基于面板门槛模型的中国雾霾(PM_(2.5))库兹涅茨曲线研究

A Haze (PM_(2.5)) Pollution Study of Kuznets Curve in China Based on Panel Threshold Model

将人均GDP和第二产业增加值占比作为门槛变量,利用门槛模型对我国232个城市分为四组进行研究,在构建的雾霾(PM_(2.5))污染EKC模型中加入产业结构、政府投入、技术投入、环境规制和绿地面积对我国雾霾(PM_(2.5))EKC本土化特征进行实证研究,发现不同的城市组别具有不同的EKC特征,研究结论如下:有12个城市的雾霾环境库兹涅茨曲线呈现倒U型,有38个城市呈现微弱正U型曲线,有36个城市呈现单调递增线性关系,有164个城市出现正U型曲线。还发现在研究样本的232个城市中,大部分城市的雾霾浓度随着经济增长而增长。其他影响因素中,环境规制和政府财政投入会降低雾霾浓度,而第二产业比例则显著影响着雾霾浓度的升高。

With the rapid urbanization and industrialization boosted in China,the problem of air pollution is becoming increasingly serious.The frequent haze pollution has become a common situation.According to global pollution data published by World Health Organization（WHO）,there are 30cities in China among the top 100cities of average PM（2.5）concentrations list.Therefore,reducing concentration of PM（2.5）and maintaining economic growth in a reasonable range have already reached a consensus by the Chinese people.As the environmental Kuznets hypothesis is a classical and persuasive interpretation of the relationship describing environmental change and economic growth,many scholars at domestic and overseas use the environmental Kuznets curve analysis framework to test and verify the impact of economic growth on atmospheric environmental pollution.The relationship between the environment and income per capita of each country may follow the similar pattern of general evolution.However,considering the differences in levels of economic development,energy structure as well as the proportion of the second industry in different countries,the inflection point value and the shape of the EKC would show strong national characteristics.As China has vast land area,cities in different economic initial conditions,economic development level,resource factor endowment and climate conditions,EKC of each country or district show different shapes.This indicates different relationship between economic growth and haze pollution.Reasonable evaluation of the relationship between economic growth and haze pollution may improve the accuracy and validity of government decision.Therefore,the paper divides the 232cities in China into four groups automatically by the threshold variable of GDP per capita and the proportion of secondary industry through the threshold model.In this paper,we add industrial structure,government investment and technology investment into the haze（PM（2.5））pollution EKC model to analyze the localization characteristics of haze EKC in China.It is found that there are different EKC characteristics in different groups.The conclusions are as follows：There are 12cities showing the inverse U curve,38cities exhibiting irregular U-shaped curves,36cities demonstrating monotonically increasing linear relationship,and 164cities showing positive U-shaped curves.It was also found that in 232cities,the haze concentration of most cities grew with economic growth.Among other factors,environmental regulation and government investment will reduce haze concentration,while the industrial structure will significantly intensify haze concentration.Researches on EKC of haze pollution which extends to the inner regions of the country level are scarce,the shortage of related literature will affect the systemic and integrity of haze pollution governance.Environmental Kuznets Curve on national level implies the assumption of homogeneity,which assume the influencing factors of the environment among the Chinese cities show the same direction.This assumption denied the differences existing in the geographical location,industrial structure,technology development,market maturity,infrastructure investment,resources endowment,and climatic conditions in different Chinese cities.But examining Kuznets curve of haze pollution in China leads to estimation errors by according to the same standard.Therefore,it is very important to examine＂heterogeneity＂of haze（PM（2.5））Kuznets curve among the Chinese cites.In order to exam haze（PM（2.5））Kuznets curve accurately,this paper develops two perspectives in terms of research methods and choice of indicators：Firstly,on the research methods,we use threshold effect model to divide research observers effectively by identifying endogenous characteristics of the data automatically,which overcome the defect of weak convergence of group,the arbitrary in setting grouping criteria,and the explicit explanation by using general linear model.Therefore,we use Hansen’s panel threshold model to divide research observer into different sample intervals,determine the threshold value and mark off the group of strong convergence.The writer divides 232cities in China into four different groups to examine the characteristics of EKC curve in China,and uses fixed effect model,the random effects model as well as mixed OLS model to exam respectively.Secondly,in terms of index selection,as it is inevitable to discharge a variety of pollutants by fossil energy consumption in the process of economic growth,the structure of the pollutants in the air is keeping change with the progress of technology.Arrow et al.（1995）argue that pollutants structure in the air are constantly change if income per capita continues to grow.In the Post-industrial era,in addition to the influence of the traditional pollutants,new technology gradually brings new type of pollution into environment.Compared with traditional air pollutants,PM（2.5）emissions can be split into primary emission source and secondary source.The contribution of quadratic transformation of PM（2.5）concentration occupies 50%,which could represent compound and complex integrated information of air pollution,and directly reflects the relationship between economic growth and haze pollution.Currently,the literature on traditional atmospheric pollution emissions is relatively abundant,the research on PM（2.5） Kuznets curve still devoid.Thus,research on the new type of pollutant in the air pollutant effect on EKC should not be ignored.Therefore,the paper applies the data from Columbia University center for social and economic data and application,based on the aerosol optical depth（AOD）released the satellite monitoring data,we accurately attained the PM（2.5）concentration data of 232cities in China's Mainland during 1998-2012.We take per capita GDP and the added value of secondary industry as threshold variables,and use threshold effect model to divide 232cities in China into four groups automatically,and add the proportion of secondary industry,the government investment,technology investment,environmental regulation and green land area of our country into haze pollution EKC model to exam＂localization＂of EKC by empirical analysis,aiming at exam the diversity and convergence characteristics of EKC.