Credit Risk and Artificial Intelligence: On the Need for Convergent Regulation
10 Pages Posted: 27 Nov 2023
Date Written: October 27, 2023
Abstract
Credit risk is probably the most regulated application field of statistical learning models worldwide, with a strong body of regulations encompassed in the Basel Accords. As the use of machine learning and Artificial Intelligence models (AI) is becoming more widespread, entities worldwide, including the European Banking Authority and local regulators in Canada, Germany, and the UK, have released opinion papers or put forward proposals on how the new regulations will need to adapt to accommodate AI tools in banking for credit risk modelling and management. These regulatory proposals, which have identified credit risk as a high-risk application area of AI, have shown a significant divergence in their views, which goes against the idea of the Basel regulations as a unifying framework. In this opinion paper, we discuss the current proposals for regulatory oversight of credit risk models in Canada, the European Union, the United Kingdom, and the United States, and show where they diverge and where they coincide. We also motivate and discuss the urgent need for the development of a unified framework, as new laws regulating AI will certainly clash with current banking regulations and initiate a first discussion on making these proposals convergent.
Keywords: Credit Risk, Artificial Intelligence, Banking Regulation
JEL Classification: G32, G28, E58
Suggested Citation: Suggested Citation