머신러닝 기법을 이용한 ESG 펀드와 일반 펀드의 주식 포트폴리오 편입 결정 요인에 대한 연구

초록

This study analyzes the determinants of stock selection decisions for ESG funds and conventional equity funds using machine learning techniques. Specifically, the analysis employs nine firm-specific variables and four ESG rating categories (Environmental, Social, Governance, and Composite ESG) to predict stock inclusion in equity fund portfolios. The analysis employs decision tree-based machine learning algorithms --- Random forest, XGBoost, and LightGBM. Based on data from Korean public active equity funds as of December 2022, we find that firm size is the most significant factor in fund inclusion decisions for both ESG and conventional funds. Although ESG ratings exhibit relatively lower importance compared to other firm-specific factors, the Environmental and Composite ESG ratings show a significantly higher impact within ESG funds than in conventional funds. The study underscores the value of machine learning techniques in uncovering the determinants of fund inclusion decisions and offers practical and policy implications for ESG investment.

키워드