Gender Salary Gap in the Auditing Profession: Trend and Explanations

ABSTRACT

This study documents the gender salary gap in the auditing profession and explains its development. Using Swedish administrative data from 2007 to 2015 for all Certified Public Accountants (CPAs), I find that the auditing profession’s overall gender salary gap has substantially narrowed during this period, and more female auditors have moved up to the top earnings group, reducing the wide salary gap at the top of the income distribution. Further analysis shows that the increase in female auditors’ client portfolio size accounts for approximately half of the decrease in the total salary gap. I also find evidence that the rise in female leadership in the Big Six firms is positively associated with the increase in female auditors’ client portfolio size. This effect is more pronounced in the middle and bottom half of firms’ hierarchy, suggesting that female representation at the top of a firm’s hierarchy has spillover benefits for lower-ranked female auditors. The implications of this study may help audit firms narrow their gender gaps and cope better with the overall talent challenges of the auditing industry.

Keywords:

• Gender gap
• Auditor
• Glass ceiling
• Female leadership

JEL Codes:

• M42
• J16

Acknowledgements

The author is deeply grateful for the helpful comments and suggestions from Beatriz Garcia Osma (editor) and two anonymous reviewers, as well as the support and advice from her PhD committee members: Henrik Nilsson (chair), Edward Lee, and Ann Vanstraelen. The author also wishes to thank Daniel Aobdia, Juha-Pekka Kallunki, Anna Dreber Almenberg, Florian Eugster, Kenth Skogsvik, Mariya Ivanova, Martin Jacob, Karolis Ramoska, Milda Tylaite, Antonio Vazquez and seminar participants at Maastricht University, the First Labor and Accounting Conference, Stockholm School of Economics, the Swedish Inspectorate of Auditors (Revisorsinspektionen), and the FIRE workshop for their helpful comments. The author wishes to thank the Swedish Inspectorate of Auditors for providing the data and the generous funding from the Torsten Söderberg Foundation, the Iwar Sjögrens Foundation, Mistra Center for Sustainable Markets (Misum), the Swedish Bank Research Fundation (BFI), Handelsbanken, and the Jan Wallanders and Tom Hedelius Foundation (Jan Wallanders och Tom Hedelius stiftelse).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental Research Materials

Supplemental data for this article can be accessed online at https://doi.org/10.1080/09638180.2022.2113550.

Online Appendix A1: How Big Six firms communicate their initiatives to promote women.

Online Appendix A2: Female share in top income groups.

Online Appendix A3: Gender salary gap in the Big Six firms.

Online Appendix A4: Robustness tests.

Online Appendix A5: Decomposing the gender salary gap among the Big Six auditors.

Online Appendix A6: Supplementary tests on the determinants of auditors’ decisions to exit the profession.

Online Appendix, Tables, A-1 to A-5.

Online Appendix, Figures, F-1 to F-5.

Notes

1 See, for example, Abdelfattah et al. (2021), Almer et al. (2021), Bustos-Contell et al. (2022), and Hardies et al. (2021).

2 See, for example, Albrecht et al. (2015), Keloharju et al. (2022), Miller (2011), Bolotnyy and Emanuel (2022), Matsa and Miller (2011), and Niederle and Vesterlund (2007).

3 For example, Athey et al. (2000) show that female superiors may mentor female subordinates better and serve as role models and advocate for lower-ranked females. Female leaders also cultivate more women-friendly cultures within their firms (Tate & Yang, 2015). Furthermore, Matsa and Miller (2011) document that women in corporate leadership roles can help other women to advance to top management positions; Kunze and Miller (2017) also show evidence supporting the notion that female leadership can have positive spillover effects on women in lower ranks of the same firm.

4 I present, in the online appendix (A1), a summary of the Big Six firms’ recent statements regarding their attempts to increase female representation at top levels.

5 Due to data limitation, I can only identify Big Six firms’ CPA job affiliations and their office locations.

6 Leading audit firms in the UK and Australia have recently reported large gender pay gaps (Deloitte, 2018; EY, 2020; KPMG, 2017; PwC, 2020). Professional accounting bodies and audit firms have also raised concerns about gender inequality and its detrimental effects on talent management (AICPA, 2019a; Deloitte, 2011; Echevarria, 2012; International Federation of Accountants [IFAC], 2021; Franzel, 2014). In the accounting literature, many studies have documented the gendered nature of public accounting, including Adapa et al. (2016), Almer and Single (2007), Anderson-Gough et al. (2005), Barker and Monks (1998), Cohen et al. (2020), Dalton et al. (2014), Edgley et al. (2016), Hull and Umansky (1997), Kornberger et al. (2010), and Lupu (2012).

7 For example, Knechel et al. (2013) focus on the determinants of Big Four partners’ compensation. In their regression analyses based on a sample of Big Four audit partners in Sweden, the FEMALE indicator shows a significantly negative sign (Table 6, p. 371), suggesting that even after controlling for job rank (as well as a battery of other variables), female auditors’ compensation at the top of the Big Four firms’ hierarchies remains discounted. Both Knechel et al. (2013) and Vandenhaute et al. (2020) find that commercial factors, such as audit partners’ client portfolio size, total audit fees collected, and the number of listed clients, are strong predictors of partner compensation.

8 The same concern has been raised elsewhere. For example, Kevin Ellis, UK chair and senior partner at PwC (UK), warned that the audit profession was unattractive to new recruits. PwC also stated that its attrition levels for newly qualified CPAs were 8% higher in 2021 than in 2020 (O’Dwyer, 2021).

9 Blau and Kahn (2017) also note that women’s job aspirations may be affected by their anticipated treatment in the labor market.

10 This is because men can work more overtime, while their household partners take care of home obligations. The effect of overwork offsets the improvement in women’s educational attainment and other forms of human capital.

11 Goldin (2014) points out that, for occupations where job flexibility is extremely costly, women are more likely to be burdened by the high penalty for childbirth, shorter working hours, and a flexible schedule. Hence, without fundamental changes in ‘how jobs are structured and remunerated’ (Goldin, 2014, p. 1091), the gender pay gap cannot be eliminated.

12 For professionals in other occupations – such as lawyers and managers – where job flexibility is also very costly, prior studies even show a large and widening gender pay gap over time (Bertrand et al., 2010; Goldin, 2014; Noonan et al., 2005; Reuben et al., 2015).

13 For example, although women now account for approximately half of the labor force, there are only 8% female CEOs running Fortune 500 firms as of 2021 (Hinchliffe, 2021). Female leadership also seems to be rare in the auditing industry. In 2019, the Financial Times reported that, among all the lead audit partners of FTSE 100 companies, only nine were women, and almost all were white (Marriage & Gross, 2019). Changes in the situation also seem to be slow: AICPA’s (2017) survey showed that 22% of partners in all public accounting firms were women, whereas the figure for 2011 was 21%. Female representation in the largest audit firms (with more than 100 CPAs) is even lower, with women accounting for 16% of the equity partners and representing merely 14% of the total equity in those firms (AICPA, 2011, 2017).

14 Several empirical studies have found evidence supporting this ‘women helping women’ effect (Bilimoria, 2006; Cook & Glass, 2014, 2015; Kirsch, 2018; Kunze & Miller, 2017; Skaggs et al., 2012; Stainback et al., 2016).

15 Historically, the Big Four culture has been characterized by male and white domination, masculinity, and patriarchy (Edgley et al., 2016). In such an environment, ‘homogeneity’ is an important element of the Big Four’s occupational branding and, therefore, the ability to ‘fit in’ is critical for auditors’ career success (Hanlon, 2004; Hechter, 2008). This results in a ‘closed circle of male homo-sociality’ that precludes the ascension of women in audit firms (Anderson-Gough et al., 2005, p. 478). Other studies in this line of research include Adapa et al. (2016), Barker and Monks (1998), Dambrin and Lambert (2012), Hull and Umansky (1997), and Lupu (2012).

16 One exception is Edgley et al. (2016), who employ the institutional logics approach in studying how diversity is perceived and enacted in audit firms. Edgley et al. find that gender imbalance appears to attract most attention on firms’ diversity agenda; however, the shift seems to be symbolic, and there is a clear discrepancy between the Big Four’s discourses and the actual data. The data gathered in their study are in the form of texts, images, photographs, and graphical representations, among others.

17 The total number of Swedish CPAs has been decreasing gradually in recent years, from 4,218 in 2000 and 3999 in 2010, to 3589 in 2015 (see Figure F-1 in the online appendix for the trend). According to SIA’s (2021) annual report, the total number of registered auditors is 3,007.

18 Similar to the Big Four firms, BDO and Grant Thornton operate as global networks of independent firms that follow the same global policies. Thus, I expect that the gender equality status within BDO and Grant Thornton is likely to be closer to that of the Big Four firms than Swedish local firms.

19 2007 is the first year for which the income data are available from the data provider.

20 STA classifies a person’s income into three categories: salary, capital, and business income. Auditors’ work-related compensation may come from both salary and capital income. Equity partners of an audit firm receive dividend payments from the firm, which is classified by STA as capital income. A limitation of this study is that I do not have access to the data required for identifying equity partners in audit firms; neither can I distinguish partnership dividends from other types of capital income, such as capital income from stock trading and real estate transactions. Therefore, the focus of this study is on the gender salary gap. Nevertheless, for robustness check in one specification, I follow Knechel et al. (2013) and identify an audit partner as one who has signed a listed company’s audit report in a certain year. Following Knechel et al. (2013), I use salary plus capital income as a crude measure of a partner’s total job-related compensation.

21 The Serrano database combines the following data sources: financial statement data and bankruptcy information for all Swedish companies from the Swedish Companies Registration Office (SCRO), the general history of company data from Statistics Sweden, and group data from Bisnode’s group register. The SCRO also keeps track of information about each company’s auditors, including their personal identity numbers and auditor tenure with the client companies.

22 Following the labor economics literature (e.g., Blau & Kahn, 2006), I do not winsorize the salary data for tests presented in this study. Since the income data source is the STA, the data quality is high. For robustness check, I winsorize the yearly salary data at the 1 and 99 percentiles for regression analyses, and the results remain largely the same for all the tests.

23 For example, the salary difference between men and women in 2007 is approximately 22%. The difference decreases to 14% in 2011, and further decreases to approximately 10% in 2015.

24 The median (25th percentile) male CPA salary is 553,060 (451,803) SEK, and the median (25th percentile) female CPA salary is 485,123 (399,637) SEK. The difference is 67,937 (52,166) SEK or 14% (13%). However, a male CPA at the 90th percentile of the male CPAs’ salary distribution earns 880,000 SEK, whereas a female CPA at the corresponding level earns only 750,605 SEK – the gap widens to approximately 130,000 SEK or 17.2%. The highest male CPA salary in the sample is 8,856,826 SEK, whereas the highest female CPA salary is 6,061,232 SEK; the difference is close to 2.8 million SEK or 46%.

25 I control for REGION because the Swedish economy is not evenly distributed across the country. The population of Sweden is concentrated in the south, while the north is heavily forested. Economic development varies across Sweden’s geographical regions and municipalities, as Edvinsson et al. (2011) show. Citizens of Stockholm (the capital and largest city), Gothenburg, and Malmö have the highest disposable income levels, whereas those living in the north have the lowest.

26 The division follows the classification scheme of the Swedish Post (Posten): 1 = Stockholm, 2 = Skåne, 3 = South East, 4 = South West, 5 = South Central, 6 = East, 7 = North Central, 8 = Southern Norrland, and 9 = Northern Norrland. The city of Gothenburg belongs to the South West region, and Malmö to the Skåne area.

27 The requirement for becoming an approved CPA was to have a bachelor’s degree, pass a professional competency exam, and have at least three years’ practical experience. The higher standards for becoming an authorized CPA included a master’s degree, at least five years’ experience, and passing a more challenging professional competency exam. After June 2013, the two-tier system was removed, and all new CPAs are presently designated as authorized CPA (auktoriserad revisor).

28 More specifically, I first calculate, for each year, individual auditor’s market share in a given industry (classified by the two-digit SNI industry code) as the sum of their clients’ total sales in the industry divided by the total sales of all companies in the industry. Second, I rank auditors in each industry by their market share and then define an auditor as an industry expert if he/she is the top-ranked or second-ranked auditor in the industry.

29 From 2007 to 2011, the average number of clients for female CPAs increased from 38 to 44 but decreased to 40 in 2015. The same pattern emerges for male CPAs: the values of NAUDITS for 2007, 2011, and 2015 are 76, 85, and 72 clients per male CPA, respectively. This pattern is probably due to the change in the Swedish audit regulation in 2010, which removed the mandatory audit requirement for micro-sized companies. After 2010, most small companies under the size threshold gradually opted out of financial statement audits (Dong et al., 2022).

30 Another factor that may negatively affect women’s work experience is childbirth interruption. In an untabulated test, I also introduce a crude adjustment to TENURE for female CPAs. More specifically, I start to count women’s tenure from 480 days (1.31 years, the total duration of paid parental leave in Sweden) after they have obtained their CPA license, assuming that all female auditors take the full parental leave period after they obtain the license. Using this crude measure for tenure with motherhood adjustment, the estimated coefficient on FEMALE is -0.142 (p < 0.01). This result shows that the size of the salary gap remains large even after adjusting for the potential impact of parental leave on female auditors’ tenure.

31 To demonstrate the difference between the sizes of the gap in 2007 and 2015, in untabulated tests, I also conduct an estimation similar to that in Panel, A. Column (7), but include only data for 2007 and 2015 and add the interaction term, FEMALE*year2015. The coefficient on FEMALE is -0.130 (p < 0.01), while the coefficient on the interaction term, FEMALE*year2015, is 0.083 (p < 0.01), suggesting that the size of the gap in 2015 is significantly smaller than that in 2007.

32 More specifically, the estimated coefficients on FEMALE are −0.165 at the 95th percentile, −0.117 at the 75th percentile, −0.0985 at the median, and −0.0982 at the 25th percentile, respectively, and all are statistically significant at the 1% level.

33 Please note that, in contrast to the OLS regression which presents a negative sign for the salary gap based on women less men’s log salaries, the Oaxaca-Blinder approach usually presents the male-female log wage difference.

34 The aggregated effect with a negative sign on ‘other career variables’ is mainly driven by the number of audit clients (LN_NAUDITS, with −0.0297 log points). This finding may indicate that it is more difficult for women to obtain large clients, i.e., given the same client portfolio size, women tend to engage with a larger number of smaller clients, which is not as profitable as a small number of large clients.

35 One might argue that if women were more likely to work part-time, they would have smaller client portfolio sizes, which might bias the proportion of the explained gap. While this is likely to be the case, it should not substantially affect my results because part-time work is not prevalent in audit firms. According to PwC’s (Sweden) 2021 annual report, among middle-senior level staff in the entire firm, approximately 5% of females are on part-time, compared to 1% of men (PwC, 2021). The proportion of part-time CPAs in my sample should be even smaller because all CPAs in this sample could be identified with at least one audit client. Therefore, I believe that the concern about part-time working should not significantly bias the results presented here.

36 Indeed, the unexplained gap in 2015 remains large. This could be due to discrimination, measurement error of control variables, or unobservable characteristics such as psychological factors and individual preferences for risk or competition. The proportion of the unexplained gap relative to the total gap (45%) as shown in Table 5 is largely consistent with Blau and Kahn (2017), who use detailed U.S. PSID data to examine the trend and explanations for the gender pay gap in the U.S. labor market from 1980 to 2010.

37 Data on firms’ board room compositions were also obtained from the Serrano database. Year 2017 is the last year for which I have access to these data.

38 Due to data unavailability (described in the third section), I could only identify the Big Six firms’ detailed information necessary for conducting the empirical analyses in this section.

39 The considerable decrease in the unexplained component of the total gap (shown in Table 5) is also likely to be affected by female leadership. As I have discussed in previous sections, changes in unobservable factors such as women’s preferences, selection into (or exit from) the profession, psychological issues, and risk aversion could all play a part in narrowing the unexplained gap.

Additional information

Funding

This work was supported by Swedish Bank Research Fundation; Jan Wallanders and Tom Hedelius Foundation; Mistra Center for Sustainable Markets.