Study On The Fluctuation Characteristics And Prediction Of National Carbon Emission Trading Price

In order to cope with the greenhouse effect and various adverse effects caused by global warming,China has committed to peak carbon dioxide emissions by 2030 and strive to achieve carbon neutrality by2060.Establishing a national carbon emission trading market is an important measure to achieve the “dual carbon” goal.The fluctuations and future trends of national carbon emission trading price affect the development direction of various industries and enterprises.Therefore,studying the national carbon emissions trading price has certain significance.This article takes the national carbon emission trading price as the research object,and conducts research from two aspects: fluctuation characteristics and prediction.In terms of the research on the fluctuation characteristics of national carbon emission trading prices,daily closing price data was collected.First,processed it with logarithmic returns,and then used GARCH family models to study its volatility.The empirical results show that the logarithmic yield series will be significantly affected by external shocks and past fluctuations,and has long-term memory;The volatility of the logarithmic yield series has a leverage effect,where the impact of bad information is greater than the impact of positive information.There is no significant correlation between logarithmic yield series and risk,which cannot reflect the potential risks in the national carbon emission trading market.In terms of predicting the national carbon emission trading price,23 factors affecting the national carbon emission trading price were selected from 7 dimensions.Firstly,Lasso regression method is used to screen and analyze the influencing factors.Then,multiple machine learning methods were used to construct regression prediction models before and after screening influencing factors.And use test set data to predict the national carbon emission trading price.Using evaluation indicators such as MSE to verify the predictive ability of the selected influencing factors on the national carbon emission trading price,and the fitting effect and prediction accuracy of machine learning prediction models.The empirical results show that among the 14 selected influencing factors,compared with foreign influencing factors,domestic influencing factors have a greater impact on the national carbon emission trading price;The 14 selected influencing factors can effectively predict the national carbon emission trading price,The model constructed using SVR predicts the test set data to obtain the minimum difference between the predicted value and the true value.The fitting effect and prediction accuracy are relatively best,followed by XGBoost and finally BPNN.Therefore,the SVR model is more suitable for regression prediction of national carbon emission trading price.Finally,based on the conclusions drawn from empirical analysis,suggestions were made to relevant departments regarding the regulation of national carbon emission trading prices,hoping to contribute to stabilizing national carbon emission trading prices and achieving the “dual carbon” goal.