芬兰经济研究所-企业动力在绿色转型中的作用：芬兰制造业的碳生产率分解（英）-2022.12-36正式版

WorkingPapers|9913.12.2022TheRoleofFirmDynamicsintheGreenTransitionCARBONPRODUCTIVITYDECOMPOSITIONINFINNISHMANUFACTURINGAbstractThispaperinvestigatestheimportanceoffirmdynam-ics,includingentryandexitandtheallocationofcar-bonemissionsacrossfirms,onthegreentransition.Usingthe2000–2019firm-levelregisterdataongreen-housegasemissionsmatchedwiththeFinancialState-NataliaKuosmanenmentdataintheFinnishmanufacturingsector,weex-EtlaEconomicResearch,Finlandaminethesourcesofcarbon-productivitygrowthandnatalia.kuosmanen@etla.fiassesstherelativecontributionsofstructuralchangeandfirmdynamics.WefindthatcontinuingfirmswereTerhiMaczulskij(Correspondingauthor)themaindriversofcarbonproductivitygrowthwhere-EtlaEconomicResearch,FinlandandasthecontributionofenteringandexitingfirmswasIZAInstituteofLaborEconomics,Germanynegative.Inaddition,theallocationofemissionsacrossterhi.maczulskij@etla.fifirmsseemstobeinefficient;itsimpactoncarbonpro-ductivitygrowthwasnegativeoverthestudyperiod.Moreover,wefindthatthereisapositiverelationshipSuggestedcitation:betweenlabor-intensivefirmsandcarbonproductivi-Kuosmanen,Natalia&Maczulskij,Terhitybutthatfirmswithalargermarketsharetendtobe(13.12.2022).“TheRoleofFirmDheGreenTransition:CarbonProductivityDecompositioninFinnishManufacturing”.ETLAWorkingPapersNo99.http://pub.etla.fi/ETLA-Working-Papers-99.pdfETLAWorkingPapers|No99TiivistelmäRakennemuutosjavihreäsiirtymä:hiilituottavuudenosatekijätSuomenteollisuudessaTässäartikkelissatutkitaanteollisuudenrakennemuu-toksenjayritystenuusiutumisenyhteyttähiilituottavuu-denkasvussa.Tutkimuksessahyödynnetäänteollisuu-denyritystasontietojakasvihuonekaasupäästöistäjatilinpäätöstiedoistavuosille2000–2019.Menetelmänäkäytetäänniinkutsuttuahajotelmamenetelmää,jonkaavullahiilituottavuudenkehitysvoidaanjakaakolmeenosaan:keskimääräiseenjatkavienyrityksienhiilituot-tavuudenmuutokseen,uusienjapoistuvienyritystenkontribuutioonsekäkasvihuonekaasupäästöjenkoh-dentumiseenyritystenkesken.Tulostenperusteellatoi-mintaansajatkaneetyrityksetolivathiilituottavuudenkasvuntärkeinveturi,kuntaasosatehokkaimmistayri-tyksistäolijostainsyystäkannattamattomiajapoistuimarkkinoilta.Lisäksikasvihuonekaasupäästöjenkoh-dentuminenyritystenkeskennäyttääolevantehotonta.Tämätarkoittaasitä,ettäpäästötkohdentuivatsaastut-tavimpiinyrityksiinjapäästöjenvähentäminentapahtuijoennestäänvähäpäästöistenyritystentoimesta.Lisäksihiilituottavuuskulkeekäsikädessätyöntuottavuudenkanssa,muttasuhdekilpailukykyynonpäinvastainen.

Ph.D.NataliaKuosmanenisaChiefResearchScientistatETLAEconomicResearch.Ph.D.TerhiMaczulskijisaChiefResearchScientistatETLAEconomicResearchandaResearchFellowatIZA.MMTNataliaKuosmanenonElinkeinoelämäntutkimuslaitoksentutkimuspäällikkö.KTT,FM,dosenttiTerhiMaczulskijonElinkeino-elämäntutkimuslaitoksentutkimuspäällikkösekäIZA:nResearchFellow.Acknowledgements:Theauthorsgratefullyacknowl-edgefinancialsupportfromtheTTFoundation.DatausedintheempiricalapplicationofthispaperwereprovidedbyStatisticsFinland.Kiitokset:HaluammekiittäähankettarahoittanuttaTeollisuudenjaTyönantajainKeskusliiton(TT)-sääti-ötä.Keywords:Carbonproductivity,Decomposition,Firmdynamics,Firm-leveldata,ManufacturingAsiasanat:Hiilituottavuus,Hajotelmamenetelmä,Rakennemuutos,Yritystasonaineisto,TeollisuusJEL:D24,L60,Q542TheRoleofFirmDynamicsintheGreenTransition:CarbonProductivityDecompositioninFinnishManufacturingIntroductionAmajorcauseofclimatechangeisgreenhousegas(GHG)emissions(Solomon,2007).AccordingtotheGovernmentProgramme,Finlandaimstobecarbonneutralbyyear2035.1Further,thenewClimateChangeAct,whichcameintoforcein2022,hassetambitioustargetsforreducingemissionsby80%by2040,comparedtothelevelsin1990(MinistryoftheEnvironment,2022).Tomeetitscarbonneutralitygoals,FinlandmuststrengthenitseffortstofightclimatechangeandreduceCO2emissions.Therefore,itiscrucialtounderstandthedrivingforcesbehindtheemissionchanges.ThecontributionofthemanufacturingsectortoFinland’stotalGHGemissionswas22%in2020(StatisticsFinland,2022).Figure1depictstheGHGemissionsgeneratedbythemanufacturingsectorbetween2000and2019.Althoughthissectorisresponsibleforthegreatmajorityoftotalemissionsatthenationallevel,ithasbeenabletodecreaseitsemissionsbyapproximately2%annuallybetween2000and2019.Theemissionlevelsremainedfairlyconstantbetween2000and2008:approximately18milliontonnesofcarbondioxideequivalent(CO2eq.)annually.Asthe2008globalfinancialcrisisturnedintoaneconomiccrisisintheeurozonecountries,industrialoutputdeclinedsharply,whichledtoasharpdeclineinemissions.Since2009,therehasbeencontinuousemissionreduction;althoughthisreductionispartlyduetothecontinuousdeclineinindustrialoutput,itisalsoduetoimprovedcarbonuse.[Figure1here]Accordingtotherequirementsofsustainabledevelopmentandeconomicgrowth,improvingcarbonproductivityisakeypathwaytoaddressingclimatechange(HeandSu,2011;LiandClimateNeutralFinland2035,MinistryoftheEnvironment:https://ym.fi/en/climate-neutral-finland-2035.3ETLAWorkingPapers|No99Wang,2019).CarbonproductivityisaperformancemeasuregenerallydefinedaseconomicoutputperunitofGHGemissions(e.g.,Sunetal.,2021;Murshedetal.,2022).2Recentworkonthisissuehasusedmostlymacro-leveldataoncountriesandregionstodecomposechangesincarbonproductivity(orinitsinverse,carbonintensity)intocomponentssuchasefficiencyandtechnologicalinnovation(seee.g.,MengandNiu,2012;HuandLiu,2016;Wangetal.,2018;Baietal.,2019).Thelackofsuitablemicrodatahaslimitedempiricalresearchinthisfield.Althoughthenumberoffirm-levelanalyseshasincreasedinrecentyears,theyarefocusedmostlyonexaminingthedeterminantsoffirm-levelfactorsandcarbonproductivitygrowth(e.g.,CaoandKarplus,2014;Jungetal.,2021;Bagchietal.,2022).Yet,nostudieshaveaddressedtheroleoffirmdynamicsinthegreentransitionor,morespecifically,theeffectsofstructuralchangeoncarbonproductivitygrowth.Thepurposeofourpaperistofillthisgap.Thisissueishighlyrelevantfordesigningeffectivepolicyresponsestoreachstringentclimategoals.Abetterunderstandingoftheunderlyingmechanismmayhelptoimproveenvironmental-policymeasuresandpromotethegreentransition.Inthispaper,weusetheoriginalfirm-levelemissionsdataonalltheFinnishmanufacturingfirmsthatbelongtotheEUEmissionsTradingSystem(EUETS).Toassesscarbonproductivityatthemicrolevelofthefirm,wematchtheadministrativeemissionsdatawiththefirm-levelFinancialStatementdatausinguniquefirm-identificationcodes.Notethatalltheinformationisregisterbased,whicheliminatestherisksofnonresponseandmeasurementerrorsassociatedwithself-reportedmeasures.Thedatainclude5,269firm-yearobservationsovertheperiod2000–2019.Weapplyastructuralchangedecompositionofcarbonproductivity,whichisbasedontheseminalstudybyOlleyandPakes(1996)andTheconceptofcarbonproductivity,definedastheratioofgrossdomesticproducttoemissionsatthenationallevel,wasfirstintroducedbyKayaandYokobori(1999).4TheRoleofFirmDynamicsintheGreenTransition:CarbonProductivityDecompositioninFinnishManufacturingitsextensionbyKuosmanenandKuosmanen(2021).3Thismethodenablesconsistentaggregationofproductivitymeasuresatthefirmleveltothoseattheindustrylevelandisapplicabletoboththelevelsandchangesofproductivityovertime.OurresultsshowaclearU-shapedtrendincarbonproductivitygrowthbetween2000and2019.Wefindthatthecontributionoffirmsthatcontinuedtooperateinthesameindustryispositiveovertheanalyzedperiodsandthatithasincreasedovertime.However,thecomponentsofentry/exitandtheallocationofemissionsacrossfirmsnearlycanceloutthepositiveeffectofnonswitchingcontinuingfirms.Thenegativecontributionofallocationimpliesthatemissionswereallocatedtowardlessproductivefirms.Itisalsoconcerningthatexitingfirmsexhibithighercarbonproductivitycomparedtosurvivingandnewenteringfirms.Moreover,wefindthatfirm-specificcharacteristics,suchasnumberofemployeesandlaborproductivity,arepositivelyrelatedtocarbonproductivitywhereasfirms’turnoverandmarketsharearenegativelyrelatedtocarbonproductivity.Therestofthepaperisorganizedasfollows.Section2situatesouranalysisinthecontextofrelevantliterature.Section3describesthedecompositionmethod.Section4presentsthedatausedinthestudy.Section5and6presentthedecompositionresultsandtheregressionresults,respectively,andSection7presentsourconclusions.Kuosmanen,Maczulskij,andKuosmanen(2022a)haveexaminedcarbonproductivitygrowthusingdataontheFinnishenergysector.5ETLAWorkingPapers|No99LiteraturereviewToagreatextent,currentresearchonthedeterminantsofcarbonproductivitygrowthisconductedusingmacro-leveldataoncountriesandregions.Forinstance,alargebodyofliteraturehasanalyzedtrendsincarbonproductivitygrowthacrosscountries(HeandSu,2011;Ekinsetal.,2012;Baietal.,2019;Xiaoetal.,2020)orinasingleeconomy,oftenaggregatedbyindustryorregionaldata(e.g.,LiandWang,2019).Thesestudiesdemonstratethatalthoughcarbonproductivityhasincreasedglobally,ithasincreasedmuchmoreindevelopedthanindevelopingcountries(HeandSu,2011;Baietal.,2019).Similarly,Xiaoetal.(2020)findthatconsumption-basedcarbonintensity(theinverseofcarbonproductivity)hasbeenhigherindevelopingcountriesandlowerindevelopedcountries.Further,Baietal.(2019)appliedconvergenceanalysisandaprobitmodeltocountry-leveldatatoexaminewhichdeterminantsconvergetodifferentgroupsofcarbonproductivitygrowth.TheirresultsindicatedthatR&DinvestmentsandGDPpercapitatendtoconvergetothegroupwithhighcarbonproductivity,whereaseconomieswithforeign-tradedependenceandhigherenergyintensitytendtoconvergetothelow-carbon-productivitygroup.LiandWang(2019)appliedspatial-analysistechniquesandpanel-datamodelstoregionaldataandquantifiedthevariationsincarbonproductivityacrossChineseprovinces.Theyfoundthattechnologylevel,tradeopenness,GDPpercapita,andforeigndirectinvestmentsenhancecarbonproductivity.Thereisalsoapositivelinkbetweenenvironmental-taxreformandcarbonproductivityinEUcountries(Ekinsetal.,2012).Somestudieswentfurtheranddecomposedchangesincarbonproductivityorcarbonintensityintounderlyingcomponents,suchastechnicalefficiencyandtechnologicalchange(e.g.,MengandNiu,2012;HuandLiu,2016).Thesestudieshaveprimarilyappliedinsights6TheRoleofFirmDynamicsintheGreenTransition:CarbonProductivityDecompositioninFinnishManufacturingfromindexdecompositionanalysisorproductiontheory.4Thefindingsofmanystudiesshowthatcarbon-productivitygrowthhasresultedmainlyfromtechnologicalchange(MengandNiu,2012;HuandLiu,2016;Wangetal.,2018;Baietal.,2019)whereastheglobalreductionincarbonintensityhasresultedprimarilyfromdecreasedenergyintensity(Liuetal.,2022)andimprovementsinthethermalefficiencyofelectricitygeneration(AngandSu,2016).Moreover,bothcapitalandlabor-energysubstitutionsandenergystructurehavedecreasedthecarbon-intensitygapbetweenJapanandChina(Lietal.,2022).Althoughstudiesthatfocusonthemacrolevelofcountriesandregionsclearlyprovideimportantinsights,itisimportanttounderstandthedrivingforcesoftheevolutionofcarbonproductivityfromtheperspectiveofthemicroleveloffirms.Eventhoughrecentyearshavewitnessedagrowthinfirm-levelstudiesofthisissue,onlyasmallbodyofresearchhasexamineditindepth.Somestudiesfocusedoncorrelationanalysesofvariousfirm-levelfactorsandcarbonproductivitygrowth(e.g.,CaoandKarplus,2014;Jungetal.,2021;Bagchietal.,2022).CaoandKarplus(2014)examinedthefirm-leveldeterminantsofcarbonintensityusingdataonChinesefirms.Theresultsshowedthatchangesincarbonintensityweredrivenlargelybychangesinenergyusebutfirmsizeandfirmownershipalsoplayedarole.Forexample,state-ownedfirmsexhibitedhighercarbonintensitycomparedtojointventures.Brännlundetal.(2014)examinedtheeffectofclimatepolicyandfoundthatCO2taxhasbeenasignificantreasonforthedeclineinSwedishmanufacturingfirms’carbonintensity.Energyconsumptionisthemaincauseofemissions.Therefore,energyintensityandcarbonintensityarerelatedbutnotsynonymous.Althoughdifferentdecompositionanalysesarealsousedtoexamineenergyintensity(e.g.,LiuandAng,2003;LinandDu,2014;TanandLin,2018),ourpaperconcentratesonstudiesthatexaminecarbonproductivity(orcarbonintensity,itsinverse).7ETLAWorkingPapers|No99Morerecently,RichterandSchiersch(2017)testedthehypothesisthatexportingfirmsperformbetterenvironmentallythannonexportingfirms.TheirresultsshowedapositiverelationshipbetweenGermanfirms’exportintensityandcarbonproductivity.Jungetal.(2021)showedthatcarbonproductivityhasbeenhigherinfirmsundertheemissions-tradingscheme.Theyalsofoundthatcarbonproductivityhasbeenhigherinmoreprofitableandinnovativefirmsandinfirmsinwhichthemanagementhasexperienceinenvironmentalfields.Bagchietal.(2022)useddataonfirmsinthemanufacturingsectorofIndiaandfoundthatespeciallyexportandtechnologicalintensitiesenhancecarbonproductivity.Lastly,CoderoniandVanina(2022)useddataonItalianfarmsandfoundanonlinearrelationshipbetweencarbonproductivityandfarms’economicperformance.Regardingtheempiricalresearchontheroleofmicro-leveldynamicssuchasthemarketentryandexitoffirms,studiesofindustryswitchingandoftheallocationofemissionsacrossfirmsare,tothebestofourknowledge,stilllacking.Ourstudyaddressesthisgapbyexaminingthecontributionofsuchmicro-levelstructuralchangesoncarbonproductivityusinguniquedataonfirmsinFinland’smanufacturingsector.Carbonproductivitydecomposition3.1DecompositionofproductivitylevelProductivitydecompositioninlevelsmeasurescomponentsofaggregateproductivity(e.g.,theproductivityofanindustryorasector).OnesuchapproachwasoriginallyproposedbyOlleyandPakes(1996),whodecomposedindustry-levelproductivityintothesumofanunweightedaverageproductivitylevelofallfirmsandacovariancecomponentrepresentingtheallocationofresourcesacrossfirms.AsinOlleyandPakes(1996)butforourcontextof8TheRoleofFirmDynamicsintheGreenTransition:CarbonProductivityDecompositioninFinnishManufacturingcarbonproductivity,wefirstdefinetheaggregatecarbonproductivityofasectorinperiodtasCt.Assumingconsistentaggregation,thesector’scarbonproductivityisashare-weightedaverageoffirm-levelcarbon-productivitymeasurescit,thatis,��∑�������. (1)InEq.(1),��������istheshareoffirmiinthetotalGHGemissionsofthesectorinyeart,and���������isthecarbonproductivityoffirmiinperiodtdefinedastheratioofthefirm’svalueadded(��)toitsGHGemissions(��).Thesector’scarbonproductivitycanbesplitintotwocomponents:��̄�∑����̄�cov,�������,(2)wherē�istheunweightedaverageofthecarbonproductivityofallthefirmsobservedinperiodtandcov��,���isacovariancetermthatcapturestheallocationofemissionsacrossfirms.Anegativecovariancetermindicatesthatlow-productivityfirmstendtohavealargershareofemissionsthanhigh-productivityfirms,whereasapositivecovariancetermindicatesthathigh-productivityfirmstendtohavealargershareofemissionsthanlow-productivityfirms.AsEq.(2)indicates,thesector’scarbonproductivitycangroweitherbecauseofincreasesintheaveragecarbonproductivityofallthefirmsorbecauseofahighercovariancevalue,whichrepresentsashiftofemissionsfromlow-productivitytohigh-productivityfirms.TheOlley–Pakesdecomposition,however,doesnotexplicitlyconsidertheentryandexitoffirmsbutattributesthesefirmstothecovarianceterm.FollowingKuosmanenandKuosmanen(2021),weclassifythesector’sfirmsintofourmutuallyexclusivegroups:9ETLAWorkingPapers|No99entrants(E)inperiodt+1,exitingfirms(X)observedinperiodtbutnotinperiodt+1,andallcontinuing(surviving)firmsS,whicharesubdividedintocontinuingnonswitchingfirms(Sn)andcontinuingindustry-switchingfirms(S–Sn).5Applyingthisclassification,thesector’scarbonproductivityinperiodtcanbewrittenasasumoffourcomponents,asfollows:���̄���̄��̄����̄��̄�����̄������������������(3).Thefirstcomponentontheright-handsideofEq.(3)istheaveragecarbonproductivityofnonswitchingcontinuingfirms.Thesecondcomponentdescribestheeffectofindustryswitching,whichisverycommoninmanyindustries(Kuosmanenetal.,2022b).Wedefineindustryswitchingasanobservedchangeinthe5-digitindustryclassificationofthefirminthemanufacturingsector.Thiscomponentisidentifiedbycomparingtheaveragecarbonproductivityofallthecontinuingfirmsandthatofthenonswitchingcontinuingfirms.Notethatwhentheswitchingeffectisnotconsideredexplicitly,itscontributionismixedwiththeeffectsofcontinuingnonswitchingfirmsandthecontributionofentryandexit.Thethirdcomponentcapturestheproductivityimpactofentryandexitbycomparingtheaveragecarbonproductivityofallthefirmsandthatofthecontinuingfirms.Finally,thefourthcomponentcapturestheallocationofemissionsacrossallthefirms.Wemeasurethiscomponentasthedifferencebetweenthesector’scarbonproductivityandtheunweightedaveragecarbonproductivityofallthefirms.Asimilarclassificationisusedinotherproductivitystudies,suchasMaliranta(2003),BöckermanandMaliranta(2007),HyytinenandMaliranta(2013),andMalirantaandMäättänen(2015).10TheRoleofFirmDynamicsintheGreenTransition:CarbonProductivityDecompositioninFinnishManufacturing3.2DecompositionofproductivitychangeDecompositionofproductivitychangemeasuressourcesofaggregateproductivitygrowth(Bailyetal.,1992;GrilichesandRegev,1995;MelitzandPolanec,2015).Usingthesameclassificationoffirms,wedecomposethesector’scarbonproductivitygrowthintofourcomponentsexpressedaspercentagechanges:�������̄����̄����̄��̄����̄���̄,(4)������������̄��������̄��������̄�������̄�������̄������������̄�wheresubscriptSreferstothesurvivingfirmsandSnreferstothesurvivingnonswitchingfirmsinperiodstandt–1.Thefirstcomponentontheright-handsideisthecarbonproductivitychangeofthecontinuingnonswitchingfirms.Thesecondcomponentmeasuresthecontributionofthecontinuingindustry-switchingfirmstoaggregatecarbonproductivitygrowth.Thethirdcomponentcapturesthecontributionoffirms’entryandexit,andthefourthcomponentcapturestheallocationofemissionsacrossfirms.Thus,thesector’scarbonproductivitygrowthisthesumofthesefourcomponents.Data4.1DatasourcesThisstudyfocusesontheFinnishmanufacturingsectorduringtheperiod2000–2019.Theanalysisisbasedonfirm-levelvaluesofcarbonproductivitycomputedastheratioofafirm’svalueadded(VA)toitsGHGemissions.Observationswithmissingvaluesandobservations11ETLAWorkingPapers|No99withzeroemissionswereexcluded,becausecarbonproductivitycannotbecomputedforthoseobservations.Thehigherthevalueofcarbonproductivity,themoreefficientthefirmisinitsuseofemissions.Toobtaintherequiredmicrodata,werelyontwodatasources.TheGHGemissionmicrodatacomefromtheNationalGreenhouseGasInventoryofStatisticsFinland.6ThisinventoryannuallyreportsGHGemissionsandremovalsandprovidesaninformationbasefortheplanningandmonitoringofclimatepolicy.UndertheUnitedNationsFrameworkConventiononClimateChange,theKyotoProtocol,andEUregulations,StatisticsFinlandisthegeneralauthorityfortheofficialstatisticsofFinlandandisresponsibleforGHG-inventorysubmissions.TheemissionsdataincludeunitsthatbelongtotheEUETSandreportbothcarbondioxideandGHGemissionsinCO2eq.attheestablishmentandfirmlevelsannually.Inthisstudy,weutilizefirm-leveldataandGHGemissionsinCO2eq.ComparingourGHG-emissionsdatawithEurostat’saggregatefiguresforthemanufacturingsector’sGHGemissions,wefindthatourdata’scoverageisabout99%thatofEurostat’sdata.OuremissionsdataarethusrepresentativeoftheentireFinnishmanufacturingsector.ThedataonVAaredrawnfromStatisticsFinland’sFinancialStatementpaneldata.Thesepaneldataprovideexhaustivecoverageofalltheindependentbusinessenterprisesinalmostallindustriesandincludethemostessentialloss-and-profit-accountandbalance-sheetdataoffirms(e.g.,industrycode,numberofpersonnel,VA,andotherfirm-relatedinformation).Allenterpriseswithatleast20employeesareincludedinthedirectdatacollection,andthedataonsmallerenterprisesandnonrespondententerprisesarederivedfromadministrativerecords,suchasbusinesstaxationregisters.Linkingthesetwosourcesofinformationthroughfirms’IDcodesallowsustocreateauniquematcheddatasetinwhichfirm-levelemissionrecordsarecombinedwiththeInformationontheGreenhouseGasInventory:https://www.tilastokeskus.fi/tup/khkinv/index_en.html.12TheRoleofFirmDynamicsintheGreenTransition:CarbonProductivityDecompositioninFinnishManufacturingbusiness-registerdatasetscontainingdetailedinformationonfirms’financialstatistics.AftermatchingtheemissionsdatawiththeFinancialStatementpaneldata,wearriveat5,269yearlyobservationsrepresenting602manufacturingfirmsoperatingin2000–2019.WedescribeoursampleinTable1.VAispresentedinmillionsofeuros,GHGemissionsinthousandsoftonnesofCO2eq.,andcarbonproductivityinthousandsofeurospertonneofCO2eq.VAandcarbonproductivityweredeflatedusingtheGDPdeflatorforFinland(with2015asthebaseyear)toallowforcomparisonacrossyears.Thenumberofobservationsinthesubsamplesvariesbetween247and276annually.Theaveragecarbonproductivityforthefirmswas31,000eurospertonneofCO2eq.in2000andincreasedto694,000eurospertonneofCO2eq.in2019.Asthetablehighlights,therearelargevariationsbetweentheaveragesduringthestudyperiod.[Table1here]4.2CarbonproductivityofthemanufacturingsectorFigure2plotsthecarbonproductivityofthemanufacturingsectorfortheperiod2000–2019calculatedbasedonourfirm-leveldata.Asnotedabove,ouremissionsdataarerepresentativeoftheentireFinnishmanufacturingsector.ItshouldbenotedthatthereisaclearU-shapedtrendincarbonproductivityovertime:itdecreasedconsiderablyfrom2000to2009andthenincreasedinmorerecentyears.Despitethislatterpositivetrend,thesector’scarbonproductivityhasyettonotreacheditshighestvaluethatwasobservedin2000(890eurospertonneofGHGemissions).TheunderlyingtrendsofVAandGHGemissionsofthemanufacturingsectorarepresentedinFigure3.Thefigurerevealsthattheincreasingtrendincarbonproductivity13ETLAWorkingPapers|No99after2009isnotsolelyaphenomenonofdecreasedemissionsandimprovementsinenvironmentalperformancebutthatitisalsoduetothedecreasingVA.AsmentionedintheIntroduction,thismaybetheresultofthedeclineofindustrialoutputbecauseofthefinancialcrisistakingplaceatthattime.[Figures2and3here]4.3AveragecarbonproductivitybysubperiodsandsubgroupsThestudyperiodcoversthe20yearsfrom2000to2019.Tobettercapturetheeffectsoffirmdynamicsoncarbonproductivity,wefocusonthreesubperiodslastingsixtosevenyears:2000–2006,2007–2012,and2013–2019.Weusethesesubperiodsforthreereasons.First,wechoosemedium-runtimeperiodsbecauseshort-runanalysis(e.g.,analysisofyearlychanges)isunabletocapturestructuralchangessuchasfirmentry,firmexit,andindustryswitching.Second,theperiodsincludedifferenteconomicup-anddownturns,includingthegrowthperiod,theGreatRecession,andthefollow-uprecessionandslowrecovery.Third,theseperiodsarecloselylinkedwiththefirstthreephasesoftheEUETS:thepilotphase,orphase1(2005–2007),phase2(2008–2012),andphase3(2013–2020).7RecallthatthedecompositionofcarbonproductivitypresentedinSection3isbasedonpartitioningthesampleoffirmsintofourmutuallyexclusivesubgroups.BeforeweThefirm-leveldataontheFinnishGHGinventoryreachbackto1999,buttheEUETSwaslaunchedinJanuary2005.Becausetheemissionallowanceswereinitiallygivenforfreeinproportiontohistoricalemissionlevels(anapproachknownas“grandfathering”;e.g.,Satoetal.,2022),therewasaneedtomonitorGHGemissionspriortothepilotphase(phase1)oftheEUETSin2005–2007.14TheRoleofFirmDynamicsintheGreenTransition:CarbonProductivityDecompositioninFinnishManufacturingpresenttheresultsinthesubsequentsections,wecomparetheaveragecarbonproductivityofthesefours