IMF-是什么導(dǎo)致貝弗里奇曲線在疫情期間在美國變高？（英）

WhatCausedtheBeveridgeCurveToShiftHigherintheUnitedStatesDuringthe

Pandemic?

GeneKindberg-HanlonandMichaelGirard

WP/24/8

IMFWorkingPapersdescriberesearchin

progressbytheauthor(s)andarepublishedto

elicitcommentsandtoencouragedebate.

TheviewsexpressedinIMFWorkingPapersare

thoseoftheauthor(s)anddonotnecessarily

representtheviewsoftheIMF,itsExecutiveBoard,

orIMFmanagement.

2024

JAN

?2024InternationalMonetaryFund

WP/24/8

IMFWorkingPaper

ResearchDepartment

WhatCausedtheBeveridgeCurvetoShiftHigherintheUnitedStatesDuringthePandemic?PreparedbyGeneKindberg-HanlonandMichaelGirard

AuthorizedfordistributionbyRafaelPortillo

January2024

IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicit

commentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseofthe

author(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.

ABSTRACT:TheBeveridgecurveshiftedsubstantiallyhigherintheUnitedStatesfollowingthestartofthe

COVIDpandemic.In2022,vacanciesreachedrecordhighsacrossallsectorswhileunemploymentfelltopre-pandemiclows.Atthesametime,thepandemichasresultedinseverelaborshortages,andweestimatethatthelaborforcewasapproximately2millionbelowtrendatthestartof2023.Weexploitstate-leveldataintheUnitedStatestofindthatlowerimmigration,higherexcessmortalityduetoCOVID,andfallingolder-worker

laborforceparticipationwereassociatedwithlargerupwardshiftsintheBeveridgecurve.Wealsofindthat

statesthathadalargeremploymentconcentrationincontact-intensivesectorshadlargerupwardshiftsintheirBeveridgecurve.WhiletheeffectofsectoralreallocationandrehiringhasbeenshownintheoreticalmodelstolifttheBeveridgecurve,weshowthatworkershortagesalsoresultinanupwardshiftintheBeveridgecurveiftheyincreasethemarginalproductoflabor.Thisresultholdsinasearchandmatchingmodelwithon-the-jobsearch,butdoesnotholdwithouton-the-jobsearch.

JELClassificationNumbers:

E24,J63,J64

Keywords:

BeveridgeCurve;COVID

Author’sE-MailAddress:

gkindberg-hanlon@;mgirard@

IMFWORKINGPAPERSWhatCausedtheBeveridgeCurvetoShiftHigherintheUnitedStatesduringthePandemic?

INTERNATIONALMONETARYFUND

WORKINGPAPERS

WhatcausedtheBeveridgecurvetoshifthigherintheUnitedStatesduringthepandemic?

PreparedbyGeneKindberg-HanlonandMichaelGirard

2

Contents

1Introduction

3

2Literature

5

3AssessingcovariatesofBeveridgecurveshiftsusingU.S.state-leveldata

7

4Factorsdrivingthelaborshortage

12

5AsearchandmatchingmodelwithOJS

16

6Conclusionandoutlook

25

AModel

27

BEstimatingthematchingfunction

38

3

1Introduction

TheBeveridgecurveintheUnitedStateshasshiftedsubstantiallyhighersincethestartoftheCOVID-19pandemic(Figure

1

).TheinverserelationshipbetweenunemploymentandvacanciesintheBeveridgecurvereflectscyclicalfactorsinfluencingthedemandforlaborandpoolofunemployedworkers,butshiftsintherelationshipcanfollowarangeastructuralchanges.Bytheendof2022,theunemploymentratehadreturnedtoitspre-pandemictroughofabout31/2%suggestingthattheeconomyhadlargelycyclicallyrecoveredfromthepandemicshock,whilethevacancyrateincreasedfrom41/2%priortothepandemictoabout61/2%.Vacancieshaverisenacrossabroadrangeofsectorsandincreaseshavenotbeenlimitedtothosethatweremostaffectedbythepandemic(Figure

1

).TheshiftintheBeveridgecurvehasimportantmacroeconomicimplications.Forexample,severalstudieshavefoundanon-linearrelationshipbetweentheratioofvacanciestounemployment,ameasureoflabormarkettightness,andthe

rateofinflation(Balletal.,

2022;

BenignoandEggertsson,

2023)

.TheshiftintheBeveridgecurvesincethepandemicimpliesatighterlabormarketthantherateofunemploymentwould

signalalone.

Figure1:Beveridgecurveshiftsandsectoraltrends

Beveridgecurvepre-andpost-COVIDSectoralvacancyandemploymentdevelopments

Vacancyrate,percentChangeinemployment

8.0

7.0

6.0

5.0

4.0

3.0

2.0

1.0

0.0

2015Jan-March2020

2020April-Dec2022

0246810121416

Unemploymentrate,percent

20%

15%

10%

5%

0%

-5%

WholeRetail

sale&

Trade

Trans.&Warehousing

Information

OtherServices

andGov.

·Leisure&Mining,Cons.,Manu.Hospitality

ProfsrcialEdu.&Health

20%30%40%50%60%70%

Changeinvacancies

Note:VacancyratereferstothetotalJOLTs-measurednumberofvacanciesasapercentageofthelaborforce.Sectoralchangeinvacanciesemploymentshowsthechangebetweentheannualaveragein2019and2022.Payrollconceptofemploymentisusedtomeasureemploymentbysector

Usingstate-leveldata,wefindsupportfortheexistingviewthatlayoffsandreallocation

drivenbythepandemichasresultedinshiftsintheBeveridgecurve,asworkerswerere-hired

4

intotheirpreviousrolesornewsectorsandlocations(Blanchardetal.,

2022;

FiguraandWaller,

2022

).Thisfactorwaslikelytobemostimportantearlyintherecoveryfromthepandemic.Wealsofindthatalargelaborshortagegeneratedbythepandemichasdrivenintensecompetitionforworkers,leadingtohighhiringintensity(andthusvacancies),increasedjob-to-jobflows,andlowerlayoffs(Figure

2

).Weestimatethatthelaborforcewasapproximately2millionbelowtrendatthestartof2023duetoCOVID-relatedmortality,lowerolder-workerparticipationrates,andlowerimmigration.Thisshortageofworkers,alongsidethelargeinitiallayoffsandreallocationeffectsdrivenbythepandemic,hasalsobeenalargecontributortotheobservedupwardshiftintheBeveridgecurve.During2021-22,theshortageofworkersrelativetotrendhasreachedashighas4million,evenasGDPreturnedtotrendinlate2021.Theeffectofincreasedexogenousseparations,suchasreallocationbetweensectorsasconsumerspendingwasredirectedbythepandemic,isalreadywell-knowntobeassociatedwithshiftsintheBeveridgecurve(

Elsbyetal.,

2015).TheeffectofworkershortagesontheBeveridgecurvehasyettobe

studiedtoourknowledge.

Wefindthatinasearchandmatchingmodel,ashockthatbooststhemarginalproductoflaborwillresultinaverticalshiftintheBeveridgecurve.Inisolation,anegativeshocktolaborsupplywillinduceariseinthemarginalproductoflaborandexcessdemandintheeconomy(apositiveoutputgap)(

Gal′?etal.,

2012

).Higherreturnsfromnewhiresincentivizesfirmstobearhighercostsofadvertisingmorevacanciesevenwhenthepossibilityofsecuringanewfirm-workermatchislow.Inaddition,whileaworkershortagewouldencourageworkerhoarding,whichshouldshifttheBeveridgecurvedownbyreducingseparationrates,thisisnotthecaseinamodelwhichincorporateson-the-jobsearch(OJShenceforth).AworkershortagewillencouragealargerproportionofworkerstoengageinOJSinthehopeofgainingbetteremploymentterms.Increasedcompetitionforfirm-workermatchesfromOJSworkerscrowdsoutunemployedworkersfromgainingemploymentforagivennumberofvacancies.Vacanciesmustbehigherinequilibriumtomaintainaparticularunemploymentrate.Inacalibratedmodel,wefindthatundercertainassumptionsabouttheriseinthemarginalproductoflaborduringthepandemic,thismechanismcouldexplainapositive0.6percentagepointshiftinthe

Beveridgecurveatlowlevelsofunemployment.

Thepaperproceedsasfollows:wefirstoutlinetheliteraturethathasexaminedthepost-COVIDchangesintheBeveridgecurveandtheliteratureontheimportanceofOJSintheoretical

frameworks.Wethenexaminethecovariatesofstate-levelheterogeneityinthescaleofthepost-

5

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

Figure2:Job-to-jobswitcheshavebeenhigh,layoffslow

Labormarketflowsdevelopments

Percentoflaborforce

9.08.0

7.0

6.0

Hirerate

Quitrate

5.0

4.0

3.0

6.05.0

4.0

2.0

1.0

0.0

3.02.01.00.0

Percentofemployment

Layoffrate

Note:Hires,quitsandlayoffsobtainedfromtheJOLTSsurvey.LaborforceandemploymentdataobtainedfromtheBLShouseholdsurvey.

COVIDshiftintheBeveridgecurve.Finally,weoutlinetheCOVID-drivenreductioninthelaborforcefortheU.S.asawhole,andhowthiscancauseanupwardshiftintheBeveridgecurveinacalibratedsearchandmatchingmodelaugmentedwithOJS,andalsoshowhow

COVID-drivenreallocationeffectscanshifttheBeveridgecurvehigherinthesamemodel.

2Literature

TheliteratureexaminingchangesintheBeveridgecurvepost-COVIDhasfocusedonboththeslopeofthecurveandshiftsinthecurve.

FiguraandWaller

(2022)arguethattheapparent

upwardverticalshiftoftheBeveridgecurveisanillusion.Specifically,theysuggestthattheBeveridgecurveissteepatlowlevelsofunemploymentandhasshiftedinwardsasseparationratesinitiallyroseandthendeclinedpostCOVID,reflectinganeedforsectorallaborreallo-cation.Intheirframework,inwardshiftsofasteeplyslopedcurvegivetheillusionofverticalshifts.Notably,theydonotaccountforOJSintheirtheoreticalframeworkorparameteresti-matesoftheBeveridgecurve.

Blanchardetal.

(2022)suggestaverticalshifthasoccurredin

theBeveridgecurveandthecurveremainsrelativelyflat,implyingahightradeoffintermsofemploymentasvacanciesfall.Theirworkisbasedonparameterestimatesthatincludedataon

job-to-jobhiresreflectingOJSandalsoreinforcedbyhistoricalexperience.

Bietal.

(

2022

)also

6

findhistoricalevidencethattheBeveridgecurveisrarelysteeplysloped.

TheoreticalandempiricalworkhasexaminedthedriversofoutwardsshiftsintheBeveridgecurveinearlierepisodes.TheBeveridgecurveshiftedsubstantiallytotherightinthe1970s,possiblydrivenbymismatchesbetweengrowthindifferentregionsoftheU.S.andlimitedlabormobility,creatingamismatchofdemandandsupplyoflaboracrossU.S.states(

Abraham,

1987

).Thereareparallelsbetweenthisanalysisofstructuralchangesinthe1970sandouranalysisofpost-COVIDchanges,inthatwenotethatashortageoflaborislikelytobeakeydriver.TheU.S.Beveridgecurvealsoshiftedoutwardsfollowingthe2008GlobalFinancialCrisis.Thiswasattributedtoadeteriorationinmatchingefficiencyandsearchintensity(

Davisetal.,

2013;

Elsbyetal.,

2010)

.

ThemostcloselyrelatedworktoourpaperspecificallyincorporatesOJSintosearchandmatchingmodelstoexplainpost-COVIDdevelopments.

CheremukhinandRestrepo-Echavarria

(2022)arguethatinadual-labormarketsetupwheretheunemployedandemployedsearch

forworkinseparatemarketstheBeveridgecurveforunemployedworkershasstayedbroadlyunchangedpost-COVID,whilethemarketforemployedworkershassteepened.Ourframeworkisdifferent,inthatweassumethatworkersandtheunemployedarecompetinginthesamemarketforjobs,andthereforethereisacrowdingouteffectforunemployedworkerswhenmoreOJStakesplace.ThisfeatureresultsinverticalshiftsintheBeveridgecurve.Inastylizedframework,

Barlevyetal.

(2023)alsonotetheimportanceofOJSindrivingshiftsin

theBeveridgecurvefollowingthepandemic.Incontrasttoourwork,theyattributetheriseinOJStoworkersreevaluatingcareerchoicesandsearchingforimprovedwork-lifebalancesafterthepandemic.OurmechanismforincreasingOJSisdifferent,withworkersinsteadincentivized

tosearchbythepayrisesavailableforjob-moversduringalaborforceshortfall.

WenotethatthesearchandmatchingliteraturehasfoundthatOJSincreasestheabilityofthisclassofmodelstomatchempiricalmomentsoflabormarketdata,butfrequentlythesemethodsarenotemployedduetotheunobservablenatureofworkersengaginginOJS.Ap-proximatelyone-thirdofhiringinvolvesjob-to-jobtransitionsinestimatesusingCPSmicrodata

(Elsbyetal.,

2015;

FallickandFleischman,

2004),similartothatsuggestedbytheJobOpenings

andLaborTurnoverSurvey(JOLTS)quitsandhiringdata.

FujitaandRamey

(2012)findthat

modelswithOJSbettermatchU.S.labormarketdynamicsrelativetosimplermodelswithonlytransitionsfromunemploymenttoemployment.Ourmodelcontainsamechanismfirstproposed

in

Pissarides

(1994),whichshowsthatinamodelwithOJSandtenure-relatedhumancapital

7

building,ariseinproductivityofworkerscanleadtoaverticalshiftintheBeveridgecurve.ItisthisfeatureofOJSmodelswhichwelinktothepost-COVIDlaborshortagetoexplainpart

oftheupwardshiftintheBeveridgecurve.

3AssessingcovariatesofBeveridgecurveshiftsusingU.S.state-

leveldata

ToassesstheempiricalsupportforourviewthatlaborshortagesandreallocationhavebeenimportantinexplainingstructuralshiftsintheBeveridgecurve,weexploitcrossstatevariationinshiftsintheBeveridgecurveintheUnitedStates.WefirstestimateareducedformBeveridgecurveseparatelyineachofthe50statesandtheDistrictofColumbiausingthefollowing

specification:

vi,t=c+βi,1D2020+βi,2trendt+βi,3ui,t+βi,4u,t+?i,t(1)

Wheretisthe(monthly)timeperiod,andiisthestate.visthevacancyrate,definedastheratioofJOLTS-measuredvacanciestothelaborforce,anduistheunemploymentrate.Thesquaredunemploymentratetermallowsfornon-linearityintherelationship,andfollowsthespecificationusedby

Abraham

(1987)

.1

Thespecificationisestimatedwithdatafrom2001toend-2022,theavailablerangeoftheJOLTSdataforvacancies.ThespecificationincludesatimetrendtoaccountfortherisingdriftofthecurveintheU.S.since2000.Thedummyvariable(DMarch2020)identifiestheshiftintheinterceptsincetheoutbreakoftheCOVIDpandemicandtakesavalueof1afterFebruary2020.OurestimatesshowaverticalshiftineachstateandD.C.post-COVID,butwithsubstantialvariation,rangingfroma1.1toa3.4percentagepointincrease

.2

ToidentifycorrelatesoftheU.S.Beveridgecurveshift,weregresstheshiftintheintercept

estimatedaboveineachstate(βi,1)onarangeofvariablesinthefollowingspecification:

1Wenotethatstate-levelunemploymentdatapublishedbytheBLSispartlymodel-basedratherthanrelyingentirelyontheCPSsurveydata(asthenationalunemploymentestimatedoes).

2Thereducedformspecificationisliabletothecritiquethatitmayincorrectlyidentifyverticalshiftsinthecurvethatactuallycapturehorizontalshiftsofasteeplyslopedcurve(

FiguraandWaller,

2022)

.However,wenotethatoncegrosshiresincludingjob-to-jobswitchesareincludedinestimatesofthematchingfunction,inthesamespecificationusedby

FiguraandWaller

(2022),thecurvelookssimilartothereducedformspecificationwith

aggregateU.S.data.

8

D=c+βX+?(2)

WhereDisavectoroftheestimatedBeveridgecurveinterceptshiftineachstate,Xcontainspotentialcovariates,oftheshiftandβisavectorofcoefficients.Theseincludevariablesrelatedtothesectoraldisruptiondrivenbythepandemic,aswellasvariablesrelatedtotheimpactof

thepandemiconthelaborforce.Specifically,weinclude:

1.I-94temporaryworkervisaadmissionsineachstate,shortfallrelativetoacontinuationofthe2019fiscalyearratefrom2020until2021.ThedataissourcedfromtheDepartmentofHomelandSecurityandexpressedasaproportionofthelaborforce.Thisdatashowsentrantstoeachstatefromholdersofwork-relatedvisasandtheirfamilies,andprovidesaproxyforthedeclineinwork-relatedimmigrationtoeachstate.Asthedatareflectin-stancesofenteringthecountryonawork-relatedvisa,theycanrecordthesameindividualmultipletimesperyear,andalsocapturedecliningtravel.ThevariableneverthelesscanreflecttherelativeimportanceofimmigrantworkersineachstateandtheeffectofCOVID

incurtailingimmigrantmovementtothestate.

2.Shareofworkersincontact-intensivesectorsin2019whichweremostvulnerabletodis-ruptioncausedbytheCOVID-19pandemic,theleisure&hospitalityandretailtrade

sectors.

3.Censusdataonnetdomesticimmigrationbetweenstatestoaccountformassmovementthatoccurredintheimmediateaftermathofthepandemic.

3

Asinthecaseofvisaissuance,

thisiscalculatedasthecumulativeshortfallin2020-21relativetothe2019rate.

4.Censusnetinternationalmigration,toaccountforabroaderrangeofinternationalmigra-tionthancapturedbypurelyworker-focusedI-94workeradmissionsandamoreaccuratereflectionofthosepermanentlymovingtoastate.Incontrast,I-94admittancecancap-turetemporaryvisitsorreturnsfromtravel(butismorespecificaboutentrantsforwork

ratherthanotherreasons,suchasforstudyandfamilyreasons).

5.Theover-65participationratechangebetween2019-21,tocaptureshortfallsinworkers

drivenbythatsharpdeclineinparticipationatthisagegroup.Thisfigureisderivedfrom

3Seeforexample:

/library/stories/2022/05/population-shifts-in-cities-and-towns-one

-year-into-pandemic.html

9

theCPSmicrodataandisnotavailableasapublishedstatistic.

6.Excessdeathsbystateforindividualsaged15-64between2020-2021,calculatedbytheCDC’sNationalCenterforHealthStatistics

.4

7.Googlemobilitydataforretailandrecreationlocations.Deviationbystatefrom2019levelsonaveragein2021.Thisindicatormayreflectthedegreeofchangingconsumer

preferencesinthestateandtheextentofbehavioralchangedrivenbythepandemic.

8.Statedomesticproductgrowth,2019-2021.Statedomesticproductgrowthreflectstherecoveryindemandacrossstatesandcontrolsfortheextenttowhichothervariablesunderconsiderationaffectdemand(i.ehighernetmigrationincreasethesupplyoflaboranddemandforgoodsandservices,butnotnecessarilytoanequivalentextentinthe

timeframeunderconsideration).

Additionalfactorsmaybeimportantindrivingstate-variationintheBeveridgecurveshift.Theseincludepandemic-inducedchangesinunemploymentinsurance,orvariationinthedesiretopursuechangestowork-lifebalance.However,wearenotawareofdatasourcescovering

cross-statevariationinthesefactors.

EndogeneitySeveralofourchosencovariatescouldbeendogenouswithrespecttotheshiftintheBeveridgecurveorreflecttheeffectsofomittedvariables.Forexample,domesticandinternationalmigrationratesineachstatecouldpartlydependontheexogenousfallinmigrationdrivenbygovernment-imposedrestrictionsandvoluntarysocialdistancing,butalsoreflectlabormarketconditionsinindividualstates.Atighterlabormarketmayinducehighermigration.Theparticipationrateofover-65salsoislikelytoreflectlabormarkettightnessaswellasexogenouseffectsrelatedtothepandemic.Giventhesmallsizeofthedataset,itssingleperiodnature,andcollinearityissues,wedonotattempttoovercometheseissues.However,inourtheoreticaldiscussionbelow,weattempttodemonstratethattheeffectsthatvariablesassociatedwithlaborshortagesandreallocationshaveareconsistentwiththedirectionofcorrelationinthese

regressions.

RegressionresultsInmanyspecifications,theI94admittanceshortfallisstatisticallysignif-

icantandnegativelysigned,suchthatashortfallisassociatedwithanincreaseintheBeveridge

4Availableat:

/nchs/nvss/vsrr/covid19/excessdeaths.htm#dashboard

10

Table1:CovariatesoftheBeveridgeCurveShiftinUSStates

Dependentvariable:

StateVerticalBeveridgeCurveShift

(1)

(2)(3)(4)

(5)

I94workeradmittanceshortfall(2020-21)

?0.143***(0.038)

?0.076*

(0.042)

?0.075*(0.042)

?0.151***(0.042)

?0.080*(0.044)

Sharevulnerablesectors(2019)

0.030*

(0.017)

0.030**

(0.015)

0.028*

(0.015)

0.036**

(0.017)

0.033**

(0.016)

Changein65+participation(2019-21)

?0.090**

(0.036)

?0.078**(0.034)

?0.081**(0.034)

?0.079**(0.036)

?0.094**(0.036)

Excessdeaths15-64(2020-2022)

0.128

(0.089)

0.172**

(0.084)

0.152*

(0.086)

0.161*

(0.087)

Googlemobility(retail&rec.,2021)

?0.022***(0.008)

?0.021**

(0.008)

?0.025***(0.008)

Statedomesticproductgrowth(2019-21)

?0.026

(0.014)

?0.035

(0.017)

Domesticnetmigrationchange(2020-21)

?0.103(0.196)

?0.213

(0.189)

Internationalnetmigrationchange(2020-21)

?0.103(0.196)

?0.213

(0.189)

Constant

0.754

(0.464)

0.7210.821*0.750(0.432)(0.439)(0.478)

0.724

(0.452)

Observations

51

515151

51

AdjustedR2

0.262

0.3600.3650.225

0.361

Note:*p<0.1;**p<0.05;***p<0.01

11

Figure3:CovariatesofBeveridgecurveshiftacrossstates

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Beveridgecurveshift

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

-8-6-4-2024

I-94visaadmittanceshortfall2020-21vs2019rate,%labor

force

Beveridgecurveshift

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

-8-6-4-2024

Changein65+participationrate

Beveridgecurveshift

15202530354045

Shareofemploymentinvulnerablesectors,2019

Beveridgecurveshift

-50-40-30-20-1001020

FallinGooglemobility,%

Note:PlotsshowtheestimatedBeveridgecurveshiftineachstateandD.C.plottedagainstselectedcovariatesincludedintheregressioninTable1

curveintercept(Table

1

).Thechangeinparticipationratesofthose65+isnegativelycorrelatedineachspecificationandstatisticallysignificantinmost,suchthatstateswithfewerolderwork-erssawlargerrisesintheBeveridgecurve.Thisisdespitetheendogeneityofthevariable(tightlabormarketsdrawingmoreindividualsintothelaborforce)workingintheoppositedirectiontothecoefficient.Ineveryspecification,ahighershareofworkersinCOVID-19vulnerablesectorspriortothepandemicwasassociatedwithalargerriseintheBeveridgecurve.LowerGooglemobilityin2021wasalsoassociatedwithlargerupwardshiftsinthecurve.Thisprovidessup-portforthe‘reallocation’effecttheory,whererisingworkerseparationstonewindustriescaused

bythepandemicthroughchangingconsumerpreferencesandcontactavoidancedrivetheBev-

12

eridgecurvehigher.Finally,excessdeathsofthose16-64(whereparticipationratesarehigh)ispositivelycorrelatedwithlargerpositiveBeveridgecurveshifts,aspredictedbyourhypothesisthatlaborshortageshavecontributedtotheBeveridgecurveshift.WefindlessevidencethatbroadermeasuresofinternationalordomesticmigrationaredriversoftheBeveridgecurveshift,possiblybecausetheycoverlargenon-workrelatedmigrantflows.Incontrast,theI94admit-tanceismorelikelytoreflectwork-relatedvisaflows.Weplotsomesimplecorrelationsbetween

theBeveridgecurveshiftandseveralofthesecovariatesinFigure

3.

4Factorsdrivingthelaborshortage

Ourreduced-formanalysissuggeststhatworkershortagescanresultinverticalshiftsintheBeveridgecurve.Inthissection,weattempttoquantifythescaleofthelaborforceshortfallforthecountryasawholeanddecomposeitintostructuraldriversrelatedtothepandemicthat

areunlikelytoreflectthecyclicalpositionoftheeconomy.

RelativetoprojectionsbytheCongressionalBudgetOffice(CBO)inJanuary2020,thereareapproximately1.8million(1.1%)fewerindividualsinthelaborforceatthestartof2023(Figure

4

).ThisestimateisbelowtheFederalReserve(

BoardofGovernorsoftheFederalReserve

System,

2023)estimateofa3.6millionshortfallattheendof2022,butaccountsformoreup

-to-datepopulationadjustmentsfromtheCensusandmorerecentparticipationdevelopments.Itisarguablyalowerboundestimate,sincesomepopulationchangesinthelatestrevisionwereduetochangestopre-COVIDpopulationstatistics.

5

Theshortfallisunlikelytoreflectcyclicalfactors;GDPandconsumptionhavebroadlyreturnedtotrend,whilethelabor