Estimation and Forecast of Carbon Emission Market Volatility Based on Model Averaging Method

The carbon market, as a market operating with carbon emission rights for core trading, plays an important role in reducing the production of greenhouse gases and controlling the risk of climate change caused by environmental pollution but also shows complex and changeable dynamic characteristics. Estimating and predicting the evolution of carbon market volatility can not only provide data for the correct measurement of market risk and calculation of value at risk but also provide instructions for the pricing of carbon-related assets and the construction of investment portfolios. In this paper, based on the EUA carbon futures price series, various GARCH models and SV models are used to describe and forecast various properties of volatility, such as spike and thick tail asymmetry and jump characteristics. Considering the complexity of the model, the model averaging methods are used to infer the whole alternative model space to reduce the uncertainty of volatility prediction, which extends the existing research. The empirical results show that the persistence of EUA carbon market volatility is strong and that there is a certain leverage effect consistent with the characteristics of traditional financial markets. However, different from the traditional financial market, the jump of volatility has expressed characteristics of smaller probability but larger amplitude. The MCS test shows that the model averaging methods present higher prediction accuracy, and the prediction errors are smaller than those of the single model. Instead of blindly selecting a single model for carbon market volatility forecasting, investors should take advantage of model averaging methods to reduce information uncertainty under different loss functions in terms of research purpose and actual needs