亞開行-中國(guó)的崛起及其對(duì)日本創(chuàng)新的競(jìng)爭(zhēng)效應(yīng)(英文)-行業(yè)資料-市場(chǎng)研究報(bào)告-綜合-可編輯版

ADBIWorkingPaperSeries

THERISEOFTHEPEOPLE’SREPUBLIC

OFCHINAANDITSCOMPETITION

EFFECTSONINNOVATIONINJAPAN

NobuakiYamashitaand

IsamuYamauchi

No.939

March2019

AsianDevelopmentBankInstitute

NobuakiYamashitaisaSeniorLecturer,RoyalMelbourneInstituteofTechnology,Australia.IsamuYamauchiisaResearchAssociateattheResearchInstituteofEconomy,TradeandIndustry(RIETI)andaJuniorAssociateProfessoratMeijiGakuinUniversity.

TheviewsexpressedinthispaperaretheviewsoftheauthoranddonotnecessarilyreflecttheviewsorpoliciesofADBI,ADB,itsBoardofDirectors,orthegovernmentstheyrepresent.ADBIdoesnotguaranteetheaccuracyofthedataincludedinthispaperandacceptsnoresponsibilityforanyconsequencesoftheiruse.TerminologyusedmaynotnecessarilybeconsistentwithADBofficialterms.

Workingpapersaresubjecttoformalrevisionandcorrectionbeforetheyarefinalizedandconsideredpublished.

TheWorkingPaperseriesisacontinuationoftheformerlynamedDiscussionPaperseries;thenumberingofthepaperscontinuedwithoutinterruptionorchange.ADBI’sworkingpapersreflectinitialideasonatopicandarepostedonlinefordiscussion.Someworkingpapersmaydevelopintootherformsofpublication.

Suggestedcitation:

Yamashita,N.andI.Yamauchi.2019.TheRiseofthePeople’sRepublicofChinaandItsCompetitionEffectsonInnovationinJapan.ADBIWorkingPaper939.Tokyo:AsianDevelopmentBankInstitute.Available:/publications/rise-prc-and-its-competition-effects-innovation-japan

Pleasecontacttheauthorsforinformationaboutthispaper.

Email:nobu.yamashita@.au

Thisstudywasundertakenasapartoftheproject“MobilityofKnowledgeandtheInnovationPerformanceofJapaneseFirms”attheResearchInstituteofEconomy,TradeandIndustry(RIETI)inTokyo,Japan.Forusefulsuggestionsandcomments,wewouldliketothankYuqingXing,MatthiasHelble,andparticipantsattheADBworkshop.

AsianDevelopmentBankInstitute

KasumigasekiBuilding,8thFloor

3-2-5Kasumigaseki,Chiyoda-ku

Tokyo100-6008,Japan

Tel:Fax:URL:E-mail:

+81-3-3593-5500

+81-3-3593-5571

info@

?2019AsianDevelopmentBankInstitute

ADBIWorkingPaper939

YamashitaandYamauchi

Abstract

Thispaperempiricallyexaminesthe“defensiveinnovation”hypothesisthatfirmswithhigherexposuretolow-wageeconomyimportcompetitionintensivelyundertakemoreinnovativeactivitybyusingahighqualityJapanesefirm-levelpaneldatasetovertheperiod1994–2005.Thenovelfeatureoftheanalysisistherelationoffirm-levelvariationsofpatentusagetoimportcompetition.TheresultssuggestthatintensifiedimportcompetitionfromthePeople’sRepublicofChinahasresultedingreaterinnovativeactivitybyJapanesefirms,consistentwiththefindingsofEuropeanfirmsinBloometal.(2016).Moreover,suchcompetitionhasalsoledtoanincreaseinnon-usedpatents.

JELClassification:O00,F10

ADBIWorkingPaper939

YamashitaandYamauchi

Contents

1.INTRODUCTION

1

2.THERISEOFTHEPRCINWORLDTRADE

2

3.DATAANDVARIABLES

6

3.1Firm-levelPatentData

6

3.2JapanIndustryProductivityData

7

4.EMPIRICALSPECIFICATIONANDRESULTS

9

5.RESULTS

10

6.CONCLUSION

13

REFERENCES

14

APPENDIX

17

1

ADBIWorkingPaper939

YamashitaandYamauchi

1.INTRODUCTION

Thispaperexaminesthe“‘defensiveinnovation”hypothesisfirstdiscussedinWood(1994)andsubsequentlyformalizedinThoenigandVerdier(2003).Asareactiontoimportcompetitionfromlow-wageeconomies,firmsindevelopedeconomieswouldrespondbyupgradingtheirinnovativeactivities,leadingto“defensiveskill-biasedinnovation.”Inabroadercontext,theeffectofcompetitionontherateofinnovationhasbeenoneofthemoststudiedareasintheliterature(e.g,Aghionetal.2005).Inthestudymostrelevanttoourpaper,Bloometal.(2016)foundthatalargesampleofEuropeanfirmsincreasedawiderangeoftheirinnovativeactivities(patenting,researchanddevelopment(R&D)expenditures,computeruse,andtheTFPgrowth),drivenbyintensifiedcompetitionfromthePeople’sRepublicofChina(PRC).Thisinnovationwasconductedwithin-firm.1

Buildingonthefoundationsetbythepreviouslymentionedstudies,thispaperexaminesthecausaleffectofintensifiedChineseimportcompetitionontheinnovativeactivitiesofapanelofJapanesefirmsfortheperiod1994–2005.Wefocusonpatentusagedataasanindicatorofinnovativeoutputs.Unlikeotherstudiesusingpatentstatistics,thisstudyaddstotheliteraturebyexploringstrategicpatentusageasresponsestoimportcompetitionfromalow-wageeconomy(thePRC).Itisgenerallyacknowledgedthatpatentstatisticsaremeaningfulproxiesforfirm-levelinnovation,butitiswellknownthatfirm-levelpatentingservesasmuchmorethanjustanindicatorofknowledgecapitaloutput(Nagaokaetal.2010).Well-knowninventorsurveys(e.g.,theRIETI-GeorgiaTechUS-Japansurvey)haverevealedthatmanyofthepatentsarenotusedtointroducenewproductsintothemarket;instead,theyareusedaseffectivestrategicinstrumentsto“block”othercompetitorsfrominnovatingorimitating.BoldrinandLevine(2013)presentanicecaseinvolvingMicrosoft—amarketincumbentwithastockpileofpatentsblockingGoogleinthesmartphonemarket).

Studyinginnovativefirms’responsestoChineseimportcompetitionprovidesaninterestingandexcellenttestinggroundforthefollowingreasons:First,overthepastdecades,thePRChasemergedasapivotalassembly-exporteconomyofhigh-techproducts(mainly,electronics),importingpartsandcomponentsfromotheradvancedeconomiesandexportingfinalproducts(includingthefamousiPhone).Accordingly,thePRC’sexportbundlehasdramaticallychangedfromlabor-intensivegoodstohigh-techproducts,exertingconsiderablecompetitivepressuresonfirmsindevelopedeconomies.Second,manyChineseexportscompeteatlowercostmarginsthanmosthigh-techproducts.Forinstance,astudybySchott(2008foundthatthePRC’sexportsimilarityindexhasbecomeclosertothatofOrganizationforEconomicCooperationandDevelopment(OECD)economies,buttheunitpricesofChineseexportshavebeenconsistentlylowerthanOECDeconomies.

ThefindingsuggeststhatChineseimportcompetitionleadsJapanesefirmstoexpandtheirinnovativeactivities,asfoundbyBloometal.(2016.Theexpansionispartlydrivenbyanincreaseinfirms’numbersofunusedpatents,whichreflectthestrategicuseofintellectualproperty(IP)protection.

Theorganizationofthispaperisasfollows.ThenextsectionpresentsanoverviewofthePRCinworldtrade.Section3discussesthedataset,followedinSection4bythe

1AmitiandKhandelwal(2013)findthatincreasedimportcompetition(measuredbyadeclineintariffs)spursaeconomy’sexportquality(measuredbythemarketshare)intheUSmarket.

2

ADBIWorkingPaper939

YamashitaandYamauchi

empiricalapproachandadiscussionofthepreliminaryfindings.Section5concludesthepaper.

2.THERISEOFTHEPRCINWORLDTRADE

Figure1-AdepictstheriseofthePRCinworldexportsfortheperiod1990–2011.In1990,thePRC’sexportsaccountedforatinyshare(around3%)ofworldexports.Sincethen,thePRC’ssharehasgraduallyincreased.Inparticular,thePRC’sexportgrowthhasrisensincetheearly2000s.Inthesecondhalfof2000,thePRChasachievedformidableexportexpansionbyovertakingGermanyforthepositionoftheworld’slargestexporter,accountingformorethan10%ofworldexports.ThePRC’sexportsharehasbeengrowingwithoutanydisruptions,whiletheworldsharesofJapan,theUnitedStatesandGermanyhavenotgrownduringthesameperiod.Atthesametime,thePRChasbecomeanimportanteconomyintheworldimportantmarket(Figure1-B).WhiletheUnitedStatesstillaccountsforthebulkofworldimports(around15–20%inworldimports),itssharehasgraduallybeendecliningsince2000.Bycontrast,thePRC’ssharehassteadilyincreasedtocloseto10%in2011.

Figure1:TheRiseofthePeople’sRepublicofChinainWorldTrade,1990–2011

(%intotalexports)

Source:UNComtrade.

WiththeriseofthePRCinworldtrade,itsspecializationhasdramaticallychanged,aswell.Figure2depictstheshareofrelativelymorecapital-andtechnology-intensiveproductslikeelectricalmachineryandhouseholdelectricappliancesascomparedtomorelabor-intensiveproductsliketextilesandtoys.Therehasbeenanotableshiftofcomparativeadvantagesfrommorelabor-intensiveproductstowardmorecapital-andtechnology-intensiveproducts.In1992,textilesandtoysaccountedforapproximately45%ofthePRC’stotalexports.However,thissharecontinuouslydeclinedanddroppedtocloseto20%in2011.Ontheotherhand,theexportshareofelectricalmachineryandhouseholdappliancesdoubleditsshare,fromlessthan15%in1992to30%in2011.Inthisproductcategory,theexportcompositionishighlyconcentratedinInformationCommunicationTechnology(ICT)products.Otherimportantproduct

3

ADBIWorkingPaper939

YamashitaandYamauchi

categoriesincludeofficemachines,andtelecommunicationssoundequipment(includingmobilephones).

Basedontheincome-weightedexportbundleofChinesegoods,somecommentatorsarguethatthisisasignthatthetechnologicalcapabilityofthePRCisrapidlyconvergingtowardthetechnologicalfrontierofadvancedOECDeconomies,andisnowdirectlycompetingwiththemintheexportmarket.However,thisshouldbeinterpretedcautiously.Allowingforintra-productspecialization,itisknownthatthePRC’sexportspecializationstillrestslargelyonthelabor-intensiveassemblystageratherthanspecializationintechnologicalcontent(AthukoralaandYamashita,2006).Inotherwords,thePRC’scomparativeadvantagesstillrestsonalabor-intensivesegmentinhigh-techproducts,eventhoughtheseproductsareexportedfromthePRC(afinalassemblyeconomy).ThisexplainswhySchott(2008)observesthattheunitpriceofChineseexportbundlesareatthelowerendofthepricerange,ascomparedtothoseofOECDeconomies(thepricecompetitivenesscomingfromthePRC’slowerlaborcosts).Insum,thebulkofChineseexportsaremass-marketcommoditiesassembledwithrelativelylowunitcostsandimportedhigh-techpartsandcomponentsfromotherindustrialeconomies(notebookcomputers,mobilephones).

Figure2:StructuralChangesinthePeople’sRepublicofChina’s

ExportProductCompositions,1990–2011

(%intotalexports)

Source:UNComtrade.

Table1displaysthetopeightandbottomeightindustriesbydegreeofChineseimportcompetition1994(thebeginningoftheestimationperiod).2Inthetextileindustry,whereChinesefirmsareconsideredtohavecomparativeadvantages,thedegreeofimportcompetitionwasalreadystrongin1994—ofJapan’simportoftextileproducts,49%camefromthePRC.Thatsharecontinuedtoincrease,reaching77%in2005.Morestrikingly,thelargestincreaseinthePRC’sshareofJapaneseimportsisinofficeandserviceindustrymachines;thissharerosefrom19%in1994to76%in2005.

2Year1990dataisusedinanexperimentalstage,buttheorder-importcompletion-exposedindustriesareroughlythesameinyear1994).

4

ADBIWorkingPaper939

YamashitaandYamauchi

Correspondingly,intheindustrieswherebythePRC’sshareincreased,therewasadeclineinthesharesofAsiannewlyindustrializingeconomies(NIEs)—Taipei,China;HongKong,China;RepublicofKorea;andSingapore)andtheUS.Inthebottomeightindustries,anincreaseinthePRC’sshareispalpable,withstronggrowthinelectronicequipmentandsemiconductordevices.ProductionnetworksbetweenJapanandthePRCmayexplainanexpansioninChineseimportinthosehigh-techindustries.

Table1:ChangeofImportCompetitionbySourceEconomies/GroupsinJapaneseManufacturingIndustry,1994and2005

1994

Asian

PRCNIEsSEAsiaUS

Manufacturing,total11.4

15.9

10.2

25.7

Top8sectorsin1994

Coalproducts

Textileproducts

Miscellaneousceramic,stoneandclayproductsRubberproducts

Leatherandleatherproducts

Electricalgenerating,transmission,distributionandindustrialapparatus

Pigironandcrudesteel

Officeandserviceindustrymachines

68.9

48.7

34.4

33.4

26.5

24.4

23.7

19.4

13.2

15.1

19.1

18.3

19.9

24.1

4.0

16.5

0.0

8.0

3.0

10.1

5.8

19.5

3.0

21.6

2.7

5.6

13.6

15.7

5.2

19.6

7.0

22.1

Bottom8sectorsin1994

Chemicalfibers

Petroleumproducts

Electronicequipmentandelectric

measuringinstruments

Pulp,paper,andcoatedandglazedpaper

Semiconductordevicesandintegratedcircuits

Printing,platemakingforprintingand

bookbinding

Tobacco

Motorvehicles

1.2

1.0

0.6

0.5

0.4

0.4

0.1

0.0

48.9

22.4

3.1

1.7

41.7

26.0

0.0

0.3

2.7

12.0

0.6

0.9

8.2

1.0

0.0

0.0

26.4

6.2

63.9

40.7

49.1

64.7

95.3

27.7

2005

PRC

AsianNIEs

SEAsia

US

Manufacturing,total

28.6

12.8

10.9

15.2

Top8sectorsin1994

Coalproducts

92.2

1.5

0.0

0.5

Textileproducts

76.5

3.5

4.0

2.0

Miscellaneousceramic,stoneandclayproducts

60.4

5.0

3.8

9.6

Rubberproducts

58.4

6.9

17.2

5.5

Leatherandleatherproducts

46.5

1.8

2.7

2.0

Electricalgenerating,transmission,distributionandindustrialapparatus

47.2

8.5

17.2

10.2

5

ADBIWorkingPaper939

YamashitaandYamauchi

Pigironandcrudesteel

29.7

6.7

1.5

1.7

Officeandserviceindustrymachines

76.2

8.2

7.7

2.7

continuedonnextpage

Table1continued

2005

Asian

PRCNIEsSEAsiaUS

Bottom8sectorsin1994

Chemicalfibers13.9

Petroleumproducts2.8

Electronicequipmentandelectric10.5

measuringinstruments

Pulp,paper,andcoatedandglazedpaper7.3

Semiconductordevicesandintegrated7.9

circuits

Printing,platemakingforprintingand13.5

bookbinding

Tobacco0.6

Motorvehicles1.4

34.2

21.1

3.5

6.5

48.2

11.2

0.1

1.9

15.5

12.8

4.0

13.4

19.1

4.5

0.1

0.9

13.3

2.5

38.8

33.5

18.9

23.3

89.6

8.8

PRC

Change94–05

AsianNIEs

SEAsia

US

Manufacturing,total17.2

–3.10.7

–10.5

Top8sectorsin1994

Coalproducts

23.3

–11.7

0.0

–2.2

Textileproducts

27.8

–11.6

–4.0

–3.6

Miscellaneousceramic,stoneandclayproducts

26.0

–14.1

0.9

–4.1

Rubberproducts

25.0

–11.5

7.2

–10.2

Leatherandleatherproducts

20.0

–18.0

–3.1

–3.3

Electricalgenerating,transmission,distributionandindustrialapparatus

22.8

–15.6

–2.3

–9.4

Pigironandcrudesteel

Officeandserviceindustrymachines

6.0

56.7

2.7

–8.4

–1.5

–13.9

–5.3

–19.4

Bottom8sectorsin1994

Chemicalfibers

12.7

–14.7

12.8

–13.1

Petroleumproducts

1.8

–1.4

0.8

–3.7

Electronicequipmentandelectric

measuringinstruments

9.9

0.4

3.4

–25.1

Pulp,paper,andcoatedandglazedpaper

6.8

4.8

12.6

–7.1

Semiconductordevicesandintegratedcircuits

7.5

6.5

10.9

–30.3

Printing,platemakingforprintingand

bookbinding

13.1

–14.8

3.5

–41.4

Tobacco

0.6

0.1

0.0

–5.8

Motorvehicles

1.4

1.7

0.9

–18.9

Source:JIP2013database.

6

ADBIWorkingPaper939

YamashitaandYamauchi

3.DATAANDVARIABLES

3.1Firm-levelPatentData

Patentstatisticsasanindicatorforinnovativeoutputshaverecentlybecomewidelyavailabletoresearchersbecauseofsignificantprogressmadeindataaccessibility(e.g.,USNBERpatent,JapanPatentOffice,PATSTAT).Patentstatisticscarryimportantinvention-relatedinformationsuchasbibliographicdata,backwardandforwardcitations,thetechnologyfields,nameofinventor,andusefulness.However,ithasbeenwell-documentedfromsurvey-basedstudiesthatnotallpatentsareinuse(butrather,are“sleeping”).InJapan,ithasbeenreportedthatapproximately60%ofpharmaceuticalpatentsarenotcurrentlyinuse(Nagaokaetal.2010.3Rather,firmsobtainpatentsasadefensiveblockingmechanisminresponsetotechnologycompetition.4“Blocking”patentsmightprotectafirm’sonce-exclusivemarketasitbecomescommercialized.Thisproject,forthefirsttimeintheliterature,empiricallyrelatesthisunexploitednatureofpatentholdingstoimportcompetitionfromalow-wageeconomy.

Forthispurpose,weextractedtherelevantdatafromaJapanesefirm-levelsurvey

—theBasicSurveyofBusinessStructureandActivity,conductedbytheMinistryofEconomy,TradeandIndustry(METIdata)5—coveringtheperiod1994–2005.

Thefirm-levelpatent-usagedataisthenmergedwithindustry-levelexposurestoChineseimportcompetition,resultinginauniquedatasetforthefollowingaspects:First,itprovidesapaneldatasetofpatentusageasitrelatestocompetitivepressures.Theavailablesurveystendtoreportsingle-yearresponses,onlydepictingthestaticnatureofpatentusage.6Usingapaneloffirm-leveldataofferstheperspectiveofwithin-firmvariationsofpatentusageinresponsetoimportcompetition.Second,thedataperiodislongenoughtocoverthePRC’schangingcomparativeadvantagefrommorelabor-intensivetomoreskill-andtechnology-intensivegoods.Third,usingpaneldataallowsfirm-specificeffectstobeincluded,because(unobserved)managerialskills(assumingtimevariantintra-firmelements)canbecontrolled,alongwithindustryandyearfixedeffects.Clearly,inacross-sectionalsetup,thiscannotbecontrolled.

Basedonfirm-levelinformation,wecreatedthepatent-usagevariablesasshowninFigure3.Inshort,foreachfirmwecountthepatentsowned(PAT),thepatentsinuse(USE)andpatentsthatarenotinuse(NON-USE).WithinPAT-USE,wehave

3Moregenerally,itismorecommoninthediscretetechnologyindustries.Inthepharmaceuticalindustry,R&Dcantakeaslongas10–15yearsbeforenewdrugscanbeintroducedintothemarket.Hence,thereisasubstantialnumberofpatentsfordrugsthatarestillintheprocessofR&Dandnotyetinthemarket.

4Itisimportanttonotethatthoseunusedpatentsmaysimplyreflectfirmsthatlacktheinternalassetstocommercialize,oraresearchingforlicensees.

5ThissurveyisgovernedbytheJapaneseStatisticsAct,andfailuretoreplyresultsinafine.Thesurveysampleisrestrictedtofirmsthathavemorethan50employeesandcapitalofmorethan30millionyen.Itcollectsfirms’accountinginformation(sales,employment,employmentcompensation,thenumberofestablishments,R&Dspending,exports,andimports).Theindustryclassificationisavailableata3-digitlevel.But,forourpurposeofanalyzingtheimpactofimportcompetition,werestrictedthesampletoonlymanufacturingfirms.Allindividualfirmsareassigneduniqueidentifiers,makingitpossibletotracktheoperationsofthesamefirmsovertime(thepaneldata).

6Motohashi(2008)usesthedatafromtheSurveyofIntellectualPropertyActivitiesbytheJapanPatentOffice(JPO)conductedin2001inordertoclassifypatentusage.Itwasfoundthatsomeofthepatentsarewithheldbyfirmswishingtousethem(orlicensethemout)inthefuture.Othersarekeptbecauseafirmneedsthemforfuturelicensingnegotiations.Thispracticeiscommonintheelectronicsindustrywherecross-licensingisoccursmorefrequently.

7

ADBIWorkingPaper939

YamashitaandYamauchi

informationforthenumberofpatentsbasedoninternalinventions(DEV),andthenumberofpatentsthatarelicensedout(LICENSE).Thesevariablesformthedependentvariablesintheregressionanalysisthatfollows.

Figure3:PatentUsageandVariableDefinitions

VariableSymbol

BriefExplanationsandDefinitions

PatentOwned

PAT

Thecountofpatentsowned(includingthosepurchasedandcross-licensed)reportedbyafirminagivenfiscalyear.Thisincludesthecumulativecountofpatentsownedbyfirms,notjustpatentsforwhichapplicationhasbeenmadeinagivenyear.

Use(includinglicensedout)In-houseinventions

-use

USEDEVNON-USE

Thosepatentscurrentlyinuse.

Patentsbasedoninternalinventionsthatareinuse.DefinedasPATminusUSE,includingblockingandfuturecommercialuse/negotiation.

Licensedout

LICENSE

Totalcountofpatentswhicharelicensedout.Domesticandinternationalsegregationisavailable,aswellastheamountofmoneyreceived.

Itisimportanttonoteseverallimitations.First,thepatentstatisticsinourdataisapatentpool—allpatentsinwhichthefirmshaveownership.Empiricalworkthatusespatentstatisticscollectedfromthepatentofficenormallycoversthosepatentsforwhichapplicationhasbeenmade,aswellthosepatentsthathavebeengrantedtothefirm.Inourdata,allpatentsarepresumablythosegranted(becausethesurveyquestionaskshowmanypatentsafirmowns,ratherthanpatentsthathavebeenappliedfororarebeinggranted).Sincepatentapplicationscanindicatefirms’innovativeefforts,ourmeasuremayunderestimatethem.

Second,ourpatentdataissimplythecount.However,otherstudiesemployingpatentstatisticsusuallyweightthepatentcounttoits(backwardandforward)citations,thuscontrollingforpatentquality.Thehigherquality(orsometimesmorebasic)inventionsattractmoreforwardcitationsthanlowerqualityinventions(sometimes,referredtoas“patentthickness”).Withouttheabilitytolinkourdataonfirm-ownedpatentswiththecitationinformation,weareunabletoaccountforthisqualitydimension.

Third,ourdatadoesnotadjustforthedepreciationrateofoutdatedpatents.Itisappropriatetoadjustforthedepreciationrateofpatents,becausesomefirm-heldpatentscanbecomeobsolete.However,withnoidentificationofthegrant(orapplication)dateofeachpatent,thedeprecationratecannotbeappliedinourdata.Wethereforelookatthegrowthrateofeachpatentusage(ratherthanasimplecount),hopingtominimizethebiascomingfromthenon-depreciationofthepatents.

3.2JapanIndustryProductivityData

IndustrylevelvariablesusedintheregressionanalysisaremainlysourcedfromJapanIndustrialProductivity(JIP)data(JIP2013)storedintheonlinedatabaseintheResearchInstituteofEconomy,TradeandIndustry(RIETI)inJapan.7TheJIPdatasetisorganizedatthe3-digitindustrylevel(52manufacturingindustries).

7http://www.rieti.go.jp/en/database/JIP2013/SeetheAppendixforfurtherdetailsonJIPdatabase.

8

ADBIWorkingPaper939

YamashitaandYamauchi

3.2.1ChineseImportCompetition

WeusethevalueofimportsoriginatingfromthePRC(IMthePRC)asashareoftotalworldimports(IMWorld)asameasureoftheexposuretoChineseimportcompetitioningivenJIPindustries(asubscriptj).

CHM=Chineseimportsj.(1)

jImportsj

WealsoemploytheconventionalmethodofconstructingChineseimportpenetrationbynormalizingChineseimportondomesticabsorption(i.e.,domesticabsorption

=valueadded+imports–exports).8

CHM=Chineseimportsj

j(ValueAddedj+Importsj-Exportsj)

(2)

3.2.2InstrumentalVariable

WhileourmotivationfortheempiricalanalysisistoestimatethecausaleffectsofChineseimportcompetitiononpatentoutputs,weencounterthepossibleendogeneityproblem:Firm-levelinnovativeactivityforreasonsotherthanChineseimportcompetitionmayalsoshapetradeflows,alteringthedegreeofimportcompetitionintheindustry(forexample,moreinnovativefirmsmightopttodomoreoffshoringtothePRCinordertofacilitatetheirinnovativehomeoperations).Forthesamereason,thereversecausalityisalsoapossibility:ImportsfromthePRCmaybecorrelatedwithindustry-widetechnologyshocks(tosomedegree,industry-specificfixedeffectsmaytakecareofthisconcern,butitmightnotbesufficient).Thismakesordinaryleastsquare(OLS)estimatorsbiasedandinconsistent.

WeusedameasureofChinese(labor)productivityasaninstrumentfortheendogenousChineseimportvariablesinthetechnologyequation.Thisimpliedvolatility(IV)strategyextractsanyexogenousvariationsaffectingChineseexportsupplycapacity,whileindirectlyaffectingthelevelofinnovativeactivityonlythroughtheintensifiedimportcompetitioninJapan.Thisinstrumentisinspiredbytheuseofaninstrumentinotherstudies:Autoretal.(2015)usedtheexposuretoChineseimportcompetitionofeightadvancedeconomies9asinstrumentstomeasureUSexposuretoChineseimports.ThemotivationfortheirIVstrategywastoextractsupply-sideproductivityelementsinChineseexportperformance.However,aspointedoutbyAutoretal.(2015),theirinstrumentfacesavaliditychallenge,wherebyindustrytechnologicalchangesamongthoseadvancedeconomiesmustbeseparateincidents.Inotherwords,thetechnologicaldiffusionsmustbelimitedacrossthosehighincomeeconomies.InourimplementationoftheIVstrategy,wedirectlyusedtheproductivitymeasure(laborproductivity)ofChineseindustries,whichundoubtedlyhasbeenbehindthesurgeinChineseexportgrowth,yetisindirectlyrelatedtofirm-levelinnovativeactivity.ThesedataareextractedfromthePRCIndustrialProductivity(CIP)database.10ThereisnostrictindustrycorrelationbetweenCIP

8“Value-added”isdefinedasthedifferencebetweengrossoutputandintermediateinputs.Grossoutp