Applying explainable AI to analyze size-sorted ESG portfolios

Abstract

This research utilizes explainable artificial intelligence (XAI) techniques to explore the relationship between size-sorted environmental, social, and governance (ESG) portfolio indices (large-cap, mid-cap, and small-cap ESG portfolios) and their corresponding base indices (large-cap, mid-cap, and small-cap portfolios) in the US stock market. Additionally, the study accounts for key market factors, risks, and uncertainties. The findings reveal several important insights. First, size-sorted ESG portfolios are most significantly connected with portfolios containing companies of similar sizes. Second, for large-cap and mid-cap ESG portfolios, after accounting for interaction effects, their base indices (large-cap and mid-cap portfolios) primarily interact with other base indices. This observation suggests that the performance of large-cap and mid-cap ESG portfolios is more closely tied to the overall performance of the aggregate stock market. Third, for the small-cap ESG portfolio, the corresponding small-cap index (small-cap portfolio) frequently interacts with the bond market and default spread. Rather than indicating identical “default risk profiles,” these interactions reflect a stronger sensitivity of small-cap ESG portfolios to credit-market conditions and default-related dynamics, consistent with theoretical arguments regarding the default-risk exposure of small firms.