November 12, 2010 Elizabeth M. Murphy, Secretary ... - SEC.gov

PF2 Securities Evaluatio ons, Inc. 48 Wall Street, Suiite 1100 New York, NYY 10005 18 4407 T 212 918 4943 F 212 91 www.pf2 2se.com

Novembeer 12, 2010 Elizabeth M. Murphy, Secretary Securitiess and Exchangge Commissio on 100 F Streeet, NE Washington, DC 20549 9‐1090 Dear Ms. Murphy and Commissioneers: Re: Do odd‐Frank Wall Street Refform and Con nsumer Prote ection Act (“D Dodd‐Frank Act,” or “DFA””) Tittle IX Subtitle e C — Improvvements to th he Regulation n of Credit Raating Agencie es (“CRAs”) Thank you u for providin ng a forum forr our comments on the SEEC initiatives under the Dod dd‐Frank Act. PF2 Securrities is a New w York‐based consulting co ompany, whicch was formed in early 200 08 as an independent alternativve for evaluatting and measuring the rissks inherent in n corporate and trust‐ DOs). With each of our preferred collateralized debt obligaation securitiees (corporate and TruPS CD founding members havving spent some part of hiis prior professsional career within a CRA A, we are particularrly sensitive to o the importaance of ratinggs accuracy an nd the deep reliance on raatings through hout our existin ng financial sttructure. Given ourr credit ratingg background and our structured financce expertise, we will limit our commentts to Sections 931 through 939 of the DFFA, and conceentrate on imp provements we believe would serve to o buffer our economy aggainst risks th hat may otherrwise arise fro om the structtured financee market. We recognize the challenges you faace in adoptin ng the requireed rules within the one yeaar time framee from DFA enactment. In this light, we hope thatt our submisssion assists yo ou in meetingg your goals and we encou urage you to contact us freeely should yo ou value furth her communiccation. Sincerely, Guillaumee Fillebeen an nd Gene Philliips Directors

© Copyright 2010 PF2 Securitie es Evaluations, Incc. All rights reserved. www.pf2se.co om


Submissiion Overview w Part 1 — Detailed Com mments parency of Ratings Perform mance (1) Seection 932.(q).(2) – Transp (2) Seection 932.(s)) – Transpareency of Credit Rating Methodologies and Information n Reviewed (3) Seection 933.(b b).(2).(B) – Excceptions to Sttate of Mind in Private Acttions Part 2 — The Case for Minimizing External Distrractions to CR RA Ratings Objectivity



Detailed Comments Section 93 32.(q).(2) – Transparency of Ratings Pe erformance Section C:: For withdraawn ratings, an additional rating attribu ute should be added to ind dicate whetheer the a loss (“SL”), or is aw security was paid‐in‐fu ull (“PIF”), suffered f waiting a loss assessment (“ALA”).1 The PIF measure is imp portant in perrforming due diligence on,, and adding transparencyy to, ratings performance: it allowss a user to ideentify situatio ons in which, for example, unmonitored d speculative‐‐ grade2 seccurities were fully paid down. Merely denoting a rating as havingg been withdrawn tends to obfuscatee the proliferaation of this substantial “TType II” ratingg error — the providing of too low a rating to an issuer that does not subsequeently default.3 Section D: We suggestt for the following three reeasons, amon ng other, thatt the Commission make the referenceed disclosuress available to the public through the Co ommission’s website: (i) th he Commissio on will be ablee to monitor the regularityy of alteration ns being mad de to CRA disclosures; (ii) th he Commissio on will be ablee to supervisee the uniform mity of the datta being delivvered; and (iii) th he users will be afforded the material advantage of being able to access the data from a ceentral, disinte erested, locattion (which ad dditionally en ngenders marrket confidence). 32.(s) – Transsparency of Credit Rating Methodologgies and Inforrmation Revie ewed Section 93 The new disclosure req quirements provide materrial improvem ments and add ditionally advvance the goal of CRA accou untability. Th he language as it stands, however, leaves the CRAs open to providing only thee vaguest of explanation ns in tandem with their acttions, as oppo osed to ensurring the proviision of informativve and meaningful justificaation. For examp ple, disclosingg that “we do owngraded seecurity ABC due to a changge in the way we treat assumptio on XYZ in our model” provvides little bassis for externaal verification n as to the validity of the on change.4 assumptio 1

At the Commission’s determiination, it may be feasible for CRAs to estimate lossess at a certain meassurement date, sayy 30 days after default, and to furtheer distinguish the abovementioned “SL” category into bands SL1 to SL5, for example. SL1 could represent an estimated loss ranging %]; SL2 could repre esent an estimated loss ranging from m (20%‐40%]; etc.. from (0%‐20% 2 peculative grade, synonymous with “junk” or “sub‐invvestment‐grade,” refers to those seccurities rated belo ow the “investmen nt‐ The term sp grade” thresh hold of Baa3, BBB‐‐ and BBB‐ for Moo ody’s, Fitch and S& &P, respectively. 3 e II error, CRAs aree mathematically incentivized to dow wngrade all ratinggs to their lowest rating category, su uch that Absent discclosure of the Type n lowly rated at th he time of default,, resulting in a stro ong accuracy ratio performance read ding for if or when theey do default, theyy would have been the ratings aggency. (Accuracy ratios reflect Typee I errors, alone — the ability for ratiings to adequatelyy separate those securities that do default hat do not.) from those th 4 ution’s purchase of a foreign company proved a creditt positive due to the (geographical) We have also, for example, seen that an institu n benefits afforded while the same rating agency foun nd in a separate siituation that a com mpany’s sale of a foreign arm was diversification


ons, Inc. PF2 Securities Evaluatio 48 Wall Street, Suiite 1100 New York, NYY 10005 18 4407 T 212 918 4943 F 212 91 www.pf2 2se.com

To protecct against poo orly justified or unverifiablee disclosures being made as to the assu umptions app plied and the data relied upo on, the Comm mission oughtt to additionaally require: (i) disclosure of the data ittself; (ii) disclosure as to whetheer the data was provided by an interestted or conflictted party; and (iii) disclosure as to alternative assumpttions that werre disregarded in favor of the chosen assumptio ons. Part (i) solves the prob blem of the CR RAs simply reeferring to wh hat they’re seeeing in the marketplace ass sufficient justification, as an alternaative to beingg able to provide concrete proof. Further, it improvees an investors’’ ability to perform due diligence and faacilitates inveestors’ capacitty to reprodu uce the ratingg. If investors are able to ve erify the accu uracy of CRA ratings, CRAs will be increaasingly pressu ured to mainttain accurate, up‐to‐date raatings. This will encouragee ratings accu uracy. Part (ii) en nables an investor to independently gauge the adeq quacy of the data being rellied upon. ws an external user to deteermine wheth her the data that supportss each Together with part (i), part (ii) allow rating assumption mayy alternativelyy have suppo orted other asssumptions, too. Part (iii) discourages raatings inflatio on: it mitigatees the possibillity that CRAss will choose to apply the on that resultts in the mostt favorable raatings outcom me. Such disclosure, in tandem with thee assumptio specifications of Sectio on 932.(s).(3),, may have beeen powerful enough to have subverted the CPDO failures. Had these req quirements been in place, the rating aggencies that ultimately pro ovided the fau ulty onouncementt: AAA ratings may have been requireed to proffer the following ominous pro “U Under assumpttion X, which we deem to bee reasonable fo or the followingg reasons, we reach a AAA raating or this security; however, if we were to app ply assumption Y, which is not unreasonablee, the resultingg fo raating would have been BBB.”5

Section 93 33.(b).(2).(B) – Exceptionss to State of Mind in Privaate Actions CRAs ought additionallly to be held responsible for “knowingly or recklessly” failing to adequately an nd consistenttly apply their then‐curren nt public meth hodology. Th he application n of asset‐leveel, sector‐wid de and econo omic assumptions ought to be identically applied beetween and among differeent security tyypes.

considered a credit benefit as it allowed the com mpany to focus in greater depth on itts local operationss. These disclosurres, or reasons, run n counter to on ne another but alo one each may seem m plausible. As such, verifiable discllosures would pro ove to be preferable alternatives. 5 ngsreform.wordpre ess.com/2010/02//10/economies‐off‐ratings‐scales‐part‐2/ http://ratin



The Case e for Minimizing Externaal Distractions to CRA Ratings Objecctivity The Finan ncial Standard ds Board (“FSB B”) commentted recently that “[the] ‘haard wiring’ of CRA ratings in standardss and regulations … is a cau use of the ‘cliiff effects’ of the sort expeerienced durin ng the recentt 6 crisis, thro ough which CRA rating dow wngrades can n amplify procyclicality and d cause systeemic disruptio ons.” he “systemic importance” of credit ratin ng agencies (SSection 931.(1)). The DFA, too, notes th Whether or not the FSB’s “Principlees for Reducin ng Reliance on CRA Ratingss” are implem mented — and our n — we will co ontinue to remain utterly opinion on the matter is beyond thee scope of this submission dependen nt on ratings accuracy, from a systemic risk perspecttive, for at leaast the near term.7 We share the Commisssion’s interesst in improving the quality of credit ratings. To this end, we urge the ntrate its enerrgies on ensuring the CRAss focus on thee very seriouss business of Commission to concen providing accurate credit ratings. In n advancing this goal, we recommend the Commissiion consider carefully removing tho ose barriers th hat act as imp pediments to transparent, accurate ratings. Specifically, CRAs have e become incrreasingly com mmercial enteerprises. Indeeed Section 93 31.(3) of the DFA recognizees that “the acctivities of creedit rating agencies are fundamentally commercial in nature.” Copious congressional testimony haas been offerred to support the hypotheesis that ratin ngs quality waas affected by the quest for market sh hare. This deccline in ratinggs standards was not limiteed to market pressuress in the arena of structured d finance.8 Nor were theirr parent comp panies’ short‐‐term profit generatio on interests lim mited to the scope of provviding ratings. Therein liess the root cau use of the problem. Informatiional Asymmetries in the Structured Fiinance Arenaa The Europ pean Commisssion appreciaates the significant role plaayed by CRA opinions in helping to me the informational asymmetry between those issuing debt instrruments and those investiing in “overcom these insttruments.”9 The presence of informatio on asymmetrries, like the absence of reggulation, enables poorly inccentivized parrties (often seellers of risk) to take advan ntage of comparatively lesss sophisticated market paarticipants (offten buyers of risk, or inveestors). 6

http://www w.financialstabilityyboard.org/publicaations/r_101027.p pdf ber of years for maarket participants to develop enhancced risk managem ment In fact, the FSB agrees that “[[in some] cases, it may take a numb capability so as to enable reducced reliance on creedit rating agenciees.” Ibid. 8 ptember 2010 drafft of “How did incrreased competition affect credit ratings?” Becker (Harrvard Business Sch hool) and Milbourn In their Sep ouis) explain that within the contextt of corporate debt ratings, their “interpretation of [th heir research] find dings is (Washington University in St Lo ere intentionally deceiving markets, but instead comp promised ratings quality at the marggin as competition not that the rating agencies we du/research/pdf/0 09‐051.pdf increased.” http://www.hbs.ed 9 nsultation on Credit Rating Agenciess,” November 2010 0 “Public Con http://ec.europa.eu/internal_m market/consultatio ons/docs/2010/cra/cpaper_en.pdf 7



The harmful effects of informationaal asymmetriees are at leastt two‐fold: firrst, they creatte the opporttunity for active adverse selection of collateral portfolios; second, th hey lead to laarge price disccrepancies between fundamental or “intrinsic”” value and trraded levels, in a time of crisis. on: The lack of disclosure inherent in structured finaance securities, in tandem with Active advverse selectio the varyin ng degrees off complexity, enabled the creation of veehicles that permitted selleers to obfuscate the true quality of secu urities being added to und derlying portfo olios. This infformational disconnect allowed (o or encourage e) sophisticateed and techno ologically savvvy institution ns in the know w to take advantagee of less soph histicated, rattings‐reliant “accredited” investors.10 Price discrepancies: Prrice discountss appear to bee most readilyy apparent “ffor those secu urities where the payoff strreams are parrticularly com mplex,” and indeed are “am mplified as wee move down the securitization chain to more opaque securities wh here there is greater scopee for informattion to be asyymmetric.”11 Ch hanging Nature of Ratings Provvided by Large Credit Rating Age ency osition by asset class, Jan. 1988 to Dec. 2009) (Compo Figure (left): ( With the securitization s maarket having grown g exponenttially over time, the informaational asymmettries that exist p posed, and con ntinue to pose, systemic s risk concern ns.

In the stru uctured finan nce market wh hich, as show wn, has grown n to constitutee a significantt portion of all onal gap. Thus, until or unless a debt issuaance, CRA rattings help to bridge the abovementioneed informatio 10

http://expectedloss.blogspo ot.com/2010/06/in n‐due‐diligence‐wee‐trust.html nce Discussion Pap pers “Could Asyymmetric Informattion Alone Have Caused the Collapse of Private‐Label Securitization?” International Finan Number 1010 0 of the Board of Governors of the Federal Reserve Syystem, October 20 010 http://www.ffederalreserve.govv/pubs/ifdp/2010//1010/ifdp1010.pdf 11



ready alteernative exists to CRAs in this space, wee strongly urgge the Commiission to conccentrate its energies on promotingg transparenccy in this spacce, and removving any barriiers to CRAs providing objective opinions.

ernative Rent‐Seeking Arm m s CRAs’ Alte DFA Section 931.(3) recognizes thatt CRAs perforrm “evaluativee and analytical services on behalf of clients” an nd it recognizzes the poten ntially harmful effect of theese for‐profit businesses as it seeks, in Section 93 32, the “separation of ratings from salees and marketting.” We urge the Commissiion to investiggate the conssequences of permitting th hese supplem mentary servicces. We conteend that these e supplementtary services are broader, deeper and more problem matic than generally considered and that they serve as a tangible distracction and matterial impedim ment to both ratings qu uality and ratiings transpareency. Moreoverr, our experie ence has show wn us that in the best of caases, though purportedly separate, “Ratings” and “Analytics” departmeents often apply the same models, and “Ratings” and “Evaluation ns” divisions often rely on the same asssumptions. To the extent investors dep pend on rating agencies fo o r analytics or evaluation ns (or both) in n addition to the ratings th hemselves, those investorss are exposed d to an amplified impact when models or economic assumptions change. In the worst of cases CRAs might bee encouraged d to be particu ularly opaquee about their rating assumptio ons and meth hodologies as a considered d means to geenerate increased rent in their other divisions. Indeed, for certain classees of securitiees in which the existing raters' methodo ologies are nance securities — we havve particularrly lacking in transparency — such as is the case for structured fin noticed an n additional reliance by investors, inclu uding governm ment entities,, on their evaaluations and analytics. Rating ageencies do nott sell only ratiings: they sell data, analytics, advisory services, prices and ns, research and more. evaluation Certain CR RAs are particcularly adept at selling anaalytical tools and consultin ng services in the space of structured d finance: the ey realize that there are lim mited alternaatives availablle, especially given the privvate nature of the market and the subseequent difficu ulty gaining siggnificant coveerage of the issuances thaat uch, the data and analytic vendor comm munity is largeely oligopolistic in nature, with came to market. As su limited su upply lending itself to high costs. In addition to the luster of high h cost data an nd analytics, some CRAs are alive to the significant advvantage it affo ords to a marrket participant if she is ab ble to anticipaate future ratting actions. 122 The rating agencies ofteen market thiss advantage when advertissing that the 12

Estimatingg the frequency or magnitude of ratiing actions is impo ortant for several market participantts. The advantagees include: (a) the calculation off margin and regullatory capital reserve requirements to satisfy credit riisk‐based guidelines; (b) the pricing of certain securities and



model theey’re selling iss the the sam me model used d by the ratin ng analyst who monitors th he security’s rating. In this way and d other ways, “Evaluations” and “Analyttics” departm ments are ablee to charge a premium for their servvices; in otherr ways, their Evaluations and Analytics departmentss are able to offer a premium m product. petitive advan ntage their ratings servicess afford for th he sale of their evaluationss and analyticcs, The comp however, unfortunatelly creates a material distraaction to the quality of theeir original rattings service. As such, wheere rating anaalysts may hisstorically havee opted to provide complimentary tran nsparency — in the form of daata, research and analyticss — in conjun nction with th heir ratings op pinions, the gains sought in their new wer divisions encourage business managgers to ratherr sell those co omplementary tools. The data n priced in succh a way as to o prove prohibitively expeensive for smaaller market and analyytics are often players an nd for regulattors, who are thereby limitted in supervvising their constituents.13 The resultingg comprom mised transparrency, again, favors the larrger and more sophisticateed institution ns as it allows for increased exploitation of market assymmetries. Thus, abseent a pressurre to sell external productss, we envision n CRAs becom ming more traansparent14 in n their assu umptions and their origins,, a situation which will allo ow investors to be better equipped in performin ng their intern nal due diligeence.15 Given thee systemic imp portance of ratings, we beelieve that thee best solutio on is to ensuree a ratings mercial distracction. While we cannot acccurately gaugge the net im mpact environment suffers minimal comm of parent companies’ supplementary divisions on the quality of credit ratings, we arguee strongly thaat they posee a material disservice. Certainly they offer no advan ntages. We implore the Commission to o immediattely require th hat CRAs disassociate, to the extent possible, from their alternatiive divisions. Ideally, CR RAs ought to act as stand‐alone companies, where the quality of their ratings can stand e pressures of shareholderrs, parent insttitutions or su upplementary rent‐seekin ng unaffected by negative businessees that serve to jeopardize or otherwisee compromisee ratings standards. Ratheer than leavin ng our econo omy in a state e that is vulneerable to the snowballing effects of selff‐perpetuatin ng economic forces,16 we ought to ffocus our energies on instituting an independent bufffer against th he realization n of imperfectt ratings. facilities whicch are based off crredit rating pricingg grids; and (c) the analysis of certain n structured finance notes whose performance — nott to bility — can be sub bject entirely to ratting changes madee to the underlying collateral securitties. mention viab 13 on that regulators did not have that information, quitee honestly, [was] equal parts cost an nd According to Prof. Joseph Mason, “[the] reaso interest. Thee cost of market daata on new produccts [was] prohibitivve to performing research on risks to the financial sysstem. For instancee, while ncy in 2006‐7 to reesearch mezzaninee RMBS CDOs with h Intex, the astrono omical cost forced d us to use only paart of working with a regulatory agen ual mortgages [was] priced similarly..” http://www.rou ubini.com/financemarkets‐ the data packkage. Performance data on individu monitor/2580 017/tbtf_is_not_aabout_size_its_abo out_information 14 5/14/rating‐agenciies‐transparent‐orr‐not/ http://ratingsreform.wordprress.com/2010/05 15 uentes, Ph.D., reco ommended the following before thee U.S. Senate vis‐à‐vis the eliminatio on of so‐called blacck‐box ratings: “… based Arturo Cifu n belief among scie entists and engineeers: never trust a ‘result’ that you don't know how to o replicate. In other words, ‘don't ju ust tell on a common d I will check it myyself.’ In short, trusst but verify.” http p://www.ft.com/cms/s/0/384bb0bee‐4d4f‐ me what the result is, tell me how you did it, and 11dd‐b527‐00 00077b07658.htm ml 16 ot.com/2008/11/illiquidity‐self‐perpetuating.html http://expectedloss.blogspo



We urge the Commissiion to engagee the CRAs — at least in th he short term — to help create a transp parent ratings en nvironment which is condu ucive to indep pendent analyysis, and an in nformed inveestment environment which values active du ue diligence, thoughtful rissk managemeent and prudeent lending. With increased transparencyy into CRA daata, assumptio ons and models, an inform med investor community might o gain confide ence in the ussage of ratinggs as complem mentary, objeective opinion ns. be able to Establishm ment of Officce of Credit Ratings or sections off our submissiion, we humb bly recommen nd that in esttablishing thee As it relates to the prio Office of Credit Ratingss — as per Seection 932.(p)) — the Comm mission focuses on candidaates who are familiar with the pros, cons and con nflicts inheren nt in the diffeerent ratings‐pay models, and who posssess an intricatte understanding of their various modees of revenuee generation. Next, the staaffers ought preferablyy to come fro om diverse fin nancial analyssis backgrounds and oughtt ideally to haave degrees of nd their differring methodological appro oaches. Last, but experiencce working wiith the ratingss agencies an equally im mportant, the e candidates selected ough ht to share the opinion thaat, if well adm ministered, rattings can provid de a material public good and they ougght to focus on extracting and maximizing the value that ratings prrovide.



Addendu um — Resou urces www.pf2se.com/Content.aspx?Type=Rese earch) Credit Ratiing Agency Research (available at http://w omies of (Rating gs) Scales Partt 1 February 1, 2010: Econo January 22 2, 2010: A Centralized Solutio on January 8, 2010: “Gamin ng” the Ratingss System, or th he Observer Efffect pecial Report: First Steps Tow ward Real Rating Agency Refo orm Septemberr 14, 2009: Sp bmissions Relevant Regulatory Sub nse to FDIC, FR RB, OCC, OTS’s “Advanced No otice of Proposeed Rules (ANPR R) on Alternatives to August 31,, 2010: Respon atings” Use of Exteernal Credit Ra ostings Related Crredit Rating Po RatingsRefform: www.ratingsreform.wordpress.com Expect[ed]] Loss: http://e expectedloss.b blogspot.com/ssearch/label/R Rating%20Agen ncy%20Reform m
