Links
Next content
Read more
Constrained machine learning models for credit default prediction: who wins, and who loses
Balancing the prediction accuracy of machine learning models with their interpretability is crucial, particularly in settings such as credit decisions, where regulatory requirements demand transparent decision-making processes. Imposing monotonicity constraints to a machine learning...
The authors use confidential loan-level data from the European Central Bank to investigate how changes in the countercyclical capital buffer requirement in Germany affect lending to firms. They find evidence showing that tightening the countercyclical capital buffer leads German banks to reduce the volume of corporate loans and increase the price of new loans. These effects take place immediately after the announcement, given 12 months before the change was implemented. Importantly, the authors find that the reduction in credit availability notably affects small and medium-sized enterprises, which experience both a significant decrease in available credit and an increase in credit costs. In contrast, large firms are not affected.
This paper is part of the Banking Supervision Policy Working Paper Series in the context of the SSM-EUI partnership on SSM Banking Supervision Learning Services. Read more.