Table Of ContentResources Policy 60 (2019) 203–214
ContentslistsavailableatScienceDirect
Resources Policy
journal homepage: www.elsevier.com/locate/resourpol
Resource abundance, industrial structure, and regional carbon emissions
T
efficiency in China
Keying Wanga,b,c,d, Meng Wua,e, Yongping Suna,b,⁎, Xunpeng Shia,b,f,g, Ao Sunj, Ping Zhangh,i
aCenterofHubeiCooperativeInnovationforEmissionsTradingSystem,HubeiUniversityofEconomics,China
bSchoolofLowCarbonEconomics,HubeiUniversityofEconomics,China
cSchoolofStatisticsandMathematics,ZhongnanUniversityofEconomicsandLaw,China
dCrawfordSchoolofPublicPolicy,AustralianNationalUniversity,Australia
eAccountingSchool,HubeiUniversityofEconomics,China
fAustralia-ChinaRelationsInstitute,UniversityofTechnologySydney,Australia
gEnergyStudiesInstitute,NationalUniversityofSingapore,Singapore
hCenterforSocialSecurityStudies,WuhanUniversity,China
iSchoolofPoliticalScienceandPublicAdministration,WuhanUniversity,China
jSchoolofEconomics,XiamenUniversity,China
ARTICLE INFO ABSTRACT
Keywords: With increasing concerns over climate change and the global consensus regarding low carbon growth, the
Resourceabundance transitionofresource-basedregionshasbecomeurgentandchallenging.WeemployaSlacks-BasedMeasure
Resourcedependence withwindowsanalysisapproachtoestimatethecarbonemissionsefficiencyandabatementpotentialofChina's
Industrialstructure provincesovertheperiodof2003–2016.ApanelTobitmodelisfurtheremployedtoanalyzethedirectand
Emissionsefficiency
indirecteffectsofnaturalresourceabundanceonemissionsefficiency.Wefindthat:(1)Thereexistsanegative
Abatementpotential
correlationbetweenresourceabundanceandcarbonemissionsefficiency.Themoreabundanttheresources,the
lowertheemissionsefficiency.(2)Althoughemissionsefficiencyandabatementpotentialaregenerallynega-
tivelycorrelated,abatementpotentialalsodependsonthescaleoftheeconomy.(3)Resourcedependenceis
unfavourablefortherationalizationandadvancementoftheindustrialstructure,whichindirectlyaffectsthe
carbonemissionsefficiency.Thesefindingsimplythatresource-basedregionsshouldmaketheimprovementof
emissionsefficiencyandtheexplorationofabatementpotentialastheirtoppriorityofactionsforalow-carbon
transition,andpromotethetransformationofindustrialstructureinordertoobtainadoubledividendinsus-
tainabledevelopmentandcarbonemissionsefficiency.
1. Introduction fossilfuelsinprimaryenergyconsumptiontoaround20%(Xianetal.,
2018). In 2017, non-fossil fuels accounts for 13.6% of China's total
Lowcarbongrowthiswidelyregardedasthekeywaytoresolvethe primary energy consumption (BP, 2018). In December 2016, the Na-
contradictory demands for economic growth and carbon emissions tional Energy Administration issued the “Revolutionary Strategy for
mitigation.Findingwaystouseresourcesincludingenergy,moreeffi- Energy Production and Consumption (2016–2030)” which states that
ciently, is a key requirement for low carbon growth. China, as the non-fossilenergywillaccountformorethanhalfofprimaryenergyin
world'slargestcarbonemitter,hasurgencytoincreaseemissionseffi- 2050 (NDRC, 2016). The nationalunified carbonmarket that was of-
ciency,reducecarbonemissions andrealizelow-carboneconomicde- ficiallylaunchedinDecember2017,couldincreasetheabatementcosts
velopment. While China has formally promised the world in the of enterprises and reduce the demand for fossil fuels (Wang et al.,
Intended Nationally Determined Contributions (INDC) to peak its 2018).
carbon emissions around 2030, it is actually trying to peak the emis- Although resource-based regions have played a major role in pro-
sionsearlierthanthisdeadline.Chinahasintegratedpoliciespertaining motingtheinitialstagesofindustrialization,implementinglow-carbon
tothecontrolofgreenhousegasemissionsintothenationaleconomic transitionisaseverechallengeforresource-basedregionswhoseeco-
andsocialdevelopmentstrategy,suchastoincreasetheshareofnon- nomic growth is often dominated by resource-intensive industries.
⁎Correspondingauthor.
E-mailaddress:[email protected](Y.Sun).
https://doi.org/10.1016/j.resourpol.2019.01.001
Received12September2018;Receivedinrevisedform29December2018;Accepted2January2019
Available online 12 January 2019
0301-4207/ © 2019 Elsevier Ltd. All rights reserved.
K.Wangetal. Resources Policy 60 (2019) 203–214
Depending on resource advantages, resource-based regions have de- differentperspectivesusingvarioustheoriesandmethods.Manystudies
veloped compatible industrial structures, and these have greatly ac- have empirically demonstrated that a large number of regions with
celeratedregionaldevelopment(Shi,2013).However,mostofthein- abundantnaturalresources,especiallycoal,oilandgas,aretrappedin
dustriesintheseregionsarelikelytobecharacterizedbyhighenergy the “resources curse” (Ahmed et al., 2016; Badeeb et al., 2017;
and emissions intensities (Feng et al., 2017). The abundance of the Brunnschweiler,2008;FriedrichsandInderwildi,2013;Gerelmaaand
naturalresourcesleadstolowpricesofresources,whichhasledtohigh Kotani,2016;ShaoandYang,2014;Songetal.,2018)
extensive and inefficient energy consumption patterns and low emis- Various approaches have been applied to evaluate the carbon
sionsefficiency(AdomandAdams,2018;Yangetal.,2018).Further- emissions efficiency and carbon abatement potentials. By applying a
more, resource intensive industries tend to cluster in resource-based dataenvelopmentanalysis(DEA)method,recentstudies(Wangetal.,
regionsandformindustryagglomeration,eventuallybecomethepillar 2013; Xian et al., 2018; Zha et al., 2016) found that even if all elec-
industries,whichfurtherleadstoresourcedependence.Afteragglom- tricity-generating units in each region were able to adopt the best
eration, non-resource intensive industries are closely attached to the practices, the nationwide 18% intensity reduction target was not fea-
resource-intensiveones,andasaresultthefurtherresourcedependence siblethroughimprovingtechnicalefficiencyinashortormediumterm.
worsensthecarbonemissionsefficiencyinresource-basedregions.As Owing to the diversity among the development patterns and natural
themajoroutputsofresource-basedregions,itisunrealistictoabandon resource endowments in China's various regions, there is significant
resource-intensive industries since a consequent growth plummet re- differenceinthecarbonemissionsperformancesattheprovinciallevels
sultinginserioussocialandeconomicproblems. (Chang et al., 2017; Yao et al., 2015). The remarkable imbalances in
However, wherenaturalresource endowmentis rich,resource de- economicdevelopment,technologygaps,policies,industrialstructures
pendenceisnotnecessarilyhigh.Naturalresourceabundanceincludes and energy consumption structures may explain the regional differ-
tworelatedcases:richendowmentandhighdependence.Thenatural ences in carbon emissions efficiency (Lin and Du, 2015; Wang et al.,
resourcesendowmentreferstothequantityofnaturalresourcesthata 2016;Yaoetal.,2015).
country or region can use for social and economic development; the Some researchers argued that rational industrial structure adjust-
natural resources dependence refers to the role of resource-based in- ment could improve resource utilization efficiency (Zhang and Deng,
dustries in the development of regional economy (Sun and Ye, 2012; 2010)andmitigatecarbonemissions(Lietal.,2017;Shaoetal.,2016).
Wuetal.,2018). Therearemanystudiesdiscussinghowtoadjusttheindustrialstructure
Given the difficulties in improving emissions efficiency and redu- soastoreducecarbonemissions,suchasincreasingtheproportionof
cingcarbonemissionsinresource-basedregions,itisatimelyandva- tertiaryindustryinGDP(Zhangetal.,2014).Tianetal.(2014)pointed
luableexercisetoinvestigatetherelationshipbetweenresourceabun- outthatdifferentsolutionsshouldbeusedtocontrolCO2emissionsin
dance,industrialstructureandcarbonemissionsefficiencysoastooffer regionswhichareatdifferentstagesintheprocessofindustrialstruc-
policysuggestionsforlowcarbontransitioninresource-basedregions. turalchange.
Astheworld'slargestproducerofcoalandthelargestemitterofcarbon, Theroleofindustrialstructureincarbonemissionscontrolmaybe
China provides an excellent case to study the topic and focusing on moreimportantinresource-basedregionsthaninotherregions.Long-
Chinaisimportantfortheglobalcommunity.Thedevelopingcountry termresourcedevelopmenthasmadeindustrialstructuresinresource-
statusandlaggedeconomicdevelopmentmeanthatChina'slessonsand basedregionsdominatedbynaturalresourcedevelopmentandprimary
experience can be useful for other developing countries that rely on processing (Sun and Ye, 2012). Such industrial structures will likely
naturalresources. lead to high emissions intensity in the resource-based regions. Fur-
Inthispaper,weapplytheSlakes-BasedMeasure(SBM)withwin- thermore,lowlevelindustrialstructuresintheresource-basedregions
dows analysis approach to estimate carbon emissions efficiency and have“lock-ineffect”and“crowd-outeffect”,whichhindertheadjust-
abatement potential of China's 30 provinces from 2003 to 2016, and ment and evolution of regional industrial structures (Li et al., 2019;
analyzethedirectandindirectimpactofresourceabundanceoncarbon Morrisetal.,2012).Suchfeaturesmakeitdifficultforresource-based
emissionsefficiencyfromtwoperspectivesofresourcedependenceand regions to achieve sustainable development in a low carbon world.
endowment.Ouranalysisisausefulextensiontotheexistingliterature However,afewresearchesshowpotentialofovercomingnegativere-
and can offer suggestions relating to low-carbon transition in China's source abundance effects. Balsalobre-Lorente et al. (2018) found that
resource-basedregions.Thecontributionsofthispaperaretwofold:1) countrieswithnaturalresourcesreducedtheirimportsofdirtyenergy
analysis of the impacts of natural resources abundance on carbon resources,whichhadapositiveeffectonCO2emissionsreduction.
emissionsefficiency;and2)analysisoftheinfluenceofnaturalresource Insummary,theexistingrelatedresearchprovidesalittlestudyof
abundance on industrial structures, and then examination of the in- carbonemissionsefficiencyandabatementpotentialwhileconsidering
directeffectsoncarbonemissionsefficiency. thenaturalresourcesabundanceandindustrialstructure.Inthispaper,
The paper proceeds as follow: Section 2 reviews the literature, by introducing two indicators which denote industrial structure and
Section3discussestheinfluencemechanismofresourceabundanceon employing the panel Tobit model, we analyze the direct and indirect
emissionsefficiency,Section4elaboratesonthemethodologyanddata, effects of resource abundance on emissions efficiency from the per-
and Section 5 presents the empirical results and discussion. The con- spectivesofresourcedependenceandendowment.
cluding section provides policy recommendations and suggestions for
furtherresearch.
3. Influencemechanismofregionalcarbonemissionsefficiency
2. Literaturereview
Thispaperwillanalyzetheimpactofnaturalresourcesabundance
Thispaperiscloselyrelatedtothreeresearchstrandsinthelitera- oncarbonemissionsefficiencyfrombothdirectandindirectchannels.
ture. The first strand is “resource curse”, which is a popular topic in On the one hand, abundant natural resources loosen the resource
resource economics. The second strand is carbon emissions efficiency constraints of enterprises, leading to the use of resources in a more
and abatement potential. The last strand is industrialstructure which extensive and inefficient manner, which directly affects the carbon
connects closely with the first two strands. Therefore, the literature emissions efficiency of resource-based regions. On the other hand,
reviewheredealswiththesethreeaspects. abundantnaturalresourceswillalsodistorttheindustrialstructureof
Formostcountriesandregions,havingabundantresourceshinder resource-based regions, making high emissions industries as pillar in-
long-runeconomicgrowthratherthanpromoteit.The“resourcecurse” dustries, which indirectly affects the carbon emissions efficiency of
has become a popular academic topic and has been discussed from resource-basedregions.
204
K.Wangetal. Resources Policy 60 (2019) 203–214
3.1. Directinfluence 4.1. TheSBMwithwindowanalysisapproach
The natural resources abundance causes relatively low resource 4.1.1. TheSlacks-BasedMeasure
prices and thus make companies’ behavior in resource-based regions The SBM with window analysis approach is employed to estimate
different from those in other regions. Due to the convenience, avail- the carbon emissions efficiency of China's 30 provinces (except for
ability and lower prices, companies located in resource-based regions Tibet)from2003to2016.UndertheframeworkofDEA,thenon-radial
face a lower risks as well as a lower costs for resource reserves. Low and non-oriented Slacks-Based Measure (SBM) can utilize input and
resourcepricesleadtoalowerwillingnesstoinvestinresource-saving output slacks directly in producing an efficiency, it has been widely
technologies and equipment (Shi, 2014). Furthermore, resource-in- appliedtoevaluatecarbonemissionsefficiencyandabatementpoten-
tensivecompanieshavetokeepcertainquantitiesofresourcereserves tial (Cecchini et al., 2018; Choi et al., 2012; Guo et al., 2017; Song
to guard against possible operational risks caused by resource et al., 2013; Zhang et al., 2017, 2015; Zhang and Choi, 2013; Zhou
shortages, which places an extra pressure on companies’ financial etal.,2006,2013).
status.Overall,theextensiveuseofresourceswillinevitablyleadtoa TakingChina's30provincesasDMUj(j=1,2…30),theSBMcanbe
declineincarbonemissionsefficiency. writtenasfollows:
1 1 m si
*=min m i=1xi0
3.2. Indirectinfluence 1+ 1 s1 srg + s2 srb
s1+s2 r=1yrg0 r=1yrb0 (1)
Abundanceofnaturalresourcesnotonlyleadstoarigidindustrial
x =X +s ;
structures,butalsoreducestheemissionsreductionsderivedfromthe 0
industrialstructuredividend(SunandYe,2012),whichinturnaffects s.t. y0g =Yg sg;
thecarbonemissionsefficiency.Theindustrialstructuresdominatedby yb=Yb +sb;
0
asingleresourcesector,thatisresourcedependence,havesqueezedthe 0,s 0, sg 0, sb 0 (2)
development space of modern manufacturing. Thus resources-based
regionsoftenfallintoarigidspecializationtrap.Thetendencyof“de- wherethevectors x Rm, yg Rs1 and yb Rs2 represent inputs,de-
industrialization”hascausedtheindustrialstructureofresource-based sirable outputs and undesirable outputs respectively. The vectors
regionstobeinastateofdistortionforalongtime,thustheycannot s Rn and sb Rs2 correspondtoexcessesininputs andundesirable
gainthe“structuraldividend”resultingfromtheoptimizationandup- outputs respectively, while sg Rs1 expresses shortages in desirable
grading of industrial structures. However, the industrial structure is outputs. The objective value satisfies0< * 1. Let an optimal solu-
shapedbymarketselectionundertheconstraintsofnaturalresources, tionoftheaboveprogrambe( *,s *,sg*,sb*).Then,theDMUjiseffi-
technologies, economic development stages and other factors within cientinthepresenceofundesirableoutputsifandonlyif *=1,i.e.,
the economic system. Each of these factors offer spontaneity, en- s *=0, sg*=0, and sb*=0.IftheDMUjisinefficient,i.e., *<1,it
dogeneityandrationalitytosomedegree.Fig.1summarizesthedirect canbeimprovedandbecomeefficientbydeletingtheexcessesininputs
andindirecteffectsofnaturalresourceabundanceoncarbonemissions and undesirable outputs, and augmenting the shortfalls in desirable
efficiency. outputsbythefollowingSBM-projection:
xˆ x s *
0 0
yˆg yg+sg*
4. Methodologyanddata 0 0
yˆb yb sb*
0 0 (3)
Two distinctive methods are employed in this study. The Slacks-
In this paper, the excess in undesirable outputs means the carbon
based Measure (SBM) with window analysis approach estimates the
abatement potential and is denoted as sb*, which is the quantity of
carbon emissions efficiency and abatement potential, while the panel
Tobit model investigates the influencing factors of carbon emissions potentialemissionsreductionofDMUjwhen *isimprovedto1.
Clearly, abatement potential measures the absolute reductions of
efficiency.
carbonemissions.Itdependsbothontheemissionsefficiencyandthe
scaleoftheeconomy.Someregionswithhighemissionsefficiency,due
Fig.1.Influencemechanism.
205
K.Wangetal. Resources Policy 60 (2019) 203–214
to the large scale of the economy, may still have larger absolute andindustrialstructure,governmentintervention,technologyinnova-
quantityofemissionsreductions.Someregionswithlowemissionsef- tion,energyprice,theurbanizationlevelandregulationareregardedas
ficiencies,duetothesmallerscaleoftheeconomy,mayhavesmaller the important factors that impact the carbon emissions efficiency in
absolutequantityofemissionsreductions.Thus,SBMprovidesascalar manystudies.
measurerangingfrom0to1thatencompassesalloftheinefficiencies Because of the externality of carbon emissions, government inter-
thatthemodelcanidentify. vention plays an important role in improving carbon emissions effi-
InestimatingcarbonemissionsefficiencybasedonSBM,weemploy ciency, and the fiscal policy is a common form of government inter-
labor, capital and energy to represent the inputs, GDP as a desirable vention through providing funds for improving of energy-saving and
output,andthetotalamountofCO2emissionsasanundesirableoutput. emissions-reductiontechnologies,encouragingenterprisestoeliminate
Specifically,laborisdenotedbythenumberofemployedpersons,ca- backward production capacity through incentives and subsidies, sup-
pitalisestimatedusingtheperpetualinventorymethod: portingthedevelopmentofcleanenergy(Priceetal.,2005).Govern-
mentintervention(GOV)isdenotedbytheratiooffiscalexpenditureto
Kit=(1 it)Kit 1+Iit (4)
fiscalrevenue.
energyisrepresentedbythetotalenergyconsumptionofeachprovince, Thetechnologyinnovationisplayingmoreandmoreimportantrole
andCO2emissionsarecalculatedwiththeenergyemissionsfactorsand in improving carbon emissions efficiency and it is essential to meet
energyconsumption. long-term emissions reduction targets (Dechezleprêtre et al., 2016;
Gallagheretal.,2006).Thetechnologyinnovation(R&D)isexpressed
astheratioofR&Demployeesinallemployedpeople.
4.1.2. Thewindowanalysisapproach
When energy prices continue to rise, the cost effect will stimulate
Duetotheadvantagesofanalyzingafrontiershiftbetweendifferent
energy conservation and emissions reduction (Fisher-Vanden et al.,
periods under a possible occurrence of a frontier crossover and in
2004; McCollum et al., 2016), while speeding up the diffusion of en-
handlingpaneldata,thewindowanalysisapproachisusedtoevaluate
ergy-savingtechnologies,reducingenergyconsumptionandimproving
carbon emissions efficiencies over time and across different regions,
carbon emissions efficiency (Jacobsen, 2015). The change of energy
sectorsandsubjects(Al-Refaieetal.,2018;Cucciaetal.,2017;Linand
price (EPI) can be signified by purchasing price indices for industrial
Tan, 2017; Shawtari et al., 2015; Sueyoshi et al., 2013; Vlontzos and
producersoffuelandpower.
Pardalos,2017;Wangetal.,2013).
In the process of urbanization, economic activities are centralized
A window with n×w observations is denoted starting at time t
(1 t T) with window width w (1 w T t), n=30 for China's andtheenergyhasbeenconsumedmassively.Ontheotherhand,the
scale effect and technique spill-over effect from the agglomeration of
30provincesandtheT=14forthe2003–2016period.Theselectionof
economic activities will reduce the intensity of energy consumption,
thewidthofthewindowwisakeypointinthewindowanalysisap-
improveenergyconsumptionefficiencyandcarbonemissionsefficiency
proach. As is most commonly done in the literature, we set w =3
(WangandZhang,2016).Thelevelofurbanization(UR)isrepresented
(HalkosandTzeremes,2009;VlontzosandPardalos,2017).
bytheproportionofurbanresidentpopulationineachprovince.
Astheclimatechangebecomemoreserious,governmentswilltake
4.2. PanelTobitmodel
stricter environmental regulations, which exert extra costs on en-
terprisesandforcethemtoadoptemissions-reductiontechnologiesand
SincethecarbonemissionsefficiencybaseonSBMiscensoredby0
clean energy. However, excessive cost may not be conducive to the
and 1, in this case, parameter estimates obtained by conventional re-
operationofenterprisesandresultinthedecreaseofcarbonemissions
gression methods (e.g. OLS) are biased. Consistent estimates can be
efficiency.Weapplytheenergy-savingandemissions-reducingtargets
obtainedbyTobitmodelproposedbyTobin(1958),whichisaspecial
foreachprovincein“Five-YearPlans”astheindicatorofenvironmental
case of the more general censored regression model (Baltagi and
regulation(Regulation).
Boozer,1997;Maddala,1987)andhasbeenusedinaverywiderange
of applications characterized by censored observations. These are re-
4.3. Variableconstructionanddatasources
cent examples (Bi et al., 2016; Brown et al., 2015; Kaya Samut and
Cafrı, 2016). We apply a random panel Tobit model to estimate the
Consideringtheregionaleconomiclandscape,resourceabundance
possible determinants of carbon emissions efficiency. The model is
and geographic features (Zhou et al., 2014), we divided China's 30
writtenas:
provinces into eight economy-geographic regions, i.e., the northeast,
lnefficiencyit = + 1lnNRDit+ 2lnrationalit+ 3lnadvancedit northcoast,eastcoast,southcoast,themiddleYellowRiver,themiddle
+ 4lnNRDit*lnrationalit+ 5lnNRDit*lnadvancedit YangtzeRiver,southwestandnorthwestregions.
+ lnPGDP + lnNRD *lnPGDP +X + + Themeasurementindicatorsofnaturalresourcesabundancecanbe
6 it 7 it it it i it
roughlydividedintotwocategories:theresourcedependenceindexand
(5)
theresourceendowmentindex.Consideringthatthispapercalculates
where efficiency indicates carbon emissions efficiency of province i in the carbon emissions efficiency from energy consumption, we choose
year t calculated by the SBM with window analysis approach. NRD theoutputvalueproportionofthecoalminingindustryandtheoiland
indicates natural resource dependence. The variables of rational and gasextractionindustryintotalindustrialoutputvaluetorepresentthe
advancedaretwoindicatorsusedtocharacterizethedevelopmentofthe degreeofnaturalresourcedependence(NRD).Thelargerthevalue,the
industrial structure. PGDP represents the level of economic develop- higherthedegreeofdependence.Atthesametime,inordertocarryout
ment and is denotedby the GDP per capita measured by theprice in robustnesstests,thispaperalsocalculatesothervariablesthatcanre-
2003. present the natural resource dependence: the natural resource depen-
Xsarethecontrolvariablevectors.Besidestheresourcedependence denceinemployment(NRDL),whichisrepresentedbytheemployment
206
K.Wangetal. Resources Policy 60 (2019) 203–214
proportionofthecoalminingindustryandtheoilandgasexploration Table1
industry in total industrial employment. Considering that this paper CarbonemissionsefficiencyinChina's30provinces(2004–2016).
mainly measuresthe efficiencyand potentialofcarbonemissions, we
Region 2004 2006 2008 2010 2012 2014 2016
usefossilenergyendowment(FEE)torepresentresourcesendowment
index,whichisrepresentedbytheratioofproductiontoconsumption MiddleYellowRiverRegion 0.479 0.520 0.580 0.589 0.590 0.581 0.525
offossilfuels.Thelargertheratio,thehigherthedegreeoffossilenergy Shanxi 0.394 0.404 0.455 0.487 0.490 0.475 0.469
InnerMongolia 0.466 0.530 0.575 0.579 0.542 0.674 0.496
endowment.
Henan 0.536 0.582 0.661 0.652 0.671 0.574 0.543
We employ two indicators to characterize the improvement of in- Shaanxi 0.519 0.565 0.631 0.641 0.656 0.599 0.591
dustrial structure: rationalization and advancement, separately mea- NorthCoastRegion 0.677 0.711 0.764 0.776 0.781 0.831 0.886
suringtheallocationefficiencyofproductionfactorsamongindustries Hebei 0.556 0.575 0.587 0.612 0.624 0.601 0.545
Shandong 0.689 0.646 0.637 0.615 0.615 0.766 1.000
and the stages of industrial structures evolution (Sun and Ye, 2012).
Beijing 0.792 0.942 0.960 0.944 0.989 0.992 1.000
Rationalizationmeanshighallocationefficiencyandcouplingquality,
Tianjin 0.673 0.682 0.872 0.935 0.896 0.967 1.000
in which economic development is enhanced and carbon emissions NortheastRegion 0.497 0.553 0.607 0.621 0.663 0.669 0.643
efficiency is improved. Rational means rationalization index of in- Liaoning 0.478 0.538 0.587 0.631 0.671 0.619 0.571
dustrialstructure(Ganetal.,2011),whichisshowninEq.(6): Heilongjiang 0.495 0.526 0.576 0.625 0.653 0.637 0.605
Jilin 0.518 0.594 0.656 0.607 0.664 0.749 0.753
n Y Y Y NorthwestRegion 0.679 0.642 0.681 0.675 0.698 0.671 0.756
Rational= i ln i/ Xinjiang 0.495 0.525 0.554 0.541 0.518 0.493 0.464
Y L L
i=1 i (6) Gansu 0.450 0.485 0.546 0.585 0.595 0.567 0.560
Ningxia 0.769 0.559 0.624 0.611 0.754 0.624 1.000
wherei=1,2,3indicatetheprimary,secondaryandtertiaryindustries Qinghai 1.000 1.000 1.000 0.964 0.928 1.000 1.000
respectively,andn=3.YandLindicatetheindustrialoutputandthe SouthwestRegion 0.625 0.655 0.710 0.728 0.680 0.679 0.728
Guizhou 0.415 0.442 0.527 0.559 0.575 0.600 0.614
industrialemploymentrespectively.Whentheeconomyisinanequi-
Guangxi 0.694 0.763 0.861 0.808 0.668 0.694 0.661
libriumstate,theproductionefficiencyofeachindustrywillconverge
(Yi/Li= YL and Rational=0). The smaller the value of Rational, the SYiucnhnuaann 00..653514 00..659618 00..761244 00..767385 00..760716 00..660974 00..678450
morereasonabletheindustrialstructure.Advancementisrepresentedby Chongqing 0.832 0.810 0.823 0.862 0.781 0.800 0.939
the ratio of gross value of the tertiary industrial sector to that of the MiddleYangtzeRiverRegion 0.640 0.674 0.757 0.767 0.757 0.751 0.783
Anhui 0.597 0.651 0.692 0.717 0.722 0.689 0.673
secondary industrial sector, and higher value means more advanced
Jiangxi 0.669 0.763 0.868 0.816 0.803 0.714 0.718
industrial structure. The development of tertiary industry plays an Hubei 0.613 0.618 0.722 0.757 0.701 0.681 0.743
important role in emissions reduction and improvement in carbon Hunan 0.681 0.662 0.746 0.778 0.803 0.922 1.000
emissionsefficiency. EastCoastRegion 0.796 0.844 0.876 0.930 0.878 0.818 0.975
The annual data that was used to estimate the carbon emissions Zhejiang 0.850 0.794 0.809 0.789 0.794 0.782 0.925
Shanghai 0.765 0.918 0.917 1.000 0.839 0.884 1.000
efficiencyallcomefromtheChinaStatisticalYearbooksof2004–2017.
Jiangsu 0.774 0.819 0.902 1.000 1.000 0.787 1.000
Other data are collected from the China Statistical Yearbook, China SouthCoastRegion 0.940 0.911 0.933 0.926 0.883 0.882 0.929
Industrial Statistical Yearbook, China Energy Statistical Yearbook, China Fujian 0.833 0.829 0.855 0.779 0.699 0.694 0.786
Population and Employment Statistics Yearbook, China Labor Statistics Guangdong 1.000 0.953 1.000 1.000 1.000 0.993 1.000
Hainan 0.988 0.951 0.944 1.000 0.948 0.960 1.000
Yearbook,andChinaScienceandTechnologyStatisticalYearbook.
Average 0.658 0.680 0.731 0.744 0.733 0.728 0.769
ThevaluesofRegulationfor2006–2010arecalculatedbasedonthe
energyconsumptionreductiontargetforeachprovinceduringthe11th
Five-Year Plan period, that for 2011–2016 are from “Work Plan for
ControllingGreenhouseGasEmissionsinthe12th(and13th)Five-Year
Plan”, and the values for 2003–2005 are set to 0 indicating no en- Theabatementpotentialistheexcessincarbonemissionswhenthe
vironmental regulation policies on carbon emissions in that period. DMUiisinefficientandneedstobedeletedasthecarbonefficiencyis
Exceptforenvironmentalregulations,thevaluesofthevariablesinthe improvedandtheDMUibecomesefficient.Fig.2showstheaggregate
modelarealllogarithms. abatementpotentialovertimein30provinces.Wecanseethat,firstly,
there are distinct differences in the abatement potential among the
provinces. Over the 11th Five-Year Plan, 12th Five-Year Plan and in
5. Empiricalresults
2016, the provinces with the largest abatement potential were Shan-
dong, Shanxi and Inner Mongolia, and the average annual abatement
5.1. Carbonemissionsefficiencyandabatementpotential
potentialwas595mt,677mtand822mtseparately,whilethesmallest
abatement potential was only 1.28 mt, 1.17 mt and 0.1 Secondly, al-
Based on the SBM with window analysis approach mentioned
thoughthecarbonemissionsformostoftheprovinceswereimproving,
above, we estimated the carbon emissions efficiency and abatement
theabatementpotentialswereincreasing,especiallyinthehighemis-
potentialforChina's30provinces.
sions regions of Inner Mongolia, Shanxi, Hebei, Xinjiang, Liaoning,
Table 1 shows the results for selected years. Firstly, the carbon
Henan,Shaanxi.Theemissionsreductionresultingfromefficiencyim-
emissions efficiencies in almost all provinces improved from
provementscannotoffsetincreasesinemissionscausedbytheexpan-
2003–2016butthereweresignificantgapsamongprovinces.In2016,
sionofproductionactivities(Choietal.,2012;Zhangetal.,2016).This
the carbon emissions efficiency in 10 provinces (Shandong, Beijing,
isthekeyforChinainachievingthetargetofcarbonemissionspeak.
Tianjin, Ningxia,Qinghai,Hunan, Shanghai,Jiangsu,Guangdongand
While emissions efficiency and abatement potential are generally
Hainan) achieved efficient production activities (the value of carbon
negatively correlated, the eight regions are categorized into four dis-
emissionsefficiencyis1),whiletheeightprovincesofXinjiang,Shanxi,
tinctgroupsaccordingtotherelationshipbetweenemissionsefficiency
InnerMongolia,Henan,Hebei,Gansu,LiaoningandShaanxihadvery
and abatement potential over the 12th Five-Year Plan (See Fig. 3 for
lower emissions efficiency (below 0.6). Secondly, there was regional
details). The first group has low efficiency in emissions with high
clustering in carbon emissions efficiency. The efficiency of carbon
abatement potential (LE-HP), including 11 provinces: Xinjiang, Inner
emissionsinthecoastalregionswasgenerallyhigherthanthatofthe
central and western regions. Not surprisingly, the coal-rich Middle
YellowRiverregionandtheNortheastregionhadthelowestemissions
efficiencies. 1Whenthecarbonemissionsefficiencyis1,thereisnoabatementpotential.
207
K.Wangetal. Resources Policy 60 (2019) 203–214
8000
The Eleventh Five-year Plan The Twel(cid:3)h Five-year Plan 2016
7000
A 6000
b
a
t
e 5000
m
e
n 4000
t
P
o
t 3000
e
n
(cid:2)
a 2000
l(m
t) 1000
0
ShanxiHebeiShandongInner MongoliaLiaoning XinjiangHenanShaanxiGuizhouZhejiangAnhuiHeilongjiang GansuSichuanJiangsuHubeiFujianNingxiaJilinYunnanHunanGuangxiJiangxiShanghai TianjinChongqingBeijingGuangdongQinghaiHainan
Fig.2.ThecarbonabatementpotentialinChina'sprovinces(mtCO2).
Fig.3.ThecarbonemissionsefficiencyandpotentialofChina'sprovincesinthe12thFive-YearPlan.
Mongolia, Heilongjiang, Liaoning, Hebei, Shanxi, Shaanxi, Henan, endowment and relatively clean energy consumption structures. This
Hubei, Sichuan and Guizhou, accounting for 65% of the abatement meansthatthetotalcarbonemissionsarelow.Thus,thereislittlepo-
potentialinthe12thFive-YearPlan.Allofthecoalabundantareasare tential for further carbon reduction even if efficiencies are improved.
inthisgroup.Reducingthecarbonemissionsfromthisgroupiscrucial The third group is high efficiency but with high potential (HE-HP):
forachievingtheemissionsreductiontargets.Thesecondgroupislow Shandong,Jiangsu,Anhui,andZhejiang.Thefourprovinceshavelarge-
efficiencybutlowpotentialregions(LE-LP):Gansu,Jilin,Yunnan,and scaleeconomiesandmoresourcesofemissions,althoughtheemissions
Guangxi. These provinces have relatively underdeveloped economies efficiencyishigh,thequantityofpotentialemissionsreductionwillstill
andlowcarbonemissionsefficiencies,buttheyhavelowfossilenergy be large. In 2016, the abatement potential in the four provinces was
208
K.Wangetal. Resources Policy 60 (2019) 203–214
Table2
Analysisofinfluencingfactorsofcarbonemissionsefficiency.
(1) (2) (3) (4) (5) (6)
Dependentvariable lnefficiency
lnNRD −0.041*** −0.029*** −0.042*** −0.117** −0.249*** −0.277***
(0.004) (0.004) (0.003) (0.047) (0.054) (0.050)
lnrational −0.032*** −0.027** −0.029**
(0.006) (0.012) (0.014)
lnadvanced 0.072*** 0.058 0.094**
(0.020) (0.039) (0.047)
lnNRD*lnrational 0.004* −0.002
(0.002) (0.003)
lnNRD*lnadvanced −0.011 −0.013
(0.009) (0.012)
lnPGDP 0.165*** 0.185*** 0.179***
(0.020) (0.021) (0.027)
lnNRD*lnPGDP 0.013*** 0.023*** 0.027***
(0.005) (0.005) (0.005)
lnGOV 0.107***
(0.025)
lnR&D 0.039**
(0.017)
lnEPI 0.143**
(0.067)
lnUR −0.028
(0.040)
Regulation 0.985**
(0.493)
_cons −0.488*** −0.569*** −0.506*** −2.113*** −2.272*** −2.844***
(0.015) (0.017) (0.015) (0.196) (0.210) (0.493)
sigma_u 0.172*** 0.147*** 0.152*** 0.162*** 0.148*** 0.170***
(0.007) (0.007) (0.006) (0.006) (0.006) (0.007)
sigma_e 0.120*** 0.117*** 0.118*** 0.105*** 0.107*** 0.105***
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004)
rho 0.672 0.613 0.625 0.705 0.655 0.723
Prob > chi2 0.000 0.000 0.000 0.000 0.000 0.000
Prob >=chibar2 0.000 0.000 0.000 0.000 0.000 0.000
n 420 420 420 420 420 420
Note:Standarderrorsareshowninparentheses.
***,**,*denotethestatisticalsignificanceat1%,5%and10%separately.
dramaticallyreducedasaresultoftheimprovementsmadeincarbon rationalization and advancement of industrial structure will increase
efficiencies. The last group is high efficiency with low potential (HE- carbonemissionsefficiency,3butthecoefficientofadvancementisno
LP):Beijing,Shanghai,Tianjin,Chongqing,Guangdong,Fujian,Hunan, longer significantincolumn (5).The cross-termsshow thatinthe re-
Jiangxi,Qinghai,Ningxia,andHainan.Theseareallsuccessfulinterms gionswithhigherresourcedependence,theadvancedindustrialstruc-
ofefficiencyimprovementandcarbonreduction. turefails topromotecarbonemissions efficiency,andtherationalin-
dustrialstructuremaynotfurtherreducecarbonemissionsefficiency.
Economic development, however, can mitigate the negative effect
5.2. Influencingfactorsofcarbonemissionsefficiency
ofnaturalresourceabundanceonemissionsefficiency.Thecross-term
coefficient of per capita GDP and its resource dependence is sig-
The panel Tobit regression results2 are shown in Table 2. From
nificantly positive in all regressions, indicating that higher levels of
column (1),it canbe seen thatthe resourcedependence (NRD)hasa
economic development can mitigate the negative impact of resource
significantnegativeeffectonthecarbonemissionsefficiency.Forevery
dependenceonemissionsefficiency.
1% increase in resource dependence, the carbon emissions efficiency
will decrease by about 0.04%. After the two industry structure in-
5.3. Robustnesstest
dicators were added to columns (2)–(3), and cross-terms of industrial
structure variables with resource dependence were added to columns
InTable3,weturntousingtheNRDLtoverifythereliabilityofthe
(4)–(6)andallcontrolvariableswerefurtherincludedincolumn(6),
previous results. The regression results are basically the same as the
thecoefficientsofresourcedependencewerestillsignificantlynegative.
previousfindings.Specifically,intheabsenceofcross-termsandcontrol
However, the values of the coefficients of resource abundance sig-
variables, the emissions efficiency dropped by 0.03% for each 1% in-
nificantlyincreasedintheregressionmodelswithcross-items,whichis
crease in resource dependence in columns (1)-(3). After adding the
inlinewithprevioustheoreticalanalyses.
cross-terms and control variables, for each 1% increase in resource
The results in columns (2)-(6) show that the increase in the
dependence in columns (4)-(6), the emissions efficiency will decrease
byapproximately0.30%.Themagnitudeoftheinfluenceissimilarto
2BeforerunningthepanelTobitmodel,weconductedthecorrelationanalysis thatofTable2.Theresultsofindustrialstructurevariablesandallcross
for independent variables and the test result showed that there is no multi- terms are also basically the same, but the significance level of some
collinearity.Weconductedfixedeffectpaneldatamodelbutdidnotfindsig- coefficientsdeclines.
nificant difference of results from the panel Tobit model. The results are
available upon request. Since the literature shows that panel Tobit model is
more efficiency than the fixed effect model for data used in this paper, we 3Accordingtoequation(7),thesmallerthevalueofRational,themorerea-
conductedouranalysisbasedontheresultsofpanelTobitmodel. sonabletheindustrialstructure.
209
K.Wangetal. Resources Policy 60 (2019) 203–214
Table3
Analysisoftheeffectsofresourcedependenceandindustrialstructureoncarbonemissionsefficiency.
(1) (2) (3) (4) (5) (6)
Dependentvariable lnefficiency
lnNRDL −0.029*** −0.026*** −0.025*** −0.256*** −0.266*** −0.330***
(0.003) (0.003) (0.004) (0.051) (0.054) (0.057)
lnrational −0.052*** −0.004 −0.019
(0.005) (0.014) (0.013)
lnadvanced 0.057*** 0.005 −0.004
(0.020) (0.047) (0.054)
lnNRDL*lnrational 0.006*** 0.003
(0.002) (0.002)
lnNRDL*lnadvanced −0.017** −0.016
(0.009) (0.010)
lnPGDP 0.177*** 0.200*** 0.198***
(0.024) (0.022) (0.029)
lnNRDL*lnPGDP 0.024*** 0.024*** 0.031***
(0.005) (0.005) (0.005)
lnGOV 0.074***
(0.025)
lnR&D 0.029
(0.018)
lnEPI 0.075
(0.069)
lnUR −0.033
(0.040)
Regulation 0.982***
(0.381)
_cons −0.505*** −0.587*** −0.482*** −2.259*** −2.429*** −2.804***
(0.016) (0.018) (0.017) (0.229) (0.233) (0.502)
sigma_u 0.178*** 0.162*** 0.177*** 0.135*** 0.140*** 0.161***
(0.008) (0.011) (0.007) (0.006) (0.006) (0.008)
sigma_e 0.121*** 0.118*** 0.119*** 0.106*** 0.107*** 0.105***
(0.005) (0.004) (0.004) (0.004) (0.004) (0.004)
rho 0.684 0.653 0.687 0.620 0.629 0.700
Prob > chi2 0.000 0.000 0.000 0.000 0.000 0.000
Prob >=chibar2 0.000 0.000 0.000 0.000 0.000 0.000
n 420 420 420 420 420 420
Note:Standarderrorsareshowninparentheses.
***,**,*denotethestatisticalsignificanceat1%,5%and10%separately.
Naturalresource abundancehasacertain correlationwithnatural energy-saving and abatement technologies, and encouraging en-
resourcedependence.However,wherenaturalresourcesareabundant, terprises to eliminate backward production capacity. As in the litera-
natural resource dependence is not necessarily high. The natural re- ture, the roleofR&Din improving carbonemissionsefficiency issig-
sources abundance refers to the quantity of natural resources that a nificant.Theimpactofenergypricesoncarbonemissionsefficiencyis
country or region can use for social and economic development; the significantly positive because the rise of energy prices will force en-
natural resources dependence refers to the role of resource-based in- terprises to introduce energy-saving technologies or reduce energy
dustries in the development of regional economy (Sun and Ye, 2012; consumption. The coefficient of urbanization is negative, but not sig-
Wu et al., 2018). To distinguish the effect of resource endowment to nificantinTables2and3,andweaklysignificantinTable4.Thereason
thatoftheresourcedependence,weuseFEE(fossilenergyendowment) couldbethattheeconomicactivitiesarecentralizedandtheenergyhas
as the main dependent variable, which is the ratio of production to beenconsumedmassivelyintheprocessofurbanizationbutthescale
consumptionoffossilfuelstomeasurethelevelofresourceendowment. effectandtechniquespill-overeffectcannotbereflectedinshort-term.
Fromtheresultsincolumns(1)to(3)inTable4,itcanbeseenthat Thesignificanteffectofenvironmentalregulationoncarbonemissions
resource endowment have a significant negative correlation with efficiencysuggestthattheenergy-savingandemissions-reductiontarget
carbonemissionsefficiency.Foreach1%increaseinresourceendow- isaneffectivetoolforcontrollingemissions.
ment,theemissionsefficiencywilldecreasebyabout0.12%.Inaddi-
tion,similartothepreviousresults,theincreaseintherationalization 5.5. Medium-termandlong-termeffects
and advancement of the industrial structure will improve the carbon
emissions efficiency. When adding the cross-terms of the industrial Theeffectofmedium-runandlong-runareindeedimportanttothe
structurevariableandresourceendowmentincolumns(4)-(6),itcanbe analysisoftheimpactofnaturalresourceabundanceoncarbonemis-
seenthatthedistortionofresourceendowmenttoindustrialstructureis sionsefficiency.Accordingtothemethodinthepreviousstudies(Arin
notasseriousasthatofresourcedependence. andBraunfels,2018;Knelleretal.,1999;Wuetal.,2018),thisstudy
furthercarriesoutthepanelTobitmodelwith5-yearmovingaverage
5.4. Testofcontrolvariableseffects datatoexaminethemedium-termeffectofnaturalresourceabundance
andindustrialstructureoncarbonemissionsefficiencyandcarriesout
As shown in Tables 2–4, the coefficients of GOV are significantly the Tobit model to analyze long-term effect with the cross-sectional
positive since fiscal expenditure can finance the improvement of dataof14-year(2013–2016)averages.
210
K.Wangetal. Resources Policy 60 (2019) 203–214
Table4
Analysisoftheeffectsofresourceendowmentsandindustrialstructureoncarbonemissionsefficiency.
(1) (2) (3) (4) (5) (6)
Dependentvariable lnefficiency
LnFEE −0.117*** −0.117*** −0.121*** −0.289** −0.406*** −0.244**
(0.013) (0.012) (0.014) (0.115) (0.139) (0.120)
lnrational −0.016*** −0.020** −0.009
(0.006) (0.009) (0.010)
lnadvanced 0.110*** 0.064** 0.081***
(0.019) (0.028) (0.031)
lnFEE*lnrational 0.009 0.025***
(0.008) (0.009)
lnFEE*lnadvanced −0.035 0.035
(0.023) (0.029)
lnPGDP 0.115*** 0.117*** 0.118***
(0.012) (0.011) (0.023)
lnFEE*lnPGDP 0.023** 0.035*** 0.025**
(0.012) (0.014) (0.012)
lnGOV −0.002
(0.024)
lnR&D 0.025
(0.017)
lnEPI 0.092
(0.068)
lnUR −0.081**
(0.040)
Regulation 0.850**
(0.376)
_cons −0.452*** −0.485*** −0.431*** −1.578*** −1.537*** −1.951**
(0.012) (0.017) (0.010) (0.113) (0.112) (0.490)
sigma_u 0.151*** 0.149*** 0.155*** 0.154*** 0.147*** 0.148***
(0.007) (0.007) (0.007) (0.006) (0.006) (0.006)
sigma_e 0.116*** 0.116*** 0.115*** 0.107*** 0.107*** 0.104***
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004)
rho 0.630 0.624 0.644 0.677 0.654 0.670
Prob > chi2 0.000 0.000 0.000 0.000 0.000 0.000
Prob >=chibar2 0.000 0.000 0.000 0.000 0.000 0.000
n 420 420 420 420 420 420
Note:Standarderrorsareshowninparentheses.
***,**,*denotethestatisticalsignificanceat1%,5%and10%separately.
Tables 5 and 6 show that the impacts of resource dependence on (shortruneffect).Thecoefficientofurbanizationbecomesignificantly
carbon emissions efficiency keep unchanged and are still significant positive from non-significant, suggesting that although the impact of
negative in both the medium-term and the long-term.4 Similar to the urbanizationoncarbonemissionsefficiencyisunclearintheshortand
results in the models with the annual data, the rationalization and medium-term,inthelongrun,theeffectagglomerationandtechnology
advanced of industrial structure promote the carbon emissions effi- spillover ofurbanization significantly improves carbonemissions effi-
ciencyinthemedium-termandlong-term(seecolumn(2)–(3)inTables ciency.
5 and 6).In thecase ofintroducing cross-items, the positiveeffect of It needs to be noticed that the coefficient of energy-saving and
rationalizationissignificantlyweakened,andthecoefficientsofcross- emissions-reduction target is significantly negative in the long-term
items furthershowthatintheresourcedependence regions,eventhe model, while it is non-significant in the medium-term model and is
developmentofindustrialstructuretowardsrationalandadvancedstill significantlynegativeinthemodelswithannualdata.Duetothefiscal
cannot improve carbon emissions efficiency (see column (4)–(6) in decentralization system,theinterestsbetweentheChina'scentraland
Tables5and6). localgovernmentsarenotentirelyconsistentandthepursuitofrapid
Inthelongrun,theresourcesdependenceofeconomicdevelopment economic growth is the primary objective of local governments for a
inresource-basedregionswillinevitablyleadto“lock-ineffect”,which long time. Under the constraints of energy saving and emissions re-
hinderstheadjustmentandevolutionofregionalindustrialstructures. duction targets, local governments are more likely to control current
Thelongerthisdevelopmentmodelismaintained,thehigherresource energy consumption through some government interventions and re-
dependence will be and the greater negative impact on carbon emis- sulting in higher carbon emissions efficiency in the short term.
sions efficiency will be. Even when resources are exhausted, the However, if the target constraints cannot be translated into effective
transformationofdevelopmentmodelarestillverydifficultduetothe environmentalregulationstopromoteinnovationandindustrialstruc-
lackofalternativeindustries. ture upgrading, they cannot continuously and effectively improve
Forthecontrolvariables,thecoefficientsofeconomicdevelopment, carbonemissionsefficiencyinthelongrun.
thegovernmentintervention,thetechnologyinnovationandtheenergy Insummary,theempirical analysisshowsthatrationalizationand
price are consistent with those in the models with the annual data advancementoftheindustrialstructureimprovethecarbonemissions
efficiencyindifferenttimehorizon.However,thesignificantlynegative
coefficientsofresourceabundanceinallmodelsandthecoefficientsof
4Wealsoexaminethemedium-termandlong-termeffectwiththevariableof interaction-terms all show that higher resource dependence not only
resourcedependenceinemployment(NRDL)andthefossilenergyendowment hinders the improvement of carbon emissions efficiency directly but
(FEE). The results are consistent with that in the Tables 5 and 6 except the alsoaffectsthecarbonemissionsefficiencybydistortingtheindustrial
coefficientsofFEEbecomelesssignificantinthelong-termeffectmodelwith structureandfurtherexacerbatesinefficiency.
thecross-items.
211
K.Wangetal. Resources Policy 60 (2019) 203–214
Table5
Middle-termanalysisofinfluencingfactorsofcarbonemissionsefficiency.
(1) (2) (3) (4) (5) (6)
Dependentvariable lnefficiency
lnNRD −0.014*** −0.016*** −0.012*** −0.237** −0.266*** −0.402***
(0.003) (0.003) (0.003) (0.047) (0.045) (0.040)
lnrational −0.011*** −0.003 0.071***
(0.004) (0.010) (0.011)
lnadvanced 0.134*** 0.080** 0.175***
(0.012) (0.033) (0.036)
lnNRD*lnrational 0.009*** 0.019***
(0.002) (0.002)
lnNRD*lnadvanced 0.002 −0.006
(0.008) (0.009)
lnPGDP 0.152*** 0.168*** 0.294***
(0.020) (0.017) (0.024)
lnNRD*lnPGDP 0.026*** 0.028*** 0.043***
(0.005) (0.004) (0.004)
lnGOV 0.090***
(0.017)
lnR&D 0.044***
(0.011)
lnEPI 0.586***
(0.107)
lnUR −0.014
(0.048)
Regulation 0.005
(0.014)
_cons −0.355*** −0.386*** −0.391** −1.839*** −1.937*** −5.756***
(0.013) (0.016) (0.011) (0.193) (0.176) (0.661)
sigma_u 0.196*** 0.134*** 0.130*** 0.193*** 0.195*** 0.142***
(0.006) (0.004) (0.003) (0.006) (0.005) (0.004)
sigma_e 0.078*** 0.070*** 0.067*** 0.068*** 0.065*** 0.054***
(0.003) (0.003) (0.003) (0.003) (0.003) (0.002)
rho 0.864 0.788 0.791 0.889 0.899 0.873
Prob > chi2 0.000 0.000 0.000 0.000 0.000 0.000
Prob >=chibar2 0.000 0.000 0.000 0.000 0.000 0.000
n 300 300 300 300 300 300
Note:Standarderrorsareshowninparentheses.
***,**,*denotethestatisticalsignificanceat1%,5%and10%separately.
Table6
Long-termanalysisofinfluencingfactorsofcarbonemissionsefficiency.
(1) (2) (3) (4) (5) (6)
Dependentvariable lnefficiency
lnNRD −0.071*** −0.068*** −0.064*** −0.856** −0.751*** −0.594**
(0.018) (0.018) (0.017) (0.327) (0.255) (0.247)
lnrational −0.051** 0.019 0.188**
(0.022) (0.040) (0.069)
lnadvanced 0.173*** −0.074 0.455**
(0.062) (0.195) (0.217)
lnNRD*lnrational 0.013 0.018
(0.008) (0.015)
lnNRD*lnadvanced −0.052 0.075
(0.045) (0.077)
lnPGDP 0.383*** 0.360*** 0.218
(0.127) (0.102) (0.132)
lnNRD*lnPGDP 0.081** 0.068*** 0.059**
(0.032) (0.024) (0.026)
lnGOV 0.315***
(0.098)
lnR&D 0.229***
(0.064)
lnUR 0.842**
(0.352)
Regulation −0.660***
(0.137)
_cons −0.606*** −0.711*** −0.566** −4.384*** −4.184*** −3.134*
(0.077) (0.090) (0.076) (1.270) (1.046) (1.582)
n 30 30 30 30 30 30
Note:Standarderrorsareshowninparentheses.
***,**,*denotethestatisticalsignificanceat1%,5%and10%separately.
212
