For many years, the role of regulatory bodies has been based on a blunt approach, one that has seen requirements and guidelines imposed on financial organisations. These were then left to figure out how to comply, no matter the clarity of such requirements or the burden they would impose once implemented in practice.
The clarity of such requirements within the model risk management space is often found lacking. This has sometimes led to inconsistencies, both internal to organisations and across the industry. More recent publications, however, have indicated a change in direction. In fact, regulators have been showing signs of pragmatism and modernisation. This is what we will discuss in this article, as we look at how parsimony and innovation are forming the basis of the EBA’s very recently published 2023 work plan.
In their publication, the EBA sets out its strategic priorities in the context of delivering a mandate within European financial markets. The work program’s scope sheds light on areas such as the implementation of Basel standards in ensuring that banks remain credible and resilient (both operationally and financially), EU stress testing , data standardisation and digitalisation.
The primary implication of this work program is that financial institutions will have to develop their own approach in following this direction, while allocating resources accordingly. However, there is also a great opportunity for financial firms to reorganise themselves to align to some key strategic work streams set forward by the EBA, which, in our view, seem somewhat connected to more effective model risk management:
(1) With market conditions becoming more complex and less predictable, the EBA’s initiative for an EU-wide approach to stress testing will require firms to run additional tests to ensure model robustness. To meet the increasing demand for model validation, many more firms may employ intelligent automation to run dynamic scenario tests with minimal effort, thereby avoiding unnecessary backlogs and minimising operational risk.
(2) As demand for model testing increases relative to firms’ limited resources, more firms will focus on standardising and automating their reporting and model documentation processes in a bid to unburden themselves. As the EBA works to simplify its data management framework via the EUCLID platform, firms can leverage new technologies to integrate existing platforms with the system using APIs.
(3) Clarity over AI guidelines. As high-risk AI applications have rapidly increased over a short space of time, the EBA is aiming to clarify existing supervisory expectations. Compared to traditional models, AI/ML models are relatively more complex, requiring greater monitoring and upkeep. To stay compliant and ensure AI/ML models perform their intended function, banks must scale their model validation efforts by investing in AI-driven platforms capable of more programmatical estimations when model performance is deteriorating.
(4) The EBA will implement a new data strategy, while also leveraging the EUCLID platform to help firms remain compliant.
While the full and long-term implications of the EBA’s new work program have yet to be seen, financial firms must be ready to adopt new practices and scale their model risk management to align themselves with the fast-changing and increasingly innovative regulatory landscape.