A0560
Title: Bayesian projection pursuit for efficient and sustainable banking
Authors: Nicola Loperfido - University of Urbino (Italy)
Cinzia Franceschini - University of Sassari (Italy)
Alessandro Berti - Urbino University Carlo Bo (Italy) [presenting]
Abstract: The European banking agency (EBA) established new and cogent guidelines (EBA-LOM) that European banks should follow when granting credit. The guidelines were set in June 2021 but have been implemented in the current year. There is a general agreement that their implementation will be gradual. The main novelty of EBA-LOM guidelines, in addition to providing precise and well-defined indications on tools for credit risk measurement, is their emphasis on environment, sustainability and governance factors (ESG factors), which are non-financial factors. For example, banks should be careful when granting credit to firms with questionable attitudes towards their workers. The EBA-LOM guidelines pose several problems to banks, which did not occur when credit is granted using balance sheet ratios only. Firstly, how should ESG factors be measured? Secondly, how can ESG data be collected from firms? Thirdly, how should these data be summarized into a single measure of credit risk? The default approach relies on software prepared by rating agencies (Moody's, Fitch, and Standard \& Poor's), which do not consider the unique characteristics of given firms. A Bayesian approach is proposed based on prior elicitation of ESG factors given by professional bank consultants, which are then merged with other information by means of projection pursuit. The approach is illustrated with a small dataset of Italian firms.
