DeepGreen: Effective LLM-Driven Green-washing Monitoring System Designed for Empirical Testing -- Evidence from China

TL;DR

DeepGreen presents a dual-layer LLM-driven framework to detect corporate green-washing from financial statements by extracting and evaluating green keywords and their implementation. It introduces GreenImplement ($G.I.$) as a core quantitative indicator derived from the ratio of implemented to claimed green actions, and validates it against Huazheng ESG scores. Empirical analysis on 204 annual reports from 68 Chinese A-share firms (2021–2023) shows $G.I.$ correlates positively with asset return ratio ($ARR$) and aligns with ESG ratings, with sizable heterogeneity by firm size. The work demonstrates that genuine green practices can boost performance and that DeepGreen offers a practical, interpretable tool for regulators and investors to monitor green-washing beyond traditional methods.

Abstract

This paper proposes DeepGreen, an Large Language Model Driven (LLM-Driven) system for detecting corporate green-washing behaviour. Utilizing dual-layer LLM analysis, DeepGreen preliminarily identifies potential green keywords in financial statements and then assesses their implementation degree via iterative semantic analysis of LLM. A core variable GreenImplement is derived from the ratio from the two layers' output. We extract 204 financial statements of 68 companies from A-share market over three years, comprising 89,893 words, and analyse them through DeepGreen. Our analysis, supported by violin plots and K-means clustering, reveals insights and validates the variable against the Huazheng ESG rating. It offers a novel perspective for regulatory agencies and investors, serving as a proactive monitoring tool that complements traditional methods.Empirical tests show that green implementation can significantly boost the asset return rate of companies, but there is heterogeneity in scale. Small and medium-sized companies have limited contribution to asset return via green implementation, so there is a stronger motivation for green-washing.

Figures (5)

• Figure 1: Overview of DeepGreen Framework.
• Figure 2: Violin Plot of Experimental Results under Different Manually Classified Variables
• Figure 3: K-Means Cluster Result.
• Figure 4: A Comparison of Differences and Advantages
• Figure 5: Word Cloud Plot of Top 100 Words