ECB -CBDC、金融穩(wěn)定和央行依賴（英）

Occasional

Paper

SeriesKnowyour(holding)limits:CBDC,financialstabilityandcentralbankrelianceBarbaraMeller,OscarSoonsNo

326Disclaimer:

ThispapershouldnotbereportedasrepresentingtheviewsoftheEuropeanCentralBank(ECB).TheviewsexpressedarethoseoftheauthorsanddonotnecessarilyreflectthoseoftheECB.ContentsAbstract23Executive

Summary12345Introduction4Literature

reviewThe

model91013171820262628293235373742Data

and

descriptive

statisticsSimulation

results5.15.2TheimpactontheEurosystembalancesheetTheimpactonbanks’

balancesheets6Alternative

model

specifications6.16.26.3AnenvironmentwithlowerexcessreservesAninterbankmarketsegmentedacrossnationalbordersA

retailbankrunscenario7ConclusionReferencesAppendixABDetailedmodeldescriptionAdditionalresultsECBOccasionalPaperSeries

No3261AbstractHowdocentralbankdigitalcurrencies(CBDC)impactthebalancesheetsofbanksandcentralbanks?Totackle

thisquestionempirically,webuiltaconstraintoptimisationmodelthatallowsforindividualbankstochoosehowtorespondtooutflowsofdeposits,basedon

costconsiderationsand

subjecttotheavailabilityofreservesandcollateral,withintheindividualbanksandsystemwide,andforagivenlevelof

liquidityrisktolerance.Wesimulatethe

impactof

afictitiousdigitaleurointroductioninthethirdquarterof2021,usingdatafromover2,000euroareabanks.Thatimpactdependsoni)thenumberofdepositswithdrawnandthespeedatwhichthisoccurs,ii)theliquidityavailablewithinthebanking

systematthetimeofthedigitaleurointroduction,

iii)theliquidityriskpreferencesofthemarketsandsupervisors,

iv)thebank’sbusinessmodel,andv)thefunctioningoftheinterbankmarket.Wefindthata€3,000digitaleuroholding

limitperperson,assuggestedbyBindseil(2020)andBindseilandPanetta(2020),wouldhavebeensuccessfulincontainingtheimpactonbankliquidityrisksandfundingstructuresandontheEurosystembalancesheet,eveninextremelypessimisticscenarios.JEL

codes:

E52,E58,

G21.Keywords:

digitalcurrency,

financialintermediation,financialstability,

liquidityrisk.ECBOccasionalPaperSeries

No3262ExecutiveSummaryCentralbanksthroughouttheworldareinvestigatingthepotentialbenefitsandrisksofintroducingCBDCsornot.Inthispaper,weproposeamodeltosimulatehowthebalancesheetsofbanksand

centralbanksmightbeimpactedbythelossofbanks’depositfundingshouldaCBDCbeintroduced.Wesimulatetheimpactofafictitiousdigitaleurointroduction

in2021usingdetailedbank-leveldata.Wheninterpretingtheresults,wepayparticularattentiontooutflowscompatible

witha€3,000holdinglimit,assuggestedbyBindseil(2020)andBindseilandPanetta

(2020).WhenaretaildepositorwithdrawsfundsfromabankinordertoholdCBDC,itsbankwillneedtotransfercentralbankreservestothecentralbank.ShouldabankholdinsufficientbanknotesandreservestomeetthedemandforCBDC,ithasdifferentoptionstoborrowreserves:shorttermorlongterm,onthesecuredorunsecuredinterbankmarketorfromthecentralbank.Abankwillchoosebetweenthoseoptionsbasedontherelativecoststheyinvolve,butalsobasedon

theimpactonitsliquiditybuffersandsubjecttotheavailabilityof

collateralandmarketliquidity.Thedepositoutflows,reserveholdingsandliquiditybuffersoftheotherbanksthereforedeterminetheavailableoptionsofeachbank.Ourconstraintoptimisationmodelcapturesthesedifferentconsiderations.Inourcasestudy,wesimulatetheimpactofafictitiousdigitaleurointroductioninthethirdquarterof2021usingeuroareabankbalance-sheetdataandillustratea)howbanksmighthaverestructuredtheirbalancesheetsinthe

immediateaftermath,andb)howmuchadditionalreservedemandtheEurosystem

wouldhavefaced.Weanalysetheimpactunderdifferentliquidityrisktolerancescenarios.Underourbaselinescenario,bankshaveanintermediateliquidityrisk

appetiteandwishtokeephalfofthebank-specificvoluntaryliquiditybufferstheyheldinexcessoftheregulatoryminimum.Inthisscenarioandwithaholdinglimitof€3,000perpersoninplace,wefindthatbankfundingstructureswouldnothavechangedextraordinarilyandnoadditionalEurosystemfundingwouldhavebeen

needed.Wealsosimulatethe

impact

ofadigitaleurowith2019data,atimewhenreserveswerelower.Moreover,weadjustourmodelspecificationto

studytheimpactofasegmentedratherthanperfectlyfunctioning

interbankmarketaswellasthe

impactofabankrunwherethereisnotinterbankmarket.Thesimulationsinthesecaseslikewiseshowthattheimpactwouldhavebeenrelativelybenign,providedadigitaleuroholdinglimitof€3,000perpersonwouldbeinplace.Marketandpolicyrates,collateralandreserveavailability,liquiditybuffersandbanks’willingnesstodrawthesedown,areallimportantdeterminantsinbanks’portfoliochoices.Thesefactorsinteractandchangeover

time.Also,the

centralbankwouldpre-emptivelytake

intoaccountanupcomingdigitaleurowhendecidingonitsoperationalframeworkanditssupplyofreserves.ToproperlygaugetheactualimpactofadigitaleuroacrosseuroareabanksandEUMemberStates,itisthereforenecessarytorepeatthesesimulationsusingdataandaccountingfortheprevailingoperationalframeworkatthetimeofapossibledigitaleurointroduction.ECBOccasionalPaperSeries

No32631IntroductionCentral

banks

throughout

the

world

are

investigating

the

potential

benefits

andrisks

of

introducing

CBDCs

or

not.

Asuccessful(retail)CBDCwouldleadto(retail)customersshiftingpartoftheirfundsawayfrombankdepositstocentralbanks.Anoften-cited

concerninthisregardisa

consequentincreaseinbanks’fundingrisksandadecrease

inbanklending(Eurosystem,2021).1

Inthispaper,weexaminetheformerandstudyhowbanksmightadjusttheirbalancesheetsinresponsetoalossofretaildepositfunding(overnighthouseholddeposits),andwhatthepotentialimplicationsmightbeforbanks’liquidityrisksandthedemandforexcessreserves.Wefocuson

theshort-termimpact,assumingbanklendingremainsconstant.When

a

retail

depositor

withdraws

funds

from

a

bank

to

exchange

them

fordigital

euros,

its

bank

will

need

to

transfer

central

bank

reserves

to

the

centralbank.

AsdetailedinAdalidet

al.(2022),abankcanobtainCBDCbytransferringeitherbanknotesorcentralbankreservestoitscentralbank.Shouldabankhaveinsufficientbanknotesandreservestomeetthedepositoutflow,itcouldacquirenewreservesontheinterbankmarketorfromthecentralbank.Thebankwouldchoosehowtoadjustitsbalancesheetbasednotonlyontherelativecostsoftheseoptions,butalsoontheirimpacton

itsliquiditybuffersandtheavailabilityofexcessreservesandcollateral.We

built

a

model

that

allows

individual

banks

to

adjust

their

balance-sheet

inreaction

to

a

retail

deposit

outflow

based

on

cost-efficiency

and

subject

totheir

own

but

also

other

banks’

liquidity

preferences,

reserve

constraints

andcollateral

availability.

Inchoosinghowtorespondtodepositoutflows,banksareconstrainedbytheirholdingsofreservesandcollateralandfaceatrade-offbetweenfundingcostsand

liquidityrisks,which

isnottrivial.2

Also,

banks’optionswhentappingtheinterbankdependnotonlyontheirownpreferencesbutalsoonthedepositoutflows,reserveholdingsandliquiditybuffersof

theotherbanks.Allowingbankstoendogenouslyselecttheirpreferredbalance-sheetadjustmentsposesthereforeacomplexoptimisationproblem.Our

main

contribution

is

to

develop

a

detailed

simulation

model

of

thechanges

to

each

and

all

banks’

liability

positions,

reserve

holdings

andregulatory

liquidity

ratios.

In

contrasttotheexistingliterature,ourmodelanddatamakeitpossibletoassesstheimpactonandoffthetwokeyliquidityrequirements,1Inthispaper,

wearenotdiscussingthe

benefitsof

introducing

adigitaleuro.Forthis,we

refertoPanetta

(2021),whoamong

otherbenefits

pointsoutthemonetaryanchor

role

of

adigitaleuro:“[C]onvertibility

intocentralbank

moneyis

thereforenecessaryforconfidenceinprivatemoney,

bothasa

meansof

paymentandas

astore

ofvalue”.2Forinstance,short-termunsecured

borrowing

hasahigherrun-offratethanovernighthouseholddeposits.

Consequently,

it

wouldincrease

theLCRdenominator(expected

outflow)andwouldnotcountas

stablefundingforthe

NSFR.

While

medium-term

securedborrowingdoesnot

negativelyimpactexpectedoutflows,itneedsto

bebackedby

collateral,whichreducestheLCRnumerator(unencumberedHQLAs)

andincreasesthe

NSFR

denominator(required

stablefunding).

Furthermore,forbothtypesof

interbankmarketfunding,thereservesof

thebankthat

provides

liquidityontheinterbankmarketdecrease,

whichlowersitsLCRnumerator

andalsoincreasesits

NSFRdenominator.ECBOccasionalPaperSeries

No3264namelytheLiquidityCoverageRatio(LCR)andtheNetStableFundingRatio(NSFR),applyingtherequisiteinformationonassethaircutsandliabilityrun-offrates.Theseregulatoryratiosconstrainbanks’fundingoptionsandarethereforeimportantdeterminantsfortheirbalancesheetcomposition,influencingbanks’choicetoreverttotheinterbankmarketorthecentralbank.Themodelalsoincorporateseachbanks’reserveholdingsandavailable

unencumberedcentralbankeligiblecollateralandcontinuouslyupdatesthisinformationduringthesimulation.Fortheinterbankmarket,securedandunsecuredfundingoptionswithdifferentmaturitiesandhaircutsandtheirimpactonliquidityratiosforbanksonbothsidesofthetransactionsarealso

includedinthemodel.Forcentralbankfunding,varioustypesofcentralbankfunding

areconsidered,includingshort-termandlong-termfundingsecuredbyhighqualityliquidassets(HQLAs),eligiblenon-HQLAs,orcurrentlynon-eligiblecollateral.Theexistingempiricalliterature,sofar,disregardseithertheNSFRorliquidityregulationaltogether,ignoresthe(limitationsofthe)interbankmarketaswellas

collateraland/orreserveavailability,andconsidersasubsetofadjustmentoptionsinisolation.We

apply

our

model

to

illustrate

the

impact

of

a

fictitious

digital

eurointroduction

in

2021

on

the

balance

sheets

of

euro

area

banks

and

theEurosystem.

Ourbaselinesimulationusesbalance-sheetassetsand

liabilitiesdataforthethirdquarterof2021–

themostrecentdataavailableatthetimeofouranalysis–formorethan2,000euroareabanks,therebyencompassingmorethan95%oftheeuroareabankingsectorassets.Intermsofliquidityrisktolerance,ourbaselinescenario(Scenario

B)assumesthatbankswouldbewillingtodrawdownhalfofthevoluntaryliquiditybufferstheyholdinexcessof

theregulatoryminimum,whichisequaltotheirmedianobservedannualchanges.

Wealsoassumethatbankswouldbewillingtoprovideliquidityontheinterbankmarket,providedthisdoesnotincreasetheirliquidityriskbeyondtheirpreferredlevels.Regardingtherelativefundingcosts,wemakethereasonableassumptionthatshort-termliquiditywouldbecheaperthanlong-termliquidity,securedfundingless

costlythanunsecuredfundingandmarketfundinglessexpensivethancentralbankfundingofasimilarmaturityprovidedthat

theoverallamountofexcessliquidityinthesystemissufficient.3We

simulate

banks’

responses

to

a

withdrawal

of

overnight

retail

deposits,focusing

on

the

most

extreme

outflows

compatible

with

a

digital

euro

holdinglimit

of

€3,000

per

person.

In

ourempiricalassessment,

wesimulatebanks’responsestoovernightretaildepositoutflows,giventhatthesedepositsarearguablytheclosestsubstituteforadigitaleuro.Bindseil(2020)andBindseilandPanetta(2020)suggestthatthemaximumdepositoutflowcouldbe

restrictedbyimposinga€3,000digitaleuroholding

limitperperson.Intheeuroarea,thatlimitwouldmeanamaximumdepositoutflowof

€1.0trillion

ifeachandeveryeuroarearesidentwereto3Regardingthepricingof

thedifferentfunding

options,wedeviated

fromthe

ratesobservedin2021.

Atthetimeofthesimulation,using

TLTROs

wouldhavebeenthedominantstrategyforallbankswithaccesstothatoptiongiventheirattractivepricingand

thatsuchoperationshavenonegativeimpactonliquidity

ratiosif

collateralised

againsteligiblenon-high-quality

liquidassets(HQLAs).TLTROs

would,infact,haveimproved

bankprofitabilitysincetheyearn

interest,whileretail

depositswerenotgenerallyrenumerated.However,

weexcludedtheoptionofTLTROs

onthegroundthattheyareunlikelytoprevail.ECBOccasionalPaperSeries

No3265adoptthedigitaleuroandwouldcontinuouslyprefundthe

digitaleurouptothemaximumholdinglimitsolelythroughtheirbankdeposits.

Clearly,itis

unlikelythatallresidentswouldfullyutilisetheirlimitof€3,000,which

couldbemorethantheirmonthlyincome,onacontinuousbasisand

solelybydepositsubstitution.Accordingly,outflowsof€1.0

trillionareassumedtobethe“mostextreme”,whilelargeroutflowsaredeemedtobe“unrealistic”.4Based

on

2021

data,

we

find

that

even

with

the

most

extreme

retail

depositoutflows,

the

digital

euro

would

have

had

little

impact

on

the

Eurosystembalance

sheet

beyond

a

swap

of

counterparties

from

banks

to

households.Withlowdepositoutflows,theEurosystemwouldnothaveneededtoprovideadditionalreservestobanks,

assumingbankswerewillingtousehalfoftheirvoluntaryliquiditybuffers.BankswouldhavepreferredcheaperinterbankfundingratherthanrecoursetotheEurosystem.Theinterbankmarketredistributesexcessreservesfrombankswithhighreserveholdingsandhigh

liquiditybufferstobanksinneedofreserves.Thequestionthatarisesis

atwhatoutflowtheEurosystemwouldneedtosupplyadditionalreservestoavoid

stressontheinterbankmarket,whichhappenswhenallbanksreachtheirliquidityrisktolerancelimit,meaningthattheywouldbereluctanttoprovide

furtherliquidityontheinterbankmarket.5

Underourbaselineliquidityrisktolerancescenario(ScenarioB),the

bankingsystemcouldhaveaccommodatedanoutflowof20%ofretaildeposits

bymerelydrawingdownexistingexcessreservesandnotrequiringadditionalreservesfromthecentralbank.The20%ofretaildepositsequateto€1.4trillion,whichexceedsthemostextremeoutflowof€1.0trillion.It

isonlyinthehighlyunlikelyeventofhigheroutflowsthatbanksneedtoobtainadditionalreservesfromtheEurosystemagainsteligiblecollateral.Naturally,ifoutflowswereextremelyhigh,somebankswouldrunoutofcurrentlyeligiblecollateral.UnderScenarioB,wefindthatitisonlywhenoutflowswouldhaveexceededaround30%ofretaildeposits,equatingto€2.1trillion(morethandoublethemostextremeoutflow),thatone-tenthofthecentralbankfundingrequiredcouldnothavebeensecuredagainstcurrentlyeligiblecollateral.Even

with

the

most

extreme

outflows,

the

shift

in

banks’

funding

structuresaway

from

retail

and

towards

wholesale

and

central

bank

funding

would

nothave

been

unusual.

Usingdataforthethirdquarterof2021andassumingthatbanksmaintainhalfoftheirvoluntaryliquiditybuffers,wefindthatevenwiththemostextremeoutflows,onlyafew

bankswouldhaveexperiencedanextremeincreaseintheirrelianceoncentralbankorwholesalefunding.Itis

onlywhenoutflowsexceeded28%ofretaildeposits,equatingtoatotalof€1.9trillion,thatamoresignificantproportionofthebankingsector(over10%intermsoftotalassets)wouldhaveexperiencedanextremeincreasein

itswholesalefundingreliancewhencomparedtohistoricalquarterlychanges.Weusehistoricaldatatoarguethataslowphase-inofthedigitaleuro,thatlastslongerthanaquarter,wouldrenderthe45Forcomparison,eurobanknotesin

circulationcurrentlyamounttoabout€1.6

trillion.TheEurosystemprovidesadditional

liquidity

whenexcessliquidityreachesthe

floorrequiredexcessliquidity

(FREL)leveltoensureasmooth

transmissionof

monetarypolicy.

Inourmodel,thispointisreachedwhenbanksareno

longer

willingtoprovidemorefundingbecauseit

wouldresult

intheirliquidity

levelsbeinglower

than

theywouldprefer.ECBOccasionalPaperSeries

No3266increasesincentralbankfundingrelianceandwholesalefundingdependenceevenmoremoderatecomparedto

historicalannualchanges.The

impact

of

a

digital

euro

with

a

€3,000

holding

limit

also

remains

moderatewhen

running

simulations

using

data

for

the

third

quarter

of

2019,

aneconomic

environment

with

less

excess

liquidity,

or

when

assuming

asegmented

interbank

market

or

a

potential

bank

run

scenario.

Clearly,iftheinputorassumptionsforour

modelweretochange,theoutcomeofthesimulatedimpactofaCBDCintroductionwouldalsochange.Toillustratethis,wefirstapplythemodeltoeuroareabalance-sheetdataforthethirdquarterof2019,whenreserveswerelessampleandtherewasthereforelessexcess

liquidity.Thissimulationshowsabenign

impactonbanksthatis

similartothatinourbaselinescenario.Therearetworeasonsforthispotentiallysurprisingfinding.First,banksheldlowerreservesbuthadmoreeligible

collateralin2019ascomparedwiththethirdquarterof2021.Theycouldthereforeusethiscollateraltoobtainreserveswhenneeded.Second,bankshadfewerretaildepositsin

2019ascomparedwiththethirdquarterof2021.Wethensimulatethemodelrelaxingtheassumptionofafrictionlessinterbankmarket.

Wefindthatwhenbanksonlyborrowandlendonanationalinterbankmarket,it

almostmakesnodifferenceto

theconclusionsdrawnfromthemorebenignbaselinescenario.Finally,weshowthatourmodelcouldbeusedtocalibratedigitaleuro

holdinglimitsthatwould

containadigitaleuro’simpactonbanks’liquidityrisksduringasystem-widestressperiod.Wefindthatadigitaleuro’simpactonbanks’liquidityrisksinabankrunscenariowouldhavebeencontainedintheeventofa€3,000holdinglimit.AllourresultsaresummarisedinTable2.A,Table2.BandChart14attheendofthispaper.Our

model

could

be

used

to

guide

policy-makers

decision-making

on

thedesign

and

timing

of

a

future

digital

euro

introduction.

Whileasuccessfuldigitaleurorequiresuptakebyeuroarearesidents,itshouldnot

beusedtoomuchtoavoidfinancialstabilityrisks(Ahnertetal.,2022).Adigitaleuro

holdinglimitmightpreventexcessiveuseofadigitaleuro.Obviously,the

simulationresultsgiveninthispapercannotbeusedtopredictthe

responseofbanksifandwhenadigitaleuroisintroduced,giventhateconomicsituations,marketratesandbankbalancesheetsaresubjecttochange,andthiswouldbeparticularlytrueif

anintroductionweretobeanticipated.Similarly,alsothecentralbankwouldpre-emptivelytakeintoaccountanupcomingdigitaleurowhendecidingonitsoperationalframeworkanditssupplyofreserves.Therefore,policy

makerswillneedtore-runour

modelclosetothetimeofadigitaleurointroductionto

gaugetheimpactonthebalancesheetsoftheEurosystemandbankswithinthechangedenvironment.Suchsimulation

coulddistinguishbetweentheimpactonbusinessmodelsand

MemberStatesandbeusefulforthecalibrationofdigitaleuroholdinglimits,ifpolicymakerswishto

imposethose.66Whendeciding

whetherornottoimposeaholdinglimitandif

soat

which

level,financial

stability

andcentralbankfootprintconsiderationswillof

coursebeonlyonepartof

the

equation.Otherconsiderationsincludetheusabilityof

a

digitaleuro(e.g.households’

averageexpectedtransactionsizeandincome)

andthe

digitaleuro’smonetaryanchor

role,amongothers.ECBOccasionalPaperSeries

No3267The

remainder

of

the

paper

is

structured

as

follows.

Chapter2presentsaliteraturereview.Chapter3presentsthemodelandChapter4thedataanddescriptivestatistics.Chapter5studiestheimpactontheEurosystemandbanks’balancesheetsofapotentialdigitaleurointroductionas

simulatedbyourmodel.Chapter6considersvariationstothedatainputandmodelspecifications,

includingalowerinitial

levelofexcessreserves,animperfect

interbankmarketandabankrunscenario.Chapter7setsout

theconclusionsandpolicyrelevance.ECBOccasionalPaperSeries

No32682LiteraturereviewA

growing

literature

uses

theoretical

models

to

study

how

banks

might

beimpacted

by

CBDCs.

Servingasabenchmark,BrunnermeierandNiepelt(2019)presentan“equivalence”result:undercertainconditions,

bankswould,intheory,beunaffectedbyadepositoutflowtoCBDCsifthecentralbankweretoredirectliquiditybackintothebankingsystem

underfavourableconditions.

Ourmodelcouldreplicatetheequivalenceresultbyassuminganenvironmentinwhichcentralbankfunding

isthecheapestadjustmentoptionanddoesnotrequirecollateral.However,wefocusontherealisticsituationwhentheequivalenceresultdoesnotholdduetoliquidityregulation,collateralrequirementsandintheabsenceof

unconventionalmonetarypolicyinstruments.

SomeoftheseaspectsarealsostudiedbyNiepelt(2020),Assenmacheretal.(2021),Burlonetal.(2022),Williamson(2022),andMu?ozandSoons(2023).Ourmodeldiffersfromthesestudiesas

ithas

afocusonliquidityriskandconsidersthebankleveloptimizationproblemratherthantherequiredmacroeconomicadjustment.Only

few

papers

have

attempted

to

quantify

the

potential

impact

of

a

CBDC

onindividual

banks.

SincenodevelopedcountryhasintroducedaCBDC,thereisnodataavailabletomeasureitsimpactonbanks.

Asmallnumberofpapersresort,however,toscenarioanalyses.Castrénetal.(2022)useanetworkapproach

toconsiderhowsector-levelbalancesheetsmightchangeunderdifferentCBDCscenarios.BIS(2021)considersastylisedmodelinwhich

thebankingsystemholdsitsliquidityratioconstantafterdepositoutflowsbyacquiringHQLAsusinglong-termwholesalefunding;however,

itmakesnoallowanceforthefactthattheavailabilityofreservesinthebanking

systemmightbeaconstraint.Gorelovaetal.(2022)considersthe

impactonliquidityratiosof

severallargeCanadianbanksifretailfundingweretobereplacedbyfundingwithahigherrun-offrate,againabstractingfromthefactthattheoverallreservesinthesystemarenotinfinite.Weaddtothesestudiesduetoourgranularbank-leveldataanddetailedsimulationmodelwhichsimultaneouslyencompassesliquidity,

collateralandreserveconstraintsat

individualbankandbankingsystem

level.Thismakesitpossibleto

considerhowindividualbankswouldselecttheirpreferredfundingoptionratherthanresortingtoastylisedsector-levelscenarioanalysis.GrossandLetizia(2023)provideupperboundestimatesofactualCBDCtake-up,underdifferentassumptionsofCBDCrenumeration.Thisnicelycomplementsouranalysis,asweareagnosticregardingtheactualtake-upandrathersimulatehowbanksmightrespondtodifferentretaildepositoutflows.PreliminaryandpartialresultsofouranalysisfeaturedinAdalidetal.(2022).ECBOccasionalPaperSeries

No32693ThemodelThisChaptersetsouttheintuitionbehindtheoptimisationmodel.Amoredetailedtechnicalpresentationofthe

modelis

containedinAppendixA.We

model

how

each

bank

would

optimally

respond

to

a

retail

deposit

outflowwhile

minimising

costs

and

allowing

for

liquidity,

collateral,

and

reserveconstraints.

Inthemodel,bankscan

intermediateCBDCdemandinthreeways:1)byusingtheir

currentcentralbankreservesorbanknotes,

2)byobtainingadditionalcentralbankreservesorbanknotesontheinterbankmarket7,or3)byincreasingcentralbankborrowing.8

Tobe

precise,ontheasset

side

ofitsbalance

sheet,abankcouldaccommodateitsretaildepositoutflowsbyreducingitsexisting

centralbankreserveholdings,therebyalsoreducingthesizeofitsbalancesheet.

Shouldthebanknothavesufficientreserves,it

couldobtainadditionalcentralbankreservesontheinterbankmarketorincreaseitscentralbankborrowing.Thebankwouldreplaceretaildepositswithwholesaleorcentralbankborrowingontheliabilitysidewithnoimpactonthesizeof

itsbalancesheet,butitsassetswouldbeencumberedifitengagedinsecuredborrowing.9

Importantly,themodelassumesaperfectlyfunctioningandfrictionlessinterbankmarket.InChapter6,

werelaxthisassumption.Not

all

funding

options

are

equally

feasible

or

desirable

given

that

they

wouldhave

an

impact

on

banks’

profitability,

liquidity

risk,

and

collateral

availability.Securedfundingis

cheaperthanunsecuredfundingandshort-termfundingislessexpensivethanlong-termfunding.Yet,drawingdownownreservesorpledgingHQLAsascollateraltoobtainsecuredfundingnegativelyimpactsbanks’liquiditypositions(reducingtheirLCR).Furthermore,usingshort-termratherthanlong-termfundingincreasesbanks’roll-overrisk(reducingtheirNSFR).Withregardtorelativeprices,weassumethatthed