




版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
NBERWORKINGPAPERSERIES
ARESOFTWAREAUTOMATIONANDTELEWORKERSSUBSTITUTES?PRELIMINARYEVIDENCEFROMJAPAN
RichardBaldwin
ToshihiroOkubo
WorkingPaper31627
/papers/w31627
NATIONALBUREAUOFECONOMICRESEARCH
1050MassachusettsAvenue
Cambridge,MA02138
August2023
WethankDavidGreenaway,RichardKneller,andJonathanDingelforconstructivecommentsandsuggestions.WealsothankNIRA(ReikoKanda,AtsushiInoue,andKozueSekijima),andKoheiAndoforsurveypreparation,datacleaning,datapreparationandexcellentassistance.FinancialsupportfromJSPS(23H00821,19H01487),KeioUniversity(ResearchProjectforNextGeneration),andNIRAisgratefullyacknowledgedbyToshihiroOkubo.TheviewsexpressedhereinarethoseoftheauthorsanddonotnecessarilyreflecttheviewsoftheNationalBureauof
EconomicResearch.
NBERworkingpapersarecirculatedfordiscussionandcommentpurposes.Theyhavenotbeenpeer-reviewedorbeensubjecttothereviewbytheNBERBoardofDirectorsthataccompaniesofficialNBERpublications.
©2023byRichardBaldwinandToshihiroOkubo.Allrightsreserved.Shortsectionsoftext,nottoexceedtwoparagraphs,maybequotedwithoutexplicitpermissionprovidedthatfullcredit,including©notice,isgiventothesource.
AreSoftwareAutomationandTeleworkersSubstitutes?PreliminaryEvidencefromJapanRichardBaldwinandToshihiroOkubo
NBERWorkingPaperNo.31627
August2023
JELNo.F6
ABSTRACT
Digitaltechnologyisreshapingworkplacesbyenablingspatialseparationofoffices,knownastelework,orremoteintelligence(RI),andbyfacilitatingautomationofservicesectortasksviaartificialintelligence(AI).ThispaperisafirstattempttoempiricallyinvestigatewhetherAIandRIarecomplementsorsubstitutesintheservicesector.Itusesaworker-levelpanelofsurveyscollectedfromaround10,000workersfrompre-COVID-19pandemictolate2022,wefindpreliminaryevidencethatsuggeststhatAIandRIarecomplementsratherthansubstitutes.TheevidencecomesfirstfromthepositivecorrelationofinvestmentsinAI-promotingandRI-promotingsoftwareatthefirmandworkerlevel,andsecondfromthepositivecorrelationofworkers'expectationsregardingteleworkandsoftwareautomation.Theevidenceisfarfromdefinitivebutsuggeststhatthecomplement-substitutionquestionisafruitfullineforfutureresearch.
RichardBaldwin
GraduateInstitute,Geneva
GenevaChem.Eugène-Rigot2
1202Genève
Switzerland
andCEPR
andalsoNBER
rbaldwin@
ToshihiroOkubo
KeioUniversity
2-15-45MitaMinato-ku
TokyoJapan
okubo@econ.keio.ac.jp
2
1Introduction
Digitaltechnologyisreshapingworkplacesinwaysthatwereunimaginablejustafewyearsago.Thisishappeningintwodistinctways.First,digitechisunbundlingofficesinthesensethatsomeservicetasksthatweretraditionallyperformedbyworkerssittingphysicallyclosetoeachothercannowbeundertakenatadistance.Second,digitechisallowingmanymoreservicesectortaskstobeautomatedbysoftware.
Teleworkisonenameforthefirstimpactofdigitaltechnologyandthekeypointisthatdigitaltechnologyismakingremoteworkincreasinglyeasy.Ofcourse,teleworkingisnotanewphenomenonandfirstappearedinthe1970s(Nilles,1975;Aguileraetal.2016).Therewerealwayssomeworkerswhoworkedremotely–‘roadwarriors’wasonenameforserviceworkerswhoroutinelyparticipatedinofficeworkwhiletravellingvialaptops,emails,andmobilephones(Bloometal.2018).SincetheCovid-19restrictionsrequiredorencouragedworkerstoworkfromhome,thissortofteleworkhasbecomefarmorecommon(Kahn,2020).Thekeytechnicalinnovationsincludeimprovedcollaborativesoftwaresuites,videoconferencingapps,andsecure,cloud-baseddocumentsharingandediting.Todatemuchoftheteleworkisdonebydomesticworkers,butthetrendtowardsusingforeignremoteworkersiswellunderway–aphenomenonthatcanbecalledtelemigration.
Thesecondmajorimpactofdigitaltechnologyonworkpatternsconcernsautomationofmanytasksrelatedtoservicejobs.TherecentbreakthroughsingenerativeAI,suchasChatGPT,aretransformingmanyofficeandprofessionaljobs.ButevenbeforeChatGPT,softwaresuchasRoboticProcessAutomation(RPA),chatbots,andautomatictranslationappswereautomatingmanytaskstraditionallyperformedbyofficeworkersandprofessionals.Collectively,suchsoftwaremightbecalledsoftwarerobots,orwhite-collarrobots(todistinguishthesefromthemorefamiliarindustrialrobots).
Thekeyinnovationherewastherapidadvanceofmachinelearning.Before2017(theyearFortuneMagazinecalledtheyearofAI),computerscouldonlyperformtaskstheautomatedhumanthinkingoftheanalytic,conscioustype–whatDanielKahnemancalls‘thinkingslow’(Kahneman,2011).Thereasonwasduetothenatureofcomputerprogramming.Writingcoderequiresthecodertounderstandthementalprocessthatisbeingtaughttothecomputer.Computerscouldnotdointuitive,unconsciousthinking–likerecognizingfacesinphotos,ortranslatinglanguages–sincehumansdidnotunderstandhowhumansthinkintuitively(westilldonot).Anewwayofwritingcomputerprogramsrelaxedthisconstraint.AformofAI,calledmachinelearning,alloweddatascientiststoestimateverylargeandverycomplexstatisticalmodelsthatmimichumanunconsciousthinking–orwhatKahneman(2011)calls‘thinkingfast’.Since2017,computersareasgoodorbetterthanhumansinsomeinstinctual,unconsciousmentaltasks—thingslikerecognizingspeech,readinghandwriting,andidentifyingobjectsinphotos.Therateofadvanceinthisfieldisastounding.
Themergerofthesetwoemergenttrends–globalisationandrobotisationofservicesectortasks–hasbeencollectivelyreferredtoasthe"GloboticsUpheaval"(Baldwin,2019).In2022,thepossibilityofdigitaltechnologydisruptingworkplacesanddisplacingjobsisnolongeraconjecture.Moreover,ineconomieswherejobsarethefoundationofeconomicprosperityandsocialstanding,thedisruptivenatureofcompetitionfromdistantteleworkersandnearbyservicetaskautomationsoftwareisareal
concern.
Thedualnatureofthedigitechimpactonthefutureofworkpromptsanimportantquestionabouttherelationshipbetweenservicesectorautomationandoffshoring:aretheycomplementsorsubstitutes?IsAIappliedinofficesasubstituteforremoteintelligence(RI)employedinoffices?Ortoputitmoredirectly,willteleworkingmakeitmorelikelythatyouremployerwillautomatepartorallofyourjob?Oneexamplewheretheyareclearlysubstitutableisinthedomainofcallcenterswhereautomationisreplacingremoteworkers.Callcenters,anintegralpartofmanyservice-sectorbusinesses,oftendeliveressentialsupportservicesusingworkerslocatedinlowercostplaceswithinthesamecountryoreveninothercountries.Inrecentyears,however,manyofthejobsinthesecallcentersarereplacingworkerswithchatbots,whicharejustAI-driveautomationsoftware.Anexample
3
presentedintheMITTechnologyReview(Hao,2020)relateshowthelocalgovernmentinOtsegoCountyNewYorkusedIBM’s‘WatsonAssistantforCitizens’torespondtothesurgeinroutinequestionsaboutCovid-19whilecuttingstaffbyhalf.Duringthepandemic,IBMsawatwo-fifthsboostintheuseofitsWatsonAssistant.TheemergenceofverycapablechatbotshasledtoareductionofhumansemployedincallcentersinthePhilippinesandIndia.Asusualitisnotafullsubstitution,sincehumansstillliketotalktohumans,andchatbotsdonothavetheanswertoeveryquestionbutaschatbotsadvance,thenumberofhumansneededwillfall.
Anothertypeofworker-substitutingautomationiscalled‘RoboticProcessAutomation’(RPA).AsJasonKingdon,whochairsanRPAcompany,BluePrism,describesthem:“Theymimicahuman.Theydoexactlywhatahumandoes.Ifyouwatchoneofthesethingsworking.…youseeittyping.Screenspop-up,youseeitcuttingandpasting.”RPAsaredesignedtoworkasanautomatedofficeworkerwhichiswhyBluePrismcallsthesoftwarerobots.ThecommercialinterestinRPAliesinthewayitcanreducethenumberofworkersinback-officeprocesses.RPAsarealsomoreconsistentthanworkersandtheycreateadigitalrecordoftheiractivity.RPAsneedhumanbackupfordifficultcases,butthenumberofworkersisreduced.Asmanyofthese‘assemblyline’informationprocessingjobswereshiftedtoteleworkduringthepandemic,RPAcanbethoughtofasbeingasubstituteofRI.
Onthecomplementside,oneobviousautomationsoftwarethatbooststheuseofremoteworkersisautomatictranslation.Withthewidespreadadoptionofmachinetranslationtechnology,remoteworkhasbecomeincreasinglyaccessibletonon-Englishspeakingworkers.Formanyservicetasks,allthatisneededisaccesstotheinternetconnection,alaptop,andtherequisiteskills.WithsoftwarelikeGoogleTranslate,DeepLTranslator,orBingMicrosoftTranslator,workerslivingincountrieswhere$5anhouraffordsamiddle-classlivingcanpotentiallybeworkmateswithofficeworkersinwealthiernations.HereAIisacomplementtoRIsincetheAIismakingtheRIeasiertodoforworkerswithoutlanguageskills.Inessence,theAImakesiteasierforforeignbasedfreelancersandremoteworkerstodelivercontentinEnglishevenwhentheyarenotnativespeakers.
LiteratureReview
TheliteratureontheimpactofComputers/Robots/ICT/AIonlaborhasbeengrowinginthelastdecades(Acemoglu,2007;Autoretal.2013;AcemogluandRestrepo,2018).ThetheoreticalworkhasfocusedontheimpactoftechnologicaladvancessuchasICTandrobotsonemploymentandwages.Autoretal.(2003),forinstance,modelledunskilledworkersasperformingmostlyroutinetasks,andskilledworkersasperformingmostlynon-routinetasks.Theimpactofautomationonthetwotypesoflabordependsuponwhetherthetechnologyissubstitutablefororcomplementarytothetwotypesoftasks.Assumingautomationisasubstituteforroutinetasksbutacomplementtonon-routinetasks,automationraisesthegapbetweenjobopportunitiesandwagesforunskilledversusskilledlabor.ThisaccountsforpolarizationintheUSlabormarket.Goosetal.(2007,2009,2014)foundthatunskilledlaborshiftedfromreplaceablejobsbyautomationtonon-replaceablebutlow-incomejobsintechnologicalchangeinEurope.Beyondthis,asshowninBrynjolfssonandMcAfee(2014)andFord
(2016),automationisdeeplyspreadoverthelabormarketinmanydimensions.FreyandOsborne
(2013)predictedaroundhalfofoccupationswillbedisruptedbyautomationinthefuture.
Turningtoremoteintelligence(RI),therearealreadysomestudiesonremoteworkbeforethepandemicinworker-levelanalysisoflaboreconomics.Remoteworkhasseveraladvantagessuchasimprovingwork-lifebalance(Dutcher,2012;CoenenandKok,2014;Kazekami,2020)andproductivities(Bartiketal.2020).WiththespreadofCovid-19,studiesontheimpactofteleworkingoneconomiesincreased(Adams-Prassl,etal.2022;2023;Alipouretal.2021;2023;Bonacinietal.,2021;KawaguchiandMotegi,2021;Morikawa,2022;Mongeyetal.2021;Okubo,2022;Okuboetal.2021).DingelandNeiman(2021)suggestedthattherearesomespecificoccupationssuitableforremotework.TheirderivedremoteworkableindexbyoccupationbasedonO*NETinformationfoundthat37%ofjobsintheUnitedStatescanbeperformedentirelyathome.
4
Asfarasweknow,nostudieshaveinvestigatedthenexusofautomationbyAIandremoteworkbyRIintheirlong-runaspects.OurpaperisaimedatinvestigatingthesubstitutabilityofAIandRItolaboratoccupationallevelandprovidingevidencefromJapaneseworker-levelsurveys.Furthermore,thesurveyweuseisuniqueinthatitdirectlyaskedworkersabouttheiruseandtheircompany’suseofAIandRIaswellastheirexpectationsconcerningtheimpactofAIandRIontheirownjobsinthefuture,andwhethertheyseeAIandRIassubstitutesorcomplementstotheirjobs.Inaddition,thesurveygathersextensiveinformationontheworkers’basiccharacteristics,thetaskstheirjobinvolves,andinformationontheirfirm.
Planofpaper
ThispaperinvestigatesthesubstitutabilityofAIandRIattheoccupationallevelinthecontextofJapanusingauniquepanelofsurveyscollectedfromabout10,000workerswithastartdatethatwasjustbeforetheCovid-19shockandcontinueduntillate2022.
Theinvestigationstartsbyexaminingthetechnicalteleworkabilityandautomatabilityofoccupationsasafirststeptolookingattheactualco-movementofteleworkandsoftwareautomation.Thisisbasedonexanteevaluationoftheoccupationdescriptions.ThemostfamousoftheseisDingelandNeiman(2020)whichexaminesthejobdescriptionsontheUS’sO*NET.AsimilarexercisehasbeendoneusingJapan’sO*NETbyKotera(2021).Whenitcomestoautomatability,welookatthefamousFrey-OsbourneindexofautomatabilitytogaugehowautomatabilityofvariousoccupationsinUS.Equivalently,FreyandOsbourne(2015)calculatedtheindexofautomatabilityforcaseofJapanincollaborationwithNRI(NomuraResearchInstitute).
Tosummarizethetechnicalfeasibilityforvariousoccupations,weintroducethe"Globoticsquadrantdiagram,"whichmapsoccupationsintoatwo-dimensionalchartbasedontheirautomatabilityandteleworkability.ThescatterchartisusedtoillustratedataforboththeUnitedStatesandJapan.Thecentralargumentisthatwhilesomeoccupationsarebothautomatableandteleworkable,manyothersaresuitedtoonlyoneorneither,asdeterminedbystandardmeasuressuchastheDingel-NeimanindexforteleworkandtheFreyandOsborneindexforautomatability.
ThesubsequentsectionofthepaperexaminestheimpactoftheCOVID-19pandemicinJapanontherelationshipbetweenAIandRIusingdatafromthesurvey,whichwasconductedbyToshihiroOkubo(KeioUniversity)andNIRA(NipponInstituteforResearchAdvancement).Whilethesurveyprovidesdetailedinformationonthenumberofworkerswhoswitchedtoteleworking,itlacksdataonthenumberofworkerswhowerereplacedbywhite-collarrobots.Thus,adirecttestofthesubstitution-versus-complementarityquestioncannotbeassessed.However,thesurveydoesprovideinformationonfirms'useofsoftwarethatfacilitatesteleworkandautomationofofficetasks,respectively.
Usinginformationonfirms’usageofpro-RIandpro-AIsoftware(asreportedbytheworkers),andtheexogeneityoftheCOVID-19shock,thepaperexamineswhetherfirms'investmentsinAI-promotingsoftwareandRI-promotingsoftwarewerepositivelyornegativelycorrelated.Giventhattheriseinteleworkwasdirectlyinducedbythepandemic,whiletheadoptionofautomationsoftwarewasnot,apositivecorrelationwouldsuggestthatAIandRIarecomplements,whileanegativecorrelationwouldsuggestsubstitution.Inshort,westudywhethercompaniesthatusedmoreteleworkinvestedinmoreorlessservice-tasksoftwareautomation,likeRPA.
Thefinalsubstantivesectionturnstoanotherindirectproxyforsubstitutionandcomplementarybehavior.Thesurveyincludesquestionstoworkersabouttheirexpectationsaboutfutureuseofteleworkandsoftwareautomation.ThisevidencestrandprovidesadditionalinsightintotherelationshipbetweenRIandAI.
Whiletheevidencepresentedinthispaperisnotconclusive,whatisclearisthatonthewholeteleworkandautomationdonotseemtobesubstitutes.Weshowthatteleworkandautomationaretechnicallypossibleformanyoccupations,butthereisnoclearevidencethatautomationwasless
5
usedinoccupationswhereteleworkrosethemost.WetakethisaspreliminaryevidencethatinthecontextoftheCovid-19shock,AIandRIareeitherindependentorcomplementsnotsubstitutes.
2TeleworkabilityandautomatabilityintheUSandJapan:The
globoticsquadrantdiagram
Thingscanbesubstitutable,complementary,orindependent.Theideaofsubstitutabilitymeanstwothingscanbereplacedbyoneanother;complementaritymeanstwothingsareusedtogether.Itisalsopossiblethatthetwothingsareindependentandhavenothingtodowitheachother.Thesedefinitionsareimportantwhenwethinkaboutofficeautomationandteleworkbecausenotalljobscanbeautomatedandnotallcanbedoneviatelework.
Thepossibilitythatautomationandremoteworkaresubstitutesorcomplementsarisesforoccupationswherebothautomationandteleworkarefeasible.Incontemplatingtheirimpact,wemustfirstdeterminewhichoccupationsarecapableofbeingautomatedandperformedremotely.Toclassifyoccupationsbasedonthissuitability,weintroducethe“globoticsquadrant”diagram.Itallowsustoassesstheteleworkabilityandautomatabilityofeachoccupation,thusdeterminingthepotentialforautomationtosubstituteforremoteworkandviceversa.
2.1Whichservicejobsareoffshoreandautomatable?
Sincetheonsetofthepandemicin2019,economistshavemadeseveralattemptstoclassifyoccupationsintoteleworkableandnon-teleworkablecategories.ThemostfamouswasbyDingleandNeiman(2020)fortheUS,whilstthebest-knowneffortwasbyFreyandOsborne(2013).ThesearethemeasuresweuseinthefirstinstancewhenlookingattheUS.
Weplotoccupationsinaquadrantdiagram,whatwecalltheGloboticsQuadrant(Figure1)
3
.Thehorizontalandverticallinesaredrawnatthemeanvalueforalloccupations,sooccupationstotheleftoftheverticallinearelessthanaverageautomatableandthosebelowthehorizontallinearelessthanaverageteleworkable.WhiletheoriginalFrey-OsborneandDingle-NeimanestimatesweredoneattheBLSoccupationlevel,itwouldbetooclutteredtopresentallinasinglediagram,sowehaveaggregatedtheBLSoccupationsintoJapan’s38occupations(NIRAclassification)weightedbyoccupationallaborforceinJapan(PopulationCensus)(SeeAppendixTable).WefirstfocusonlyonUSdata.BelowwecontrasttheseresultswithsimilarmeasuresforJapan.
3OriginalideaandbasicconceptofthediagramonpolicyperspectivewerefirstlyproposedbyOkubo(2022c).
6
Figure1:TheUSgloboticsquadrant:occupationsbyautomatabilityandteleworkability
Source:Authors’elaborationofdatafromDingel-Neiman(telework),andFrey-Osbourne(automatability).Note:Eachpointrepresentsanoccupation;x-axisshowstheautomatabilityscore(from0to1),andy-axisisteleworkableScore(0to1).OccupationsgroupedintoJapan’sNIRA38aggregates.
Toreadthediagram,keepinmindthefactthatoccupationsintheNortheastquadrantareexposedtoaboveaverageoffshorability(takingteleworkabilityasaroughindicatorofoffshorability)andtoaboveaverageautomation.ThoseintheSouthwestareexposedtobelowaverageteleworkabilityandautomatability.Intheothertwoquadrants,theoccupationshaveaboveaverageexposuretoteleworkabilityorautomatabilitybutbelowaverageexposuretotheother.ThelabelsinthediagramshowthenumberofUSworkerswithoccupationsineachofthequadrants.
Thekeytakeawaysaresimple.First,occupationsarespreadacrossallfourquadrants,sotheimpactofadvancingdigitaltechnologywillvarygreatlybyoccupation.TherecanbenouniversalanswertothequestionofwhetherAIandRIarecomplementsorsubstitutes.Forparticularoccupations,AIandRIcanbesubstitutesorcomplements,butnotasageneralstatementaboutalloccupations.
Second,thereareabout12millionworkersintheoccupationsfoundintheNortheastquadrant,whichisabout10%ofoccupationsclassified(Table1).ThesearejobswhereAIandRIarebothrelevantandthuspotentiallycomplementsorsubstitutestoeachother.Theoccupationsinthequadrantaregeneralclerical,outdoorservice,andofficeapplianceworkers.
Thereareabout21millionworkersinoccupationsintheNorthwestquadrant(Table1).Thesejobsaredifficulttoautomatebuteasilyundertakeninremoteplaces.Theoccupationsareaccountancyclerks,productionrelatedclerical,agriculturalforestryandfisheryengineers,foodanddrinkcookingstaffandservingcustomers,transportationandmachineoperation,agricultureforestryandfishery,manufacturingprocess,carryingcleaningpackagingandrelated,sales,constructionandmining,familylifesupportandcareservice.
7
Table1:USoccupationsandjobs(millions)inUSgloboticsquadrants
Quadrant
Jobs
Quadrant
Jobs
Northwesttotal
20.7
Northeasttotal
11.7
Accountancyclerks
2.5
Generalclerical
1
Production-relatedclerical
0
Outdoorservice
1.4
Agriculture,forestry,andfisheryengineers
0.6
Officeapplianceoperators
4.6
Foodanddrinkcooking,staffservingcustomers
6.6
Transportandpostclerical
0.9
Transportandmachineoperation
1.7
Salesclerks
1.8
Agriculture,forestryandfishery
0.8
Managerofresidentialfacilitiesandbuildings
1.4
Manufacturingprocess
1.6
Financeandinsuranceprofessionals
0.6
Carrying,cleaning,packaging,andrelated
5.4
Sales
0.1
Constructionandmining
0.2
FamilyLifeSupportandCareService
1.1
Source:Authors’elaborationofdatafromDingel-Neiman(telework),andFrey-Osbourne(automatability).Note:Eachpointrepresentsanoccupation;x-axisshowstheautomatabilityscore(from0to1),andy-axisisteleworkableScore(0to1).
OccupationsintheSouthwestcornerarebothdifficulttoautomateanddifficulttoundertakefromaremotelocationsotheissueofsubstitutabilityandcomplementaritydoesnotreallyarise.Manyoftheseareinfactlistedashavingzeroteleworkability,sotheissueofsubstitutesversuscomplementsdoesnotarise.
Third,thereisnoclearcorrelationbetweenteleworkabilityandautomatability.Forinstance,wedonotseemostprofessionsintheNortheastandSouthwestquadrantsaswouldbethecaseifautomationandteleworkabilitywerepositivelycorrelatedacrossprofessions.Thisstressestheneedfornuancewhenthinkingabouttheimpactofdigitaltechnologyonthefutureofwork.
Fourth,thelackofaclearpositiveornegativecorrelationissuggestiveofalackofeconomy-widesubstitutabilityorcomplementarity.Ifalltheoccupationshadlinedupinapositivelyslopedline,wecouldsaythatinareduced-formsense,teleworkandautomatabilitytendedtobedonetogether.Whenyouseealotofteleworkability,youseealotofautomatability.Bycontrast,aclearnegativecorrelationwouldhavesuggestedareduced-formsubstitutability.Occupationstendedtobeeitherautomatableandnotveryteleworkable,orviceversa.
2.2JapanesequadrantcontrastwithUS
ThescoresinFigure1arebasedonananalysisofthetasksinvolvedinthevariousoccupationsanalyzedintheUS.AsimilaranalysishasbeendoneforJapan.Specifically,weusetheteleworkabilityindexofKotera(2020)basedonrecalculatingDingelNeiman’sremoteworkindexbyusingtheJapan’sO*NET.WealsousetheautomatabilityindexbyFreyandOsbourne(2015,NRIreport),recalculatedastheJapanesecase.
8
Teleworkability
22
2223
10
24
26
29
32
3529
8
0.00
0.00
23
1.00
27
8
35
28
Automatiability
1.00
US
JP
0.50
10
30
28
0.50
32
24
273737
30
26
Overall,Japaneseoccupationsarejudgedasbothlessteleworkableandlessautomatable.Onthescale
ofzerotoone,theUSaverageautomatablescoreis0.51whileitis0.32inJapan.Forteleworkability,theaveragesare0.47and0.41.
Figure2:AutomatabilityandTeleworkable,officeandwalkingserviceoccupations:USvsJapan
Teleworkability
1.00
25
25
1
19
12
19
6
2
12
11
20
20
US
JP
1
11
6
2
0.50
0.00
0.000.501.00
Automatiability
#OfficeWorkOccupations#WalkingServiceOccupations
1Administrativeandmanagerialworkers8Publichealthnurses,midwives,andnurses
2Researchers10Professionalsocialwelfareworkers
6Dataprocessingandcommunicationengineers22Salesclerks
12Financeandinsuranceprofessionals23Outdoorserviceworkers
19Generalclericalworkers24Transportandpostclericalworkers
20Accountancyclerks26Salesworkers
25Officeapplianceoperators27WorkersinFamilyLifeSupportandCareService
28Occupationalhealthandhygieneserviceworkers
29Foodanddrinkcooking,staffservingcustomers
30Managerofresidentialfacilitiesandbuildings
32Securityworkers
35Transportandmachineoperationworkers
37Carrying,cleaning,packaging,andrelatedworkers
Source:Authors’elaborationofdatafromDingel-Neiman(telework),andFrey-Osbourne(automatability).Note:Eachpointrepresentsanoccupation;x-axisshowstheautomatabilityscore(from0to1),andy-axisisteleworkableScore(0to1).
Figures2and3compareUSandJapaneseteleworkabilityandautomatabilityscoresbydistinguishingoccupationsbythreebroadcategoriesofserviceoccupations:officeworkers,professionalserviceworkers,andwalkingserviceworkers(seelistunderthefigures).
Figure2(leftpanel),showsthecomparisonforofficeworkers,itindicatesthattheresultsfortheUSandJapanarenotverydifferentwhenitcomestooffice-basedoccupations.AllbuttwooftheoccupationsareinthesamequadrantforboththeUSandJapanesecase.Moreover,thereisnotapervasivedifferenceineitherdimension.Forexample,threeoftheeightoccupationsarereportedasmoreteleworkableinJapan,andfouraremoreautomatable,buttwoarebothmoreautomatableandteleworkable,andthreehavelowerscoresinJapanonbothdimensions.Bycontra
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 2024年杭州市富阳区卫健系统事业单位招聘考试真题
- 小学《国家学生体质健康标准》评分标准表
- 氧化应激机制-第1篇-洞察及研究
- 2025年四川省遂宁市中考语文试题含解析
- 2025年重庆育才成功学校英语七下期中学业质量监测模拟试题含答案
- 青少年法制教育讲座稿
- 技术创新过程中的商业秘密保护
- 柴油市场调研与购销策略合同
- 常州二手房交易税费减免及补贴合同
- 提升学习效率的数据结构学习法
- 荆州中学2024-2025学年高二下学期6月月考历史试题答案
- 外科换药拆线技术规范
- 2025年四川泸州市中考数学试卷真题及答案详解(精校打印)
- 2025年中考考前最后一卷化学(武汉卷)(全解全析)
- 2026届高考语文复习:直击2025年语文高考阅读客观题关键词比对
- 江西中考语文试题及答案
- 19S406建筑排水管道安装-塑料管道
- 武汉理工大学船舶建造工艺学期末考试试卷试题二
- 动力电池电气元器件选型报告
- 人教小学英语四年级下册单词表
- 精选小学数学教师招聘与进城考试试题(10套)
评论
0/150
提交评论