Expected credit losses in their first real test
Loan loss provisions appear on 23 recent papers in the finance and accounting journals, IFRS 9 on 5 and CECL on 3. The expected-loss rules were designed after 2008 to make banks provision earlier. The 2022–2024 rate shock and the pandemic were their first real tests. Did provisions move earlier, or just more?
What prompted this
Since October 2025, 31 papers carry a loan-loss-provision, expected-credit-loss, IFRS 9 or CECL keyword (19 in finance journals, 11 in accounting): "loan loss provisions" on 23, "expected credit loss" on 6, "IFRS 9" on 5, "CECL" on 3. Twenty of the 31 carry a procyclicality, crisis, lending, capital or discretion keyword, so the question is on the table but with few papers.
Questions
- Did banks under IFRS 9 and CECL raise provisions at the start of the pandemic and the rate shock, and release them later, in a way that stabilized or amplified lending?
- Do the forward-looking scenarios banks use for provisions differ systematically from consensus forecasts, in the direction of smoother earnings?
- Did the "day one" provisioning for new loans reduce lending to riskier borrowers, as critics predicted?
Data that could help
Bank provisions by stage and scenario disclosures from annual reports, loan-level data, and macro forecasts at the time.
Next step
Compare quarterly provision changes of IFRS 9 banks with those of banks still under incurred-loss rules during 2020 and 2022–2023.