Ratings based inference and credit risk: detecting likely-to-fail banks with the PC-Mahalanobis method

This paper proposes a new approach of how to test the validity of bank ratings assigned by Rating Agencies. An innovative Early Warning System (EWS) is introduced that allows to unveil prodromic signals of instability for selected individual banks, and possibly forecast bank failures. A forward-looking, credit risk model that is based on financial ratios is designed to assess the financial position of rated banks. This approach allows to discriminate between banks that are in a stable, financially healthy position, and banks that are possibly going to become insolvent (likely-to-fail banks). Our empirical results are compared with the official ratings assigned by RAs to the same intermediaries. Our findings reveal incoherent positions and possibly incorrectly rated banks. We argue that our method can be easily implemented by financial regulators.