RESEARCH · EU
ECB publishes machine-learning framework for probability-of-default and stress testing
The European Central Bank has released technical research on using ML and AI to estimate probability of default (PD) and conduct stress tests across portfolios. Framework aims to improve accuracy and speed of credit risk modeling vs. traditional econometric methods.
WHY IT MATTERS
ECB-endorsed ML PD models signal regulatory acceptance of AI for core credit-risk calculations; banks must begin validating internal ML stress-test engines against ECB benchmarks or face competitive/capital disadvantage.