Fair Value Accounting and Earnings Persistence: Evidence from International Banks

Using hand-collected data from a sample of 210 international banks during the period 2009 to 2013, we investigate whether fair value exposure, the proportion of financial assets measured at fair values, is associated with earnings persistence and whether the reliability of fair value measurements influences earnings persistence. We also examine whether the association between fair value measurements and earnings persistence is a function of institutional factors such as legal enforcement, the audit environment, and country-level auditor industry expertise. Results suggest that the use of fair values for balance sheet financial instruments enhances earnings persistence. Also, we find that the nondiscretionary fair value Level 1 assets (measured with observable inputs) are positively associated with earnings persistence, whereas the Level 2 assets (measured with indirectly observable inputs) and Level 3 assets (measured using unobservable inputs) are not associated with earnings persistence. We provide further evidence that there is a strong association between factors reflecting countrywide institutional structures and the predictive power of fair values based on discretionary measurement inputs (Level 2 and Level 3 assets) and we find that the moderating effect from these institutional factors is greater for Level 3 assets than for Level 2 assets. Additional tests suggest that the association between fair value estimates and earnings persistence is moderated by the classification of fair value assets (that is, through profit and loss versus other comprehensive income) and the reliability of fair value estimates.