Green finance and the efficiency of energy transition policies: An integrated MCDA and SHAP approach for OECD countries

This study examines how green finance strategies influence energy transition performance in 31 OECD countries. It combines a structured, literature-based indicator framework with WSM, TOPSIS, SMAA robustness analysis, and interpretable machine learning via Random Forest, XGBoost, and SHAP, and double machine learning for orthogonalized effect estimation. The results reveal significant differences among countries, with Denmark ranking as the top performer and several others constrained by fossil fuel dependence and structural barriers. The XGBoost model predicts the SMAA-based country ranking with an R 2 value of 0.447. SHAP analysis identifies real GDP per capita, renewable energy public RD&D, and green patents as the three strongest predictors of transition performance rankings, followed by environmentally related tax revenues and energy productivity. Double machine learning shows that environmental tax revenues carry a robust orthogonalized effect on the transition rank, stable across nuisance learners and control sets. The 31 OECD countries fall into three strategic profiles: 8 renewable-dominant, 8 intermediate-transition, and 15 fossil-dominant. These findings demonstrate that effective green finance strategies require coherence among fiscal instruments, innovation support, energy productivity, and country-specific structural conditions. The study provides a comparative framework for designing differentiated policy strategies aligned with national transition capacities.