Ind AS 109 ECL model in practice — testing expected credit loss

Testing expected credit loss under Ind AS 109 means testing three distinct layers: first, whether financial assets were correctly allocated to Stage 1 (12-month ECL), Stage 2 (lifetime ECL, where credit risk has increased significantly since initial recognition) or Stage 3 (credit-impaired, lifetime ECL); second, whether the probability of default, loss given default and exposure at default inputs used to compute the ECL are supportable and consistent with the entity's own historical loss experience adjusted for forward-looking information; and third, whether any management overlay applied on top of the model's raw output is separately justified, since overlays are where unsupported judgement most often hides.

Testing the stage allocation

1. 01Recompute days-past-due triggers against the entity's own policyConfirm the rebuttable presumption that significant increase in credit risk occurs at 30 days past due (and default at 90 days past due) is applied as stated, and any rebuttal of that presumption is specifically documented and supportable.
2. 02Test qualitative triggers for stage migration, not just quantitative onesA customer in financial distress, a covenant breach, or a credit rating downgrade should move a receivable or loan to Stage 2 or 3 even where days-past-due alone would not — check these aren't being missed because the model is purely aging-based.
3. 03Sample-test individual large exposures for stage 3 identificationA material receivable or loan with objective evidence of impairment (restructuring, significant financial difficulty, breach of contract) should be in Stage 3 regardless of what an aggregate provisioning matrix would otherwise compute.

Testing the ECL computation itself

1. 01Trace the historical loss rate used to actual collection/write-off dataRecompute the loss rate from the entity's own aged receivables and actual historical write-offs over a representative period, rather than accepting a provision matrix rate without recalculation.
2. 02Test the forward-looking adjustmentMacroeconomic overlays (GDP growth, sector-specific outlook) applied to historical loss rates should be based on reasonable and supportable information available without undue cost or effort at the reporting date — a generic adjustment with no link to the entity's actual exposure sector should be challenged.
3. 03Test any post-model management overlay separately from the model outputAn overlay added or removed at management's discretion at period end needs its own rationale and evidence trail — recompute the model's unadjusted output first, then evaluate the overlay as a distinct estimate.