國民經(jīng)濟研究局-新冠疫情與全球收入不平等（英文）-2021.5-20正式版

NBERWORKINGPAPERSERIES

COVID-19ANDGLOBALINCOMEINEQUALITY

AngusDeaton

WorkingPaper28392

/papers/w28392

NATIONALBUREAUOFECONOMICRESEARCH

1050MassachusettsAvenue

Cambridge,MA02138

January2021,revisedFebruary2021

IamgratefulforcommentsandassistancetoTimBesley,Fran?oisBourguignon,AnneCase,WilliamEasterly,ChicoFerreira,IanGoldin,PennyGoldberg,GitaGopinath,RobJoyce,BrankoMilanovic,ChrisPapageorgiou,SamPreston,PaulSchreyer,JoeStiglitzandNickSternforcommentsandassistance.Errorsaremyown.IacknowledgefinancialsupportfromtheNationalInstituteofAgingthroughNBER,AwardNumberP01AG05842.TheviewsexpressedhereinarethoseoftheauthoranddonotnecessarilyreflecttheviewsoftheNationalBureauofEconomicResearch.

NBERworkingpapersarecirculatedfordiscussionandcommentpurposes.Theyhavenotbeenpeer-reviewedorbeensubjecttothereviewbytheNBERBoardofDirectorsthataccompaniesofficialNBERpublications.

?2021byAngusDeaton.Allrightsreserved.Shortsectionsoftext,nottoexceedtwoparagraphs,maybequotedwithoutexplicitpermissionprovidedthatfullcredit,including?notice,isgiventothesource.

COVID-19andGlobalIncomeInequality

AngusDeaton

NBERWorkingPaperNo.28392

January2021,RevisedFebruary2021

JELNo.F01,I14,O11

ABSTRACT

ThereisawidespreadbeliefthattheCOVID-19pandemichasincreasedglobalincomeinequality,reducingpercapitaincomesbymoreinpoorcountriesthaninrich.Thissuppositionisreasonablebutfalse.Richcountrieshaveexperiencedmoredeathsperheadthanhavepoorcountries;theirbetterhealthsystems,higherincomes,morecapablegovernmentsandbetterpreparednessnotwithstanding.TheUSdidworsethansomerichcountries,butbetterthanseveralothers.Countrieswithmoredeathssawlargerdeclinesinincome.Therewasthusnotonlynotrade-offbetweenlivesandincome;fewerdeathsmeantmoreincome.Asaresult,percapitaincomesfellbymoreinhigher-incomecountries.Countrybycountry,internationalincomeinequalitydecreased.Whencountriesareweightedbypopulation,internationalincomeinequalityincreased,moreinlinewiththeoriginalintuition.ThiswaslargelybecauseIndianincomesfell,andbecausethedisequalizingeffectofdecliningIndianincomeswasnotoffsetbyrisingincomesinChina,whichisnolongeragloballypoorcountry.ThatthesefindingsarearesultofthepandemicissupportedbycomparingglobalinequalityusingIMFforecastsinOctober2019andOctober2020.

AngusDeaton

SchoolofPublicandInternationalAffairs

127JulisRomoRabinowitzBuilding

PrincetonUniversity

Princeton,NJ08544-1013

andNBER

deaton@

0.Introduction

TheCOVID-19pandemichasthreatenedthelivesandlivelihoodsofless-educatedandless-wellpaidpeoplemorethanthoseofmoreeducatedandbetterpaid,manyofwhomcanstaysafelyathomeandcontinuetowork.TheincreaseindomesticincomeinequalityhasbeenoffsetbylargescalegovernmentincomesupportprogramsintheUSandinmanyothercountries.

Internationalincomeinequalityisanothermatter,andthereisawidespreadbeliefthatthepandemichasorwillincreaseinequalitiesinincomebetweencountries.Inoneofmanysuchexamples,GoldinandMuggah(2020),writingfortheWorldEconomicForumsay“inequalityisincreasingbothwithinandbetweencountries.”UNDP(2020)writes“Thevirusisruthlesslyexposingthegapsbetweenthehavesandthehavenots,bothwithinandbetweencountries.”Stiglitz(2020)laysouttherationale:“COVID-19hasexposedandexacerbatedinequalitiesbetweencountriesjustasithaswithincountries.Theleastdevelopedeconomieshavepoorerhealthconditions,healthsystemsthatarelesspreparedtodealwiththepandemic,andpeoplelivinginconditionsthatmakethemmorevulnerabletocontagion,andtheysimplydonothavetheresourcesthatadvancedeconomieshavetorespondtotheeconomicaftermath.”

Thisargumentseemscompelling,butitisgoodtocheckthedata,whichiswhatIdointhispaper.Idemonstratethatglobalinequality—definedasthedispersionofpercapitaincomebetweencountriestakingeachcountryasaunit—hascontinuedonitspre-pandemicdownwardtrend,andhasifanythingfallenfasterasaresultofthepandemic.Thisfindingisfragile,issensitivetooutcomesinsmalleconomies,andbearslittlerelationshiptowhatwemightreasonablycareabout,whichisinternationalinequalityinmateriallivingstandards.Itmayalsobetemporary.Alternatively,globalinequalitycanbemeasuredwitheachcountryweightedbyitspopulation,andbythismeasure,between-countryincomeinequalityhasincreased,largelybecauseIndiahasdonesomuchworsethantherichcountriesoftheOECD,inlinewiththeintuitivenotionofpoorcountries

2

sufferingthelargestincomeloss.TherelativesuccessofChinaduringthepandemichascontributedtoareductionin(population-weighted)inequality,butChinaistodaynolongeragloballypoorcountry,soChina’sexceptionallypositiveoutcomedidnotoffsettheinequalityincreasingeffectofthefallinincomeinIndia.TherapidgrowthofChinahas,fordecades,decreasedpopulation-weightedbetween-countryinequality,becauseithasliftedmorethanbillionpeopleupfromthebottomoftheworldincomedistribution.ButChinaisnolongeragloballypoorcountry,sothatwhenitgrowsmorerapidlythanothercountries,asitdidin2020duringthepandemic,itdoesrelativelylittletodecreaseglobalinequality.

Forreasonsthatareonlypartiallyunderstood,andmayincludemeasurementerror,poorercountriessufferedfewerCOVIDdeathspercapitain2020thandidrichercountries.Moreover,eachcountry’slossinpercapitanationalincomebetween2019and2020wasstronglyrelatedtoitspercapitaCOVIDdeathcount.Thesetwofactstogetherhavemeantthatpercapitaincomeshave,onaverage,fallenmoreincountrieswithhigherpercapitaincomesin2019;the97poorestcountrieslostanaverage5percentoftheir2019percapitaGDP,whiletherichest96countries,withanaveragepercapitaincomesixandaquartertimeslarger,lostanaverageof10percent.Thisneednothavenarrowedinternationalincomeinequality,butinfactitdidso.Countrybycountry,withtinycountriescountingthesameasgiantcountries,percapitaincomesareclosertooneanothernowthanin2019.

China(butnotIndia)hadfewdeathsandexperiencedpositiveeconomicgrowthin2020.Beforethepandemic,China’srapidgrowthhadliftedmorethanabillionpeopleupfromthebottomoftheglobalincomedistribution,andhaslongbeenresponsibleforareductioninglobalincomeinequalitywheneachcountryisweightedbyitspopulation.ButthiseffecthasbeenattenuatingasChina’sincomehasrisen.Today,outoftheworld’spopulationof7.8billion,4.4billionliveincountrieswhosepercapitaincomeislowerthanChina,whileonly2.0billionlivein

3

countrieswhosepercapitaincomeishigherthanChina.Duringthepandemic,theChineseeconomygrewwhilemostothereconomiesshrank,andwhilethisreducedpopulation-weightedglobalinequalitybetween2019and2020,theeffectwasnotlargeenoughtooffsettheinequality-increasingeffectof(muchpoorer)India’slossofincome,andpopulation-weightedglobalinequalityincreased.

Contrarytopre-existingtrends,thepandemicreducedglobalunweightedinequality,andincreasedglobalpopulation-weightedinequality.ThatmyfindingsareconsequencesofthepandemicissupportedbycomparinginequalitymeasuresusingIMFincomeestimatespre-andpost-pandemic.

ItisimportanttobeclearaboutwhatIamandamnotclaiminghere.Myresultssaynothingaboutwhetherthedegreeofsufferinghasbeenlargerorsmallerinpoorcountries;inparticular,theyareconsistentwiththepandemicincreasingpovertyaroundtheworld,inparticularwithestimatesthatbetween88and115millionpeoplewillbepushedintopoverty,WorldBank(2020).Evenifallcountrieshadthesamedeclineinpercapitaincome,thepoorercountrieswouldhavehadlargerincreasesinpovertybecausetheyhavemanymorepeopleneartheglobalpovertyline.Asitis,weknowfromDecerfetal(2020)that,comparedwithrichercountries,thesufferingfromthepandemichashitpoorcountriesmoreintermsofpoverty,andlessintermsofmortality.Allofmyresultscomefromnationalaccountsdata,andthereisalonghistoryofnationalaccountsdataonconsumptionandincomedifferingfromconsumptionandincomeasrecordedinthehouseholdsurveydatathatareusedfortheassessmentofpovertyandwithin-countryinequality.Beyondthat,GDPpercapitaisoftenapoorindicatorofmateriallivingstandardsifonlybecauseGDPcontainsmuch—suchasprofitsaccruingtoforeigners—thatarenopartofdomesticconsumption,evenasmeasuredinthenationalaccounts.

4

Myfindingsmaybetemporary.Thepandemicisnotdone,therearemoredeathstocome,andtheymayfallmoreheavilyonpoorercountries.Indeed,giventhatthepandemicstartedalongtrade-routes,andaffectedurbanbeforeruralareas,itisplausiblethatcurrentpatternswillcontinuetochange.Itisalsopossiblethatdeathsareseverelyunderstatedinpoorcountries,someofwhomdonothaveregularvitalstatisticssystemsthatcomprehensivelyreportdeathseveninnormaltimes.Mycalculationsusedatauptotheendof2020,beforevaccineshadanychancetoaffectoutcomes,andtheysaynothingaboutthehowvaccineswillbedistributedbetweencountries.Itisentirelyplausiblethatrichcountrieswillrecovermorerapidlyin2021andbeyond,whichwillwidenglobalinequalities.

Myresultsconcerntwodistinctmeasuresofinternationalincomeinequality,thedispersionofpercapitaincomebetweencountries,witheachcountryasaunitofobservation,andthedispersionofpercapitaincomebetweencountries,butwhereeachcountryisweightedbypopulation.Milanovic(2011,Chapters1and2)hasusefullylabeledtheseinequalitymeasuresasConcept1andConcept2respectively.Concept1treatseachcountryasanindividualandcalculatesinequalitybetweenthose“individuals.”Concept2pretendsthateachpersonintheworldhastheircountry’spercapitaincome,andthencalculatesinequalityamongallthesepersons.BothConcept1andConcept2arebetweencountrymeasuresandbothignorewithincountryinequality.Thedistributionofincomebetweenallpersonsintheworld,whichMilanoviccallsConcept3inequality,startsfromConcept2,butthenaddsinthedistributionofincomewithincountrieswhichisalsochangingbecauseofthepandemicandthepolicyresponsestoit.Becausebetweencountryinequalitiesinpercapitaincomearelargerthanwithincountryincomeinequalities,changesinConcept2inequalityareoftenagoodguidetochangesinConcept3inequality.

Itisentirelypossiblefortheglobaldistributionofincomeamongallpersonsintheworldtohavewidenedwhileoneorbothofthebetweencountrymeasureshavebeendecreasing.Inrecent

5

years,largelybecauseoftherapidgrowthinpercapitaincomesinIndiaandChina,population-weightedbetween-countryinequality(Concept2)hasfallen,whileunweightedinequality(Concept1),whichroseuntilaround2000,hasfallensincethen,Milanovic(2016,Figure4.1).Atthesametime,beforethepandemic,thefallinweightedbetween-countryinequalityhasbeenaccompaniedbyrisinginequalitywithinmanycountries,withtheneteffectthattheglobaldistributionofincomebetweenallthepeopleintheworldhasbecomemoreequal,seeagainMilanovic(2016,Figure3.1)andDeaton(2013,262).ButtheenrichmentofChinahasdiminishedthesizeofthecontributionthatitshighgrowth(andlargepopulation)hasmadetonarrowingtheglobaldistributionofincomeamongallpersons;ifacountrygrowsfastenoughforlongenough,itwillinevitablybecomerich.

1.Income,incomegrowth,anddeathsfromCOVID-19

IusedataontotaldeathspermillionfromOurWorldinData,asofDecember31st,2020.Dataonrealnationalincomepercapita,expressedin2017international(PPP)dollarsaretakenfromtheIMFWorldEconomicOutlookofOctober2020,fromtheWorldBank’sGlobalEconomicOutlookofJanuary2021andfromitsWorldDevelopmentIndicatorsdatabase.TheIMFdata,whichismymainsource,covers193countries;incomedataaremissingforSyriaandSomalia.

Figure1showsthescatterplotacrosscountriesofthelogarithmofdeathspermillionagainstthelogarithmofincomeperheadin2019;thereare169countrieswithnon-missingvaluesofbothvariables.Theareasofthecirclesareproportionaltopopulation.ThecirclesareshowninblackfortheOECDcountriesandinredforthecountriesnotintheOECD.Thepopulation-weightedregressionlineisshownasthedashedline;itsslopeis0.83(t=4.9).Theunweightedregressionlineissomewhatsteeper,0.99(8.6).

6

ThereisnorelationshipbetweenpercapitaincomeandCOVIDdeathspermillionwithintheOECD,weightedorunweighted,sothepositiverelationshipisdominatedbytherelationship

2020

8

Dec31,

6

millionasof

4

deathsper

2

COVID

0

Logof

-2

OECDinblack

SouthAfrica

BrazilMexico

USA

Ireland

Afghanistan

India

Australia

Japan

Korea

DRC

Nigeria

China

Singapore

Thailand

Vietnam

Burundi

Tanzania

Taiwan

6

8

10

12

logofpercapitaincome(2017ppp$)

Figure1:COVID-19deathspermillionandpercapitaincomein2019:brokenlineisthepopulation-weightedregressionline,areasofcirclesproportionaltopopulation

betweenOECDandnon-OECDcountries,aswellasbytherelationshipwithinthenon-OECDitself.Amongthelatter,muchdependsonIndiaandChina.Ignoringpopulationsize,thecountry-by-countryrelationshipinthenon-OECDisclosetothatforallcountries.Weightedbypopulationsize,therelationalsoexistswithinthenon-OECDifChinaisexcluded;China’slowdeathtollisanoutlier,anditspopulationisthelargestintheworld,soitsinclusionannulstherelationship.

Mymainpurposehereismeasurement,butthepositiverelationshipinFigure1,previouslydocumentedbyGoldbergandReed(2020),raisesimportantissues,ifonlybecauseitcontradictssomanypre-suppositions.Studiesofglobalhealthandglobalincome,eversincePreston’sfamous1975paper,haveuniversallyfoundthathigherincomecountrieshavebetterhealth;theyhavebetterpublicandprivatehealthsystems—bothofwhichcostmoney—andusuallyhavegovernmentsthat

7

aremoreeffectiveatprotectingtheirpopulation’shealth.SuchisthebasisforStiglitz’argumentabove.Moreformally,thereisacomprehensive2019studyofglobalhealthsecurity1byJohnsHopkins,theNuclearThreatInitiativeandtheEconomistIntelligenceUnit.Thatstudypublishedasetofglobalhealthindexesfor195countriesbasedon140questionsthatmeasurecountrycapacityinsixdimensions,(i)preventionoftheemergenceandreleaseofpathogens,(ii)earlydetectionandreportingforpandemicsofpotentialinternationalconcern,(iii)rapidresponseandmitigationofthespreadofapandemic,(iv)sufficiencyandrobustnessofthehealthsystemtotreatthesickandprotecthealthworkers,(v)commitmentstoimprovingnationalcapacity,financing,andadherencetonorms,and(vi)riskenvironmentandvulnerabilitytobiologicalthreats.Thesearepresentedseparately,andalsoaggregatedintoanoverallindex,GHS(2020,6).

Inlinewiththe“healthiswealth”presupposition,theoverallindexcorrelates0.65withthelogarithmofpurchasingpowerparitypercapitaincomeover166countries;forthefirstfourindexeslistedabovethecorrelationsare0.62,0.48,0.48,and0.64.Theindexesarealsopositivelycorrelatedwithdeathspermillion,0.47fortheoverallindex,and0.47,0.41,0.31,and0.48ofthefirstfour.Inspiteofbeingdesignedtobehelpfulfor“highconsequencepandemicthreats,suchasafast-spreadingrespiratorydiseaseagentthatcouldhaveageographicscope,severity,orsocietalimpactandcouldoverwhelmnationalorinternationalcapacitytomanageit,”(p7),andinspiteoftheevidentcareandthoroughnessofthereport,countriesthatdidbetterontheindexesexperiencedmoredeathsthanthosethatdidworse.Itseemsthatevendistinguishedandcarefulexpertscouldnotpredicttheinternationalpatternsofdeathsinthepandemic,atleastthroughtheendof2020,norisitclearthatanycountrycouldhavebeenadequatelypreparedforwhathappened.Ascountrieslearnlessonsfromthepandemic,andtrytopreparebetterforthefuture,theywillpresumablyhavetotakemeasuresatleastsomeofwhicharedifferentfromthoseproposedintheGHSreport.

IamgratefultoBrankoMilanovichbringingthisreporttomyattention.

8

TheFigureshowsthesmallnumberofdeathsinChina,aswellasinotherEastAsiancountries,whetherintheOECDornot.TheverylownumbersofdeathsinBurundiandTanzaniaaremostlikelyduetoundercounts;TanzaniastoppedreportingcasesinMayclaimingthatithadconqueredthevirusandbothitandBurundihaverejectedoffersofvaccine,WallStreetJournal(2021).

Misreportingaside,thelownumberofdeathsinpoorcountrieshasbeenlinkedbyGoldbergandReed(2020)toobesity,tothefractionofthepopulationover70,andtothedensityofpopulationinthelargesturbancenter.HeuvelineandTzen(2021)provideage-adjustedmortalityratesforeachcountrybyusingcountryage-structurestopredictwhatdeathswouldhavebeeniftheage-specificCovid-19deathrateshadbeenthoseoftheUnitedStates;theratioofpredicteddeathstoactualdeathsisthenusedtoadjusteachcountry’scrudemortalityrate.ThisprocedurescalesupmortalityratesforcountriesthatareyoungerthantheUS(Peruhasthelargestageandsexadjustedmortalityrate)andscalesdownmortalityratesforcountrieslikeItalyandSpain(whichhadthehighestunadjustedrate)thatareolderthantheUS.IfFigure1isredrawnusingtheadjustedrates,thepositivesloperemainsthoughthe(unweighted)slopeisreducedfrom0.99to0.47(t=4.7).Inpoorcountries,manychildrensufferfromill-health—diarrhealdisease,respiratoryinfections,undernutrition—thatcouldraisetheriskofdeathconditionaloninfection,sothattheymaynotgetasmuchbenefitofayoungagestructureaswouldtheUS.

Poorcountriesarealsowarmercountries,wheremuchactivitytakesplaceoutside,andtherearerelativelyfewlargedensecitieswithelevatorsandmasstransit.ItisalsopossiblethatAfrica’slong-standingexperiencewithinfectiousepidemicsstooditingoodsteadduringthisone.Countrieswithmoredevelopedeconomieshavehigherdegreesofintermediationandahigherfractionofservices,bothofwhichmakeinfectioneasier.Butsuchexpoststoriesareworthlittlewithoutmore

9

seriousanalysis,andagain,theseriousandthoroughanalysisintheGHSindexreportpredictedjusttheopposite.

Perhapsthemostsurprisingresultinthefigureistherelativelyhighnumberofdeathsamongthehighest-incomecountries.Therehasbeenmuch(well-deserved)condemnationoftheTrumpadministration’shandlingoftheepidemic,butdeathspermillionintheUSarenoworsethaninseveralotherrichcountriesandnotmuchworsethanpredictedfromtheglobalpattern.Statementsaboutthedisproportionofdeathsandpopulation(theUShasonlyfourpercentoftheworld’spopulationbuttwentypercentofthedeaths,orthattheUShasmorethan30timesasmanydeathsasPakistan)areconsequencesofthepatterninthefigure,includingthesmallnumberofdeathsinChina,andtelluslittleabouthowwellorbadlythepandemicwashandledintheUSorelsewhere.DeathsintheUSareabovetheregressionlineoflogarithmofdeathspermilliononthelogarithmofincomepercapita,butbythatmeasuretheUSdidaboutaswellasSweden,andbetterthanHungary,Spain,Poland,Portugal,Italy,theUnitedKingdom,andFrance.(Belgiumistheworstofall,likelybecauseofitsmorecomprehensivemeasureofCOVID-19deaths.)Troesken(2015)arguesthattheUShaslongbeenpronetoinfectiousdisease;in1900,afterasafeandeffectivevaccinehadbeenavailableformorethanacentury,andinspiteofalreadybeingtheworld’srichestcountry,theUSdidworsethanotherrichcountriesinpreventingsmallpoxdeaths.Troeskenargues(p176)thatthiswas“notdespitebeingrichandfree,butpreciselybecauseitwasrichandfree.”

Formypurposeshere,thereisnoneedtotrytoestablishcauses.Largescalemisreportingisanothermatter,andagainInotethat,evenwithperfectreporting,thedynamicsofthepandemicwillalmostcertainlychangepatternsovertime.

ThesecondpartofthestoryistherelationshipbetweenpandemicdeathsandgrowthinpercapitaGDPin2020.Here,Irelyonforecastdata,twosetsofwhichareavailable,onefromtheIMF

10

inOctober2020,andonefromtheWorldBankinearly2021.IusetheearlierIMFnumbershere;theWorldBanknumbersareclose,andthecross-countrycorrelationsbetweenthetwosetsofestimatesis0.945.AnobviousconcernisthattheBankandtheFundusedthedeathcountstoforecastthechangeinincome.ButforecastsconstructedinOctoberandinJanuaryundoubtedlyhavealargecomponentofactual,asopposedtoforecast,data.Weshouldalsoworryifdataonthepandemicwerenotincorporatedintotheforecasts.Again,themostseriousconcernisaboutmisreporting,andaboutbadGDPforecaststhatarebasedonbaddataondeaths.

.1

income

0

changeinlog

-.1

predicted

-.2

-.3

China

Vietnam

Egypt

Taiwan

Tanzania USA

Thailand

India

Fiji

Venezuela

-2

0

2

4

6

8

logarithmofdeathspermillion

Figure2:Predictedgrowthofpercapitaincome2019-20anddeathspermillion:populationweightedregressionshownasbrokenline,areasofcirclesproportionaltopopulation

Figure2showstheIMF’spredictedgrowthratesfrom2019to2020plottedagainstdeathspermillion.China,withfewdeaths,showspositivegrowth;theUS,withmanydeaths,showsnegativepredictedgrowth.Therearemanycasesthatarenotontheline,atleastsomeofwhichhavenothingtodowithCOVID,but,aswouldbeexpected,therearesimilarlyslopedscattersofcountriesoverallandbothinandoutoftheOECD.The(weighted)regression—shownasthe

11

dottedline—hasaslopeof-0.015(t=10.2),sothatpredictedgrowthdecreasesbyoneandahalfpercentagepointsforeveryunitincreaseinthelogarithmofdeathspermillion;theslopeofthe

weightedregressionis-0.007(t=4.0).(IhaveexcludedLibyaandGuyanafromthefigureandthecalculations;neitherisexceptionalindeaths,butLibyahasapredictedlogchangeofpercapita

incomeof-1.1,andGuyanaapredictedlogchangeof+0.23,numbersthatarenotonly(absolutely)verylargebutpresumablyunrelatedtothepandemic.Ihavealsorepeatedthesecalculationsusing,notgrowthforecastfor2020inOctober2020,buttherevisiontothe2019to2020growthforecastbetweenthe2019and2020editionsoftheWorldEconomicOutlook,theideabeingtoisolatethereductioningrowthassociatedwiththepandemic.Thecorrespondingfigureandregressionresultsaresimilartotheoriginals,albeitwithlowerlevelsofpredictedgrowthsothat,forexample,allrevisionstogrowtharenegative.)

ItisperhapsnotsurprisingthatdeathsfromCOVID-19shouldbringeconomicdestruction,northattherelationshipshouldbetighterthantherelationshipbetweendeathsandincomein2019.But,onceagain,thatthereshouldbethisrelationshipwasnotobviousbeforethepandemic.Indeed,intheearlydays,therewasmuchdiscussionofthevalueoflifeandaboutasupposedtrade-offbetweendeathsandincome,thatlockdownswouldsavelivesbutdestroyeconomies.AspreviouslynotedbyWolf(2020)wholookedattheadvancedcountriesplusIndiaandChina,thereisnoevidenceinthesecross-countrydatafortheexistenceofanysuchtrade-off.Instead,theroutetogrowthliesthroughstoppingdeaths.Itisnotamatterofyourmoneyoryourlife,butyourmoneyandyourlife.Thisshouldnotbetakenasanargumentinfavorofgovernment-orderedlockdownsbecausevoluntarysocialdistancinginthefaceofinfectionanddeathhasalsobeenimportant,andperhapsmoreso,IMF(2020,Chapter2)andGoolsbeeandSyverson(2020).

Figure3closesthecircle.ItplotstheincomechangesfromFigure2againstthe2019levelsofincomeinFigure1;itshowsthatrichercountrieshadslower(ormorenegative)growthin2020

12

thandidpoorercountries.TheslopeoftheunweightedregressionlineintheFigureis-0.010

(t=3.3),sothateveryunitincreaseinlogincomeshavesonepercentagepointoffofthepredictedgrowthrate.Giventhedisparateexperiencesofthetwogiants,IndiaandChina,theweighted

regressionhasaninsignificantsmallslopeof-0.003(t=0.8).Chinaisgrowingbecause,inspiteofitsrelativelyhighincome,ithasseenfewdeathswhileIndia,withmoredeathspermillionthanothercountriesatitsincomelevel,showsa10.2percentdeclineinincome.Eachcountryisanoutlier,butinoppositedirections.WhenIrunthesame(unweighted)regressionusingthe2019predictedgrowthratesascalculatedbytheIMFbeforethepandemic,theslopeisstillnegative,butsmallandinsignificant-0.001(t=1.4).

.1

income

0

changeinlog

-.1

predicted

-.2

-.3

China

USA

India

7

8

9

10

2017

11

logpercapitaincomein2019,PPP

$

Figure3:Growthofpercapitaincome,2019-20,andpercapitaincomein2019:lineisunweightedregressionline,areasofcirclesproportionaltopopulation

Ignoringpopulationsize,thenegativerelationshipbetweengrowthin2020andincomein2019existsfortheworldasawhole,andwithinthenon-OECDcountries.WithintheOECD,the

13

betteroffcountriesgrewfasterin2020,buttheregressioncoefficientisnotsignificantlydifferentfromzero,ascanbeseeninthefigure.

Thathigherincomecountriesexperiencethelargestdecreasesinincomeonaveragedoesnot,inandofitself,implythattherewasadecreaseininequalityinpercapitaincomesbetweencountries;therelationshipinFigure3isnotexact,anddeviationsfromthelinealsoaffectinequality.

Figure4show