IFRS 9 replaced IAS 39's incurred-loss model with a forward-looking expected credit loss (ECL) model — a loss allowance is recognised before a default actually happens, based on reasonable and supportable information about the future, not just historical experience. Which of the two available approaches applies depends on the type of financial asset.

The three-stage general approach

For most financial assets — loans, debt investments, most receivables outside the simplified-approach categories below — IFRS 9 uses a three-stage model:

What "significant increase in credit risk" actually means

This is a relative test, not an absolute one: it compares the risk of default at the reporting date to the risk of default at initial recognition — it is not simply "is this asset risky." A loan that was high-risk from day one and remains equally high-risk has not necessarily migrated to Stage 2; a loan that started very low-risk and has deteriorated meaningfully may have, even if its absolute risk level is still moderate. There is a rebuttable presumption that credit risk has increased significantly when contractual payments are more than 30 days past due.

The simplified approach for trade receivables

For trade receivables and contract assets that do not contain a significant financing component, the simplified approach is mandatory: measure the loss allowance at an amount equal to lifetime ECL from initial recognition, with no staging assessment at all. For trade receivables with a significant financing component, and for lease receivables, the simplified approach is available as a policy choice.

Building a provision matrix

A provision matrix is the standard practical expedient for the simplified approach:

  1. Group receivables by shared credit risk characteristics — customer type, geography, product line, or credit rating, whichever actually correlates with historical loss patterns in your business
  2. Within each group, bucket by days past due (current, 1-30, 31-60, 61-90, 90+)
  3. Calculate a historical loss rate for each cell, from actual write-off experience over a representative period
  4. Adjust each rate for current and forward-looking macroeconomic factors that are reasonably expected to affect collectability — this step is what distinguishes IFRS 9 from a purely backward-looking bad-debt estimate, and it is not optional
  5. Apply the adjusted rates to the current outstanding balance in each cell

Where this typically goes wrong

The IFRS Advisor on Alpha Agentic AI Advisor builds provision matrices and stage assessments directly from your aged receivables and historical loss data — flagging where a cell's grouping looks inconsistent with its own loss history rather than accepting the grouping as given. Start a free trial to run it on your own portfolio.