Public Audit Oversight and Audit Pricing: Evidence from the EU

Abstract

We examine the audit pricing consequences of auditor inspections under the public oversight regime in the EU. Employing a staggered differences-in-differences design, we document an inspections audit fee increase during the post-inspection period when companies’ auditors are subject to inspections by the national Public Oversight Body (POB). However, this effect masks significant cross-sectional variation. Specifically, we find that the increase in audit fees attributable to inspections is concentrated among POBs with more adequate human resources, where inspections last longer or occur both at the auditor’s and the regulator’s premises. Also, the effect of inspections on audit fees is evident only when the POB prohibits inspectors from joining an audit firm immediately after their departure or when the oversight system is funded by multiple stakeholders. Overall, our findings suggest that audit costs as reflected in audit fees increase for clients of inspected auditors but only when inspections are more laborious, independent, and rigorous.

KEYWORDS:

• POB
• Public oversight
• Inspections
• Audit fees

JEL codes:

• G18
• G38
• M41
• M42
• M48

Acknowledgements

We thank Clive Holderness, Clive Lennox, Christian Leuz, Peter Pope, Nemit Shroff, Irem Tuna, Ann Vanstraelen and seminar participants at university of Essex, Nottingham, Warwick, Lancaster and Bocconi as well as seminar participants at the 6th Audit Quality Workshop for their comments on earlier versions of the paper. We also thank Annalisa Prencipe and Juan Manuel Garcia Lara for helping us obtain a response from the Italian and Spanish POB, respectively. We are particularly grateful to the Editor Beatriz García Osma and two anonymous reviewers for their valuable guidance.

Disclosure statement

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

Supplemental Data and Research Materials

Supplemental data for this article can be accessed on the Taylor & Francis website, https://doi.org/10.1080/09638180.2022.2145981.

Figure 1: Inspection effects on audit fees in event-time

Table OA1: Inclusion of auditor-year fixed effects

Table OA2: Cross-sectional analyses using only the treatment sample

Table OA3: EU public audit oversight systems and inspections characteristics

Table OA4: Univariate analysis (by treatment country)

Table OA5: Stacked and two-stage DID regressions

Table OA6: Sensitivity analyses

Table OA7: PSM analysis

Table OA8: Additional analyses on audit change and audit fees

Table OA9: Additional analyses on audit quality/financial reporting quality

Table OA10: POB annual budget

Table OA11: POB enforcement

Notes

1 We explain our focus on resources and independence in Section 3.3 and Appendix 2.

2 Starting from 2005, EU-listed companies have adopted International Financial Reporting Standards (IFRS).

3 Country-level data on market capitalization is available at: https://data.worldbank.org/indicator/CM.MKT.LCAP.CD.

4 See Lohlein (2016) for a recent PCAOB literature review, and Gipper et al. (2020).

5 The 2006/43/EC Directive was updated in May 2014, leading to the 2014/56/EU Directive. In principle, the amendments were expected to be implemented from June 2016 onward.

6 Conditional on audit failure, litigation costs represent the probability of auditor lawsuit and the strictness of the legal liability regime (Choi et al., 2008). Christensen et al. (2021) document that publicly available firm-level PCAOB inspection reports with audit deficiencies provide useful information to lawyers and ultimately increase the number of auditor lawsuits. However, concurrent studies (Carson et al., 2021; Hanlon & Shroff, 2022) and our data (see Appendix 2) reveal that in the EU the vast majority of POBs disclose only aggregate inspection reports. Therefore, it is highly unlikely that lawyers rely on publicly available POB information when deciding to file lawsuits against inspected auditors. Moreover, the shift to public audit oversight does not alter country-level legal infrastructure. Thus, our discussion regarding the potential effect of auditor inspections on litigation costs focuses on the likelihood of audit failure.

7 We note that audit risk anticipated by auditors can affect fees through planned incremental audit effort and/or through risk premia built into audit fee negotiations. In common with prior literature, we are unable to observe the fundamental determinants of audit fees (audit effort and risk perceptions).

8 Arguably, other dimensions may also play an important role in the effectiveness of public oversight systems. For example, IFIAR (2011) provide seven core principles regarding the structure and operations of audit regulators. In addition to the factors examined in our study, IFIAR (2011) highlight the importance of cooperation among national audit regulators, comprehensive enforcement and high standards of ethical conduct. Similarly, in addition to transparency, Carson et al. (2021) examine the type of oversight, the frequency of inspections and the type of enforcement. Ideally, we would like to examine as many dimensions as possible. However, driven by economic theory and policy recommendations as well as by practical reasons related to data collection (e.g., trade-off between quantity and quality of data obtained via surveys) we focus on the aforementioned dimensions of public oversight. Nonetheless, we also investigate the conditioning audit fee effects of POB enforcement, given its vital role in regulatory effectiveness (see Section 4.5.3).

9 Consistent with this argument, our data show that there is some overlap among countries with well-staffed oversight systems, countries with inspections of longer duration and countries where inspections take place at both the auditor’s offices and the POB’s premises (see Online Appendix (OA) – Table OA3).

10 Hendricks et al. (2022) provide some evidence in line with the human capital hypothesis. They find that the number of former PCAOB employees hired by large audit firms enhances regulatory compliance (i.e., reduces the number of deficiencies reported in firms’ future inspection reports) but does not improve subsequent audit quality.

11 Arguably, if funding by audit firms is not discretionary but instead is mandated by law, then POB oversight is likely to be independent from the audit profession. However, in line with the capture theory, we argue that the relative level of such funding may significantly affect audit firms’ leverage in affecting the political survival of the regulatory officials as well as the agency’s budget.

12 Naturally, an alternative proxy for oversight independence is the POB composition, and in particular the percentage of independent members. However, our data reveal that all POB members are non-practitioners, and thus independent, for all sample countries. Given the lack of cross-country variation we are not able to examine the aforementioned oversight feature.

13 To mitigate any related concerns, we assess the sensitivity of our baseline findings to the inclusion of auditor-year fixed effects in addition to industry and country indicators. Auditor-year fixed effects control for potential audit firm-wide changes affecting all clients. As reported in Table OA1, our results remain unchanged.

14 In this setting, the INSPECTION indicator effectively corresponds to the product of a firm indicator variable and a post-adoption indicator variable. To enhance readability, we do not write out the product of the two aforementioned indicators but rather refer to their product as INSPECTION. Since we include firm and year fixed effects, we suppress the respective main effects.

15 We repeat all cross-sectional tests after using only the treatment group. Table OA2 shows that our inferences persist.

16 Given the importance of SIZE as a determinant of audit pricing, we also include SIZE² to capture any nonlinearities between the former and audit fees (Hay et al., 2006).

17 The initial email was sent in the winter of 2018, and follow-up reminders were sent throughout 2018 and 2019. All correspondence with POBs, with a few exceptions, is written and is available on request from the authors. Approval was granted by the Research Ethics Committee.

18 We also use country profiles provided by IFIAR to obtain information about the national law that transposed the 2006/43/EC Directive.

19 Article 34 of the 2006/43/EC Directive specifies that companies are subject to the public oversight requirements of the country in which they are domiciled (‘home-country’). Hence we exclude non-domiciled companies from each country. We use the nationcode item from Worldscope (code WS06027) as well as ISIN codes to determine domicile.

20 We identify companies trading on unregulated markets of relevant countries using hand-collected information from the stock exchange website in each country.

21 We exclude Cyprus and Poland from the treatment group because they did not mandate POB inspections during our sample period. We also exclude Norway, which apparently has implemented a public audit oversight system since 1995. Finally, we drop Greece, Latvia, Luxembourg and Romania due to insufficient required data.

22 We report and discuss oversight and inspections characteristics for which we received clear answers with sufficient cross-country heterogeneity (for further details see Appendix 2). Our cross-sectional tests are also based on these characteristics.

23 The number of sample countries in Panel D is lower because Czech Republic and Italy (Czech Republic, Italy and Netherlands) did not provide information regarding inspections’ duration for large auditors (small auditors).

24 We calculate the economic significance of INSPECTION based on the antilog of the 0.043 estimate.

25 We create seven event-specific datasets based on the commencement of POB inspections by a given country. In our sample, POB inspections start at seven different years, namely year 2004 (UK), 2005 (France), 2007 (Belgium, Germany and Netherlands), 2009 (Czech Republic, Finland and Portugal), 2010 (Austria and Italy), 2011 (Spain) and 2016 (Ireland and Slovenia). In each event-specific dataset, we restrict the control group to firms that never adopt POB inspections so the control group are ‘clean’ controls within the dataset. These seven datasets are then stacked together, and a DiD regression of Equation (1) is estimated on this stacked dataset. As multiple datasets are stacked together, the sample size is increased significantly.

26 Specifically, using only the control sample we estimate the first-stage regression of audit fees with year-, firm-fixed effects and size. Then, we obtain the predicted residual for all companies (treatment and control) from this first-stage regression. In the second stage, we regress the residual against INSPECTION and all other control variables and fixed effects included in Equation (1).

27 We omit the indicator t-2, which serves as benchmark period.

28 We balance the treatment and control sample on the first, second and third moment of all control variables used in Equation (1). Entropy balancing does not allow the inclusion of firm fixed effects in the 2nd stage regression. We therefore use year-, industry- and country- fixed effects. However, we also perform propensity score matching on all control variables used in Equation (1) with replacement and based on caliper 0.05 and 0.01, and after employing firm- and year- fixed effects in the 2nd stage regression (see Table OA7).

29 Eierle et al. (2021) identify several country-level determinants of audit pricing. Given that time-invariant country factors are controlled by our firm fixed effects we focus on time-variant variables with available data, which we obtain from World Bank. To maintain the same sample size we fill in missing values using the average of non-missing values for each variable. But, we note that results remain unchanged when we repeat our analysis using a smaller sample.

30 Poland established the national POB in 2009 but mandated POB inspections in 2017.

31 Following Christensen et al. (2013) we include a binary variable equal to 1 for Finland and UK for all observations ending on and after 2005; for Germany and Netherlands for all observations ending on or after 2006; and for Ireland for all observations ending on or after 2007.

32 Auditor changes may be driven by supply forces (i.e., the auditor rejects a particular client) or demand forces (i.e., a client refuses to be audited by a particular auditor). Our data does not allow us to distinguish between the two alternative explanations.

33 As reported in Table 2, Panel A, there is only one country where the number of inspectors is less than five. Similarly, there is only one country where inspectors range between 25 and 50. Accordingly, we group these countries with their immediately closest cohort.

34 As reported in Table 2, Panel C, there is only one country where the number of audit firms regulated ranges between five and 10. Accordingly, we group this country with its immediately closest band.

35 In this case, we cannot split INSPECTION_{Employees_High} into two further subsets depending on the number of audit firms regulated because there is no variation in the latter for this INSPECTION subgroup.

36 As reported in Table 2, Panel D, there is only one country, where inspections of small auditors may last less than one week. Similarly, there is only one country, where the duration of small auditor inspections can be more than eight weeks. Accordingly, we group these two countries with their immediately closest band.

37 A possible concern is that we always obtain the same group of countries when sorting our sample using the inspections conditioning variables; as a result, the reported conditional audit fees effects of inspections may be driven by a single set of countries. However, as described in the text, while there is naturally some level of overlap of countries across the various clusters, there is also considerable heterogeneity.

38 However, we note that this analysis is subject to two caveats: a) the budget data relates only to 2017 and therefore is time-invariant; and b) the sample is reduced by 30% given that six out of 13 treatment countries did not provide any budget information.

39 We classify the enforcement type of our treatment countries based on information reported by Carson et al. (2021; see their Table 1). Accordingly, we employ a sample of nine treatment countries (instead of 13) due to missing data. Moreover, in our treatment sample only the German POB oversees the enforcement process; therefore, we group Germany with countries, whose POB has no enforcement power.

40 Please note that if the answer to question 2 is c, all remaining questions refer to the current POB, which was established in accordance with the EU Directive 2006/43/EC and replaced the previous POB.

41 For example, have there been any changes in relation to questions 4, 5, 6 and 7 following the amendments brought about with the 2014 EU Regulation and Directive?

42 According to EU Directive 2006/43/EC, a ‘non-practitioner’ means any natural person who, for at least three years before his or her involvement in the governance of the public oversight system, has not carried out statutory audits, has not held voting rights in an audit firm, has not been a member of the administrative or management body of an audit firm and has not been employed by, or otherwise associated with, an audit firm.