Initial Evidence of U.S. Audit Partner Identification in Form AP Shows Limited Impact on Audit Quality Free

After six years of contentious debate, the Public Company Accounting Oversight Board (PCAOB) adopted Rule 3211 (Auditor Reporting of Certain Audit Participants), which requires disclosure of audit engagement partner names in Form AP for audit reports issued on or after January 31, 2017 (PCAOB 2015b). The PCAOB argued that disclosure of partner identities would increase partners' sense of accountability, citing important psychological effects in “the act of signing itself,” and drawing parallels to the CFO and CEO certifications of the Sarbanes-Oxley Act of 2002 (PCAOB 2009, 6). The PCAOB also cited improvements in the transparency of audits, which would provide useful information to outside parties and encourage accounting firms to improve the quality of all engagement partners (PCAOB 2009, 5). Institutional investors were in favor of the proposed rule because they could better evaluate audit quality at the individual level rather than the audit-office level (e.g., CII 2009). In contrast, practicing CPAs largely opposed the ruling, stating that partners already have a strong sense of accountability and audit quality was already sufficiently high given firms' quality control systems that are subject to a rigorous inspection and enforcement process (e.g., Deloitte 2009; EY 2009; Grant Thornton 2009; KPMG 2009; PwC 2009).

While the initial proposal suggested that engagement partners sign their name in the audit report, the PCAOB eventually compromised in the final ruling by implementing a separate disclosure of partner identities in Form AP with no mention of the partner in the report. Instead, registered accounting firms submit Form AP directly to the PCAOB within 35 days of the date that the auditor's report is first included in a document filed with the SEC (PCAOB 2015b). Given the contentious debate and the removal of “the act of signing” in the final ruling, Cunningham, Li, Stein, and Wright (2019; hereafter CLSW) examine whether the new Form AP disclosure had immediate effects on the quality of audit services in the U.S.¹ Such analysis enhances our understanding of this important regulatory rule and its impact on practice.

The PCAOB's initial Concept Release on Requiring the Engagement Partner to Sign the Audit Report (PCAOB 2009) received 23 comment letters with significantly different views. As previously described, institutional investors were largely in favor of the proposal, whereas practicing CPAs were largely against it. In addition to arguing that accountability and audit quality were already sufficiently high, accounting firms also stated that partner name disclosure in Form AP may give financial statement users a false impression that audits are the responsibility of one individual rather than a team effort.² Moreover, accounting firms strongly argued that disclosure in the audit opinion could change the perceived legal liability of the audit partner, causing audit partners to perform unnecessary procedures to mitigate potential increases in personal risk (e.g., Grant Thornton 2009; PwC 2009).

Following the initial concept release, the PCAOB went through two iterations of proposed rulings. The first proposal recommended that auditors continue signing the firm name but identify the partner name in a separate sentence within the audit report with additional disclosure in the firm's annual Form 2 (PCAOB 2011). The second proposed ruling retained partner identification in the audit report but removed the Form 2 reporting requirement, citing that this disclosure would not be timely and would impose additional costs on accounting firms (PCAOB 2013). Consistent with the initial concept release, institutional investors and practicing CPAs continued to express divided feedback with each of these iterations (e.g., CII 2012; Deloitte 2012; EY 2014; McGladrey 2014; PwC 2012). Moreover, there was growing speculation among board members for the proposed purpose and form of audit partner identification.³

On June 30, 2015, nearly six years after the initial concept release, the PCAOB solicited comments on a proposal that would remove the partner identity from the audit report altogether and instead require accounting firms to file the partner's name in a newly-created “Form AP” (PCAOB 2015a). In contrast with previous proposals, practicing CPAs were more supportive of this approach, while institutional investors felt that it lacked the accountability effect that would inherently increase audit quality (e.g., Deloitte 2015; Grant Thornton 2015; KPMG 2015; CII 2015). For example, disclosure in Form AP removed the timely transparency of partner names in the audit report and it removed psychological effects related to “the act of signing” the audit report (PCAOB 2009, 6).⁴ Ultimately, this version of the PCAOB's proposal for audit partner identification culminated in the PCAOB adopting Rule 3211 on December 15, 2015, which was subsequently approved by the SEC on May 9, 2016. The requirements in Rule 3211 became effective for audit reports issued on or after January 31, 2017. On the final adoption date in 2015, Steve B. Harris, former PCAOB board member, provided the following commentary in the PCAOB Open Board Meeting (Harris 2015):

Concurrent research compares audit quality before and after Form AP and concludes that audit quality improved as a result of Form AP (e.g., Burke, R. Hoitash, and U. Hoitash 2019; Dao, Xu, and Liu 2019). However, it is possible that audit quality would improve during the same time period regardless of whether Form AP was implemented, whether it be due to new technologies, continuous improvements in audit quality, quality control processes, or ongoing PCAOB inspection and enforcement mechanisms. These “normal course of business” improvements would make it difficult to attribute baseline changes surrounding Form AP implementation to the Form AP disclosure itself. An alternative approach to mitigate this concern is to compare changes across two types of audits: (1) audits where Form AP is expected to have a significant impact (treatment group), and (2) audits where Form AP is not expected to have a significant impact, yet these audits are performed in a similar macroeconomic environment and time period (control group). The Form AP effect should be the portion of changes observed in the treatment group that are above and beyond any changes observed in the control group. This technique is often referred to as “difference-in-differences” analysis because we examine whether changes between pre- and post-Form AP are different between treatment and control groups (Roberts and Whited 2013; Atanasov and Black 2016).

A critical component to difference-in-differences analysis is being able to identify a suitable control group. The treatment and the control group should exhibit “parallel trends” in audit quality in the years leading up to the treatment effect so that we can assume that the path taken in the control group could have also occurred in the treatment group absent the Form AP disclosure, all else being equal. Since the effective date of Form AP disclosure was the same for all audits of public companies, this was a difficult but not impossible task. We describe three different control groups used in our study, each of which has supporting details provided in CLSW.

For our first control group, we exploited the common practice of disclosing audit partner identities at the annual shareholders meeting (PCAOB 2009). We argue that companies previously disclosing the partner identity at the annual meeting and disseminating this information on the company's website should not be significantly impacted by Form AP. Similar to Form AP, the disclosure of the audit partner at the annual meeting occurs after the audit report date and the information is accessible for those seeking to find it (company website versus PCAOB website). Thus, the behavior of these partners should already reflect the accountability effect of being publicly disclosed.

We collect data from company websites to identify companies that disclose the identity of audit engagement partners during the fiscal year prior to the Form AP effective date. If the PCAOB rule achieves its intended purpose of increasing accountability for the audit partner, leading to improved audit quality, we expect the rule to have a greater accountability impact on the group of partners that were not previously identified to the public in annual meeting archives on companies' websites. In tests using the early adopters control group, audit partners identified in annual meeting archives before Form AP represent our control group and partners not previously identified in annual meeting archives before Form AP represent our treatment group. In both the treatment and control group, we use companies with audit reports filed between January 31, 2017 and June 29, 2017, to avoid the confounding effects of additional required disclosures in Form AP (i.e., disclosure of other firms participating in the audit) that became effective June 30, 2017.⁵

Next, we identify a control group of companies with an audit report issued immediately before the Form AP effective date of January 31, 2017. These companies should exhibit similar trends in audit quality as companies filing an audit report immediately after the effective date of Form AP, without the accountability impact of being identified in Form AP disclosures. In these tests, we use these “pseudo adopter” companies with audit reports filed between July 1, 2016 and January 30, 2017 as the control group and compare them to companies with audit reports filed between January 31, 2017 and June 29, 2017 as the treatment group.

Finally, as an alternative to the two U.S. control groups, we consider companies in the United Kingdom (U.K.) as the third control group. The U.K. had already enacted audit partner identification prior to the beginning of our sample period. Specifically, the Companies Act in the U.K. required the audit engagement partner to sign his or her name to the audit report for fiscal years ending in April 2009 or later (Carcello and Li 2013). We are not aware of significant changes to U.K. audit or accounting standards in the year before or after the Form AP effective date in the U.S. Therefore, any changes in U.K. audit quality during this period should represent global changes in audit quality that occur in the normal course of business. In tests using the U.K. control group, we classify companies in the U.K. as the control group and companies in the U.S. as the treatment group with both groups having audit reports filed between January 31, 2017 and June 29, 2017.⁶

Since actual audit quality is unobservable, we estimate or “proxy” for audit quality using a wide variety of measures commonly used in academic research, all of which are based only on publicly available data. First, we proxy for audit quality using discretionary accruals (Kothari, Leone, and Wasley 2005; Carcello and Li 2013; Gutierrez, Minutti-Meza, Tatum, and Vulcheva 2018). The intuition behind this measure is that a portion of accruals (i.e., the difference between cash flows from operations and reported net income) is expected as part of normal operations, while the remaining portion relates to discretionary changes to these accruals. Higher (lower) audit quality should result in lower (higher) discretionary accruals. Second, because some changes in discretionary accruals may still be allowed “within GAAP,” a restatement would be a clear indicator of material errors in a prior filing that would indicate lower quality audits. Unfortunately, at the time of data analysis for CLSW, there was not yet enough time to examine whether the audited financial statements immediately before or after Form AP were eventually restated. Instead, we use the “F-score,” which is the predicted probability of a material accounting error based on financial and non-financial characteristics (Dechow, Ge, Larson, and Sloan 2011). Dechow et al. (2011) created this measure based on characteristics that are predictive of AAERs and material fraud. Higher (lower) values for the F-score should indicate lower (higher) audit quality. Third, we consider material weaknesses in internal control over financial reporting as a proxy of audit quality. Since the reporting of a material weakness is not necessarily revealing of audit quality in its own right, we first run a prediction model for the likelihood of reporting a material weakness and then compare whether the actual internal control opinion agrees with the predicted internal control opinion (Ge, Koester, and McVay 2017; Bhaskar, Schroeder, and Shepardson 2019). This measure provides a better estimate of the quality of audit services; for example, we can assess whether the auditor requires disclosure of a material weakness in settings in which a material weakness is more likely to occur.

In additional tests, we also consider how stock returns and company losses are related, commonly referred to in academic research as “timely loss recognition” (Basu 1997). We also consider whether the company meets or just (barely) beats analyst forecasts, which is a sign that the company may have manipulated earnings to narrowly achieve the benchmark (Reichelt and Wang 2010; Reid, Carcello, Li, and Neal 2019). Further, we examine whether the company reports impairment losses or special items relative to expectations based on the nature of the assets being held, as well as whether the company records adequate valuation allowances for deferred tax assets (Laurion, Lawrence, and Ryans 2017). The combination of three control groups, nine different dimensions of audit quality, cross-sectional tests based on company characteristics (e.g., size, complexity, bankruptcy risk, etc.), and robustness tests resulted in 156 difference-in-differences tests disclosed in CLSW.

Before considering tests involving the control groups, our results suggest that audit quality trended upward on average when comparing the year before and after implementation of Form AP disclosures as illustrated in Figure 1, Panel A.

However, as discussed previously, this increase in audit quality between periods may not necessarily be attributed to disclosures of partner identities in Form AP. To determine whether any portion of this increase can be attributed to Form AP, we compared changes in audit quality (i.e., pre- versus post-Form AP) for each of the treatment groups to changes in audit quality for each of the control groups. As a reminder, these control groups should not have been affected by Form AP but should illustrate changes in audit quality that occur in the normal course of business. Therefore, any change in the treatment group that is above and beyond the change in the control group is likely attributed to the new disclosures in Form AP. But if the changes are not statistically different from one another, it suggests that changes observed pre- to post-Form AP are likely due to changes in the normal course of business.

Across all three of our treatment groups, we most often find that the increase in audit quality after the Form AP effective date is not statistically different than the increase in audit quality for each of the control group of companies. This general finding can be observed in Panel B of Figure 1, which shows that both treatment and control groups increased at similar rates after the implementation of Form AP. Only a minor number of tests (six tests, or less than 4 percent) identify an improvement in audit quality that may be attributable to Form AP; however, those situations are limited to specific dimensions of audit quality and to specific types of engagements with weak internal controls and higher risk of litigation.

Broadly, these results suggest that improvements in audit quality around the Form AP effective date may be due to general trends over time, due to any number of factors unrelated to Form AP.

The rationale behind Rule 3211 and the development of Form AP was to improve audit quality by “enhancing and influencing a leader's sense of individual accountability and acceptance of responsibility for a team effort he or she has led by signing his or her name to a most commonly reviewed report” (Harris 2015). Public accounting firms argued that accountability was already substantially high and that the Form AP requirement would not significantly increase it (e.g., Deloitte 2009; EY 2009; Grant Thornton 2009; KPMG 2009; PwC 2009). Using a difference-in-differences design around the Form AP effective date, we fail to find consistent, robust evidence of an increase in audit quality when comparing treatment groups to control groups.

Our failure to find robust and consistent evidence of an on average immediate change in audit quality attributable to U.S. partner identification may be due to several factors, statistical and non-statistical. Statistically, we recognize that a lack of significant results does not guarantee that there really is “no effect.” It could be that that the sample sizes are not large enough to detect a difference between the two groups. While we believe that we have provided evidence in CLSW to assuage concerns about low sample sizes, we cannot eliminate the concern completely.

Unrelated to statistics, at least two factors may be driving our failure to find evidence of an increase in audit quality attributable to partner identification in Form AP. First, accounting firms argued that partner accountability was already sufficiently high prior to mandatory disclosure, in which case we would not expect to see any additional improvement in audit quality attributable to partner identification in Form AP. Second, the final adoption of Form AP required audit partner disclosure outside of the audit report, which reduces partners' perceived legal liability and removes psychological effects specific to “the act of signing itself” (PCAOB 2009, 6). As a result, disclosure in Form AP may not be a strong enough identification mechanism to change partners' sense of accountability. We cannot disentangle or completely eliminate these alternative explanations using publicly available data.

We recognize that CLSW is only one of many studies that will inevitably investigate the effects of Form AP. To be clear, our study only provides limited evidence on the immediate impacts of Rule 3211. We hope that future researchers will examine other possible benefits (and costs) of Form AP disclosure. For example, in CLSW, we suggest that future research may examine:

• longer term impacts of Rule 3211;

• whether investors make trading decisions based on partner name disclosure;

• whether audit committees make audit partner or audit firm selection decisions based on public disclosures of audit partners' other clients (including the number of other clients);

• whether companies seek out or avoid partners with a reputation for “tough” calls (e.g., a material weakness or impairment);

• whether partners perform higher quality or more conservative audits for other public clients following negative public disclosures where the partner's identity may be investigated, such as restatement announcements;

• whether public accounting firms' employees change their behavior for seeking out partnership roles that require Form AP identification;

• whether there is any change in the personal liability rate for audit partners; and

• whether there are any unanticipated costs to partner disclosure (e.g., personal threats).

We encourage the profession to consider a wide variety of research types and research methods when doing post-implementation economic analyses and weighing the pros and cons of Form AP partner identification. We also encourage the profession to work with a wide variety of research teams to analyze audit quality indicators or other benchmarks of audit quality based on data internal to the accounting firms, which would allow researchers to triangulate results based on publicly available proxies of audit quality. Stone (2018, 108) states, “scientific ‘truth' exists only in the accumulation of multiple investigations, ideally across methods and investigators.” Practitioners evaluating the evidence that is likely to follow from Form AP might consider:

• What was the sample composition of the study? Look for descriptive statistics such as industry composition, company size, company profitability, company complexity, audit firm type, etc. Studies with descriptive statistics that roughly mirror the general U.S. population can likely be generalizable to a wide variety of companies. Studies with limited samples can still be useful but should only be generalizable to similar types of companies. In CLSW, the samples range from using all available companies with publicly available data to using only companies in the S&P 1500.⁷

• Are results robust to different types of samples or measurements of key variables? In CLSW, audit quality was one of the main purposes of the study. Between the main analyses and additional tests (including Online Appendices), we had nine different measures of audit quality and three different samples for a total of 156 tests. To understand a meaningful relationship, and not just a spurious one, look for a consensus of results across multiple measures and methods.

• Was everyone in the sample subject to the rule change at the same time; if so, how do the researchers separate the effect of the “event” from general trends occurring from the “normal course of business”? In our case, we identified early disclosers, filers immediately before the effective date, and filers in another country. Even though each of these samples may have specific limitations, inferences across all three should provide strong evidence of an effect, if one exists.