FinanceandEconomicsDiscussionSeries

FederalReserveBoard,Washington,D.C.

ISSN1936-2854(Print)ISSN2767-3898(Online)

TradingCostsv.IndicativeLiquidityintheOf-the-RunTreasury

Market

OlegSokolinskiy

2025-049

Pleasecitethispaperas:

Sokolinskiy,Oleg(2025).“TradingCostsv.IndicativeLiquidityintheOf-the-RunTrea-suryMarket,”FinanceandEconomicsDiscussionSeries2025-049.Washington:BoardofGovernorsoftheFederalReserveSystem,

/10.17016/FEDS.2025.049

.

NOTE:StafworkingpapersintheFinanceandEconomicsDiscussionSeries(FEDS)arepreliminarymaterialscirculatedtostimulatediscussionandcriticalcomment.TheanalysisandconclusionssetfortharethoseoftheauthorsanddonotindicateconcurrencebyothermembersoftheresearchstafortheBoardofGovernors.ReferencesinpublicationstotheFinanceandEconomicsDiscussionSeries(otherthanacknowledgement)shouldbeclearedwiththeauthor(s)toprotectthetentativecharacterofthesepapers.

1

TradingCostsv.IndicativeLiquidity

intheOff-the-RunTreasuryMarket*

OlegSokolinskiy?

July7,2025

Abstract

Thispaperestimatestradingcostsintheoff-the-runTreasurymarketusingcompre-hensivetransactionsdataandmachinelearningtechniques.Theanalysisrevealsseveralkeyfindingsthatenhancetheunderstandingoftheoff-the-runTreasurymarketliquid-ity.First,theindicativebid-askspreadisshowntobeabiasedmeasureofliquidity,evenwhennotconsideringtransactionvolume.Specifically,bid-askspreadssystematicallyoverstatetradingcostsofmoreseasonedTreasuries,andtheliquidityofbenchmark,on-the-runsecuritiesaffectshowoff-the-runbid-askspreadsmaptotradingcosts.Second,thepaperdemonstratesthattradingcostsmayscalenon-monotonicallywithtransactionvolume,whichsuggestsselective,opportunisticliquidity-taking.Additionally,transac-tionsizehasgreatereffectonoff-the-runsecurities’tradingcostswhenbenchmark,on-the-runliquidityislower.Finally,indicativebid-askspreadsmaynotablyoverstatetrad-ingcostsforlargerordersofrelativelylessliquidsecurities.Thesefindingscontributetoourunderstandingofactualliquidityintheoff-the-runTreasurymarket,whilehigh-lightingthelimitationsofatraditionalliquiditymeasure.Byprovidingamorenuancedviewoftradingcosts,thisstudycontributesvaluableinsightsforsupportingfinancialstabilityandoptimalassetallocation.

Keywords:liquidity,Treasurymarket,off-the-run,effectivebid-askspread

JELcodes:G10,G12

*Thisarticlerepresentstheviewsoftheauthor,andshouldnotbeinterpretedasreflectingtheviewsoftheBoardofGovernorsoftheFederalReserveSystemorothermembersofitsstaff.IamgratefulforhelpfulcommentsfromanddiscussionswithDavidBowman,DobrislavDobrev,MaximilianDunn,SebastianInfanteBilbao,PeterJohansson,DonKim,EdithLiu,AndrewMeldrum,MariusRodriguez,andMinWei.

?BoardofGovernorsoftheFederalReserveSystem.Email:oleg.v.sokolinskiy@

2

1Introduction

TheU.S.Treasurymarketisthedeepestfixedincomemarketintheworld.Marketliquidity

–abilitytotransactsignificantamountswithouttemporarilydislocatingprices–allowstheTreasurymarkettoserveasabenchmarkforvaluationandasourceofnearrisk-freecollat-eral.However,TreasurymarketfunctioningbecameaconcernattheheightoftheCOVID-19crisis(e.g.,see

Logan

,

2020

;

FlemingandRuela

,

2020

;

Duffieetal.

,

2023

).

1

TheMarch2023SVBcollapseandtheApril2025tradetensionsledtotwomoreepisodesofseverelystrainedTreasurymarketliquidity.Furthermore,thesupplyofTreasurysecuritieshasbeenincreas-ingrelativetotheprimarydealers’abilitytointermediateinthismarket(see

Duffie

,

2023

,

2025

).Asaresult,thefunctioningoftheTreasurymarketislikelytoremainakeyfocusforbothregulatorsandinvestors.

TheTreasurymarkethastwomajorsegmentsthatdifferintheirmicrostructureand,con-sequently,liquidity.First,thereareon-the-runTreasuries–mostrecently-issuedsecuritiesforagivenmaturitydate.On-the-runTreasurieshavemuchhighertradingvolumesthanotherTreasuries,withasignificantproportionoftradesbeingoncentralizedmarketswithfirmquotes.

2

However,on-the-runTreasuriesareonlyasmallfractionofthenotionalofalloutstandingmarketableTreasurydebt.Second,thereareoff-the-runTreasuries–allTrea-surysecuritiesthatarenoton-the-run.Incontrasttoon-the-runsecurities,off-the-runTrea-

suriestradeinfrequentlyinadecentralized,bilateralmarketprovidingindicativequotes.

3

Duetothismicrostructureoftheoff-the-runTreasurymarket,itsliquidityisdifficulttogauge.

Inthispaper,Isuggestanestimateoftheeffectivethebid-askspreadmeasure(see

Dem-

1Adrianetal.(2025)considersTreasurymarketliquidityoveralongertimelinestartingfromtheGFC.

2Afirmquotewithanassociatedvolumeisacommitmentofaliquidityprovidertotradeinthatsizeif

requestedtodoso.

3Indicativequotesarenotfirmcommitmentstotrade,andthepriceofapotentialtradeismaybesetupon

furthernegotiations.

3

setz

,

1968

)fortheoff-the-runTreasurymarketliquiditylargelybasedonactualtrades.

4

Thecorecomponentsofthesuggestedmethodfortheassessmentofactualtradingcostsintheoff-the-runTreasurymarketare(i)relianceonoff-the-runTreasuries’transactiondataalongwithfirmquotesforon-the-runTreasuriesand(ii)state-of-the-artmachinelearningapproachtofittingthetermstructureofbenchmark,on-the-runrates.Themethodreliesontheon-the-runTreasuries’quotesasthesourceofmostrecentinformationonchangesinthebenchmarkinterestratetermstructure.Inturn,periodicsnapshotsofindicativequotesenabletheestimationofidiosyncraticpricecomponentsofparticularoff-the-runsecurities.Themachinelearning-basedmethodof

Filipovi′cetal.

(

2022

)forfittingthetermstructureofinterestrates–thesecondcriticalcomponentofthesuggestedmethod–deliverssufficientaccuracyandstabilitytoestimatethetermstructureofbenchmark,on-the-runratesateachmomentwhenatradeinanoff-the-runsecurityoccurs.

5

Havingconstructedameasureoftradingcostsforoff-the-runTreasuries,Iexplorehowitdeviatesfromindicativeliquidity.First,Idocumentbiasesinindicativebid-askspreadsthatareunconditionalontransactionvolume.Inparticular,indicativebid-askspreadstendtosystemicallyoverstatetradingcostsformoreseasonedsecurities.Anotherbiasofindicativebid-askspreadsresultsfromtheirfailuretoreflectbenchmark,on-the-runTreasurymarketliquidity.Whenbenchmark,on-the-runliquidityisdegraded,indicativebid-askspreadsforoff-the-runsecuritiestranslateintogreatertradingcosts.Thisbiasisofparticularsignifi-canceforassessingfinancialstability–amarketdysfunctionresultingfromprohibitivetrad-ingcostsintheoff-the-runTreasurymarketmayoccuratloweroff-the-runbid-askspreadlevelsthanasimplerlinearregressionmodelmaysuggest.

Second,overcomingtheinherentlimitationofindicativequotes,Iexploretheeffectoftransactionsizeonprice.Whileindicativebid-askspreadsdonotprovideinformationon

4Bid-askspreadisthedifferencebetweenthelowestpriceatwhichasecuritymaybeboughtandthehighestpriceatwhichthesecuritymaybesold.

5SubjecttoconsideringtradesthatoccurduringtheBrokerTecplatform’sactivehours.

4

howtradingcostsscalewiththesizeofanorder,effectivebid-askspreadsareassociatedwithparticulartransactionvolumes.Withinanon-parametricframeworkformodelinghowtradingcostsvarywithtransactionvolume,Iestimatethecorrespondingscalinglawscondi-tionaloneitherbenchmarkliquidityorrelativeliquidityofthespecificoff-the-runsecurity.Ishowhowtheeffectofvolumeontradingcostsintheoff-the-runmarketvarieswiththeliquidityconditionsinthebenchmark,on-the-runTreasurymarket:worsebenchmarkliq-uiditytranslatesintohighertradingcostsintheoff-the-runmarket,evencontrollingforthecorrespondinglywideroff-the-runsecurities’indicativebid-askspreads.Finally,Idemon-stratethatindicativebid-askspreadsmaynotablyoverstateexpectedtradingcostsforlargertransactionvolumesofrelativelylessliquidoff-the-runsecurities.Moregenerally,Iobtainevidenceinfavorofselectiveliquiditytaking–endogeneityoftradinginresponsetoavail-ableliquidity.

Literature

Intheremainderofthissection,Iplacethispaperwithintheframeworkoftheexistentresearch.Inparticular,therearethreestrandsofresearchthatthispapercontributesto:(i)liquiditymeasurementinthedecentralized,bilateralTreasuryoff-the-runmarket;(ii)differencesbetweenindicativeandactualliquidity;and(iii)scalingoftradingcostswithtransactionvolume.

First,thesuggestedmethodforthecalculationoftheeffectivebid-askspreadcontributestotheliteratureonliquiditymeasurementintheoff-the-runTreasurymarket.Theopacityoftheoff-the-runTreasurymarketshiftedresearchfocustotheliquidityintheon-the-runTreasurymarket,forwhichthereisampleexcellentresearchby

Fleming

(

2001

)amongoth-ers.Incontrast,liquidityintheoff-the-runmarketisthesubjectoffarfewerstudies.Duetosparsenessoftradesinspecificoff-the-runTreasuries,theindicativebid-askspread(orsometransformthereof)isaprominentmeasureofTreasurymarketliquidity,asin

Pasquariello

5

andVega

(

2009

)and

Goyenkoetal.

(

2011

).Anotherstrandofliteraturereliesonliquiditymeasuresthataremoresimilartopriceimpact.Amongthesepapers,

Babbeletal.

(

2004

)isconceptuallyclosesttotheapproachIadoptinthispaper.

Babbeletal.

introducesapricepressuremeasureintheoff-the-runTreasurymarket,whichisthedifferencebetweenasyn-thetic‘on-the-runprice’–priceofthesecurityobtainedbydiscountingcashflowsusingtheinterestratetermstructurefromon-the-runTreasuries–andtheactualtradepriceforrela-tivelylowvolumetransactions.Similarto

Babbeletal.

,Ialsouseon-the-runTreasuriesasthebenchmarkformostup-to-datepricing,butfitthetermstructureusingthemoreflexiblemodelof

Filipovi′cetal.

(

2022

)insteadofthe

NelsonandSiegel

(

1987

)parametricmodel.

6

Theapproachof

Babbeletal.

alsoreliesontheidentificationofwhetherthetradewasini-tiatedbythebuyorsellsidebasedontheassumptionthat$1mmtradeshaveanegligibleeffectivebid-askspread.Incontrast,usingindicativequotesIcircumventthenecessitytoidentifythetrade-initiatingsideand,instead,directlyestimatethespreadbetweencashflowequivalenton-andoff-the-runsecurities.Amongotherpapersinthestrandoflitera-turegoingbeyondindicativebid-askspreadsare

FurfineandRemolona

(

2002

)and

Diazand

Escribano

(

2017

).

FurfineandRemolona

usesthe

Hasbrouck

(

1991

)priceimpactmeasuretoestimateliquidityoverlongerperiodsbasedondailyreturns.

DiazandEscribano

usesasetofmeasuresincludingthe

Amihud

(

2002

)measureandtheirversionofthe

Jankowitsch

etal.

(

2011

)dispersionmeasure.Finally,thespreadbetweenon-andoff-the-runTreasuriesiscommonlyreferredtoastheliquiditypremium,asin

DiazandEscribano

.However,thisspreadcapturesnotonlyliquiditybutdifferingfinancingcostsofon-andoff-the-runsecu-rities,(

Krishnamurthy

,

2002

),aswellasrepocounterpartyrisk,(

LiuandWu

,

2017

).

Second,thispapercontributestotheliteratureondifferencesinindicativeandactualliq-uidity.Inequitymarkets,

BlumeandGoldstein

(

1992

)demonstratesnotabledifferencesbe-tweentheeffectiveanddisplayedspread,withthedifferencesbeingaffectedbytheposted

6

Andersen

(

2007

)arguesthatparametricmodelsdonotallowforsufficientflexibilityandarenotincommonusebymarketmakers–acriticalrequirementforobtainingabenchmark,on-the-runinterestratetermstructure.

6

volumeavailableatthequotes.Inthemarketforcollateralizedloanobligations,

Hender-

shottetal.

(

2024

)showsthatindicativequotesinover-the-countermarketsmayleadtoanoverestimateofavailableliquiditywhenfailedattemptstotradeareignored.Inthispa-per,Iuncoverseveralbiasesinindicativebid-askspreads.Inparticular,indicativebid-askspreadsoverestimatetradingcostsforseasonedandrelativelylessliquidsecurities.More-over,Ishowthatrelianceontheregularrelationshipbetweenindicativeandeffectivebid-askspreadsleadstoanunderestimateoftheliquiditydeteriorationintheoff-the-runTrea-surymarketduringperiodsoflowbenchmark,on-the-runTreasurymarketliquidity.Thisevidenceaugmentstheresultsin

FurfineandRemolona

(

2002

)thatoff-the-runsecurities’liquidity–asmeasuredbypriceimpact–deterioratesproportionatelymorethanon-the-runsecurities’liquidityduringperiodsofmarketstress.

Finally,thispapercontributestoresearchonhowtradingcostsscalewithtransactedvolumeindecentralized,bilateralmarkets.ThemajorityofexistentempiricalresultsareforCLOB-drivenmarkets,withthenotableexceptionof

Babbeletal.

(

2004

).

Bouchaudet

al.

(

2009

),

Farmeretal.

(

2006

),and

LilloandFarmer

(

2004

)amongothersfindthattradingcostsarenon-linearintransactionvolume.AnotabledistinctionofthispaperfrommostempiricalstudiesinthefieldisthatIallowforthescalinglawtovarywithmarketcondi-tionsandrelativeliquidityofthesecurity.Therefore,Iconsiderconditionalscalinglaws.Furthermore,Ifindevidenceinsupportofselectiveliquiditytaking–largetradesgetexe-cutedwhenliquidityishigh.Thisfindingcomplements

Hendershottetal.

(

2024

)thatdrawsattentiontounobservedfailedtradeswhenavailableliquidityislow.

Therestofthepaperisorganizedasfollows.InSection

2

,Isuggestamethodforcal-culatingtheeffectivebid-askspreadforoff-the-runTreasuries.InSection

3

,Iexploretherelationshipbetweenactualandindicativeliquiditybycomparingthestatisticalpropertiesofeffectivebid-askspreadsandpriceimprovementsrelativetoindicativebid-askspreads.

7

Inconclusionofthatsection,Iidentifyvariousbiasesinindicativebid-askspreadsusingthelinearmixedeffectsmodelingframework.InSection

4

,Iextendthelinearmixedeffectsmodeltoobtainscalinglawsfortheconditionaldependenceoftheeffectivebid-askspreadontransactedvolume.Iallowthescalingwithvolumetodependonbenchmark,on-the-runTreasurymarketliquidityinSection

4.1

andonrelativeliquiditycharacteristicsofsecuritiesinSection

4.2

.InSection

5

,Ireiteratethemainresultsofthepaperandsuggesttopicsforfurtherresearch.Finally,inaseriesofappendices,Iprovidesomebackgrounddetails,aswellasexploretherobustnessandextensionsofresultstovariouscloselyrelatedTreasurymarketsegments.InAppendix

A

Iprovideabriefoverviewofon-andoff-the-runTreasurymarketmicrostructure.Then,Iconsider(i)theeffectofAlternativeTradingSystems(ATS)intermediationonliquidityinAppendix

B

,(ii)interconnectednessofliquidityinthedealer-to-clientanddealer-to-dealersegmentsinAppendix

C

,and(iii)executionqualityofretailclients’tradesinAppendix

D

.

2EffectiveBid-AskSpreadforOff-the-RunTreasuries

Themethodologicalframeworkofthispaperhastwocomponents–(i)analgorithmforcal-culatingtheeffectivebid-askspreadforoff-the-runTreasuries,and(ii)alinearmixedmodelframeworkthatallowsforrandomeffectsassociatedwitheachsecurity,whileenablingnon-parametricexplorationofhowtheeffectivebid-askspreadscaleswithordersize.IcoverthefirstcomponentinthissectionanddeferthedescriptionofthemixedmodelingframeworkuntilthetimeitisappliedinSection

3.3

.

2.1AdvantagesoverCompetingMeasureofLiquidity

Theeffectivebid-askspread(see

Demsetz

,

1968

)isthedifferencebetweenthetradepriceofanassetanditsfairvalue.Thewidertheabsolutevalueoftheeffectivebid-askspread,the

8

greaterthetradingcostsare.

7

Theeffectivebid-askspreadisameasureofactual,experi-

encedliquidityforeachindividualtransaction.Consequently,transactioncharacteristics–liketradedvolume–arelinkedtoeachobservationoftheeffectivebid-askspread.

Theeffectivebid-askspreadhasthreemainadvantagesoverindicativebid-askspreads.

First,theeffectivebid-askspreadisbasedonactualtransactionsandnotmereindicationsofwhereagenerictradecouldoccur(absentpossiblenegotiations).Second,theeffectivebid-askspreadallowstheexplorationofhowtradingcostsscalewithvolume,whichisentirelylatentinindicativebid-askspreads.Theremaynotbeapreciseordersizeassociatedwiththequote.Thisisparticularlythecaseforstreamedquotes.

8

Ingeneral,a‘normalmarketsize’mayserveastheimpliedpotentialtransactionvolume.However,whatis‘normal’maychangedependingonmarketconditions.Finally,duetothebilateralnatureoftradingintheoff-the-runTreasurymarketandmultiplecompetingintermediaries,obtainingreliablebestindicativebidandaskpricesforalloff-the-runTreasuriesmaybeprohibitivelyexpensiveevenforsomemarketparticipants.Insummary,marketmicrostructureconsiderationsfavortheeffectivebid-askspreadovertheindicativebid-askspreadasameasureofliquidityintheoff-the-runTreasurymarket.

Theeffectivebid-askspreadisalsoamorefittingmeasureforoff-the-runTreasuryliq-uiditythanpriceimpact.Priceimpact–howmuchtradeororderflowofagivenmagnitudeoverasetexecutionhorizonaffectsthemarketprice–isaprominentmetricinCentralLimitOrderBook(CLOB)markets.

9

OnCLOB-basedplatforms,marketparticipantssplittheirtotaldesiredtransactionintomultiplesmallerchildordersthattheysubmitovertimewhiletargetingaparticularaverageexecutionprice,withpossibleopportunisticdeviationsinre-sponsetotheCLOBdynamics.Thismannerofexecutionissupported,ifnotrequired,by

7Undertheassumptionthatpositive(negative)differencebetweenthetradepriceofanassetanditsfunda-mentalvaluecorrespondstobuyer-initiated(seller-initiated)transactions.

8Asnotedabove,marketmakersmaystreamquotestoclientsviatheirin-houseplatformsorexternalAlter-nativeTradingSystems(ATS).Incontrasttotherequestforquote(RFQ)protocol,clientsdonotcontactdealerstoobtainquotesand,thus,hidetheirinterestinapotentialtrade.

9BrokerTecplatformisaprominentexampleofaCLOB-basedmarketforon-the-runTreasuries.

9

thelimitedimmediatelyavailableliquidityontheCLOB.Instarkcontrast,suchexecutionwouldbeconsideredabreachofexpectedbehaviorinthebilateralmarketforoff-the-runTreasuries(see

Wood

,

2018

).Ifaclientsplitsalargeorderamongmultiplemarketmakers,thepriceoftheassetislikelytomoveagainstthemarketmakersastheinformationsubse-quentlysipsintothemarket.Ifaclientweretoengageinsuchordersplitting,itmayexpecttoreceiveconsiderablylessfavorablequotesinitsfutureinteractionswithmarketmakers.

10

Thus,theatomicnatureoftheeffectivebid-askspreadthatdoesnotincorporateflowsoveraperiodoftimeismoresuitablefortheoff-the-runTreasuriesmarket.Inaddition,asnotedin

Hendershottetal.

(

2024

),accurateestimationofpriceimpactatadailyfrequencyrequiresmoretransactionsthancommonlyoccurinmanyindividualoff-the-runTreasuries.

2.2EstimationAlgorithm

Theanalyticaldefinitionoftheeffectivebid-askspread,adaptedfrom

Hagstr?mer

(

2021

),is:

2Di(Pit,ade-Ptair),(1)

wherePit,adeisthetransactionpriceandPtairisthefairvalueofsecurityiattimet;Diis

equalto1forbuyer-initiatedand-1forseller-initiatedtrades.Thescalingby2expressestheeffectivebid-askspreadonthesamescaleastheindicativebid-askspread.Fortransac-tionpricesIrelyonthecomprehensiveTreasuryTradeReportingandComplianceEngine(TRACE)datafromtheFinancialIndustryRegulatoryAuthority(

FINRA

).SinceTreasuryTRACEdatadonotcontainanexplicitidentifierofthesideinitiatingatransaction,Imaketheassumptionthataggressivebuying(selling)leadstopositive(negative)deviationsofthe

10Thesamelogicdoesnotapplyincentralized,CLOB-basedmarkets.Whilethegradualexecutionofalargeorderwouldalsomovethepriceagainstmarketmakersthatprovidedliquidityfortheearlierchildorders,theanonymityonCLOB-basedmarketsdoesnotallowmarketmakerstoretaliateagainstclientsinfutureinterac-tionsbasedontheirpastbehavior.

10

tradepricefromthefairvalue,thatis,Iset:

Inordertomaketheeffectivebid-askspreadcomparableforsecuritiesofdifferentmaturi-ties,Iintroducetheduration-normalizedeffectivebid-askspread:

wheredurationi,tmeasuresthesensitivityofsecurityi’spricetochangesininterestratesattimet(itsmodifieddurationmultipliedbyitsprice).Normalizationbydurationexpressestheeffectivebid-askspreadintheyieldspace,ratherthaninthepricespace–duration-normalizedeffectivebid-askspreadistheparallelshiftoftheinterestratetermstructurethatwouldcausetwicethedollardifferencebetweenthetransactionpriceandthefairpriceofasecurity.Suchnormalizationoftheeffectivebid-askspreadbydurationenablescomparisonandunifiedmodelingofliquidityacrossthecurve.Forbrevity,Ihenceforthrefertotheduration-normalizedeffectivebid-askspreadinEq.(

2

)astheeffectivebid-askspread.

Whilethedefinitionoftheeffectivebid-askspreadisstraightforward,itsestimationiscomplex.Thetradepriceisobserved,butthefairvaluehastobeestimated.Themostcom-monestimateofthefairvalueisthemid-pointbetweenthebestbidandaskquotesatthetimeofthetrade.However,

Hagstr?mer

(

2021

)showsthatthisfairvalueestimatebiasestheeffectivebid-askspreadmeasure.Moreover,unlikeintheequitymarket,bestbidandaskpricesareopaqueforoff-the-runTreasurysecurities.Tradinginoff-the-runTreasuriesisbi-lateralandrelativelyreliablequotesfromaspecificdealerareavailableonlyviatherequestforquoteprotocol;streamedquotestendtobeanotablyrougherindicationofprice.

11

Thus,

11Clientscannotroutinelyrequestquotesfromtheentireuniverseofdealerstoassessthetruebestbidandaskquotes.Requestingquoteswithoutacertainamountoftradingwouldviolatethemarketetiquette–inmanycases,generationofoff-the-runTreasuryquotesisnotfullyautomatedandis,thus,costlytothedealer.

11

someindicativequotesmaydifferfromthebestavailablebidsandoffers.Also,conversa-tionswithmarketparticipantsindicatethatmarketmakersdifferintherelativefirmnessoftheirindicativequotes–withsomebeingmorereliableindicationsofthepriceatwhichatradecanoccur.Furthermore,inbilateralrelationshipstheidentitiesofcounterpartiesareknownandmarketmakersmayshadetheirquotesbasedontheirassessmentofhowin-formedtheclientis(see

Wood

,

2018

).Consequently,thereisnosinglepricethatwouldbeavailabletoallmarketparticipants.

GuidedbytheTreasurymarketmicrostructure,Iestimateoff-the-runTreasuries’fairval-uesbasedontransparentfirmquotesforon-the-runTreasuriesfromtheleadingBrokerTecATSandindicativeNPQSquotesfromtheFederalReserveBankofNewYork.Ingreaterdetail,tocalculatethefairvalueofanoff-the-runTreasury,Ptair,Ideconstructitintothebenchmarkandidiosyncraticpricecomponents,asfollows.

Thebenchmarkcomponentoftheoff-the-runTreasury’spriceisthepriceofthehypothet-icalon-the-runsecuritywithequivalentcashflows–itreflectstheinterestratetermstruc-tureunderconditionsofhighliquidityandlowfinancingcosts.Anidiosyncraticcomponentoftheoff-the-runTreasury’spricecapturestheeffectsoflowerliquidityandlessfavorablefinancingofaspecificsecurityrelativetotheon-the-runbenchmark.Thebenchmarkcom-ponentreflectstheinformationinnewsannouncementsandflowsoffunds.Consequently,thebenchmarkcomponentmaybehighlyvariableintraday.Incontrast,thefinancingcostandliquiditypremiaarerelativelystableintraday–thesepremiaprimarilydependonthefinancingandliquidityconditionsovertheremaininglifeofthebond.

12

Duetohigherliquidityandtradingvolumesintheon-the-runTreasurymarket,itisthelocusofTreasurypricediscovery.Thus,toestimatethebenchmarkpricecomponentIuse

12Cheapfinancingandhighliquiditybenefittheinvestorovertheholdingperiod,andtheirvaluesovertheremaininglifeofthesecuritywilllikewiseaffectfutureholdersofthesecurity.Consequently,financingandliquiditycharacteristicsovertheremaininglifeofthesecuritydeterminethecorrespondingpremia.Thatsaid,valueofliquiditycanspikeduringmarketstress.InSection

3.1

,Iintroduceaforecastaccuracyfilterthatmitigatestheeffectofsuchspikesonliquiditymeasurement,albeitatthecostofretainingfewerobservations.

12

firmon-the-runTreasuries’quotesfromoneoftheleadinginter-dealerbroker(IDB)Bro-

kerTecAlternativeTradingSystem(ATS).

13

Crucially,firmon-the-runTreasuries’quotesareavailableatahighfrequency,whichprecludespricingbasedonstaleinformation.Anotherargumentinfavoroftheon-the-runTreasurymarketasthesourceofthebenchmarkpricecomponentisthatoff-the-runTreasuriescanbequot