The effect of the current expected credit loss model on banks’ loan loss recognition timeliness

Abstract

The switch from the incurred credit loss (ICL) to the current expected credit loss (CECL) model is a momentous change in bank accounting in the U.S. that aims to improve banks' loan loss recognition timeliness. In this paper, we examine whether the switch achieves the intended objective. Using novel hand-collected data on CECL adoption by public U.S. banks, we find that banks that voluntarily adopt the CECL model during the COVID-19 pandemic improve their loan loss recognition timeliness. This effect is more pronounced for riskier banks or banks with a higher proportion of loans individually evaluated for impairment, suggesting that eliminating the ICL's post-lending "trigger event" requirement for recording loan losses enhances banks' loan loss recognition timeliness. The effect also is more pronounced for banks that use the CECL transition provision to mitigate concerns about inadequate regulatory capital after recognizing additional loan losses. In addition, we document that CECL-adopting banks make a larger day-one adjustment to their loan loss allowance if, under the ICL regime, their loan loss recognition was less timely, consistent with these banks experiencing a larger catch-up effect in their loan loss allowance at the start of CECL adoption. Finally, we find that CECL-adopting banks reduce their lending, possibly due to concerns about having to record large expected loan losses during the COVID-19 pandemic. Overall, our study offers new insights into how the switch to a more forward-looking credit loss model affects banks' accounting practices.