A0334
Title: Exploring the predictive capacity of ESG sentiment on official ratings: A few-shot learning perspective
Authors: Elena Toenjes - Justus-Liebig-University Giessen (Germany) [presenting]
Christoph Funk - Justus-Liebig-University Giessen (Germany)
Christian Haas - Frankfurt School of Finance & Management (Germany)
Abstract: Environmental, social, and governance (ESG) criteria are increasingly central to corporate reporting. Natural language processing (NLP) techniques, specifically a RoBERTa-based few-shot model, are applied to conduct aspect-based sentiment analysis (ABSA). The analysis targets ESG-related entities and their sentiments within EURO STOXX 50 company reports, mapping them to an ESG sub-category to assess their impact on ESG ratings. Ratings data are sourced from established providers, including Refinitiv, Standard \& Poor's, and potentially Bloomberg. Furthermore, to explore potential reciprocal influences on these variables, a panel vector auto-regressive (PVAR) model is employed, which facilitates the modeling of bidirectional interactions. The combination of advanced NLP methods and comprehensive data integration aims to provide detailed insights into the dynamics between company disclosures and rating providers' ESG scores.
