Model users must identify the sources of risk, assess the magnitude of risk, and mitigate or control the risk.
Regulators have articulated clear expectations for model risk management. In this environment, any business that uses models must:
- Implement effective approaches to model governance and model risk management
- Develop models to meet regulatory expectations and business requirements
- Conduct model validations to ensure that model risks are managed appropriately and models continue to meet business needs
Promontory’s team has experience in all key dimensions of model governance. Our areas of specialty include:
- Developing policies on model risk
- Documentation of model governance frameworks
- Methodologies for model risk tiering
- Design of model inventory systems
- Design of model documentation and validation templates
No other firm offers the depth of insight, experience, and understanding that Promontory’s team brings to model governance engagements.
We develop all of our models in accordance with client policies, regulatory requirements, and industry best practices. Our approach prioritizes sound model risk management and is sensitive to the implementation of appropriate controls during the development process, including development data, key assumptions, and statistical modeling. All of our model development engagements specifically include:
- Consideration and documentation of the conceptual and statistical foundations for the model design
- Consideration and documentation of the data used for model development
- Appropriate development testing, including benchmarking, back-testing, and sensitivity analysis
- Results from model implementation and preliminary model performance
Model validation is an integral component of effective model risk management. Supervisory guidance requires model validation to encompass three core elements:
- Evaluation of conceptual soundness
- Process verification and ongoing monitoring
- Performance tracking and outcomes analysis, including back-testing
Promontory’s approach to model validation and review is structured around these three elements and ensures that each is covered appropriately.
Evaluation of Conceptual Soundness
Promontory assesses developmental evidence to evaluate the appropriateness of each specific model’s development methodology. Our evaluations are informed by the client’s unique business exposures and focus on the intended model use, data selection, variable selection, model selection, statistical analysis, and — in the case of stress testing — scenario design. We draw on our extensive experience with model documentation to ensure that it is clear and effective.
Process Verification and Ongoing Monitoring
Promontory typically employs selected sample-based testing to determine whether a client’s model and data systems are functioning as designed. In doing so, our reviewers look for evidence that the implementation of system components has been adequately tested, that internal processes are structured to control operational risk, and that the model’s production environment continues to be aligned with its design.
Performance Tracking and Outcomes Analysis
Promontory reviews the performance-tracking and -reporting frameworks used by clients to measure whether their models are performing as expected. As part of this analysis, we ensure that appropriate actions are linked to potential or actual performance triggers.
Additional Resources and Publications
Engagement Case Studies
Click here to download a PDF featuring selected case studies of quantitative services engagements.