2020. The year of change for risk. Covered in our first risk prediction, we expect AI implementation to surge. In this blog, we explore why lenders are keen to review model risk practices this year.
Last year, we saw a significant increase in demand for support on model validation and model monitoring as more lenders seek services that highlight model deficiencies to ultimately satisfy regulators and maximise profitability.
Statistical models underpin many processes in banking and financial services. Their ability to accurately predict losses has led to them forming the foundation of regulation such as IFRS 9, ICAAP and IRB. Whilst using models inevitably reduces risk due to human error, conversely, it introduces a new risk due to potential inaccuracies within their prediction: model risk.
The regulator acknowledges the pivotal role that models play in regulation and key business decisions and so has placed a larger spotlight on model risk. As a result, we have seen a significant rise in demand for our modelling and regulation expertise to support firms to satisfy model risk objectives.
What should a lender do?
Firms must have robust model risk practices in place by quantifying and monitoring the levels and enforcing appropriate mitigating actions with the end objective of maintaining a stable economy. They should enforce levels of model governance that are commensurate to the size of the firm to appropriately measure and control for model risk.
Appropriate model validation techniques, such as model monitoring, should be applied to all models with outputs reviewed and challenged by a team of independent experts to ensure the models are fit for purpose and compliant.
This is typically done by either establishing an internal model validation team within the firm’s second line of defence or alternatively outsourcing it to a qualified third party to ensure the independence criteria is satisfied.
What are the benefits of change?
- Satisfy increasing regulatory demands
- Assessing model risk enables you to optimise profitability through maximising the accuracy of models that make key lending decisions