Relevant and reliable?

By Steve Fortin
Illustration: John Sapsford

Fair value accounting for financial instruments: what can be concluded from academic evidence

Following in the footsteps of the Financial Accounting Standard Board and the International Accounting Standard Board, the CICA’s Canadian Accounting Standard Board recently proposed new recognition and measurement rules for financial instruments (proposed sections 3855 and 3865 of the CICA Handbook). While the proposed standards will maintain historical cost accounting for some securities (held-to-maturity securities, loans and receivables and most financial liabilities), it will impose measurement for securities held for trading (speculative) purposes, securities that are deemed “available-for-sale,” and all derivatives.

Most accountants will agree that fair values are potentially more relevant than historical cost, particularly for financial instruments. The difficult question is the reliability of fair-value estimates. Reliability concerns are particularly serious for instruments that are not market traded, since they must be evaluated by management.

The measurement of financial instruments has attracted extensive interest from the academic community, especially its reliability for purposes of financial reporting  and its relevance for investors. Evidence from prior research suggests that reliable and relevant measures can be obtained, but important caveats do apply. We now review such evidence.

Research methodologies
Following the introduction of SFAS 107, 115, 119 in the US, fair-value estimates of financial instruments have become readily available. By statistically relating these fair-value disclosures to observed stock market values, researchers infer the information content, or value relevance, of fair-value disclosures. Value relevance jointly implies both relevance for investor's decision-making and reliability of measurement.

However, to assess value relevance, one must first determine how accounting information maps, or is reflected, in stock prices. A widely used valuation model, the balance-sheet valuation model, is based on the accounting equation. In this model, the market value of a stock equity is assumed to be explained by the sum of the book values of the firm’s individual assets and liabilities. The model can be expressed algebraically (see chart below).

In this model, e is the portion of firm value that is not explained by the model, often akin to goodwill. When examining research questions concerned with supplementary fair-value disclosures, the model is transformed to isolate the book value of financial instruments from the book value of other assets and liabilities. Supplementary fair-value disclosure variables are also added to the model. The valuation model thus becomes:

The fair value of financial instruments shows incremental value relevance, over and above their recorded book value, if ß4 is different from zero in the second equation. For example, assume that a firms’ financial instruments have a book value of $0 but a fair value of $1,000. If investors rely on fair-value disclosure on a one-to-one basis, we then expect ß4 to be equal to one. That implies that $1,000 worth of financial instruments translate into $1,000 of stock market valuation for the firm.

This model ignores income statement information, which is hardly consistent with real-life observations. Combining both balance sheet and income statement numbers in the valuation model captures both the current value of the firm (balance sheet information) and its growth pros-pects (income statement information).

After statistically estimating a valuation model similar to the second equation, failure to document value relevance would usually be interpreted by researchers as a lack of reliability, since the potential for relevance for investors is almost always supported by a strong theoretical basis. Unfortunately, this is not that simple, as the lack of relevance could also be due to the use of an inappropriate valuation model for the firm, to several statistical issues presenting a challenge when estimating valuation models on data, or to the omission of other relevant variables from the valuation model.

Review of extant literature pertaining to financial instruments
Mary Barth examined the value relevance of fair-value disclosures of investment securities held by banks. From 1971 to 1990, banks reported supplementary fair-value information about their portfolio of investment securities in their annual report. She documents that these fair-value disclosures are reflected in stock market values, using a valuation model roughly equivalent to second equation. Focusing on the income statement, professors Ahmed and Takeda found that both realized and unrealized gains and losses on investment securities directly affect stock returns.

Several authors examined the value relevance of mandatory fair-value disclosures of loans, investment securities, long-term debt and off-balance-sheet securities made by banks under SFAS 107. Karen Nelson, Elizabeth Eccher and Barth generally found fair-value disclosures made by banks in footnotes are value relevant (particularly the fair value of investment securities). Using a comprehensive valuation model, Barth found value relevance for most financial instruments (loans, securities and long-term debt). Using less complete models, Nelson and Eccher found that only securities fair-value disclosures are value relevant. These studies have one point in common: all failed to show value relevance for derivatives securities, raising serious questions about derivatives fair values relevance or reliability.

Mohan Venkatachalam, in a concurrent study, determined the lack of value relevance for derivatives is due to shortcomings of derivatives disclosure under SFAS 107. SFAS 107 did not require firms to indicate whether their derivatives portfolio was an asset or a liability. SFAS 119, published in 1995, which is similar to CICA Handbook Section 3860, corrected this deficiency. Using SFAS 119 disclosures, Venkatachalam showed that the fair value of derivatives is significantly associated with firms’ stock market values.

Kathy Petroni and James Wahlen found that the fair value of equity investments and US Treasury investment is value relevant for a sample of property-liability insurers, but municipal and corporate bond fair values are not. They argued that mu-nicipal and corporate bonds may be too lightly traded for markets to produce relevant and reliable estimates of market values for accounting purposes.

Caveats of valuation relevance research
What can be concluded from this research? It does appear that fair-value estimates of financial instruments obtained or computed by management are value relevant. Does this directly imply that standard-setters should adopt fair-value measurement?

The first shortcoming of this methodology relates to the lack of conclusion with respect to the identification of a cause-effect relationship. Did the market use accounting information released to compute firm value or did it use information from other sources? If the accounting information reflects the information used by the market in valuing the firm, both the market data and the accounting information may appear correlated, even though market participants may not have used the accounting information at all. Should standard-setters only require the reporting of information if it is directly used by market participants? Should firms report information that is already reflected in prices, even though alternate sources for this information exist? These questions are the object of much debate between academics. Robert Holthausen and Ross Watts made the identification of a cause-effect relationship a desired, if not a necessary, condition. On the other hand, Barth et al. claim the confirmation value of accounting information is well grounded in the conceptual framework of accounting, using association studies as important supporting evidence for standard-setters.

The second shortcoming relates to the ad-hoc nature of the valuation models used in the empirical tests. No valuation model is perfectly representative of an investors’ valuation model. All are at best an approximation of an investor’s true model, and their performance will vary depending on the empirical setting examined. Moreover, the models hold under any accounting policies chosen by the firm, somewhat reducing their usefulness in sorting out accounting alternatives. As long as the accounting policies chosen by the firm produce numbers that can be systematically related to the market value of equity, the valuation models’ parameters (i.e. the regression coefficients) will adjust upward or downward as needed. This greatly complicates the interpretation of any regression results about the relevance of accounting numbers. Holthausen and Watts claimed this is so serious as to eliminate the usefulness of the evidence. Barth, while recognizing the limitations imposed, claimed that researchers and standard-setters alike are aware of these limitations and properly control for them when interpreting their results.

Third, value relevance research ignores other potential users of financial state-ments, such as lenders, employees, suppliers, etc. Financial statements are prepared for a general purpose audience, and should arguably serve all users. There is currently little evidence with respect to the relevance of fair values to other groups of users. Still, equity investors are important users of financial statements, and while their needs should probably not be the only concern of the standard setters, they also cannot be ignored.

Conclusion
There is strong evidence reported that the fair value of financial instruments is associated with stock market equity values. This result has been examined by most standard setting bodies, as indicated by the membership of at least one academic on the CASB, the FASB and the IASB. All have adopted a fair-value measurement standard recently (Section 3855 of the CICA Handbook, SFAS 115 and 133, IAS 39), thus indirectly acknowledging that research findings support the adoption of the fair-value measurement standard.

Association with stock market values should not be the only consideration examined by standard-setters, due to the shortcomings of this methodology. The most important shortcoming of the extant research is probably its ignorance of other financial statements users. The recent political uproar surrounding the proposed adoption of IAS 39 by the European Union is a good reminder that financial statements also serve these other users. Some member states of the European Union, and more particularly France and Spain, have been heavily opposed to the adoption of IAS 39. They are concerned the fair-value adjustments will create volatility in banks’ stock market values. Only a watered-down version of IAS 39, which excludes banks from its mandatory application, could be approved. This is intriguing, since most of the research evidence is based on banks and indicates banks’ shareholders value this information. Accordingly, there must be other concerns than the relevance for stockholders that standard-setters have to weigh when making standard-setting decisions.

Steve Fortin, CA, PhD, is with the faculty of management at McGill University

Technical editor: Michel Magnan, PhD, FCA, associate dean, external affairs, Lawrence Bloomberg chair in accountancy, John Molson School of Business, Concordia University

References
Ahmed, A.S., Takeda, C., 1995. “Stock market valuation of gains and losses on commercial banks’ investment securities: An empirical analysis.” Journal of Accounting and Economics 20: pp. 207-225.
Barth, M.E., 1994. “Fair-value accounting: Evidence from investment securities and the market valuation of banks.” The Accounting Review 69: pp. 1-25.
Barth, M.E., Beaver, W.H., Landsman, W., 1996. “Value relevance of banks’ fair value disclosures under SFAS No. 107.” The Accounting Review 71: pp. 513-537.
Barth, M.E., Beaver, W.H., Landsman, W., 2001. “The relevance of the value relevance literature for financial accounting standard setting: another view.” Journal of Accounting and Economics 31: pp. 77-104.
Eccher, A., Ramesh, K., Thiagarajan, S.R., 1996. “Fair value disclosures bank holding companies.” Journal of Accounting and Economics 22: pp. 79-117.
Holthausen, R.W., Watts, R.L., 2001. “The relevance of the value relevance literature for financial accounting standard setting.” Journal of Accounting and Economics 31: pp. 3-75.
Jeffery, C., 2004. “French triumph on IAS 39.” Risk (August): p. 46.
Nelson, K.K., 1996. “Fair value accounting for commercial banks: An empirical analysis of SFAS No. 107.” The Accounting Review 71: pp. 161-182.
Venkatachalam, M., 1996. “Value-relevance of banks derivatives disclosures.” Journal of Accounting and Economics 22: pp. 327-355.