Green Finance and Economic Growth Across Global Financial Centres: Evidence from Econometric and Machine Learning Methods

As financial systems undergo green transformation, systematic performance evaluation becomes increasingly important for policy formulation and strategic planning. Drawing on global surveys of finance professionals and a validation framework incorporating 127 instrumental indicators across sustainability, business climate, infrastructure, and human capital, Z/Yen publishes the Global Green Finance Index (GGFI). Using GGFI measures of both depth and quality for 99 international financial centres (after data preprocessing; note that GGFI 16, October 2025, covers 94 centres), this study examines their relationship with economic growth from 2018H1 to 2024H2. Economic growth is measured by real GDP per capita growth rates derived from the World Bank (2026) Databank. A panel vector autoregression (PVAR) model is employed to estimate dynamic interactions between green finance development and economic growth. Model stability and specification validity are confirmed through AR root tests for joint stationarity and LM tests for residual serial correlation. Impulse response functions show that real GDP per capita growth initially fluctuates following shocks to GGFI depth and quality during the first six semiannual periods, then converges toward a persistent positive response. Cumulative impulse responses remain non-negative, and Granger causality tests indicate a one-way causal relationship running from GGFI to economic growth, with only weak evidence of feedback from economic growth to green finance quality (p < 0.10). Robustness checks using purchasing-power-parity-adjusted GDP per capita growth yield consistent results. To account for potential nonlinearities and cross-regional heterogeneity, the analysis further applies Double Machine Learning (DML) with Random Forest and Gradient Boosting learners. The DML estimates indicate a statistically significant positive average marginal effect of GGFI on global GDP growth. However, continent-level analysis reveals heterogeneity. The positive association is robust and highly significant in the Asia-Pacific region, while statistically insignificant in Europe and North America during the sample period (average marginal effect ≈ 1.55–1.57 basis points per unit increase in GGFI). These findings suggest that the growth effects of green finance are not uniform across financial centres but vary with regional structural conditions and development stages. A dummy variable capturing the Covid-19 period (2020H1-2023H2) does not produce statistically significant effects across model specifications.