Geopolitical risk, green finance, and corporate environmental performance: evidence from global uncertainty shocks

Abstract

The growing importance of sustainable finance has intensified interest in understanding how green finance contributes to corporate environmental performance (CEP). However, prior studies have largely assumed a stable relationship between green finance and environmental outcomes, giving limited attention to the disruptive effects of geopolitical uncertainty. This study examines the effect of green finance on CEP and investigates whether geopolitical risk alters the effectiveness of green finance across different institutional and industrial contexts. The study uses a global panel dataset comprising 15,872 firm-year observations across multiple countries and industries. The empirical analysis applies fixed-effects estimation, moderation analysis, System Generalized Method of Moments (System GMM), and panel threshold regression. Industry heterogeneity and governance effects are also examined. In addition, machine learning techniques, including Random Forest, Gradient Boosting Regression Trees (GBRT), XGBoost, and SHAP analysis, are employed to assess predictive performance and identify the relative importance of key determinants of CEP. The findings show that green finance significantly improves corporate environmental performance, while geopolitical risk has a significant negative effect on environmental outcomes. The interaction between green finance and geopolitical risk is negative and statistically significant, demonstrating that geopolitical uncertainty weakens the environmental effectiveness of green finance. Panel threshold analysis identifies a critical geopolitical risk threshold of τ = 142.8, beyond which the positive effect of green finance declines substantially. Energy-intensive industries are found to be more vulnerable to geopolitical shocks than non-energy-intensive sectors. Governance quality strengthens the positive environmental effects of green finance and serves as an important resilience mechanism under uncertain conditions. Among the machine learning models, XGBoost achieves the highest predictive accuracy (R 2 = 0.72), while SHAP analysis identifies green finance, geopolitical risk, and governance quality as the most influential predictors of CEP. The results demonstrate that the effectiveness of green finance is conditional rather than uniform across geopolitical and institutional environments. The study contributes to the sustainable finance literature by introducing and empirically validating the Geopolitical Adaptive Green Finance Effectiveness Theory (G-AGFET), which proposes that the environmental effectiveness of green finance depends on geopolitical stability, governance quality, and institutional resilience. The findings suggest that policymakers should strengthen governance frameworks and design adaptive green finance mechanisms that remain effective during periods of geopolitical uncertainty. Investors and corporate managers should similarly incorporate geopolitical risk and institutional quality into sustainable investment and environmental strategy decisions.