Industry

IFRS 9 expected credit loss in SAP Analytics Cloud: staging, ECL and model structure

· 2 min read · SAC Templates Hub

IFRS 9 made impairment forward-looking: instead of waiting for a loss, banks must book an expected credit loss from day one and update it as risk migrates. Reporting it in SAP Analytics Cloud means modelling stages, drivers and a calculation that must stay correct at every level of aggregation. Here is the structure that holds up to audit.

The building blocks

IFRS 9 sorts exposures into three stages. Stage 1 (performing) carries a 12-month ECL; Stage 2 (significant increase in credit risk) and Stage 3 (credit-impaired) carry a lifetime ECL. The loss itself is built from three drivers: PD (probability of default), LGD (loss given default) and EAD (exposure at default). The headline figure, ECL, is their product, discounted.

Dimensions: stage, portfolio, vintage

Model a Stage dimension (1, 2, 3), a Portfolio or product dimension (mortgages, consumer, corporate), an Internal rating grade, a Vintage or origination-year dimension, plus the mandatory time dimension. This granularity is what lets risk teams see ECL move as exposures migrate between stages, and drill into the portfolio driving the change.

ECL is a calculated measure, never a sum of drivers

Store PD, LGD and EAD as their own measures, then define ECL = PD × LGD × EAD as a calculated measure computed at the displayed level. Like every regulatory ratio, the coverage ratio (ECL over gross exposure) must be recomputed from aggregated components, not obtained by averaging the rows. The same discipline underpins the ratios in Basel III reporting, which sits right next to IFRS 9 on most banks' regulatory stack.

Stocks vs flows: the aggregation trap

Gross exposure and the ECL allowance are balances — point-in-time stocks. Across the time dimension they must use a LAST aggregation, not a sum of months, or your year-end allowance comes out twelve times too high. Stage transitions (the flow of exposures moving from Stage 1 to Stage 2) are sums. Mixing these up is the most common modelling error; our deep-dive on the time-dimension aggregation exception shows how to set it.

Versions for scenarios and overlays

IFRS 9 ECL is inherently scenario-based (base, upside, downside, with weights). A Planning model with a Version dimension lets you hold each macro scenario and any management overlay side by side and measure the weighted ECL — something a read-only Analytics model cannot do.

Start from a maintained regulatory base

The bounded structure — stages, drivers, the LAST exceptions, the calculated ECL — is where the time and the audit risk concentrate. Begin with the free banking regulatory template to see the approach, and for a full IFRS 9 build use our verified & maintained kits, updated at every revision of the standard. Insurers running IFRS 17 alongside will recognise the same movement-based design.

Save time with a ready-to-use template

64 SAP Analytics Cloud templates for 16 industries, already structured following these best practices.

Explore the catalog →

Also worth reading