IMF-菲利普斯曲線變陡了嗎

HasthePhillipsCurve

BecomeSteeper?

AnilAri,DanielGarcia-MaciaandShrutiMishra

WP/23/100

IMFWorkingPapersdescriberesearchin

progressbytheauthor(s)andarepublishedto

elicitcommentsandtoencouragedebate.

TheviewsexpressedinIMFWorkingPapersare

thoseoftheauthor(s)anddonotnecessarily

representtheviewsoftheIMF,itsExecutiveBoard,

orIMFmanagement.

2023

MAY

*ThispaperbenefitedfromtheguidanceofS.AliAbbas,MarkFlanaganandLauraPapi,andhelpfulcommentsfromTommasoAquilante,HelgeBerger,OyaCelasun,RuoChen,RomainDuval,ChristianGlocker,MyrtoOikonomou,SaskiaterEllen,BoromeusWanengkirtyo,andseminarparticipantsattheBankofEnglandandHisMajesty'sTreasury.WearegratefultoGloriaLiforexcellentresearchassistance.

?2023InternationalMonetaryFund

WP/23/100

IMFWorkingPaper

EuropeanDepartment

HasthePhillipsCurveBecomeSteeper?

PreparedbyAnilAri,DanielGarcia-MaciaandShrutiMishra*

AuthorizedfordistributionbyS.AliAbbas

May2023

IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.

ABSTRACT:ThispaperanalyzeswhetherstructuralchangesintheaftermathofthepandemichavesteepenedthePhillipscurvesinadvancedeconomies,reversingtheflatteningobservedinrecentdecadesandreducingthesacrificeratioassociatedwithdisinflation.Particularly,analysisofgranularpricequotedatafromtheUKindicatesthatincreaseddigitalizationmayhaveraisedpriceflexibility,whilede-globalizationmayhavemadeinflationmoreresponsivetodomesticeconomicconditionsagain.Usingsectoraldatafrom24advancedeconomiesinEurope,higherdigitalizationandlowertradeintensityareshowntobeassociatedwithsteeperPhillipscurves.Post-pandemicPhillipscurveestimatesindicatesomesteepeningintheUK,Spain,Italyandtheeuroareaasawhole,butatmagnitudesthataretoosmalltoexplaintheentiresurgeininflationin2021–22,suggestinganimportantroleforoutwardshiftsinthePhillipscurve.

RECOMMENDEDCITATION:Ari,A.,Garcia-Macia,D.,andMishra,S.(2023).HasthePhillipsCurveBecomeSteeper?IMFWorkingPaper,WP/23/100

JELClassificationNumbers:E31,E52,O33,F41

Keywords:PhillipsCurve;Inflation;De-globalization;Digitalization;Structuralchange

Author’sE-MailAddress:

aari@

,

dgarciamacia@

,

sm4371@

WORKINGPAPERS

HasthePhillipsCurveBecomeSteeper?

PreparedbyAnilAri,DanielGarcia-MaciaandShrutiMishra

1

1ThispaperbenefitedfromtheguidanceofAliAbbasandMarkFlanagan,andhelpfulcommentsfromTommasoAquilante,RuoChen,RomainDuval,ChristianGlocker,MyrtoOikonomou,SaskiaterEllen,BoromeusWanengkirtyo,andothercounterpartsattheBankofEnglandandHisMajesty'sTreasury.

IMFWORKINGPAPERS

HasthePhillipsCurveBecomeSteeper?

INTERNATIONALMONETARYFUND2

Contents

1.INTRODUCTION

3

2.RELATEDLITERATURE

5

3.EMPIRICALFRAMEWORKANDDATA

7

3.1SectoralPhillipsCurveEstimation

7

3.2DigitalizationandPriceFlexibility

8

3.3DataSourcesandConstruction

8

4.EMPIRICALRESULTSANDROBUSTNESS

10

4.1SectoralPhillipsCurveRegressions

1

0

4.2DigitalizationandPriceFlexibility

14

5.POST-PANDEMICPHILLIPSCURVEESTIMATION

15

6.CONCLUSION

16

FIGURE

1.PhillipsCurveSlope

4

TABLES

1.CountriesandSectorsIncludedinthePanelDataset

9

2.BaselineEmpiricalResults

12

3.RobustnessCheck

13

4.DigitalizationandPriceFlexibilityResults

14

References

17

IMFWORKINGPAPERS

HasthePhillipsCurveBecomeSteeper?

INTERNATIONALMONETARYFUND3

1.Introduction

ThePhillipscurve,definedastheempiricalrelationshipbetweenunemploymentandinflation,illustratesthetrade-offbetweenmaintainingpricestabilityandachievingfullcapacityutilization.TherecenthistoryofthePhillipscurvehasbeenapuzzle.ManystudieshaveconcludedthatthePhillipscurvehasflattenedinadvancedeconomiesinrecentdecades(seee.g.,Blanchard,2016;DelNegroandothers,2020amongmanyothers),whileothershavedocumentedthemutedreactionofinflationtolargechangesinunemploymentduringandaftertheGreatRecession(CoibionandGorodnichenko,2015;Heiseandothers,2022).

Theliteraturehasproposedanumberofexplanationsforthisphenomenon.First,theimplementationofinflationtargetingmayhaveincreasedthecapacityofmonetarypolicytoneutralizedemandshocks(whichwouldmoveinflationandunemploymentinoppositedirections),leavingsupplyshocks(whichmovetheminthesamedirection,shiftingthePhillipscurveinwards/outwards)asthemainsourceofaggregatevolatility(Broadbent,2020;Bergholt,andothers,2023).Second,firmeranchoringofinflationexpectationsmayhavereducedsecond-roundeffectsofrelativepriceshocks(Borioandothers,2021).Third,globalizationmayhavereducedtheresponsivenessofinflationtodomesticeconomicslackandunemployment(Heiseandothers,2022),includingbyraisinglabormarketflexibilitythroughareductioninlaborbargainingpower(Lombardiandothers,2020).Fourth,increasedmarketpowermayhaveaugmentedfirms’capacitytoabsorbcost-pushshocksintheirmarkups(Baqaee,FarhiandSangani,2021).

Someofthesestructuraltrendsmayhavestartedtoreverseinrecentyears.Therecentriseininflationtohistoricallyhighlevelsinadvancedeconomiesrisksweakeningtheanchoringofinflationexpectationsachievedoverthepasttwodecades.Ade-globalizationscenariomayalsomakethedomesticoutputgapmorerelevantagain,aswellasrestoringsomelaborbargainingpower(GoodhartandPradhan,2020).NewstructuralfactorsmayhavealsocontributedtoasteepeningofthePhillipscurvereducingnominalrigidities.Notably,morewidespreaddigitalization,includinginresponsetotheCOVID-19pandemic,hasraisedtheshareofonlineretail,wherepricesaretypicallymoreflexible(GorodnichenkoandTalavera,2017;Cavallo,2018).

Figure1demonstratesboththeflatteningofthePhillipscurveinrecentdecadesandmorerecentsignsofsteepening.TheempiricalcorrelationbetweenunemploymentandinflationwhichhadgonefrombeingnegativetobroadlyzerointheUSandtheeuroarea,andpositiveintheUK,hasagainturnednegativesincethepandemic.

IMFWORKINGPAPERS

HasthePhillipsCurveBecomeSteeper?

INTERNATIONALMONETARYFUND4

andstaffcalculations

Figure1:PhillipsCurveSlope

(10yearrollingcorrelationoftheunemploymentrateandCPIinflation)

Source:ONSIMF

Inviewofthesedevelopments,thispaperanalyzestheimpactofrecentstructuralchangesintradeintensityanddigitalizationontheslopeofthePhillipscurve.Todothis,wefirstestimateasectoralPhillipscurveusingquarterlydatafrom17sectorsin24advancedeconomiesinEuropeover2012Q1–2019Q4.

1

WefindthatbothstructuralfactorshaveastatisticallysignificantimpactinsteepeningthePhillipscurve.Thesefindingsarerobusttocontrollingforvarioussourcesofnon-linearitiesandendogenousmonetarypolicyresponsestoshocks.

ToshedfurtherlightonthemechanismthroughwhichdigitalizationsteepensthePhillipscurve,wecomplementthisanalysiswithpricequotedatafromtheUK(i.e.,individualpricesforgoodsinspecificestablishments).Particularly,weusemonthlydatafromJanuary2008toJune2021toshowthatgreatere-commerceintensityatthesectorallevelisassociatedwithasignificantlyhigherfrequencyofindividualpricechanges.

2

ThisfindingisconsistentwithpreviousresultsfromGorodnichenkoandTalavera(2017)andCavallo(2018).

1EstimatingsectoralPhillipscurvesallowsustoexploitmuchmorevariationinthestructuralvariablesweconsider,increasingthepowerofourestimations.Theimplicitassumptionthatunderliesthisidentificationstrategyisthatproductionfactorsarenotperfectlymobileacrosssectors,allowingforoutputgapstodifferacrosssectors.

2Thisfindingisrobusttodefiningpriceflexibilitywitheitherpositiveornegativepricechangesonlyandincludingsectorandyearfixedeffects.

IMFWORKINGPAPERS

HasthePhillipsCurveBecomeSteeper?

INTERNATIONALMONETARYFUND5

Finally,weapplythesectoralPhillipscurveresultstoinferchangesinthepost-pandemicPhillipscurveforfivelargeEuropeaneconomies–France,Germany,Italy,SpainandtheUK—aswellasaeuroareaaggregate.WhilewefindthatthePhillipscurvehasindeedsteepenedintheUK,Spain,Italy,andtheeuroarea,theextentofsteepeningislimited,andpost-pandemicPhillipscurvesremainlargelyflat.Thisconclusionholdseveninahypotheticalde-globalizationscenariowheretradeintensityfallssubstantially.ItalsosuggestsanimportantroleforoutwardsshiftsinthePhillipscurve(whichcouldbeduetohigherinflationexpectations,supplyshocks,orotherstructuralchanges)inexplainingthesurgeininflationin2021–22.

Theremainderofthepaperisorganizedasfollows.Section2reviewstherelatedliterature.Section3describestheempiricalstrategyanddata.Section4providestheresultsandrobustnesschecks.Section5estimatesPhillipscurveslopesintheaftermathofthepandemicandinade-globalizationscenario.Section6concludes.

2.RelatedLiterature

OurpaperiscloselyrelatedtotheliteratureontheslopeofthePhillipscurve.Blanchard(2016),Mishkin(2007)andPowell(2018)documentthattheUSPhillipscurvehasflattenedinthe1980s,whileHaldaneandQuah(1999)reachasimilarconclusionfortheUK.LeducandWilson(2017)andMurphy(2018)documentthatthewagePhillipscurvehasalsobecomeflatter.DelNegroandothers(2020)andCoibionandGorodnichenko(2015)considertheflatteningofthePhillipscurveasapossibleexplanationformissingdeflationduringtheGreatRecession,andHeiseandothers(2022)useittoprovideanexplanationformissinginflationinthe2010s.Abdihandothers(2018)arguethathighinflationpersistencecanaccountformissinginflationpuzzlesintheeuroarea.

FitzgeraldandNicolini(2014),McLeayandTenreyro(2020),andHooper,MishkinandSufi(2020)arguethatendogenousmonetarypolicyresponsesmayresultinanunderestimationoftheslopeofthePhillipscurveandshowthatestimatesbasedonUSregionaldataindicateasteeperPhillipscurve.Incontrast,IMF(2021)showsthatinstrumentingforchangesinunemploymentwithmonetarypolicyshocksdoesnotleadtoasteeperPhillipscurveslopeestimateforadvancedeconomies.Hazellandothers(2022)findevidenceofamodestflatteningoftheUSPhillipscurveusingregionaldata.Similarly,Tuckett(2018)usesregionaldatatoshowthattheUKPhillipscurvehasbecomeflatter.Whilewerelyoneconomy-widedatainourstudy,20ofthe

24economiesinoursampleareeithermembersoftheeuroareaorhaveacurrencypegwiththeeuro,andthuscanbethoughtofas“regions”withinamonetaryunion.Usingthesub-sampleofeuro(pegged)countriesandthesameapproachastheabovestudies,weshowthatourfindingsarerobusttocontrollingformonetarypolicyresponses.

Otherstudiesarguethatthetime-varianceintheslopeofthePhillipscurvemaybeattributedtonon-linearities.BarnesandOlivei(2003)proposeapiecewise-linearPhillipscurvespecificationwhereinflationrespondsmorestronglytounemploymentwhenunemploymentisbelowathreshold,andDoserandothers(2023)showthatsuchaspecificationcanaccountformissingdisinflationduringtheGreatRecession.Speigner(2014)findsthatlong-termunemploymenthasasignificanteffectoninflationonceconvexitiesareallowedforduringthe

IMFWORKINGPAPERS

HasthePhillipsCurveBecomeSteeper?

INTERNATIONALMONETARYFUND6

estimationprocess.Wealsoconsidernon-linearPhillipscurvespecificationsinourempiricalanalysisandfindthatourresultsalsoremainrobusttoaccountingforthese.

TheliteraturehasalsoofferedvarioustheoreticalexplanationsfortheflatteningofthePhillipscurve.Nalewaik(2016)andBorioandothers(2021)attributeittofirmeranchoringofinflationexpectations.BallandMazumder(2011)arguethatadeclineinthelevelandvariabilityofinflationhasreducedpriceflexibilitygivencostlypriceadjustment.DalyandHobjin(2014)emphasizesimilarinteractionsbetweenlowinflationanddownwardnominalwagerigidities,whileAfrouziandYang(2020)focusontheroleofrationalinattentioninpricesettingdecisions.Baqaee,FarhiandSangani(2021)andRubbo(2022)insteadrelatetheflatteningofthePhillipscurvetomarkupsandintermediateinputs,bothofwhichdampenthepassthroughofshocksintoprices.Lombardiandothers(2020)suggestthattheflatteningisduetoaweakeninginlaborbargainingpower,whichmayinturnberelatedtodecliningunionizationandglobalization.Auerandothers(2017)andHeiseandothers(2022)alsoidentifyglobalizationasadrivingfactorbutproposedifferenttransmissionchannelsthatreducetheresponsivenessofinflationtothedomesticoutputgap.Auerandothers(2017)focusontheroleofinternationalinput-outputlinkages,whileHeiseandothers(2022)considerimportcompetition.

Toourknowledge,thispaperisamongthefirsttostudyasteepeningofthePhillipscurveinthepost-pandemicera.Weconsidertwostructuralfactors—de-globalizationanddigitalization—aspotentialdriversofasteeperPhillipscurve.Alfonsoandothers(2021)documentthegrowthofe-commerceinrecentyearsanditsaccelerationduringthepandemic,andGorodnichenkoandTalavera(2017)andCavallo(2018)showthatpricesinonlinemarketsaremoreflexible.GlockerandPiribauer(2021)findthatmonetarypolicyshockshavelessimpactonoutputwhentheshareofonlineretailtradeishigh,inlinewithourresults.Witt(2019)andRazin(2020)arguethattradeglobalization(asmeasuredbyshareoftradeinworldGDP)hasreversedcoursesincetheglobalfinancialcrisis.Ouruseoftradeintensityasaproxyforde-globalizationissimilartoHeiseandothers(2022),whorelyonimportpenetrationasaproxy.Whilepost-pandemicdatadonotregisteradeclineintradeintensityinlargeEuropeaneconomiesandtheeuroarea,ade-globalizationscenarioisarelevantriskgiventensionsbetweentheUSandChinaandRussia’swarinUkraine,whichhaveraisedconcernsaboutthereliabilityofsupplychains(Handley,KamalandMonarch,2020).

Finally,thispaperalsorelatestorecentstudiesthatprovideabottom-upaccountingoftherecentsurgeininflationinEurope,includingbyestimatingPhillipscurves(McGregorandToscani,2022;IMF,2022;Babaandothers,2023).ThesestudiesfindthatPhillipscurvesestimatedonhistoricaldataandchangesinenergyandothercommoditypricescanonlyexplainabouthalfoftheincreaseininflationin2022inadvancedandemergingEuropeaneconomies.

3

ThesefindingsareconsistentwithourestimatesofflatPhillipscurveslopesforlargeEuropeaneconomiesinthispaper.

3Relatedly,Hodgeandothers(2022)useaDSGEmodeltoshowthat,evenwithasteeperPhillipscurveslope,policyactionswouldnothaveexplainedtheinflationsurgeineithertheeuroareaortheUS.

INTERNATIONALMONETARYFUND7

3.EmpiricalFrameworkandData

ThissectionfirstintroducestheempiricalframeworksforsectoralPhillipscurveestimatesandtheanalysisoftheimpactofdigitalizationonpriceflexibility.Thenitdescribesthedatasourcesanddataconstruction.

3.1SectoralPhillipsCurveEstimation

Theanalysisisbasedonapanelof17sectorsin24advancedeconomiesinEurope,atquarterlyfrequencyover2012Q1–2019Q4.

4

Thebaselinespecificationreflectsastandardopeneconomy,expectations-augmentedPhillipscurverelationship(seee.g.,Kamber,MohantyandMorley,2020)andcanbewrittenasfollows:

πi,s,t=ai+as+at+ai,s+Fi,s,t+γEt[冗i,t+4]+∑=1入k冗i,s,t?k+e」si,t+6Xi,s,t+pi,s,tXi,s,t+ei,s,t.(1)

wherei,s,tdenotecountry,sector,andtimesuchthat(ai,as,at,ai,s)arerespectivelycountry,sector,time,andcountry-sectorfixedeffects.πi,s,trepresentsinflation,i,s,tistheoutputgapandEt[冗i,t+1]capturesinflationexpectationsasisstandardintheliterature.WealsofollowGaliandGertler(1999)inintroducinglaggedinflationtermstocontrolforadaptivecomponentsofinflationexpectationsaswellaslaggedchangesinotherdeterminantsofinflation.Additionally,asisstandardforopeneconomieswithalargeproportionofimportedgoodsandservices,wealsocontrolforappreciationintherealeffectiveexchangerate」si,t.ThevectorXi,s,tcapturesstructuralchangesassociatedwithtradeintensityanddigitalization,andei,s,tistheresidual.

Thekeycoefficientsthatweaimtoestimatearedenotedbythevector,whichcapturestheimpactofstructuralfactorsontheslopeofthePhillipscurve.Particularly,anegativeestimateforthetradeintensitycoefficientandapositiveestimateforthedigitalizationcoefficientinthisvectorwouldindicatethatde-globalizationanddigitalizationareassociatedwithasteeperPhillipscurve.

Wealsoconsideranumberofrobustnesstests.First,werepeatthebaselineestimatesin(1)butclusterthestandarderrorsatthelevelofsector-countrycombinations.Second,ourestimatesmaybebiasedbyendogenousmonetarypolicyresponsestoinflationandtheoutputgap.Toaddressthisconcern,wealsoestimatespecification(1)usingdataonlyfromcountriesthatareintheeuroareaorthathaveanexchangeratepegwiththeeuro.Sincethesecountrieshaveacommonmonetarypolicy,endogenousmonetarypolicyresponseswouldbecapturedbytimefixedeffects.AthirdconcernwithouridentificationisthatthePhillipscurvemaybenon-linear.Tocheckwhetherouranalysisisrobusttonon-linearities,wealsoconsideraspecificationwithquadratictermsfortheoutputgapandinflationexpectations,suchthat

πi,s,t=ai+as+at+ai,s+Fi,s,t+F2s,t+γEt[冗i,t+4]+y2Et[冗i,t+4]2e」si,t+6Xi,s,t+

+∑=1入k冗i,s,t?k+pi,s,tXi,s,t+ei,s,t(2)

4Seesection3.3foralistofthecountriesandsectorsincluded.Addingasectoraldimensiontothepanelstrengthenstheidentificationbyallowingtocontrolforsectorandcountry-sectorfixedeffects.

INTERNATIONALMONETARYFUND8

3.2DigitalizationandPriceFlexibility

WefurtherinvestigatethemechanismthroughwhichdigitalizationaffectstheslopeofthePhillipscurvebyusingmonthlydatafromtheUKbetweenJanuary2008andJune2021toanalyzewhether(off-line)pricesarerevisedmorefrequentlyinsectorswithhighere-commerceintensity.Ourspecificationisasfollows:

Fs,t=as+at+FEs,t+es,t

(3)

wheres,tdenotesectorandyearandasandatarerespectivelysectorandyearfixedeffectsasbefore.Fs,tdenotespriceflexibility,definedasthemonthlyrateatwhichpricequotespforindividualitemsinagivenshoparechangedwithinsectorsinyeart:

Fs,t≡∑jeS∑meTI(pj,m?pj,m?1≠0)

(4)

wheremindicatesthemonthandjthepricequote,andNs,tisthenumberofpricequotesinthesampleforagivensector-year.Forexample,apriceflexibilityof?meansthatthepriceofanitemchangeseverytwomonthsonaverage.Es,trepresentse-commerceintensity,definedastheshareofenterprisesreceivinge-commerceorderswithinasector,andes,tistheresidual.

AnestimateofF?>0wouldthensuggestthatgreatere-commerceintensityisassociatedwithhigherpriceflexibility.WhilethiscouldbeinterpretedasadirecteffectofdigitalizationasinGorodnichenkoandTalavera(2017)andCavallo(2018),analternativeinterpretationwouldbethate-commercepenetrationactsasadeflationaryforce,triggeringpricereductionsthatthencreateafalseimpressionofincreasedpriceflexibility.Asarobustnesscheckforsucha“deflationarybias,”wealsorepeatouranalysisbyre-definingFs,ttoalternatelyexcludeallpositiveandnegativepricechanges,asthefrequencyoftheformerwouldnotincreasefromdeflationarypressure.

3.3DataSourcesandConstruction

Table1reportsthecountriesandsectorsincludedinthepaneldatasetforthePhillipscurveestimates,whichcoverstheperiod2014Q4–2019Q4atquarterlyfrequency.

5

,

6

Werestrictouranalysistopre-pandemicdataasweconsiderourestimatesofinterest(capturedin)tobedeepparametersthatarestableovertimebutthatmayhavebeentemporarily“dislodged”bypandemic-relateddisruptions.Theinitialtimepointandsectoralbreakdowninourdatasetaregivenbydataavailability.

Weproxyfordigitalizationwiththeshareofenterpriseswithe-commercesalesamountingtoatleast1percentoftotalturnover,whichisavailablefromEurostatatthesectoralbreakdownshowninTable1.Weaggregateothersectoralvariableswithmoregranularsectoraldatauptothislevel.Theseincludesectoralinflation,measuredusingimpliedGDPdeflators;thesectoraloutputgap,constructedbyapplyingaHodrick-Prescott

5Amongthecountries,Czechia,Norway,SwedenandtheUKhaveindependentmonetarypolicies,leadingtotheirexclusionfromrobustnesschecksforendogenousmonetarypolicyreactions.Theremainingcountriesareeuroareamembers,exceptforDenmarkwhichhasanexchangeratepegwiththeeuro.

6In2019,theincludedsectorsaccountedonaveragefor42percentoftotalgrossvalueaddedforthecountriesinthedataset.

IMFWORKINGPAPERS

HasthePhillipsCurveBecomeSteeper?

INTERNATIONALMONETARYFUND9

filtertorealgrossvalueadded(GVA)andtakingtheresidualinpercentageterms;andsectoraltradeintensity(aproxyfor(de-)globalization),whichistheratioofsectoralimportsandexportstosectoralGVA.ThesevariablesareallobtainedfromEurostat,withsectoralimportsandexportsonlyavailablefrom2012Q1onwards.Inflationexpectationsandrealeffectiveexchangeratesareonlyavailableatthecountrylevel.Weproxyforinflationexpectationswithone-year-aheadCPIinflationforecastsfromConsensusEconomicsandobtain(CPI-based)realeffectiveexchangerate(REER)datafromtheIMF’sWorldEconomicOutlookdatabase.

Table1:CountriesandSectorsIncludedinthePanelDataset

Countries:

Sectors:

NACERev.2

sectorcode

C10-C12C13-C15C16-C18C19

C20 C21C22-C23

C24-C25

C26C27-C28C29-C30

C31-C33

H49-H53

I

J58-J60

J61

J62-J63

Austria,Belgium,Cyprus,Czechia,Denmark,Estonia,Finland,France,Germany,Greece,Ireland,Italy,Latvia,Lithuania,Luxembourg,Malta,Netherlands,Norway,Portugal,Slovakia,Slovenia,Spain,SwedenandtheUnitedKingdom.

Sectorname

Manufactureoffoodproducts;beveragesandtobaccoproducts

Manufactureoftextiles,wearingapparel,leatherandrelatedproducts

Manufactureofwood,paper,printingandreproduction

Manufactureofcokeandrefinedpetroleumproducts

Manufactureofchemicalsandchemicalproducts

ManufactureofbasicpharmaceuticalproductsandpharmaceuticalpreparationsManufactureofrubberandplasticproductsandothernon-metallicmineralproducts

Manufactureofbasicmetalsandfabricatedmetalproducts,exceptmachineryandequipment

Manufactureofcomputer,electronicandopticalproducts

Manufactureofelectricalequipmentandmachineryandequipmentn.e.c.Manufactureofmotorvehicles,trailers,semi-trailersandofothertransportequipmentManufactureoffurniture;jewelry,musicalinstruments,toys;repairandinstallationofmachineryandequipment

Landtransportandtranspor