Banks have relied on models for analysis and decision-making for quite some time. But what has changed is the number of models, the expanding areas they're being used and their complexity, which has prompted closer scrutiny from banking regulators over the past few years.
In the past, models were often used to manage market liquidity, portfolios and credit risk. Today, models are used enterprise-wide in areas such as compliance, AML, fraud, trade surveillance and marketing. It's no surprise the primary drivers for increased model use are increased data availability and the advanced technology solutions that can harvest and process the data.
The increased use of models and data naturally results in increased risk for financial institutions. Those risks originate from the increased complexity of the models but also from the frequency of data inputs. As a result, a financial institution's underlying data quality and complexity become a very important factor.
There's also model aggregation risk. As a risk manager or a line of business manager, you should understand how the output of one model may affect the output of others. Risk in the aggregate is all about the potential for interconnectedness between all models.
Model governance functions should be just as sophisticated as the model ecosystem within your financial institution. Are there sufficiently skilled staff to evaluate complex models?
Re-evaluate your validation and testing frequency using a risk-based approach. Ensure the validation staff has an appropriate level of independence to challenge authority. Is there a mechanism in place for reporting unresolved risk issues to senior management and the board?
Next, you should establish a solid model risk-management infrastructure. Partner with a technology provider that can help establish a model risk-management solution. A good solution will allow you to do several things, including control access, manage model versions and establish auditable model documentation. It will also help you understand the aggregate risk of model concentrations and the links between those models. Controlled workflows manage model validations.
Using a solution that leverages artificial intelligence and machine learning may enable you to gain insights and manage your business at lower levels than previously possible. For example, what if you had the potential to improve liquidity management by understanding liquidity risk at an individual line of business? Or what if you could understand your consumers' behavior to improve marketing efforts by offering well-suited and timely products?
Finely tune credit underwriting processes and offer credit terms at more customized levels that reflect the customer's true risk. The benefits and use cases available by partnering with a known and trusted technology partner for model risk management are virtually immeasurable.