Discover how transparency data bias affects ESG ratings and how AI could improve data collection, verification and rating reliability.
Published:
March 2025
Author: Xinxiu Liu, GFN Member
Abstract:
ESG rating plays a crucial role in capital markets, as they evaluate company's non-financial performance to gauge risk and opportunities, providing valuable reference for various stakeholders. Amidst growing regulatory demands, the surge in ESG data disclosure has led to transparency data bias, as the volume of disclosed data can influence rating outcomes, potentially compromising the accountability of ESG rating agencies. The findings reveal awareness of this bias among the interviewees and an acknowledgement of its impact. The study also uncovers the underlying causes to transparency data bias, which stems from the quality of the data, the data collection process and the applied methodology. While AI, including NLP, offers promising solutions for data collection, its application within ESG rating agencies remains underexplored. Despite the challenges associated with integrating AI into the rating system, interviewees provided insights into the capabilities and potential possibilities of AI tools in supporting data source, data collection, and verification processes. They also offer practical recommendations for developing a fundamental database to train and finetune data sets, ensuring the robustness of the entire system, and validating the performance of tools.
