itsual Model Risk Governance
Overview
Model risk arises when models are wrong, misused, or poorly implemented. Governance ensures material models are identified, validated, monitored, and used within their intended scope.
When to Use
- Building or maturing MRM frameworks
- Inventorying models across credit, fraud, AML, marketing, AI
- Independent validation and periodic review
- Regulatory exam preparation on models
Core Practices
- Maintain a model inventory with owners, purpose, and materiality tier
- Require documentation: development, data, assumptions, limitations
- Perform independent validation proportionate to risk
- Monitor performance, stability, and outcome drift in production
- Control changes through approval and re-validation triggers
- Restrict use outside approved scope
Principles
- Materiality drives depth of challenge — not every spreadsheet is equal
- Developer cannot be the sole validator of high-risk models
- Ongoing monitoring matters as much as point-in-time validation
- AI/ML models need the same accountability outcomes, adapted methods
Verification
- Material models are inventoried and tiered
- Validation and monitoring cadence matches tier
- Limitations and use boundaries are explicit