effect of energy development and technological innovation...

Research ArticleEffect of Energy Development and Technological Innovation onPM25 in China A Spatial Durbin Econometric Analysis

Xiaohong Liu

Doctor of Management Environmental Economics Business College of Nanjing Xiaozhuang University No 3601Hongjing Road Jiangning District Nanjing Jiangsu 211171 China

Correspondence should be addressed to Xiaohong Liu amylxhong163com

Received 18 September 2018 Revised 17 November 2018 Accepted 29 November 2018 Published 16 December 2018

Academic Editor Alicia Cordero

Copyright copy 2018 Xiaohong LiuThis is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Using the panel data of 29 provinces and regions in China this paper analyzes the temporal and spatial characteristics of PM25concentration The spatial Durbin model is used to empirically examine the impact of energy development and technologicalinnovation on PM25 The results show that the interprovincial PM25 in China showed an inverted ldquoNrdquo trend There is a spatialagglomeration of interprovincial PM25 and the spatial autocorrelation is significant The spatial autoregression coefficients ofadjacent weights and distance weights are significantly positive and PM25 pollution in adjacent areas has a positive effect onthe region From the view of energy development the increase in energy intensity will significantly increase the concentrationof PM25 in adjacent areas which will increase the global PM25 concentrationThe increase in the proportion of coal consumptionincreases the concentration of PM25 in the region and globally The increase in energy consumption prices will reduce the PM25concentration in the region and globally From the view of technological innovation the increase in sales revenue of new productsof industrial enterprises above the scale will reduce the concentration of PM25 in the region by 01285 percentage points and theconcentration of PM25 in adjacent areas by 03058 percentage points which will reduce the global PM25 concentration by 04343percentage points The increase in the number of granted patent applications will reduce the PM25 concentration of the adjacentregions and globally The increasing level of infrastructure for technological innovation in the region will reduce the concentrationof PM25 in adjacent areas The increase in technical market turnover will reduce the concentration of PM25 in the region and theconcentration of PM25 in adjacent areas which will reduce the global PM25 concentration In order to reduce the concentrationof PM25 in China we should continue to promote PM25 regional collaborative governance In terms of energy development it isnecessary to reduce energy intensity optimize energy structure implement energy substitution and rationalize energy prices Interms of technological innovation it is essential to increase sales revenue of new products of industrial enterprises raise number ofgranted patent applications vigorously improve level of infrastructure for technological innovation and enhance technical marketturnover

1 Introduction

In the first quarter of 2013 China experienced severe hazyweather that affected 13 million square kilometers and800 million people Since then this haze has continued toshroudmost parts of China Energy consumption directly orindirectly increases the degree of haze pollution How doesenergy development affect haze pollution At the same timeChina is speeding up the construction of an innovative coun-try Studying energy development and technology innovationon haze pollution have important implications for building abeautiful China

The main components of haze pollution are PM25 andPM10 Particles with an aerodynamic equivalent diameter of10120583m or less are called PM10 Particles with diameters lessthan 25120583m are called PM25 Energy development includesenergy intensity energy structure and energy prices Theliterature has studied the impact of energy intensity onhaze pollution For example Cheng et al (2017) found thatChinarsquos energy intensity has a significant positive impacton PM25 and should limit the rapid growth of energyintensity [1] Liu and Jiang (2017) found that a decrease inenergy intensity would reduce PM10 concentrations in theregion [2] Regarding energy structure contamination with

HindawiDiscrete Dynamics in Nature and SocietyVolume 2018 Article ID 2148318 12 pageshttpsdoiorg10115520182148318

2 Discrete Dynamics in Nature and Society

haze Ma and Zhang (2014) asserted that in the long runchanging the energy consumption structure and optimizingthe industrial structure are the keys to controlling haze Inthe short term reducing the use of inferior coal is an effectivemethod [3] Hao and Liu (2015) indicated that to reducehaze pollution in China the energy structure in secondaryindustries should be improved by reducing coal consumptionand increasing the consumption of clean energy such asnatural gas and renewable energy [4] Wei and Ma (2015)pointed out that adjusting energy structure and technologicalprogress are the fundamental means to control haze [5] Li(2016) found that the cause of haze pollution in China was thebackwardness of energy technologies the higher proportionof coal consumption and the greater proportion of heavyindustries [6]Wang X et al (2018) found that clean energyconsumption had negative effects on PM25 concentrationsin China [7]

Technological innovation is the introduction of newproducts services or elements in the production process orservice operations [8] Technological innovation accelerateseconomic growth by increasing productivity and reducesenergy consumption in the production process by increasingenergy efficiency ultimately reducing pollution emissions [9]The literature has investigated the impact of technologicalinnovations on atmospheric pollution For example Popp(2003) used patent data to study the innovation of the UnitedStatesrsquo flue gas desulfurization devices (ldquoscrubbersrdquo) andshowed that technological innovations since 1990 increasedscrubber removal efficiency [10] Shi and Lai (2013) found thatthe diffusion of low-carbon and green technological inno-vations promotes regional economic development but savesenergy through low-carbon dioxide emissions or zero energyemissions [11] Yang and Li (2017) studied the relationshipbetween technological change and carbon emissions basedon Chinarsquos 1997ndash2010 panel data for 30 provinces and foundthat technological progress is the greatest contributor to CO2emission reductions [12] Wang et al (2018) estimated theimpact of Chinese technical knowledge on the intensity ofindustrial air pollutants based on the learning curve andenvironmental learning curve theories and found a causalrelationship between technical knowledge and the intensityof industrial air pollutants [13]

The literature has also investigated the impact of tech-nological innovations on haze For example Chang et al(2017) applied regional optimization methods to analyze theimpact of technological development in Shanghai on PM25pollution and found that decreased fossil energy consump-tion and increased use of clean energy can reduce PM25contamination [14] Xu and Lin (2018) believe that Chinarsquosenergy-saving emission reduction technology continues tolag behind developed countries whichmakes reducing PM25pollution in China is a difficult problem to overcome [15]Zhang et al (2016) suggested that technological innovationshould be strengthened to prevent and control haze pollutioninBeijing [16] Liu (2018) found that technological innovationreduces PM10 at a regional level and indirectly reduces PM10in adjacent regions through knowledge spillover effects [17]

The literature has also discussed the impact of energydevelopment and technological innovation onhaze pollution

but the following deficiencies have been observed Firstlyfrom the perspective of energy development the literature onthe impact of energy structure on haze pollution is abundantbut investigations of the impact of energy development com-binedwith technological innovation on the haze pollution arescarce Secondly the literature on the impact of technologicalinnovation on environmental pollution has been increasingbut the literature on the impact of technological innovationon haze pollution especially on PM25 is relatively scarceThirdly from the perspective of research methods most ofthe literature has analyzed the impact of technological inno-vation on haze pollution from the perspective of ordinarypanels but investigations from the perspective of spatialmeasurement are still lacking Finally there is a dearth ofliterature on the use of technical innovation infrastructureindicators to measure technological innovation which isusually measured by patents or RampD input indicators

Based on the deficiencies in the literature this paperintends to expand on the following aspects First energydevelopment and technological innovation are included inthe same framework to examine the impact of energy devel-opment and technological innovation on PM25 Secondly thespatial Durbin model with an adjacent weight matrix anddistance weightmatrix is used to analyze the impact of energydevelopment and technological innovation on PM25 Finallyindicators such as technological innovation infrastructure areused to refer to technological innovation

2 Methodology and Data

21 Global Spatial Autocorrelation The global spatial auto-correlation can effectively describe the overall situation ofPM25 changes at the interprovincial level and is usuallyanalyzed by the Global Moranrsquos I index This index was firstproposed by Moran (1950) [18] The Global Moranrsquos I indexreflects the similarities between units in each region andadjacent regions through a spatial weight matrix (Anselin1988) [19] Its calculation formula is as follows

1198721199001199031198861198991015840119904 119868 =119899sum119899119894=1 sum119899119895=1119908119894119895 (119909119894 minus 119909) (119909119895 minus 119909)sum119899119894=1 sum119899119895=1119908119894119895sum119899119894=1 (119909119894 minus 119909)

2

=sum119899119894=1 sum119899119895=1119908119894119895 (119909119894 minus 119909) (119909119895 minus 119909)

1198782sum119899119894=1 sum119899119895=1119908119894119895

(1)

where 1198782 = (1119899)sum119899119894=1(119909119894minus119909)2 is the variance of the obser-vations 119909 = (1119899)sum119899119894=1 119909119894 is the average of all spatial unitobservations xi and xj represent the ith and jth spatial unitobservations respectively and n is the total number of spatialunits wij is a spatial weight matrix The range of the GlobalMoranrsquos I Index is [-1 1] Igt0 indicates a positive spatialautocorrelation of interprovincial PM25 in China The largerthe value is the stronger the agglomeration is that is the highvalue zone is adjacent to the high value zone and the lowvalue zone is adjacent to the low value zone Ilt0 indicatesa negative autocorrelation of the interprovincial PM25 inChina that is the high value zone is adjacent to the low

Discrete Dynamics in Nature and Society 3

value zone I=0 indicates no spatial autocorrelation in theprovincial PM25 in China

The significance was tested using a standardized statisticZ with the following formula

119885 = 119868 minus 119864 (119868)radic119881119860119877 (119868) (2)

In (2) E(I) and VAR(I) are the expected values andvariances of Moranrsquos I index respectively At a significantlevel of 5 if |119885| gt 196 there is a significant spatial auto-correlation

Anselin (1996 2002) [20 21] believes that the GlobalMoranrsquos I index scatter plot can show the spatial autocor-relation The first and third quadrants illustrate the positivespatial correlation of the observations The second andfourth quadrants describe the negative spatial correlation ofobservations

22 Spatial Regress Models The spatial panel model mainlyincludes spatial lag model SLM spatial error model SEMand spatial Durbin model SDM (Anselin et al 2008) [22]According to Elhorst (2012) the spatial lag model SLMformula is as follows [23]

119910119894119905 = 120575119873

sum119895=1

119908119894119895119910119895119905 + 120572 + 119909119894119905120573 + 120583119894 + 120582119905 + 120576119894119905

119894 = 1 119873 119905 = 1 119879(3)

where 119910119894119905 = (y1119905 y2119905 y3119905 y119873119905)119879 it is an Ntimes1 dimen-sional vector composed of explanatory variables sum119873119895=1119908119894119895119910119895119905is the interaction effect between the interpreted variable yitand the adjacent elements yjt andwij are an NtimesNnonnegativespatial weight matrix 120575 is the response parameter of theendogenous interaction effect 120572 is a constant item 119909119894119905 isan N times K exogenous explanatory variable matrix 120573 is thematching response parameter it is an independent identicaldistribution error term subject to (0 1205902) distribution 120583irepresents the spatial effect 120582t is the time fixed effect

This paper uses two types of spatial weight matrices Oneis the traditional binary spatial adjacent matrix hereinafterreferred to as the adjacent matrix

119908119894119895 =

1 119891119900119903 119894 = 119895 119886119899119889 119899119890119894119892ℎ1198871199001199031198941198991198920 119891119900119903 119894 = 119895 119886119899119889 119899119900119905 - 119899119890119894119892ℎ1198871199001199031198941198991198920 119891119900119903 119894 = 119895

(4)

The other is the distance weight dij is the distancebetween the geographical centers of the two regions

119908119894119895 =

11198891198941198952

119894 = 119895

0 119894 = 119895(5)

where d is the distance between the geographical centerof the two regions

The spatial error model SEM which describes the spatialdisturbance correlation and spatial overall correlation isgiven by

119910119894119905 = 120572 + 119909119894119905120573 + 120583119894 + 120582119905 + 120601119894119905

120601119894119905 = 120588119873

sum119895=1

119908119894119895120601119894119905 + 120576119894119905(6)

where 120601119894119905 is the spatial autocorrelation error and 120588 isthe spatial error correlation coefficient which estimates thedegree of influence of the error shock of adjacent units withregard to the dependent variable on the observation value ofthe unit The other parameters are the same as defined above

The spatial Durbin model SDM which includes thespatial lag value of the explanatory variables is given by

119910119894119905 = 120575119873

sum119895=1

119908119894119895119910119895119905 + 120572 + 119909119894119905120573 +119873

sum119895=1

119908119894119895119909119894119895119905120579 + 120583119894 + 120582119905 + 120576119894119905 (7)

where both 120579 and 120573 are Ktimes1 dimensional parametervectorsTheother parameter implication is the same as aboveDue to the introduction of the spatial weight matrix thespatial econometricmay produce endogenous variables If theordinary least squares estimation is still used in the spatialeconometric model a bias factor will be produced thus themaximum likelihood method is used for estimation

23 Model Specification To analyze the impact of energydevelopment technological innovation and other factors oninterprovincial PM25 in China the following model wasconstructed

ln (11987511987225119894119905) = 120572 + 1205731 ln (119864119894119894119905) + 1205732 ln (119862119904119894119905)

+ 1205733 ln (119875119901119894119894119905) + 1205734 ln (119873119901119894119894119905)

+ 1205735 (ln119875119886119905119890119899119905) + 1205736 ln (119868119899119891119903119886119894119905)

+ 1205737 ln (119872119886119903119896119890119905119894119905) + 120576119894119905

(8)

In formula (8) PM25 is the particulate matter that canenter the lung or fine particulate matter Ei is the energyintensity which is the ratio of the energy consumption ofeach province to the regional GDP Cs is the energy structurewhich is the proportion of coal consumption in energyconsumption Ppi is the price of energy consumption whichis represented by the producer price index of industrial pro-ducer Npi is the sales revenue of new products of industrialenterprises above the scale Patent is the amount of patentapplication authorization Infra is technological innovationinfrastructure that is characterized by internet penetrationrate Market is the turnover of the technology market

24 Data This article examines 29 provincial administra-tive regions in mainland China Due to the lack of dataon Tibet and Chongqing it does not consider Tibet andChongqing The sources of the raw data are the 2004ndash2017China Statistical Yearbook China Energy Statistical Year-book China Science and Technology Statistical Yearbookstatistical yearbooks of various provinces and districts andstatistical bulletins on the national economic and socialdevelopment of various provinces and districts over the yearsChinarsquos monitoring of PM25 started late that is well after


Table 1 Descriptions of all variables in econometric model

Variable Unit Mean Maximum Minimum Std Dev ObservationPM25 120583gm3 375951 15400 2170 225786 406Ei Tce104 Yuan 1194 4520 0270 0705 406Cs 72776 4518000 9810 221860 406Ppi No unit 101991 125250 82410 6185 406Npi 104 Yuan 27244930 287E+08 3855100 44506476 406Patent piece 2425841 2699440 70000 4527393 406Infra 30420 77830 2150 19767 406Market 104 Yuan 1518325 39409752 1885000 4029395 406

Table 2 Correlations analysis and VIF tests

VIF PM25 Ei Cs Ppi Npi Patent Infra MarketPM25 - 1000Ei 37015 -03579lowastlowastlowast 1000Cs 10148 -00440 01386lowastlowastlowast 1000Ppi 14093 -04122lowastlowastlowast 03714lowastlowastlowast 00706 1000Npi 64635 03280lowastlowastlowast -04463lowastlowastlowast -00366 -02206lowastlowastlowast 1000Patent 69443 02866lowastlowastlowast -04200lowastlowastlowast -00364 -02164lowastlowastlowast 09335lowastlowastlowast 1000Infra 56643 04299lowastlowastlowast -05900lowastlowastlowast -00857lowast -05087lowastlowastlowast 05298lowastlowastlowast 05305lowastlowastlowast 1000Market 13815 02702lowastlowastlowast -03164lowastlowastlowast -00488 -01736lowastlowastlowast 02619lowastlowastlowast 03321lowastlowastlowast 04799lowastlowastlowast 1000Notelowast lowast lowast lowast indicates significance at 10 and 1 levels respectively

the problem began Therefore to acquire PM25 data theannual mean value data from 2003 to 2010 was mainly basedon the measurement of global aerosol optical depth fromsatellite-borne equipment operated by a Columbia Universityresearch team and the Battle Research Institute (Donkelaar etal 2010) [24] The data of the population-weighted averageannual PM25 for all provinces of China from 2003ndash2010 arealso derived from these sources PM25 data from 2011ndash2012were calculated based on monitoring data from the ChinaEnvironmental Monitoring Center The years 2013ndash2016 arefrom the China Statistical Yearbook (2014ndash2017) Referringto Ma et al (2016) provincial capital data are used instead ofprovincial data [25] The descriptive statistics of the relevantvariables are shown in Table 1

The correlation coefficient and multicollinearity test ofthe variables are shown in Table 2 It can be observed thatin addition to the correlation coefficient between Patent andNpi being as high as 09335 other correlation coefficients arelower than 06 and some correlation coefficients do not passthe significance test Using the variance inflation factor VIFto test formulticollinearity the results show that aVIF greaterthan 1 is less than 695 and the mean is 3797 Since the VIFvalues are all less than 10 there is no multicollinearity

3 Temporal and Spatial Characteristics ofInterprovincial PM25 in China

31 Temporal Variation Characteristics of the InterprovincialPM25 From the perspective of the temporal characteristicsthe interprovincial PM25 in China showed an inverted ldquoNrdquotrend that is a slow downward trend before 2012 an upwardtrend after 2012 a peak after reaching a peak in 2013 and

000

2000

4000

6000

8000

10000

12000

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

The NortheastThe North CoastThe East CoastThe Middle Yellow RiverThe Middle Yangtze RiverThe South CoastThe NorthwestThe Southwest

Figure 1 Time change trend of PM25 in China

then a downward trend These trends are in line with Chinarsquossituation Chinarsquos haze pollution showed a major increasein 2013 Since then the country has committed to themanagement of haze pollution and haze pollution has showna downward trend

According to the traditional eight regional division meth-ods the interprovincial PM25 in China is divided intoeight regions (Figure 1) The descriptive statistics of PM25


Table 3 Descriptions of PM25

Region Mean Maximum Minimum Std Dev ObservationsNortheast 299595 81000 7530 248257 42North Coast 541728 154000 27700 276504 56East Coast 393197 78000 21510 162013 42Middle Yellow River 406401 108000 11430 247028 56Middle Yangtze River 441114 94000 26870 172208 56South Coast 206838 53000 2170 129610 42Northwest 290419 88000 12530 208145 56Southwest 371357 96000 19360 140458 56

No data

N

8560000 - 26580000

26580001 - 38830000

38830001 - 50380000

50380001 - 640000000 500 km

Figure 2 Spatial distribution of PM25 average value in China

concentrations in the eight regions are shown in Table 3The average PM25 concentrations in the northern coastand middle Yangtze River are higher than the nationalaverage Except for 2009 the average PM25 concentrationin the middle Yellow River is higher than that of the wholecountry The mean value of PM25 in the southern coastalareas is lower than the national average As presented inTable 3 the average annual PM25 concentration in China is375951120583gm3 In the eight regions the mean concentrationof PM25 in the northern coast is 541728 ranking first Theaverage concentration of PM25 in the middle Yangtze Riverand middle Yellow River was higher than 40120583gm3 rankingsecond and third respectively The average concentration ofPM25 in the eastern coast ranks fourth The mean PM25

concentrations in these four regions are higher than thenational average The average concentration of PM25 inthe southern coastal areas is the lowest from low to highfollowed by the northwest northeast and southwest regionsin that order The PM25 concentrations in these four regionsare lower than the national average

32 Spatial Differentiation Characteristics ofthe Interprovincial PM25

321 Spatial Distribution Pattern of the Interprovincial PM25Using ArcGIS102 software according to the quartile methodthe distribution map of PM25 in China from 2003 to 2016 isplotted (Figure 2)


27

16

05

minus06

minus17

minus28

271605minus06minus17minus28

lagg

ed P

M2_

5

PM2_5

Moranrsquos I 0420279

Figure 3 Moranrsquos I index scatter plot of PM25 average in China

From low to high the interprovincial PM25 average isdivided into four levels The first-level provinces are HainanFujian Inner Mongolia Ningxia Heilongjiang and QinghaiThe second-tier provinces are Guizhou Jilin Gansu YunnanGuangdong Zhejiang Xinjiang Shanghai Liaoning Jiangxiand Shanxi The third-level provinces are Guangxi ShaanxiHunan Tianjin Beijing Anhui Hubei Sichuan and JiangsuThe fourth level is Henan Shandong and Hebei It can beobserved that the mean values of PM25 in the northerncoast and middle Yellow River are relatively high and thatof Guizhou Yunnan and the northeast in the south coastand southwest regions are relatively low The interprovincialPM25 in China shows a trend of high and low in the east andthere is a spatial agglomeration

322 Global Spatial Autocorrelation Analysis of the Inter-provincial PM25 By using GeoDA95 software and rookneighboring Moranrsquos I index of the interprovincial PM25average value in China from 2003 to 2016 is 04203 After 999permutations the P value is 00020 and the normal statisticZ value is 36473 which is larger than the critical value 196 ofthe normal distribution function at the 005 significant leveland indicates that the interprovincial PM25 average valuespatial autocorrelation is significant

Moranrsquos I index scatter plot is used to further examine thestate of spatial agglomeration (Figure 3) The first quadrantis the HndashH agglomeration Ten provinces and autonomousregions namely Hebei Shaanxi Shandong Henan JiangsuBeijing Tianjin Anhui Hubei and Shanxi are in this quad-rant account for 3448 of all the investigated provincesand are mainly located in the north coast and middle YellowRiver The third quadrant is the LndashL agglomeration Tenprovinces and autonomous regions namely Inner MongoliaFujian Guangdong Zhejiang Shanghai Gansu Ningxia

Heilongjiang Jilin and Xinjiang are in this quadranttogether with Hainan which spans the second and thirdquadrants and accounts for 3793 of all the provincesThese provinces are mainly located in the southern coastthe eastern coast and the northeast The second quadrant isLndashH with five provinces and regions in Jiangxi LiaoningYunnan Guizhou and Qinghai and Hainan The fourthquadrant is an agglomeration of HndashL and Sichuan Hunanand Guangxi are located in this quadrant In general theproportion of provinces with the same spatial autocorrelationin theHndashHandLndashL quadrants is 7241 In the LndashHquadrantand the HndashL quadrant the provinces with different spatialautocorrelations accounted for only 3103 Based on theseresults it can be observed that the interprovincial PM25regional integration in China is significant

4 Empirical Research

Based on the spatial agglomeration of PM25 the spatialeconometric model is used to analyze the impact of energydevelopment and technological innovation on PM25

41 Spatial Diagnostic Test In Section 2 as aforementionedthere are generally three types of spatial panel models Todetermine which model to use a nonspatial interactionmodel is used for spatial diagnostic tests based on the modelselection mechanism proposed by Anselin (2005) [26] Theresults are shown in Table 4 For the adjacent weights theLM-lag and LM-error tests show that both the hypothesis ofno spatially lagged dependent variable and the hypothesis ofno spatially auto-correlated error termmust be rejected at 1significance The time fixed effect did not pass the robust LM-lag test The spatial fixed effect did not pass the significancetest of robust LM-error According to Elhorst (2014) [27] thespatial Durbin panel model was chosen The measurementmethod is the maximum likelihood estimation proposed byElhorst (2003) [28]

For the distance weight the LM-lag and LM-errortests show that both the hypothesis of no spatially laggeddependent variable and the hypothesis of no spatially auto-correlated error termmust be rejected at 1 significance Therobust LM-lag test shows that the pooled OLS and time fixedeffects do not pass the significance test Similar to the robustLM-error test only the pooled OLS passes the significancetest thus the spatial Durbin panel model is chosen in thesame way as the adjacent weights The LR test is used toselect the fixed effects The results indicate that at the 1level of significance the null hypothesis that the spatial fixedeffects are jointly insignificant must be rejected Similarly thenull hypothesis that the time-period fixed effects are jointlyinsignificant must be rejected According to Baltagi (2005)[29] the two-way fixed effects model is selected which is alsoknown as the two-way fixed effects model

42 Empirical Analysis The results of the spatial Durbinmodel with two-way fixed effects are shown in Table 5 TheWald and LR tests of the two weights show that at the 1level of significance the null hypothesis of the spatial Durbin


Table 4 Estimation results of nonspatial panel model

Variables Pooled OLS Spatial fixed effects Time-period fixed effects two-way fixed effectsAdjacent weight

Intercept 115064lowastlowastlowast (00000) mdash mdash mdashlnEi 03473lowastlowastlowast (00001) -11951lowastlowastlowast (00000) 05160lowastlowastlowast (00000) 03958lowastlowastlowast (00078)lnCs 02146lowastlowastlowast (00037) 01257lowast (00722) 00781(01749) 00851lowastlowast (00395)lnPpi -25530lowastlowastlowast (00000) -23159lowastlowastlowast (00000) -11728lowast (00781) -12583lowastlowastlowast (00009)lnNpi -00179(06843) -01685lowastlowastlowast (00009) 01261lowastlowastlowast (00006) -01664lowastlowastlowast (00000)lnPatent 02509lowastlowastlowast (00000) 02093lowastlowastlowast (00015) 01263lowastlowastlowast (00024) -00204(06597)ln Infra -00352(04382) -02274lowastlowastlowast (00007) -03931lowastlowastlowast (00000) 00364(05603)lnMarket 00776lowastlowastlowast (00017) -00598(01106) 01053lowastlowastlowast (00000) -00565lowastlowast (00112)R2 04091 04705 04592 01661LM-lag 2017888lowastlowastlowast (0000) 3400601lowastlowastlowast (0000) 704194lowastlowastlowast (0000) 666052lowastlowastlowast (00000)Robust LM-lag 46368lowastlowast (0031) 369528lowastlowastlowast (0000) 12878(0256) 607071lowastlowastlowast (00000)LM-error 2198807lowastlowastlowast (0000) 3031079lowastlowastlowast (0000) 822136lowastlowastlowast (0000) 393528lowastlowastlowast (00000)Robust LM-error 227288lowastlowastlowast (0000) 00006(0980) 130820lowastlowastlowast (0000) 334547lowastlowastlowast (00000)

Distance weightIntercept 115064lowastlowastlowast (00000) mdash mdash mdashlnEi 03473lowastlowastlowast (00001) -11951lowastlowastlowast (00000) 05160lowastlowastlowast (00000) 03958lowastlowastlowast (00078)lnCs 02146lowastlowastlowast (00037) 01257lowast (00722) 00781lowastlowastlowast (01749) 00851lowastlowast (00395)lnPpi -25530lowastlowastlowast (00000) -23159lowastlowastlowast (00000) -11728lowastlowastlowast (00000) -12583lowastlowastlowast (00009)lnNpi -00179(06843) -01685lowastlowastlowast (00009) 01261lowastlowastlowast (00006) -01664lowastlowastlowast (00000)lnPatent 02509lowastlowastlowast (00000) 02093lowastlowastlowast (00015) 01263lowastlowastlowast (00024) -00204(06597)ln Infra -00352(04382) -02274lowastlowastlowast (00007) -03931lowastlowastlowast (00000) 00364(05603)lnMarket 00776lowastlowastlowast (00017) -00598(01106) 01053lowastlowastlowast (00000) -00565lowastlowast (00112)R2 04091 04705 04592 01661LM-lag 1022239lowastlowastlowast (0000) 3231789lowastlowastlowast (0000) 136513lowastlowastlowast (0000) 358038lowastlowastlowast (0000)Robust LM-lag 06919(0406) 462787lowastlowastlowast (00000) 11820(0277) 104315lowastlowastlowast (0001)LM-error 1366806lowastlowastlowast (00000) 2769055lowastlowastlowast (00000) 129372lowastlowastlowast (00000) 276293lowastlowastlowast (0000)Robust LM-error 351486lowastlowastlowast (00000) 00053(0942) 04678(00494) 22570(0133)

Fixed effects Statistics DOF P valueLR-test Spatial fixed effects 5385783 29 00000

Time-period fixed effects 4417073 14 00000Notelowast lowastlowast lowast lowast lowast indicates significance at 10 5 and 1 levels respectively

model SDM into the spatial lag model SLM is rejected andthe spatial Durbin model SDM is also rejected as the spatialerror model SEM therefore a spatial Durbin model SDM isusedThe estimated value of the Hausman test of the adjacentweight is 956264 the degree of freedom is 15 and the valueof P is 0Therefore the random effect is rejected and the two-way fixed effect model is used Elhorst (2014) indicated thatthe coefficients of the spatially lagged dependent variable andindependent variables are quite sensitive to bias correctionsTherefore in the following analysis of adjacent weights thetwo-way fixed effect model with bias correction is selected(the third column of data in Table 4) The estimated value ofthe distance weight Hausman test is 600659 the degree offreedom is 15 and the value of P is 0Therefore themodel alsorejects the random effect and selects the bi-directional fixedeffectmodelThat is distance weights are used to describe thethird column of the second half of Table 5

Comparing the estimation results in Tables 5 and 4 itcan be observed that the R2 of the noninteractive effect

model is smaller whether it is the adjacent weight or thedistance weight The noninteractive effect of an adjacentweight and distance weight two-way fixed model R2 areboth 01661 The two-way fixed effect of the spatial Durbinmodel R2 is 09398 and the distance weight error correctiontwo-way fixed effect Durbin model R2 is 09297 This paperillustrates that the use of a spatialDurbinmodel to analyze theimpact of energy development and technical innovation onPM25 is reasonableThe spatial auto-regression coefficients ofadjacent weights and distance weights are 04268 and 05168respectively all pass the 1 level significance test indicatingthat the PM25 of adjacent regions have a positive impact onlocal region A decrease of one percentage point in PM25 inadjacent regions results in a decrease in PM25 in this areaby 04268 05168 percentage points respectively PM25 hasa spatial spillover effect This is consistent with the robustLM test of the spatial autocorrelation of Table 4 Additionallythe auto-regression coefficient of the distance weights issignificantly larger than the adjacent weight because the


Table 5 Estimation results of spatial Durbin model

Variables Two-way fixed effects Two-way fixed effects (with bias correction) Random spatial effects and time-period fixed effectsAdjacent weight

120575 03679lowastlowastlowast (00000) 04268lowastlowastlowast (00000) 04329lowastlowastlowast (00000)lnEi 01058(04343) 00980(04896) 00411(07629)lnCs 00899lowastlowastlowast (00089) 00880lowastlowast (00147) 00882lowastlowast (00143)lnPpi -10526lowastlowastlowast (00018) -10389lowastlowastlowast (00003) -10866lowastlowastlowast (00021)lnNpi -01079lowastlowastlowast (00000) -01043lowastlowastlowast (00002) -00893lowastlowastlowast (00015)lnPatent 00626(01363) 00657(01356) 00905lowastlowast (00310)ln Infra 00410(04668) 00454(04419) 00467(04255)lnMarket -00452lowastlowast (00170) -00439lowastlowast (00000) -00471lowastlowast (00161)Wlowast ln Ei 10147lowastlowastlowast (00004) 09882lowastlowastlowast (00010) 04913lowast (00676)Wlowast lnCs 00696(02970) 00608(03844) 00434(05334)Wlowast ln Ppi -04512(04771) -03386(06104) -05460(04074)Wlowast lnNpi -01558lowastlowastlowast (00076) -01414lowastlowast (00206) -00786(01850)Wlowast ln Patent -03095lowastlowastlowast (00000) -03052lowast (00000) -02093lowastlowastlowast (00016)Wlowast ln Infra -02437lowastlowast (00474) -02467lowast (00557) -02543lowastlowast (00463)Wlowast lnMarket -00869lowastlowast (00252) -00802lowastlowast (00486) -00864lowastlowast (00312)Teta mdash mdash 00701lowastlowastlowast (00000)R2 09391 09398 08902log-likelihood 1414070 1414070 -8405862Wald spatial lag 718978lowastlowastlowast (0000) 615451lowastlowastlowast (0000) 461114lowastlowastlowast (0000)LR spatial lag 677774lowastlowastlowast (0000) 677774lowastlowastlowast (0000) mdashWald spatial error 876175lowastlowastlowast (0000) 753315lowastlowastlowast (0000) 545307lowastlowastlowast (0000)LR spatial error 816844lowastlowastlowast (0000) 816844lowastlowastlowast (0000) mdash

Hausman test Statistics DOF P value956264 15 00000

Distance weightΔ 04449lowastlowastlowast (00000) 05168lowastlowastlowast (00000) 04399lowastlowastlowast (00000)lnEi 04862lowastlowastlowast (00000) 04867lowastlowastlowast (00011) 03066lowastlowast (00245)lnCs 00752lowastlowast (00421) 00737lowast (00570) 00795lowastlowast (00426)lnPpi -08146lowastlowast (00282) -07859lowastlowast (00431) -11188lowastlowastlowast (00030)lnNpi -01472lowastlowastlowast (00000) -01441lowastlowastlowast (00000) -01198lowastlowastlowast (00000)lnPatent 00152(07189) 00160(07184) 00646(01287)ln Infra 00163(07785) 00203(07379) 00219(07171)lnMarket -00591lowastlowastlowast (00006) -00579lowastlowastlowast (00057) -00492lowastlowast (00187)Wlowast ln Ei -04203(02145) -04437(02106) -05212lowast (00864)Wlowast lnCs 03411lowast (00764) 03277(01039) 03314(01030)Wlowast ln Ppi -10727(01067) -09562(01689) -00258(09525)Wlowast lnNpi -02430lowastlowastlowast (00013) -02238lowastlowastlowast (00046) -01340lowast (00685)Wlowast ln Patent -02628lowastlowast (00210) -02619lowastlowast (00280) -00715(05086)Wlowast ln Infra -02226(01075) -02215(01261) -00901(05263)Wlowast lnMarket -00024(09589) 00021(09653) -00085(08572)teta mdash mdash 00782lowastlowastlowast (00000)R2 09287 09297 08679log-likelihood 1054847 1054847 -6497897Wald spatial lag 243571lowastlowastlowast (00000) 205623lowastlowastlowast (00045) 97989(02003)LR spatial lag 232803lowastlowastlowast (00015) 232803lowastlowastlowast (00015) mdashWald spatial error 332143lowastlowastlowast (00000) 284091lowastlowastlowast (00000) 147569lowastlowast (00392)LR spatial error 302017lowastlowastlowast (00000) 302017lowastlowastlowast (00000) mdash


Notelowast lowastlowast lowast lowast lowast indicates significance at 10 5 and 1 levels respectively


Table 6 Decomposition estimates of the direct indirect and total effects

Variables Direct effect Indirect effect Total effectDistance weight

lnEi 02323(01155) 16573lowastlowastlowast (00010) 18896lowastlowastlowast (00006)lnCs 01012lowastlowast (00158) 01616(01815) 02628lowast (00718)lnPpi -11418lowastlowastlowast (00062) -12932(02219) -24350lowast (00507)lnNpi -01285lowastlowastlowast (00003) -03058lowastlowastlowast (00055) -04343lowastlowastlowast (00012)lnPatent 00270(05692) -04464lowastlowastlowast (00006) -04194lowastlowastlowast (00040)ln Infra 00153(07962) -03766lowast (00806) -03613(01181)lnMarket -00560lowastlowast (00183) -01605lowastlowast (00250) -02165lowastlowast (00136)

Adjacent weightlnEi 04738lowastlowastlowast (00059) -03154(06591) 01584(08446)lnCs 01243lowastlowast (00180) 07165lowast (00894) 08408lowast (00663)lnPpi -09929lowastlowast (00159) -26540lowastlowast (00396) -36469lowastlowast (00114)lnNpi -01819lowastlowastlowast (00000) -05782lowastlowastlowast (00011) -07601lowastlowastlowast (00001)lnPatent -00189(06949) -05025lowastlowast (00443) -05214lowast (00559)ln Infra -00126(08562) -04191(01590) -04317(02006)lnMarket -00595lowastlowast (00164) -00442(06603) -01037(03700)Note lowast lowastlowast lowast lowast lowast indicates significance at 10 5 and 1 levels respectively

spillover of PM25 is related to the geographical location Evenwhen not geographically adjacent a certain area affects otherareas Therefore this paper uses two types of weight matrixnamely the adjacent weight and distance weight to analyzethe spatial spillover effect of provincial PM25 in China inorder to reflect the objective reality comprehensively andaccurately

Since the significance of each variable estimation coeffi-cient in the nonspatial model is not the same as in the spatialeconometric model the coefficient in Table 5 cannot be com-pared with the Table 4 LeSage and Pace (2009) believe thatdirect and indirect effects explain the true spatial spillovereffect of each variable Therefore this paper decomposes thedirect and indirect effects of explanatory variablesThe resultsare shown in Table 6 Due to the feedback effect the directeffects of the variables in Table 6 are different than in Table 5Elhorst (2014) believes that the feedback effect is partly dueto the coefficient estimate of the spatially lagged dependentvariable and partly to the coefficient of the spatially laggedvalue of the explanatory variable itself The decompositionresult of the adjacentweight is better than the distanceweightand the coefficient signs are roughly the same Therefore theresults of the adjacent weights are used for analysis

The direct effect of energy intensity is positive but ithas not passed the significance test indicating that thepositive effect of energy intensity on haze pollution in theregion is not obvious The indirect effects and total effectsof energy intensity are significantly positive The increase inenergy intensity will significantly increase the concentrationof PM25 in adjacent areas and subsequently increase theglobal PM25 concentration The direct effects and total effectsof energy structure are significantly positive The increase inthe proportion of coal consumption increases the PM25 con-centration in the region and globally The proportion of coalconsumption increase 1 percentage will increase the PM25

concentration in the region by 01012 percentage points andincrease the global PM25 concentration by 02628 percentagepoints Therefore in order to reduce PM25 concentrationsChina should reduce the proportion of coal consumption

The direct effect and total effect of energy consump-tion price are significantly negative The increase in energyconsumption price decreases the PM25 concentration in theregion and globally The energy consumption price increase1 percentage will reduce the concentration of PM25 in theregion by 11418 percentage points and reduce the globalPM25 concentration by 24350 percentage points It showsthat the rise of energy consumption price can effectivelyreduce the PM25 concentrationThe direct indirect and totaleffects of the sales revenue of new products of industrialenterprises are significantly negative The increase in salesrevenue of new products will reduce the concentration ofPM25 in the region by 01285 percentage points reducethe concentration of PM25 in adjacent regions by 03058percentage points and reduce the global PM25 concentrationby 04343 percentage points It shows that the sales rev-enue of new products of industrial enterprises has achievedremarkable results in reducing PM25 concentration Theindirect effects and total effects of patents are significantlynegative The increase in the number of patent applicationsgranted in the region will reduce the PM25 concentrationof the adjacent regions and globally The indirect effect ofInfra is significantly negative indicating that the increasein infrastructure for technological innovation in the regionwill reduce the concentration of PM25 in adjacent areasThe direct indirect and total effects of the turnover of thetechnology market are significantly negative The increase inthe turnover of the technology market will not only reducethe concentration of PM25 in the region but also the concen-tration of PM25 in the adjacent region through the spillovereffect significantly reducing Global PM25 concentration


Therefore in order to reduce PM25 concentration it isnecessary to increase the number of patent applicationsgranted expand technological innovation infrastructure andincrease turnover of the technology market in China

5 Conclusion and Policy Implications

This paper uses the panel data of 29 provinces and regions inChina (excluding Tibet and Chongqing) to analyze the spatialand temporal characteristics of PM25 The spatial Durbinmodel is used to empirically examine the impact of energydevelopment and technological innovations on PM25 Thefollowing are the main findings of this research

From the perspective of temporal characteristics theinterprovincial PM25 in China shows an inverted ldquoNrdquo trendThe average concentrations of PM25 in the northern coastmiddle Yangtze River middle Yellow River and eastern coastare higher than the national average The average concentra-tions of PM25 in the south coast northwest northeast andsouthwest regions are lower than the national average Theinterprovincial PM25 in China is high in the east and low inthe west and a spatial agglomeration is observed The PM25has a significant autocorrelation and the regional integrationtrend is significant The spatial autoregression coefficientsof adjacent weights and distance weights are significantlypositive and haze pollution in adjacent regions has a positiveimpact on local region There is a spatial spillover effectof interprovincial PM25 in China The indirect effects andtotal effects of Energy intensity are significantly positive Theincrease in energy intensity will significantly increase theconcentration of PM25 in adjacent areas and subsequentlyincrease the global PM25 concentration The direct effectsand total effects of Energy structure are significantly positiveThe increase in the proportion of coal consumption increasesthe PM25 concentrations in the region and globally Thedirect effect and total effect of energy consumption priceare significantly negative The increase in energy consump-tion price reduces the PM25 concentration in the regionand globally The direct indirect and total effects of thesales revenue of new products of industrial enterprises aresignificantly negative The increase in sales revenue of newproducts in local region will reduce the concentration ofPM25 in the region by 01285 percentage points and thatin adjacent regions by 03058 percentage points which willreduce the global PM25 concentration by 04343 percentagepoints The indirect effects and total effects of patents aresignificantly negative The increase in the number of patentapplications granted in the region will reduce the PM25concentrations of the adjacent regions and globally Theindirect effect of Infra is significantly negative indicating thatthe increase in infrastructure for technological innovationin the region will reduce the concentration of PM25 in theadjacent areas The direct indirect and total effects of theturnover of the technology market are significantly negativeThe increase in the turnover of the technology market willnot only reduce the concentration of PM25 in the regionbut also reduce the concentration of PM25 in the adjacentregion through the spillover effect significantly reducingglobal PM25 concentration

Based on the main findings of this study this paperproposes the following recommendations for China Firstcontinue to promote the regional collaborative managementof PM25 Based on the spatial spillover effect of PM25 thecontrol of PM25 requires joint prevention [7] After theoutbreak of smog pollution the Chinese government imple-mented strong measures to control haze pollution includingjoint prevention and control of the BeijingndashTianjinndashHebeiregionThe inverted ldquoNrdquo curve of Chinarsquos PM25 indicates thatthese governance measures are effective Additionally con-tinue to promote regional collaborative governance and forma joint force to control haze pollution The concentrationsof PM25 in the southern coastal and southwest regions arerelatively low and these regions can work together to consol-idate existing achievements and present these positive effectsthroughout the country The concentrations of PM25 in thenorthern coastal areas and the middle Yellow River are rel-atively high The provinces and regions within these regionsmust negotiate cogovern work together share resources andinformation and reduce their respective concentrations ofPM25

Secondly reduce energy intensity [30] An increase inenergy intensity will increase the concentration of PM25thus it is necessary to increase energy efficiency and reduceenergy intensity through technological innovations that isdirectly by reducing the concentration of PM25 and indirectlyby improving the level of technological innovation to reducePM25 pollution Taxation and other economic means willencourage enterprises to carry out technological transfor-mations to improve their technological level and reduceenergy intensity research and develop clean technologiesimplement clean production processes and eradicate thegeneration and emission of PM25 from the source

Again optimize energy structure and implement energysubstitution An increase in the proportion of coal con-sumption increases the concentration of PM25 in the regionand globally Therefore China should reduce its proportionof coal consumption and optimize its energy structureAlthough the rapid reduction of Chinarsquos coal consumptionis a difficult task in the short term it is possible to managethe pollution control of bulk coal by formulating coal qualitystandards using gasification and purification methods togenerate electricity and employing liquid fuels to utilize coalin a cleaner manner [31] Additionally as soon as possibleachieve coal-fired power generation on behalf of coal whileimproving coal quality reducing the use and supply of coaloptimizing its energy consumption structure increasing theratio of clean energy and renewable energy implementingenergy alternatives supporting the development of newenergy and promoting the consumption of new energy

Thirdly increase energy prices The consumption ofenergy is in line with the law of demand that is anincrease in energy prices will reduce energy consumptionand subsequently reduce PM25 concentrations At presentthe price of energy in China is low and this price distortiondoes not reflect the cost of energy This phenomenon is oneof the reasons for the high energy consumption in ChinaTherefore China should completely change the phenomenonof lower energy prices and play a regulatory role in the


marketmechanism the price should reflect the cost of energyuse [32 33] Along with the overall advancement of Chinarsquosresource tax reform China must speed up the implementa-tion of environmental protection tax legislation establish anenvironmental tax system and realize the combined effect ofresource taxes energy taxes and environmental taxes

Finally improve the level of technological innovation Asmentioned the sales revenue of new products of industrialenterprises patent turnover of the technology market andother factors can directly or indirectly reduce the concentra-tion of PM25 and technological innovation is an effectivemethod of reducing PM25 Therefore China must seizethe opportunity to build an innovative country enhancedisruptive technological innovation and improve the level oftechnological innovation China can learn from the literatureon developed countries and consider effective and advancedgovernance technology that can be absorbed and reformedand applied to Chinarsquos PM25 governance [34] Talent isthe basis of technological innovation thus it is possible toincrease technological innovation personnel through incen-tive measures Efficiency improves the level of technologicalinnovation and the capital scale of technological innovationTechnological innovation requires financial support thusthe financing channels for technological innovation must bebroadenedAdditionally expanding the scale of technologicalinnovation capital also promotes the expansion of technolog-ical innovation and production scale

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that they have no conflicts of interest

Acknowledgments

This paper is sponsored by Qing Lan Project in JiangsuProvince (17EYD006) the social science fund project inJiangsu Province the Major Program of National SocialScience Foundation of China [15ZDA053]

References

[1] Z Cheng L Li and J Liu ldquoIdentifying the spatial effects anddriving factors of urban PM25 pollution in Chinardquo EcologicalIndicators vol 82 pp 61ndash75 2017

[2] X H Liu andK Jiang ldquoAnalyzing effect of urbanization on hazepollution based on static and dynamic spatial panel modelrdquoNongye Gongcheng Xuebao vol 33 no 20 pp 218ndash225 2017

[3] LMa andX Zhang ldquoSpatial effects of regional air pollution andthe impact of industrial structurerdquo China Population Resourcesamp Environment vol 7 pp 157ndash164 2014

[4] Y Hao and Y-M Liu ldquoThe influential factors of urban PM25concentrations in China A spatial econometric analysisrdquo Jour-nal of Cleaner Production vol 112 pp 1443ndash1453 2016

[5] W Wei and Ma X ldquoOptimal policy for energy structureadjustment and haze governance in chinardquo China PopulationResources amp Environment vol 25 no 7 pp 6ndash14 2015

[6] M Li ldquoHaze pollution control strategies in China from theperspective of energy conservation and emission reductionrdquoNature Environment and Pollution Technology vol 15 no 3 pp887ndash893 2016

[7] X Wang G Tian D Yang W Zhang D Lu and Z Liu ldquoRe-sponses of PM25 pollution to urbanization in Chinardquo EnergyPolicy vol 123 pp 602ndash610 2018

[8] FDamanpour andWM Evan ldquoOrganizational Innovation andPerformance The Problem of ldquoOrganizational Lagrdquordquo Adminis-trative Science Quarterly vol 29 no 3 p 392 1984

[9] K Sohag R A Begum S M Syed Abdullah and M JaafarldquoDynamics of energy use technological innovation economicgrowth and trade openness in Malaysiardquo Energy vol 90 pp1497ndash1507 2015

[10] D Popp ldquoPollution Control Innovations and the Clean Air Actof 1990rdquo Journal of Policy Analysis andManagement vol 22 no4 pp 641ndash660 2003

[11] Q Shi and X Lai ldquoIdentifying the underpin of green andlow carbon technology innovation research A literature reviewfrom 1994 to 2010rdquo Technological Forecasting amp Social Changevol 80 no 5 pp 839ndash864 2013

[12] L Yang and Z Li ldquoTechnology advance and the carbon dioxideemission in China ndash Empirical research based on the reboundeffectrdquo Energy Policy vol 101 pp 150ndash161 2017

[13] W Wang B Yu X Yao T Niu and C Zhang ldquoCan tech-nological learning significantly reduce industrial air pollutantsintensity in ChinamdashBased on a multi-factor environmentallearning curverdquo Journal of Cleaner Production vol 185 pp 137ndash147 2018

[14] Z Chang K Pan and H Zhu ldquoPower generation systemoptimization with emission co-benefits analysis A case study ofShanghairdquo in Advances in Energy Systems Engineering SpringerInternational Publishing 2017

[15] B Xu and B Lin ldquoWhat cause large regional differences inPM25 pollutions in China Evidence from quantile regressionmodelrdquo Journal of Cleaner Production vol 174 pp 447ndash4612018

[16] H Zhang S Wang J Hao et al ldquoAir pollution and controlaction in Beijingrdquo Journal of Cleaner Production vol 112 pp1519ndash1527 2016

[17] X H Liu ldquoDynamic evolution spatial spillover effect oftechnological innovation and haze pollution in Chinardquo Energyamp Environment vol 229 no 6 pp 968ndash988 2018

[18] P A Moran ldquoNotes on continuous stochastic phenomenardquoBiometrika vol 37 pp 17ndash23 1950

[19] D A Griffith and L Anselin ldquoSpatial econometrics methodsand modelsrdquo Economic Geography vol 65 no 2 p 160 1989

[20] L Anselin ldquoThe Moran scatter plot as an ESDA tool to assesslocal instability in spatial associationrdquo Taylor and Francis pp111ndash125 1996

[21] L Anselin ldquoUnder the hood issues in the specification andinterpretation of spatial regression modelsrdquo Agricultural Eco-nomics vol 27 no 3 pp 247ndash267 2002

[22] L Anselin J Gallo and H Jayet ldquoSpatial panel econometricsrdquoine Econometrics of Panel Data Springer Berlin Heidelberg2008

[23] J P Elhorst ldquoMatlab software for spatial panelsrdquo InternationalRegional Science Review vol 37 no 3 pp 389ndash405 2014


[24] A van Donkelaar R V Martin M Brauer et al ldquoGlobalestimates of ambient fine particulate matter concentrationsfrom satellite-based aerosol optical depth development andapplicationrdquo Environmental Health Perspectives vol 118 no 6pp 847ndash855 2010

[25] L M Ma S L Liu and X Zhang ldquoStudy on haze pollutioninduced by energy structure and transportation based on spa-tial econometric model analysisrdquo Finance amp Trade Economicsvol 37 no 1 pp 147ndash160 2016

[26] L Anselin ldquoExploring spatial data with GeoDa a workbookrdquohttpgeodacenterasuedusystemfilesgeodaworkbookpdf2005

[27] J P Elhorst Spatial econometrics Springer Briefs in RegionalScience Springer Heidelberg 2014

[28] J P Elhorst ldquoSpecification and estimation of spatial panel datamodelsrdquo International Regional Science Review vol 26 no 3pp 244ndash268 2003

[29] Baltagi B H Econometric Analysis of Panel Data vol 7 WileyNew York Chichester 3rd edition 2015

[30] K Luo G Li C Fang and S Sun ldquoPM25 mitigation in ChinaSocioeconomic determinants of concentrations and differentialcontrol policiesrdquo Journal of Environmental Management vol213 pp 47ndash55 2018

[31] G Du S Liu N Lei and Y Huang ldquoA test of environmentalKuznets curve for haze pollution in China Evidence from thepenal data of 27 capital citiesrdquo Journal of Cleaner Productionvol 205 pp 821ndash827 2018

[32] D-X Niu Z-Y Song and X-L Xiao ldquoElectric power sub-stitution for coal in China Status quo and SWOT analysisrdquoRenewable amp Sustainable Energy Reviews vol 70 pp 610ndash6222017

[33] K Cheng ldquoSpatial overflow effect of haze pollution in Chinaand its influencing factorsrdquo Nature Environment and PollutionTechnology vol 15 no 4 pp 1409ndash1416 2016

[34] J Chen C Zhou S Wang and J Hu ldquoIdentifying the socioeco-nomic determinants of population exposure to particulatemat-ter (PM25) in China using geographically weighted regressionmodelingrdquo Environmental Pollution vol 241 pp 494ndash503 2018

Hindawiwwwhindawicom Volume 2018

MathematicsJournal of


Mathematical Problems in Engineering

Applied MathematicsJournal of


Probability and StatisticsHindawiwwwhindawicom Volume 2018

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of



Engineering Mathematics

International Journal of


Operations ResearchAdvances in

Journal of


Function SpacesAbstract and Applied AnalysisHindawiwwwhindawicom Volume 2018

International Journal of Mathematics and Mathematical Sciences


Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Hindawiwwwhindawicom Volume 2018Volume 2018

Numerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisAdvances inAdvances in Discrete Dynamics in

Nature and SocietyHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Dierential EquationsInternational Journal of

Volume 2018


Decision SciencesAdvances in


AnalysisInternational Journal of


Stochastic AnalysisInternational Journal of

Submit your manuscripts atwwwhindawicom


haze Ma and Zhang (2014) asserted that in the long runchanging the energy consumption structure and optimizingthe industrial structure are the keys to controlling haze Inthe short term reducing the use of inferior coal is an effectivemethod [3] Hao and Liu (2015) indicated that to reducehaze pollution in China the energy structure in secondaryindustries should be improved by reducing coal consumptionand increasing the consumption of clean energy such asnatural gas and renewable energy [4] Wei and Ma (2015)pointed out that adjusting energy structure and technologicalprogress are the fundamental means to control haze [5] Li(2016) found that the cause of haze pollution in China was thebackwardness of energy technologies the higher proportionof coal consumption and the greater proportion of heavyindustries [6]Wang X et al (2018) found that clean energyconsumption had negative effects on PM25 concentrationsin China [7]

Technological innovation is the introduction of newproducts services or elements in the production process orservice operations [8] Technological innovation accelerateseconomic growth by increasing productivity and reducesenergy consumption in the production process by increasingenergy efficiency ultimately reducing pollution emissions [9]The literature has investigated the impact of technologicalinnovations on atmospheric pollution For example Popp(2003) used patent data to study the innovation of the UnitedStatesrsquo flue gas desulfurization devices (ldquoscrubbersrdquo) andshowed that technological innovations since 1990 increasedscrubber removal efficiency [10] Shi and Lai (2013) found thatthe diffusion of low-carbon and green technological inno-vations promotes regional economic development but savesenergy through low-carbon dioxide emissions or zero energyemissions [11] Yang and Li (2017) studied the relationshipbetween technological change and carbon emissions basedon Chinarsquos 1997ndash2010 panel data for 30 provinces and foundthat technological progress is the greatest contributor to CO2emission reductions [12] Wang et al (2018) estimated theimpact of Chinese technical knowledge on the intensity ofindustrial air pollutants based on the learning curve andenvironmental learning curve theories and found a causalrelationship between technical knowledge and the intensityof industrial air pollutants [13]

The literature has also investigated the impact of tech-nological innovations on haze For example Chang et al(2017) applied regional optimization methods to analyze theimpact of technological development in Shanghai on PM25pollution and found that decreased fossil energy consump-tion and increased use of clean energy can reduce PM25contamination [14] Xu and Lin (2018) believe that Chinarsquosenergy-saving emission reduction technology continues tolag behind developed countries whichmakes reducing PM25pollution in China is a difficult problem to overcome [15]Zhang et al (2016) suggested that technological innovationshould be strengthened to prevent and control haze pollutioninBeijing [16] Liu (2018) found that technological innovationreduces PM10 at a regional level and indirectly reduces PM10in adjacent regions through knowledge spillover effects [17]

The literature has also discussed the impact of energydevelopment and technological innovation onhaze pollution

but the following deficiencies have been observed Firstlyfrom the perspective of energy development the literature onthe impact of energy structure on haze pollution is abundantbut investigations of the impact of energy development com-binedwith technological innovation on the haze pollution arescarce Secondly the literature on the impact of technologicalinnovation on environmental pollution has been increasingbut the literature on the impact of technological innovationon haze pollution especially on PM25 is relatively scarceThirdly from the perspective of research methods most ofthe literature has analyzed the impact of technological inno-vation on haze pollution from the perspective of ordinarypanels but investigations from the perspective of spatialmeasurement are still lacking Finally there is a dearth ofliterature on the use of technical innovation infrastructureindicators to measure technological innovation which isusually measured by patents or RampD input indicators

Based on the deficiencies in the literature this paperintends to expand on the following aspects First energydevelopment and technological innovation are included inthe same framework to examine the impact of energy devel-opment and technological innovation on PM25 Secondly thespatial Durbin model with an adjacent weight matrix anddistance weightmatrix is used to analyze the impact of energydevelopment and technological innovation on PM25 Finallyindicators such as technological innovation infrastructure areused to refer to technological innovation

2 Methodology and Data

21 Global Spatial Autocorrelation The global spatial auto-correlation can effectively describe the overall situation ofPM25 changes at the interprovincial level and is usuallyanalyzed by the Global Moranrsquos I index This index was firstproposed by Moran (1950) [18] The Global Moranrsquos I indexreflects the similarities between units in each region andadjacent regions through a spatial weight matrix (Anselin1988) [19] Its calculation formula is as follows

1198721199001199031198861198991015840119904 119868 =119899sum119899119894=1 sum119899119895=1119908119894119895 (119909119894 minus 119909) (119909119895 minus 119909)sum119899119894=1 sum119899119895=1119908119894119895sum119899119894=1 (119909119894 minus 119909)

2

=sum119899119894=1 sum119899119895=1119908119894119895 (119909119894 minus 119909) (119909119895 minus 119909)

1198782sum119899119894=1 sum119899119895=1119908119894119895

(1)

where 1198782 = (1119899)sum119899119894=1(119909119894minus119909)2 is the variance of the obser-vations 119909 = (1119899)sum119899119894=1 119909119894 is the average of all spatial unitobservations xi and xj represent the ith and jth spatial unitobservations respectively and n is the total number of spatialunits wij is a spatial weight matrix The range of the GlobalMoranrsquos I Index is [-1 1] Igt0 indicates a positive spatialautocorrelation of interprovincial PM25 in China The largerthe value is the stronger the agglomeration is that is the highvalue zone is adjacent to the high value zone and the lowvalue zone is adjacent to the low value zone Ilt0 indicatesa negative autocorrelation of the interprovincial PM25 inChina that is the high value zone is adjacent to the low




119885 = 119868 minus 119864 (119868)radic119881119860119877 (119868) (2)




119910119894119905 = 120575119873

sum119895=1

119908119894119895119910119895119905 + 120572 + 119909119894119905120573 + 120583119894 + 120582119905 + 120576119894119905

119894 = 1 119873 119905 = 1 119879(3)



119908119894119895 =

1 119891119900119903 119894 = 119895 119886119899119889 119899119890119894119892ℎ1198871199001199031198941198991198920 119891119900119903 119894 = 119895 119886119899119889 119899119900119905 - 119899119890119894119892ℎ1198871199001199031198941198991198920 119891119900119903 119894 = 119895

(4)


119908119894119895 =

11198891198941198952

119894 = 119895

0 119894 = 119895(5)



119910119894119905 = 120572 + 119909119894119905120573 + 120583119894 + 120582119905 + 120601119894119905

120601119894119905 = 120588119873

sum119895=1

119908119894119895120601119894119905 + 120576119894119905(6)



119910119894119905 = 120575119873

sum119895=1

119908119894119895119910119895119905 + 120572 + 119909119894119905120573 +119873

sum119895=1

119908119894119895119909119894119895119905120579 + 120583119894 + 120582119905 + 120576119894119905 (7)



ln (11987511987225119894119905) = 120572 + 1205731 ln (119864119894119894119905) + 1205732 ln (119862119904119894119905)

+ 1205733 ln (119875119901119894119894119905) + 1205734 ln (119873119901119894119894119905)

+ 1205735 (ln119875119886119905119890119899119905) + 1205736 ln (119868119899119891119903119886119894119905)

+ 1205737 ln (119872119886119903119896119890119905119894119905) + 120576119894119905

(8)












000

2000

4000

6000

8000

10000

12000

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016








No data

N

8560000 - 26580000

26580001 - 38830000

38830001 - 50380000

50380001 - 640000000 500 km







27

16

05

minus06

minus17

minus28


lagg

ed P

M2_

5

PM2_5




















































Data Availability




Acknowledgments


References











































Journal of












Journal of







Volume 2018






Volume 2018











119885 = 119868 minus 119864 (119868)radic119881119860119877 (119868) (2)




119910119894119905 = 120575119873

sum119895=1

119908119894119895119910119895119905 + 120572 + 119909119894119905120573 + 120583119894 + 120582119905 + 120576119894119905

119894 = 1 119873 119905 = 1 119879(3)



119908119894119895 =

1 119891119900119903 119894 = 119895 119886119899119889 119899119890119894119892ℎ1198871199001199031198941198991198920 119891119900119903 119894 = 119895 119886119899119889 119899119900119905 - 119899119890119894119892ℎ1198871199001199031198941198991198920 119891119900119903 119894 = 119895

(4)


119908119894119895 =

11198891198941198952

119894 = 119895

0 119894 = 119895(5)



119910119894119905 = 120572 + 119909119894119905120573 + 120583119894 + 120582119905 + 120601119894119905

120601119894119905 = 120588119873

sum119895=1

119908119894119895120601119894119905 + 120576119894119905(6)



119910119894119905 = 120575119873

sum119895=1

119908119894119895119910119895119905 + 120572 + 119909119894119905120573 +119873

sum119895=1

119908119894119895119909119894119895119905120579 + 120583119894 + 120582119905 + 120576119894119905 (7)



ln (11987511987225119894119905) = 120572 + 1205731 ln (119864119894119894119905) + 1205732 ln (119862119904119894119905)

+ 1205733 ln (119875119901119894119894119905) + 1205734 ln (119873119901119894119894119905)

+ 1205735 (ln119875119886119905119890119899119905) + 1205736 ln (119868119899119891119903119886119894119905)

+ 1205737 ln (119872119886119903119896119890119905119894119905) + 120576119894119905

(8)












000

2000

4000

6000

8000

10000

12000

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016








No data

N

8560000 - 26580000

26580001 - 38830000

38830001 - 50380000

50380001 - 640000000 500 km







27

16

05

minus06

minus17

minus28


lagg

ed P

M2_

5

PM2_5




















































Data Availability




Acknowledgments


References











































Journal of












Journal of







Volume 2018






Volume 2018

















000

2000

4000

6000

8000

10000

12000

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016








No data

N

8560000 - 26580000

26580001 - 38830000

38830001 - 50380000

50380001 - 640000000 500 km







27

16

05

minus06

minus17

minus28


lagg

ed P

M2_

5

PM2_5




















































Data Availability




Acknowledgments


References











































Journal of












Journal of







Volume 2018






Volume 2018











No data

N

8560000 - 26580000

26580001 - 38830000

38830001 - 50380000

50380001 - 640000000 500 km







27

16

05

minus06

minus17

minus28


lagg

ed P

M2_

5

PM2_5




















































Data Availability




Acknowledgments


References











































Journal of












Journal of







Volume 2018






Volume 2018









27

16

05

minus06

minus17

minus28


lagg

ed P

M2_

5

PM2_5




















































Data Availability




Acknowledgments


References











































Journal of












Journal of







Volume 2018






Volume 2018
















































Data Availability




Acknowledgments


References











































Journal of












Journal of







Volume 2018






Volume 2018



























Journal of












Journal of







Volume 2018






Volume 2018








effect of energy development and technological innovation...

Documents