IFRS 9 shifted the recognition of impairment from incurred loss to expected loss. The conceptual change is well known. What tends to be underestimated is its operational consequence: an institution no longer records what happened, but what its model projects will happen. And a model is, above all, a machine that consumes data.

Three stages, three different requirements

Classification into stages —from initial recognition to credit impairment— requires keeping the history of every operation over time. Determining whether there has been a significant increase in credit risk is not a snapshot: it is a comparison against the moment of origination, which may be years back.

This makes mandatory something many institutions treated as optional: preserving, with quality and without ambiguity, the historical traceability of the portfolio.

Where implementations fail

• Insufficient historical depth. Series that are too short or have methodological breaks that invalidate the calibration.
• Forward-looking information without governance. Macroeconomic scenarios go into the model, but nobody documents who approved them or the criteria used to weight them.
• Segmentations that age. Risk groupings defined once and never reviewed, even though the portfolio has changed.
• Adjustments outside the system. Manual corrections applied to the model output with no record or reproducible justification.

A supervisor rarely disputes the formula. It disputes the evidence: what data went in, who validated it and why the result was adjusted.

From calculation to report

The model output has to reach the financial statements, the disclosures and the prudential reports, and match across all three. That reconciliation is where the most time is lost when the calculation lives in an analytical environment separate from the reporting process.

Integrating both worlds —model and reporting— on a common data layer eliminates reconciliation as a task and turns it into a property of the process. It is the approach we apply at banks, financial companies and credit unions subject to IFRS 9 and local regulation. Let's talk about your implementation.