Portal de Periódicos da CAPES

Risk reporting quality: implications of academic research for financial reporting policy

2012; Taylor & Francis; Volume: 42; Issue: 3 Linguagem: Inglês

10.1080/00014788.2012.681855

2159-4260

Stephen G. Ryan,

Financial Markets and Investment Strategies

Abstract In this paper, I survey empirical research on the relevance of firms’ financial report information for the evaluation of their risk. I recommend that financial reporting policymakers require or encourage firms to enhance their risk reporting quality in four ways. First, firms should report comprehensive income statements that: (1) use fair value or a similarly information-rich accounting measurement attribute and (2) separate the components of comprehensive income that are primarily driven by variation in cash flows from those that are primarily driven by variation in costs of capital. Such comprehensive income statements would provide users of financial reports with the flexibility to calculate alternative summary accounting numbers and to perform different types of risk assessment analyses. Second, firms should conduct and disclose the results of back-tests of prior significant accrual estimates, indicating any identified trends in and drivers of revisions to those estimates, and describing the effects of those revisions on current or future summary accounting numbers. Third, firms should aggregate and present risk disclosures in tabular or other well-structured formats that promote the usability of the information. Identifying existing best disclosure practices and encouraging new best practices are the most natural way to do this. Fourth, for model-dependent risk disclosures, firms should disclose the primary historical and forward-looking attributes of the models and their implementation in practice, sensitivity of the model outputs, and benchmarking of the models to standard portfolios of exposures. Keywords: disclosurefinancial reportingrisk reporting Acknowledgements I prepared this paper for the 19–20 December 2011 ICAEW Information for Better Markets Conference. I appreciate useful comments from Gauri Bhat, Leslie Hodder, Lisa Koonce, Jim Ohlson, and Dushyant Vyas. The comment from an anonymous reviewer is also appreciated. Notes In a number of recent US standards – for example, FAS 166, Statement of Financial Accounting Standards No. 166, Accounting for Transfers and Servicing of Financial Assets and Extinguishments of Liabilities, June 2009, paragraph 16-A (ASC 860-20-50) – the FASB states disclosure requirements in terms of general objectives before requiring specific disclosures. I believe this is a good idea, although it remains to be seen how great an impact it will have on firms’ disclosures. To illustrate, assume a firm holds only two exposures, whose economic performances are denoted A and B, so that firm's performance is A+B. The variance of the firm's performance equals the variance of A plus the variance of B plus two times the covariance of A and B. Perfect offsetting occurs (i.e. the variance of the firms’ performance is zero) if the covariance of A and B equals minus the average of the variances of A and B. For reasons of space, I do not attempt in this paper to evaluate individual research studies based on the specification of the regression models or the statistical power of the research designs. I personally prefer studies that more fully incorporate accounting structure and economic or decision-making context into their models and designs. Libby et al. (Citation2006) conduct an experimental study demonstrating auditors’ lesser concerns about disclosed than recognised information. For simplicity, throughout the paper I use the term ‘cost of capital’ to refer to any discount rate (e.g. for any financial instrument held by the firm), not just the discount rates for the firm's debt or equity capital. See, for example, Penman (Citation2010) and Pope (Citation2010). From accountants’ perspective, the term ‘CFB’ is unfortunate, as its estimation involves accrual accounting numbers. A more accurate term would be ‘accounting beta’; in fact, this term is used in the empirical literature summarised in Ryan (Citation1997) and Section 2. Despite this, I use the former term in conformance with the usage in the recent literature. I do not dispute that more and less risk-relevant accounting measurement attributes exist in specific economic or decision-making contexts. For example, long-standing literatures in accounting argue and provide evidence that: (1) the conservative deferral of revenue and gains is a prudent response to uncertainty about the realisation of income (Penman Citation2011) and (2) that conservative accounting is associated with lower bankruptcy risk (Biddle et al. Citation2011). My point is that a measurement attribute that yields highly risk-relevant financial report information in one context may not in another. For example, Penman explains that conservatism can yield earnings growth that is ‘not to be paid for,’ i.e. that should not increase valuation multiples for earnings. I direct the reader to ICAEW Financial Reporting Faculty (Citation2011) for a broad-ranging discussion of the issues related to risk reporting quality. See Campbell et al. (Citation2011) and Kravet and Muslu (Citation2011) for evidence that firms that provide longer lists or expanded discussion of risk factors in financial reports exhibit higher risk, as proxied by return variance, beta, and bid–ask spread. See Beyer et al. (Citation2010) for a recent survey of research on voluntary disclosures. It is unclear to me whether this finding would remain if researchers used more recent data and approaches to estimate accounting beta. As discussed in Section 3, Cohen et al. (Citation2009) find that CFB has strong predictive power for standard beta, although they do not run a horse race assessing the relative predictive powers of CFB and earnings variance for standard beta. Statement of Financial Accounting Standards No. 151, Inventory Costs – an amendment of ARB No. 43, Chapter 4, November 2004. As a certifiable accounting dinosaur, I refer to US GAAP standards that predate the Accounting Standards Codification (ASC) by their original type of standard and number designation, as with FAS 151 here. I indicate the relevant portion of the ASC the first time I refer to a standard. On the other hand, the research on implied costs of equity capital and CFBs deemphasises the distinctions among operating risk, operating leverage, and financial leverage made in the earlier research, although some accounting-based valuations models distinguish net operating assets (which are associated with operating risk and operating leverage) from net financial liabilities (which is associated with financial leverage). To the best of my knowledge, no extant accounting-based valuation models distinguish operating risk from operating leverage. See Preinreich (Citation1941), footnote 14. In addition, RIM has been used to develop trading strategies based on the difference between modelled (based on accounting numbers) and observed market values of equity. The risk-assessment and trading strategy literatures appear to largely explain the same phenomena; in particular, the results of Cohen et al. (Citation2009) and Nekrasov and Shroff (Citation2009) strongly suggest that the estimated abnormal returns generated by the RIM-derived trading strategies can be significantly explained by RIM-derived CFBs. RIM may also be derived from the discounted free cash flow valuation model using both the clean surplus relation and the cash conservation relation (i.e. free cash flow equals operating cash flow minus each of capital expenditures, cash dividends, and principal and interest payments to debtholders). See Ohlson (Citation1995) for the derivation of this version of RIM. Easton (Citation2007) provides a comprehensive discussion of the conceptual and practical issues involved in ICOC estimation. In particular, he discusses the fundamental issue of the dependence of estimates of ICOC on assumed growth rates. Admittedly, this is a rather low bar for success given the dismal performance of standard finance approaches to estimating the cost of equity capital to date. The market discount factor mt ,t+s reflects investors’ state-dependent willingness to defer consumption from period t to period t+s. The expectation in period t of mt ,t+s – which is probability-weighted across all possible future states that might occur in period t+s – equals 1/(1+r f t , t+s ). Intuitively, a security that pays a constant amount in each possible future state must yield the risk-free rate. More negative covariances with mt ,t+s indicate greater systematic risk because mt ,t+s is inversely related to expected consumption in period t+s. More positive covariances with consumption indicate greater systematic risk. Cohen et al.’s (Citation2009) approach to estimating CFB is somewhat less directly motivated by RIM than is Nekrasov and Shroff (Citation2009)'s approach. Cohen et al. estimate CFB by correlating time-aggregated ROE with time-aggregated market ROE. They allow the time-aggregation horizon to vary from 1 to 15 years, under the idea that a longer horizon allows errors in the measurement of ROE to be more fully mitigated. Roughly speaking, this yields CFBs that reflect a weighted-average of the covariances in Equation (2) across the time-aggregation horizon. I do not describe a related and far larger body of ‘value-relevance’ research that examines the explanatory power of accounting numbers for the market value of equity or share returns. Statement of Financial Accounting Standards No. 107, Disclosures about Fair Value of Financial Instruments, December 1991. Francis (Citation1990), Barth et al. (Citation1995), and Bernard et al. (Citation1995) provide evidence that fair value accounting for a portion of firms’ economic sheets yields more variable owners’ equity and net income than does amortised cost accounting, and that this variance is only partly reflective of economic variance. International Financial Reporting Standard No. 7, Financial Instruments: Disclosures, August 2005 (but frequently amended), paragraphs 25–27, requires similar disclosures. Hodder et al. (Citation2006) directly estimate gains and losses on non-term deposits, an important financial instrument for banks, because FAS 107 does not require banks to disclose the fair values of these deposits. I believe that Hodder et al.’s (Citation2006) findings regarding the risk-relevance of fair value gains and losses likely generalise to firms’ revisions of accrual estimates generally. For example, Petroni et al. (Citation2000) provide evidence that discretionary revisions of loss reserves by property-casualty insurers are positively associated with the insurers’ beta and return variance. Even if the fair value option had existed then, as it does now under Statement of Financial Accounting Standards No. 159, February 2007, relatively few firms have elected that option to date. Statement of Financial Accounting Standards No. 157, Fair Value Measurements, September 2006. Most SEC disclosure rules are not included in the ASC. International Financial Reporting Standard No. 7, Financial Instruments: Disclosures, August 2005 (frequently amended). I do not discuss operational risk in this paper, despite considerable disclosure of this risk in banks’ financial reports. This risk is outside my expertise, has a quite different nature from the risks I discuss, and is not subject to any research related to the themes in this paper of which I am aware. Financial Reporting Release (FRR) 48, Disclosure of Accounting Policies for Derivative Financial Instruments and Derivative Commodity Instruments and Disclosure of Quantitative and Qual­itative Information about Market Risk Inherent in Derivative Financial Instruments, Other Financial Instruments, and Derivative Commodity Instruments (1997), 17 CFR 229.305. Statement of Financial Accounting Standards No. 161, Disclosures About Derivative Instruments and Hedging Activities – An Amendment of FASB Statement 133, March 2008. Statement of Financial Accounting Standards No. 133, Accounting for Derivative Instruments and Hedging Activities, June 1998. Statement of Financial Accounting Standards No. 140, Accounting for Transfers and Servicing of Financial Assets and Extinguishments of Liabilities, September 2000; Statement of Financial Accounting Standards No. 166, Accounting for Transfers and Servicing of Financial Assets and Extinguishments of Liabilities, June 2009. For example, mortgage banks’ fixed-rate mortgage servicing rights are sensitive to interest rates through their direct effect on discounting and indirect effect on prepayment. Ideally, mortgage banks’ disclosed sensitivities to interest rates would reflect both effects. Instead, their disclosed sensitivity to interest rates reflects only the discounting effect and they make separate disclosures of prepayment sensitivity. American Institute of Certified Public Accountants (AICPA) Statement of Position 94-6, Disclosure of Certain Significant Risks and Uncertainties, December 1994. In addition, FASB Staff Position SOP 94-6-1, Terms of Loan Products That May Give Rise to a Concentration of Credit Risk, November 2005 (ASC 825-10-55) indicates that non-traditional mortgage products may constitute a credit risk concentration. (This FSP is misdesignated; it primarily provides guidance for FAS 107, not SOP 94-6, because mortgages are financial instruments.) Statement of Financial Accounting Standards No. 5, Accounting for Contingencies, March 1975. In my experience, financial institutions evidence little or no compliance with this disclosure requirement. International Accounting Standard No. 37, Provisions, Contingent Liabilities, and Contingent Assets, September 1998. SEC, Securities Act Industry Guide 3, Statistical Disclosure by Bank Holding Companies, http://sec.gov/about/forms/industryguides.pdf. Statement of Financial Accounting Standards No. 118, Accounting by Creditors for Impairment of a Loan-Income Recognition and Disclosures – An Amendment of FASB Statement No. 114, October 1994. FASB Staff Position FAS 133-1 and FIN 45-4, Disclosures About Credit Derivatives and Certain Guarantees: An Amendment of FASB Statement No. 133 and FASB Interpretation No. 45; and Clarification of the Effective Date of FASB Statement No. 161, September 2008. International Financial Reporting Standard No. 7, paragraphs 27A-B, requires similar disclosures. SEC Regulation S-K, Management's Discussion and Analysis of Financial Condition and Results of Operations 17 CFR 229.303. A related literature examines the value-relevance of banks’ maturity or repricing gap disclosures, e.g. Flannery and James (Citation1984), Schrand (Citation1997), and Ahmed et al. (Citation2004). Statement of Financial Accounting Standards No. 119, Disclosure About Derivative Financial Instruments and Fair Value of Financial Instruments, October 1994. This standard was superceded by FAS 133. This dearth of research may reflect the difficulty of proxying for banks’ credit risk using their financial report disclosures, as discussed in Knaup and Wagnerz (Citation2009). Banks have diverse loan portfolios and other credit risky exposures, including complex ones such as retained interests from securitisations and credit derivatives. Beatty and Liao (Citation2011) provide evidence that banks that make timelier provisions extended more loans during the financial crisis. The results for commodity price risk discussed in Section 6.a are also strong but apply to fewer firms. As discussed in Ronen and Ryan (Citation2010), when markets are illiquid, fair value is just one sensible measurement basis among several others with similar informational richness but different approaches to dealing with illiquidity. I am not wedded to fair value as it is currently defined, but I believe whatever measurement attribute is chosen should be applied consistently across exposures. See page 119 of Annual report 2010, JP Morgan Chase & Co, available at http://files.shareholder.com/downloads/ONE/1750571523×0x458380/ab2612d5-3629-46c6-ad94-5fd3ac68d23b/2010_JPMC_AnnualReport_.pdf (accessed 28 March 2012). See page 139 of Annual report 2007, Countrywide Financial Corporation, available at http://www.sec.gov/Archives/edgar/data/25191/000104746908002104/a2182824z10-k.htm (accessed 28 March 2012).