Greenwashing Detection via NLP and its Correlative Effect on Municipal Bond Yields and Corporate Sustainable Debt Markets in the US

• Billy Elly

  Lautech

  Author

The proliferation of environmental, social, and governance (ESG) investment products has created perverse incentives for issuers to exaggerate sustainability credentials, a phenomenon known as greenwashing. This practice undermines market efficiency by distorting price signals in both municipal bond and corporate sustainable debt markets. This study develops and validates a Natural Language Processing (NLP) framework employing a fine-tuned FinBERT model to systematically detect greenwashing in corporate earnings call transcripts and municipal bond official statements. The detection methodology achieves an overall classification accuracy of 90.2% (F1-score: 89.4%), with greenwashing intensity scores ranging from 0 to 2 across a sample of 30,000+ disclosures. Regression analysis reveals that a one-unit increase in greenwashing intensity correlates with a 12–18 basis point increase in municipal bond yields and a 22–35 basis point premium on corporate green bonds, after controlling for credit risk and market factors. The effect is most pronounced in the utilities sector and for issuers with weaker governance structures. These findings provide empirical evidence that markets penalize perceived greenwashing through higher borrowing costs, offering actionable implications for regulators, investors, and policymakers seeking to preserve integrity in sustainable finance markets.

• PDF

Copyright (c) 2026 Billy Elly (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.