A Tale of Two Banks: When Credit Loss Models Meet Economic Crises

Journal of Accounting Research, Forthcoming

52 Pages Posted: 12 Nov 2025

See all articles by Chen Chen

Chen Chen

Monash University - Department of Accounting

Difang Huang

Chinese Academy of Sciences (CAS) - Academy of Mathematics and Systems Science (AMSS)

Date Written: February 28, 2022

Abstract

Policymakers and researchers are concerned that the expected credit loss (ECL) approach may exacerbate procyclicality. Using administrative loan-level and firm-level data in China, we find that banks adopting the ECL model reduced their credit supply and became more prudent in lending decisions after the onset of the COVID-19 pandemic, compared to banks using the incurred credit loss (ICL) approach. Our findings are more pronounced for banks that experienced greater loan loss provisions induced by ECL and for firms with higher credit risk. The credit contraction persisted throughout our sample period. We further document that firms more exposed to ECL banks experienced larger reductions in loans, assets, liabilities, and revenue after the pandemic began than those more exposed to ICL banks. These findings support the conjecture that the ECL approach may exacerbate procyclicality.

Keywords: expected credit loss, loan loss provisions, procyclicality, real effects, COVID-19

Suggested Citation

Chen, Chen and Huang, Difang, A Tale of Two Banks: When Credit Loss Models Meet Economic Crises (February 28, 2022). Journal of Accounting Research, Forthcoming, Available at SSRN: https://ssrn.com/abstract=5719602 or http://dx.doi.org/10.2139/ssrn.5719602

Chen Chen

Monash University - Department of Accounting ( email )

Building H, Level 3
Caulfield campus
Melbourne, Victoria 3800
Australia

Difang Huang (Contact Author)

Chinese Academy of Sciences (CAS) - Academy of Mathematics and Systems Science (AMSS) ( email )

Beijing
China

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
183
Abstract Views
526
Rank
422,345
PlumX Metrics