FROM: TO: Dr. Steven Glover (Professor at Brigham Young University) Dr. Brian Bratten (Assistant Professor at the University of Kentucky) Dr. Nathan Cannon (Assistant Professor at Texas State University) Dr. Brant Christensen (Assistant Professor at the University of Missouri- Columbia) Financial Accounting Standards Board DATE: February 17, 2016 SUBJECT: Comments on Proposed Accounting Standards Update, Fair Value Measurement (Topic 820): Disclosure Framework Changes to the Disclosure Requirements for Fair Value Measurement We appreciate the opportunity to submit our comments with respect to the Board s Proposed Accounting Standards Update (hereafter referred to as the Proposal). We are pleased to see that the Board is addressing the importance of communicating measurement uncertainty to financial statement users as we believe this will result in disclosures that are more faithfully representative. Our comments below are in response to questions outlined in the Proposal and are based heavily on academic research (see the references to research studies in Appendix A for more information). Question 1: Would the proposed amendments result in more effective, decision-useful information about fair value measurements? If not, please explain why. Unobservable and Observable Inputs For Level 3 fair value measurements, the Proposal requires companies to report quantitative information about the range, weighted average, and time period associated with significant unobservable inputs (Section 820-10-50-2(bbb)). Research suggests that even very slight and reasonable changes to inputs used in fair value estimates can materially change the final point estimate (Christensen, Glover, and Wood 2012). As such, fair value estimates may contain measurement uncertainty with a reasonable range of possible reported values (i.e., outcomes considered reasonably likely by experts) equal to multiples of materiality. In addition to affecting recorded figures on the Balance Sheet, the significant uncertainty often flows through to summary figures such as net earnings and earnings per share (EPS). Therefore, we commend the Board for requiring information regarding the reasonable distribution around inputs used in Level 3 measurements. This will result in a better representation of the measurement than only providing the point estimate or the point estimate and a narrative. The Proposal is especially helpful by requiring the disclosure of both a reasonable range and a weighted average for the observable inputs because they provide complementary information about the estimate and related uncertainty. While a range highlights measurement uncertainty (see further discussion about measurement uncertainty below), the weighted-average provides more complete information about how the range was used, as the midpoint of the range of an input may be a poor estimate of the weighted-average (Bratten, Jennings, and Schwab 2016).
While we agree that providing such information about unobservable inputs is crucial, it would seemingly be incomplete without similar disclosures for any significant observable inputs used in management s valuation models. Paragraph 820-10-50-1D(a) of the Proposal states that the objective of the standard is to provide users with the valuation techniques and inputs that a reporting entity uses to arrive at its measures of fair value, including judgments and assumptions that the entity makes. This objective does not limit the scope of the standard to only unobservable inputs. Without disclosure of all significant inputs to Level 3 valuation models, observed and unobserved alike, users are provided incomplete information regarding the fair value estimates. Finally, if preparers only disclose certain inputs it will be more difficult for users to compare inputs across preparers or to effectively evaluate management s track record and consistency over time, which is a characteristic highly valued by users (Peterson, Schmardebeck, and Wilks 2015). We encourage the Board to expand on its quantitative disclosure requirement to include all significant inputs included in management s discussion of its valuation model, not just those that are unobservable. Further, the desire to provide users with information regarding management s judgments and assumptions raises the question of whether the same disclosure requirements should be extended to Level 2 assets and liabilities. Proposed Range Disclosure Currently, the Proposal is silent as to the nature of the input range to be disclosed by management. We recommend that guidance require the range to contain inputs considered reasonably likely by experts. In 2014, HSBC utilizes a 95 percent confidence interval surrounding their quantitative uncertainty analysis (see Appendix B). However, such a wide range includes values that are unlikely to occur and has triggered user concern that such ranges would be unnecessarily wide (see Proposal BC37). A more useful range of uncertainty to disclose would be a range that experts agree is reasonably likely. To improve consistency and comparability across preparers, the board could choose to define a reasonable range as, for example, the 50 percent confidence interval. In disclosures it would be described as the range of outcomes experts consider reasonably likely. Granger (1996) describes a 95 percent confidence interval as embarrassingly wide, and academic research generally finds narrower ranges more useful to users than wider ones (Davis-Friday et al. 2004; Du et al. 2011). Of course nothing would preclude a preparer from also providing the 95 percent confidence intervals to represent what they consider the full range of likely and unlikely outcomes. Question 2: Are the proposed disclosure requirements operable and auditable? If not, which aspects pose operability or auditability issues and why? Audit firms use quantitative materiality benchmarks in evaluating whether the financial statements as a whole are fairly stated in all material respects. However, in a recent study, audit managers and senior managers with expertise in auditing fair values report that over 70 percent of complex fair value measurements contain measurement uncertainty with a reasonable range of possible reported values (i.e., outcomes considered reasonably likely by experts) that is equal to or exceeds quantitative audit materiality (hereafter referred to as
significant measurement uncertainty ) (Cannon and Bedard 2015). Current financial reporting standards require these volatile items to be reported as single point estimates, with no requirement to disclose the reasonable range of this measurement uncertainty. Audit firms are currently required, however, to offer an opinion on whether the financial statements present fairly, in all material respects, the financial position of the company. Taken together, research indicates that fair value measurements may not be auditable under these current standards in situations of significant measurement uncertainty (see Bratten, Gaynor, McDaniel, Montague, and Sierra (2013) for a review of fair-value related audit research and a more detailed discussion of factors limiting the auditability of such estimates under current standards). This situation is especially problematic when one considers that a single set of financials statements has multiple point estimates with significant uncertainty, some of which have a direct effect on bottom line figures such as net income and EPS. The result is that net income and EPS will also have reasonable ranges that are many multiples of what users and auditors have traditionally defined as a material amount. To address auditability in instances where measurement uncertainty exceeds materiality, it seems critical for auditors to not only opine on the single point estimate, but also on the reasonable range of uncertainty associated with the estimate. Thus, as discussed further below, we support a requirement for preparers to provide a quantitative measurement uncertainty analysis on the disclosed reasonable range in order to allow auditors to opine on the reported range of outcomes. This would provide additional transparency to users, improve comparability across financial statements, and highlight the challenges faced by auditors required to provide assurance that point estimates and summary figures (e.g., net income, EPS) reported in the financial statements are fairly stated in all material respects. The Public Company Accounting Oversight Board (PCAOB) is currently in the process of developing a new standard on the audit of fair value measurements and other estimates (PCAOB 2014). We encourage the Board to work with the PCAOB in aligning financial reporting and auditing standards. Question 4B: Should entities other than public business entities (for example, employee benefit plans and not-for-profit organizations) also be exempt from the proposed amendments mentioned in Question 4A? If yes, please describe why and which disclosures they should be exempt from. No. While it is likely that many employee benefit plans (such as defined contribution benefit plans) and non-for-profit organizations do not have material assets or liabilities utilizing recurring Level 3 fair value measurements, to the extent that they do, these disclosure requirements certainly seem applicable and would likely be important to users of these entities financial statements. Question 6: The proposed amendments to paragraph 820-10-50-2(bbb) require that a reporting entity disclose the time period used to develop significant unobservable inputs. What would be the costs associated with including this disclosure? Would this disclosure provide more effective, decision-useful information?
There is potential value to this information, since it provides additional context for the range and weighted average of the inputs used by management. In a stock option setting, Bartov, Mohanram, and Nissim (2007) find evidence that when weighting historical vs. implied (forward-looking) information to determine the volatility estimate, managers appear to place more weight on the information that would minimize their reported expense. Similarly, in a fair value measurement setting, managers could be faced with choosing a time period for which to evaluate inputs that may result in biased inputs, and disclosing the time period used could serve as a safeguard to curb this bias. However, disclosure of the time period used to develop inputs is somewhat ambiguous. We suggest the Board provide guidance regarding what to do in cases when the firm does not use only historical information to develop the inputs (i.e., inputs could be derived from contractual agreements and forward-looking assessments of likely outcomes). Question 7: Are there any other disclosures that should be required by Topic 820 on the basis of the proposed Concepts Statement or for other reasons? Please explain why. Communicating Measurement Uncertainty Research suggests that investors find qualitative signals of increasing measurement uncertainty relevant in making investment decisions (Christensen, Glover, and Wolfe 2014; Kachelmeier, Schmidt, and Valentine 2015). However, given the wide variation in the extent of measurement uncertainty across Level 3 assets and liabilities, requiring only a qualitative discussion of measurement uncertainty as outlined in 820-10-50-2(g) falls short of helping users understand the magnitude of measurement uncertainty that exists. IFRS 13.93.h(ii) goes beyond this narrative disclosure and requires, for financial assets, an additional quantitative disclosure of the effect of significant changes in fair value due to changes in inputs. In its deliberations, the IASB concluded that the usefulness of communicating the potential variability of fair value measurements outweighed the potential difficulty and costs (IFRS 13-BC208). While we are sensitive to concerns voiced by preparers regarding the cost of a quantitative measurement uncertainty analyses (described in Proposal BC9, 36-37, etc.), we agree with the additional quantitative disclosure required by IFRS 13. User demand for such information is strong (as described in BC35), and the fact that companies reporting under IFRS have been successfully providing this level of disclosure for several years now provides strong evidence that such a requirement is, in fact, operable. Further, we assume that preparers already have much of the necessary information available due to the quantitative sensitivity analyses performed to support their valuation approach and to determine the appropriate point estimate eventually recorded in financial statements. Regarding usefulness, preparers have argued that the [quantitative] information provided would not be decision useful because the range of reasonably possible Level 3 fair values would be extremely wide and, thus, would be meaningless ( BC37). However, we argue that it is precisely because of the possibility of wide reasonable ranges of fair values that these disclosures are useful and important (see our suggestion for defining a reasonable
range above). Without these disclosures, users will not have the tools necessary to determine the varying degrees of risk inherent in Level 3 fair value measurements and the resulting effect on summary measures such as net income and earnings per share. Recent research also suggests that more complete disclosure of fair value ranges preemptively encourages management to be more conservative in their reporting decisions (Majors 2015). Thus, not only will a quantitative measurement uncertainty analysis allow users to recognize risk more readily, but also it has the potential to reign in future aggressive reporting behavior. In implementing the quantitative measurement uncertainty analysis, recent research suggests that disclosing the effect of a significant change in the fair value measurement on net income is more effective at communicating risk to users than simply showing the change in the underlying value of the fair value measurement (Cannon 2015). We encourage the Board to enact quantitative measurement uncertainty analysis requirements similar to those in IFRS 13 to include the effect of uncertainty on net income and reported fair value. As it is presented currently by companies reporting under IFRS (see example from HSBC and Standard Chartered Bank in Appendix B), such information could be included by adding additional columns. For example, in Case C of the Proposal (page 21), after disclosing the reasonable range of fair values due to reasonable changes in the inputs, an entity could also disclose the effect on net income of the measurement uncertainty. We agree with the Board that the quantitative disclosure should not extend to the interrelationships among the inputs as this level of disclosure could quickly become overwhelming to users. Rather, the current narrative disclosure regarding interrelationships outlined in 820-10-50-2(g), coupled with an explanation that quantitative measurement uncertainty analyses are performed holding all other variables constant, should be sufficient. Materiality The Proposal references the Proposed Accounting Standards Update, Notes to Financial Statements (Topic 235): Assessing Whether Disclosures Are Material. This proposed standard update to Topic 235 revises the definition of materiality and limits disclosures to those deemed to be material by management. We urge the Board to further consider how materiality is applied in the context of significant measurement uncertainty and to provide preparers with guidance on how to evaluate the materiality of fair value disclosures with measurement uncertainty, particularly when the uncertainty is larger than what has traditionally been considered material and when such uncertainty flows through to net income and earnings per share. To summarize, we support the Board s initiative to require the disclosure of reasonable ranges, weighted averages, and time periods from inputs used in Level 3 models, although additional guidance would be helpful related to the disclosure of time periods. However, we also encourage the Board to require quantitative disclosure of all significant inputs included in managements Level 3 valuation models, including observable inputs. Further, we encourage the Board to expand the measurement uncertainty analyses to include quantitative analyses of the effect of measurement uncertainty on reported fair value measurements and net income. Finally, we
recommend the Board carefully consider the role and application of materiality when reporting fair value measures with significant measurement uncertainty. We appreciate the opportunity to offer our comments. Kind regards, Dr. Steven Glover, Brigham Young University Dr. Brian Bratten, The University of Kentucky Dr. Nathan Cannon, Texas State University Dr. Brant Christensen, The University of Missouri-Columbia
Appendix A: References Bartov, E., Mohanram, P., and Nissim, D. (2007). Managerial Discretion and the Economic Determinants of the Disclosed Volatility Parameter for Valuing ESOs. Review of Accounting Studies, 12(1): 155-179. Bratten, B., Gaynor, L. M., McDaniel, L., Montague, N. R., and Sierra, G. E. (2013). The audit of fair values and other estimates: The effects of underlying environmental, task, and auditor-specific factors. Auditing: A Journal of Practice & Theory, 32(sp1), 7-44. Bratten, B., Jennings, R., and Schwab, C.M. (2016). The Accuracy of Disclosures for Complex Estimates: Evidence from Reported Stock Option Fair Values. Accounting, Organizations and Society, Forthcoming. Cannon, N. (2015) Fair Value Measurement under High Uncertainty: The Effects of Disclosure Format and Management Aggressiveness on Users Risk Assessments. Working paper, Texas State University. Available at http://ssrn.com/abstract=2731367 Cannon, N. and Bedard, C. (2015). Auditing Challenging Fair Value Measurements: Evidence from the Field. Working paper, Texas State University and Bentley University. Available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2220445 Christensen, B. E., Glover, S. M., and Wood, D. A. (2012). Extreme estimation uncertainty in fair value estimates: Implications for audit assurance. Auditing: A Journal of Practice & Theory, 31(1), 127-146. Christensen, B. E., Glover, S. M., and Wolfe, C. J. (2014). Do Critical Audit Matter Paragraphs in the Audit Report Change Nonprofessional Investors' Decision to Invest? Auditing: A Journal of Practice and Theory, 33 (4): 71-93. Davis-Friday, P. Y., Liu, C. S., & Mittelstaedt, H. F. (2004). Recognition and disclosure reliability: Evidence from SFAS No. 106. Contemporary Accounting Research 21(2) Summer 2004, 399-429. Du, N., Budescu, D.V., Shelley, M.K., & Omer, T.C. (2011). The appeal of vague financial forecasts. Organizational Behavior and Human Decision Processes, 114(2011), 179-189. Granger, C. W. J. (1996). Can we improve the perceived quality of economic forecasts? Journal of Applied Econometrics, 11, 455 473. Kachelmeier, S., Schmidt, J., and Valentine, K. (2015). The Disclaimer Effect of Disclosing Critical Audit Matters in the Auditor s Report. Working paper, The University of Texas at Austin. Available at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2481284 Majors, T. (2015). The Interaction of Communicating Measurement Uncertainty and the Dark Triad on Managers Reporting Decisions. The Accounting Review, forthcoming. Peterson, K, Schmardebeck, R. and Wilks, T.J. (2015) The Earnings Quality and Information Processing Effects of Accounting Consistency. The Accounting Review, 90 (6): 2483-2514. Public Company Accounting Oversight Board. (2014). Staff Consultation Paper: Auditing Accounting Estimates and Fair Value Measurements. August 19, 2014.
Appendix B: Examples of IFRS 13 Quantitative Measurement Uncertainty Analyses 2014 HSBC Holdings
2014 Standard Chartered Bank 2015-350