Carefully tailoring the model risk management (MRM) processes to model tiers has become an increasingly important goal for financial institutions worldwide. However, this process can be confusing and daunting in practice due to competing expectations of developers, users, and regulators with limited resources. During the RiskMinds event in late May, many industry leaders shed light on this complex process and introduced new ideas for a risk-based approach to MRM.
The general consensus was that the pandemic significantly stress-tested the model tiering approach, and there is a clear need to overhaul it. Additional metrics such as the probability of a model breaking or not functioning at all and materiality tied to the notion of loss among others need to be devised over and above materiality, complexity, and model use. The expanded model tiering system could then better guide the development of improved approaches to combat model failure during crises. It will also help introduce a “proportionate to risk” model risk management framework where additional controls are instituted for critical models. Some of the additional controls that the event speakers mentioned were related to flexible governance and ongoing performance monitoring, increased use of stress testing, parallel use of alternative/ challenger models, use of kill switches in the models, and last but not the least establishing expert teams who would be go-to teams when models break down.
Some speakers also mentioned that the revamped risk-based approach is required to release and rationalize resources from models with lower residual risk. The continuing rise of talent scarcity and cost pressures with added responsibilities since the pandemic has overburdened the model risk function. Flexibility in setting up validation and monitoring cycles in times of crises was discussed so that models with lower residual risk and fewer or no validation findings can be given some breathing room. There were also discussions around ways to define and hence limit essential testing and documentation requirements for lower-tiered models. One speaker mentioned how a global investment bank recommended specific validation tests that tend to be time-consuming should be left out for lower-tier models while at another global investment bank, model risk governance has set SLAs on Turnaround Times for model validation exercises and have set them significantly lower for tier 3 and 4 models.
The creation of all these varied responses and decisions regarding the risk-based approach to model risk management has been left to the discretion of banks and their MRM Governance functions. MRM governance is expected to initiate dialogue among various stakeholders internally – model developers, users and owners, validators, and risk and audit committee to revamp the risk-based approach to MRM. Essentially MRM Governance teams will have to define non-negotiable principles for redirecting resources from lower-tier models while creating a clear benefit framework for instituting additional controls for critical higher-tier models.
To learn more about a risk-based approach to model risk management, contact MRMSolutions@evalueserve.com.