The effect of audit partner style on financial statement comparability

ABSTRACT

Several studies provide evidence that auditors play an important role in the implementation of accounting comparability. We extend this line of research by examining whether comparability in accounting earnings between two firms is affected by auditor industry expertise measured at the partner level. Using 45,741 pairs of firms over the period 2003–2016, we find that the accounting comparability between two firms in a pair is higher when they are audited by the same expert partner than when they are audited by the same non-expert partner. The findings suggest that the degree of accounting comparability within a partner’s clientele is influenced by the industry expertise of the individual audit partner. Specifically, expert partners are more likely to provide consistent implementation of auditing standards and enforcement of accounting standards across their client base. These findings are robust to various sensitivity tests including using alternative measures of comparability, industry expertise and pair-matching methods.

KEYWORDS:

• Accounting comparability
• audit partner
• industry expertise

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. By 2017, several countries have required engagement partner signature or the disclosure of partner identity (e.g., Australia, Taiwan, Sweden, Germany, France, Belgium, and Luxembourg).

2. https://pcaobus.org/News/Releases/Pages/SEC-approves-transparency-Form-AP-051016.aspx.

3. The United States has mandated the disclosure of partner identity beginning from 2017, and Korea has also mandated the disclosure of partner name for financial statements with the reporting period ending after 15 December 2018. Refer to Carcello and Li (2013) and Public Company Accounting Oversight Board (PCAOB) (2015) for further details on the regulation and controversy surrounding the regulation.

4. To respond to the demand for enhanced comparability of accounting information across countries, 166 jurisdictions have adopted IFRS (www.ifrs.org) and related studies suggest that the adoption increases accounting comparability. For example, Barth et al. (2012) find that non-U.S. firms adopting IFRS show greater comparability than U.S. firms which apply their domestic accounting standards.

5. According to De Franco, Kothari, and Verdi (2011), comparability is measured by the average difference between a firm’s earnings-return relationships within the industry. De Franco, Kothari, and Verdi (2011) state that economic events are mapped into financial statement through accounting system. Therefore, if two firms have comparable accounting systems, they produce similar financial statements for a given set of economic events. In the construction of comparability measure, they use stock return as a proxy of economic events and earnings as a proxy for financial statement outcomes.

6. Note that the definition of the accounting comparability in De Franco, Kothari, and Verdi (2011) and several subsequent studies is different from that used in Francis, Pinnuck, and Watanabe (2014). De Franco et al.’s (2011) definition refers to the similarities of the effect of accounting earnings on returns between a specific firm and the industry average. In contrast, Francis et al.’s (2014) definition refers to the difference in the accounting accruals between two firms in a pair. Thus, De Franco, Kothari, and Verdi (2011) measure is estimated for each firm observation but Francis, Pinnuck, and Watanabe (2014) measure is estimated for each pair of firms.

7. Although it is not directly related to the notion of comparability dealt in this study, there are several studies which examine the effect of audit partners on various settings (e.g., partners’ experience, rotation). Please refer to Bae, Choi, and Lee (2014), Chi et al. (2015), Park and Chung (2015), Wang, Yu, and Zhao (2015), and Choi et al. (2019a) for such examples.

8. Upon this assumption, engagement partner information can be valuable for information users such as investors and financial analysts to evaluate financial statement. In the process of making an investment decision in a certain industry, investors need to compare financial data of one company with its alternatives to evaluate them. If the audit partner of firm A is the same as the partner of firm B, information about that partner could be useful for investors to understand the nature of accounting treatments of the two firms. Therefore, accounting comparability at the engagement partner level is important for information users to enhance the effectiveness and efficiency of the evaluation procedure.

9. As Bedard (1991) suggests, the quality of audit decision making is evaluated with three measures, which are consistency with professional standards, consistency with the firm’s standards, and consensus among auditors, under assumption that audit quality increases with consistency with which standards are applied.

10. Analyses of partner-level accounting comparability require the construction of firm-pairs within the same audit firm. If we compare firms in a pair which are audited by two different audit firms, the difference in accruals between the two client firms may reflect audit firm-level characteristics rather than partner-level characteristics. Thus, we construct firm-pairs within the same audit firm which makes it impossible to compare Big 4 and non- Big 4 combinations. As a result, empirical results show the relative differences of audit partners against other partners working in the same audit firm.

11. We additionally use total accrual difference as an alternative dependent variable. However, in all empirical analyses, we do not find significant coefficient for our test variables when we use the total accrual difference. While both discretionary accrual and total accruals are the joint outcome of client firms’ characteristics and audit partners’ choices on appropriate accounting treatments, total accruals are more likely to be related to clients’ operating characteristics than discretionary accruals are. Given that the role of auditor is restricting discretionary accruals choices by managers and persuading the client firms to choose the more appropriate accounting methods, discretionary accrual difference is a more proper measure for accounting comparability between two firms in a pair.

12. There is a lack of theoretical background or empirical evidence for the appropriate control variables which should be included in a regression determining earnings comparability (Lang, Maffett, and Owens 2010). Sohn (2016), the only study that uses financial statement comparability measure of De Franco, Kothari, and Verdi (2011) as the dependent variable, controls for asset size, leverage, regulated industry dummy, intangible asset, standard deviation of earnings, book-to-market ratio, operating cycle, and capital intensity in his 1^st stage regression which determines comparability. Meanwhile, Francis, Pinnuck, and Watanabe (2014) control for variables widely used in literatures related to the similarities of earnings. They control for economic fundamentals such as volatility of operation, and propensity to manage earnings such as market-to-book ratio and leverage. We also control for the variables associated with economic fundamentals of firms following Francis, Pinnuck, and Watanabe (2014), the study which has more similar research settings with this study.

13. Given that we use Korean data, one may question the generalizability of the empirical findings of this study. Since Korea is known as one of the countries with relatively poor investor protection and weak legal liability of auditors (La Porta, Lopez-De-Silanes, and Shleifer 2006; Brown, Preiato, and Tarca 2014), it is likely that Korean auditors provide relatively low-quality audit service, compared with other countries with strong legal regime. Therefore, if we find any significant effect of audit partners in Korea, it is evidence that audit partner plays an important role even in countries with weak legal regime. Thus, the effect documented in this study is likely to be more pronounced in countries with strong legal regime, where auditors provide higher-quality audit service and auditors are perceived to be more influential than those in weak legal regime countries.

14. Changing the cut-off value to 5 or 8 does not qualitatively change our empirical findings.

15. We repeat our empirical analyses with the sample including the firm-year observations which an audit partner audited only one client firm in the year. Unreported empirical results are qualitatively similar. In this study, the expression ‘qualitatively similar’ implies that the empirical results for the test variable is significant in the same direction and that all the inferences remain the same.

16. Following Choi and Bae (2016), we use KSIC one-digit industry code for the purpose of making firm-pairs. With the restriction of two-digit industry code, only 6,860 firm-pairs survive in the pair-matching process because the average number of firms within an industry based on two-digit KSIC code is much smaller than that based on one-digit industry code. Except for this case, other industry classification used in this study is based on KSIC two-digit industry code.

17. The number of pairs are much larger than the number of clients in each year. For example, if there exist 3 (4 or 5) clients in an industry and all of them are audited by the same audit firm, the number of pairs in the industry is 3 (6 or 10). The exact number of pairs can be calculated by using the notation of nC2, where n represents the number of clients in an industry audited by the same audit firm in the year and C represents ‘Combination’. Thus, if none of the two clients in the industry are audited by the same audit firm in a year, the number of pairs is zero and thus none of the observations in the industry for the year are used in the analysis.

18. The mean values of IndExpert1 and IndExpert2 are not exactly 25% and 10%, respectively, because we define these variables before excluding the 2,202 firm-year observations in which an audit partner audited only one client firm in a year. Since the remaining firm-year observations include only firm-year observations in which an audit partner audited at least 2 client firms in a year, the average proportion of expert partner increases in the final sample.

19. Note that we match a pair of client firms audited by the same auditor. Thus, although about 80 percent of the audit partners work in Big 4 audit firms (as reported in Panel A), the number of pairs audited by Big 4 audit firms increases to 96 percent. The difference arises because Big 4 audit firms audit more clients than non-Big 4 audit firms.

20. Exceptionally, we notice that the standard deviations of mb_diff and std_growth_diff are relatively large, implying the possibility that these two variables may have some outliers. Our empirical results are qualitatively similar when we repeat our analyses after excluding these two variables individually or altogether.

21. In terms of economic significance, the estimated coefficient on Same_partner, 0.35 indicates that, on average, clients with the same audit partner exhibit a 36% higher level of comparability than clients with different audit partners (the mean value of DA_diff is 0.103 in Table 3).

22. Considering that the mean value of DA_diff is 0.103, the negative coefficients on Same_partner*IndExpert1 (coefficient = −0.52) and Same_partner*IndExpert2 (coefficient = −1.00) imply that comparability (DA_diff) increases by about 54% (103%) when the two firms are audited by the same expert partner at the top quartile (decile).

23. The result using IndExpert2 is qualitatively similar to the tabulated one. We do not separately tabulate the results for simplicity.

24. During the sample period, 9 partners migrated from one Big N audit firm to the other Big N audit firms (223 firm-years), 8 partners changed their audit firms from Big N to non-Big N (136 firm-years), and 1 partner changed his or her audit firm from non-Big N to Big N (10 firm-years). There is no partner who migrates from non-Big N to non-Big N audit firm.

25. In extracting firm-pairs audited by the same partner who changed audit firms, we match a pair of firms audited by different audit firms and audited in different years. Therefore, this analysis contains a potential limitation that firms from different years are matched in a pair.

26. In our main analysis, we construct pairs in the same industry and year, and audited by the same audit firm. However, in this analysis, we match a pair of firms in the same industry, but audited by different audit firms and audited in different years. We do not include firm-pairs which are composed of the same client firm. That is, we exclude cases in which a partner audited the same client firm after moving to a different audit firm from our migration sample. It leads to, among 3,961 firm-pairs, 60 firm-pairs (1.51%) which are audited by the same partner.

27. We cannot report the regression results using IndExpert2 since there is no observation that is classified under the definition of IndExpert2.

28. For example, if different audit firms adopt a similar partner assignment strategy, the partners who change audit firms are likely to be assigned to clients with similar characteristics even after the change.

29. The absolute value of the difference between the predicted earnings of firm i and firm j is not aggregated at the firm level within the industry since we measure comparability at the firm-pair (firm i and firm j) level. Also, we do not multiply (−1) by the average of the absolute value of the difference between the predicted earnings of firm i and firm j. Therefore, smaller difference of predicted earnings between two firms in a pair indicates greater comparability between two firms in this study.

30. We find that the statistical significance of the coefficient on the interaction term between Same_partner and IndExpert1(2) becomes weaker in which IndExpert1(2) is defined as an indicator variable which equals 1 if the market share of individual audit partner based on audit fee at the top quartile(decile).

31. Exceptionally, the result using partner migration sample becomes insignificant when we use the continuous market share measure.

32. Consistent with this finding, it is relatively difficult to build up industry expertise for audit partners in non-Big N audit firms. Craswell, Francis, and Taylor (1995) states that a fee premium for industry expertise reflects demand for higher quality Big N audits and product differentiation within Big N audit firms. DeFond, Francis, and Wong (2000) also report that industry expertise of Big N is different from that of non-Big N, because industry expertise of non-Big N capture market share through providing lower price audit. Therefore, even though a non-Big N auditor has large market share and thus it is classified as an industry expert, the non-Big N auditor is less likely to provide a higher-quality audit service.