prices代理猎手

TheChoiceofCapitalInstrumentsLARRYD.
WALLANDPAMELAP.
PETERSONWallisanofficerinthefinancialsectionoftheAtlantaFed'sresearchdepartment.
PetersonisaprofessorinthefinancedepartmentofFloridaStateUniversity.
TheythankJamesAng,GeorgeBenston,CurtHunter,SteveSmith,RaymondSpudeck,andotherparticipantsinworkshopsatEmoryUniversityandFloridaStateUniversityandatthe1990FinancialManagementAssociationand1991SouthernFinanceAssociationmeetingsforcommentsonapriorversionofthispaper.
ANUMBEROFVERYLARGEBANKINGORGANIZATIONSHAVERECENTLYPROPOSEDORCONSUM-MATEDMERGERS,INCLUDINGCHEMICALANDCHASEMANHATTAN,CITICORPANDTRAVELERS,ANDNATIONSBANKANDBANKAMERICA.
THESECOMBINATIONSHAVEINCREASEDTHEIMPOR-TANCEOFHAVINGASYSTEMOFBANKSUPERVISIONANDREGULATIONTHATPROTECTSTHETAXPAYERSANDTHEFINANCIALSYSTEMWHILEAVOIDINGTHEIMPOSITIONOFUNNECESSARYCOSTSONBANKS.
THISARTICLEFOCUSESONTHECOSTSIMPOSEDBYONEOFTHEPRIMARYTOOLSOFCURRENTBANKSUPER-VISIONANDREGULATION—THEMEASUREMENTANDREGULATIONOFCAPITALADEQUACY.
4FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998Bankcapitalratioshavebecomeoneoftheprinci-palmeasuresofabank'sfinancialcondition.
Capitalratioshavelongbeenanimportantregulatoryconsider-ation,buttheirimportancehasrecentlygrownpartlyasaconsequenceofinternationaleffortstoharmonizebanksupervisoryrulesandpartlybecauseoftheinclu-sionofpromptcorrectiveactionprovisionsintheFederalDepositInsuranceCorporationImprovementAct(FDICIA).
Oneimportantconcernisthattheemphasisoncapitalregulationwillincreasebanks'costsandmakethemrelativelylesscompetitivewithotherfinancialserviceproviders.
Thecapitalregula-tionsmayimposecostsonbankstotheextentthatthecontrolsreducethesubsidyvalueofthefederalsafetynet;however,thisincreaseincostsisanintendedcon-sequenceoftheregulations,designedtooffsetFederalDepositInsuranceCorporation(FDIC)liability.
1Amoretroublingquestioniswhethertheregula-tionsimposecoststhatarenotnecessaryforachievingthegoalsoftheregulation.
Inparticular,thecurrentstandardseffectivelyforcebankstomaintainminimumlevelsofequitycapitalasmeasuredinaccountingval-ues.
Yetavarietyofstudieshavesuggestedthatmain-taininghigherequitycapitallevelsatthecostofreduceddebtlevelsiscostly—forexample,inreducingthetaxshieldassociatedwithcorporateinterestpay-ments.
Ifequityismoreexpensivethandebt,regulatorsshouldreconsiderthelimitstheyimposeonsubstitut-ingdebtforequity.
InanearlierarticleinthisEconomicReview,WallandPeterson(1996)surveyedtheexistingliteratureonbanks'responsestobindingcapitalregulations.
Theyfoundthatempiricalevidencesupportsthehypothesisthatcapitalregulationsexerciseabindinginfluenceon5FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998banks'capitalpositions.
Theythenexaminedbanks'twooptionsforrespondingtobindingregulations:(1)actionsthatincreaseregulatorycapitalratiomeasureswithoutreducingabank'sriskoffailure,whichWallandPetersoncallcosmeticchanges,and(2)actionsthatincreasereg-ulatoryratiosandreducetheriskoffailure,whichtheycalledeffectiveincreasesincapital.
2Theevidencetheysurveyindicatesthatstockmarketparticipantsseethroughcosmeticchangesthatartificiallyraisecapitalandthattheyreducethestockpriceofbanksengaginginsuchsteps.
Banksmayeffectivelyincreasetheircapi-talthroughtheissuanceofnewstock,butthisactionalsoreducestheprice.
Oneexplanationofbothfindingsisthatthemarketmayinterpretcosmeticactionsandnewequityissuanceasindicatingthatthebankexpectsweakfutureearningsandthusmusttakeotherstepstosatisfytheregulatoryrequirements.
WallandPeterson'sreviewmaybeinterpretedassuggestingthatcapitalregulationsimposeunnecessarycostsonthebankingsystem,butthisconclusionisnotobvious.
Financialmarketparticipantsseemtobeusingabank'sdecisiontoissuecapitalasamethodofinferringthebank'sfutureearningpower.
Ifmarketparticipantsarerationalandfinancialmarketsarecompetitive,theseinferencesshouldbecorrectonaverage,soitisnotnec-essarilythecasethatcapitalregulationshavebeencost-lytothebankingsystemasawhole.
3However,eveniffinancialmarketparticipantsmakecorrectinferencesonaverage,theymaystillbeunabletofullyseparatebankswithgoodprospectsthathavehadbadluckfrombankswithbadprospects.
4Ifmarketpar-ticipantscannotperfectlyseparatethetwosetsofbanks,thenthemarketpriceofgoodbanksmayfallmorethanitotherwisewouldandthemarketpriceofbadbanksmayfalllessthanitotherwisewould.
Ineffect,thecapitalregulationscausegoodbankstoissuecapitalatalowerpricethantheyshould,andgoodbanks'lossesareoffsetbybadbanks'beingabletoissuecapitalatahigherprice.
Themarketmaymakeinferencesaboutabank'sconditionfromeitherdebtorequityissues.
However,astheresidualclaimantonabank'svalue,thevalueofcommonequityismostsensitivetomarketmisestima-tion,whileclaimsonafixedportionofthebank'scashflow—thatis,debt—arelesssensitivetomisestima-tion.
Hence,capitalregulationsthatoccasionallyforcebankstoissuenewequitymayimposehighercostsongoodbanksthancapitalregulationsthatallowthebanktosubstituteadebtissue.
Thisarticlefocusesonthequestionofwhetherexistingcapitalregulationsareimposingunnecessarycostsonbanks.
Thefirstsectionreviewsthere-gulatoryandmarketin-fluencesthathavebeenhypothesizedtoinflu-encebanks'decisionsaboutissuingcapital.
Thesecondsectiondescribesthemodelusedinthisresearch.
Theresultsreportedinthethirdsectionpro-videnewevidenceonthecostsassociatedwithnewcapitalissuesbybanks,therebyshed-dingadditionallightontheprivatecostsofcapitalregulation.
Thefinalpartana-lyzessomereasonswhyregulatorsmightchoosetosetminimumequitycapitalrequirements.
Thenewevidenceisobtainedbyanalyzingthedeterminantsofwhichnewsecurity,ifany,abankingorganizationissuestomeetcapitalregulations.
Issuanceofcapitalinstrumentsmayimposeavarietyofcostsonbanks,dependingontheinstrumentchosen.
Capitalreg-ulationshavealwayscountedcommonequityandatleastsometypesofpreferredstockincalculatingtheratios.
However,includingsometypesofdebtsecuritiesaswellmayenhancetheabilitytodistinguishamongthedifferenttheoriesofcapitalstructure.
Undertheexistingtier1risk-basedandleveragecapitalregula-tions,notypeofdebtsecurityisasubstituteforequity;however,thecapitalregulationsfirstadoptedinDecember1981allowedaspecialtypeofdebtcalledmandatoryconvertibledebttosubstituteforequityinprimarycapital(theequivalentofthecurrenttier1capitalmeasure).
Thus,toincludedebtsecurities,thediscussionlooksbackattheissuancedecisionsmadeundertheprimarycapitalregulationsofthe1980s.
1.
Boththecalculationofthecapitaladequacymeasureandtherequiredlevelhavebeenthesubjectofongoingdebate.
ForacriticalanalysisoftheexistingrulesseetheShadowFinancialRegulatoryCommittee(1996,Statements84,96,110,112,124,and126),PeekandRosengren(1997),andJonesandMingo(1998).
2.
Anexampleofacosmeticchangewouldbesellingassetsthathaveappreciatedinvaluebutnotthosethathavedecreasedinvaluetoincreasecapitalasmeasuredbyregulatoryaccountingevenifthesalereducedthebank'seconomiccapital.
Anexampleofaneffectiveactionistheissuanceofnewcapitalbyabank.
3.
Admittedly,theconclusionthatcapitalregulationsmayhavenotbeencostlymightbeweakenedbytheinclusionofriskaversiononthepartofinvestors.
4.
Inthiscasethetermbadluckisusedasawayofreferringtobanksthathappenedtoobtainanearningsdrawfromthelowertailofthedistribution.
Mostoftheprivatecostsandbenefitsassociatedwithdifferentcapitalstruc-turesarisebecauseofthedifferencesbetweendebtandequity.
6FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998TheresultsoftheempiricalanalysissupportWallandPeterson'sconclusionthatasymmetricinformationcostsareanimportantpartoftheissuanceofaddition-alcommonequity.
Theresultsalsosuggest,though,thattheoptiontoissuedebtsecuritiesasasubstituteforequitymaybemorevaluabletolargebanksthantosmallerbanks,thelatterbeingsignificantlylesslikelytoissuemandatoryconvertibledebt.
5Theanalysisoftheregulatoryimplicationsofallowingbankstosubstitutedebtforequityinthecapi-talstructuresuggeststhatproperlystruc-tureddebtisasgood,orbetterthan,equityinaddressingmostreg-ulatoryconcerns.
Theareaofprimaryregula-toryconcerninwhichequityislikelytobesuperioristhatofmin-imizingtheriskoffail-ureafterabankhasalreadyincurredaloss.
However,subordinateddebtmaybemoreef-fectiveindiscouragingbanksfromtakingexcessiveriskandthereforemayreducetheprobabilitythatabankbecomesfinanciallydistressed.
Moreover,evenifafailureshouldoccurtheregulatorsretainothertoolsforreducingthecoststosociety.
Thus,theoneadvantageofequityoverdebtfromaregulatoryper-spectivemaynotbethatimportant.
TheoreticalDeterminantsoftheCapitalIssuanceBanksareprivatecorporationsthatoperateinaspecialregulatoryenvironment.
Asprivatecorpo-rations,theircapitalstructuredecisionsaresub-jecttothesameinfluencesasothercorporations.
Theseinfluencesincludefactorsthatwouldleadtoanoptimalequity-to-debtratioinastaticsettingaswellasdynamicadjustmentcostssuchasthecostsofissuingnewequity.
Theregulatoryenvironmentmodifiestheprivatecostsandbenefitsofdifferentcapitalstructuresintwoimpor-tantways.
First,depositinsurancereducesthesensitivityofinsured-depositinterestratestoanorganization'srisk-inessbyguaranteeingrepaymentevenifthebankshouldfail.
TheFDIC'shistoricalpracticeofextendingtheseguaranteestootherliabilitiesthatlackdejureinsurancecoveragemayalsoreducethesensitivityoftheseclaimstothebank'sriskiness.
6Thelowersensitivityofliabilityratestoabank'sriskinessreducestheamountofcapitalshareholderswouldwantthebanktoholdforanygivenlevelofportfoliorisk.
Thesecondregulatoryinfluenceisthatofcapitalregulations.
Theseregulationsareone-sided:regulatorsrequirebankstomaintainminimumlevelsofcapital,buttheyvirtuallyneverobjecttoabankmaintainingcapitalratiosinexcessofitsneeds.
RegulatoryInfluences.
ThetheoryofsecurityissuanceforU.
S.
banksincorporatesboththetheoryofcapitalstructurefornonfinancialcorporationsandtheuniquefeaturesofbanks.
Oneofthemostimportantfea-turesofbanksisthattheirdepositsareinsuredbythefederalgovernment.
7Aconsequenceofdepositinsur-anceisthatthecostofalargeportionofabank'sfundsisrelativelyinsensitivetochangesinthebank'srisk,cre-atinganincentiveforbankstotakegreaterrisks.
Thefederalgovernmentattemptstolimittheexposureofitsdepositinsuranceagencybyimposingavarietyofregu-lationsonbanksandbyrequiringbankstoundergoperi-odicexaminations.
Capitalregulationisanimportanttypeofregula-tion.
U.
S.
bankregulatorshavelongbeenconcernedwithbankcapitaladequacy.
Thecapitalregulationsdur-ingthe1970swereenforcedonacase-by-casebasis,suc-cessfullypreventingmostbanksfromloweringtheircapitalratiostoalevelsignificantlybelowtheirpeers'duringthisperiod.
Butregulatorsdidnotpreventtheindustryasawholefromreducingitscapital(Marcus1983).
The1981capitalguidelinesweredevelopedtostopthereductionincapitalratiosandtoincreasetheratiosatthelargestU.
S.
bankingorganizations.
Thecapitalguidelinesannouncedin1981bytheFederalReserveSystemforbankholdingcompaniesdefinetwotypesofcapital:primarycapitalandtotalcap-ital.
Primarycapitalincludescommonstock,perpetualpreferredstock,retainedearnings,loan-lossallowance,andmandatoryconvertiblesecurities.
Totalcapitalincludesprimarycapitalpluslimited-lifepreferredstockandsubordinateddebt.
Thestandardsalsodefinethreecategoriesofbankorganizations:multinationalorgani-zations(theseventeenlargestbankholdingcompa-nies),regionalorganizations(allotherbankswithassetsinexcessof$1billion),andcommunityorganiza-tions(thosewithassetsoflessthan$1billion).
The1981guidelinesdonotspecifynumericalstan-dardsforthemultinationalorganizations,but1981statementsexpressedtheregulators'expectationthatthesefirmswouldincreasetheircapitalratios.
Regionalorganizationswereexpectedtomaintainaminimumprimarycapital-to-total-assetsratioof5percent,where-ascommunityorganizationswererequiredtomaintaina6percentratio.
Theregulatorsalsostatedthatbank-ingorganizationsweregenerallyexpectedtooperateatcapitallevelsabovetheseminimalstandards.
TheregionalbankstandardwasextendedtocoverthemultinationalorganizationsinJuneof1983.
8Thepri-marycapitalstandardforallbankingorganizationswassetat5.
5percentinMarch1985.
Ifequityismoreexpensivethandebt,regulatorsshouldreconsiderthelimitstheyimposeonsubstitutingdebtforequity.
7FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998Theeffectofthe1981primarycapitalguidelineshasbeentoplacealowerboundontheprimarycapitallevelofbankingorganizations.
Further,thelimitsontheamountofmandatoryconvertibledebtincludedinprimarycapitalsetalimitonthemaximumtotal-debt-to-total-assetsratio.
Althoughthe1981standardsappeartohavebeeneffectiveinraisingcapitallevels,theregulationsalsoseemedtobedistortingbanks'portfoliodecisions.
Inpar-ticular,thestandardsdidnotdistinguishamongtherisk-inessofdifferentassetsandalsofailedtoexplicitlyincorporateoff-balance-sheetexposuresintothecapitalrequirements.
Subsequenttotheimpositionofthe1981standards,bankswereobservedrespondingtotheappar-entincentivescreatedbythecapitalregulations—notonlyweretheyincreasingcapitalbuttheywerealsoreducingtheirholdingsofhighlyliquid,low-riskassetsandincreasingtheirexposuretooff-balance-sheetcon-tracts.
InJuly1988thecentralbanksandbankregulatorsofthemajorindustrialnationsreachedaninternationalagreementtoimplementcapitalguidelinesthattookmoreaccurateaccountofthecreditrisksassociatedwithbanks'on-andoff-balance-sheetportfolios.
Interimrisk-basedcapitalstandardstookeffectin1990,withthefullstandardstakingeffectattheendof1992.
Asapartoftherisk-basedcapitalguidelines,thenarrowerdefinitionofcapitalexcludedmandatoryconvertibledebt,reducingitsvalueasasubstituteforequityincomplyingwiththecapitalguidelines.
9Thus,eventhoughtheprimarycapi-talstandardsarenolongereffective,morecanbelearnedabouttherelativecostsofdebtandequityarisingfrommarketforcesbyanalyzingbankcapitaldecisionsundertheprimarycapitalregulationsofthe1980s.
MarketInfluences.
Marketforcescouldpotentiallyimposevaryingcostsbasedonboththelevelofabank'scapitalandchangesinthebank'scapitalstructure.
ThetheoreticalstartingpointforanalyzingmarketforcesisModiglianiandMiller's(1958)demonstrationthatafirm'scapitalstructure—thatis,itsmixofdebtandequi-ty—doesnotaffectitsvalueinperfectmarkets.
Animpli-cationofthismodelisthatsecuritiespricesareanunbiasedestimateoftheirintrinsicvalue,sothetimingandtypeofsecuritysoldbythefirmdonotaffectthevalueofthefirm.
ModiglianiandMiller'sworknotonlyestab-lishedtheconditionsunderwhichcapitalstructureisirrelevantbutalsotoldfinancialeconomistsunderwhatconditionscapitalstructuremayberelevant.
10Buildingonavari-etyofstudiesanalyzingnonfinancialcorpora-tions'optimalcapital,OrglerandTaggart(1983)developamarketmodelofoptimalcapitalstructureforbanks.
11Intheirmodel,thebenefitstobanksoflowercapitalratiosaremorefavor-abletaxtreatmentandanincreaseinthevalueofdepositinsurance.
Theoffset-tingcostsoflowercapitalratiosarethe(eventual)disec-onomiesofscaleinproducingdepositservicesandthedeadweightcostsofbankruptcythatarepartiallybornebythebank'sowners.
12Flannery(1994)arguesthatagencycostsalsomaybeanimportantdeterminantofbankcap-italstructures.
13Lowercapitalratiosimposedesirable5.
Onelimitationoftheempiricalanalysisisthatthemodelhasproblemsidentifyingwhybankswouldissuepreferredstockratherthanmandatoryconvertibledebt.
6.
Althoughdejuredepositinsurancecoveragewaslimitedto$100,000perdepositorinadomesticbranch,theFDICgener-allyprovided100percentcoverageofalldepositsandsometimesguaranteednondepositliabilitiesduringthetimeperiodofthisarticle'ssample.
However,the1991passageofFDICIAinitiatedavarietyofstepstoreducethegovernmentsubsidytofailedbanks.
BankregulatorsappeartobegenerallyfollowingthroughonFDICIA,anddepositinsurancecoveragehasbeenlimitedformostofthebankfailuressincetheact'spassage.
However,theeffectivenessofthesestepsinpracticehasnotyetbeenfullyresolvedbecausenoneoftheverylargebanksthatwereeligibleforinclusioninthisstudy'ssamplehavefailedsincetheadoptionofFDICIA.
SeeWall(1993)foradiscussionofFDICIAanditsapplicationtolargebanks.
7.
DepositinsuranceoriginateswiththeFederalDepositInsuranceAct,BankingActof1933(48Stat.
162[1933]).
TheFDICprovidesinsurancefordeposits,accompaniedbyregulatoryandexaminingfunctionstomonitorthisinsurancefunction.
8.
Priorto1983bankholdingcompanycapitalregulationswerebasedontheFederalReserve'sgeneralsupervisoryauthori-ty.
In1983theFederalReservewasgivenaspecificstatutorymandatebytheInternationalLendingSupervisionActof1983(PublicLaw98-181)torequirebankingorganizationstomaintainadequatecapitallevels.
9.
SeeWall(1989)foradiscussionofthe1981guidelinesandtheirreplacementbytherisk-basedcapitalguidelines.
10.
SeeMiller(1995)foradiscussionoftherelevanceofthe"M&M"propositionstobanking.
11.
Forexample,seeModiglianiandMiller(1963),DeAngeloandMasulis(1980),andMasulisandTrueman(1988)onincometaxesandBaxter(1967)andKrausandLitzenberger(1973)onbankruptcycosts.
12.
Diseconomiesofscaleexistifanincreaseinvolumeresultsinanincreaseinaverageunitcosts.
Deadweightlossesofbank-ruptcyrefertocoststhatarisesolelybecauseofthebankruptcyandprovidenosocialvalue—legalcosts,forexample.
13.
SeeJensenandMeckling(1976),Barnea,Haugen,andSenbet(1981),andJensen(1986)foradiscussionofagencycostsinmoregeneralsettings.
Allowingbankstoissuedebtratherthanequitymayreducetheircostsofcomplyingwiththecapitalstandards.
8FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998limitsonmanagementandreducetheneedforshare-holdermonitoring.
14Conversely,lowercapitalincreasestheincentivesforbankshareholderstohavemanagersundertakeriskierprojectsandtorejectsomelow-riskinvestments.
Thesecostsofreducedcapitalmaybemiti-gated,Flanneryargues,byhavingthebankissuedepositswithveryshortmaturitiessothatdebtholdersmaytakeeffectiveactionifthebankadoptsahigh-riskinvestmentstrategy.
Thus,Flannery'sanalysisarguesthatbanksshouldissueveryshort-termdebtandmaintainlowcap-italratios(althoughtheywouldnotnecessarilybeunder-capitalizedbyregulatorystandards).
ShrievesandDahl(1992)andHughesandMester(1994)pointtoanotheragencyproblemthatmayinflu-encebanks'capitalstructure—managerialriskaver-sion.
Mostindividualsarethoughttoberiskaverse,andthereisnogoodreasonforthinkingthatbankmanagersareinherentlymoreriskaversethantheaverageshare-holder.
However,bankmanagershaveproportionatelyfarmoreoftheirtotalwealth(includinghumancapital)investedintheirbankthanmostshareholdersand,asaconsequence,havemoretolosefromthebank'sfailure.
Thus,bankmanagersmaychoosehighercapitallevelsthanwouldbeoptimalfromtheshareholder'sperspective.
HughesandMesterestimatebankcostfunctionsthatallowformanagerialriskaversionandfindsupportforsuchriskaversion.
Animplicitassumptionofthestatictrade-offmodelsofcapitalstructureisthatthecostofadjustingabank'scapitalstructureiszero.
Recentworkthatfocusesoninformationasymmetriesbetweenmanagersandinvestorssuggests,however,thattheprocessofadjustingthecapitalratiomayconveyimportantinformationtoshareholders.
Animportantpartoftheanalysesofinfor-mationasymmetrieshasfocusedontheissuanceofnewsecuritiesbycorporations.
MyersandMajluf(1984)exam-ineafirm'sdecisiontoissuedebtorequityandconcludethattheannouncementtoissueequityconveysnegativeinformationtothemarketaboutthefirm'svalue.
Themar-ketmayovervalueboththedebtandequityofafirm.
However,ifthemarketoverestimatesthevalueofafirm,thatoverestimationwillhaveaproportionatelylargerimpactonequitybecauseequityhastheresidualclaimonfirm'svalue.
Thus,ifmanagementbelievestheintrinsicvalueofafirmislessthanitsmarketvalue,existingshareholdersbenefitifthefirmissuesequity.
Otherwise,existingshareholdersarebestservedeitherbythefirmissuingdebtorforgoinganynewsecurityissue.
Prospec-tivenewshareholdersrealizetheincentiveofexistingshareholderstohavethefirmissuenewequityonlyifitisovervaluedand,hence,interpretanewequityissueasanadversesignalaboutfirmvalue.
Thismodelsuggeststhatfirmsgenerallyprefertoissuedebtratherthanequity.
Oneversionofthisanalysisholdsthatfirmsfollowapeckingorderindeterminingwhichsecuritiestoissue.
Afirmwillissuedebtuntilfurtherdebtissuancewouldbecome"excessively"costly,andthenitwillissueequity.
Thus,avarietyofhypotheseshavebeenofferedre-latingtothecostandbenefitsofdifferentlevelsofequi-tyandchangesintheequitylevel.
Mostofthesecostsandbenefitsarisefromimportantdifferencesbetweendebtandequity.
First,interestpaymentsondebtreceivemorefavorabletaxtreatmentthandividendsonequity.
Second,equitymayabsorblosseswithoutcausingthefirmtoenterfinancialdistressandbankruptcywhereasbankruptcyisoftenrequiredbeforedebtholderswillacceptreducedpayments.
Thisseconddifferencehasfourimplications:(a)higherlevelsofdebtfinancing,holdingotherfactorsconstant,increasetheexpectedcostsoffinancialdistress,(b)higherlevelsofdebtfinancingincreasetherisktomanagers'humancapital,(c)higherlevelsofdebtmayencouragemoreefficientmanagement,and(d)higherlevelsofdebtgiveequityholdersanincentivetopreferariskierinvestmentstrategy.
Athirddifferencebetweendebtandequityisthatthemarketislesslikelytoviewdebtissuanceasanadversesignal.
Onefurtherissueisthatofapossiblescaleeffectinthecostassociatedwithsecurityissuance.
Smallerfirms(nonbankingaswellasbanking)arelesslikelytohavepubliclyissuedsecurities,andthosehavingpubliclyissuedsecuritiesarelikelytohavealessdiversesetoftypesofsecurities.
Apossibleexplanationisthatsmall-ersecurityissuestendtobelessliquid,inpartbecausethecostsofanalyzingasecurityoftenincreaseataratethatislessproportionatetothesizeoftheissueandinpartbecausetheissuemaybeheldbyasmallersetofinvestors.
Whatevertheexplanation,theimplicationisthatthesmallerbankingorganizationsinthesamplestudiedmaybelesslikelytoissuepreferredstockormandatoryconvertibledebtthantoexpandthesizeoftheiroutstandingcommonstockissue.
TheEmpiricalModelSpecification.
Themodelofsecuritychoicepresent-edhereusesmultinomiallogit.
Roughly,themodelmaybethoughtofassimultaneouslyestimatinglin-earregressionmodelstoestimatetheprobabilitythataparticulartypeofsecuritywillbeissued(seetheboxforspecifications).
Inthiscasetheconcernistoexplainthedecisiontoissueoneofthreesecurities:commonstock,preferredstock,ormandatoryconvertibledebt.
Becauseinamultinomiallogitmodeloneoftheoutcomesisdeter-minedbythedecisionmadeforalltheotheroutcomes,themodelrequiresspecifyingoneofthepossiblechoicesasthebasecaseandconsideringtheprobabilityoftheothercasesrelativetothebasecase.
Forexample,ifabankdecidesnottoissuecommonorpreferredstockthenitmustissuemandatoryconvertibledebt.
Mandatorycon-vertibledebt,thefocusofthisarticle,isthebasecaseinthemodeldevelopedbelow.
9FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998Nogenerallyacceptedformalmodelincorporatesallthefactorsdiscussedinthetheorysectiontoexplaincor-poratesecurityissuancedecisions.
Theresearchreportedherefollowspriorstudiesofnonbankingcorporations'securityissues,mostnotablyMarsh(1982)andJung,Kim,andStulz(1996),indevelopingempiricalproxiesforthetheoreticalconcepts.
Thediscussiondecomposesthesecurityissuancedecisionintofourparts:taxes,financialdistress,securitytimingandpeckingorder,andcostsrelatedtoissuesize.
Table1providesasummaryofthevariablesandpredictedsignsdiscussedinthissection.
ProxyforTaxes.
Taxesmayaffectthecapitalstructuredecisionsincetheissuanceofadebtsecurity,vis-à-visanequitysecurity,hasdifferenttaximplicationsfortheissuer.
Becauseinterestisdeductiblefortaxpur-poses,theuseofdebtfinancinggenerallyincreasesthevalueofthefirm.
Thegreatertheeffectivetaxrate(ETR),themorevaluablethetaxdeductionand,hence,thelesslikelythefirmistoissueeithertypeofequity.
Thus,theexpectedsignonthecoefficientsonETRisnegativeintheequationsforboththeprobabilityofissuingpreferredstockandtheprobabilityofissuingcommonstockrela-tivetomandatoryconvertibledebt.
ProxiesforFinancialDistress.
Therelevantmea-sureoffinancialdistresscostsforthepurposesofdeter-miningoptimalsecurityissuanceistheexpectedcostsborneexantebythefirm'sexistingshareholders.
Thesecostsincludethosebornebythefirm'sprivatecreditors,giventhatthesecreditorsdemandahigherinterestratetocompensateforhigherrisklevels.
Theprobabilityofdistressisaffectedbythefirm'sbusinessrisk,whichisinturnaffectedbyrevenueriskandoperatingrisk.
Ideally,thebusinessriskismeasuredbysomevariablethatisindependentofcapitalstructure;forexample,goodproxiesforindustrialfirmswouldbethehistoricalvariabilityofsalesandoperatingearnings.
However,forbankingfirmswhose"production"isrelatedtothemanagementofinterestraterisk,theriskassoci-atedwithoperationsismorecomplex.
Sincethereisnocomprehensivemeasureofthebusinessriskofabankthatisindependentofcapitalstructure,thevariabilityinpretaxoperatingincome(VOI)overthepriorfouryearsisusedasariskmeasure.
Theproblemofvariabilityinprofitabilityaffectingcapitalstructureissomewhatmiti-gatedbythefactthatbankcapitalstructuresvarywithinnarrowboundsrelativetoindustrialcorporations.
TheexpectedsignofthecoefficientonVOIispositiveinboththepreferredandcommonstockequations.
Anotheraspectofabank'sriskisthatinducedbythecapitalstructure.
Oneproxyforabank'sfinancial14.
Jung,Kim,andStulz(1996)suggestthatowner/manageragencyconflictsresultinsomenonbankfirmsissuingequity.
Theirreasoningisthatsomemanagersmaywanttoundertakecertainprojectseventhoughtheprojectshaveanegativenetpresentvalue(NPV)toshareholders.
Assumingthesemanagershaveeffectivecontrolofthefirm,theirprimaryconcernishowbesttofundthenegativeNPVprojects.
Iftheprojectisfundedwithdebt,thentheexpectedvalueofthepaymentofinterestandprincipalonthedebtmayexceedtheexpectedreturnsonthenegativeNPVprojects,eventuallyresultinginashrinkageoftheresourcesunderthemanager'scontrol.
However,iftheprojectsarefundedbyequity,thenthecombinationofthenewequityandnegativeNPVprojectsneednotreducetheseresources.
Theimplicationsofthishypothesis,assumingitistrue,forregulatingbanks'equitycapitalareunclear.
Ontheonehand,capitalregulationmayreducethecosttomanagersofissuingnewequitybyallowingthemtoclaimtoskepticalshareholdersthatanewissueisrequiredtomeetregulatoryrequirements.
However,regulatoryequitycapitalminimumsmayalsoincreasethecoststomanagersiftheirfirmshouldincurasubstantiallossandfallintoviolationofthecapitalrequirements.
Thechoiceofissuingoneofthethreeformsofprimarycapitalisapolychotomous,discretedecision.
Thedecisionismodeledsuchthattheprobabilitiesthatbankichoosestoissuemandatoryconvertibledebt(m),pre-ferredstock(p),orcommonstock(c)arerepresentedbyPmi,Ppi,andPci,respectively,suchthatPmi+Ppi+Pci=1.
Theprobabilityofissuerichoosingoneformofcapi-tal—forexample,preferredstock—canbecharacterizedasPmi=Pr[Choosepreferredstock|Factorsaffectingchoice],(1)or,alternatively,Pmi=F(Xij),(2)whereXisthevectoroffactorsthatinfluencethechoiceforbanki;jisthecoefficientmatrixforfactorsforeachjalternativeformofcapital;Pristheprobabilityoperator;andFisthecumulativedensityfunction.
Eachequationisestimatedcross-sectionallyusingthemultinomiallogitpackageofLIMDEP.
BOXEstimationTechnique10FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998riskbasedonaflowmeasureofitscapitalpositionisitsabilitytomeetrecurring,fixedchargesforwhichthefixedcoverageratio(FCR)istheproxy.
Thegreatertheabilityofthefirmtosatisfyitsfixedfinancialobliga-tions(thatis,thegreaterthefinancialrisk),thelowertheprobabilityoffinancialdistressandthelesslikelythebankistoissuecommonorpreferredstock.
15Analternativewaytomeasuretheriskinducedbyabank'scapitalstructureisbasedonitsstockofcapi-tal.
Ameasureofthecapitalstructureistheratioofthebookvalueofthefirm'scommonequitytothebookvalueofitstotalassets,BEA.
Theexpectedsignonthecoefficientsforbothtypesofequityissuancerelativetomandatoryconvertibledebtissuanceisnegative;thatis,ahigherexistingequityratioimplieslowerprobabilitylevelsofissuingcommonorpreferredstock.
Bankshareholdersareconcernedaboutthatportionofbankruptcycoststhatisbornebythefirm'sprivatecreditorssincehigherbankruptcycostsimplygreaterriskpremiumsonthebank'soutstandingdebt.
Theshareofthecostsbornebyprivatecreditorsdependsinlargepartontheextenttowhichabank'sliabilitiesarecoveredbydejureordefactodepositinsurance.
Thus,bankswithhighlevelsofexplicitandimplicitinsurancearelikelytofacesignificantlylowerprivatecostsoffinancialdis-tressthanbankswithlowerlevelsofcoverage.
Thisstudyusedtwoproxiesforthelevelofcoverage.
Theproxyforexplicitcoverageistheratioofuninsuredlia-TABLE1ExplanatoryVariablesVariablesDefinitionsExpectedSignPp/PmPc/PmETREffectivetaxrateforthemostrecentyear––VOIStandarddeviationoftheratioofpretaxoperatingincome++(totalassets)overthepriorfouryearsFCRFixedchargecoverageratio,evaluatedatthemostrecentfiscal––year-endBEABookvalueofcommonequitydividedbybookvalueoftotalassets––UNLRatioofuninsuredliabilitiestototalassetsforthemostrecent++fiscalyear-endTBFBinaryvariablethathasavalueof1iftheissuerisoneoftheten––largestbanks,0ifotherwiseB1DUMBinaryvariablethatequals1iftheratioofthemarketvalueof––equitytothebookvalueofequityattheendofthemostrecentfiscalyear-endislessthan1and0ifotherwiseB1MBKRatioofthemarketvalueofequitytothebookvalueofequityat++theendofthemostrecentfiscalyear-endifthemarket-to-bookratioislessthan1and0ifotherwiseA1MBKRatioofthemarketvalueofequitytothebookvalueofequityat++theendofthemostrecentfiscalyear-endifthemarket-to-bookratioisgreaterthanorequalto1and0ifotherwisePCHPricechangeofthecommonstockoverthemostrecentfiscalyear++precedingtheissuanceannouncementCSIBinaryvariablethathasavalueof1iftheissuerissuedcommon––stockwithinthepasttwelvemonthsand0ifotherwisePSIBinaryvariablethathasavalueof1iftheissuerissuedpreferred–+stockwithinthepasttwelvemonthsand0ifotherwiseMCIBinaryvariablethathasavalueof1iftheissuerissuedmandatory++convertibledebtwithinthepasttwelvemonthsand0ifotherwiseLGMKTLogofthemarketvalueoftheissuer'scommonequity–11FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998bilitiestototalassets(UNL).
Thisratioisexpectedtohaveapositiveeffectontheprobabilitythatabankwillissueeithertypeofequityrelativetotheprobabilityitwillissuemandatoryconvertibledebt.
Theproxyfortheimplicitcoveragegrantedlargebanks,TBF,isabinaryvariableequaltooneifthebankisoneofthetenlargestandzerootherwise.
Thisvariableisaproxyforthefactthatthelargestbanksareconsidered"toobigtofail"andhencearelikelytohave100percentdefactodepositinsurance.
ThecoefficientsonTBFinthecom-monandpreferredequityequationsareexpectedtohavenegativesigns.
16ProxiesforSecurityIssuanceTimingandPeckingOrder.
Oneproxyfortimingconsiderationsistheratioofthebank'smarketvaluetoitsbookvalue.
Bankman-agersclaimtobereluctanttoissuecommonequitywhenthisratioislessthanone(WallandPeterson1991),sayingthatitresultsindilution.
17Thisobjectionisnotsupportedbyfinancetheory,however.
Thebookvalueofafirm'sstockisirrelevanttoitsfinancialdeci-sions,andstockshouldbeissuedifthenetpresentvalueofadditionalinvestmentsresultingfromtheissueexceedsthevalueofthestockissued.
Market-to-bookratiosmayhavegreaterrelevanceforbankingbecausemostbankassetsareshort-termfinancialassetswhosemarketvalueshouldbeclosetotheirbookvalue.
Abank'shavingamarket-to-bookvaluebelowonesuggestsexpostthatitsmanagementhasmadebaddecisions,andthemarketmaybereluctanttogivethesemanagersaddition-alcapital.
Thus,bankswithlowmarket-to-bookratiosmaybelesslikelytoissuenewequity,especiallynewcom-monstock.
Twovariablesareusedasproxiesfortheeffectofmarket-to-bookratiosoflessthanone:abinaryvariabletocaptureanyleveleffectsofaratiolessthanoneandaslopetermforbankswitharatiooflessthanone.
Specifically,thebinaryvariableB1DUMtakesavalueofoneifthemarket-to-bookratioislessthanoneandzerootherwise,andtheslopevariableB1MBKtakesavalueofthebank'smarket-to-bookvalueiftheratioislessthanone.
BothB1DUMandB1MBKhaveavalueofzeroforbanksabovethestandards.
Bankswithratiosthatarebelowonearelesslikelytoissueequity,sotheexpectedsignonB1DUMisnegativeinbothequations.
However,asB1MBKincreasestowardavalueofone,theprobabilityofissuingequitymayincrease,suggestingthatthecoeffi-cientsonB1MBKinbothequationsmaybeexpectedtohaveapositivesign.
Banksmayalsotakeaccountoftheirmarket-to-bookratioinmakingsecurityissuancedeci-sionsifthisratioisgreaterthanone.
Thus,thevariableA1MBKisalsoincludedinthemodel,whereA1MBKtakesavalueofthebank'smarket-to-bookratioiftheratioisgreaterthanorequaltooneandzerootherwise.
ThesignofthecoefficientsonA1MBK,likethatonB1MBK,isexpectedtobepositive.
Anothermeasureofwhetherabank'sstockmaybeperceivedbyabank'smanagerstobeover-orundervaluedistherecentmovementinitsstockprice.
Ifmanagement'spercep-tionofabank'svaluechangesmoreslowlythanthemarket's,thengreaterlevelsofstockpriceappreciationmaybeassociatedwithahigherprobabilitythatmanagementperceivesthebank'sstocktobeovervalued.
AproxyfortherecentpricechangeinthestockisPCH,whichisthepricechangeofthecommonstockoverthemostrecentfiscalyearprecedingtheissuanceannouncement.
TheexpectedsignofthecoefficientonPCHispositiveforbothtypesofequity.
Animplicationofthepeckingorderhypothesisisthattheprobabilitythataparticulartypeofsecurityisissuedmayberelatedtoitsownpastissuance.
Threedummyvariablesdesignatepreviousissuanceswithinthelasttwelvemonths:CSI(issuanceofcommonstock),PSI(issuanceofpreferredstock),andMCI(issuanceofmandatoryconvertibledebt).
Ifpreferredstockissueistreatedassomethingbetweencommonstockandmanda-toryconvertibledebt,thepeckingorderhypothesisdeliv-ersunambiguoussignsfortheprobabilityofissuingcommonandpreferredstockrelativetotheprobabilityofissuingmandatoryconvertibledebt.
18Underthepeckingorderhypothesistheprobabilityofissuingcommon15.
Thecoverageratiomayalsobeinterpretedasameasureofthebank'sfreecashflow.
Thecostofissuingnewpreferredstockormandatoryconvertibledebtmaybereducedtotheextentthatitreducesthebank'sfreecashflow.
Thefree-cash-flowinterpretationofthecoverageratiosyieldsthesamepredictionastheriskinterpretationoftheratios:theprobabilityofafirmissuingdebtorpreferredstockisexpectedtobeanegativefunctionofFCR.
16.
Theexactsizecutofffortoo-big-to-failstatusisunknownandmaychangeovertime.
However,thetenlargestbanksmayberegardedasareasonableproxyformembershipinthiselitegroup.
17.
OsbornandEvansgiveanexampleofthecommonviewthatbanksshouldnotissuestockatpricesbelowbookvalue:"Equityissuesaredifficultforthemoneycenterbankssincemostaretradingbelowbookvalue"(1988,47).
18.
Preferredstockmaybethoughtofasanintermediatecasebecause,likedebt,itcommitshealthybankstomakingafixedannu-alpaymentandbecause,likecommonstock,itpermitsthefirmtosuspendpaymentsintimesofseverefinancialdistress.
Banksposespecialprob-lemsintermsofdebtmaturitybecausealargefractionoftheirassetsisinvestedinassetsthateitherhaveashortmatu-rityoraretradedinliquidmarketsorboth.
12FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998stockrelativetotheprobabilityofissuingmandatoryconvertibledebtisapositivefunctionofMCIandPSIandanegativefunctionofCSI.
Also,theprobabilityofissuingpreferredstockisapositivefunctionofMCIandanegativefunctionofPSIandCSI.
ProxiesforRelativeCostsofIssuance.
Thecostsofissuingnewsecuritytypesarehypothesizedtobeadecreasingfunctionofabank'ssize.
Giventhatallbanksinthesamplehavepubliclytradedcommonstock,thishy-pothesisimpliesthattheprobabilityofissuingpreferredstockandmandatoryconvertibledebtisanincreasingfunctionoffirmsize.
Aproxyforfirmsizeisthemarketvalueofthefirm'soutstandingcommonequity.
However,theeffectofbanksizeonissuancecostislikelytodecreasewithsize,implyinganonlinearrelationshipbetweenthesizeofthefirmandthecostofissuance.
Thus,thenaturallogofbanks'marketvalue(LGMKT)isusedasaproxyfortherelativecostofissuance.
Thisproxyisexpectedtohaveanegativecoefficientontheprobabilityofissuingcommonstock,buttheexpectedrelationshipwiththeprobabilityofissuingpreferredstockrelativetoissuingmandatorycon-vertibledebtisambiguous.
MethodologyandData.
Theboxprovidesthespec-ificationofthemodelandadiscussionofthemultinomi-allogit.
Thesampleconsistsofstockissuancesfrom1983through1986.
ThesampleofbankingorganizationsistakenfromthebanksincludedintheexpandedannualindustrialdatafilesofStandardandPoor'sCompustat.
19AllaccountingdataexceptformandatoryconvertibledebtoutstandingandtheprimarycapitalratioareobtainedonanannualbasisfromtheBankHoldingCompanyFinancialSupplement(FRY-9),collectedbytheFederalReserveSystem,andfromCompustat.
MarketvaluationdataareobtainedfromtheUniversityofChicago'sCenterforResearchinSecurityPrices(CRSP)databases.
DataonthetimingandamountofsecuritiesissuedareobtainedfromIrvingTrust'sCapitalSecuritiesIssued:CommercialBankingandLehmanBrothers'FinancingsbyUnitedStatesBanksandBankHoldingCompaniessince1976for1983through1987.
WhenIrvingTrust'spub-licationwasinadequatefordeterminingwhetheradebtissuedqualifiedasmandatoryconvertible,thestatusofthesecurityissuewasalsoreviewedinMoody'sBankingandFinanceManualandindividualbankingorganiza-tions'annualreports.
Thefinalsampleconsistsof152observations.
Table2providesabreakdownofthesamplebysecuritytypeandyear.
Table3presentsdescriptivestatisticsforeachofthecontinuousvariablesforeachtypeofissuance:com-monstock,preferredstock,andmandatoryconvertibledebt.
Thevalueofthemarket-to-bookratio(MBK)ispre-sentedratherthanB1DUM,B1MKT,andA1MKT.
Thesesubstitutionsfacilitatecomparisonoftheaveragecapi-talpositionandaveragemarket-to-bookratioofthethreetypesofissuance.
NotealsothatthemeanvaluesofthevariablesTBF,CSI,PSI,andMCImaybeinterpret-edastheproportionofbanksthataretoobigtofail,thathaveissuedcommonstock,thathaveissuedpreferredstock,andthathaveissuedmandatoryconvertibledebt,respectively.
Severalcross-sectionaldifferencesstandoutinTable3.
First,mandatoryconvertibledebthasthehigh-esteffectivetaxrate.
Second,mandatoryconvertibledebtismorelikepreferredstockthanlikecommonstockalongmanydimensions,includingvolatilityofoperatingincome,deviationfromregulatoryprimarycapitalrequirements,uninsuredliabilities,too-big-to-failstatus,andmarket-to-bookandmarketvalue.
EstimationResultsLogitRegression.
ThelogitestimationresultsappearinTable4.
Thetableprovidestheestimatedcoeffi-cientsandtheirt-statisticsforeachrelativeproba-bility(relativetoissuingmandatoryconvertibledebt).
Theexplanatorypowerofthemodelisstatisticallysignificant,withthepercentageofcorrectlypredictedwithin-samplecasesof61.
18.
Allcoefficientsareinsignif-icantintheequationestimatingtheprobabilityofissuingpreferredstockrelativetoissuingmandatoryconvertibledebt.
Moreover,amongtheobservationsinwhichbanksactuallyissuedpreferredstock,themodelpredictedthat43.
59percentwouldhaveissuedmandatorycon-vertibledebtwhereasitpredictedthatonly33.
33per-centwouldhaveissuedpreferredstock.
20TABLE2SamplebyTypeofCapitalandYearofIssuanceMandatoryConvertibleDebtPreferredCommonYearIssuanceStockIssuanceStockIssuanceTotalSample1983114132819842981047198519813401986892037Total57395615213FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter199819.
Thedatafilesincludetheprimary,secondary,tertiary,fullcoverage,andresearchfiles.
Standard&Poor'sindicatesthatthisuniversecontainsallbankswith"significantinvestorinterest.
"20.
Theremainingobservationswerepredictedtohaveissuedcommonstock.
Thesignandsignificanceofseveralvariablesinthecommonstockequationareconsistentwiththetimingoftheissuanceofsecurities.
Thebinaryvariableforabankhavingamarket-to-bookratiooflessthanone(B1DUM)ismarginallysignificantlynegative(significantonlyatthe10percentlevel),suggestingthatsuchbanksmaybemorelikelytoissuecommonstock.
Thecoefficientonthemarket-to-bookratioofbankswitharatiooflessthanone(B1MBK)issignificantlypositive,suggestingthatbanksaremorewillingtoissuecommonstockastheirmarket-to-bookratioincreases.
Further,thecoeffi-cientsonbothpreviouspreferredstockissuance(PSI)andpriormandatoryconvertibleissuance(MCI)arepositive,suggestingthatbanksswitchtoissuingcom-monstockafterexploitingopportunitiestoissuepre-ferredstockandmandatoryconvertibledebt.
Thecoefficientonthelogofthefirm'smarketvalueissignificantwithanegativesign,suggestingthatsmall-erbanksaremorelikelytoissuecommonstockthanmandatoryconvertibledebt.
Thisresultsupportsthehypothesisthatmandatoryconvertibleissuesaremoreexpensiveforsmallerbanks.
Overall,theseresultssupportpriorfindingsinsug-gestingthatallowingbankstoissuedebtratherthanequi-tymayreducetheircostsofcomplyingwiththecapitalstandards.
Inparticular,theseresultssupportthehypoth-esisthatallowingbankstoissuedebtmayreducethecoststogoodbanksofbeingpooledwithbadbanks.
TABLE3DescriptiveStatisticsbyTypeofCapitalaMandatoryConvertibleDebtPreferredStockCommonStockVariableIssuanceIssuanceIssuanceETR0.
25420.
21560.
1656(0.
1364)(0.
2725)(0.
1297)VOI0.
0098820.
010490.
008266(0.
003098)(0.
003353)(0.
003658)FCR1.
1411.
12091.
158(0.
0614)(0.
04225)(0.
05962)BEA0.
046990.
048420.
05285(0.
009169)(0.
009867)(0.
009454)UNL0.
51580.
48630.
3593(0.
2134)(0.
2307)(0.
1902)TBF0.
36840.
30770.
07142(0.
4867)(0.
4676)(0.
2599)MBK0.
94620.
88421.
083(0.
2005)(0.
2400)(0.
2287)PCH0.
25600.
23560.
3144(0.
2333)(0.
1663)(0.
1878)CSI0.
15790.
23080.
1607(0.
3679)(0.
4268)(0.
3706)PSI0.
17540.
15380.
1607(0.
3837)(0.
3655)(0.
3706)MCI0.
22810.
15380.
1786(0.
4233)(0.
3655)(0.
3865)MKT1682.
1588.
550.
7(1688.
)(1915.
)(878.
3)aMeanvalues;standarderrorsareinparentheses14FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998TABLE4MultipleLogitRegressionEstimationResultsVariableLog(Pp/Pm)Log(Pc/Pm)ETR1.
10250.
1782(0.
5350)(0.
9287)VOI85.
41150.
94(0.
3856)(0.
6084)FCR–13.
227–0.
2140(0.
1029)(0.
97691)BEA53.
949.
949(0.
2151)(0.
8126)UNL–0.
6195–0.
4683(0.
7882)(0.
8315)TBF0.
73561.
499(0.
4735)(0.
1968)B1DUM2.
565–7.
879(0.
5207)(0.
0548)cB1MBK0.
168010.
82(0.
9473)(0.
0019)aA1MBK2.
4502.
209(0.
4047)(0.
3813)PCH–0.
7419–0.
1272(0.
5538)(0.
9119)CSI0.
78830.
2213(0.
2056)(0.
7362)PSI0.
15402.
020(0.
83012)(0.
0120)bMCI0.
063391.
5462(0.
9309)(0.
0400)bLGMKT–0.
3990–1.
914(0.
3511)(0.
00003)aConstant10.
888.
700(0.
2283)(0.
2599)LogL–128.
6018X2(28)72.
55099PercentPredicted61.
18Theprobabilitythatthecoefficientisnotequaltozeroinatwo-tailedt-testisshowninparenthesesbelowthecoefficient.
LogListhelogofthelikelihoodatmaximum,X2istheChi-squareddistributedstatisticforthetestofallnoninterceptcoefficientsnotequaltozero,andpercentpredictedisthepercentageofcorrectlypredictedwithin-samplecases,basedonthelargestprobabilityusingesti-matedcoefficients.
TheChi-squaredteststatisticissignificantatthe5percentlevelofsignificance.
aindicatescoefficientdifferentfromzeroatthe1percentlevelofsignificancebindicatescoefficientdifferentfromzeroatthe5percentlevelofsignificancecindicatescoefficientdifferentfromzeroatthe10percentlevelofsignificance15FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter199821.
DepositorpreferencewaspassedasapartoftheOmnibusBudgetReconciliationActof1993.
Underthisprovision,allofthedepositorsatabank,insuredanduninsured,wouldbeplacedaheadofallnondepositliabilityholdersintheeventofabank'sfailure.
ThisprovisionreducestheFDIC'sexpectedlossesbecauseintheeventtheagencymakespaymentstodepos-itorsafterabankfailuretheFDICassumesthesamepriorityclaimontheremainingassetsasthedepositorsdid.
However,theseresultsalsosuggestthatthebenefitsofexpandingthedefinitionofcapitalwillnotaccrueequal-lytoallbanks.
Largerbanksthatcanissuesufficientvol-umesofnewsecuritiesaremorelikelytosubstitutedebtforequitythanaresmallerbanks.
Further,theestimatedmodelprovidesnoinsightintobanks'choiceofmandato-ryconvertibledebtversuspreferredstock.
OptimalCapitalStructureandRegulatoryConcernsThesurveyoftheoreticalanalysesabovesuggeststhatmostoftheprivatecostsandbenefitsassociat-edwithdifferentcapitalstructuresarisebecauseofthedifferencesbetweendebtandequity.
ThesurveyofempiricalresultsinWallandPeterson(1996)aswellasthenewresultspresentedintheprevioussectionsuggestthatasignificantpartofthiscosttakestheformoftrans-fersfromgoodbankstobadbanks.
Theimplicationisthatregulatorscouldminimizethecostofmeetingthecapitalguidelinestogoodbanksbyallowingbankstosubstituteuninsureddebtforequity.
Thecapitalstandardswouldnotbecostlesstobanksbecausehighercapitalstandardswouldstillreducethedepositinsurancesubsidytorisktaking.
However,thecapitalstandardsmayimposelittleornoadditionalprivatecoststotheextentthattheyallowfirmstousedebtratherthanequity.
Uninsureddebtisapotentiallyviablesubstituteforequityinlimitingdepositinsurancelosses.
Intheory,allnondepositliabilitiesbecameabuffertotheinsurancefundwiththeenactmentofdepositorpreferencein1993.
21Further,alldepositorswithmorethan$100,000ondepositshouldshareanyremaininglosseswiththeFDICundertheleastcostlyresolutionprovisionsofFDICIA.
However,cap-italregulationscontinuetofocusonalimitedsetofequityanddebtobligations.
Onepossibleexplanationforthecontinuingfocusonequityandcertaindebtcontractsisconcernabouttheextenttowhichdepositsover$100,000andnonsubordi-natedliabilitieswouldreduceFDIClossesintheeventofafailure.
TheFDICmay,withtheconcurrenceoftheSecretaryoftheTreasuryandtheFederalReserveBoard,extenddepositinsurancetodepositsover$100,000.
Further,nondepositliabilitiesthatarenotcontractuallysubordinatedtodepositsmaybegivencollateraltoreducethelossesontheseclaimsshouldthebankfail.
Horvitz(1984)andBenstonandothers(1986)aswellasrecentspeechesbyFederalReserveGovernorFerguson(1998)suggestanalternativethatdoesnothavetheproblemsassociatedwithdepositorpreference.
Theyrecommendtheincreaseduseofatypeofdebtcalledsub-ordinateddebt—debtthatisjuniororsubordinatedtoallotherliabilitiesifabankshouldfail.
Ifsubordinateddebtissuchaneasysolution,whydoregulatorsnotallowbankstosubstituteitforequityRegulatorystandardshaveinfactallowedpartialsubstitution.
Boththe1981standardsandthecurrentstandardsallowsubordinateddebtasanelementoftotalcapital.
However,bothstandardslimitthesubstitutionbyimposingadditionalrequirementsforanarrowerdefinitionofcapitalthatdoesnotincludeordi-narysubordinateddebt.
Theserequirementsarethepri-marycapitalguidelinesunderthe1981standardsandthecurrenttier1risk-basedandleveragestandards.
Thus,therealquestioniswhyregulatorsdonotallowunlimitedsub-stitution.
Threepossibleobjectionsexisttotheuseofsub-ordinateddebt.
Twooftheseobjectionsmaybeeasilyaddressedwithinthecontextofthestandardsandtheirimplementation.
Thethirdismorefundamental.
ThefirstobjectionisthatsubordinateddebtmaynotprotecttheFDIC.
Subordinateddebtdoesnothavedejuredepositinsurancecoverage,butsubordinated-debthold-ershavereceiveddefactoinsurancecoverageduringsomepriorbankfailures,suchasthatofContinentalIllinoisinJuly1984.
FlanneryandSorescu(1996)examinetheex-tenttowhichsubordinatedobligationsofbankingorgani-zationsreflectedtheriskinessoftheissuingorganizationbetween1983and1991.
TheirfindingssuggestthatthepricesearlyintheirsampleperiodembedasignificantprobabilitythattheFDICwouldextenditscoveragetoincludeuninsureddepositors.
Thesolutiontoproblemsposedbydefactoinsur-ancecoverageissimple;however,theFDICshouldnotextenddepositinsurancetocoversubordinatedliabili-ties.
Indeed,inmorerecentfailurestheFDIChasnotcoveredsubordinated-debtholdersatfailedbanks.
ConsistentwiththechangeinFDICpoliciesisFlanneryandSorescu'sfindingthatsubordinated-debtholderspricedindividualbankingorganizations'defaultriskdur-ingthelaterpartoftheirsampleperiod.
Asecondobjectionisthatthematuritystructureofdebtmayalsobeimportantindeterminingbanks'behav-ior.
Asnotedabove,Flanneryarguesthatthematurityofabank'sdebtobligationsisimportantinminimizingcon-flictsbetweenownersandcreditors.
Banksposespecialproblemsintermsofdebtmaturitybecausealargefrac-tionoftheirassetsisinvestedinassetsthateitherhaveashortmaturityoraretradedinliquidmarketsorboth.
Thus,banksareinapositiontosubstantiallychangetheriskinessoftheirinvestmentportfolioinamatterofmonthsordays(orperhapsevenhoursinafewcases).
Yettheregulationsforsubordinateddebttobeincluded22.
RedemptionofthesubordinateddebtinWall'sproposaliscontingentonthebankremainingincompliancewiththecapi-talstandardsafterredemption.
Thus,thesubordinated-debtholderscouldnotavoidtakinglossesfromabank'sfailuremerelybyrequestingredemptionimmediatelypriortoitsfailure.
23.
Forasurveyofthisliteraturearguingthatthemacrocostsofbankfailureneednotbehigh,seeBenstonandKaufman(1995).
16FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998incapitalratiosgenerallyrequirethatthedebthaveanaveragematurityatissuanceofatleastfiveyears.
Onceagain,asolutionseemsclear:alloworrequirebankstoissuesubordinateddebtwithashortmaturity.
Benstonandothers(1986)advocatethatbanksregularlyhavesubordinateddebtissuesrollingoverandthatsomesmallpercentagemightberedeemable.
Evanoff(1992)developedaproposalinwhichpartoftheoutstandingsub-ordinateddebtmaturesonaregularbasis(suchaseverysixmonths).
Calomiris(1997,1998)providesforbothreg-ularrolloversandlimitsontheratethedebtcouldpayabovetherisklessrateofinterest.
Wall(1989)developedanentireproposalhecalledputtablesubordinateddebtthatwouldallowsubordinated-debtholdersto"put"theirdebtbacktothebank,ineffectsimulatingthedisciplineimposedbydemanddepositorsintheabsenceofdepositinsurance.
22Thesevariousproposalsforredemptionofsubordinateddebteitheratregularintervalsorupondemandbysubordinatedcreditorswouldallowsubordi-natedcreditorstoeffectivelysubstitutedebtforequityinprotectingtheFDICwhilegivingsubordinatedcreditorsamechanismforprotectingtheirowninterestsfromrisk-increasingstrategiesbyequityholders.
Thethirdpossibleobjectiontosubordinateddebtarisesfromthegoalofcapitalrequirements.
IfthegoalofcapitalrequirementsistoprotecttheFDIC,thenitispos-sibletostructuresubordinatedobligationsthatwillfulfillthisobjective.
However,subordinatedobligationsareunlikelytohelpifthegoalofcapitalrequirementsistoreducetheprobabilityoffailureafterabankhasincurredsignificantlosses.
Subordinateddebtdoesnotprovideacushionthatcanabsorblosseswithoutcausingfailure.
Ifthepromisedpaymentstosubordinatedcreditorsarenotmadeinatimelymanner,thenthebankisilliquidandwillbeclosed.
Allowingorrequiringbankstoissuesubordinat-edobligationsthathaveashortmaturity,thatarepartial-lyrolledoveronaregularbasis,orthatareputtableonlyincreasestheriskthatobligatedpaymentstosubordinated-debtholderswillpushaweakbankintofailure.
However,theargumentthatsubordinateddebtincreasesabank'sprobabilityoffailureafteritincursalargelossdoesnotnecessarilyimplythatsubstitutingsub-ordinateddebtforequitywouldmakethebankingsystemlessstable.
Equityholdersreceiveboththelargerpayoutassociatedwithrisksthatsucceedaswellaspartofthelossesifthegamblefails.
Subordinated-debtholderscan-notobtainahigherrateofreturnthantheirpromisedinterestratebutareexposedtofailedgambles.
Thus,sub-ordinatedcreditorsarelikelytoprovidegreaterincen-tivesforbankstoavoidtakingexcessiverisksexante.
Thus,Horvitz(1984)pointsoutthatgreaterrelianceonsubordinateddebtislikelytoreducetheexanteproba-bilitythatabankwilltakeexcessiverisksthatwouldraisetheprobabilityofitsfailure.
Moreover,whyshouldregulatorscareaboutthefailureofanindividualbankThefailureofanyindividualbankisnotapublicpolicyproblemperse.
AbankfailurebecomesaproblemonlyifitcausessignificantlossestotheFDICorsignificantlyreducesaggregatereal(nonfinancial)eco-nomicactivity.
Properlystructuredsubordinateddebtpro-tectstheFDICfromlossesinamannersimilartoequityatinsolventbanks.
Moreover,avarietyofstudieshaveexam-inedtheconsequencesofbankfailurefortherealeconomy,andmanyofthesestudiesarguethatmostoftheadverseconsequencesofabank'sfailurefortherealeconomymaybeoffsetwithappropriatemonetarypolicy.
23ConclusionBanksaroundtheworldareorhavebeenunderintenseregulatorypressuretoraiseormaintaincapitallevelsinresponsetointernationalrisk-basedcapitalguidelines.
ThisarticleexaminesthefactorsthatdeterminethetypeofcapitalbankingorganizationswillraisebystudyingU.
S.
banks'responsetotheprimarycapitalguidelinesannouncedinDecember1981.
Theempiricalfindingssuggestthatasymmetricinformationandthecostsassociatedwithsmallissuesizeareimportantdeterminantsofthesecurityissuancedecision.
Bankregulatorsmayreducethecostofasym-metricinformationbyallowingbankstoissuequalifyingdebtsecuritiestocomplywithallpartsofthecapitalreg-ulation.
However,totheextentthatthecostofissuingnewtypesofsecuritiesishigh,sucharegulatorychangemaybeoflittlevaluetosmallerbankingorganizations.
Giventhepotentialofsubordinateddebttoreducethecostsofregulatorycomplianceforatleastsomebanks,whatjustificationmightbegivenfortheexistingfocusonequitycapitalThisdiscussionconsidersthreepossiblereasons:subordinateddebtmaynotprotecttheFDIC,thematuritystructureofdebtisimportantinminimizingthecostsofconflictsbetweenownersandcreditors,andsub-ordinateddebtisunabletoreducetheprobabilityofabank'sfailureafteritabsorbssubstantiallosses.
Thefirsttwoobjectionsmaybeeasilyaddressedduringtheregu-latoryimplementationofnewrulespermittingtheuseofsubordinateddebt.
Thethirdobjectionholds,butitignorestheroleofsubordinateddebtinreducingtheprobabilitythatabankwillincursubstantiallossesandtheroleofothermechanismsinlimitingtheimpactofabank'sfailureontherealeconomy.
17FederalReserveBankofAtlantaECONOMICREVIEWSecondQuarter1998BARNEA,AMIR,ROBERTA.
HAUGEN,ANDLEMMAW.
SENBET.
1981.
"MarketImperfections,AgencyProblems,andCapitalStructure.
"FinancialManagement10:7–22.
BAXTER,NEVINS.
1967.
"Leverage,RiskofRuin,andtheCostofCapital.
"JournalofFinance22:395–403.
BENSTON,GEORGEJ.
,ROBERTA.
EISENBEIS,PAULM.
HORVITZ,EDWARDJ.
KANE,ANDGEORGEG.
KAUFMAN.
1986.
PerspectivesonSafeandSoundBanking.
Cambridge,Mass.
:MITPress.
BENSTON,GEORGEJ.
,ANDGEORGEG.
KAUFMAN.
1995.
"IstheBankingandPaymentsSystemFragile"JournalofFinancialServicesResearch9:209–40.
CALOMIRIS,CHARLESW.
1997.
ThePostmodernBankSafetyNet:LessonsfromDevelopedandDevelopingEconomies.
Washington,D.
C.
:AEIPress.
———.
1998.
"BuildinganIncentiveCompatibleSafetyNet:SpecialProblemsforDevelopingCountries.
"JournalofBankingandFinance(forthcoming).
DEANGELO,HARRY,ANDRONALDW.
MASULIS.
1980.
"LeverageandDividendIrrelevancyunderCorporateandPersonalTaxation.
"JournalofFinance35:453–64.
EVANOFF,DOUGLASD.
1992.
"PreferredSourcesofMarketDiscipline:Depositorsvs.
SubordinatedDebtHolders.
"FederalReserveBankofChicagoWorkingPaperNo.
21.
FERGUSON,ROGERW.
,JR.
1998.
"TheChangingBankingEnvironmentandEmergingQuestionsforPublicPolicy.
"RemarksbeforetheOhioBankersDayConference,Columbus,Ohio,March19.
FLANNERY,MARKJ.
1994.
"DebtMaturityandtheDeadweightCostofLeverage:OptimallyFinancingBankingFirms.
"AmericanEconomicReview84:320–31.
FLANNERY,MARKJ.
,ANDSORINM.
SORESCU.
1996.
"EvidenceofBankMarketDisciplineinSubordinatedDebentureYields:1983–1991.
"JournalofFinance54:1347–77.
HORVITZ,PAULM.
1984.
"SubordinatedDebtIsKeytoNewBankCapitalRequirement.
"AmericanBanker,December31.
HUGHES,JOSEPHP.
,ANDLORETTAJ.
MESTER.
1994.
"BankManagers'Objectives.
"FederalReserveBankofPhiladelphiaWorkingPaper94-8,April.
JENSEN,MICHAELC.
1986.
"AgencyCostsofFreeCashFlow,CorporateFinance,andTakeovers.
"AmericanEconomicReview76:323–29.
JENSEN,MICHAELC.
,ANDWILLIAMH.
MECKLING.
1976.
"TheoryoftheFirm:ManagerialBehavior,AgencyCosts,andOwnershipStructure.
"JournalofFinancialEconomics3:305–60.
JONES,DAVID,ANDJOHNMINGO.
1998.
"IndustryPracticesinCreditRiskModelingandInternalCapitalAllocations:Impli-cationsforaModels-basedRegulatoryCapitalStandard.
"Paperpresentedatconference,FinancialServicesattheCrossroads,FederalReserveBankofNewYork,February26–27.
JUNG,KOOLYUL,YONG-CHEOLKIM,ANDRENM.
STULZ.
1996.
"Timing,InvestmentOpportunities,ManagerialDiscretion,andtheSecurityIssueDecision.
"JournalofFinancialEconomics42:159–85.
KRAUS,ALAN,ANDROBERTLITZENBERGER.
1973.
"AState-PreferenceModelofOptimalFinancialLeverage.
"JournalofFinance28:911–22.
MARCUS,A.
J.
1983.
"TheBankCapitalDecision:ATimeSeriesCrossAnalysis.
"JournalofFinance38:1217–32.
MARSH,PAUL.
1982.
"TheChoicebetweenEquityandDebt:AnEmpiricalStudy.
"JournalofFinance37:121–44.
MASULIS,RONALDW.
,ANDBRETTTRUEMAN.
1988.
"CorporateInvestmentandDividendDecisionsunderDifferentialPersonalTaxation.
"JournalofFinancialandQuantitativeAnalysis23:369–83.
MILLER,MERTONH.
1995.
"DotheM&MPropositionsApplytoBanks"JournalofBankingandFinance19(June):439–89.
MODIGLIANI,FRANCO,ANDMERTONH.
MILLER.
1958.
"TheCostofCapital,CorporationFinance,andtheTheoryofInvestment.
"AmericanEconomicReview48:261–97.
———.
1963.
CorporateIncomeTaxesandtheCostofCapital.
"AmericanEconomicReview53:433–43.
MYERS,STEWARTC.
,ANDNICHOLASS.
MAJLUF.
1984.
"CorporateFinancingandInvestmentDecisionsWhenFirmsHaveInformationThatInvestorsDoNotHave.
"JournalofFinancialEconomics13:187–221.
ORGLER,YAIRE.
,ANDROBERTA.
TAGGARTJR.
1983.
"ImplicationsofCorporateCapitalStructureTheoryforBankingInstitutions:ANote.
"JournalofMoney,Credit,andBanking15:212–21.
OSBORN,NEIL,ANDGARRYEVANS.
1988.
"Cooke'sMedicine:KillorCure"Euromoney(July):35–54.
PEEK,JOE,ANDERICS.
ROSENGREN.
1997.
"HowWellCapitalizedAreWell-CapitalizedBanks"FederalReserveBankofBostonNewEnglandEconomicReview(September/October):40–50.
SHADOWFINANCIALREGULATORYCOMMITTEE.
1996.
"ShadowFinancialRegulatoryCommitteeStatementsNumbers70–134.
"DecembersupplementinJournalofFinancialServicesResearch110,editedbyGeorgeG.
Kaufman.
SHRIEVES,RONALDE.
,ANDDREWDAHL.
1992.
"TheRelationshipbetweenRiskandCapitalinCommercialBanks.
"JournalofBankingandFinance16:439–57.
WALL,LARRYD.
1989.
"CapitalRequirementsforBanks:ALookatthe1981and1988Standards.
"FederalReserveBankofAtlantaEconomicReview74(March/April):14–29.
———.
1993.
"TooBigtoFailafterFDICIA.
"FederalReserveBankofAtlantaEconomicReview78(January/February):1–14.
WALL,LARRYD.
,ANDPAMELAP.
PETERSON.
1991.
"ValuationEffectsofNewCapitalIssuesbyLargeBankHoldingCompanies.
"JournalofFinancialServicesResearch5:77–87.
———.
1996.
"Banks'ResponsestoBindingRegulatoryCapitalRequirements.
"FederalReserveBankofAtlantaEconomicReview81(March/April):1–17.
REFERENCES