DRAFT: COMMENTS INVITED

Financial Crisis Inquiry Commission

Preliminary Staff Report

CREDIT RATINGS AND THE FINANCIAL CRISIS June 2, 2010

ThispreliminarystaffreportissubmittedtotheFinancialCrisisInquiryCommission(FCIC) andthepublicforinformation,review,andcomment.Commentscanbesubmittedthrough theFCICswebsite,www.fcic.gov/contact. ThisdocumenthasnotbeenapprovedbytheCommission. ThereportprovidesbackgroundfactualinformationtotheCommissiononsubjectmatters thatarethefocusoftheFCICspublichearingonJune2,2010.Inparticular,this reportprovidesinformationoncreditratings.Staffwillprovideinvestigative findingsaswellasadditionalinformationonthesesubjectmatterstotheCommissionover thecourseoftheFCICstenure. DeadlineforComment:July14,2010

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Table of Contents I. Introduction ............................................................................................................................... 3 II. Background on Credit Ratings .................................................................................................. 4 A. A Brief History of the Credit Rating Industry .................................................................... 4 B. The Meaning of Credit Ratings........................................................................................... 4 C. The Use of Ratings in Regulation ....................................................................................... 5 1. Determining Capital Requirements............................................................................... 5 2. Restrictions on Asset Allocation ................................................................................... 6 3. Statutory References to Credit Ratings ......................................................................... 7 D. The Use of Ratings by Firms, Investors, and Other Private Entities .................................. 8 E. The Regulation of Rating Agencies .................................................................................... 9 III. The Rating of RMBS and CDOs ............................................................................................ 10 A. The Structure of Structured Products ................................................................................ 10 B. How Pooling and Tranching Can Create AAA Securities ................................................ 13 C. Rating Methodology ......................................................................................................... 14 1. RMBS ......................................................................................................................... 14 2. CDOs........................................................................................................................... 19 3. Monitoring RMBS and CDOs .................................................................................... 20 D. Market Share of Major Rating Agencies in RMBS and CDOs ........................................ 20 E. Volume of Rated RMBS and CDOs ................................................................................. 22 IV.Credit Ratings during the Financial Crisis .............................................................................. 24 A. Rise in Mortgage Defaults ................................................................................................ 24 B. Downgrades and Impairments of RMBS and CDOs ........................................................ 26 C. Downgrades of Other Financial Institutions ..................................................................... 39

Page 1 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT References ..................................................................................................................................... 42 Appendix 1: Moodys Ratings Idealized Expected Loss Rates ................................................... 43 Appendix 2: Sensitivity of Moodys RMBS Model to Pool Characteristics ................................ 44 Appendix 3: Moodys Model for Rating ABS CDOs ................................................................... 45 Appendix 4: Downgrades and Impairments of ABS CDOs ......................................................... 47

Page 2 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

I.

Introduction

Thepurposeofthispreliminarystaffreportistopresentbackgroundontherolethat creditratingagencies(RAs)mayhaveplayedinthefinancialcrisis.MostsubprimeandAlt Amortgageswereheldinresidentialmortgagebackedsecurities(RMBS),mostofwhich wereratedinvestmentgradebyoneormoreRA.Furthermore,collateralizeddebt obligations(CDOs),manyofwhichheldRMBS,werealsoratedbytheRAs.Between2000 and2007,Moodysrated$4.7trillioninRMBSand$736billioninCDOs.Thesharprisein mortgagedefaultsthatbeganin2006ultimatelyledtothemassdowngradingofRMBSand CDOs,manyofwhichsufferedprincipalimpairments.Lossestoinvestorsandwritedowns onthesesecuritiesplayedakeyroleintheresultingfinancialcrisis. InflatedinitialratingsonmortgagerelatedsecuritiesbytheRAsmayhavecontributedto thefinancialcrisisthroughanumberofchannels.First,inflatedratingsmayhaveenabled theissuanceofmoresubprimemortgagesandmortgagerelatedsecuritiesbyincreasing investordemandforRMBSandCDOs.IffewerofthesesecuritieshadbeenratedAAA, theremayhavebeenlessdemandforriskymortgagesinthefinancialsectorand consequentlyasmalleramountoriginated.Second,becauseregulatorycapital requirementsarebasedinpartontheratingsoffinancialinstitutionsassets,theseinflated ratingsmayhaveledtogreaterriskadjustedleverageinthefinancialsystem.Hadthe ratingsofmortgagerelatedsecuritiesnotbeeninflated,financialinstitutionswouldhave hadtoholdmorecapitalagainstthem.Onarelatedpoint,theratingsofmortgagerelated securitiesinfluencedwhichinstitutionsheldthem.Forexample,hadlesssubprimeRMBS beenratedAAA,pensionfundsanddepositoryinstitutionsmayhaveheldlessofthem. Finally,therapiddowngradingofRMBSandCDOsbeginninginJuly2007mayhave resultedinashocktofinancialinstitutionsthatledtosolvencyandliquidityproblems. Inaddition,downgradesofmonolinebondinsurerssuchasAmbacandMBIAandother providersofcreditprotectionsuchasAIGtriggeredcollateralcallsbuiltintoinsuranceand derivativecontracts,exacerbatingliquiditypressuresatthesealreadytroubledfirms.This ledtoratingsdowngradesofthesecuritiesthesefirmsinsured,promptingincreased capitalrequirementsatthefirmswhichheldthesesecuritiesandinthecaseofmoney marketmutualfundsonlypermittedtoholdhighlyratedassetssalesofassetsintoan alreadyunstablemarket. SectionIIofthereportprovidesgeneralbackgroundonthecreditratingindustry.Section IIIdescribeshowRMBSandCDOswererated.SectionIVdescribesthecollapseofcredit ratingsonRMBSandCDOsduringthefinancialcrisis.

Page 3 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

II.

BackgroundonCreditRatings
A. ABriefHistoryoftheCreditRatingIndustry

Creditratingagencies(RAs)haveexistedforwelloveracentury,andthemarkethasbeen dominatedbyafewlargeplayerssinceitsinception.Thefirstsecuritiesratingswere issuedin1909,whenJohnMoodypublishedabookofratingsforU.S.railroadbonds.He extendedhisratingstoutilitiesandindustrialbondsthefollowingyear.Additional entrantsfollowed,withPoorsPublishingCompanyissuingitsfirstratingsin1916, StandardStatisticsCompanysixyearslaterin1922,andFitchPublishingCompanysoon afterin1924.In1962,Dun&BradstreetboughtMoodys;nearly40yearslater,in2000,it spunMoodysoffasanindependentpubliccorporation.StandardandPoorsmergedin 1941,andthenwasacquiredbyMcGrawHillin1966.Fitchmergedwithanumberof smallerratingsagencies:IBCAin1997(throughthemergerwithIBCA,FitchRatings becameownedbyFimalacS.A.),Duff&Phelpsin2000,andThomsonBankWatchin2000. Thus,by2000,thethreemajorRAsremainingwereMoodys,Standard&Poors(S&P),and Fitch. Animportantchangetothestructureoftheindustrywastheevolutioninthe1970sfroma subscriberpaysmodel,inwhichbondinvestorspaytheagenciesforaccesstotheiranalysis andratings,toanissuerpaysmodel,inwhichthebondissuerschooseandpaytheRAsthat ratetheirbonds.

B.

TheMeaningofCreditRatings

Creditratingsprovideameasureofthecreditworthinessofdebtsecuritiestoinvestors. EachoftheRAsconsidersanumberoffactorstodetermineratings,includingfirmand securityspecificfactors,marketfactors,regulatoryandlegalfactors,andmacroeconomic trends.Theirratingsintendtoprovideameansofcomparisonofcreditriskacrossasset classesandtime. Theratingsfromdifferentagenciesmeasureslightlydifferentcreditriskcharacteristics. S&PandFitch,forexample,basetheirratingsontheprobabilityofdefault.Moodys,in contrast,basesitsratingsonexpectedloss,whichisequaltotheproductof(1)the probabilityofdefaultand(2)theproportionoftheinvestmentthatinvestorsonaverage loseintheeventofdefault.However,investorsandregulatorstendtoviewtheratingsof themajorRAsasroughlyequivalent.Table1showstheratingscalesofMoodys,S&P,and Fitchbycreditqualitycategory.Theratingsaredividedintotwocategories:bondsrated BBBandaboveareconsideredinvestmentgrade;bondsratedbelowBBBarespeculative

Page 4 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT grade(sometimesalsocalledjunk).WhenratingastructuredproductlikeanRMBSor CDO,theRAsestimatetheprobabilityofdefaultorexpectedlossofthebondandcompare ittobenchmarksforeachoftheirratings.Thefiveyearexpectedlossbenchmarksfor MoodysratingsareinAppendix1.Theexpectedlossoverfiveyearsonabondthat MoodysratesAaaislessthan0.0016%,whichfora$1millionbondamountsto$16.Fora bondratedBaa1itislessthan0.605%,whichfora$1millionbondamountsto$6,050. Table1:MeaningofCreditRatings
CreditQuality Highestcreditquality Highcreditquality Strongpaymentcapacity Adequatepaymentcapacity Possibilityofcreditrisk Significantcreditrisk Highcreditrisk Defaultislikely/imminent Indefault CreditRatingAgency Moodys S&P Investmentgrade Aaa AAA Aa1toAa3 AA+toAA A1toA3 A+toA Baa1toBaa3 BBB+toBBB Speculativegrade Ba1toBa3 BB+toBB B1toB3 B+toB Caa1toCaa3 CCC+toCCC Ca CC,C C SD,D Fitch AAA AA A BBB BB B CCC CC,C D

C.

TheUseofRatingsinRegulation

TheRAsratingsofbondsandentitieshavebeenusedinfinancialregulationsince1931, whentheOfficeoftheComptrolleroftheCurrency(OCC)requiredthatbankholdingsof publiclytradedbondshavearatingofBBBorbetterinordertobeallowedtobecarriedat bookvalue.PubliclytradedbondswithratingsbelowBBBwouldhavetobevaluedonthe banksbalancesheetatadiscount.Thetwoprimaryusesofcreditratingsinfinancial regulationarein(1)determiningcapitalrequirementsoffinancialinstitutionsand(2) restrictingfinancialinstitutionsassetallocations. 1. DeterminingCapitalRequirements

CreditratingshavebeenusedasadeterminantofcapitalrequirementsintheUnitedStates since1951,whentheNationalAssociationofInsuranceCommissionersbeganimposing highercapitalrequirementsonlowerratedbondsheldbyinsurers.However,the watershedeventintheuseofindependentcreditratingsinfederalregulationoccurredin

Page 5 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT 1975,whentheSecuritiesandExchangeCommission(SEC)modifieditsminimumcapital requirementsforbrokerdealerstotakeintoaccounttheriskinessoftheirportfoliosand baseditsassessmentofriskinessonbondratings.Fearingthatbrokerdealerdemandfor highbondratingswouldleadtoinflatedcreditratingsfromtheRAs,theSECcreatedanew designation,thenationallyrecognizedstatisticalratingorganization(NRSRO),andonly recognizedthebondratingsofRAsdesignatedasNRSROs.TheSECinitiallyrecognized Moodys,S&P,andFitchasNRSROs,butadditionalfirmshavebeenrecognizedovertime. Creditratingshavemorerecentlyplayedaroleintheregulatorycapitaldeterminationsof U.S.bankandthriftregulatorsaswell.Regulatorspermitbanksandthriftstouseratingsas anelementintheirinternalassessmentsofthecreditqualityoftheassetstheyhold. Further,ratingsareusedasapartoftheregulatorycapitalcalculationsforcertainclasses ofassets.In2001,U.S.bankandthriftregulatorspassedanewcapitalrulegoverningasset securitizationondepositoryinstitutions.Amongotherprovisions,thenewrulesetrisk weightsforRMBSandotherassetbackedsecuritiesbasedontheratingofthebond.Risk weightsforbondsratedAAAorAAwerebelowweightsforlowerratedbonds.Institutions wererequiredtoholdcapitalinproportiontotheriskweights.Forexample,relativetothe requiredcapitalforAAAorAAsecurities,BBBsecuritiesrequiredfivetimesgreatercapital andBBsecuritiesrequiredtentimesgreatercapital. Internationally,creditratingshaveevengreaterforceindeterminingcapitaladequacy. TheBaselIIstandards,aninternationalattempttostandardizecapitalrequirementsthat followedthe1988BaselIstandards,allowbankstorelyonexternalcreditratingsto determinetheriskweightsforthecapitalrequirementsassociatedwithvariousexposures. TheEuropeanUnionsCapitalRequirementsDirective,adoptedin2006,introducedthe riskweightedcapitalrequirementsinBaselIItofinancialinstitutionswithintheEU.For U.S.firms,BaselIIhadnotbeenimplementedwhenthefinancialcrisisbegan. 2. RestrictionsonAssetAllocation

AnumberofU.S.regulationsusecreditratingstodeterminethepermissibilityofcertain classesofinvestments.Forexample,theSECmakestheuseofratedsecuritiesattractiveby restrictingmoneymarketmutualfundstosecuritiesthathavereceivedcreditratingsfrom anytwoNRSROsinoneofthetwohighestshorttermratingcategoriesorcomparable unratedsecurities.Similarly,in1989thefederalgovernmentrelaxedaDepartmentof Laborrulerestrictingpensionfundsfrominvestinginassetbackedsecuritiesincluding RMBStoallowthemtoinvestinsecuritiesratedAorhigher.Additionally,bank regulatorsrestrictpermissibleinvestmentsecuritiesbynationalbanksusingcreditrating cutoffs.Forexample,underOCCregulationsoninvestmentsecurities,anationalbankmay

Page 6 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT buyandsellforitsownaccountinvestmentcompanysharesthatmeetotherrequirements providedthatthesharesareratedinvestmentgradeorthecreditequivalentofinvestment grade. 3. StatutoryReferencestoCreditRatings

Whilemuchoftheuseofcreditratingsinfinancialregulationhasoccurredviaregulatory actionsbyfederalagencies,Congresshasexplicitlyprescribedtheuseofratingsinsome statutes.Primarily,creditratingshavebeenusedinordertodefinetermsinlegislation. Forexample,P.L.98440,SecondaryMortgageMarketEnhancementAct,whichpermitted federallycharteredfinancialinstitutionstoinvestinmortgagerelatedsecurities,included inthedefinitionofmortgagerelatedsecuritiesarequirementthatthesecurityberatedin oneofthetwohighestcategoriesbyanNRSRO.Similarratingsdependentdefinitionshave beenusedinlegislationtodefineallowableinvestmentsbyfederallycharteredfinancial institutionsinsmallbusinessrelatedsecurities,1FederalHousingAdministration eligibilitytoenterintopartnershipsorothercontractualarrangementsincluding reinsuranceandrisksharingagreementswithaqualifiedhousingfinanceagency,2and todelineatetheexemptionofcertaincompaniesfromprovisionsoftheInvestment CompanyActof1940.3Further,P.L.106102,GrammLeachBlileyActof1999,inaddition tootherreferencestoratingsagencies,gaveauthorizationtoconductinsubsidiaries certainactivitiesthatarefinancialinnaturetoqualifyingbankswhich,amongother requirements,cannothavefewerthan1issueofoutstandingeligibledebtthatiscurrently ratedwithinthe3highestinvestmentgraderatingcategoriesbyanationallyrecognized statisticalratingorganization. Incertaincases,legislationhasincorporatedcreditratingsinotherways.Forexample,the FederalDepositInsuranceActwasamendedin1989inP.L.10173,FinancialInstitutions Reform,Recovery,andEnforcementAct,todisallowsavingsassociationsfromacquiring orretaininganycorporatedebtsecuritynotofinvestmentgrade,anddefinesinvestment gradeasratedinoneofthe4highestratingcategoriesbyatleastonenationally recognizedstatisticalratingorganization.P.L.109171,FederalDepositInsuranceReform
Rated in 1 of the 4 highest rating categories by at least 1 nationally recognized statistical rating organization; P.L. 103-325: Riegle Community Development and Regulatory Improvement Act of 1994. 2 carries the designation of `top tier' or its equivalent, as evaluated by Standard & Poors or any other nationally recognized rating agency; receives a rating of `A' for its general obligation bonds from a nationally recognized rating agency; or otherwise demonstrates its capacity as a sound and experienced agency based on, but not limited to, its experience in financing multifamily housing, fund balances, administrative capabilities, investment policy, internal controls and financial management, portfolio quality, and State or local support; P.L. 102-550: Housing and Community Development Act of 1992. 3 any debt security that is rated investment grade by not less than 1 nationally recognized statistical rating organization; 15 USC 80a-6(a)(5)(A)(iv)(I))
1

Page 7 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Actof2005,mandatedthattheFDICuseratingsasonesourceofinformationforestimating riskofloss.Finally,asapartoftheSarbanesOxleyActof2002,theSecuritiesand ExchangeCommissionwasrequiredtoconductastudyoftheroleofcreditratingagencies intheoperationofthesecuritiesmarkets.Thestudyidentifiedanumberofissuesfor futureexamination,includingdisclosureofratingsprocesses,potentialconflictsofinterest, anticompetitivepractices,regulatorybarrierstoentry,andtheneedforadditionalSEC oversight.

D.

TheUseofRatingsbyFirms,Investors,andOtherPrivateEntities

Ratingagenciesinfluenceeconomicactivitythroughmarketchannelsaswell.Most prominently,manyprivatecontractsrelyuponcreditratingstoprotectcreditors.Ratings triggers,forinstance,canmandatethatadebtorpostcollateral,orgivecreditorstheright todemandimmediaterepaymentofdebtsinfull,intheeventofadowngrade.4InJanuary 2008suchtriggersmademonolinebondinsurerMBIAliablefor$2.9billionintermination paymentsand$4.5billionincollateralandleftmanyofthesecuritiesithadinsured vulnerabletodowngrades.5A2004surveybytheAssociationforFinancialProfessionals revealedthat87percentofrespondingorganizations,whicharemostlylargecorporations, withoutstandingdebthadbeenrequiredtomaintainaspecifiedratingfromatleastoneof thefourNRSROsinexistenceatthetime.6 Ratingsalsofigureintothedecisionsofprivateandpublicentitiestoextendcreditor purchasesecurities.Manyinstitutionalinvestors,suchaspensionfundsoruniversity endowments,donothavetheresourcestoevaluateallofthesecuritiestheypurchase, whichinanycasewouldbeduplicativeoftheagencieswork.Inaddition,theseinvestors donothaveaccesstothesameinformationthattheratingsagenciesdo.Consequently, theyusecreditratingsasasubstituteforthemoregranularinformationtheywould otherwisehavetogather.AsformerMoodysManagingDirectorJeromeFonshas acknowledged,subprimeRMBSandtheiroffshootsofferlittletransparencyaroundthe compositionandcharacteristicsoftheloancollateral.Loanbyloandata,thehighest levelofdetail,isgenerallynotavailabletoinvestors.Methodsofcreditanalysis,headded,

4SecuritiesandExchangeCommission,ReportontheRoleandFunctionofCreditRatingAgenciesinthe OperationoftheSecuritiesMarkets(Jan.2003),29. 5MBIACommentsontheImpactoftheMoodysDowngradeofMBIAsInsuranceFinancialStrengthRatingto A2onitsAsset/LiabilityManagementBusiness,pressrelease(June20,2008). 6AssociationforFinancialProfessionals,2004CreditRatingAgencySurvey:ReportofSurveyResults(Oct. 2004).

Page 8 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT arequitetechnical,oftenrelyingonadvancedstatisticaltechniquesandtherefore beyondthegraspofmanyinvestors.7 However,notallinvestorsandcreditdefaultswapprotectionwritersreliedexclusively uponratings.AIG,forexample,usedinternalmodelstocalculatetheexposureitwould agreetoassumeonthesuperseniorcreditdefaultswapsitwrote.8However,provisionsin theunderlyingcontractsdidstipulatethatratingsdowngradeswouldtriggercollateral calls,whichwereaproximatecauseofthefirmsproblemsthatledtogovernment intervention.

E.

TheRegulationofRatingAgencies

TheprimaryregulationofRAsoccursviarecognitionasanNRSRObytheSEC.In1975, whentheSECbeganrelyingoncreditratingstodeterminethecapitaladequacyofbroker dealers,itrecognizedS&P,Moodys,andFitchasNRSROsandlaterrecognizedfour additionalNRSROsbetween1982and1991(Duff&Phelps,McCarthy,Crisanti&Maffei, IBCA,andThomsonBankWatch).TheprocedureforapprovinganewNRSROwasnot explicitlydefinedinstatute.Theprincipalrequirementwasthattheagencybenationally recognizedbythepredominantusersofratingsintheUnitedStatesasanissuerofcredible andreliableratings.FollowingtherequestofanRAtobecomeanNRSRO,theSEC investigatedtheRAwithregardstointernalprocesses,financialresources,and organizationalstructure.IfanRAwasapproved,theSECissuedanoactionletterstating thatitwillnotrecommendenforcementactiontotheCommissionifratingsfromthe ratingagencyareconsideredbyregisteredbrokerdealerstoberatingsfromanNRSROfor purposesofapplyingtherelevantportionsofthenetcapitalrule.9 Followingtheexpansionoftheuseofcreditratingsinregulation,theSECbeganareviewof thepotentialneedforgreaterregulatoryoversightbeginninginaConceptReleaseissuedin 1994,whichsolicitedpubliccommentontheCommissionsuseofNRSROratings. FollowingoveradecadeofdiscussiononRAreform,CongresspassedCreditRatingAgency ReformActof2006,P.L.109291,whichdefinedanNRSROandofferedmoreconcrete directionregardingtherecognitionprocessofNRSROsbytheSEC.Introducingthefirst oversightormonitoringofNRSROs,thenewapplicationprocessrequired,amongother things,ananalysisofthehistoricalperformanceofcreditratingsbytheapplicant,rating
7QuotedinOhioPolice&FirePensionFundetal.v.Standard&Poorsetal.(USDistrictCt.,SouthernDistrictof 8AmericanInternationalGroupInvestorMeeting,ThomsonStreetEventstranscript(Dec.5,2007),11. 9

Ohio,2009),caseno.2:09cv1054.

U.S. Securities and Exchange Commission (2003) Report on the Role and Function of Credit Rating Agencies in the Operation of the Securities Markets: as required by Section 702(b) on the Sarbanes-Oxley Act of 2002.

Page 9 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT proceduresandmethodologies,policiesregardingpotentialmisuseofmaterial, organizationalstructure,potentialconflictsofinterest,andalistofthetwentylargest issuersandsubscribersthatuseitsratingsbyrevenueintheyearprecedingtheapplication date.ThelawtookeffectinJune2007,andexistingNRSROswererecertifiedthroughthe applicationprocess.However,thelawfocusedonmandatorydisclosureofratingagencies methodsandexpresslyprohibitedtheSECfromregulatingthesubstanceofthecredit ratingsortheproceduresandmethodologiesbywhichanyNRSROdeterminescredit ratings.10 Overtheyears,ratingsagencieshaveacquiredstrongdefensesagainstbeingheldliablefor erroneousratingsthroughprivatelitigation.Mostprominently,Rule436(g)(1)ofthe SecuritiesActof1933explicitlyexemptsthemfromliabilityformisstatementsmadein connectionwithsecuritiesregistrationstatements.MoodysandS&Phavealsosucceeded inusingtheFirstAmendmentasashieldagainstliabilityandinquashingsubpoenaswith theargumentthattheyareprovidingjournalisticservicesprotectedbytheFirst Amendmentorreportersprivilege.11Moreover,courtcaseshaveacceptedtheNRSROs disclaimersthattheirratingsaremerelyopinions,withonejudgegoingsofarastorule in1999thataninvestorsrelianceonthemwasnotreasonabledespitethefactthat regulatorsrelieduponthesesameratings.12

III.

TheRatingofRMBSandCDOs
A. TheStructureofStructuredProducts

RMBSandCDOsaretypesofstructuredproducts.Generally,structuredproductsinvolve thepoolingofassetsintoaspecialpurposevehicle(SPV)andthetranchingofthebonds issuedbytheSPVintoclassesofsecuritieswithparticularpaymentrights.Thecashflows oftheSPVsassetsareusedtomakethepromisedpaymentstotheSPVsbondholders.13 AcrucialgoalofthetranchedcapitalstructureoftheSPVistocreatesomebondsthatare deemedlowriskandcanreceiveinvestmentgraderatingsfromtheratingagencies.One keytoolusedtoachievethisissubordination.TheclassesofsecuritiesissuedbytheSPV areorderedaccordingtotheirpriorityinreceivingdistributionsfromtheSPV.The structureissetuptooperatelikeawaterfall,withtheholdersofthemoreseniortranches beingpaidpriortothemorejunior(orsubordinate)tranches.Themostseniorsetof
10

15 U.S.C.78-o7(c)(2),

11See,forexample,InrePanAmCorp.,161BR577(SDNY,1993). 12Quinnv.McGrawHill,168F3rd331(7thCir.,1999). 13RMBSandCDOsarediscussedinmoredetailintheFCICsApril7,2010,PreliminaryStaffReport,

SecuritizationandtheFinancialCrisis.

Page 10 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT tranchesreferredtosimplyasseniorsecuritiesrepresentthelowestriskand consequentlypaythelowestinterestrate.Theyaresetuptobepaidpriortoanyofthe classesbelowandaretypicallyratedAAA.Seniorsecuritiestypicallymakeupthemajority ofbondsissuedbytheSPV.Thenextmostseniortranchesarethemezzaninetranches. Thesecarryhigherriskandpayacorrespondinglyhigherinterestrate.Themostjunior trancheinthestructureiscalledtheequityorresidualtrancheandreceiveswhatevercash flowisleftoverafterallothertrancheshavebeenpaid.Thesetranches,whicharetypically notrated,sufferthefirstlossesonanydefaultsofassetsinthecollateralpool. Figure1providesanotionalbalancesheetforatypicalRMBS.Theentityholdsapoolof residentialmortgagesthatmakeprincipalandinterestpayments.InthecaseofanRMBS, theassetsareresidentialmortgages.

Figure1:BalanceSheetofanRMBS
Assets
first Principal and interest payments next claim

Liabilities AAAseniorbonds:80% AandAAbonds:15% MezzanineBBBbonds:3% BBBondsandEquitytranche:2%

Mortgages

next claim

last

ACDOcanholdapoolofcollateralthatincludesmanytypesofassets.CDOsthatcontain RMBSorothertypesofassetbackedsecuritiesarecalledABSCDOs.ABSCDOsthatlargely containthemezzaninetranchesofRMBSarereferredtoasmezzanineCDOs;those containinghigherratedRMBSarecalledhighgradeCDOs. Figure2providesanotionalbalancesheetforatypicalmezzanineABSCDO.Inthis example,theAAAseniorbondsmakeup76percentoftheprincipalamountofdebtissued bytheSPV,AAAbondsaccountfor14percent,mezzanineBBBbondsmakeup6percent, andtheresidualtrancheamountsto4percent.

Page 11 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

Figure2:ConstructionofamezzanineABSCDO
Mortgages

RMBS

AAA AA BBB Equity

CDO Balance Sheet

Assets Principal and interest Liabilities

Mortgages

AAA senior bonds: 76%

Mortgages

AAA AA BBB Equity

Pool of

BBBrated RMBS

A - AA bonds: 14%

Mortgages

Mezzanine BBB bonds: 6%

AAA AA BBB Equity

Equity tranche: 4%

TheCDOsdescribedabovethatholdactualbondsascollateralarereferredtoascashflow CDOs.AnothertypeofCDO,calledasyntheticCDO,isbasedoncreditdefaultswaps(CDS) ratherthanactualbonds.ACDSisaderivativecontractinwhichthebuyerpaysastream ofperiodicpaymentstotheseller,andinreturnthesellerhastopayoutsomeamountif somereferencesecuritydefaults.TheseCDScontractsarestructuredsothatthesynthetic CDObondsroughlymimicthepaymentsofacashflowCDOthatholdscashpositionsin thosesameRMBS.SyntheticCDOsthatarebasedonRMBS,theinvestorsintheCDO effectivelyactasthesellerofcreditdefaultswapsonapoolofRMBSasreferencesecurities tosomesetofcounterparties.HybridCDOsalsoexistthatarebasedinpartonholding bondsascollateralandinpartonCDS.14

14

FCIC staff will release in the near future a preliminary staff report on derivatives that will describe synthetic CDOs in greater detail.

Page 12 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

B.

HowPoolingandTranchingCanCreateAAASecurities

Thegoalofthepoolingofassetsandtranchingandsubordinationofliabilitiesinstructured productsistotransformrelativelyriskycollateralassets(e.g.,mortgagesorBBBrated mezzanineRMBStranches)intoasetofbondsthatincludesverylowriskbonds.Acrucial assumptionforthistobepossibleisthattheassetsheldbytheSPVarenotperfectly correlatedsothattherearebenefitstodiversification. Toseethis,considerthefollowingsimpleexample.15SupposeaCDOholdstwoidentical$1 bondsasassets.Ifabonddoesnotdefault,itpays$1,butifitdoesdefaultitpaysnothing. Supposeeachbonddefaultswithprobabilityequaltoonehalf(0.5).Furthermore,suppose theCDOissuesbothajuniorandseniortrancheofbonds,eachofwhichpay$1.Thesenior tranchegetsthefirstclaimonthecashflowsoftheCDOsothatitonlydefaultsifbothofthe bondsheldascollateraldefault.Incontrast,thejuniortranchesuffersthefirstlossand defaultsifeitherofthecollateralbondsdefaults. Toseetheimportanceofcorrelation,considerfirstthecaseinwhichthetwobondsheldas collateralareperfectlycorrelated.Thismeansthatthetwobondseitherbothdefaultor bothpayoutitisnotpossibleforonetodefaultandtheothernottodefault.Thismeans thatwithprobabilityonehalftheCDOhasnomoneytopayeitherofitsbonds,andwith probabilityonehalftheCDOcanpayboththejuniorandseniorbonds.Withperfect correlationinitsassetsdefaults,then,theCDOhasnottransformeditscollateralassets intoanylowerriskbonds.Boththebondsheldascollateralandthebondsissuedbythe CDOallhaveaprobabilityofdefaultequaltoonehalf. Considernowthealternativeextremeassumptioninwhichthebondsheldascollateralare uncorrelated.Thismeansthatwhetheronebonddefaultsdoesnotprovideany informationaboutwhethertheotherbonddefaults.Underthisassumption,therearethree possibleamountsofcashpaidoutbytheCDOscollateral:$0,$1,and$2.TheCDOssenior tranchewillonlydefaultiftheCDOreceives$0,whichoccurswhenbothbondsdefault. Becausetheyareuncorrelated,bothbondsdefaultwithprobability0.5x0.5=0.25.The CDOreceives$1ifjustoneofthebondsdefault,whichoccurswithprobability0.5.Inthis scenariotheseniortrancheispaidbutthejuniortranchedefaults.Finally,theCDO receives$2ifneitherofthebondsdefault,whichoccurswithprobability0.25.Thisisthe onlyscenarioinwhichthejuniortranchegetspaid,sothejuniortranchedefaultswith probability0.75.

15ThisexampledrawsonCoval,etal(2009).

Page 13 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Thesetwocasesrepresenttheextremesofperfectcorrelationandzerocorrelation.More realistically,collateralassetsaretypicallyimperfectlycorrelated.Thismeansthatknowing thatoneofthebondsdefaultsmakesitmorelikely(butnotcertain)thattheotherbond willdefault.Commonfactorsthatinfluencetheperformanceofbondsareonesourceof correlation.Forexample,whenhousingpricesgodown,itcanaffecttheperformanceof manyRMBS. Withsuchimperfectcorrelation,poolingandtranchingcanproduceaseniortrancheof securitiesthatarelowerriskthantheCDOsunderlyingassets,asinthecaseofzero correlationconsideredabove.Thehigherthecorrelationoftheassetsdefaults,however, thelessthepoolingandtranchingcanreducethedefaultriskofthestructuressenior tranche.Intuitively,theseniorCDObondwilldefaultonlyiftheCDOscollateralpool experiencesmassivelosses.Awelldiversified(i.e.,lowcorrelation)collateralportfoliois unlikelytosustainsuchlargelosses.Butahighlyconcentrated(i.e.,highcorrelation) collateralportfoliohasahigherprobabilityofrealizingverylargelossesbecauseifone assetdefaults,manyotherassetsarealsomorelikelytodefault.

C.

RatingMethodology

WedescribenowthemethodologyusedtorateRMBSandCDOsaccordingtotherating agenciespubliclyavailabledocumentation.Thesefinancialinstrumentsarecomplex,and theratingagenciesmethodologiesarecomplex.WhatfollowsistheFCICstaffsattemptto distilltheessenceofthesemethodologies,whichassumessomefamiliaritywithstatistics andomitsmanydetails. 1. RMBS

TheRAsmethodologyforratinganRMBSbeginswithanassessmentoftheriskinessofthe mortgagesheldascollateralintheRMBS.HerewedescribeMoodysapproachtorating subprimeRMBSfromroughly1996throughtheendof2006asdescribedinMoodyspublic documentation.16 Theanalysiswouldbeginwithanevaluationofthecreditriskofthemortgagesinthe RMBSscollateralpool.Moodysusedasetofquantitativerelationshipsthatrelate mortgagecharacteristicstodefaultprobabilityanddefaultseverity(i.e.,thelossonthe mortgagegiventhatitdefaults)tocalculateanexpected(i.e.,mean)poollossestimate, whichisthetotallossexpectedtooccuroverthelifetimeofthepoolasafractionofthe originaltotalprincipalbalanceofthepool.AkeyadjustmentthatMoodysmadetoits
16

Moodys (1996) lays out the basic analytic framework.

Page 14 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT expectedpoollossestimatewasbasedontheoriginatoroftheloans.Lossestimatesfor poolsofmortgagesmadebyoriginatorsthatMoodysdeemedtobelowerriskwere adjusteddownward.Fororiginatorsdeemedhighrisk,lossestimateswereadjusted upward. Becausethepoollossisuncertain,Moodysalsoassumedsomevariationinthepoolloss expressedasastandarddeviationandadjustedthatstandarddeviationbasedonthe numberofloansinthepool,theirgeographicdiversification,andotherfactors.Thepool lossdistributionwasthenmodeledasalognormaldistribution(anexampleofwhichis shownbelowinFigure3)withthisestimatedmeanandstandarddeviation. Table2belowprovidesalistoftheprincipalpoolcharacteristicsthatMoodysconsidered inestimatingthemeanpoolloss.Appendix2showsthesensitivityofmeanpoollossto changesinvariouspoolcharacteristicsusingMoodysmodel.Forexample,increasing averageLTVinthehypotheticalpoolAby10pointsincreasesmeanpoollossfrom0.82% to1.39%;forpoolBitincreasesmeanpoollossfrom2.96%to4.76%. Table2:PrincipalFactorsConsideredbyMoodysSubprimeRMBSModelasof2004 Factor Loantovalueratio(LTV) Combinedloantovalueratio (CLTV) Creditscore Debttoincomeratio(DTI) Originatorquality Effect HigherLTVincreasesdefaultprobabilityanddefault severity. HigherCLTVincreasesdefaultprobabilityanddefault severity. Highercreditscorelowersdefaultprobability. HigherDTIincreasesdefaultprobability. Expectedlossestimatesaremultipliedbyanoriginator factor,withhighqualityoriginatorsgettingafactorless thanone,andlowqualityoriginatorsgettingafactor greaterthanone. Loansthatwereoriginatedalongertimepriortothe RMBSbeingissuedaregenerallyconsideredtohavelower probabilityofdefaultanddefaultseverityduetothe accumulationofequity.Moodysfactorsinhousingprice depreciationonseasonedloansaswell,whichcan increasetheirdefaultprobabilityanddefaultseverity.

Loanseasoning

Page 15 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Timetoforeclose Localhousingmarket projections Propertytype Housevaluerelativetolocal market ARMvs.FRM Balloonmortgages Interestonlyandnegatively amortizingloans Loanpurpose Owneroccupancy Mortgageinsurance Coupon Poolsize Geographicdiversification Mortgageservicer Moodysincorporatesvariationacrossstatesinthetimeto foreclose,whichaffectsdefaultseverity. Moodysestimateofahousingmarketsprospectsare representedbyafactorthatraisesorlowersaloans defaultprobabilityanddefaultseverity. Attachedhousingandmultifamilypropertiesareassigned ahigherdefaultseverity. Thehigherapropertysrelativeprice,thehigherthe assigneddefaultseverityanddefaultprobability. ThedefaultprobabilityofARMsisadjustedupward. Defaultprobabilityisadjustedupward. Defaultprobabilityisadjustedupward. Homepurchasesandrefinanceloansareassignedlower defaultprobabilitythanhomeequityloans. Defaultprobabilityisadjusteddownward. Presenceofmortgageinsurancereducesdefaultseverity. Higherinterestrateloanshavehigherexpectedloss. Largerpoolshavelowerstandarddeviation. Poolswithgreatergeographicdiversificationhavelower standarddeviation. Moodysratesservicerqualityincollectingpaymentsand effectivenessinmodifyingdelinquentloansandadjusts thecreditenhancementsneededforagivenratingslevel. Moodysadjustsupwardthecreditenhancementrequired forloansthatcontainpartialornodocumentationofthe borrowersincomeorassets. Secondliensareassumedtohavehigherexpectedloss thanfirstliens.

Documentationstatus

Lientype

Sources:Moodys(1996),Moodys(2001),Moodys(2004a),Moodys(2004b).

Page 16 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Moodysexplainsthedatausedtocreatetheirpoollossmodelasfollows: Wedrawonhistoricaldatafrommanysources,beyondrated loanpools,todeterminetherelationshipbetweenloan characteristicsandcreditrisk.Keydatasourcesarethe MortgageBankersAssociation;FannieMaeandFreddieMac; privatemortgageinsurers;andadatabaseofratedloanpools trackedbyMoodys.Defaultfrequencyandlossseveritydata werecollectedattheindividualloanlevelaswellasatthepool level.17 Figure3

Source:Moodys(1996). Figure3illustratesalognormalprobabilitydensityfunctionforthelossesincurredina poolofloansinahypotheticalexample.Thehorizontalaxisindicatesthetotaldollar amountoflossesincurredbythepool.(Forexample,CSandXCSindicatedifferentdollar amounts.)Thedensityfunctionindicatedbythecurveillustratestheprobabilityofeach dollaramountofloss.Foragiventranche,thecreditsupportlevelisthedollaramountof lossesthatcanbesustainedbeforetheparticulartranchebeingratedbeginstoexperience adeclineinpayments.AhypotheticallevelforonetrancheisindicatedbyCSinFigure3. Forexample,amorejuniortranchebeginstoexperienceadeclineinpaymentsafter

17Moodys(1996).

Page 17 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT relativelyminorlossesintheoverallpool.Hence,thecreditsupportlevel,orCS,forthe juniortranchewouldbetotheleftofthecreditsupportlevelofthemoreseniortranche. IfthepoolofmortgagessustainslossesgreaterthanCS,thenthereareunsupportedlosses equaltothedifferencebetweenthetotallossestotheoverallpoolandthecreditsupport levelforthetranche.Forexample,ifapoolexperiencesalossof$3millionduetodefaults, andacreditsupportlevelforaparticulartrancheis$1million,theunsupportedlossfor thistrancheis$2million.Moodysthencalculatestheprobabilityofeachpossiblescenario oflossestoestimatetheexpectedlosstothetranche.Theexpectedlosstothetrancheis equaltothesumacrossalllossoutcomesgreaterthanCSoftheproductoftheunsupported lossesandtheprobabilitythatthelossoutcomeoccurs. Finally,Moodysassignsaratingtoeachtranche.Itdoesthisbyfirstcalculatingtheeffect ontheyieldofthebondfromtheseexpectedlosses.Inasense,theexpectedloss discountstheexpectedpaymentsonthesecurity.Moodyscomparesthisdiscounttoa standardizedscheduleofratingsanddiscountstoassignratingstothetranches. In2003,MoodysadoptedanewmodelforratingprimejumboandAltAmortgagepools calledMoodysMortgageMetrics(M3),whichincorporatedlossesassociatedwith macroeconomicfactors.Inpublicdocumentation,Moodysdescribedtheadoptionofthe model.WhilethepublicdocumentsarenotcompletelyclearonMoodysmethodologyand whatpreciselychangedovertime,in2003MoodysstatedthatithasrefineditsRMBS modelfor[a]setoflarge,geographicallydiverseJumboAandAltApoolsfromestablished originatorsandratedservicerstothepointofdelegatingthebulkofthedeterminationof thesecreditsupportlevelstothemodel.18Itwasnotadoptedatthattimeforuseon subprimepools.ItwasbasedonadatasetfromLoanPerformance,Inc.,containingover 500,000mortgages.TheM3modelsimulatestheperformanceofeachloaninthepool through1,250differenteconomicscenarios.Theseeconomicscenariosaregenerated usingMoodysprojectionsofinterestrates,statelevelunemploymentscenarios,andstate levelhomepriceappreciationscenarios.Inthesimulation,homepricesbroadlytrend upwardatapproximately4%peryearonaverage.19Moodysusestheresulting distributionofpoollossestoestimatetheexpectedlosstoeachtrancheandassignits rating. AsimilarmodelwasthendevelopedforusewithsubprimepoolscalledM3Subprime.Its usewasphasedintoMoodysratingsprocessinlate2006.

18Moodys(2003). 19Moodys(2003p.10).

Page 18 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Inadditiontothequantitativeanalysisdescribedabove,Moodyslegalanalystsconsidered whethertherewereanylegalissuesthatwouldaffectthebondspayments.20 Whentheinitialquantitativeandqualitativeanalysiswascomplete,theleadanalystonthe dealconvenedaratingcommitteecomposedofotheranalystsandmanagerstodetermine theratingsfortheRMBSsbonds.Theleadanalystpresentedanoverviewofthe transactionandMoodysquantitativeandqualitativeanalysesofthetransaction,andafter deliberationandpotentialadjustments,thecommitteevotedonMoodysratingsforthe bonds. 2. CDOs

TheRAsrateABSCDOsusinginformationabouttheratingandtypeofeachbondheldin thecollateralpooltoestimatetheprobabilitydistributionoflossestothecollateral.Here wedescribeMoodysapproach.Themainingredientsusedtoestimatethecollateralpool lossdistributionarethedefaultcorrelationamongthecollateralbondsandthecollateral bondsdefaultratesandrecoveryrates.Thedefaultcorrelationisthedegreetowhichthe defaultofoneofthecollateralbondsimpliesahigherprobabilityofdefaultoftheother bondsthattheCDOholdsascollateralassetsitsimportanceisdescribedinsectionIII.B. above.Theratingagenciesuseamixofassumptionsandhistoricaldatatodeterminethe defaultcorrelationbetweendifferenttypesofassets.Defaultratesandrecoveryratesare setbasedontheratingsofthecollateralbonds. OnceMoodyshasestimatedacollaterallossdistribution,itthencalculatestheexpected lossestoeachoftheCDOstranches. Thefinalstepinthequantitativeanalysisistocomparetheexpectedlossforeachtranche toasetofbenchmarksinordertodeterminethemodeledratingofeachtranche.Appendix 2providesthefiveyearidealizedexpectedlosspercentagesforeachofMoodysratings. Themodeledratingforeachtrancheisthehighestratingforwhichitsexpectedlossover theappropriatetimehorizonislessthantheidealizedexpectedlossforthatrating. Inadditiontothisquantitativeanalysis,Moodyslegalanalystsconductaqualitative analysis,examiningtheCDOslegaldocumentationandadjustingratingsforanyrisks associatedwithfeaturesoftheCDO.Intheend,thepublishedratingassignedtothebond issetbyaratingscommittee,whichconsidersboththequantitativeandqualitative analysisoftheCDOsbonds.Asaresult,thepublishedratingcandifferfromthemodeled rating.
20Moodys(2001).

Page 19 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT 3. MonitoringRMBSandCDOs

AfterissuinganinitialratingonanRMBSorCDO,theratingagenciesmonitorthebondsto determinewhethertheyperformasexpectedorinsteadshouldbeeitherdowngradedtoa lowerratingorupgradedtoahigherrating. TomonitorratingsonRMBSonamonthlybasis,Moodysusesquantitativescreenstoflag RMBSwhoseperformancesuggestsaneedtoconsiderchangingtheirratings.21One importanttoolisthepipelinelosscalculation.Thistakesasinputsthefractionof mortgagesinthepoolinvariousstagesofdelinquencyordefault(e.g.,current,30day delinquent,90daydelinquent,inforeclosure,etc.)andusingassumedrollratesatwhich eachdelinquencytypewillultimatelyresultinaloss,anewexpectedlossforthepoolis calculated.Moodysthencalculatesaratioofeachtranchescreditenhancementtothe updatedexpectedlossandflagstranchesforwhichthisratiosuggeststhatthetranches currentratingisinappropriate.Ananalystisthenassignedtodoanindepthreviewofthe bondsandMoodysmaypubliclyannouncethatthebondsareonawatchlistforpossible downgrade.Aratingcommitteeultimatelydecideswhethertodowngradethebonds. MoodysprocedureformonitoringCDOsisdescribedinMoodys(2002).Moodys regularlyreviewswhethertheCDOnotescontinuetoconformtotheguidelinesthatwere thebasisoftheinitialratings.22Itmonitorsmonthlytrusteereportslookingforany deteriorationinthequalityoftheCDOscollateralpool.IfMoodysdetectsanyproblems withtheCDOscollateral,itwillcontacttheCDOsmanagertoinvestigateanddetermine whetheritisnecessarytodowngradetheCDO.IfMoodysdeterminesthatthereisa chancethattheCDOsratingswillchange,theMoodysratingcommitteewillplacethe affectedsecuritiesonawatchlist.ItwillthenproceedtomodelthecashflowsoftheCDO basedonthecurrentcollateralpool,giventheirperformancetodate.Onceithasupdated itsexpectedlossanalysis,theratingcommitteewilldeterminewhetheraratingactionis appropriatefortheCDOssecurities.

D.

MarketShareofMajorRatingAgenciesinRMBSandCDOs

Moodys,S&P,andFitchdominatethemarketforratingbothRMBSandCDOs.Figure4 belowshowsthefractionofnewRMBSthatwereratedbyeachofthethreeRAsineach quarterfrom2000to2007.Themarketsharesaddtogreaterthan100%becausemost RMBSareratedbymorethanoneRA.WhilethethreeRAshadsimilarmarketsharesof

21SeeMoodys(2003b). 22Moodys(2002,p.2).

Page 20 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT between50%and70%in2000,overthenextsevenyearsMoodysandS&Pssharesgrew toover90%,whileFitchssharefelltoaround40%.

Figure5showsthesamemarketshareplotforCDOs.From2003to2004Moodyssharein CDOratingsdroppedfromover90%tobetween60%and70%.Bythethirdquarterof 2005,though,Moodysshareincreasedbackover90%.

Page 21 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

E.

VolumeofRatedRMBSandCDOs

Forsimplicity,wefocushereonthevolumeofratingsbyMoodys,whichasdescribed abovecoveredmostoftheRMBSandABSCDOmarkets.Figures6and7belowdepictthe amountofRMBSandABSCDOs,respectively,newlyratedbyMoodysineachratinglevel foreachquarterfrom2000through2007.Bothmarketsgrewdramaticallyoverthe period.Atotalof72,461tranchesofRMBS,totaling$4.7trillion,wereratedbyMoodys duringthisperiod.TheCDOmarketwassmaller,with5,650tranchesofABSCDOs,totaling $736billion,ratedbyMoodysovertheperiod.23Thebulkofthesebonds42,625 tranchesofRMBS(90%bydollaramount)and2,160tranchesofCDOs(84%bydollar amount)wereratedAaa.

There were also unrated tranches of RMBS and CDOs, typically just the equity tranche of each SPV, issued over this period.

23

Page 22 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Figure6:VolumeofnewRMBSratingsfromMoodys

Page 23 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Figure7:VolumeofnewABSCDOratingsfromMoodys

IV.

CreditRatingsduringtheFinancialCrisis
A. RiseinMortgageDefaults

Mortgagerelatedsecuritiesperformedworsethanexpectedduetoalargefallinhousing pricesandalargeincreaseinmortgagedefaultsbeginningin2006.Nominalhousing pricespeakedin2005andin2006begantodeclinenationally.Figure8showsnational nominalhousingpricesoverthisperiodasmeasuredbytheCaseShillerIndex.Bythe beginningof2007,nationalhousingpricesweredownabout2%fromtheirpeak.Bythe beginningof2008,theyweredownatotalof15%.Bythefirstquarterof2009,housing priceshadfallen31%nationally.

Page 24 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

InOctober2006,withthehousingmarketdownturnwellunderway,Moodys Economy.com,whichisaseparateMoodyssubsidiaryfromtheMoodysunitthatrates structuredproducts,issuedareportauthoredbyChiefEconomistMarkZandititled HousingattheTippingPoint:TheOutlookfortheU.S.ResidentialRealEstateMarket. Basedonastructuraleconometricmodelofhousingsupplyanddemand,thereport concludedthat: Nearly20ofthenationsmetroareaswillexperienceacrashin houseprices:adoubledigitpeaktotroughdeclineinhouse pricesThesesharpdeclinesinhousepricesareexpected alongtheSouthwestcoastofFlorida,inthemetroareasof ArizonaandNevada,inanumberofCaliforniaareas, throughoutthebroadWashington,D.C.area,andinandaround Detroit.Manymoremetroareasareexpectedtoexperience onlyhousepricecorrectionsinwhichpeaktotroughprice declinesremaininthesingledigits.Inadditiontothesome30 metroareasthatarealreadyexperiencingpricedeclines,the structuraleconometricapproachidentified70othermetro areasthatwillsoonexperienceameasurabledeclineinprices.

Page 25 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Itisimportanttonotethatpricedeclinesinvariousmarkets areexpectedtoextendinto2008andeven2009. Withover100metroareasrepresentingnearlyonehalfofthe nationshousingstockexperiencingorabouttoexperience pricedeclines,nationalhousepricesarealsosettodecline. Indeed,oddsarehighthatnationalhousepriceswilldeclinein 2007;thefirstdeclineinnominalnationalhousepricessince theGreatDepression.24 Accompanyingthisdeclineinhousingpriceswasasharpincreaseinmortgage delinquencies.Figure9belowshowsthefractionofmortgagespastdueandinforeclosure overtime.Theriseindelinquencieswasinitiallyconcentratedinsubprimemortgagesand includedanabnormallylargenumberofearlypaymentdefaultsinwhichtheborrower defaultsinthefirstthreemonthsofhismortgage.

B.

DowngradesandImpairmentsofRMBSandCDOs

Tocharacterizewhathappenedtocreditratingsduringthefinancialcrisis,weagainfor simplicityfocusonMoodysratings.Theinitialriseinmortgagedefaultswasconcentrated insubprimemortgagesthathadbeenoriginatedin2005and2006.

24Moodys(2006).

Page 26 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT OnJan.18,2007,MoodysissuedaSpecialReport,EarlyDefaultsRiseinMortgage Securitizations,thatnotedthatmortgagesbackingsecuritiesissuedinlate2005andearly 2006havehadsharplyhigherratesofforeclosurethanpreviouslyissuedsecuritiesat similar,earlypointsintheirlives.25Theseforeclosureswereconcentratedinsubprime mortgagepools. OnMarch7,2007,MoodysissuedaSpecialReport,ChallengingTimesfortheUS SubprimeMortgageMarket,whichstatedthat: Itisgenerallytooearlytopredictultimateperformanceforthe subprimemortgageloansoriginatedin2006andthebonds securedbysuchloans.Anumberoffactorswilldeterminethe ultimatelosses.Homepriceappreciationandrefinancing opportunitiesavailableinthenextfewyearsareexpectedto havethebiggestimpact.Economicfactors,suchasinterest ratesandunemployment,willalsoplayasignificantroleaswill lossmitigationtechniquesemployedbyloanservicers.26 Moodysattributedthepoorperformanceof2006subprimeloansprimarilytotherecent slowdowninhomepriceappreciationandtheintroductionofriskymortgageproducts overthepastseveralyearsNevertheless,Moodysassertedthatwebelievethat performancewouldneedtodeterioratesignificantlyforthevastmajorityofbondsratedA orhighertobeatriskofloss.Onaverage,forlowerratedBaabondstobeatriskofloss, performancewouldhavetocontinuetodeclinematerially.Moodysconcludedthatitis generallytoosoontotellwhetherultimatelosseswillmateriallyexceedouroriginalloss expectationsfor2006securitizedsubprimemortgagepools. OnMarch23,2007,MoodysissuedaSpecialComment,titledTheImpactofSubprime ResidentialMortgageBackedSecuritiesonMoodysRatedStructuredFinanceCDOs:A PreliminaryReview,whichdiscussedtheimplicationsofthesubprimemortgagecrisisfor CDOs.Moodyshadperformedananalysisoftheeffectofhypotheticalscenariosof subprimeRMBSportfoliodeteriorationonCDOs.MoodysfoundthatforCDOsthat containedlargeconcentrationsofRMBSascollateral,thepotentialdowngradeimpacton theCDO[bonds]wassevereinsomecases10ormorenotches.27Forexample,fora CDOwith100%ofitscollateralinsubprimeRMBS,if10%ofsubprimeRMBScollateral immediatelydefaultedwitha20%recoveryandtheremainderwasimmediately
25Moodys(2007a).

26Moodys(March7,2007). 27Moodys(March23,2007,p.2).

Page 27 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT downgradedfournotches,thenMoodysestimatedthatalloftheCDOsbondswouldbe downgraded,withthenonAAAbondsdowngradedtospeculativegrade.28 OnApril20,2007,Moodysissuedareportstatingthat: Eventhoughearlydelinquenciesforthe2006securitizedloans arecloselytrackingthoseofloansbackingdealsissuedin2000 .whichhadcumulativelossesofapproximately6%after72 monthsearlylossesforthe2006loansaretrendinghigher thanthoseofthe2000loans.Whiletheemploymentoutlook todayisstrongerthanthatactuallyexperiencedinthepost 2000period,theoutlookforothermajordriversofmortgage losseshomepriceappreciation,interestrates,and refinancingopportunitiesforsubprimeborrowersfacingrate/ paymentrestsislessfavorable.Asaresult,Moodysis currentlyprojectingthatcumulativelossesforloansbacking 2006subprimesecuritizationswillgenerallyrangebetween 6%and8%,thoughparticularlystrongorpoorperforming poolsmayfalloutsideofthisrange.Barringcumulative losseswellinexcessofcurrentexpectations,wedonotexpect amaterialnumberofdowngradestobondsratedAorhigher. ThefirstmassdowngradeofRMBSoccurredonJuly10,2007.InthemorningS&P announcedthatitwasplacing612subprimeRMBStranchesissuedinlate2005through 2006onwatchforpossibledowngrade.ThatafternoonMoodysfollowedbydowngrading 399tranchesof2006vintagesubprimeRMBSandplacinganadditional32trancheson watchforpossibledowngrade.Thedowngradedsecuritiestotaled$5.3billioninvalueand constituted1.3%of2006vintagefirstlienRMBS.29Allbutoneofthedowngradedtranches wasoriginallyratedBaaorbelow.Moodysalsodowngraded52tranchesofsecondlien subprimeRMBSissuedin2005. Thenextday,onJuly11,2007,Moodysplaced184tranchesofCDOsbackedprimarilyby RMBS,withoriginalfacevalueofapproximately$5billion,onwatchforpossible downgrade.

expectationsaboutthefutureperformanceofsubprimeRMBS. 29MoodysJuly12,2007,teleconferencepresentation,p.12.

28Moodys(March23,2007,p.6).MoodysdidnotintendthescenariostorepresentMoodysactual

Page 28 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT OnJuly12,2007,S&Pdowngraded498ofthe612tranchesithadplacedonwatchtwo daysearlier.ThemajorityofthetrancheswereratedBBBorlower,but8AAArated trancheswereincluded.Atthatpointnominalhousingpriceshadfallenapproximately4% nationallyfromtheirpeakatthebeginningof2006.30 Thesecondhalfof2007sawacontinuedslideinhousingpricesandcontinuedmass downgradesofRMBSandCDOs.Figures10and11belowshowthemonthlyvolumeof downgradesofRMBSandABSCDOs,respectively. Figure10:RMBSdowngradesbyMoodys

30

S&Ps Case-Shiller Index.

Page 29 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Figure11:ABSCDOdowngradesbyMoodys

Bythefirstquarterof2008,housingpriceshadfallen15%fromtheirpeak. Figure12belowshowsthedowngradesandultimateprincipalimpairmentssufferedbythe 2006vintageofRMBSthatwereratedAaaandBaa.Thedashedlinesdepictthe cumulativefractionofeachsetofbondsdowngradedateachpointintime;downgrades indicateMoodyschangedexpectationsaboutthebondsperformance.Thesolidlines depictthecumulativefractionofeachsetofbondsthathavesufferedanyprincipal impairment(i.e.,forwhichpromisedprincipalpaymentsonthebondswerenotmade). TheBaatranchesofRMBSweredowngradedfirst,withaparticularlylargejumpinOctober 2007.Bythemiddleof2008over90%ofBaatrancheshadbeendowngraded. Impairmentsofthosetranchesoccurredafterthedowngrades,withlargejumpsin impairmentsoccurringinApril2008andOctober2008.ImpairmentsofBaatranches

Page 30 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT leveledoutbythemiddleof2009,withover80%oftranchesfailingtomakepromised principalpaymentstoinvestors. DowngradesofAaatranchesofRMBSdidnotbegininearnestuntilthemiddleof2008and continuedsteadilyuntilthemiddleof2009,whentheyleveledoffwithabout80%of tranchesdowngraded.ImpairmentsofAaatranchesoccurredatmuchlowerlevels,witha numberofbondsdefaultinginthespringof2009,bringingthetotalfractionofAaaRMBS impairedtojustunder10%. Figure12:Moodysdowngradesandimpairmentsof2006vintageRMBS.

Figure13belowdepictsthedowngradesandimpairmentsoftheAaaandBaanotesfrom ABSCDOs,whichcontainlargeamountsofRMBSascollateral.BaaCDOdowngradesbegan afewmonthsafterBaaRMBSdowngradesbeganbutotherwisetracktheBaaRMBSclosely. ImpairmentsofBaaCDOstrackdowngradesveryclosely,suggestingthattrancheswere

Page 31 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT beingdowngradedataboutthesametimetheywerefailingtomakeprincipalpayments. AaaCDObondsweredowngradedlaterthantheBaaCDOnotes,butbytheendof2008 morethan90%ofbothAaaandBaaCDOnoteshadbeendowngraded.Over90%ofBaa CDObondsand80%ofAaaCDObondswereultimatelyimpaired. Appendix4containsfiguresthatfurtherbreakoutthedowngradesandimpairmentsof ABSCDOsintovarioussubgroups,suchasmezzanineCDOs,highgradeCDOs,andsynthetic CDOs. Figure13:Moodysdowngradesandimpairmentsof2006vintagecashflowandhybridABSCDOs.

Page 32 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Intotal,$2.5trillionworthofRMBSand$564billionworthofCDOshavebeendowngraded sinceJanuary2007.31 OnewaytoassessthetimingofcreditratingagencydowngradesofRMBSrelativeto contemporaneousmarketviewsregardingtheriskofsubprimemortgagesecuritiesisby comparingthedowngradesof2006vintageRMBStotrendsinthepriceoftheABXindex. TheABXisaseriesofbenchmarkindicesforthepriceofcreditdefaultswaps(CDS)on subprimeRMBS.ACDSisaderivativecontractthatrequiresthebuyertomakeaseriesof paymentstotheprotectionsellerinreturnforprotectionagainsttheeventthatthe referencebonddefaults.LaunchedbyMarkitinJanuary2006,eachABXindexreferences 20RMBSthatareratedbyMoodysandS&Pandissuedinthesixmonthspriortothe launchoftheindex.32Eachindexvintageconsistsoffiveindividualsubindices,each referencingexposurestothesame20underlyingsubprimemortgagesecuritizationsat differentratinglevels:33AAA,AA,A,BBB,BBB.Therefore,eachABXindexreflectsthe tradingpriceofcreditdefaultswapsonRMBSwithinacertainratinglevelfora6month vintage.AdeclineintheABXindexcorrespondstoanincreaseinCDSrates,whichinturn reflectsanincreaseintheprobabilityassessedbyinvestorsthattheunderlyingRMBSwill default. Figures14and15depictthepriceofeachratingsbasedABXindexduring20072010for the2006H2vintageand2007H1vintage,respectively.Theindexforeachvintage representstranchesfromapoolofRMBSoriginatingintheprevioushalfyear.For example,theABX062AAAindexreferencestrancheswithanoriginalratingofAAA originatedinthefirsthalfof2006,whereastheABX071BBBindexreferencestranches withanoriginalratingofBBBoriginatinginthelatterhalfof2006. AsearlyasFebruary2007,the2006H2ABXand2007H1ABXindicesforlowerratings beganexperiencingnoticeabledrops,signalinganexpectedincreaseintherateofdefault onlowinvestmentgradeRMBSoriginatedin2006.OnFebruary27,2007,the2007H1 BBBpriceindexwastradingat67.13,downmorethan30pointsfromthestartingpriceof 100.00.Bycomparison,Figure12showsthattheearliestdateatwhichBaatranchesof RMBSweresignificantlydowngradedwasOctober2007,almosteightmonthsafterthe initialdropinthecorrespondingABXpriceindex.

These are the total principal amount of securities that were ever downgraded; securities that were downgraded multiple times are only counted once. 32MarkitABXMarketingPresentation,January2006. 33Thereferencesecuritiesaresortedintotheseindicesbasedonthelesserofthetworatingsassignedby MoodysandS&P.Oncecreated,indexcompositionremainsfixed,whichmeansthattheunderlyingcredit qualitycanmigratetoratingslowerthanindicatedbytheindexname.

31

Page 33 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

Figure14:ABXHomeEquityindexpricesbyratingforthe2006H2vintage.

Similarly,theAAAABXindicesexperiencedtheirfirstsignificantdeclinesinearlyAugust andlateNovemberof2007,withthe2007H1vintagedecliningmoreprecipitouslythanthe 2006H2vintage.Incontrast,downgradesofAaatranchesofRMBSwerevirtually nonexistentuntilthemiddleof2008.

Page 34 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

Figure15:ABXHomeEquityindexpricesbyratingforthe2007H1vintage.

In2007through2008,thefrequencyofdowngradesofRMBSandCDOsreachedrecord levels.BenmelechandDlugosz(2009)comparedthemassdowngradesofRMBSandCDOs duringthistimetothehistoricalrateofratingchangesoncorporatebonds.Between1983 and1996,thenumberofratedcorporatebondsthatexperienceddowngradesremainedat roughlysimilarproportions.Moreover,theaveragechangeincreditratingsofoutstanding bondsperyearduetoupgradesanddowngradesremainedessentiallystable;from1983to 1996,theaveragedowngradeneverexceeded1.6notchesinanygivenyear.These modesttransitionsinthecreditratingsofcorporatebonds,evenduringmajor macroeconomicevents,contrastwiththerapidityandmagnitudewithwhichRMBSand CDOtranchesweredowngradedin2007and2008.Evenwhencorporatebonds underwentsignificantdowngradingduringthebankruptcywaveandrecessionin2000 2002,wherethenumberofoutstandingbondsdowngradedatleastonceincreasedfrom 12%to30%,theaveragechangeincreditratingwhentherewasanupgradeordowngrade

Page 35 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT wasonly1.8notches.Incontrast,theaveragedowngradeofstructuredfinancesecurities byMoodysin2007and2008was4.7and5.6notches,respectively. Throughout2006andthefirsthalfof2007,Moodyscontinuedtoratelargevolumesof newCDOsandRMBSdespitemarketeventssuggestingacontinuedriseindelinquencyand foreclosureratesandmassdowngradesofCDOsandRMBS.Itwasnotuntilthesummerof 2007,aroundthetimeofthefirstwaveofmassdowngradesofRMBSandCDOs,thatthe amountofnewissuancesbegantodecline.MoodysgaveAaaratingstobillionsofdollars ofnewCDOsandRMBSevenafterthosemassdowngrades.Outofatotalof$119billionin RMBSratedsincethedowngradesofJuly10,2007,90%wereratedAaa.Similarly,outofa totalof$51billioninCDOsMoodysratedsinceJuly10,2007,88%wereratedAaa. Figure16belowshowsthetotalnumberofdollarsofnewlyratedRMBSandCDOsper monthfromthebeginningof2006throughthefirsthalfof2008incomparisontothe progressionofthemortgagecrisisaschronicledbyseveralMoody'sreports.

Page 36 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Figure16:NewlyratedCDOs/RMBSpermonth

Note:Figure16displaysthesumofdollarsissuedinRMBSandCDOs.Theshadedcolorsrepresent proportionsofthetotalamountissuedinRMBSandCDOs,respectively.

ThelastbatchofRMBSandCDOsthatwereratedpriortotherapiddeclineofthehousing marketwereissuedinthesecondhalfof2007.Figures17and18depictthedowngrades andultimateprincipalimpairmentssufferedbythe2007H2vintageofRMBSandCDOs, respectively.WhileMoodysmadesomeadjustmentstotheirratingmethodsduringthe precedingmonthsastheperformanceof2006securitiesdeclinedprecipitously,these figuresshow,incomparisontoFigures12and13,thatthefinalvintageperformedsimilarly poorlyincomparisontoRMBSandCDOsoriginatingfromthepreviousyear.Bymid2009, virtuallyalloftheseRMBSoriginallyratedBaaaswellasAaahadbeendowngraded.

Page 37 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Figure17:Moodysdowngradesandimpairmentsof2007H2vintageRMBS

Page 38 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT Figure18:Moodysdowngradesandimpairmentsof2007H2vintage cashflowandhybridABSCDOs

C.

DowngradesofOtherFinancialInstitutions

InadditiontoratingRMBSandCDOs,thecreditratingagenciesratedthedebtofmany financialinstitutionsthatplayedkeyrolesinthefinancialcrisis.Insomecasestheratings ofthoseinstitutionslaggedmarketevents.InspiteofthewarningsofOctober2006and theirowndowngradesofRMBSinJuly2007,theNRSROsdidnotreevaluatethecompanies thatheldorinsuredthosesecuritiesuntilNovember2007attheearliest,thelone exceptionbeingLehmanBrothers,whichtwooftheagencieshaddowngradedinJune.In thecaseofBearStearns,thefirmenjoyedinvestmentgraderatingsdaysbeforeJPMorgan ChaseacquireditwiththehelpoftheU.S.Treasury.Similarly,thecreditratingagencies gaveLehmanBrothersratingsintheuppermediumrangeofinvestmentgradetheweek beforeitfiledforbankruptcy.Bothfirmscreditdefaultswapshadbeentradingatspreads Page 39 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT consistentwithjunkbonds(Ba1)sinceJuly2007.Uptodaysbeforeitsbailout,AIG receiveduppermediuminvestmentgraderatingsbyallthreemajorcreditratingagencies; Moodys,S&P,andFitchratedthefirmA2,A,andArespectively.34Nevertheless,it warrantsattentionthatapartfromLehmanBrothersnoneofthesefirmseverdefaulted uponitsdebts.Therefore,totheextentthattheagencieswerepresumingthatthese institutionsweretoobigtofail,theirratingsmayhavebeenaccurate.Table3summarizes thecreditratingsofseveralmajorfinancialinstitutionsthedaybeforemajorevents occurred. AmbacandMBIAwerealsodowngradedwellafterthefinancialcrisishadbegun.Eachof thesebondinsurershaddisclosedinpublicfilingsthatithadinsuredbillionsofdollarsin subprimeRMBS,diversifiedCDOs,andCDOsquareds.Nevertheless,inDecember2007 S&PaffirmeditsAaaratingsofthetwocompanies,althoughitassignedthemanegative outlook.Inmakingthisdetermination,S&PcitedthequalityofthefirmsCDO underwriting,thebackendedtimingoftheirpotentialliabilities,andtheiraccessto opportunitiestoraisecapital.Theagencyinturnbaseditsevaluationofthefirms underwritingupontheAAAandsuperAAAratingsoftheirinsuredCDOs.35Theagencies didnottakedefinitiveactiontorevisetheirratingsofthesefirmsuntilJanuaryofthe followingyear,whenS&PandFitchdowngradedAmbacandMBIA,andMoodysplaced themoncreditwatch. Table3:Creditratingsofmajorfirmsattimeofbankruptcy,acquisitionorbailout.
Firm Lehman Bros. AIG Date 9/15/08 Filedforbankruptcy. Event Moodys A2 S&P A Fitch A+

9/16/08

Receivedan$85billionloanfromtheFederal Reserveinexchangefora79.9percentequitystake. Received$20billioninequityandguaranteeson $300billionofitsassetsfromtheUSTreasury. StruckdealtobeacquiredbyBankofAmerica.

A2

Citigroup Merrill Lynch

11/23/08

Aa3

AA

AA

9/14/08

A2

A+

34DavidEvansandCarolineSalas,FlawedCreditRatingsReapProfitsasRegulatorsFail,Bloomberg(Apr. 35SeeDickP.Smith,DavidVeno,andRobertE.Green,DetailedResultsofSubprimeStressTestofFinancial

29,2009).

Guarantors,Standard&PoorsCommentaryReport(Dec.19,2007),14.

Page 40 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

Wachovia 9/29/08 Announcedagovernmentforcedsale to Citigroup (laterWellsFargo). Offereda$25billionloanfor28daysbytheFederal Reserve. A1 A+ A+

3/14/08 Bear Stearns

A2

A+

PurchasedbyJPMorganChasewiththehelpofa 3/16/08 governmentguaranteeonthefirmsmosttoxic securities.

Baa1

BBB

BBB

Source:Bloomberg

Page 41 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

References
Benmelech,EfraimandJenniferDlugosz,2009.TheCreditRatingCrisis,NBERWorking Paper15045. Coval,Joshua,JakubJurek,andErikStafford,2009,TheEconomicsofStructuredFinance, JournalofEconomicPerspectives23(1):325. Moodys,Nov.8,1996,MoodysApproachtoRatingResidentialMortgagePassThrough Securities. Moodys,Jan.19,2001a,MoodysApproachtoRatingResidentialMortgageServicers. Moodys,July11,2001b,TheCombinedUseofQualitativeAnalysisandStatisticalModels intheRatingofSecuritizations. Moodys,Sept.30,2002,MoodysApproach:TheCDOMonitoringProcess. Moodys,April1,2003a,MoodysMortgageMetrics:AModelAnalysisofResidential MortgagePools. Moodys,Aug.13,2003b,OverviewofMoodysRMBSMonitoringProcess. Moodys,May5,2004a,MoodysApproachtoRatingInitialperiod,InterestOnly MortgagesinPrimeRMBS. Moodys,Nov.12,2004b,MoodysApproachtoCodingJumboandAltAResidential MortgageDocumentationPrograms:UpdatedMethodology. MoodysEconomy.com,Oct.2006,HousingattheTippingPoint:TheOutlookfortheU.S. ResidentialRealEstateMarket. Moodys,Dec.11,2008,Announcement:Moodysupdatesitskeyassumptionsforrating structuredfinanceCDOs. Moodys, March 2, 2009, Moodys Approach to Rating SF CDOs.

Page 42 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

Appendix1:MoodysRatingsIdealizedExpectedLossRates
Rating 5yearIdealized ExpectedLossRate Aaa 0.0016% Aa1 0.0171% Aa2 0.0374% Aa3 0.0781% A1 0.1436% A2 0.2569% A3 0.4015% Baa1 0.6050% Baa2 0.8690% Baa3 1.6775% Ba1 2.9040% Ba2 4.6255% Ba3 6.5230% B1 8.8660% B2 11.3905% B3 14.8775% Caa1 19.9726% Caa2 26.8125% Caa3 38.4017% Ca 55.0000% C 100.0000% Source:Moodys(2009).

Page 43 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

Appendix2:SensitivityofMoodysRMBSModeltoPoolCharacteristics

Source:Moodys(1996).

Page 44 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

Appendix3:MoodysModelforRatingABSCDOs
WefocushereonMoodysapproachtoratingcashflowCDOs.Itbeginswithaquantitative analysisoftheperformanceofeachofthetranchesofsecuritiesissuedbytheCDO.36The goalofthequantitativeanalysisistoestimatetheexpectedlossofeachofthetranches.This isdonebyspecifyingasetofscenariosforlossesexperiencedbythecollateralassetsand thento: 1. Specifytheprobabilityofeachlossscenario. 2. Foreachscenario,calculatethelosstoeachtrancheoftheCDO. 3. CalculatetheexpectedlosstoeachtrancheoftheCDObyaveragingtheproductof (a)theprobabilityofeachlossscenarioand(b)thelosstothetrancheunderthat lossscenario. Toestimatethelossdistributionofthecollateralassetpoolinstep1,forstatictransactions MoodysreliesonaGaussiancopulabasedMonteCarlosimulationoftheperformanceof thecollateralpool,whichisastatisticaltechniqueformodelingdefaultcorrelation.Each RMBSinthepoolisassignedadefaultprobabilityandrecoveryrate(i.e.,theexpected amountoftheassetsvaluethatisretainedintheeventtheassetdefaults).Moodys assumptionsabouttherecoveryrateofeachcollateralbondareafunctionofthesector type(e.g.,RMBS,ABS,etc.),thethickness(i.e.,amountofprincipalcontainedin)ofthe bondstranche(thickertrancheshavegreaterrecoveryrates),andthebondsrating (higherratedbondsareassumedtohavehigherrecoveryrates).Thedefaultprobability assumedforanassetisbasedonitsMoodysrating(whichrepresentsexpectedloss)and theassetsassumedrecoveryrate. Formanagedtransactionsorforstatictransactionsthathavenotyetbeenfullyrampedup (i.e.,forwhichnotallofitscollateralhasbeenacquired),Moodysmodelsthecollateral assetportfolioasasetofidenticalcorrelatedbonds,tospecifythecollateralloss distribution.ForthesetypesofCDOs,thecollateralmanagerisboundbycovenantsthat specifyboundsontheultimateweightedaveragecharacteristicsofthecollateralpool, includingitsaveragerating,recoveryrate,andlife.Thesecovenantedaverage characteristicsareusedtospecifytheparametersofthemodel.Thefourparametersof thatdistributionarethe(1)commondefaultprobability;(2)numberofrepresentative assets;(3)commonrecoveryrate;and(4)acorrelationassumption.Whenthecollateral

36Moodys(2005)describesMoodysquantitativeanalysisofcashflowCDOs.

Page 45 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT isacquired,thecollateralmanagerhastocheckthecollateralpoolscorrelationassumption byrunningMoodysmodelinputtingtheCDOsactualcollateral. Oncethecollaterallossdistributionisspecified,eachisrunthroughacashflowmodelthat incorporatestheCDOssubordinationstructuretocalculateanexpectedlosstoeachofthe CDOstranches.

Page 46 of 47

FINANCIALCRISISINQUIRYCOMMISSION CREDITRATINGSANDTHEFINANCIALCRISIS PRELIMINARYSTAFFREPORT

Appendix4:DowngradesandImpairmentsofABSCDOs

Page 47 of 47