Mitigating company characteristics as external biases in ESG valuation models

Abstract

We study the impact of firms’ characteristics such as firms’ size, sector, and geographical activity area in ESG scorings. These characteristics may create biases in the scoring models. For example, larger firms may show on average better ESG scores than smaller firms. We propose a methodology to mitigate these biases and compute a score that is free from any factors including firms’ characteristics. We also propose a framework to detect outperforming companies regardless of any factors. Our methodology can break down final ESG scores into a component related to these external factors and a component unrelated to them.