Korean evidence on auditor switching for opinion shopping and capital market perceptions of audit quality

ABSTRACT

This study examines whether auditor switching for opinion shopping (OS) influences investors’ perceptions of audit quality (proxied by earnings response coefficients; ERCs). To the extent that audit quality deteriorates after firms switch auditors for OS, investors are less likely to rely on these firms’ audited earnings. This may result in a negative association between auditor switching for OS and ERCs. However, when the intention for auditor switching is undetected by investors, ERCs would do not differ depending on whether firms switch auditors for OS or not. We find that firms switching auditors for OS have lower ERCs than non-switching firms or those that switch for other reasons, suggesting that investors perceive auditor switching for OS as evidence of impaired auditor independence.

KEYWORDS:

• Opinion shopping
• auditor change
• auditor independence
• audit quality
• capital market perceptions

Disclosure statement

No potential conflict of interest was reported by the authors

Data availability

All data are publicly available from the sources mentioned in the text.

Notes

1. These regulatory bodies continuously discuss whether mandatory auditor change or rotation systems can prevent auditor switching for OS and call for related research in academia on this issue.

2. In this paper, a non-clean audit opinion refers to a MAO defined in Korean Standard on Auditing (KSA) No. 705. We interchangeably use non-clean (clean) and modified (unmodified) opinions.

3. By contrast, a few studies fail to find a significant relation between non-audit services and ERCs (Higgs and Skantz 2006; Lim and Tan 2008; Ghosh, Kallapur, and Moon 2009).

4. In December 1989, Korea adopted the auditor designation system, which applies to specific firms for which investor protection is strongly required. The Korean Securities and Futures Commission, which is a regulatory body equivalent to the SEC in the United States, designate auditors for firms that did not appoint their auditors for the year, unduly changed the auditors, committed accounting fraud in recent periods, or planned for initial public offer, and so on (Financial Supervisory Service 2017). Considering that these firms involuntarily change auditors, we remove them from our sample to avoid measurement errors.

5. For reference, in a case where the use of the going concern basis of accounting is appropriate but where a related material uncertainty exists, the auditor shall express an unmodified audit opinion with an explanatory section indicating the existence of the material uncertainty related to the going concern issue (KSA No. 570). We cannot identify the details of these explanatory paragraphs because the KIS-VALUE database does not provide detailed information on them. However, to the extent that the audit opinion is unmodified, we view that the attached going-concern issue may not significantly trigger the auditor-switching decisions motivated by OS.

6. Our findings do not alter materially when we apply the alternative cut offs including −0.1%, −0.2%, −0.3%, −0.4%, and −0.6%. When we use the cut offs of 0, −0.7%, −0.8%, −0.9%, and 1%, our inferences are also consistent with those from the main results, albeit with weak statistical significance.

7. In our sample, 36 observations have a MAO (i.e., MAO = 1): 17 observations have a qualified opinion, two observations have an adverse opinion, and 17 observations have a disclaimer of opinion. These statistics reveal that auditors issue MAO less frequently in Korea than in the United States (Chung et al. 2019) and in the United Kingdom (Lennox 2000).

8. Except for the correlation between E and C_E, none of the other correlations are greater than 0.4. Thus, it is less likely that our regression results are seriously affected by multicollinearity.

9. Our matching procedure is based on the auditor-switching sample to control for the systemic differences in firm and auditor characteristics between switching and non-switching firms.

10. Specifically, we limit the sample to have an auditor change for year t and to have all data for the next consecutive two years. As a result, we have 22 and 406 observations for OS switching and non-OS switching samples, respectively. The ERC of the OS switching sample is statistically insignificant for t, t + 1 and t + 2, although the difference in t and t + 1 is statistically significant at the 10% level (F-statistics = 3.36). The statistical insignificance may be caused by the small sample size given the sample restriction.

11. When we further divide firms switching auditors to non-Big 4 into two groups depending on whether the incumbent auditor is Big 4 or non-Big 4, we find inconsistent results. The negative market reaction is more prevalent for OS-motivated auditor change from non-Big4 to non-Big4 auditors. However, there are very small number of switching observations from Big4 to non-Big4 (12.33%, N = 9), which may influence our results. In addition, Lawrence, Minutti-Meza, and Zhang (2011) argue that the inability of non-Big 4 firms to obtain affordable insurance coverage may increase the audit effort of non-Big 4 firms relative to Big 4 firms because smaller audit firms have difficulty to obtain a similar level of backing from insurance companies. This suggests that ex-post audit quality may vary among non-Big4 firms, making investors perceive that the opportunities for OS are higher for the auditor changes between non-Big 4 firms.

Additional information

Funding

This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2019S1A5A8032900)