ChinaEconomicReview21(2010)482–(.).Inbothgraphs,theverticalaxisshowsthemeasuredexportsophisticationofacountry(EXPYandESIaretwoexportsophisticationindices,whichwewilldiscussinSection2),andthehorizontalaxisshowsacountry'spercapitaGDP(PCGDP).ThegraphsrevealapositiverelationshipbetweentheexportsophisticationindexandPCGDP,withChina(CHN)(2006)as“special”.'soutlierstatusisdeterminedbythevaluesoftwoparameters:therelativelyhighlevelofChina'sexportsophisticationandtherelativelylowlevelofChina':arethesetwoparametersproperlyAnevenmorestartlingfeatureosophisticationofChina'sexportsfarexmeasured?ThesophisticationindicesusedSchott(2004)foundthatevenwiththem ChinaEuropeInternationalBusinessSchool(CEIBE-mailaddress:xubin@'-951X/$–seefrontmatter': edbyRodrik(2006)andSchott(2008),,leatherandfootwear”,,thisindustry'sexportradio,%ofChina''srapidupgradeonChina's14manufacturingindustriesrankedbyR&,“Textiles,&D-intensiveindustry,%ofChina'%.TakingitsplacearetwoR&D-intensiveindustries,“Electricalmachinery;ent”and“Machineryandequipment;of ce,accountingandcomputingmachinery”,:IsChinaspecial?BinXU ChinaEuropeInternationalBusinessSchool,ChinaarticleinfoabstractArticlehistory:RecentstudieshavefoundthatChinais“special”“Chinaisspecial”,productqualityhasnotbeenfullyAccepted19April2010consideredinthemeasurementofsophistication,whichhascausedanoverestimationofthesophisticationofChina',theaverageincomeofChinahasbeenusedtoJELclassi cation:measuretheexportcapabilityofChina,whichhascausedanunderestimationofChina',:'(2006)andSchott(2008)areconstructedfromproduct-leveltradedataostdetailedproductclassi cation(HS10-digit),theproductvarietiesfromdifferentS),699HongfengRoad,Pudong,Shanghai201206,.:+862128905602;fax:+(2006)....
ce,accountingandcomputingmachinery(29,30)(24)−(27),re nedpetroleumproducts,nuclearfuel(23)(25)-metallicmineralproducts(26)−,beveragesandtobacco(15,16)−(28),paper,printing,andpublishing(20,21,22)−,jewelry,musicalinstruments,sportsgoods,gamesandtoys,andothermanufacturing(36,37)−,apparel,leatherandfootwear(17,18,19)−:R&DintensityistheratioofR&DexpendituretogrossoutputofChinese rmsintherespectiveindustryin2001,&DexpenditureandgrossproductarefromChina'snationalsurveysof '.,(2008)viewedthispricedifferencewithinthesameproductcategoryasre ecting“within-productsophistication”.HefoundthatwhileChina'sacross-productexportsophistication(measuredbyEXPYorESI)isexceptionallyhigh,,althoughmanyofChina'sexportedgoodsbelongtosophisticatedcategories,,onewouldoverestimatethesophisticationofChina',thesophisticationofacountry'sexportstructureisdeterminedbythecountry''sexportcapability,itcanserveasthesinglevariableinpredictingthecountry',forcountrieswithhighregionalheterogeneity,'scoastalprovinces,whichaccountforover90%ofChina'sexports,',“Chinaisspecial” ndingisthatChina'sexportsappear“special”becausethesophisticationmeasuresdonotfullyconsiderproductqualityandbecauseChina'(,2008), ndthatChina'soutlierstatusisreducedafterintroducingproduct-qualityorPCGDPadjustmentstotheanalysis,buttheindividualeffectofeachisnotlargeenoughtorejectthe“Chinaisspecial”, ndingisusefulinthecontextofthedebateonthenatureofChina'(2006)consideredChina'soutlierstatusasevidencethatChina'sexportexperiencewasneither“asimplestoryofspecializationaccordingtocomparativeadvantage,”nor“astraightforwardstoryofexportgrowthachievedthroughtradeopennessandfreemarketforces.”Our ndingshowshoweverthatChina''sexportsophistication,anumberofrecentpapershaveexploredthereasonsbehindthehighsophisticationofChina'(2010) ndthatimprovementsinhumancapitalandgovernmentpoliciesintheformoftax-favoredhigh-techzoneshavebeenkeydeterminantsofChina'(2010)arguethatthehighsophisticationofChina',they ndnoevidencethatChina'(2010)alsoarguethattheobservedhighsophisticationofChina''stotalexportsincreasedsigni cantlyfrom1992to2005,,foreign rmsarealsoconsideredamajorforcebehindtheriseofChina'(2009) ndthatforeign rmsfromadvancedcountrieshavecontributedsigni cantlytotheincreaseofthesophisticationofChina'“Chinaisspecial”,'.
rstoneisan“incomecontent2constructedbyRodrik(2006)andHausmann,HwangandRodrik(2007).Thisindex,denotedbyEXPY,measurestheticationofcountryc'sexportsasEXPY=∑SPRODYð1Þciicj∈IilarindiceshavebeenconstructedbyMichaely(1984)andLall,WeissandZhang(2006).
esthesameequation(,Eq.(3)).5ForqualityofChina'sexports ,neitherEXPYnorESInecessarilyre ectthesophisticationofa5y',-PRODYgoodbecauseofthemultinational rstdevelopedbyFingerandKreinin(1979).Schott(2008)(2010)(2006)doesnotspecifythisregressionequationexplicitly,() ':,(IND)andMexico(MEX) ttedline,suchasHongKong(HKG).Thustherecouldbemany“CountryXisspecial”period1992–()and(),theestimatedcoef cientsofCHNarepositiveandstatisticallysigni cantatthe1%level,()and(),theeffectofCHNisestimatedforthreesub-periods(1992–1995,1996–2000,and2001–2005).Theresultsshowthattheestimatedcoef cientsofCHNremainpositiveandstatisticallysigni cantatthe1%level,andthepointestimatesshowadecliningtrendovertime,aswasalsonotedbyRodrik(2006).Togetamoreconcreteideaabouthow“special”Chinais,wedisplayinTable3thegroupoftencountriesatPCGDPlevelssimilartoChina',China'sEXPY(7356)andESI()werebothsigni cantlyhigherthanthegroupaverage(4250;);Indiaalsoappearstohavebeenanoutlier(7358;).In2004,,China'sEXPY(12,524)andESI()werebothsigni cantlyhigherthanthegroupaverage(7225;);Venezuelaalsoappearstohavebeen“special”(10,985;).logðEXPY⁄ESIÞ=α+βlogðPCFDPÞ+γCHN+εð4ÞctcttctctCHNisadummyvariableforChinaandε(realGDPpercapita).(2),thesophisticationofgoodiisgivenbythejweightedaverageoftheincomelevelsofallcountriesinsetCthatexportthegood,wheretheweightvariable(s/∑s)re ectsiijintheimportanceofgoodiincountryj'-incomecountries,'sexportsisrevealedbythe“incomecontent”,ExportSimilarityIndex(ESI),computestheoverlapofacountry'sexportswiththatofasetofadvanced3countries:ESI=∑minðS;SÞð3Þcdicidi∈IInEq.(3),ESIisthesimilaritybetweencountryc'sexportbundleandcountryd',ESIrevealsthesophisticationofcountryc'“Chinaisspecial”resultTworecentstudies(Rodrik,2006;Schott,2008)examinedtherelationshipbetweenacountry'sexportsophistication(EXPYorESI) (2006)andSchott(2008)maybereplicatedusingthefollowingregressionequation:
()()()() ()() CHN(1992–1995)()() CHN(1996–2000)()() CHN(2001–2005)()() ()()−()−()Year :absolutevaluesoftstatisticsareinparentheses. Signi cantat1%.Table3ExportsophisticationofcountriesinChina',itisalsoratherpossiblethatacountryexportsahighlysophisticated &Dintensityfor2-digitISICindustriesofOECD6countries,,andonlythreeresource-intensiveindustries(fuel,rubber,andbasicmetals)showestimatesthatarestatisticallyinsigni &,however,donotcaptureoneimportantdimensionofexportsophistication,.,(2004),evenatthemostdisaggregatedproduct-leveldataofinternationaltradestatistics(10-digitHScode),,therewere26countriesexporting“Linetelephonesetswithcordlesshandsets”(HS=8,517,110,000)$,$44,$38,$55,andJapanexported463274setsat$(2008)pointedoutthatsuchpricedifferencessignalproduct-qualitydifferences,andhede nedthesophisticationrelatedtoproductqualityas“within-productsophistication”incontrasttothe“across-productsophistication”measuredbyEXPY/&Dintensityofprevious veyearsasindustryR&Dvariableforcurrentyear,&D-GDPratioaveragedover1996–2004astheR&Dvariablefor2005,we cantatthe1%level..f
ce,accountingandcomputingmachinery(29,30)***222Electricalmachinery;radio,televisionandcommunicationequipment(31,32)**210Medical,precisionandopticalInstruments;watchesandclocks(33)**197Transportequipment(34,35)***226Furniture,jewelry,musicalinstruments,sportsgoods,gamesandtoys,andothermanufacturing(36,37)*226Notes:thesampleis18OECDcountriesin1992–2004(seetheAppendixfordatadetail).Industry-levelPRODYistheweightedaverageofHS10-levelPRODY,&Dintensityistheaverageofprevious veyears.***,**,*indicatestatisticalsigni canceatthe1%,5%,and10%(HS=8,517,110,000),(θ=1/2)CHN11,300,,183,,510,,457,,210,,,,,,,,,,,,,,,,,,,399Notes: -productqualitydifference,weconstructthefollowingrelative-priceindex:picq=ð5ic∑ð pÞininn∈CiInEq.(5),,theweight beingcountryn'sexportshareofgoodiinallcountries'exportsofgoodi,whichre (2004)thatforthesameproduct,thereisapositivecorrelationbetweenthepriceofacountry'“Linetelephonesetswithcordlesshandsets.”Wemayinferfromthispositive.ÞiOthernon-metallicmineralproducts(26)***226Basicmetals(27),apparel,leatherandfootwear(17,18,19)***226,paper,printing,andpublishing(20,21,22)***226re nedpetroleumproducts,nuclearfuel(23)(24)***213randplastics(25)&(ISICcodes)CorrelationObservationsFoodproducts,beveragesandtobacco(15,16)***226Textil
:logðqÞ=α+βlogðPCGDPÞ+γCHN+εð6Þ-squNotes: SnshipbetweenpriceandPCGDPthatrichcountriesgenerallyexportproductsofhigherquality,,Schott(2008)“linetelephonesetswithcordlesshandsets”,.(6),αisaproduct-year (),theestimatedcoef cientofCHNisnegativeandstatisticallysigni cantatthe1%(), cientsofCHNinthethreesub-periodsareallnegativeandicallysigni cantatthe1%level,-adjustedmeasureofsophisticatione ndingthatthepricesofChina'sexportsareexceptionallylowisimportanttotheevaluationofChina',thenEXPY/ESIoverestimatestheexportsophisticationofChina.-priceregressions,HS10products,1992–2005.()() ()() −() (1992–1995)−() (1996–2000)−() (2001–2005)−() ant−()−()ct-year xedeffectsYesYesvations2,607,8792,607, cantat1%.
()()()()()()() ()()()()()()−() ()()()()()()() ()()()()()()()HKG−−−−−−− ()()()()()()() ()()()()()()()Year “special”),whileHongKonghasanEXPYlevelsigni θincreases(.,asmorequalityadjustmentisintroducedtothesophisticationmeasure),theestimatedcoef cientsofCHN,INDandMEXallfalsteadily,,whenθ=2/3,Chinaisnolongeranoutlierintermsofthestatisticalsigni canceoftheChinadummy,, cantlyhigherthanthepredictedlevels(sotheylogðQEXPYÞ=α+βlogðPCGDPÞ+γCHN+γIND+γMEX+γHKG+εð8Þcttct1234ctInEq.(7),(q)isthequalitymultiplier,whereθ(θ=0),QPRODY=-qualityvariety('scottonshirtsmadeinItaly,q=)thatbelongstoaproductcategoryoflowsophistication(HS=6,105,100,010,“Men'sshirtsofcotton,knit”,PRODY=3323in2001),andtherelativelylowsophisticationofalow-qualityvariety(,q=)thatbelongstoaproductcategoryofhighsophistication(HS=8,528,304,000,“Videoprojectors,CLR,non-hidef”,PRODY=25093in2001).'sexportsophisticationtoqualityadjustmentToseetheimportanceofqualityadjustmentinevaluatingChina'sexportsophistication,weperformasensitivitytestusingthefollowingregressionequation: nethefollowingquality-adjustedindexofsophistication:θQPRODY=ðqÞ×PRODYð7ÞiciciθNotes:. Signi cantat5%. Signi cantat1%.Howdoweincorporateproductqualityinthemeasurementofsophistication?-productqualitydiffere
:China'sStatisticalYearbook,. Municipalitydirectlyunderthecentralgovernment(equivalenttoprovince).increasinglyanoutlierasθincreases,whichisconsistentwiththefactthatalargeportionofHongKong'sexportscomefrommainlandChinasoHongKong'sexportsophisticationisadjusteddownwardafterconsideringthequalityofitsexports.θWhatisapropervalueofθ?Table8showsthesizeofthequalitymultiplier(q)atdifferentvaluesofθ.Forexample,anadjustmentofθ=1/2impliesthefollowingeffects:foraproductatthe1stpercentileoftheqdistribution,QPRODY=×PR-8ODY;foraproductatthe99thpercentileoftheqdistribution,QPRODY=דcorrect”valueofθ.However,wecanruleoutθ=θ=1sinceitseemstointroducetoomuchqualityadjustment,makingQPRODYonly2%ofPRODYfortheproductsatthe1stpercentileoftheqdistribution,θforChinais2/3abovewhichtheestimatedcoef cientofCHNisstatisticallyinsigni ,wechooseθ=1/,withthismoderatequalityadjustment,China,IndiaandMexicoremainoutliersinthecross-countryrelationshipbetweenexportsophistication(measuredbyQEXPY)θ,ourresultsshowthatthe“Chinaisspecial”'',theconsiderableheterogeneityamongChineseregionsandthefactthatmostofChina'''sPCGDP;thenumbersrevealthattheincomeleveloftherichestregion(Shanghai)was13timesthatofthepoorestregion(Guizhou),forexample,the1-percentileproducthasq=,;the99-percetileproducthasq=,,thelastcolumnofTable5showsQPRODY(θ=1/2)of“Linetelephonesetswithcordlesshandsets”.,t
cientofCHNremainspositiveandstatisticallysigni cant,()ofTable11showstheresultsofreplacingEXPYwiththequality-adjustedsophisticationmeasure(QEXPY),theestimatedcoef cientofCHNremainspositiveandstatisticallysigni cant,().Inregression(), ,weseethatChinadisappearsasanoutlierinthecross-countrycomparisonofthesophisticationofexportsoncethemeasurementofsophisticationconsidersproductqualityandthemeasurementofacountry'λtoseethesensitivityoftheresultsofregression().Wecreportinregression()oneoftheexperimentsinwhichλ=-andlow-incomecountriesincludingc10Schott(2008)notedtheimportanceofconsideringtheinter-regionalrelativeendowmentdisparitieswithinChina,',,thereportedincomelevelofcoastalprovincesmaynotre 'sPCGDP(Table9). rstexperiment,weassumeλ=-incomecountriesandλ=,,weusePCGDPfromEq.(10) (),(),',eccer,isthatwedonothavedatatocomputeλ,-incomecountriesandrelatively11overallPCGDP, ningλastheratioofccountryc'sexport-region-weightedPCGDPtothecountry'soverallPCGDP,wehavePCGDP=λPCGDPð11Þecccq.(10),Rdenotesthesetofexportregions,PCGDPisthepercapitaGDPlevelofregionr,'sPCGDPissigni cantlyhigherthanChina'(SD)3686(1211)6640(2702)()Notes:PCGDPisChina''sprovincialPCGDP,%ofChina'(secondcolumn)andhighsharesofforeigndirectinvestment(lastcolumn).ItisevidentthatChina'soverallGDP10signi cantlyunderestimatesthecapitalandtechnologycapacitypossessedbyChina',weconstructthefollowingincomevariablethatusesprovincialexportsharesasweights:PCGDP=∑SPCGDPð10Þerrr∈R
HKG−()−()−()−()−()ofthewholecountryhasbeenusedtomeasureexportcapacity,whichhascausedanunderestimationofthecapacityofChina'stoaproperevaluationofChina'“Chinaisspecial”,severalrecentstudieshavestartedtoexplorethereasonsbehindChina'sincreasinglysophisticatedexportstructure,andhaveexaminedtherolesofforeigndirectinvestment,processingtrade,,IndiaandMexicosuggestthatdifferentcountriesmayhavedifferentreasonstobe“special”,BillKeithler,Shang-JinWei,JeffreyZax,andparticipantsinconferencesandseminarsinBeijingUniversity,FudanUniverisity,NankaiUniversity,ShanghaiUniversity,ShanghaiUniversityofFinanceandEconomics,SoutheastUniversity,andXiamenUniversityforusefulcommentsonearlierdraftsofthepaper,andSallyHu,(No70773121) ndingisnotsurprisingasalargeportionofHongKong'smanufacturingexportsarefrommainlandChina..“special”“Chinaisspecial”result,'smeasuredexportsophisticationisparticularlysensitivetoproductqualityincomparisonwithotherdevelopingcountries,'soverallPCGDPisapoormeasureofitsendowmentandtechnologycapacitytoexportsophisticatedgoods, ()()()()()Year :absolutevaluesoftstatisticsareinparentheses. Signi cantat10%. Signi cantat5%. Signi cantat1%.China,andλ= cantatthe10% rmthattheoutlierstatusofChinaisconsiderablyreduced(ifnotcompletelyremoved) cientsonINDandMEXremainpositive,sizableandstatisticallysigni cantinregressions()and().Thissuggeststhatthemeasurementbiasesareamoresigni ' uentialpapersbyRodrik(2006)andSchott(2008)foundthatChinais“special”,weprovideanevaluationofthetwokeyparametersthatestablishtheoutlierstatusofChinaincross-countrycomparisons,namelytheexportsophisticationindexandthecountry' ndthatitwastheimpropermeasurementofthetwoparametersthatledtothe“Chinaisspecial”,productqualityhasnotbeenfullyconsideredinthesophisticationmeasurementwhichhascausedanoverestimationofChina',theaverageincome()()()()()logEXPYlogEXPYlogQEXPY(θ=1/2)logQEXPY(θ=1/2)logQEXPY(θ=1/2) ()() ()() () ()()()()() ()()()()() ()()()()()
ce.(1995).“:UnitValuesVaryWidelyforIdenticallyClassi edCommodities,”ReportGAO/,Ari,&Gangnes,Byron(2010).Electronicsproductionupgrading:isChinaexceptional?AppliedEconomicsLetters,17(issue5),477−,Zhi,&Wei,Shang-Jin(2010).WhataccountsfortherisingsophisticationofChina'sexports?InRobertFeenstra,&Shang-JinWei(Eds.),China'sGrowingRoleinWorldTrade.:,Bin,&Lu,Jiangyong(2009).Foreigndirectinvestment,processingtrade,andthesophisticationofChina',20(2009),425−439.
