Adapting your model risk framework for evolving conditions

Ryan Michalik, Ziyu Zhang
Adapting your model risk framework for evolving conditions

How can you apply lessons learned to help your models weather future turbulence?

The COVID-19 pandemic was a stark wake-up call. Model risk officers and managers can take proactive steps before the next shockwave arrives.

Recent economic disruptions exposed weaknesses in many models once considered largely reliable. Fire drills erupted at many banks and credit unions, including the following scenarios:

  • Current expected credit loss model owners scrambled to adjust estimates when model limitations were identified and output didn’t align to expectations.
  • Many fraud models produced a higher volume of false positives as customer behaviors drastically changed during the pandemic.
  • Price and credit stresses triggered overreactions in many risk models.

Volatility will happen again, and you can better prepare your risk models.

No risk management team wants to face unexpected challenges. That’s why every experience should be viewed as a lesson in understanding the limitations of the organization’s models and enhancing practices to mitigate model risk.

Laying the groundwork now can provide an advantage when another disruption strikes.

Our consultants can help you mitigate model risk against future disruptions.

What steps should banks and credit unions take now to strengthen their model risk frameworks?

What steps should banks and credit unions take now to strengthen their model risk frameworks? - No one can predict the future, but you can identify issues sooner.

No one can predict the future, but organizations can position themselves to identify issues sooner. They can also establish a model risk framework to better use their models in a volatile environment.

Establish a robust, ongoing monitoring framework

A standardized, well-documented process that includes back-testing actual results against established performance metrics can help improve model reliability.

If organizations only test when they suspect a problem, it is already too late. Scheduled monitoring can help spot deterioration earlier and provide an opportunity to calibrate.

Review overrides, overlays, and adjustments

Models are simplifications of reality. Overrides, overlays, and adjustments are often made for factors identified but not explicitly captured in the model. Adjustments to output can become more prevalent in a period of economic uncertainty.

Establishing a model risk framework to document, support, and track adjustments can provide transparency to management and users of the model. It also facilitates effective discussion and challenge of the model performance and adjustments applied.

If adjustments are continual or significant in volume or impact, the model itself might need revisiting.

Identify model interconnectedness

Identify model interconnectedness

When models depend on each other and on common data, an issue can trigger exponential effects across the organization.

All model and data interconnectedness should be mapped and updated to show downstream risk to other models and reports.

Model risk managers can use this view to quickly assess the potential knock-on effect of an issue. They can then communicate the issue to interested stakeholders and address mitigation plans accordingly.

Review your sensitivity analysis procedures

The past few years proved how outlying events can produce nonintuitive results, which can happen when models are developed on historical data that didn’t capture newer, extreme events. A good follow-up question in this situation is, “What other outlying events might our models not be prepared to process?”

A well-documented routine can help prevent unpleasant surprises in the future. Now is the time for banks and credit unions to re-examine the boundaries of their models and the stress factors their risk teams have applied through testing. As part of this process, they should include both potential new factors and deviations based on experiences from recent disruptions.

Improve collaboration between model owners, model risk teams, and management - Positioning the model risk team as an advocate can facilitate greater adoption of model risk practices.

Improve collaboration between model owners, model risk teams, and management

Model risk management, model owners, and managers all share the same objective of using models to make well-informed business decisions. Everyone can benefit by removing obstacles to collaboration and support.

Positioning the model risk team as more of an advocate, in contrast to a compliance function, can facilitate greater adoption of model risk practices by model owners. Model owners might feel more empowered to work with the model risk team if they know that they have someone who can advise management about when resources should be allocated to remediate issues.

Maintain a schedule of model validations

All models due or past due for validation should be a priority. Then, organizations should follow a schedule for each model that includes appropriate risk-based activities between full validations.

Keeping a disciplined schedule, along with prompt remediation of findings, can better position the organization to use its models in volatile times.

We can help you pave your path toward model stability.

Our consultants can help your teams power model risk management through new economic landscapes. Schedule a consultation to discuss the technologies, processes, and solutions that best fit your organization’s needs.

Need a checkup or forward-moving plan for your model risk management?

Contact us

Ryan Michalik
Ryan Michalik
Principal, Financial Services Consulting
Ziyu Zhang
Ziyu Zhang
Financial Services Consulting