Estimation and forecast of carbon emission market volatility based on model averaging method

Understanding volatility is essential for risk management and green investment decision-making in the carbon market. However, existing studies lack a unified framework for modeling and estimating carbon market volatility, and predictions are often affected by model uncertainty. Using data from EU emission allowances, we estimate parameters for multiple GARCH models via the Sequential Monte Carlo method and improve forecasting accuracy with model averaging techniques. Our results reveal that carbon market volatility is characterized by spikes, thick tails, asymmetry, and jumps. Based on Model Confidence Set test, model comparison demonstrates that averaged models consistently outperform individual models across various loss criteria. By integrating information from multiple models, the model averaging approach simplifies model selection and plays a pivotal role in supporting volatility timing strategies.