Speech

GSE Risks St.LouisSocietyofFinancialAnalysts St.Louis,Missouri January13,2005 PublishedintheFederalReserveBankofSt.LouisReview,March/April2005,87(2,Part1),pp.85-91 Almost two years ago, in a speech at a FannieMaeandFreddieMacfacefivemajor conference hosted by the Office of sourcesofbusinessrisk:creditrisk,prepayment FederalHousingEnterpriseOversight risk,interestrateriskfrommismatchedduration (OFHEO), I argued that government- ofassetsandliabilities,liquidityrisk,andopera- sponsored enterprises (GSEs) specializing in the tionalrisk.Asixthrisk,so-calledpoliticalrisk, mortgage market, especially Fannie Mae and arisesfromthepossibilityofregulatoryorstatutory Freddie Mac, exposed the U.S. economy to sub- revisionsthatcouldadverselyaffectthosewho stantialrisk,primarilybecausetheircapitalposi- holdthefirms’debtorequity.I’lldiscussthese tions are thin relative to the risks these firms risksinturn,devotingmuchmoretimetosome assume (Poole, 2003). I had a number of specific thanothers.Alongtheway,Iwillalsodiscussan risks in mind, but did not elaborate the nature of extremelyimportantpointconcerningthefre- these risks. My purpose here is to provide that quencyofoccurrenceoflargeinterestratechanges. elaboration. I will concentrate on risks facing Thisissueiscriticaltounderstandingtherisksof anystrategyinvolvingincompletehedging. Fannie Mae and Freddie Mac, but it should be understood that the Federal Home Loan Banks raise many of the same issues. AnunderstandingoftherisksfacingFannie CREDIT RISK MaeandFreddieMac—whichIwillsometimes Creditriskoccursbecausehomeownerscan refertoas“F-F”tosimplifytheexposition—is anddodefaultonmortgageloans.Eventhough importantfromtwoperspectives.First,investors defaultratesonmortgagesintheUnitedStates shouldbeawareoftheserisks.Althoughmany arelow,inrecentyearslessthan1percent,they investorsassumethatF-Fobligationsareeffec- arenotzeroandvaryconsiderablyacrossregions. tivelyguaranteedbytheU.S.government,the Creditriskonmortgagescanbehandled,asinfact factisthattheguaranteeisimplicitonly.Iwill FannieandFreddiedoveryeffectively,through notattempttoforecastwhatwouldhappenshould apolicyofgeographicdiversificationandofnot eitherfirmfaceasolvencycrisis,becauseIjust buyingasignificantnumberofhighloan-to-value donotknow.WhatIdoknowisthattheissueis mortgages,aswellasthroughtheuseofmortgage apoliticalone,andpoliticalwindschangein insuranceandguarantees. unpredictableways. Inassessingcreditrisk,itisimportantnotto Asecondreasontounderstandtherisksis focusjustonnationalaverageconditions.For thatsoundpublicpolicydecisionsdependon example,althoughaveragehousepricesinthe suchunderstanding.Toreducethepotentialfor UnitedStateshavenotdeclinedyeartoyearsince afinancialcrisis,risksneedtobemitigated. theGreatDepression,1priceshavedeclinedin 1 Thisstatementmayormaynotbestrictlyaccurate.AnnualdataonnationalaveragenewhomepricesfromtheU.S.Censusstartin1963and showsmalldeclinesinthelate1960sandearly1990s.Annualdataforthemediansalespriceofexistingsingle-familyhomesfromthe NationalAssociationofRealtorsstartin1968anddonotexhibitanyannualdeclines. 1 FINANCIALMARKETS particularsignificantmarkets.Someexamples However,formanyyearsF-Fhavebeenaccu- wouldbeBoston1989-92,LosAngeles1991-96, mulatingaportfoliooftheirownMBSsand SanFrancisco1991-95,andTexas1987-88.More directlyownedindividualmortgages.Forthetwo formally,thedispersionofchangesinhouseprices firmstogether,theseportfoliosareverylarge, andnotjustthenationalaverageisrelevantfor amountingtoover$1.5trillionattheendof2003. judgingmortgagedefaultrisk. Thus,F-Fassumeprepaymentriskbyholding Giventhathousepricesdosometimesdecline theseassets. inparticularmarkets,itispossiblethatageograph- Underthemostconservativefinancialstrategy, icallydiversifiedportfolioofmortgagescould FannieandFreddiecouldmitigatecompletelytheir suffersignificantlosses.Therefore,todetermine prepaymentriskbyissuinglong-termcallable thecapitalafirmneedstoholdagainstcreditrisk bondstofinancetheirholdingsoflong-termmort- requiresnotonlyanalysisofthegeographical gageassets.Withsuchastrategy,thecashinflow diversificationintheportfoliobutalsoananalysis fromtheassetsmatchesexactlythecashoutflow ofrisksandlikelylossesgivenforeclosurein requiredtoservicetheliabilities,andinterestrate varioushousingmarkets.FromeverythingIknow, andprepaymentriskareperfectlyhedged. FannieandFreddiedoafinejobofmanaging creditrisks,butIamnotonewhobelievescredit riskscanbeignored. A DIGRESSION ON FINANCIAL ENGINEERING PREPAYMENT RISK Inpractice,bothFannieandFreddiemake limiteduseoflong-termcallablebonds.Rather, FannieMaeandFreddieMacissuemortgage- theyissuenon-callablelong-termbondsanda backedsecurities(MBS)againstpoolsofconform- significantamountofshort-termdebt.Doingso ingmortgages—mortgageswithdollarvalueat exposesF-Ftoprepaymentriskandinterestrate orbelowtheconforminglimitthatqualifiesthe riskfromamismatchofdurationofassetsand mortgagesforF-Foperations.Allsuchmortgages liabilities.Theythenusevariousdevicestoman- havenoprepaymentpenaltiesandaretherefore agetheriskscreated. subjecttoprepaymentrisk. BeforediscussingthewaysF-Fmanagepre- Infinancelingo,thesefixed-ratemortgages paymentandinterestraterisk,itisworthnoting carryacalloption.Intheeventthatinterestrates thatthemoreelaborateportfoliopolicyhasnoth- fallduringthelifeofthemortgage,thehomeowner ingwhatsoevertodowiththemortgagemarket canexercisetheoptiontorefinancethemortgage, perse.Considerthisanalogy:Aninvestment effectivelycallingtheoutstandinghighinterest companycouldownaportfoliooflong-termcor- ratemortgageandreplacingitwithanewlower poratebonds,mostofwhichbecomecallableat interestrateobligation.Historically,theexercise somepointbeforematurity.Wheninterestrates ofthisoptionwasconstrainedbyrelativelyhigh transactioncosts.Inrecentyears,however,trans- fall,corporationscallsuchbondsandrefinance actioncostshavefallenconsiderablysothatthe withlower-ratebonds.Thephenomenonisexactly calloptioninthetypicalfixedratemortgage thesameasthatobservedinthemortgagemarket, instrumentcomesin-the-moneywithrelatively exceptthatcorporatebondshaveacertainnumber smalldeclinesinmortgagerates.Suchrefiactivity ofyearsofcallprotectionwhenissuedandpaya hasbeensubstantialinrecentyears. callpremiumwhencalled. WhenFannieandFreddieissueMBSstobe AsfarasIknow,therearenoclosed-end heldbytheinvestingpublic,buyersofthebonds investmentcompaniesthatholdaportfolioof assumetheprepaymentrisk.FannieandFreddie corporatebonds,financedbytheirownissuesof servicetheMBSsandguaranteethem,thusassum- shortandlongdebt.Thereason,Iconjecture,is ingthecreditrisk. thatthereisnoimpliedfederalguaranteeonsuch 2 GSERisks obligations,whichmeansthataninvestment longer-termandlesslikelyrateshocks;(3)Imple- companycouldnotearnasatisfactoryspread mentadditionalhedgesasinterestratelevels fromholdingaportfolioofmarketablecorporate change,andtheunlikelybecomeslikely”(Jaffee, bondsfinancedbyitsownobligations. 2003,pp.16-17).Theterm“dynamichedge”refers TheGSEs,however,havethebenefitofthe toastrategythatinvolvescontinuousrebalancing impliedfederalguarantee,whichmakestheir ofthefirm’sportfolioinanattempttomaintain financialengineeringprofitable.Becauseofthe acceptableriskexposures.Adynamichedging impliedguarantee,F-Fcanoperatewithasmall strategycanbequitesuccessfulwhenpricesmove capitalpositionandissuetheirownobligations continuously,insmallsteps,butisincreasingly atratesthatarelittleabovethosepaidbytheU.S. ineffectivethelargerarepricediscontinuities,or Treasury.ThespreadoverTreasuriesissmallerat pricejumps. theshortendofthematuritystructurethanatthe Theadvantageofusingderivativesand longend,whichiswhyF-Fissuelargeamounts imperfectdynamichedgingtomanageinterestrate ofshort-termdebt.Thisfinancialengineeringhas riskisthatthesestrategiesarelesscostlythanthe littletodowiththemortgagemarket,exceptthat perfecthedgeandperformequallywellwhenthe F-Fareauthorizedtoholdmortgagesratherthan interestratevolatilityismoderate.Thedisadvan- corporatebondsintheirportfolio.Thefinancial tageisthatpotentiallossesassociatedwiththe engineeringhasnothingtodowiththemortgage unlikelyriskscanbeverylarge. marketperseandeverythingtodowiththe • Becauseofimperfectdynamichedging,F-F impliedfederalguarantee. maysufferasignificantlosswheneverthere areunexpectedandlargeinterestratemove- mentsineitherdirection.Formalmodelsof INTEREST RATE RISK dynamichedgingassumepricecontinuity FannieandFreddiecreateinterestrateriskfor anddonotworkwellwhenpricesjump themselvesbyfinancingtheirportfoliothrough discretelybylargeamounts. amixtureoflong-termnon-callablebondsand • FannieMaeandFreddieMacareexposed short-termobligations.Bothfirmshaveobligations tothecounterpartydefaultriskintheir duewithinoneyearintheneighborhoodof50 derivativecontracts.Thecounterparty percentoftotalliabilities. defaultriskpersemaybesmallbecause Havingcreatedprepaymentandinterestrate bothfirmsrequireallcounterpartiestopost riskbynotmatchingthecharacteristicsoftheir collateralonaweeklybasis.However,ata obligationstothecharacteristicsoftheirmortgage timeofdisruptedfinancialmarkets,itwould assets,F-Fmustthenpursuesophisticatedhedging beverycostlytoreplacetheswappositions strategies.Theyemploydebtandinterestrate ofadefaultingcounterpartybecausethe swapstocreatesyntheticlong-termobligations— othercounterpartiesarelikelytohave ashort-termobligationplusafixed-payswapeffec- similarproblems. tivelycreatesacashflowobligationthatmimics thatofalong-termbond.Theyalsouseoptions— inparticular,swaptions—tohedgetheprepayment JUDGING THE SCALE OF risk. INTEREST RATE RISK Finally,likemanylargefinancialfirms, FannieMaeandFreddieMacemployastrategy Withouthighlydetailedinformationaboutthe ofimperfectdynamichedging,whichinvolves hedgingstrategiespursuedbyF-F,itisimpossible threesteps:“(1)Maintainverycompletehedges toofferaquantitativeassessmentofthescaleof againstthelikely,near-term,interestrateshocks; interestraterisktowhichthefirmsareexposed. (2)Uselesscompletehedgesorevennohedgesfor However,thefactthathedgingisincompleteraises 3 FINANCIALMARKETS Figure 1 Trading Day Percentage Changes in On-the-Run 10-Year Treasury Note Percentchange 6 Dayswhenon-the-runbondchanged are omitted 3 0 –3 –6 78 80 82 84 86 88 90 92 94 96 98 00 02 04 May May May May May May May May May May May May May May 1 1 1 1 1 1 1 1 1 1 1 1 1 1 NOTE:Dashedlinesshowarangeof±3.5standarddeviations. warningflags.Thereasonisthatstandardhedging theyaretradedinahighlyliquidmarket,are strategiesrelyontheassumptionthatchangesin employedextensivelyinhedgingstrategies.Price securitiespricesfollowanormaldistribution— changesfortheTreasurybondforabout25years thefamiliarbell-shapedcurve.TheBlack-Scholes areshowninFigure1.Theverticalaxismeasures formulaforpricingoptionsassumes,forexample, thedailypercentagepricechange,andthedashed thatassetpricesfollowanormaldistribution. bandsdefinearangeplusandminus3.5standard Tojudgerisk,westartbycomputingthestan- deviationsfromthemean. darddeviationfromalonghistoryofpricechanges Thefirstthingtonoteinthisfigureisthefre- insomeparticularmarket.Thenormaldistribu- quencyoflargechanges.Roughly0.75percentof tionisthebaselinecase.Whatweinfactobserve theTreasurybondpricechangesinthesampleare are“fattails,”bywhichwemeanthatthereare greaterinabsolutevaluethan3.5standarddevia- manymorelargepricechanges—changesoutin tions,morethan16timesthenumberofsuch thetailsofthedistribution—thanexpectedwith outliersthatwouldbeexpectedfromanormal anormaldistributionofthecalculatedstandard distributionofpricechanges.Letmerepeat— deviation.Failuretotakeadequateaccountoffat thereare16timesmorepricechangesinexcess tailsisresponsibleformanyfailuresoffinancial of3.5standarddeviationsthanexpectedwiththe firmsovertheyears,suchasthe1998failureof normaldistribution.Assuming250tradingdays LongTermCapitalManagement. inayear,onaveragebondpricechangesofthis Akeysecurityinthecontextofthemortgage orgreatermagnitudeinabsolutevalueoccurtwice marketisthe10-yearon-the-runTreasurybond. peryearinsteadofonceevery8years.Thenormal Long-termmortgagesarepricedoffthe10-year distributionprovidesagrosslymisleadingpicture Treasury,andTreasurybondsthemselves,because oftheriskoflargepricechanges.Reallylarge 4 GSERisks changesof4.5ormorestandarddeviations—the clearlybyRichardPosnerinhisrecentWallStreet onesthatcanbreakahighlyleveragedcompany— Journalop-edarticle(Posner,2005,p.A12)on occuronly7timesinamillionunderthenormal theIndianOceantsunami.Posnerwrites: distribution,butthereare11suchchangesinthe TheIndianOceantsunamiillustratesa 6,573dailyobservationsinthefigure. typeofdisastertowhichpolicymakers Asecondpointtonotefromthefigureisthat paytoolittleattention—adisasterthat largechangestendtoclustertogether.Itappears hasaveryloworunknownprobabilityof thatmarketsgothroughperiodsofrelativevola- occurring,butifitdoesoccurcreates tilityandotherperiodsofrelativetranquility. enormouslosses…Thefactthatacatas- Clusteringisimportantbecauseafirmmaybe tropheisveryunlikelytooccurisnota rockedseveraltimesinquicksuccessionbylarge, rationaljustificationforignoringtherisk unanticipatedpricechanges.Incompletehedges ofitsoccurrence. againstlargepricechangesexposeafirmtocas- cadingfailure. Ofcourse,thelossofscoresofthousandsof livesinthetsunamiisnottobecomparedtothe Thefattailsphenomenonhasbeendocu- lossesfromafinancialcrisis.Nevertheless,the mentedforawiderangeoffinancialinstruments twodisastercasesillustrateanotherimportant overmanydifferentsampleperiods.Benoit pointaboutriskmanagement.Inthecaseofthe MandelbrotandRichardHudsonrefertothese tsunami,nothingcanbedoneabouttheprobabil- featuresas“wildrandomness”(Mandelbrotand ityofoccurrence;lossmitigationdependson Hudson,2004,p.32).Theyconclude: installingwarningsystems.Inthecaseoftherisk Extremepriceswingsarethenormin offinancialcrisis,thekeypolicyinterventionis financialmarkets—notaberrationsthat toreducetheprobabilityoftheevent,bysuch canbeignored.Pricemovementsdonot methodsasincreasingtheamountofcapitalfirms followthewell-manneredbellcurve hold. assumedbymodernfinance;theyfollow Iamalsoarguingthattheriskoffinancial amoreviolentcurvethatmakesthe problemsatFannieMaeand/orFreddieMacis investor’sridemuchbumpier.Asound notasremoteasitmightseem,becauseofthefat tradingstrategyorportfoliometric tailsofthedistributionofpricechangesinasset wouldbuildthiscold,hardfactintoits markets.Thesetwoobservations—enormous foundations. potentialcostsandaprobabilityoffailurehigher thancommonlyrealized—implythattherisksof RobertEnglecharacterizesreturnsinfinancial verylargeeventsmustbeidentifiedandcarefully marketsthisway:“Returnsarealmostunpredict- analyzedthroughextensive“stresstesting.”Then, able,theyhavesurprisinglylargenumbersof adequatecontrolsmustbeinstitutedtomitigate extremevalues,andboththeextremesandquiet theidentifiedrisks. periodsareclusteredintime.Thesefeaturesare ThisisexactlytheapproachthatMandelbrot oftendescribedasunpredictability,fattails,and andHudsonrecommend:“Sowhatistobedone? volatilityclustering”(Engle,2004,p.407). Forstarters,portfoliomanagerscanmorefre- quentlyresorttowhatiscalledstresstesting.It meanslettingacomputersimulateeverythingthat MANAGING INTEREST RATE RISK couldpossiblygowrong,andseeingifanyofthe InmyspeechtotheOFHEOconference possibleoutcomesaresounbearablethatyouwant almosttwoyearsago,Iemphasizedtheriskof torethinkthewholestrategy”(Mandelbrotand systemic,worldwidefinancialcrisisshouldeither Hudson,2004,p.267). FannieMaeorFreddieMacbecomeinsolvent. Bythiscriterion,incompletehedgingof Theargumentwasthesameasthatstatedso longer-termandlesslikelyinterestrateshocksis 5 FINANCIALMARKETS notanadequateriskmanagementstrategyfor casesseniorexecutiveshaveleftthefirmsand GSEs.Capitalratiosthatarenottestedagainst auditattestationshavebeenquestioned.Bothfirms extremeeventsdonotadequatelymitigatethe havebeenrequiredtorestateearningsforanumber interestrateriskfacedbysuchinstitutions. ofyears.InvestigationsbyOFHEO,theSEC,and theDepartmentofJusticeareongoing. Accountingproblemswerenotonmyradar LIQUIDITY RISK screenwhenIfirstbecameconcernedaboutGSE risk.Therecentrevelationsareanotherexample FannieMaeandFreddieMacmustrollover ofourinabilitytopredictshocksthatwillimpact roughly$30billionofmaturingshort-termobliga- ourfinancialsystem.EventhoughtheassetsF-F tionseveryweek.Atatimeofdisruptedfinancial holdarerelativelysimple—residentialrealestate markets,thecreditmarketsmightrefusetoaccept mortgagesandmortgage-backedsecurities—the theF-Fpaper.“FannieMaeandFreddieMac firmsthemselvesarecomplexorganizations recognizethisriskandbothfirmsindicatethey becauseoftheirscaleandthefinancialengineer- maintainsufficientliquiditytosurviveforsome ingtheyemploy.Theaccountingproblemsprovide time(3monthsorlonger)withoutaccessto anexampleofoperationalrisk;otheraspectsof rollovermarkets…[However,]theU.S.General F-Foperations,suchastheautomatedunderwrit- AccountingOffice(1998)hasalsopointedoutthat ingprocedures,arealsosubjecttooperational holdingsecuritiesintheirinvestmentportfolios risk.Itremainstobeseenhowtheaccounting forliquiditypurposesrepresentsahighlyprofit- restatementswillaffectthemarket’sviewofF-F ablearbitragefor[bothfirms],sincethereturnon earningsandcapitaladequacy.Clearly,though, theassetsexceedsthecostoftheagencybonds F-Fneedtoholdcapitalagainstoperationalrisk. usedtofundthepositions”(Jaffee,2003,p.16). Therefore,ifFannieMaeandFreddieMacare unabletosellnewdebt,thentheymayalsobe POLITICAL AND REGULATORY unabletocarryoutsalesofthe“liquid”securities fromtheirinvestmentportfolio. RISK Idiscussedliquidityriskatsomelengthina Fromanarrowmarketperspective,akeyissue speechlastspring(Poole,2004).Iwon’trepeat iswhetherthefederalgovernmentwouldbailout thatanalysis,butthebottomlineissimple:The FannieMaeand/orFreddieMacshouldthesol- FederalReservehasadequatepowerstoprevent vencyofeitherfirmbethreatened.Butthatistoo thespreadofaliquiditycrisis,butcannotpre- narrowaperspective,evenforaholderofF-F ventasolvencycrisisshouldFannieorFreddie obligations. exhausttheircapital.Intheeventofasolvency Iftherewereasolvencycrisis,theoutcome crisis,themarketwouldbecomeunreceptiveto wouldcertainlyinvolveextensivechangesinthe Fannieand/orFreddieobligations;theywould powersandcharacteristicsofthefirms.Institu- havedifficultyrollingovertheirmaturingdebt. tionsholdingF-Fobligations,directorguaranteed, Moreover,theiroutstandingobligationswould wouldmostlikelyhavetoaltertheirportfolio declineinpriceandtheirmarketswouldbecome practices.Moreover,evenifthefederalgovern- lessliquid.Beyondthat,itishardtosayexactly mentbailedoutF-F,theirobligationsmightbe whatelsemighthappen. redeemedeventuallybutceasetotradeactively inliquidmarkets.Finally,thereisofcourseno guaranteethatthefederalgovernmentwouldin OPERATIONAL RISK factbailoutF-F.Manyobservers,myselfincluded, Inthepasttwoyears,therehavebeensurpris- believethatabailoutwouldnotbeagoodidea. ingnewsreportsofaccountingirregularities,first Thebottomlineisthatthereissubstantial atFreddieandmorerecentlyatFannie.Inboth uncertaintyoverthefutureregulatorystructure 6 GSERisks thatwillapplytoFannieMaeandFreddieMac bethatthehighlyvolatileinterestrateenviron- andoverthelikelybehaviorofthegovernment mentoftheearly1980sisextremelyunlikelyto shouldthesolvencyofeitherfirmcomeinto recur,butIwouldliketoseeF-Fmaintaincapital question. positionsthatwouldenablethefirmstowithstand suchanenvironmentanyway. OnethingIthinkIknowforsureisthis:An CONCLUDING REMARKS investorwhoignorestherisksfacedbyFannie MaeandFreddieMacundertheassumptionthat Mypurposehasbeentoprovideanoutlineof afederalbailoutiscertainshouldtherebeaprob- alltherisksfacingFannieMaeandFreddieMac. lemismakingamistake. Therearesixriskstoconsider:creditrisk,pre- paymentrisk,interestrateriskfrommismatched durationofassetsandliabilities,liquidityrisk, REFERENCES operationalrisk,andpoliticalrisk.Muchmore couldbesaidabouteachoftheserisks,butI Engle,Robert.“RiskandVolatility:Econometric thoughtitwouldbeusefultodiscusseachof ModelsandFinancialPractice.”AmericanEconomic thembrieflyinordertohaveacompletecatalog. Review,June2004,94(3),pp.405-20. I’veparticularlyemphasizedtheimportance offacinguptotheimplicationsoflow-probability Jaffee,Dwight.“TheInterestRateRiskofFannieMae events.Alowprobabilitymustnotbetreatedas andFreddieMac.”JournalofFinancialResearch, 2003,24(1),pp.5-29. ifitwereazeroprobability.Moreover,extensive evidencefrommanydifferentfinancialmarkets, Mandelbrot,BenoitandHudson,RichardL.The reinforcedbysimilarfindingsincommodity (Mis)BehaviorofMarkets.NewYork:BasicBooks, markets,indicatesthatpricechangesinasset 2004. marketsarecharacterizedbyfattails.Theproba- bilityoflargepricechangesismuchhigherthan Poole,William.“HousingintheMacroeconomy.” suggestedbythefamiliarnormaldistribution.In FederalReserveBankofSt.LouisReview,May/June thecaseofthe10-yearTreasurybond,changes 2003,85(3),pp.1-8. of3.5standarddeviationsormoreare16times morefrequentthanexpectedunderthenormal Poole,William.“TheRisksoftheFederalHousing distribution. Enterprises’UncertainStatus.”PanelonGovernment Moregenerally,theprobabilityofshocksof SponsoredEnterprisesandTheirFuture.In manysortsmaybehigherthanonewouldthink. Proceedings:40thAnnualConferenceonBank Theaccountingproblemsthatsurfacedatboth StructureandCompetition,May2004,Federal FannieandFreddiewouldsurelyhavebeen ReserveBankofChicago,pp.464-69. assignedaverylowprobabilitytwoyearsago. Unlikethesituationinfinancialmarkets,where Posner,RichardA.“TheProbabilityofaCatastrophe...” awealthofdatapermitssomeformalprobability WallStreetJournal,January4,2005,p.A12. estimates,theprobabilityofothersortsofevents ismuchmoredifficulttojudge.Forthisreason, U.S.GeneralAccountingOffice.Government SponsoredEnterprises:FederalOversightNeedfor IbelievethatthecapitalheldbyF-Fshouldbe NonmortgageInvestments.GAO/GDD-98-48. ataleveldeterminedprimarilybythecushion Washington,DC:1998. requiredshouldanunlikelyeventoccurrather thanbyanestimateoftheprobabilityitself.Itmay 7

William Poole (2005, January 12). Speech. Speeches, Federal Reserve. https://whenthefedspeaks.com/doc/speech_20050113_poole

@misc{wtfs_speech_20050113_poole,
  author = {William Poole},
  title = {Speech},
  year = {2005},
  month = {Jan},
  howpublished = {Speeches, Federal Reserve},
  url = {https://whenthefedspeaks.com/doc/speech_20050113_poole},
  note = {Retrieved via When the Fed Speaks corpus}
}