Expected Credit Loss models built to withstand auditor and regulator challenge, PD, LGD, EAD and staging, methodologically sound and properly documented.
IFRS 9’s expected credit loss framework requires forward-looking, probability-weighted estimation of credit losses, a significant methodological shift from the incurred-loss models it replaced. LeapWise builds and validates ECL impairment models: probability of default (PD), loss given default (LGD) and exposure at default (EAD) estimation, staging criteria and movement, and the underlying risk-policy documentation that supports the model in front of auditors and regulators.
We work with banks, non-bank financial companies, and corporates with material receivables or lending exposure, building models that are methodologically defensible, properly documented, and genuinely usable by your risk and finance teams, not just delivered as a one-time report.
We assess available historical data, portfolio segmentation and existing methodology as the starting baseline.
PD, LGD and EAD methodologies are designed or refined to fit your data quality, portfolio type and regulatory context.
Staging criteria are defined and documented, along with the full model methodology paper required for audit.
The model is tested, results validated, and your team is walked through ongoing application and updates.
Both. We can design a new model where none exists, or review, validate and refine an existing model that is facing challenge or needs updating.
Historical default and recovery data where available, portfolio segmentation, and existing credit policies. We assess data readiness as the first step and work with what is available.
Yes, for corporates with trade receivables rather than a lending book, we apply the simplified ECL approach appropriate to that exposure type.
That is a core objective, the methodology paper and supporting documentation are built specifically to withstand audit review.
Reach out for a direct conversation about your business, no obligation, no generic sales pitch.