Environmental, Social, and Governance Taxonomy Simplification: A Hybrid Text Mining Approach

Abstract

Currently, environmental, social, and governance (ESG) reporting is mostly voluntary, granting companies the discretion to choose the information to disclose and the standards to follow, resulting in a lack of comparability across ESG reports. Efforts to combine standards for global comparability are static and may not fit the everchanging, industry-specific nature of ESG topics. This paper proposes a hybrid methodology for extracting simplified, ex post, and dynamic taxonomies based on existing ESG standards and reports to improve the comparability of ESG reporting. This hybrid methodology, which combines text mining techniques with manual processing, balances the efficiency of automatic processes with the effectiveness of human judgment. An example of deriving a simplified environmental taxonomy from European companies’ ESG reports and the Global Reporting Initiative (GRI) standards illustrates the proposed methodology. The methodology could help regulators to develop comparable taxonomies and detect greenwashing and enable various stakeholders to compare companies’ ESG performance.