Research On The Driving Mechanism Of Carbon Emissions Based On Panel Quantile Regression Model

With the rapid development of the world economy,the process of urbanization and industrialization is accelerating.The rapid global climate change caused by the continuous growth of carbon emissions has seriously endangered the living environment as well as health and safety of human beings.Due to the huge pressure of international carbon emission reduction and the need of sustainable development strategy,the transformation of low-carbon development is urgent.In order to address the climate change and promote energy conservation and emission reduction,it is conducive to providing information support and decision-making reference for making practical and reasonable policy recommendations through exploring the influencing factors and the driving mechanism of carbon emissions.At the same time,the non-normal and asymmetric characteristics of energy economic data are difficult to meet the basic assumptions of the mean panel regression model and the quantile regression provides a powerful tool to solve this problem.Quantile regression is a linear function regression model based on the conditional distribution of dependent variable to fit independent variable.It has good fitting effect for non-normal and heteroscedasticity data.It can describe the influence of independent variable on different conditional distribution of dependent variable.Based on the panel data model and quantile regression theory,this paper systematically models the driving factors of carbon emissions in four scales:national,regional,provincial and sector and quantitatively analyzes the influence of driving factors on the whole conditional distribution of carbon emissions,especially focusing on the influence in the upper tail and lower tail.It can effectively overcome the limitations of the traditional mean panel data model in theory and get more accurate and comprehensive estimation results.It provides technical support and analysis for guiding the scientific decision-making of emission reduction policy classification in the empirical analysis.Firstly,based on the energy-emissions relationship,the panel quantile causality test,which aims at the traditional Granger test only reflects the directional information between variables from the perspective of the mean,is constructed to analyze the causality between renewable and non renewable energy consumption,economic growth and carbon emissions in BRICs countries in different quantile intervals.This study employs nonparametric method to estimate the panel quantile Granger causality with quantile non causality sup-Wald test and compares the results with the mean causality based on the Panel Error Correction Model(PECM).The results indicate that the causal relationship from renewable energy consumption to carbon emission is only significant in low quantile intervals including[0.05,0.5],[0.2,0.4]and[0.4,0.6]in Russia.The causal relationship from non renewable energy consumption to carbon emission exists in the quantile intervals[0.05,0.5],[0.05,0.2]and[0.2,0.4]of Brazil,South Africa and most of the quantile intervals of Russia,India and China.The causal relationship from economic growth to carbon emissions exists in all quantile intervals of BRICs countries.The heterogeneity of causality between variables in different conditional distributions is verified.Secondly,based on the economy-emissions relationship,the fixed panel quantile regression model,which aims at the biased parameter estimation of panel data model in heterogeneous distribution,is constructed to analyze the impact of foreign trade and FDI on carbon emissions in the eastern,middle and western region of China.The simultaneous equations including carbon emission equation and economic growth equation are constructed to examine the indirect impact of foreign trade and FDI on carbon emissions through promoting economic growth with the use of the two-stage least square method(TSLS).The parameters of panel quantile regression model are estimated based on penalty function and the results are compared with those of mean regression model estimated by the least square method.The direct impact of foreign trade on the high-emission provinces in the eastern region is significantly negative and the positive indirect impact is greater than the negative direct impact in low-emission provinces,and less than the negative direct impact in high-emission provinces.The total impact of foreign trade significantly promotes carbon emissions in low-emission provinces in the eastern region while it restrains the carbon emissions in high-emission provinces.The direct impact of foreign trade in the central region is not significant and the negative indirect impact in the quantile below 30%is greater than the positive direct impact and the indirect impact of the quantile above 30%is positive.Therefore,the total impact of foreign trade is conducive to curbing emissions of low-emission provinces in the central region while significantly increasing carbon emissions of high-emission provinces.The positive impact of foreign trade in the western region is significant for the quantile below 50%and the negative indirect impact is greater than the positive direct impact at all quantiles.As a result,the total impact of foreign trade is positive.The direct,indirect and total effects of FDI on carbon emissions in the eastern,central and western region of China are similar to those of foreign trade.The results also verify that the EKC hypothesis between economic growth and carbon emission holds in the eastern region of China.Thirdly,based on the policy-emissions relationship,the Moran index is calculated by using the spatial autocorrelation test.The spatial panel quantile regression model,which aims at the spatial spillover effect of cross-section individuals,is constructed to analyze the impact of different types of emission reduction policies on China's provincial carbon emissions.Based on Bayesian inference,MCMC algorithm is designed to estimate the parameters of spatial panel quantile regression model and the results are compared with those of spatial lag panel regression model.The results show that there is a significant spatial dependence of carbon emissions.The command and control emission reduction policy is an effective way to promote emission reduction in low-emission provinces while it shows the green paradox effect in high-emission provinces.The market-oriented emission reduction policy has a significant negative impact on carbon emissions at all quantiles and the emission reduction effect is more obvious at high quantiles.The positive emission reduction effect of market-oriented emission reduction policies is verified.Finally,based on the extended STIRPAT model,the study employs the fixed dynamic panel quantile regression model to analyze the driving factors of carbon emission intensity(CEI)in China's transport sector.The instrument variable method is used to estimate the dynamic panel quantile regression model and the results are compared with those of the system generalized moment estimation method(SYS-GMM).The results demonstrate that there exists a significant dynamic effect in the changes of CEI.Economic growth significantly promotes the growth of CEI in the transport sector in low-CEI provinces while it effectively curbs the CEI of high-CEI provinces.The positive impact of energy intensity on CEI increases with the increase of quantile levels.The number of private cars and cargo turnover are the driving factors for the increase of CEI in low-CEI provinces.The coefficient of urbanization is significant and positive at the most of quantiles,indicating that improving the urbanization level is not conducive to the reduction of CEI in most of provinces in China.Besides,the results also verify the validity of EKC hypothesis between economic growth and CEI in the transport sector of low-CEI provinces.