At Night All Cats Are Gray, but at Day They Are Not: Default (PD) Forecasts Capturing Italian Banks’ Idiosyncrasy

Abstract

Although the probability of default (PD) modeling has reached a great maturity in both academia and business, for the Italian case we demonstrate that banks' available PD models would be misleading if today applied directly to Italian banks. We argue that what determines the PD of Italian banks, rather than the liquidity, are the return on assets (ROA), the financial leverage and the BCC category of the bank. Furthermore, we demonstrate that the conventional approach dominates the more trendy machine learning (ML). Finally, we demonstrate that model's performance could be used as a supervisory tool for retrospective analysis of the bank's position. Moreover, we bring positive evidence on the BCC 2016 reform in Italy.

Keywords: bank failure, adaptive lasso, logistic regression, CART, probability of default, random forest, machine learning, model selection

JEL Classification: C52, C53, D22, G21, G28, G33