Carbon Trading Price Forecasting In China Considering Market Linkage Effect

Based on the background of greenhouse gas emissions caused by industrialization,China introduced the carbon trading mechanism in 2011 and rapidly established and developed eight regional carbon trading markets in a decade,promoting regional energy saving and emission reduction through carbon trading quotas,carbon taxes and other regulated carbon trading measures,further deepening the green environmental goal of sustainable development.In 2021,the establishment of a national carbon trading market to promote carbon trading from a regional to a unified A national market will be set up to stimulate carbon trading from the supply side to the demand side,which will expand the market from regional high-polluting enterprises to those with a demand for carbon emissions trading across the country,thus becoming an extension and unification of the national carbon trading policy.However,while the carbon trading mechanism is highly valued,the current carbon trading market structure is characterized by strong regional aggregation and high price volatility.In summary,this paper examines three representative carbon trading markets in Guangdong,Hubei and Shanghai,taking into account the volatility of each market and the interconnectedness of each other,and collects a variety of influencing factors to extract factors with information characteristics and construct a carbon trading price prediction model.Firstly,this paper considers the volatility characteristics of the carbon trading market.This paper constructs a DCC-GARCH model that responds to the volatility and dynamic correlation of the price series.On one hand,purpose of this research is to observe carbon trading prices’ volatility.On the one hand,the linkage effect between carbon markets are valued as well.The empirical study shows that Hubei,Guangdong and Shanghai carbon trading pilots are all characterised by volatility aggregation.The Hubei and Guangdong markets are characterised by stable volatility and long persistence,while the Shanghai market is characterised by high volatility and short persistence.The dynamic correlation coefficient shows that there is a weak positive correlation between the carbon trading market returns of Guangdong and Hubei,and Guangdong and Shanghai,which may be attributed to the volatility of the market and the weak linkage effect due to the diverse characteristics of the market construction time and trading subjects,reflecting the strong regional structural characteristics of the carbon trading market.Secondly,this paper constructs the influencing factors for carbon trading prices.As a matter of fact,this paper collects factors which lead to changes in carbon prices from multiple aspects,considering market linkages,energy prices,regional climate environment,overall macroeconomic and regional economic development indices,and domestic and international related carbon trading factors.LASSO model is used to filter factors that do not have influence on prices.The study found that market linkages,energy factors,climate factors and macroeconomic factors all have a certain degree of influence on the prices of the three carbon trading markets but not to the same extent.Meanwhile,the Chinese financial market has no influence on the price of carbon emissions.Based on the above analysis,the influence of the Chinese financial market was excluded.In order to avoid multicollinearity among the factors,the factors of each market were extracted by PCA with 88% information coverage for subsequent price prediction.In terms of carbon trading price prediction,this paper decomposes the carbon trading price series by CEEMDAN method and reconstructs the series into highfrequency,medium-frequency and low-frequency series,and then introduces external influencing factors to the reconstructed series for LSTM model prediction.The prediction accuracy of the CEEMDAN-LASSO-LSTM combined model was found to be higher than that of the traditional decomposition method or machine learning prediction model by comparing the two perspectives of model decomposition method and prediction model selection with the prediction accuracy of the CEEMDANLASSO-LSTM combined model.In summary,there is strong regional heterogeneity and volatility in the carbon trading prices of Hubei,Guangdong and Shanghai,with weak positive linkage effects among the markets.The CEEMDAN carbon trading price decomposition reconstruction method and LSTM prediction model based on LASSO factor screening can better predict the carbon trading price trend and promote the sound development of the national carbon trading market.Therefore,this paper proposes some suggestions such as promotion on the optimization of the national unified carbon trading market system,establishment on risk control system of the regional carbon trading market,building the channel between carbon trading prices and enterprise costs,and the development of carbon financial products.Therefore,these measures can promote the further maturation of the carbon trading market,enhance market activity,implement environmental economic policies,and achieve the goal of carbon neutrality and carbon peaking.