Jobgoogleadsence

NBSWorkingpaper7/2019Thelimitsofforeign-ledgrowth:DemandfordigitalskillsbyforeignanddomesticfirmsinSlovakiaJanDrahokoupil,BrianFabowww.
nbs.
skNárodnábankaSlovenska2019research@nbs.
skThispublicationisavailableontheNBSwebsitewww.
nbs.
sk/en/publications-issued-by-the-nbs/research-publicationsTheviewsandresultspresentedinthispaperarethoseoftheauthorsanddonotnecessarilyrepresenttheofficialopinionoftheNationalBankofSlovakia.
ISSN2585-9269(online)3Thelimitsofforeign-ledgrowth|NBSWorkingpaper7/2019Thelimitsofforeign-ledgrowth:DemandfordigitalskillsbyforeignanddomesticfirmsinSlovakiaJanDrahokoupil1,BrianFabo2Abstract:ThispaperaddressesdemandforskilledlabourinSlovakia,acountrythatischaracterizedbyahighdegreeofeconomicintegrationthroughinwardforeigninvestmentandthroughinternationalbackwardlinkageswithinglobalvaluechains.
Developingexistingapproachestopoliticaleconomyandglobalproductionnetworks(GPNs),ourframeworkdistinguishesbetweendemandfordigitalskillsontwolevels:occupationalstructure;andskillcontentwithinoccupationaltypes.
Inthisway,wecanassessnotonlywhatkindofworkersarehiredbycompanies,butalsowhatkindofspecificskillsarerequiredfromtheseworkers.
Usingalargedatasetonvacanciesfromaleadingjobportal,combinedwithadministrativedataoncompanysizeandownership,weshowthatforeignandmixed-ownershipcompaniesgenerallyadvertiseforhigherskilledoccupationsthandomesticfirms,buttheirskillrequirementsforthesejobsarelowerthaninsimilarjobsindomesticcompanies.
Foreigncompanieshavehigherskillrequirementsonlyinsomeblue-collarjobslinkedtoassemblyandcomponentmanufacturing.
Forwhitecollaroccupations,domesticcompaniesaremorelikelytorequiredigitalskills.
ThefindingsconfirmourexpectationsaboutthepositionofSlovakiaasacountryinanintegratedperiphery,wheremultinationalcompaniesareheavilypresentbutrarelybringcomplexactivities.
Ourkeypolicyimplicationisthatforeigndirectinvestmentintheintegratedperipherybringsonlyalimitedpotentialfortechnologytransfers.
Keywords:skills,foreigndirectinvestment,FDI,digitalization,globalproductionnetworks,jobvacanciesJELClassification:J24O331EuropeanTradeUnionInstitute(Brussels),JDrahokoupil@etui.
org2NationalBankofSlovakiaandComeniusUniversity(Bratislava),brian.
fabo@nbs.
skThelimitsofforeign-ledgrowth|NBSWorkingpaper|7/20194IntroductionHavingexperiencedanimpressivecatch-upgrowthrelativetotheEUaveragein2003-2008,theprocessofeconomicconvergenceinSlovakiasloweddown,ifnotstalled,afterthecrisisof2008.
Thegrossnationalincomeperpersonremainedrelativelystablesince2013,ataround76%oftheEUaverage.
Amoreintensiveuseofskilledworkersisakeyrequirementforincreasingproductivitytounderpinasustainedprocessofconvergence.
Digitalskillshavebecomeparticularlyrelevantinthecontextofthetechnologicalchangethatcharacterizedlastdecades.
Theeconomiccatchupinthe2000swassupportedbytheinflowsofforeigndirectinvestment(Havlatetal.
,2018).
Slovakia,asothercountriesintheregion,reintegratedintotheworldeconomyafterthebreakdownoftheComeconthroughproductionnetworksorganizedbyMNCs(MyantandDrahokoupil,2011).
MNCsmodernizedtheindustrialbasethroughacquisitionsandreorganizingsupplierrelations.
Greenfieldinvestmentthenbroughtnewactivities,notablyinmetalandelectronicsmanufacturing,complementingexistingindustries.
Foreigninvestorsalsotookcontrolofamajorpartofnon-tradeableservicessuchastelecomsandbanking.
Inthe'dependent'(NlkeandVliegenthart,2009)or'FDI-based'(MyantandDrahokoupil,2011)economies,theMNCs,ratherthandomesticinstitutionsandactors,haveamajorroleingeneratinginnovationandskillsthatunderpintheircomparativeadvantage.
Importantly,dualeconomicstructurescharacterizethedependenteconomiesinCEECs,withlargegapsbetweendomesticandforeignsectors.
Foreign-controlledcompaniesgeneratemuchhighervalueaddedthandomesticcompanies.
Furthermore,foreign-controlledcompaniesaretechnologyleaderswithmostoftheinnovation,andresearchanddevelopment(R&D),concentratedinforeign-controlledsectors(e.
g.
RadosevicandCiampiStancova,2018;Knell,2017).
InnovationandproductivitygapsbetweenforeignanddomesticsectorsarewidelyseenasachallengefortheregioninsustainingupgradingandconvergencewiththerestoftheEU.
Atthesametime,economicintegrationthroughglobalproductionnetworks(GPNs)offersopportunitiesforeconomicupgradinganddevelopment(Hendersonetal.
,2002;Mahutga,2014;WorldBank,2017).
Establishinglinksbetweenforeign-controlledinnovationleaders,domesticcompaniesandinstitutionscanbringindustriesintheregionclosertoinnovationfrontiers(KravtsovaandRadosevic,2012).
ParticipationinGPNscanleadtohigheroutputandproductivityasMNCsincreasedemandforskilledworkersandtheirtrainingandastrainedworkersmovetolocalfirms(EngelandTaglioni,2017).
However,despitetheirtheoreticalimportance,thereislittleresearchonthedifferencesinskillusebetweenforeignanddomesticcompaniesinSlovakiaandotherFDI-basedeconomies.
Weaddresssuchdifferencesinthispaper.
Morespecifically,wedevelopaframeworkonskilluseinFDI-basedeconomiesandanalysemicro-leveldataonthedemandforskillsinSlovakia.
Characterizedbyoneofthehighestdegreesofdependenceonforeigninvestment,thecountrycorrespondscloselytotheidealtypicalmodelofadependentmarketeconomy.
Ourpapercontributestotheliteratureonglobalproductionnetworksandcomparativepoliticaleconomybydrawingonananalyticalapproachdevelopedinempiricallaboureconomics.
ThelatterprovidesanempiricalunderstandingoflabourdemandbycompaniesthatismissinginpoliticaleconomyapproachestoGPNs.
Inturn,weenhancetheunderstandingofferedinempiricallaboureconomicsbyincorporatingglobalproductionnetworksandcomparativepoliticaleconomyperspectives.
Thelatterframeworksprovideaconceptualizationofthestructureandorganizationofvalue-addingactivitiesthatistypicallynotconsideredinempiricallaboureconomics.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/20195Webeginbyderivingexpectationsofthedifferencesinskilldemand,concerningbothoccupationalstructureandskillusewithinoccupationaltypes,betweenforeignanddomesticsectors.
Whiletheforeignsectorcanbeexpectedtodemandhigher-skilledoccupations,weexpectMNCstohavehigherskilledrequirementsthandomesticcompaniesonlyinalimitedsegmentofprofessions.
Ouranalyticalstrategyreliesoninformationfromjobvacanciesobtainedfromajob-searchportalthatwematchwithinformationoncompaniesfromtheSlovakorganizationsregister.
3Ourdatasetcontainsdetailedinformationthatallowsustodistinguishskilldemandonthetwolevels.
Wecanalsotakeintoaccountthestructureoftheeconomy,comparingbetweendifferenttypesofcompanies.
Wefocusondigitalskillsthatareassociatedwiththeuseofnewtechnologyandwithanexecutionofcomplextasks.
TheseindicateapotentialforinnovationandproductivityspilloversfromFDI.
Theempiricalfindingsbroadlysupportourexpectations.
Accordingly,foreign-ownedcompaniesgeneratemorejobsinhigher-skilledoccupations,butthespecificskillrequirementsforthesejobsarelowerthaninsimilarjobsindomesticcompanies.
Theexceptionrepresentsanarrowsetofoccupationsrelatedtocomparativeadvantageinassemblyandcomponentmanufacturing.
Whileourexpectationsandfindingsonthedifferenceswithinoccupationsmayseemcounter-intuitive,theycorrespondtotheresearchonthestructureofproductionnetworksinmanufacturingthatwediscussindetailinthefollowingsection.
Theexpectationsalsocorrespondtothemorestylizedobservationsintheservicesector.
Forinstance,intheITsector,largeforeignplayersincludemultinationalsthatoutsourcerelativelysimpleactivitiessuchasremotenetworksupport.
Incontrast,majordomesticITcompaniesincludeleadingantiviruscompaniesthatrelyonownresearchanddevelopment.
Yet,theseverydifferentactivitiesfallunderthesameoccupationcategoriesandwouldbeignoredbyanoccupationalanalysis.
Thedifferencescanbeonlyidentifiedthroughthemoredetaileddataonskilldemand.
GlobalproductionnetworksanddependentdevelopmentinCentralandEasternEuropeThedistinctionbetweenforeign-ownedanddomesticcompaniesplaysakeyroleintheresearchonGPNsandglobalvaluechains.
Whileforeignownershipindicatesthecontrolofstrategicresources(firm-specificadvantages),domesticcompaniesinless-developedcountriesarelikelytointegrateintoGPNsinasubordinateroleassuppliersofsimplercomponentsorservicesthatrelyonlow-skilledlabour(e.
g.
HumphreyandMemedovic,2003).
IntegrationintoGPNscontrolledbymultinationalsisthenseenasanopportunityforless-developedregionstoupgrade(Hendersonetal.
,2002;Coeetal.
,2004;WorldBank,2017).
Economicdevelopmentisthusconceptualizedastheoutcomeofthestrategiccouplingbetweenregionalassetsandtheneedsofglobalproductionnetworks.
Transfersofknowledge,suchassuperiorprocessandproducttechnologyandmarketingskills,betweenmultinationalcorporationsanddomesticfirmsthusrepresentsakeymechanismofdevelopment(BlomstromandKokko,2001;DunningandLundan,2008;PonteandSturgeon,2014).
Thepresenceofspilloversalongthevaluechain,andthenatureofthelinksbetweenactorsinGPNs,hasdominatedtheresearchagendaoftheGVC/GPNapproaches.
4However,suchapproacheshavebeencriticisedfortreatingthefirmasablackbox,withlittleattentionpaidtotheprocessesoflearningandskilldevelopmentandupgradingwithinthefirm(RamirezandRainbird,2010).
3TheregistryisanadministrativedatasourcemaintainedbytheStatisticalOfficeofSlovakia.
4EmpiricalfindingsonthedevelopmentalconsequencesofintegrationthroughGPNsremainsomewhatinconclusive,ifnotcontradictory.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/20196GPNsplayedamajorroleinthereintegrationoftransitioneconomiesinEasternEuropeinthelate1990sandearly2000s(Drahokoupil,2008;MyantandDrahokoupil,2011).
ForeigninvestorscontroltheproductionnetworksthroughwhichtradeablesectorsinCEECsareorganized.
TheFDIinflowshaveallowedtheregiontoreconstituteitselfastheindustrialheartlandofEurope,specializinginthelowvalue-addedstagesofglobalvaluechains(e.
g.
LeitnerandStehrer,2014).
Giventheirdependenceonthetransferofcapital,technologiesandoperationalskillsthroughGPNs,CEECshavedevelopedthefeaturesofan'integratedperiphery'(Pavlínek,2018).
Alargepartofnon-tradeableservices,suchasbankingortelecommunications,isalsoforeign-owned.
TheextentofthedependenceonFDIisillustratedinTable1.
Thetablealsoincludesthemostrecentdataontheimportcontentofexports,ameasureofintegrationintoglobalvaluechainsthroughinternationalbackwardlinkages.
Table1ForeigncontrolandintegrationintovaluechainsImportcontentofexports,%,2014FDIstock,%ofGDP,2017Foreigncontrolofenterprises,%,2016InwardOutwardValueaddedEmploymentSlovakia48.
1954.
93.
348.
128.
5Czechia46.
6165.
111.
643.
327.
6Hungary47.
3161.
2*20.
0*51.
425.
7Poland32.
9842.
95.
436.
829.
3Germany25.
3524.
240.
924.
811.
2UK21.
8757.
558.
128.
019.
0France26.
2731.
852.
816.
411.
0EU19(EMU)26.
57EU2824.
615.
3Notes:Theimportcontentofexportsisdefinedasforeignvalue-addedingrossexportsdividedbytotalgrossexports.
Sources:OECD(2019),Importcontentofexports(indicator).
doi:10.
1787/5834f58a-en,Eurostat[tec00105,fats_g1a_08],*datafromtheHungarianNationalBankwereusedasthecomparativedatasetsdonottakeintoaccounttheeffectsofspecialpurposeentitiesandtransfercapitalthatdistortsHungarianFDIdata(AntalóczyandSass,2015)CEECshavethusbeenconceptualizedas'dependent'or'FDI-based,second-rank'marketeconomiesinthepoliticaleconomyliterature(NlkeandVliegenthart,2009;MyantandDrahokoupil,2011respectively).
Inthedependentmarketeconomymodel,multinationalcorporations,ratherthandomesticinstitutionsandactors,playakeyroleingeneratingtheinnovationandskillsthatunderpincomparativeadvantage.
FDI-basedeconomiesarecharacterizedbydualeconomicstructuresthatconformtotheexpectationofGPNtheories:foreign-controlledcompaniesexhibithigherproductivityandinnovationintensitythandomesticcompanies.
Thelatterareintegratedintoglobalproductionnetworksaslower-tiersuppliersofmultinationalcorporations.
Foreign-ownedcompaniesalsopayconsistentlyhigherwagesthandomesticcompanies(Gottvaldetal.
,2013,controllingforrelevantcharacteristics).
Thedualnaturealsocharacterizesinnovationsystemsintheregion,withforeign-ownedcompaniesaccountingformostR&Dspendingandinnovation(Knell,2017;RadosevicandCiampiStancova,2018).
Infact,innovationinFDI-basedeconomiestendstoberestrictedtoupgradingoftheproductionprocessratherthanR&D.
Moreprecisely,twoparallelinnovationsystemscanbeidentifiedintheregion(Radosevicetal.
,2010).
ThereisthelargeFDI-centredsystem,targetedtowardsdownstreamactivitiesinproductionsuchasthedevelopmentofproductionprocesses.
Incontrast,thedomesticsystem,howeverweak,isR&Dbased:itThelimitsofforeign-ledgrowth|NBSWorkingpaper|7/20197consistsofahandfulofnewtechnologycompaniesspecializinginupstreamknowledge-intensiveservices.
ThenatureofdualeconomiesinCEECshasbeenestablishedthroughempiricalresearchthatcomparesthecomplexityofactivitiesinforeignanddomesticcompaniesandthenatureofthelinksbetweenthem.
Forinstance,studiesoftheautomotivesector,whichdominatesindustrialstructuresinSlovakia(seeTable2),haveshowngapsbetweendomesticandforeigncompanies.
Atthesametime,however,highervalue-addedfunctionsinforeignsubsidiarieshavebeenfoundalsotobeweaklydeveloped,withmostR&D-relatedactivitiesconcentratedoutsideofthecountry(Pavlínek,2016).
EconometricevidencehaspointedtobothpositiveandnegativespillovereffectsofFDIonproductivity(Jacobsetal.
,2017).
Moredetailedsectoralstudiesofspilloversandlinkagesintheautomotiveindustryshowonlytenuousanddependentlinkagesbetweenforeignsubsidiariesanddomesticfirms,withunconnectedforeignsubsidiariesthatoperateasassemblyplatformsbeingverticallyintegratedintoexternally-organizedGPNsthatareweaklyembeddedintheSlovakeconomy(FereníkováandFifeková,2006;Pavlínek,2018).
Intheservicesector,inwardinvestmentshavetargetednotonlynon-tradeableservices,buthavealsocreatedasubstantialsegmentofbusinessservicesprovidersgearedtowardsclientsabroad.
Thesegmentgeneratesconsiderabledemandinprofessionaloccupations,buttheskillintensityoftheseactivitiestendstoberelativelylow(CapikandDrahokoupil,2011;Mezihorak,2018).
Table2Sectoralstructure,2016Manufacturing,%Motorvehicles,%ValueaddedEmploymentValueaddedEmploymentSlovakia36.
631.
48.
24.
6Czechia39.
935.
38.
54.
6Hungary38.
527.
77.
93.
5Poland33.
128.
73.
32.
1Germany34.
325.
36.
42.
9UK15.
513.
11.
60.
8France22.
718.
61.
91.
4EU2826.
621.
42.
91.
8Sources:Eurostat[sbs_na_sca_r2]However,despitetheirimportanceinthetheoreticalframeworkslinkingGPNswithdevelopment,empiricalevidenceondifferencesinskilluseincompaniesremainsweak.
5ResearchonGPNsinintegratedperipheriestendstofocuson(thelackof)thetransferofskillsandknow-howandthelinkagesbetweenfirms,butthereislittleevidenceonwhatcanbetransferredasfarastheskillsusedincompaniesareconcerned(i.
e.
theskillgap).
DemandforskillsandoccupationsindualeconomiesThenatureofthedemandforskillshasbeenanalysedbyempiricalapproachesinlaboureconomics.
Thefieldhasundergoneashiftfromcanonicalmodelsassumingtwodistinctgroupsofworkers(skilledandunskilled)performingdifferentandimperfectlysubstitutabletaskstowardsframeworksconsideringinteractionsamongworkerskills,jobtasks,evolvingtechnologiesandshiftingtradingopportunities(AcemogluandAutor,2011;FreyandOsborne,2017).
Someofthebestempiricalstrategieshaveanalysedtaskcontentinjobsderivedfrom5Pavlínekandíalová(2016)offersomesmallsampleevidenceonskilluseinautomotivesuppliersinCzechia.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/20198theoccupationalclassification–forEuropeancountries,typically3-digitISCOinEU-LFS(e.
g.
Arntzetal.
,2016;KeisterandLewandowski,2017).
KeisterandLewandowski(2017)identifiedacomparativelyhighandgrowingshareofroutinecognitivetasksinCEECs,includingSlovakia.
ThismightbelinkedtoacomparativelylowinvestmentinICTintheregion(Arntzetal.
,2016)andalsotothegrowthinthebusinessservicesoutsourcingsectoridentifiedabove.
Amoregranularunderstandingofthedemandforskillsinthelabourmarketcanbeobtainedfromdataonjobvacancies.
Jobvacancyresearchhasalongtraditionineconomics(HoltandDavid,1966),butithasgrownasmoredatahasbecomeavailablewiththeshifttowardsonlinejobadvertising(AskitasandZimmermann,2015).
Suchresearchhasfocusedonthechangingdemandforindividualskillsonthelabourmarketandonthenatureofskillrequirements(Modestinoetal.
,2016;WadeandParent,2002).
AsimilarfocuscanbefoundinjobvacancyresearchontheSlovaklabourmarket:thisincludesananalysisoftheskillrequirementsforindividualgroupsofworkers(tefánik,2012;MytnaKurekovaandiliníková,2015)aswellasthewagepremiumassociatedwithforeignlanguageskills(Faboetal.
,2017).
Inthispaper,weextendsuchanalysisbycomparingthedemandforskillsbetweenfirms,henceaddressingthekeystructuralpropertyofdualeconomiesintheintegratedperiphery.
Statisticallyverifiableexpectationsonthedifferencesinthedemandforskillsbetweendomesticandforeign-ownedsectorscanbederivedfromGPNandpoliticaleconomyframeworks.
Atthelevelofoccupationalstructure,foreigncompaniesarelikelytoemploymoreworkersinhigher-skilledoccupationsastheyspecializeinmorecomplexactivitiesinthevaluechainsthandomesticcompanies.
Inthedependentmarketeconomymodel,MNCshaverelativelylowskillrequirements;theyareattractedtotheregionbythecombinationofrelativelylowlabourcostsandaskilledpopulationwithsubstantialknowledgeofthemediumleveloftechnologythatisusedincomponentmanufacturingandtheassemblyofindustrialgoods(NlkeandVliegenthart,2009).
However,weexpectthemtodemandmorecomplexoccupationsthandomesticfirmsasthelattertypicallyrepresentsecondorthird-tiersuppliersinmanufacturing,specializinginsimpleractivities.
Intheservicesector,demandformoreskilledprofessionaloccupationsislikelytobeassociatedwiththebusinessservicesoutsourcingprovidersthattendtoemploygraduates.
However,thepatternmaynotbereproducedoncewecomparethedemandforspecificskillswithintheoccupationalgroups.
Higherskillrequirementscanbeexpectedfromforeignfirmsonlyinoccupationsrelatedtoassemblyandcomponentmanufacturing(i.
e.
ISCO6–ISCO8).
Activitiesofforeign-ownedfirmsthatimporttechnologiesandoperationalskillstendtobelimitedtoassemblyoperationsandtheproductionofgenericandlabour-intensivecomponents,orthelabour-intensiveproductionoflow-volumespecialmodels(e.
g.
Pavlínek,2018).
Inthissegment,innovationstravelsfromforeigncompaniestodomesticonesandMNCsthuscanbeassumedtobethefirsttoincreasedemandforskillsrelevanttonewtechnologies.
Inotheroccupationalgroups,weactuallyexpectdomesticcompaniestohavehigherskilledrequirements.
First,domesticcompanies,whiletypicallyspecializinginlesscomplexactivities,needtocoveralsostrategicfunctionsthatMNCsarelikelytoservicefromtheirheadquarters.
Domesticcompaniesarethuslikelytohavehigherskillrequirementsinmanagerialandprofessionaloccupations(ISCO1andISCO2).
Second,professionalsemployedbybusinessservicesoutsourcingproviders,typicallyforeign-owned,arelikelytohaverelativelylowskillrequirements(henceanexpectationofarelativelylowerlevelofdemandfromforeigncompaniesfordigitalskillsinISCO2).
Finally,thedomesticR&Dsegmentisalsolikelytobeassociatedwithhigherskilldemandfromdomesticcompaniesinmorecomplexoccupations(ISCO1,ISCO2andalsoISCO3Techniciansandassociateprofessionals).
However,theoveralleffectsshouldbesmallgiventheweaknessofthedomesticR&Dsegment.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/20199Insum,wedistinguishbetweenskilldemandontwolevels:occupationalstructure;andskillcontentwithinindividualoccupations.
Atthelevelofoccupationalstructure,foreigncompaniesarelikelytoemploymoreworkersinhighlyskilledoccupations,astheytendtospecializeinmorecomplexactivitiesinthevaluechainsthandomesticcompanies.
However,atthelevelofspecificskilldemandwithinoccupations,foreignfirmscanbeexpectedtohavehigherskillrequirementsonlyinoccupationalcategoriesrelatedtoassemblyandcomponentmanufacturing.
AnalyticalstrategyWeanalyzetheuseofskillsindualeconomiesthroughacasestudyofSlovakia,acountrythatischaracterizedbyahighdegreeofdependentintegrationthroughinwardforeigninvestmentandinternationalbackwardlinkagesintoglobalvaluechains,relativeeventootherCEECs.
AsshowninTable1,thehighdegreeofdependenceonglobalvaluechainsinSlovakiaisnotuniqueintheregion.
Inanycase,SlovakiaapproximatescloselytheFDI-based,dependentmarketeconomymodel.
6Wefocusondigitalskillsasthesearelikelytobeassociatedwiththeuseofnewtechnologyandwiththeexecutionofcomplextasks.
Asdiscussedabove,highertechnologyandskillintensityshouldindicateapotentialforproductivityspilloversfromFDItodomesticfirms.
Wederiveinformationonskilldemandfromjobvacanciespostedonaleadingjobsearchportalduring2011-2017.
Vacancydatarefertojobopenings(flows)ratherthanthejobsincompanies(stocks).
However,thelong-termperspectivemaywellweakentheimportanceofsuchadistinction.
Demandforskillsexpressedthroughhiringisalsolikelytoreflectdemandsrelatedtotheintroductionofnewtechnology.
Jobvacanciesdataprovidemoredetailedinformationontheskillsrequiredforthejobthanavailablesurveysandcoveralongtimeperiod.
Moreover,thedataincludeacompanyidentifier,inconnectionwithwhichwehavebeenabletolinktoadministrativedata(viaafirmregistry)toobtaininformationonownershipofthecompanyanditssize.
Ananalysisofjobvacanciespostedbydifferenttypesofcompaniesrepresentsanoveltyofourapproachtojobvacancyresearch.
Thesevariablesaretypicallynotincludedinanalysisasaresultofthelackofavailabledata,despitetheassumedtheoreticalimportanceofownershipandcompanysize(forforeignownership,see,e.
g.
,Faboetal.
,2017).
Incontrast,empiricalcomparisonofdomesticandforeigncompaniesiscommonintheGPNanalyses.
However,thesetendtorelyonsmall-Nsamplesandcasestudiesofindividualfirmsorfirmnetworksbeforeextrapolatingconclusionstothelevelofnationalindustries(seeDickenetal.
,2001,p.
89).
Weanalysethenatureofdualeconomiesusingnational-leveldatathatwecanbreakdownalsototheindustrylevel.
OurlevelofanalysisthuscorrespondstothelevelsatwhichtheimplicationsofintegrationintoGPNsfordevelopmentneedtobetheorized(Bair,2005;Mahutga,2014).
Morespecifically,weexaminealargesampleofjobvacanciespostedontheleadingSlovakjobportal,Profesia.
7Wefirstanalysethedifferencesindemandforindividualoccupationsandthencomparedemandfordigitalskillswithinoccupations.
Inthiscontext,wedifferentiatebetweenbasic,intermediateandadvanceddigitalskills.
Ourclassificationfollowsamethodologyestablishedbythetestingofanexhaustivelistofdigitalskillsonawidescale6NlkeandVliegenthart(2009)thusgaveSlovakiaasanexampleofacoherentdependentmarketeconomy.
7Profesiahasanexceptionallyhighmarketshare,withabout80percentofonlinejobvacanciesadvertisedthere(tefánik,2012).
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201910ofonlinevacanciesfromtheUS(Beblavetal.
,2016a).
ThecompletelistoftestedskillsislistedinTable11intheAppendix.
Technically,wecomputemarginaloddsatthemeanthroughlogisticregressionwithabinaryvariabledenotingthepresenceofadigitalskillrequirementinavacancyastheresponsevariable.
Astheexplanatoryvariables,wehavedummiesforcompanycharacteristics–size;ownership(domestic,mixedandforeign);locationinthecapitalregionofBratislava8;andsector9–aswellastheminimumrequirededucationalattainmentlistedinthevacancy10andtheyearlyfixedeffect.
WepredictmodelsseparatelyforeachISCOoccupationalgrouptoalleviatebiasesarisingfromanimplicitassumptionoftheemployerthatajobcandidatewillhavearequiredskillonthebasisofhavinganother,explicitlylisted,requirement(Carnevaleetal.
,2014).
11Thisway,wecanlimitcomparisonstojobswithsimilaroverallskillrequirementsandhiringcultures.
Wederiveinformationonskillrequirementfrom'tags'associatedwithindividualvacancies(asinFaboetal.
,2017).
Theskilltagsusedbytheportalbroadlycoverknowledgeofforeignlanguages,alargesetofdifferentdigitalskillsandsomejob-specificskillssuchasaccounting.
12Theadvantageofourapproachisthatouranalysisisbasedonwhatemployersthemselvesspecifyastheskillrequirementsforadvertisedjobs.
Ourapproachthusallowsavoidingerrorsinextractingskillrequirementsfromthetextoftheadvertisements,ascommoninotherstudies(e.
g.
Beblavetal.
,2016b;Kureková-Mtnaetal.
,2016;Maurer-Fazio,2012).
Suchparsingiserror-proneastheskillsdemandedaretypicallynotpresentedinastandardizedmanner.
Ontheotherhand,thetextsofadvertisementsratherreflectthelinguisticandstylisticapproachesappliedbydifferentemployers(Bosellietal.
,2018;Carnevaleetal.
,2014).
InthecaseofProfesia,professionalpositionsinthefieldsofICTorfinanceareoftenadvertisedinEnglish,whileSlovakiscommoninotherprofessionsandsectors.
Manyadvertisementsarestructured,butidiosyncraticstylesarealsooftenused.
Forinstance,someadvertisementsaimingtorecruityoungworkerstendtouseverycolloquiallanguageandopttoskiptraditionallistsofskillrequirements.
Finally,inordertotestforthevalidityofmeasuringthedemandfordigitalskillsthroughjobvacancies,weestimateregressionmodelsalsowiththedemandforforeignlanguageskillsasaresponsevariable(Table13intheAppendix).
Weassumethatapositiveeffectofforeignownershipcanbeexpectedwithahighdegreeofconfidenceforthistypeofskill.
Inlinewithourexpectations,wefindahigherdemandforforeignlanguagesfromforeigncompaniesforalloccupationalcategories,exceptforprofessionaljobs,forwhichthepropensityofmixedownershipcompaniestorequireforeignlanguageproficiencywasfoundtobenarrowlylowerthanamongdomesticcompanies.
DataWebaseouranalysisonacompletedatasetobtainedfromtheProfesiajobportalcoveringtheperiod2011-2017.
13AcomparisonbetweentheintakeofvacanciesonProfesiawiththeinflowofvacanciesreportedbythePublicEmploymentAgency(UPSVAR)isshowninFigure1.
Thenumberofvacanciespostedontheonlineportalishigherthanthatfoundinthepublic8WhichappearstobetheonlyrelevantconsiderationforregionalskillbiasinSlovakia(tefánik,2012)9SizeandownershipinformationisobtainedbymatchingeachindividualemployerIDwithcompanyregistryadministrativedata.
Locationandsectorareself-reportedbytheemployer.
10Employerssignalvacancyrequirementsusingsimilartagstothoseusedtosignalskillrequirements.
11Forinstance,aworkerwithadegreeisunlikelytolackthebasicabilitytointeractwiththeinternet.
12Thetagsareassignedtoavacancyatthetimeofitspostingbytheprospectiveemployerwhochoosesfromalistof220checkboxesrepresentingindividualskills.
13Atthetimeofanalysis,thecompletedatafor2018werenotavailableandthuswedecidednottoincludetheincompleteyearinordertoavoidintroducingpotentialbiascausedbyseasonaleffects.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201911registry.
14Thetwoindicatorscovarytoalargeextent(r=0.
95),indicatingthattheyrespondtothesamesetofmacroeconomicvariables.
Nonetheless,neitherofthetwofiguresislikelytocovertheentireuniverseofvacancies,inparticularbecausetheobligationofemployerstoreportvacanciestoUPSVARwasdiscontinuedin200315.
Infact,thestructureofvacanciespublishedonlinediffersfromthatoftheemploymentagency:jobsadvertisedonProfesia16tendtobemorecommonlyfoundinwhitecollaroccupationalgroups(ISCO2-4),whiletheemploymentagencyismorelikelytoregisterbluecollarjobs(ISCO7-8).
Figure1InflowofvacanciespostedontheProfesiajobportalandregisteredbythePublicEmploymentAgencySources:Profesia,UPSVaR.
However,asshowninTable3,thedifferencesbetweenthetwodatasetshavediminishedovertime.
TherehasbeenadecreaseintherelativeshareofprofessionaljobsintheProfesiadataset,whiletheshareofplantandmachineoperatorsandassemblershasgrownwithallothercategoriesremaininglargelystable.
Whilethehistoricalbiasofhigh-skilled,whitecollarpositionsbeingmorelikelytobeadvertisedonlineisinlinewiththeexistingliterature(Carnevaleetal.
,2014),ashifttowardsmorebluecollarvacanciesmightbeinterpretedasashifttowardsarelativelymorecomprehensivecoverageoftheactualpopulationofvacanciesintheeconomy.
AcomparisonofthesharesofindividualsectorsisreportedinTable4.
Asexpected,therelativeshareofinformationandcommunicationstechnologies(ICT)ishigherintheonlinedatabase,whiletherelativeshareoftransportation,andmanufacturingandindustry,ishigherinthedatabasemaintainedbytheemploymentagency.
14Forthepurposesofthecomparison,weexcludearmedforcesoccupations.
ThelattertendtobecompletelyabsentintheProfesiadatasetbutarewellrepresentedintheUPSVARdata.
15Furthermore,sometimesin2015thereisaclearbreakintheseriesinUPSVARdatacausedbytheadoptionofpolicyofactivesolicitationofjobvacanciesfromcompanies.
16Profesiausesitsownoccupationalqualificationwhichis,however,heavilybasedonSlovaknationaloccupationalclassificationKZAMwhich,inturn,istransferabletothe2008updateintheISCOclassification.
0500010000150002000025000300002011201220132014201520162017ProfesiaUPSVARThelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201912Table3ShareofcivilianoccupationalgroupsinvacanciespublishedontheProfesiaportalcomparedwiththePublicEmploymentAgencyregisterISCO20112012201320142015Managers(ISCO1)5%(1%)4%(-1%)4%(-1%)3%(-3%)3%(0%)Professionals(ISCO2)29%(15%)28%(14%)27%(13%)25%(8%)23%(11%)Techniciansandassociates(ISCO3)15%(-5%)15%(-1%)15%(-2%)15%(0%)14%(-1%)Administrative(ISCO4)14%(3%)14%(4%)14%(3%)13%(1%)13%(2%)Serviceandsales(ISCO5)22%(9%)22%(-4%)22%(1%)21%(0%)21%(-1%)Skilledagricultural,forestryandfisheryworkers(ISCO6)0%(-1%)0%(-1%)0%(-1%)0%(-1%)0%(-1%)Craftsandtrades(ISCO7)6%(-8%)7%(-4%)7%(-4%)7%(-5%)7%(-8%)Operatorsinproduction(ISCO8)7%(-10%)7%(-6%)7%(-5%)9%(-3%)12%(-3%)Elementaryoccupations(ISCO9)3%(-4%)4%(0%)4%(-2%)5%(0%)6%(-2%)Note:FiguresgiveninbracketsrepresentthedifferenceintheshareofindividualoccupationalgroupsinthevacanciesregisteredbyUPSVRTable4ShareofsectorsinvacanciespublishedontheProfesiaportalcomparedwiththePublicEmploymentAgencyregisterSector2011201220132014201520162017Manufacturingandindustry21%(-6%)21%(-6%)21%(-7%)21%(-12%)21%(-12%)22%(-6%)24%(-9%)Construction3%(-4%)3%(-5%)3%(-3%)3%(-3%)3%(-1%)4%(-4%)4%(-3%)Wholesaleandretail19%(3%)21%(6%)21%(6%)22%(9%)22%(10%)23%(10%)23%(7%)Transportation3%(-9%)4%(-9%)3%(-9%)4%(-8%)4%(-9%)4%(-10%)5%(-10%)Accommodationandrestaurants5%(1%)6%(4%)6%(0%)6%(1%)7%(2%)7%(2%)7%(1%)ICT21%(17%)19%(15%)19%(15%)17%(14%)18%(15%)16%(13%)15%(12%)Financeandinsurance10%(4%)10%(2%)9%(2%)8%(-1%)8%(-2%)9%(0%)8%(3%)Businessandconsultancy4%(-1%)5%(1%)5%(1%)5%(-4%)5%(-1%)4%(-2%)5%(2%)Administration9%(7%)9%(7%)9%(4%)9%(7%)8%(5%)7%(3%)6%(3%)Education,healthcareculture3%(-14%)4%(-13%)4%(-8%)4%(-4%)3%(-7%)3%(-6%)3%(-6%)Note:FiguresgiveninbracketsrepresentthedifferenceintheshareofindividualoccupationalgroupsinthevacanciesregisteredbyUPSVRThefullProfesiadatasetcontainedapproximately1,250,000uniquevacancies17althoughwehaveremovedalargepartfromtheanalysis.
Ouranalysiswasthenperformedonapproximately640,000privatesector,non-military,non-agriculturaljobvacanciespostedonProfesiabetween2011and2017(Figure2).
Thebiggestlossofobservationswasassociatedwiththeremovalofadvertisementspostedbypersonnelagencies.
Thesevacanciesweresubmittedbycompaniesthat,inthevastmajorityofcases,werenotlookingtohireworkersthemselvesbutwereratherrecruitingcandidatesforothercompanies18.
Thesevacanciescould,therefore,notbelinkedtoaspecificemployerinmostcases.
Thesecondmostfrequent17Publicationofthesevacanciesispaidfor,soemployershaveanincentivenottopostduplicates.
Nonetheless,itisnotpossibletoruleoutthepossibilitythatsomevacancieswerenotfulfilledandthenrepeatedly(re-)posted.
Profesiadoesnotremovesuchduplicatepostingsfromthedatasetbecauseitisnotstraightforwardtoseewhetheravacancyisactuallyapositionthatisrepeatedlyofferedor,rather,anewofferforaverysimilar,standardizedposition.
18Notmuchisknownabouttheroleofagenciesonthelabormarket.
Recentqualitativeresearchshowsitiswidespreadpredominantlyinsmallersupplierfirmsratherthanmajorcompaniesandtendtocovermainlyunskilledwork(Kahancováetal.
,2017).
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201913reasonforremovalwastheexclusionofvacanciesnotpostedbyaprivatesectorcompanybutbypublicinstitutions,collectives,churchesandothernon-privateorganizationsor,inveryrarecases,thosethatcouldnotbematchedwithaspecificemployer.
Additionalreasonsforexclusion,affectingamuchsmallershareofobservations,wereforvacanciesadvertisedabroad19orforskilledagriculturalworkersandmilitaryoccupations.
Finally,asmallnumberofobservationswereremovedwheretherewereunlikelycombinationsofoccupationandeducation,suchasthoseforprofessionaljobsrequiringonlyelementaryeducationormachineoperatorjobsavailableonlytoholdersofadegree.
20Figure2Thefullsamplesizebeforeandafterdataselection(effectivesample)ResultsWefirsttestourexpectationsonthelevelofoccupationalstructure.
Thedifferencesindemandbetweendomestic,mixedandforeigncompaniesforindividualoccupationsarepresentedinFigure3.
Inlinewithourexpectations,internationally-ownedcompaniesdemandmoreskilledoccupationalcategoriesthandomesticfirms.
Firmswithmixedownershiprankinbetween.
19AustrianandCzechcompanies,inparticular,advertiseskilledjobsonProfesiaduetoshortagesofskilledworkersinthesecountries.
20Theseunlikelycombinationsconstitutedcellswithalownumberofobservationsthatwouldintroducebiasesintotheregressionresults.
Weremovedthembycross-tabulatingthecategoriesforoccupationandeducationalrequirementsandkeepingonlythosevacancieslocatedincellswithatleast5,000observations.
Thisapproachremovesallapparentlyunlikelycombinationswithoutinfluencinganyseeminglymeaningfulones.
0500001000001500002000002500003000002011201220132014201520162017FullsampleAnalyzedsubsampleThelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201914Figure3OccupationalstructurepercompanyownershipHowever,thesedifferencesmayberelatedtootherstructuraldifferences,suchascompanysize(foreigncompaniesarepredominantlylarger:seeFigure5).
Wethuscontrolfortheeffectsofcompanycharacteristics,includinglocationinthecapitalcityofBratislava,ownershipstructure,sizeandeconomicsector,throughasimplelogisticregressionthatestimatesdemandforbluecollarpositions(ISCO5,7,8and9).
TheformalspecificationoftheregressionispresentedinEquation1.
(>4)=(,,,,)(1)AsshowninTable5,foreign-ownedcompaniesare,indeed,6.
5percentlesslikelytoofferbluecollarpositions.
Interestingly,companieswithmixedownershipstructuresareslightlymorelikelytoofferbluecollarpositionsthanaredomesticonesonceindividualcharacteristicshavebeenaccountedfor.
ThisstandsincontrastwiththepictureobtainedfromthedescriptivedatainFigure3.
Table5Regressionestimatesforthedeterminantoftheadvertisedpositionbeing'bluecollar'Coef.
Std.
Err.
ZP>z[95%Conf.
Interval]Capital-.
1842046.
0015912-115.
770.
000-.
1873232-.
1810859Mixedownership.
0135704.
00261485.
190.
000.
0084456.
0186953Foreignownership-.
0652037.
0018891-34.
520.
000-.
0689064-.
061501110-99empl.
-.
0348587.
0022663-15.
380.
000-.
0393005-.
0304168100-999empl.
-.
0610685.
0024231-25.
200.
000-.
0658177-.
05631931,000+empl.
-.
0831817.
0028461-29.
230.
000-.
0887599-.
0776035Note:BluecollaroccupationisunderstoodasonebelongingtoISCO5,7,8and9.
Controlledforyearandsectorfixedeffects.
Educationrequirementsexcludedduetoendogeneity.
N=513,046Asfarasthedemandfordigitalskillswithinindividualoccupationsisconcerned,weexpectedforeignfirmstohavehigherdemandfordigitalskillsonlyinISCO6-8occupationalcategories.
Thedescriptiveanalysis,presentedinTable6,confirmsthisexpectation,butonlyasfarastheintermediateskillsareconcerned.
Infact,theintermediatelevelofdigitalskillsseemsmost0%20%40%60%80%100%PrivatedomesticForeignMixedManagersProfessionalsTechniciansandassociatesAdministrativeServiceandsalesCraftsandtradesOperatorsElementaryoccupationThelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201915relevanttomorecomplexactivitiesintheblue-collarprofessions.
Theredoesnotappeartobeanydifferencebetweentherequirementsforbasicandadvanceddigitalskillsfor'bluecollar'workers;andnodifferencebetweenthedemandforanycategoryofdigitalskillsforserviceandsales,andelementaryoccupationworkers.
Thepatternsinwhitecollaroccupationsareconsistentwithourexpectationsinalllevelsofdigitalskills.
Domesticfirmsexhibithigherdemandforbasicandintermediatedigitalskillsinallwhitecollaroccupations(ISCO1-4),notjustinISCO1-3aswehypothesized.
Foradvanceddigitalskills,thisdynamicispreservedonlyforprofessionals,withotherwhite-collaroccupationsbeingequallylikelytodemandadvanceddigitalskillsregardlessofformofcompanyownership.
Suchapatterncorrespondstoexpectationsrelatedtotheeffectofthepresenceofthebusinessservicesoutsourcingsector.
Table6DigitalskillrequirementsbreakdownperISCOoccupationgroupBasicdigitalskillsIntermediatedigitalskillsAdvanceddigitalskillsDomForMixDomForMixDomForMixManagers(ISCO1)24%15%19%77%51%50%4%5%4%Professionals(ISCO2)12%7%5%37%27%28%31%16%15%Techniciansandassociates(ISCO3)21%10%11%66%48%45%9%11%8%Administrative(ISCO4)23%11%13%73%46%60%4%4%6%Serviceandsales(ISCO5)15%15%11%38%36%40%1%1%1%Craftsandtrades(ISCO6)9%10%8%18%34%23%2%3%2%Operatorsinproduction(ISCO7)4%4%5%10%18%11%1%1%1%Elementaryoccupations(ISCO8)4%2%1%9%8%4%0%1%0%Note:Bolddenotesthelargestvaluewithineachparticularoccupationalandskillgroupaslongastheyareatleastfivepercentagepointshigherthanthesecondhighestvalue.
Italicsdenotethesmallestvaluewithineachparticularoccupationalandskillgroupaslongastheyareatleastfivepercentagepointslowerthanthesecondlowestvalue.
Inthefinalstep,wetestifthedifferencesindemandholdwhenwecontrolforotherfactorsthatmightdeterminethelikelihoodofthedemandfordigitalskills.
Theformallogitmodelispresentedinequation2.
Werunthemodelseparatelythreetimeswithbasic,officeandadvancedcomputerskills(s)asthedependantvariable(y).
()=(,,,,,)(2)Themodelsconfirmthepatternsthatwereapparentinthedescriptivestatistics(seeTable7).
Ourtheoreticalexpectationsontheeffectsofforeign-ownedfirmsarethusbroadlyconfirmed.
Theeffectsofmixedownershipgointhesamedirectionasthatforforeign-ownedonesbuttendtobeweaker.
Formanagers(ISCO1),weseethatthedemandfordigitalskillsismuchhigherindomestically-ownedcompanies.
Thedifferenceisparticularlypronouncedforintermediatedigitalskills,whichare17percentlesslikelytobeindemandinthecaseofmixedownershipand20percentlesslikelytobedemandedinforeign-ownedcompanies;whilethedemandforadvancedskillsismarginallyhigherformixedandforeign-ownedcompanies.
Thisrelationshipholdsalsoforthecategoriesoftechnicianandassociateprofessionals(ISCO3)andadministrativeworkers(ISCO4),althoughtheeffectisnotnearlyashighasforintermediateskills.
Professionaljobs(ISCO2)advertisedbydomesticcompanieslikewiseexhibitahigherprobabilityofrequiringintermediatedigitalskills.
Thiseffectisstrongerforadvanceddigitalskills(by3percentinthecaseofmixedownershipand9percentinthecaseofforeignownership).
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201916Asfaras'bluecollar'occupationsareconcerned,domestically-ownedcompaniesarelesslikelytodemanddigitalskillsinISCO5and7occupations,confirmingourexpectationofhigherskilllevelsbeingdemandedbyforeigncompaniesintheseareas.
Thedifferenceisgreatestintheintermediateskillscategory,inwhichspecificallycraftandtradesworkersare11percentmorelikelytoberequiredtohavemasteredintermediatedigitalskillsiftheyseekworkinaforeign-ownedcompany(4percentinthecaseofmixedownership)comparedtotheirpeersapplyingforajobinadomesticcompany.
Theeffectsforserviceandsalesworkers,aswellasholdersofelementaryoccupations,aremixedandinconclusive.
Finally,lookingatthecontrolvariables,wemakeseveralobservations.
ThedigitalskillsrequirementsaregenerallylowerforpositionsadvertisedinthecapitalcityofBratislava,probablyasaresultofthetightlabormarket.
Interestingly,whenacollegedegreeisrequired,thedigitalskillsrequirementstendtobelowerthanforpositionsrequiringonlyhighschooldegreewithfinalexamination.
Apossiblecausebehindthisapparentcontradictionisthatthedegreerequirementtoalargeextentsufficestosignaltheimplicitdemandforskills,whichinturnsmakesitlessimportantforemployerstoexplicitlyspecifyit.
Theevidencefortheeffectofcompanysizeismixed.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201917Table7Regressionresultsfordigitalskills,byoccupation,fullsampleISCO12345789BasicSkillsCapital-0.
0527***-0.
0323***-0.
0434***-0.
0312***-0.
00786***-0.
0009480.
0124***0.
000178Mixedownership-0.
0509***-0.
0216***-0.
0518***-0.
0430***-0.
00807***0.
0128***0.
0107***-0.
00842***Foreignownership-0.
0798***-0.
00834***-0.
0454***-0.
0473***-0.
0107***0.
0132***0.
00389***-0.
00718***10-99empl.
0.
0211***0.
0191***-0.
0420***0.
002390.
00329*0.
0142***0.
00931***-0.
00187***100-999empl.
0.
0455***-0.
0154***-0.
0708***-0.
0189***0.
0175***-0.
0180***-0.
00713***-0.
000119>1,000empl.
-0.
0812***-0.
0595***-0.
128***-0.
124***-0.
118***-0.
0546***-0.
0378***-0.
0183***Primary-0.
0611***-0.
0522***-0.
0112***Withoutexam-0.
0609***-0.
0683***-0.
0459***-0.
0135***-0.
00704***College-0.
0387***-0.
0221***-0.
0466***-0.
0362***-0.
0256***OfficeSkillsCapital-0.
0700***-0.
0706***-0.
0593***-0.
0290***-0.
0192***0.
0325***0.
0344***-0.
00635***Mixedownership-0.
173***-0.
0678***-0.
154***-0.
0678***0.
0554***0.
0371***0.
00142-0.
0131***Foreignownership-0.
200***-0.
0330***-0.
112***-0.
135***-0.
0422***0.
113***0.
0605***0.
0116***10-99empl.
0.
0543***0.
0890***0.
0153***-0.
0338***0.
00738**0.
0230***0.
00633*-0.
00516**100-999empl.
-0.
02070.
0462***-0.
0517***-0.
144***-0.
0381***0.
0172**-0.
00352-0.
0100***>1,000empl.
-0.
223***-0.
0197***-0.
162***-0.
354***-0.
286***-0.
0294***-0.
0138***-0.
0624***Primary-0.
381***-0.
225***-0.
0978***Withoutexam-0.
364***-0.
249***-0.
277***-0.
0943***-0.
0526***College-0.
0306***-0.
0132***-0.
0599***-0.
0418***0.
0282***AdvancedSkillsCapital0.
00223-0.
0259***-0.
00796***-0.
00509***0.
000666***0.
001090.
00267***-0.
000310***Mixedownership0.
00936**-0.
0330***0.
00657**0.
0124***0.
00207***0.
00330**0.
00366***0.
000510**Foreignownership0.
0164***-0.
0909***0.
00968***0.
00929***0.
00230***0.
00271***0.
00228***0.
000730***10-99empl.
-0.
00659*0.
0447***0.
00230-0.
00783***9.
23e-05-0.
00537***-0.
00196***0.
000100100-999empl.
0.
000670-0.
0803***-0.
0233***-0.
0107***-0.
000238-0.
00326***-0.
00162***-0.
000388**>1,000empl.
-0.
0266***-0.
146***-0.
0434***-0.
0130***-0.
00371***-0.
00372**-0.
00467***-0.
000833***Primary-0.
00805***-0.
00814***-0.
00153***Withoutexam-0.
0527***-0.
00436***-0.
0237***-0.
00523***-0.
00150***College0.
00562**-0.
0746***-0.
00857***0.
0117***5.
56e-05Observations21,213151,5197,89793,745170,37941,75752,87935,176Note:AllregressionsarecontrolledforyearandsectorfixedeffectsAsourexpectationsonthedifferencesbetweenwhite-collarandblue-collaroccupationsarebasedontothestylizedfactsfrommanufacturingandservice-sectoroutsourcing,wecomparealsosector-specificmodelsthatincludeonlyadvertisementsbyemployersoperatinginmanufacturingandITrespectively.
Thepatternofoccupationalbreakdownindomesticandforeignownedcompaniesinthesetwosectorsindeedlookscompletelydifferent(seeTable79).
Inmanufacturing,domesticcompaniesemploymorecraftandtradesworkers,whileforeignownedcompaniesoffermoreprofessionalandassociateprofessionalpositions.
ThesituationisreversedintheITsector,wherethedomesticcompaniestendtoadvertiseprofessionalposition,whileforeignownedonesofferrelativelymorepositionsinserviceandsalesdomain.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201918Table8Regressionresultsfordigitalskills,byoccupationISCO12345789ManufacturingDomestic4%15%19%9%12%19%19%3%Foreign3%21%23%11%7%14%19%1%Mixed3%21%21%12%6%15%19%2%ITDomestic1%70%12%8%6%2%0%0%Foreign1%62%12%17%6%0%0%0%Mixed3%67%11%6%13%0%0%0%AsshowninError!
Referencesourcenotfound.
,theresultscorrespondtoourexpectationsandalsototheaggregatemodelspresentedabove.
Morespecifically,foreignmanufacturingcompanieshavehigherrequirementsforintermediatedigitalskillsinbluecollarpositions(ISCO6-8)andlowerrequirementsforthewhitecollaroccupations.
Incontrast,foreigncompaniesinIThavelowerrequirementsacrosstheoccupationalcategories,withsignificantdifferencesinwhitecollaroccupations.
DomesticITcompanieshaveparticularlyhigherskillrequirementsinISCO2andISCO3,indicatingamuchhigherlevelofsophisticationindomesticITcompanies.
Theeffectismostpronouncedintheadvancedskillcategory.
Table9Regression-basedaveragemarginalcoefficientsfordemandfordigitalskillsinforeignownedcompaniescomparedtodomesticownedones,byoccupationforITandmanufacturingsectorsISCO12345789ManufacturingBasc-8.
41%-5.
05%-4.
62%-2.
62%-9.
75%3.
16%0.
67%1.
21%Office-7.
49%-4.
46%-6.
16%-1.
14%-2.
68%11.
30%3.
92%3.
05%Advanced4.
81%2.
18%3.
17%6.
18%0.
77%-0.
30%0.
63%0.
00%ITBasic-6.
48%-0.
39%-2.
32%-1.
81%-3.
58%Office-3.
15%-3.
51%-4.
28%-14.
30%-11.
70%Advanced0.
28%-24.
00%-13.
40%-0.
53%-0.
31%Note:controlledforbeinglocatedinBratislava,sizeofthecompany,educationrequirementsandyearfixedeffect.
Coefficientsinboldsignificantat5%threshold.
Finally,totestthestabilityoftheestimates,weranthemodelseparatelyforeachanalyzedyear,controllingforoccupationalcategories(seeTable10).
Wefindtheeffectsofacompanybeingforeign-ownedretainthesamedirectionaswellasapproximatelythesamemagnitudeovertime,althoughthereappearstohavebeenanarrowingofthedifferenceinthelasttwoyears,inparticularforintermediatedigitalskills.
Thismightbeasignofanarrowingofthedifferencesbetweenforeignanddomesticsectors,butchangeoveralongerperiodwouldneedtobeobservedtosupportsuchaconclusion.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201919Table10Developmentofthedifferenceindemandfordigitalskillsbetweenforeignanddomesticfirmsovertime2011201220132014201520162017Basicskills-0.
0635***-0.
0357***-0.
00829***-0.
0109***-0.
0206***-0.
0125***-0.
0167***Officeskills-0.
0580***-0.
0747***-0.
0572***-0.
0570***-0.
0443***-0.
0294***-0.
0134***Advancedskills-0.
0142***0.
00218*-0.
00959***-0.
00829***-0.
0102***-0.
00375***-0.
00136**EffectsofforeignownershiprelativetodomesticConclusionForeign-ownedfirmsgeneratemorejobsinhigherskilledoccupations,confirmingthattheyaddhighervalueaddedthanthedomesticfirms.
Atthesametime,however,wefoundthat,forcomparablejobs,domesticfirmsaretheonesmorelikelytoseekdigitally-skilledworkers.
Thesituationisdifferentinblue-collarjobswhereforeign-ownedcompaniesaretheonesthataremorepronetorequiredigitalskills.
Foreigncompaniesarethustechnologyleadersinanarrowsegmentofmanufacturingactivitiesrelatedtoassemblyandcomponentproduction,withapotentialfortechnologytransferfromforeigntolocalfirms.
Overall,therangeofactivitiesperformedinlocalbranchesofMNCsthusappearslimited,withrelativelylowskillcontent.
Asourrobustnesscheckshows,workersareratherrequiredtobeabletocommunicateinforeignlanguagesandthusbesmoothlyintegratedintheinternationaldivisionoflabour.
Thepatternremainslargelystableovertime,confirmingthestabilityofSlovakia'spositionintheintegratedperiphery.
Thepolicyimplicationsofourfindingsarethreefold.
Firstly,theconsiderablepublicresourcesinvestedinattractingFDItoSlovakiamightleadtosomemodernmanufacturingjobs,buttheyareunlikelytocontributetoabroadermodernizationanddigitalizationoftheSlovakeconomy.
ItmightbeprudenttoconsidershiftingsomeresourcestosupportingthedomesticR&Dsector.
Secondly,theco-existenceofhigherwages(normallyassociatedwithhigherhumancapital)paidbyforeign-ownedcompanies,withthedigitalskillacquisitionbeingconcentratedindomesticfirms,mightresultinamisallocationofhumancapital.
Thirdly,theconcentrationofthemostgiftedwhite-collarworkersinMNCs,wheretheirengagementwithdigitaltechnologiesmightbelimited,meansthattheseworkersmightnotbeaccessiblefordigitallymoreadvanceddomesticcompanies.
Moreover,thosetalentedworkersmightencountera"glassceiling"withintheSlovakbranchesofMNCs,underwhichtheyareunabletoaccessthehighvalueopportunitiesassociatedwiththeintensiveuseofdigitalskills,andthismightdrivethemtoseekthoseopportunitiesabroadintheheadquartersoftheircompany.
ThismightnegativelyaffecttheconvergenceoftheSlovakeconomyandlivingstandardsinSlovakia.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201920AppendixTable11DependentvariabledefinitionsBasicskillsIntermediateskillsAdvancedskillsMacOSInternetMicrosoftWindowsMS-DOSMicrosoftAccessMicrosoftExcelMicrosoftWordMicrosoftPowerPointSAPAutoCADCorelDRAWCorelPHOTO-PAINTAdobePhotoshopMicroStationIBMLotusNotesT602MicrosoftFrontPageMicrosoftOutlookAdobeIllustratorAdobePageMakerQuarkXPressMagicAutodesk3dsMaxPro/ENGINEERSAPMicrosoftProjectOpenOfficeOracleFormsCATIASolidWorksCADAdobeInDesign602OfficeNXI-DEASACADSolidEdgeAutodeskInventorAllplan-NemetschekCENKROSJeevesSIMATICSTEP7OmronMovexMicrosoftDynamicsNAVArchiCADAutodeskRevitArchitectureAllplanArConGalileoAmadeusARISHumanPOHODAASPC++PerlServeradministrationHTMLIBMInformixJavaJavaScriptOracleDatabaseSybaseAdministrationUNIX/LinuxUNIX/LinuxVisualBasicAdministrationWindowsDelphiMicrosoftSQLServerMicrosoftVisualFoxProAdministrationNovellNetWarePascalInterBasePHPPervasiveSQLTango2000XMLIBMiAdministrationOpenVMSCOBOLFortranAdaPL/1PowerBuilderSmalltalkAdobeDreamweaverSASAdobeFlashMySQLWAPOpenGLDirectXSASASP.
NETC#AdministrationLotusNotesSQLLotusScriptJavaEERPGPL/SQLProgress4GLThelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201921AutodeskMayaSketchUpMicrosoftVisioAdobePremiereProAdobeAfterEffectsOMEGAALFAOpenOffice.
orgWriterOpenOffice.
orgCalcOpenOffice.
orgImpressOpenOffice.
orgBaseOpenOffice.
orgDrawSDLTradosWordfastBloombergMicrosoftSharePointABRATIAPortalWinCCOAEPLANMoneyS3MoneyS4GISAdministrationLAN/WAN.
NETVisualBasic.
NETUMLPythonErwinVisioStudioMicrosoftVisualStudioABAPMFCAssemblerwin32APICShellActionScriptAJAXIBMDB2XHTMLAdministrationMySQLAdministrationOracleDatabaseAdministrationMicrosoftSQLServerVMwareESXMicrosoftHyper-VRubyCCSSObjective-CPostgreSQLGoogleAnalyticsGoogleAdWordsGoogleAdSenseAdministrationIBMDB2AdministrationSAPObjectPascalWordPressSwiftRNODE.
JSAngularReactJSBashHaskellVHDLMatlabDrupalOpenCartPrestaShopMagentoThelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201922Table12IndependentvariabledefinitionsCompanylocationCapitalIsthejoblocatedintheBratislavaregion(dummy)CompanyownershipDomestic(referencecategory)Mixed(dummy)Foreign(dummy)CompanysizeSmall1,000empl.
-0.
225***-0.
0229***-0.
0729***-0.
110***-0.
284***-0.
231***-0.
203***-0.
196***Priamary-0.
230***-0.
186***-0.
209***Withoutexam-0.
358***-0.
225***-0.
362***-0.
143***-0.
163***Collage0.
334***-0.
001460.
231***0.
113***0.
377***SectordummiesYESYESYESYESYESYESYESYESYeardummiesYESYESYESYESYESYESYESYESObservations21,213151,5197,89793,745170,37941,75752,87935,176Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201925ReferencesAcemoglu,D.
,Autor,D.
,2011.
Skills,TasksandTechnologies:ImplicationsforEmploymentandEarnings,in:Ashenfelter,D.
C.
andO.
(Ed.
),HandbookofLaborEconomics.
Elsevier,pp.
1043–1171.
https://doi.
org/10.
1016/S0169-7218(11)02410-5Antalóczy,K.
,Sass,M.
,2015.
Throughaglassdarkly:Thecontentofstatisticaldataonforeigndirectinvestment.
Külgazdaság59,34–61.
Arntz,M.
,Gregory,T.
,Zierahn,U.
,2016.
TheRiskofAutomationforJobsinOECDCountries(OECDSocial,EmploymentandMigrationWorkingPapersNo.
189).
https://doi.
org/10.
1787/5jlz9h56dvq7-enAskitas,N.
,Zimmermann,K.
F.
,2015.
Theinternetasadatasourceforadvancementinsocialsciences.
InternationalJournalofManpower36,2–12.
https://doi.
org/10.
1108/IJM-02-2015-0029Bair,J.
,2005.
GlobalCapitalismandCommodityChains:LookingBack,GoingForward.
Competition&Change9,153–180.
https://doi.
org/10.
1179/102452905X45382Beblav,M.
,Fabo,B.
,Lenaerts,K.
,2016a.
DemandforDigitalSkillsintheUSLabourMarket:TheITSkillsPyramid(SSRNScholarlyPaperNo.
ID3047102).
SocialScienceResearchNetwork,Rochester,NY.
Beblav,M.
,Kureková,L.
M.
,Haita,C.
,2016b.
Thesurprisinglyexclusivenatureofmedium-andlow-skilledjobs:EvidencefromaSlovakjobportal.
PersonnelReview45,255–273.
https://doi.
org/10.
1108/PR-12-2014-0276Blomstrom,M.
,Kokko,A.
,2001.
Foreigndirectinvestmentandspilloversoftechnology.
InternationalJournalofTechnologyManagement22,435–454.
https://doi.
org/10.
1504/IJTM.
2001.
002972Boselli,R.
,Cesarini,M.
,Marrara,S.
,Mercorio,F.
,Mezzanzanica,M.
,Pasi,G.
,Viviani,M.
,2018.
WoLMIS:alabormarketintelligencesystemforclassifyingwebjobvacancies.
JIntellInfSyst51,477–502.
https://doi.
org/10.
1007/s10844-017-0488-xCapik,P.
,Drahokoupil,J.
,2011.
ForeignDirectInvestmentsinBusinessServices:TransformingtheVisegrádFourRegionintoaKnowledge-basedEconomyEuropeanPlanningStudies19,1611–1631.
Carnevale,A.
,Jayasundera,T.
,Repnikov,D.
,2014.
Understandingonlinejobadsdata:atechnicalreport.
GeorgetownUniversity.
Coe,N.
M.
,Hess,M.
,Yeung,H.
W.
,Dicken,P.
,Henderson,J.
,2004.
"Globalizing"regionaldevelopment:aglobalproductionnetworksperspective.
TransactionsoftheInstituteofBritishGeographers29,468–484.
https://doi.
org/10.
1111/j.
0020-2754.
2004.
00142.
xDicken,P.
,Kelly,P.
F.
,Olds,K.
,Yeung,H.
W.
-C.
,2001.
Chainsandnetworks,territoriesandscales:towardsarelationalframeworkforanalysingtheglobaleconomy.
GlobalNetworks1,89–112.
https://doi.
org/10.
1111/1471-0374.
00007Drahokoupil,J.
,2008.
GlobalizationandthestateinCentralandEasternEurope:Thepoliticsofforeigndirectinvestment.
Routledge,London.
Dunning,J.
H.
,Lundan,S.
M.
,2008.
MultinationalEnterprisesandtheGlobalEconomy.
EdwardElgarPublishing.
Engel,J.
,Taglioni,D.
,2017.
Themiddle-incometrapandupgradingalongglobalvaluechains,in:WorldBank(Ed.
),GlobalValueChainDevelopmentReport2017:MeasuringandAnalyzingtheImpactofGVCsonEconomicDevelopment.
WorldBankGroup,Washington,DC,pp.
119–139.
Fabo,B.
,Beblav,M.
,Lenaerts,K.
,2017.
TheimportanceofforeignlanguageskillsinthelabourmarketsofCentralandEasternEurope:assessmentbasedondatafromonlinejobportals.
Empirica44,487–508.
https://doi.
org/10.
1007/s10663-017-9374-6Fereníková,S.
,Fifeková,M.
,2006.
FDIspilloversinSlovakia:Trendsfromthelastdecadeandrecentevidencefromautomotiveindustry,Ekonomickasopis.
JournalofEconomics54,633–651.
Frey,C.
B.
,Osborne,M.
A.
,2017.
Thefutureofemployment:HowsusceptiblearejobstocomputerisationTechnologicalForecastingandSocialChange114,254–280.
https://doi.
org/10.
1016/j.
techfore.
2016.
08.
019Gottvald,J.
,Rievajová,E.
,ipikalová,S.
,2013.
DeterminantsofIndividualWagesintheSlovakRepublic.
Ekonomickasopis(JournalofEconomics)61,672–689.
Havlat,M.
,Havrlant,D.
,Kuenzel,R.
,Monks,A.
,2018.
EconomicconvergenceintheCzechRepublicandSlovakia(No.
34),EconomicBrief.
EuropeanCommission,Brussels.
Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201926Henderson,J.
,Dicken,P.
,Hess,M.
,Coe,N.
,Yeung,H.
W.
-C.
,2002.
GlobalProductionNetworksandtheAnalysisofEconomicDevelopment.
ReviewofInternationalPoliticalEconomy9,436–464.
Holt,C.
,David,M.
,1966.
Theconceptofjobvacanciesinadynamictheoryofthelabormarket,in:TheMeasurementandInterpretationofJobVacancies.
NBER,pp.
73–110.
Humphrey,J.
,Memedovic,O.
,2003.
TheGlobalAutomotiveIndustryValueChain:WhatProspectsforUpgradingbyDevelopingCountries,UNIDOSectorialStudiesSeriesWorkingPaper.
Jacobs,E.
J.
,Zámborsk,P.
,Sbai,E.
,2017.
MutualProductivitySpilloversinSlovakia:AbsorptiveCapacity,theTechnologyGap,andNonlinearEffects.
EasternEuropeanEconomics.
Kahancová,M.
,Martiková,M.
,Sedláková,M.
,2017.
NegotiatingWageInequalityInSlovakia(No.
21),CelsiResearchReport.
CELSI,Bratislava.
Keister,R.
,Lewandowski,P.
,2017.
AroutinetransitioninthedigitaleraTheriseofroutineworkinCentralandEasternEurope.
Transfer:EuropeanReviewofLabourandResearch23,263–279.
https://doi.
org/10.
1177/1024258917703557Knell,M.
,2017.
R&DinternationalisationandlocalinnovationintheVisegradGroupaftertheFDIpeak,in:Galgóczi,B.
,Drahokoupil,J.
(Eds.
),CondemnedtoBeLeftbehindCanCentralandEasternEuropeEmergefromItsLow-WageModelEuropeanTradeUnionInstitute,Brussels,pp.
111–131.
Kravtsova,V.
,Radosevic,S.
,2012.
AresystemsofinnovationinEasternEuropeefficientEconomicSystems36,109–126.
https://doi.
org/10.
1016/j.
ecosys.
2011.
04.
005Kureková-Mtna,L.
,Beblav,M.
,Haita,C.
,Thum,A.
-E.
,2016.
Employers'skillpreferencesacrossEurope:betweencognitiveandnon-cognitiveskills.
JournalofEducationandWork29,662–687.
https://doi.
org/10.
1080/13639080.
2015.
1024641Leitner,S.
M.
,Stehrer,R.
,2014.
TradeIntegration,ProductionFragmentationandPerformanceinEurope-BlessingorCurseAComparativeAnalysisoftheNewMemberStatesandtheEU-15(No.
397),wiiwResearchReports.
TheViennaInstituteforInternationalEconomicStudies,wiiw.
Mahutga,M.
C.
,2014.
Globalmodelsofnetworkedorganization,thepositionalpowerofnationsandeconomicdevelopment.
ReviewofInternationalPoliticalEconomy21,157–194.
https://doi.
org/10.
1080/09692290.
2013.
779932Maurer-Fazio,M.
,2012.
EthnicdiscriminationinChina'sinternetjobboardlabormarket.
IZAJMigration1,12.
https://doi.
org/10.
1186/2193-9039-1-12Mezihorak,P.
,2018.
Competitionforcontroloverthelabourprocessasadriverofrelocationofactivitiestoasharedservicescentre.
HumanRelations71,822–844.
https://doi.
org/10.
1177/0018726717727047Modestino,A.
S.
,Shoag,D.
,Ballance,J.
,2016.
Downskilling:changesinemployerskillrequirementsoverthebusinesscycle.
LabourEconomics41,333–347.
https://doi.
org/10.
1016/j.
labeco.
2016.
05.
010Myant,M.
,Drahokoupil,J.
,2011.
Transitioneconomies:PoliticaleconomyinRussia,EasternEurope,andCentralAsia.
Wiley-Blackwell,Hoboken,NJ.
MytnaKurekova,L.
,iliníková,Z.
,2015.
Low-SkilledJobsandStudentJobs:Employers'PreferencesinSlovakiaandtheCzechRepublic(SSRNScholarlyPaperNo.
ID2655094).
SocialScienceResearchNetwork,Rochester,NY.
Nlke,A.
,Vliegenthart,A.
,2009.
Enlargingthevarietiesofcapitalism:TheemergenceofdependentmarketeconomiesinEastCentralEurope.
WorldPolitics61,670–702.
Pavlínek,P.
,2018.
GlobalProductionNetworks,ForeignDirectInvestment,andSupplierLinkagesintheIntegratedPeripheriesoftheAutomotiveIndustry.
EconomicGeography94,141–165.
https://doi.
org/10.
1080/00130095.
2017.
1393313Pavlínek,P.
,2016.
WhosesuccessThestate–foreigncapitalnexusandthedevelopmentoftheautomotiveindustryinSlovakia.
EuropeanUrbanandRegionalStudies23,571–593.
https://doi.
org/10.
1177/0969776414557965Pavlínek,P.
,íalová,P.
,2016.
Linkagesandspilloversinglobalproductionnetworks:firm-levelanalysisoftheCzechautomotiveindustry.
JEconGeogr16,331–363.
https://doi.
org/10.
1093/jeg/lbu041Thelimitsofforeign-ledgrowth|NBSWorkingpaper|7/201927Ponte,S.
,Sturgeon,T.
,2014.
Explaininggovernanceinglobalvaluechains:Amodulartheory-buildingeffort.
ReviewofInternationalPoliticalEconomy21,195–223.
https://doi.
org/10.
1080/09692290.
2013.
809596Radosevic,S.
,CiampiStancova,K.
,2018.
InternationalisingSmartSpecialisation:AssessmentandIssuesintheCaseofEUNewMemberStates.
JournaloftheKnowledgeEconomy9,263–293.
https://doi.
org/10.
1007/s13132-015-0339-3Radosevic,S.
,Savic,M.
,Woodward,R.
,2010.
KnowledgebasedEntrepreneurshipinCEE:ResultsofaFirmLevelSurvey.
KnowledgeIntensiveEntrepreneurshipandInnovationSystemsEvidencefromEurope.
London,UK:Routledge.
Ramirez,P.
,Rainbird,H.
,2010.
Makingtheconnections:bringingskillformationintoglobalvaluechainanalysis.
Work,EmploymentandSociety24,699–710.
https://doi.
org/10.
1177/0950017010380641tefánik,M.
,2012.
Focusedinformationonskillsdemandusinginternetjobsearchdata(withresultsforSlovakuniversitygraduates).
Wade,M.
R.
,Parent,M.
,2002.
RelationshipsBetweenJobSkillsandPerformance:AStudyofWebmasters.
JournalofManagementInformationSystems18,71–96.
https://doi.
org/10.
1080/07421222.
2002.
11045694WorldBank,2017.
Globalvaluechaindevelopmentreport2017:MeasuringandanalyzingtheimpactofGVCsoneconomicdevelopment.
WorldBankGroup,Washington,DC.