U. Bindseil - Evaluating Monetary Policy Operational Frameworks (2016)

Embargoed until presentation time of 10:55 a.m.
Mountain Daylight Time, August 26, 2016
Evaluatingmonetarypolicy
operationalframeworks
U.Bindseil*
10August2016
This paper first discusses objectives of operational framework design and related assessment criteria.
Second, it reviews the pre2007 experience and summarizes central bankers beliefs at the eve of the
financial crisis. Third, it tries to derive additional lessons from the last nine years, including: what the
operational framework might contribute exante to support financial stability, the design of the LOLR
function,theroleoftheZLB,andtheimpactofregulatorychange.Thepaperthenoutlineshowagood
long term framework may look like reviewing the key monetary policy instruments one by one. To
translatethetheoreticalconsiderationsintoapracticalevaluationtool,theideaofscorecardsisexplored.
JELcodes:E42,E43,E44,E52,E58
Key words: Operational framework, monetary policy implementation, central bank market operations,
standingfacilities,openmarketoperations,reserverequirements.
* European Central Bank, DGMarket Operations. Opinions expressed are those of the author and not
necessarilythoseoftheECB.IwishtothankJeanPierreDanthine,ArturoDiezCaballero,TroyDavig,John
Groom, Paul Chilcott, Tobias Linzert, Antoine Martin, Arne Kloster, Simon Potter, Benjamin Sahel, Ylva
SovikandOlivierVergoteforhelpfuldiscussionsand/orcomments.Allremainingerrorsaremine.
Embargoed until presentation time of 10:55 a.m. Mountain Daylight Time, August 26, 2016
1. Introduction
Monetarypolicybecomesrealthroughfinancialoperationsofthecentralbankwiththerestoftheworld
andinparticularwithfinancialinstitutions.Signalingoropenmouthoperationsofcentralbanksare
only relevant if they affect expectations regarding actual financial market operations. Therefore, a
perceptionthattheroleofcentralbankmarketoperationshadinsomesensetoberediscoveredthrough
thefinancialcrisiswouldbesurprising.Centralbankfinancialmarketoperationshavebeenthecentral
banksportaltorealityeversincethecreationofcentralbanksinthe17thcentury.Operationalframeworks
(OFs),i.e.thefinancialinstrumentsandrelatedrulesandpracticesformonetarypolicyimplementation
canbedesignedtobemoreorlesseffective,transparentandefficient.Inthelattercase,notonlythe
achievementofacentralbanksmonetarypolicyobjectivesisputatrisk,butinadditiontheefficiencyand
stabilityofthefinancialsystemcanbeundermined.ConsiderfourexamplesoftherelevanceofOFsfrom
thelast150years.
First,before1914,intheclassicalageofmetalbasedcurrencies,twocentralbankmarketoperations
issueswereextensivelydebatedandeventuallybetterunderstood:(i)thequestionhowtoaddressabank
runthroughcentralbankcreditandoutrightpurchaseoperations(thelenderoflastresort,astreatedby
Bagehot,1873),and(ii)howtomakebankrateeffective,i.e.howtoensurethatthereisactualcentral
bankcontrolofshorttermmarketrates,andnotonlyoftheirownpolicyrates(seeinparticularKing,
1936).Onboth,historyillustratesabundantlyhowapoorunderstandingofissuesrelatedtocentralbank
operationsledtowrongdecisionsandeconomiccosts.Despitetheimprovedunderstandingtowardsthe
end of that period, both topics remain a challenge until today, as for example (i) moved again to the
forefrontin2008,and(ii)wasthekeyissueintheFedsliftoffin2015.
Second,theunstable25yearsafter1914werecharacterizedbywarfinancing,goldstandardsuspension
andinflation,andlaterbythetemporarycostlyrestoringofthegoldstandardinaworldofunprecedented
internationalimbalancesandeventuallythedeflationandbreakdownoffinancialstabilityintheearly
1930s.Centralbanksimprovisedandinnovatedconsiderablyontheirmarketoperations,withmoreor
lesssuccess.Whiletheeventualfailuretorestorenormalityandsustainablegrowthisexplainedalsowith
macroeconomicproblems(suchastheinabilitytoswiftlysolvewardebtandreparationissues),alsopoor
centralbankoperationsmadetheircontribution,suchasthefailuretocollaboratewithregardtointer
central bank credit operations (e.g. Toniolo, 2005), and renewed hesitations of central banks to act as
lenderoflastresort(LOLR),inparticularduringthebankrunsoftheearly1930s.Moreover,thisperiod
witnessedthebirthofreservepositiondoctrine(e.g.Meigs,1962),whichwoulddeterminethe(arguably
misleading)textbookversionofmonetarypolicyimplementationforalmosttherestofthecentury.
Third,theperiod19601990couldbecalledthebaroqueageofOFs,whichfromtodaysperspective
seemstohavebeenoverburdenedwithanexcessivelycomplexmonetarypolicyimplementation(reserve
requirements with various ratios that were frequently changed, various excess reserves concepts that
wereallsupposedtomatter,poorlydefinedstandingfacilities,andlastbutnotleastvariousdirectcontrol
measuressuchase.g.ceilingstoretaildepositinterestratesthatbankscouldcharge).Thereserveposition
doctrineremainedthetextbooklogicofmonetarypolicyimplementationandtransmission.Kneeshawand
VandenBergh(1989)provideasurveyoftheevolutionofOFssincethe1970s.Therootcausesofthe
oversophisticationofOFsduringthisperiodseemtohavebeenunrealisticacademicdoctrinescombined
withanoverestimationofcentralbankspolicymakingcapabilities.
19902007couldbecalledthemodernageofOFs,inwhichmuchsimplificationcouldbeachievedand
thepre1914viewthatmonetarypolicyimplementationisaboutcontrollingshortterminterestrateswas
fully rehabilitated. For example, the FED eventually, after 80 years, set again the interest rate on the
2
DiscountFacility(whichisaLombardfacilityininitialterminology)toabovetheFedfundstarget.The
designandlaunchoftheOFforthemultijurisdictioneuroareawasanoteworthyeventandcanprobably
beconsideredasasuccessresultingfromcarefulpreparatorydiscussionandreviewofexperience(see
e.g.EuropeanMonetaryInstitute,1997).
Insum,gettingcentralbanksOFsrightalwaysmatteredandwillcontinuetomatter.Therecentcrisis
confirmedthis,andiscertainlyagoodoccasiontotakestockandtorevisithowtodesignandassessOFs.
RelevantliteratureonOFscanbefoundinparticularinthefollowingfourformats:First,officialorsemi
official descriptions of OFs and monetary policy implementation by central banks: Some central banks
explaintheiroperationalframeworksinbookchapterslikee.g.chapter2ofDanmarksNationalbank(2009)
or Chapter 4 of ECB (2011). Meulendyke (1998) describes US monetary policy implementation in the
secondhalfofthe20thcentury.TheFedNewYorkpublishesanannualreportontheconductofmonetary
policyoperations(seee.g.FederalReserveBankofNewYork,2016).Second,comparativestudiesofOFs
includingassessments:Borio(1997)remainsthemostcomprehensivesurveyataspecificmomentintime,
andalandmarkasfirstsystematiccomparativestudyofmodernframeworks.IMF(2004)focusesonthe
caseofemergingmarkets.Ho(2008)providesacomparisonofOFsinAsiancountriesandselectedother
economies.AlsoBindseilandNyborg(2008)surveyedframeworksjustbeforethebreakoutofthefinancial
crisis.Blencketal(2001),Bindseil(2004a)andFriedmanandKuttner(2010)limitthecomparisontoafew
central banks. Sellin and Sommar (2014, chapter 5) provide a recent systematic comparative study. A
number of further surveys focus only on one aspect of the operational framework, like the collateral
framework (e.g. Chailloux et al 2008; Cheun et al, 2009). Third, academic work: There is a significant
numberoffocusedacademicstudiesonspecificmonetarypolicyframeworkorimplementationtopics.
Someliteratureissurveyede.g.inBindseil(2004),(2014a)orSellinandSommar(2014,chapter2).The
practitionermayfindsomeoftheacademicstudiestoprovideusefulintuitionandguidanceonrealworld
decisions.Fourth,speechesbycentralbankersonpostcrisisOFdesign,suchasCoeur(2013)orPotter
(2016).
Thecurrentpaper,drawingfromthesefourtypesofliteraturesonOFs,willprovideageneralreviewof
the objectives and options of OF design. Section 2 first briefly reviews the pre2007 experience and
summarizesthebeliefsonOFdesignattheeveofthefinancialcrisis,and,second,triestoidentifykey
lessonsfromthecrisis.Onthatbasis,section3revisitstheobjectivesofOFdesignandcriteriahowto
assessandcomparetheperformanceofalternativeOFs.Sections4to6eachgodeeperintooneareaof
OFdesign:section4revisitstechniquestocontroltheoperationaltarget(liquiditymanagement),section
5dealswiththecentralbankbalancesheetstructureandtheroleofanoutrightportfolio,whilesection6
coversthecollateralframeworkandtheLOLR.Section7drawsconclusions.
WhenaimingatderivinguniversalprinciplesforOFdesignorevenuniversalOFspecifications,onanumber
ofoccasionsthispaperwillreachapparentlimitsrelatingtofundamentallydifferentphilosophiesofcentral
banksorofindividualauthors.AsGeorgSimmeldescribeditinhisfamousintroductiontothePhilosophy
ofMoney(2014/1900,51):Everyareaofresearchhastwoboundariesmarkingthepointatwhichthe
processofreflectionceasestobeexactandtakesonaphilosophical character.Thepreconditionsfor
cognitioningeneral,liketheaxiomsofeveryspecificdomain,cannotbepresentedandtestedwithinthe
latterdomain,butrathertheycallforascienceofamorefundamentalnature.Thegoalofthisscience,
whichislocatedininfinity,istothinkwithoutpreconditionsagoalwhichtheindividualsciencesdeny
themselvessincetheydonottakeanystepwithoutproof,thatis,withoutpreconditionsofasubstantive
andmethodologicalnature.Three(somewhatrelated)philosophicalissues,oratleastissuesbeyondthe
science of OF design, will be encountered in this paper, namely with regard to (i) what attitude the
centralbankshouldtaketowardsfinancialexposurestothegovernment;(ii)whatneutralitymeansin
3
termsofcentralbankcollateraleligibilityandassetallocationdecisions,and(iii)howambitious,activist
andcomplexityfriendly(or,alternatively,skeptical,humbleandsimplicityoriented)acentralbankshould
beinitsOFdesign.Ofcourseallofthesethreequestionsseemtobesuitableforbeinganalyzedandjudged
furtheronthebasisofamorefundamentalanalysis,butforthepresentpaperitwasconcludedtoaccept
themasgivenphilosophicalviews.
InAnnex1,theconclusionsfromeachofthesections4to6aretranslatedintoscorecardswhichare
constructedtobeusedforstandalone,orcomparativeassessmentsofexistingOFs.Thisisanattemptto
verify that the findings indeed can be translated into realworld guidance. The scorecards are lists of
statements, which,iffullyaffirmed,giveahighscore,andiffullyrejected,a lowone.Finally,Annex2
describesasimulationtoolthatisextensivelyusedinsection4tounderstandhowvariousparameters
determinetheeffectivenessofthecontroloftheovernightinterestrate(whilepreservingmoneymarket
activity).
2. The2007consensusandlessonsfromthelastnineyears
Following others (e.g. Blanchard et al, 2014, when discussing postcrisis macroeconomic thinking), this
section first tries to summarize the 2007 consensus on OFs (subsection 2.1) to then review the main
lessonsfromthecrisis(subsections2.2).
2.1The2007consensusonOFdesign
AnyreviewofOFsisconfrontedwiththeriftcreatedbythefinancialcrisisthatstartedin2007.Thepre
2007worldhadwitnessedconvergenceinthesearchofthebestOF,andOFsdeterminedmonetarypolicy
implementation practices. In contrast, after 2007, monetary policy implementation practice was
dominated by countless idiosyncratic nonstandard measures and the OF concept had therefore less
relevance.Thissubsectionprovidesabriefsummaryofwhatappearstohavebeenthe2007consensus
amongstcentralbanksonOFdesign.Sevenkeyelementscanbehighlighted.
1) Desirability of having a single and well defined operational target for daytoday monetary policy
implementation.Whiletherehadbeenalotofconfusionontheconceptbeforethe1990s,in2007a
clearunderstandingoftheoperationaltargethademerged,beingavariablethat(i)canbecontrolled
onadaybydaybasisbythecentralbankwithoutcreatingunduevolatilityinotherfinancialmarket
prices;(ii)isameaningfulstartingpointofthetransmissionmechanism,i.e.itsimpactontheultimate
targetofmonetarypolicyisrelevantandsufficientlypredictable;(iii)canbedecidedbythemonetary
policy decision making body, be communicated to the public and gives sufficient guidance for
implementationofficers.
2) Thebestoperationaltarget(atleastforadvancedeconomies)isashortterminterestrate.Bindseil
(2004b)describesthe20thcenturystruggletorediscoverthis.SellinandSommar(2014,124)orHo
(2008, section 3) provide overviews of the details of operational target choices by central banks.
Blanchardetal(2014,5)summarizesthemacroeconomicbeliefsbehindthe2007consensusonshort
terminterestratesasoperationaltarget.
3) OFsshouldinprinciplebesimple.Inthe1980s,atrendsetintoovercomethecomplexitiesofthe
baroqueageofmonetarypolicyimplementationandin2007thereseemedtobeaconsensusthat
OFs should be parsimonious. The trend could have continued further, as there was still scope for
streamliningofOFs,butitwasstoppedbythecrisis.
4) Aseparationprinciple(ordichotomy)appliesbetween(i)themacroeconomicanalysisandthesetting
oftheoperationaltargetlevel,and(ii)itsimplementationthroughmonetarypolicyoperations.Afull
dichotomy means that the overnight interest rate is the single gate between monetary policy
implementation and the rest of the transmission mechanism. Conditions for this being viable (and
optimal)are:(i)distancefromeffectivelowerboundfortheovernightrate;(ii)sufficientarbitrageand
stabilityofrelationsbetweenovernightrateandrestoftheinterestratestructure;(iii)normalmarket
accessofindebted(solvent)agents.
5) Creditopenmarketoperationswithallotmentvolumessetbythecentralbankarethestandardtoolto
steerliquidityconditions(i.e.tosteerthescarcityofbanksreserveswiththecentralbank,andthereby
theshortterminterbankinterestrate).Creditoperationshavematuritiesuptothreemonthsandboth
variablerateandfixedratetenderswereused(e.g.BindseilandNyborg,2008,753).
6) Reserve requirements are (in advanced economies) a tool for smoothing daily autonomous factor
liquidity shocks without having to take recourse to daily open market operations. Reserve
requirementsareremuneratedatclosetothetargetrate(i.e.atclosetomarketrates;anexception
wastheUSinwhichremunerationofexcessreserveswasonlyallowedinprinciplebyCongressin2006
andeffectiveasof1October2008).Reservemaintenanceperiodshavealengthbetweenoneweek
andonemonth.Foremergingeconomies(oftenwithlargeforeignreserves)othermonetarycontrol
functionsof(nonremunerated)reserverequirementsremainedrelevant(overviewsine.g.Bindseil
andNyborg,2008,757,orHo,2008,12).Anotherpartofthecentralbankingcommunity(Sweden,
Canada, New Zealand, Australia, etc.) had opted for daily open market operations and no reserve
requirementsystem,andwereabletoachievesimilarorevenbettercontrolofshortterminterest
rates(SellinandSommar,2014).
7) Asymmetricstandingfacilitiescorridoraroundthetargetinterestrateisthepreferredapproachtothe
daytoday control of the overnight interest rate. The choice of the width of the standing facilities
corridorisseenasareflectionofatradeoffbetweendesiredinterestratecontrolandtheneedto
avoidthatthecentralbankbecomesanintermediaryinthemoneymarket.Twovariantshademerged
(e.g. Bindseil and Nyborg, 2008, 256), in combination with the existence or not of reserve
requirements:(i)Symmetricnarrow(25basispoints)standingfacilitiescorridorwithadailyopen
market operation at the target rate such as to guide the interbank market rate. This simple and
effectiveapproachwasappliede.g.inAustralia,Canada,andNewZealand.(ii)Reserverequirements
withaveraging,relativelywidersymmetriccorridor(100basispoints).Largemonetaryareashave
shownapreferenceforthissecondapproach.TheUSFedremainedagainanexceptionaswithouta
depositfacilityortheabilitytoremunerateexcessreserves,itcouldobviouslynotoptforasymmetric
corridorapproach.
Inthreeotherimportantfields,convergenceofviewsandpracticeshadnottakenplacein2007,partially
linkedtotheheterogeneousphilosophicalviewsonmonetarypolicyimplementationmentionedinthe
introduction:
1) Roleof anoutrightmonetarypolicy portfolioand theoptimalsizeof the bankingsystemsliquidity
deficit visvis the central bank to be covered through credit open market operations. While for
exampletheUSFedonlycoveredaround5%ofitsassetsidethroughcreditopenmarketoperations;
fortheEurosystemthisnumberstoodataround50%
2) Sovereignexposure:whilesomecentralbanks(theUSFed)considereditnaturaltofilllargepartsof
theirassetsidewithGovernmentdebt(revealingthattheyseethemselvesinsomesenseaspartof
theofficialsector,sothattheirapproachimpliesaleanconsolidatedofficialsectorbalancesheet),
othersbelievethat,atleastintheirowncase,theimpliedinterconnectednessbetweenthecentral
bankandthegovernmentisproblematic.
5
3) Collateralandcounterpartysets:First,somecentralbanksoptedforanarrow(e.g.USFed)andothers
forabroad(e.g.Eurosystem)collateralsetforopenmarketcreditoperations.Second,centralbanks
haddifferentapproacheswithregardtothepoolingversussegregationofcollateralsets.Whilee.g.
theFedhadverydistinctsetsforcreditopenmarketoperationsversusdiscountwindowlending,the
Eurosystemhadthesamesetforallitslendingoperationstowardsbanks.Third,somecentralbanks
reliedonabroadcounterpartysetforopenmarketcreditoperations(e.g.Eurosystem),whileothers
onanarrowone(e.g.Fed).
After2007,wewitnessedtheglobalfinancialcrisis,countlessnonconventionalmonetarypolicymeasures
bycentralbanks,most centralbanksofadvancedeconomies hittingthe zerolowerbound,largescale
assetpurchaseprogramsimplyingbankingsystemsinexcessliquidityforyears,andthenewregulatory
universe that was created to prevent future crises. Analyzing monetary policy implementation during
theseyearsasanapplicationofpreviouslyexistingOFsdoesnotseemparticularlyuseful.Theneedto
differentiate the pre2007 world from what came afterwards is also illustrated by the term of
unconventional (or nonstandard) monetary policy measures, which did not seem to have existed
before2007(althoughofcourseunconventionallenderoflastresortactivitiesaredescribedextensively
alreadyinBagehot,1873,andattemptsbycentralbanktotargetlongterminterestratesalsohavebeen
seenpreviously).Conventionalmonetarypolicymeasuresandtoolscouldbedefinedasthosewhichaim
atthecontrolofthestandardoperationaltarget,i.e.shortterminterestrates.Nonconventionalmeasures
mustthereforeaimatsomethingdifferentthantheshortterminterestrate,butate.g.(i)termspreads
(or,equivalently,longtermriskfreerates);(ii)liquidityandcreditspreads(or,equivalently,interestrates
onvariousnonriskfreeinstruments);(iii)financialstabilityforthesakeofsupportingthemonetarypolicy
transmissionmechanism.
2.2Lessonsfromthelastnineyears
Thenormalizationoffinancialconditionswillnotsimplybringusbackto2007viewsonOFdesign.First,
thefinancialsystemwillnotmovebacktoits2007state,duetothecrisisexperienceitself,regulationand
furthertechnologicalchange.Second,thescaleoftheeventsculminatingin2008inspiredtheviewsof
bothpractitionersandacademicsonwhatanOFneedstoachieve.Considerbelowfivekeyrequirements
thatOFswillneedtobeabletocopewithinthefuture,whichallhadlessprominencebefore2007.
OFsneedtosupporttheabilitytoaddresspossiblefuturecrisesforcefullyandquickly
The economic and monetary developments since 2007 seem to have validated the forceful and quick
reactions of central banks to the crisis. If anything, the persistence of lowgrowth and disinflationary
tendenciesofthelastyearssuggestthatcentralbankscouldpossiblyhavehesitatedevenlessinmoving
short term interest rates quickly towards the effective lower bound and to launch unconventional
measures.Forthefuture,thisimpliestheimportanceoftheOFsupportingtheabilityofthecentralbank
toactforcefullyandquicklyinthecaseofarenewedseriouscrisis.Therefore,afutureOFshouldalsobe
assessedagainstitsabilitytoallow,andbecompatiblewith,thevariousnonstandardmeasuresthatmight
beneededagainatsomestage.ThisincludesthatafutureOFshouldsupportthepreservationofrelevant
humancapitalofcentralbankstaff,possiblybyconfrontingstaffregularlywithmoredemandingpartsof
financialmarkets(e.g.creditmarkets,lessliquidinstruments;thebestwaytopreservehumancapitalis
obviouslytooperateinthesemarkets).Theabilitytoactquicklyandthepreservationofknowledgeshould
coverinparticularthosemeasuresthatseemtohavebeeneffective,suchas:(i)LOLRandmarketmaker
oflastresort(MMLR)operations;(ii)largescaleassetpurchaseprograms;(iii)attractivelongtermcredit
operations(includingfixedratefullallotmentandtargetedvariants);(iv)currencybridges(LOLRin
foreigncurrencybasedonswapagreementsbetweencentralbanks).Thepotentialfuturecrisismeasures
couldbeclassifiedasbeingpartofthelongtermOF,ortheycouldbeseenassomethingoutsidetheOF,
6
i.e. on the shelf, that could be activated at any time but which is too unlikely to be displayed in
permanenceasanelementoftheOF.
PostcrisisOFshouldallowtopushsomewhatlowertheeffectivelowerbound(ELB)
On 30 March 2016, Ireland issued a 100 year bond at a yield of 2.35% and US 10 year treasury yields
reachedanewhistoricallowinJuly2016atbelow1.40%,bothillustratingthepresentconsensusthat
interest rates will stay low for long. Reasons for that include in particular expectations of lower trend
growth (secular stagnation), ageing society and the implied current excess of savings, and the high
credibilityofcentralbanksasinflationfightersgainedgraduallysincetheearly1980s(seee.g.Blanchard
etal,2014,andHaldane,2015,forrelateddiscussions).Moreover,thelastyearshaveremindedusthat
financialintermediationspreadsincreasesubstantiallyduringfinancialcrises,requiringextracentralbank
accommodation to compensate this effect. Blanchard et al (2014, 8) believe that in 2008, the policy
adequateshortterminterestratewouldhavebeenaslowasbetween3%to5%.Finally,newregulation
may contribute to structurally increase financial intermediation spreads. These factors imply a higher
likelihoodthatthezerolowerbound(ZLB)forshortterminterestrateswillbeinthefuturearecurring
constraint to monetary policy. The ZLB relates in particular to the problem that banknotes are not
remunerated,andthatincaseofnegativeinterestratesappliedondeposits,thedepositorsmaywithdraw
depositsandhoardcash.Thisisinefficientandalsoincreasestheliquiditydeficitofthebankingsystem
visvisthecentralbank(andtherebyreducesliquiditybuffersofbanks).Differentsolutionsoutsidethe
designoftheOFhavebeenproposedforovercomingthis,suchasinparticular(i)increasingtheinflation
objectivefrom2%to4%soastohittheZLBlessfrequently(e.g.Blanchardetal2014),(ii)abolishingpaper
money,orfindingasolutiontoapplynegativeratestoit(e.g.Goodfriend,2000andBuiter,2009).The
majorityviewamongstcentralbankersisthatbothofthesemeasureswouldbetooradicalorproblematic.
Therefore,thequestionariseswhetherinthecontextoftheOFdesign,theabilitytogoslightlynegative,
i.e.topushratestothesocalledeffectivelowerbound(ELB,withELB<0)canbesupported.Indeed,the
experienceofthelastfewyearshasbeenthatnegativeinterestratepolicies(NIRP)aretosomeextent
feasibleandeffective,withsomecentralbanksgoingaslowassettingtheoperationaltargetinterestrate
to75basispoints,withoutthiscreatinganacuteincreaseofthedemandforbanknotes,atleastsofar.
Indeed,NIRPmayberegardedascontinuationofstandardmonetarypolicy.Thereisalsonoreasonto
believe, that in negative territory the nominal interest rate loses its ability to serve as allocation
mechanism.Thelastyearshaveshownthatfinancialmarketfunctioningdoesnotchangefundamentally
innegativeterritory.Ofcoursetheremaybesometemporarytransitioneffectandadjustmentneeds,e.g.
relatingtoITsystemsandcontractualandlegalframeworkssupportingnegativeinterestrates.Moreover,
securitiesmarketsmayneedsomeadjustmentstoeitheravoidsecuritieswithnegativecoupons(byrelying
more on issuance above par), or by making negative coupons possible. Also the question of negative
interestratesonindexedretailmortgageloansraisesadjustmentissues.(Alltheseissuesobviouslyneed
tobesolvedonlyonce,i.e.ifhittingtheZLBbecomesarecurringphenomenon,theseinvestmentscan
beamortizedoverseveralcycles).
Besides the ELB determined by banknotes, another ELB, kicking in even earlier, has recently gained
prominence.Theargumentrelatestotheobservationthatifretaildepositratescannotbecomenegative,
then bank profitability may suffer from negative rates. This could lead banks to increase rather than
decreasetheirlendinginterestratesasthecentralbankcutsitsoperationaltargetinterestratefurther.It
istobenotedthatsucheffectsalsooccurwithlowrates,i.e.lowpositivecapitalmarketratesalsoreduce
theseignorageincomeofbanksfromissuingdepositsrelativetocapitalmarketrates.Therefore,models
of the transmission mechanism of monetary policy in bank based systems should generally take into
accountthatloweringinterestrateslowersseignorageincomeofbanks,andthereforeexertsdownward
pressureonprofitswithpossiblesideeffectsonthetransmissionofpolicyratestobanklendingrates.
BrunnermeierandKoby(2016)presentamicroeconomicanalysisofthisissue.
7
Incaseoflargeexcessreservesofbankswiththecentralbank(e.g.reflectingalargescaleassetpurchase
program) combined with negative remuneration of excess reserves (such as applied in e.g. Denmark,
Sweden,Switzerlandortheeuroarea),bankprofitabilitycouldsufferinadditionfromthenegativecarry
betweenbankliabilitiesandtheseexcessreserves.Anumberofcentralbankshaveacknowledgedthisto
beapotentialissueandhavetriedtoaddressitwithanOFinnovation,namelythroughsocalledexcess
reserves tiering systems which exempt a part of the excess reserves from the application of negative
interest rates. The idea is that a tiering system would allow the combination of (i) negative rates still
effectiveatthemarginandthereforepassedontomoneyandcapitalmarkets,and(ii)theexemptionof
parts of the excess reserves moderating negative effects on bank profitability that could weaken the
effectivenessofNIRP.Bydisconnectingthetwo,thetransmissionofNIRPtobanklendingratescouldbe
improved,andtheELB(atwhichfurtherratecutsarenolongereffectiveintermsofloweringfundingcosts
oftherealeconomy)couldbelowered.Theneedtostudyfurthertieringsystemsandtopossiblyhave
blueprintsforthemisthereforeoneconcreteOFlessonfromtherecent(orongoing)ZLBexperiences.If
tieringsystemsallowpushingdowntheELBby25,50oreven100basispoints,thentheyreallywould
makeadifferenceinaworldinwhichtheELBisconsideredamajormacroeconomicissue(asprominent
macroeconomistshaveargued).
Avoidcontributingtothebuildupoffuturefinancialcrisis
Onekeyelementoftheofficialsectorsstrategytomakearepeatofthe2008financialcrisislesslikelyhas
beenBaselIII(withinparticularhighercapitalrequirements,theideaofcountercyclicalcapitalbuffers,
theleverageratio,andliquidityregulation).However,someobservershavealsoraisedthequestionwhat
rolemonetarypolicyanditsimplementationmighthaveplayedintermsoffavoringtheexcessiveleverage
which is said to have contributed to the crisis. The US Fed has for example been criticized for holding
interestratestoolowinthefirsthalfofthelastdecade,andforhavingoverlookedthebuiltupofthesub
prime mortgage problem (see e.g. Danthine, 2012, for a survey of views regarding central bank
contributionstothebuildupofthefinancialcrisis).Thatsaid,theFedspre2007OFhasnotbeencriticized
forhavingcontributedtothecrisis.Incontrast,BuiterandSiebert(2005)arguedthattheECBcollateral
frameworkwouldhavecontributedtoartificialsovereignspreadcompressionintheeuroareaandthereby
couldhaveunderminedmarketdiscipline(BindseilandPapadia,2006,providedsomecounterarguments).
Nyborg(2015)alsoarguesthattheECBcollateralframeworkcouldbedetrimentaltofinancialstabilityas
itwouldunderminemarketdiscipline.
Moregenerally,itcouldbearguedthattheextenttowhichbanksanticipatetohaveaccesstothecentral
bankaslenderoflastresortwillobviouslyinfluencebanksintheirdecisionstoundertakematurityand
liquiditytransformation(seee.g.Bindseil,2013).Therefore,asthe2008crisisisusuallyinterpretedasa
consequence of excessive such transformation (seen as a symptom of moral hazard), one may ask
whether perceptions of banks regarding central banks readiness to act as LOLR (as also builtin for
example in the collateral framework) could have been part of the problem. The facts do not seem to
supportthisintuition.TheoriginoftheliquiditycrisishavebeeninparticularintheUS,andtheFedhada
rather restrictive LOLR framework. Also, the crisis then spread to countries with very diverse LOLR
frameworkswithoutthelatterseemingtohaveexplanatorypower.
ThedesignoftheOFwouldideallybesupportivetothebankingsystemsabilitytoprovidematurityand
liquiditytransformationattheserviceofsociety,whilenotgoingasfarastofacilitateexcessiveleverage
andmoralhazard.Ifprominenceisgiventothemoralhazardproblem,thenanOFcouldincludecentral
banktoolsprovidingincentivesagainstsuchabehaviorinadditiontoregulation.Thiswillbetakenupin
section6.
Technicalprogressthattookplacebothbeforeandafter2007
8
The last decades have seen unprecedented IT developments, which strongly affect a practical and
informationintensiveactivitylikemonetarypolicyimplementation.First,therecouldbedirecteffectsof
ITprogressonOFdesigninthesensethattechnologicalconstraintsatthelevelofthecentralbankto
constructanoptimalOFcanbeovercome.Second,progressinITandcommunicationhavesomeindirect
effectsontheOF,forinstanceviaimpliedprogressinmarketinfrastructures,paymentsystems,settlement
links,etc.whichwouldchangethedemandforexcessreservesandcollateralscarcity.
Copewiththenewregulatoryenvironmentanditsimplicationsonfinancialmarkets.
Postcrisis prudential and market regulation have profound effects on market functioning and market
liquidity.Ramificationsofliquidityregulationprovedcomplexandsomebelievethatliquidityregulation
couldmakeliquiditystressedbanksrelymore,andnotless,oncentralbanks(despitethefactthatthe
initial intention was the opposite; see e.g. Bindseil and Lamoot, 2011, or ECB, 2016). CGFS (2015, 2)
concludesasfollowswiththisregard:Manyofthenewregulationswillincreasethetendencyofbanks
totakerecoursetothecentralbankasanintermediaryinfinancialmarketsatrendthatthecentralbank
caneitheraccommodateorresist.Weakenedincentivesforarbitrageandgreaterdifficultyofforecasting
thelevelofreservebalances,forexample,mayleadcentralbankstodecidetointeractwithawidersetof
counterparties or in a wider set of markets. Also, market liquidity is expected to decline at least in a
numberofmarketsegmentswithimplicationsonOFdesign.ForexampleCGFS(2015,15)predictsthat
money market turnover will shrink as a consequence of new regulation while the volatility of money
market rates will increase. Negative effects of regulation on bond market liquidity have also recently
receivedconsiderableattention(e.g.PwC,2015,CGFS,2016,TrebbiandXiao,2015)withmixedevidence.
QuestioninOFdesignrelatingtothenewregulatoryenvironmentanditsmarketimpactwillthusbe:(i)
WhataretheimplicationsonOFdesignofapossiblestrongadditionaldemandbybanksforreserveswith
the central bank for the purpose to fulfill liquidity regulation? For example, can this undermine the
optimalityofthesymmetriccorridorapproach?ThisquestionisanalyzedforexamplebyBechandKeister,
2015, who conclude that indeed central banks may have to adjust the way they control the overnight
interestrate.(ii)Ifcertainmoneymarketsegmentsshrinkparticularlyduetoregulation(e.g.theunsecured
moneymarket),couldthisimplytheneedtochangethechoiceoftheoperationaltarget(e.g.movetoa
securedinterbankrate)?(iii)Shouldthecentralbanktoleratethatrecoursetocentralbankcreditbysome
banksincreasesasaconsequenceofregulation,orshoulditgoagainstiteitherbyadjustingtheOFto
strengthendisincentives,orbyrequestingchangestoregulation?(iv)ShouldcentralbanksintheOFdesign
andmonetarypolicyimplementationpractice,facilitateornotbanksfulfilmentofBaselIIIliquidityratios?
There is surprisingly little clarity amongst central bankers how to answer in particular the last two
questions.
3. GuidingprinciplestoevaluatealternativeOFs
HowshouldoneevaluatetheperformanceofprecrisisOFs,andwhatobjectivesshouldweaimatwhen
designingthefuturepostZLB/postexcessliquidityframeworks?Onsomeoccasions,centralbankshave
been explicit on the principles that they applied to derive their OFs, such as for example the Federal
ReserveSystem(2002,11),ortheECB(2011,9495).Borio(1997,910)arguesthatframeworks(i)need
tobeeffective(i.e.allowtohaveadequatecontrolovermonetaryconditions),(ii)thattheirdesignalso
mattersbecauseithassignificantimplicationsfortheorganizationandfunctioningofmoneyandeven
capital markets as well as for asset price volatility and (iii) that they need to be explained well as
misunderstandingsaboutmonetarypolicyimplementationareoftenthesourceofmonetaryeconomic
misconceptions.Inthepast,ithashoweveralsooftenbeenarguedindifferentvariantsthatitisdifficult
to come up with general criteria to guide the design of OFs, whereby this view has been articulated
accordingtothefollowingvariants.
Relativism.Accordingtothisview,theoptimalframeworkdependsoncircumstances.Forexamplethe
BoardofGovernors(1994,35)arguedthatIngeneral,nooneapproachtoimplementingmonetary
policyislikelytobesatisfactoryunderalleconomicandfinancialcircumstances.Theactualapproach
hasbeenadaptedatvarioustimesinlightofdifferentconsiderations,suchastheneedtocombat
inflation,thedesiretoencouragesustainableeconomicgrowth,uncertaintiesrelatedtoinstitutional
change,andevidentshiftsinthepublicsattitudestowardtheuseofmoney.Wheneconomicand
financialconditionswarrantclosecontrolofamonetaryaggregate,moreemphasismaybeplacedon
guidingopenmarketoperationsbyafairlystricttargetingofreserves.Inothercircumstances,amore
flexible approach to managing reserves may be required. Such arguments seem to have been
overstretchedinthepastandrandomorhistoryrelatedheterogeneityacrosscentralbanksofsimilar
monetaryareasoracrosstimeweretooconvenientlyrationalized.
Agnosticism. The lack of clear conclusions on OF design in the literature seems to reflect a lack of
convictionorfaithintoourabilitytocometofirmconclusionsinthisfield.AstheBankofEngland
(2015) put recently in its research agenda, there remain big questions in many fields of central
banking,includinginmonetarypolicyimplementation.Theinherentdifficultiesinanysocialscience
andspecificallythemultipleequilibriumsandnonlinearityinfinancialmarketsshouldindeedsuggest
ahumbleattitudeonoptimalOFdesign.Ontheotherside,muchofmonetarypolicyimplementation
appearstobeofmanageablecomplexity,relativetootherissuesineconomicsorscience,andthere
donotseemtobegoodreasonstobeparticularlyfatalistic.
Claimofirrelevance:Accordingtothisviewitdoesnotreallymatteranyway;onlytheoutcomematters
intermsof monetarystance,and thiscanbeachievedthrough manyOFs.WhileBorio(1997)was
stronglyaffirmativewithregardtotheusefulnessofstudyingOFdesign,Borio(2001,5)seemstobe
closer to such an irrelevance view: Just as there are a hundred ways to skin a cat, so there area
hundredwaystoimplementmonetarypolicy.Thesemaydifferconsiderablyintermsoftheinterest
ratesthatarethefocusofpolicy,therangeofinstrumentsemployed,thefrequencyofoperations,
thespectrumofcounterpartiesandothertechnicalelements.Suchdifferencesreflectamixtureof
purelyhistoricalfactorsanddifferentviewsregardingthefinebalancebetweentheprosandconsof
thevariouschoices.
AnypaperondesigningOFsshouldobviouslybemotivatedbythedesiretocontradictsuchviewsandto
work towards a universal idea of how a good OF should be designed. In the rest of this section, nine
evaluationcriteriaofOFs(orobjectivesofOFdesign)willbepresented,thatwillguidethesubsequent
discussioninsections46.Wegroupthenineevaluationcriteriaintothreecategories:monetarypolicy,
financial,andfundamental,asvisualizedinthefollowingfigure.WearelookingideallyforanOFinthe
10
intersectionofthenineobjectives.Inpractice,therewillbetradeoffs,andhowthesewilltranslateinto
goodcompromiseswillalsodependontheenvironment.
Figure1:Evaluationcriteria(objectives)ofOFdesignandcriteriaforOFassessment
Effectivecontroloftheovernightinterestrate(ONR)
ThefirstandmostobviouscriteriaagainstwhichtoevaluateanOFisitseffectivenessfromtheperspective
of the degree of control of the operational target of monetary policy. In normal times, this means
essentially the ability to control short term interest rates and can be measured by comparing an
announced operational target interest rate with the actual interest rate. An early example of a cross
comparisonofthemeanandvolatilityofdeviationsfromactualshortterminterestratesfromtargetrates
is Ayuso, Haldane and Restoy (1997) and a comprehensive update can be found in Sellin and Sommar
(2014, 128 and 132). These studies indicate that systematic deviations of overnight rates from the
operationaltargetrateseemtobelimitedinalmostallcases(intheonedigitbasispointarea),suggesting
theeffectiveness,onaverage,ofinterestratecontrol.Mostofthetime,thevolatilityofthespreadisalso
limited, and when it is not (such as in the UK before 2006), such volatility is still not systematically
transmittedbeyondtheshortestmaturity.Interestingly,Ayuso,HaldaneandRestoy(1997)hadstillcome
todifferentconclusions,namelythat(p.20):wehavefoundasignificantvolatilitytransmissioneffect
from overnight to longer term money market rates for France, Spain and the UK. This evidence is
supportiveoftheimportanceattributedbymostcentralbankstoachievingareasonablelevelofstability
in the very shortterm money markets. It follows that since overnight interest rate volatility is not
completely internalized in that market, it is likely to influence real saving and investment decisions.
Whiletheseconclusionsdonotseemtobevalidatedbymorerecentdata(e.g.bySellinandSommar,
2014),theyrecallwhythecontrolofshortterminterestratesmattersandistobeseenasanimportant
objectiveofOFdesign.
11
Supportsstanceandtransmissionofmonetarypolicy
ThissecondmonetarypolicyobjectiveofOFdesignislessobviousthanthepreviousone,inparticularif
onebelievesthattheovernightinterestrateshouldbetheonlyoperationaltargetofmonetarypolicyand
that the rest of the transmission mechanism is, in normal times, not a matter of monetary policy
implementation,butanissueofmonetarymacroeconomicsthatfindsitswaybacktomonetarypolicy
implementationonlyindirectly,i.e.viaitsimpactonsettingtheoptimalleveloftheovernightinterestrate.
However, as sections 5 and 6 will argue, monetary conditions are also inevitably affected by other
elementsoftheOF,namelythelenderoflastresort(LOLR)contentoftheOF,andtheduration,creditand
liquidityriskcontainedinapossibleoutrightportfolioheldbythecentralbank.Therefore,theseother
elementsoftheOFshouldatleastnotinterferewiththestanceandtransmissionofmonetarypolicy,and,
ifapplicable(e.g.atthezerolowerbound),shouldpossiblysupportit.
Leanness(efficiency,parsimony)
Aframeworkshouldallowachievingtheobjectivesofmonetarypolicyimplementationwithfewsimple
instruments, avoiding complexities and redundancies. How opposed one feels towards complexity is a
matteroftaste,orofphilosophy,asmentionedalreadyinsection1.Inanycase,mostwouldtodayagree
thatframeworksshouldavoidthebaroquearchitectureandmultitudeofinstrumentsthatprevailedinthe
1960sand1970s.Synonymsforleaninthissenseareefficientandparsimonious.Moreover,acomplex
OFisoftenalsoopaque,andbenefitsmoresophisticatedmarketparticipantswhoareinabetterposition
tounderstand,orwhohavetheeconomiesofscalemakingitworthinvestinginunderstanding.Itthereby
encouragesinvestmentintoforeknowledgeinthesenseofHirshleifer(1971)whichisinefficientfrom
societys perspective. Compared with the preceding decades, monetary policy implementation
approachesintheperiod19902007wererelativelywellfocusedandtransparent(seesubsection2.1).
Still, particularly central banks from the large monetary areas had preserved in 2006 the luxury of a
numberofcomplexities.First,theuseofreserverequirementswithaveragingtoabsorbliquidityshocks
createdanontrivialintertemporalstructureofreservedemand.Second,operationalframeworkswere
oftenspecifiedinsuchawaythatsignificantdiscretionwasexertedinconductingopenmarketoperations
(seethenextpointonautomation).
Automation(strictrelianceonrules,avoidanceofdiscretioninmonetarypolicyimplementation)
Automation,i.e.beingstrictlyrulebased,requireshavingachievedahighlevelofunderstandingofthe
economic relationships at stake. Discretion may sometimes be unavoidable, but often actually reflects
inabilityonthesideofthecentralbanktounderstandwellandpreprogramitsinteractionwiththe
market.Rulesversusdiscretionhasbeenanimportanttopicinmonetarymacroeconomicsfordecades
(e.g.seeFischer,1988,Taylor,1999).Comparedtomacroeconomics,monetarypolicyimplementationin
advancedlargemonetaryareasdoesnotappearsocomplexthatitcouldnot,atleasttoaverylargeextent,
berulebased.Asoneexample,centralbanksbefore2007normallyusedauctionproceduresinwhichthe
allotment decision contained discretionary elements. When bidding in such auctions, banks had to
speculateaboutwhatquantitythecentralbankwouldallot.Itseemspreferablethatcentralbanksrelyon
automatedallotmentprocedures,suchasfixedratefullallotment,orvariableratetenderswithprefixed
allotmentvolumederivedmechanicallyfromforecastsofautonomousfactors.
Universality
A framework should ideally allow an efficient control of the operational target and achieve other
objectives across different financial and macroeconomic environments. Ideally, a universal framework
could thus be applied to diverse settings across various dimensions (high/low/negative interest rates,
monetary tightening/monetary easing phase, liquidity surplus/deficit of the banking system; financial
stability/stress phase, large/small economy, advanced/emerging/developing economy, free/managed
12
float,oligopolisticvs.competitivebankingsystem,more/lessregulation).Universalityisdesirablebecause
itavoidsrenewedandneverendingtrialanderroracrosstimeandacrossmonetaryareas.
Often,centralbankshavedefendedchangesintheirframeworksordifferencesrelativetoframeworksin
othercountriesbysayingthatoneapproachcannotfitall,andthattheyhadtoadapttotheenvironment
inwhichtheyoperate(whatwasreferredtoabovetoasrelativistic).Alsononcentralbankauthorshave
doubtedthefeasibilityofauniversalOF.Borio(1997)arguesthatchangesinoperatingprocedureshave
beensubstantial,havingbeendrivenbysignificantchangesthathavetakenplaceinthestructureof
financialmarketsaswellasinthebroadereconomicandpoliticalenvironment.Ho(2008)askshow
much diversity remains? And perhaps more interestingly, why does this diversity remain? (p.2). She
concludes(p.3):Thispaperfindsthat,whileanumberofcommonthemesandpracticescanbeidentified,
thereisnouniquebestwaytoimplementmonetarypolicy.Evenamongjustthefourmajorindustrial
economycentralbanksinthesample,considerabledifferencesstillexist,reflecting,interalia,differences
in the domestic financial environment, history, legal and regulatory constraints, and even political
philosophy.
Overall,itappearsthatalargepartofpastdifferencesbetweenOFsoflargeadvancedmonetaryareas
(suchastheUS,Japan,theUKandtheeuroarea)lackedagoodjustification,i.e.couldnotbeattributed
todifferentenvironments.Sometimesitisarguedthatdifferencescouldbejustifiedbythedissimilarityof
bankbasedandcapitalmarketsbasedfinancialsystems,butthelinktothedetailsoftheOFhasnever
been elaborated in detail Whether OF universality could also cover the case of very different financial
systems(e.g.ofsignificantlylessdevelopedeconomies)isadmittedlylessclear.
Honesty
Borio (1997) notes that monetary policy implementation has given rise to potential misconceptions
amongpartsoftheacademicprofessionwherebyheseemstohaveinmindwhatBasilMoore(1989)
calledverticalism(andMeigs,1962,reservepositiondoctrine).AlsoBorioconsidersthataproper
understandingofoperatingprocedurescouldthrowlightontheultimatepowerofthecentralbankto
affectmonetaryconditions,onitssource,changingcharacteristicsandreach.Onemaywanttotranslate
thatintoifacademicauthorsandcentralbankersfailtohavetherightunderstandingofthebasiclogicof
monetarypolicyoperations,howconfidentcanonebethattheirmonetarymacromodelsmakesense?
AhonestOFcanbedefinedasonethatcanbeexplainedsincerelyandconvincinglynotonlytoexpertsin
monetaryeconomics,butalsotomicroeconomists,financialmarketexperts,bankersandcentralbank
practitioners. Examples of nonhonest communication on monetary policy implementation have been
noted on a few occasions. Consider briefly two 20th century examples which have both to do with
verticalism:
1) Theinitialfabricationofreservepositiondoctrineafter1919.FriedmanandSchwartz(1963,250),who
areotherwisenosupportersofinterestratesasoperationaltargets,areastonishedbythefactthat
e.g.inthe1921annualreportsoftheBoardofGovernors,explicitdiscussionoftheFedsaggressive
hikingofinterestratesafterNovember1919andtheimplieddeflationandrecessionisavoided:Itis
hardtoescapetheconclusionthatthisisdesignedtoturnasidecriticismwithouteithermeeting
them or making explicit misstatements For example, in the whole ninepage section, neither the
wordsdiscountratenoranysynonymsoccursItisnaturalhumantendencytotakecreditforgood
outcomesandseektoavoidtheblameforbad.
2) TheVolckerepisodeofNonborrowedreservestargeting(19791982).AlthoughVolckeristheheroof
breakingthetrendtowardseverhigherinflationinthe1970sandlaunchingatrendofeverfalling
inflation rates, todays views on the Feds operational target choice in the 197982 episode (non
borrowedreserves)tendtobecritical.Some,asforinstanceGoodhart(2001)andMishkin(2004)
arguethatthewholeapproachwasjustaboutavoidingtheFedtotakeresponsibilityforthenecessary
13
stronghikingorinterestratestobringdowninflation,andtheassociatedeconomiceffectssuchasa
strongriseinunemployment.InthewordsofGoodhart(2001),theepisode,ifproperlyanalyzed,
reveal that the Fed continued to use interest rates as its fundamental modus operandi, even if it
dressedupitsactivitiesunderthemaskofmonetarybasecontroltherewasadegreeofplayacting,
evendeceptionThesmokescreencreatedbyVolckerwouldthushavebeensimplyanecessary
conditionforbringinginflationtoanendunderconditionsofimperfectcentralbankindependence.
Thepricewouldhavebeentogiveupatleasttemporarytheprinciplesofhonesty.
Financialefficiency
Financial efficiency is about ensuring, for a given achievement of the other objectives, an adequate
financialreturnontheenormousresourceswithwhichthecentralbankisnormallyendowed(relatingto
banknoteissuance,reserverequirements,andcapital).Iftwoframeworksallowforthesamedegreeof
achievementofpolicyobjectives,butthefirstleadstohigherfinancialreturnsthanthesecond,thenthe
secondcannotbeefficient.Actually,thecentralbankhasconsiderableleewayfromthepolicyperspective
inchoosingitsassets,asthescarcityofcentralbankreserves(andhencetheirprice,theovernightrate)
canbesteeredessentiallyatthemargin(andevenatthemarginitcanbesteeredwithdifferentfinancial
instruments).Financialrisktakingcanalsobesubsumedunderfinancialefficiencyofcentralbankasset
choices:expectedreturnshavetobeputinrelationtorisktaking,wherebythecentralbankspecificities
(notbeingsubjecttoliquidityrisk,havingalonginvestmenthorizon,havingprobablynooutstandingskills
andagilityincreditriskmanagement)shouldbedulyreflected.
Financialneutrality,includingexposurestothegovernment
ChoiceswithregardtotheOF(includingthecollateralframework)andcentralbankassetcomposition
shouldavoidhavingdistortingeffectsontherelativepricesoffinancialassets1arelevantissuebecause
of the (potential) scale of central bank assets and the significance of central banks in financial stress
situations.Intwodimensions,opposingphilosophiescontinuetoprevailonhowinpracticetounderstand
theobjectiveofneutrality.Thefirstdimensionrelatestotherelationshipbetweenthecentralbankand
financialliabilitiesofthegovernment,whilethesecondhastodowiththegeneralfinancialpreferences
ofthecentralbank.Considerthetwosubsequently.Withregardtothetreatmentofgovernmentliabilities,
the central bank independence view states that the central bank should treat the government as a
normalissuerandcertainlynotoverweight,orpreferablyunderweightitsroleinoutrightportfoliosorin
thecollateralframework.Otherwise,itwouldsubsidizethegovernmentbyloweringthegovernments
fundingcostsrelativetootherdebtors,andthecentralbankwillmakeitselfdependentofthegovernment
orevenfallunderfiscaldominance.Undertheoppositeconsolidatedofficialsectorview,thecentral
bankshouldaimataleanconsolidatedofficialsectorbalancesheet.Thisapproachimpliesapreference
ofthecentralbankforexposurestothegovernment.Largeholdingsofprivateassetswouldbeconsidered
under this approach to distort private debt instruments prices to the upside relative to state debt.
Unfortunately,bothviews,whichmayalsobecalledtheGerman2andtheUSphilosophies,leadto
opposite conclusions on neutrality. With regards to the second dimension, the revealed financial
1
Ofcoursethemonetarypolicystanceisnot,andshouldnotbe,neutralonrelativeassetprices.
TheGermantraditionseemstooriginatefromtheexperienceofmonetaryfinancingoftwoworldwarsandthe
subsequentalmosttotallossofpurchasingpowerofmoney,oncein1923,andoncein1945/1948.TheECBsdoctrine
maybeconsideredasadescendentfromtheGermanone.Independentofhowthecentralbankperceivestherisks
offiscaldominance,ifitoperatesinaneconomywithinsufficientamountsofeconomywideriskfreeassets(likethe
Eurosystem),theCBispartofstatesectorapproachdoesnotworkinpracticebecause(i)thereareinsufficient
assetswhichhavethecreditqualityasthecentralbankand(ii)anydecisiononhowtoinvestamongthegovernment
sectorassetsmaycreateaccusationsofpromptinggovernmentdebtunionthroughthebackdoor.Hence,thebest
approachinpracticeforsuchacentralbankistohaveanassetportfolio(outrightandcollateral)whichspansthefull
setofassetsintheeconomy,possiblywithacutoffatagivenrisklevel(e.g.investmentgrade).
2
14
preferences,thefirstviewstatesthatacentralbankbehavesneutrallyifitactslikeanaverageinvestor
(average investor doctrine), while the second view would argue that the central bank is unique (in
particularitisneversubjecttoliquidityrisks,andisconsideredcreditriskfreebyitscounterparties)and
should reflect this in idiosyncratic preferences towards financial assets (idiosyncratic fundamentals
doctrine)asthealternativetoignoretheseuniquefeatureswouldbeartificialandinthissensedistorting.
Insum,withrespecttofinancialneutralityincludingtheattitudetowardsexposurestothegovernment,
each central bank needs for the time being to be assessed against the philosophical principles it has
adopted.Thisisnottodenythatsomedoctrineseventuallymakemoresensethanothers,inparticularin
thecontextofspecificcentralbanks.
Financialstabilityandfinancialmarketfunctioning
TheOFshouldsupportfinancialstability.Forexample,theOFshouldsetincentivesforbankstoruna
sound and secure funding model, and it should contribute to, and not undermine, the functioning of
variousfinancialmarketsegments.Fivedimensionsofthecreditoperationsofthecentralbankappear
importantinthisrespect:(i)thevarietyofassetswhichareeligibleascollateralandthehaircutsapplied;
(ii)thesetofcounterpartieswhichhaveaccesstorefinancing;(iii)thesizeand/orfrequencywithwhich
thecentralbankoffersrefinancingopportunitiestobanks;(iv)thereadinessofthecentralbanktoconduct
socalledemergencyliquidityassistance(ELA)operations,ortoprovidemarketwideassistance;(v)
the readiness of the central bank to possibly act as market maker of last resort through outright
purchasesinsecuritiesmarkets.Itiscrucialtodistinguishtheexanteandtheexpostdimensionsinthis
respect.Exante,theanticipationbybanksanddebtorsofasupportivecentralbankattitudemayonthe
positive side support the maturity and liquidity transformation services that the financial system can
delivertosociety,butonthenegativesideleadpossiblytooverleveragingandunderpricingofrisk.Ex
post,supportivemeasurestendtobeusefultopreservefinancialstabilityandavoidlargescaleeconomic
damagerelatingtowidespreaddefaultinacrisisandtheassociateddestructionoforganizationalcapital.
Supportive ex post measures might also be counterproductive if they contribute to zombify the
economy,i.e.allowtheunduepreservationofbanksandcorporateswithlowproductivityorunsound
structures.
ItisthereforeoftenarguedthattheidealOFwithrespecttosupportingfinancialstabilityisonethatallows
thecentralbanktobesupportiveinasystemiccrisis,whileatthesametimeensuringtotheextentpossible
thatthisisnotexcessivelyanticipatedbymarketparticipantsandtranslatedintoexcessiveleveraging(see
e.g.AcharyaandViswanathan,2011).Therearetwochallengesintranslatingthisintopracticaladviceto
centralbanks:first,whatisexcessiveleverage,andsecond,howtopreventfinancialinstitutionsfrom
building expectations regarding ex post central bank supportiveness and to behave accordingly? One
answertothesecondchallengeisconstructiveambiguity,whichhoweverhasanumberofdrawbacks.
Anotheranswer,atleastintheory,tobothchallenges,isBaselIIIregulation,andinparticularliquidity
regulationandtheleverageratio.First,theregulationestablisheswhatitdeemstobeexcessiveleverage
andliquiditytransformation,andsecondregulationpreventsbankstogobeyondit.Ofcourse,designing
good liquidity regulation is challenging. Liquidity transformation is a subtle economic activity, and
standardizingandcentralizingimportantpartsofitthroughregulationwillcomewithefficiencylosses(e.g.
Knig,2015).Therefore,itwillremainimportantthattheOFcontributesitselftotherightbalancebetween
supportivenessandcounterincentivesagainstexcessiverelianceontheLOLR,asacontributiontofinancial
stabilityandefficiency,andnotleavealltheburdentoregulation.Intheory,regulatorsandcentralbankers
shouldjointlyassessthisandcometocommonconclusionsthatarethenreflectedincomplementaryOF
andregulatoryapproaches.ForexampleCGFS(2015)explorestentativelysomeoftherelatedissues.
Relatingcloselytotheexantecontributiontofinancialstability,theoperationalframeworkshouldnot
undermineincentivesfor activeinterbankand capitalmarkets. Thisshouldbeachievednormallyifthe
15
centralbankdoesnotconstantlyintermediatethefinancialsystemandhasanaccordinglyshortbalance
sheet,andonlytransactswiththemarkettoprovideinnettermsthecounterpartofthemonetarybase.
Ifthecentralbankbuys,forinstance,largeamountsofassetsoutrightsuchthatallbanksareinexcess
liquidity(suchasisthecaseformanycentralbanksin20102016),thenallbankswilltendtouseatthe
marginfordaytodayliquiditymanagementthecentralbanksdepositfacility(orremunerationofexcess
reserves),andtherewillbelittleinterbankmarketactivity.Also,ifthecentralbankliquidityprovisionis
extremely elastic for any single bank (also because of a very wide collateral set), then the banks are
temptedtosteertheirliquiditytoalargeextentbyincreasingorreducingtheirrecoursetocentralbank
credit,insteadofadjustingpermanentlythroughinterbankandcapitalmarkets.Itshouldbetakeninto
accountthatasolidinterbankandcapitalmarketaccesscannotbegainedatshortnotice,asthereadiness
ofinvestorsandotherbankstohaveexposuretoacertainnamehastobebuiltupovertime.Nyborg
(2015) also pleads strongly for OFs, and in particular collateral frameworks, to not undermine market
discipline.
Anactiveinterbankmoneymarketseemsdesirableforanumberofreasons:First,itallowsmeasuring
easilytheshortterminterestrateandtherebythedegreeofachievementoftheoperationaltargetof
monetary policy. To some extent, short term interest rates can also be measured in money markets
beyondthetraditionalinterbankmarket(e.g.corporatetobankmoneymarketsandshorttermbankdebt
markets),asfarastheyexistent.Recenteffortsonthemeasurementofshorttermmoneymarketratesin
e.g.theUSandtheeuroareahavegoneintothisdirectionandsuggestthattraditionalinterbankmarkets
mightpossiblybelesscrucialthanassumedpreviouslyforthispurpose(forthecaseoftheeuro,seee.g.
EMMI, 2015). Second, it has been argued that interbank money markets are useful because they
incentivizebankstocrossmonitoreachother,whichstrengthensthemarketmechanism(e.g.Blasqueset
al,2016).Ofcourse,incentivesformonitoringbanksshouldalsobetosomeextenteffectivethroughother
bank funding markets (e.g. short term paper, bond and equity markets). But it seems plausible that
interbankmarkets,asapeertopeermarket,havesomethingparticularinthisrespect.Ifmonitoringis
effective and long term relationships (and sufficient limits) are in place, this can also support financial
stabilityastemporarycapitalmarketfundingstressordepositoutflowsaffectingasinglebanksthatare
notduetofundamentalscouldbecompensatedbyinterbankmoneymarkets.Finally,anactiveinterbank
moneymarketprovidesevidenceofnettingoftheliquidityneedsofthebankingsystembeforerecourse
istakentocentralbankcredit.Inthissense,anactivemoneymarketisequivalenttotheavoidanceof
centralbankintermediationbetweenbanksandtheadditionalexposureofthecentralbankassociated
withit.
Conclusionsonobjectivesandassessmentcriteriaofoperationalframeworks
ThissectiondevelopedasetofexanteobjectivesofOFdesign,whichatthesametimecanserveasex
postevaluationcriteriaforframeworksinpractice.Themeasurabilityofthefulfillmentoftheassessment
criteria varies. Still, in most cases it appears relatively straightforward to assess concretely features of
existingframeworksagainsttheseobjectivesandcriteria.Ofcourseoftenparameterchoicesbycentral
banksinvoketradeoffsbetweenobjectives.Forexample,therewillbetradeoffsbetweeneffectiveness
ofthecontroloftheoperationtargetandpreservingmoneymarketactivity(seesection4),orbetween
leannessanduniversality,etc.Onemaythereforehavetogiveweightstothedifferentobjectivesinorder
toderiveanoptimalOFforagivenenvironment.Thenextthreesectionsofthispaper,devotedeachto
onekeyaspectofOFdesign,willbeguidedbytheobjectivesdevelopedinthepresentsection.
4. Controloftheoperationaltarget
16
4.1Introduction
Central bank liquidity management is the use of open market operations to control the scarcity of
reservesofbankswiththecentralbankinawaytohavetheirpricecloseorequaltotheoperationaltarget
level. Assummarizedinsubsection2.1,accordingtothe2007consensus,thisis mosteffectivelydone
within a corridor set by standing facilities (central bank credit operations at preannounced rates that
bankscantakerecoursetoattheirinitiative).Modelsofcontrollingshortterminterestratesthroughopen
marketoperationswithacorridorsetbystandingfacilitiescanbefounde.g.inWoodford(2003),Whitesell
(2006),EnnisandKeister(2008),FriedmanandKuttner(2010)orBindseil(2014).Inpractice,acorridor
approachneedsspecificationalongthefollowingfivekeydimensions:(i)Reserverequirementsofacertain
sizewithaveragingornoaveraging(intheformercasenormallylessthandailyopenmarketoperations,
in the latter case daily open market operations); (ii) Symmetric versus asymmetric setting of the
operational target within the corridor (and associated to this, symmetric or asymmetric liquidity
management);(iii)Widthofthecorridor(e.g.50,100,or200basispoints);(iv)Openmarketoperations:
fixedratefullallotmentorallotmentamountsetbythecentralbank;(v)Sizeofliquiditydeficitofthe
bankingsystemthatiscoveredwithcentralbankshorttermcreditoperations.
Theassessmentoftheeffectivenessofchoicesmadeshouldbedoneagainstthecriteriaestablishedin
section 3. The rest of section 4 proceeds as follows. Subsection 4.2 revisits reserve requirements with
averagingandcomestotheconclusionthattheymaynotbepartoftheoptimallongrunOF.Subsection
4.3restateswhyasymmetriccorridormaybesuperiortoanasymmetricone.Subsection4.4recallsthe
fundamentaltradeoffbetweenmarketturnoverandinterestratecontrolinherentinthesettingofthe
width of the standing facilities corridor. Subsection 4.5 discusses tender procedures. Subsection 4.6
analysesthesizeoftheliquiditydeficitofbankstobecoveredthroughcentralbankcreditoperations.
Subsection4.7discussestargetratelimitedaccess(TARALAC)facilities,whichmayberegardedasan
alternativetoreserverequirementsornarrowcorridors.Subsection4.8concludes.
4.2Reserverequirementwithaveraging
The various objectives and specifications of reserve requirements are recalled e.g. in Goodfriend and
Hargraves(1987)orBindseil(2014,chapter8).Currently,thefollowingthreeobjectivesarestillregarded
aspotentiallyrelevant:(i)Averagingtostabilisetheovernightratewithinthereservemaintenanceperiod
withouttheneedtoconductdailyopenmarketoperations.Thisisthemostprominentobjectiveofreserve
requirementsinindustrialisedcountries.Itobviouslyrequiresthatanaveragingperiod(i.e.thereserve
maintenanceperiod)isdefined.(ii)Changestructuralpositionofthebankingsystemvisvisthecentral
bank.Forexample,ifacentralbank(e.g.ofanemergingeconomy)hascumulatedlargeforeignreserves,
itmayuserequiredreservestoabsorbtheimpliedexcessreservesofthedomesticbankingsystem.(iii)
Unremuneratedreserverequirementsmaystabilise(orcounterincentivise)moneycreationbytaxing
it.Thismay,againinthecaseoflargeforeignreserves,helpthecentralbanktocounterbalancealarge
negativecarry(andhencestructuralunprofitability).
AsurveyofthespecificationsofreserverequirementsystemscanbefoundintheOFsurveysmentioned
insection1.Closeto2007,therehadbeenanumberofremarkableinnovationspioneeredinparticularby
theBankofEngland(voluntarysystem;endofmaintenanceperiodnarrowcorridor;fulfilmentcorridor).
AlsotheUSFedhadeventuallygottheapprovalfromCongresstoremuneraterequiredreserves.TheECB
improveditssystemafterfouryearsbyaligningintermeetingperiodswithreservemaintenanceperiod,
whichstoppedinstabilitiesinreservedemandandthebiddingbehaviourofbanks(asdescribede.g.in
Ayuso and Repullo, 2003). While overall, progress towards an efficient and welldesigned reserve
requirementaveragingsystemhadbeenachieved,itcanstillbearguedthatreserverequirementswith
averagingareafundamentallycomplexwaytoachievemoreovernightstabilitywithoutdailyopenmarket
17
operations.AsPerezQuirosandRodriguez(2006)haveillustrated,calculatingthroughtheoptimalreserve
fulfilment path over a maintenance period is far from obvious even under simplifying model
assumptions.Indeed,everydayofthereservemaintenanceperiodisdifferent,andtheinterestratepath
duringtheremainderofareservemaintenanceperioddependsonhistorywithinthereservemaintenance
period.ItshouldthereforebeexploredwhetherinfutureOFsthesecomplexitiescouldnotbeovercome
byrelying,insteadofonreserverequirements,onothertoolstoachieveasufficientdegreeofinterest
ratestability(e.g.dailyopenmarketoperations,narrowercorridor,TARALACfacilities)
4.3Thesymmetriccorridorapproach
Thehistoryofthoughtonthissubjectandtheevolutionofthecorridorapproachinpracticeissummarized
e.g.inBindseilandJablecki(2011)orinSellinandSomar(2014).Thesymmetriccorridorapproachhad
becomethestandardin2007andindeedseemstohavespecificadvantages,whichcanbeillustratedby
reviewingfouralternativestoit:
I.
Mixedasymmetricapproaches(e.g.inwhichthetargetrateisatvsdistancetothetwostanding
facilityrates).Thisapproachsuffersrelativetothesymmetriconefromrequiringthecentralbankto
predict(inthecaseofthecentralbankmakingtheopenmarketoperationallotmentdecisions)higher
ordermomentsofautonomousfactorshockstosteerratesprecisely.Moreover,intradaypatternsof
autonomousshockvolatilitydomatter.Inotherwords,amixedasymmetriccorridorapproachismore
complexforboththecentralbankandformarketparticipants,andwilllikely,despitehigherresources
devotedbybothsidestooptimizingwithinthisframework,leadtomoreinterestratevolatility.Even
withafullexanteunderstandingoftheprobabilitydistributionofautonomousfactorshocks,expost
theONRwillobviouslynotshowasymmetricdistributionwithinthecorridor.
II.
Fullyasymmetricapproaches(floororceilingapproaches)inwhichthetargetrateissettobeequal
(orveryclose)tooneofthestandingfacilityrates.Comparedtothisapproach,thesymmetriccorridor
approach has the advantage to better support interbank money market activity. Under a fully
asymmetricapproach,bydefinitionbankshaveonlyweakornoincentivestotradeinthemarketas
themarketrateisverycloseorequaltotherelevantstandingfacilityrate(thisisequivalenttosaying
thatbecausemostbanksareinthesameliquidityposition,thepricingofthefewtradeswiththefew
bankswhoareontheothersideofthemarketwillreflectthehugeaggregateimbalance).Thereisno
solutiontothisprobleminafullyasymmetricapproachandinparticularnotinthenewregulatory
worldinwhichinterbankexposuresaresubjecttohighercapitalcosts,sothatthehurdleforinterbank
transactionsisanywayhigh.
III.
Approacheswhichsteerovernightrateswithinacorridoroftwosamesidedstandingfacilityrates.The
classicapproachofcentralbanksupto1914consistedtosteershortterminterbankratesinacorridor
setbytwoliquidityprovidingstandingfacilities:adiscountfacility(inwhichbankscouldselltradebills
satisfying certain criteria) and a Lombard facility (an overnight lending facility based on a broad
collateralset),pricedat100basispointsabovethediscountfacility.In2015,theFedannouncedand
startedtoapplyasitspostliftoffapproachasomewhatsimilarlogicbasedontwoliquidityabsorbing
facilities.Specifically,theFedsetatargetrangeforthefederalfundsratethatis25basispointswide.
TheIOER(Interestonexcessreserves)rateandtheofferingrateassociatedwithanovernightreverse
repofacilityconstitutethetopandbottomofthetargetrange,respectively.Whilethistwosame
sidedstandingfacilitiesapproachhasprobablybeenoptimalinthepre1914worldandfortheFeds
liftofffromtheZLBinthecontextoflargeexcessreserves,itseemstohavedrawbacks(comparedto
thesymmetriccorridorapproach)whichmakeitlessobviousasafuturelongtermoptimalapproach.
Indeed,thisapproachreliesontheideathatthelessattractivelypricedofthetwofacilitiesisaccessible
withoutbindinglimitsintermsofcounterpartiesandcollateral(thelatterrelevantonlyforliquidity
providingoperations),whilethemoreattractivelypricedofthetwofacilitymustbeconstrainedby
18
some access limitations (only a limited set of counterparties, in the case of liquidity providing
operationspossiblyalsoscarcercollateral).Interbankrateswillfluctuatebetweenthetwo,andthe
positionwithinthecorridorwilldependon:(i)Howstrongareeffectivelytheaccessconstraintstothe
moreattractivelypricedfacility(howlimitedisthenumberofcounterparties,howconstrainedare
theyinarbitraging,incaseofaprovidingfacilityhowscarceeligibleassetsare).(ii)Howbigisthe
liquidity deficit or surplus that needs to be covered by the two facilities (the lower the necessary
recourse,theclosertheinterbankratewillbetotheconstrained,moreattractivelypricedfacility).In
thecaseofthepre1914OFs,mainlyavailabilityofeligibletradebillsforthediscountfacilitywasa
constraintthatoftenpushedtheinterbankovernightratetolevelsabovethediscountfacilityrate.In
thecaseoftheFedsliftoff,asPotter(2015)explains,bankonlyaccesstoIOER,creditlimitsimposed
by cash lenders, and other impediments to market competition, and the costs of balance sheet
expansionassociatedwitharbitrageactivityallcontributetothefedfundsratetobebelowtheIOR
rate(fortheexantedesignoftheliftoffframework,seeFrostetal,2015).Obviously,itischallenging
tosteertheovernightrateverypreciselythroughthisapproachasthepositionoftheovernightrate
withinthecorridorisdrivenbyvariousinstitutionalandquantitativefactorswhohaveeffectswhich
aredifficulttopredictwithprecision.Thisbeingsaid,theFedhasbeenimpressivelysuccessfulwith
thistechniqueinachievingastablefederalfundsrateafterliftoff(seee.g.FederalReserveBankof
NeyYork,2016,1112).
IV.
Finally, approaches with only one standing facility as the Feds pre2007 approach are effectively
asymmetric mixed approaches as one can interpret them as including a deposit facility at a zero
interestrate.Dependingontheabsoluteleveloftheoperationaltarget,andassumingthatthecredit
facilityrateisalwaysonehundredbasispointsabovethetargetrate,theasymmetryoftheeffective
corridorobviouslyvarieswiththestanceofmonetarypolicy,whichseemsinconvenientandwhich
violatestheseparationprinciple.
Onemaythereforeconcludethatinrelativeterms,thesymmetriccorridorcontinuestoperformwellto
reachtheobjectivesofOFdesigndiscussedinsection3.
4.4Keyparametersofcorridorapproaches
Corridorsystemsrequireinparticularthespecificationofatleastoneparameterwhichthendetermines
thevolatilityofovernightratesandthemoneymarketturnover(andassociatedwiththelatter,thedegree
ofintermediationbythecentralbank).Inthecaseofsymmetricapproaches,thekeyparameteristhe
widthofthecorridor.Inthecaseofafullyasymmetricapproach(e.g.afloorsystem),thisistheaverage
amountofrecoursetothefacilityattheoperationaltargetinterestrate(thewidthofthecorridorisalso
relevant,butnotasimportantasinthesymmetriccorridorapproach).InthecaseofaTARALACfacility
(seesubsection4.7),itisboththewidthofthecorridorsetbypenaltyratesandthelimit(thequota)
applyingtotheTARALACfacility.
In section 6.2 of Bindseil (2014), based on Bindseil and Jablecki (2011), a basic simulation approach is
explainedtounderstandtheroleofthewidthofthecorridorforthecontroloftheovernightrateandthe
preservationofinterbankmoneymarketactivity.Intherestofsection4ofthecurrentpaper,asimilar
approachwillbeusedtocaptureafewtradeoffsinthedesignoftheframeworkforthecontrolofthe
operationaltarget(seeAnnex2foradescriptionofthesimulationtoolcoveringalsothecaseofaTARALAC
facility).Thefollowingfigureshowstheresultofasimulationoftheimpactofthe(symmetric)corridor
width(xaxis)onovernightratevolatilityandmoneymarketturnover(Yaxis).
Figure 2: Expected money market turnover and interest rate volatility as functions of corridor width (underlying
specification:noTARALACfacility,volatilitiesofallliquidityshockvariablesaresetto0.5;levelofinterbankmarket
transactioncostsis10basispointsseeAnnex2).
19
Theconcave turnoverandlinearvolatility (bothasfunctionofcorridorwidth)suggest thatthere isan

optimuminteriorsolutionforthechoiceofthecorridorwidth.
This simulation approach can also be used to better understand the drawbacks of an asymmetric
approach,suchasafloorsystem,whichwerealreadymentionedinsubsection4.3.Thefollowingtable
assumesthatthecentralbankwantstomovefromasymmetriccorridorapproach(widthofcorridor:200
basispoints)toafloorsysteminwhichtheovernightrateissupposedtobeat5basispointsabovethe
depositfacility.Allliquidityshockshaveaninitialvolatilityof0.5(seeannex2forexplanations).ColumnI
showsthesystemunderthesymmetriccorridorapproach.Theneutralopenmarketoperationsvolumeis
1,theovernightratevolatilityis0.29andthemoneymarketvolumeis0.19.ColumnIIshowsthesystem
underthenewasymmetricapproach.Theopenmarketoperationsvolumeneedstobeincreasedto4in
ordertoachieveanaverageinterestratearound5basispointsclosetothedepositfacility.Themoney
market volume declines by around 70%. Scenario IIIV confirm the other weakness of asymmetric
approaches: volatility patterns of liquidity factor shocks impact on the level of overnight rates, and
therefore any change of volatility patterns requires an adjustment of the volume of open market
operationstoensurethattheoperationaltargetremainsattheadequatelevel.InscenarioIII,itisassumed
that the premoney market aggregate autonomous factor shock has 4 times the volatility of the post
moneymarketaggregateshock,whileinscenarioIVitistheotherwayround.Inbothcases,theopen
marketoperationsvolumerequiredtokeeptheovernightrate5basispointsabovethedepositfacility
ratehastoincreaseinscenarioIIIto5.8andinscenarioIVto9.Againinbothcases,moneymarket
volumesareevenlowerthaninscenarioII.InscenarioV,theaggregateautonomousfactorshocksare
againlikeinscenarioIandII,buttherelativeliquidityshockhasdoubledinsize.Inthiscasetheadequate
openmarketoperationvolumemovesto5.5.Thatvolatilitypatternsofliquidityshocksmatterforthe
adequate allotment volume is an issue as changes of volatility of liquidity shocks are unlikely to be
predictable and therefore will undermine the precision of interest rate control under any asymmetric
approach(recallthatunderasymmetricapproach,volatilitypatternsdonotmatterfortheadequateopen
marketoperationsvolume).Ofcoursethisproblemcouldbesolvedbyconstantlyprovidinghugeexcess
liquidity,suchasintheexampleconstantlysettingtheopenmarketoperationvolumeto10.Then,allthe
changes of volatility patterns captured in the scenarios below do not matter for the control of the
overnightrate(theaverageovernightratewillneverexceed2basispoint),butthepriceisthatinterbank
moneymarketactivitywillbequasinonexistent.
20
Table1:Symmetriccorridorversusfloorapproach;forlatter:impactofrelative
volatilityofliquidityshocksonadequateopenmarketoperationvolume
Scenario:
II
III
IV
OMOVolume
1.0
4.0
5.8
AFshockpreMMStddev
0.5
0.5
AFshockpostMMStddev
0.5
0.5
RelLiquShockStddev
0.5
MMtransactioncost
V
9.0
5.5
2.0
0.5
0.5
0.5
2.0
0.5
0.5
0.5
0.5
1.0
0.1
0.1
0.1
0.1
0.1
2.00
2.00
2.00
2.00
2.00
Interestratevolatility
0.29
0.09
0.18
0.02
0.14
AverageMMvolume
0.19
0.05
0.03
0.01
0.10
Averageinterestrate
0.99
0.05
0.05
0.05
0.05
Widthofcorridor
Moneymarketimplications
Onemaywonderinhowfarthetimingofadailyopenmarketoperationcaninfluencethevolatilityofthe
overnightrateandtheturnoverinthemoneymarket.Thetimingoftheopenmarketoperationcanbe
relevant for the extent to which the morning autonomous factor shock has been revealed and can
thereforebecompensatedbythecentralbankintheopenmarketoperationvolume.Thisisequivalentto
a decline of the effective volatility of the premoney market aggregate autonomous factor shock. The
following figure shows how the money market volume and the overnight rate volatility evolve if the
effectivevolatilityofthemorningaggregateshockdeclinesgradually,startingfromtheparametervalues
ofscenarioIintable1.Theresultshowninfigure3seemsverypromising,i.e.itseemsthatasolutionhas
beenfoundtoobtainafullcontrolofovernightrateswithoutimpairingmoneymarketvolumes.Eventually
whatmattersisthatwhentradingintheinterbankmarket,banksdonotbelievetohaveanyinformation
suggestingbiasedendofdayliquidityconditions.Thiscanindeedbeachievedtosomedegreebygood
forecastingofautonomousfactorshocksandanottooearlyconductoftheopenmarketoperationinthe
courseoftheday. Theremarkable effectivenessincontrollingovernightinterestrates by anumberof
centralbanksnotapplyingreserverequirementsandaveraginghaslikelybeenbasedonthisapproach
(e.g.AustraliaandCanadaaccordingtoSellinandSommar,2014,128).
21
Figure3:MMvolumeandONRvolatilityasafunctionofthevolatilityofthemorningaggregateautonomousfactor
shock (that can be influenced by the timing of the daily open market operation, i.e. the later the open market
operation, the lower the volatility of the morning shock that is not compensated by the open market operation
volume)
Finally,itisimportanttonotethatreducingbothautonomousfactorshocksbyimprovingforecastsand
therebyanticipatingbetter(andreflectinginopenmarketoperationsvolumes)boththemorningandthe
afternoon autonomous factor shocks also has positive effects, although less linearly on the side of
overnight volatility and somewhat more on the side of money market volume. The following figure
illustratesthisbyvaryingthesizeofthenonanticipatedabsoluteliquidityshock,foragivenlevelofthe
relativeliquidityshocks.Thefigurevariesonthexaxisthevolatilityofbothabsoluteliquidityshocks,while
assumingfortherestoftheparametersthesamevaluesasinthepreviousbasecase.
Figure4:MMvolumeandONRvolatilityasafunctionofthevolatilityofnonanticipatedabsoluteliquidityshocksfor
givenrelativeliquidityshocks(underlyingspecification:noTARALACfacility,volatilitiesofrelativeliquidityshocksat
0.5;levelofinterbankmarkettransactioncostsis10basispointsseeAnnex2).
22
Alsothisresultconfirmstheusefulnessofaggregateautonomousfactorforecastingfortheovernightrate
volatility/moneymarketvolumetradeoff.
4.5Tenderprocedures,maturitystructureandsizeforcreditoperations
Superiorityofautomatedallotmentmechanisms
Thefundamentalproblemwithdiscretionaryallotmentdecisionsisthatbankswillhavetotrytoanticipate
notonlyhowotherbankswillbid,butalsotheallotmentdecisionofthecentralbank.However,thinking
through the equilibrium between the central banks use of discretion and the bidding behaviour of
counterparties is almost impossible. Neither bidders nor the central bank are likely able to solve this
probleminameaningfulwayandtocoordinatewellonanequilibrium.Inaddition,incentivesarecreated
tospendresourcesonforeknowledge,moresophisticatedplayersarerewarded,andoveralltransparency
isreduced.Also,itisnotclearwhatdiscretioncanreallyaimat(foranalysisofcentralbankauctionssee
e.g. Vlimki, 2003; Ayuso, and Repullo, 2003, Nautz and Oechsler, 2003 or Bindseil, Nyborg and
Strebluaev,2009).
Optionsforautomatedtenders
Therearetwooptionsregardingautomatedtenderprocedures:Infixedratefullallotment,theinterest
rateissetbythecentralbankandthebanksgetwhatevertheybidfor(assumingthattheyhavesufficient
collateral).Thedifferencetoastandingfacilityisthatthetenderisonlyofferedatspecificpointsintime.
ThefixedratefullallotmentapproachhasbeenappliedbytheEurosystemtoalmostallcreditoperations
since2008.Thesecondautomatedtenderprocedureisvariableratefixedvolumeasappliedforexample
bytheEurosysteminallitslongertermrefinancingoperations(LTRO)between1999and2008(thetender
volumeispreannouncedbeforethebiddingstarts,andthemarginalrateisthereforeautomaticallyset
bytheintersectionbetweentheverticalsupplycurveandthedemandcurve).Themainadvantagesofthe
fixedratefullallotmentapproachare:(i)Fixingtherateseemsalogicalconsequence(intheintrapolicy
meetingperiod)iftheovernightrateistheoperationaltarget.(ii)Itisevensimplerthanvariablerate
fixedvolume(noneedforbidderstothinkaboutatwhatratetobidorabouthowotherbidderswillbid,
noneedforthecentralbanktopredictliquidityneedsandtocalibrateallotmentamounts).(iii)Incasethe
newBaselIIIregulationmakesbanksdemandforexcessreservesunpredictableassomehaveargued,the
fixedratefullallotment approach allows banks to address autonomously the issue. Advantages of the
variableratefixedvolume approach are: (i) It solves a specific potential problem of the fixedratefull
allotmentapproachthatbankscannotcoordinatebidsinawaytoensurethattheaggregatebidmatches
theactualaggregatedliquidityneedsevenifacommonforecastofaggregateneedsisavailable.Atleast,
theaggregatebidinfixedratefullallotmentoperationswillcontainsomenoisetermwhichwillaffect
liquidityconditions(seee.g.Vlimki,2003).Thisproblemwillbematerialiftherelativeliquidityshocks
arelargeanddifficulttodistinguishattheindividualbanklevelfromaggregateshocks.(ii)Somemayargue
thatthefixedratefullallotmentapproachmakeslifeofbankstooeasyintermsofcentralbankreliance
(however,themechanismsagainstsuchexcessreliancecouldbemoretargetedthanintroducingvariable
rate tenders for this purpose). (iii) Some authors have argued that the variable rate tenders are more
marketorientedandallowthecentralbanktoextractmoreinformationonmarketsandbanks.Overallit
seemsprematuretoconcludeforshorttermcreditoperationsinauniversalmannerinfavourofoneor
theotherautomatedtenderprocedures,althoughtherelativeadvantagesofthefixedratefullallotment
approachmayhavegrownbecauseofthedifficultiesforthecentralbanktopredictthedemandforexcess
reserves resulting from new Basel III regulation (see CGFS, 2015). variableratefixedvolume tenders
shouldinanycasebeusedforlongerterm creditoperations,i.e.thosegoingbeyondthe forthcoming
meetingofthepolicydecisionmakingbody.
Frequencyandmaturitiesofcreditoperations;numberofcreditoperationsoutstandinginparallel
23
Aneffective butparsimoniousapproachcanprobablybebased ontworegularoperations.Withsome

dailyliquiditybufferavailable,thiscouldbeacombinationofoneweekandthreemonthscreditoperations
(as the Eurosystem practiced it between 2003 and 2007, whereby the threemonths operations were
conductedonamonthlybasissothatalwaysthreeofthemwereoutstanding).Intheabsenceofabuffer,
dailyoperationswithovernightmaturityareneededinaddition.Relevantfactorsforchoosinganoverall
time architecture of regular tender frequencies and maturities are: (i) Time series properties of
autonomousfactorsandoftheirforecasterrors.(ii)Totalsizeofexpectedaverageliquiditydeficittobe
coveredwithcreditoperations.Thelargertheliquiditydeficit,themoreoperationsofdecentsizecanbe
outstanding in parallel. (iii) Perception of rollover risks by banks: if allotments per counterparty are
considereduncertain(asitisthecaseundervariableratetenders)banksmayhaveapreferenceforlimiting
rolloverrisks,suggestingbenefitsofkeepingthesizeofsingleoperationslimited.
In any case, parsimony should remain a guiding principle. Also, misconstructions should be avoided in
which the tender sizes drift apart, as it can happen when overlapping credit operations (e.g. weekly
operationswithtwoweekmaturity)areusedtoaddresstemporaryautonomousfactorfluctuations.
Sizeofliquiditydeficitofbankstowardscentralbanktobecoveredbyshorttermcreditoperations
Thesizeoftheliquiditydeficitofbankstowardsthecentralbankthatneedstobecoveredbyshortterm
creditoperationsisdeterminedbythesumofoutrightportfolioholdingsandlongtermcreditoperations,
relative to autonomous factors. The larger the liquidity deficit to be covered by short term credit
operations,the(i)highertherolloverriskforbanks(unlesstheoperationsareconductedasfixedrate
fullallotment,whichminimizesrolloverrisks);(ii)theshorter(everythingelseequal)theaverageduration
ofcentralbankassets(withitsconsequencesontheriskreturncharacteristicsofcentralbankbalance
sheets and the central banks contribution to maturity transformation); (iii) the more broadbased the
directtransmissionofpolicyratestomarketrates(whichmaybeimportantinlessefficientmarkets);(iv)
thelargerthebuffersforshorttermchangesinautonomousfactorsthatensurethatshorttermcredit
operationsremainofmeaningfulsize;(v)thelargerthepotentialforrelativecentralbankintermediation
withintheLOLRfunctionofthecentralbank(seesection6).Thesevariousdimensionscreatetradeoffs
whichneedtobelookedatcarefullyforanyspecificenvironment.
Amountofreservesprovidedvialongertermcreditoperations
The optimal amount of reserves provided via longer term credit operations should also be considered
withinspecifictradeoffs,andinparticular:(i)foragivenamountoftotalcreditoperations,banksperceive
longertermoperationstoreducerolloverrisks;(ii)longertermoperationsmaybeappreciatedbybanks
inthepostcrisisworldastheysupportthecompliancewithliquidityratios;(iii)alargershareoflonger
termcreditoperationsinallcreditoperationsreducestheflexibilityofthecentralbanktoreactto(non
anticipated)largeautonomousfactorchanges.
4.6Targetrate,limitedaccess(TARALAC)facilities
Reserverequirementswithaveragingprovidebuffersagainstdailyautonomousfactorshockswhichallow
thecentralbanktoachieveahighdegreeofstabilityoftheovernightinterestratewithoutdailyopen
marketoperations.However,asnotedabove,reserverequirementswithaveragingarecomplexandhave
dynamicpropertieswithinthereservemaintenanceperiod.Thissectionthereforebrieflyadvocatesan
approachadvancedforexamplebyBindseilandWrtz(2008,3137)asTARALACandactuallyapplied
since2007inNewZealandasTieringsystemandsince2013inNorwayasQuotaSystem.Athird
termthatalsohasonsubstanceasimilarmeaningistheoneofadailyremuneratedreserverequirement
systemwithfulfilmentbandi.e.inwhichpenaltyratesdonotkickinimmediatelyincaseofunderor
overfulfillmentofrequiredreserves,butonlyifthefulfilmentfallsoutofaband,suchase.g.+/10%
24
aroundtheexactreserverequirementfigure.AnadvantageofaTARALACfacilityrelativetoareserve
fulfilmentbandcouldobviouslybethattheformerdoesnotneedtobebasedonareserverequirement
system.
InNewZealand,everybankgetsitsreserveswiththecentralbankremuneratedatthetargetrateforan
amountuptoitstierallocationandbeyondthatatthelevelofadepositfacilityrate,i.e.100basispoints
below(Nield,2008).Tierallocation[is]determinedprimarilybyrevealeddemandandbehaviorinthe
paymentsystemindividuallyallocatedtiersupto$100million,$250million,thenincrementsof$250
millionto$1500million,andinstepsof$500millionthereafter.(Nield,2008,13).
NorgesBankhadpriorto2011afloorsystem,butfoundthatitmadetheredistributionofreservesinthe
interbankmarketfunctionpoorly(NorgesBank,2016).Thetwoobjectivestomovetoaquotasystem
wereaccordinglytostopthegrowthinbankreservesandgeneratemoreinterbankmarketactivityin
theovernightmarket.NorgesBank(2014,6)alsonotesthatthedrawbackofasymmetriccorridormodel
wouldbethatitplacesconsiderabledemandsonliquiditymanagementbythecentralbank,whichmust
at all times steer towards zero reserves in the banking system. According to Norges Bank (2016), the
quotasystemisacompromisebetweenafloorsystemandacorridorsystemonlyacertainamountof
eachbanksdepositswiththecentralbankaquotaisremuneratedatthekeypolicyrate.NorgesBank
(2016) explains further that Like the floor system, the quotabased system is a system with surplus
reserves(thecentralbankaimstokeepbankingsystemreservesabovezero).Anyforecastingerrorswill
thereforeaffectmarketratestoalesserextentthanunderacorridorsystem.
RecallverybrieflythelogicofaTARALACfacility(followingBindseilandWrtz,2008,3133).Assumeagain
thatthecentralbankoperatesinasymmetricapproach,i.e.thetargetinterestratei*isinthemidofa
corridorsetbythecreditanddepositfacilities.Moreoverassumethati*=0.5andiD=0,iC=1(therateof
the deposit facility and of the credit facility, respectively) and that there is only one TARALAC facility,
namelyacreditfacilityofferedataninterestratei=0.5.Letthequota(i.e.themaximumamountofcredit
providedbythestandingfacilityatthetargetrate)ofarepresentativebankbe.Forriskneutralbanks,
theinterbankrateshouldsimplybetheweightedaverageofthethreepossibleendofdaymarginalcosts
offunds,theweightsbeingtheperceivedrespectiveprobabilitiesofrecoursetoeachofthethreefacilities.
Thefundamentalequationdeterminingtheinterbankrateisi=P(short)iC+P(TARALAC)i*+P(long)iD,
wherebyP(short)istheprobabilityperceivedbymarketparticipantstohavetotakerecoursetothe
credit facility at dayend, P(TARALAC) is the probability that the TARALAC facility is used, but not
exhausted,andP(long) isthe probabilityofendingthedaywithexcessreservesbeyond ,i.e.with
recoursetothedepositfacility.InterbankratesshouldbeequaltothetargetinterestrateifP(short)=
P(long).Forsymmetricprobabilitydistributions,thismeansthatopenmarketoperationsneedtohave
injectedanamountofreservesequaltothesumofautonomousfactorsminusonehalfofthequota(
/2).Obviously,thelargerthequota(),thelargertheprobabilitythatitwillbesufficienttoabsorbthe
shocks and therefore the lower the probability that banks will need to have recourse to either of the
penaltyfacilities,andthemorestabletheovernightratewillbe.Ifratestabilitywastheonlycriterion,
then the higher the quota, the better. Unfortunately, also activity in the interbank overnight trading
dependsonthequota,andthehigherthequota,thelowerincentiveswillbetotradeinthemarket.In
fact,thewholeideaoflimitingtheaccesstoafacilityatthetargetratereflectstheideathatsuchalimit
supports the interbank market. The money market turnover interest rate volatility tradeoff can be
simulatedagainwiththeonedaymodelexplainedinannex2.Thefollowingfigureprovidestheresults
byvaryingonthexaxisthesizeofthequota.
25
Figure5:Moneymarketturnoverandovernightratevolatilityasfunctionsofthequota.(Liquidityshockvolatilities
havebeensetto0.5,interbanktransactioncoststo10basispointsandthewidthoftheinterestratecorridorto100
basispointsseeAnnex2).
BoththemoneymarketvolumeandthevolatilityoftheONRdeclinemonotonouslywiththeTARALAC
quota.Anelementofconcavityinthemoneymarketvolumecouldagainsuggestthataninterioroptimum
of the quota size exists. It could be argued that the merits of a TARALAC approach relative to reserve
requirementswithaveragingalsodependonthetimeseriespropertiesofautonomousfactorinnovations
and the volatility patterns of autonomous factors. Reserve requirements with averaging may perform
relativelywellwhenthecumulativeshocksoverseveraldaysaresmallrelativetothesumoftheabsolute
valuesoftheindividualdailyshocks.Ifdailyinnovationshaveanegativeserialcorrelation,thenaveraging
bydefinitionworkswell.Ifincontrasttheshockstendtohaveapositiveserialcorrelation,thenaveraging
hasnoadvantagetoadaily,historyfreebuffer.
Finallywebrieflyexplorewhetherthecombineduseofthetwoparameterswidthofthecorridorand
size of the TARALAC facility really make a difference. The basic test compares the frontier of MM
turnover/ONRvolatilitycombinationsthatcanbeachievedwithoutaTARALACfacilitywiththeonethat
canbeachievedwithaTARALACfacility.Specifically,wecomparethefrontiersforthecaseofallliquidity
shock volatilities being equal to 0.5, and interbank transaction costs being 10 basis points. Different
corridorwidthsaresimulatedfortwoalternativecaseswithregardtoaTARALACfacility:(i)NoTARALAC
facility;(ii)aTARALACfacilitywithaquotaof1.Thefollowingfigureseemstosuggestthatinthischosen
case,theTARALACfacilitydoesnotallowpushingouttheefficientfrontier,i.e.itwouldbearedundant
toolinthecaseoftheseenvironmentalparametervaluesandassumptions.TheabsenceofaTARALAC
facilitycouldsimplybecompensatedbyanarrowingofthecorridor.Ofcourse,thisredundancyresult
doesnotneedtobeuniversalbutmaydependonthespecificsetting.
26
Figure6:Moneymarketturnoverandovernightratevolatility:efficientcombinationsthatcanbeachievedbyvarying
the width of the corridor, (i) without TARALAC facility; (ii) a TARALAC facility with a quota =1. (Liquidity shock
volatilities have been set to 0.5, interbank transaction costs to 10 basis points and the width of the interest rate
corridorto100basispointsseeAnnex2)
Finally,itisinterestingtocheckhowaTARALACfacilityandthewidthofthecorridorinteractforalower
thandailyfrequencyofopenmarketoperations.Thiscanbetestedwiththesimulationtoolbycumulating
allautonomousfactorshocksbetweenthelastopenmarketoperationandtherelevantmoneymarket
session.Ifweassumethattheopenmarketoperationonday1targetstheaverageamountofautonomous
factorsoveranndayshorizon,thentheassumptionthatautonomousfactorsfollowamartingaleimplies
thatthesameopenmarketoperationvolumeischosenregardlessofwhetherthehorizonisonedayorn
dayswithn>1.Assumethattheaggregateautonomousfactorshocksinthemorningandafternoonof
eachdayeachhaveavolatilityofdandthatalltheseindependentshockscumulateovertime.Thenifthe
openmarketoperationisonthemorningofday1,atthemomentofthemarketsessionondaynthere
havebeen(n1)2+1independentandidenticallydistributedshockswithvolatilitydcumulating,meaning
thatthepremoneymarketshockhasacumulatedvolatility
1 2 1.Ifwetakeforexamplethe
caseofafivebusinessdaysperiodcoveredbytheopenmarketoperation,thiswouldmeanthatonday5
everythingisexactlyasiftheopenmarketoperationisinthemorningofday5,butthemorningliquidity
shockhasavolatilityof3 whiletheafternoonaggregateautonomousfactorshockhasavolatilityofd.
Thiswillobviouslymeanmorevolatility.Thefollowingchartshowshow,forastandingfacilitiescorridor
of200basispoints,themoneymarketturnoverovernightratevolatilitypairevolveoverthecourseofa
5daysperiodcoveredbyasingleopenmarketoperation.Thestartingpointofeachlinefromtheleft
represents day 1, the end point on the right represents day 5. This is shown for four quota sizes of a
TARALACfacility(0,1,2,3).Regardingtheotherparameters,weassumethatthesingle(morningaggregate,
afternoon aggregate, relative) autonomous liquidity factor shocks have a volatility of 0.5 and that
interbanktransactioncostsare10basispoints.
27
Figure7:Moneymarketturnoverandovernightratevolatilityoverafivedaysperiodcoveredbyoneopenmarket
operationandforfouralternativelevelsofaTARALACquota.Leftendofeachlineisdayone,rightendofeachlineis
day5.(Liquidityshockvolatilitieshavebeensetto0.5,interbanktransactioncoststo10basispointsandthewidthof
theinterestratecorridorto
100basispoints).
TARALACquotasseemtohaveadditionalbenefitsinthissetting.Forthedayoftheconductoftheopen
marketoperation(leftendofthefourscatterlines),weobservetheusualpatternthatahigherTARALAC
quotaimprovesovernightratestabilitybutreducesmoneymarketturnover.Forthesubsequentdays,
higherTARALACquotasallowtoavoidthefurtherdroppingofmoneymarketvolumes.Inotherwords,the
more days one open market operation has to cover, the more beneficial the TARALAC facility could
becomeatleastunderspecificparameterconstellations.
4.7Counterpartiestocentralbankcreditoperations
Thesetofcounterpartiesqualifyingforthecentralbanksstandardmonetarypolicycreditoperationsis
notonlyakeyissueforthecentralbanksLOLRframework(seesection6),butcanalsobeseenfroma
more narrow monetary policy perspective. In particular, the question arises whether the setting of
counterpartycriteriaisrelevantfortheabilitytocontroltheovernightinterestrate,orforthesmoothness
ofthetransmissionofmonetarypolicybeyondthat.Ifmarketsareperfectlyefficient,thecounterparty
eligibilitycriteriaandthesizeofthecounterpartysetprobablydonotmattertoomuch,apartmaybefor
two considerations: (i) the set should be large enough to ensure fully competitive behavior of eligible
counterparties when acting as intermediaries between the central bank and the rest of the financial
system.(ii)Ifbiddersareindividuallysubjecttonoisysignalsontheotherwisecommonvalueofthegood
inquestion(centralbankreserves),thenitmayalsobeusefultohavearelativelyhighnumberofbidders
inordertoreducethevolatilityofmarginalbidratesinvariableratetenders(independentlyofthedegree
ofexantecompetition).
If markets are modestly efficient, and e.g. intermediation costs are relatively high (maybe also due to
regulation) and unstable, then additional arguments for a broader counterparty set emerge. If
intermediarieshaveincreasingandunstablemarginalcosts,thenincreasingthenumberofentitieswith
28
direct access to the central bank reduces total intermediation costs and stabilizes the first part of the
transmission mechanism. Moreover, the information content of an operational target rate, such as an
overnight interest rate, is likely to be improved, because those entities that do not participate to the
interbankmarketandthosewhodoarelesslikelytohaveverydifferentvaluationsofovernightfunds.
Sofaronlypotentialadvantagesofabroadercounterpartysetwerementioned.Ofcoursetherearealso
costs.Indeed,wedonotobservecentralbankstograntaccesstoanextremelybroadsetofcounterparties,
such as including e.g. corporates and natural persons. Key costs are: (i) A more diverse and large
counterpartysetrequiresmoreduediligence(knowyourcounterparty)andmonitoringworkonthe
sideofthecentralbank.(ii)Counterpartiesneedtobeabletosubmiteligiblecollateral,i.e.itispointless
tograntaccesstoentitiesthatdo not haveorare unabletohandle eligible collateral (unlesstheyare
counterpartiesonlyintermsofdepositingwiththecentralbank).(iii)Grantingaccesstoentitiesthatare
nothighlyactiveinvariousfinancialmarketsegmentsandthathaveparticularlyhightransactioncosts
mayeventuallyunderminetheeffectivenessoftransmissionasliquiditybecomesfragmented.Thiswill
implyoverallhigherexcessreserves,lesspredictabilityofaggregateliquidityneeds,moreabsolutecentral
bankintermediationandmayeventuallyreduceeconomicefficiencyandendangerfinancialstability.
Inafinancialcrisis,broadeningthesetofcounterpartieshavingaccesstocentralbankcreditwilllikely
makesense,astheelementswithinthetradeoffdescribedchangetowardstheprosofabroadset.
Additionalconsiderationsapplyforthegrantingofaccesstoadepositfacility.Inparticular,itappearsthat
granting access to a deposit facility to a broad set of entities beyond banks could undermine financial
stability(seee.g.Garratetal,2015,section6.1).Whenthereisageneralperceptionofbankingsector
risks, a broad access of nonbanks to depositing with the central bank is an invitation for withdrawing
fundsfromthebankingsystem.Forexample,depositswithcentralbanksbyofficialsectorentities(e.g.
Governmentsandforeigncentralbanks)significantlyincreasedin2008andcontributedtofundinggaps
ofbanksthatthecentralbanksthenneededtofillthroughLOLRoperations.Onesolutiontothispotential
problemcouldbetoputlimitsondepositsofnonbanks.
Inthenewregulatoryenvironment,thetradeoffhasprobablychangedtowardscentralbankstendingto
increase somewhat the size of their counterparty set to reflect the higher costs of intermediation in
wholesale money markets. It is noted that the precrisis starting points of central banks were very
heterogeneous, with the Fed conducting open market operations only with primary dealers, while for
exampletheEurosystemwasreadytooperatewithanyregulatedmonetaryfinancialinstitutionthatwas
willingtofulfillbasicoperationalcriteria.
29
5. Balancesheetstructureandoutrightportfolio
5.1Introductionthereturntoaleanbalancesheet
Duringthefinancialcrisisanditszerolowerboundaftermath,centralbankbalancesheetsofindustrialised
countrieshavegainedsubstantiallyinlength.In the9 monthsafter theLehman default,central banks
intermediatedthefinancialsystemthroughLOLRoperations,whileasofthesecondhalfof2009,central
bankslengthenedtheirbalancesheetmainlythroughlargescaleassetprogrammes(LSAPSinthecase
oftheECB,LSAPSdominatedthebalancesheetonlyasof2015).Incontrast,aperfectlyleancentral
bankbalancesheetcanbedefinedasonewhichwouldhaveatotallengthclosetothevalueofbanknotes
issued.Whileincrisistimes,thereareanumberofjustificationsforlengtheningthecentralbankbalance
sheet through LSAPS or LOLR operations (e.g. Curdia and Woodford, 2010), a lean balance sheet is in
principlepositiveinnormaltimesasitsuggeststhatthecentralbankfocusesonthecoreofitsmandate.
Moreover, a lean balance sheet is a sign of wellfunctioning financial markets and a healthy economy
becausethecentralbankisneitherusedasintermediarybythebankingsystem,nordoesthecentralbank
seeaneedtoengageinspecialcrisismeasuressuchasLSAPs.Ofcourse,thereareanumberofpotential
goodreasonswhycentralbankbalancesheetsmaynotbeperfectlylean,suchas,ontheliabilityside:(i)
reserverequirementsof banks (ifone believesinthemasausefulinstrumentofmonetarypolicy);(ii)
capitalandreservesofthecentralbank(asbufferagainstfinancialrisks).Ontheassetside,besidesLSAPs,
theneedtoholdlargeforeignreserves(e.g.asprotectionagainstfuturedepreciationpressure,orasresult
ofpastpreventionofappreciationofthecurrencyinapegormanagedfloatexchangeratesystem)may
alsobeadriverofbalancesheetlengthening.Therearealsosometimeslessidealexplanationsforlong
balancesheets,suchaswhenthecentralbankisnotindependentandisinstructedbytheGovernmentto
monetise its unsustainable debt or rescue a nonviable industry. An outstandingly lean central bank
balancesheetwastheoneoftheFedprecrisis,wherethetotalbalancesheetlengthwasonlyaround1.1
timesthetotalamountofbanknotesincirculation.Ingeneral,theobjectiveofaleanbalancesheetshould
remainvalidinthefuturelongtermOF,evenifmonetaryandFXpolicies,orinsomecasesauxiliarycentral
banktasks,canjustifysomelengthening.Aswillbearguedfurtherbelow,theideathatthecentralbank
permanentlyinjectsmonetaryaccommodationthroughalongerbalancesheetwithsubstantialholdings
ofaportfoliooflessliquidassetswithlongmaturityandpossiblysomecreditriskinessdoesnotappear
sufficientlyconvincing.
5.2WhattodowiththeNPC(notpolicyconstrained)centralbankassets?
Forlargeadvancedmonetaryareas,thereisusuallyalargepartofassetsthatarenotpolicyconstrained
(NPCassets):theoperationaltargetcanbeachievedatthemarginwithsomerepooperations,andthe
needtoholdforeignreservesistypicallylimited.Therefore,thecentralbankseemstohaveleewayfor
otherconsiderationsonassetcomposition(eveniftotalassetsarenotexpandedbeyondthesumofthe
monetarybaseandcapitalandreservesofthecentralbank).Forexample,boththeFedandtheECBissue
banknotesofmorethanatrillionintheirrespectivecurrencies,buthaveforeignreservesforpotential
interventionpurposesintheorderofmagnitudeoffarlessthan10%ofthis.Also,theneedforopenmarket
creditoperationstosteerreservescarcityandthereforetheovernightrateshouldnotexceedsay10%of
theamountofbanknotesincirculation.Thiswouldimplythatforaround80%ofthe(minimal)balance
sheetlength,assetcompositioncanpotentiallybechosenonthebasisofconsiderationsbeyondmonetary
andexchangeratepolicies,asdiscussedinthefollowing.
DurationofNPCassets
One service of the financial system to society is to undertake maturity transformation, as investors
typicallywanttoholdshortterm,liquidassets,whileeconomicprojectsoftenneedfinancingforalong
periodoftime.Itcouldbearguedthatthecentralbankcancontributetothisbyholdinglongduration
30
assetsinitsbalancesheet,unlessspecificreasonsspeakagainstit.Forexample,itcouldinparticularbe
arguedthatthecentralbankcantakeconsiderabledurationriskintoitsbalancesheetascentralbanks
liabilitiestendtoberatherstable(atleastaslongasassetscanbeaccountedforatamortisedcosts).Over
thelast70years,banknotedemandhasbeensubjecttonoconsiderabledownwardsurprises,andforthe
last15yearsevenhasgrown(atleastintheUSandtheeuroarea)abovenominaltrendgrowthofGDP.
Thankstothetermspread,alongeraveragedurationofcentralbankassetswouldalsosupporttheincome
ofthecentralbankandthesizeofitsprofitdisbursementstotheGovernment.Itgoeswithoutsayingthat
a central bank with a substantial longduration outright portfolio should never put into question a
necessarytighteningofmonetarypolicybecausethismightinflictshorttermlossesonitviaitsoutright
portfolio(incasethattheaccountingtreatmentofthesecuritieswouldforeseemarkingtomarkets).
LiquidityofNPCassets
Related to the previous point, it could be argued that central banks should also hold assets of lower
liquidity. First, as just mentioned, central banks tend to have long term liabilities. Second, even if
banknoteswouldsuddenlyshrink,centralbankswouldneverbeforcedtoselltheirassetsastheycan
alwaysabsorbexcessliquiditythroughliabilitysideoperations.Thereforeholdinglessliquidassetscan
again be argued to be a contribution to the financial systems role to provide liquidity transformation
servicestosociety,or,takingadifferentperspective,itallowsgeneratingadditionalcentralbankincome
thatwilleventuallybesharedwiththetaxpayer.
DiversificationandcreditriskinessofNPCassets
AccordingtotheCAPM,everyinvestorshouldholdacombinationofthemarketportfolioandtheriskfree
asset. This idea could also be applied to the NPC assets of a central bank. Of course, in reality, most
investorsdonotholdthemarketportfolio,asthecostsofduediligenceinrealityputlimitstotheoptimum
degreeofdiversification.Onecouldarguethatthecentralbankshouldgorelativelyfarinitsdiversification
in view of the scale of its NPC assets. On the other side, the central bank as public institution has
organisationalconstraintsthatwillmakeitalessagileandsophisticatedinvestor,implyingthatitshould
beprudentandnotoverestimateitsabilitytorunahighlydiversifiedportfolio.
ExposuretogovernmentofNPCassets
Asexplainedinsection2,therearestilltwoopposingfundamentalphilosophiesamongstcentralbanks
regarding the role of government debt in the outright holdings and collateral sets of central banks
(consolidated official sector vs. central bank independence). Whichever of the two (or an
intermediate)philosophyisadoptedshouldhavelargeimplicationsontheNPCassets.Thediversification
considerationsmadeinthepreviousparagraphobviouslybecomelessrelevantundertheconsolidated
officialsectorapproach.
NeutralityofNPCassets
Theformertwoissues(diversificationandexposuretotheGovernment)alsodeterminewhatonemay
consider as neutral or nondistorting NPC asset allocation. In fact there is a further philosophical
questionthatisrelevantinthiscontext:whetheronebelievesthatitisdistortingifthecentralbank,as
largeandidiosyncraticplayer,impactsonrelativefinancialmarketprices.Accordingtoafirstview,the
NPCinvestmentapproachisnondistortingifitreflectstheidiosyncraticeconomiccharacteristicsofthe
centralbankasaninvestor.Proponentsofthisviewwillarguethatthisallowsputtingtheidiosyncratic
privilegesofcentralbanksattheserviceofsocietyalsointheareaofNPCassetallocation.Thesecond
view in contrast states that the NPC investment approach is nondistorting if the central bank broadly
behaveslikeanaverageinvestor.
Withinwelldefinedassetclasses,theconceptofneutralityseemsmucheasiertodefine.Forexample,
ifacentralbankhasconcludedtoholdaGovernmentbondportfolio,thenitseemsneutraltoreflect
31
thematuritystructureofoutstandingbondsproportionallyinthecentralbankportfolio.Suchneutrality
shouldbethestartingpoint,anddeviationsfromsuchneutralitymaybeconsideredifspecificarguments
suggestso.Similarly,incaseacentralbankrunsacorporatebondportfolio,itseemsuncontroversialthat
neutralityofissuerweightsisdefinedwithreferencetoabenchmarkofoutstandingeligiblebonds.
Expertisebuildingensuringthatthecentralbankunderstandsfinancialmarketsandstabilityissues
OnefurtherpossibleguidingprincipleinspecifyingNPCassetsofcentralbankscouldbetoensurethatthe
central bank builds up and maintains handson expertise in financial market segments that play an
important role in monetary policy transmission and that can interfere with transmission in case of
impairment.Handsoncapitalmarketsexpertisecouldbeseenaspreconditionforinterveningwithout
delayinanyofthesemarkets,i.e.ensuretheeffectivenessofthecentralbankwhenaseriousfinancial
crisis breaks out. Indeed, central banks were easily able to launch monetary policy motivated asset
purchaseprogrammessoquicklyin2009becausetheywereinvestedintomostofthese(orsimilar)asset
classesbeforehand.
NointerferenceofNPCassetmanagementwithmonetarypolicy
ObviouslythemanagementofNPCassetsshouldnotinterferewithmonetarypolicyimplementation.For
example,largepurchasesorsalesofNPCassetsshouldnotcreateautonomousfactorshocksthatimpact
ontheovernightrate.
Shouldoutrightportfoliosaimatimpactingthemonetarypolicystance?
Threepossiblereasonscouldargueinfavoroftheideathatoutrightportfolioswithduration,creditand
liquidityriskshouldpermanentlyaddmonetaryaccommodation:
Monetaryincomeargument(linkedtoargumentsstatedearlier).CallDCLacompositemeasureofthe
duration,creditandliquidityriskinacentralbanksoutrightportfolio,andassumetwoalternative
portfoliospecificationswithDCL1>DCL2.Then,thereshouldbetwopairsofoutrightportfolio/short
terminterestratecombinations,(DCL1,i1*)and(DCL2,i2*),leadingtothesamemonetaryconditions,
withi1*>i2*.Isthereanypointinpreferringonetotheother?OnemightargueinfavorofDCL1asit
seems to generate higher average income for the central bank and therefore eventually for the
taxpayer, which appears as social welfare contribution of the central bank to society (the higher
incomeresultsfromboththehigheroutrightportfolioincomeandfromthehigheraveragelevelof
shortterminterestrates).
Fasterpossibleeasingincaseofacrisis.Inaworldoflowgrowth,itmaybegoodtoprovidemore
breathingspacefortheshortterminterestratetobeloweredincaseneeded.Itcouldbearguedthat
injectingaccommodationthroughaloweringofshortterminterestratesismucheasierandfaster
thandoingsobybuildingupaportfolioofnotsoliquidandpossiblycreditriskybonds.Thereforeit
wouldbebettertodofirstandpermanentlywhatismoretimedemanding,andhavethenmoreroom
forwhatcanbedoneveryquickly.However,thisargumentseemstobeweakenedbythefactthatthe
announcement effect of large scale asset purchase programs tends to be powerful and allows for
quasiimmediateeffectiveness.
Supports financial stability. Woodford (2016) argues that outright portfolios can provide monetary
accommodationwithlessrisktofinancialstabilitythanconventional(interestrate)policies.Thisis
because outright portfolios with duration, credit and liquidity risk reduce risk premia, and thereby
dampenincentivesforfinancialintermediariestoexcessivelytakesuchrisks.
Whileallthreepointsmightbesufficienttoprovidefurtherargumentsinfavourofhavingapartofcentral
bankassetsinvestedintosecuritieswithsomeduration,creditandliquidityrisk,itislessobviousthatthey
wouldalsojustifyapermanentlengtheningofthecentralbankbalancesheet.Thiscouldbeviewedasa
return to the overambitious and oversophisticated monetary policy ambition of what was called in
section1theBaroqueageofmonetarypolicyimplementation(the1960sand1970s).Inaddition,itseems
32
clearthatvariationsofoutrightportfoliosizeandcompositionshouldnotbeconsideredastoolforvarying
thestanceofmonetarypolicyovertime.Adjustingshortterminterestratesshouldbetheonlytoolfor
thatpurposeinnormaltimes.
6. ThecollateralframeworkandtheLOLR
Thelenderoflastresort(LOLR)functionofthecentralbanktakesinparticularthreeforms:first,andthis
isthemainfocushere,someLOLRelementsarebuiltintothenormaltimesOF(e.g.elasticityofindividual
banksrecoursetocentralbankcreditasdeterminedbythecollateralframework).Second,theLOLRcan
taketheformofadecisionofthecentralbanktoexpand,incrisistimes,theLOLRcontentoftheOF(e.g.
temporarily broaden the collateral set). Ex ante, banks will have some beliefs on the readiness of the
centralbanktotakesuchmeasuresinacrisis.Third,theLOLRcantaketheformofidiosyncraticcredit
operationstosinglebanksoutsidetheapplicableOF,calledemergencyliquidityassistance(ELA)bythe
Eurosystemandbysomeothercentralbanks.Alsowithregardtothisthirdelement,bankswillhavebeliefs
ontheattitudeofthecentralbankthatwillmaterializeonceaneedforELAoccurs.Thethreecomponents
togetherwillaffect,togetherwithregulationandsupervision,theexantedecisionsofbanksontheextent
ofmaturityandliquiditytransformationtheyengagein.Section6.3willdiscussfurthertherespectiverole
ofthesethreecomponentsoftheLOLR.
Themostimportantaspect(butnottheonlyone)determiningtheLOLRcontentoftheOFisthesizeof
theeligiblecollateralset.Foragiveninitialrecourseofabanktocentralbankcredit,thissetdetermines
thebanksavailabilityofunencumberedcentralbankeligiblecollateral,andtherebytheextenttowhich
bankscancloseemergingfundinggapsthroughanincreasedrecoursetostandardcentralbankcredit.This
isthereasonfortreatingcollateralandtheLOLRfunctioninonesection.Insubsection6.1,thecollateral
framework is discussed on a standalone basis, i.e. without seeing it particularly as a part of the LOLR
framework. Subsection 6.2 turns to the LOLR question. Subsection 6.3 develops the idea of an over
proportionalborrowingframework(OPBF).
6.1Thecollateralframework
Whilehavingbeenoverlookedforalongtimeoutsidecentralbanks,theimportanceofthecentralbank
collateralframeworkforthefinancialsystemandformonetarypolicyhasrecentlyreceivedmoreattention
(e.g. Chapman et al, 2011, Bindseil, 2013, Cassola and Koulischer, 2015). Beyond the LOLR (treated in
section6.2),threemainobjectivesinthedesignofcollateralframeworksmaybedistinguished.
1. Risk protection, which is the basis for allowing unconstrained counterparty access to central bank
credit operations, including standing facilities (instead of imposing limits and foreseeing a credit
approvalprocessforeachcentralbankcredit).
2. Sufficiencyofcollateralforthemajorityofbankssothatcollateralscarcitydoesnotinterferewiththe
effectivenessoftheframeworkincontrollingtheoperationaltarget.Forexample,collateralscarcity
shouldbesufficientlyremotesoastoallowarealsymmetriccorridorapproach.Ifcollateralistoo
scarce, then neutral liquidity would imply that interbank rates are clearly above the middle of the
corridorsincethetruecostofaccessingthecreditfacilitywouldbesignificantlyhigherthanjustthe
interestratecharged.
3. Market neutrality, i.e. avoid that the collateral framework distorts relative asset prices. With this
regard,oncemoretwooppositephilosophiescanbeobservedamongstcentralbanks:Accordingto
thefirstview,thecentralbankshouldget(atleastinnormalcreditoperations)astandardinterbank
33
marketcollateralset,otherwiseitisasignofunduearbitragingofthecentralbank(oradverse
selection,orGreshamslawappliedtocentralbankcollateraluse).Accordingtothesecondview,
thecentralbankisspecialandinaworldofscarcecollateral,itshouldacceptlessliquidcollateralsince
(i)itcanimposehaircuts(whicharenotanaseffectivetooltoaddressriskconcernsifthecollateral
receiver is also subject to default risk); (ii) it is not liquidity constrained. The central bank best
contributestosocietyifitreflectsthesecomparativeadvantagesinitscollateralpolicies.Thechoice
of the doctrine will have a market impact, but both views claim neutrality. E.g. Nyborg (2015) is a
proponentofthefirstview.
Onthebasisofthesethreeobjectives,thefollowingthreequestionsrelatingtothedesignofthecollateral
frameworkemergeaskey.
Howbroadshouldacollateralsetbe?
Ifweorderpotentialcollateralfromthemosttotheleastsuitable,whereistheoptimumcutoffpoint
betweenwhatshouldbeeligibleandwhatshouldnot?Argumentsinfavourofabroadcollateralsetare:
(i)Sufficiencyofcollateraltominimisepossibleinterferenceofcollateralscarcitywithmonetarypolicy
implementation. (ii) Substantial LOLR elasticity built into normal framework supports ex post financial
stabilityandexanteahighcontributiontotheabilityofthefinancialsystemtocontributetomaturityand
liquiditytransformation.(iii)Potentialforcollateraldiversification,i.e.suchastoreducecentralbankrisk
taking (benefiting from this advantage may require to impose actual diversification of collateral pools
throughe.g.limits).(iv)Avoidanceofprivilegedtreatmentofsomeissuers;minimisecollateraleligibility
premiumandthepotentiallyassociateddistortionofprices.Argumentsagainstatoobroadcollateralset
are:(i)abroadsetmaybeperceivedbybanksasinvitationtorelyexcessivelyoneasycentralbankaccess,
insteadofdevotingresourcestomaintainingasoundmarketaccessorabroaddepositbase(itmaycreate
moral hazard, if one wants to use this term); (ii) A narrow set reduces complexity, credit risk
management issues and due diligence work for the central bank; (iii) A broad set requires more
diversificationofhaircutsandmoremonitoring;(iv)Ifthesetisnarrowingoodtimes,thenthereismore
andeasierbroadeningpotentialincrisistimes,whenneededforfinancialstabilityreason.
Whatshouldbethelevelofhaircuts?
Haircutsshouldprotectagainst(i)valuationmistakes;(ii)thepossibleexogenousdropofvalueduringthe
liquidation period; (iii) possible negative effects of the central banks collateral liquidation in case of
counterpartydefaultonassetvalues.Onestraightforwardapproachtocalibratinghaircutswouldbethat
foreveryeligibleasset,onechoosesahaircutthatatacertaingivenconfidencelevel(e.g.99%)willnot
leadtolosseswhenthecentralbankwilloptimallyliquidatetheassetafteracounterpartydefault.This
approachwouldensureexpostriskequivalenceinthesensethatafterhaircuts,alltypesofassetsare,
from the central bank perspective, equally risky (at least according to the measure chosen) when
submittedbycounterpartiesascollateral.Haircutscouldalsobehigherforassetsthataremorecostlyfor
thecentralbanktoassessandtohandle,inordertodiscouragetheuseofsuchassets,orforassetswith
regardtowhichcentralbanksfearhighuseascollateralandthereforeconcentrationrisks.However,there
wouldseemtobemoreprecisetoolstoaddressthis.Thefirstissuecouldbeaddressedbyimposingfees
to make counterparties internalise costs arising to the central bank when accepting certain types of
collateral.Thesecondissuecouldbeaddressedbylimitsorconcentrationriskaddonstohaircuts(which
wouldkickinifactualconcentrationrisksarise).
Howsegregatedshouldcollateralsetsbe?
Central banks take very different approaches in terms of segregating or not their collateral sets and
assigningthemtospecifictypesofoperations.Thefollowingthreetypesofoperationsaresegregatedin
termsofcollateralpoolbyapartofthecentralbankcommunity:(i)Shorttermcreditoperationstocontrol
theoperationaltarget(overnight,uptooneweek);(ii)Longtermcreditoperations(providinglongterm
34
fundingtobankssurprisinglycalledbytheBankofEnglandLiquidityinsurance);(iii)Liquidityproviding
standingfacilities(discountwindowintheUS,marginallendingfacilityintheeuroarea).Centralbanks
whodonotsegregatecollateralsetswouldprobablyargueasfollows:First,segregationmeansadditional
complexityandthereforeneedsstrongjustification.Second,haircutdifferentiationshouldensureexpost
riskequivalenceandtherebymakecentralbanksindifferentaboutwhattypeofcollateralisusedbythe
bank.Third,ifstillbanksoverusecertainassets,thisisnotaproblemperseasthecentralbankisspecial
anditmaybeefficientthatitendsuponaveragewithe.g.lessliquidcollateral.Finally,ifcentralbanks
endwithan unwarranteddegreeofconcentrationtosomecollateralcategory,thenitis easierandas
effectivetosetlimits(e.g.totheshareofonecollateralclassinthecollateralportfoliosubmittedbya
bank).Centralbanksinsistingontheneedtosegregatecollateralsetswillarguethatpoolingcollateralwill
inanycaseleadtooveruseofilliquidcollateralandthatthiswouldbeaformofmarketdistortion.
The Fed traditionally had a very narrow set of eligible collateral for standard open market credit
operations. It never had in normal times a second broader set of collateral for e.g. longer term credit
operations. However, it had a very broad collateral set for discount window borrowing, and discount
windowborrowingwasperceivedasrelativelyclosetoemergencyliquidityassistance(ELA).TheBankof
Englandintroducedpostcrisis(atleast)threecollateralsetsforregularoperations.Thecollateralsetfor
end of day overnight credit and the one for longer term credit operations were set as identical (but
theoretically could be distinct), and are broader than the one for the conduct of shorter term credit
operationsservingthecontroloftheovernightrate.TheECBandtheBoJhaveonlyonesetofcollateral
forallmonetarypolicycreditoperations,i.e.forcreditoperationsaimingatthecontroloftheoperational
target,forendofdayrecoursetoovernightcredit,andforlongertermrefinancingoperations.Finally,all
centralbankshavenonexplicitcollateralsetsfordiscretionaryELAoperations.Actually,inanELAcredit
operationthecentralbankcouldpotentiallyacceptanybankassetascollateral,ifthereissufficientlegal
certaintyregardingthepledgetothecentralbank.Normally,inELAoperationsthecentralbankwould
take an active approach and select the assets it would like to receive as collateral, or just take all the
unencumberedassetsforthispurpose.
Itmayappearremarkablethatnocentralbankseemstohaveoptedforaportfolioapproachtoriskcontrol,
i.e.anapproachwithhaircutsdependingalsoonoverallcollateralportfoliocharacteristics,soastoreflect
portfolio diversification or concentration effects. IT progress could suggest this to be reconsidered. Of
coursetherulesoftheportfolioapproachwouldneedtobetransparent,suchastoprovidethenecessary
predictabilitytobanks.3
It may be noted that the ECB recently introduced an approach under which haircuts are applied whenever a
counterparty submits an own covered bond. This is an example of haircut addons depending on counterparty
collateralissuercorrelation.
35
6.3Thelenderoflastresort(LOLR)withintheOF
Bagehots1873analysisoftheLOLRisstillconsideredvalidtodayanditisrarelyquestionedthattheLOLR
functionisapartofcentralbankingandbeneficialtosociety,andthatitsoptimalspecificationreflects
tradeoffsbetweenfinancialstability,centralbankrisktaking,moralhazard,preventingcostsofdefaults
and contagion, etc. An extensive modern modelbased literature has taken up these topics and has
providedsomemoreconceptualclarification.Asmentionedabove,theLOLRfunctionofthecentralbank
may take three forms, namely the builtin LOLR content in the normaltimes OF, the readiness of the
centralbanktoenhancetheLOLRcontentoftheOFincrisistimes,andthereadinessofthecentralbank
toprovideELAtoindividualbanks.Considerthethreeformsonebyone.
LOLRbuiltintothenormaltimesOF
ThereareatleastsevendeterminantsoftheLOLRcontentoftheOF.First,asmentionedearlier,collateral
availabilityprovidesafirstnaturallimittocentralbankcreditattheindividualbanklevel.Thesizeofthe
collateralsetshouldbeviewedinrelationtotheliquiditydeficitofthebankingsystemtobecoveredby
centralbankcreditoperations.Forexample,inthecaseoftheEurosystem,thenominalvalueofeligible
securitiesisaroundEUR13trillion,ofwhicharoundEUR5trillionisheldbybanks,againstaEUR0.5trillion
liquiditydeficitoftheeuroareabankingsystemtobecoveredbycreditoperations.Thisimpliesthata
representative bank could extend recourse to central bank credit approximately 10 times relative to
proportionalitybeforehittingcollateralconstraints.Second,thesizeoftheliquiditydeficittobecovered
bycreditoperationsdeterminesthepotentialrelativecentralbankintermediation.Relativecentralbank
intermediationoccurswhenliquidity constrainedbankscrowdoutlessconstrainedbanksfromcentral
bank credit without the latter yet ending in a liquidity surplus towards the central bank. The spread
betweentheinterestrateschargedoncentralbankliquidityprovidingoperationsandtheremuneration
ofexcessreservesdoesnotprovidecounterincentivesagainstrelativecentralbankintermediation.Third,
thespreadbetweenthecentralbanklendingrateandtherateatwhichexcessreservesareremunerated
determines the cost of absolute central bank intermediation as it kicks in once the strong banks are
crowded out completely from central bank credit operations. Fourth, the central bank may impose in
theory limits on banks recourse to central bank credit (beyond the collateral limit mentioned as first
point).ForexampletheReichsbankinJune1931imposedafreezeoncentralbankcreditforbanks,which
ifanythingcontributedtothesubsequentrunontheGermanbankingsystem.Fifth,thecentralbankmay
consider an overproportional borrowing framework (see subsection 6.4), variable rate tenders, or
segregated collateral sets to provide additional counterincentives against overreliance on the central
bank.Sixth,activestigmatisationordestigmatisationthroughcentralbankcommunicationwillimpacton
thepropensityofbankstorelyontheLOLR.Seventh,itmatterswhoisabletobenefitfromdirectLOLR.
TheperimeterofinstitutionsdirectlybenefittingfromtheLOLRelementsinthenormalOFdependson
thecounterpartyframeworkofthecentralbank.Blanchardetal(2014,14)notes:Thecrisishasforced
centralbankstoextendtheirliquiditysupporttonondeposittakinginstitutionsandinterveneddirectly
(withpurchases)orindirectly(throughacceptanceoftheassetsascollateral)inabroadrangeofasset
markets.Thequestioniswhetherthesepoliciesshouldbekeptintranquiltimes,whichBlanchardetal
seemtoansweraffirmatively.Forexample,accesstocentralbankcreditwillbeofinterestfornonbank
centralclearingcounterparties(CCPs).Grantingsuchaccessmayhoweverraiseissuesofequaltreatment
ofnonbankstocentralbankcredit(i.e.implyingtheneedtodefineobjectivecriteria),andmakesmore
complexthecounterpartyframework.
ReadinessofcentralbanktoaddLOLRcontenttotheOFincrisistimes
TheimpactoftheLOLRonbankbehaviorwillnotbelimitedtotheLOLRcontentoftheOFinnormaltimes.
Whatmattersaswell(forabankthatisnotcompletelymyopic)isthebanksliquidityinascenarioof
financialmarketstress.Anticipatingthiscasealsoincludesbuildingexpectationsonthereadinessofthe
36
centralbanktoaddLOLRcontenttotheOFincrisistimes,e.g.throughatemporarybroadeningofthe
collateral framework (as indeed almost all central banks did in 2007/2008). Expectations will be
determinedbyhistoricalexperienceandforwardlookingcentralbankcommunication,whichbanksmay
findcredibleornot.
Readinessofcentralbanktoprovideemergencyliquidityassistance(ELA)toindividualbanks
ELAcanbedefinedasaLOLRactivityforthebenefitofindividualbankswithafundingproblemthatisnot
rulebased,atleastnotinthesensethatitisgovernedbytheOF.OfcoursealsoELAneedstotakeplace
withinsomelegalframeworkandwithinthemandateofthecentralbankLimitationstoELAprovisioncan
resultfrom:(i)ELAcollateralrequirements(normallyELAcollateralsetsshouldbewiderthanstandard
creditoperationscollateralset).(ii)PricingofELA,i.e.whatsurchargerelativetomonetarypolicycredit
operationsisimposed.(iii)LimitationsonthedurationtoprovideELA.(iv)Limitsonthetotalshareofa
banksbalancesheetthatmaybefinancedtemporarilythroughELA.(v)PossiblerequirementthatELAis
onlygrantedifthecentralbankisprotectedinadditionbyagovernmentguarantee.Beyondadditional
riskprotection,thismaybeconsideredusefulasitrequiresanelectedgovernmenttoconfirmitsbacking
ofELAoperations(butitshouldnotdelayveryurgentandobviousELAprovisionbythecentralbank).(vi)
ELAcounterpartyset:AsbydefinitionELAisexantenotsubjecttoclearrulesandconstraints,alsothe
questionarisestowhatextentnonbanksmaybenefitfromit.Thosecouldbenonbanksthathaveaccess
tocentralbankfacilitiesinarulebasedmanner(seeabove),butalsoentitiesthataccordingtotheregular
frameworkdonothaveanycentralbankaccess.Whilebyconstructionitseemswrongtoexantedefine
theperimeterofexpansionoftheELAcounterpartyset,itseemsusefulforcentralbankstothinkthrough
theconceptual,proceduralandlegalissuesthatwouldariseifduringacrisissuchELAwouldbeconsidered.
HowmuchoverallLOLRshouldacentralbankprovide?,
AssumenowforamomentthatthecontributionofthethreeLOLRformsisnottheissue,asonewould
simplyassumethattheoptimalmixischosen,includingonalltheunderlyingdetails.Stillonecouldask
whatthetotalLOLRprovisionbythecentralbankshouldbe.Itisusefultothinkfirstabouttwoextreme
LOLRchoicesofthecentralbank.(i)MaximumLOLR:acceptinthenormaltimesOFallassetsofbanksas
collateralatfairvalues.Thiswouldallowsolventbankstofinancealltheirassetswiththecentralbank,if
desired,andnosolventcounterpartycouldeverdefaultforliquidityreasons.Furthermore,thewidthof
thestandingfacilitiescorridorissettozeroandtherearenosurchargesforoverproportionalborrowing.
TheothertwoformsofLOLR(expandingtheLOLRcontentoftheOFincrisistimesandELA)wouldnot
reallybeneededasallthesecouldaddisalreadycoveredinthenormalOF.Thisapproachwouldmaximize
theabilityofbankstoprovidematurityandliquidityservicestosociety,andinthissensemightbeargued
tobewelfaremaximizing.(ii)MinimumLOLR:thecentralbankimplementsmonetarypolicyonlyagainst
riskfreeassets,whichmayeitherbedefinedascentralgovernmentpaper,orashighlyliquidAAArated
paper.Itlargelycoversitsassetsidethroughoutrightholdingsoftheriskfreeasset.Thecentralbankmay
conductatthemarginsomereposagainstriskfreeassets,butinabilateralwayinwhichitchoosesits
counterpartiesandalwaysgoesforthemostsecureones.InthisOF,bankshavenodiscretionaryaccess
tothecentralbankatalltoclosepossiblefundinggaps,i.e.theOFhaszeroLOLRcontent.Moreover,the
centralbankwouldfullyprecommittoneverchangetheLOLRcontentofitsOFnortoeverprovideELA.
ProponentsofsuchanapproachmayarguethatasupportiveLOLRwillleadtoasmanyfinancialcrisisas
averytightone,butcrisiswillbemoremessybecausewhentheyoccurthefinancialleveragewillbemuch
higher(fourwheelvehiclesmakeyougetstuckinareaswhicharemoredifficulttoaccesswhenyouneed
toberescued).
ThelargemajorityofcentralbankersbelievethattheoptimalLOLRisinbetweenthesetwoextremes(and
regulation will have to play an additional role to achieve an overall optimum for society). The LOLR
strengthenstheabilityofthefinancialsystemtoprovidematurityandliquiditytransformationasservices
37
tosociety.Atthesametime,puttingsomelimitstotheLOLRroleisbeneficialforsociety,tohavesome
protectionagainstinformationasymmetriesandmoralhazard,toavoidrelyingexcessivelyontheabilities
of supervisors and auditors, and generally to preserve stronger incentives to maintain funding market
accessandtherebymarketdiscipline.
Assumeforamomentthatwecaptureinaunitinterval[0,1]thesupportivenessoftheLOLRframework
ofacentralbankandletthemostrestrictiveframeworkdescribedaboveberepresentedby0andthe
mostforthcomingframeworkby1.Again,itisofcourseasimplificationtoassumethatdesigningtheLOLR
isaonedimensionalproblem.ActuallyonecanimaginemappingtheLOLRunitintervalintoatleastfive
effectsofinterestwhichshouldnotbeexpectedtobeidentical,althoughoftenthisseemstobeimplicitly
assumed:
(1) Socialwelfareistheultimatemeasureofinterestandcanbeequatede.g.withtheextenttowhich
theLOLRframeworkcontributestofinancialconditionsleadingtomaximumeconomicgrowthinthe
mediumtolongterm,i.e.throughthefinancialandeconomiccycle.ForexampleKeister(2015)maps
theLOLRsupportivenessintosocialwelfare,andBindseilandJablecki(2013)mapitintogrowth.The
lattershowthatwhile differentshapesoftherelationshipbetweenLOLRintensityand growthare
possible, it is likely that the relationship is a concave function with interior maximum (i.e. an
intermediateLOLRapproachmaximizesgrowth).
(2) Centralbankrisktakingisnormallyexpectedtoincreasemonotonouslyinthe[0,1]LOLRunitinterval,
wherebyitwilltypicallyliftofffromveryclosetozeroonlybeyondacertainthreshold.Bindseiland
Jablecki(2013)alsoprovideanexampleinwhichtherelationshipisaconvexfunctionwithinterior
minimum,illustratingBagehotsintuitionthatsometimesonlythebraveplanisthesafeplanforthe
centralbank,i.e.moreforthcominglendinginacrisisreducescentralbankrisktakingrelativetoa
restrictiveapproach.
(3) Leverageofbanksandtheirabilitytoprovideliquidityandmaturitytransformationshouldincrease
monotonouslyacrosstheLOLRunitinterval.Obviously,regulationmaylimitleveragetolowerlevels.
(4) Financial fragility will probably first decrease, and then increase across the LOLR unit interval,
suggesting that a measured LOLR can stabilize the financial system while a too liberal one could
eventuallyleadtoparticularlydeepfinancialcrises.
(5) Marketdisciplineandfundingmarketfunctioningcanbethoughtofaseitherfallingmonotonously,or
as mirroring the financial fragility curve, i.e. it would benefit from some moderate LOLR, but is
underminediftheLOLRisexcessive.Bindseil(2013)showsthatwhenassetliquiditydeterioratesafter
someexogenousshock,thenasupportiveLOLRcanpreservefundingmarketaccessforsolventbanks,
butnotforinsolventbanks,whilearestrictiveLOLRwillimplyarunalsoonsolventbanks,implying
thatamoresupportiveLOLRcanbeconducivetoamoreeffectivemarketmechanismthanavery
restrictiveone.
Theexactshapesofallfivecurveswilldependonthefinancialenvironment.Thefollowingfigureprovides
an example of how the five curves may look like (the shapes that are considered most likely under
conditions of developed financial markets are drawn). In this example, the xvalue of the highest
effectiveness of market discipline is strictly positive, but below the xvalue of the minimum financial
fragility,whichitselfisbelowthepointofmaximumsocialwelfare(ormaximumlongtermgrowth).
38
Figure 8: Possible relationships between LOLR intensity of the OF and five impacted variables (social
welfare,centralbankrisktaking,leverage,financialfragility,marketdiscipline).
WhatshouldbetherelativecontributionsofthethreeformsofLOLR?
AssumethatacertainoptimaltotalLOLRprovisionhasbeenchosenwhatrelativeroleshouldthethree
formsofLOLRplay?Centralbankersconsensusisthatallthreeformsshouldplayarole.Today,relative
tothepre2007consensus,preferencemighthavemovedsomewhattowardsmoreimportanceofclear
rulesexante(andcertainlymoreimportanceisassignedtothecomplementaryroleofregulation).
WhatdistinguishesthelasttwoformsofLOLRfromthefirstistheuncertaintyattachedtothem,whichin
thepasthassometimesbeenconsideredasadvantageousconstructiveambiguity,preventingthatbanks
factorinsuchsupportexante.Indeed,manymodelscometotheconclusionthatideallytheLOLRappears
harsh ex ante but is soft ex post (e.g. Acharya, V.V. and S. Wiswanathan, 2011, or Keister, 2016). The
conceptofconstructiveambiguitymayhoweverbeconsideredtosufferfromthefollowingthreewell
known problems: First, ambiguity does not prevent banks from building expectations. Therefore,
constructiveambiguityisnotauniversalsolutiontosolvethetensionbetweentheexantepreferenceof
central bankstonothave banksrelyexcessivelyon theLOLRandthe meritsofaliberalLOLRexpost.
Constructive may instead only create additional randomness, but not be able to bias expectations
systematically. Second, central banks are anyway bound to consistency in their ELA provision as a
consequenceoftherequirementthattheofficialsectorneedstoactinaccordancewithequaltreatment
principles.Therefore,fulldiscretionisanywaynotpossible,atleastnotinELA.Third,onecouldarguethat
designingarulebasedframeworkforcesthecentralbanktoreallythinkthroughtheproblemsthatitwants
tosolveandhowitwantstodoso,whileincontrast,emphasizingconstructiveambiguityallowscentral
bankstoavoidsuchefforts..
Nevertheless,itwouldseemfromapracticalperspectivewrongandoverambitioustoaimatafullexante
setofcontingentrulesonfutureliquidityprovisionincrisissituationsthatwouldallbewritteninthestone
oftheOF,andtoprecommitagainstexpostflexibility.Whilethemechanicsoffinancialcrisishavesome
common patterns,thedetailsofeach crisistendtobedifferent,andeverystresssituationinfinancial
marketshasidiosyncraticelementsthatcannotbeanticipated.Expost,itcanbeassessedwhetherLOLR
ofform2or3addstocentralbankrisktaking,whetheritvalidatesinsomesensepastmoralhazardof
banks,andtowhatextentoverallitappearstobeintheinterestofsociety.Ifthefinancialstressoriginates
39
fromfactorsthatneitherbanks,northecentralbank,norbankingsupervisorscouldhaveanticipated,then
onemaycometotheconclusionthatthereisnoparticularmoralhazardissue,anditwillbeunlikelythat
the precrisis OF is necessarily adequate in terms of LOLR with regard to the nonanticipated crisis
situation.RiskendogeneitymayalsoimplythatadditionalexpostLOLRmaydecrease,andnotincrease
financialriskstothetaxpayer.Suchflexibility doesnotimply perseatoodovishapproach,asstillthe
centralbankmaysetthehurdleforexpostLOLRmeasureshigh,andcommunicatethistothefinancial
system. Also, such flexibility does not preclude that the central bank aims at increasing the relative
importance of rule based elements at the detriment of discretionary elements. Indeed, for the above
mentionedreasons,arulebasedapproachtoLOLRincorporatedintotheOFhassignificantadvantages,
anditshouldbeviewedasapermanentobjectiveofcentralbankstolearnfrompastcrisisandtoreflect
onhowarulebasedapproachtoLOLRmaycontributetopreventfuturecrisis.Pursuingthisaimshould
howevernotbeunderstoodasreflectingabeliefthattheaimcanbereachedtosuchanextentthatone
couldgenerallyprecommitagainstexpostflexibilityandLOLRmeasuresofform2and3.
Thechoicebetweenforms2and3ofLOLR,i.e.betweenanexpoststrengtheningofLOLRelementsinthe
OFversusELAwillbemainlydeterminedbythescopeofliquidityproblemwithinthebankingsystem.
Whileinprinciplearulebasedapproachbenefittingeveryoneequallyseems preferable,itcanalsobe
clearlyinefficientiftheliquidityproblemislimitedtooneorfewbanks.Moreover,theprovisionofELA
canbeassociatedwithspecificrequeststowardsthebenefittingbank,whichmaybeconsideredimportant
toaddressmoralhazard.Therefore,bothformsofexpostLOLRshouldremainpartofanoverallapproach.
The devil lies in the detail, and beyond these general questions on the overall LOLR provision and the
contributionofthethreedifferentformsofLOLR,amultitudeofmorespecificissuesemergesintheLOLR
field.Considerbelowanumberofsuchissues.
Bagehotsonlythebraveplanisthesafeplanandriskendogeneity
EversinceBagehotitisacknowledgedthatthecentralbankcannotmakeitsLOLRchoicesasifitwasan
atomistic investornotinfluencing the properties (e.g.inparticulardefault probabilities)ofthesystem.
Often,beingmoreforthcomingasaLOLRafteranegativefinancialstabilityshock(e.g.broadentheeligible
assetsettolessliquidassets)willdecreasefinancialrisktakingbythecentralbank,insteadofincreasing
it. It will therefore also eventually benefit the taxpayer. Risk endogeneity should lead to a more
forthcoming LOLR, i.e. the welfaremaximizing LOLR framework will be more supportive than the one
obtainedifriskendogeneityisignored.
LOLRandmonetarypolicy
TheLOLRbecomesanimportantnonstandardmonetarypolicymeasureatthelatestwhenthecentral
bank has hit the zero lower bound. Indeed, the readiness of the central bank to act as LOLR affects
intermediationspreadsofbanks,inparticularinaliquiditycrisis,andtherefore,foragivenlevelofthe
shorttermriskfreeinterestrate,theaveragefundingcostsoftherealeconomy(e.g.Bindseil,2013).
LOLRjustificationsiftheZLBisnotyetbinding
Fromamonetarypolicycentricperspective,onecouldaskwhethertheLOLRisatallneededwhenthe
central bank policy interest rate is not at the ZLB. In equilibrium, the ability of the banking system to
provide liquidity and maturity transformation to society would be lower, but the effect on financial
conditionscouldbecompensatedbythecentralbanksettinglowermonetarypolicyinterestrates.Two
counterargumentsagainstthisviewcanbeformulated:first,theloweraveragecentralbankinterestrate
leads,everythingelseequal,tolowercentralbankincome,whichmayberegardedasameasureofthe
socialcostsofsuchapolicy(ascentralbankprofitsareeventuallysharedviathegovernmentwithsociety).
Second,theLOLRplaysaroleintheoptimizationtowardsthetradeoffbetweenzombificationonone
40
side,andthecostsofdefaultsofproductivecompaniesontheotherside,suchasoutlinedinBindseiland
Jablecki(2013).InthissenseatoorestrictiveLOLRwouldleadtolowereconomicgrowthrateasregularly
too many resources get lost in liquidity induced defaults and this holds independently of the
appropriatenessofmonetaryconditionsfromtheperspectiveofpricestability.
Is it useful that the LOLR adds monetary accommodation in normal times such as to allow a higher
averagelevelofshorttermriskfreeinterestrates?
The relationship between the LOLR and interest rate policy in crisis times is discussed for example by
Freixasetal(2009).ShouldOFsaimatasufficientLOLRcontentinnormaltimesfromtheperspectiveof
lifting the operational target interest rate further away from the ZLB, even if the ZLB is not an acute
constraint (in analogy to a similar question raised in subsection 5.2 on the central banks outright
portfolio)? The LOLR function has an impact on the ability of banks to provide liquidity and maturity
transformation,i.e.onbankintermediationspreads.AssumethatL1andL2aretwoalternativeintensities
oftheLOLRfunctionwithinanOF,withL1havinghigherintensitythanL2(i.e.L1facilitatesmoreliquidity
andmaturitytransformationbybanks),andassumethat(L1,i1*)isconsideredtoprovideforanadequate
stanceofmonetarypolicy(i.e.foradequatemonetaryconditions).Then,alternatively,thereexistsanother
leveloftheoperationaltargetrate,i2*,withi2*<i1*,whichshouldallowtoreachanidentical,adequate
stance of monetary policy by being combined with L2: (i2*, L2). Is there any point in preferring one
combination to the other? Assume for a second that from a financial stability and central bank risk
management perspective, L1 an L2 are identical. Then, one might argue outside monetary policy
considerationsinfavorofL1asthehigheraverageincomeitgeneratesforthecentralbankandtherefore
eventually for the taxpayer is a measure of a social welfare contribution that the LOLR can provide to
society(ofcoursetheseresultsneedtoholdthroughthefinancialcycle).Beyondsuchmonetaryincome
arguments (which, again, actually capture a part of the LOLR contribution to social welfare), it seems
difficulttoidentifyargumentstousetheLOLRcontentoftheOFasatoolforstructurallyinfluencing,or
forsteeringthroughtimethestanceofmonetarypolicy.
Moralhazardandcentralbanklosses
A popular theme in papers on the LOLR is moral hazard, but the concept often remains vague. One
pragmaticviewisthatmoralhazardonlymaterializesifthecentralbankfacesactuallossesfromitscentral
bankoperations(orseesitsprofitreducedfromit).Thisinterpretationalsohastheadvantagethatitwould
reduce the complexity of the LOLR design problem by one dimension and map something vague and
complex(moralhazard)intosomethingconcreteandmoremeasurable(centralbankrisktakingevenif
somewhat complicated by endogeneity). If central banks are worried about moral hazard, they could
adjustthepricingoftheirLOLRservices(seesubsection6.4),ortheycouldtightenriskcontrolmeasures
(innormaltimes,tonotbeprocyclical)sothattheprobabilityofcentralbankcreditlossesdeclineseven
further.
ExcessivestigmatizationoftheLOLR?
Sometimescentralbanksworrythatbanksattachexcessivestigmatotakingrecoursetodifferentforms
ofLOLR.Forexample,recoursetotheDiscountWindowisconsideredtoremainstigmatizedintheUS
althoughtheFedwouldwanttochangethissince2002.Also,inanumberofcreditopenmarketoperations
of central banks during the financial crisis, aversion of banks to participate materialized such that the
accommodationthattheoperationsaimedatcouldnotbefullyachieved.Excessivestigmatizationseems
togointheoppositedirectionasmoralhazard,i.e.centralbanksmayfindthewillingnessofbankstotake
recoursetotheLOLReitherexcessiveorinsufficient.Centralbanksshouldthereforehavetoolsinhandto
adjust the willingness of banks to come to LOLR operations. Stigmatization through communication is
difficult to control, to finetune, and to revert. Therefore, central banks should avoid the verbal
41
stigmatizationofLOLRoperationsandinsteadrelyonformonetarydisincentivesthroughe.g.aframework
suchastheonedescribedinthenextsubsection.
LOLRandliquidityregulation
SomehavearguedthattheonlysolutiontothecentralbankersAngstfromthedangersononesideof
beingexploitedbybanks(moralhazard),andontheothersideofexcessivestigmatizationoftherecourse
totheirLOLRoperations(preventingbankstousethem),wouldbealiberalLOLRcombinedwithtight
liquidityregulation.AgoodcombinationofappropriateincentivestotakerecoursetotheLOLRandwell
deigned regulation likely allows achieving the best outcome for society. As already noted in section 3,
managing liquidity risk is a core activity of banking, and it seems unlikely that centralizing this subtle
activitythroughliquidityregulationcanbedonewithoutefficiencycosts.Thereforeliquidityregulation
must not be overburdened, and central banks providing some welldesigned and rule based economic
counterincentivestowhattheydeemanexcessiverelianceontheLOLRisanimportantcontributionto
reducetheburdenputonregulation.Infact,thecombinationofregulation(inallitsdetails)andtheLOLR
(in all its details as well) will jointly determine (i) the ability of banks to provide liquidity and maturity
transformationasservicestosociety;(ii)financialstability;(iii)centralbankrisktakingandbankspossible
moralhazard.Alotofworkisaheadtobetterunderstandtheserelationshipsandwhatconclusionsto
drawonthejointdesignoftheLOLRandofliquidityregulation.
6.4Overproportionalborrowingframework
Integratingasocalledoverproportionalborrowerframework(OPBF)intotheOFcouldbeaworthwhile
additioninordertoprovideforastrongandindependenttooltocontrolforperceivedoverrelianceof
individualbanksoncentralbankcreditandtheLOLR.Moreover,suchanapproachcouldbeinstrumental
toassignagreatershareoftheLOLRtotheOF,insteadoftouncertainexpostdecisions.Possiblereasons
todislikeoverrelianceofsinglebanksoncentralbankcreditaremultiple,andinparticular:(i)financial
risktakingfromconcentratedexposurestoweakerbanks,(ii)contributiontozombificationandweakening
ofmarketmechanism,(iii)weakeningofmonetarypolicytransmissionasatoohighshareofcentralbank
creditmaygetabsorbedbybankswithoutmuchmarketinterconnectedness.OPBFshavebeenappliedin
thepast,asillustratedbythefollowingthreeexamples.First,theIMFforeseessurchargesforlargeand
long borrowing relative to a countrys quota. Second, e.g. Goldenweiser (1949) mentions that in 1919
someFederalReserveBanksadoptedsystemsofratesgraduatedinproportiontotheamountborrowed
byanindividualbank.Third,theBundesbankappliedfordecadesaproportionalitybasedlimitsystem
(Rediskontkontingente)fortherecourseofbankstoafavorablypricedcreditfacility.ThelogicofanOPBF
canbestbeintroducedthroughasimplesystemoffinancialaccounts.Assumeafinancialsystemwithtwo
bankswhichmaybedifferentinsizebutalsointermsofrelativeshareofmarket(includingdeposits)vs.
centralbankfunding.
Figure9:FinancialaccountstoillustratemechanicsofOPBF
Bank1
LendingtocorporatesL1=D1+B1
HouseholddepositsD1
CreditfromCBB1 0
Bank1
LendingtocorporatesL2=D2+B2
Householddeposits D2
CreditfromCBB2 0
Centralbank
CreditoperationsB=B1+B2
BanknotesB
LetL=L1+L2bethetotallengthofthebankingsystembalancesheet.LetP*=B/Ldefineproportional
recoursetothecentralbank.LetPi=Bi/Libetheactualvalueoftheproportionalitymeasureofbanki.Let
42
P#bethethresholdforPibeyondwhichthecentralbankappliesanextrafee.Thecentralbankshouldgive
somecostfreeleewaytoborrowoverproportionally,i.e.P#>P*.ForPibeyondP#,surchargeswouldkick
in.ThesurchargefunctionbeyondP#couldbeanynondecliningpositivefunctionofoverproportionality.
Considerthefollowingexample:L1=L2=100(totallendingbybankstoeconomy:200);D1=95;D2=85
(totaldeposits=180);B1 =5;B2=15 (totalbanknotesB=20);ProportionalborrowingP*=10%ofbank
balancesheets;P1=5%;P2=15%.WithoutanOPBFtherewouldbeinthisexamplenofinancialincentives
forthebankstoequalizetheircentralbankrecourse,asthecorridorwidthdoesnothelpagainstrelative
centralbankintermediation.AssumehowevernowthatP#=12%andthatbeyondthatanextrainterest
rateof1%ischarged.Now,thebankingsystemhasincentivestorebalancerecoursetocentralbankcredit.
ForexampleBank1cangiveaninterbankloanof3tobank2.Alternatively,bank2couldbecomemore
aggressiveinattractingdeposits,forexamplebyincreasingitsbranchnetwork,orbypayinghigherinterest
ratesondeposits.Itwouldneedtoattractatotalof3ofextradeposits(whichwouldbewithdrawnby
depositors from bank 1). A reduction of the recourse by bank 2 to central bank credit by 3 (and a
correspondingincreasebybank1)wouldbringbothbankstobelowtheproportionalitythresholdsetby
thecentralbank.
Thesurchargeformulamaybespecifiedinrelationtotheaveragerecoursetocentralbankcreditovera
certainperiod.Thatwouldreflecttheintuitionthatshorttermpeaksincentralbankreliancearelessofa
problemthanpermanentlargescalereliance.
AdvantagesofanOPBF
Asmentionedatthebeginningofsubsection6.4,centralbankshavegoodreasonstodislikeconcentration
oftheircreditoperationstofewbanks.KeyadvantagesofanOPBFrelativetoothertoolstolimitover
proportionalityarethefollowingones.First,anOPBFisbydefinitionfocusedontheproblemitself,and
adjustingithaslesssideeffects(e.g.likeincaseonewouldadjustthecollateralframeworkforthesakeof
reducingtheleewayforoverproportionality).Second,anOPBFisunconstrainedinspecifyingasurcharge
formula,includingaveragingoverachosenperiod.Surchargescanbebilledconvenientlytobanksafter
anypossibleaveragingperiod.Finally,anOPBFdrawsattentionofbanksandotherstakeholderstothe
proportionalityissue,andallowspolicymakerstoreflecttheirviewsonthissubjectthroughawelldefined
framework(essentiallythesurchargefunction).TherelativeeffectivenessofanOPBFasatooltolimit
excessive reliance of single banks on central bank credit can further be illustrated by reviewing the
weaknessesofalternativeinstrumentsinpursuingthisobjective.
- Moralsuasion/Stigmatization:asalreadyarguedinsubsection6.3,thisisbydefinitionanimprecise
toolasitdoesnotrelyonquantificationandisdifficulttocontrolandtoadjustwhenneeded.
- Variable rate tenders are also a crude tool to control for overproportional borrowing. It may be
partiallyeffective,butitworksviathecreationofallotmentuncertainty.Itdoesnotappearparticularly
eleganttouseuncertaintytoprovideincentives,asuncertaintytypicallycomeswithsomecosts.
- The width of the standing facilities corridor is not an effective tool against relative central bank
intermediation (i.e. the crowding out of other banks from central bank credit). Assume a banking
systemwith10banksofequalsize,andthatinitiallyeachbankhas10%ofthetotalcentralbankcredit
volume.Ifthefundingdislocationislimitedtoonebank,thenthisbankcanextenditscentralbank
creditbyafactorof10beforethewidthofthecorridorstartstomatter.Clearly,anOPBFisamore
effectivetooltopreventoverproportionalityasitcankickinatanytimeearlierthanthat,aswished.
Returningtotheexample,ifthefundingdislocationismoresystemwide,andsay5ofthe10banks
sufferfromliquidityoutflowsandtheother5frominflows,thenthecorridorwidthwillkickinearlier
as an effective additional incentive. However, in this case typically central banks will prefer to be
forthcoming and toreducestigmatization,andthereforemay consider tighteningthewidthofthe
standingfacilitiescorridor.
43
ConstructiveambiguityonELA(inconjunctionwithalimitedregularcollateralset)mayalsobeviewed
as a technique to limit overreliance on central bank credit. It is a tool relying on the creation of
uncertainty,whichforreasonsmentionedearlierdoesnotseemideal.
PossibledrawbacksandchallengesofanOPBF
Anumberofpossibledrawbacksandchallengesmayhavedeterredcentralbanksfromamoresystematic
relianceonOPBFs.
First,anOPBFconstitutesanadditionalframeworkthatneedstobemaintained,andinthissenseitadds
complexity.Additionalcomplexitycanbeminimisedifthecentralbankcanrelyonexistingproportionality
measures,suchasareservebasedefinedfora(previouslyexisting)reserverequirementframework.
Second,centralbanksmayconsiderthatbanksunderliquiditystresstendtobepotentiallyalsosubjectto
aweakersolvencysituation.Thereforeimposinghighsurchargesonthemcouldseemcounterproductive
intermsofbringingthembacktomarkets.However,surchargesshouldnotrapidlymakesuchadifference
fortheoverallsolvencysituationofabank.Also,ifsolvencyistheissue,thensolvencyrelatedmechanisms
should address it. Moreover, a surcharge system that is put in place in advance creates the necessary
certainty allowing banks to factor it in in their liquidity risk management and pricing of liquidity risk.
Therefore,consequencesonprofitabilityandsolvencyofoverrelianceoncentralbankcreditshouldnot
comeasasurprisetobanks.
Third,centralbanksmayactuallyfindthatstigmaissuesundulydeterbanksfromtakingrecoursetocentral
bankcredit,andtheymayfeelthataddingsurchargeswouldmakecounterincentivesevenmoreexcessive,
throughdirecteffectsandthroughindirecteffectsviaevenmorestigmatization.However,ifthiswould
beuniversallytrue,thantheentiremoralhazarddebaterelatingtoexcessiverecoursetothecentralbank
duringthefinancialcrisisshouldnotexist.
Fourth, an OPBF needs to find a solution for the case that some banks have no intention to come
themselvestothecentralbanktoaskforcentralbankcreditbecausee.g.theydonothaveafulltreasury
departmentwithsufficientoperationalabilities.Thismaybethecaseforsmallsavingsandmutualbanks
whichrelyonaspecialisedbankhubtoaccesscentralbankcredit.Or,bankinggroupsmayhavechosen
oneentitywithinthegrouptoaccessthecentralbankforthesakeofanefficient,centralisedtreasury.
This issue can be solved by allowing banks who indeed have a specific group structure to pool their
surchargefreecentralbankcreditallowance.Theentitieswhodonotwanttocometothecentralbank
wouldneedtodeclaretheirreadinesstosharetheirallowancewiththerelatedentityaccessingcentral
bankcreditontheirbehalf.ThishasbeenpracticedforexamplerecentlyinthecaseoftheEurosystems
TLTROs,and hasprovedtobemanageable (seeECBpressreleaseof3July2014,annex,section1).In
addition,thedifferencebetweenstrictproportionalityP*andthethresholdP#atwhichsurchargeskick
inalsoallowstoalsocopewiththisphenomenon.
Fifth,itmaybenotedthatanOPBFseemssomewhatlesscompatiblewithvariableratefixedvolumethan
with fixedratefullallotment operations, as the former imply an ex ante uncertainty on the amount
allotted.Therefore,bankssubjecttoapotentiallybindingOPBFmaybidmorecarefullyinvariablerate
fixedvolumetenders.However,thisissueshouldbelimitediftheimportanceofsinglevariableratefixed
volumetendersinthetotalcentralbankcreditisnottoolarge.Moreover,again,thedifferencebetween
P*andP#shouldhelp.
44
Conclusions
Overall,anOPBFappearsaspotentiallyattractivetooltoaddressoverrelianceoncentralbankcreditan
issue that got great attention after the crisis. An OPBF allows assigning the prevention of over
proportionality toafocusedandeffectivetool.Of course,asortof OPBF couldalso beappliedbythe
bankingsupervisor,i.e.thebankingsupervisorwouldstarttoaskquestionsorapplysupervisorymeasures
when central bank reliance exceeds P#, etc. This raises again the question on the optimal overall
combinationofincentivesettingbytheOFandbybankingregulationandsupervisionsuchastoachieve
financialstabilityandanadequateextentofleverageandmaturityandliquiditytransformationbybanks.
Putting all the burden to regulation and supervision likely has high efficiency costs in view of the
decentralisednatureofinformationintheeconomy.
7. Conclusions:Designingandevaluatingoperationalframeworks
ThispaperaimedatreviewingtherealworlddesignandevaluationofOFs.Themainconclusionscanbe
summarizedasfollows.
1. ThereisasetofuniversalobjectivesofOFdesignvalidacrossenvironments.OFsshouldbeeffective,
lean, automated; they should support financial stability, financial market functioning, and financial
efficiency; the OF should be in some sense neutral with regard to relative financial prices; the OF
shouldbehonest,andideallyuniversal,i.e.portabletodifferentenvironmentsandacrosstime.
2. The search for universal objectives of OF design still encounters some limits because of divergent
philosophicalviewsonanumberofissues.Inparticular:(i)shouldcentralbanksbeseenprimarilyas
partoftheofficialsectororistheirdetachmentfromthegovernmentkeyandneedsreflectioninthe
avoidanceof(overproportional)exposurestofinancialinstrumentsissuedbythegovernment?;(ii)
should the central bank express in its collateral and asset allocation decision its idiosyncracies or
shoulditaimatbehavinglikeanaverageinvestor?Inaddition,viewsonOFdesign(asmaybemore
generally views on monetary policy) will always depend on the individuals inclination towards
ambitiousness,activismandacceptanceofcomplexity.Onetypeofcentralbankersandacademicswill
tendtostressthelimitsofcentralbanksabilitytodesignandoperatecomplexandambitiousOFsand
monetarypolicies,referringalsotothegeneralfailuresofthesophisticatedapproachesofthe1960s
and 1970s. The other type will reject such scepticism as fatalism and will argue in favour of being
ambitiousandoptimistonthehumanabilitytodesignamorecomplexframeworkforthegoodof
society.
3. The universality of optimal OF design finds some limits even under the assumption of universal
objectivesintheheterogeneityofenvironmentsunderwhichtheOFshavetooperate.Therefore,
one should not aim at absolute universality in OF design. Still, this does not put into question the
observationthatpastheterogeneityofOFsamongstadvancedlargecurrencyareaswasmostlydueto
historyandduetodifferentviewsanddoctrines,i.e.theexplanationsthatheterogeneityacrossthese
currency areas (or across time) were necessary reflections of differences (or changes) in the
environmenttendtobeunconvincing.
4. Thekeyoperationaltargetofmonetarypolicyshouldremainashorttermmoneymarketrate,without
excludingthatthiscouldbesomecompositerateorasecuredrate,orarateencompassingthemoney
marketbeyondthepureinterbankmarket.
5. Therearenosufficientreasonstodoubtonthemeritsofthesymmetriccorridorapproach,evenif
moneymarketswouldbesomewhatlessefficientinthefutureduetofinancialregulation;
45
6. OFsshouldonlyrelyonautomatedtenderproceduresforopenmarketcreditoperations,i.e.tender
procedures in which after the announcement of the conditions, the mapping from bids to tender
resultsisautomatic.
7. Ifthedemandforexcessreservesremainsunpredictableinthenewregulatoryenvironment,fixed
ratefullallotment operations may be the preferable automated tender procedure relative to
variableratefixedvolume. The latter is however the appropriate tender procedures for credit
operationsmaturingbeyondthenextmeetingofthepolicydecisionmakingbody.
8. Doubts can be raised on the efficiency of reserve requirements with averaging because of their
complexity and memory within the maintenance period. Other tools could be more elegant and
simpler(morefrequentopenmarketoperations,morenarrowcorridor,TARALAC).
9. Thecontroloftheovernightinterestratecanbeimprovedby(i)narrowingtheinterestratecorridor
setbystandingfacilities;(ii)introducing(andincreasingthequotaof)aTARALACfacility;(iii)increasing
thefrequencyofopenmarketoperations;(iv)conductingopenmarketoperationslaterintheday;(v)
betterautonomousfactorforecasting.Whilethefirsttwoalsohavethenegativesideeffecttoreduce
interbankmarketactivity,thelasttwoavoidthisandthereforeshouldinanywaybepursued.
10. Central banks should invest into autonomous factor forecasting to improve the control of the
overnight rate for a given level of money market turnover. Actually if good autonomous factor
forecastingandtimingofopenmarketoperationswithinthedayensuresthatbanksperceiveatthe
timetheytradeininterbankmarketsthatliquidityconditionsarebalanced,thenafulleffectiveness
ofinterestratecontrolcanbeachievedwithoutharmingmoneymarketturnover.Thisexplainsthe
remarkable overnight interest rate stability that a number of central banks could reach pre2007
withoutaveragingandreserverequirements.
11. ATARALACfacilitysystemappearsaspotentiallyattractivetool,butsuperiorityintermsofachieving
better combinations of money market activity and overnight rate stability still has to be better
established.
12. AleanCBbalancesheetremainsdesirable,i.e.thelengthofthebalancesheetofthecentralbank
shouldbeorientedtonotexceedsubstantiallythesumofthemonetarybaseandthecentralbanks
capitalandreserves,unlesspolicyneedsprovideveryconvincingargumentstodoso.
13. Thereareseveralgoodjustificationsforcentralbankshavinginnormaltimesanoutrightportfolioof
securitieswithsomedurationanddiversificationacrossissuers,evenif,asindicatedintheprevious
point,thesizeofthisportfolioshouldnormallynotlengthenthecentralbankbalancesheetandleave
enoughbreathingspaceforopenmarketcreditoperationstosteermoneymarketconditions.
14. Collateralframeworksshouldavoidprocyclicality,i.e.theyshouldbeconservativeinnormaltimes,
sothattheycanremainstableinstressedtimeswhenriskparameters(volatility,valuationuncertainty,
etc.) deteriorate. Collateral frameworks should protect the central bank at a very high confidence
level.Thishoweverdoesnotimplythatthecentralbankshouldonlyacceptthemostliquidassets,as
thecentralbankcanimposehaircuts(acceptedbyitscounterpartiesinviewofthecreditriskfree
natureofthecentralbank)andtakewhatevertimeisoptimalforassetliquidation.Itremainsopen
whetheritisworthaddingthecomplexityofmorethanonecollateralpool,suchastoallowprice
differentiationofcentralbankcreditacrosscollateralsets.
15. TheLOLRfunctionofthecentralbankplaysarolebothinanexanteandanexpostsense.Exante,
the LOLR contributes to the ability of the financial system to provide maturity and liquidity
transformationservicestosociety;expost,i.e.whenassetandmarketfundingliquiditydeteriorated
duetosomeexogenousshock,itcanpreventeconomicdamagethatresultsfromdefaultsandforced
assetliquidation.FromtheperspectiveofOFdesign,itisimportanttonotethattheexanteeffectsof
theLOLRwillbebasedonthreecomponents:theOFsbuiltinelasticityofindividualbanksrecourse
tothecentralbanktocentralbankcredit;perceptionsofbanksregardingthepropensityofthecentral
banktoadjusttheOFinasystemiccrisistostrengthenitsLOLRcontent(e.g.broadenthecollateral
set);perceptionsofbanksregardingthepropensityofthecentralbanktoprovideemergencyliquidity
46
16.
17.
18.
19.
20.
21.
assistance(ELA)tosinglebanks.Thefirsttwoelementsareasimportantasthelastone,andtherefore
theLOLRshouldbeconsideredakeydimensionofOFdesign.Asbanksmayalsohaveatendencyto
overleverageandtoengageinexcessiveliquidityandmaturitytransformation(andactuallythiswas
themainimpressionfromthe20072008liquiditycrisis),theOFrelatedelementsoftheLOLRshould
buildinincentivesthatcontributetoguidebankstotheappropriateamountofliquidityrisktaking.
Thisshouldinanycaseincludecounterincentivesagainstregulardisproportionalexanterelianceon
thecentralbank.Obviously,alsoregulationshouldplayarole,andinanidealworldtheLOLRaspects
oftheOFandliquidityregulationwouldbedesignedinsomeconsistentway.
IfonesimplifiestheLOLRspecificationproblemforamomentinawaythatitwouldconsistinchoosing
onedimensionallytherightLOLRintensityinanOF,andcapturethechoicebetweentwoextreme
frameworks (with minimum and maximum LOLR content) in a unit interval [0,1], then the welfare
maximizing LOLR intensity is likely to be an interior point, i.e. none of the extreme frameworks is
optimal.Itisimportanttonotethatthewelfaremaximizingpointintheunitintervalwillnormallynot
necessarilybetheonewhichmaximizesfinancialstability,ormarketdiscipline.Possiblycentralbank
risktakingisnegligibleintheoptimumLOLRspecification,butnotnecessarily.
Better understanding the LOLR should allow strengthening its rulebased elements, which means
buildingittoalargerextentintotheOF,andreducingthecontributionoftheunpredictableadditional
elements that come in only in concrete systemic or idiosyncratic stress situations. The
constructivenessofambiguityalwayshaditslimits,andevenmoreintodaysworldofhighdemand
towardsthetransparencyofofficialsectordecisionmaking,theneedforequaltreatmentacrosscases
(alsotopreventlegalactionagainstthecentralbank)andagenerallycriticalattitudeofthepublic
towardscentralbanking.
Anoverproportionalborrowingframework(OPBF)isonefocusedwaytocontroloverrelianceonthe
LOLRandpreventconcentrationinexposuresfromcentralbankcreditoperationsinafocusedand
rulebasedway.
Innormaltimes,thestanceofmonetarypolicyshouldbesteeredthroughshortterminterestrates,
andnotbyvariationofothercentralbankinstrumentsthatdetermineoverallmonetaryconditions
andfinancingcostsoftherealeconomy,suchastheLOLRcontentoftheOF,orthecredit,liquidity
anddurationriskthatthecentralbankholdsintheformofoutrightportfolios.
Argumentsfromamonetarypolicyperspectiveforastructurallevelshiftoftheshortterminterest
ratethroughaconstantdoseofaccommodationinjectedbythesetwootherinstrumentsarealsonot
obvious.
Theexpostassessment(individually,orincomparativetermsforseveralcentralbanks)ofactualOFs
canbesupportedbyscorecards,asoutlinedinannex1.Atleast,thesescorecardscandirectquickly
totherelevantquestionsforevaluatingexistingOFsandidentifyingopportunitiesforimprovement.
Beyondthese21conclusions,stillwithinthesearchforuniversalqualitiesofOFs,onemayalsolookfora
moreparsimoniousoverallOFdesign,i.e.asortofreengineeringfromscratch,eliminatingcomplexities
thatmaybeleftoversfromhistory.Ideally,inareengineeredleanOF,eachinstrumentwouldpursue
clearlyoneobjectiveandwouldnotstronglyinterferewithotherinstrumentsandobjectives.Thefollowing
figuretriestocapturethecurrentusualmappingofinstrumentstoobjectives(wherebyonlyasubsetof
toolsandobjectivesiscovered).
47
Figure10:TraditionalmappingofinstrumentsintoOFobjectives.
Tools:
Reserverequirements
Creditoperations(through
variableratetenders)
Standingfacilities
Collateral
Objectives
Interestratecontrol
Incentives,preservemarketnetting
Centralbankprotection
InsuchanOF,therelationsbetweeninstrumentsandobjectivesarenotbijectiveinmostcases:Reserve
requirementsinadvancedeconomiesnowareoneamongstseveraltoolstoensureaneffectivecontrolof
overnightrates,whilepreservingsomemoneymarketactivity.Theconductofcreditoperationsthrough
variableratetendersaimsbothatachievingtheoperationaltargetshortterminterestrate,andbean
incentivemechanismthatinvitesbankstobecompetitiveandnottooverrelyonthecentralbank.Also
standingfacilitiesfacethemultipleobjectivesofsteeringinterestrates,whilepreservingincentivesfor
tradingfirstinmarkets.Collateralpoliciesarenotonlyconsideredatoolforfinancialriskprotectionofthe
centralbank,butalsoasonetolimittheabilityofbankstorelyoverproportionallyoncentralbankcredit.
Aradicallycleanedupmappingofinstrumentstoobjectivecouldbethefollowingone.
Figure11:ReorganizedmappingofinstrumentsintoOFobjectives.
Tools:
Objectives
Zerowidthcorridorfacility
Interestratecontrol
Overproportionalrateaddons
Incentives,preservemarketnetting
CBprotection
Collateral
Interestratecontrolwouldbeachievedeffectivelybyazerowidthstandingfacilitiescorridor.Incentives
to not rely disproportionally on the central bank would be based on an overproportional borrower
framework which could also encompass penalties for banks who deposit too large amounts of excess
reservesatthecentralbank.Collateralwouldaimonlyatcentralbankriskprotectionandsufficiency.The
propertiesofsuchabijectiveframeworkwouldneedtobestudiedfurther.Forthetimebeing,theidea
canserveasabenchmarkforamoremoderatestreamliningofOFs.
DespiteprogressinourunderstandingofOFs,alsosometrialanderrorwillprobablyremainnecessaryto
findoutwhatOFwillworkbestforthecentralbanksoflargemonetaryareasaftertheeventualliftoff
fromthezerolowerboundandthenormalizationofbalancesheets,i.e.inafewyearsfromnow.OFdesign
willhopefullybeapproachedevenmoreconsciouslybycentralbanksthaninthepast,reflectingwellthe
importanceoftheOFindeterminingthestartingpointofthetransmissionmechanismandinshapingthe
financialsystem.
48
Listofreferences
Acharya,V.V.andS.Wiswanathan(2011),Leverage,Moralhazard,andliquidity,JournalofFinance,66,99138.
Ashcraft,A.,N.Grleanu,andL.H.Pedersen(2011),TwoMonetaryTools:InterestRatesandHaircuts,NBERMacroeconomics
Annual,2010,Chapter3,Volume25,editedbyD.AcemogluandM.Woodford,143180.
Ayuso,J.andRepullo,R.(2003),AmodeloftheopenmarketoperationsoftheECB,EconomicJournal,113,889902.
Ayuso,J.,A.G.Haldane,Restoy,F.(1997),VolatilityTransmissionalongtheMoneyMarketYieldCurve,Weltwirtschaftliches
Archiv,133,5675.
Bagehot,W.(1873),LombardStreet:Adescriptionofthemoneymarket,London:H.S.King.
BankofEngland(2011):TheFrameworkfortheBankofEnglandsOperationsintheSterlingMoneyMarkets,December2011.
BankofEngland(2013),LiquidityinsuranceattheBankofEngland:developmentsintheSterlingmonetaryframework.
BankofEngland(2015),OneBankResearchAgenda,DiscussionPaper,February2015.
Bech,M.andT.Keister(2015),Liquidityregulationandtheimplementationofmonetarypolicy,WorkingPaper.
Bindseil,U.(2004a),Monetarypolicyimplementation,OxfordUniversityPress.
Bindseil,U.(2004b),Theoperationaltargetofmonetarypolicyandtheriseandfallofreservepositiondoctrine,ECBWP372.
BindseilU.andF.Papadia(2006),Creditriskmitigationincentralbankoperationsanditseffectsonfinancialmarkets:Thecase
oftheEurosystem,ECBOccasionalPaperSeries,No.49.
Bindseil,U.andNyborg,K.(2008),Monetarypolicyimplementation,inX.Freixas,P.Hartmann,C.Mayer(eds.):Handbookof
European Financial Markets and Institutions, p 742778, Oxford University Press, New York. (Also available as Working Paper
versionunderssrn:Discussionpaper10/2007,March2007,DiscussionPaperSeriesoftheNorwegianSchoolofEconomicsand
BusinessAdministration)
Bindseil,U.andF.Wrtz(2008),Efficientanduniversalframeworks(EUF)formonetarypolicyimplementation,WorkingPaper,
August2008(paperpresentedattheAusschussfrGeldtheorieundGeldpolitikdesVereinsfrSocialpolitik,FrankfurtamMain,
20February2009,http://www.wiwi.unifrankfurt.de/professoren/wieland/vfs/papers2009/Bindseil_Wuertz.pdf).
Bindseil,U.,KjellNyborgandI.A.Strebulaev(2009),RepoAuctionsandtheMarketforLiquidity.JournalofMoney,Creditand
Banking,41:13911421.
Bindseil,U.(2013),Centralbankcollateral,assetfiresales,regulation,andliquidity,ECBWPNo.1610.
Bindseil,U.(2014),Monetarypolicyoperationsandthefinancialsystem,OxfordUniversityPress.
Blanchard,Olivier,GiovanniDellAriccia,PaoloMauro(2014),RethinkingMonetaryPolicy,IMFStaffPositionNote,SBP/10/03.
Blasques,F.,F. Bruning,andI.vanLelyveld(2015),Adynamicnetworkmodeloftheunsecuredinterbanklendingmarket,
FederealReserveBankofBostonWorkingPaperSeries,No.1603.
BoardofGovernors(1994),TheFederalReserveSystem:PurposesandFunctions,Washington,DC:BoardofGovernorsofthe
FederalReserveSystem.
Borio,C.E.V.(1997),Monetarypolicyoperatingproceduresinindustrialcountries,BISconferencepapers,Vol.3,286368.
Borio,C.E.V.(2001).Ahundredwaystoskinacat:comparingmonetarypolicyoperatingproceduresintheUnitedStates,Japan
andtheeuroarea,BISPapersNo.9.
Brunnermeier,M.K.andYannKobi(2016),ThereversalrateEffectivelowerboundonmonetarypolicy,Slidespresentedin
aseminarattheECBon16March2016.
Buiter,W.andSiebert,A(2005).HowtheEurosystemsTreatmentofCollateralinitsOpenMarketOperationsweakensfiscal
disciplineintheEuroZone(andwhattodoaboutit),CEPRdiscussionpaperNo.5387.
Buiter,W.H.2009.NegativeNominalInterestRates:ThreeWaystoOvercometheZeroLowerBound,NBERWP15118.
Carlson,MarkA.andDuyganBump,BurcuandNelson,WilliamR.(2015),WhyDoWeNeedBothLiquidityRegulationsanda
LenderofLastResort?APerspectivefromFederalReserveLendingDuringthe200709USFinancialCrisis,BISWPNo.493.
Cassola,Nuno,andF.Koulischer(2015),Thecollateralchannelofopenmarketoperations,workingpaper.
CGFS(2015),Regulatorychangeandmonetarypolicy,CGFSPaperNo.54.May2015.
CGFS(2016),Fixedincomemarketliquidity,CGFSPaperNo.55.January2016.
Chapman,J.T.E,J.Chiu,andM.Molico(2011),Centralbankhaircutpolicy,AnnalsofFinance,7,319348.
Cheun,S.,I.vonKppenMertes,andB.Weller(2009),ThecollateralframeworksoftheEurosystem,theFederalReserveand
theBankofEngland,andthefinancialmarketturmoil,ECBOccasionalPaper,No.107.
49
Coeure,B.(2013),Wheretoexitto?Monetarypolicyimplementationafterthecrisis,Speechatthe15thGenevaConference
ontheWorldEconomy:Exitstrategies:timetothinkaboutthem,Geneva,3May2013.
Curdia,V.andM.Woodford(2010),Thecentralbankbalancesheetasaninstrumentofmonetarypolicy,Paperpreparedfor
theCarnegieRochesterConferenceonPublicPolicy,TheFutureofCentralBanking,April1617,2010
DanmarksNationalbank(2009),MonetarypolicyinDenmark,2009.
Danthine,JeanPierre(2012),Reconcilingpriceandfinancialstability,SpeechgivenatZrichUniversity,24January2012.
EuropeanCentralBank(2005).TheImplementationofMonetaryPolicyintheEuroarea:GeneralDocumentationonEurosystem
MonetaryPolicyInstrumentsandProcedures,FrankfurtamMain.
ECB(2013),TheEurosystemcollateralframeworkthroughoutthecrisis,ECBmonthlybulletin,July2013.
ECB(2015),ThefinancialriskmanagementoftheEurosystemsmonetarypolicyoperations,FrankfurtamMain.
EMMI(2015),ConsultativepositionpaperontheevolutionofEuribor,EuropeanMoneyMarketsInstitute,30October2015.
Ennis, H. M. and Keister, T. (2008), Understanding monetary policy implementation, Federal Reserve Bank of Richmond
EconomicQuarterly,Vol.94,3,253263.
EuropeanCentralBank(2011),ThemonetarypolicyoftheECB,FrankfurtamMain.
EuropeanMonetaryInstitute(1997),ThesinglemonetarypolicyinStageThreethespecificationoftheoperationalframework,
FrankfurtamMain.
FederalReserveBankofNewYork(2016),Domesticopenmarketoperationsduring2015,AreportpreparedfortheFederalOpen
MarketCommitteebytheMarketsGroupoftheFederalReserveBankofNewYork,April2016.
Fischer,Stanley(1988),Rulesvs.discretioninmonetarypolicy,NBERWorkingPaper,No.2518.
Freixas,X.,A.MartinandD.Skeie(2009),Bankliquidity,interbankmarketsandmonetarypolicy,FederalReserveBankofNew
YorkStaffReport,No.371.
Friedman,M.(1982),Monetarypolicy:Theoryandpractice,JournalofMoney,CreditandBanking,14,98118.
Friedman,BenjaminMandK.N.Kuttner(2010),Implementationofmonetarypolicy:howdocentralbankssetinterestrates,
NBERWorkingPaper16165,July2010(AlsopublishedasachapterintheHandbookofMonetaryEconomics.
Frost,J.,L.Logan,A.Martin,P.E.McCabe,F.M.Natalucci,J.Remache(2015),OvernightRRPOperationsasaMonetaryPolicy
Tool:SomeDesignConsiderations,BoardofGovernorsoftheFederalReserveSystem(U.S.),FinanceandEconomicsDiscussion
Series201510.
Garratt,R.,A.Martin,JMcAndrews,E.Nosal(2015),Segregatedbalanceaccounts,FederalReserveBankofNewYorkStaff
Reports,No.730.
Goodfriend,M.(2000),Overcomingthezerolowerboundoninterestratepolicy,JournalofMoney,CreditandBanking,32,
10071035.
Goodfriend,M.andM.Hargraves(1987),Ahistoricalassessmentoftherationalesandfunctionsofreserverequirements,in:
M.Goodfriend,Monetarypolicyinpractice.Richmond:FederalReserveBankofRichmond.
Haldane,AndreG.(2015),Howlowcanyougo?,Speechgivenon18September2015atPortadownChamberofCommerce.
Hirshleifer, J. (1971), The private and social value of information and the reward to inventive activity, American Economic
Review,61,561574.
Ho,Corinne(2008),Implementingmonetarypolicyinthe2000s:operatingproceduresinAsiaandbeyond,BISWPNo.253.
International Monetary Fund (2004). Monetary Policy Implementation at Different Stages of Market Development, paper
preparedbythestaffoftheMonetaryandFinancialSystemsDepartment,26thOctober2004.
Keister,T.(2016),BailoutsandFinancialFragility,ReviewofEconomicStudies,Vol.83(2016)460480.
Kneeshaw,J.Z.,andP.vandenBergh(1989),Changesincentralbankmoneymarketoperationproceduresinthe1980s,BIS
EconomicPapersno.23.
Knig,P.(2015),Liquidityrequirements:Adoubleedgedsword,InternationalJournalofCentralBanking,Dec.2015,129168.
MarketsCommittee(2013),Centralbankcollateralframeworksandpractices,ReportbyaStudyGroupchairedbyGuyDebelle
andestablishedbytheBISMarketsCommittee.
Meigs,J.A.(1962),Freereservesandthemoneysupply,Chicago:UniversityofChicagoPress.
Meulendyke,A.M.(1998),USMonetaryPolicyandFinancialMarkets,NewYork:FederalReserveBankofNewYork.
Mishkin,F.(2004),Theeconomicsofmoney,bankingandfinancialmarkets,7thed.,Reading,MA:AddisonWesley.
Nautz,D.andOechsler,J.(2003),TherepoauctionsoftheEuropeanCentralBankandthevanishingquotapuzzle,Scandinavian
JournalofEconomics,105,20720.
50
Nield, I. (2008), Evolution of the reserve banks liquidity facilities, Reserve Bank of New Zealand: Bulletin, Vol. 71, No. 4,
December2008.
NorgesBank(2014),BanksassessmentofNorgesBanksliquiditymanagementsystem,No.4,2014.
NorgesBank(2016),BackgroundforthesystemformanagingbankreservesinNorway,Websiteinformation.
Nyborg,K.(2015),Centralbankcollateralframeworks,SwissFinanceInstitute,ResearchPaperSeriesNo.1510,February2015.
PerezQuiros,G.andH.Rodriguez(2006),ThedailymarketforfundsinEuropehassomethingchangedwiththeEMU?,Journal
ofMoneyCreditandBanking,38,91118.
Potter,S.(2015),Moneymarketsandmonetarypolicynormalization,April15,2015,remarksattheMoneyMarketeersofNew
YorkUniversity,NewYorkCity.
Potter,S.(2016),Implementingmonetarypolicypostcrisis:Whathavewelearned?Whatdoweneedtoknow?,speechat
conferenceorganisedbyColumbiaUniversitySIPAandtheFederalReserveBankofNewYork,May4,2016.
PwC(2015),Globalfinancialmarketsliquiditystudy,August2015.
Rogoff,K.(2014),Costandbenefitstophasingoutpapercurrency,WorkingpaperpresentedatNBERMacroeconomicsAnnual
Conference,April11,2014.
Selin,Peter,andPerAsbergSommar(2014),TheRiksbanksoperationalframeworkfortheimplementationofmonetarypolicy
areview,SverigesRiksbank.
Simmel,G.(2004/1900),Thephilosophyofmoney,editedbyD.Frisby,Thirdenlargededition,Routledge,LondonandNewYork.
Taylor,JohnB(1999),Ahistoricalanalysisofmonetarypolicyrules,inJ.Taylor(ed.),Monetarypolicyrules,319348,University
ofChicagoPress.
Toniolo,G.(2005),CentralbankcooperationandtheBIS,19301973,CambridgeUniversityPress,Cambridge.
Trebbi,FandK.Xiao(2015),Regulationandmarketliquidity,NBERWorkingPaperSeries,No.21739.
Vlimki,Tuomas(2003),Centralbanktenders:threeessaysonmoneymarketliquidityauctions,BankofFinlandStudies.
Woodford,M.(2003),Interestandprices.Foundationsofatheoryofmonetarypolicy,Princeton:
PrincetonUniversityPress.
Woodford,M.(2016),QuantitativeeasingandfinancialstabilityCEPRdiscussionpaper11287.
51
Annex1:Scorecards
Inthisannex,theconclusionsfromeachofthesections4to6aretranslatedintoscorecardswhichare
constructedtobeusedforstandalone,orcomparativeassessmentsofexistingOFs.Thisisanattemptto
translatetheconclusionsdrawninthesesectionsintosomethingevenmoredirectlyapplicableinpractice
andmaybetoverifythatthefindingsindeedcanbetranslatedintorealworldguidance.Thescorecards
consistoflistsofstatements,which,iffullyaffirmed,giveahighscore,andiffullyrejected,alowone.The
scorecardsactuallycontainfourdifferenttypesofstatements:
Statements on a quantitative success measure. The measures could in principle be mapped in a

predetermined way into a score. For example it could be predetermined that the statement The
controloftheovernightinterestrateisprecisegetsascoreof5ifthedailyvolatilityisbelow3basis
points,of4ifitisbetween4and5basispoints,etc.
QualitativestatementsontheefficiencyandeffectivenessoftheOFdesignsuchasMonetarypolicy
implementationisautomatedinbetweenmonetarypolicycommitteemeetings.Onecoulddevelop
guidanceonhowtodecideonthescoresforsuchstatements.
Statementsaboutwhetherthecentralbankhaspursuedwelldefinedobjectivesandhasapplied
sound analysis in designing and parametrizing its OF. For example, it can be asked whether the
followingstatementistrueThelenderoflastresortfunctionbuiltintotheOFhasbeendesignedin
awaytocontributetofinancialstabilitywhilepreventingmoralhazardandthedesignwasbasedon
stateoftheartanalysis.Answeringthisquestioncouldbeeitherdoneonthebasisofareviewof
publicationsofthecentralbank,oronthebasisalsoofinternaldocumentation,dependingonwhether
onehasaccesstothelatter.
Statementsonthesoundnessofthepracticalimplementationofaframework,coveringissuessuch
as the quality of documentation, transparency, IT systems, understanding of decision makers,
existenceofcontingencyplans,operationalrisks,etc.Examplesofsuchstatementsare:Theeconomic
logic of the monetary policy implementation framework is truthfully communicated and well
understoodbymarketsandthepublic;Internalprocessesrelatingtocollateralassessmentarewell
documented;ITsystemsareresilientandcontingencyproceduresareinplace;Thecentralbank
iswellpreparedintermsofresourcesandITsystemstolaunchnonstandardmeasureswhenneeded.
These questions are actually rather general and independent of the specification of the OF, and
thereforearekeptataminimuminthescorecardsoutlinedinthispaper.
Thetotalnumberofscorecardquestionswaskeptlimitedbutcaneasilybeexpanded.Also,tonotoverdo
it,noannotativeguidancewasaddedonhowtomapanswersintoscores.Still,scorecardswereincluded
toexploretheideatocondenseconclusionsontherightdesignofOFsintoapracticalassessmenttool.
Controloftheoperationaltarget(section4)
Assumingthatthesymmetriccorridorapproachistoprevailforthereasonsindicated,thecentralbank
hastostillchoosethefollowingsetsofparametersinparallel:(i)Widthofthecorridorsetbythetwo
standingfacilitiesatpenaltyrates;(ii)sizeofTARALACquota(if any), or,alternatively,sizeofreserve
requirementsandthelengthoftheaveragingperiod(again,ifatall);(iii)frequencyandintradaytiming
of open market operations, and whether the latter are conducted as fixedratefullallotment or as
variableratefixedvolume.Thecriteriathatshouldguidethechoicescanbederivedfromtheprinciples
setinsection3.Inpractice,thefocusshouldbeinparticularoneffectivenessofthecontroloftheovernight
rate, preserving money market turnover (and prevent absolute central bank intermediation and a
lengtheningofthecentralbankbalancesheet)andoverallsimplicity.Howthecombinationsofwidthof
corridor, TARALAC quota size (or reserve requirement) and timing and frequency of open market
operationsinfluencetheovernightratevolatilityandmoneymarketactivitywassimulatedfromanumber
ofanglesabove,butcouldbedonemoresystematicallyinfutureresearch.Inthisway,anefficientfrontier
52
ofturnovervolatilitycombinationscanbederived,whichcanthenbematchedwiththepreferencesof
thecentralbankonthesetwoparameters,aswellaswiththesimplicitycriterion.Onecouldalsotryto
calibratethesesimulationswiththeactualvolatilitypatternsofautonomousfactors,interbanktransaction
cost,etc.observedinamonetaryareasuchastobeabletoconsiderthesimulationsasdirectlyrelevant
foraspecificcase.ThefollowingscorecardcouldbeusedtoevaluateOFswithregardtotheirapproachto
controltheoperationaltargetofmonetarypolicy.
ScorecardI:Controlofoperationaltarget(overnightrate)andconductoftenders
Score
(15)
1.
2.
Thevolatilityofovernightinterestratesislow
Thevolatilityofovernightinterestrates(intheintermeetingperiod)isnottransmittedalongthe
yieldcurve,i.e.innovationsinovernightratesnormallyarenotcorrelatedwithinnovationsin
mediumandlongertermmoneymarketrates
3. Thespreadofovernightraterelativetotargetratetendstobesmallandstable
4. Forwardovernightratesintheperioduntilthenextmeetingofthepolicydecisionmakingbodyare
veryclosetotheleveloftheinterestratetarget
5. Incaseofasymmetriccorridorapproach,recoursetostandingfacilitiesisusuallyonesidedexpost,
butexantesymmetric.
6. Stabilizationofovernightratesreliesoninstrumentsthataresimple,transparentandautomated
7. Monetarypolicyimplementationisnormallyautomaticinbetweenmeetingsofdecisionmaking
bodies(i.e.nocentralbankdecisionmakingneeded,oratleastnotmorethane.g.mechanically
mappingaforecastofautonomousfactorsintoavariableratefixedvolumetendervolume)
8. Controlofovernightratesisachievedwithoutsignificantimpairmentofmoneymarketactivity
9. Theapproachtothechoice,andthetechniqueofcontroloftheoperationaltargetisdocumented
andexplainedtothepublicinahonestmanner
10. Neithertenderannouncementsnorallotmentdecisionsareusedtosignalthestanceofmonetary
policy
11. Whencalibratingtheirbidsintenderoperations,banksdonotneedtothinkaboutbidding
behaviorofotherbanks
12. Afterthetenderannouncement,thecentralbankdoesnotneedtotakeanyfurtherdecision
(exceptasformality),andbiddingbehaviorismappedautomaticallyintoallotment
13. ITinfrastructureandprocessesfortender procedures areefficientandreliable
14. Creditoperationswithmaturitybeyondthenextmeetingofthepolicydecisionmakingbodyare
conductedasvariableratetenderwithpreannouncedvolume
15. Tenderratesandrelevantmarketratesarecloselylinkedexante
16. Theannouncementoftenderresultshasnormallynoimpactonshorttermmarketrates
17. Thenumberofoutstandingtendersatanypointintimetendstobeparsimonious
18. Thereisnoindeterminacyonsharingofliquiditydeficitamongstoutstandingcreditoperations
19. Thecounterpartysetisbroadenoughtoensure(i)competitiveintermediationbetweenthecentral
bankandtherestofthefinancialsystemand(ii)thatbiddinginvariableratetenders(ifapplicable)
isnotoverlynoisy.
20. Thecounterpartysetisnottoobroadinthesenseofunderminingtheroleofuniversalbanks(with
broadaccesstokeyfinancialmarketsincludingthemoneymarketandtoborrowers)inmanaging
andnettingfundingneedsvisvisthecentralbank.Thecentralbankdoesnotprovideabsolute
intermediationinnormaltimes.
21. Thecounterpartyframeworkdiscriminatesbetweensuitableandunsuitablemonetarypolicy
counterparties,theformerbeingthosefinancialinstitutionswithbroadaccesstokeyfinancial
marketsincludingthemoneymarketandtoborrowers.
22. Accessofnonbankstodepositingwiththecentralbankdoesnotcreaterisksofbankdis
intermediationincaseofageneralnegativeperceptionofthebankingsystem.
Average
53
Balancesheetstructureandoutrightportfolio(section5)
Section5concludedthatwhilecentralbankbalancesheetsshouldreturntowardsaleanshapeinwhich
theirlengthisnotinflatedbylargescalesecuritiesholdings,theremaybegoodreasonstoholdamedium
sized, more diversified outright portfolio. Such a portfolio contributes to central bank income without
necessarilycreatingsubstantialrisks(thankstodiversificationeffects)andstrengthensexpertisewithin
thecentralbankonbondmarkets,whichcanbeessentialinafinancialcrisis.Theremainingliquiditydeficit
ofthebankingsystemvisvisthecentralbankshouldbeofsufficientsizetoallowthecentralbankto
steershortterminterestrateswithcreditoperations.Thefollowingscorecardaimsattranslatingsomeof
theconclusionsofthissectionintopracticalorientationsforassessingexistingOFs.
Score
ScorecardII:Balancesheetstructureandoutrightportfolio
(15)
1.
Thebalancesheetislean,i.e.itisnotmuchlongerthanwhatisimpliedbythemonetarybaseand
centralbankcapitalandreserves
2. Outrightportfoliosdoleaveenoughroomtocontrolshortterminterestrateseffectivelyatthe
marginthroughcentralbankcreditoperations
1. Thecentralbankhascontroloftheaverageduration,credit,liquidityanddurationrisksinits
balancesheet
3. NPCassetsareinvestedinawaytoprovideanefficientriskreturncombination
4. NPCassetsareinvestedinawaytomaintainexpertiseinthecentralbankthatisusefulasit
ensuresthatcentralbankstaffunderstandsfinancialmarketfunctioning,whichmaysupportthe
abilitytolaunchquickly,whenneeded,certainnonconventionalmeasures
5. ThecentralbankistransparentonitsNPCassetcompositionandexplainsittothepublic
6. ThecentralbanksNPCassetcompositiondoesnotdistortfinancialmarketprices(i.e.isneutral,
accordingtothedefinitionofneutralitychosenbythecentralbank).Thecentralbankisconscious
ofitsneutralitydefinitionandmakesitknowntothepublic
7. NPCassetsdonotinterferewithmonetarypolicy
Average
Collateralframework(subsection6.2)
Collateral frameworks should protect central banks against credit risk, should ensure that collateral
scarcitydoesnotinterferewithmonetarypolicy,andshouldavoiddistortingmarkets.
Score
ScorecardIII:Collateralframework
(15)
1.
2.
3.
4.
5.
6.
7.
8.
Thecollateralframeworkensuresataveryhighconfidencelevelthatthecentralbankwillnot
sufferlossesincaseofacounterpartydefault
Thecollateralframeworkensuresthatcollateralscarcitydoesnotinterferewiththesmooth
implementationofmonetarypolicy
Thecollateralframeworkdoesnotunderminetheefficiencyandactivityinthemoneyandcapital
market,e.g.byprovidingincentivestobankstoretainissuedsecuritiesinsteadofplacingthem
withanexteriorinvestorbase)
Thecollateralframeworkavoidsunjustifieddiscriminationacrossissuertypes
Haircutcalibrationissystematicandachievesposthaircutriskequivalenceintermsoftherisk
measurethatisdeemedmostrelevant.
Thecollateralframeworkdoesnotleadtoanundesiredconcentrationincollateraluserelativeto
theeligibleset.
Incasetheuseofsometypesofcollateralisparticularlycostlyforthecentralbank,thenthese
costsareinternalisedbychargingfeestobanks.
Thecollateralframeworkistechnicallyefficient,i.e.eligibilitychecks,use,monitoring,etc.are
automatedandwellorganised.
54
9.
Thecollateralframeworkwastestedforalleligibleassetclassesinthecontextofcounterparty
defaultsinthesensethattherelevantassetsweresuccessfullyliquidatedandpriorassumptions
onliquidationwerevalidated.
10. Thecollateralframeworkistransparentandwellexplainedtowardsthepublic
11. Thecollateralframeworkisfelttoleadtoanadequatedegreeofelasticityofregularcentralbank
credittoindividualbanks(adequatecontributiontotheCBsLOLRfunction)
Averagescore
LOLRframework(subsection6)
TheLOLRframeworkaimsexanteatstrengtheningtheabilityofthebankingsystemtoprovideliquidity
and maturity transformation services to society, and ex post to preserve financial stability and avoid
unnecessary economic damage due to default and other forms of forced liquidation. It has to address
moralhazardandtheendogeneityofriskinthenegativeandpositivesense.Notonlyatotalextentofthe
LOLRfunctionshouldbechosen,butalsotheoptimalcontributionstoitfromthenormalOF,perceptions
ofthereadinessofthecentralbanktochangetheOFincrisistimetowardsmoreLOLRcontent,andfinally
perceptionsofthecentralbankpropensitytoprovideELA.
ScorecardIV:LOLRframework
Score
(15)
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
TheoverallextentofbuiltinLOLRelasticitywithinthenormalOFprovidesadequatebreathing
spacetobankfundingincasesoftemporarytensionsinbankfundingmarketsortemporary
depositoutflows
TheoverallextentofbuiltinLOLRelasticitywithinthenormalOF doesnotleadtoundue
concentrationofcreditoperationswithfewcounterpartieswithweakermarketaccess.
Instead,mechanismsensurethatcreditoperationsarespreadamongstbankswhichhavea
widerangeofmarketaccessandbankingbusiness
AccessconditionstoCBcreditwithintheregularOFisnotprocyclical
ThecentralbankiswellprotectedinLOLRagainstfinancialrisks(throughcollateralpolicies
and/orguarantees),soastokeepminimaltheprobabilityofcentralbanklosses
Thereareadequatesafeguardsagainstmoralhazardofbankswithregardtotheiraccessto
theLOLRfunctionofthecentralbank
TheLOLRframeworkisbalancedtowardsthetradeoffbetweenzombification(survivalof
unproductivefirms)andcostsofundesirabledefaultsofviablefirms
TheLOLRframeworkcanbeadjustedinacrisistoprovideadditionalelasticityreflectingthe
higherexogenousliquidityuncertaintyinsuchacontext
RecoursetotheLOLRisnotexcessivelystigmatised
TheassignmentoftheLOLRfunctiontoregularvs.ELAcreditoperationsbalancestheprosand
consofpredictabilityversusconstructiveambiguity
TheELAframeworkfollowsdefinedproceduresandisdocumentedinaninternalmanual
ELAprovisionissufficientlyrulebasedsuchastopreventsuccessfullawsuitsagainsttheCB
becauseofanallegedlackofequaltreatment
ThedegreeoftransparencyonELApoliciesandproceduresisadequatetobalanceclarityand
constructiveambiguity
Therolesofotherstakeholders(e.g.thegovernment,thedepositinsurancescheme,the
resolutionmechanism,etc.)arewelldefinedintheinternalELAmanualandthereismutual
understandingbetweentheseofficialsectorentitiesontherespectiverolesandpolicies.
Average
55
SomegeneralobjectivesofOFdesign(concludingsection)
ThefollowingfinalscorecardtakesupagainsomemorefundamentalobjectivesofOPFdesign.
ScorecardV:FocusandstructureofOFandachievementoffundamentalobjectives
1
2
3
4
5
6
Score
(15)
TheOFisdesignedinawaythateachinstrumenttendstobewellfocusedonobjectives.
TheOFreliesoverallonfeweffectiveinstrumentstoachieveitsobjectives
Theobjectivesofthedifferentinstrumentsareclearandhowtheyinteracttowardsachievingall
mainobjectivesoftheOFiswellanalyzedanddocumentedcentralbankinternally.
Theobjectivesofdifferentinstrumentsandhowtheyinteracttowardsachievingallmainobjectives
oftheOFiswelldocumentedandexplainedtotheoutsideworld,andisgenerallywellunderstood.
TheOFispresentedinanhonestwaytothepublic
Theframeworkisuniversal,i.e.itappearsportabletodifferentenvironments
Annex2:Simulatingthecontrolofinterestrateswhilepreservingmoneymarket
activity
Thesimulationtoolisbasedonasimpletwobanks,onedaymodel.Thetwobanksaresubjecttoliquidityshocks
beforeandafterthetradingsession,wherebythereareshocksofaggregateandofrelativenature.Variantsofthe
toolhadbeenusedinBindseilandWrtz(2008)tosimulateaTARALACfacilityandinBindseilandJablecki(2013)to
simulatethewidthofthecorridor(seealsoBindseil,2014,section6.2).Thevariantpresentedhereallowstocapture
theimpactof(a)thewidthofthestandingfacilitiescorridor,(b)theexistenceandquotaofaTARALACfacility,(c)
thefrequency,and(d)tenderprocedureofopenmarketoperationson:(i)overnightratevolatility,(ii)interbank
tradingvolume,and(iii)centralbankbalancesheetlength/averagerecoursetothefacilities.Thetimelineofevents
everydayisasfollows:
i.
Centralbankopenmarketoperation.Inthemorning,thecentralbankadjustsitscreditprovisionbymeans
ofanopenmarketoperation(OMO)inawaytomatchexpectedautonomousfactors(Bforbanknotes),
i.e.OMO=B.TheactualautonomousfactorsatdayendaredefinedasB+d1+d2.Bisthedeterministic
componentandlevelofautonomousfactorsinthemorning,whiled1,d2arestochasticshockshittingeach
bankinthecourseoftheday,withE(d1)=E(d2)=0andwithasymmetricdensityfunction.SinceOMO=B
in the morning, the total bank reserves R will be equal to zero in the morning. There are no reserve
requirements.
ii.
Morning aggregate and relative liquidity shocks. After the central bank operation, a first stochastic
componentofautonomousfactorsoccursandbecomespubliclyknown:d1.Atthesametime,adepositshift
shock takes place, k, which is neutral in terms of aggregate liquidity, but reflects that e.g. deposits of
householdsandcorporatesmovefromonebanktotheother.Forthesimulation,weassumethatthetwo
shocksareN(0,d1),N(0,k),respectively.
iii.
Interbanktradingsession.Atmidday,atradingsessiontakesplace,inwhichtheinterbankrateissetinthe
middleofthevaluationsattributedbythetwobankstocentralbankmoneyundertheassumptionthatno
furtherinterbankmarkettradingcouldtakeplace(i.e.thepriceissuchthattherentfromtradingisshared
equally between the two banks). These valuations under a notrade assumption are for each bank the
weightedaveragesofthethreestandingfacilityrates,theweightsbeingtheperceivedprobabilitiesofthe
banktobeatthemargininrecoursetotherespectivefacility.Theinterbanktradeyistheamountlentfrom
bank1tobank2(i.e.anegativevalueofyindicatesthatbank2lendstobank1).Weassumethatinterbank
tradingissubjecttotransactioncosts,implyingthatifthegainsfromtradingaretoosmall(becausethepre
tradevaluationsofthetwobanksaretoosimilar),thenthereisnotrading.
iv.
Afternoonaggregateliquidityshock.Intheafternoon,theeventualenddaylevelofautonomousfactorsis
revealed, as the second stochastic autonomous factor component d2 occurs, which is assumed in the
simulationtobeN(0,d2).
56
v.
Dayendandrecoursetostandingfacilities.Toendthedaywithzeroreserves,bankswilltakerecoursefirst
totheTARALACfacility,andiftheimbalanceexceededthequotaof/2perbank,alsotothestanding
facilitiesatpenaltylevel.Wewillassumethatthe(penalty)depositfacilityrateiszero.Therefore,thewidth
ofthecorridorisequaltotherateofthecreditfacility.
The daily timeline is summarized in the figure below. The end of day financial accounts representation is shown
afterwards.Theinterbanktradingyis,aswillbeshownbelow,afunctionof(a)thewidthofthecorridor;(b)thesize
ofapossibleTARALACfacility;(c)thestandarddeviationofthevariousliquidityshocks(asweassumeforthesake
ofsimplicitynormallydistributedshockswithexpectedvalueofzero,theuniqueparameterdescribingeachofthe
liquidityshockstochasticvariableisitsstandarddeviation),and(d)theinterbankmarkettransactioncosts.Notethat
forthesimplicityofthepresentationinthefinancialaccounts,interbanklendingispresentedasaNETassetitemof
bank1andasanetliabilityitemofbank2.Obviouslytherelativeliquidityshockcouldalsoinducetheoppositecase.
FigureA1:Dailytimelineofcentralbankoperationsandinterbanktrading
Autonomous factor
Trading session
Autonomous factor
shocks d1, k
in
shocks d2
which
is
determined
End-day
recourse
TARALAC
facilities
Central
bank
conducts OMO
End-day recourse
penalty standing
facilities
FigureA2:Financialaccountsatendoftheday(notethatd=d1+d2)
RealAssetsEDB
DepositsBank1D/2d/2+k
DepositsBank2D/2d/2k
BanknotesB+d
Households/Investors
HouseholdEquity
Bank1
Householddeposits/debt D/2d/2+k
OMOcreditfromcentralbankB/2
RecoursetoTARALACcredfac.min(/2,max(0,(kd/2y))
Recoursetopenaltycreditfacilitymax(0,(kd/2y)/2)
Bank2
LendingtocorporatesD/2+B/2
Householddeposits/debtD/2d/2 k
Netinterbanklendingy
OMOcreditfromcentralbankB/2
RecoursetoTARALACdep.Fac.min(/2,max(0,kd/2+y)
RecoursetoTARALACcred.Fac.min(/2,max(0,(kd/2+y))
Recoursetopenaltydep.Fac.max(0,kd/2+y/2)
Recoursetopenaltycred.Fac.max(0,(kd/2+y)/2)
CentralBank
OpenmarketoperationcreditoperationB
Banknotes
B+d
TARALAC credfac min(/2,max(0,(kd/2y))+min(/2,max(0,(k TARALACdep.Fac.min(/2,max(0,kd/2y)+min(/2,max(0,kd/2+y)
d/2+y))
Penaltydep.Fac.max(0,kd/2y/2)+max(0,kd/2+y/2)
Penaltycreditfacilitymax(0,(kd/2y)/2)+max(0,(kd/2+y)/2)
LendingtocorporatesD/2+B/2
Netinterbanklendingy
RecoursetoTARALACdep.Fac.min(/2,max(0,kd/2y)
Recoursetopenaltydep.Fac.max(0,kd/2y/2)
Howwillthedifferentpolicyparametersaffecttheinterbankmarketandaveragecentralbankbalancesheetlength?
Iftherearenointerbanktransactioncosts,thenbankswouldalwaystradeintheinterbankmarketuntiltheinterbank
57
shockisfullyoffset.Incontrast,ifthereisacostCMMassociatedwithtransactinginthemoneymarket,thenthe
interbanktradeywilldependbothonCMMandontheotherparameters.IfCMM>0andthepenaltycorridorhasa
widthofzero,whichisequivalenttothecasethattheTARLACfacilityhasaquasiinfinitequota,thentherewillnever
beanytrading.Ingeneral,whatthebankingsystemneedstominimisewhenestablishingtheinterbanktradeisthe
sumofinterbanktransactioncostsplustheexpectedcostsofatwosidedrecoursetopenaltystandingfacilities.If
bankstotallyreversethedepositshiftshock,thenthesecondtypeofcostsiscertainlyzero(asthetwobankswill
endthedayeitherbothinthecreditfacility,orbothinthedepositfacility).Butiftransactioncostsarematerial,this
is suboptimal as at the end, the incremental savings relating to the second type of cost will be lower than the
interbanktransactioncost.
CallAVitheantemoneymarketvalueofdepositsofbankiandPVithepostmoneymarketvalueofdeposits
ofbanki.Then,bankswilltradeifAV1AV2CMMandinthiscasethetradingvolumewillbesuchthatPV1PV2
=CMMasbeyondthistheadditionaltransactioncostsexceedthemarginalrentfromtrading.Theexanteandexpost
valuesofdepositsforthetwobanksareasfollows:
AV1=P(kd/2</2)iC+P(/2kd/2/2)i*+P(kd/2>/2)iD
AV2=P(kd/2</2)iC+P(/2kd/2/2)i*+P(kd/2>/2)iD
PV1=P(kd/2y</2)iC+P(/2kd/2y/2)i*+P(kd/2y>/2)iD
PV2=P(kd/2+y</2)iC+P(/2kd/2+y/2)i*+P(kd/2+y>/2)iD
In the simulation, the calculus is further simplified by assuming that when choosing y the banks can linearly
interpolatebetweentheirpretradevaluations.
WhileitisobvioushowtosimulatethewidthofthecorridorandtheexistenceandquotaofaTARALACfacility,the
otherpolicychoices(frequencyandtimingofOMO,fixedratefullallotmentvs.variableratefixedvolume)canbe
simulatedbytakingassumptionsonhowtheyimpactthevolatilitiesofliquidityshockswithinthemodel.Thecaseof
alessthandailyfrequencyofopenmarketoperationcantobecapturedbyincreasingthevolatilityd1ofthepre
MM aggregate AF shock d1 in a way that this increase reflects the estimated increase of autonomous factor
uncertaintyandvolatilityfromthelongerperiodbetweentheconductoftheopenmarketoperationandtheactual
dayend.Conceptuallythisshockcapturesalsowhathascumulatedintermsofautonomousfactorimbalancefor
somedays,i.e.backtothelastopenmarketoperation.Thechoicebetweenfixedratefullallotmentversusvariable
ratefixedvolumecanbereflecteddependingonthebeliefsonehaswithregardtothenoisetermseachofthe
tender procedures injects, in addition to the actual stochastic autonomous factor process. Views can again be
reflectedinadditionalelementsoftheAFshockd1.Forexample,ifonebelievesthatthefixedratefullallotment
suffersfromtheproblem(innormal,liquidityneutraltimesandunderthesymmetriccorridorapproach)ofnoisybid
aggregation,andifonebelievesthatthevariableratefixedvolumeapproachistheeffectiveresponsetoit,then
onewouldsimplyreflectthefixedratefullallotmentwithavolatilityaddontotheAFshockd1.Thetimingofthe
openmarketoperation
58

U. Bindseil - Evaluating Monetary Policy Operational Frameworks (2016)

Cargado por

Información del documento

Descripción original:

Título original

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

U. Bindseil - Evaluating Monetary Policy Operational Frameworks (2016)

Cargado por

Copyright:

Formatos disponibles

Embargoed until presentation time of 10:55 a.m.

Mountain Daylight Time, August 26, 2016

Theconcave turnoverandlinearvolatility (bothasfunctionofcorridorwidth)suggest thatthere isan

Aneffective butparsimoniousapproachcanprobablybebased ontworegularoperations.Withsome

Statements on a quantitative success measure. The measures could in principle be mapped in a

También podría gustarte