AI REQUIRES MODEL RISK MANAGEMENT THOUGHT LEADERSHIP
The Bank of England’s Prudential Regulation Authority (PRA) released its supervisory statement (SS)1/23 on model risk management (MRM) principles for banks. In its consultation, it said it had found “evidence of poor MRM when reviewing Firms’ applications for internal regulatory model permissions and when reviewing approaches to expected credit loss accounting under IFRS 9”. Given the changing environmental and digital landscape, and increasingly sophisticated modelling methods, the PRA said it expected the use and complexity of models to grow. It cited the need to quantify financial risk linked to climate change, and to the use of artificial intelligence (AI) and machine learning (ML).
What is driving the change?
In a competitive market, firms must continually innovate and evolve to meet customers’ changing expectations and deliver a strong return on investment (ROI) for shareholders, so they are deploying more models, and more varied models, to gain an advantage.
Globally, around 28 percent of institutions are running more than 100 models, according to PwC, and the numbers and sophistication of these models are only going to increase further in the coming years. It is understood that firms need the right level of oversight and governance to ensure they adhere to the latest regulations.
The economic impact of the coronavirus (COVID-19), the war in Ukraine, the integration of environmental, social and governance (ESG) and climate risk, and the expansion of AI and ML within firms have all given the UK regulator the stimulus it needed to bring a holistic MRM framework to the forefront. Firms have already made great progress in mitigating model risk, but the principles set out by the PRA should bring greater clarity and set the standard across the financial services industry.
Oct-Dec 2023 Issue
SAS Institute