Fair Value Measurement for Long-Lived Operating Assets: Research Evidence

Abstract

This paper reviews research evidence on the usefulness of fair value measurement of long-lived operating assets for financial reporting. Although economically material, these assets have been sidelined in the current fair value debate that largely centers on financial assets, from which operating assets differ in important respects. Our review provides several insights. First, fair value measurement is pervasive for investment property, while nearly non-existent for PP&E and intangible assets. Second, external fair value appraisers help enhance the decision usefulness of fair values. Third, the determinants of fair-value-related reporting choices vary by context and are not yet well-understood. Fourth, fair value measurement is largely useful when it comes to the valuation role of financial reporting, but this usefulness varies across recognized and disclosed fair values. Fifth, the contracting implications of fair value measurement for long-lived operating assets are severely under-researched. Overall, whereas extant research provides relevant insights, it is derived from diverse settings employing different measures and research designs, and does not yet provide an accumulation of evidence sufficient for final clarity about the determinants and consequences of fair value measurement for long-lived operating assets. Based on this conclusion, we develop several suggestions for future research in this area.

Keywords:

• Fair value
• Value relevance
• Contracting
• Operating assets
• Economic consequences

JEL classifications:

• G12
• G30
• G32
• M41

Acknowledgements

We thank Igor Goncharov, Katharina Hombach, Gilad Livne, Maximilian Müller, Eddie Riedl, and workshop participants at LMU Munich School of Management for helpful comments and suggestions.

Supplemental Data and Research Materials

Supplemental data for this article can be accessed on the Taylor & Francis website, https://doi.org/10.1080/09638180.2018.1511816

Notes

1 For relevant reviews, see, for example, Ryan (2011), Laux (2012), and Beatty and Liao (2014).

2 By long-lived operating assets we refer to long-lived non-financial assets such as PP&E, investment property, biological assets, and intangible assets – asset classes that non-financial firms typically use in their operations.

3 We note that ‘reliability’, the term used in Ball (2006), no longer features among the qualitative characteristics of financial information in the IFRS Conceptual Framework, having been replaced by the notion of ‘faithful representation’. However, the concern can be restated as one about lack of ‘verifiability’, an enhancing qualitative characteristic that ‘helps assure users that information faithfully represents the economic phenomena it purports to represent. Verifiability means that different knowledgeable and independent observers could reach consensus, although not necessarily complete agreement, that a particular depiction is a faithful representation’ (IFRS Conceptual Framework, par. 2.30). This idea reflects a notion of (potential) estimation error, or measurement uncertainty. We thank an anonymous reviewer for this point. In the remainder of the paper, we do continue to refer to ‘reliability’ (only) where the underlying studies do.

4 Determination of fair values under International Financial Reporting Standards (IFRS) follows a so-called ‘fair value hierarchy’ defined in IFRS 13, Fair Value Measurement (IASB, 2011): ‘To increase consistency and comparability in fair value measurements and related disclosures, this IFRS establishes a fair value hierarchy that categorises into three levels (…) the inputs to valuation techniques used to measure fair value. The fair value hierarchy gives the highest priority to quoted prices (unadjusted) in active markets for identical assets or liabilities (Level 1 inputs) and the lowest priority to unobservable inputs (Level 3 inputs)’ (IASB 2011, par. 72). ‘Level 2 inputs are inputs other than quoted prices included within Level 1 that are observable for the asset or liability, either directly or indirectly’ (IASB 2011, par. 81).

5 We are indebted to Igor Goncharov for this excellent point. Note, however, that the presence of synergies differs even among different asset classes within the realm of operating assets. For example, whereas a set of PP&E items that form a car maker’s assembly line clearly work synergistically together, the presence of synergies is less obvious within a portfolio of heterogeneous investment property assets held by a real estate firm.

6 Following Hodder et al. (2014), we use the term fair value measurement rather than fair value accounting. Hodder et al. (2014, p. 146) refer to fair value accounting as ‘an ambiguous, indeterminate phrase that appears to mean different things to different accounting researchers’. In particular, we consider fair value-related evidence irrespective of the presentation format (i.e., recognition on the balance sheet or disclosure in the footnotes) and irrespective of how fair value changes are treated (i.e., recognition in profit or loss, or direct recognition in a revaluation reserve within equity – thus by-passing profit or loss).

7 We do not include in our review evidence on fair value measurement related to goodwill impairment. For example, the impairment test under IFRS (IAS 36 Impairment of Assets) requires firms to derive the recoverable amount of an asset (or a cash-generating unit), that is, the higher of its fair value less cost to sell and its value in use. However, fair values are only recognized if they fall short of carrying amounts. For a summary of evidence on asset impairment, see, for example, Riedl (2004) and Sellhorn (2004), and on goodwill impairment specifically, Boennen and Glaum (2014).

8 For a discussion of the role of evidence from academic research in standard setting, see, for example, Barth (2007), Fülbier, Hitz, and Sellhorn (2009), Singleton-Green (2010), Ewert and Wagenhofer (2012), and Trombetta, Wagenhofer, and Wysocki (2012).

9 For a more in-depth discussion of alternative measurement concepts, see Whittington (2007).

10 Only a limited number of studies included in this review refer to the three-step fair value hierarchy, as its introduction by standard setters postdates the sample period of most earlier studies.

11 Results in Altamuro and Zhang (2013) and Lawrence, Siriviriyakul, and Sloan (2016) suggest that level 3 fair values are not necessarily less useful to investors than level 2 fair values. However, as these studies analyze fair values of financial assets, it is unclear whether these results apply to long-lived operating assets.

12 Under the revaluation model, some negative fair value changes are recognized in profit or loss, i.e., to the extent they represent impairment losses under IAS 36, Impairment of Assets.

13 However, IAS 40 requires fair values of investment property assets to be disclosed in the footnotes. Under this ‘cost model plus fair value disclosure’, only the fair value (level) is disclosed, not the corresponding fair value changes. In contrast, the cost models within IAS 16 and IAS 38 do not require footnote disclosure of fair values.

14 For a more in-depth overview of fair value measurement in IFRS, see Cairns (2007).

15 For detailed descriptions of historical developments of fair value measurement, see Barlev and Haddad (2003), Alexander (2007), Zeff (2007), Georgiou and Jack (2011), Whittington (2015), and Watts and Zuo (2016).

16 Australia and the U.K. represent two important settings that have been used in research on asset revaluations. For example, Benson, Clarkson, Smith, and Tutticci (2015) review evidence on financial reporting from the Asia-Pacific region and cover studies on asset revaluations in Australia. Domestic accounting standards in several other countries also allow asset revaluations. For instance, Barlev et al. (2007) examine asset revaluations using firms from 35 countries.

17 In contrast to FRS 102, the accounting standard preceding FRS 102 (SSAP 19 Accounting for Investment Properties) mandated a revaluation model for investment property.

18 U.S. GAAP does not provide a separate accounting standard for investment property, but recently considered the adoption of an accounting standard mandating fair value measurement for such assets. The FASB issued the corresponding exposure draft (Proposed Accounting Standards Update Real Estate – Investment Property Entities (Topic 973)) in October 2011, but decided to remove the project from its agenda in January 2014.

19 We thank an anonymous reviewer for this point.

20 Note that, in this sense, the term ‘consequences study’ is something of a misnomer for value relevance studies, as associations per se do not establish causality. For a more detailed discussion of the value relevance approach, see, for example, Barth et al. (2001) and Holthausen and Watts (2001).

21 Again, note that the term ‘reliability’ is no longer consistent with the status of the IFRS Conceptual Framework.

22 Whereas standard-setting implications of value-relevant fair values are limited, they do include the following: First, these statistical associations speak to the confirmatory value, or feedback value, of fair value information. Second, assuming external validity, they suggest that fair values reflect important economic constructs for public firms. Because non-public firms do not have observable stock prices, this suggests that fair values can be used as indicators of value or changes in value, which is useful in contracting and monitoring. (We are grateful to an anonymous reviewer for the second point.)

23 For an overview of event study methodology and approaches, see, for example, Campbell, Lo, and MacKinlay (1997) and Kothari and Warner (2007).

24 For an in-depth discussion of issues related to causal inference, see, for example, Gassen (2014) and Gow, Larcker, and Reiss (2016).

25 The SSCI covers more than 3,000 journals from 55 social science disciplines (Thomson Reuters, 2014, p. 4). We base our search on the SSCI as it represents a broad and transparent source of relevant literature. As further described below, while we explicitly point out the studies we have identified via our SSCI search approach (see Appendix A), we also refer to studies outside the scope of the SSCI (e.g., studies included in journals or books not covered by the SSCI, as well as pertinent working papers) throughout the review.

26 This additional search approach identifies studies that do not exhibit the keywords used in our initial keyword search. Further, for some journals, the SSCI does not cover all years. Our additional search approach identifies relevant studies published (in an SSCI journal) in a year not covered by the SSCI. In this step, we also identify some studies examining the footnote disclosure of discounted cash flow (DCF) estimates for proved oil and gas reserves under SFAS 69 Disclosures about Oil and Gas Producing Activities and ASC 932 Extractive Activities—Oil and Gas, respectively (e.g., Boone, 2002; Patatoukas, Sloan, & Zha, 2015). We do not include these studies in our review, as these DCF disclosures are not defined as fair values (i.e., not based on an exit price notion as is the definition in IFRS 13 and ASC 820). Further, they are characterized by industry-specific accounting requirements (deviating from IFRS 13 and ASC 820). In particular, there is a mandate to apply a uniform discount rate of 10%. Also, the DCF estimates are based on current oil and gas prices, i.e., prices derived in active markets.

27 For example, Kvaal and Nobes (2010), Kvaal and Nobes (2012), and Nobes and Stadler (2013) provide evidence on firms’ accounting choices under IFRS. Examining a broad set of accounting choices, these studies do not specifically focus on fair value measurement for long-lived operating assets. However, the studies also examine firms’ fair value-related accounting choices for property, plant and equipment and investment property and, thus, provide relevant evidence on the economic relevance of fair value measurement for long–lived operating assets (discussed in Section 3.2).

28 For example, with regard to asset revaluations of long-lived operating assets under Australian GAAP, Easton, Eddey, and Harris (1993, p. 12) document that for their 72 industrial sample firms, the number of firms revaluing their assets ranges from 25 to 44 over the period from 1981 to 1990, indicating that asset revaluations represented a common economic phenomenon under Australian GAAP at this time.

29 The IASB provides jurisdiction profiles, i.e., country-level profiles describing whether and to what extent (e.g., use in consolidated and/or separate financial statements) a country mandates or allows firms to use IFRS. These profiles are available at http://www.ifrs.org/use-around-the-world/pages/jurisdiction-profiles.aspx.

30 This indicates that the publicly traded real estate industry considers the fair value model best practice. Also, the industry association of European real estate firms, the European Public Real Estate Association (EPRA), recommends the use of the fair value model (EPRA, 2014, p. 20). While studies on the real estate industry point out that several firms which initially employed the cost model switched to the fair value model over time (e.g., Israeli, 2015, p. 1490; Müller et al., 2015, p. 2423), they also document that some firms decide to not to follow the industry best practice, instead continuing to employ the cost model.

31 According to Cairns et al. (2011, p. 14), the one Australian firm not measuring its biological assets at fair value recognizes these assets as part of PP&E (measured at depreciated cost).

32 Possible exceptions could include homogeneous, interchangeable items such as taxi licenses, fishing licenses, or greenhouse gas emission rights.

33 Based on a similar argument, IFRS and U.S. GAAP abstain from requiring fair value measurement for financial assets held to maturity as well as loans and receivables carried under an initiate-and-hold business model.

34 Note that, consistent with conceptual frameworks at the time, these studies referred to ‘reliability’ as the pertinent qualitative characteristic of fair value information. As discussed above, this term has been replaced by ‘faithful representation’. We use ‘reliability’ where the studies being discussed use it, noting that the terms are not equivalent.

35 NAV, computed as fair value of investment property less fair value of liabilities, represents a widely used metric in the real estate industry (Liang & Riedl, 2014, p. 1155).

36 However, consistent with the reduction in performance measures, conducting an asset revaluation increases firms’ book value of total assets and equity and, thus, (potentially) their visibility, e.g., for regulators. Brown et al. (1992, p. 40) also acknowledge that firm size is a noisy measure for political cost and use proneness to strikes as an additional measure. They find some evidence that firms operating in industries which are more prone to strikes are more likely to perform an asset revaluation.

37 We return to the debt contracting aspect in section 3.4.2.

38 Quagli and Avallone (2010) recognize validity concerns with regard to the interpretation of the market-to-book ratio in the real estate industry and point out that under depreciated cost measurement (prior to mandatory IFRS adoption) the relation between information asymmetry and the market-to-book ratio might be distorted. They state that in this context high market-to-book ratios might indicate ‘growth opportunities associated with a fair estimation of investment properties and therefore with a lower information asymmetry’ (Quagli & Avallone, 2010, p. 479).

39 Recall that the term ‘consequences study’ is a misnomer for value relevance studies, as they cannot establish causality.

40 For in-depth reviews of the evidence on the value relevance of financial asset fair values, see Barth et al. (2001, pp. 83–84), Ryan (2011, pp. 277–285), and Beatty and Liao (2014, pp. 357–358).

41 Illustrating this issue, Lawrence, Siriviriyakul, and Sloan (2016) replicate portions of Song et al. (2010), providing evidence that lower coefficients on Level 3 assets are the result of correlated omitted variables. (We thank an anonymous reviewer for pointing this out.)

42 For a review of the evidence on the use of financial reporting information in debt contracts, see, for example, Shivakumar (2013) and Christensen, Nikolaev, and Wittenberg-Moerman (2016).

43 Net worth represents a balance sheet-based financial covenant. It is computed as total assets less total liabilities and can incorporate additional adjustments, such as the exclusion of certain assets (e.g., Demerjian 2001, p. 183).

44 In a related study on goodwill adjustments, Frankel, Seethamraju, and Zach (2008) document a negative relation between firms’ exposure to goodwill and the likelihood to exclude goodwill adjustments in covenants, suggesting that goodwill information is not perceived as detrimental. However, Frankel et al. (2008) also document that the exclusion of goodwill adjustments has increased after the promulgation of SFAS 141 Business Combinations and 142 Goodwill and Other Intangible Assets.

45 In contrast to the findings in Goncharov et al. (2014), evidence for U.S. banks’ financial assets in Ettredge, Yang, and Yi (2014) documents a positive association between audit fees fair value exposure. Ettredge et al. (2014) also find that this association is increasing in fair value level, i.e., level 1, level 2, and level 3, consistent with the audit of the latter requiring greater effort. However, as discussed above, a causal interpretation must be made with caution.