Forecasting carbon–coal price co-movement evolution and its implications for emission reduction: An AI-based complex network approach

Carbon markets are a central instrument for emission reduction, but their effectiveness cannot be fully understood without considering the co-movements between carbon prices and fossil energy prices, which embed important signals for emission-reduction incentives but have received little attention in existing research. Accordingly, this study aims to analyze and forecast the evolutionary patterns of carbon–coal price co-movement modes and to evaluate their implications for emission reduction. To achieve this goal, the study integrates a complex network representation of price co-movement modes with AI-based learning tools, including GraphSAGE and XGBoost, to model and predict transitions among these modes. Empirically, the framework is applied to the EU ETS, China’s national ETS, and seven regional pilot markets. The results reveal that the EU ETS and China’s national ETS follow opposite evolutionary trajectories in carbon–coal price co-movement structures. In the EU ETS, co-movements favorable for emission reduction have historically dominated, but the forecasted evolution indicates an increasing prevalence of unfavorable patterns. By contrast, China’s national ETS is currently characterized by unfavorable co-movements, yet the expected transition toward more favorable patterns suggests a strengthening of emission-reduction incentives. In addition, China’s regional carbon–coal price co-movement pattern exhibit pronounced heterogeneity across pilot markets. By providing forward-looking signals on how carbon–coal price co-movement patterns are likely to evolve, this study enables the anticipation of shifts in emission-reduction incentives. This study offers forward-looking insights into carbon market development and supports policies that promote the transition toward green energy.