Replication Package: Say-Do Gap in the Age of AI — External Media Verification of ESG Disclosure Quality and Firm Value in IBEX 35 Firms

AI時代の言動ギャップ：IBEX35企業のESG開示品質と企業価値の外部メディア検証 (AI 翻訳)

Alfredo Merlet

Zenodo (CERN European Organization for Nuclear Research)データセット2026-08-05#AI×ESGOrigin: EU経営インパクト: 資金調達対象セクター: cross_sector

DOI: 10.5281/zenodo.21813917

原典: https://doi.org/10.5281/zenodo.21813917

🤖 gxceed AI 要約

日本語

スペインIBEX35企業を対象に、AI（BETO）を用いてESG開示の「言動ギャップ」指数を構築。メディア検証による未検証コミットメントを検出し、企業価値（Tobin's Q等）との関連を分析。ESG開示の信頼性評価にAIを活用する先駆的枠組みを提供。

English

This replication package constructs a Say-Do Gap Index for IBEX 35 firms (2019-2025) using a fine-tuned Spanish BERT (BETO) to detect unverified ESG commitments from media sources. It links the index to firm value (Tobin's Q, ROA, ROE) via two-way fixed-effects models, offering a novel AI-driven approach to ESG disclosure quality assessment.

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

日本ではSSBJ開示が始まり、ESG情報の信頼性検証が課題。本手法はメディア検証による開示品質評価のAI活用事例として、日本の開示監査や投資家対応に示唆を与える。

In the global GX context

Globally, this addresses the growing need for AI-based assurance of ESG disclosures, complementing ISSB/CSRD frameworks. The Say-Do Gap methodology offers a scalable way to verify corporate commitments, relevant for regulators and investors seeking to combat greenwashing.

👥 読者別の含意

🔬研究者:Provides a reproducible AI pipeline for ESG disclosure quality measurement and its valuation impact.

🏢実務担当者:Offers a tool to benchmark and improve ESG disclosure credibility, potentially enhancing investor trust.

🏛政策担当者:Demonstrates a media-based verification approach that could inform ESG assurance standards.

📄 Abstract（原文）

This repository contains the code and data panels used to construct and analyze the Say-Do Gap Index for the 34 constituent firms of the IBEX 35 (Spain) covered in this study, 2019-2025, as reported in the accompanying paper submitted to Meditari Accountancy Research (Special Issue: AI in Sustainability Accounting and Assurance). It includes: (1) the full pipeline for GDELT/BigQuery corpus extraction, (2) the manually annotated training set and targeted enrichment procedure for the "unverified commitment" taxonomy category, (3) the BETO (Spanish BERT) fine-tuned classifier scripts, (4) the firm-year Say-Do Gap Index panel, (5) the merged financial panel (Tobin's Q, ROA, ROE), and (6) the two-way fixed-effects econometric analysis scripts. See README.md for full file-by-file documentation.

🔗 Provenance — このレコードを発見したソース

openalex https://doi.org/10.5281/zenodo.21813916first seen 2026-08-07 05:08:17 · last seen 2026-08-07 05:08:18

🔔 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。