Research on the Assessment of Greenwashing in Corporate Environmental Information Disclosure Based on Large Language Models and the Dynamic Mechanism of Green Financing

Abstract

This study is situated against the backdrop of global green economic transformation and examines the dynamic internal mechanisms linking greenwashing behavior in corporate environmental information disclosure with green financing policies. Drawing upon an interdisciplinary theoretical framework, the study constructs and empirically tests a model addressing the systematic discrepancy between the information conveyed in Environmental, Social, and Governance reports and actual green performance. The research gap lies in the quantitative identification of greenwashing phenomena and in the systematic analysis of the dynamic transmission mechanism by which green financing exerts both incentive and constraint effects over different time horizons. Employing large language models for in‐depth semantic extraction and sentiment quantification of corporate environmental disclosures, and integrating a combined econometric model based on the Slack-Based Measure and the Malmquist index for precise measurement of actual green practices, a comprehensive greenwashing index is constructed. Subsequently, dynamic overlapping difference-in-differences and symbolic regression models are utilized to conduct a multidimensional, multi-period dynamic analysis of the short-term incentive and long-term constraint effects of green financing policies on corporate greenwashing behavior. The findings, underpinned by a methodological fusion of natural language processing techniques and econometric modeling, provide a novel empirical perspective on the dynamic mechanism linking green financing policies to corporate environmental governance and offer a data-driven theoretical basis for policymakers.