In accordance with IAS 39 and its replacement IFRS 9, financial liabilities are accounted for at amortised cost or fair value. When they are accounted for at fair value, any changes in fair values (which can arise from changes in the credit risks of the reporting entity or changes in market conditions, such as movements in risk-free interest rates) are recognised as gains and losses in the statement of profit or loss.
When the reporting entity’s own credit risk increases, the fair values of financial liabilities issued by the entity decrease, with a corresponding gain reported in the profit or loss statement. The reverse happens when the reporting entity’s own credit risk decreases. The reporting of a gain (or loss) when the entity’s credit risk increases (or decreases) creates a counterintuitive effect on the profit or loss statement.
Several studies suggest that investors may misinterpret the financial statement effects of changes in own credit risk. For example, Barth, Hodder and Stubben (2008) found that the stock returns of firms with more debt were less negatively related with their own credit risk changes.
The impact of this counterintuitive effect could be significant in the case of large gains and losses in fair value arising from own credit risk of liabilities, especially for banks. For example, JP Morgan Chase reported a US$2bn fair value gain from its debt as a result of worsening credit risk in 2008, and losses when its credit risk subsequently improved.
In response to concerns about the counterintuitive treatment of an entity’s own credit risk in financial liability measurement, the International Accounting Standards Board (IASB) specified in IFRS 9 that for financial liabilities optionally designated at fair value, the portion of fair value change attributable to changes in the entity’s own credit risk is to be recognised in other comprehensive income (OCI), with no subsequent recycling.
The final completed version of IFRS 9 was issued in 2014. Early adoption was allowed for entities to report fair value changes arising from their own credit risk in OCI for financial liabilities designated at fair value. A similar amendment to US GAAP in the form of Accounting Standards Update 2016-01, Recognition and Measurement of Financial Assets and Financial Liabilities, took effect in fiscal years beginning after 15 December 2017 with early application permitted.
We conducted a study to examine the company characteristics that determined voluntary early adoption of reporting fair value changes arising from changes in own credit risk through OCI. We focused on banks as they would have significant financial liabilities. We extracted a list of US and non-US banks (including Canadian, Australian and European) with stock ticker codes, total assets and total liabilities data from Bureau van Dijk’s Osiris database.
Early adoption was allowed for entities to report fair value changes arising from their own credit risk in OCI for financial liabilities designated at fair value
We then determined whether a bank adopted early, and the date of that adoption, by carrying out keyword searches in the annual reports of non-US banks and the 10-K reports of US banks.
For example, note 1 on page 73 of Australia and New Zealand Banking Group reads: ‘The classification and measurement requirements for financial liabilities under AASB 9 are largely consistent with AASB 139 with the exception that for financial liabilities designated as measured at fair value, gains or losses relating to changes in the entity’s own credit risk are included in other comprehensive income. This part of the standard was early adopted by the Group from 1 October 2013.’
We also hand-collected financial data such as total assets, total liabilities, leverage and net profit after tax from annual financial statements along with one-year probabilities of default from Bloomberg as the measure of the banks’ credit risks.
Only 29 out of the total Osiris database of 1,044 banks went for early adoption of reporting fair value changes arising from changes in own credit risk in OCI. We matched these banks to a control sample of 29 banks that were not early adopters, and used a probabilistic empirical model, the logit model, to examine the relationship between the propensity to adopt early and banks’ characteristics in the year before early adoption.
Early adopters appeared to be larger and of higher credit risk than the non-early adopters. As riskier banks tend to have higher earnings volatility, this suggested that banks adopted early in order to minimise earnings volatility by reporting fair value changes in own credit risk of a financial liability in OCI instead of net income, as allowed under IFRS 9.
We also found that 66% of the early adopter banks recorded losses from own credit risk during the year of early adoption. The recognition of losses from own credit risk in OCI instead of profit or loss gave rise to positive income statement effects.
Six out of seven US early adopters and 13 out of 22 non-US early adopters recorded losses from own credit risk. Fair value gains or losses from own credit risk classified in profit or loss are not disclosed separately for non-early adopters.
An overwhelming majority of US early adopters and a smaller majority of non-US early adopters therefore avoided losses from own credit risk by early adoption. Of the six US early adopters that recorded a positive income effect in the first year, five of them also continued to record a positive income effect in the second year. Thus, US banks appeared to adopt early to avoid reporting losses in their profit or loss statements.
Banks’ early adoption of fair value accounting for financial liabilities designated as fair value through profit or loss is consistent with the notion that they did so to minimise earnings volatility on larger and riskier liabilities. A further conclusion is that US banks were likely to adopt early in order to avoid reporting losses in the profit or loss statement.