Bayesian learning models to measure the relative impact of ESG factors on credit ratings

Abstract Artificial intelligence methods, based on machine learning models, are rapidly changing financial services, and credit lending in particular, complementing traditional bank with platform lending. While technologies improve user experience possibly lower costs, they may increase risks and, the model that derive from inaccurate rating assessments. In this paper, we will show how to reduce such risks, using a S.A.F.E. statistical model, which improves: Sustainability, taking environmental, social governance factors into account; Accuracy, building maximises predictive accuracy; Fairness, merging ESG scores different data providers, improving their representativeness; Explainability, clarifying relative contribution of each score accuracy.