
Model Validation
Validate all aspects of your model design, implementation, performance and compliance.
Add value beyond compliance by achieving a complete understanding of risk.
Model validation, both at the point of inception and as part of periodic review, is central to best practice model governance, minimising model risk and ensuring regulatory compliance. Whether built internally, or by a third party, the validation exercise ensures the models are fit for purpose. This includes an assessment of the suitability of modelling techniques, data handling and selection, the model construct, modelling assumptions, regulatory considerations, and model implementation.
Irrespective of the type of model being validated, we adopt a consistent approach to achieve the best possible outcome.
Through consultancy-led model validation, we have supported clients in the validation of models used at all stages of the credit lifecycle, including operational and behavioural scorecards, Fraud, Affordability, Provisioning (IFRS9), Stress Testing, Capital (IRB) and Finance Models. Our approach ensures that the appropriate level of challenge is provided, depending on the organisational requirements, the intended model use, and related regulatory standards (e.g, IFRS9, CRR, MCOB). The outputs of the reviews can be shared with the relevant regulatory body to demonstrate that the models have been subjected to sufficient scrutiny.
Scoping
Every model validation project commences by defining and agreeing upon the scope of the review. As we often augment our client’s 2nd line team, roles and responsibilities should be clarified from the outset.
Review & Assess
The model review process can then commence, which typically entails a desktop review and a review of model development and/or implementation code.
An initial high-level review will identify any missing information or early red flags. This is followed up with a deep-dive review which culminates with a set of queries and clarification points. These are shared with the model owner for responses, which are communicated back to the review team.
Equipped with all required information, the model can be assessed against the agreed criteria using either the client’s or Jaywing’s own model validation policy. Our initial view of findings can be forwarded to the model owner with an assessment of the quality of the model established.
Report & Governance
The final phase involves finalising the list of findings and assigning severity statuses (e.g., high, medium, low) in accordance with policy.
These final findings will influence a decision to apply model remediations or enhancements. A model validation report captures the review process, findings, conclusions and recommendations. This report is reviewed by 1st line, with a summary and the key points presented at governance committees.
Our market-leading experience in IFRS 9 has led us being nominated to validate all aspects of our client’s model design, its implementation, model performance and compliance.
Taking stock of the FCA’s interim findings for remediating the credit information market
The FCA's proposals to improve the credit information market could have significant implications for credit reference agencies, lenders, and borrowers alike. In Steve's latest article, he reviews the interim report/discussion paper and outlines potential remediations - including what they could mean for the future of CRAs.
(Machine) learning to live with CP6/22
Steve Finlay, one of the UK’s leading data scientists, examines how AI can be leveraged for model risk management and CP6/22 regulations.
How to put customers at the forefront of your fraud strategy
As customers continue to demand simple, frictionless experiences and digital banking becomes the norm, how can lenders provide excellent customer journeys whilst mitigating fraud?