Uncovering lobbying strategies in sustainable finance disclosure regulations using machine learning.

Purpose: We analyse lobbying behaviour by using Machine Learning approaches. In the context of Sustainable Finance Disclosure Regulation (SFDR), we gain detailed insights, assign these to existing strategies, and measure how strongly which participant influences the regulation.

Study Design/methodology/approach: We use tri-gram analysis, sentiment analysis, and similarity analysis as methods to obtain insights into the political commentary process of European Supervisory Authorities (ESAs) drafts dealing with SFDR.

Findings: Our metadata helps to identify stakeholders and lobbying strategies. We found that the most negative comments came from the regulated, who argued strongly subjectively in a very objective environment of ESG disclosure. We also identified typical lobbying strategies based on arguments, persuasion, and classic cost-benefit considerations.

Originality/value: We generate emotion values and synthesise detailed argument differences and show that modern algorithms can contribute to the identification of interest groups and lobbying strategies. Furthermore, we generate similarity values of arguments that can be taken into account in the analysis of the success of a lobbying strategy.