeZoort, T., D. Hermanson and R. Houston (2003), Audit committee support for auditors: the effects of materiality justification and accounting precision, Journal of Accounting and Public Policy, Vol.22, pp. 175199. Eagly, A. and L. Carli (2003), The female leadership advantage: an evaluation of the evidence, The Leadership Quarterly, Vol.14, pp. 807834. Erhardt, N., J. Werbel and C. Shrader (2003), Board of director diversity and firm financial performance, Corporate Governance: An International Review, Vol.11, pp. 102110. Farrell, K. and P. Hersch (2005), Additions to corporate boards: the effect of gender, Journal of Corporate Finance, Vol.11, 85106. Fondas, N. (1997), Feminization unveiled: management qualities in contemporary writings, Academy of Management Review, Vol.22, pp. 257282. Fondas, N. and S. Sassalos (2000), A different voice in the boardroom: how the presence of women directors affects board influence over management, Global Focus, Vol.12, pp. 1322. Francis, J. and D. Simon (1987), A test of audit firm pricing in the small client segment of the U.S. audit market, The Accounting Review, Vol.62, pp. 145167. Francis, J. and D. Wang (2005), Impact of the SEC's public fee disclosure requirement on subsequent period fees and implications for market efficiency, Auditing: A Journal of Practice and Theory, Vol.24, pp. 145160. Goodwin-Stewart, J. and P. Kent (2006), Relation between external audit fees, audit committee characteristics and internal audit, Accounting and Finance, Vol.46, pp. 387404.

34

Green, C., N. Jegadeesh and Y. Tang (2009), Gender and job performance: Evidence from Wall Street, Financial Analysts Journal, Vol.65, pp. 6578. Hay, D., R. Knechel and N. Wong (2006), Audit fees: a meta-analysis of the effect of supply and demand attributes, Contemporary Accounting Research, Vol.23, pp.141 191. Heckman, J. (1978), Dummy endogenous variables in a simultaneous equation system Econometrica, Vol.46, pp. 931959. Hillman, A., C. Shropshire and A. Cannella (2007), Organizational predictors of women on corporate boards, Academy of Management Journal, Vol.50, pp.941952. Huang, H-W., K. Raghunandan and D. Rama (2009), Audit fees for initial audit engagements before and after SOX, Auditing: A Journal of Practice & Theory, Vol. 28, pp. 171190. Huse, M. and A. Solberg (2006), Gender-related boardroom dynamics: how Scandinavian women make and can make contributions on corporate boards, Women in Management Review, Vol.21, pp. 113130. Huse, M., S. Nielsen and I.M. Hagen (2009), Women and employee-elected board members, and their contributions to board control tasks, Journal of Business Ethics, Vol.89, pp. 581597. Jehn, K., G. Northcraft and M. Neale (1999), Why differences make a difference: a field study of diversity, conflict and performance in workgroups, Administrative Science Quarterly, Vol.44, pp. 741763. Jianakoplos, N. and A. Bernasek (1998), Are women more risk averse?, Economic Inquiry, Vol.36, pp. 620630. Johnson, J. and P. Powell (1994), Decision making, risk and gender: are managers different?, British Journal of Management, Vol.5, pp. 123138. Knapp, M. (1987), An empirical study of audit committee support for auditors involved in technical disputes with client management, The Accounting Review, Vol.62, pp. 578 588.

35

Knechel, R. and M. Willekens (2006), The role of risk management and governance in determining audit demand, Journal of Business Finance and Accounting, Vol.33, pp. 13441367. Krishnan, G. and L. Parsons (2008), Getting to the bottom line: An exploration of gender and earnings quality, Journal of Business Ethics, Vol.78, pp. 6576. Kumar, A. (2010), Self-selection and the forecasting abilities of female equity analysts, Journal of Accounting Research, Vol.48, 393435. Larcker, D. and T. Rusticus (2010), On the use of instrumental variables in accounting research, Journal of Accounting and Economics, Vol.49, pp. 186205. Lee, H., V. Mande and R. Ortman (2004), The effect of audit committee and board of director independence on auditor resignation, Auditing: A Journal of Practice & Theory, Vol.23, pp. 131146. Lee, H. and V. Mande (2005), The relationship of audit committee characteristics with endogenously determined audit and non-audit fees, Quarterly Journal of Business & Economics, Vol.44, pp. 93112. Levin, I., M. Snyder and D. Chapman (1988), The interaction of experimental and situational factors and gender in a simulated risky decision-making task, Journal of Psychology, Vol.122, pp. 173181. Maznevski, M. (1994), Understanding our differences: performance in decision-making groups with diverse members, Human Relations, Vol.47, pp. 531552. Mitra, S., M. Hossain and D. Deis (2007), The empirical relationship between ownership characteristics and audit fees, Review of Quantitative Finance and Accounting, Vol.28, pp. 257285. Nielsen, S., and M. Huse (2010), The contribution of women on boards of directors: going beyond the surface, Corporate Governance: An International Review, Vol.18, 136 148. Powell, M. and D. Ansic (1997), Gender differences in risk behaviour in financial decision making: an experimental analysis, Journal of Economic Psychology, Vol.18, pp. 605 627.

36

Raghunandan, K. and D. Rama (2006), SOX section 404 material weakness disclosures and audit fees, Auditing: A Journal of Practice and Theory, Vol.25, pp 99114. Robinson, G. and K. Dechant (1997), Building a business case for diversity, Academy of Management Executive, Vol.11, pp. 2131. Rose, C. (2007), Does female board representation influence firm performance? The Danish evidence, Corporate Governance: An International Review, Vol.15, pp. 404 413. Sarbanes-Oxley Act (2002), Public Law No: 107-204, Washington D.C.: Government Printing Office. Schubert, R. (2006), Analyzing and managing risks on the importance of gender difference in risk attitudes, Managerial Finance, Vol. 32, pp. 706715. Securities and Exchange Commission, SEC (2003), Strengthening the Commission's Requirements Regarding Auditor Independence, Financial Reporting Release No. 68. Washington, D.C.: SEC. Simon, D. and J. Francis (1988), The effects of auditor changes on audit fees: tests of price cutting and price recovery, The Accounting Review, Vol.63, pp. 255269. Simunic, D. (1980), The pricing of audit services: theory and evidence, Journal of Accounting Research, Vol.22, pp. 161190. Stewart, J. and L. Munro (2007), The impact of audit committee existence and audit committee meeting frequency on the external audit: perceptions of Australian auditors, International Journal of Auditing, Vol.11, pp. 5169. Turley, S. and M. Zaman (2004), The corporate governance effects of audit committees, Journal of Management and Governance, Vol. 8, pp. 305332. Vafeas, N. and J. Waegelein (2007), The association between audit committees, compensation incentives, and corporate audit fees, Review of Quantitative Finance and Accounting, Vol. 28, pp. 241255. Watson, J. (2002), Comparing the performance of male- and female-controlled businesses: relating outputs to inputs, Entrepreneurship Theory and Practice, Vol.26, pp. 91100.

37

Watson, J. and M. McNaughton (2007), Gender differences in risk aversion and expected retirement benefits, Financial Analysts Journal, Vol.63, pp. 5262. Williams, K. and C. OReilly (1998), Demography and diversity in organizations, in B. Staw and R. Sutton (eds.), Research in Organizational Behavior (JAI Press), pp. 77 140. Wood, W., D. Polek and C. Aiken (1985), Sex differences in group task performance, Journal of Personality and Social Psychology, Vol.48, pp. 6371. Yermack, D. (1996), High market valuation of companies with a small board of directors, Journal of Financial Economics, Vol. 40, pp. 185211.

38

Table 1. Number of firms by SIC codes.

SIC Code 1000-1400 1500-1700 2000-3900 4000-4900 5000-5100 5200-5900 7000-8900

Industry Description Mining Construction Manufacturing Transportation, communications, electric, gas, and sanitary services Wholesale trade Retail trade Services Totals

Firms with Firms with Firms in female AC female AC industry chair experts 47 6 14 13 2 3 514 59 127 134 25 95 113 941 9 2 13 11 102 30 10 28 33 245

Firms with female AC members 18 9 292 77 20 56 55 527

The table presents the number of firms by standard industry classification (SIC) codes. The sample consists of 941 firm-year observations of the S&P 500 firms. Financial institutions (SIC codes between 6000 and 6900) are excluded from the sample.

39

Table 2. Descriptive statistics.

Panel A. Summary statistics for all firms (n=941) Variable Mean Median Audit fees (in $ millions) 5.341 5.101 Total assets (in $ millions) 11,053.3 10,802.0 Inventory and receivables to total assets 0.229 0.210 SQ of 2-dig sic codes 1.861 1.732 Foreign sales to total sales 0.311 0.309 Current ratio 1.756 1.425 Leverage 0.247 0.237 Return on assets 8.037 8.247 Loss 0.087 0.000 Material weakness in internal controls 0.022 0.000 Restatement 0.033 0.000 Nonaudit fee to audit fee 0.217 0.173 Initial year audit 0.036 0.000 Committee expertise 0.424 0.333 Committee meetings 9.250 9.000 Committee size 4.187 4.000

Std.dev. Max 0.860 94,300.7 1.132 797,701.2 0.145 0.855 0.502 2.828 0.260 1.383 1.169 10.454 0.168 1.420 9.281 51.696 0.282 1.000 0.148 1.000 0.178 1.000 0.204 2.497 0.187 1.000 0.368 1.000 3.350 30.000 1.028 8.000

Min 0.493 784.8 0.013 1.000 0.000 0.090 0.000 -68.059 0.000 0.000 0.000 0.000 0.000 0.000 0.000 2.000

Panel B. Summary statistics for firms with female AC representation Variable Firms with female AC chair (n=102) Audit fees (in $ millions) 5.293 4.911 0.767 Total assets (in $ millions) 11,736.8 12,338.6 0.994 Inventory and receivables to total assets 0.228 0.191 0.143 SQ of 2-dig sic codes 1.871 2.000 0.637 Foreign sales to total sales 0.332 0.301 0.268 Current ratio 1.787 1.475 1.244 Leverage 0.257 0.253 0.153 Return on assets 5.489 8.064 12.821 Loss 0.118 0.000 0.324 Material weakness in internal controls 0.000 0.000 0.000 Restatement 0.020 0.000 0.139 Nonaudit fee to audit fee 0.209 0.161 0.203 Initial year audit 0.039 0.000 0.195 Committee expertise 0.581 0.500 0.356 Committee meetings 8.539 8.500 3.266 Committee size 4.176 4.000 0.883

23.700 58,834.8 0.659 2.828 0.878 8.765 0.686 31.451 1.000 0.000 1.000 1.386 1.000 1.000 16.000 7.000

0.998 1,369.9 0.026 1.000 0.000 0.275 0.000 -56.179 0.000 0.000 0.000 0.002 0.000 0.000 0.000 3.000

40

Table 2. Continued.
Variable Mean Median Std.dev. Max Min Firms with female AC experts (n=245) Audit fees (in $ millions) 5.862 5.565 0.858 51.085 0.725 Total assets (in $ millions) 11,458.5 10,388.9 1.129 202,296.4 1,369.9 Inventory and receivables to total assets 0.245 0.206 0.156 0.826 0.043 SQ of 2-dig sic codes 1.862 2.000 0.544 2.828 1.000 Foreign sales to total sales 0.317 0.282 0.276 0.953 0.000 Current ratio 1.767 1.446 1.180 9.995 0.275 Leverage 0.254 0.264 0.163 0.760 0.000 Return on assets 7.268 7.904 8.963 31.451 -29.939 Loss 0.118 0.000 0.324 1.000 0.000 Material weakness in internal controls 0.041 0.000 0.198 1.000 0.000 Restatement 0.037 0.000 0.188 1.000 0.000 Nonaudit fee to audit fee 0.218 0.173 0.187 1.386 0.000 Initial year audit 0.041 0.000 0.198 1.000 0.000 Committee expertise 0.672 0.667 0.315 1.000 0.167 Committee meetings 9.278 9.000 3.856 30.000 0.000 Committee size 4.249 4.000 0.936 7.000 3.000 Firms with female AC members (n=527) Audit fees (in $ millions) 5.717 5.498 0.837 51.085 0.493 Total assets (in $ millions) 12,650.9 12,500.0 1.134 276,368.2 784.8 Inventory and receivables to total assets 0.226 0.199 0.149 0.826 0.013 SQ of 2-dig sic codes 1.898 2.000 0.528 2.828 1.000 Foreign sales to total sales 0.292 0.274 0.264 1.383 0.000 Current ratio 1.675 1.381 1.120 9.995 0.090 Leverage 0.248 0.247 0.158 0.769 0.000 Return on assets 7.332 7.898 9.317 39.050 -56.179 Loss 0.104 0.000 0.306 1.000 0.000 Material weakness in internal controls 0.025 0.000 0.155 1.000 0.000 Restatement 0.028 0.000 0.166 1.000 0.000 Nonaudit fee to audit fee 0.220 0.176 0.184 1.386 0.000 Initial year audit 0.036 0.000 0.187 1.000 0.000 Committee expertise 0.437 0.333 0.347 1.000 0.000 Committee meetings 9.175 9.000 3.471 30.000 0.000 Committee size 4.429 4.000 1.036 8.000 2.000 Panel A reports descriptive statistics for the S&P 500 firms. Financial institutions (SIC codes between 6000
and 6900) are excluded from the sample. Panel B reports descriptive statistics for firms with female representation on the audit committee.

41

Table 3. Correlations.
Variable AFEE TA INVREC SQSIC FOREIGN LIQ LEV ROA LOSS ICW RESTATE NAFRATIO INITIAL CEXP CMEET CSIZE FCHAIR FEXPERT FMEMB FCE FCM
TA INVREC SQSIC FOREIGN LIQ LEV ROA LOSS ICW RESTATE NAFRATIO INITIAL CEXP CMEET CSIZE FCHAIR FEXPERT FMEMB FCE FCM FTOM

0.690

-0.025 -0.243

0.217 0.128 0.162

0.386 -0.267 -0.033 -0.367 0.120

0.039 -0.099 0.095 -0.108 0.064

0.059 -0.024

0.177 0.035

-0.051 0.007 -0.045 -0.029 -0.039 -0.020 0.002 -0.023 -0.010 -0.004

0.049 0.059 0.025 0.067 0.076 -0.017 -0.050 -0.038 0.002 -0.016 -0.006

0.008

0.065

0.185 0.084

0.122 0.193

-0.025 0.019 -0.002 0.007 0.029 0.009 0.021 -0.096 0.038 -0.053 -0.011 -0.014 0.006 0.149 -0.074 -0.004

0.064 0.019

0.090 -0.015 -0.033 0.135 0.007 0.027 0.006 0.020 0.049 0.024

-0.014 0.045 0.106 0.075 -0.093 -0.069 0.041 -0.025 -0.035 -0.053 -0.011 -0.009 0.024 0.056 -0.070 0.196 0.307 0.261 0.309 0.310 0.596

0.015 -0.001 -0.046

0.110 -0.086

0.043 -0.032 -0.038 -0.040 0.043 0.093

0.054 -0.038 -0.016 0.165

0.067 -0.024 0.002 0.083

0.023 -0.136

0.016 -0.007 0.104 0.159 -0.169

-0.041 -0.017 -0.010 -0.030 0.105

0.200 -0.224 -0.272

0.101 -0.162

0.014 -0.081 0.005 -0.079 0.023 0.002

0.046 -0.051 0.010 -0.031 0.032 0.057

0.014 -0.071 0.141 0.006

-0.056 -0.002 -0.038 -0.228 0.112 0.073 -0.049 0.128 0.054

-0.590 -0.115 0.106

-0.075 -0.085 -0.086 0.061 0.009 0.121 0.014

-0.049 -0.086 0.066 0.074 0.069

-0.036 -0.061 0.009 0.011

0.102 -0.064 0.135 -0.014 0.026

0.018 -0.047 -0.031 -0.010 -0.007

-0.006 -0.002 -0.022 -0.049 0.093 0.005

-0.019 -0.037 0.003 0.015 0.401

0.003 -0.021 0.008 0.020

0.015 -0.006 -0.026 0.000 -0.001 0.039 0.202 0.045 0.098

-0.006 -0.143 -0.108

0.005 -0.025 0.036 0.447

-0.081 -0.065 0.024 0.596 0.302 0.184 0.589

0.266 -0.021 0.309 0.526 0.898 0.528 0.277

The table reports pairwise correlations for the variables used in the regressions. (i) The dependent variable AFEE is the natural logarithm of audit fees. (ii) Firm-specific control variables are defined as follows: TA is the natural logarithm of total assets, INVREC is inventory and receivables to total assets, FOREIGN is the proportion of foreign sales to total sales, LIQ is the current ratio, LEV is total debt to total assets, ROA is the return on assets, LOSS is a dummy variable that equals one if the firm has incurred loss during the fiscal year, ICW is a dummy that equals one if there is a material weakness in internal controls, RESTATE is a dummy that equals one if there is a restatement, NAFRATIO is the ratio of nonaudit fees to audit fees, INITIAL is a dummy that equals one for the initial-year audit, CEXP is the proportion of experts to committee members, CMEET is the number of audit committee meetings, and CSIZE is the number of audit committee members. (iii) The dummy variables for female representation in the audit committee are defined as follows: FCHAIR equals one if the chair of the committee is a female, FEXPERT equals one if the committee has at least one female expert, FMEMB equals one if the committee has at least one female member, FCE equals one if the committee has a female chair and at least one female expert, FCM equals one if the committee has a female chair and at least one additional female member, and FTOM equals one if the committee has at least two female members.

42

Table 4. Regression results.

Variable Constant Control variables: TA + Exp. sign Model 1 2.651 *** (7.531) Model 2 2.650 *** (7.459) 0.507 (34.61) 0.540 (4.039) 0.170 (4.757) 1.366 (19.71) -0.045 (-2.857) 0.235 (2.659) -0.004 (-2.114) 0.028 (0.408) 0.687 (5.913) -0.307 (-3.24) -0.124 (-1.466) 0.012 (0.152) *** *** *** *** *** ** * Model 3 2.665 *** (7.618) 0.506 (34.94) 0.537 (4.019) 0.170 (4.748) 1.361 (19.36) -0.046 (-3.042) 0.235 (2.671) -0.004 (-1.834) 0.033 (0.481) 0.696 (5.956) -0.305 (-3.19) -0.117 (-1.367) 0.013 (0.158) *** *** *** *** *** *** Model 4 2.658 *** (7.570) 0.506 (34.81) 0.543 (4.054) 0.169 (4.718) 1.359 (19.32) -0.046 (-2.991) 0.236 (2.681) -0.004 (-1.825) 0.033 (0.485) 0.703 (6.062) -0.305 (-3.19) -0.117 (-1.374) 0.013 (0.160) *** *** *** *** *** *** Model 5 2.647 *** (7.465) 0.507 (34.64) 0.546 (4.085) 0.169 (4.737) 1.365 (19.70) -0.045 (-2.889) 0.239 (2.717) -0.004 (-1.946) 0.029 (0.425) 0.693 (5.972) -0.309 (-3.230) -0.122 (-1.432) 0.012 (0.146) *** *** *** *** *** *** * Model 6 2.640 *** (7.519) 0.507 (34.98) 0.555 (4.109) 0.170 (4.753) 1.354 (19.40) -0.046 (-2.985) 0.239 (2.712) -0.004 (-1.964) 0.029 (0.431) 0.698 (6.008) -0.306 (-3.19) -0.122 (-1.424) 0.019 (0.236) *** *** *** *** *** *** * Model 7 2.623 *** (7.434) 0.507 (34.86) 0.565 (4.160) 0.170 (4.754) 1.351 (19.27) -0.046 (-2.984) 0.233 (2.642) -0.004 (-1.937) 0.027 (0.399) 0.698 (6.008) -0.308 (-3.220) -0.119 (-1.383) 0.016 (0.200) *** *** *** *** *** ** *

0.506 *** (34.84) INVREC + 0.542 *** (4.051) SQSIC + 0.169 *** (4.724) FOREIGN + 1.358 *** (19.49) LIQ -0.046 *** (-2.977) LEV + 0.234 ** (2.658) ROA -0.004 (-1.839) LOSS + 0.034 (0.499) ICW + 0.703 *** (6.086) RESTATE + -0.306 *** (-3.22) NAFRATIO -0.117 (-1.369) INITIAL 0.013 (0.158) Audit committee control variables: CEXP + 0.068 (1.265) CMEET + 0.020 *** (3.898) CSIZE + 0.036 ** (2.056) Female representation variables: FCHAIR FEXPERT FMEMB FCE FCM FTOM Annual fixed effects Adjusted R 2 F-stat. Yes 0.696 127.465 ***

*** ***

*** ***

*** ***

*** ***

*** ***

*** ***

0.088 (1.610) 0.019 *** (3.699) 0.037 ** (2.121) -0.136 *** (-2.883)

0.039 (0.655) 0.020 *** (3.905) 0.034 * (1.935)

0.067 (1.240) 0.020 *** (3.899) 0.035 * (1.910)

0.090 (1.627) 0.020 *** (3.727) 0.036 ** (2.080)

0.079 (1.457) 0.020 *** (3.800) 0.037 ** (2.092)

0.076 (1.392) 0.020 *** (3.852) 0.041 ** (2.304)

0.041 (1.059) 0.009 (0.274) -0.119 ** (-2.300) -0.156 ** (-2.444) -0.085 * (-1.782) Yes 0.696 120.760 ***

Yes 0.698 121.587 ***

Yes 0.696 120.446 ***

Yes 0.695 120.268 ***

Yes 0.697 121.104 ***

Yes 0.697 120.987 ***

43 The table reports the estimates of alternative versions the following regression specification: AFEE i ,t 1TAi , t 2 INVREC i ,t 3 SQSIC i ,t 4 FOREIGN i , t 5 LIQ i ,t
6 11 15

LEV i ,t

ROAi ,t
12

LOSS i ,t

9 13

ICW i ,t CEXPi ,t

10

RESTATE i ,t CMEET i ,t

NAFRATIO i ,t

INITIAL i ,t

14

CSIZE i ,t i ,t 16 FEMALE i ,t where AFEEi,t is the natural logarithm of audit fees, TAi,t is the natural logarithm of total assets, INVRECi,t is inventory and receivables to total assets, FOREIGNi,t is the proportion of foreign sales to total sales, LIQi,t is the current ratio, LEVi,t is total debt to total assets, ROAi,t is the return on assets, LOSSi,t is a dummy variable that equals one if the firm has incurred loss during the fiscal year, ICWi,t is a dummy that equals one if there is a material weakness in internal controls, RESTATEi,t is a dummy that equals one if there is a restatement, NAFRATIOi,t is the ratio of nonaudit fees to audit fees, INITIAL i,t is a dummy that equals one for the initial-year audit, CEXPi,t is the proportion of experts to committee members, CMEET i,t is the number of audit committee meetings, and CSIZEi,t is the number of audit committee members. The female representation dummy variables are defined as follows: FCHAIRi,t equals one if the chair of the committee is a female, FEXPERTi,t equals one if the committee has at least one female expert, FMEMBi,t equals one if the committee has at least one female member, FCEi,t equals one if the committee has a female chair and at least one female expert, FCMi,t equals one if the committee has a female chair and at least one additional female member, and FTOMi,t equals one if the committee has at least two female members. The t-statistics (reported in parentheses) are based on Whites heteroskedasticity consistent standard errors. ***, **, and * denote significance at the 0.01, 0.05, and 0.10 levels, respectively.

44

Table 5. Matched-firm regressions.

Variable Constant Control variables: TA + Exp. sign Model 1 2.731 *** (3.218) Model 2 3.130*** (6.869) 0.496*** (26.21) 0.453** (2.518) 0.204*** (3.916) 1.373*** (13.92) -0.055** (-1.977) 0.021 (0.176) -0.006* (-1.823) -0.063 (-0.718) 0.549*** (3.740) -0.389*** (-3.217) -0.291*** (-3.658) 0.035 (0.382) -0.079 (-1.074) 0.019*** (2.779) 0.004 (0.165) Model 3 2.615 *** (3.229) 0.490 *** (33.89) 0.407 *** (3.184) 0.152 *** (4.314) 1.516 *** (21.18) -0.041 ** (-2.083) 0.377 *** (3.455) -0.002 (-1.346) -0.003 (-0.048) 0.679 *** (4.726) -0.514 * (-1.849) -0.184 ** (-2.549) 0.513 * (1.877) 0.061 (1.180) 0.026 *** (5.027) 0.056 *** (3.160) Model 4 2.904 *** (2.984) 0.506 *** (13.47) 0.713 ** (1.982) 0.182 * (1.946) 1.372 *** (7.917) -0.055 (-1.083) 0.250 (0.801) -0.008 (-1.139) -0.383 * (-1.675) 0.528 ** (2.435) -0.254 (-1.336) -0.318 *** (-2.760) -0.045 (-0.260) -0.002 (-0.022) 0.023 ** (2.192) -0.027 (-0.504) Model 5 3.526* (1.905) 0.474*** (6.991) 0.788 (1.483) 0.100 (0.956) 1.843*** (5.588) -0.310** (-2.398) -0.493 (-0.826) -0.015 (-1.580) -0.129 (0.503) 0.030 (0.093) -0.231 (-1.099) -0.302* (-0.652) 0.265 (1.303) -0.009 (-0.053) 0.025 (1.585) 0.159* (1.998) Model 6 2.963 *** (2.976) 0.493 *** (13.24) 0.401 (1.290) 0.141 ** (1.964) 1.746 *** (9.890) -0.095 * (-1.831) -0.124 (-0.375) -0.012 ** (-2.199) -0.013 (-0.075) 0.721 (1.470) 0.018 (0.159) -0.184 (-1.436) 0.012 (0.106) 0.159 (1.468) 0.019 ** (2.197) 0.094 ** (2.439)

0.513 *** (15.60) INVREC + 0.763 ** (2.392) SQSIC + 0.174 ** (2.229) FOREIGN + 1.346 *** (8.103) LIQ -0.056 (-1.149) LEV + 0.133 (0.497) ROA -0.008 (-1.526) LOSS + -0.307 * (-1.667) ICW + 0.509 ** (2.533) RESTATE + -0.294 * (-1.748) NAFRATIO -0.224 * (-1.785) INITIAL -0.013 (-0.095) Audit committee control variables: CEXP + 0.047 (0.484) CMEET + 0.018 * (1.817) CSIZE + 0.002 (0.053) Female representation variables: FCHAIR -0.145 ** (-2.026) FEXPERT FMEMB FCE FCM FTOM Annual fixed effects Adjusted R2 F-stat. Yes 0.663 24.105 ***

0.042 (0.917) -0.028 (-0.880) -0.113 (-1.398) -0.111 (-0.829) -0.161 ** (-2.326) Yes 0.674 25.289 ***

Yes 0.684 60.621***

Yes 0.667 115.574 ***

Yes 0.663 19.455 ***

Yes 0.690 10.137***

45 The table reports the estimates of alternative versions the following regression specification using matched-firm samples: AFEE i ,t 1TAi , t 2 INVREC i ,t 3 SQSIC i ,t 4 FOREIGN i , t 5 LIQ i ,t
6 11 15

LEV i ,t

ROAi ,t
12

LOSS i ,t

9 13

ICW i ,t CEXPi ,t

10

RESTATE i ,t CMEET i ,t

NAFRATIO i ,t

INITIAL i ,t

14

CSIZE i ,t i ,t 16 FEMALE i ,t where AFEEi,t is the natural logarithm of audit fees, TAi,t is the natural logarithm of total assets, INVRECi,t is inventory and receivables to total assets, FOREIGNi,t is the proportion of foreign sales to total sales, LIQi,t is the current ratio, LEVi,t is total debt to total assets, ROAi,t is the return on assets, LOSSi,t is a dummy variable that equals one if the firm has incurred loss during the fiscal year, ICWi,t is a dummy that equals one if there is a material weakness in internal controls, RESTATEi,t is a dummy that equals one if there is a restatement, NAFRATIOi,t is the ratio of nonaudit fees to audit fees, INITIAL i,t is a dummy that equals one for the initial-year audit, CEXPi,t is the proportion of experts to committee members, CMEET i,t is the number of audit committee meetings, and CSIZEi,t is the number of audit committee members. The female representation dummy variables are defined as follows: FCHAIRi,t equals one if the chair of the committee is a female, FEXPERTi,t equals one if the committee has at least one female expert, FMEMBi,t equals one if the committee has at least one female member, FCEi,t equals one if the committee has a female chair and at least one female expert, FCMi,t equals one if the committee has a female chair and at least one additional female member, and FTOMi,t equals one if the committee has at least two female members. The t-statistics (reported in parentheses) are based on Whites heteroskedasticity consistent standard errors. ***, **, and * denote significance at the 0.01, 0.05, and 0.10 levels, respectively.