ESG and Financial Distress: The Role of Disclosure Quality in Predictive Accuracy
ESGと財務危機: 予測精度における開示品質の役割 (AI 翻訳)
Iulia Florentina Voicila Voicila, Elena UrquiaGrande
🤖 gxceed AI 要約
日本語
スペインと英国の87,225社の非上場企業データを用い、ESG指標が財務危機予測精度を向上させる条件を分析。英国では標準化された開示によりESG情報が予測力を高めるが、スペインでは不完全な開示のため効果なし。AI/MLモデルの優位性も実証。
English
Using 87,225 private firms from Spain and the UK, this study shows ESG indicators improve financial distress prediction only when disclosure quality is high (UK). In Spain, fragmented ESG data yields no improvement. Machine learning models, especially neural networks, extract more value from robust ESG 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
SSBJが有報でのサステナビリティ開示基準を整備する中、本論文は開示の質が財務的価値に直結することを示す。日本企業のESGデータの比較可能性向上が投資家の早期警戒モデルに有用である可能性を示唆。
In the global GX context
The cross-country evidence directly supports ISSB and CSRD goals of standardized, comparable ESG disclosure. It demonstrates that disclosure quality is not just a reporting issue but a prerequisite for ESG metrics to have financial materiality in risk prediction.
👥 読者別の含意
🔬研究者:Empirical evidence that disclosure quality moderates the predictive value of ESG for financial distress, reinforcing materiality theory.
🏢実務担当者:Highlights the necessity of consistent ESG reporting to enable reliable early-warning models for credit risk or investment decisions.
🏛政策担当者:Justifies regulations that mandate standardized, auditable ESG disclosure to unlock the financial system's ability to price sustainability risks.
📄 Abstract（原文）
This study investigates whether environmental, social, and governance (ESG) indicators improve the prediction of multiclass financial distress beyond traditional accounting ratios. Using a cross-country dataset of 87,225 privately held firms from Spain and the United Kingdom, we estimate multinomial logistic regression, random forest, and multilayer perceptron models, each evaluated with and without ESG information. We show that the predictive value of ESG indicators is not universal but seriously dependent on disclosure quality. In Spain, where environmental subdimensions are fragmented and exhibit missingness, ESG variables provide no incremental improvement in precision, recall, or F1‑scores. By contrast, in the United Kingdom, where ESG reporting is consistent and standardized, ESG indicators generate clear and systematic gains across all models, with the largest improvements observed in the distress category.Machine‑learning models, particularly neural networks, outperform multinomial logistic regression and extract more predictive value from ESG variables when disclosure is robust. These findings demonstrate that disclosure quality is a necessary condition for ESG information to contain meaningful distress prediction signals, offering new empirical evidence on the financial materiality of sustainability metrics. The results have direct implications for regulators and practitioners, highlighting that ESG-based early warning models are effective only when supported by comparable and high‑quality non‑financial reporting.
🔗 Provenance — このレコードを発見したソース
- openaire https://doi.org/10.2139/ssrn.6516488first seen 2026-06-11 05:08:39
🔔 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。