A Survey of Research on Fair Value Accounting for Financial Institutions^*

3.1 Application of FVA by Nonfinancial Firms

For a typical industrial firm reporting under US GAAP, little of the firm's balance sheet will be reported at FV. Generally, only investments of surplus cash in trading securities and derivatives will use FVA at each reporting date. Other assets will be recorded at FV only when written down through the impairment process. For a similar industrial firm reporting under IFRS, in addition to the above, firms will carry biological assets at FV, as well as investment properties, PP&E, and intangible assets (if elected). Since few firms elect to carry PP&E and intangible assets at FV because of estimation challenges and investor skepticism (Christensen & Nikolaev, 2013), FVA under IFRS has little practical difference from US GAAP, except in specific industries such as farming (due to biological assets) and real estate (due to investment properties).

McDonough et al. (2020) provide descriptive statistics on the extent of FVA application in US nonfinancial firms. In 2017, nonfinancial firms in the United States reported 19% of their assets and 5% of their liabilities at FV. Most nonfinancial sectors reported only 3%–5% of their assets at FV, whereas the health care and business equipment industries reported the highest proportion of FV assets, with 44% and 20% of total assets, respectively. The health care industry also reports the highest proportion of FV liabilities, with 9% of total liabilities. Most of the FV assets for US nonfinancial firms are measured using Level 1 or Level 2 inputs, while FV liabilities are typically measured using Level 2 or Level 3 inputs.

3.2 Application of FVA by Financial Institutions

FVA affects financial institutions to a much greater extent than industrial firms due to extensive holdings of financial instruments. Derivatives and investments that are held-for-trading (HFT) are recorded at FV, with unrealized FV changes recorded in net income, while available-for-sale (AFS) investments are recorded at FV, with unrealized FV changes recorded in other comprehensive income (OCI). Investments that are held-to-maturity (HTM) are carried at amortized cost. Similarly, most loans are carried at amortized cost (except for impaired loans, which are written down to FV). While SFAS 159 and IAS 39 allow banks to report certain loans at FV, most banks do not make such elections (Laux & Leuz, 2010).

Financial institutions include bank holding companies, commercial banks, savings banks, investment banks, insurance companies, brokerage firms, and advisory and other financial services companies. We now discuss statistics from several papers based on different samples of financial institutions; in particular, we present statistics from samples of financial institutions, bank holding companies, and banks. The statistics for financial institutions are based on samples containing all types of financial services firms. While there is a considerable overlap between banks and bank holding companies, bank holding companies report consolidated statements that include non-bank subsidiaries.

For large US bank holding companies from 2004 to 2006, as an average percentage of total assets, HFT assets were 7%, AFS assets were 15%, HTM assets were 0.1%, and loans and leases were 47% (Laux & Leuz, 2010). For large US investment banks, HFT comprised 16% of total assets (Laux & Leuz, 2010). By 2017, US bank holding companies on average reported trading assets of 0.3%, AFS securities of 15%, HTM assets of 3%, and net loans and leases of 70% (McDonough et al., 2020). Notably, US bank holding companies report almost no trading assets on average in 2017, while large US banks carry 3% in trading assets (McDonough et al., 2020)—both of which are significantly lower than in the pre-GFC period. These statistics are largely similar for non-US banks reporting under IFRS.

In 2017, McDonough et al. (2020) showed that US financial institutions reported 27% of their assets at FV, which is greater than the proportion reported by nonfinancial firms (other than firms in the health care industry). Focusing on US banks, specifically, the proportion of assets carried at FV is similar to nonfinancial firms, with less than 1% HFT (3% for large banks and less than 1% for smaller banks) and 15% AFS (McDonough et al., 2020). Similar statistics are found for a longer sample period from 1984 to 2013 (Laux & Rauter, 2017). Only large commercial banks report a higher proportion of FV assets (29%) compared to the average proportions for nonfinancial firms (19%) (Laux & Rauter, 2017).

Liabilities are rarely carried at FV unless they are out-of-the-money derivatives, liabilities associated with a portfolio of assets, or liabilities designated at FV using the FV option under SFAS 159 or IAS 39. In 2017, US financial institutions reported only 3% of their liabilities at FV, which is lower than the 5% of liabilities reported at FV by nonfinancial firms (McDonough et al., 2020). US bank holding companies reported an even lower proportion of 0.5% of their liabilities at FV. While SFAS 159 and IAS 39 open the door for reporting liabilities at FV, the actual proportion remains low (Couch et al., 2017; Schneider & Tran, 2015; Xu, 2019). US banks and bank holding companies rely heavily on Level 2 inputs to measure FV assets and liabilities. In 2017, these firms used Level 2 inputs for approximately 90% of FV assets and 80% of FV liabilities.

Overall, the pure form of FVA where all assets and liabilities are carried at FV is rare. Even a sample of large US commercial banks and bank holding companies carried at most 30% of assets at FV (Laux & Leuz, 2010; McDonough et al., 2020). This percentage is much lower for small financial institutions and most industrial firms. In recent years, the percentage of assets and liabilities reported at FV has declined even for large financial institutions (McDonough et al., 2020). Level 2 inputs are the most common inputs used to measure FV assets and liabilities, while Level 3 inputs are the least common in measuring FV assets, and Level 1 inputs are the least common for FV liabilities.

4.1 Earnings and Balance Sheet Volatility

FVA has the potential to increase volatility in financial statements since changing market conditions are reflected as they occur. In contrast, HCA inherently provides more smoothing of financial statements due to the general use of the original purchase price until disposal. Furthermore, changes to recorded values under HCA are often consistent and systematic (e.g., depreciation or amortization). FVA increases the volatility of an average bank's earnings by 26% (Barth et al., 1995). The increased volatility is theorized to come from (1) inherent volatility, (2) estimation error volatility, and (3) mixed measurement volatility (Barth, 2004).

Inherent volatility relates to changes in the underlying economic value of financial statement items. The reflection of such volatility is generally believed to provide useful information to investors; however, such volatility can be misleading for HTM assets (ECB, 2004). Furthermore, inherent volatility may be artificial for insurance companies because of the use of reinsurance and the time horizon of asset holdings, whose values are believed to be mean reverting over the long term (Trainar, 2008). Estimation error volatility stems from errors in determining FV, particularly for FVs relying on unobservable inputs (i.e., Level 3). All being equal, such estimation errors will increase volatility under FVA relative to HCA due to the lack of estimation in the latter. Finally, mixed measurement volatility results from using a mixture of FVA and HCA, which is common in practice. Using FV for assets but not for the corresponding liabilities can cause greater volatility relative to a pure FV model.5 Barth (2004) argues that the volatility associated with FVA is not necessarily bad since true changes in value provide information to financial statement users to assist in their capital allocation decisions. The challenge is in minimizing the volatility arising from estimation error and mixed measurement.

Using pro-forma analysis, pure FV for US banks is found to result in net income (comprehensive income) that is five times (three times) more volatile than under existing accounting rules (Hodder et al., 2006; McInnis et al., 2018). However, this incremental volatility is positively associated with market beta and the volatility of stock returns, suggesting that pure FV captures components of firm risk that are priced by investors (Hodder et al., 2006). In a simulation exercise, the ECB (2004) also finds increased volatility under a pure FV model.

Both US GAAP (SFAS 159) and IFRS (IAS 39) offer an FV option (FVO) for financial instruments. By using the FVO, companies can choose to report a financial instrument using FV rather than another measurement base such as amortized cost. The FVO is meant to alleviate the mixed-measurement volatility described above. Lending support to this idea, earnings volatility is found to decrease for financial institutions that adopted the FVO to reduce an accounting mismatch (Fiechter, 2011), and simulations provide further support (IMF, 2008). On the downside, US banks are found to use the FVO in an opportunistic way (J. S. Song, 2008). In addition, earnings volatility increased for US financial institutions that used the FVO, relative to non-adopters (Couch et al., 2017). The authors find that the increase in earnings volatility is primarily driven by firms that used the FVO for assets only, which would not necessarily mitigate the mismatch arising from mixed measurement.

Overall, academic research finds that FVA generally increases the volatility of earnings and the balance sheet, despite standard setters providing the FVO to reduce the volatility resulting from mixed measurements. In addition, contrary to the intended purpose of the FVO, some firms appear to use it in a manner that creates more, rather than less, measurement mismatch, which likely induces more volatility. Whether or not the increased volatility provides incremental information for capital market participants remains an open question. Stock market reaction studies (e.g., Beatty et al., 1996) indicate that investors view the volatility associated with FVA as negative; however, such reactions could stem from other factors associated with FVA, such as anticipated management behavior.

4.2 Regulatory Capital

Much of the contagion concerns surrounding FVA are predicated on increases in the volatility of regulatory capital, which will increase the likelihood of capital benchmark violations. Theoretical work typically presumes a direct relationship between FVs and regulatory capital; however, many differences exist in practice (see Laux, 2012). Some reporting regimes use different rules for financial reporting and regulatory capital, while other regimes start with financial reporting under US GAAP or IFRS and make adjustments to determine regulatory capital.6 Common adjustments include the risk weighting of assets and the exclusion of unrealized gains stemming from changes in an entity's own credit risk. In addition, unrealized losses on debt securities classified as AFS often do not affect regulatory capital unless such losses are realized or considered other-than-temporary. In summary, regulatory capital rarely has a direct one-to-one relationship with financial reporting.

In the United States, banks are governed by the following rules for regulatory capital: FV losses on HFT securities and AFS equity securities affect regulatory capital. FV losses on AFS debt securities do not affect regulatory capital unless such losses are deemed as other-than-temporary impairments (OTTIs). Similarly, losses on HTM securities, which are carried at amortized cost for financial reporting, affect regulatory capital only when impaired. Loans carried at amortized cost affect regulatory capital through the deduction of banks' loan loss provisions. Under the US implementation of Basel III, advanced approach banks are required to include unrealized gains and losses (UGL) for AFS assets in regulatory capital.7 Regulators have recently proposed to expand this requirement to all banks with at least $100 billion in total assets (Federal Reserve System, 2023). US insurance companies are subject to risk-based capital requirements, meaning that the capital ascribed to a given asset varies based on the credit quality of the underlying asset.

Analytical papers such as Allen and Carletti (2008) and Plantin et al. (2008) presume pure FVA and a direct relationship between financial reporting and regulatory capital—both presumptions are simplifications, as discussed above. Consequently, empirical work provides a necessary link to the impact of FVA in practice. In early work, Barth et al. (1995) evaluate the effect of carrying investment securities at FV rather than HC on regulatory capital. Using US banks from 1971 to 1990, they estimate regulatory capital and presume that FV changes have a direct impact. In this context, Barth et al. find that earnings would be more volatile and thus hypothetical violations of regulatory capital minimums would occur more frequently under FVA.

In an international context, Bernard et al. (1995) conducted an empirical study of Danish banks from 1976 to 1989 that operated in an environment where FVA is used for regulatory capital. The authors find that earnings became more volatile after incorporating FV changes; however, due to banks carrying regulatory capital in excess of required minimums, the authors assign only a 2% probability of an annual FV adjustment causing a capital violation.

Merrill et al. (2012) examine the selling behavior of US insurance companies from 2006 to 2009. The authors argue that the regulatory capital of insurers can be doubly impacted by decreasing values of investment assets: first, decreases in FV can cause direct reductions in regulatory capital. Second, since the credit quality of those investment assets has correspondingly declined, regulatory capital is further reduced. These two effects were found to motivate the selling of investment assets at distressed prices. Similarly, Ellul et al. (2011) find that regulatory capital constraints motivate US insurance companies to sell corporate bonds that have been downgraded in an effort to improve regulatory capital.

While these studies provide evidence that FVA can lead to higher regulatory capital volatility and increased likelihood of fire-sale of investments with deteriorating FVs, other studies present evidence of a limited impact of FVA on regulatory capital. When examining the 14 largest US banks in 2008, Shaffer (2010) finds that FV write-downs caused reductions to Tier 1 regulatory capital of between 0.1% and 3.9%. Dividends paid by these firms had a greater impact on regulatory capital than FV write-downs. Overall, the author's limited sample indicates FVA does not bear the primary blame for the GFC. Fiechter et al. (2017) examine the 2008 amendments of IAS 39 that allowed reclassification of FV assets to HC. Using a sample of 160 European bank holding companies, they find that “too-big-to-fail” banks made less use of this option to reclassify from FV to HC, presumably because they enjoyed political protection from regulatory pressure. However, smaller banks used the reclassification option to a greater extent to protect their regulatory capital. Hence, their results indicate that the negative effects of FVA on regulatory capital may be disproportionately experienced by smaller banks with less systemic importance.

In a more comprehensive study, Badertscher et al. (2012) test the impact of FVA on banks' regulatory capital during the GFC. The authors examine AFS and HTM securities and recognize that FV changes for AFS and HTM assets only impact regulatory capital if decreases in FV are deemed as OTTIs.8 While FV changes on AFS equity securities do impact regulatory capital, such securities have little impact for their sample of firms. Using 150 US banks from 2004 to 2008, the authors find that OTTIs had only a minor impact on regulatory capital. Specifically, removing the effect of OTTIs would have increased the median Tier 1 capital ratio from 9.9% to 10.0%. The authors conclude that bad debt expense on uncollectible loans had a much greater impact on capital ratios than FVA (bad debt expense moves the median capital ratio from 9.9% to 10.7%). Furthermore, the bulk of OTTIs were recorded in the second half of 2008, which weakens the argument that FVA caused the GFC since they occurred in the latter part of the crisis.

In summary, the literature finds mixed evidence on the effect of FVA on regulatory capital. On the one hand, theoretical and empirical evidence indicate that (1) FVA can lead to greater regulatory capital volatility and increased likelihood of violations and (2) firms may engage in suboptimal fire-sale of affected assets to minimize the impact of FVA on regulatory capital ratios. This strand of literature relies on the assumption that there is a direct link between FVA and regulatory reporting; however, there are many differences in the historical application of FVA for these two purposes. On the other hand, the literature documents that FVA has little actual effect on the likelihood of regulatory capital violations. This stream of literature demonstrates that other factors, such as dividend payments and loan loss provisions, have greater effects on regulatory capital, particularly for large and systemically important banks.

4.3 Other Consequences of FVA

Recent studies consider additional consequences of FVA that include its role in earnings management and regulatory capital management, its role in compensation contracting, and various economic consequences of FVA. The first stream of literature examines the role of FVA in earnings management and regulatory capital management. For a sample of US bank holding companies, Bratten et al. (2020) find that banks with greater proportions of FV assets and liabilities substitute accrual-based earnings management through loan loss provision with transaction-based earnings management through timing the realized gains and losses on investment sales. Zhao (2019) also finds that banks trade off managing earnings by using loan loss provisions versus using securitization gains (securitization gains increase with the proportion of retained interest, which is measured at FV).

Dong and Zhang (2018) examine earnings management by US banks using AFS securities. They find evidence that banks reporting UGLs through OCI use selective selling of AFS securities to manage earnings in contrast to banks that report UGLs through net income. Robinson et al. (2018) provide evidence of capital-constrained banks using Level 3 FV estimates to manage earnings to stay just above key Tier 1 capital ratios.

DeFond et al. (2020) examine the effect of FVA on the association between earnings and executive compensation (i.e., pay-performance sensitivity). They find evidence of a negative effect of FVA on pay-performance sensitivity, indicating that FVA could impair the usefulness of earnings in managers' performance evaluation. In contrast, Henderson (2022) finds that FVs are used efficiently for compensation contracting by considering the reliability of the FV estimates.

Other studies examine the relationship between FVA and dividend payout policy. E. Chen and Gavious (2016) study dividend policy changes around IFRS adoption by nonfinancial Israeli firms. They find evidence of increased dividend payouts associated with unrealized FV gains in the post-IFRS period. X. Chen et al. (2020) examine a sample of financial firms in Australia and find a positive association between FV adjustments (i.e., net UGLs) and dividend policy. These papers may indicate an unintended consequence of FVA—namely, that firms make dividend payments based on unrealized FV gains.

Other recent studies further examine the economic consequences of FVA. Chircop and Novotny-Farkas (2016) find that banks affected by Basel III requirements to include AFS investment UGLs in regulatory capital decrease their investments in risky AFS securities. These findings are in line with Basel III requirements inducing a reduction in bank risk-taking (Acharya & Ryan, 2016). Wang and Zhang (2017) study the effect of FVA on public debt structure using new debt issuances by US firms from 2008 to 2013. They find that the use of Level 2 and Level 3 FVs is associated with lower financial reporting quality, and as a result, with higher agency cost of debt.

Gopal and Gutierrez (2019) examine the effect of FVA on loan origination in syndicated loan markets in a setting where non-bank lenders use FVA and traditional banks use HCA. Relative to HCA lenders, they find that FVA lenders are more susceptible to market price changes and reduce (increase) lending to a larger extent when market prices fall (rise). Bleck and Gao (2023) examine the impact on the lending decisions of US banks when FVA directly affects regulatory capital. They find FVA may negatively affect the ex ante quality of loan origination.

Taken together, these studies demonstrate that FVA can have significant economic consequences of either a positive or negative nature. However, more research is needed to develop a better understanding of these consequences for financial and regulatory capital reporting.

5.1 Value Relevance

At a conceptual level, if FV is the sole measurement basis, then “equity value is read from the balance sheet, with no further analysis needed, and the income statement reports realizations for determining value at risk” (Penman, 2007, p. 42). Thus, pure FVA would provide complete valuation information. Whether financial reporting should provide complete valuation information or provide the raw material that feeds into investors' own valuation processes is more of a philosophical question that has no clear answer at present. Nevertheless, in practice, FVA applies only to a limited number of assets and liabilities—far from Penman's scenario of comprehensive FV. Furthermore, Ball (2006) argues that “fair value incorporates more information into the financial statements” (pp. 19–20). Incorporating more information means that financial statements become more useful for assessing a company's value and for contracting with parties such as lenders (Ball, 2006).

Theoretical research on the value relevance of FVA is sparse. Using a one-period, one-asset analytical model, Luo (2019) shows that under conservative accounting, reported earnings are value-irrelevant and that informativeness increases as the volatility of reported earnings approaches the volatility of the underlying economic earnings. This finding implies greater value relevance of FVA (relative to HCA) since FVA better aligns the volatility of reported earnings with underlying economic volatility. However, Gao and Jiang (2020) show that FVA provides more information than HCA only when managers have little discretion.

Much of the empirical academic research supporting FVA draws its support from value relevance studies that demonstrate a stronger correlation between FVs (changes in FV) and stock prices (stock returns). Such studies focus on valuation as the primary purpose of financial statements. Consequently, stronger correlation between accounting data and stock values indicates increased usefulness for valuation purposes (Barth et al., 2001).

Numerous empirical studies starting in the mid-1990s have documented that FVs are superior to HCs for explaining the market value of firms. Early work explored investment assets and found enhanced value relevance for FVs. Barth (1994) finds securities' FVs are more value-relevant than HCs. Further research also finds value relevance for loans and core deposits (Barth et al., 1996) and derivatives (Venkatachalam, 1996). Beyond investment assets, FVs of pensions (Landsman, 1986) and stock options (Aboody et al., 2004) are found to be value-relevant.

Additional empirical research extending to long-lived assets provides mixed results. In an Australian setting, Barth and Clinch (1998) find that FV changes for intangible assets are value-relevant but FV changes for PP&E are not. Using a different methodology in the same setting, Easton et al. (1993) find value relevance for both intangible assets and PP&E. Liang and Riedl (2014) compare analyst forecast accuracy for investment property firms between UK firms that use FV versus US firms that use HC. They find that analyst net asset value (NAV) forecasts are more accurate for UK firms. Since investment property firms' NAVs are closely related to market value of equity, this finding supports the value relevance of FVA. Beyond value relevance, Aboody et al. (1999) find that increases in FV for PP&E are related to improvements in future operating performance. Since FVs theoretically represent the present value of future cash flows, this finding validates the predictive ability of PP&E fair value estimates.

More recently in a European setting, Liao et al. (2020) document that FVs are more value-relevant than HCs during the GFC, but not before. Similarly, Adwan et al. (2020) find that the value relevance of book equity increases during the GFC for firms with more exposure to FVA before the crisis. Bratten et al. (2016) find that UGLs recognized in OCI predict future profitability for up to 2 years ahead. This finding holds even for UGLs recognized during the GFC.

In contrast, in a US bank setting, Eccher et al. (1996) find no value relevance for FVs of deposits and only limited support for FVs of loans. Similarly, Nelson (1996) finds no support for the value relevance of deposits, loans, or long-term debt. McInnis et al. (2018) examine US banks over the period 1996–2013 and find that financial statements under full FVA are less value-relevant than those under the prevailing GAAP. They find the value relevance of FV net income improves when it is disaggregated into transitory and persistent components, suggesting that persistent FV changes are more useful to decision-makers. A possible explanation for the conflicting findings is that McInnis et al. (2018) compare full FVA with US GAAP (i.e., partial FVA) rather than comparing with full HCA.

Overall, most studies in this literature provide evidence supporting the improved value relevance of FVA relative to HCA. An exception is Gao and Jiang (2020), who show that the value relevance of FVA depends on the level of managerial discretion. We argue that this result does not demonstrate that FVs lack value relevance per se; rather, their results highlight concerns about FVs' reliability and the potential for manipulation. In Section 6.1, we provide a detailed discussion of reliability concerns. As another exception, McInnis et al. (2018) show that net income and equity measured at “full fair value” are not more value-relevant compared to current US GAAP; however, no firms use a full FV model in practice.

5.2 Risk Relevance

More recently, research has considered the relationship between FVA and various firm risk measures. The literature on the effect of FVA on earnings volatility is discussed above in Section 4.1. Here we discuss studies that focus on testing FVA's ability to assess firm risks, such as market risk, credit risk, and default risk. We seek to determine whether the literature can help us answer this question: Does FVA help external stakeholders assess a firm's risk exposure to a greater extent than HCA?

Hodder et al. (2006) find that higher volatility in full FV net income is positively associated with the volatility of stock returns, market beta, and long-term interest rate beta. In a US setting, Blankespoor et al. (2013) find that FVs provide more information for explaining credit spreads for financial instruments. Using non-US commercial banks cross-listed in the United States, Duh et al. (2012) find that FVs better capture firms' credit risks. Schneider and Tran (2015) and Fontes et al. (2018) find that FVA is associated with lower information asymmetry. Fontes et al. (2018) find that the reduction in information asymmetry is greater when banks also recognize gains and losses arising from their own credit risk and when they provide related narrative disclosures. Mohrmann and Reipe (2019) show that banks with larger proportions of Level 3 FV assets exhibit higher stock volatility and default risk. In summary, these studies show that FVA is associated with market-based risk measures, suggesting that FVA provides information relevant for assessing firm risk.

Overall, the above studies generally suggest that FVA provides decision-useful information for shareholders and creditors. FVs provide more timely information through regular updates of the value of assets and liabilities. Furthermore, stakeholders can analyze the characteristics of the FV time series to better determine firm risk characteristics. As a result of the above, more information is provided through the FV model (Ball, 2006). Nevertheless, numerous criticisms of FVA have been made, including reliability challenges, and encouragement of procyclicality and contagion.

6.1 Reliability

Historically, academics argued that FVA's key challenge was ensuring the reliability of FV estimates. Relative to HCs, FVs may be less reliable because values are not determined from realized transactions. Benston (2008) notes that it is difficult to establish FVs; managers must determine “imaginary prices that might be offered by hypothetical independent acquirers of its assets and liabilities who are participants in non-existent markets” (p. 103). Auditors must then find sufficient and appropriate ways to challenge and verify these values. Estimation errors can be made when FVs are required but lack traded market prices (Ball, 2006). As a specific example, Magnan (2009) points out that approximately 1 month after releasing its earnings numbers, Credit Suisse revised its 2007 operating income downward by 10% because additional work led to changes in FV estimates. This example demonstrates that estimates can be subject to a wide margin of estimation error, leading to wide swings in estimates from period to period. Financial institutions are most affected by these swings since they hold many assets and liabilities at FV, and such instruments are often complex and difficult to value.11

Furthermore, managers can exercise a significant amount of discretion when they establish FV estimates. Such management discretion can be used to make FV estimates that achieve a stated goal, such as reaching an earnings target or bonus threshold. Ball (2006) believes that FVs can be more influenced by managers in illiquid markets, which negatively affects reliability, and that FVs will be more relevant than HCs only when based on observable liquid market prices. Overall, estimation errors or biases in FV estimates reduce the reliability of FVA by introducing greater uncertainty in the reported FVs or significant directional differences between reported FVs and underlying “true” FVs.

We next explore the primary approaches used in the empirical literature to investigate the reliability of FVA. The first approach employs “direct” tests of reliability by comparing FV estimates to future realized prices, subsequent revisions, restatements of FV estimates, or associations with other measures of uncertainty for assets or liabilities reported at FV. The second approach relies on the moderating effect of reliability on the value relevance of reported FVs.

6.2 Procyclicality and Contagion

The GFC of 2007–2009 moved the focus away from criticizing the reliability of FVs to concerns about procyclicality and contagion. Procyclicality of FVA is typically described as exacerbation of boom-bust cycles (e.g., Adrian & Shin, 2010; Novoa et al., 2009; Plantin et al., 2008). In boom times, unrealized FVs increase, which leads to higher equity and lower leverage based on market values. The decreased leverage allows firms to acquire additional assets to move leverage back to a target level. The additional purchases cause asset prices to rise, which perpetuates the cycle. In a bust cycle, falling asset prices cause lower equity and higher leverage. The higher leverage necessitates asset sales or recapitalization. The asset sales cause further asset price drops, leading to a feedback effect that can cause a crisis. Further affecting price drops can be illiquidity: in periods where falling prices cause asset sales, the preponderance of sellers leads to a lack of potential buyers.

In contrast to procyclicality, contagion effects stem from asset prices deviating from fundamental value (Cifuentes et al., 2005; Laux & Leuz, 2009). In times of crisis, a firm may be required to sell assets at prices that are below fundamental value; these “fire-sale” prices then become the reported FVs for other firms (Shleifer & Vishny, 2011). The low FVs may cause firms to violate regulatory capital requirements. Thus, one firm's distress gets transferred to numerous other firms that carry the same or similar assets, causing contagion. Such distress sales that occur during a time of crisis can be accompanied by a tightening of credit (Brunnermeier, 2009), which can exacerbate financial distress and lead to more distress sales. While the very definition of FV considers only “orderly” transactions (IFRS 13, IASB 2011, para. 9)—which would typically exclude fire sales—firms (or their auditors) may be hesitant to ignore recent transactions when determining FV. The exclusion of distress sales when determining FV was made explicit by the FASB's staff position issued in April 2009 (FSP FAS 157-4, FASB, 2009). Investor reaction to FAS 157-4 is discussed below.

The debate about whether FVA creates or exacerbates procyclicality and contagion is ongoing with the academic literature finding mixed evidence. First, we discuss studies providing support for the claim that FVA causes procyclicality and contagion, including papers documenting investors' reaction to FAS 157-4. We then discuss the studies that dispute the causal effect of FVA on procyclicality and contagion.

7.1 Procyclicality, Contagion, and When Markets Are Not Orderly

Laux and Leuz (2009) identified several areas that required additional work, including whether FVA contributed to the GFC. Researchers have attempted to answer this question, but more work is needed to clearly understand if and how FVA contributed to the financial crisis. What are the precise mechanisms through which FVA leads to contagion in a financial crisis? How exactly would HCA (or other alternatives) help in avoiding or mitigating crises? The S&L crisis in the 1980s, the GFC of 2007–2009, and the 2023 US banking failures all came after significant and rapid changes in macroeconomic conditions; however, much of the FVA literature abstracts away from macroeconomic effects when considering the role of accounting in crises. More research is needed to understand whether accounting choices have incremental effects after considering macroeconomic variables. For example, during the 2023 failure of SVB, many blamed HCA for helping the bank hide its unrealized losses until it was too late. Little consideration has been given to the potential impact had FVA been used for the assets in question. Would FVA have exacerbated the problem by creating contagion effects since many other banks held similar assets on their balance sheets?

Moreover, much of the extant research fails to consider the role of banks' business models in the procyclicality of bank leverage, for which FVA is blamed for aggravating. For example, banks play a crucial role in the business cycle through their lending decisions. Bank lending generally expands during a boom because it is less risky and more rewarding to lend, and contracts during a bust because it is more risky and less rewarding to lend. This can create a mechanical relationship between bank leverage and the FVs of assets and liabilities on banks' balance sheets. While Laux and Rauter (2017) consider the role of the business cycle, more research is needed.

Under the current exit value−based definition of FV (SFAS 157 and IFRS 13), FV measurement requires estimation of the market values of assets if they are not actively traded or if markets are deemed disorderly. When estimating the exit value for assets, firms should consider the best possible use for the assets by the firm or a third party. However, managers may struggle to estimate the value of their assets from a third-party perspective, particularly for nonfinancial or long-duration assets. In this scenario, managers may be forced to rely on value-in-use when estimating asset FVs. We found no studies exploring the extent to which firms rely on value-in-use as opposed to exit value when estimating FVs. We believe such research will contribute to the literature on the relevance and reliability of FVA and inform standard setters about the feasibility of the current definition of FV.

7.2 FVO

The limited number of existing studies on the FVO provide mixed evidence on its effectiveness in reducing measurement mismatch and hence earnings and balance sheet volatility. Further research is needed to fully understand the extent and nature of how the FVO is applied. How extensively is the FVO used? Does the application of the FVO vary systematically across firms? To what extent is the FVO used opportunistically as opposed to increasing informativeness? Are investors able to undo the effects of opportunistic FVO use?

7.3 Earnings and Balance Sheet Volatility

The literature on the effect of FVA on the volatility of financial statement elements is mostly focused on earnings volatility. While the evidence on FVA's effect on earnings volatility is mixed and requires more research, there are several other potentially fruitful avenues for future work. Researchers could examine the effect of FVA on the volatility of other financial statement elements and metrics that may be based on those elements. For example, how does FVA affect the volatility of assets, liabilities, and ratios such as the current ratio? The volatility of these measures can have significant implications for firms since ratios are often used in debt contracting. For instance, would FVA increase the volatility of the current ratio, which could lead to firms holding more short-term assets to avoid covenant violations? If this is the case, does FVA improve firm resilience or create inefficiencies since, ceteris paribus, short-term assets may provide lower expected returns relative to long-term assets?

In addition, the question of whether FVA-induced financial reporting volatility is reflective of underlying economic realities and hence increases the informativeness of financial statements remains unresolved. Most of the existing research focuses on excessive downside volatility during crises, such as the GFC. Future researchers could examine the effect of FVA during normal times. Does it affect financial reporting volatility during periods of economic tranquility? What about during economic euphoria (i.e., bubbles)? In general, financial statements accurately reflecting fluctuations in companies' underlying economics may simply be bringing the volatility to the surface. The challenge is that market values can significantly diverge from fundamental values on a short-term basis. However, HCA is not immune from this problem since costs will often diverge from firm fundamentals as time moves away from the original transaction date. Future research could address these concerns by examining how financial reporting volatility under FVA and HCA maps to volatility in the underlying economic fundamentals. This line of research can also contribute to the question of whether FVA contributes to contagion during crises.

7.4 Risk Relevance

Research on risk relevance of FVs is relatively recent and includes only a few studies. We believe this is one of the most fruitful avenues for future research. FVs of assets and liabilities are closely tied to the underlying risks in those instruments, including firm-specific and macroeconomic risks. For example, the FV of an investment portfolio of corporate bonds may fluctuate due to changes in the credit risk of the individual bond issuers, due to changes in market interest rates, or due to changes in the liquidity characteristics of the bonds affecting the investors' abilities to buy/sell the bonds without significant price impact. Future research could examine the extent to which FVA reflects the above and other risks facing the reporting firm and to what extent FVA affects ex ante risk-taking behavior.

The relationship between FVA and credit risk warrants further investigation. Much of the literature focuses on the relevance of FVs for equity investors, while credit markets are significantly larger in magnitude. Do FVs provide superior information to HCs for assessing an entity's ability to repay? How often and in what capacity are FVs used in contracting with the providers of debt capital?

7.5 Disentangling Relevance versus Reliability

Much of the FVA literature deals with relevance versus reliability, which are the two fundamental characteristics of accounting information. However, the methodology used in most studies to test the reliability of FVA cannot fully disentangle reliability from relevance. For example, the weak association between market value of equity and Level 3 FVs could be driven by lower perceived reliability or lower relevance. This is not a criticism of any particular study but rather a comment on the difficulty of examining reliability and relevance in isolation. Relatively few studies (discussed in Section 6.1.1) have attempted to directly test the reliability of FVA by using realized prices when assets are subsequently sold or other market-based measures of reliability such as bid-ask spread. We encourage future researchers to identify such settings where they can study only one of the two fundamental characteristics in isolation. For example, ceteris paribus, stock price volatility or analyst forecast dispersion could be used as proxies for the perceived reliability of financial statements. However, it remains difficult to rely on market-based measures to isolate the effects of reliability since market participants make decisions based on all available inputs. We believe experimental research may be a fruitful avenue for this testing since experiments could be designed to hold one of the effects constant while varying the other.

Recent studies have examined many of the research questions posed by earlier surveys of the FVA literature. Landsman (2007) suggests that cross-country institutional differences are likely to affect the relevance and reliability of FVs, and several studies discussed in this review have provided insight. However, more work is needed to reach consensus and to understand the precise mechanisms through which institutional differences affect FVA. For example, future researchers could examine whether cross-country institutional differences directly affect the quality of FV estimates or whether they influence investors' abilities to process FV information.

7.6 Insurance Companies

Much of the existing research has focused on FVA's impact on banks, with limited attention being directed toward insurance companies—even though the impact of FVA can be heightened for these firms due to the long time horizon of their assets and liabilities (Trainar, 2008). Additional research could determine whether findings using banks extends to insurance companies and whether FVA's challenges affect insurance companies to a greater or lesser extent.

7.7 Alternatives to FVA

FVA is often criticized without providing a superior alternative accounting treatment. In general, two measurement options exist: (1) HC (which generally includes a concept that reduces the value down to FV if impaired) and (2) FV. Future research could develop new measurement concepts that share in the benefits of FVA (such as high relevance) while limiting the downsides (such as the potential for opportunism). We note that financial firms often use different measurement approaches for financial reporting versus for determining regulatory capital. Could a single measurement base exist that satisfies the needs of both sets of users? Such a measurement base would reduce costs and enhance comparability across jurisdictions. As an example, accounting for investments in long-term bonds using amortized cost has been criticized for delaying loss recognition when market interest rates increase and sale is likely. Similarly, using FV has been criticized for increasing earnings and asset volatility when the asset in question may be held to maturity. Could a measurement middle ground exist that minimizes the criticisms levied against both HC and FV?