Auditing, Reasoning Systems, Reporting Frameworks, and Accounting Policy Risk: A Response to Alexander

This is a response to Alexander's (2010) comments on aspects of Smieliauskas et al. (2008), henceforth SCA. In SCA, we contended that the integration of accounting theory and standards with auditing theory and standards would make audited financial reporting more defensible and understandable to users of financial reports and bring more substance to the verifiability concept of financial reporting.

In SCA we made an important distinction between accounting risk (henceforth AccR) and audit risk (henceforth AudR) in the theory of audited financial reporting. Here we clarify the nature of AccR and AudR and explain the relevance of each for various accounting theories. We use AudR as a substitute designation for the RMM concept used in SCA and Alexander (2010). This is because the more restricted sense of RMM in ISA 315, Identifying and Assessing the Risks of Material Misstatements Through Understanding the Entity and its Environment, reflects only some parts of audit evidence gathering risk.

Alexander (2010) states that he reviewed the original SCA paper ‘from my own accounting theory perspective[since the SCA paper] was significantly, and in my view, mainly, conceptualized from an auditing theory perspective’ (p. 447, emphasis added). Our response is that it should not be surprising that SCA adopted an ‘auditing theory perspective’, since the focus of SCA was on the audit opinion in an auditor's report. Alexander's statement may help to reconcile differences in conclusions and offer the strength of more than one perspective in dialogue about the wording of the auditor's report.

The remainder of this response outlines the general reasoning approach we adopted. We then proceed to clarify the distinction between AudR and AccR and the roles of each of these risk concepts in different reporting frameworks. Thereafter, we clarify how Alexander's proposed accounting policy risk can be integrated with existing reporting frameworks.

A REASONING SYSTEM FOR AUDITING FINANCIAL REPORTING

Generally, to avoid contradictions, a system of reasoning in financial reporting needs to be coherent and consistent. Moreover, such a reasoning system should be persuasive but not mislead financial statement users. This means attention should be paid to the concepts and principles used, their role in the reasoning process and their relationship to reporting objectives. Three generic approaches can be used in developing such a reasoning system for auditing financial reports.

Approach 1: Use the reasoning system underlying auditing theory and standards as the starting point and extend that reasoning system to financial reporting. This approach, emphasizing the verifiability concept, was used implicitly in SCA. The rationale is that if the numbers in financial reports are not verifiable, the resulting reporting system will not be very useful.

Approach 2: Use some variant of accounting theory and standards and impose a consistent reasoning on auditing theory. This is the approach of Alexander (2010): his ‘conceptualizations’ are from an ‘accounting theory perspective’.

Approach 3: Start afresh. Develop an entirely novel, but coherent reasoning system for audited financial reporting.

We use approach 1 in this response (and in SCA) because:

The increasing importance of audits in successfully implementing fair value accounting and related estimates has been emphasized by Magnan and Thornton (2010). They note that the mark to market model in fair value accounting ‘is a defensible method of reporting provided auditors can attest to the credibility of management's estimates’ (2010, pp. 23–24, emphasis added). International standards on auditing already provide guidance on financial reporting frameworks. They highlight the crucial distinction between compliance of general purpose reports and the fairness of presentation of general purpose reports (ISA 200).

Alexander (2010) focuses on the effects of the financial reporting framework on the AccR and AudR concepts discussed in SCA. He notes two potential problems with AccR as defined in SCA. First, that AccR will be very high in many situations; and second, that AccR will be influenced by the financial reporting framework used (e.g., exit value, company-specific vs exit value, market-specific value). To address this issue, Alexander proposes a concept of accounting policy risk that is based on his contention that ‘many numbers in IFRS (and other) financial statements are not, and are not designed to be, estimates of ultimately realizable amounts’(p. 449).

For reasons explained earlier, SCA pursue an auditing theory approach. Nonetheless, it is important to engage with Alexander on his ‘accounting theory ground’. Alexander argues that since IFRS (and all other systems of acceptable accounting) contain standards that include a wide variety of approaches to assigning money amounts to balance sheet items, then many (if not most) of these approaches are not technically designed to reflect ‘what is realized ultimately’. While this is correct, it is irrelevant, since these approaches yield proxies (crude ones in many cases) for what is realizable.

Because we adopt approach 1 in analysing such situations, we are taking a risk-based perspective to accounting issues. Such a response is consistent with the risk-based approach of auditing standards, including the risks associated with accounting estimates as per ISA 540. Alexander also seems to be taking a risk-based approach to financial reporting through his concepts of accounting risk and accounting policy risk. The principal difference between Alexander (2010) and SCA lies in the relative emphasis accorded to ‘future realized values’. To SCA, ‘future cash flows’ and ‘future benefits’ are essentially comparable terms.

FAIRNESS OF PRESENTATION FRAMEWORK

We argue that the fairness of presentation framework requires consideration of future realized values, even if the generally accepted accounting principles invoked in preparing financial reports generally ignore future realized values. Otherwise, a compliance framework is sufficient to deal with what is purely factual information, such as observable market values in perfect and complete markets. In a compliance framework the auditor reports only whether ‘the financial statements are prepared, in all material respects, in accordance with an applicable financial reporting framework’ (ISA 200, para. 11). From an audit standards perspective, it is important not to blend an applicable compliance reporting framework with a fairness of presentation reporting framework. This distinction seems to be to be at the core of Zeff's (2007) ‘two-part’ audit opinion proposal. Below, we clarify the importance of maintaining the distinction.

Consistent with ISA 200.13, fairness of presentation frameworks offer a potentially crucial distinction between the past (including to the date of the financial statements) and the future. This is critical to the distinction SCA make between the AudR and AccR concepts. SCA justified this distinction by going back to the Cohen Commission Report (Cohen, 1978). As well, a persuasive justification of the past vs future distinction, on which these risks are based, is provided in Beaver (1991). Further clarification in explaining that past vs future is what distinguishes cash flow reporting from accrual accounting is provided in Glover et al. (2005). Past vs future is also what distinguishes pure historical cost from accrual accounting, pure entry values (whether market or entity-based) from accrual accounting, and pure exit values (whether market or entity-based) from accrual accounting.

So, a fundamental issue appears to be how important it is to emphasize future events in financial reporting. We contend that social norms imply that there should be a strong emphasis on future events. The importance of Beaver (1991) and the research cited there and in SCA is the evidence these provide on the social norms applicable to financial reporting. For example, Beaver (1991, p. 132) includes the following in his ‘major observations’:

When combined with the social norms noted in SCA (p. 237) that are reflected by the Gricean maxims (Grice, 1989), Beaver's observations suggest a way to resolve the paradox regarding the appearance of determinacy in accounting numbers. Using the significant risk concept for future outcomes of ISA 540, the SCA proposal is tantamount to equating their acceptable level of AccR to insignificant levels of risk due to future outcomes.

Basic research on how the human mind works indicates that the past vs future distinction is critical in reasoning and language (e.g., see Pinker 2007, pp. 188–233). The criticality of this distinction includes users of financial statements too. This is indicated by the basic concepts of assets and liabilities which are defined as future economic benefits or future cash flows (see IFAC concept of asset F.49(a) and liabilities F.49(b) at http://www.iasplus.com/standard/framewk.htm). However, proposed changes to the conceptual framework reduce this forward looking orientation (e.g., see http://www.iasb.org/Current+Projects/Conceptual+Framework/). Nonetheless, given these considerations, the future should not be ignored in fairness of presentation reporting frameworks. Therefore, uncertainties with respect to the future should not be ignored. This is the justification for the SCA concept of AccR. Accordingly, if the auditor's opinion is expressed in the words present fairly, then this implies consideration of acceptable levels of AccR. This is consistent with insignificant estimation uncertainties of ISA 540 and leads to reasonably achievable outcomes or reasonably predictable realizable values.

Alexander recognizes correctly that under many reporting frameworks the AccR can be ‘massive’ (p. 449). That is why SCA introduced the concept of acceptable AccR as a central feature of fairness of presentation. This is implicit in audit standards for what the AICPA (2008) and ISAE 3400, The Examination of Prospective Financial Information, term ‘prospective financial information’ or future oriented financial information (henceforth FOFI). FOFI can include forecast earnings and forecast financial statements many years into the future. Audit reports on FOFI do not use the term ‘present fairly’ even though the assumptions may be considered reasonable (but not necessarily achievable) by the auditor (ISAE 3400.29–.30). A reasonable range of realized amounts in a volatile environment may be too wide to provide a reliable indication of what will be realized. The fairness of presentation framework implies some sufficiently high probability of being realized, which is equivalent to a low probability of not being realized, that is, sufficiently low AccR.

AccR measures are similar to acceptable AudR measures in that they both deal with material misstatement: AudR with the risk of factual misstatements, and AccR with material forecast errors. However, AccR and AudR are not necessarily at the same levels because it is the socially acceptable level of the combination of the two types of risks for financial reporting that matters. This is consistent with the significant risk concept associated with accounting estimation uncertainty of ISA 540.

Further supporting this perspective is the characterization of fairness of presentation of financial reporting frameworks (ISA 200.13(a)). Essentially these consist of two conditions:

Ronen (2008, pp. 184, 187, 189, 193, 197–8) is also of the view that the auditor should be held responsible for ultimately realizable amounts and therefore should control, or at least appropriately disclose, AccRs in financial reports. Nonetheless, Ronen does not mention AccR explicitly. It is implied. Essentially, he proposes to provide auditors with incentives to properly disclose AccRs through an insurance mechanism that would require auditors to compensate investors when estimates are not realized. Such a proposal seems to have merit only in fairness of presentation reporting frameworks. Insurance in compliance frameworks is less likely to lead to protection of investors since compliance frameworks only tell investors whether certain rules were followed or facts reported. Compliance frameworks do not necessarily tell them whether future realizable amounts are reasonably reported. This suggests that AccRs and their control in financial reporting are a key judgment in meeting user needs.

INTEGRATING ALEXANDER'S PROPOSED ACCOUNTING POLICY RISK INTO PROPOSED REPORTING FRAMEWORKS

We now address the effects of different reporting frameworks on the issues identified by Alexander. If Alexander does not want to commit to making any predictions of the future, then a compliance reporting framework suffices. Presumably, he is content for this to be made clear to users when the financial reporting framework is disclosed. Interestingly, if we assume that the pure form of transactions and events up to the date of the balance sheet are used, all the measurement attributes identified by Alexander involve only factual information available at the audit report date.

Consider pure factual reporting cases with no AccR. In these conditions AudR is the only consideration. Thus, Alexander's concept of accounting policy risk has no role because it is future-oriented, not past-oriented. The only reporting issue is whether the relevant facts under the reporting framework are reported truthfully. This is a compliance reporting matter. The risk of non-compliance is simply AudR, as determined by the auditor. AudR is defined relative to the misreporting of facts, defined by the reporting framework. For example, even though exit price market values are justified, based partly on the market's ability to predict future benefits from an asset, what is relevant to the auditor is verification of the accuracy of the market values reported. If market values do not exist, the auditor has verified a market value of zero. Only if there is a forecast element in the reporting framework is the auditor obliged to go beyond the facts and consider additional reasonable assumptions in order to approximate a market value (e.g., level 2 and 3 measures of fair value accounting under FAS 157, Fair Value Measurements). This is when AccR becomes relevant, and when mere compliance with reporting certain relevant facts becomes insufficient. When there is a forecast element in the reporting framework, the fairness of presentation takes on substantive meaning.

Generally, auditors must consider the objectives of the reporting framework. Almost inevitably, this involves some predictive ability (e.g., as noted by Beaver, 1991). For example, auditors need to consider subsequent events in financial reporting to assure fairness of presentation, even though technically the financial statements end on a given date. Additionally, there are many enthymemes (silences or hidden assumptions) in accounting theory, as Alexander acknowledges (p. 453). Moreover, ISA 540 recognizes the potential for high estimation uncertainty and significant risk with estimates. These are interpreted as management judgment misstatements (ISA 540.A118). We characterize them as arising from unreasonable assumptions about the future. The unreasonableness of assumptions is assessed by the auditor (ISA 540.A118). Unreasonable assumptions increase the AccR for the resulting estimates that are forward looking.

Once the relevance of future realizable values and AccR is admitted to a theory of financial reporting, a new dimension to financial reporting complexity is introduced. It is no longer sufficient for the auditor to conclude ‘the financial statements have been prepared, in all material respects, in accordance with’ the applicable reporting framework. This is because a new qualitative feature to financial reporting has been added. The fairness of presentation framework now required by ISA 700, The Independent Auditor's Report on a Complete Set of General Purpose Financial Statements, substantiates the need for the wording ‘present fairly, in all material respects’. In particular, SCA propose that ‘present fairly’ (or its ISA 700 equivalent ‘true and fair view’) means that AccR is within some acceptable level of the reported amounts being good proxies for future realizable values. Equivalently, the reported amounts are regarded as ‘achievable’ with sufficiently high probability. Otherwise, the auditor should disclaim achievability, just as is done in audits of FOFI, described above. Under this system of reasoning, AccR indicators that are ‘massive’ according to Alexander (p. 449) are (or should be) avoided in the financial reporting if they are to ‘present fairly’. Otherwise, the auditor should issue a compliance audit report because the accounting policy (applicable reporting framework of ISA 200) results in an AccR that is too high (or, equivalently, just a ‘one-part’ opinion on mere conformity with GAAP per Zeff (2007, pp. 6–7). Thus, under the SCA proposal, not all reported amounts under a given measurement system are acceptable for fairness of presentation—only those amounts with acceptable AccRs.

Alexander notes that the problem with AccR is that it can be huge in many situations. If AccR is measured as a probability, then its maximum value is 1. If an acceptable AccR were set at 1, then any amount with a predicted element is acceptable. Such a reporting system would probably be considered misleading. To prevent this, SCA proposed that financial reporting frameworks based on present fairly should incorporate acceptable AccR as a key feature of the financial reporting system. When auditors do not use a fairness of presentation reporting framework, such as in compliance reporting or FOFI reporting, acceptability of AccR is not relevant because there is no explicit or implicit claim about the achievability of reported numbers. Instead, the claim made is only that the basis of their derivation is described appropriately.

By specifically acknowledging the role of predictions in financial reporting and the effects on AccR, Alexander's concept of accounting policy risk becomes relevant for auditors. Such an accounting policy risk could vary according to the account in question. For example, the AccR for cash is probably unchanged by the accounting policy selected—although this is not always the case (e.g., in respect of policies regarding translation of foreign currency cash into a reporting currency). However, incorporating accounting policy risk as a probability requires careful consideration of risk concepts and their relationships. For example, one cannot simply add the risks, as Alexander suggests, and expect to obtain a valid overall risk measure. Such aggregation can lead easily to aggregate risks much greater than 1, thereby violating basic probability axioms. On this point, a proposal made by Smieliauskas (2008) regarding identification of fraudulent forecasts in financial reporting merits consideration. Smieliauskas (2008) introduced the concept of a benchmark range based on acceptable AccR, proposed a way to properly aggregate audited financial reporting risks, and proposed a general way of constructing ranges of acceptable forecasts for financial reporting. Any accounting estimate involving assumptions about the future that fall outside the benchmark range is considered unacceptable for the financial reporting framework. This is because it would result in unacceptably high AccR (or in the terminology of ISA 540.11, ‘significant estimation uncertainty risk’).

This reasoning can be extended to Alexander's accounting policy risk. We can equate the AccR resulting from following a policy with the acceptable AccR used for constructing the benchmark range of reasonable amounts from following that policy (e.g., the reasonable range concept of ISA 540.A87–A94). An auditee recording a value outside this benchmark range creates a misstatement relative to the policy. This misstatement is the difference between the auditee's proposed amount and the nearest point on the benchmark range. There are then two estimates of AccR: one associated with the benchmark range; and one associated with the auditee's estimated value outside the range. This would be one way of operationalizing Alexander's concept of accounting policy risk, using the concept of AccR. Thus, if current GAAP can result in what Alexander refers to as ‘massive’ AccRs, then one way to escape this bind is to require the use of recorded amounts with acceptable AccRs in a fairness of presentation reporting framework. In such a situation, the accounting policy risk could be close to 1, but a fairly presented number within the policy could be within the acceptable (and presumably much lower) AccR level, when it exists.

Consistent with the reasoning approach adopted in SCA, our fundamental point remains that AccR (more specifically, acceptable AccR) should be a central concept in a fairness of presentation financial reporting framework.