Credit loss models that stand up to auditors, regulators and boards.

The general approach applies three stages. Stage 1 exposures carry a 12-month ECL, Stage 2 exposures (those with a significant increase in credit risk since origination) carry a lifetime ECL, and Stage 3 exposures are credit-impaired. The simplified approach, permitted for trade receivables, contract assets and lease receivables, always recognises lifetime ECL and removes the need to track SICR. Most corporates use the simplified approach with a provision matrix; banks and lenders must apply the general approach to loans and debt instruments.

We combine quantitative and qualitative triggers. Quantitative triggers compare the lifetime PD at reporting date with the lifetime PD at origination, typically using a relative threshold (for example a doubling or tripling of PD) together with an absolute floor. Qualitative triggers include watch-list status, forbearance, covenant breaches and adverse sector news. IFRS 9 also carries a rebuttable presumption that credit risk has increased significantly when contractual payments are more than 30 days past due, and a presumption of default at 90 days. Our calibration tests each trigger against historical migration data so that the staging is evidence-based rather than arbitrary.

Data scarcity is the norm in the GCC outside the largest banks. For PD we use transition matrices where rating or ageing histories exist, otherwise we map internal grades to external rating agency default studies or use Pluto-Tasche low-default-portfolio techniques. For LGD we use workout recoveries where available, otherwise collateral haircuts and time-to-recovery assumptions benchmarked to regional experience. For EAD we apply credit conversion factors to undrawn limits. Every proxy is documented with the rationale and a sensitivity so auditors can see the impact of the assumption.

IFRS 9 requires ECL to reflect reasonable and supportable forward-looking information. We regress historical default or loss rates on macroeconomic variables, then project ECL under multiple scenarios. In Saudi Arabia and Bahrain the most explanatory variables are typically Brent crude prices, non-oil GDP growth, inflation, policy interest rates and, for real estate portfolios, transaction volumes and price indices. Scenarios are probability-weighted (commonly base 50 to 60 percent, downside 20 to 30 percent, upside 15 to 25 percent) and the weights are reviewed each period with documented rationale.

SAMA's IFRS 9 guidance expects a board-approved ECL policy, documented models with independent validation, clear SICR criteria, forward-looking scenarios with governance over the weights, management overlays that are justified and time-bound, and regular back-testing. The Central Bank of Bahrain has similar expectations. We build the governance framework alongside the model so both are examination-ready.

Yes. A well-designed provision matrix takes an ageing report as input, applies segmented historical loss rates adjusted for forward-looking factors, and produces the lifetime ECL and the journal entry. We typically deliver it in Excel with locked formulas, a control sheet and a one-page run guide. The forward-looking adjustment is refreshed annually or when conditions change materially.

Yes. Independent validation covers conceptual soundness, data integrity, implementation testing, back-testing, sensitivity analysis and documentation. We issue a validation report with findings graded by severity and a remediation plan. This is often requested by audit committees, by SAMA-regulated entities, and by groups consolidating subsidiaries with different models.

It depends on data availability and on what you need the model to do. A provision matrix for a corporate with clean ageing and write-off history is a short engagement; a general-approach model for a lender with PD, LGD and EAD components, macroeconomic overlays and validation is a longer one. We agree the plan and the milestones after the data assessment, and we tell you at that point what is realistic for your reporting deadline.