Is the Public Oversight of Auditors Effective? The Impact of Sanctions on Loss of Clients, Salary and Audit Reporting

Abstract

This study examines the consequences of sanctions against individual Swedish auditors issued by the Supervisory Board of Public Accountants (SBPA). The results provide no support for individual auditor client loss after receiving a sanction. However, we find that Big 4 auditors have a lower salary after the sanction than before. Finally, we do not find that auditors become more conservative in their reporting after being sanctioned. Collectively, our results support that public oversight sanctions have relatively limited consequences for auditors of private companies.

Acknowledgements

We would like to thank Ann Vanstraelen (the associate editor), an anonymous reviewer and Jörgen Hellström for their guidance and constructive comments. This paper has also benefitted from comments of the participants at the 37th Annual Conference of the European Accounting Association, the EIASM Workshop on Audit Quality (2014) and seminars at Umeå School of Business and Economics.

ORCiD

Stefan Sundgren http://orcid.org/0000-0002-2496-108X

Notes

1 Research at the individual auditor level is particularly interesting considering the recent change in for example EU regulation to disclose the name of the engagement partner in the audit report (IAASB, 2013).

2 29 certified auditors had their licences withdrawn during the same period. We are unable to research this group because they have left the profession, but the number of sanctions indicates that the risk of losing one's licence is not negligible.

3 FAR is the professional institute for authorized auditors, approved auditors and other qualified professionals in the accountancy sector in Sweden.

4 The quality controls conducted by FAR should meet all the requirements stated by the EU.

5 In the period 2005–2009 a total of 674 disciplinary cases were opened: 177 (26.3%) were initiated as a result of inspections by SBPA or FAR, 145 (21.5%) as a result of tips from tax authorities, 169 (25.1%) based on tips from clients and 183 (27.2%) based on tips from others.

6 SBPA opened a total of 674 disciplinary investigations from 2005 to 2009 (SBPA Annual Report, various issues). Two hundred and ninety-five or 44% of these cases led to the issuing of disciplinary sanctions. Thirteen auditors received multiple sanctions. From 2005 to 2009 the average number of qualified auditors was 4083. These auditors are allowed to audit both private and public companies but not all of them are audit partners. In fact, only about 30% of qualified auditors working in the Big 4 audit firms in Sweden are audit partners. Individual decisions in disciplinary cases can be found using the following link: http://www.revisorsnamnden.se/rn/search/praxis.html.

7 Receiving the name of sanctioned auditors is free of charge. By telephoning or emailing the SBPA, the names of sanctioned auditors are received within a short period of time. The Chief Legal Advisor at SBPA informed us that they receive telephone calls from clients weekly asking for this information. Typically, the request for this information peaks during periods when general meetings are held.

8 Sundgren and Svanström (2013) investigated 267 out of a total of 274 disciplinary sanctions issued in the period 2005–2009 and showed that 68 sanctions (25%) were related to auditor reporting.

9 The model includes net income to total assets, total liabilities to total assets and the current ratio. The ratios include a number of observations with extreme values. We winsorized the ratios with 1% in each tail before the probability was calculated.

10 A company can appoint a person or an audit firm as its auditor. The data only includes observations when a person is hired as the auditor. We do not have any information about the auditor-in-charge of an assignment if an audit firm is formally hired. A person is appointed by on average 87.0% of the companies over the 2006–2012 period. Large clients more commonly appoint an audit firm.

11 Publicly available databases do not include audit firm affiliation on an annual basis. However, based on the sample used to test hypothesis three, we constructed a file with the auditor's identity and audit firm for the years 2006–2011. This data included 16,273 auditor-years. In 120 of these, the auditor had signed reports for two or more audit firms. In those cases, we assumed the auditor to be affiliated to the firm at which the majority of the reports had been signed. Thirteen auditor-years were excluded because the majority rule was not applicable.

12 We also attempted to exclude all observations that filed for bankruptcy less than 180 days after the balance sheet date. This left us with 3000 observations and the results are qualitatively similar to those reported in Table 7.

13 The Pearson goodness of fit and the deviance goodness of fit were significant at the <.01 level for the full sample and the sub-samples with Big 4 and non-Big 4 auditors, suggesting that a Poisson regression is inappropriate.

14 We also run the regressions with the logarithm of the total sales of all clients instead of total assets. These results are qualitatively similar.

15 We also estimated the models with interactions between Big 4 and SANCTION as well as Big 4 and AfterSANCTION as an alternative to the analyses of the sub-samples of Big 4 and non-Big 4 auditors in Table 4. The results are qualitatively similar.

16 We winsorize AvSOLV and AvASSETS at 1% in both tails in the main analyses. The results are qualitatively similar when the variables are trimmed.

17 With the logarithm as the dependent variable, the change can be calculated as and in our case b₁ is the coefficient of AfterSANCTION (see Cameron & Trevedi, 2010, p. 88).

18 We also attempted to estimate a model with interactions between Big 4 and SANCTION as well as Big 4 and AfterSANCTION. The conclusions that can be drawn from this model are very similar to those from the separate analyses of the firms audited by Big 4 and non-Big 4 firms: the coefficient of AfterSANCTION is insignificant, suggesting that a sanction does not have any significant impact on the salary in non-Big 4 firms. The sum of the coefficients of AfterSANCTION and Big4*AfterSANCTION is negative and significant at the .10 level (p-value .052), suggesting that auditors at Big 4 firms have a lower salary after the sanction. Finally, the coefficient of Big4*AfterSANCTION is negative and significant at the .05 level (p-value .018), which supports H2b.

19 We winsorize AvSOLV and AvASSETS at 1% in both tails in the main analyses. The results are qualitatively similar when the variables are trimmed. For example, the coefficient (p-value) of AfterSANCTION in the analysis of the sub-sample with auditors at Big 4 firms is −.144 (.048).

20 The results are also insignificant when the dependent variable is trimmed 5% in each tail, that is, the centile with the smallest and largest values of LnW_t − LnW_t−1 were omitted.

21 We use the ado file for two-way clustering in Stata written by Guan and Petersen. The file is available at: http://www.kellogg.northwestern.edu/faculty/petersen/htm/papers/se/se_programming.htm (retrieved January 2015).

22 As an alternative way to test H3b, we estimated a model with interactions between Big4 and SANCTION as well as Big 4 and AfterSANCTION. We then studied the average marginal effects from the logistic regression models and the results do not support H3b. Thus, results with models including interactions are qualitatively similar to the ones in Table 6.

23 We classified audit reports with going concern opinions into three groups: emphasis of matter opinions, qualified/adverse opinions and reports that were difficult to classify. The final category comprises 43 observations in which the audit reports included remarks suggestive of a going concern opinion, but the statements were much more imprecise than the examples the standard provides. This category is not classified as going concern opinions in the main analyses. However, the results are qualitatively similar when the observations are classified as going concern opinions.

Additional information

Funding

Tobias Svantröm and Stefan Sundgren acknowledge financial support from Jan Wallanders och Tom Hedelius Stiftelse [grant numbers P2014-0088:1 and P2015-0104:1].