Banks deploy financial and operational models to support business decisions and increase operational efficiencies. Risk modeling often leverages artificial intelligence and machine learning modeling techniques, with algorithms too complex for non-subject matter experts to verify for accuracy. For risk managers, validating working models is essential to spot erroneous outputs or anomalies. But what are best practices and current trends in risk model validation?
RMA surveyed model risk managers across the banking industry, to gather input on vendor model validation and third-party risk management practices. In April 2022, RMA published the executive summary of the Survey of Model Risk Management, Vendor Model Validation, and Third-Party Model Risk Management whitepaper.
Some top-level findings include:
- Cost and talent are the biggest concerns for an operational expansion.
- Vendor documentation needs to be rigorous.
- Regulators are paying more attention to AI/ML models and non-model tools.
As banks use new tools daily, risk managers see the heightened material impact that models bring to their organizations and business partners. The scope of model risk management includes multiple risk domains: operational risk management, such as third-party risks and legal and compliance risks; enterprise risk management, such as cybersecurity and technology risks, and the need for model risk governance.
Model validation work also has a high barrier of entry due to the challenges of developing a model and the black-box-testing nature of the validation process.
The full survey results are available only for RMA institutional members who participated in the survey. Speak with us to learn more about the full survey results faster than your peers.