ESG Disclosure and Corporate Tax Avoidance: The Moderating Effects of State Ownership and Financial Constraints-Evidence from Vietnamese Non-Financial Firms
ESG開示と企業の税回避行動：国有企業と財務制約の調整効果－ベトナム非金融企業の証拠 (AI 翻訳)
Hieu Thanh Nguyen, Hoa Minh Pham, Anh Thao Nguyen, Linh Khanh Long, Ngoc Minh Nguyen, Anh Duc Phan
🤖 gxceed AI 要約
日本語
本研究は、ベトナムの非金融上場企業118社を対象に、ESGパフォーマンスが税回避行動に与える影響を分析。ランダム効果モデルにより、ESG開示の質（個別E,S,Gおよび複合指標）が実効税率で測った税回避と負の相関があることを確認。特にS（社会）柱の効果が顕著だが、複合指標の影響が最も強い。また、財務制約や国有企業ではこの軽減効果が弱まる。機械学習（CatBoost）を用いた分析では、ESG変数を組み込むことで税回避の予測精度が52.92%に向上。
English
This study examines the impact of ESG disclosure on tax avoidance of 118 Vietnamese non-financial listed firms (2020-2024). Using random effects models, it finds that ESG performance (individual E, S, G pillars and a composite index) is negatively associated with tax avoidance, measured by effective tax rate. The social pillar shows the strongest individual effect, but the composite index has the most substantial impact. State ownership and financial constraints weaken this mitigating effect. A CatBoost machine learning model incorporating ESG achieves 52.92% predictive accuracy for tax avoidance, outperforming a non-ESG XGBoost model (38.14%).
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
ベトナムの非金融企業を対象とした本研究成果は、日本企業がベトナム市場で事業展開する際の税務戦略やESG開示の重要性を示唆。また、日本におけるESGと税務の関連研究や、SSBJ開示基準の実務対応にも参考となる。
In the global GX context
This paper provides novel evidence from an emerging market on the relationship between ESG disclosure and corporate tax avoidance, highlighting the moderating roles of state ownership and financial constraints. It demonstrates the value of machine learning in ESG-related predictive modeling, relevant to global disclosure standards and ESG integration into corporate strategy.
👥 読者別の含意
🔬研究者:The paper offers empirical evidence on the ESG-tax avoidance nexus and demonstrates the application of machine learning (CatBoost) for predictive modeling in this context, valuable for scholars in corporate governance and sustainability accounting.
🏢実務担当者:Corporate tax and sustainability teams can use the findings to understand how ESG performance may reduce tax avoidance, particularly under different ownership and financial conditions, informing integrated reporting and risk management.
🏛政策担当者:Regulators can consider the implications for tax policy and ESG disclosure mandates, especially how state ownership and financial constraints affect corporate behavior, contributing to the design of effective sustainability reporting frameworks.
📄 Abstract（原文）
This study investigates the impact of ESG performance on the tax avoidance behavior of 118 non-financial listed firms in Vietnam from 2020 to 2024. Employing a Random Effects Model (REM), empirical results reveal that sustainability reporting quality-measured by individual E, S, and G pillars and a composite ESG index-is negatively associated with corporate tax avoidance, proxied by the Effective Tax Rate (ETR). Among these, the social (S) pillar exerts the most pronounced effect; however, individual component impacts remain less substantial than the comprehensive ESG index. Furthermore, findings indicate that the mitigating effect of ESG on tax avoidance significantly weakens when firms face financial constraints or operate under state ownership. Notably, applying machine learning techniques demonstrates that a CatBoost algorithm integrating the ESG variable achieves 52.92% predictive accuracy for tax avoidance, outperforming an XGBoost model lacking ESG inclusion (38.14%). Additionally, feature importance analysis of financial and non-financial variables highlights ROA as the dominant financial predictor (35.5%), while ESG contributes a notable 10.35% to the model's explanatory power. Ultimately, these findings provide vital insights for policymakers and investors regarding the interplay between sustainability commitments, ownership structures, and corporate tax strategies.
🔗 Provenance — このレコードを発見したソース
- openaire https://doi.org/10.20944/preprints202604.1159.v1first seen 2026-06-11 05:17:02
🔔 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。