Trading ESG vs. Trading E, S, and G Separately: An Exploratory Research

Environment, Social, and Governance (ESG) criteria become a relevant factor in the investment universe. We develop an AI-based algorithm that uses public data, mainly Web-based information, to assign E, S, and G ratings to companies. Using our scoring procedure, we construct portfolios, comprising 50 firms each from the S&P 500 index, 50 firms with the highest scores and 50 with the lowest scores for 4 scoring categories: ESG, E, S, and G, for the years 2018-2021. We find that, except in 2021, high-ESG score portfolios consistently outperform low-ESG score portfolios. In particular, we observe that the shares of high G-score companies outperform low G-score portfolios, with the largest difference in performance between high and low-score portfolios. The data support the hypothesis that indicators of good corporate governance can identify better performing firms. We also note the outperformance of high S-rated portfolios in 2018–2020. We find that the E-portfolios behave differently from the S and G portfolios. Due to data constraints, we view this paper as exploratory only, and further research is due to validate our findings.