Prof. Dr. Valeriya Dinger, Prof. Dr. Peter Grundke, Valerii Amosov, Julian Hüßing and Julian Meyer have published a joint research paper entitled "Textual analysis and sustainability scores: A feasibility study" in the "Zeitschrift für Bankrecht und Bankwirtschaft".
Link to the article: www.zbb-online.com/heft-2-2025/zbb-2025-136-textuelle-analysen-und-nachhaltigkeitsscores-eine-machbarkeitsstudie/
The feasibility study focuses on the question of the extent to which the sustainability of companies can be automatically derived from publicly accessible texts in terms of taxonomy conformity - i.e. the conformity of economic activities with the requirements of the EU Taxonomy Regulation. The natural language processing (NLP) model ClimateBERT, which specializes in climate issues, is used for this purpose. In a first step, the semantic similarity between text passages of the EU Taxonomy Regulation and over 4,000 media reports from DAX companies is analyzed. The results show a significant positive correlation between the semantic similarity and the reported taxonomy conformity. In a second step, an exemplary application of the developed models to three small and medium-sized enterprises (SMEs) is carried out. This analysis shows the potential, but also the challenges of the methodology - especially with regard to the limited data availability and quality for SMEs.
