From Kpis to ESG: Addressing Redundancy and Distortions in ESG Scores

Abstract

We investigate the construction of Environmental, Social, and Governance (ESG) scores, focusing on Refinitiv's (acquired by the London Stock Exchange Group) methodology. We uncover critical challenges, including the inflation of correlations caused by missing data imputation and redundancy among Key Performance Indicators (KPIs). By employing various association measures empirically and theoretically, we demonstrate imputing missing values with zeros distorts relationships between KPIs. We then propose an optimization algorithm to identify an optimal subset of KPIs capable of replicating pillar scores. Our findings reveal that a small subset of KPIs can closely replicate Refinitiv’s pillar scores, highlighting that many of the 180 KPIs used are redundant. While no single subset of KPIs consistently tracks the ESG pillar scores, reducing redundancy offers a path to streamline ESG ratings, improving their efficiency and interpretability without sacrificing depth.

Keywords: ESG, Imputation, KPI, Redundancy, Sustainability