Carbon price prediction incorporating media attention to low-carbon news

ZHA Donglan; LI Jing; CAO Yang; YAO Zhengyi

Abstract：Accurate carbon price prediction can offer valuable references for carbon market participants in formulating production and operation plans as well as investment and trading strategies.Taking the spot carbon quota price in the China's national carbon market as research subject,this study constructs a low-carbon news media attention index(CPI)using daily data from five major media outlets and uses grey correlation analysis method to conduct correlation analysis among variables.Carbon price prediction and explanations analysis were carried out by combining the extreme gradient boosting(XGBoost),Shapley additive explanations(SHAP)and cumulative local effects(ALE).The results indicate that CPI is strongly correlated with the carbon price,and incorporating CPI can significantly decrease the prediction error and enhance the accuracy of trend judgment.The global contribution of CPI to carbon price prediction is larger than those of crude oil prices,stock market indices,and exchange rates.Moreover,the marginal effect of CPI on the carbon price exhibits a nonlinear characteristic.Carbon market participants ought to pay attention to the effects of low-carbon news reports on price shock in carbon markets.