Analysis Of Influencing Factor And Prediction To Transaction Price Of Carbon Emission

The construction of the national carbon emission market was official launched on December 19,2017.However,there are essential difference between the starting and the trading.At present,our country is in the bottom of the carbon emission transactions,facing the imperfect market mechanism.International voice is inadequate and the motivation is still insufficient.In order to realize the total goal of energy saving and emission-reduction,the formal trading of the national carbon market and avoiding risks reasonably for enterprises,from the perspective of the government and enterprises,we take 6 polit regions as the research objects,taking the transaction data from January 1,2017 to June 30,2018 as the training set and July 1,2018 to September 30,2018 as the test set,and mainly explore the mechanism of the influencing factors of carbon price in China and the prediction of the carbon price.On the one hand,through Lasso regression we screened out the important factors influencing carbon price from the four fundamentals of macro-economy,energy price,climate and environment and the international market,and established the static panel regression model for six pilot regions in China based on the selected important factors.The results show that there are significant intercept term differences in the influence of influencing factors on carbon price in various pilot areas.Coal price,natural gas price,extreme weather and air quality index have significant influence on carbon price in China.The price of fossil fuels has a negative impact on the price of carbon,which decreases successively from the price of coal,natural gas and the crude oil.The macroeconomy has a weak impact on the carbon market and the stock market has a weak linkage.On the other hand,we predicted the price of carbon emission for the period of future time based respectively on the Lasso selection of important variables to establish the gray BP neural network model,on ARMA innovation model and Prophet model based on the carbon price volatility itself.Compared with the characteristics of various prediction methods and prediction error,we can concluded that the gray BP neural network model suits for short-term carbon price trend and volatility forecasting,which will be more accurate.And the prediction error of 15 phases is less than the time series model.In addition,compared with the traditional time series prediction method,in the long-term forecast Prophet model is more robust.According to the empirical results,we put forward the following suggestions for carbon emissions trading market participants.Firstly according to the principle of the effect of influencing factors,we should gradually realize the ‘dominance of coal,natural gas utilization,clean energy in full swing’ path of development,improve the ‘fossil fuels-carbon market’ restraint mechanism,gradually change the role of government from leading to the market mechanism to reducing market intervention.Secondly,Proposal for carbon price forecasts is using gray BP neural network model for short-term forecasting of the carbon price,but we need control the time in the period of 15.However,for the long-term forecasts,we suggest using the Prophet model.The operation convenience and decomposition characteristics of model can help researchers have deep exploration for a single factor and support carbon market personnel better understand the carbon market.To summary,the selection of variables by Lasso regression is more objective.Also through comparing the prediction methods based on the influence factors with those based on time series,the advantages and disadvantages of each method are clarified,and the application of Prophet model in the carbon emission market is enriched.