Comments of the Auditing Standards Committee of the Auditing Section of the American Accounting Association on PCAOB Staff Consultation Paper, Auditing Accounting Estimates and Fair Value Measurements: Participating Committee Members: Free

Dear Board Members:

The Auditing Standards Committee of the Auditing Section of the American Accounting Association is pleased to provide comments on the PCAOB Staff Consultation Paper “Auditing Accounting Estimates and Fair Value Measurements.”

The views expressed in this letter are those of the members of the Auditing Standards Committee and do not reflect an official position of the American Accounting Association. In addition, the comments reflect the overall consensus view of the Committee, not necessarily the views of every individual member.

We hope that our attached comments and suggestions are helpful and will assist the Board. Please feel free to contact the subcommittee chair should the Board have any questions about our comments and suggestions.

Respectfully submitted,

Auditing Standards Committee

Auditing Section, American Accounting Association

The Committee commends the PCAOB (“the Board”) for considering an update to the auditing standards dealing with the auditors' responsibility with respect to audits of fair value estimates and other types of accounting estimates. The following presents a number of specific comments or suggestions, organized by the questions posed by the Board in the Staff Consultation Paper “Auditing Accounting Estimates and Fair Value Measurements” (the Consultation Paper).

• Question 1: 

  Does the information presented above reflect aspects of current practice? Are there additional aspects of current practice, of both larger and smaller firms—including centralized testing, the use of third parties, or specific challenges to auditing accounting estimates and fair value measurements—that are relevant to the staff's consideration of the need for standard setting in this area?

We note that the Board has taken into account the several published academic studies relevant to auditing accounting estimates and fair value measurements. We would like to point out a few more recent academic studies currently in working paper form. Each is a survey and/or interview of experienced auditors with fair value measurement experience. The purpose of each study is to gain insight and codify current practice, as well as identify areas in which auditors could use more guidance or the standards could be clarified: Cannon and Bedard (2014); Glover, Taylor, and Wu (2014a); Glover, Taylor, and Wu (2014b); and Griffith (2014). Where relevant, we reference findings from these studies in our responses to other Consultation Paper questions.

• Question 4: 

  Do accounting estimates and fair value measurements have sufficiently common attributes that the audit procedures should be included within a single standard? Are there limitations to the approach of having a single standard address both auditing accounting estimates and fair value measurements?

We do not believe that combining accounting estimates and fair value measurements into a single standard would be beneficial. The nature of accounting estimates and fair value measurements can be quite different. So too are the skill sets required to provide a reasonable level of assurance. For example, assessing the net realizable value of accounts receivable requires very different tools than assessing the net realizable value of goodwill. Audit engagement teams achieve a reasonable level of assurance of an accounts receivable allowance by considering collection history, credit policies, subsequent cash receipts, and other traditional accepted procedures. Assessing goodwill for impairment can require a very different skill set, one not traditionally found on audit engagement teams, and is likely assigned to in-house valuation specialists or third-party experts whenever the “Step Zero” qualitative analysis fails or the cushion is small. The skill sets are so different there are now specialized Master's programs dedicated entirely to training valuation specialists at the nexus of accounting and finance, such as the M.Acc. Valuation program at Vanderbilt University.

• Question 13: 

  In circumstances where the company uses information obtained from a third party, are there matters—such as information systems at third parties, controls that management has over the work of third parties, and controls at third parties—not currently addressed in AU Sec. 324, Service Organizations, or other standards that the staff should consider?

PCAOB inspection reports find that auditors rely heavily on the work of third-party specialists. Auditors might rely on third-party experts because clients frequently use third-party experts. For example, Cannon and Bedard's (2014) survey of experienced auditors finds that for the most challenging and difficult-to-audit fair value measurement estimates, clients use third-party experts about 60 percent of the time and auditors consult with external (internal) valuation specialists 5.1 percent (99 percent) of the time. While some studies find that valuation experts can increase the reliability of fair value measurements for investors (e.g., Barth and Clinch 1998; Muller and Riedl 2002), the Board might consider cautioning auditors about judgment problems that can occur when using their clients' third-party specialists' reports. For example Salzsieder's (2014) experimental study finds that if there are no regulatory controls preventing them from doing so, managers will opinion shop for third-party experts whose FV estimates enhance the manager's wealth. The Board might consider requiring filers to indicate whether they obtained fair value estimates from multiple third-party experts because Salzsieder finds that when auditors are informed that managers have engaged in opinion shopping, they exercise more scrutiny over the fair value estimates reported. Thus, that research suggests if there is adequate disclosure, the audit process can be effective in curbing management opportunism.

Joe, Vandervelde, and Wu (2014) provide further evidence about the challenges auditors encounter when using the client's third-party expert's report in a high client risk scenario. They found that, when the client's internal control risk was high, auditors' testing of the client's fair value estimate was influenced by the amount of information (i.e., the level of detail) included in the third-party expert's report. When the client's expert's report had a higher amount of quantitative data, auditors focused more audit effort testing the detail and objective inputs, while neglecting to test subjective inputs to the client's fair value estimate than when the client's expert's report had a lower amount of quantitative data. Joe et al. (2014) note that because the failure to test the subjective inputs to the client's fair value measurement is a recurring deficiency in the PCAOB inspection reports, and because audit standards do not support client data driving the nature or extent of auditors' testing, the Board is likely to judge that audit quality is impaired whenever the content of the client's third-party expert's report influences auditors' testing. Evidence from Joe et al. (2014) also suggests that the Board might want to rethink its approach of issuing audit alerts as a way to improve or change the way audits are conducted. In a second experiment, Joe et al. (2014) found that simply reminding auditors that the PCAOB wants them to test more subjective inputs to fair value estimates does not mitigate auditors' tendency to allow the data in the client's expert's report to drive their testing. In that study, after receiving an alert reminding them of the PCAOB's preference for more testing of subjective inputs, the auditors still allocated a higher proportion of audit effort to testing objective inputs when the client's expert's report included a higher amount of quantified data than when the report had a lower amount of quantified data. However, the Joe et al. (2014) study indicates that the PCAOB alerts do change auditors' testing in important ways. Specifically, reminding auditors that the PCAOB wants them to test more subjective inputs to fair value measurements leads auditors to allocate more total hours to test the client's fair value estimate, which leads to more hours to test subjective inputs to fair value estimates and greater use of the CPA firm's in-house valuation specialists.

Research indicates that the Board should consider cooperating with the SEC to ensure that auditors have sufficient access to the underlying data used to support the client's fair value estimates. Cannon and Bedard (2014) and Glover et al. (2014a) report that auditors often are frustrated by an inability to access key data when evaluating the inputs to fair value estimates prepared by third-party experts (e.g., pricing services). For instance, Cannon and Bedard (2014) find that experienced audit experts reported that in 23.2 percent of the most challenging audit cases, the client's third-party expert used a proprietary valuation model, which auditors were prohibited from examining.

While there are no studies that examine directly the impact of the quality of the third-party expert's internal control on audit judgment, studies examining how the client's internal controls influence audit judgment offer some insights. Two studies find that auditors attend to the level of the client's internal controls when making fair value judgments and that attention to the client's internal controls can impact audit effectiveness. Joe et al. (2014) found that when the client's entity-level internal controls were strong, auditors did not differentiate in the nature and extent of testing of the client's fair value estimate contingent on the contents of the third-party expert's report. However, when the client's entity-level controls were weak, the nature and extent of auditors' testing of the client's fair value estimate was influenced by the content of the third-party expert's report. Brown-Liburd, Mason, and Shelton (2014) find that when the client's internal controls are effective, auditors assess the inherent risk of misstatement of the fair value estimates to be low and judge the client's third-party expert to be competent and reliable. That result did not hold when the client's internal control was ineffective. Thus, their results suggest that reliance on the client's internal control strength can lead to over reliance on the third-party expert. These studies suggest that if auditors had information about third-party experts' internal controls it could negatively influence audit quality when evaluating the client's fair value measurement.

• Question 14: 

  Is the potential amendment to Auditing Standard No. 12 described above clear and appropriate for both accounting estimates and fair value measurements? Are there other factors that would be relevant in the auditor's evaluation of the degree of complexity of judgment in the recognition or measurement of an accounting estimate or fair value measurement (e.g., the use of a third party for the determination of a price)?

Several recent studies surveying auditors (managers, senior manager, and partners) suggest that it is not uncommon for auditors to encounter wide measurement uncertainty when auditing fair value measurements, such that the range of estimation uncertainty exceeds the materiality threshold. For instance, Cannon and Bedard (2014) find that over 70 percent of audit managers and senior managers surveyed report auditing complex fair value measurements in which the measurement uncertainty exceeds materiality. While the Consultation Paper points out that “wide range of measurement uncertainty” increases the “degree of complexity or judgment” of an accounting estimate, and thus risk, it does not provide clear guidance to auditors regarding how they should respond when dealing with estimates with extreme measurement uncertainty. Moreover, Glover et al. (2014a) find that 93 percent of audit partners with fair value measurement expertise surveyed are in favor of additional clarity and guidance from standard setters and regulators regarding auditing of fair value measurements. In particular, the audit partners report wanting to see more guidance regarding what constitutes an acceptable reasonable range of estimation uncertainty. While the committee believes that the Board's proposed amendments to paragraph 71 of Auditing Standards No. 12 are appropriate for both accounting estimates and fair value measurements, we encourage the Board to consider including additional guidance and clarification for auditors to help them to determine the appropriate responses when wide measurement uncertainty exists. Further, with respect to other relevant factors auditors should consider in their evaluation of complexity or judgment in the recognition or measurement of an accounting estimate, and in turn significant risks, the committee believes the Board should consider adding indicators of possible management bias as another factor auditors should consider (such guidance would be consistent with guidance provided in ISA 540).

• Question 19: 

  Should a potential new standard include specific audit procedures related to auditing disclosures of accounting estimates (e.g., disclosures on levels within the fair value hierarchy)?

Management's classification of fair value assets under ASC 820 is scrutinized by investors and, consequently, by auditors. When an argument can be made to classify a particular security as either a Level 2 or a Level 3 security (i.e., the security's classification is “at-margin”) there is room for managerial discretion and opportunism in fair value classification decisions. The market considers Level 3 inputs to be less reliable and riskier than Level 2 inputs and, accordingly, ASC 820 requires more disclosure for Level 3 assets. Given this increased disclosure requirement, and the fact that the market discounts Level 3 securities more than Level 2 securities, companies have incentives to use a Level 2 over a Level 3 classification to avoid being punished by the market (e.g., Laux and Leuz 2009; Laux and Leuz 2010). For example, studies find that Level 3 assets have greater information asymmetry when compared to Level 2 assets (Riedl and Serafeim 2011), the market discounts Level 3 securities significantly more than Level 2 securities (e.g., Beck 2012; Cullinan and Zheng 2014; Song, Thomas, and Yi 2010; Kolev 2009), and that the discount observed for Level 3 securities ranges from 20 to 30 percent of reported assets (Laux and Leuz 2010). Because fair value classification is important to investors, auditors dedicate significant effort in conducting ASC 820 “leveling” procedures¹ to evaluate the appropriateness of management's classifications. Earley, Hoffman, and Joe (2014) note that auditors have a significant role in assuring the appropriateness of these classifications because the reasoning underlying management's classifications is not disclosed publicly and because there are so few disclosures required for Level 2 securities.

Earley et al. (2014) find that for at-margin securities, auditors are more skeptical of management's classification choice when the client chooses the incentive-aligned Level 2 classification than when management chooses the Level 3 classification, which is not aligned with its incentives. Earley et al. (2014) suggest that auditors have internalized the warnings by the PCAOB to scrutinize management's fair value reporting, and therefore question classification choices that appear aligned with management's incentives. Given the potential for managerial opportunism, and because the classification hierarchy is so important to market participants, the Board should consider cooperating with the SEC to encourage more disclosures around both Level 2 and Level 3 assets and to consider whether more disclosure about management's rationale for the ASC classification would be informative to investors.

Recent inspection reports (e.g., the 2013 inspection of KPMG and the 2013 inspection of PwC) suggest that auditors might have difficulty with other aspects of the classification task. For example, the 2013 KPMG inspection report (PCAOB 2014a) finds deficiencies in evaluating the internal controls over Level 2 versus Level 3 classifications (Issuer A) and the classification of Level 2 securities (Issuer G). The 2013 PwC inspection report (PCAOB 2014b) notes deficiencies in testing the disclosures for hard-to-value Level 2 and Level 3 securities (Issuer H). The importance of auditors' role in providing assurance over the ASC 820 classifications and the findings in the recent firm inspection reports indicate that additional Board guidance related to fair value classifications would be helpful to auditors and the market participants who value auditors' assurance.

• Question 22: 

  Are there specific factors that affect the auditor's selection of approaches related to testing accounting estimates? What considerations would be appropriate for the auditor to take into account when determining which approach (or combination of approaches) for testing accounting estimates should be selected?

Existing research finds some differences in terms of auditors' use of the three substantive testing approaches available. Griffith, Hammersley, and Kadous (2014) interviewed audit partners and senior managers experienced in auditing complex estimates and found that auditors report frequently using the first substantive testing approach: testing management's process; however, auditors report infrequently on developing independent estimates and/or reviewing subsequent events and transactions. On the other hand, two more recent surveys of experienced auditors by Cannon and Bedard (2014) and Glover et al. (2014b) find that auditors report frequently using all three approaches or a combination of approaches when responding to significant risks. In light of differences in current practice as shown in the existing research, the Board should consider providing more guidance regarding instances in which auditors should use more than one substantive testing approach, in particular when developing an independent estimate is necessary. The Board might also want to consider providing additional guidance regarding necessary substantive testing approach(es) when management's estimates are determined by a third-party pricing service or valuation specialist when compared to the estimates derived directly from management or its internal specialists. Further, the Board should consider the approach taken in ISA 540 by providing specific additional substantive procedures necessary to respond to “significant risks.”

• Question 28: 

  Would a requirement for the auditor to determine which assumptions used by management are significant assumptions present difficulties in practice? Should the staff consider a requirement for the auditor to identify assumptions not used by management that might be important for recognition or measurement of this accounting estimate?

We believe that it is a good idea for the auditor to more explicitly identify assumptions management explicitly or implicitly uses in tests of reasonableness of the accounting estimates. It certainly will not be onerous, but rather will lead to a more clear identification of how the auditor assesses reasonableness of the management's assumptions and the estimate as a whole. More clear documentation of these assumptions in the working papers has the potential to reduce the likelihood that the audit work would be subject to future criticism during the PCAOB inspection process, as an auditor's judgment process on the matter will be more evident. More clear documentation of assumptions by management will help identify sources of biases and errors in the relevant estimates (Martin, Rich, and Wilks 2006). Prior research documents that “fair value and other estimates based on management's subjective models and inputs contain estimation uncertainty or imprecision that is many times greater than materiality” (Christensen, Glover, and Wood 2012). For accounting estimates more critical to the audit process (such as testing of the reasonableness of the bad debt or inventory obsolescence reserves), it is probably best for the auditor to document how she/he assesses the estimates' sensitivities to different assumptions involved and to more clearly define the range of the available accounting estimates, depending on these assumptions. This is especially important in light of the recent research documenting that the auditor is more likely to be switched when a client has to record a goodwill write down (Ayres, Neal, Reid, and Shipman 2014). Moreover, the standard should probably stipulate that the auditor clearly explain in her/his work papers why she/he concluded that certain assumptions are more reasonable or not.

With respect to the assumptions not explicitly considered by management, we believe it is sufficient that the auditor only considers other assumptions that could have a material impact on the estimate and provide clear explanations why the auditor believes this to be the case. If management does not concur that these assumptions need to be considered, an auditor should clearly document perceived sources of this disagreement and assess whether they give rise to a potential material misstatement.

• Question 29: 

  Is the potential requirement suggested above clear and appropriate for both accounting estimates and fair value measurements? Are there other specific characteristics of significant assumptions that should be included?

We recommend the following clarifications:

1. (1) 

   The Board may consider clarifying what it means by unobservable data (i.e., whether here it follows the definition used in determining whether a particular investment is classified as Level 3 under the ASC 820 hierarchy).

2. (2) 

   In addition, it is not entirely clear what the Board means by [assumptions] that “are based on the company's intent and ability to carry out specific courses of action?” If the Board is alluding to the fact that the audit client can manipulate assumptions in order, say, to manage earnings to meet earnings targets, then more explicit language would be helpful. The assessment of risk of opportunistic manipulation of the assumptions should be done in the context of the overall assessment of risk of material misstatement (RMM) and presence of “red flags” suggesting higher likelihood of RMM.²

3. (3) 

   Perhaps, the Board could also consider requiring more explicit auditor's consideration of any assumption that could cause material changes in the accounting estimate, and where the range of possible outcomes is likely to be uncertain. For example, for Level 3 investments without available liquid spot markets (such as private equity holdings), a wide variety of discount rates can be used to arrive at a valuation estimate. Using this example, the Board might want to recommend that auditors' working papers clearly include documentation on the range of available discount rate options and explain exactly why they agree on a particular discount rate choice.

• Question 30: 

  Are the suggested factors described above appropriate for evaluating the reasonableness of significant assumptions? Are there other factors the auditor should assess when evaluating the reasonableness of significant assumptions relevant to accounting estimates?

We believe that the list of factors, while reasonable, appears a bit too general. For example, it is unclear what is meant by “relevant industry, regulatory … factors, including economic conditions.” We believe that the standard, when released, needs to specify clearly under what circumstances the auditor needs to be particularly attentive in testing managerial assumptions (e.g., similar to the list of fraud risk factors stipulated in SAS 99/AU 316). We further believe that the proposed standard needs to emphasize that the auditor needs to develop a good sense of whether management has a good track record of deriving assumptions that, ex post, are highly correlated with ultimate economic outcomes. In particular, if an auditor observes that management's assumptions are persistently off mark and are, say, optimistically biased, then the auditor needs to be particularly skeptical toward those assumptions in the future.

• Question 36: 

  Are the potential requirements described above for evaluating audit evidence from events or transactions that occur subsequent to the measurement date through the date of the auditor's report, appropriate for both accounting estimates and fair value measurements?

Yes. We do believe that the guidance provided by the Board is appropriate for accounting estimates and fair value measurements. Martin et al. (2006) note that the many of the assumptions used in measuring fair value are similar to those used to generate other accounting estimates. That is, both fair value measurement and estimates attempt to quantify the future outcome of one or more past transactions or events. With regard to subsequent events, AU Section 560 requires auditors to evaluate events and transactions subsequent to the end of audit fieldwork, but prior to the issuance of the auditor's report. Therefore, evidence obtained from these events and transactions can provide guidance on both fair value measurements and accounting estimates as of the balance sheet date.

• Question 37: 

  Are there additional factors that should be taken into consideration when evaluating the relevance of the audit evidence obtained from events or transactions that occur subsequent to the measurement date through the date of the auditor's report?

As mentioned above, subsequent events can provide evidence about fair value measurements and accounting estimates at the balance sheet date. Specifically, these subsequent transactions can provide a direct substantiation of estimates and fair value measurements. As noted in the potential standard, however, some subsequent events or transactions could reflect changes in the company's circumstances or economic condition changes occurring after the balance sheet date and would not constitute audit evidence about the fair value measurement or estimate at the balance sheet date. As such, it is important to carefully consider the nature of the transaction when employing the use of subsequent events to evaluate fair value measurements and accounting estimates. For example, the sale of land held for investment shortly after the fiscal year end can provide sufficient evidence about its net realizable value (an accounting estimate). However, the fair value measurement of complex financial instruments, specifically Level 3 assets, might not be evaluated through any subsequent transactions. Glover, Prawitt, and Taylor (2009) note that accounting firms use standards to develop audit methodologies; therefore, we believe it would be beneficial in the updated standards to emphasize the limited value of subsequent events in auditing fair values and accounting estimates. Griffith et al. (2014) note that the few firms use subsequent events as a primary means of audit evidence for accounting estimates and fair value measurements. On the other hand, Cannon and Bedard (2014) and Glover et al. (2014b) find that auditors report the use of subsequent events as evidence when auditing complex estimates and fair value measurements. We believe that the value of subsequent events as a supplement or confirmation should not be ignored by audit firms.

• Question 38: 

  Would the potential requirements described above address procedures performed by audit firms that use a centralized testing approach? Would these requirements create issues in practice for smaller firms?

As noted by Griffith et al. (2014), auditing fair value measurement and accounting estimates requires a high level of professional skepticism and valuation knowledge. Because auditors often lack valuation knowledge, they often rely on third-party and in-house specialists. This guidance should extend to firms that use a centralized testing approach as well. While the use of external valuation specialists, including a centralized or national-level pricing desk, could have a positive effect on audit quality through increased reliability, it could have a negative effect on audit quality if auditors overly rely on specialists. The more important issue is that the audit team gathers appropriate evidence to audit fair value measurements and accounting estimates. As noted by Martin et al. (2006, 288) it is important for auditors to understand “valuation models, significant assumptions, audit procedures, and possible biases” when auditing fair value measurements and accounting estimates.

Recent research provides evidence that greater use of fair value measurements is associated with higher audit fees (Ettredge, Xu, and Yi 2014). Requiring the use of an independent third party could differentially impact smaller firms that do not have access to a centralized pricing desk. We believe that mandating the use of an independent (third-party) specialist is not as important as emphasizing the use of professional skepticism and judgment about estimates provided by management. This may involve employing a specialist to gather evidence sufficient to audit such accounts.

• Question 39: 

  Should the potential new standard require the auditor to use a third party that is different from the third party used by management? Would such a requirement present challenges for certain types of accounting estimates and fair value measurements?

As noted by Bratten, Gaynor, McDaniel, Montague, and Sierra (2013), the PCAOB, in its inspection reports, has been less concerned about the identity of the third-party specialist than in the auditors' overreliance on the assumptions made by management and that specialist. It is clear that third parties can provide benefits to financial statement users because a 2010 survey by Deloitte (2010) finds that 73 percent of asset managers surveyed believe pricing services are reliable.

We believe that professional skepticism should be employed in the use of third-party specialists, whether the same one employed by management, or a different one. If the auditor carefully evaluates the techniques, assumptions, and other inputs to valuation models in order to address the reasonableness of managements' estimates, then the use of a third party can be beneficial (Harvard Law School Forum on Corporate Governance and Financial Regulation 2011).