Explore CKC Based On Generalized Additive Mixed Models Considering Unobservable Time-Related Effects

Economic growth is the focus of attention above all countries in the world,but the environmental pollution caused by economic growth is also receiving increasing attention from governments and people around the world.Among them,the carbon dioxide emissions that cause the greenhouse effect are the focus of research.How to cope with the continuous growth of carbon dioxide emissions is a serious challenge for all governments.As the country with the largest global carbon dioxide emissions,China is not only facing the pressure of public opinion brought about by carbon dioxide emission reduction in the international arena,but also being restricted by the material demand of the economic upswing in domestic.Therefore,how to reduce the carbon dioxide emissions while maintaining the stability of economic growth is an urgent problem that we need to solve.In this context,it is particularly important to study the intrinsic relationship between carbon dioxide emissions and economic growth.This paper focuses on the analysis of the nonlinear relationship between economic effects and unobservable time-related effects and carbon dioxide emissions under consideration of space-time heterogeneity.By establishing a non-parametric model to fit the effects of economic effects and time effects on carbon emissions,we explore the trend of carbon emissions from the national and inter-provincial regions and whether there is an "inverted U-shaped" CKC.The main contents and conclusions of this paper include the following points:Firstly,the measurement of carbon dioxide emissions and economic development indicators.Based on the data of 28 provinces in China from 1997 to 2016,according to the calculation method of carbon dioxide emissions proposed by IPCC,this paper obtains the summary data of carbon dioxide emissions from eight types of energy fossil combustion in the province.In the simple fitting of the relationship between per capita carbon emissions and per capita real GDP by using local weighted regression scatter smoothing method,it is found that there is no"inverted U-shaped" CKC in China from the national point of view.There are"inverted U-shaped" CKC in Beijing and Shanghai in the eastern part of China,and other regions are similar to the national curve.At this time,from the national perspective as a whole,there is no turning point in carbon dioxide emissions.Secondly,comparative analysis is carried out under the premise of setting spatial factors,and whether time-related factors are added to observe whether the model will affect CKC.Different from setting parameter in most previous literatures,this paper uses the GAMs to analyze the CKC considering the relationship between carbon emission and economic development level is not a specific linear relationship or monotonic relationship.In the model,the economic effect is used as smooth term structure,the fitting function is not limited and the variable is more flexible.Considering the differences between provinces and time zones,the relationship may be different between carbon emissions and economic development levels.Therefore,we add the spatial heterogeneity and temporal heterogeneity to the model as dummy variables,and compare the analysis of the results.The study found that both the temporal and spatial dummy variables have a significant impact on carbon emissions in the presence of significant economic effects of the smooth term structure.And under the premise of setting spatial factors,adding time-related factors will make the model regression overall higher and smoother.This just confirms that unobservable time-related factors do have an impact on carbon emissions and CKC.Thirdly,under the premise of spatial heterogeneity,economic effects and time-related effects and independent variables of residual effects of residuals are established in the GAMM model,and five hypothetical models are ed.Combined with the empirical results of the previous paper,spatial heterogeneity and temporal heterogeneity do have an impact on the CKC fitting.Therefore,considering the spatial heterogeneity,according to the comprehensive standards of economic development level and geographical location,China is divided into three regions of East,Central and West,and analyzed by provinces.Considering the temporal heterogeneity,the unobservable time-related effects are set together with the economic effects as independent variables in the model,and the five hypothesis comparison models are established in combination with spatial heterogeneity.In addition,in order to solve the unique correlation and randomness of panel data,ARMA is used to extract the residual term structure and set it as a random effect in the model.In summary,it is required that the model can contain both the fixed effect of the smooth term and the random effect of the residual term,so the GAMMs model that can satisfy the condition at the same time is selected.Fourthly,select the optimal model for each region from the five hypothetical models by using the AIC criterion.And analyze how economic effects and time-related effects affect carbon emissions.Using the AIC criterion,the optimal model of the three regions of East,Central and West is selected and the smoothing term analysis is performed.It is found that the optimal model selection of the three regions of East,West and West has the spatial effects of spatial heterogeneity and space homogeneity.That is,the impact of economic development level on carbon emissions varies among regions,but the unobservable time-related effects have the same effect on carbon emissions in various regions.From the perspective of carbon emission peaks,although there is no "inverted U-type" CKC from the perspective of the country and most inter-provincial regions,the carbon emissions of individual provinces have shown a trend.The growth rate of carbon emissions in other regions has also shown a slowdown.