美聯(lián)儲-金融壓力下的公司債券發(fā)行：全球視角

BoardofGovernorsoftheFederalReserveSystem

InternationalFinanceDiscussionPapers

ISSN1073-2500(Print)

ISSN2767-4509(Online)

Number1390

May2024

CorporateBondIssuanceOverFinancialStressEpisodes:AGlobalPerspective

ValentinaBruno,MicheleDathan,YuriyKitsul

Pleasecitethispaperas:

Bruno,Valentina,MicheleDathan,andYuriyKitsul(2024).“CorporateBondIssuanceOverFinancialStressEpisodes:AGlobalPerspective,”InternationalFinanceDiscus-sionPapers1390.Washington:BoardofGovernorsoftheFederalReserveSystem,

/10.17016/IFDP.2024.1390

.

NOTE:InternationalFinanceDiscussionPapers(IFDPs)arepreliminarymaterialscirculatedtostimu-latediscussionandcriticalcomment.TheanalysisandconclusionssetfortharethoseoftheauthorsanddonotindicateconcurrencebyothermembersoftheresearchstafortheBoardofGovernors.ReferencesinpublicationstotheInternationalFinanceDiscussionPapersSeries(otherthanacknowledgement)shouldbeclearedwiththeauthor(s)toprotectthetentativecharacterofthesepapers.RecentIFDPsareavailableontheWebat/pubs/ifdp/.ThispapercanbedownloadedwithoutchargefromtheSocialScienceResearchNetworkelectroniclibraryat.

1

CorporateBondIssuanceOverFinancialStress

Episodes:AGlobalPerspective

ValentinaBruno

MicheleDathan

April18,2024

YuriyKitsul*

Abstract

Weuseamergedglobaldatasetofsecurity-levelcorporatebondissuanceandfirm-levelfinancialstatementdatatoshowthat,incontrasttoearlierperiodsoffinancialstress,duringtheCOVIDpandemicnonfinancialfirmsaroundtheworldweremorelikelytoissuebonds,drivenbyaboominlocal-currency-denominatedissuance.Weobserveadistinctcross-regionaldifferenceinthecharacteristicsofissuingfirms,findingthatinadvancedeconomiesissuanceduringCOVIDwasdrivenbylessriskyfirms,aspredictedbyexistingtheories;inemergingmarkets,onlyissuanceofU.S.dollardenominatedbondscamefromlargerorlessriskyfirms.

JELCodes:F30,G15,G30.

Keywords:corporatebonds,issuance,COVID,crises

*BrunoiswithAmericanUniversityandcanbereachedatbruno@.DathanandKitsularewiththeBoardofGovernorsoftheFederalReserveSystemandcanbereachedatmichele.h.dathan@andyuriy.kitsul@.WearegratefulforcommentsfromRicardoCorrea,AyoungPark,conferenceparticipantsatFMA2023andtheAEA2024PosterSession,andseminarparticipantsattheFederalReserveBoard.WearegratefulforexcellentresearchassistancefromAustinAdams,ThomasGeorgeandBillLang.TheanalysisandconclusionsarethoseoftheauthorsanddonotreflecttheviewsoftheBoardofGovernorsoftheFederalReserve.

2

1Introduction

Duringperiodsoffinancialmarketstress,investorshoardliquidity,disengagefromriskandflocktohigh-qualityassets,makingitmoredifficultforfirmstoraisecapitalandsecurefunding.Forexample,duringthedepthsoftheglobalfinancialcrisis,onlythehighestqualityfirmswereabletoaccessexternalcapitalmarkets,andeventhosefirmshadtoissuebondsthatinvestorsdeemedaslessrisky,suchasthosewithshortermaturitiesandmore

security(Ereletal.,

2012).

Suchflight-to-qualitybehaviorbybondinvestorscanresultinaninefficientallocationofcapitaltofirms,leadingtoforegoneinvestmentopportunitiesand

othersocialcosts(CaballeroandKrishnamurthy

(2008),

Vayanos

(2004))

.

Incontrast,thispatternwasnotobservedduringtheCOVIDpandemic,evenastheglobaleconomyenteredintounchartedwatersamidunprecedenteduncertainty.Asdocumentedin

Hallingetal.

(2020)and

BeckerandBenmelech

(2021),issuanceofcorporatebondsby

U.S.firmsboomedinspringof2020,partlyreflectingextraordinarymonetaryandfiscalsupport.Suchresilienceofprimarycorporatebondmarketshelpsalleviatefirms’financialconstraints,allowsthemtopursueattractiveinvestmentopportunitiesandbetterwithstand

futureshocks(HanandQiu,

2007)

.However,elevateddebtissuance,especiallyamongriskierfirms,mayprecedefuturecreditcrunches,significantwideningofcreditspreadsand

deteriorationoffirms’financialhealthandcreditquality(GreenwoodandHanson,

2013)

.Therefore,itisimportanttounderstandhowfirms’financialratioshaveevolvedthroughthe

pandemic.

Furthermore,financialmediaandmarketparticipantspointtoasimilarbondissuanceboomoutsideoftheU.S.(forexample,

Toole

(2021),

Lonski

(2021),and

Wheatley

(2020)).Indeed,

asshowninFigure

1,comparingcumulativenonfinancialcorporatebondissuancebyyearin

differentregionsfrom2015to2021,bondissuanceboomedeverywherearoundtheworldin

2020despitethepandemic(redlines);almostallregionsexperiencedarecordlevelofannual

3

050010001500

issuance,withtheonlyexceptionbeingnon-Chinaemergingmarketeconomies(EMEs),

whichalsosawrecordissuancebutin2021(bluelines).

Figure1:Totalvalueofcorporatebondissuancebyyearfor2015-2021

AnnualU.S.dollarvalueofnonfinancialcorporatebondissuancefor2015to2021.Includes

issuancefromfirmsinourmatchedsample.Source:RefinitivWorkspace.

Euroarea

OtherAEs

U.S.

Amtissued(USDbn)

0100200300400500

0200400600

123456789101112month

123456789101112month

123456789101112month

China

Amtissued(USDbn)

0100200300400

123456789101112month

OtherEMEs

0100200300400

123456789101112month

2020

2015-2019

2021

However,academicliteraturehasdevotedmuchlessattentiontoissuancepatternsoutsideoftheU.S.andensuingimplicationsforissuingfirms,despiterapidlyincreasingimportance

ofcorporatebondmarketsabroadoverrecentdecades.1

Inthispaper,weattempttoclose

thisgapbyusingamergedglobaldatasetonsecurity-levelcorporatebondissuanceand

1Debtsecuritiesofnon-financialcorporationsasapercentageofGDPalmostdoubledbetween2009and

2020,reaching6.8%inadvancedeconomiesand2.2%inemergingeconomies(Aldasoroetal.,

2021).Within

oursample,thenewissuecorporatebondmarketoutsidetheU.S.reacheds1.35trillionin2021,comparedtos830billionintheU.S.

4

firm-levelfinancialstatementstoexaminehownonfinancialcorporatebondissuancepatternsduringCOVIDpandemiccomparedtothoseobservedinpreviousperiodsoffinancialstress(theglobalfinancialcrisisandthetapertantrum)acrossfirmsandcountries.Furthermore,weexaminefinancialhealthratiosforfirmsfromvariousregionsaroundtheworldthatissued

bondsduringCOVIDandidentifyhowtheyusedtheproceedsfromsuchissuance.

Inparticular,weaskthefollowingquestions:HowdidcorporatebondissuanceevolveovertheCOVIDpandemicindifferentregions,andhowdoessuchevolutioncomparetopreviousperiodsoffinancialstress?WethenexaminethecharacteristicsoffirmsthatissuedbondsduringCOVID,aswellaswhatthesefirmsusedtheraisedcashforandhowhealthytheylookinapost-pandemicworld.Comparingtheevolutionofcorporatedebtissuanceandfirmfinancialsacrossstressperiodsandcountriesshouldhelpusunderstandbothfirms’responsestostressandhowgovernmentpolicieshelpedthemcope,aswellashowtheongoingremoval

ofpolicyaccommodationmayaffectthecorporatesectoraroundtheworld.

Inordertoformallyquantifyhowthedifferentperiodsofstressimpactedthebondmarket,weexamineaglobalpanelofmorethan60,000firmsfrom45countriesovera16yearperiodthatincludesthethreeperiodsofacuteeconomicdistress(COVID,tapertantrumandtheglobalfinancialcrisis).WethendelvedeeperintothesubsampleoffirmsthatissuedbondsduringCOVID,andcomparetheircharacteristicstoissuersinthenon-stressyearsleadingupto2020

.2

Finally,weexaminethepost-COVIDoutcomesforthesampleoffirmsthat

contributedtotheissuanceboomduringCOVIDcomparedtothosethatdidn’t.

Wefirstusethispaneldatatoestimatefirms’propensitytoissuebondsduringnormalandstressperiods.Wefindthatincontrasttotheglobalfinancialcrisisandthetapertantrum,firmsinallregionsweremorelikelytoaccessthecorporatebondmarketduringCOVID

relativetonon-stressperiods.ThisincreasedpropensitytoissuebondsduringCOVIDis

2ThedefinitionoftheCOVIDperiodusedthroughoutthispaperaimstocaptureonlyaninitial,especiallyuncertain,periodofthepandemic(MarchtoJune2020,inclusive)ratherthantheentiredurationofthepandemic.

5

concentratedinissuanceofbondsdenominatedinlocalcurrency,ratherthanU.S.dollardenominatedbonds.Furthermore,theamountissuedandnumberofissuedcorporatebondswasalsohigherduringCOVIDrelativetoprecedingyears,incontrasttopreviousperiods

ofstress.

Second,consistentwiththetheorythatcapitalprovidersbecomemorecautiousinbadtimes,wefindthatincreasedissuanceduringCOVIDwasdrivenbylessriskyfirmsinadvancedeconomies,asproxiedbyfirmsize,leverageandprofitability.Innon-Chinaemergingmarkets,issuanceofU.S.dollardenominatedbondsduringCOVIDwasdonebylargerandlessriskyfirms,similartothecaseofadvancedeconomyfirms.Incontrast,non-Chinaemergingmarketissuanceoflocalcurrencybonds(thecurrencysegmentwhereissuanceincreasedsignificantly)wasnotdonebysaferfirmsduringCOVID,thoughtheywerelargerduring

thetapertantrum.

Third,weexplorepotentialreasonsthatmayexplainthesecorporatebondissuancepatterns,withafocusontheunprecedentedcentralbankpolicysupportthatoccurredglobally.Bond-buyingprogramsinmanyeconomiesexpandedcentralbankbalancesheetsdramatically,andweshowthattheadditionofproxiesforthisactivityandothermacroeconomicconditionsdampenstheimportanceoftheonsetoftheCOVIDpandemicinourregressionanalysis;in

otherwords,monetarypolicysupportpartlyexplainstheboominCOVIDissuance.

Inourfinalsetofanalysis,weexaminetheevolutionoffirmfinancialratiosaroundtheCOVIDpandemictoidentifyuseofbondissuanceproceedsandassessiffirmsappearover-levered.WeshowthatChinesefirms,inparticular,whoissuedduringCOVID,havehigher

leveragethantheirnon-issuingpeers.

RelatedLiterature.Thereisafast-growingliteraturethatexamineshowfirmsreactedtotheonsetofCOVIDduring2020,includingtheiraccesstocapitalandfundingmarkets.After

initiallyrelyingoncreditlinesina“dashforcash”(AcharyaandSteffen,

2020),companies

6

wereabletosuccessfullytapcapitalmarkets,althoughlessfinanciallyconstrainedandhigher-credit-qualitycompanieswere,atleastinitially,moresuccessfulinraisingcapitalthanothers(e.g.

Hallingetal.

(2020))

.Inaddition,bondissuancewasmoreresilientthansyndicated

loanissuance(BeckerandBenmelech,

2021),anddebtfinancingwasmoreprevalentthan

equityissuanceearlyinthepandemic(Hotchkissetal.,

2020).Asmallernumberofpapers

examinewhatfirmsdidwiththeproceedsfrombondsissuedduringCOVID;

Darmouniand

Siani

(2022)showtheyusedfundstorepayloansandhoardcashratherthanrealinvestment,

likelydrivenbyprecautionarymotives(PaganoandZechner,

2022)

.Thesecashholdingshelpedfirmskeeptheirnetleveragestableevenamidaborrowingbinge;grossleverage

ratiosexhibitedanincreasingtrendevenpriortothepandemic(Benmelech,

2021).

Pagano

andZechner

(2022)arguethatforlistedcompaniesevengrossleverageratiosdeclinedas

equitycapitalincreasesdominatedtheirborrowing-inducedbalance-sheetexpansion.

Ourpaperexamineswhethersomeofthesepatternsholdaroundtheglobeandhowtheycomparetopriorperiodsoffinancialstress.Toourknowledge,non-U.S.evidenceonissuesrelatedtofirms’capitalraisingandfinancialdecisionsduringtheCOVIDpandemicisquitescant,withtheexceptionofsomeanalysisofEuropeanfirms.Inparticular,

Darmouni

andPapoutsi

(2021)investigatewhetherinEuropeitwascorporatebondsoflargersafer

issuersorthoseofnewmarketentrantsthatweresubjectofsell-offbybondinvestorsanddowngradesbyratingcompanieswhile

PaganoandZechner

(2022)examinecapitalraising

activitiesandfinancialdecisionsoffirmsinbothU.S.andEuropeanddocumentbroadly

similarpatternsacrossthetworegions.

Ourstudycontributestotheliteraturealongthreedimensions.First,toourknowledge,ourpaperisthefirstcomprehensiveexaminationofCOVID-periodcorporatebondissuanceacrossallregions,includingEMEs.WhiletheissuancesurgeintheU.S.andEuropehasbeendocumentedandisperhapsnotsurprising,lessisknownaboutEMEs,thegovernments

ofwhichprovidedlessmonetaryandfiscalsupport.EMEsareofgrowingimportanceintheir

7

ownright.Moreover,issuancepatternsinEMEsmayhelpunderstandcross-borderspillovers

ofadvancedeconomymonetaryandfiscalpolicies.

Second,wecomparebondissuancepatternstothoseobservedoverearlierperiodsoffinancialstress,includingtheglobalfinancialcrisisandthetapertantrum,helpingshedlightontheroleunprecedentedworldwidepolicysupportdistinguishingtheCOVIDpandemicmighthaveplayedinsupportingcorporatecreditmarkets.Inthisrespect,thecloseststudyis

BeckerandBenmelech

(2021);itcomparescapitalraisingduringCOVIDpandemictothose

duringtheglobalfinancialcrisisinthecontextoftheU.S.

Third,weexaminenotonlybondissuancebutalsolinkfirm-levelissuancetofirm-levelfinancialandrealoutcomes,againusingourcomprehensivecross-countryfirm-leveldataset

.3

Stabilizingcorporateborrowingratesandfacilitatingfirms’accesstobondmarketsisoneoftheinitialstepsthroughwhichcreditmarketsupportpoliciesaffecttheeconomy.TheseeffectshavebeenstudiedbothincontextofCOVIDpoliciesinthepaperscitedabove

andinthecontextofearliercreditmarketsupportprogramsbyECBandBOE(D’Amico

andKaminska

,

2019;

Todorov,

2020)

.However,longer-termmacroeconomicandfinancial-stabilityimplicationsofsuchpolicieswilldependonwhatcompaniesdowiththeraisedfunds.Trackingfirms’financialandrealoutcomesovertimeisthefirststeptounderstandingsuch

longer-termeffects.

3Otherpapersthatemployissuance-firmmatcheddatasetsforfirmsoutsideoftheU.S.include

Gozzi

etal.

(2010),

Gozzietal.

(2015),

Cortinaetal.

(2018),and

Didieretal.

(2021)

.

8

2EmpiricalSpecification

2.1DataandVariablesofInterest

WeusedataoncorporatebondissuancetransactionsfromRefinitivWorkspaceforInvest-mentBankingforasampleperiodofJanuary1,2005toDecember31,2021.Thedatabasecoverstransaction-leveldetailsforbondofferingsfromcompaniesaroundtheworld.Weexcludeconvertibledebenturesandpreferredshares,aswellas$0issuanceandsecurities

withlessthan1yeartomaturityatissuance.

Wethenmatchbondissuancedatawithfirm-levelannualfinancialstatementsfromRefini-

tivWorldscopefortheyears2004-2020(financialdataislaggedtothefiscalyearbeforebondissuance).Wematchfirstbasedontheultimateparent’sRefinitivInstrumentCode(RIC),whichcanbefoundinbothRefinitivWorkspaceandRefinitivWorldscope.Wethensupplementadditionalmatchesbasedonfirmname(Issuer/BorrowerNameFullinRefinitivWorkspaceandNameinRefinitiv).Someofthematchesinthissecondstepincludesub-sidiaryfirmsthathavefinancialstatementsinWorldscopewhiletheultimateparentdoes

not(e.g.PetroChina).

Wefocusonbondissuanceactivitybynon-financialcorporationsbyexcludingfromWorld-scopeallfirmswithanSICcodebetween6000and6999.Wealsoexcludepublicadminis-trationfirmswithanSICcodeabove9000.Finally,weexcludefirmsincountrieswithfewer

than10uniqueissuersoverthesampleperiod.

Ourfinalmatcheddatasetincludes60,421firmsfrom45countries,ofwhich6,058issuedatleastonebondduringoursampleperiod(see

AppendixA

forabreakdownbycountry).Ourmatcheddatasetof80,375bondstotalsUS$25.8trillioninfacevalue,whichrepresents75.5%ofthenumberofand97.3%ofthefacevalueofnon-financialcorporatebondsinRefinitiv

Workspace.

9

Ourmainvariablesofinterestconcernmonthlybondissuanceatthefirmlevel,comparingissuancedecisionsandfirmcharacteristicsduringmonthswithfinancialstresscomparedtomonthswithnostress(“normaltimes”).Wedefineourindependentvariablesofinterestasthreedummiesfortheperiodsoffinancialstress,whichtakeonavalueof1inthefollowing

months:

.COVIDpandemic:MarchtoJune2020,inclusive;

.Tapertantrum:May2013toApril2014inclusive;and

.Globalfinancialcrisis:December2007toJune2009,inclusive.

Thesemonthswerechosenbasedonthelikelyimpactonthebondmarkets.TheGFCmonthswerechosenbasedontheNBER-definedrecession;thetapertantrumincludestheMay2013testimonybyBenBernankesignalingtheunexpectedstarttotheendofquantitativeeasing,andincludesthesubsequentslow-downinChinaandotherEMEs;andtheCOVIDmonths

includethefourmonthswheretheVIXindexwasatanaveragemonthlylevelabove30.

Wefirstexaminebondissuancedecisions,suchasthepropensitytoissueabond(issuerdummy,

whichtakesonavalueof1ifthefirmissuesatleastonebondinamonthand0otherwise),

andthedollaramountandnumberofbondsissuedbyafirminamonth.Issuanceamountsaremeasuredinconstant2011U.S.dollars.Wethenlookatthecharacteristicsofthebondsissued,suchastheweightedaveragetimetomaturityofnewbondsissuedbyafirminamonth,andtheproportionofbondsissuedbyafirmthatareratedandinvestmentgrade

rated.

Wethenturntocharacteristicsofbondissuersasofthefirm’spreviousfiscalyearend,suchaslogassets(logofUSDtotalassetvalue,using2011constantU.S.dollars),bookleverage(totaldebtdividedbytotalassets),andprofitability(netincomedividedbytotalassets).

Thelattertwovariablesarewinsorizedatthe5%and95%level.

10

Wetheninvestigatethetimeseriesofmacroeconomicvariablesaroundourperiodsoffinan-cialstress.Atthecountry-monthlevel,weexamineacountry’saverage10-yeargovernmentyield,thesizeofthecentralbank’sbalancesheetrelativetoGDP,andflowsintoacoun-try’sbondmutualfundsasapercentageofsuchfunds’assetsundermanagement.Atthemonthlylevelbutwithnocountryvariation,weexaminetheleveloftheWu-XiaShadow

FederalFundsRate(WuandXia,

2016),theleveloftheUSDbroaddollarindex,andthe

monthlyVIXlevel.

Inafinalsetofanalysis,welookattheevolutionoffirmfinancialratiosaftertheonsetoftheCOVIDpandemic.Inadditiontothepreviouslydefinedbookleverage,wealsolookatthepercentageofshort-termdebtrelativetototaldebt;theproportionoffirmswithinterestcoverageratios(definedasearningsbeforeinterest,taxes,depreciationandamortizationdividedbyinterestexpense)lessthan2;andcash,capitalexpendituresanddividends,all

dividedbytotalassets.

Weperformouranalysisforfiveseparategeographicalareas:theeuroarea,otheradvancedeconomies(AEs),theUnitedStates,China,andotheremergingmarketeconomies(EMEs).InChinaandotherEMEs,inadditiontolookingatallbondissuance,weexamineinsome

testslocalcurrencyissuanceandUSDissuance.

Table

1

showssummarystatisticsforthefullsampleperiod,brokendownbyregion.

InPanelA,wedescribethecharacteristicsofbondissuance,includingtheregionalbreakdownofthe6,058uniqueissuingfirms.Oursampleincludes17,542bondsfromnonfinancial

corporationsintheU.S.4

Forfirm-monthsthatincludeissuance,theaverageamountofbondsissuedislargestintheU.S.atapproximately$992million,followedcloselybythe

euroareaat$984million.OtherAEs,ChinaandotherEMEsaresmallerat$579million,

4Thisnumbercanbecalculatedasthenumberoffirm-monthswithissuance(10,142)multipliedbytheaveragenumberofbondswhenissuing(1.73).Oursampleissimilartothe17,379bondsissuedbetween2002and2020asdescribedin

BeckerandBenmelech

(2021)

.

11

Table1:Summarystatistics

Sample:

Euroarea

Other

AEs

U.S.

China

Other

EMEs

PanelA:Bondissuancedetails

Numberofuniqueissuingfirms

468

1,286

1,499

975

1,830

Numberoffirm-monthswithissuance

4,713

8,964

10,142

6,821

13,420

Averagenumberofbonds

0.02

0.01

0.02

0.01

0.01

Averagenumberofbonds(>0)

1.96

2.01

1.73

1.46

1.91

Averagebondsize(sm)

s10.1

s2.6

s8.8

s3.3

s1.3

Averagebondsize(>s0m)

s1,078.9

s646.3

s1,008.3

s376.8

s268.3

Weightedaveragematurity

8.53

8.95

11.44

6.10

5.80

Ratedshare

67.2%

39.9%

89.1%

12.2%

15.8%

IGratedshare

54.6%

32.3%

53.6%

4.1%

10.1%

PanelB:Bondissuercharacteristics

Logassets

10.25

9.73

9.31

9.13

8.97

Bookleverage

35.6%

36.8%

38.3%

36.6%

37.1%

Profitability

3.2%

3.3%

3.7%

3.0%

2.9%

PanelC:Allfirmcharacteristics

Numberofuniquefirms

4,043

18,360

12,073

6,407

19,538

Numberofuniquefirm-months

503,946

2,257,255

1,155,809

773,922

2,719,365

Logassets

5.36

4.28

4.07

5.94

4.97

Bookleverage

24.8%

18.7%

27.3%

21.1%

23.9%

Tangibility

23.4%

30.1%

24.7%

26.5%

31.1%

PanelD:Macroeconomicconditions

Levelof10-yearyield

2.88%

2.12%

2.92%

3.45%

4.76%

ChangeinUSDbroadindex

0.05%

0.05%

0.05%

0.05%

0.05%

WuXiashadowFedFundsrate

0.69%

0.69%

0.69%

0.69%

0.69%

MonthlychangeinVIX

2.10%

2.10%

2.10%

2.10%

2.10%

Changein1-mtheuroareaOISatECBann’ts

0.19%

0.19%

0.19%

0.19%

0.19%

Flowintobondfundsas%ofAUM

0.39%

0.41%

0.32%

5.06%

0.74%

12

$370millionand$242million,respectively.Intermsofnumberofbondsissued,formonths

withpositiveissuancetheaveragenumberofbondsissuedis2.01inotherAEs,1.96intheeuroarea,1.91inotherEMEs,1.73intheU.S.and1.46inChina.Averageinitialtimetomaturityislongerinadvancedeconomies(U.S.at11.5years,otherAEsat9.0yearsandeuroareaat8.5years)thaninemergingeconomies(6.1and5.8yearsinChinaandotherEMEs,respectively).ThemajorityofbondsintheU.S.andeuroareaarerated,whileonly40%ofbondsinotherAEsarerated.Only12%and16%ofbondsinChinaandotherEMEs

arerated.

PanelBshowscharacteristicsofbondissuingfirmswhereasPanelCshowscharacteristicsofallfirms.Notsurprisingly,bondissuersare,onaverage,largerandmoreleveredthanthefirmsinthefullsample.Intermsofbondissuers,euroareafirmsarethelargestandleastlevered,whilefirmsinotherEMEsaresmallestandhavethesecondhighestleverage(U.S.issuershavethehighestleverage).Inthefullsample,Chinesefirmsarethelargestandhavethesecondlowestleverage,whileU.S.firmsarethesecondsmallestandhavethehighest

leverage.

Finally,PanelDshowsthattheaverage10-yeargovernmentyieldsoverthesampleperiodarelowestinotherAEs,followedbytheeuroareaandtheU.S.Non-ChinaEMEshavethehighestaveragesovereignyields.Chinahasexperiencedthelargestaverageflowsintobondmutualfunds,althoughtheaverageispositiveacrossallregions.Overthesampleperiod,theaveragemonthlychangeintheUSDbroaddollarindexis0.05%,theaverageshadowFederalFundsratesis0.69%,theaveragemonthlychangeinVIXis2.10%andtheaveragechangeinthe1-montheuroareaovernightindexswaparoundEuropeanCentralBankannouncements

is0.19%.

13

2.2Regressionspecifications

Inordertotesttherelationshipbetweenissuancedecisionsandperiodsoffinancialstress,

ourmainregressionspecificationtakesonthefollowingform:

yi,t=α+β*COVID+δ*TT+γ*GFC+ηi+?,(1)

whereyi,tisoneofissuerdummy,dollaramountissued(convertedusingconstant2011U.S.dollars),ornumberofbondsissued.Thesevariablesaredefinedforthefullsampleoffirm-months.Weruntheanalysisonanunbalancedpanel,basedonthefactthatsomefirmsenterandexitthesample,butforrobustnesswealsore-runtheanalysisusingonlythefirmswithdataforallyears;theresultsremainqualitativelythesame.β,δandγmeasurewhetherissuancedecisionschangeduringstressmonthsrelativetotherestofthesampleperiod.Wecontrolforfirmfixedeffectsinallspecificationswithηi.Inordertoproperlyaccountforthe

cross-sectionaldependenceinourpaneldata,weuseDriscoll-Kraaystandarderrors(Driscoll

andKraay,

1998)

.

WethenrunEquation

1

onthesampleoffirm-monthswithpositiveissuance.Thedependentvariablesyi,tthatweexplorearedollar-amountweightedaveragematurityofbondsissued

inafirm-monthandshareofbondsissuedinafirm-monthwithaninvestmentgraderating.

Wenextexaminewhetherthecharacteristicsoffirmsthatissueduringstressperiodsaredifferentthanfirmsthatissueinnon-stressperiods,usingthefollowingregressionspecifica-

tion:

yi,t=α+β*COVID+δ*TT+γ*GFC+ηj+ρt+?,(2)

14

whereyi,tisthefirmcharacteristicofinterestforfirmithatissuesabondinmontht.Thefirmcharacteristicsweexaminearelogassets(inU.S.dollars,convertedusingconstant2011exchangerates),bookleverage,andprofitability.Thecoefficientsβ,δandγmeasurethedifferenceinaveragefirmcharacteristicsinstressperiodsrelativetonon-stressperiods.Wecontrolforindustryfixedeffects(using1-digitSICcode)withηjandcontrolfortimetrend

withyearfixedeffectsρt.Weclusterourstandarderrorsattheindustrylevel.

3EmpiricalResults

3.1Whatdoescorporatebondissuancelooklikeduringperiods

offinancialstress?

AsshowninFigure

1

intheintroduction,totaldollarissuanceofnonfinancialcorporatebondssurgedtorecordornear-recordlevelsin2020acrossallregions,despitetheCOVIDpandemic.Afterabriefstallintheco