Structural changes and volatility forecasting in China’s carbon markets: A hybrid approach integrating GARCH-SVR with LASSO-weighted windows method

As a newly established and rapidly developing financial market, China’s carbon markets are highly susceptible to structural changes driven by policy adjustments, external shocks, and immature trading mechanisms, which pose critical challenges for volatility forecasting. This study focuses on volatility forecasting of China’s carbon markets, proposing a novel hybrid forecasting approach to address these challenges. We first apply the hybrid GARCH-SVR method that combines the strengths of GARCH models in capturing volatility clustering with SVR’s ability to model nonlinear dynamics. To address model uncertainty arising from structural changes, we further develop a LASSO-weighted window method, where weights of GARCH-SVR forecasts across different windows are determined via LASSO regression. Empirical results show that LASSO-weighted window approach outperforms standalone GARCH models and GARCH-SVR in forecasting accuracy. Robustness tests with alternative windows and volatility proxies confirm this superiority. Importantly, economic value evaluation based on portfolio performance demonstrates that our method enhances asset allocation efficiency, providing practical guidance for carbon asset management.