IFRS 9 (Financial Instruments) and the ECL Model#
During the 2008 Global Financial Crisis, banks suffered catastrophic collapses. A major contributor to this disaster was the accounting framework at the time (IAS 39), which relied on the "Incurred Loss" model.
Under that old model, a bank could not record a loss on a mortgage until the borrower actually missed a payment (the loss was incurred). Even if the economy was visibly crashing and unemployment was skyrocketing, banks were forced to keep the loans valued at 100% on their balance sheets until the actual default occurred. It was a classic "too little, too late" scenario.
To ensure this never happened again, the IASB issued IFRS 9, introducing the revolutionary Expected Credit Loss (ECL) model.
The Predictive Power of ECL#
IFRS 9 forces banks and corporations to look into the future. You no longer wait for a default. The moment a bank issues a loan, they must immediately model the probability of that loan defaulting in the future and record an "Expected Credit Loss" provision on day one.
The Three-Stage Approach#
The ECL model classifies loans into three stages:
- Stage 1 (Performing): The loan is healthy. The bank must record a provision equal to the expected credit losses resulting from default events possible within the next 12 months.
- Stage 2 (Underperforming): The loan's credit risk has increased significantly since it was issued (e.g., the borrower lost their job, or the macro-economy is in a recession). The bank must drastically increase the provision to cover expected losses over the entire lifetime of the loan.
- Stage 3 (Non-Performing): The loan has actually defaulted. The provision remains based on lifetime expected losses, but interest revenue is severely curtailed.
Implementing the IFRS 9 ECL model requires massive integration of macroeconomic data and predictive AI algorithms. It forces banks to take the hit to their P&L before the crisis fully materializes, providing a massive buffer to the global financial system.