IMF-在家工作在發(fā)展中經(jīng)濟(jì)體會繼續(xù)嗎？（英）

WillWorkingfromHomeStickinDevelopingEconomies?MarinaConesaMartinez,FutoshiNarita,

andChrisPapageorgiouWP/23/112IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutive

Board,orIMFmanagement.2023MAY?2023InternationalMonetaryFundWP/23/112IMFWorkingPaperResearchDepartmentWillWorkingfromHomeStickinDevelopingEconomies?PreparedbyMarinaConesaMartinez,FutoshiNarita,

andChrisPapageorgiou*Authorizedfordistributionby

ChrisPapageorgiouMay2023IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.

TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,or

IMFmanagement.ABSTRACT:Indevelopingeconomies,ashifttoworkingfromhomeduringtheCOVID-19pandemicvariessubstantially.Anincreaseinteleworkingdaysperweekrangesfrom0.7to17.6percentagepointsacross10developingcountriescoveredbyanonlinesurveytoabout500respondentspercountry.Anestimatedincomediscountassociatedwithteleworkdisappearedtemporarilyattheonsetofthepandemic.Acalibratedmodelindicatesthatworkers’preferencestoteleworkmaylargelydependontheireducationalattainments.Whetherteleworkwillsustaininthesecountriescoulddependonobstaclestotelework,particularlyforworkerswithlesseducation,andadegreeofeconomy-wideexternality.RECOMMENDEDCITATION:ConesaMartinez,Marina,FutoshiNarita,andChrisPapageorgiou.2023.“WillWorkingfromHomeStickinDevelopingEconomies.”IMFWorkingPaper

23/XX.JELClassificationNumbers:Keywords:E24,J22,O15,O33COVID-19;developingeconomies;externality;workingfromhomeMConesaMartinez@;FNarita@;CPapageorgiou@Author’sE-MailAddress:*Acknowledgment:WethankNickBloom,MarinaMendesTavares,FerdinandoMonte,MachikoNarita,CarloPizzinelli,XinTang,andotherIMFcolleaguesfortheirhelpfulcomments.Thispaperispartofresearchprojectson“macroeconomicpolicyinlow-incomecountries”(IATIIdentifier:GB-1-202960)supportedbytheU.K.’sForeign,CommonwealthandDevelopmentOffice(FCDO)andthepartnersintheIMF’sCOVID-19CrisisCapacityDevelopmentInitiative(CCCDI)—Belgium,Canada,China,Germany,Japan,Korea,Spain,Singapore,andSwitzerland;andon“macroeconomicresearchonclimatechangeandemergingrisksinAsia”supportedbytheGovernmentofKorea.TheviewsexpressedhereinarethoseoftheauthorsandshouldnotbeattributedtotheIMF,itsExecutiveBoard,itsmanagement,orthesupportingpartners.Weareresponsibleforanyremainingerrors.WORKINGPAPERSWillWorking

fromHomeStickinDevelopingEconomies?Preparedby

MarinaConesaMartinez,FutoshiNarita,andChrisPapageorgiou11

Acknowledgment:WethankNickBloom,MarinaMendesTavares,FerdinandoMonte,MachikoNarita,CarloPizzinelli,XinTang,andotherIMFcolleaguesfortheirhelpfulcomments.Thispaperispartofresearchprojectson“macroeconomicpolicyinlow-incomecountries”(IATIIdentifier:GB-1-202960)supportedbytheU.K.’sForeign,CommonwealthandDevelopmentOffice(FCDO)andthepartnersintheIMF’sCOVID-19CrisisCapacityDevelopmentInitiative(CCCDI)—Belgium,Canada,China,Germany,Japan,Korea,Spain,Singapore,andSwitzerland;andon“macroeconomicresearchonclimatechangeandemergingrisksinAsia”supportedbytheGovernmentofKorea.TheviewsexpressedhereinarethoseoftheauthorsandshouldnotbeattributedtotheIMF,itsExecutiveBoard,itsmanagement,orthe

supportingpartners.Weareresponsibleforanyremainingerrors.IMFWORKINGPAPERSWillWorkingfromHomeStickinDevelopingEconomies?Contents1.Introduction

32.Teleworkindevelopingeconomiesduringthepandemic

5Figure1.Cross-countrydifferenceinashifttoteleworkingattheonsetofthepandemic

7Figure2.Transitionacrossworkarrangementsindevelopingeconomiesuponthepandemic8Figure3.Estimatedincomedifferentialsbyworkarrangement(percent)

93.Preferencesorprofitability:Amodelofteleworkchoice10Figure4.Preferenceshiftsimpliedbyteleworksharesandestimatedrelativeincome

(percent)11Figure5.ComparingBrazilandVietnamwithoutorwithexternality

124.Conclusion13AppendixA.Onlinesurveydetails14AppendixTableA1.Demographicssummary—2nd

roundonly

15AppendixTableA2.Incomesummary—2nd

roundonly16AppendixTableA3.Teleworkandpre-pandemicfundamentals—2nd

roundonly17AppendixFigureA1.Backgroundinformationonmainincomeearners—2nd

roundonly18AppendixFigureA2.Employmentstatusofmainincomeearners—2nd

roundonly19AppendixFigureA3.Teleworksituationsindevelopingcountries—2nd

roundonly19AppendixFigureA4.Teleworksituationsbyworker/householdcharacteristics—2nd

roundonly20AppendixFigureA5.Notonlyteleworkbutalsoparttimejobsincreasedduringthepandemic21AppendixFigureA6.Teleworkshareandrecoveryinworkdaysfromthepandemic-relatedshocks

...

22AppendixFigureA7.ReflectionofCOVID-19infectionriskwhenchangingjobs,byeducation

22AppendixB.Characteristicsassociatedwithtelework23AppendixTableB1.Teleworksharesandworkers’characteristics

25AppendixTableB2.Incomelevelsandworkers’characteristics26AppendixTableB3.Estimatedincomedifferentialsbyworkarrangementandeducation27AppendixTableB4.Severeincidentsrelatedtothepandemicandrespondent’scharacteristics28AppendixC.Modeldetails

29C.1Model29C.2Calibration33C.3Derivation34References38INTERNATIONALMONETARYFUND2IMFWORKINGPAPERSWillWorkingfromHomeStickinDevelopingEconomies?1.IntroductionTheCOVID-19pandemichadchangedthewayweworkglobally,butadatagapexistsonhowsignificantitwasindevelopingeconomies.Aswiftshifttoworkingfromhomeattheonsetofthepandemiciswelldocumentedformanyadvancedeconomies,especiallyintheUnitedStates(e.g.,Barrero,Bloom,Davis2021a;Bick,Blandin,Mertens2022;Jaumotteandothers2023).Fordevelopingeconomies,however,onlylimiteddatahavebeenavailableregardingtheextenttowhichpeopleactuallyadjustedtheirworkarrangements(exceptforcotemporaneousresearchbyAksoyandothers2022).1Tonarrowthisdatagap,thispaperpresentsevidenceofhowpeopleadoptedteleworkin10developingcountries,coveredbyanewonlinesurveydatasetcompiledbytheIMFResearchDepartment.2

Thesurveydatasetincludesabout500respondentsfromeachofthefollowing10countries:Argentina,Brazil,Ecuador,India,Indonesia,Pakistan,Peru,SouthAfrica,Türkiye,andVietnam.Howmanydaystherespondentsworkedperweekwithoutcommutingwasaskedinearlyandlate2021togatherretrospectiveinformationaboutbeforeandaftertheonsetofthepandemic,inadditiontobackgroundcharacteristicsofrespondents.Thedatasetshowsasubstantialcross-countrydifferenceinthedegreeoftheshifttoworkingfromhome.Severalcountriesshowedalargerincreasethanothersintheshareofteleworkdaysperweek,ranging11.7-17.6percentagepoints(Argentina,Ecuador,Peru).Adjustmentwasverylimitedinothercountries(Indonesia,Türkiye),rangingfrom0.7to1.2percentagepoints.Inmostcases,theoverallteleworklevelwasmuch

belowtheU.S.level,andparttimejobswereincreased,likelyasanalternativewaytoadjusttothepandemicsituation.ButsomepatternsaresimilartothoseintheUnitedStates.Forexample,teleworkdaysincreasedmoreforworkerswithhighereducationalattainments.Thedatasetalsorevealsthatremoteworkwasprevalentforlow-incomehouseholdsbeforethepandemicinmanyofthesecountries(particularlyPeru,Vietnam).ForU.S.data,Oettinger(2011)findsthattherewasasizablediscountof25-27percentinwagesassociatedwithhome-basedworkasof1980,butitalmostdisappearedasof2000.Inthecaseofthe10countriescoveredbytheonlinesurveydataset,anestimatedincomediscountwassizableasof2019,at33percent,butitalmostdisappearedattheonsetofthepandemicandthenpartiallyreversedto-17percentinlate2021.Theseestimatedassociationsareconsistentwiththestrongincreaseintheteleworksharebyhigh-incomehouseholds.Butthen,thequestioniswhetheritisbecausethepandemic-relatedconcernsmadepeopleprefertoworkremotely

moreorbecausetherelativeprofitabilityofteleworkjobsincreaseduponthepandemic,comparedtoon-sitejobsthatweremoreexposedtoCOVID-19infectionrisksormoresubjectto

containmentmeasures.Asimplemodelisdevelopedtoparsimoniouslydisentangle

theteleworkchoiceintoworkers’preferencesandjobs’relativeprofitability.ThemodelisbuiltontheideaofBick,Blandin,andMertens(2022)touseanequilibriummodeltoseparate“teleworkadoption”and“teleworksubstitution”fromobservedteleworkbehaviors,respectivelycorrespondingtoapreferenceshifttowardteleworkandachoicebasedonprofitability1

Aksoyandothers(2022)conductanonlinesurveyonworkfromhomein27countriesincludingseveraldevelopingcountries.Itcoversthreecountries(Brazil,India,Türkiye)thatarealsocoveredinthesurveydatasetinthispaper.2

Thiscustomizedonlinesurveywas

implementedbyNielsen,aglobalmarketresearchcompany.TheworkisfinanciallysupportedbytheU.K.’sForeign,CommonwealthandDevelopmentOffice(FCDO),theGovernmentofKorea,andthepartnersin

theIMF’sCOVID-19CrisisCapacityDevelopmentInitiative(CCCDI)—Belgium,Canada,China,Germany,Japan,Korea,Spain,Singapore,andSwitzerland.INTERNATIONALMONETARYFUND3IMFWORKINGPAPERSWillWorkingfromHomeStickinDevelopingEconomies?(e.g.,highon-sitejobcostsduetosocialdistancing).Unliketheirmodel,ourmodelassumesasearch-matchingfrictioninthelabormarketsothatjobprofitability

istightlylinkedtoworkers’earnings,makingtheteleworkchoicebasedonasimplecomparisonbetweenprofitabilityandworkers’willingnesstoworkremotely.Undersomesimplifyingparametricassumptions,calibrationcanbeeasilydonefrom

anobservedteleworkshare,totalemploymentshare,andanestimatedrelativeincomefromteleworkcomparedwithon-sitework.Themodelalsoincorporatesexternalityinprofitabilitydependingontheeconomy-wideshareofthesameworkmodality(on-siteorremote).Ontheonehand,localagglomerationspilloversmayincrease

theprofitabilityofin-officejobsifmorejobsaredoneinon-siteofficespace,asmodeledbyDelventhal,Kwon,andParkhomenko(2022).Ontheotherhand,strategiccomplementaritiesamongfirmsmayexistbecauseteleworkjobsmaybecomemoreviableiftheircounterpartyfirmsorclientshavealsoadoptedremoteworkarrangements,asdiscussedbyBarrero,Bloom,andDavis(2021a).3

Themodeldemonstratesthatthiscoordinationmechanismcaninducediverseteleworkresponsestosmalldifferencesinfundamentals,suggestingthatwhetherteleworkwillsustaininthefutureorshrinkbacktopre-pandemiclevelsmaydependonrelativelysmalldifferencesinobservablefeaturessuchasthelevelof

accesstointernetoreducationalattainment.Theseinsightsinformanongoingdebateonthefutureofremotework.TheliteratureonworkingfromhomehasstronglyexpandedsincetheonsetoftheCOVID-19pandemic.Theliteraturehasfocusedonadvancedeconomies,inparticulartheUnitedStates.4

Akeyquestioniswhetherworkingfromhomewillstickgoingforward,evenafterthepandemicsituationends(Barrero,Bloom,Davis2021a).Workingfromhomehasimplicationsonproductivity,inequality,andresilience,aswellasinflationpressures(Barrero,Bloom,Davis2021b,Barreroandothers2022).5

Thecalibratedmodelanalysisofthispapershedslightonapossiblekeymechanism,onwhichthesustainabilityofworkingfromhomedepends.Thispaperalsocontributestotheliteratureon

developmenteconomicsbynarrowingthedatagaponteleworkevidenceindevelopingeconomies.Intensiveeffortshavebeenmadetogatherinformationonthepandemicsituationsindevelopingeconomies.6

Butdirectinformationontheactualteleworkbehaviorshasstillbeenscarce(exceptfortherecentstudybyAksoyandothers2022).Withthelimiteddataavailability

manyresearchersimputetheindexofteleworkabilityusingtheU.S.O*NETdata,followinganinfluentialstudyby3

Barrero,Bloom,andDavis(2021a,p.15)giveanexamplethat“[workingfromhome]becomesmoreviableforlawfirmswhenmoreoftheirclientsworkremotely.”Strategiccomplementaritymaygenerateasituationwhereequilibriumoutcomescandiffereveniffundamentalsaresimilar(Cooper,John1988).SeealsoMino(2017)forasurveyofequilibriumindeterminacy,explainingstrategiccomplementarityasasourceofindeterminacy.4

FortheUnitedStates,inadditiontoseveralsurveystudies(e.g.,Barrero,Bloom,Davis2021;Bick,Blandin,Mertens2022;Brynjolfssonandothers2020;Footeandothers2021;Ozimek2020),theU.S.CensusBureaualsopublishesdataontelework(HouseholdPulseSurvey:/programs-surveys/household-pulse-survey.html),andtheU.S.BureauofLaborStatisticsaddedfivequestionstotheCurrentPopulationSurvey(/cps/)onteleworkintthepandemiccontext.5

SeealsopapersandarticlesavailableattheWFHResearch’swebsite(/).6

ForstudiesusingtheWorldBank’sHigh-frequencyPhoneSurveys(WorldBank2021),seealso,forexample,Josephson,Kilic,andMichler(2021),Kuglerandothers(2021),Khamisandothers(2021),Kimandothers(2021),Narayanandothers(2022),anddePaz,Gaddis,andMuller(2021).OthersurveysstudiesincludePERC(2021),Kansiimeandothers(2021),andDurizzoandothers(2021)forsub-SaharanAfrica;Bottan,Hoffmann,andVera-Cossio(2020)andArteaga-Garavitoandothers(2020)forLatinAmericaandtheCaribbean;ShinozakiandRao(2020)forthePhilippines;and

Eggerandothers(2020)forninedevelopingcountries.ForIndia,Gupta,Malani,andWoda(2021)usemonthlyhouseholdsurveyby

theCentreforMonitoringIndianEconomyPvt.Ltd.(CMIE).Onremotelearning(insteadofworking),Asanovandothers(2021)provideevidenceinEcuador.INTERNATIONALMONETARYFUND4IMFWORKINGPAPERSWillWorkingfromHomeStickinDevelopingEconomies?DingelandNeiman(2020),toprovideindirectevidenceformanydevelopingeconomies.7

TheevidenceprovidedbythispapercomplementsthosestudiesbasedonimputationfromtheU.S.evidence.Therestofthepaperisorganizedasfollows.Section2documentsevidenceof

teleworkbehaviorsinthe10developingcountriescoveredbytheonlinesurvey.Section3considersasimplemodelto

separatetwofactorsbehindteleworkbehaviors—preferencesorprofitability.Section4concludes.AppendixAprovidesdetailsofthenewonlinesurvey,AppendixBpresentsregressionresultsassociatedwithtelework

anditsimplicationsduringthepandemic,andAppendixCexplainsthedetailsofthemodel.2.Telework

indevelopingeconomiesduringthepandemicAnonlinesurveywasconductedtobetterunderstandtheimpactofthepandemicindevelopingeconomies.ThesurveywasdesignedbytheIMFResearchDepartmentincollaborationwiththeNielsenCompanyin10developingeconomies(Argentina,Brazil,Ecuador,India,Indonesia,Pakistan,Peru,SouthAfrica,Türkiye,Vietnam).Atotalof5,058respondents—withanevendistributionacrossthe10countries—participatedinthesurveybetweenOctoberandNovember2021.Thequestionnaireaddressedvariousaspects,includingdemographicandhouseholdcharacteristics,employmentstatusandworkarrangements,pandemic-relatedincidents(e.g.,illness,lackoffood),perceptionsofincomeandprices,andpoliticalsupport.Thesurveygatheredyearlyandquarterlydataforpre-pandemicandpandemicperiods,byaskingretrospectivequestionsbasedontherespondents’recollections.SeeAppendixAfordetailsanddescriptivestatistics.Thequestionnaireisavailableonline.Thesurveyaskedthefollowingquestionstomeasurethefrequencyofteleworkbeforeandaftertheonsetofthepandemic.First,thesurveyasked(Q8a,Q8b,Q8c),“Howmanydaysperweekdidyouusuallywork?”andthenasked(Q9a,Q9b,Q9c),“Howmanydaysperweekdidyouusuallycommutetowork?”showingthefollowingmultiple-choiceresponses:8????Zeroorhadneverworked1-2days3-4days5daysormoreThequestionswereaskedforthefollowingthreeperiods,basedontheWHO’sdeclarationofaglobalpandemiconMarch11,2020,andoneyearafterthedeclaration:???Beforeend-March2020(i.e.,beforetheonsetofthepandemic)Betweenend-March2020andend-March2021(i.e.,duringthefirstyearofthepandemic)Afterend-March2021(i.e.,afteroneyearofthepandemic)7

See,forexample,Brussevich,Dabla-Norris,andKhalid(2020),Gottlieb,Grobovsek,Poschke,andSaltiel(2021),Hasan,Rehman,andZhang(2021),Hatayama,Viollaz,andWinkler(2020),andIMF(2020).ManyresearchersusetheO*NETdatabase(e.g.,Boeri,Caiumi,Paccagnella2020,Famiglietti,Leibovici,Santacreu2020;Mongey,Pilossoph,Weinberg2021).TheAmericanTimeUseSurveyisanotherdatasourceinmeasuringteleworkability(e.g.,Alonandothers2020,Hensvik,LeBarbanchon,Rathelot2020,PapanikolaouandSchmidt2022).8

Theuseofranges(e.g.,1-2days)wasintendedtoreducethefatigueofsurveyrespondents,withanadditionalconsiderationthatthissurveyaskedretrospectivequestionsbasedonrespondents’recollections,whichmaybemore

burdensome.ThesequestionsfollowthoseoftheReal-timePopulationSurvey(RPS)byBick,Blandin,Mertens(2022),althoughtheRPSasked

fortheexactnumberofdays,insteadofranges.Thereisnodistinctionbetweenwhetherpeopleworkfromhomeasself-employedornot.INTERNATIONALMONETARYFUND5IMFWORKINGPAPERSWillWorkingfromHomeStickinDevelopingEconomies?Ifthesurveyrespondentswerenotthemainincomeearneroftherespondent’shousehold(checkedbyQ18),thenthesamequestionswererepeated(Q25a-Q26c)togatherinformationaboutthemainincomeearner’steleworksituation.9

Wefocusontheteleworksituationofmainincomeearnersbecausetheirworkingsituationsaremorecloselyrelatedtohousehold-levelincomethatwemainlyanalyze.10Thedegreeofteleworkingismeasuredintwoways:(1)theshareofteleworkdaysperweekand(2)acategoricalindicatorofwhetherthepersonworkedremotelyallthetime,partofthetime,ornoneofthetime.Thefirstoneiscalculatedasoneminustheratioofcommuteworkdaysoverthetotalworkdaysperweek,takingthemid-pointoftherangeresponses(e.g.,1.5daysif1-2dayswasselected).ThesecondoneisaclassificationvariablethatfollowsBick,Blandin,andMertens(2022).Itis“Commuteonly”iftherespondentdidnotteleworkduringtheweek,“Workfromhomeonly”ifsheorheworkedfullyremotely,and“Workfromhomesomedays”forpartialtelework.Thelattertwocategoriesareoftencombinedas“sometelework”inanalysis.Lookingatthedataacrosscountriesrevealsdiverseteleworkexperiencesbeyondthevariationsin

featuresobservedatthecountrylevel(Figure1).Anincreaseinteleworkwaslargeandmorethan10percentagepointsforArgentina,Ecuador,andPeru,whereasitwasverylimitedforIndonesiaandTürkiyebyjustaboutonepercentagepoint.Thereisalmostzerocorrelation

(0.08)betweenthepre-pandemiclevelsandtheincreasesuponthepandemic.AlthoughIndiamaybeanoutlier,teleworkdaysperweekbothbeforeandaftertheonsetofthepandemicweregenerallypositivelycorrelatedwiththepopulationshareofinternetusersasof2019andtheaverageyearoftotalschoolingforpeopleofage15-64asof2015(seeAppendixTableA3fordatadetails).WhenexcludingIndia,thecorrelationcoefficientsare0.46and0.70forinternetusersandeducationalattainment,respectively.EvenincludingIndia(stillexcludingtheUnitedStates),theincreaseinteleworkdaysuponthepandemicpositivelycorrelatedwiththesepre-pandemicfundamentals(withcorrelationcoefficientsof0.41and0.69,respectively).ButasimplemultivariateregressiononthesetwofeaturesandthelogarithmofGDPpercapitashowsthatabout70percent(or60percentexcludingIndia)ofthecross-countryvariationintheteleworkshareuponthepandemicisunexplained.Afteroneyearfromtheonsetof

thepandemic,teleworklevelsweremostlykeptunchanged.Theadjustmentsatthisstagewereminorformanycountries.Argentinasawareversalofitslargestincreaseamongthe10countriesby8percentagepoints,althoughmaintainingtheshareofteleworkdaysat20percent.ForIndia,eventhoughthecountrysawawidespreadinfectionoftheDeltavariantinApril-June2021,thelevelofteleworkdidnotchangemuch.Vietnamsawarelativelylargeincreaseintheshareofteleworkdaysto23percent.Variationsunexplainedbythefundamentalsremainatabout70percent(butreducetoabout40percentifIndiaisexcluded).9

Q18asked“Whoisthemainincomeearnerinyourhousehold?Ifthereismorethanonepersonwhoearnssimilaramounts,pleasepickoneamongthem,forwhichyouhavethemostinformation.”withthefollowingchoiceresponses:“Yourself,”“Someoneelsethanyourself,”or“Noone”.Lessthan1percentofrespondentschose“Noone”whilechoosingcurrentlybeing“employed.”Weincludetheserespondentsinthesampleofmainearners.10

Teleworkwasgenerallymoreprevalentamongrespondentswhoarenotmainearners.INTERNATIONALMONETARYFUND6IMFWORKINGPAPERSWillWorkingfromHomeStickinDevelopingEconomies?Figure1.Cross-countrydifferenceinashifttoteleworkingattheonsetofthepandemic(Percent,unlessindicatedotherwise)605040302010014121086420USAARGTURBRAVNMZAFECUPERIDNINDPAKInternetusers(per10ppl),rightaxis

YearsofEducationalAttainment,rightaxis

TeleworkY0

TeleworkY1

TeleworkY2Sources.IMFCOVID-19CustomTracker(2nd

round),Barrero,Bloom,Davis(2021a),BarroandLee(2013,updatedinSeptember2021),andWorldDevelopmentIndicators(WorldBank2022a).Notes.Thedotsshowthecountry-levelsample-weightedaveragesofteleworkdaysperweekinpercentoftotalworkdays,basedontheinformationonthemainincomeearneroftherespondent’shousehold(Q25a-Q26c)iftherespondentwasnotthemainincomeearner(Q18),andotherwise,basedontheinformationonrespondentsthemselves(Q8a-Q9c).Label“Y0”correspondstobeforeMarch2020,“Y1”correspondstoApril2020-March2021,and“Y2”correspondstoApril2021-October/November2021.SeeAppendixTableA3fortheunderlyingdata.Followingthetransitionofworkersacrossdifferentworkarrangementsshowsthatteleworkjobsincreasedattheonsetofthepandemicbutanincreaseinparttimejobswasalsoremarkable(Figure2).Almost20percentofrespondentswhodidnotworkfromhomebeforethepandemic

shiftedtoaworkarrangementwheretherewassometelework.Althoughtheproportionofworkersthatkeptsometypeofteleworkarrangementremainedatasimilarleveluntillate2021,therewererelativelysizablereshufflesbetweenon-siteandremoteworkers.Inparticular,almosthalfoftheteleworkersinthepre-pandemicperiodshiftedtofullyon-sitejobsattheonsetofthepandemic.

Itisnotclear

whethernewhiresin2021didteleworkmoreornot.Manyworkersalsoshiftedtoparttimejobs,includingthosewhoworkedremotely,andmorethantwothirdsofparttimejobswereon-site(AppendixFigureA5).Therearemoreworkerswhoneededtoadjusttheirworkarrangementsfromfulltimetoparttimejobsthanthosewhoadjustedtoteleworktopreservefulltimejobs.Apotentialexplanationisthatteleworkmighthavebeendifferentinessencepriortothepandemicindevelopingeconomiessothatadjustmentswerestillrequiredduetotheforcedincreaseinsocialdistancing.INTERNATIONALMONETARYFUND7IMFWORKINGPAPERSWillWorkingfromHomeStickinDevelopingEconomies?Figure2.Transitionacrossworkarrangementsindevelopingeconomies

duringthepandemic(Percent)PanelA.Remotevs.on-siteworkPanelB.Parttimevs.fulltimeworkSource.IMFCOVID-19CustomTracker(2nd

round).Notes.Foreaseofexposition,yearlabel“2019”correspondstobeforeMarch2020,“2020”correspondstoApril2020-March2021,and“2021”correspondstoApril2021-October/November2021.Thenumbersinparenthesesindicatethesharestothetotal(withpopulationweights).Thistypeofchartisoftencalled“Sankeyplot,”producedbytheStatacommandsankey_plot(Rios-Avila2022).SeeAppendixFigureA5forthetransitioncoveringtheintersectionofremoteandparttimework.Thecharacteristicsofworkerswhotendtoincreaseteleworkdaysmoreduringthepandemicarefoundtobeinlinewithexistingstudies.Thoseincludeworkerswithhighereducationalattainments,olderworkers,

thoselivingwithelderlypeople(reflectinghigherhealthrisksfromCOVID-19infection),andthoseworkedonjobsthatareeasytoconductfromhome(see

regressionresultsinAppendixBandchartsinAppendixFigureA4).Theworkersintheindustriesconsideredtobeessentialtotheeconomyorsociety

tendedtoteleworkless.Maleworkersteleworklessthanfemaleworkers(notstatisticallysignificantly).ThesepatternsareconsistentwiththeU.S.data(Barrero,Bloom,Davis2021a,Bick,Blandin,Mertens2022).Also,asobservedintheUnitedStates(Oettinger2011,Bick,Blandin,Mertens2022),low-incomepeopleworkedmoreinworkarrangementswithoutcommutingbeforethe

pandemic(particularlyPeru,Vietnam),althoughhigh-incomehouseholdsincreasedteleworkdaysmorestronglysincetheonsetofthepandemic.Therewasasizableincomediscountassociatedwithworkingremotelybeforethepandemic,butitdisappearedtemporarilyattheonsetofthepandemic,untilitpartiallyreemergedafterward(Figure3,PanelA).Theestimatedincomediscountassociatedwithworkingremotely,a