Aggregating ESG scores: a Wasserstein distance-based method

The evaluation of the Environmental, Social and Governance (ESG) profile of companies is gaining more and more importance in the credit and financial system and is made more challenging by the availability of alternative - and often divergen t- ESG ratings. In addition, the contribution of the three dimensions (E, S and G) to the final evaluation is not disclosed by the raters. This paper proposes an approach for aggregating the three dimensions constituting ESG ratings by means of optimal transport from the perspective of the Wasserstein distance. An empirical exercise, conducted on a dataset related to Small and Medium Enterprises (SMEs), shows that the proposed aggregated indicator represents a statistically sound and explainable tool for the users of ESG ratings, especially when non-homogenous evaluations are provided. Our proposal is also compared to Principal Component Analysis (PCA), a state of the art machine learning algorithm widely employed in the literature concerning the building of synthetic indicators.