← 論文一覧に戻る

Satellite-Verified Greenwashing: ESG Claim Credibility Among US Heavy Emitters

衛星で検証されたグリーンウォッシング:米国高排出企業におけるESG主張の信頼性 (AI 翻訳)

Pieter de Jong, Inga Timmerman, Alona Bilokha

Crossrefプレプリント2026-01-01#AI×ESGOrigin: US経営インパクト: 資金調達対象セクター: cross_sector
DOI: 10.2139/ssrn.6827435
原典: https://doi.org/10.2139/ssrn.6827435

🤖 gxceed AI 要約

日本語

本稿は、衛星リモートセンシングとNLPテキスト分析を組み合わせ、米国の高排出企業233社のESG開示と実際のメタン排出との乖離(Claim-Reality Divergence index)を測定。ESG主張の強度が高いほど乖離が大きいことを実証し、グリーンウォッシュが投資家向け開示を通じて行われることを示した。

English

This study introduces the Claim-Reality Divergence (CRD) index, combining satellite methane data (Sentinel-5P TROPOMI) with NLP analysis of SEC filings and earnings calls for 233 US heavy emitters (2018-2023). A one standard deviation increase in ESG text intensity is associated with a ~0.66 standard deviation increase in the CRD index, revealing that greenwashing operates through selective investor-facing disclosure rather than falsifying regulatory data.

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

📝 gxceed 編集解説 — Why this matters

日本のGX文脈において

日本においても、SSBJ基準や有報でのサステナビリティ情報開示が進む中、企業の開示と実態の乖離を検証する独立した手段として衛星データの活用が注目される。本稿の手法は、日本の排出量取引制度や投資家対応におけるグリーンウォッシュ対策に示唆を与える。

In the global GX context

This paper provides a scalable, independent verification tool for greenwashing detection using satellite data, directly relevant to SEC climate disclosure rules and global ISSB/CSRD implementation. It demonstrates that text-based ESG claims can diverge from physical emissions, supporting regulatory efforts to close the gap between disclosure and reality.

👥 読者別の含意

🔬研究者:Demonstrates a novel approach combining satellite remote sensing and NLP to quantify greenwashing, with robust empirical evidence on the divergence between ESG claims and actual emissions.

🏢実務担当者:Provides a benchmarking method using independent satellite data to assess credibility of ESG disclosures; useful for compliance and investor relations teams evaluating greenwashing risk.

🏛政策担当者:Supports the case for mandatory, standardized emissions disclosure and satellite-based verification; offers quantitative evidence for regulators designing greenwashing enforcement mechanisms.

📄 Abstract(原文)

Verifying corporate sustainability claims is difficult when available benchmarks rely on firm-supplied or commercially processed data. For energy-intensive industries, satellite remote sensing offers an independent alternative: methane emissions from oil and gas, utilities, and chemicals are large enough to detect from orbit, making satellite-based verification both feasible and credible. This study introduces the Claim-Reality Divergence (CRD) index, which triangulates ESG language intensity from SEC filings and earnings call transcripts against facility-level methane reported to the EPA Greenhouse Gas Reporting Program and atmospheric methane column densities observed by the Sentinel-5P TROPOMI satellite, a physical measurement source that firms cannot influence or selectively report. Using 1,937 firm-year observations across 233 publicly listed US heavy emitters from 2018 to 2023, a one standard deviation increase in ESG text intensity is associated with a 0.635 to 0.694 standard deviation increase in the CRD index across eleven specifications, robust to financial controls, emissions baseline controls, and firm fixed effects. Annual 10-K-based scoring confirms the greenwashing premium is a cross-sectional phenomenon: persistent differences in ESG communication posture across firms predict claim-reality gaps, but year-to-year within-firm variation does not. Taken together, the results indicate that greenwashing among US heavy emitters operates through selective investor-facing disclosure rather than falsification of regulatory data, and that satellite-based benchmarking provides a scalable, independent verification tool with direct implications for emissions disclosure regulation.

🔗 Provenance — このレコードを発見したソース

🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録(無料・3キーワードまで)。

gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。